&EPA
United States
Environmental Protection
Agency
Office of Air and Radiation
Office of Policy
November 1999
EPA-410-R-99-001
The Benefits and Costs
of the Clean Air Act
1990 to 2010
EPA Report to Congress
November 1999
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[This page left blank intentionally.]
-------
Executive
Summary
Section 812 of the Clean Air Act Amendments
of 1990 requires the Environmental Protection
Agency to periodically assess the effect of the Clean
Air Act on the "public health, economy, and envi-
ronment of the United States," and to report the
findings and results of its assessments to the Con-
gress. This Report to Congress, the first of a series
of prospective studies we plan to produce every two
years, presents the results and conclusions of our
analysis of the benefits and costs of the Clean Air
Act during the period from 1990 to 2010. The main
goal of this report is to provide Congress and the
public with comprehensive, up-to-date information
011 the Clean Air Act's social costs and benefits, in-
cluding improvements in human health, welfare, and
ecological resources.
The first report that the EPA created under the
section 812 authority, The Benefits and Costs of the
Clean Air Act: 1970 to 1990, was published and con-
veyed to Congress in October 1997. This retrospec-
tive analysis comprehensively assessed the benefits
and costs of all requirements of the 1970 Clean Air
Act and the 1977 Amendments, up to the passage of
the Clean Air Act Amendments of 1990. The re-
sults of the retrospective analysis showed that the
nation's investment in clean air was more than justi-
fied by the substantial benefits that were gained in
the form of increased health, environmental qual-
ity, and productivity.
The Clean Air Act Amendments of 1990 built
upon the significant progress made by the original
Clean Air Act of 1970 and its 1977 amendments in
improving die nation's air quality-. The amendments
utilized the existing structure of the Clean Air Act,
but strengthened those requirements to tighten and
clarify implementation goals and timing, increase the
stringency of some requirements, revamp the haz-
ardous air pollutant regulatory program, refine and
streamline permitting requirements, and introduce
new programs for the control of acid ram precur-
sors and stratospheric ozone depleting substances.
Because the 1990 Amendments represent an incre-
mental improvement to the nation's clean air pro-
gram, the analysis summarized in this report was
designed to estimate die costs and benefits of die 1990
Amendments incremental to those assessed in the
retrospective analysis. Our intent is that this report
and its predecessor, the retrospective, together pro-
vide a comprehensive assessment of current and ex-
pected future clean air regulatory programs and their
costs and benefits.
This first prospective analysis consists of a se-
quence of six steps. These six steps, listed in order
of completion, are:
(1) estimate air pollutant emissions in 1990,
2000, and 2010;
(2) estimate the cost of emission reductions aris-
ing from the Clean Air Act Amendments;
(3) model air quality based on emissions esti-
mates;
(4) quantify air quality related health and envi-
ronmental effects;
(5) estimate the economic value of cleaner air;
and
(6) aggregate results and characterize uncertain-
ties.
The methodology and results for each step are
summarized below and described in detail in the
chapters of this report.
Air
Estimation of reductions in pollutant emissions
afforded by the 1990 Clean Air Act Amendments
(CAAA) serves as the starting point for this study's
subsequent benefit and cost estimates. We focused
our emissions analysis on six major pollutants: vola-
tile organic compounds (VOCs), nitrogen oxides
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
(NOJ, sulfur dioxide (SO,,), carbon monoxide (CO),
coarse participate matter (PM]0), and fine particu-
late matter (PM2S). For each of these pollutants we
forecast emissions for the years 2000 and 2010 un-
der two different scenarios: a) the Pre-CAAA sce-
nario that assumes no additional control require-
ments would be implemented beyond those that
were in place when the 1990 CAAA were passed;
and b) the Post-CAAA scenario that incorporates
the effects of controls which, when we formulated
the scenario, we expected would be likely to occur
as a result of implementing the 1990 Amendments.
Emissions estimates for both the Pre-CAAA and
Post-CAAA scenarios reflect expected growth in
population, transportation, electric power genera-
tion, and other economic activity by 2000 and 2010.
We compare the emissions estimates under each of
these scenarios to estimate the effect of the CAAA
requirements on future emissions.
The results of the emissions phase of the assess-
ment indicate that the 1990 Clean Air Act Amend-
ments significantly reduce future emissions of air
pollutants. Substantial reductions will be achieved
for the two major precursors of ambient ground-
level ozone: volatile organic compounds (YOCs) and
oxides of nitrogen (NOJ. Relative to the Pre-CAAA
scenario, estimated VOC emissions under the Post-
CAAA case are 35 percent lower by 2010. This
change in emissions is due largely to VOC reduc-
tions from motor vehicles and area sources (e.g., dry
cleaners, commercial bakeries, and other widely dis-
persed sources).
The NOx emission reduction under the Post-
CAAA scenario represents the greatest proportional
emissions change estimated in our analysis. For the
vear 2010, the Post-CAAA NO emissions estimate
• X
is 39 percent lower than the Pre-CAAA estimate,
representing a decrease in emissions of almost 11
million tons. Nearly half of this reduction is from
utilities, largely as a result of the particular NOx
emissions cap and trading program we assumed un-
der the Post-CAAA scenario. The remaining reduc-
tions are attributable to cuts in motor vehicle and
non-utility point source emissions.
Carbon monoxide (CO) emissions contribute
directly to concentrations of carbon monoxide in
the environment. The 2010 Post-CAAA estimate
for CO emissions is 81.9 million tons, 23 percent
lower than the Pre-CAAA projection. The reduc-
tion in CO emissions is mostly due to motor ve-
hicle emission controls.
The CAAA also wfill achieve a substantial re-
duction in precursors of fine particulate matter
(PM25). Sulfur dioxide (SO,,) is an important precur-
sor of PM. By 2010, SO, emissions are 31 percent
lower under the Post-CAAA scenario. Of the 8.2
million ton difference between Pre- and Post-CAAA
SO_ estimates, 96 percent is attributable to additional
control of utility emissions through a national cap-
and-trade program involving marketable SO_ emis-
sion allowances. Oxides of nitrogen, discussed above,
are also important fine PM precursors.
We project the 1990 Clean Air Act Amendments
to have more modest effects on emissions of par-
ticulate material which is emitted in solid form (i.e.,
"primary" or "direct" PM10 and PM25 emissions).
Overall, emissions of primary PM]0 and PM,g arc
each approximately four percent lower in 2010 un-
der the Post-CAAA scenario than under the Pre-
CAAA scenario. Although the incremental effects
of the Clean Air Act Amendments on primary PM
emissions will be relatively small, PM in the atmo-
sphere is comprised of both directly emitted primary
particles and particles that form in the atmosphere
through secondary processes as a result of emissions
of SO,, NO , and organic compounds. These PM
species, formed by the conversion of gaseous pollut-
ant emissions, are referred to collectively as "second-
ary" PM. Because, as noted above, the 1990 Amend-
ments achieve substantial reductions in these gaseous
precursor emissions, the Amendments have a much
larger effect 011 PM]0 and PM25 levels in the atmo-
sphere than might be apparent if only the changes
in directly emitted primary particles are considered.
Our estimate of the costs of the Clean Air Act
Amendment provisions is based on an evaluation of
the increases in expenditures incurred by various
entities to meet the additional control requirements
incorporated in the Post-CAAA case. These costs
include operation and maintenance (O&M) expen-
ditures —which includes research and development
(R&D) and other similarly recurring expenditures—
plus amortized capital costs (i.e., depreciation plus
-------
Executive Summary
interest costs associated with the existing capital
stock). Relative to the Pre-CAAA case, Post-CAAA
scenario total annual compliance costs for Titles I
through V are approximately $19 billion higher by
the year 2000, rising to |27 billion by the year 2010.
Compliance with Title I, Provisions for Attain-
ment and Maintenance of National Ambient Air
Quality Standards (NAAQS), accounts for §14.5
billion, or over half, of the estimated increase in year
2010 compliance costs. Compliance with mobile
source emissions control provisions under Title II
of the Clean Air Act Amendments accounts for an
additional 30 percent of the total costs, or S9 billion
annually by 2010. Provisions to control acid depo-
sition and emissions of stratospheric ozone deplet-
ing substances account for most of the remainder of
the costs.
These direct compliance costs provide a good,
but incomplete, measure of the total effect of the
Clean Air Act Amendments 011 the U.S. economy.
A complete picture of the indirect impacts of these
costs would include changes in employment and
prices as well as impacts that might be experienced
among customers of the firms that must incur these
costs. While these indirect effects could be impor-
tant, we believe the direct cost estimates provide a
good initial measure of the effect of the Clean Air
Act Amendments on the U.S. economy, as well as
an appropriate metric for comparison with the di-
rect benefits reported here.
Table ES-1
Summary Comparison of Benefits and Costs (Estimates in millions 1990$)
Titles I through V
Annual Estimates
2000
2010
Monetized Direct Costs:
Low3
Central
High3
$19,000
$27,000
Monetized Direct Benefits:
Lowb
Central
Highb
$16,000
$71,000
$160,000
$26,000
$110,000
$270,000
Net Benefits:
Low
Central
High
($3,000)
$52,000
$140,000
($1,000)
$83,000
$240,000
Benefit/Cost Ratio:
Lowc
Central
Highc
less than 1/1
4/1
more than 8/1
less than 1/1
4/1
more than 10/1
aThe cost estimates for this analysis are based on assumptions about future changes in factors such as consumption
patterns, input costs, and technological innovation. We recognize that these assumptions introduce significant
uncertainty into the cost results; however the degree of uncertainty or bias associated with many of the key factors cannot
be reliably quantified. Thus, we are unable to present specific low and high cost estimates.
b Low and high benefits estimates are based on primary results and correspond to 5th and 95th percentile results from
statistical uncertainty analysis, incorporating uncertainties in physical effects and valuation steps of benefits analysis.
Other significant sources of uncertainty not reflected include the value of unqualified or unmonetized benefits that are
not captured in the primary estimates and uncertainties in emissions and air quality modeling.
0 The low benefit/cost ratio reflects the ratio of the low benefits estimate to the central costs estimate, while the high ratio
reflects the ratio of the high benefits estimate to the central costs estimate. Because we were unable to reliably quantify
the uncertainty in cost estimates, we present the low estimate as "less than X," and the high estimate as "more than Y",
where X and Y are the low and high benefit/cost ratios, respectively.
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
To estimate benefits, the results of the emissions
analysis served as the principal input to a linked se-
ries of models. We used these models to estimate
changes in air quality, human health effects, ecologi-
cal effects, and, ultimately, the net economic ben-
efits of the Clean Air Act Amendments. The goals
of these steps in the analysis were to estimate the
implications of changes in emissions resulting from
compliance with the Clean Air Act Amendments
on criteria pollutant air quality throughout the lower
48 states, and the impacts on human health and the
environment that result from these changes.
We focused our air quality modeling efforts on
estimating the impact of Prc- and Post-CAAA emis-
sions on ambient concentrations of ozone, PM1(),
PM_., SO,, NOx, and CO and on acid deposition
and visibility in future years. We found that the
majority of the total monetized benefits, however,
is attributable to changes in participate matter con-
centrations and, more specifically, to the effect of
these ambient air quality changes on avoidance of
premature mortality. We estimate that 2010 Post-
CAAA PM10 and PM25 concentrations in the east-
ern U.S. will average about 5 to 10 percent lower
than 2010 Pre-CAAA concentrations, with some
areas of the eastern U.S. experiencing much greater
reductions of up to 30 percent. The air quality mod-
eling also indicates a substantial overall reduction in
future-year PM]0 and PM,5 concentrations through-
out the western U.S., including most population
centers, following implementation of the Clean Air
Act Amendments.
The direct benefits of the air quality improve-
ments we estimated under the Post-CAAA scenario
include reduced incidence of a number of adverse
human health effects, improvements in visibility, and
avoided damage to agricultural crops. The estimated
annual economic value of these benefits in the year
2010 ranges from $26 to $270 billion, in 1990 dol-
lars, with a central estimate, or mean, of $110 bil-
lion. These estimates do not include a number of
other potentially important effects which could not
be readily quantified and monetized (i.e., converted
to dollar terms). These excluded effects include a
wide range of ecosystem changes, air toxics-related
human health effects, and a number of additional
health effects associated with criteria pollutants.
In addition, these results reflect the particular
choices wfe made with respect to interpretations of
the available scientific and economic literature and
adoption of paradigms for representing health and
environmental changes in economic terms. We re-
fer to these results, then, as our "primary" estimates;
however, in the text of this report we also present
some alternative results which reflect other available
choices for models or assumptions.
One particularly important assumption of our
primary analysis is that correlations between in-
creased air pollution exposures and adverse health
outcomes found by epidemiological studies indicate
causal relationships between the pollutant exposures
and the adverse health effects. Future research may
lead to revisions in this assumption as well as other
key assumptions, data, and models we use to esti-
mate the benefits and costs of the Clean Air Act.
Such revisions may in turn imply significant changes
111 the estimates of Clean Air Act costs and benefits
presented here and in past and future assessments.
In our judgment, however, the primary results re-
flect the best currently available science and the most
up-to-date tools and data we had at our disposal —
and the most reasonable assumptions we could
adopt— as each step of die analysis was implemented.
Cleaner air also yields benefits to ecological sys-
tems. This first section 812 prospective analysis de-
votes a great deal of effort to characterizing and,
where possible, quantifying and monetizing the im-
pacts of air pollutants 011 natural systems. Our in-
creased effort is 111 part a result of the findings of the
retrospective analysis, where we identified a better
understanding of ecological effects as an important
research direction for the first prospective and sub-
sequent analyses. Quantified benefits of CAAA pro-
grams reflected in the overall monetized benefits
include: increased agricultural and timber yields; re-
duced effects of acid rain on aquatic ecosystems; and
reduced effects of nitrogen deposited to coastal estu-
aries. Many ecological benefits, however, remain
difficult or impossible to quantify, or can only be
quantified for a limited geographic area. The mag-
nitude of quantified benefits and the wide range of
unquantified benefits nonetheless suggest that as we
learn more about ecological systems and can con-
duct more comprehensive ecological benefits assess-
ments, estimates of these benefits could be substan-
tially greater.
IV
-------
Executive Summary
We developed separate estimates for the Title
VI provisions of the CAAA designed to protect
stratospheric ozone. Stratospheric ozone is the layer
of the atmosphere that protects the planet from the
harmful effects of ultraviolet radiation (UV-b). Our
primary estimate of the cumulative benefits of Title
VI is $530 billion. Using the same uncertainty esti-
mation procedure as for other parts of the analysis,
we estimate Primary Low and Primary High esti-
mates of 8100 billion to $900 billion, respectively.
These estimates partially reflect potential averting
behaviors, such as remaining indoors or increasing
use of sunscreens or hats, which may mitigate the
effects of the UV-b exposure increases estimated in
the Pre-CAAA case.
to
Based on the specific tools and techniques we
employed, our primary estimate of the net benefit
(benefits minus costs) over the entire 1990 to 2010
period of the additional criteria pollutant control
programs incorporated in the Post-CAAA case is
S510 billion. Our results imply that the monetizable
benefits alone exceeded the direct compliance costs
by four to one. For many of the factors contribut-
ing to this net benefit estimate (especially physical
effects and economic valuation estimates), we were
able to generate quantitative estimates of uncertainty.
By statistically combining these uncertain estimates,
we were able to develop a range of net benefit esti-
mates which provide a partial indication of the over-
all uncertainty surrounding the central estimate of
net benefits. This range, reflecting a 90 percent prob-
ability range around the mean, or central estimate,
is negative $20 billion (implying a small probability
that costs could exceed monetized benefits) to posi-
tive $1.4 trillion.
The estimates for Title VI also indicate that cu-
mulative benefits ($500 billion) well exceed cumula-
tive costs ($27 billion). The time period of our Title
VI analysis (175 years) suggests that these estimates
are very uncertain. Nonetheless, the conclusion that
benefits well exceed costs holds even at our Primary
Low estimate of benefits (the low end of the 90 per-
cent probability range, or $100 billion), and regard-
less of discount rate used to generate the cumulative
estimates from the perspective of the present.
The assumptions necessitated by data limitations,
by the current state of the art in each phase of the
analytical approach, by the need to predict future
conditions, and by the state of current research 011
air pollution's effects imply that both the mean esti-
mate and the 90 percent probability range around
the central estimate are uncertain. While alterna-
tive choices for data, models, modeling assumptions,
and valuation paradigms may yield results outside
the range projected in our primary analysis, we be-
lieve based on the magnitude of the difference be-
tween the estimated benefits and costs that it is un-
likely that eliminating uncertainties or adopting rea-
sonable alternative assumptions would change the
fundamental conclusion of this study: the Clean Air
Act Amendments' total benefits to society exceed
its costs.
The uncertainties in the primary estimates and
the controversies which persist regarding model
choices and valuation paradigms nonetheless high-
light the need for a variety of new and continued
research efforts. Based on the findings of this study,
the highest priority research needs are:
• Improved emissions inventories and inven-
tory management systems
• A more geographically comprehensive air
quality monitoring network, particularly for
fine particles and hazardous air pollutants
* Use of integrated air quality modeling tools
based on an open, consistent model archi-
tecture
* Development of tools and data to assess the
significance of wetland, aquatic, and terres-
trial ecosystem changes associated with air
pollution
• Increased basic and targeted research on the
health effects of air pollution, especially par-
ticulate matter
* Continued development of economic valu-
ation methods and data, particularly valua-
tion of changes in risks of premature mor-
tality associated with air pollution
Properly directed and funded, such research
would improve the results of future analyses of the
benefits and costs of the Clean Air Act.
The CAA requires "EPA to consult with an out-
side panel of experts during the development and
interpretation of the 812 studies. This panel of ex-
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
perts was organized in 1991 under the auspices of
EPA's Science Advisory Board (SAB) as the Advi-
sory Council 011 Clean Air Act Compliance Analy-
sis (hereafter, the Council). Organizing the review
committee under the SAB ensured that highly quali-
fied experts would review the section 812 studies in
an objective, rigorous, and publicly open manner
consistent with the requirements and procedures of
the Federal Advisory Committee Act (FACA).
Council review of the present study began in 1993
with a review of the analytical design plan. Since
the initial June 1993 meeting, the Council has met
many times to review proposed data, proposed meth-
odologies, and interim results. While the full Coun-
cil retains overall review responsibility for the sec-
tion 812 studies, some specific issues concerning
physical effects and air quality modeling were re-
ferred to subcommittees comprised of both Council
members and members of other SAB committees.
The Council's Health and Ecological Effects Sub-
committee (HEES) met several times and provided
its own review findings to the full Council. Simi-
larly, the Council's Air Quality Modeling Subcom-
mittee (AQMS) held in-person and teleconference
meetings to review methodology proposals and
modeling results and conveyed its review recommen-
dations to the parent committee.
An interagency review was conducted, during
which a number of analytical issues were discussed.
Conducting a benefit/cost analysis of a major stat-
ute such as the Clean Air Act requires scores of meth-
odological decisions. Many of these issues are the
subject of continuing discussion within the economic
and policy analysis communities and within the
Administration. Key issues include the treatment
of uncertainty in the relationship between particu-
late matter exposure and mortality; the valuation of
premature mortality; the treatment of tax interac-
tion effects; the assessment of stratospheric ozone
recovery; and the treatment of ecological and wel-
fare effects. These issues could not be resolved within
the constraints of tins review. Thus, this report re-
flects the findings of the EPA and not necessarily
other agencies of the Administration.
VI
-------
Table of Contents
Summaiy[[[ /
Air Pollutant Emissions i
Compliance Costs ii
Human Health and Environmental Benefits iv
Comparing Costs to Benefits v
Review Process v
1: Introduction[[[ 1
Background and Purpose 1
Relationship of This Report to Other Regulatory Analyses 1
Requirements of the 1990 Clean Air Act Amendments 2
Analytical Design and Review 3
'Target loanable [[[ 3
Key Assumptions 3
Analytic Sequence 4
Review Process[[[ 6
Report Organization [[[ 7
2: [[[ 9
Overview Of Approach[[[ 9
Scenario Development 11
Emissions Estimation Results[[[ 11
Comparison of Emissions Estimates With Other Existing Data 18
Uncertainty In Emission Estimates [[[ 19
3; [[[ 23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Air Quality Model Results[[[ 39
O^one 39
P articulate Matter[[[ 41
Visibility 42
Acid Deposition[[[ 45
S02, NO, N02, and CO 45
Uncertainty in the Air Quality Estimates 47
5: of .......................... 51
Analytical Approach 51
Air Quality [[[ 51
Population 52
Concentration-Response functions 52
Key Analytical Assumptions [[[ 52
Exposure Analysis 54
Selection and Application
of C-R Functions 55
PMAlelated Mortality 57
Health Effects Modeling Results [[[ 60
Avoided Premature Mortality Estimates 60
Non-Fatal Health Impacts [[[ 62
Avoided Health Effects of Other Pollutants [[[ 62
Avoided Effects oj Air Toxics 62
Avoided Health Effects for Provisions to Protect Stratospheric O^one........................................... 6?
Uncertainty in the Health Effects Analysis [[[ 65
Effects....................... 69
Valuation of Benefit Estimates [[[ 69
Valuation oj Premature Mortality 70
Valuation of Specific Health 'Effects[[[ 72
Stratospheric 0%pne Provisions 74
Results of Benefits Valuation 74
Valuation Uncertainties 75
Mortality Risk Benefits Transfer[[[ 76
7: and .................................... 81
-------
Table of Contents
Summary of Quantitative Results [[[ 97
Uncertainty 98
8: of and Benefits,,,,,,,,,,,,,......,,,,,,,,,,,,,,,,,,,... 99
Monetized Benefits of the CAAA[[[ 99
Overview of Benefits Analyses 99
Summary of Monetised Benefits for Human Health and Welfare Effects.................................... 100
Annual Beneji/s Estimates 101
Aggregate Monetised Rene/its [[[ 103
Aggregate Benefits of Title VI Provisions 103
Comparison of .Monetized Benefits and Costs 104
Cost-Effectiveness Evaluation[[[ 106
Major Sources of Uncertainty 106
Quantitative Analysis of Physical Effects and Valuation Uncertainties 107
Measurement Error and Uncertainty in Direct Cost Inputs 109
PM Mortality Valuation Based on Life-Years East 109
PM Mortality Incidence Using the Dockery Study[[[ 110
Uncertainties in Title VI Health Benefits Analysis 110
Uncertainties in Emissions and Air Quality Steps[[[ / / /
Omission of Potentially Important Benefits (Categories 113
Alternative Discount Kates[[[ 113
Scenario Development[[[ A-2
Comparison of the Base Year Inventory and Emissions Projections
with Other Existing Data[[[ A-7
Posl-C-AAA Emissions Estimates and EPA Trends Data A-7
Prospective Analysis and GCVTC Emissions Estimates[[[ A-12
Prospective Analysis PM2.5 Emissions Estimates and Observed Data A-13
Industrial Point Sources A-14
Overview of Approach [[[ A-14
Base Year Emissions A-15
Growth Projections[[[ A-15
Control Scenarios A-16
-Emissions Summary [[[ A-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Growth Projections A-30
Control Scenarios A-30
Emissions Summary A-31
Area Sources A-35
Overview of Approach [[[ A-35
Base Year Emissions A-35
Growth Projections A.-36
Control Scenarios A-37
Emissions Summary [[[-A-37
Reasonable Further Progress Requirements A-41
Mercury Emission Estimates [[[ A-48
Medical Waste Incinerators (MW1) A-48
Municipal Waste Combustors (MWCs) [[[ A48
Electric Utility Generation A-49
Hazardous Waste Combustion[[[ A-49
Chlor-alkali Plants A-49
Uncertainties in the Emission Estimates A-51
Base Year Emission Estimates[[[ A-51
Growth Forecasts A-52
Future Year Control Assumptions[[[ A-5'3
References[[[A-55
B: Costs[[[ B-1
Introduction[[[ B-1
Summary of Methods [[[ B-1
EKCAM Model B-1
IPM Model[[[ B-2
Additional Methods B-2
Annali'^ation of Costs[[[ B-2
-------
Table of Contents
C: Air [[[ C-1
Introduction[[[ C-1
Overview of Section 812 Prospective Modeling Analysis[[[C-2
-Air Quality Models and Databases C-2
Methodology for the Combined Use of Observations and .Air Quality Modeling Results......... C-3
Estimating the Effects of the CAAA on Ozone Air Quality C-4
Overview of the UAM and UAM-V Photochemical Modeling Systems.............................. C-5
UAM C-5
UAM-V................................................^^^
Regional-Scale Modeling of the Eastern U.S C-6
Regional-Scale Modeling of the Western U.S. [[[ C-12
Urban-Scale Modeling of the San Francisco Bay Area C-15
Urban-Scale Modeling of the Los Angeles Area C-18
Urban-Scale Modeling of the Maricopa County (Phoenix Area) C-20
Calculation of Ozone Air Quality Profiles C-23
Overview of the Methodology [[[ C-23
Description of the Observation Dataset C-23
Calculation of Percentile-Based Adjusted factors C-24
Use of Adjustment Factors to Modify Observed Concentrations C-25
Calculation of 0-^one Profiles[[[ C-25
Estimating the Effects of the CAAA on Particulate Matter C-38
Overview of the RADM/RPM Modeling System C-38
Application of RADM/RPM for the Eastern U.S C-38
Overview of the REMSAD Modeling System C-41
Application of REMSAD for the Western U.S. [[[ C-44
Calculation of PM Air Quality Profiles C-48
Estimating the Effects of the CAAA on Visibility C-64
RADM/RPM and Visibility C-64
'RADM/'RPM Modeling RW//&T[[[ C-64
REMSAD and Visibility[[[ C-66
REMSAD Modeling Results C-66
Acid Deposition C-69
Overview of the RADM Modeling System [[[ C-69
RADM Modeling Results C-69
Estimating the Effects of the CAAA on Sulfur Dioxide,
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Types of Health Studies [[[ D-5
Chamber Studies [[[ D-5
Epidemiological Studies [[[ D-6
Selection of C-R Functions [[[ D-7
C-R Function General Issues [[[ D-7
C-R Function Selection Criteria [[[ D-12
Mortality [[[ D-16
Chronic illness [[[ D-26
Hospital Administration [[[ D-30
Minor ///KWJ-[[[^
C-R Functions Linking Air Pollution and Adverse Health Effects .................................... D-58
Carbon Monoxide [[[ D-58
Nitrogen Dioxide [[[ D-61
Osyne [[[ D-64
P articulate Ma^r [[[ ..D-70
Sulfur Dioxide [[[ D-80
Modeling Results [[[ D-82
Uncertainty [[[ D-82
Sensitivity Analyses [[[ D-83
References [[[ D-93
E; of Criteria Pollutants.. .......................... E-1
Introduction [[[ E-1
Ecological Overview of the Impacts of Air Pollutants Regulated by the CAAA ............... E-3
Effects of Atomospheric Pollutants on Natural Systems [[[ E-3
Acidic Deposition [[[ E-4
Nitrogen Deposition [[[ .E-4
Hazardous Air Pollutant Deposition [[[ E-5
Troposheric O^one [[[ "Fi-7
-------
Table of Contents
Timber Production Impacts From Tropospheric Ozone [[[ E-45
Ecological Effects of O^pne E-45
Modeling Timber Impacts from 0-^one[[[ E4-6
Ecological Results E-47
Economic Impacts[[[ E-48
Caveats and Uncertainties E-48
Carbon Sequestration Effects E-50
Caveats and Uncertainties E-52
Aesthetic Degradation of Forests E-52
Forest Aesthetic Effects front Air 'Pollutants[[[ E-53
Economic Value of Changes in Forest Aesthetics E-58
Extending Economic Estimates to a broader Area E-59
Caveats and Uncertainties E-61
Toxification of Freshwater Fisheries E-61
Impacts of Toxic Air Emissions[[[ E-62
Illustration of Economic Cost to Anglers E-63
Caveats and Uncertainties E-65
(Conclusion and Implications [[[ E-65
Summary of Quantitative Results E-66
R£commendations of Future Research [[[ E-68
References[[[ E-70
F: on ..................... F-1
Introduction[[[F-l
Ozone Concentration Data [[[F-l
Calculation of the SUM06 Statistic F-2
October to April O^one Concentration Data[[[ F-2
Yield Change Estimates F-3
Exposure-Response Functions F-3
Calculation of O^one Indices F-5
Calculation of County Weights F-5
Calculation of Percent Change in Yield[[[ F-6
Economic Impact Estimates F-7
Conclusions [[[F-8
References F-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Cost and Benefits Results with Adjusted Parameters[[[ G-25
Five Percent Discount Ra/e G-26
Three Percent and Seven 'Percent Sensitivity 'Vests [[[ G-27
Two Percent Discount Rate G-30
Undiscounted "Benefits[[[G-30
Limitations and Uncertainties[[[ G-34
'Long-Term Discounting G-34
Corfj[[[^
Benefits G-37
References G-39
H: of of
Pollutants[[[H-1
Methods Used to Value Health and Welfare Effects H-1
Valuation of Specific Health Endpoints H-3
Valuation of Premature Mortality Avoided [[[ H-3
Mortality Valuation Methodologies H-6
Valuation Strategy Chosen for this Analysis[[[Vi-11
Valuation of Hospital Admissions Avoided H-14
Valuation of Chronic Bronchitis A voided[[[ 11-15
Valuation of Chronic Asthma Avoided H-16
Valuation of Other .Morbidity Endpoints Avoided H-17
Valuation of Welfare Effects [[[ 11-18
Visibility Valuation H-18
Results of Valuation of Health and Welfare Effects H-27
Uncertainties in the Valuation Estimates H-31
Relative Importance of Different Components of Uncertainty[[[ 11-3 /
-------
Tables
Table .ES-1 Summary Comparison of Benefits and Costs (Estimates in millions 1990$) in
Table 2-1 .Major Emissions Source Categories 10
Table 2-2 Summary of National Annual Emissions Projections 12
Table 2-3 Summary of Source Category of National Annual Emission Projections to 2010
(thousand tons) 14
Table 2-4 Airborne Mercury Emission Estimates 15
Table 2-5 Key Uncertainties Associated with Emissions Estimation 21
Table 3-1 Summary of Direct Costs for Titles 1 to V of CAAA, By Title and Selected
Provisions 26
Table 3-2 Results of Quantitative Sensitivity Tests 32
Table 3-3 Key Uncertainty Associated with Cost Estimation 33
Table 4-1 Overview of Air Quality Models 37
Table 4-2 Comparison of Visibility in Selected Eastern Urban Areas 43
Table 4-3 Comparison of Visibility in Selected Eastern .National Parks 43
Table 4-4 Comparison of Visibility in Selected Western Urban Areas 44
Table 4-5 Comparison of Visibility in Selected Western .National Parks 44
Table 4-6 Median Values of the Distribution of ratios of 2010 Post-CAAA/Pre-CAAA
Adjustment Factors 46
Table 4-7 Key Uncertainties Associated with Air Quality Modeling 48
Table 5-1 Human Health Effects of Criteria Pollutants 53
Table 5-2 Summary of Considerations Used in Selecting C-R Functions 56
Table 5-3 Change in Incidence of Adverse Health Effects Associated
with Criteria Pollutants in 2010 (Pre-CAAA minus Post-CAAA) - 48 State U.S.
Population (avoided cases per year) 61
Table 5-4 Mortality Distribution by Age in Primary Analysis (2010 only),
Based on Pope et al. (1995) 62
Table 5-5 Major Health Benefits of provisions to Protect Stratospheric Ozone 64
Table 5-6 Key Uncertainties Associated with Human Health Effects Modeling 65
Table 6-1 Health Effects Unit Valuation (1990 Dollars) 70
Table 6-2 Summary of Mortality Valuation Estimates (millions of $1990) 72
Table 6-3 Results of Human Health Benefits Valuation, Post-CAAA 2010 75
Table 6-4 Valuation of CAAA Benefits: Potential Sources and Likely Direction of Bias 76
Table 6-5 Key Uncertainties Associated with Valuation of Health Benefits 79
Table 7-1 Classes of Pollutants and Ecological Effects 83
Table 7-2 Interactions of Mercury and. Ozone with Natural Systems at Various Levels of
Organization 84
xv
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 7-3 Interactions Between Nitrogen Deposition and Natural Systems
at Various Levels of Organization 85
Table 7-4 Interactions Between Acid Deposition -and Natural Systems
at Various Levels of Organization 86
Table 7-5 Ecological Effects of Air Pollutant[[[ 88
Table 7-6 Summary of Endpomts Selected For Quantitative Analysis 89
Table 7-7 Estimated Displaced Costs For Three Estuaries[[[ 91
Table 7-8 Annual Economic Impact of Acidification in 2010 92
Table 7-9 Quantified and Unquantified Ecological and Welfare Effects
of Title VI Provisions 96
Table 7-10 Summary of Evaluated Ecological Benefits [[[ 97
Table 7-11 Summary of Other Welfare Benefits 97
Table 7-12 Key Uncertainties Associated with Ecological Effects Estimation................................. 98
Table 8-1 Criteria Pollutant Health and Welfare Benefits in 2010.................................................. 102
Table 8-2 Present Value of Monetized Benefits for 48 State Population 103
Table 8-3 Summary of Quantified primary Central Estimate Benefits and Costs ...................... 104
Table 8-4 Summary comparison of Benefits and Costs 105
Table 8-5 Summary of Impact of Alternative Methods for Calculating Costs and Benefits..... 108
Table 8-6 Effect of Alternative Discount Rates on Primary Central Estimates 114
Table A-l Base Year Inventory - Summary of Approach A-3
Table A-2 Analysis Approach By Major Sector[[[ A-4
Table A-3 Projection Scenario Summary By Major Sector A-5
Table A-4 Comparison of Emissions: Prospective Analysis and GCVTC Study ..................... A-l 3
Table A-5 Industrial Point Source Control Assumptions For The Post-CAAA Scenario A-l 7
Table A-6 Industrial Point Source Emission Summaries By Pollutant
For 1990, 2000, and 2010 A-19
Table A-7 Utility Emission Summary[[[ A-23
Table A-8 BEA Growth Forecasts by Major Source Category: Nonroad Engines/Vehicles ..A-26
Table A-9 Nonroad National Emission Projections By Source Category ...................................A-28
Table A-10 Applicability of Mobile Source Control Programs A-32
Table A-ll National Highway Vehicle Emissions By Vehicle Type...............................................A-33
Table A-12 Area Source Emission Summary By Pollutant For 1990, 2000, and 2010 A-39
Table A-l3 2000 Rate of Progress Analysis [[[ A-43
Table A-14 2010 Rate of Progress Analysis A-45
Table A-15 Discretionary Control Measures Modeled For ROP/RFP ..........................................A-47
Table A-16 Airborne Mercury Emission Estimates A-51
Table B-l Summary of Cost Estimates By Emissions Source B-3
-------
List of Tables and Figures
Table 13-8 Cost Estimates of Motor Vehicle Program[[[ B-17
Table B-9 Cost Summary of Area Source NOx and PM Controls B-19
Table B-10 2000 Rate of Progress Analysis [[[ B-21
Table B-ll 2010 Rate of Progress Analysis B-22
Table B-12 Summary of Cost Estimates By CAAA Title[[[ B-24
Table B-13 Title I National Rules, Point, and Area Source VOC Control Costs B-26
Table B-14 Summary of Costs For Title I[[[ B-26
Table B-15 Summary of Title II Motor Vehicle and Non-road
Engine/Vehicle Program Costs [[[ B-27
Table B-16 Title III, MACT Standards, Point and Area Source VOC Control Costs B-28
Table B-17 Annual Cost of Title IV [[[ B-29
Table B-18 Potential Effects of Uncertainty' on Cost Estimates B-33
Table B-19 Factors Affecting Cost of Major CAAA Provisions [[[ B-37
Table B-20 Rate-of-Progress Cost Sensitivity Summary B-42
Table B-21 Area Source PM Control Cost Sensitivity Analysis, Year 2000 ................................. B-43
Table B-22 Results of Sensitivity' Analysis of LEV Cost E-45
Table B-23 Discount Rate Sensitivity Analysis For 2010 Cost Estimates...................................... B-46
Table C-'l Emission Totals by Component for each Scenario
for the OTAG Domain (tpd) C-10
Table C-2 Emissions Totals by Component for each Scenario
for the Entire U.S. (tpd) C-13
Table C-3 Emissions Totals by Component for each Scenario
for the San Francisco Bay Area Entire U.S. (tpd) C-'l7
Table C-4 Emission Totals by Component for each Scenario for Los Angeles (tpd)................ C-l9
Table C-5 Emissions Totals by Component for each Scenario for Pheomx (tpd) C-22
Table C-6 Comparison of CASTNet -and RPM Average Concentration of SO4 ......................C-40
Table C-7 Comparison of CASTNet and RPM Average Concentrations
and Fractions of NO3 [[[ C-41
Table C-8 REMSAD Output File Species C-43
Table C-9 Chemical Speciation Schemes Applied for REMSAD [[[ C-45
Table C-10 Emission Totals by Component for Each Scenario for the Entire U.S. (tpd) C-46
Table C-ll Background Species Concentration used for REMSAD Initial
and Boundary Conditions C-47
Table C-12 Geographical Regions of the U.S. [[[C-50
Table C-13 Comparison of Visibility in Selected Eastern Cities, Metropolitan Areas,
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table D-3 Studies and Results Selected for Adverse Effects in Fetuses, Infants,
and Young Children D-25
Table D-4 Summary of Selected Studies for Chronic Illness [[[ D-28
Table D-5 Studies Used to Develop Respiratory Admissions Estimates D-32
Table D-6 Summary of Hospital Admissions Studies - Respiratory Illnesses ............................. D-33
Table D-7 Studies Used to Develop Cardiovascular Admissions Estimates D-39
Table D-8 Summary of Hospital Admissions Studies - Cardiovascular Illness .......................... D-40
Table D-9 Studies Used to Develop Asthma Emergency Room Visits D-42
Table D-10 Summary of Selected Studies for Emergency Room Visits
- Asthma and Acute Wheezing D-43
Table D-ll Summary of Selected Studies for Emergency Room Visits
- All-Cause, All-Respiratory, COPD, and Bronchitis D-46
Table D-12 Studies Used to Develop Minor Illness Estimates[[[ D-51
Table D-13 Summary of Selected Studies for Minor Illness D-52
Table D-14 Summary of Selected Studies for Asthmatics [[[ D-55
Table D-15 Summary of C-R Functions for Carbon Monoxide D-59
Table D-16 Summary of C-R Functions for Nitrogen Dioxide [[[ D-62
Table D-17 Summary of C-R Functions for Ozone D-65
Table D-18 Summary of C-R Functions for Particulate Matter[[[ D-71
Table D-19 Summary of C-R Functions for Sulfur Dioxide D-81
Table D-20 Change in Incidence of Adverse Health Effects Associated with Criteria Pollutants
(Pre-CAAA minus Post- CAAA) - 48 State U.S. Population within 50 km of a
Monitor (avoided cases per year)[[[ D-85
Table D-21 Change in Incidence of Adverse Health Effects Associated with Criteria Pollutants
(Pre-CAAA minus Post- CAAA) - 48 State U.S. Population
(avoided cases per year) D-87
Table D-22 Mortality Distribution by Age in Primary Analysis, Based
on Pope et al. (1995) D-89
Table D-23 Illustrative Estimates of the Impact of Criteria Pollutants on Mortality
— 48 State U.S. Population within 50 km of a Monitor (cases per year D-90
Table D-24 Illustrative Estimates of the Impact of Criteria Pollutants on Mortality
— 48 State U.S. Population (cases per year) D-90
Table D-25 Comparison of Alternative Lag Assumptions for Premature Mortality
Associated with PM Exposure D-91
Table E-l Classes of Pollutants and Ecological Effects E-3
Table E-2 Interactions Between Acid Deposition and Natural Systems
at Various Levels of Organization E-10
Table E-3 Interactions Between Nitrogen Deposition and Natural Systems
at Various Levels of Organization E-ll
Table E-4 Interactions of Mercury and Ozone With Natural Systems
at Various Levels of Organization E-12
Table E-5 Ecological Impacts with Identifiable Human Service Flows........................................ E-16
-------
List of Tables and Figures
Table E-10 Land Use Prevalence and Pass-Through Figures[[[ E-23
Table E-ll Nitrogen Loading From Atmospheric Deposition E-23
Table E-12 Estimated Avoided Costs for three Estuaries[[[ E-27
Table E-13 Avoided Cost for Atlantic Coast E-29
Table E-14 Summary of pH-Based Effects Threshold[[[ E-38
Table E-15 Acidification Results - 2010 E-40
Table E-16 Annual Economic Impact of Acidification in 2010 [[[ E-41
Table E~17 Cumulative Economic Benefits of Acidification from 1990 to 2010 E-42
Table E-18 Cumulative Cost of pH Stabilization from 1990 to 2010 ............................................. E-43
Table E-19 Difference in Commercial Timber Growth Rates With
and Without the CAAA[[[ E-48
Table E-20 Carbon Flux B CAAA versus No-CAAA Air Quality Scenarios E-52
Table E-21 Typical Impacts of Specific Pollutants on the Visual Quality of Forests.................. E-54
Table E-22 Forests Affected by Regional Pollution E-55
Table E-23 Summary of Monetized Estimates of the Annual Value of Forest
Quality Changes E-60
Table E-24 Illustrative Value of Avoiding Forest Darn-age in the United States .......................... E-60
Table E-25 Summary of National Data on Toxicity Sampling for Fishing Advisories E-63
Table E-26 Estimates of the Welfare Cost of Toxification in New York State ............................ E-63
Table E-27 Summary of Evaluated Ecological Benefits E-67
Table F-l Ozone Exposure-Response Functions for Selected Corps (SUM06) F-4
Table F-2 Relative Percent Yield Course [[[F-7
Table F-3 Ozone Exposure-Response Functions for Selected Corps (SUM06) F-8
Table G-l Six Major Sections of Title VI G-3
Table G-2 Phaseout Scenario in Clean Air Act Section 604 and Phaseout Scenario
in Amendments Added Under Clean Air Act Section 606 G-5
Table G-3 Scope of Title VI Cost Estimates[[[G-9
Table G-4 Benefits of Section 604, 606, and 609 G-16
Table G-5 Benefits of Section 608 [[[ G-20
Table G-6 Benefits of Section 611 G-21
Table G-7 Sections 604 and 606: Valuation of Total Benefits from 1990 to 2165,
With a Two Percent Discount Rate G-23
Table G-8 Adjustment Strategy for Key Parameters[[[ G-25
Table G-9 Costs and Benefits of Title VI G-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table H-3 Unit Values Used for Economical Valuation of Health
and Welfare Endpoints H-21
Table H-4 Primary Estimates of Health and Welfare Benefits Due
to Criteria Pollutants - Post-CAAA 2000 H-29
Table H-5 Primary Estimates of Health and Welfare Benefits Due
to Criteria Pollutants - Post-CAAA 2010 H-30
Table H-6 Sensitivity Analysis of Alternative Discount Rates
on the Valuation of Reduced Premature Mortality H-37
Table H-7 Elasticity Values for Conducting Sensitivity Analysis of Income Effect................. H-39
Table H-8 Illustrative Adjustment to Estimates of WTT to Avoid Morbidity H-40
Table H-9 Illustrative Adjustment to Estimates of the Value of Statistical Life......................... H-41
Figures
Figure 1-1 Analytic Sequence For First Section 812 Prospective Analysis 4
Figure 2-1 Pre- and Post-CAAA Scenario VOC Emissions Estimates 16
Figure 2-2 Pre- and Post-CAAA Scenario NOx Emissions Estimates............................................... 16
Figure 2-3 Pre- and Post-CAAA Scenario SO2 Emissions Estimates 16
Figure 2-4 Pre- and Post-CAAA Scenario CO Emissions Estimates. ................................................ 17
Figure 2-5 Pre- and Post-CAAA Scenario PM10 Emissions Estimates 17
Figure 2-6 Pre- and Post-CAAA Scenario PM2.5 Emissions Estimates. ........................................... 17
Figure 2-7 1990 Primary PM2.5 Emissions by EPA Region 19
Figure 4-1 Schematic Diagram of the Future-Year Concentration Estimation Methodology. ... 39
Figure 4-2 Distribution of Monitor-Level Ratios for 95th Percentile
Ozone Concentration: 2010 Post-CAAA/Pre-CAAA 40
Figure 4-3 Distribution of Combined RADM/RPM- and REMSAD-Denved
Monitor Level Ratios for Annual Average PM10 Concentrations:
2010 Post-CAAA/Pre-CAAA[[[ 42
Figure 4-4 Distribution of Combined RADM/RPM- and REMSAD-Denved
Monitor-Level Ratios for Annual Average PM2.5 Concentrations:
2000 Post-CAAA / 2000Pre-CAAA 42
Figure 4-5 Distribution of Monitor-Level Ratios of Summer SO2 Emissions:
2010 Post-CAAA/ 2010 Pre-CAAA 45
Figure 4-6 Distribution of Monitor-Level Ratios of NO Summer Emissions:
2010 Post-CAAA/ 2010 Pre-CAAA 45
Figure 4-7 Distribution of Monitor-Level Ratios of NO2 Summer Emissions:
2010 Post-CAAA/ 2010 Pre-CAAA 46
-------
List of Tables and Figures
Figure 7-1 Annual Economic Welfare Benefit of Mitigating Ozone Impacts
on Commercial Timber: Difference Between Pre- and Post-CAAA 93
Figure 8-1 Monte Carlo Simulation Model Primary Benefits Results
for Target Years - Titles I Through V . [[[ 101
Figure 8-2 Analysis of Contribution of Key Parameters to Quantified Uncertainty 107
Figure A-l Comparison of Pre-CAAA, Post-CAAA, and Trends
VOC Emissions Estimates [[[ A-8
Figure A-2 Comparison of Pre-CAAA, Post-CAAA, and Trends
NOx Emissions Estimates [[[A-l0
Figure A-3 Comparison of Pre-CAAA, Post-CAAA, and Trends
SO Emissions Estimates [[[ A-l 0
Figure A-4 Comparison of Pre-CAAA, Post-CAAA, and Trends
CO Emissions Estimates [[[ A-l 1
Figure A-5 Comparison of Pre-CAAA, Post-CAAA, and Trends
PM10 Emissions Estimates [[[ A-l 1
Figure A-6 1990 Primary PM2.5 Emissions by EPA Region A-13
Figure C-'l Schematic Diagram of the Future-year Concentration
Estimation Methodology [[[ C-4
Figure C-2 Difference in Daily Maximum Simulated Ozone Concentration (ppb)
for the 15 July 1995 OTAG Episode Day: 2010 pre-CAAA90
minus base 1990 C-26
Figure C-3 Difference in Daily Maximum Simulated Ozone Concentration (ppb)
for the 15 July 1995 OTAG Episode Day: 2010 post-CAAA90
minus base 1990 [[[C-27
Figure C-4 Difference in Daily Maximum Simulated Ozone Concentration (ppb)
for the 15 July 1995 OTAG Episode Day: 2010 post-CAAA90
minus pre-CAAA90 C-28
Figure C-5 Difference in Daily Maximum Simulated Ozone Concentration (ppb)
for the 8 July 1995 Western U.S. Simulation Day: 2010 pre-CAAA90
minus base 1990 [[[C-29
Figure C-6 Difference in Daily Maximum Simulated Ozone Concentration (ppb)
for the 8 July 1995 Western U.S. Simulation Day: 2010 post-CAAA90
minus base 1990 C-30
Figure C-7 Difference in Daily Maximum Simulated Ozone Concentration (ppb)
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-lla Distribution of Monitor - Level Ratios for 95th Percentile
Ozone Concentration 2000 Pre-CAAA90/1990 Base-Year C-35
Figure C-lib Distribution of Monitor - Level Ratios for 95th Percentile
Ozone Concentration 2000 Post-CAAA90/1990 Base-Year C-35
Figure C-12a Distribution of Monitor - Level Ratios for 95th Percentile
Ozone Concentration 2010 Pre-CAAA/1990 Base-Year C-36
Figure C-12b Distribution of Monitor - Level Ratios for 95th Percentile
Ozone Concentration 2010 Post-CAAA/1990 Base-Year C-36
Figure C-13a Distribution of Monitor - Level Ratios for 95th Percentile
C)2one Concentration 2000 Post-CAAA/2000 Pre-CAAA C-37
Figure C-13b Distribution of Monitor - Level Ratios for 95th Percentile
C)2one Concentration 2010 Post-CAAA/2010 Pre-CAAA C-37
Figure C-14 80-km RADM Domain[[[C-54
Figure C-15 Comparison of Simulated and Observed Seasonal PM10 Concentration
(ug/m3) for REMSAD for the Western U.S.: Spring 1990...........................................C-55
Figure C-16 Comparison of Simulated and Observed Seasonal PM10 Concentration
(ug/m3) for REMSAD for the Western U.S.: Summer 1990 .......................................C-55
Figure C-17 Comparison of Simulated and Observed Seasonal PM10 Concentration
(ug/m3) for REMSAD for the Western U.S.: Fall 1990................................................C-55
Figure C-18 Comparison of Simulated and Observed Seasonal PM10 Concentration
(ug/m3) for REMSAD for the Western U.S.: Winter 1990 .........................................C-55
Figure C-19 Difference in Seasonal Average PM10 Concentration (ug/m3) for the Summer
REMSAD Simulation Period (1-10 July 1990) for 2010: post-CAAA90
minus pre-CAAA90 C-56
Figure C-20 Difference in Seasonal Average PM25 Concentration (ug/m3) for the Summer
REMSAD Simulation Period (1-10 July 1990) for 2010: post-CAAA90
minus pre-CAAA90 [[[C-57
Figure C-21a Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-10 Concentration:
2000 Pre-CAAA90 / 1990Base-Year C-58
Figure C-21b Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-10 Concentration:
2000 Post-CAAA/1990 Base-Year[[[ C-58
Figure C-22a Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-10 Concentration:
2010 Pre-CAAA/1990 Base-Year C-59
Figure C-22b Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-10 Concentration:
2010 Post-CAAA/1990 Base-Year[[[ C-59
Figure C-23a Distribution of Combined RADM/RPM- and REMSAD-derived
-------
List of Tables and Figures
Figure C-24a Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-2.5 Concentration:
2010 Pre-CAAA/1990 Base-Year [[[ C-61
Figure C-24b Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-2.5 Concentration
2010 Post-CAAA/1990 Base-Year C-61
Figure C-25a Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-10 Concentration:
2000 Post-CAAA/2000 Pre-CAAA[[[C-62
Figure C-25b Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-10 Concentration:
2010 Post-CAAA/2010 Pre-CAAA C-62
Figure C-26a Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-2.5 Concentration:
2000 Post-CAAA/ 2000 pre-CAAA[[[C-63
Figure C-26b Distribution of Combined RADM/RPM- and REMSAD-derived
Monitor-level Ratios for Annual Average PM-2.5 Concentration
2010 Post-CAAA/2010 Pre-CAAA C-63
Figure C-27 Seasonal Average Deciview for the Summer REMSAD Simulation
Period (1-10 July 1990): Base 1990 (Western U.S. Only) C-67
Figure C-28 Difference in Seasonal Average Deciview for the Summer REMSAD
Simulation Period (1-10 July 1990): 2010 Pre-CAAA90 Minus Base 1990
(Western United States Only)[[[C-68
Figure C-29 Annual Sulfur Deposition 1990 Base Case Scenario C-71
Figure C-30 Annual Nitrogen Deposition 1990 Base Case Scenario ................................................. C-72
Figure C-31 Annual Sulfur Deposition 2010 Pre CAAA Scenario C-73
Figure C-32 Annual Sulfur deposition 2010 Post CAAA Scenario [[[C-74
Figure C-33 Annual Nitrogen Deposition 2010 Pre CAAA Scenario C-75
Figure C-34 Annual Nitrogen deposition in 2010 Post CAAA Scenario ........................................ C-76
Figure C-35 Distribution of Monitor-Level Ratios of Summer SO2 Emissions:
2010 Post-CAAA / 2010 Pre-CAAA [[[C-78
Figure C-36 distribution of Monitor-Level Ratios of Summer NO Emissions:
2010 Post-CAAA / 2010 Pre-CAAA [[[C-78
Figure C-37 Distribution of Monitor-Level Ratios of Summer NO2 Emissions:
2010 Post-CAAA / 2010 Pre-CAAA [[[C-79
Figure C-38 Distribution of Monitor-Level Ratios of Summer CO Emissions:
2010 Post-CAAA / 2010 Pre-CAAA [[[C-79
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure E-5 Chesapeak Bay SAV 1978-1996 [[[ E-32
Figure E-6 Percentage of Acidic Surface Waters in the National Surface
Water Survey Regions[[[ E-36
Figure E-7 Acidification of Freshwater Ecosystems E-37
Figure E-8 Annual Economic Welfare Benefit of Mitigation Ozone Impacts
on Commercial Timber E-49
Figure E-9 U.S. Major Forest Types Affected By Air Pollution-Induced Visual Injuries ......... E-57
Figure G-'l Schematic of Cost and Benefit Analysis of Title VI[[[ G-12
Figure G-2 Annual Human Health Benefits from Sections 604 and 606 (Discounted at 5%).. G-29
Figure G-3 Annual Undiscounted Human Health Benefits of Sections 604 -and 606................. G-33
Figure H-'l Hypothetical Survival Curve Shift[[[ H-4
-------
Acronyms and Abbreviations
uEq microequivalents
(ig microgram
ACT average cost per ton
AGSIM AGricultural Simulation .Model
AIC Akaike information criterion
AIRS Aerometric Information Retrieval
System
ANC acid neutralizing capacity
AN OVA analysis of variance
AOD airway obstructive disease
AP-42 EPA's Compilation of Air Pollu-
tion Emission Factors
ATDM aerosol and toxics deposition
module
AQM air quality modeling
A QMS Air Quality Modeling Subcommit-
tee
ATLAS Aggregate Timber Land Assessment
System
b light extinction coefficient
ext o
BAAQM.D Bay Area Air Quality Management
District
BACT best available control technology
BAF bioaccumulation factor
BARCT best available retrofit control
technology
BCF bioconcentration factor
BEA Bureau of Economic Analysis
BID background information document
BIES Biogenic Emissions Inventory
System
BLS Bureau of Labor Statistics
BMP best management practice
BNR biological nutrient removal
BS black smoke
C-R concentration-response
CAA Clean Air Act
CAAA Clean Air Act Amendments
CAP I Clean Air Power Initiative
CAPMS Criteria Air Pollutant Modeling
System
GARB California Air Resources Board
CAS AC
CASTNet
CB
GEM
CES
CFG
CFFP
CGE
CI
CO
GOH
CO I
COPD
CRC
CRF
CTG
CV
dbh
DDT
DOE
dV
E-GAS
EC
EGU
EMFAC
ER
EPA
EPS
ERCAM
ERL
FACA
FAPRI
FGM
FDA
Clean Air Science Advisory Board
Clean Air Act Status and Trends
Network
chronic bronchitis
continuous emissions monitoring
constant elasticity of substitution
chlorofluorocarbon
Clean Fuel Fleet Program
computable general equilibrium
compression ignition
carbon monoxide
coefficient of haze
cost of illness
chronic obstructive pulmonary
disease
capital recovery cost
capital recovery factor
control technique guideline
contingent valuation
diameter at breast height
dichlorodiphenyl-trichloroethane
Department of Energy
deciview
Economic Growth Analysis System
elemental carbon
electrical generating unit
emission factors model
emergency room
Environmental Protection Agency
Emissions Processing System
Emission Reduction and Cost
Analysis Model
Environmental Research Labora-
tory
Federal Advisory Committee Act
Food and Agricultural Policy
Research Institute
Fuel Consumption Model
Food and Drug Administration
forced expiratory volume in one
second
xxv
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
FGD flue gas desulfurization
FHWA Federal Highway Administration
FMVCP Federal .Motor Vehicle Control
Program
FORCARB forest carbon model
FORTRAN formula translation
FR Federal Register
GCVTC Grand Canyon Visibility Transport
Commission
GDP gross domestic product
GIRAS Geographic Information Retrieval
Analysis System
GIS geographic information system
GNP gross national product
GSP gross state product
H + hydrogen ions
ha hectare
HAP hazardous air pollutant
HARVCARB harvested carbon model
HB FC hydrobromofluorocarbons
IIC hydrocarbon
H C F C hydrochlorofluorocarboii
HDDV heavy-duty diesel vehicle
HDGV heavy-duty gasoline vehicle
HDV heavy-duty vehicle
HEES Health and Ecological Effects
Subcommittee
Hg mercury
IIIV-1 human immunodeficiency virus
type one
HNO3 nitric acid
HPMS Highway Performance Monitoring
System
HS_O4 sulfunc acid
I/M inspection and maintenance
ICI iiidustrial/commercial/iiistitutioiial
ICD International Classification of
Disease
ID identification code
IMPROVE Iiiterageiicy Monitoring of
PROtected Environments
IP.M Integrated Planning Model
kg kilogram
km kilometer
kWh kilowatt hour
LAER lowest achievable emission rate
Ib pound
LDAR leak detection and repair
LDDT light-duty diesel truck
LDDV light-duty diesel vehicle
LDGT light-duty gasoline truck
LDGV light-duty gasoline vehicle
LEV low emission vehicle
LRS lower respiratory symptom
LTO landing and takeoff operations
m meter
m3 cubic meter
MACT maximum achievable control
technology
MAG Mancopa Association of Govern-
ments
MAGIC Model of Acidification of Ground-
water 111 Catchments
.M C motorcycle
MCF methyl chloroform
MDL method detection limit
MM4 mesoscale model four
MMBtu million British thermal units
MRAD minor restricted activity day
Models-3 Third Generation Air Pollution
Modeling System
MOD memorandum of understanding
.MOBILE mobile source emission factor
model
MPO metropolitan planning organization
MWC municipal waste combustor
MWI medical waste incinerator
N nitrogen
NAA nonattainment area
NAAQS National Ambient Air Quality-
Standards
NAPAP National Acid Precipitation Assess-
ment Program
NASA National Aeronautics and Space
Adminis tration
NCAR National ("enter for Atmospheric
Research
NCLAN National Crop Loss Assessment
Network
NE northeast
NEMS National Energy Modeling System
NERC North American Electric Reliabil-
ity Council
NESHAP National Emission Standards for
Hazardous Air Pollutants
XXVI
-------
Acronyms and Abbreviations
NET National Emission Trend
NH ammonia
NIIANES National Health and Nutrition
Examination
NIH National Institutes of Health
NMOC non-methane organic compound
N (3 nitrogen oxide
NO2 nitrogen dioxide
NOx nitrogen oxides
NP national park
NPI National Particulates Inventory
NPP net primary productivity
NPV net present value
NSPS new source performance standard
NSR new source review
NSWS National Surface Waters Survey
NYSDEC New York Department of Environ-
mental Conservation
O3 ozone
O&M operation and maintenance
OBD onboard diagnostic
O C organic carbon
ODS ozone-depleting substance
OMB Office of Management and Budget
OMS Office of Mobile Sources
OPPE Office of Policy, Planning and
Evaluation
ORIS Office of the Regulatory Informa-
tion System
OSD ozone season daily
OTAG Ozone Transport Assessment
Group
OTC Ozone Transport Commission
OTR Ozone Transport Region
P-i-G plume-in-grid
PAN peroxyacetyl nitrate
Pb lead
PCB polychlorinated biphenyl
PCDD polychlorinated dibenzo-p-dioxin
PCDF polychlorinated dibenzofurans
PCE perchloroethylene
pH logarithm of the reciprocal of
hydrogen ion concentration, a
measure of acidity
PM paniculate matter (both PM10 and
PM25)
PM.|() particulate matter less than or equal
to 10 microns in diameter
PM particulate matter less than or equal
to 2.5 microns in diameter
PnET Net Photosynthesis and Evapo-
Transpiration model
POC parameter occurrence code
POTW publically owned treatment works
ppb parts per billion
ppm parts per million
PRYL percentage relative yield loss
PRZM Pesticide Root Zone Model
PSU Pennsylvania State University
QALY quality adjusted life years
R&D research and development
RACT reasonable available control tech-
nology
RAD restricted activity day
RADM Regional Acid Deposition Model
RELMAP Regional Lagrangian Model of Air
Pollution
REMSAD Regulatory Modeling System for
Aerosols and Acid Deposition
RE rule effectiveness
RFG reformulated gasoline
RHC reactive hydrocarbon
RIA regulatory impact analysis
RFP reasonable further progress
RO2 peroxy radical
ROP rate of progress
RPM Regional Particulate Model
RUM Random Utility Model
RYP Reid vapor pressure
S sulfur
SAB Science Advisory Board
SAS Statistical Analysis Software
SAV submerged aquatic vegetation
SCAQMD South Coast Air Quality Manage-
ment District
SCAQS South Coast Air Quality Study
SCC Source Classification Code
SCR selective catalytic reduction
SEDS State Energy Data Systems
SI spark ignition
SIC Standard Industrial Classification
SIP State Implementation Plan
SO_ sulfur dioxide
xxvn
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
SO A secondary organic aerosol
SoCAB South Coast Air Basin
SOCMI synthetic organic chemical manu-
facturing industry
SUM06 sum of hourly ozone concentra-
tions at or above 0.06 ppm
TAG total annualized costs
TAP temporal allocation factors
TAMM Timber Assessment Market .Model
TBRP Tampa Bay Estuary Program
TCDD tetrachlorodibenzo-p-dioxin
TEQ toxic equivalency
TLEV transitional low emission vehicle
tpd tons per day
TREGRO tree growth model
TSDF treatment, storage, and disposal
facility
TSP total suspended particulates
[JAM Urban Airshed M'odel
URS upper respiratory symptoms
USD A United States Department of
Agriculture
ULEV ultra-low emission vehicle
USGS United States Geological Survey
UV ultraviolet
VMT vehicle miles traveled
VNA Voronoi neighbor averaging
VOG volatile organic compound
VR visual range
VSL value of statistical life
VSLY value of statistical life year
WE FA Wharton Economic Forecasting
Associates
WHO World Health Organization
WrLD work-loss days
WT A willingness-to-accept
WTP willingness-to-pay
XO, halogenated peroxy radical
yr year
ZEV zero emission vehicle
XXViil
-------
Acknowledgments
This project is managed under the direction of
Robert Perciasepe, Assistant Administrator and Rob-
ert D. Brenner, Deputy Assistant Administrator for
the EPA Office of Air and Radiation. The principal
project manager is Jim DeMocker, Senior Policy
Analyst, EPA Office of Air and Radiation/Office of
Policy Analysis and Review. Brian Heninger, EPA
Office of Policy/Office of Economy and Environ-
ment, directed the ecological assessment; and Sam
Napolitano, EPA Office of Air and Radiation/Of-
fice of Atmospheric Programs directed the electric
utility emissions and cost analyses. Robin Dennis,
EPA Office of Research and Development/National
Exposure Research Laboratory directed the RADM-
RPM air quality modeling. Al McGartland, Direc-
tor of the EPA Office of Economy and Environment
in the Office of Policy, and David Gardiner, former
Assistant Administrator for the Office of Policy' pro-
vided guidance and support.
Many EPA staff contributed or reviewed portions
of this document, including Bryan Hubbcll, John
Bachmann, Ron Evans, Rosalina Rodriguez, Scott
Mathias, Ann Watkins, Rona Bimbaum, Karen Mar-
tin, Dons Price, Drusilla Hufford, Jeff Cohen, Joe
Somers, Carl Mazza, Brett Snyder, and Tom Gillis.
A number of contractors developed key elements
of the analysis and supporting documents. Jim
Neumann of Industrial Economics, Incorporated
managed the overall integration and coordination of
the analytical work and documentation and also made
considerable substantive analytical contributions.
Other contractor members of the 812 Project Team
included Bob Unsworth, Henry Roman, Jared
Hardner, Naomi Kleckner, Nick Live say, Lauren
Fusfeld, Andre Cap, Stephen Everett, Jon Discher,
and Mike Hester of Industrial Economics, Incorpo-
rated; Lelaiid Deck, Ellen Post, Lisa Akcson, Ken-
neth Davidson, and Don McCubbin of Abt Associ-
ates; Sharon Douglas, John Langstaff, Robert
Iwamiya, Belle Hudischewsky, and John Calcagm of
ICF Incorporated, and John Blaney of ICF' Consult-
ing; and Jim Wilson, Erica Laich, and Dianne Crocker
of Pechan-Avanti Associates.
Science Advisory Board review of this report is
supervised by Donald G. Barnes, Director of the SAB
Staff. The Designated Federal Official for the SAB
reviews is Angela Nugent. Other SAB staff who as-
sisted 111 the coordination of SAB reviews include jack
Fowie, Robert Flaak, and Jack Kooyoomjian. Diana
Pozuii provided administrative support to the SAB.
The SAB Council is chaired by Maureen Crop-
per of the World Bank. SAB Council members serv-
ing during the final review of this report include A.
Myrick Freeman of Bowdoin College, Gardner
Brown, Jr. of the University of Washington, Paul
Lioy of the Robert Wood Johnson School of Medi-
cine, Paulette Middleton of the RAND Center for
Environmental Sciences and Policy, Donald Fuller-
ton of the University of Texas — Austin, Lawrence
Goulder of Stanford University, Jane Hall of Cali-
fornia State University — Fullerton, Charles Kolstad
of the University of California at Santa Barbara, and
Lester Lave of Carnegie-Mellon University. Alan
Krupnick of Resources for the Future served as a
Consultant to the Council. In addition, several mem-
bers of the SAB Council whose terms expired during
the development of the study provided valuable ad-
vice and ideas in the early stages of project design and
implementation. These former members include
Richard Schmalensee of MIT, William Nordhaus of
Yale University, Paul Portncy of Resources for the
Future, Kip Viscusi of Harvard University, Ronald
Cummings of Georgia State University, Thomas
Tietenberg of Colby College, Wallace Gates of the
University of Maryland, Wayne Kachel of MELE As-
sociates, Robert .Mendelsohn of Yale University, and
Daniel Dudek of the Environmental Defense Fund.
William Smith, a liaison to the Council from the SAB
Environmental Processes and Effects Committee also
provided valuable advice regarding the ecological as-
sessment.
XXIX
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The SAB Council is supported by two technical
subcommittees. The first of these subcommittees,
the Health and Ecological Effects Subcommittee is
chaired by Paul Lioy. Members who participated in
the final review of this report included Morton
Lippmann of New York University Medical Center,
George T. Wolff of General .Motors, A. Myrick Free-
man, Timothy Larson of the University of Washing-
ton, Joseph Meyer of the University of Wyoming,
Robert Rowe of Stratus Consulting, George Taylor
of George "Mason University, jane Hall, "Michael
Kleinman of the University of California at Irvine,
and Carl Shy of die University of North Carolina at
Chapel Hill. Several former members who provided
valuable advice in the early stages of the study in-
clude Bernard Weiss of the University of Rochester
.Medical Center, David V. Bates of the University of
British Columbia, Gardner Brown, and Lester Lave.
The second technical subcommittee, the Air
Quality Modeling Subcommittee is chaired by
Paulette Middleton. Members serving during the fi-
nal review of this report include Philip Hopke of
Clarkson University, James IT. Price, Jr. of the Texas
Natural Resource Conservation Commission, Harvey
Jeffries of the University of North Carolina — Chapel
Hill, Timothy Larson, and Peter Mueller of the Elec-
tric Power Research Institute. A former member who
helped guide the analysis in its early stages was George
T. Wolff.
The project managers wish to convey special ac-
knowledgment and appreciation for the valuable con-
tributions of A. Myrick Freeman. As a charter mem-
ber of the Council and as Vice Chair of the "Health
and Ecological Effects Subcommittee, Dr. Freeman
provided wise and excellent counsel throughout the
entire course of SAB review of both this study and
the preceding retrospective study.
This report could not have been produced with-
out the support of key administrative support staff.
The project managers are grateful to Barbara Morris,
Nona Smoke, Eunice javis, Gloria Booker, and
Wanda Farrar for their timely and tireless support
011 this project.
XXX
-------
CD
Section 812 of the 1990 Clean Air Act Amend-
ments requires the EPA to develop periodic Reports
to Congress that estimate the benefits and costs of
the Clean Air Act (CAA). The first report EPA
created under this authority, The Benefits and Costs
of the Clean Air Act: 1970 to 1990, was published and
conveyed to Congress in October 1997. This retro-
spective analysis comprehensively assessed benefits
and costs of requirements of die 1970 Clean Air Act
and the 1977 Amendments, up to the passage of the
Clean Air Act Amendments of 1990. The results of
the retrospective analysis showed that the nation's
investment in clean air was more than justified by
the substantial benefits that were gamed in the form
of increased health, environmental quality, and pro-
ductivity. The aggregate benefits of the CAA dur-
ing the 1970 to 1990 period exceeded costs by a fac-
tor of 10 to 100 times.
Before the retrospective analysis was complete,
we began the process of assessing the prospective
benefits and costs of the Clean Air Act Amendments
(CAAA), covering the period 1990 to 2010. This
report, the first of a series that we plan to produce
every two years, is the result of our prospective analy-
sis of the 1990 Amendments.
Similar to the retrospective analysis, this docu-
ment has one primary and several secondary objec-
tives. The mam goal is to provide Congress and the
public with comprehensive, up-to-date information
on the CAAA's social costs and benefits, including
health, welfare, and ecological benefits. Data and
methods derived from the retrospective analysis have
already been used to assist policy-makers in refining
clean air regulations over the last two years, and we
hope the information continues to prove useful to
Congress during future Clean Air Act reauthoriza-
tions. Beyond the statutory goals of section 812,
EPA also intends to use the results of this study to
help support decisions on future investments in air
pollution research. In addition, lessons learned in
conducting this first prospective will help better tar-
get efforts to improve the accuracy and usefulness
of future prospective analyses.
of This
to
The Clean Air Act Amendments of 1990 aug-
ment the significant progress made in improving the
nation's air quality through the original Clean Air
Act of 1970 and its 1977 amendments. The amend-
ments built off the existing structure of the original
Clean Air Act, but went beyond those requirements
to tighten and clarify implementation goals and tim-
ing, increase the stringency of some federal require-
ments, revamp the hazardous air pollutant regula-
tory program, refine and streamline permitting re-
quirements, and introduce new programs for the
control of acid ram and stratospheric ozone depleters.
Because the 1990 Amendments represent an addi-
tional improvement to the nation's existing clean
air program, the analysis summarized in this report
was designed to estimate the costs and benefits of
the 1990 CAAA incremental to those costs and ben-
efits assessed in the retrospective analysis. In eco-
nomic terminology, this report addresses the mar-
ginal costs and benefits of the 1990 CAAA. Our
intent is that this report and its predecessor, the ret-
rospective analysis, together provide a comprehen-
sive assessment of current and expected future clean
air regulatory programs and their costs and benefits.
Because of the time and resources necessary to
conduct this type of comprehensive prospective as-
sessment, however, and the ongoing refinements in
Clean Air Act regulatory programs, the estimates
presented in this report do not reflect some recent
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
major developments in EPA's clean air program.
The prospective analysis, for example, does not cap-
ture the benefits and costs of EPA's recent revision
of the participate matter and ozone National Ambi-
ent Air Quality Standards (NAAQS), the recently
proposed Tier II tailpipe standards, or the recently
finalized regional haze standards. Neither costs nor
benefits of those actions are reflected in the estimates
presented here. In most cases, Regulatory Impact
Analyses (RIAs) for those actions did incorporate
the section 812 prospective Post-CAAA scenario as
their starting point, or baseline, from which the ac-
tions were assessed, and in most respects the RIAs
used a methodology consistent with that used here.1
As a result, cost and benefit estimates presented in
those RIAs can be considered incremental to the
primary estimates presented in this document.
In addition to omitting these actions from the
assessment, this first prospective analysis required
locking in a set of emissions reductions to be used in
subsequent analyses at a relatively early date (late
1996), and as a result we were compelled to forecast
the implementation outcome of several pending pro-
grams. The most important of these was the then-
ongoing Ozone Transport Assessment Group
(OTAG) recommendations for achieving regional-
scale reductions of emissions of ground-level ozone
precursors. The NOx control program incorporated
in the Post-CAAA scenario may not reflect the NOx
controls that are actually implemented in a regional
ozone transport rule. We acknowledge and discuss
these types of discrepancies and their impact on die
outcome of our analysis in the document.
Finally, despite our efforts to comprehensively
evaluate the costs and benefits of all provisions of
the Clean Air Act and its Amendments, there re-
main a few7 categories of effects that are not addressed
by either the retrospective or prospective analyses.
For example, this first prospective analysis does not
assess the effect of CAAA provisions on lead expo-
sures, primarily because the 1990 Amendments do
' There arc minor differences in the assumptions used to
construct the Post- CAAA scenario for this analysis and the
baseline used in the PM and ozone NAAQS RTA. For example,
the RIA baseline incorporates the effects of 7- and 10-year MACT
rules that are not reflected here, because of the timing of the
two analyses, and the RIA used a 95 percent rule-eiiecliveness
assumption. In most respects, however, the analyses are com-
patible.
not include major new provisions for the control of
lead emissions. The vast majority of lead emissions
sources present in 1970 were addressed by programs
initiated under the original Clean Air Act and the
1977 Amendments; evaluation of the costs and health
benefits of these programs were important elements
of the retrospective analysis. In the retrospective,
however, wre wrere unable to quantify the potentially
substantial ecological benefits of controls on lead
emissions. While this first prospective analysis re-
flects a significantly greater investment in quantify-
ing ecological effects, for the reason stated above we
did not assess the ecological effects of lead in this
analysis either. As a result, the ecological effects of
this persistent pollutant, past emissions of which may
continue to be released from soils for many years,
are not captured by either the retrospective or pro-
spective analyses. In addition, lead previously de-
posited to soils may be re-entrained in the air as road
dust, dust plumes from construction excavations, and
other particulate matter emission processes subject
to 1990 CAAA controls. Reductions in this re-en-
trainmen t of, and potential exposure to, pre-1990
emitted lead due to post-1990 control programs,
however, are not reflected in either the section 812
retrospective (1970 to 1990) or prospective (1990 to
2010) benefit analyses.
Air
The first prospective analysis, despite the limi-
tations discussed above, presents a comprehensive
estimate of costs and benefits of all titles of the 1990
Clean Air Act Amendments. The 1990 Amendments
consist of the following eleven titles:
* Title I. Establishes a detailed and graduated
program for the attainment and maintenance
of the National Ambient Air Quality Stan-
dards.
* Title II. Regulates mobile sources and es-
tablishes requirements for reformulated gaso-
line and clean fuel vehicles.
* Title III. Expands and modifies regulations
of hazardous air pollutant emissions; and
establishes a list of 189 hazardous air pollut-
ants to be regulated.
-------
Chapter 1: Introduction
* Title IV. Establishes control programs for
reducing acid rain precursors.
• Title V. Requires a new permitting system
for primary sources of air pollution.
• Title VI. Limits emissions of chemicals that
deplete stratospheric ozone.
• Title VII. Presents new provisions for en-
forcement.
Titles VIII through XI. Establishes miscel-
laneous provisions for issues such as disad-
vantaged business concerns, research, train-
ing, new regulation of outer continental shelf
sources, and assistance for people who lose
their jobs as a result of the Clean Air Act
Amendments.
As part of the requirements under Title VIII,
section 812 of the Clean Air Act Amendments of
1990 requires the EPA to analyze the costs and ben-
efits to human health and the environment that are
attributable to the Clean Air Act. In addition, sec-
tion 812 directs EPA to measure the effects of this
statute on economic growth, employment, produc-
tivity, cost of living, and the overall economy of the
United States.
Target
The prospective analysis compares the overall
health, welfare, ecological and economic benefits of
the 1990 Clean Air Act Amendment programs to
the costs of these programs. By examining the over-
all effects of the Clean Air Act, this analysis comple-
ments the Regulatory Impact Analyses (RIAs) de-
veloped by EPA over the years to evaluate individual
regulations. Resources were used more efficiently by
recognizing that these RIAs, and other EPA analy-
ses, provide complete information about the costs
and benefits of specific rules. Within this analysis,
costs can be reliably attributed to individual pro-
grams, but the broad-scale approach adopted in the
prospective study precludes reliable re-estimation of
the benefits of individual standards or programs.
Similar to the retrospective benefits analysis, this
study calculates the change in incidences of adverse
effects implied by changes in ambient concentrations
of air pollutants. However, pollutant emissions re-
ductions achieved contribute to changes in ambient
concentrations of those, or secondarily formed, pol-
lutants in ways that are highly complex, interactive,
and often nonlinear. Therefore, benefits cannot be
reliably matched to provision-specific changes in
emissions or costs.
Focusing 011 the broader target variables of over-
all costs and overall benefits of the Clean Air Act,
the EPA Project Team adopted an approach based
on construction and comparison of two distinct sce-
narios: a "Pre-CAAA" and a "Post-CAAA" scenario.
The Pre-CAAA scenario essentially freezes federal,
state, and local air pollution controls at the levels of
stringency and effectiveness which prevailed in 1990.
The Post-CAAA scenario assumes that all federal,
state, and local rules promulgated pursuant to, or in
support of, the 1990 CAAA were implemented. This
analysis then estimates the differences between the
economic and environmental outcomes associated
with these two scenarios. For more information on
the scenarios and their relationship to historical
trends, see Chapter 2 and Appendix A of this docu-
ment.
Similar to the retrospective analysis, we made
two key assumptions during the scenario design pro-
cess to avoid miring the analytical process in endless
speculation. First, as stated above, we froze air pol-
lution controls at 1990 levels throughout the Pre-
CAAA scenario. Second, we assumed that the geo-
graphic distributions of population and economic
activity remain the same between the two scenarios,
although these distributions do change over time
under both scenarios to reflect expected patterns of
high and low population and economic growth
across the country.
The first assumption is an obvious simplifica-
tion. In the absence of the 1990 CAAA, one would
expect to see some air pollution abatement activity,
either voluntary or due to state or local regulation.
It is conceivable that state and local regulation would
have required air pollution abatement equal to -or
even greater than- that required by the 1990 CAAA;
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
particularly since some states, most notably Califor-
nia, have in the past done so. If one were to assume
that state and local regulations would have been
equivalent to 1990 CAAA standards, then a cost-
benefit analysis of the 1990 CAAA would be a mean-
ingless exercise since both costs and benefits would
equal zero. Any attempt to predict how states' and
localities' regulations would have differed from the
1990 CAAA would be too speculative to support
the credibility of the ensuing analysis. Instead, the
Pre-CAAA scenario has been structured to reflect
the assumption that states and localities would not
have invested further in air pollution control pro-
grams after 1990 in the absence of the federal CAAA.
Thus, this analysis accounts for all costs and ben-
efits of air pollution control from 1990 to
2010 and does not speculate about the frac-
tion of costs and benefits attributable exclu-
sively to the federal CAAA. Nevertheless,
it is important to note that state and local
governments and private initiatives are re-
sponsible for a significant portion of these
total costs and total benefits. In the end,
the benefits of air pollution controls result
from partnerships among all levels of gov-
ernment and with the active participation
and cooperation of private entities and indi-
viduals.
The second assumption concerns chang-
ing demographic patterns in response to air
pollution. In the hypothetical Pre-CAAA
scenario, air quality is worse than the actual
1990 conditions and the projected air qual-
ity in the Post-CAAA scenario. It is pos-
sible that under the Pre-CAAA scenario
more people, relative to the Post-CAAA
case, would move away from the most
heavily polluted areas. Rather than specu-
late on the scale of population movement,
the analysis assumes no differences in demo-
graphic patterns between the two scenarios.
Similarly, the analysis assumes no differences
between the two scenarios with respect to
the spatial pattern of economic activity.
Analytic
The analysis comprises a sequence of six
basic steps, summarized below and described
in detail later in this report. These six steps, listed in
order of completion, are:
(1) emissions modeling
(2) direct cost estimation
air quality modeling
health and environmental effects estimation
economic valuation
(6) results aggregation and uncertainty charac-
terization
Figure 1-1 summarizes the analytical sequence
used to develop the prospective results; we describe
the analytic process in greater detail below.
Figure 1-1
Analytic Sequence for
First Section 812 Prospective Analysis
Comparison of Benefits
and Costs
-------
Chapter 1: Introduction
The first step of the analysis is die estimation of
the effect of the 1990 CAAA on emissions sources.
We generated emissions estimates through a three
step process: (1) construction of an emissions inven-
tory for the base year (1990); (2) projection of emis-
sions for the Pre-CAAA case for two target years,
2000 and 2010, assuming a freeze on emissions con-
trol regulation at 1990 levels and continued economic
progress, consistent with sector-specific Bureau of
Economic Analysis economic activity projections;
and (3) construction of Post-CAAA estimates for the
same two target years, using the same set of economic
activity projections used in the Pre-CAAA case but
with regulatory stringency, scope, and timing con-
sistent with EPA's CAAA implementation plan (as
of late 1996). The analysis reflects application of
utility and other sector-specific emissions models
developed and used in various offices of EPA's Of-
fice of Air and Radiation. These emissions models
provide estimates of emissions of six criteria air pol-
lutants2 from each of several key emitting sectors.
We provide more details in Chapter 2 and Appen-
dix A.
The emissions modeling step is a critical compo-
nent of die analysis, because it establishes consistency
between the subsequent cost and benefit estimates
that we develop. Estimates of direct compliance costs
to achieve the emissions reductions estimated in the
first step are generated as either an integral or subse-
quent output from the emissions estimation mod-
els, depending on the model used. For example, the
Integrated Planning Model used to estimate emissions
and compliance costs for the utility sector develops
an optimal allocation of reductions of sulfur and ni-
trogen oxides taking into account the regulatory flex-
ibility inherent in the Title IV trading schemes for
emissions allocations. In a few cases, for example
the Title V permitting requirements, we estimate
public and private costs incurred to implement the
2 The six pollutants are participate matter (separate esti-
mates for each of PM|f and PM25), sulfur dioxide (SO,), nitro-
gen oxides (NOJ, carbon monoxide (CO), volatile organic com-
pounds (VOCs), and ammonia (NH,). One of the CAA criteria
pollutants, ozone (O^), is formed in the atmosphere through the
interaction of sunlight and ozone precursor pollutants such as
NOv and VOCs. Ammonia is not a criteria pollutant, but is an
important input to the air quality modeling step because it af-
fects secondary particulate formation. The sixth criteria pollut-
ant, lead (Pb), is not included in this analysis since airborne
emissions oi lead were virtually eliminated by pre-1990 Clean
Air Act programs.
regulatory requirements through analysis of the rel-
evant RIAs conducted to support promulgation of
the rules.
Emissions estimates also form the first step in
estimating benefits. After the emissions inventories
are developed, they are translated into estimates of
air quality conditions under each scenario. Given
the complexity, data requirements, and operating-
costs of state-of-the-art air quality models, and the
project's resource constraints, the EPA Project Team
adopts simplified, linear scaling approaches for some
gaseous pollutants. However, for particulate mat-
ter, ozone, and other air quality conditions that in-
volve substantial non-linear formation processes and/
or long-range atmospheric transport and transfor-
mation, the EPA Project Team invests die time and
resources needed to use more sophisticated model-
ing systems. For example, we exercise EPA's Re-
gional Acid Deposition Model/Regional Particulate
Model (RADM/RPM) to estimate secondarily
formed particulate matter in the eastern U.S.
Up to this point of the analysis, modeled condi-
tions and outcomes establish the Pre-CAAA and
Post-CAAA scenarios. However, at the air quality
modeling step, the analysis returns to a foundation
based 011 actual historical conditions and data. Spe-
cifically, actual 1990 historical air quality monitor-
ing data are used to define the baseline conditions
from which the Pre-CAAA and Post-CAAA sce-
nario air quality projections are constructed. We
derive air quality conditions under the Pre-CAAA
scenario by scaling the historical data adopted for
the base-year (1990) by the ratio of the modeled Pre-
CAAA and base-year air quality. Wre use the same
approach to estimate future-year air quality for the
Post-CAAA scenario. This method takes advantage
of the richness of the monitoring data on air qual-
ity, provides a realistic grounding for the benefit
measures, and yet retains analytical consistency by
using the same modeling process for both scenarios.
The outputs of this step of the analysis are profiles
for each pollutant characterising air quality condi-
tions at each monitoring site in the lower 48 states.
The Pre-CAAA and Post-CAAA scenario air
quality profiles serve as inputs to a modeling system
that translates air quality to physical outcomes (e.g.,
mortality, emergency room visits, or crop yield
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
losses) through the use of concentration-response
functions. Scientific literature on the health and
ecological effects of air pollutants provides the source
of these concentration-response functions. At this
point, we derive estimates of the differences between
the two scenarios in terms of incidence rates for a
broad range of human health and other effects of air
pollution by year, by pollutant, and by geographic
area.
In the next step, we use economic valuation
models or coefficients to estimate the economic value
of the reduction in incidence of those adverse effects
amenable to monetization. For example, a distribu-
tion of unit values derived from the economic litera-
ture provides estimates of the value of reductions in
mortality risk. In addition, we compile and present
benefits that cannot be expressed in economic terms.
In some cases, we calculate quantitative estimates of
scenario differences in the incidence of a
nonmonctizcd effect. In many cases, available data
and techniques are insufficient to support anything
more than a qualitative characterization of the change
in effects.
Next, we compare costs and monetized benefits
to provide our primary estimate of the net economic
benefits of the 1990 CAAA and associated programs,
and a range of estimates around that primary esti-
mate reflecting quantified uncertainties associated
with the physical effects and economic valuation
steps. The monetized benefits used in die net ben-
efit calculations reflect only a portion of the total
benefits due to limitations in analytical resources,
available data and models, and the state of the sci-
ence. For example, in many cases wfe are unable to
quantify or monetize the potentially large benefits
of air pollution controls that result from protection
of the health, structure, and function of ecosystems.
In addition, although available scientific studies dem-
onstrate clear links between air quality changes and
changes in many human health effects, the available
studies do not always provide the data needed to
quantify and/or monetize some of these effects.
Finally, we present a limited set of alternative
benefit estimates which reflect methods, models, or
assumptions that differ from those we used to de-
rive the primary net benefit estimate. We also quan-
tify some of the uncertainties surrounding these al-
ternative estimates. In addition, beyond those vari-
ables for which alternative results are estimated, we
conduct sensitivity analyses for a number of vari-
ables that may influence the primary net benefit es-
timate. The primary estimate and the range around
this estimate, however, reflect our current interpre-
tation of the available literature; our judgments re-
garding the best available data, models, and model-
ing methodologies; and the assumptions we consider
most appropriate to adopt in the face of important
uncertainties.
In addition, throughout the report at the end of
the chapter we summarize the major sources of un-
certainty for each analytic step. Although the im-
pact of many of these uncertainties cannot be quan-
tified, we qualitatively characterize the magnitude
of effect on our net benefit results by assigning one
of two classifications to each source of uncertainty:
potentially major factors could, in our estimation,
have effects of greater than five percent of the total
net benefits; and probably minor factors likely have
effects less than five percent of total net benefits.
The CAA requires RPA to consult with an out-
side panel of experts during the development and
interpretation of the 812 studies. This panel of ex-
perts was organized in 1991 under the auspices of
EPA's Science Advisory Board (SAB) as the Advi-
sory Council 011 Clean Air Act Compliance Analy-
sis (hereafter, the Council). Organizing the review
committee under the SAB ensured that highly quali-
fied experts would review the section 812 studies in
an objective, rigorous, and publicly open manner
consistent with the requirements and procedures of
the Federal Advisory Committee Act (FACA).
Council review of the present study began in 1993
with a review of the analytical design plan. Since
the initial June 1993 meeting, the Council has met
many times to review proposed data, proposed meth-
odologies, and interim results. While the full Coun-
cil retains overall review responsibility for the sec-
tion 812 studies, some specific issues concerning
physical effects and air quality modeling were re-
ferred to subcommittees comprised of both Council
members and members of other SAB committees.
The Council's Health and Ecological Effects Sub-
committee (ITEES) met several times and provided
-------
Chapter 1: Introduction
its own review findings to the full Council. Simi-
larly, the Council's Air Quality Modeling Subcom-
mittee (AQMS) held in-person and teleconference
meetings to review methodology proposals and
modeling results and conveyed its review recommen-
dations to the parent committee.
An interagency review was conducted, during
which a number of analytical issues were discussed.
Conducting a benefit/cost analysis of a major stat-
ute such as the Clean Air Act requires scores of meth-
odological decisions. Many of these issues are the
subject of continuing discussion within the economic
and policy analysis communities and within the
Administration. Key issues include the treatment
of uncertainty in the relationship between particu-
late matter exposure and mortality; the valuation of
premature mortality; the treatment of tax interac-
tion effects; the assessment of stratospheric ozone
recovery; and the treatment of ecological and wel-
fare effects. These issues could not be resolved within
the constraints of this review. Thus, this report re-
flects the findings of the EPA and not necessarily-
other agencies of the Administration.
The remainder of the main text of this report
summarizes the key methodologies and findings our
prospective study.
• Chapter 2 summarizes emissions modeling
and key elements of the regulatory scenarios.
• Chapter 3 discusses the direct cost estima-
tion.
• Chapter 4 presents the air quality modeling
methodology and sample results.
• Chapter 5 describes the approaches used and
principal results obtained through the hu-
man health effects estimation process.
* Chapter 6 describes the human health effects
economic valuation methodology and re-
sults.
• Chapter 7 summarizes the ecological and
other welfare effects analyses, including as-
sessments of commercial timber, agriculture,
visibility, and other categories of effects.
* Chapter 8 presents the aggregated results of
the cost and benefit estimates and describes
and evaluates important uncertainties in the
results.
Additional details regarding the methodologies
and results are presented in the appendices and in
the referenced supporting documents.
* Appendix A provides additional detail on the
sector-specific emissions modeling effort.
• Appendix B covers the direct costs.
• Appendix C provides details of the air qual-
ity models used and results obtained.
* Appendix D presents the human health ef-
fects estimation methodology and results.
* Appendix E describes the ecological benefits
estimation methods and results.
• Appendix F presents the agricultural benefits
estimation methodology and results.
• Appendix G provides details of the strato-
spheric ozone analysis.
• Appendix H describes the methods and as-
sumptions used to value quantified effects
of the CAA in economic terms.
* Appendix 1 describes areas of research which
may increase comprehensiveness and/or re-
duce uncertainties in effect estimates for fu-
ture assessments.
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[Thispage left blank intentionally.]
-------
Emissions
Estimation of pollutant emissions, a key com-
ponent of this prospective analysis, serves as the start-
ing point for subsequent benefit and cost estimates.
We focused the emissions analysis on six major pol-
lutants: volatile organic compounds (YOCs), nitro-
gen oxides (NOJ, sulfur dioxide (SO,), carbon mon-
oxide (CO), particulate matter with an aerodynamic
diameter of 10 microns or less (PM]0), and fine par-
ticulate matter (PM25).1 For each of these pollut-
ants we projected 1990 emissions to the years 2000
and 2010 under two different scenarios: a) the Pre-
CAAA scenario which assumes no additional con-
trol requirements would be implemented beyond
those m place when the 1990 Amendments were
passed; and b) the Post-(]AAA. scenario which incor-
porates the effects of controls authorized by the 1990
Amendments. We compare the emissions estimates
under each of these scenarios to forecast the effect
of the CAAA requirements on future emissions.
This chapter consists of four sections. The first
section provides an overview7 of our approach for
developing the Pre- and Post-CAAA control sce-
narios and projecting emissions from 1990 levels to
2000 and 2010. The second section summarizes our
emissions projections for the years 2000 and 2010
and presents our estimates of changes in future emis-
sions resulting from the implementation of the 1990
Amendments. The third section compares these re-
sults writh other estimates that are based upon more
1 We also estimated ammonia (NH,) emissions. NH, in-
fluences the formation of secondary PM (PM formed as a result
of atmospheric chemical processes). We used NH, emissions
estimates as an input during the air quality modeling phase of
the prospective analysis when estimating future-year ambient
PM concentrations. However, we did not examine the human
health and environmental effects of exposure to NH In addi-
tion to NH,, we also estimated mercury (Hg) emissions. We
qualitatively evaluated the effects of Hg emissions on ecologi-
cal systems, but we did not examine the impact of Hg on hu-
man health. We did not estimate the effect of the CAAA on
lead (Pb) emissions. By 1990 most major airborne Pb emission
sources were already controlled and the CAAA has minimal
additional impact on Pb emissions.
recent emissions data. Finally, we conclude this chap-
ter writh a summary of the key uncertainties associ-
ated writh estimating emissions.
Overview Of
We projected emissions for five major source
categories: industrial point sources, utilities, nonroad
engines/vehicles, motor vehicles, and area sources
(see Table 2-1).2 The basic method involves esti-
mating emissions in the 1990 base-year, adjusting
the base-year emissions to reflect projected growth
in the level of pollution-generating activity by 2000
and 2010 in the absence of additional CAAA require-
ments, and modifying these projections to reflect
future-year control assumptions. The resulting esti-
mates depend largely upon three factors: the method
for selecting the base-year inventory, the indicators
used to forecast growth and the effectiveness of fu-
ture controls, and the specific regulatory programs
incorporated in the Pre- and Post-CAAA scenarios.
We constructed the base-year inventory using
1990 emissions levels. For all of the air pollutants
examined in this analysis except particulate matter,
we selected emissions levels from Version 3 of the
National Particulates Inventory (NPI) to serve as the
baseline. This inventory consists of emissions data
compiled primarily by the National Acid Precipita-
tion Assessment Program (NAPAP), EPA's Office
of Mobile Sources (OMS), and the Federal Highway
Administration (FHWA). For both PM,5 and PMin,
however, wre updated NPI estimates to incorporate
changes in the methodology used to calculate fugi-
tive dust emissions. Adoption of this new technique,
also used to develop EPA's National Emission Trend
1 We estimated utility and industrial point source emis-
sions at die plant/facility level. We estimated nonroad engine/
vehicle, motor vehicle, and area source emissions at the county
level.
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 2-1
Major Emissions Source Categories
Source Category
Examples
Industrial Point Sources
boilers, cement kilns, process heaters, turbines
Utilities
electricity producing utilities
Nonroad Engines/Vehicles
aircraft, construction equipment, lawn and garden equipment,
locomotives, marine engines
Motor Vehicles
buses, cars, trucks (sources that usually operate on roads and
highways)
Area Sources
agricultural tilling, dry cleaners, open burning, wildfires
(NET) PM,5 and PM.|() inventory, leads to lower
estimates of fugitive dust emissions and therefore of
overall primary PM.J
Once we established the base-year inventory, we
projected emissions to the years 2000 and 2010, ac-
counting for the influences expected to cause future
emissions to differ from 1990 levels. For all but util-
ity sources, we rely on an emissions analysis using
the Emissions Reduction and Cost Analysis Model
(ERCAM) which incorporates the effects of the level
of pollution-generating activity and the stringency
and success of regulations designed to protect air
quality. In this analysis, we view changes in eco-
nomic growth as an important indicator of future
activity levels and thus, future emissions. We used
1995 Bureau of Economic Analysis (BEA) Gross
State Product (GSP) projections to forecast the
growth of emissions from industrial point sources.
We relied on BEA GSP projections as well as data
on BEA predicted changes in population to estimate
future emissions from nonroad and area sources.4
We used BEA population growth as an indicator of
the increase in nonroad emissions from recreational
marine vessels, recreational vehicles, and lawn/gar-
den equipment as well as an indicator of the increase
in area source solvent emissions (e.g., VOC emis-
sions from dry cleaners). For motor vehicle sources,
we estimated the growth in activity based primarily
on the projected increase in vehicle miles traveled
(VMT). Wre develop future VMT estimates using
the EPA MOBILE fuel consumption model.
3 Primary PM consists of directly emitted particles such as
wood smoke and road dust. Secondary PM forms in the atmo-
sphere as a result of atmospheric chemical reactions.
4 The growth forecast for area source agricultural tilling is
based oil projections oi acres planted, not BEA GSP and popu-
lation projections.
We estimated the impact of CAAA regulations
on industrial point source, nonroad, motor vehicle,
and area source emissions based on expected control
efficiency and rule effectiveness. Control efficiency
represents the percentage reduction in emissions
anticipated as a result of the implementation of the
CAAA, assuming full compliance and successful
operation of all control mechanisms. The rule ef-
fectiveness factor accounts for equipment malfunc-
tion, non-compliance, and other circumstances that
influence the overall effectiveness of air pollution
regulations. We selected a rule effectiveness of 80
percent as the standard for this analysis which we
applied to stationary source NOx and VOC con-
trols.5 Rule effectiveness was not calculated for mo-
bile source controls as an adjustment factor separate
from the emissions rates estimated for the various
vehicle classes.
To estimate future utility source emissions, we
relied on the Integrated Planning Model (IPM). This
optimization model forecasts, for the 48 contiguous
states and the District of Columbia, emissions from
all existing utility power generation units, as well as
from independent power producers and other co-
generation facilities that sell wholesale power and
are included in the North American Electric Reli-
ability Council (NERC) data base for reliability plan-
ning. The model considers future capacity7 additions
by both utilities and independent power producers
which might cause an increase in emissions. In addi-
tion, the model is capable of producing baseline air
3 At the time we selected the general rule effectiveness for
use in this analysis, 80 percent was the standard factor applied in
air pollution modeling. More recent analyses have used higher
rule effectiveness values. If a higher rule effectiveness value had
been used in this analysis, emissions reduction estimates would
be larger and the estimated benefits associated with air quality
improvements would be greater.
10
-------
Chapter 2: Emissions
emissions forecasts and estimates of air emissions
levels under various control options at the national
and NERC regional and subregioiial level. We used
IPM to estimate base-year (1990) utility source emis-
sions and to project future-year (2000 and 2010)
emissions under both the Pre- and Post-CAAA sce-
narios.
Using emissions analysis or IPM, we estimated
future emissions for each of the five major source
categories under both the Pre- and Post-CAAA sce-
narios. While the selection of the base-year inven-
tory, emission growth factors, and rate of regula-
tory effectiveness all influence the emissions projec-
tions, the difference between Pre- and Post-CAAA
estimates is primarily determined by the difference
in control assumptions incorporated in the two pro-
jection scenarios.
We developed two contrasting emissions con-
trol scenarios, the Pre-CAAA scenario and the Post-
CAAA scenario. The Pre-CAAA scenario maintains
the air pollution regulatory requirements which ex-
isted in 1990 through the 2000 and 2010 analytical
period and serves as a baseline against which we
measure the changes in emissions projected under
the Post-CAAA scenario.6 This latter scenario as-
sumes the implementation of the 1990 Clean Air Act
Amendments and incorporates the influences of the
following provisions:
• Title I VOC and NOx reasonably available
control technology (RACT) and reasonable
further progress (RFP) requirements for
ozone nonattamment areas;
• Title II motor vehicle and nonroad engine/
vehicle provisions;
• Title III 2- and 4-ycar maximum achievable
control technology (MACT) standards;
* Title IV SO, and NO emissions programs
for utilities;
6 We also attempted to incorporate in the Pre-CAAA
(baseline) scenario the non-CAAA regulations and policies we
expect will have a significant effect on emissions between 1990
and 2010. For example, the TPM, which we used to estimate
utility emissions, incorporates the effect of the deregulation of
railroad rates on SO, emissions. IPM accounts for the influence
of the future cost of low-sulfur coal prices expected to occur as
a result oi lower railroad rales. The impact oi prescribed burn-
ing policies ior private and federally owned lands on PM emis-
sions is also incorporated in the Pre-CAAA scenario.
• Title V permitting system for primary
sources of air pollution; and
• Title VI emissions limits for chemicals that
deplete stratospheric ozone.'
The Post-CAAA scenario also assumes the imple-
mentation of region-wide NOx controls and a cap-
and-trade system designed to reduce emissions dur-
ing the summer months from large utility and in-
dustrial sources in the 37 easternmost states that com-
prise the Ozone Transport Assessment Group
(OTAG) domain.8 In addition, the Post-CAAA sce-
nario incorporates the effects of a similarly designed
trading program for the 11 northeast states that com-
prise the Ozone Transport Region (OTR). This trad-
ing program is consistent with Phase II of the Ozone
Transport Commission (OTC) Memorandum of
Understanding (MOU).9 \Vc provide more detailed
discussion of both Pre- and Post-CAAA scenario
development in Appendix A.
Results
The results of the Pre- and Post-CAAA projec-
tions indicate that the 1990 Clean Air Act Amend-
ments will likely have a significant effect on future
emissions of air pollutants. Table 2-2 displays both
base-year (1990) and future-year (2000 and 2010)
emissions estimates for the modeled scenarios along
with the percent change from Pre- to Post-CAAA
VOC, NOx, SO2, CO, PM1(), and PM2, projections.
A more detailed breakout of 2010 Pre- and Post-
CAAA emissions estimates, displaying emissions for
each major source category, is contained in Table 2-
3. Figures 2-1 through 2-6 show the emissions pro-
jections for each of the pollutants examined in this
analysis.
Emissions projections for VOC, NOx, SO,, and
CO, displayed in Figures 2-1 through 2-4, follow
' For a more detailed discussion of the CAAA provisions
incorporated in the Post-CAAA scenario, see Appendix A.
8 The NO control program incorporated in the Post-
CAAA scenario may not reflect the NO., controls that are actu-
ally implemented in a regional ozone transport rule.
' The Posl-CAAA scenario does not incorporate any in-
fluences oi file recently revised PM and o^one NAAQS regula-
tions or any impact of the recently proposed Tier II tailpipe
standards.
11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 2-2
Summary of National Annual Emissions Projections
(thousand tons)
Pollutant
VOC
NOX
SO2
CO
Primary
PM10
Primary
PM2.5
1990
Base-
Year
22,715
22,747
22,361
94,385
28,289
7,091
2000
Pre-
CAAA
24,410
25,021
24,008
95,572
28,768
7,353
2000
Post-
CAAA
17,874
18,414
18,013
80,919
28,082
7,216
2000
%
Change
-27%
-26%
-25%
-15%
-2%
-2%
2010
Pre-
CAAA
27,559
28,172
26,216
107,034
28,993
7,742
2010
Post-
CAAA
17,877
17,290
18,020
81,943
28,035
7,447
2010
%
Change
-35%
-39%
-31%
-23%
-3%
-4%
Notes: Totals reflect emissions for the 48 contiguous States, excluding Alaska and Hawaii.
Percent change between Pre-CAAA and Post-CAAA scenarios.
similar patterns. Pre-CAAA estimates indicate emis-
sions of these pollutants would increase, on average,
by almost 20 percent from 1990 to 2010. These in-
creases reflect the expectation that anticipated growth
in activity levels in the relevant emitting sectors will
more than offset reductions achieved by pre-1990
control programs. While we predict relatively steady
growth in emissions in the absence of the 1990
Amendments, projections show emissions of these
four pollutants would increase at a slightly faster rate
over the last ten years of the 20 year projection pe-
riod.
Post-CAAA estimates of VOC, NOx, SO,, and
CO emissions for the modeled regulatory scenarios
decrease significantly from 1990 to 2000 and then
plateau, remaining relatively constant from 2000 to
2010. The initial decrease is triggered by the imple-
mentation of die CAAA and the associated controls.
After cleaner means of production are adopted, bet-
ter emissions control technologies are implemented,
and other required changes and improvements are
made, emissions reduction slows and in some in-
stances stops all together; emissions may even begin
to increase. Although the Post-CAAA estimates for
each of the above mentioned pollutants show little
or no change in the level of emissions from 2000 to
2010, an overall comparison of our Pre- and Post-
CAAA projections indicates that during this time
period the 1990 Amendments continue to have an
increasingly beneficial effect on emission levels.
Comparison of Pre- and Post- CAAA emissions
estimates reveals that by 2010, estimated VOC emis-
sions will be 35 percent lower as a result of the imple-
mentation of the CAAA than they would have been
if no new control requirements, beyond those in
place in 1990, were mandated. This sizeable change
in emissions attributable to the Amendments is due
largely to estimated VOC reductions from motor
vehicle and area sources. The 2010 Post-CAAA es-
timate for these two source categories combined is
8.2 million tons lower than the Pre-CAAA projec-
tion, a total which accounts for 84 percent of the
predicted difference in VOC emissions estimated
under the two scenarios.
Based on the regulatory programs incorporated
in the Post-CAAA scenario, we project that NOx
emissions will be reduced by the greatest percent-
age. Comparison of projections for the year 2010
indicates the Post-CAAA NOx estimate is 39 per-
cent lower than the Pre-CAAA estimate, represent-
ing a decrease in emissions of 10.8 million tons. We
project nearly half of this reduction will come from
utilities, while the remaining portions will come from
cuts in motor vehicle and non-utility point source
emissions.
12
-------
Chapter 2: Emissions
Figure 2-3 shows that by 2010 we anticipate SO,
levels will be 31 percent lower than they would have
been under the Pre-CAAA scenario. We project 96
percent of the 8.2 million ton difference between
Pre- and Post-CAAA estimates will result from regu-
lation of utilities, while the remaining reduction
comes from motor vehicles.
We estimate 2010 Post-CAAA CO emissions
will be 81.9 million tons, 23 percent lower than the
Pre-CAAA projection. Much of this reduction we
project will be achieved as a result of nonattainment
(Title I) and motor vehicle provisions (Title II) of
the 1990 Amendments. The more influential pro-
grams (m order of importance) are expected to be
enhanced vehicle emission inspections, wintertime
oxygenated fuel use, and LEV program adoption.
Figures 2-5 and 2-6 indicate that the 1990 Clean
Air Act Amendments have more modest effects on
primary PM10 and PM25 emissions.10 For both of
these pollutants, Pre-CAAA projections increase at
a slow rate from 1990 to 2010. Post-CAAA emis-
sions estimates for primary PM10 and PM2_, how-
ever, follow different paths. While we estimate
implementation of the CAAA will cause primary
PM10 levels to slowly decrease from 1990 to 2010,
Post-CAAA projections indicate primary PM23 emis-
sions will actually rise despite the influence of the
CAAA. Overall, however, emissions of primary
PM10 and PM,5 both will be approximately four per-
cent lower in 2010 than they would have been with-
out the CAAA.11
The significant influence of area source emissions
on primary PM emissions levels, combined with the
limited regulation of this major source category,
explains the limited effect of the CAAA on primary
particulate matter emissions. According to data used
in this analysis, area sources account for over 90 per-
cent of primary PMin emissions and over 80 percent
*c EPA projected PM1(, and PM,/5 levels holding natural
source emissions of particulate matter constant at 1990 levels.
The estimates presented in Figures 2-5 and 2-6 have been ad-
justed; these estimates represent total PM emissions minus natu-
ral source emissions (wind erosion).
:" Directly emitted PM, such as fugitive dust, is referred to
as primary PM. Secondary PM is not directly emitted, but rather
iorms in the atmosphere. NO^ and SO0 are two examples oi
secondary PM precursors.
of primary PM,_ emissions.12 As a result, even the
successful reduction of motor vehicle and nonroad
emissions have only a slight impact on overall pri-
mary PM]0 and PM25 estimates developed for this
study.13 Furthermore, the CAAA's most significant
primary PM area source controls target emissions in
counties not in compliance with the National Am-
bient Air Quality Standards (NAAQS).14 Currently,
however, there are fewer than 85 counties in the
country that are not in attainment with the national
standards. Emissions changes in these areas are ca-
pable of having only a minor influence on the over-
all primary PM level in the United States. Even
minor changes in primary PM emissions leading to
minor changes in the concentrations of this pollut-
ant, however, are significant. In the subsequent
portions of this analysis, sizable benefits are estimated
to result from small reductions in PM concentra-
tions in the atmosphere.
The seemingly small impact on direct PM emis-
sions resulting from implementation of the CAAA
depicted in Figures 2-5 and 2-6 can be misleading.
While these figures illustrate the impact of the 1990
CAAA on primary PM emissions, it is important to
remember that ambient PiVt concentrations are in-
fluenced by the presence of both primary and sec-
ondary PM. VOCs, NOx, and SO,, all pollutants
regulated by the CAA, are secondary PM precur-
sors. The reduction in the emissions of these three
pollutants also leads to lower overall PM concentra-
tions in the atmosphere. The complete impact of
the CAAA on PM thus is not fully captured by Fig-
ures 2-5 and 2-6. Additional discussion of the influ-
ence of the CAAA on PM and ambient air quality is
provided in Chapter 4 and Appendix C.
As part of this prospective analysis we also esti-
mated future-year NH emissions The 1990 Amend-
ments, however, do not include provisions designed
;2 A.s discussed on pages 18 and 20 and in Table 2-5, how-
ever, some recent data indicate that the composition data used
in this analysis may underestimate the contribution from mo-
tor vehicle carbonaceous emissions.
13 The difference between 2010 Pre- and Post-CAAA esti-
mates for PM and PM motor vehicle emissions is 31 percent
and 39 percent respectively. Hie difference between 2010 Pre-
and Post CAAA estimates for PM]0 and PM2R nonroad emis-
sions is 19 percent and 20 percent respectively.
!" The PA1 NAAQS referred to here is the 50 ug/m' (an-
nual mean) 150 ug/m3 (daily mean) standard.
13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 2-3
Summary by Source Category of National Annual Emission Projections to 2010
(thousand tons)
Source
Pollutant Category
VOC Utility
Point
Area
Nonroad
Motor Vehicle
TOTAL
NOX Utility
Point
Area
Nonroad
Motor Vehicle
TOTAL
CO Utility
Point
Area
Nonroad
Motor Vehicle
TOTAL
SO2 Utility
Point
Area
Nonroad
Motor Vehicle
TOTAL
Primary Utility
PM10 Point
Area
Nonroad
Motor Vehicle
TOTAL
Primary Utility
PM25 Point
Area
Nonroad
Motor Vehicle
TOTAL
1990
37
3,500
10,000
2,100
6,800
23,000
7,400
2,900
2,200
2,800
7,400
23,000
330
6,000
12,000
14,000
62,000
94,000
16,000
4,600
1,000
240
570
22,000
280
930
26,000
340
360
28,000
110
590
5,800
290
290
7,100
2010
Pre-CAAA
49
4,200
13,000
2,600
7,300
28,000
9,100
3,600
3,000
3,400
9,100
28,000
450
7,400
14,000
19,000
66,000
107,000
18,000
6,000
1,500
240
770
26,000
310
1,200
27,000
410
300
29,000
120
750
6,300
360
230
7,700
2010
Post-CAAA
50
3,500
8,500
1,900
3,900
18,000
3,800
2,200
3,000
2,700
5,600
17,000
460
7,400
14,000
18,000
42,000
82,000
9,900
6,000
1,500
240
410
18,000
280
1,200
26,000
340
210
28,000
110
750
6,100
290
140
7,400
% Change
2%
-1 9%
-36%
-28%
-46%
-35%
-58%
-39%
-1%
-20%
-39%
-39%
2%
0%
0%
-4%
-37%
-23%
-44%
0%
0%
0%
-47%
-31%
-9%
0%
-3%
-1 9%
-31%
-3%
-8%
0%
-2%
-20%
-39%
-4%
NOTES: Table may not sum due to rounding. Percentage change was calculated prior to rounding.
14
-------
Chapter 2: Emissions
to regulate Nil, As a result, the Pre- and Post-
CAAA estimates follow a similar upward trend. We
estimate NH3 emissions will increase roughly 55
percent from 1990 to 2010. Although we do not
estimate the costs and benefits associated with NH3
controls and changes in NH ambient concentrations
as part of this analysis, estimation of NTT. emissions
is an important part of the prospective study. NH3
is a secondary PM precursor, and we relied on fu-
ture-year NH3 emissions estimates as model input
to help us estimate PM concentrations.
We also estimated the effect of CAAA provi-
sions on mercury (Hg) emissions for five separate
Hg emissions sources: medical waste incinerators
(MWI), municipal waste combustors (MWCs), elec-
tric utility plants, hazardous waste combustors, and
chlor-alkali plants.13 Together, these sources account
for 75 to 80 percent of national anthropogenic air-
borne Hg emissions. In this analysis we qualitatively
examine the effects of mercury emissions reductions
on ecological systems (see Chapter 7 and Appendix
E). We do not, however, evaluate the impact of Hg
on human health.
Table 2-4 displays, for each emission category,
base-year (1990) and future-year (2000 and 2010) Pre-
and Post-CAAA emissions estimates. The table also
shows the difference between Pre- and Post-CAAA
estimates for each projection year. Overall, the re-
sults of this analysis indicate that the 1990 Amend-
ments will provide a reduction in Hg emissions of
44.2 tons per year (tpy) in the year 2000 and a reduc-
tion of 56.2 tpy in 2010. These changes represent a
35 percent reduction in airborne mercury emissions
for the year 2000 and a 42 percent reduction for 2010.
We estimate that most of the reduction will be the
result of New Source Performance Standards for
MWI and MWCs.
Table 2-4
Airborne Mercury Emission Estimates
2000 Emissions (tons)
Source Category
Medical Waste Incin.
Municipal Waste Comb.
Electric Utility Generation
Hazardous Waste Comb.
Chlor-Alkali Plants
1990
Emissions
(tons)
50
54
51.3
6.6
9.8
Pre-
CAAA
17.9
31.2
63.0
6.6
6.0
Post-
CAAA
1.3
5.5
61.1
6.6
6.0
Total CAAA Benefits (Reductions)
Diff.
16.6
25.7
1.9
0
0
44.2
2010 Emissions
Pre-
CAAA
22.6
33.8
68.5
6.6
2.0
Post-
CAAA
1.6
6.0
65.4
3.0
1.3
(tons)
Diff.
21.0
27.8
3.1
3.6
0.7
56.2
^ With the exception of electric utility plant Hg emissions
that were estimated using 1PM., we relied on previously gener-
ated estimates (typically from recently conducted RTAs) to evalu-
ate the impact oi the CAAA on Hg emissions. For a more com-
plete discussion of the methodology, see Appendix A.
15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure 2-1
Pre- and Post-CAAA Scenario VOC Emissions
Estimates
K 20 -
I V
W c
E 1 15-
§!
£ 10
Figure 2-2
Pre- and Post-CAAA Scenario NOX Emissions
Estimates
30
£ 25-
o
i_20-
.c tn
-------
Chapter 2: Emissions
Figure 2-4
Pre- and Post-CAAA Scenario CO Emissions
Estimates
120
o
i-
100 -
80
5?
in =
c ° 60
40
20
1980
1990
2000
Year
Re-CAAA
Fbst-CAAA
30
25
20 +
o —
= I 15 ••
| t
w 10 |
5 --
1980
1990
2000
Year
- Pre-CAAA •
- Fbst-CAAA
2010
Figure 2-6
Pre- and Post-CAAA Scenario Primary PM25
Emissions Estimates
Figure 2-5
Pre- and Post-CAAA Scenario Primary PM10
Emissions Estimates
30
25 -
20-
(/> c
c | 15 +
c i
o — 10-
1980
1990 2000
Year
2010
- Re-CAAA —•— Rost-CAAA
17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
of
Comparison of the emissions projections gener-
ated by the prospective analysis to historical emis-
sions estimates drawn from the National Air Pollut-
ant and Emissions Trends reports (Trends) provides
a check on the reasonableness of our emissions in-
ventories. In addition, comparison of emissions pro-
jections from the prospective analysis with those of
the Grand Canyon Visibility Transport Commis-
sion (GCVTC) study of western regional haze pro-
vides an initial test of the sensitivity of emissions
projections to base-year inventories and growth as-
sumptions. Analysis of PM emissions and compari-
son of estimated and observed PM data also help us
evaluate the prospective study's emissions estimation
methods.
'Trends reports contain historical estimates of
annual VOC, NOx, SO,, CO, and PM10, emissions.
While the most recent report only provides emis-
sions data through the first half of the 1990s, com-
parison of these estimates from 1990 to 1996 with
emissions trends projected under the Post-CAAA
scenarios reveals that emissions figures from both
are similar. The disparity that does exist between
die two sets of estimates largely stems from the fact
that the Post-CAAA scenario trend lines running
from 1990 to 2000 consist of only two data points.
As a result, Post-CAAA trend lines cannot capture
yearly fluctuations in emissions and the exact tim-
ing of emissions cuts. Only for NOx are the Trends
and Post-CAAA estimates significantly different; this
is because the Trends report is still in the process of
incorporating the State's periodic emission inventory
into the NET database. As a result, Trends values
do not capture all the NOx emission reductions that
have occurred since 1990. For example, significant
reductions attributable to reasonable available con-
trol technology- (RACT) requirements for major sta-
tionary source NO emitters areas are not reflected
X
in the Trends figures.
The Grand Canyon Visibility Transport Com-
mission conducted an air pollution analysis for West-
ern States that projected emissions for selected pol-
lutants, including NOx, SO_, and PM25, from 1990
base-year levels for the year 2000 and every tenth
subsequent year up to 2040. GCVTC estimates of
future-year emissions levels differ from Post-CAAA
projections. This disparity results from the use of
different base-year inventories in the two studies and
from specific regional reductions not incorporated
in the prospective analysis scenarios. Despite the
difference in GCVTC and Post-CAAA estimates,
the change in the level of emissions from 1990 to
2010 predicted by the two studies is similar. Com-
parison of both sets of projections illustrates the sen-
sitivity of future-year emissions estimates to the base-
year inventory.
The 1997 National Air Quality and Emissions
Trends Report provides a summary of PM,5 con-
centration speciation data. This report shows the
relative contribution of the major PM emissions
source components (crustal material, carbonaceous
particles, nitrate, and sulfatc) to ambient PM,5 con-
centrations in urban and nonurban areas through-
out the U.S.16 Comparison of primary PM,5 emis-
sions estimates generated for this analysis with the
observed concentration data presented in the 1997
report indicates that die ratio in die prospective study
of crustal material to primary carbonaceous particles
is high. At least part of this apparent overestima-
tioii of crustal material and underestimation of car-
bonaceous participates, however, is due to the fact
that much of the emitted crustal material quickly
settles and does not have a quantifiable impact on
ambient air quality. In this analysis, we apply a fac-
tor of 0.2 to crustal emissions to estimate the frac-
tion of crustal PM25 that makes its way into the
"mixed layer" of the atmosphere and influences pol-
lutant concentrations. Figure 2-7 displays the
breakout of primary PM_5 into its adjusted crustal
and carbonaceous (elemental carbon and organic
carbon) components. The figure divides crustal
material into two subcategories, fugitive dust or in-
dustrial sources, based on the source of the material
and also shows the fraction of primary PM,,,. that is
16 Crustal material is directly emitted from fugitive dust
sources such as agricultural operations, construction, paved and
iinpaved roads, and wind erosion as well as from some indus-
trial sources such as metals processing. Carbonaceous particles,
as defined in the 1997 National Air Quality and Emissions
Trends Report," are emitted directly and as condensed liquid
droplets irorn iuel combustion, burning oi forests, rangelands,
and fields; oil highway and highway mobile sources (gas and
diesel); and certain industrial processes".
18
-------
Chapter 2: Emissions
Figure 2-7
1990 Primary PM25 Emissions by EPA Region (tons/year)
Region 1
i
i
Region 2 ^^ Crustal- Fugitive Dust Sources
Crustal - Industrial Sources
fcjion 3 £3 otherprimary
Elemental Carbon
Organic Carbon
o
100,000 450,000 850,000
neither crustal nor carbonaceous. The ratios of ad-
justed crustal material to primary carbonaceous par-
ticles presented in Figure 2-7 are in line with the
observed PM25 concentration data presented in the
1997 report.
Uncertainty In Emission
Estimates
Table 2-5 provides a list of sources of uncertainty
associated with estimating base-year emissions, the
expected direction of bias introduced by each un-
certainty (if known), and the relative significance of
each uncertainty in the overall 812 benefits analysis.
The emissions estimates presented in the prospec-
tive analysis are characterized by three major sources
of uncertainty: estimation of the base-year inven-
tory, prediction of the growth in pollution-generat-
ing activity, and assumptions about future-year con-
trols.
Base-year emissions were estimated using emis-
sions factors that express the relationship between a
particular human/industrial activity and the level of
emissions. The accuracy of base-year emissions esti-
mates varies from pollutant to pollutant, depending
largely on how directly the selected activity and
emissions correlate. We likely estimated 1990 SO2
emissions with the greatest precision. Sulfur diox-
ide emissions are generated during combustion of
sulfur-containing fuel and are directly related to fuel
sulfur content. In addition, we were able to verify
these estimates through comparison with Continu-
ous Emission Monitoring (GEM) data. As a result,
we were able to accurately estimate SO2 emissions
using emissions factors based on data on fuel usage
and fuel sulfur content. Nitrogen oxides are also a
product of fuel combustion, allowing us to estimate
emissions of this pollutant using the same general
technique used to estimate SO2 emissions. However,
the processes involved in the formation of NOx
during combustion are more complicated than those
involved in the formation of SO2; thus, our NOx
emissions estimates are more variable and less cer-
tain than SO2 estimates.
Volatile organic compounds, like SO2 and NOx,
are products of fuel combustion; however, these
compounds are also a product of evaporation. To
estimate evaporative emissions of this pollutant we
19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
used emissions factors that relate changes in emis-
sions to changes in temperature. Because future
meteorological conditions are difficult to predict,
the uncertainty associated with forecasting tempera-
ture influences the uncertainty in our VOC emis-
sions estimates. The likely significance of this un-
certainty, in terms of its impact on the overall mon-
etary benefit present in this analysis, is probably
minor.
In this analysis we estimated primary PM,5 emis-
sions based on unit emissions that may not accu-
rately reflect the composition and mobility of par-
ticles. The ratio of crustal to carbonaceous particu-
late material, for example, likely is high as a result of
overestimation of the fraction of crustal material,
primarily composed of fugitive dust, and underesti-
mation of the fraction of carbonaceous material.
Because the CAAA has a greater impact on emis-
sions sources that generate carbonaceous particles
(mobile sources) than on sources that mainly emit
crustal material (area sources), we likely underesti-
mate the impact of the CAAA on reducing PM2_,
thereby reducing monetary benefits estimates. The
uncertainty associated with estimating the partition
of PiVl,. emissions components could conceivably
have a major impact 011 the net benefit estimate;
compared to secondary PM2_ precursor emissions,
however, changes in primary PM25 emissions have a
relatively small impact on PM related benefits.
We estimated future-year emissions levels based
on expected growth in pollution-generating activi-
ties. Inherent uncertainties and data inadequacies/
limitations exist in forecasting growth for any fu-
ture period. Also, the growth indicators we used in
this analysis may not directly correlate with changes
in the factors that influence emissions. Both of these
factors contribute to the uncertainty associated wfith
this study's emissions results. For example, the best
indicator of pollution-generating activity is fuel use
or some other measure of input/output that most
directly relates to emissions. The key BRA indica-
tor used in this analysis, GSP, is closely correlated
with the pollution-generating activity associated with
many manufacturing industry processes (iron and
steel, petroleum refining, etc.). However, a good
portion of industrial sector emissions are from boil-
ers and furnaces, whose activity is related to produc-
tion, but not as closely as to product output. Activi-
ties such as fuel switching may produce different
emission patterns than those reflected in the results
of this study.
Our future-year control assumptions are also a
source of uncertainty. Despite our efforts to mini-
mize this uncertainty, whether each of the Post-
CAAA controls will be adopted, whether Post-
CAAA control programs will be more or less effec-
tive than estimated, and whether unanticipated tech-
nological shifts will reduce future-year emissions are
all unknown. For example, the Post-CAAA scenario
includes implementation of a region-wide NOx con-
trol strategy designed to regulate the regional trans-
port of ozone. However, the control program as-
sumed under the Post-CAAA scenario may not re-
flect die NOx controls that are actually implemented
in a regional ozone transport rule.
20
-------
Chapter 2: Emissions
Table 2-5
Key Uncertainties Associated with Emissions Estimation
Potential Source of Error
Direction of Potential Bias for
Net Benefits Estimate
Likely Significance Relative to Key
Uncertainties in Net Benefit
Estimate*
PM2 5 emissions are largely
based on scaling of PM-io
emissions.
Overall, unable to determine
based on current information,
but current emission factors are
likely to underestimate PlVh.s
emissions from combustion
sources, implying a potential
underestimation of benefits.
Potentially major. Source-specific
scaling factors reflect the most careful
estimation currently possible, using
current emissions monitoring data.
However, health benefit estimates
related to changes in PlVh.5 constitute
a large portion of overall CAAA-related
benefits.
Primary PlVb.s emissions
estimates are based on unit
emissions that may not
accurately reflect composition
and mobility of the particles.
For example, the ratio of
crustal to primary
carbonaceous particulate
material likely is high.
Underestimate. The effect of
overestimating crustal emissions
and underestimating
carbonaceous when applied in
later stages of the analysis, is to
reduce the net impact of the
CAAA on primary PlVb.5
emissions by underestimating
PlVhs emissions reductions
associated with mobile source
tailpipe controls.
Potentially major. Mobile source
primary carbonaceous particles are a
significant contributor to public
exposure to PIvb.s. Overall, however,
compared to secondary PlVb.5
precursor emissions, changes in
primary PIvh.s emissions have only a
small impact on PlVb.s related benefits.
The Post-CAAA scenario
includes implementation of a
region-wide NOX emissions
reduction strategy to control
regional transport of ozone
that may not reflect the NOX
controls that are actually
implemented in a regional
ozone transport rule.
Unable to determine based on
current information.
Probably minor. Overall, magnitude of
estimated emissions reductions is
comparable to that in expected future
regional transport rule. In some areas
of the 37 state region, emissions
reductions are expected to be
overestimated, bur in other areas, NOX
inhibition of ozone leads to
underestimates of ozone benefits
(e.g., some eastern urban centers).
VOC emissions are dependent
on evaporation, and future
patterns of temperature are
difficult to predict.
Unable to determine based on
current information.
Probably minor. We assume future
temperature patterns are well
characterized by historic patterns, but
an acceleration of climate change
(warming) could increase emissions.
Use of average temperatures
(i.e., daily minimum and
maximum) in estimating
motor-vehicle emissions
artificially reduces variability in
VOC emissions.
Unable to determine based on
current information.
Probably minor. Use of averages will
overestimate emissions on some days
and underestimate on other days.
Effect is mitigated in Post-CAAA
scenarios because of more stringent
evaporative controls that are in place
by 2000 and 2010.
21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 2-5 (continued)
Key Uncertainties Associated with Emissions Estimation
Potential Source of Error
Direction of Potential Bias for
Net Benefits Estimate
Likely Significance Relative to Key
Uncertainties in Net Benefit
Estimate*
Economic growth factors used
to project emissions are an
indicator of future economic
activity. They reflect
uncertainty in economic
forecasting as well as
uncertainty in the link to
emissions.
Unable to determine based on
current information.
Probably minor. The same set of
growth factors are used to project
emissions under both the Pre-CAAA
and Post-CAAA scenarios, mitigating
to some extent the potential for
significant errors in estimating
differences in emissions.
Uncertainties in the
stringency, scope, timing, and
effectiveness of Post-CAAA
controls included in projection
scenarios.
Unable to determine based on
current information.
Probably minor. Future controls could
be more or less stringent, wide-
reaching (e.g., NOX reductions in
OTAG region - see above), or
effective (e.g., uncertainty in realizing
all Reasonable Further Progress
requirements) than projected. Timing
of emissions reductions may also be
affected (e.g., sulfur emissions
reductions from utility sources have
occurred more rapidly than projected
for this analysis).
* The classification of each potential source of error reflects the best judgement of the section 812 Project Team. The
Project Team assigns a classification of "potentially major" if a plausible alternative assumption or approach could
influence the overall monetary benefit estimate by approximately five percent or more; if an alternative assumption or
approach is likely to change the total benefit estimate by less than five percent, the Project Team assigns a classification
of "probably minor."
22
-------
Costs
The costs of complying with the requirements
of the Clean Air Act Amendments (CAAA.) of 1990
will affect all levels of the U.S. economy. The im-
pact, initially experienced through the direct costs
imposed by regulations promulgated under the
amendments, will also be seen in patterns of indus-
trial production, research and development, capital
investment, productivity, employment, and con-
sumption. The purpose of the analysis summarized
in this chapter is to estimate die incremental change
in annual compliance costs from 1990 to 2010 that
are directly attributable to the 1990 Clean Air Act
Amendments.
This chapter consists of four sections. The first
section summarizes our approach to estimating di-
rect compliance costs. In the second section we
present the results of the cost analysis. We first re-
port the total costs of Titles I through V and then
present estimates for major individual provisions.
We also briefly discuss our derivation of Title VI
costs. In the third section, we provide a qualitative
discussion of the potential magnitude of social costs
and other impacts associated with the Amendments
to characterize the potential welfare loss not cap-
tured in the direct cost approach. We conclude the
chapter with a discussion of the major analytic un-
certainties and include the results of quantitative sen-
sitivity tests of key data and assumptions.
to
Direct
As discussed in the previous chapter, the first
step of the prospective analysis required the devel-
opment of emission estimates for the base-year, 1990,
and for the two target years in our analytic time
period, 2000 to 2010. We developed two scenarios,
Pre-CAAA and Post-CAAA, that reflect three key
parameters: (i) base-year inventory selection, (li) in-
dicators of forecasted economic growth, and (iii) ef-
fects of future year controls and selected CAAA pro-
visions. The Pre-CAAA scenario applies the strin-
gency and scope of air pollution regulations as they
existed 111 1990 and projects emissions and costs to
2000 and 2010. This scenario establishes a baseline
that represents projected emission levels and con-
trol costs 111 the absence of the 1990 Amendments.
Under the Post-CAAA scenario, costs are based on
compliance with selected CAAA provisions. To-
gether these two scenarios form the foundation upon
which the incremental costs and benefits of comply-
ing with the 1990 Amendments are estimated. For
more information on the development of these sce-
narios, see Chapter 2.
We closely integrate the modeling of direct com-
pliance costs with emissions projections by main-
taining consistency among control assumptions (i.e.
emissions scenarios) used as inputs in the cost esti-
mation modeling and in the analysis of emissions
projections and benefits. Wre use two models to es-
timate costs, Emission Reduction and Cost Analysis
Model (ERCAM) and Integrated Planning Model
(IPM). These models generate cost estimates for the
Post-CAAA scenarios in two projection years, 2000
and 2010. The estimates are calculated relative to
costs under the same year Pre-CAAA scenario, so
estimates represent incremental costs of compliance
wfith the 1990 Amendments.
We use ERCAM to estimate costs associated with
regulating particulate matter (PM), volatile organic
compounds (VOCs), and non-utility source oxides
of nitrogen (NOJ.1 The model is essentially a cost-
accounting tool that provides a structure for modi-
fying and updating changes in inputs while main-
1 This model was developed by E. H. Pechan & Associ-
ates, Inc. io iacililale EPA's analysis oi emissions control.
23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
taming consistency with the emission and cost analy-
ses. Cost scenarios and assumptions are developed
for each non-utility source category (e.g., point, area,
nonroad, and motor vehicle sources) and in response
to specific provisions and emission targets. The
model estimates costs based on inputs such as cost
per ton, source-specific cost equations, incremental
production, and operating cost estimates. For this
analysis, we collected data and inputs from informa-
tion presented in regulatory impact assessments
(RIAs), background information documents (BIDs),
regulatory support documents, and Federal Regis-
ter notices.
To estimate the costs of reducing utility NOx
and sulfur dioxide (SO,,) emissions, we use the Inte-
grated Planning Model (IPM). IPM allows us to es-
timate the control costs of several pollutants while
maintaining consistent control scenarios and eco-
nomic forecasts of the electric power industry. It
assesses the optimal mix of pollution control strate-
gies subject to a series of specified constraints. Key
inputs and constraints in the model include targeted
emissions reductions (on a seasonal or annual basis),
costs and constraints of control technology, and eco-
nomic parameters (e.g., forecasted demand for elec-
tricity, power plant availability/capacity, costs of
fuel, etc.)
To assess the costs of reducing emission of pol-
lutants or sectors not covered by our two models,
we estimate costs using the best available cost equa-
tions or other types of analyses. For example, we
estimate non-utility SO emission control costs for
point sources by applying source-specific cost equa-
tions for flue gas desulfurization (FGD)/scrubber
technology to affected sources in 2000 and 2010.
While we do not explicitly model CO attainment
costs, we include in the analysis the costs of pro-
grams designed to reduce CO emissions, such as oxy-
genated fuels and a cold temperature CO motor ve-
hicle emission standard. Finally, to estimate costs
of the rate of progress/reasonable further progress
(ROP/RFP) provisions, requirements under Title I
that require ozone nonattainment areas to make
steady progress toward attainment, we first estimate
the emissions reduction shortfall that must be
achieved in each target year in each nonattainment
area, and then apply a cost per ton estimate from a
schedule of measures that could be applied locally
to meet the necessary ROP/RFP requirement. For
more detail on the specific methods used to estimate
compliance costs for each pollutant and source cat-
egory, see Appendix B.
The cost estimates in this chapter are the incre-
mental costs of the 1990 Amendments (i.e. the dif-
ference between pre- and Post-CAAA cost estimates).
We present die results as total annualized costs (TAC)
in 2000 and 2010. Annualized costs include both
capital costs, such as costs of control equipment, and
operation and maintenance (O&M) costs.2 They
do not represent actual cash flow in a given year,
but are rather an estimate of average annual burden
over the period during which firms will incur costs.
In annualizing costs, we convert total capital invest-
ment to a uniform scries of total per-year equivalent
payments over a given time period using an assumed
real cost-of-capital at five percent. Wrc then add
O&M and other reoccurring costs to the annualized
capital cost to arrive at TAC. The discounted sum
of these annual expenditures is equal to the net
present value of total costs incurred over the time
period of this analysis.3
Results
Total annual compliance costs for Titles I
through V of the 1990 Amendments in the year 2000
will be approximately $19.4 billion; the estimate in-
creases to $26.8 billion in the year 2010. These costs
reflect "annualized" operation and maintenance
(O&M) expenditures (which includes research and
development (R&D) and other similarly recurring-
expenditures) plus amortized capital costs (i.e., de-
preciation plus interest costs associated with the ex-
2 For a few VOC source categories, we estimate that capi-
tal investment will not be necessary; for these sources, compli-
ance costs reflect O&M costs only.
J We recalculate the control cost estimates from regulatory
documents that use a seven or ten percent discount rate so that
the costs will be consistent with the five percent discount rate
assumption used in this analysis. We also calculate cost using
three percent and seven percent discount rales, as sensitivity tests;
ior detail see the discussion oi uncertainty later in this chapter,
in Chapter 8, and in Appendix B.
24
-------
Chapter 3: Direct Costs
isting capital stock) for the particular year.4 We
present cost estimates by title and emissions source
category (point sources, area sources, utilities,
nonroad engines and vehicles, and motor vehicles)
111 Table 3-1.
In some cases, assigning costs to a single CAAA
title is complicated by the fact that there are rules
issued pursuant to more than one title.5 In addi-
tion, with the passage of the 1990 Amendments, the
States were given greater discretion in developing
CAAA compliance strategies. For example, the
States can determine how best to meet progress re-
quirements and are responsible for creating permit
programs (under Title V). As a result, a significant
portion of the costs also represent State-level strate-
gies and decisions for reducing emissions.
Title I, Provisions for Attainment and Mainte-
nance of National Ambient Air Quality Standards
(NAAQS), represents pollution controls (of VOC,
NO , and PM emissions) implemented primarily by
point and area sources. Title I provisions also ac-
count for State programs designed to meet progress
requirements. By 2010, we project the costs of Title
1 provisions will account for over half of total CAAA
direct compliance costs ($14.5 billion). An additional
34 percent of estimated total costs ($9 billion) is at-
tributed to regulating mobile source emissions un-
der Title II. Collectively, the combined direct com-
pliance costs of these two titles is $16 billion in 2000
and $23 billion by 2010.
The remaining three titles account for less than
20 percent of total CAAA direct costs. We estimate
that Title III provisions, which target hazardous air
pollutant (HAP) emissions, will cost $840 million
by the year 2010. This estimate represents total an-
nualized capital costs (TACs) for individual two- and
four-year MACT standards. While the majority of
this estimated cost reflects reducing VOC emissions
4 Capita] expenditures are investments, generating a stream
of benefits and opportunity costs over an investment's lifetime.
In a cost-benefit analysis, the appropriate accounting technique
is to annualize capital expenditures. This technique involves
spreading the costs of capital equipment uniformly over the use-
ful life of lie equipment, by using a discount rate to account, for
the time value of money. In this analysis, all capital expendi-
tures were annualized using a real five percent interest rate.
5 In those cases, we generally assigned costs to a single title
based upon implementation dates and the year by which emis-
sion reductions are expected.
(since I LAP emissions were not included as part of
the Section 812 base- year inventory), Title III costs
do include some costs of final MACT rules that regu-
late non-VOC HAP emissions.
In order to estimate the costs associated with
Title IV, we considered the implications of pollu-
tion abatement controls (for SO2 and NOJ on the
electric power industry's operation of generation
units and how, over time, this would affect the de-
mand for electricity. The annual compliance esti-
mate for Title IV costs is $2.3 billion in 2000. This
estimate decreases to $2.0 billion by 2010. This de-
crease reflects, in part, the future compliance cost
savings resulting from the SO, allowance trading
program.
Title V costs arc associated with new operating
permit programs. The estimate accounts for approxi-
mately one percent of total costs projected under
the Post-CAAA 2010 scenario. States are expected
to implement Title V permit programs by 2005. The
estimate reflects the costs of State-developed pro-
grams during die first five-year implementation pe-
riod. These costs include incremental administra-
tive costs incurred by the permitted sources, State
and local permitting agencies, and EPA. The esti-
mate excludes federally-implemented State programs
and state programs which were already established
in the baseline.
Our presentation of cost estimates for the strato-
spheric ozone protection provisions of Title VI is,
by necessity, different from other tides. Ideally, one
should compare the costs of actions taken in a given
year to the benefits attributable to these actions. For
Title VI, a cost-benefit comparison of any given year
requires assumptions that result in potentially mis-
leading figures. The difficulty is due to the differing
time horizons and the complexity of the process by
which ozone-depleting substances (ODSs) cause ad-
verse effects on human health and the environment.
Title VI provisions incur costs over significantly
varying time horizons; for example, the cost analy-
sis of Sections 604 and 606 provisions spans 85 years
(from 1990 to 2075). At the same time, the analysis
of Section 611 extends from 1994 to 2015. In re-
sponse to this analytic difficulty, we base our com-
parison of Title VI costs to Title VI benefits on net
present values.
25
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 3-1
Summary of Direct Costs for Titles I to V of CAAA, By Title and Selected Provisions
(Annual Costs in million 1990$)
Title/Provision
Primary Cost
Estimate Percentage of
2000 Total Costs
Primary Cost
Estimate Percentage of
2010 Total Costs
Title 1- Provisions for Attainment and Maintenance ofNAAQS
Stationary NOX Controls, Utility Industry
Progress Requirements
PM NAAQS Controls
California LEV
National LEV
High Enhanced I/M
Other Title 1 Programs
Title 1: Total Costs
$ 790
1,200
1,900
320
180
1,100
3,100
$ 8,600
4%
6%
10%
2%
1%
6%
16%
44%
$ 2,500
2,500
2,200
1,100
1,100
1,400
3,700
$14,500
9%
9%
8%
4%
4%
5%
14%
54%
Title II- Provisions Relating to Mobile Sources
California Reformulated Gasoline
NOX Tailpipe/Extended Useful
Life Standard
Other Title II Programs
Title II: Total Costs
Title III- Hazardous Air Pollutants
Title III: Total Costs
Title IV- Acid Deposition Control
Title IV: Total Costs
Title V- Permits
Title V: Total Costs
Total Annual Cost
$2,000
1,500
3,900
$ 7,400
$780
$2,300
$300
$19,400
10%
8%
20%
38%
4%
12%
2%
100%
Note: Totals may not sum due to rounding. Only major provisions are listed under each
listed here are nonetheless included in the totals by title and the overall total.
$2,400
1,700
4,900
$ 9,050
$840
$2,040
$300
$26,800
title - other, less costly
9%
6%
18%
34%
3%
8%
1%
100%
provisions not
The net present value of Title VI program costs
reflect selected actions and their associated costs from
Sections 604, 606, 608, 609, and 611. Examples of
these actions include: replacement of ozone-deplet-
ing chemicals with alternative technologies and ma-
terials; recycling and storage of unused chlorofluo-
rocarbons; labeling; training; and administration.
Using a discount rate of five percent and a 85-year
time horizon (from 1990 to 2075), we estimate the
net present value of Title VI costs is $27 billion. For
illustrative purposes, we calculated an annualized
estimate of Title VI costs. It is, however, important
to recognize that these estimates may overestimate
actual compliance costs in those years, especially in
the year 2000, because of the phased nature of imple-
mentation— see Appendix G for more details. Our
annualized estimate of total Title VI costs is Si.4
billion. This value reflects an annualized equivalent
value of costs incurred over 85 years (from 1990 to
2075) using a five percent discount rate.
Provisions
Our analysis indicates eight provisions will ac-
count for approximately 54 percent of the total di-
rect compliance costs estimate for 2010. Six are Title
I provisions that affect stationary sources and vehicle
26
-------
Chapter 3: Direct Costs
emissions. The remaining two provisions target
mobile sources under Title II. These provisions are:
• PM NAAQS controls6,
• Electric power industry compliance (station-
ary NOx control),
• Progress Requirements,
• California Low Emission Vehicle
(LEV)program,
• National Low Emission Vehicle (LEV) pro-
gram,
• High Enhanced Inspection and Maintenance
(I/A/1) program,
• California Reformulated Gasoline, and
NOx Tailpipe/Extended Useful Life Stan-
dard".
The 1990 CAAA regulates stationary source
emissions primarily under Title I. Among the rel-
evant provisions, PM NAAQS, utility industry com-
pliance with NOx standards, and progress require-
ments are the main sources of Title 1 costs. From
2000 to 2010, we estimate the control costs of all
three provisions will increase by at least a factor of
two. Under the Post-CAAA scenario developed for
the emissions analysis, the utility industry's compli-
ance with NO emission standards affects all electric
X
generation units using fossil fuels. Existing oil and
gas units face Reasonable Available Control Tech-
nology (RACT) requirements and all new units must
comply with more stringent New Source Perfor-
mance Standards (NSPS) and New Source Review
(NSR) requirements. By 2010, estimated costs for
stationary NOx controls more than triple ($790 mil-
lion to f2,500 million). The cost estimate indicates
that the provision will be the single largest source of
CAAA direct costs. The second largest component
of total costs in 2010 is attributed to progress re-
quirements. Annual compliance costs with progress
requirements double from 2000 to 2010 ($1.2 bil-
lion and $2.5 billion, respectively). Among the three
provisions, the annual costs associated with PM
NAAQS compliance exhibits the least amount of
growth. We estimate annual costs for PM NAAQS
compliance will grow from SI.9 billion in 2000 to
S2.2 billion in 2010.
5 We estimate the PM NAAQS provision costs based on
compliance with standards that were ill effect prior to 1997 revi-
sions (62 Fed. Reg. 38,652, 1997).
Among the provisions regulating vehicle emis-
sions, only the national and California LEV pro-
grams exhibit a trend of increasing direct costs of
the same magnitude as seen with the costs of regu-
lating stationary sources under Title I. The com-
bined cost of national and California LEV programs
is $2.2 billion in 2010. For the California LEV pro-
gram, the increase in cost is largely a function of
higher per vehicle cost estimates (e.g., zero emission
vehicles (ZEV) are mandated in the year 2003). Our
cost analysis of the national LEV program assumes
that only the Northeast Ozone Transport Region
(OTR) states will incur costs in the year 2000. By
2010, however, we expect that the program will af-
fect areas outside of the OTR. As a result, 2010
national LEV costs increase with the expected ex-
pansion and increased volume of vehicle sales. Un-
like many of the other provisions, high enhanced I/
M costs do not exhibit significant growth from 2000
to 2010. We estimate this provision accounts for
approximately six percent of total costs in 2000 and
five percent in 2010. These costs, however, are un-
certain pending State decisions regarding the design
of their programs.
Among the analyzed Title II provisions, we at-
tribute nearly 15 percent of total annual direct costs
to the California reformulated gasoline (REG) pro-
gram and NOx Tailpipe/Extended Useful Life Stan-
dard. Although the reformulated gasoline program
affects only California, the state accounts for nearly
ten percent of annual gasoline sales in the United
States. We estimate compliance costs of SI.9 billion
in the year 2000. As the program enters Phase 2,
estimated costs grow to $2.4 billion. The trend in
cost associated with NOx Tailpipe/Extended Useful
Life Standard is very different. While costs increase
slightly between the years 2000 and 2010, the
provision's share of total cost slightly decreases.
of Other
In an ideal setting, a cost-benefit analysis would
not only identify, but also quantify and monetize,
an exhaustive list of social costs associated with a
regulatory action. This would include assessing how
regulatory actions targeting a specific industry or set
of facilities can alter the level of production and con-
sumption in die directly affected market and related
27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
markets. For example, regulation of emissions from
the electric utility industry that results in higher elec-
tricity rates would have both supply-side and de-
mand-side responses. In secondary markets, the in-
creased electricity rates affect production costs for
various industries and initiate behavioral changes
(e.g., using alternative fuels as a substitute for elec-
tric power). With each affected market, there are
also associated externalities that should be included
in estimating social costs. Returning to the utilities
example, the externalities associated with electric
power generation versus nuclear power generation
can be very different. The mix of externalities could
change as consumers substitute nuclear power for
electric power. It is frequently difficult to accurately
characterize one or all of these dimensions of mar-
ket responses and estimate the resulting social costs.
There are three generally practiced approaches
to calculating costs associated with regulation: (i) di-
rect compliance cost, (ii) partial equilibrium model-
ing, and (ni) general equilibrium modeling. Direct
compliance cost estimates are calculated differently
than the economic welfare impact estimates that re-
sult from partial or general equilibrium modeling; a
direct cost estimate is often the most straightforward
of the three approaches. This method estimates com-
pliance expenditures or, in economic terms, how an
industry's or firm's marginal cost curve shifts due to
increased production costs associated with regula-
tory compliance. As a result, this method does not
account for firm responses and market responses,
such as adjustment of production levels and product
prices. The other two methods measure changes in
producer and consumer welfare, and incorporate
these types of adjustments.
The direct cost approach likely overstates actual
compliance expenditures, but may have an ambigu-
ous relationship to total social costs. There are two
primary reasons for the overstatement of compli-
ance expenditures. First, the direct cost approach
does not account for market responses. As a result,
total direct cost estimates reflect the incremental cost
per unit of output multiplied by the generally higher,
prc-regulation quantity produced. Second, a direct
cost approach tends to make the simplifying assump-
tion that firms rely 011 static pollution abatement
technology, when in fact the presence of compli-
ance costs provides an incentive to innovate. Sev-
eral ex post cost analyses suggest that the marginal
cost curve may not necessarily shift by the full
amount of the pollution abatement. For example,
firms may respond by altering production processes
to more efficiently reduce emissions.' Social cost
estimates, however, may include other costs not re-
flected in direct cost estimates (discussed below),
thereby offsetting the tendency for direct cost esti-
mates to overstate expenditures.
Measuring net welfare changes due to regulatory-
action requires either partial or general equilibrium
modeling. These more complicated approaches es-
timate social costs by accounting for a wider range
of market consequences associated with compliance
with pollution abatement requirements. The par-
tial equilibrium approach is particularly appropri-
ate when social costs are predominantly incurred in
directly affected markets. It requires modeling both
supply and demand functions in the affected eco-
nomic sector. Therefore, measures of social cost
reflect behavioral responses by both producers and
consumers in a specific market and do not necessar-
ily reflect how those changes affect related markets.
In cases where the regulatory action is known
to have an impact on many sectors of the US
economy, the general equilibrium model is a more
appropriate approach to estimating social costs. Like
the partial equilibrium model, the general equilib-
rium model estimates social costs by accounting for
direct compliance costs and producer and consumer
market behavior. The general equilibrium model
can capture first-order effects that occur in multiple
sectors of the economy, and may also provide in-
sight into unanticipated indirect effects in sectors that
might not have been included in the scope of a par-
tial equilibrium analysis.
The relationship of general equilibrium estimates
to estimates from the other two cost approaches is
not always clear. General equilibrium estimates have
a broader basis from which to estimate social costs
and can reflect the net welfare changes across the
full range of economic sectors in the U.S. Partial
equilibrium modeling tends to understate full social
costs because of its restricted scope (i.e., generally
limited to one industry). Total direct cost estimates
are likely to overstate costs in the primary market
because they do not reflect consumer and producer
responses. This is demonstrated in comparisons of
'Morgcnstcm et a!. (1998) estimate the ratio of incurred
abatement expenditures to estimated direct costs can be as low
as 0.8.
28
-------
Chapter 3: Direct Costs
estimates generated using a direct cost approach and
a partial equilibrium approach. The extent to which
a direct cost estimate will overstate or understate a
social cost estimate from a general equilibrium model
depends on the magnitude of the "ripple effects" in
economic sectors not targeted by a regulation.8
In the 812 retrospective analysis (RPA, 1997),
we recognized that the Clean Air Act has a perva-
sive impact on the US economy and opted for the
general equilibrium approach. The retrospective
nature of the analysis, however, provided us with
fairly well-developed historical data sets of goods and
service flows throughout the economy. These data
sets facilitated the development of detailed, year-by-
year expenditures m all sectors of the economy, from
which we modeled producer and consumer behav-
ior and estimated net social costs. In the retrospec-
tive, our central estimate of total annualized direct
costs, from 1970 to 1990, was $523 billion. In com-
parison, we estimated the aggregate welfare effects
to be between $493 and $621 billion.9
For the prospective analysis, however, we adopt
a direct compliance cost approach. Although the
general equilibrium approach may represent a more
theoretically preferable method for measuring so-
cial costs, we use the simpler direct cost modeling
method for three reasons:
• First, we believe that the direct cost approach
provides a good first approximation of the
CAAA's economic impacts on various sec-
Current regulatory analyses that apply partial equilib-
rium modeling or general equilibrium modeling tend to mea-
sure costs with the assumption that markets are currently oper-
ating under optimally efficient conditions. Emerging literature
suggests that a full accounting of the social costs and efficiency
impacts of environmental regulations could also include an as-
sessment of the incremental costs that reflect existing market
distortions, such as those imposed by the current tax code. The
distortions introduced by existing taxes, in combination with
new regulatory requirements, are collectively referred to as the
tax-interaction eiiect. One oi the major conclusions oi this
emerging literature is that the social cost oi environmental policy
changes can be substantially higher when pre-existing tax distor-
tions are taken into account. Our direct cost estimates do not
reflect quantification of this effect, in part: because of the emerg-
ing nature of this literature and in part because existing esti-
mates of the magnitude of the tax-interaction effect are calcu-
lated as increments to social costs and are not necessarily appli-
cable adjustments to direct cost estimates.
9 Estimates are in 1990 dollars. The retrospective states,
"Tn general the estimated second order macroeconomic effects
were small relative to the size of the U.S. economy." The rate
of long term GNP growth between the control and no-control
scenarios amounted to roughly one-twentieth of one percent
less growth.
tors the U.S. economy. Comparison of the
direct cost approach to the partial equilib-
rium modeling suggests that the direct cost
approach likely overstates costs to the en-
tity that incurs the pollution control cost ex-
penditure. As discussed earlier, the direct
cost approach does not reflect adjustments
to prices and quantities that might mitigate
the effects of regulation. Recent research
analysing ex ante and ex post cost estimates
of regulations suggests that ex ante analyses
are far more likely to overstate than under-
state costs.10 However, direct cost estimates
may also understate the effects of long-term
changes in productivity and the ripple effects
of regulation on other economic sectors that
are captured by a general equilibrium ap-
proach. The magnitude of those other ef-
fects, including potential magnification of
social costs by existing tax distortions, may
be substantial.
Second, we believe that the closer approxi-
mation of social costs that might be gained
through a general equilibrium approach
could be compromised by the difficulty and
uncertainty associated with projecting future
economic and technological changes. The
general equilibrium approach could provide
many insights that the direct cost approach
cannot, but also introduces a significant level
of additional uncertainty.
Third, the focus of the present analysis is a
comparison of direct costs and direct ben-
efits. To provide a balanced treatment of
costs and benefits in a general equilibrium
framework, the social cost model must be
designed and configured to reflect the indi-
rect economic consequences of both costly
and beneficial economic effects. None of
the general equilibrium models available in
the time frame of this study could be config-
ured to support effective analysis of the full
range of specific direct costs and, especially,
direct benefits of the 1990 Clean Air Act
Amendments.
"° See, for example, Harrington et al (1999), referenced in
Appendix B, for a comparative analysis of ex ante and ex post
regulatory cost estimates.
29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
* Fourth, undertaking a general equilibrium
modeling exercise remains a very resource-
intensive task. For the purposes of compar-
ing costs to benefits we concluded that more
detailed modeling would not be the most
cost-effective use of the project resources.
Uncertainty in
Estimates
Overview
As we note at the beginning of this chapter, ex-
plicit and implicit assumptions regarding changes in
consumption patterns, input costs, and technologi-
cal innovation are crucial to framing the question of
the CAAA's cost impact. Given the nature of this
prospective study, there is no way to verify the ac-
curacy of the assumptions applied to future scenarios.
We can envision other plausible analyses with esti-
mates that differ from results in this chapter. More-
over, for many of the factors contributing to uncer-
tainty, the degree or even direction of the bias is
unknown or cannot be determined. Nevertheless,
uncertainties and/or sensitivities can be identified
and in many cases the potential measurement errors
can be quantitatively characterised. In this section
of the chapter, w7e first discuss several quantitative
sensitivity analyses undertaken to characterize the
impact of key assumptions on the ultimate cost analy-
sis. We conclude the chapter with a qualitative dis-
cussion of the impact of both quantified and
unquantified sources of uncertainty.
In order to characterize the uncertainty in the
cost estimates, we conducted sensitivity analyses on
the key parameters and analytic assumptions of six
major provisions. The provisions are the following:
* Progress Requirements,
* California Reformulated Gasoline,
• PM NAAQS Controls,
* LEV program (the National and California
programs combined),
* Non-utility Stationary Source NOx Con-
trols, and
* ^TOX Tailpipe/Extended Useful Life Stan-
dard.
We selected these provisions because they are
among the most significant sources of CAAA costs,
yet cost estimates for each of the provisions incor-
porate significant uncertainties. Collectively, these
provisions account for nearly 50 percent of total di-
rect compliance cost estimates for 2010. Table 3-2
summarizes the methods we used to conduct the cost
sensitivity analyses and the results.
For each test, we developed three estimates for
one or more components of costs affecting the total
cost estimate for a given provision: (1) a central esti-
mate, equal to the 2010 primary cost estimate re-
ported in this chapter11, (2) a low estimate; and (3) a
high estimate. The low7 and high estimates assess
the potential magnitude of the effect of the
component(s) on the provision's costs and conse-
quently, total CAAA costs, using reasonable alter-
native assumptions for each cost component. For
progress requirements, PM NAAQS controls, and
stationary source NOx controls, the cost projections
are based on models of future emissions controls.
Accurately identifying the set of adopted controls is
a key source of uncertainty. For example, cost-ef-
fective control measures for complying with progress
requirements have not yet been identified and the
sensitivity test suggests the potential for substantial
variability in progress requirement compliance costs.
In the case of motor vehicle provisions, there are
two significant sources of uncertainty, projecting
future car sales and forecasting accurate per vehicle
costs.
The results indicate that the sensitivity of our
primary cost estimates (central estimates) is not uni-
form across provisions. In addition, low and high
estimates may vary by as much as a factor of two.
These sensitivity analyses demonstrate the potential
effect of altering selected assumptions and data. We
do not assign probabilities to the likelihood of the
alternative. In other words, it would be inappropri-
ate simply to add up the array of low7 and high esti-
mates to arrive at an overall range of uncertainty
around the central estimates, because it is unlikely
that a plausible scenario could be constructed where
all the estimates are concurrently either at the high
11 The one exception is the central estimate of progress
requirements. Our sensitivity analysis which is based on more
recent cosl iniormadon indicates thai our primary estimate is
more reflective oi a high estimate. See Appendix B for more
details.
30
-------
Chapter 3: Direct Costs
or low end of their individual plausible ranges. A
better interpretation of these results is that uncer-
tainty in key input parameters can have a significant
effect on the overall uncertainty of our estimates of
direct compliance costs and ultimately the net ben-
efits calculation.
In addition to examining specific provisions, we
conducted a sensitivity analysis of the cost of capital
used throughout the analysis. Cost estimates pre-
sented earlier in this chapter reflect application of a
cost of capital (for the purposes of aimualizmg total
capital costs) of five percent. We also examined the
effect on cost estimates for those provisions which
involve significant capital expenditures and where
we could recalculate annualized costs from the avail-
able information. These provisions include non-util-
ity and area source estimates for VOC, NOx, and
PM control. The alternative estimates use three and
seven percent for the cost of capital. Results indi-
cate that cost estimates are only moderately sensi-
tive to the discount rate. The provisions evaluated
have a total annualized capital cost of approximately
S3 billion in 2010. Varying the cost of capital gener-
ated alternative estimates of S2.8 billion (three per-
cent) and $3.1 billion (seven percent).12
Of
to Uncertainty
There are a wide range of other factors which
contribute to uncertainty in the overall cost esti-
mates. In most cases, the effect of these other fac-
tors cannot be quantified, though some may have
significant influences on our overall net benefits es-
timate. We present a summary of these factors in
Table 3-3 below, and provide a characterization of
the potential effect of each uncertainty on the pri-
mary estimate of the net benefits (i.e., if costs are
overestimated, net benefits are underestimated). The
two most important factors are the potential impact
of innovation on the ultimate control costs incurred
and the conservative assumptions we made to esti-
mate RFP costs.
•2 Note that these calculations reflect the use of alternative
discount rates to estimate annual costs. The use of alternative
rates to calculate the total net present value of costs incurred
through the full 1990 lo 2010 study period is examined sepa-
rately in Chapter 8, where we compare total costs lo tola! ben-
efits.
The regulatory documents which provide cost
inputs to ERCAM and the IPM contain the most
recent data available, given existing technological
development. Between 2000 and 2010, however,
advancements in control technologies will allow
sources to comply with CAAA requirements at
lower costs. For example, we anticipate technologi-
cal improvements for complying with the multiple
tiers of proposed emission standards during the
phase-in of nonroad engine controls will likely lead
to reduced costs. In addition, the costs for certain
control equipment may decrease over time as demand
increases and technology innovation and competi-
tion exert downward pressure on equipment prices.
For instance, selective catalytic reduction (SCR )
costs have decreased over the past three years as more
facilities begin to apply the technology. We also
believe that even in the absence of new emission stan-
dards, manufacturers will eventually upgrade engines
to improve performance or to control emissions
more cost-effectively; firms will institute technolo-
gies such as turbocharging, aftercooling, and vari-
able-valve timing, all of which improve engine per-
formance.
There is considerable uncertainty surrounding
the development of States' control plans for meet-
ing ozone NAAQS attainment requirements. We
base the RFP cost estimate on the assumption that
ozone nonattaiiimeiit areas (NAAs) will take credit
for NO reductions for meeting progress require-
ments. Additional area-specific analysis would be
necessary to determine the extent to which areas find
NOx reductions beneficial in meeting attainment and
progress requirement targets. Trading of NOx for
VOC'- to meet RFP requirements may result in dis-
tributions of VOC'- and NO emission reductions
X
which differ from those used in this analysis. In
response to these uncertainties, we adopted a con-
servative strategy for estimating the costs of RFP
reductions in the primary analysis. Wre use a rela-
tively high cost per ton reduced estimate of $10,000
for all required reductions. Since the time we con-
ducted our primary cost analysis more information
has emerged suggesting controls could cost much less,
perhaps as little as $3,500 (see Table 3-2 and Appen-
dix B for more details). In our sensitivity analysis of
this variable, we incorporate the more recent cost
per ton estimates. The analysis suggests that the
$10,000 per ton reduced may in fact be more repre-
31
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 3-2
Results of Quantitative Sensitivity Tests
Provision
Primary Cost
Estimate in 20101
(billions 1990 $)
Strategy for Sensitivity Analysis
Range of Estimates
from Sensitivity Test
(billions 1990 $)
Progress
Requirements
California
Reformulated
Gasoline
PM NAAQS Controls
LEV costs (California
and National
Combined)
Non-Utility Stationary
Source NOX Costs
$2.46
$2.45
$2.22
$2.16
$2.15
Vary unit costs for unidentified
measures
Vary incremental fuel costs and
gasoline sales estimates
Vary model attainment plan
assumptions and cost per ton
estimates
Vary per vehicle costs and
projections of vehicle sales
Vary unit-level cost per ton
$1.07-
( centra I,
$1.4-
$0.09 to
$1.08-
$1.1 -
$2.46
$1.15)
$3.5
$3.35
$2.48
$3.2
NOX Tailpipe/Useful
Life Standards
$1.65
Vary per vehicle costs and vehicle
sales data
$0.83-$2.48
Note:
1 In all cases, except progress requirements, the Post-CAAA 2010 primary cost estimates is equal to the central
estimate in the sensitivity analysis. For more details on the sensitivity analysis of progress requirements and other
provisions, see Appendix B.
scntativc of an upper bound cost estimate, rather than
a central estimate as our primary cost analysis re-
flects. The result of our conservative approach indi-
cates that we may overstate RFP costs by a factor of
two in 2010.
One other factor is also worth noting, although
its impact is likely to be less important than the pre-
vious two factors. Under the 1990 CAAA, EPA
created economic incentive provisions in several rules
to provide flexibility for affected facilities that com-
ply with the rules. These provisions include bank-
ing, trading, and emissions-averaging provisions.
Flexible compliance provisions tend to lower the cost
of compliance. For example, the emissions-averag-
ing program grants flexibility to facilities affected
by the marine vessels rule, the petroleum refinery
National Emission Standard for Hazardous Air Pol-
lutants (NESHAP), and the gasoline distribution
NRSHAP. These facilities can choose which sources
to control, as long as they achieve the required over-
all emissions reduction. In many of the cost analy-
ses, EPA does not attempt to quantify the effect that
economic incentive provisions will have on the over-
all costs of a particular rule. In these cases, to the
extent that affected sources use economic incentive
provisions to minimize compliance costs, costs may
be overstated. The major trading programs autho-
rized under the Amendments, however, governing
sulfur and nitrogen oxide emissions reductions from
utilities and major non-utility point sources, are re-
flected 111 the cost estimates presented here.
32
-------
Chapter 3: Direct Costs
Table 3-3
Key Uncertainties Associated with Cost Estimation
Potential Source of Error
Direction of
Potential Bias
for Net Benefits
Likely Significance Relative to Key Uncertainties
on Net Benefits Estimate
Costs are based on today's
technologies. Innovations
in future emission control
technology and
competition among
equipment suppliers tend
to reduce costs overtime.
Underestimate Probably minor. Available evidence suggests that estimates
of pollution control costs based on current engineering can
substantially overestimate the ultimate cost incurred,
resulting in understating net benefits.2
Uncertainty of final State
strategies for meeting
Reasonable Further
Progress (RFP)
requirements.
Underestimate Probably minor. We apply a conservative estimate for costs
of RFP measures. Available evidence for identified RFP
measures suggests costs could be as much as 70 percent
lower than this value. The bias most likely results in
significantly understating net benefits.
Errors in emission
projections that form the
basis of selecting control
strategies and costs in
both the IPM and ERCAM
models.
Unable to
determine based
on current
information
Probably minor. In many cases, emissions reductions are
specified in the regulations, suggesting that errors in the
estimation of absolute levels of emissions under Pre- and
Post-CAAA scenarios may have only a small impact on cost
estimates. The effect on net benefits is unknown.
Exclusion of the impact of
economic incentive
provisions, including
banking, trading, and
emissions averaging
provisions.
Underestimate Probably minor. Economic incentive provisions can
substantially reduce costs, but the major economic programs
for trading of sulfur and nitrogen dioxide emissions are
reflected in the analysis.
Incomplete
characterization of certain
indirect costs, including
vehicle owner opportunity
costs associated with
Inspection and
Maintenance Programs
and performance
degradation issues
associated with the
incorporation of emission
control technology.
Overestimate Probably minor. Preliminary evidence suggests that the
opportunity costs of vehicle owners is most likely small
relative to other cost inputs.3 In addition, it is will vary from
State to State and is subject to a variety of influencing
factors. The potential magnitude of indirect costs associated
with performance degradation is more uncertain, because
few data currently exist to quantify this effect.
33
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 3-3 (continued)
Key Uncertainties Associated with Cost Estimation
Potential Source of Error
Choice to model direct
costs rather than social
costs
Use of costs for rules that
are currently in draft form
(i.e., not yet finalized).
Exclusion of costs of 7-
year and 10-year MACT
standards and the
residential risk standards
for the 2- and 4-year
MACT standards.
Direction of
Potential Bias
for Net Benefits
Unable to
determine based
on current
information
Unable to
determine based
on current
information
Unable to
determine based
on current
information
Likely Significance Relative to Key Uncertainties
on Net Benefits Estimate1
Probably minor. The relationship of social cost to direct cost
estimates is influenced by multiple factors that operate in
opposite directions, suggesting the magnitude of the net
effect is reduced. Social cost estimates can reflect the net
welfare changes across the full range of economic sectors in
the U.S, and so may yield higher estimates of costs than a
direct cost approach. In addition, social cost estimates can
be constructed to reflect the potentially substantial cost-
magnifying effect of existing tax distortions. Direct cost
estimates, however, are likely to overstate costs in the
primary market because they do not reflect consumer and
producer responses. The extent to which a direct cost
estimate will overstate or understate a social cost estimate
depends on the magnitude of the "ripple effects" in economic
sectors not targeted by a regulation. In addition, assessment
of the effect on net benefit estimates must also account for
any economy-wide effects of direct benefits (e.g., the broader
implications of improving health status, and improving
environmental quality).
Probably minor. Rules that are most important to the overall
cost estimate are largely finalized. For example, there is
some uncertainty as to how the cap-and-trade program
through the SIP process will lower NOx emissions in an
efficient manner. The expected effect on net benefits is
minimal.
Probably minor. Costs for the 7- and 1 0-year MACT
standards are likely to be less than for the 2- and 4-year
standards included in the analysis and the need for, and
potential scope and stringency of, future Title III residual risk
standards remain highly uncertain. For consistency, benefits
of the 7- and 10-year standards and the residual risk
standards are also excluded.
Note:
1 The classification of each potential source of error reflects the best judgement of the section 812 Project Team. The
Project Team assigns a classification of "potentially major" if a plausible alternative assumption or approach could influence
the overall monetary benefit estimate by approximately five percent or more; if an alternative assumption or approach is
likely to change the total benefit estimate by less than five percent, the Project Team assigns a classification of "probably
minor."
2 For more detail, see Harrington et al (1999), referenced in Appendix B.
3 Preliminary evidence based on Arizona's Enhanced I/M program indicates that major components of the programs costs
are associated with test and repair costs rather than the costs of waiting and travel for vehicle owners. (Harrington and
McDonnell, 1999.) To date, Enhanced I/M programs have been implemented in only four States.
34
-------
Air Quality
Modeling
Air quality modeling links changes in emissions
to changes in the atmospheric concentrations of pol-
lutants that may affect human health and the envi-
ronment. A crucial analytical step, air quality mod-
eling is one of the more complex and resource-in-
tensive components of the prospective analysis. This
chapter outlines how we estimated future-year pol-
lutant concentrations under both the Pre- and Post-
CAAA scenarios using air quality modeling results
and ambient monitor data. The first section of the
chapter begins with a discussion of some of the chal-
lenges faced by air quality modelers and a brief de-
scription of the models we used in this analysis. The
following section provides an overview of the gen-
eral methodology we used to estimate future-year
ambient concentrations. This methodology section
includes a description of how we used modeling re-
sults to adjust monitor concentration data and esti-
mate ambient concentrations for the years 2000 and
2010. The third section of this chapter summarizes
the results of the air quality modeling and presents
the expected effects of the CAAA on future-year
pollutant concentrations. A discussion of the key
uncertainties associated with air quality modeling
concludes the chapter.
Overview of Air
Models
Air quality modelers face two key challenges in
attempting to translate emission inventories into pol-
lutant concentrations. First, they must model the
dispersion and transport of pollutants through the
atmosphere. Second, they must model pertinent at-
mospheric chemistry and other pollutant transfor-
mation processes. These challenges are particularly
acute for those pollutants that are not emitted di-
rectly, but instead form through secondary processes.
Ozone is the best example; it forms in the atmo-
sphere through a series of complex, non-linear chemi-
cal interactions of precursor pollutants, particularly
certain classes of volatile organic compounds (VOCs)
and nitrogen oxides (NOJ. We faced similar chal-
lenges when estimating P.M concentrations. Atmo-
spheric transformation of gaseous sulfur dioxide and
nitrogen oxides to participate sulfates and nitrates,
respectively, contributes significantly to ambient
concentrations of fine particulate matter. In addi-
tion to recognizing the complex atmospheric chem-
istry relevant for some pollutants, air quality mod-
elers also must deal with uncertainties associated with
variable meteorology and the spatial and temporal
distribution of emissions.
Air quality modelers and researchers have re-
sponded to the need for scientifically valid and reli-
able estimates of air quality changes by developing a
number of sophisticated atmospheric dispersion and
transformation models. Some of these models have
been employed in support of the development of
federal clean air programs, national assessment stud-
ies, State Implementation Plans (SIPs), and individual
air toxic source risk assessments. In this analysis,
we used several of these well-established models to
develop a picture of future changes in air quality re-
sulting from the implementation of the 1990 CAAA.
We focused our air quality" modeling efforts on
estimating the impact of Pre- and Post-CAAA emis-
sions on future-year ambient concentrations of
ozone, PM10, PM2S, SO,, NOx, and CO and on fu-
ture-year acid deposition and visibility. The ideal
model for this analysis would be a single integrated
air quality model capable of estimating ambient con-
centrations for all criteria pollutants throughout the
U.S. Although EPA is working to develop such a
model, at die time of this analysis the model was not
sufficiently developed and tested. In the absence of
a single integrated model, we employed the Urban
Airshed Model (UAM) in our analysis of ozone and
used both the Regional Acid Deposition Model/Re-
gional Particulate Model (RADM/RPM) and the
Regulatory Modeling System for Aerosols and Acid
35
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Deposition (REMSAD) model to assess PM10, PM25,
acid deposition and visibility- All three of these mod-
els are three-dimensional grid models which require
emissions and meteorological data as input. Each of
these models calculate pollutant concentrations by
simulating the physical and chemical pollution for-
mation processes that occur in the atmosphere.
We conducted separate HAM, RADM/RPM,
and REMSAD model runs for the 1990 base-year
and each future-year projection scenario. The pri-
mary model input used for each run consisted of
emissions estimates corresponding to the year and
scenario being modeled (as described in Chapter 2
and Appendix A) and historical meteorological data
corresponding to a past time period, referred to as a
simulation period. We selected previous ozone epi-
sodes, i.e., multi-day events characterized by weather
conditions conducive to ozone formation and trans-
port (and as a result, characterized by multi-day spans
with higher than average ozone concentrations), to
serve as the simulation periods for (JAM model runs.
Although ozone concentrations during these simu-
lation periods exceed the seasonal average, because
die simulation periods for both the eastern and west-
ern U.S. cover roughly a two week span, ozone con-
centration peaks are largely offset by the surround-
ing lows. Overall, the selected simulation periods
reasonably represent summertime ozone forming
meteorological conditions and ozone concentrations.
RADM/RPM simulation periods used to model PM,
acid deposition, and visibility were chosen using a
random selection process, while separate simulation
periods at the beginning of each of die four seasons
were chosen for REMSAD.
Table 4-1 provides an overview of the air qual-
ity models used in this analysis. We modeled con-
centrations of all pollutants across the 48 contigu-
ous states; however due to the lack of an integrated
model, separate air quality models were used to esti-
mate ozone and PM for the eastern and western U.S.
Table 4-1 shows the domain for each model and the
simulation periods selected for use with each model
and provides an overview of the spatial resolution
of the models used as part of this analysis. The finer
the resolution (i.e., the smaller the grid cells) the
better the model can capture the effects of localized
changes in emissions and weather conditions 011
ambient air quality. Recognizing the relationship
between grid cell resolution and the certainty of re-
sults, we endeavored to estimate pollutant concen-
trations in more populated areas using higher reso-
lution models. For this reason, we used the fine grid
UAM-IV, an urban-scale model, to estimate ambi-
ent ozone levels in selected western cities. Similarly,
w7e used an intermediate resolution grid (12 km x 12
km) to model ozone in "inner OTAG" states where
population density is high and ozone transport is a
major problem.1
Using the three-dimensional grid cell models,
CAM, RADM/RPM, and REMSAD, we estimated
grid-cell specific, hourly ozone and daily PM1(), and
PM25 concentrations for each day of the relevant
simulation periods. We conducted separate model
runs for the 1990 base-year and 2000 and 2010 fu-
ture-year Pre- and Post-CAAA scenarios. Using
these results, wfe ultimately projected the impact of
the CAAA on ozone and PM ambient levels.
We relied on the same models used to predict
PM concentrations to estimate changes in future-year
acid deposition and visibility. For each model grid-
cell we predicted daily acid deposition levels and vis-
ibility. Estimates for each day of the simulation
period were generated for the base-year and both
projection years under the Pre- and Post-CAAA sce-
narios.
Wre estimated future-year Pre- and Post-CAAA
ambient SO2, NO, NO,, and CO concentrations by
adjusting 1990 concentrations using future-year to
base-year emissions ratios. This technique assumes
a linear relationship between changes in emissions
in an area and changes in that area's ambient con-
centration of the emitted pollutant.2 Although this
technique does not take into account pollutant trans-
port or atmospheric chemistry, we believe linear scal-
ing generates reasonable approximations of ambient
concentrations of gaseous pollutants such as SO2,
NO , and CO.
1 The Ozone Transport Assessment Group (OTAG) con-
sists of the 37 easternmost states and the District of Columbia.
The "inner OTAG" region is comprised of the more eastern
(and more populated) states within the OTAG domain.
' It is important to emphasize that the correlation expected
is between changes in emissions and changes in air quality. Di-
rect correlations between the absolute emissions estimates and
empirical air quality measurements used in the present analysis
may not be strong due to expected inconsistencies between the
geographically local, monitor proximate emissions densities ai-
iecting air quality data.
36
-------
Chapter 4: Air Quality Modeling
Table 4-1
Overview of Air Quality Models
Air Quality
Measure Region Model
Spatial Resolution
Simulation Period
Ozone Eastern UAM-V
U.S.
a) 12 km x 12 km grid for "Inner
OTAG Region"
b) 36 km x 36 km grid for remainder
of 37-state OTAG region
July 20-30, 1993 and July 7-
18, 1995
Ozone Western UAM-V
U.S.
56 km x 56 km grid (regional scale) July 1-10, 1990
covering the 11 westernmost states
(states west of North and South
Dakota, including western Texas)
Ozone San UAM-IV
Francisco
Bay Area
4 km x 4 km (urban scale) grid
covering the San Francisco Bay
Area, the Monterrey Bay Area,
Sacramento, and a portion of the
San Joaquin Valley
Aug. 3-6, 1990
Ozone Los UAM-IV
Angeles
Area
5 km x 5 km grid covering the
South Coast Air Basin from Los
Angeles to beyond Riverside and
including part of the Mojave Desert
June 23-25, 1987 and Aug.
26-28, 1987
Ozone Maricopa UAM-IV
County
(Phoenix)
Area
4 km x 4 km grid covering
urbanized portion of Maricopa
County
Aug. 9-10, 1992 and June 13-
14, 1993
Particulate Eastern RADM/RPM
Matter U.S.
80 km x 80 km grid (coarse
resolution) covering eastern North
America from the Rocky Mountains
eastward to Newfoundland,
Canada and the Florida Keys (see
Fig. C-14 in Appendix C)
30 randomly selected 5-day
periods spanning a four-year
period
Particulate Western REMSAD
Matter U.S.
56 km x 56 km grid covering the 11
westernmost states
ten-day period for each of four
seasons:
May 1-10,
July 1-10,
Oct. 1-10, and
Dec. 1-10
Visibility
Visibility
Acid
Deposition
Sulfur
Dioxide
Oxides of
Nitrogen
Carbon
Monoxide
Eastern
U.S.
Western
U.S.
Eastern
U.S.
U.S.
U.S.
U.S.
RADM/RPM
REMSAD
RADM
linear scaling
linear scaling
linear scaling
(same as PM)
(same as PM)
(same as RADM/RPM)
56 km x 56 km REMSAD grid
covering 48 contiguous states
56 km x 56 km REMSAD grid
covering 48 contiguous states
56 km x 56 km REMSAD grid
covering 48 contiguous states
(same as PM)
(same as PM)
(same as RADM/RPM)
not applicable
not applicable
not applicable
37
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The air quality modeling component of the 812
prospective analysis involved the application of a
variety of complex, sophisticated air quality model-
ing tools and techniques. Overall, however, the
method we used to estimate the impact of changes
in emissions on air quality wfas relatively straight
forward. We began by gathering 1990 air quality
monitor data for the six criteria pollutants analyzed
in this study. These observational data served as the
air quality baseline for both the Pre- and Post-CAAA
scenarios. We then estimated 2000 and 2010 con-
centrations of each pollutant under each emissions
scenario by applying adjustment factors to the 1990
monitor data. The adjustment factors for each fu-
ture-year projection scenario were based on the rela-
tive change in pollutant concentration between 1990
and the desired future-year, as predicted by air qual-
ity simulation modeling. This section presents an
overview of the methodology we used to estimate
future-year ambient concentrations. For a more
detailed description, please refer to Appendix C.
The diagram in Figure 4-1 illustrates the meth-
odology used to estimate ozone and PM concentra-
tions. First, we compiled distributions of observed
pollutant concentrations recorded at each air qual-
ity monitor in 1990. We obtained these data from
EPA's Aerometric Information Retrieval System
(AIRS), a publicly accessible database of air quality
information. Separately, we then developed distri-
butions of estimated concentrations for each pollut-
ant in 1990 using 1990 emissions data and the appro-
priate air quality model. Unlike the 1990 observed
concentrations that were measured at monitoring
sites, the 1990 estimated concentrations were calcu-
lated at the center of each cell of a grid covering the
domain of the applicable air quality model. Using
future-year emission inventory estimates for the Pre-
CAAA and Post-CAAA scenarios (developed as de-
scribed in Chapter 2 and Appendix A) and the ap-
propriate air quality models, we next developed dis-
tributions of model-estimated concentrations at each
grid cell for each of four future-year projection sce-
narios: 2000 Pre-CAAA, 2010 Pre-CAAA, 2000 Post-
CAAA, and 2010 Post-CAAA. These results were
used to derive adjustment factors for each air qual-
ity monitor, based on the simulation results for the
grid cell in which the monitor is located. The fu-
ture-year/scenario adjustment factor for each pol-
lutant represents the ratio of the simulated future-
year/scenario concentration to the 1990 model-esti-
mated concentration. These factors wfere calculated
by matching future-year and 1990 concentrations at
regular intervals in each distribution. Finally, four
sets of model-derived adjustment factors were applied
to the distribution of observed 1990 concentrations
at each monitor to forecast distributions of concen-
trations for each of the four future-year projection
scenarios. It is these concentrations that serve as
inputs into the CAAA benefits modeling.
An illustrative example follows. Assume the
median observed concentration of Pollutant A at
Monitor X in 1990 w7as 0.24 ppm. Air quality mod-
eling for the grid cell in which Monitor X is located
predicts a median Pollutant A concentration of 0.30
ppm in 1990 and 0.15 ppm in 2010 under the post-
CAAA scenario. The 2010 Post-CAAA adjustment
factor for the median Pollutant A concentration
would be 0.5, and the predicted 2010 Post-CAAA
median concentration at Monitor X would be 0.5
(=0.15/0.30) times the 1990 monitor value of 0.24
ppm, or 0.12 ppm.
Our approach for forecasting concentrations of
SO_, NOx, and CO involved the same basic approach
described above. However, instead of applying
model-derived adjustment factors to the 1990 ob-
served distribution of concentrations, we adjusted
the 1990 distribution using the ratio of future-year
emissions to 1990 emissions in the vicinity of the
monitor for each of the four future-year projection
scenarios. For more information about this ap-
proach, please refer to Appendix C.
38
-------
Chapter 4: Air Quality Modeling
Figure 4-1
Schematic diagram of the future-year concentration estimation methodology
1990
(Base Case)
AQ Model
Predictions
Cone
yx\\\\\ Ipost-CAAAl-K
\\ \ \ \ \x ^ '
x\ x \ \ XX
Ratio to Base Case
Cone
XX X X X \~\
XX X X X XX
X X X X X X X
XX X X X XX
XX \\X\\
Ratio to Base Case
Adjustments
Factors
cityl
AQ
Observations
AQ
Results
Concentration
distributions
Concentration
distributions
Concentration
distributions
NOTE: Figure illustrates how model results and observations are used to produce the air quality profiles (concentration distributions) for the
benefits analysis. The figure shows model runs at the top; four sets of "ratios" of model results in space in the middle; and frequency
distributions of pollutant monitor concentrations and the space-dependent scaling of these by the ratios of the model predictions on the bottom.
Air Quality
This section presents a summary representation
of the air quality modeling results. We discuss the
model-simulated concentration estimates and the
adjusted future-year concentration predictions with
a focus on the change in air quality resulting from
the implementation of the 1990 CAAA.
Ozone
We modeled ozone concentrations separately for
the eastern U.S., western U.S., San Francisco Bay
area, Los Angeles area, and Mancopa County (Phoe-
nix, AZ) area. Examination of base-year and future-
year model concentration estimates shows expected
increases in Pre-CAAA ozone concentrations and
expected decreases in Post-CAAA ozone concentra-
tions in the eastern U.S. In this part of the country,
UAM-V predicts Prc-CAAA ozone concentration
increases will occur primarily over the states of Vir-
ginia, North Carolina, Kentucky, Tennessee, Geor-
gia, and Alabama; while Post-CAAA decreases will
be more widespread. Comparison of Pre- and Post-
CAAA model estimates shows that, wfith the excep-
tion of a few isolated areas, ambient ozone levels
throughout the East will be reduced in the year 2010
as a result of the CAAA. These lowrer levels are
largely due to significant reductions in area source
and motor vehicle VOC emissions and utility, point
source, and motor vehicle NO emissions.
Regional-scale model results for the western U.S.
indicate that ozone concentrations in this portion
of the country, just as in the eastern U.S., will gen-
erally increase from the 1990 base-year under the
Pre-CAAA scenario and decrease from 1990 levels
under the Post-CAAA scenario. In the West, we
anticipate widespread changes under both scenarios;
however, wfc project that the increases in Pre-CAAA
ozone concentrations and decreases in Post-CAAA
model concentrations will be smaller than the pre-
39
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
dieted changes in ambient ozone lev-
els in the eastern U.S Furthermore,
comparison of 2010 Pre- and Post-
CAAA model estimates shows that
future-year western ozone concentra-
tions will be lower as a result of the
1990 Amendments, but UAM-V re-
sults indicate that the reductions in the
West will likely be about half the size
of the reductions in the eastern por-
tion of the country. The difference
between the change in western ozone
concentrations and the change in east-
ern ozone concentrations is largely
due to the more aggressive NO con-
trols expected in the East. Specifi-
cally, the Post-CAAA scenario incorporates the ef-
fects of a NOx cap-and-trade system for the eastern
U.S. (OTAG region). Another reason for the dif-
ference between the modeled change in eastern and
western ozone concentrations is that we estimated
ozone levels in the East and West using different
model grid resolutions. The coarser the resolution,
the less responsive the model concentration estimates
are to localized changes in emissions. Thus, the
smaller estimated change in western ozone concen-
trations than in eastern ozone concentrations may,
in part, be attributable to die fact that UAM-V grid-
cells covering the western U.S. are larger than those
covering the eastern U.S.
Western urban-area modeling results differ from
the regional scale results described above. Examina-
tion of Pre- and Post-CAAA modeling estimates
shows that, in some portions of the urban centers of
San Francisco and Los Angeles, future-year Post-
CAAA ozone concentrations are expected to be
higher than Pre-CAAA estimates. This ozone
"disbenefit" is the result of inhibiting a complex
chemical reaction termed "NOx scavenging," during
which a reduction in NOx, an ozone precursor, leads
to an increase in ozone production instead of the
typical decrease.3 In the area immediately surround-
ing the two cities, however, and in Maricopa County,
'' Scavenging occurs in areas, typically cities, with limited
VOC and abundant NO... In VOC-limited areas where there is
a relatively high NO concentration (regions where the concen-
tration of VOC, not NO dictates the amount of ozone that
can be iorrned), these two ozone precursors (VOC and NO)
compete to react with a particular gaseous compound. To pro-
duce ozone, this compound must combine with VOC. As a
result, if the compound joins with NO.., ozone production is
impeded.; thus, a decrease in NOv lead.s to an increase in ozone
concentrations.
Figure 4-2
Distribution of Monitor Level Ratios for 95th Percentile Ozone
Concentrations: 2010 Post-CAAA/Pre-CAAA
_ 60
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
model results show that scavenging is not expected
to be influential, if it occurs at all, and future-year
Post-CAAA ozone concentration estimates are pre-
dicted to be lower than Pre-CAAA estimates.
As described above, we used the UAM-V model
results to calculate adjustment factors for each of the
four future-year projection scenarios. Wre estimated
future-year monitor-level ozone concentrations by
applying these factors to 1990 observed concentra-
tions. Examination of the distribution of adjusted
monitor concentration ratios for 95th percentile
ozone concentrations is one means of analyzing the
impact of the CAAA on air pollution. The distri-
bution of ratios of 2010 Pre-CAAA to 1990 base-
year ozone concentrations reveals that the majority
of future year Pre-CAAA ozone concentration esti-
mates are between zero and 10 percent greater than
1990 levels, with most concentrations falling in the
middle of this range. The distribution of ratios of
2010 Post-CAAA to 1990 base-year shows that in
nearly all areas of the U.S. ozone concentrations will
be lower in 2010 than in the base-year; in the major-
ity of the country, future-year concentrations will
be five to 20 percent lower than in the base-year.4
The histogram in Figure 4-2 depicts the distribution
of ratios of 2010 Post-CAAA ozone estimates to 2010
Pre-CAAA ozone estimates. Most of the ratios in
the distribution are less than one, with a median of
0.883. This indicates that the 95th percentile level
Post-CAAA concentrations, with few exceptions, are
lower than the corresponding Pre-CAAA values.
The smaller the ratio, the greater the difference be-
tween future-year scenarios.
4 See Appendix C for histograms illustrating the change in
ozone concentrations iroin the base-year.
40
-------
Chapter 4: Air Quality Modeling
To model Pre- and Post-CAAA participate mat-
ter (PM1() and PM23) concentrations, we used
RADM/RPM for the' eastern U.S. and REMSAD
for the western U.S. Results from both models show
PM concentrations are expected to be lower under
the Post-CAAA scenario than under the Pre-CAAA
scenario. This projected improvement in air quality-
is widespread throughout the eastern U.S., with 2010
Post-CAAA PM estimates in some parts of the East
up to 15 to 30 percent lower than 2010 Pre-CAAA
estimates. In the West, projected reductions in fu-
ture-year PM concentrations (Pre-CAAA minus
Post-CAAA) are largely restricted to urban areas.5
The broad scale improvement in eastern PM con-
centrations is driven largely by reductions in utility
source sulfur dioxide emissions throughout this por-
tion of the country.6 In the West, however, sulfur
dioxide emissions have a much smaller impact on
overall PM concentrations. Western PM concen-
trations are more significantly influenced by area,
motor vehicle, and nonroad source emissions of ni-
trogen oxides and directly emitted PM. These
sources are more concentrated in urban areas. As a
result, the impact of the CAAA on PM concentra-
tions in the West is primarily restricted to urban
areas.
Examination of the distribution of adjusted
monitor-level concentration ratios for annual aver-
age PM concentrations reveals that 2010 Pre-CAAA
PM10 and PM25 estimates are both higher than 1990
base-year estimates in almost all areas of the coun-
try. Pre-CAAA 2010 PM]0 and PA^. estimates are
generally zero to 10 percent greater than 1990 base-
year estimates. The average estimated increase in
PM9g concentrations, however, is slightly larger than
the average estimated increase in PM1Q.' The esti-
mated change in PM concentrations from the base-
year to 2010 under the Post-CAAA scenarios is less
uniform. While the majority of areas experience a
reduction in annual average PM10 and PM25 concen-
trations, m a number of areas ambient PM levels,
more frequently PM, _, increase from the base-year
under the Post-CAAA scenario. On average, how-
ever, 2010 Post-CAAA PM10 and PM25 concentra-
tions are between zero and five percent and zero
and 10 percent, respectively, lower than 1990 base-
year concentrations.8
As shown in Figures 4-3 and 4-4, the percentage
reduction in PM25 concentrations across the U.S.
between the Pre- and Post-CAAA scenarios vary
more widely than the percentage reduction in PM.|0.
In the emissions analysis we focus on the impact of
the CAAA 011 anthropogenic emissions and, so, hold
natural source PM emissions constant at 1990 levels.
Natural source emissions make up a much larger
portion of PM]0 concentrations than PM,5 concen-
trations and dampen the influence of changes in an-
thropogenic emissions on ambient PM10 concentra-
tions.
Comparison of the two distributions in Figures
4-3 and 4-4 shows that, despite the greater variation
of PM,5 reductions, the percentage reduction in PM25
concentrations are larger on average than the per-
centage reduction in PM,0 concentrations. The rea-
son for this difference is two fold. First, as described
above, PM,5 concentrations are more susceptible to
the influence of changes in anthropogenic emissions,
which are regulated by the CAAA. Second, the
CAAA provisions that influence PM emissions (regu-
lations that focus on secondary PM precursors such
as NOx, and SO2, and primary PM sources such as
diesel engine exhaust standards) affect the fine par-
ticulate (PM, j) subset of PM]0 to a much greater ex-
tent than the coarser fraction that makes up the rest
of PM1Q. As a result of these two factors, the pro-
jected difference in ambient concentrations between
the Pre-CAAA and Post-CAAA scenarios reflect a
larger percentage reduction in PM25 than PMin.
5 Outside the larger urban areas in the West, REMSAD
results show little or no change in PM concentrations between
Pre- and Post-CAAA estimates.
" Sulfur dioxide is a secondary PM precursor.
In some of She figures in this chapter the Pre-CAAA and
Post-CAAA scenarios are referred to as Pre-CAAA90 and Post-
CAAA90, respectfully.
8 See Appendix C ior histograms illustrating the change in
PM concentrations from the 1990 base-year to each of the Pre-
CAAA and Post-CAAA iuture vear scenarios.
41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure 4-3
Distribution of Combined RADM/RPM and REMSAD Derived
Monitor Level Ratios for Annual Average PM10 Concentrations:
2010 Post-CAAA/Pre-CAAA
60
ST g
£ 2 30 -
20 -
10 -
0
median: 0.946
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure 4-4
Distribution of Combined RADM/RPM and REMSAD Derived
Monitor Level for Annual Average PM25 Concentrations:
2010 Post-CAAA/Pre-CAAA
60 -
> SO'
c
0)
£ 2 30"
g&
1 20-
* 10"
1 1 a . 1 1
o
median: 0.919
n
1
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Visibility
We also relied on RADM/RPM and REMSAD
to estimate the impact of the CAAA on future-year
visibility- Tables 4-2 and 4-3 compare the mean an-
nual visibility (expressed in deciviews)9 in selected
eastern urban areas and National Parks, respectively,
as estimated bv RADM/RPM under the 1990 base-
year and 2010 Pre- and Post-CAAA
scenarios. Comparison of these val-
ues reveals that, in the eastern U.S.,
we anticipate that future-year visibil-
ity in both urban and rural areas is
projected to improve under the Post-
CAAA scenario. RADM/RPM pre-
dicts that Post-CAAA visibility in
2010 will not only be better than Pre-
CAAA visibility, but also, in many
areas, it will be better than the vis-
ibility 111 the 1990 base-year. This
improvement in visibility is attrib-
utable to reductions in the concen-
tration of gaseous and suspended par-
ticles, such as PM, that scatter and
absorb light, and thus influence vis-
ibility.
Visibility in the West is also sig-
nificantly better under the Post-
CAAA scenario than under the Pre-
CAAA scenario (see Tables 4-4 and
4-5). Base-year model runs show that
visibility in the western U.S. is the
poorest in larger metropolitan areas
such as Los Angeles, CA; San Fran-
cisco, CA; Denver, CO; and Phoe-
nix, AZ. Under the 2010 Pre-CAAA
scenario, REMSAD estimates that,
throughout much of the West, vis-
ibility will remain relatively un-
changed from the base-year, and in
some cases will even improve. In the
metropolitan areas, however, the
model predicts visibility degradation.
Under the Post-CAAA scenario, however,
REMSAD estimates widespread improvement in
future-year visibility in the West. In both metro-
politan and non-urban areas, deciview levels esti-
mated for 2010 are lower under the Post-CAAA sce-
nario than under the Pre-CAAA scenario. The
model suggests Los Angeles and Las Vegas will ex-
perience the greatest improvement.
9 The deciview is a measure of visibility which captures (lie
relationship between air pollution and human perception of vis-
ibility. When air is free of the particles that cause visibility deg-
radation, the DeciView Haze Index is zero. The higher the
deciview level, the poorer the visibility; a one io two deciview
change translates lo a just noticeable change in visibility for most
individuals.
42
-------
Chapter 4: Air Quality Modeling
Table 4-2
Comparison of Visibility in Selected Eastern Urban Areas
Mean Annual Deciview*
Area Name
State
1990
Base-Year
2010
Pre-CAAA
2010
Post-CAAA
Atlanta Metro Area
GA
20.9
22.8
20.0
Boston Metro Area
MA
13.2
14.0
11.9
Chicago Metro Area
IL
17.5
19.1
17.0
Columbus
OH
16.5
17.7
15.1
Detroit Metro Area
Ml
16.0
18.5
15.3
Indianapolis
IN
20.1
21.1
19.0
Little Rock
AR
15.0
17.2
15.1
Milwaukee Metro Area
Wl
15.6
18.4
15.3
Minn.-St. Paul Metro Area
MN
10.1
12.4
10.3
Nashville
TN
20.4
21.5
19.0
New York City Metro Area
NY/NJ
15.2
18.0
13.9
Pittsburgh Metro Area
PA
15.8
16.9
14.2
St. Louis Metro Area
MO
16.5
17.8
16.0
Syracuse
NY
12.4
13.2
11.5
Washington, DC Metro Area
DC/VA/MD
17.5
19.2
16.3
*For cities or metropolitan areas not contained by a single RADM/RPM grid cell, the visibility measure
presented in this table is a weighted average of the mean annual deciview level from each of the grid
cells that together completely contain the selected area. Weighting is based upon the spatial
distribution of an area over the various grid cells.
Table 4-3
Comparison of Visibility in Selected Eastern National Parks
Mean Annual Deciview*
Area Name
Acadia NP
Everglades NP
Great Smoky Mtns. NP
Shenandoah NP
State
ME
FL
TN
VA
1990
Base-Year
11.1
7.6
20.4
16.5
2010
Pre-CAAA
12.0
9.2
22.3
17.8
2010
Post-CAAA
10.4
6.9
19.6
15.2
*For national parks not contained by a single RADM/RPM grid cell, the visibility measure presented in
this table is a weighted average of the mean annual deciview level from each of the grid cells that
together completely contain the selected area. Weighting is based upon the spatial distribution of an
area over the various grid cells.
43
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 4-4
Comparison of Visibility in Selected Western Urban Areas
Area Name
Denver
Las Vegas
Los Angeles
Phoenix
Salt Lake City
San Francisco
Seattle
State
CO
NV
CA
AZ
UT
CA
WA
1990
Base-Year
19.4
14.6
22.7
15.4
12.5
24.4
20.5
Mean Annual Deciview*
2010
Pre-CAAA
22.6
17.9
24.6
17.1
14.8
26.1
22.2
2010
Post-CAAA
21.0
15.2
22.0
15.3
13.4
24.6
21.0
*For cities not contained by a single REMSAD grid cell, the visibility measure presented in this table is a
weighted average of the mean annual deciview level from each of the grid cells that together completely
contain the selected area. Weighting is based upon the spatial distribution of an area over the various grid
cells.
Table 4-5
Comparison of Visibility in Selected Western National Parks
Area Name
Glacier NP
Grand Canyon NP
Olympic NP
Yellowstone NP
Yosemite NP
Zion NP
State
MT
AZ
WA
WY
CA
UT
1990
Base-Year
11.2
8.3
11.1
9.0
11.5
8.0
Mean Annual Deciview*
2010
Pre-CAAA
11.9
8.8
11.8
9.7
13.2
9.0
2010
Post-CAAA
11.5
8.3
11.7
9.5
12.2
8.4
*For national parks not contained by a single REMSAD grid cell, the visibility measure presented in this
table is a weighted average of the mean annual deciview level from each of the grid cells that together
completely contain the selected area. Weighting is based upon the spatial distribution of an area over the
various grid cells.
44
-------
Chapter 4: Air Quality Modeling
Figure 4-5
60
5 50 H
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure 4-7
Distribution of Monitor - Level Ratios of N02 Emissions
60
I
50 -
30-
20-
10-
0
median: 0.575
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20
Ratio
Note: 2.7 percent of the distribution of ratios is less than 0.40.
Figure 4-8
Distribution of Monitor - Level Ratios of CO Emissions
60
8 50 -
4C
s-
| 30 -
ff
£ 20 H
I 10 H
o
median: 0.720
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20
Ratio
Note: 15.7 percent of the distribution of ratios is less than 0.40.
Comparison of Pre- and Post-CAAA emission-
based adjustment factors also helps illustrate the ef-
fect of the 1990 Amendments 011 ambient pollution
concentrations. The ratio of 2010 Post-CAAA ad-
justment factors to 2010 Pre-CAAA adjustment fac-
tors shows the impact of the 1990 Amendments on
ambient concentrations relative to the baseline sce-
nario. Ratios less than one indicate
that we estimate that future-year con-
centrations of SCX, NO, NO2, and
CO are lower under the Post-CAAA
scenario than under the Pre-CAAA
scenario.
Figures 4-5 through 4-8 show the
distribution of 2010 Post-CAAA to
2010 Pre-CAAA ratios for summer-
time SO_, NO, NO,, and CO respec-
tively. These figures illustrate the re-
gional variation in die influence of die
1990 Amendments on ambient con-
centrations of these pollutants. Al-
though we estimate concentrations in
some areas will increase under the
Post-CAAA scenario relative to Pre-
CAAA estimates, the median sum-
mertime 2010 Post- to Pre-CAAA ra-
tios for SO,, NO, NO,, and CO arc
0.90, 0.67, 0.58, and 0.72 respectively.
These values, each less than one, indi-
cate that the central tendency for sum-
mertime 2010 Post-CAAA concentra-
tion estimates of these four pollutants
is to be lower than 2010 Pre-CAAA
estimates.
Table 4-6 displays the median val-
ues of the distribution of Post- to Pre-
CAAA ratios for the summer months
described above and the remaining three seasons. Just
as for die summer; spring, autumn, and winter me-
dian values are less than one. Averaged over all four
seasons, we estimate a median reduction in SO2,
NO, NO2, and CO concentrations of approximately
9, 33, 40, and 25 percent respectively. RACT re-
quirements, tailpipe emissions standards, and NOx
emissions trading account for the bulk of the reduc-
Table 4-6
Median Values of the Distribution of Ratios of 2010 Post-CAAA/
Pre-CAAA Adjustment Factors
Spring
Summer
Autumn
Winter
SO2
0.904
0.892
0.916
0.924
NO
0.669
0.666
0.677
0.686
NO2
0.598
0.575
0.614
0.626
CO
0.790
0.720
0.756
0.692
46
-------
Chapter 4: Air Quality Modeling
tion in NO and NO2 concentrations. Title I
nonattainment area controls and Title II motor ve-
hicle provisions are responsible for much of the
change in CO concentrations, while regulation of
utility and motor vehicle emissions account for ma-
jority of the decrease in SO0 concentrations.
Uncertainty in Air Quality
Estimates
.Many sources of uncertainty affect the precision
and accuracy of the projected changes in air quality
presented in this study. These uncertainties arise
largely from potential inaccuracies in the emissions
inventories used as air quality modeling inputs and
potential errors in the structure and parameteriza-
tion of the air quality models themselves. For ex-
ample, we estimated changes in PM concentrations
in the eastern U.S. based exclusively on changes in
the concentrations of sulfate and nitrate particles.
By not accounting for changes in organic and pri-
mary particulate fractions, we likely underestimate
the impact of the CAAA on PM concentrations.
Also, by using separate air quality models for indi-
vidual pollutants and different geographic regions,
as opposed to a single integrated model, we were
unable to fully capture the interaction among air
pollutants or reflect transport of pollutants or pre-
cursors across the boundaries of the models cover-
ing the western and eastern states. The direction
and magnitude of bias these limitations impose on
net benefits estimate presented in this analysis can
not be determined based on current information.
Some model-related uncertainties, however, may
be mitigated because this analysis uses the air qual-
ity modeling results in a relative, not absolute, sense.
We focus 011 the change in air quality between the
Pre- and Post-CAAA scenarios and not on the am-
bient concentrations projected by the individual
models themselves. Therefore, uncertainties that
affect a model's ability7 to accurately predict the rela-
tive change in concentration of a pollutant from one
scenario to another are more important in the con-
text of this study than those that affect only the ab-
solute model results.
The relatively coarse grid cells used to model
ozone in most areas of the U.S. represents a poten-
tial source of uncertainty affecting a model's sensi-
tivity to changes in emissions. Grid size affects chem-
istry, transport, and diffusion processes that in turn
determine the response of pollutant concentrations
to changes in emissions. The less accurately a model
can predict the impact of changes in emissions 011
ambient levels, the greater the uncertainty associ-
ated with predicted differences between Pre- and
Post-CAAA concentration estimates.
Table 4-7 presents the most important specific
sources of uncertainty and Appendix C further de-
scribes the uncertainties associated with air quality
modeling. While the list of potential errors presented
in Table 4-7 is not exhaustive, it includes discussion
of those factors with die greatest likelihood of con-
tributing to any potential bias in the primary net
benefit estimates.
47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 4-7
Key Uncertainties Associated with Air Quality Modeling
Potential Source of Error
Direction of
Potential Bias for
Net Benefits
Estimate
Likely Significance Relative to Key
Uncertainties in
Net Benefit Estimate*
PM-io and PM2.5 concentrations in
the East (RADM domain) are
based exclusively on changes in
the concentrations of sulfate and
nitrate particles, omitting the
effect of anticipated reductions in
organic or primary particulate
fractions.
Underestimate. Potentially major. Nitrates and sulfates
constitute major components of PM, especially
PM2.s, in most of the RADM domain and
changes in nitrates and sulfates may serve as a
reasonable approximation to changes in total
PM-io and total PM2.5. Of the other components,
primary crustal particulate emissions are not
expected to change between scenarios;
primary organic carbon particulate emissions
are expected to change, but an important
unknown fraction of the organic PM is from
biogenic emissions, and biogenic emissions are
not expected to change between scenarios. If
the underestimation is major, it is likely the
result of not capturing reductions in motor
vehicle primary elemental carbon and organic
carbon particulate emissions.
The number of PM2.s ambient
concentration monitors
throughout the U.S. is limited. As
a result, cross estimation of PM25
concentrations from PM-io (or
TSP) data was necessary in
order to complete the "monitor-
level" observational dataset
used in the calculation of air
quality profiles.
Unable to
determine based on
the current
information.
Potentially major. PM2.s exposure is linked to
mortality, and avoided mortality constitutes a
large portion of overall CAAA benefits. Cross
estimation of PM25, however, is based on
studies that account for seasonal and
geographic variability in size and species
composition of particulate matter. Also, results
are aggregated to the annual level, improving
the accuracy of cross estimation.
Use of separate air quality
models for individual pollutants
and for different geographic
regions does not allow for a fully
integrated analysis of pollutants
and their interactions.
Unable to
determine based on
current information.
Potentially major. There are uncertainties
introduced by different air quality models
operating at different scales for different
pollutants. Interaction is expected to be most
significant for PM estimates. However,
important oxidant interactions are represented
in all PM models and the models are being
used as designed. The greatest likelihood of
error in this case is for the summer period in
areas with NOx inhibition of ambient ozone
(e.g., Los Angeles).
Future-year adjustment factors
for seasonal or annual monitoring
data are based on model results
for a limited number of simulation
days.
Overall, unable to
determine based on
current information.
Probably minor. RADM/RPM and REMSAD
PM modeling simulation periods represent all
four seasons and characterize the full seasonal
distribution. Potential overestimation of ozone,
due to reliance on summertime episodes
characterized by high ozone levels and applied
to the May-September ozone season, is
mitigated by longer simulation periods, which
contain both high and low ozone days. Also,
underestimation of UAM-V western and UAM-
IV Los Angeles ozone concentrations (see
below) may help offset the potential bias
associated with this uncertainty.
48
-------
Chapter 4: Air Quality Modeling
Table 4-7
Key Uncertainties Associated with Air Quality Modeling (continued)
Direction of
Potential Source of Error
Comparison of modeled and
observed concentrations
indicates that ozone
concentrations in the western
states were somewhat under-
predicted by the UAM-V model,
and ozone concentrations in
the Los Angeles area were
underestimated by the UAM-IV
model.
Ozone modeling in the eastern
U.S. relies on a relatively
coarse 12 km grid, suggesting
NOX inhibition of ambient ozone
levels may be under
represented in some eastern
urban areas. Coarse grid may
affect both model performance
and response to emissions
changes.
UAM-V modeling of ozone in
the western U.S. uses a
coarser grid than the eastern
UAM-V (OTAG) or UAM-IV
models, limiting the resolution
of ozone predictions in the
West.
Emissions estimated at the
county level (e.g., area source
and motor vehicle NOX and
VOC emissions) are spatially
and temporally allocated based
on land use, population, and
other surrogate indicators of
emissions activity. Uncertainty
and error are introduced to the
extent that area source
emissions are not perfectly
spatially or temporally
correlated with these indicators.
The REMSAD model under-
predicted western PM
concentrations during fall and
winter simulation periods.
Potential Bias for Likely Significance Relative to Key
Net Benefits Uncertainties in
Estimate Net Benefit Estimate*
Unable to
determine based
on current
information.
Unable to
determine based
on current
information.
Unable to
determine based
on current
information.
Unable to
determine based
on current
information.
Unable to
determine based
on current
information.
Probably minor. Because model results are
used in a relative sense (i.e., to develop
adjustment factors for monitor data) the
tendency for UAM-V or UAM to underestimate
absolute ozone concentrations would be unlikely
to affect overall results. To the extent that the
model is not accurately estimating the relative
changes in ozone concentrations across
regulatory scenarios, the effect could be greater.
Probably minor. Though potentially major for
eastern ozone results in those cities with known
NOX inhibition, ozone benefits contribute only
minimally to net benefit projections in this study.
Grid size affects chemistry, transport, and
diffusion processes which in turn determine the
response to changes in emissions, and may also
affect the relative benefits of low-elevation
versus high-stack controls. However, the
approach is consistent with current state-of-the-
art for regional-scale ozone modeling.
Probably minor. Also, probably minor for ozone
results. Grid cell-specific adjustment factors for
monitors are less precise for the west and may
not capture local fluctuations. However,
exposure tends to be lower in the predominantly
non-urban west, and models with finer grids
have been applied to three key population
centers with significant ozone concentrations.
May result in underestimation of benefits in the
large urban areas not specifically modeled (e.g.,
Denver, Seattle) with finer grid.
Probably minor. Potentially major for estimation
of ozone, which depends largely on VOC and
NOX emissions; however, ozone benefits
contribute only minimally to net benefit
projections in this study.
Probably minor. Because model results are
used in a relative sense (i.e., to develop
adjustment factors for monitor data) REMSAD's
underestimation of absolute PM concentrations
would be unlikely to significantly affect overall
results. To the extent that the model is not
accurately estimating the relative changes in PM
concentrations across regulatory scenarios, or
the individual PM components (e.g., sulfates,
primary emissions) do not vary uniformly across
seasons, the effect could be greater.
49
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 4-7
Key Uncertainties Associated with Air Quality Modeling (continued)
Potential Source of Error
Direction of
Potential Bias for
Net Benefits
Estimate
Likely Significance Relative to Key
Uncertainties in
Net Benefit Estimate*
Lack of model coverage for acid
deposition in Western states.
Underestimate Probably minor. Because acid deposition tends
to be a more significant problem in the eastern
U.S. and acid deposition reduction contributes
only minimally to net monetized benefits, the
monetized benefits of reduced acid deposition
in the western states would be unlikely to
significantly alter the total estimate of
monetized benefits.
Uncertainties in biogenic
emissions inputs increase
uncertainty in the AQM
estimates.
Unable to
determine based on
current information.
Probably minor. Potentially major impacts for
ozone outputs, but ozone benefits contribute
only minimally to net benefit projections in this
study. Uncertainties in biogenics may be as
large as a factor of 2 to 3. These biogenic
inputs affect the emissions-based VOC/NOX
ratio and, therefore, potentially affect the
response of the modeling system to emissions
changes.
* The classification of each potential source of error reflects the best judgement of the section 812 Project Team. The
Project Team assigns a classification of "potentially major" if a plausible alternative assumption or approach could
influence the overall monetary benefit estimate by approximately five percent or more; if an alternative assumption or
approach is likely to change the total benefit estimate by less than five percent, the Project Team assigns a classification
of "probably minor."
50
-------
Human Health
Effects of
Criteria Pollutants
Health benefits resulting from improved air qual-
ity constitute a significant portion of die overall ben-
efits of the Clean Air Act Amendments of 1990. As
part of die prospective analysis of these amendments,
we have identified and, where possible, estimated
the magnitude of the health benefits that Americans
are likely to enjoy in future years as a result of the
CAAA. These health benefits are expressed as
avoided cases of air-pollution related health effects
such as premature mortality, heart disease, and res-
piratory illness. This chapter presents an overview
of our approach to modeling these changes in ad-
verse health effects, discusses key assumptions asso-
ciated with this approach, and summarizes model-
ing results for major health effect categories. Al-
though this chapter focuses predominantly 011 the
human health effects associated with exposure to
criteria pollutants, the final section of this chapter
presents a discussion of the effects associated with
air toxics and stratospheric ozone.
In general, this analysis finds that the CAAA
will result 111 significant reductions in mortality, res-
piratory illness, heart disease, and other adverse
health effects, with much of these reductions result-
ing from decreases in ambient particulate matter con-
centrations.
We estimate the impact of the CAAA on hu-
man health by analyzing the difference in the ex-
pected incidence of adverse health effects between
the Pre-CAAA and Post-CAAA regulatory sce-
narios. As described in Chapter 2, the Pre-CAAA
scenario assumes no further controls on criteria pol-
lutant emissions besides those already in place in
1990, while the Post-CAAA scenario assumes full
implementation of the 1990 CAAA. For each regu-
latory scenario, we use the Criteria Air Pollutant
Modeling System (CAPMS) to estimate the incidence
of health effects for 1990 (base-year), 2000, and 2010.
Modeling die incidence of adverse health effects re-
sulting from exposure to criteria air pollutants re-
quires three types of inputs: (1) estimates of the
changes in air quality for the Pre- and Post-CAAA
scenarios in 2000 and 2010; (2) estimates of the num-
ber of people exposed to air pollutants at a given
location; and (3) concentration-response (C-R) func-
tions that link changes in air pollutant concentra-
tions widi changes in adverse health effects. We dis-
cuss each of diese inputs in greater detail below.
Air Quality
The development of criteria pollutant concen-
tration estimates for use in the CAPMS model con-
sists of two steps. First, air quality modeling and
1990 base-year monitoring data arc used to project
ambient pollution levels at monitors throughout the
48 contiguous states. Second, because air quality
momtors are neither uniformly nor pervasively dis-
tributed across the country, concentration data at
monitors are extrapolated to non-monitored areas
in order to generate a more comprehensive air qual-
ity data set covering the 48 contiguous states and the
District of Columbia.
The projections of criteria pollutant concentra-
tions at air pollution monitors are developed as sum-
marized in Chapter 4 and described in detail in Ap-
pendix C. Briefly, baseline 1990 concentrations at
each monitor are adjusted using monitor- and pol-
lutant-specific adjustment factors to produce esti-
mates of concentrations in 2000 and 2010 for each
regulatory scenario. Each adjustment factor reflects
the relative change in the concentration of a pollut-
ant in a specific geographic area between 1990 and
the target year, as predicted by air quality modeling.
51
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
To develop pollutant concentration estimates for
the entire continental U.S. we extrapolate the 1990
monitor data and the future-year estimates to the
eight kilometer by eight kilometer CAPMS grid cells
that cover the 48 contiguous states. Within each of
these cells, we calculate an estimated pollutant con-
centration using data from nearby monitors accord-
ing to a distance-weighted averaging method de-
scribed in Appendix D. We then use these grid cell
pollutant concentration estimates to predict changes
in health effects among the population residing
within each cell.
Health benefits resulting from the CAAA are
related to the change in air pollutant exposure expe-
rienced by individuals. Because the expected changes
in pollutant concentrations vary from location to
location, individuals in different parts of the coun-
try may not experience the same level of health ben-
efits. This analysis apportions benefits among indi-
viduals by matching the change in air pollutant con-
centration in a CAPMS grid cell with the size of the
population that experiences that change.
As a result, we require an estimate of the distri-
bution of the U.S. population among CAPMS grid
cells. The grid-cell-specific population counts for
1990 are derived from U.S. Census Bureau block level
population data. Grid cell population estimates for
future years are extrapolated from 1990 levels using
the ratio of future-year and 1990 state-level popula-
tion estimates provided by the U.S. Bureau of Eco-
nomic Analysis.1
Concentration-Response Functions
We calculate the benefits attributable to the
CAAA as the avoided incidence of adverse health
effects. Such benefits can be measured using C-R
functions specific to each health effect. C-R func-
tions are equations that relate the change in the num-
ber of individuals in a population exhibiting a "re-
sponse" (in this case an adverse health effect such as
respiratory disease) to a change in pollutant concen-
tration experienced by that population. The C-R
functions used in CAPMS generate changes in the
incidence of an adverse health effect using three val-
ues: the grid-cell-specific change in pollutant concen-
tration, the grid-cell-specific population, and an esti-
mate of the change in the number of individuals that
suffer an adverse health effect per unit change in air
quality.2 As described in Appendix D, we derive
this last factor, as well as the specific form of the C-
R equation, from the published scientific literature
for each pollutant/health effect relationship of in-
terest.
Using the appropriate C-R functions, CAPMS
generates estimates for each grid-cell of the change
in incidence of a set of adverse health effects result-
ing from the incremental change in exposure between
the Pre- and Post-CAAA scenarios in 2000 and 2010.
For each health effect, CAPMS then generates na-
tional health benefits estimates by summing the an-
nual incidence change across all grid cells.
Each criteria pollutant evaluated in the 812 pro-
spective analysis has been associated with multiple
adverse health effects. The published scientific lit-
erature contains information that supports the esti-
mation of some, but not all, of these effects. Thus,
it is not possible currently to estimate all of the hu-
man health benefits attributable to the CAAA. In
addition, for some of the health effects we do quan-
tify, the current economic literature does not sup-
port the estimation of die economic value of these
effects. For each of the criteria pollutants we evalu-
ate in this analysis, Table 5-1 presents the health ef-
fects that are quantitatively estimated and those that
can not currently be quantified. The sixth criteria
pollutant, lead (Pb), is not included in this analysis
since airborne emissions of lead were virtually elimi-
nated by pre-1990 ('lean Air Act programs.
Analytical
The modeling of health benefits attributable to
the CAAA involves numerous judgments and as-
sumptions to address data limitations and other con-
straints. Each of these analytical assumptions affects
both the accuracy and precision with which we can
estimate health benefits of the CAAA, but some as-
1 U.S. Bureau of Economic Analysis. 1995. BEA. Regional
Projections to 2045: Volume 1, States. U.S. Department of Com-
merce. Washington, DC. July.
•" An estimate of the baseline incidence of the adverse health
eiiect may also be required ior certain C-R functions.
52
-------
Chapters: Human Health Effects of Criteria Pollutants
Table 5-1
Human Health Effects of Criteria Pollutants
Pollutant
Quantified Health Effects
Unquantified Health Effects1
Ozone Respiratory symptoms
Minor restricted activity days
Respiratory restricted activity days
Hospital admissions-
All Respiratory and
All Cardiovascular
Emergency room visits for asthma
Asthma attacks
Mortality
Increased airway responsiveness to stimuli
Inflammation in the lung
Chronic respiratory damage / Premature aging of the lungs
Acute inflammation and respiratory cell damage
Increased susceptibility to respiratory infection
Non-asthma respiratory emergency room visits
Particulate Mortality*
Matter Bronchitis - Chronic and Acute
(PM-io, New asthma cases
PM2.s) Hospital admissions -
All Respiratory and
All Cardiovascular
Emergency room visits for asthma
Lower respiratory illness
Upper respiratory illness
Shortness of breath
Respiratory symptoms
Minor restricted activity days
All restricted activity days
Days of work loss
Moderate or worse asthma status
(asthmatics)
Neonatal mortality*
Changes in pulmonary function
Chronic respiratory diseases
other than chronic bronchitis
Morphological changes
Altered host defense mechanisms
Cancer
Non-asthma respiratory emergency room visits
Carbon Hospital Admissions -
Monoxide All Respiratory and
All Cardiovascular
Behavioral effects
Other hospital admissions
Other cardiovascular effects
Developmental effects
Decreased time to onset of angina
Non-asthma respiratory emergency room visits
Nitrogen Respiratory illness
Oxides Hospital Admissions -
All Respiratory and
All Cardiovascular
Increased airway responsiveness to stimuli
Chronic respiratory damage / Premature aging of the lungs
Inflammation of the lung
Increased susceptibility to respiratory infection
Acute inflammation and respiratory cell damage
Non-asthma respiratory emergency room visits
Sulfur Hospital Admissions -
Dioxide All Respiratory and
All Cardiovascular
In exercising asthmatics:
Chest tightness,
Shortness of breath, or
Wheezing
Changes in pulmonary function
Respiratory symptoms in non-asthmatics
Non-asthma respiratory emergency room visits
f Some of the unqualified adverse health effects of air pollution may be associated with adverse health endpoints that we
have quantitatively evaluated (e.g., chronic respiratory damage and premature mortality). However, it is likely that the value
assigned to the quantified endpoint may not fully capture the value of the associated health effect (e.g., chronic respiratory
damage may result in significant pain and suffering prior to mortality). As a result, we include such effects separately in the
unquantified health effects column.
JAppendix D includes detailed discussion of the scientific evidence for these potential health effects and includes illustrative
benefit calculations for them. Current uncertainties in our understanding of these effects do not support including these
quantitative estimates in the overall CAAA benefits estimate. However, ozone-related mortality may be implicitly quantified in
the overall analysis as part of the PM mortality estimate because of the significant correlation between ozone and PM
concentrations.
* This analysis estimates avoided mortality using PM as an indicator of the criteria air pollutant mix to which individuals were
exposed.
53
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
sumptions introduce greater uncertainty into the
results than others. This section characterizes these
key assumptions and the associated uncertainties to
allow the reader to gain a better understanding of
the potential for misestimation of avoided health
effects. In addition, health benefits are presented as
ranges to reflect the aggregate effect of the uncer-
tainty in key variables (see Results section below).
This section discusses the most important analytical
assumptions of this modeling effort, grouped into
the following categories: (1) exposure analysis, (2)
selection and application of C-R functions, and (3)
estimation of changes in PM-related mortality.
The key analytical assumptions involved in esti-
mating exposure to criteria air pollutants relate to
two steps: die extrapolation of air quality data from
monitors and the mapping of population data to air
quality data.
As discussed above, actual ambient air pollution
data arc available only for a limited number of moni-
tor sites that arc not uniformly distributed across
the U.S. Thus, to estimate the impact of air pollu-
tion changes on the health of the U.S. population,
data from monitors are extrapolated to the cells of a
grid that covers the 48 contiguous states and are
matched with population data for each grid-cell.
Essentially, the extrapolation method uses data from
die closest set of monitors surrounding a grid-cell to
compute a weighted average concentration for that
cell. Monitors closer to the grid cell are assumed to
yield a more accurate estimate of air quality in the
cell; thus data from these monitors receive more
weight than data from more distant monitors when
calculating an air quality estimate for the cell.3 The
resulting estimates are uncertain because the geogra-
phy, weather, land use, and other factors influenc-
ing air pollution may differ significantly between a
grid cell and the monitor or monitors used to gener-
ate estimates of air quality, especially as the moni-
tor-to-grid-cell distance grows.4 As a result, they may
3 Specifically, monitor data are weighted based, oil the
inverse of the distance between the monitor and the grid-cell
center. Addition;!] information on the extrapolation method is
provided in Appendix D.
4 In order to address this issue for long-distance extrapola-
tion (i.e., grid cells greater than 50 kilometers from a monitor),
the method is modified to also incorporate air quality modeling
predictions for the source and target locations. See Appendix D
for details.
not sufficiently capture local variation in air pollu-
tion levels (e.g., hot spots).
However, since the uncertainty in these extrapo-
lated values is inversely proportional to the density
of monitors in a given area, and since air quality
monitors are more prevalent in high pollution areas
than in low pollution areas, this extrapolation
method estimates the air quality in high pollution
areas (where the potential benefits of the CAAA are
greatest) with greater certainty than in low pollu-
tion areas. Thus, grid-cell ozone estimates in the
eastern U.S., where ozone levels and ozone moni-
tor density are higher, are likely to be more accurate
than those in the west, where monitor coverage is
more sparse. Also, estimates of concentrations of
criteria pollutants, which are measured by a greater
number of monitors nationwide (PM, ozone, SO,,),
are expected to be less uncertain than estimates for
CO and NO , which are measured bv considerably
X
fewer monitors.
Air pollutant concentration changes are mapped
to grid-cell population data derived from U.S. Cen-
sus bureau data, and extrapolated to future years
using population growth estimates from the U.S.
Bureau of Economic Analysis. There arc two key
assumptions associated with this population map-
ping. First, we assume the population in each grid
cell grows at the same rate as the state population as
a whole. As a result, exposures (and potential ben-
efits) in individual grid cells may be either under- or
over-estimated if population growth varies from the
state average during the 1990 to 2010 period. This
uncertainty is likely to be more significant in larger
states such as California and Texas, which may have
more geographic variability in growth patterns.
Also, the effect of this assumption may be less sig-
nificant for large population centers because their
growth rate better approximates the growth rate of
the state as a whole. Second, we assume in the ex-
posure analysis that the population in the grid cell is
similar in terms of its activity patterns and demo-
graphic characteristics to the populations in the stud-
ies from which the C-R functions are derived. This
is a potentially significant uncertainty which is dis-
cussed further in the next section and in Appendix D.
54
-------
Chapters: Human Health Effects of Criteria Pollutants
ofC-R Functions
We rely on the most recent available, published
scientific literature to ascertain the relationship be-
tween air pollution and adverse human health ef-
fects. The uncertainties underlying those published
studies and our method for selecting studies that
could be used to derive C-'R functions likely con-
tributes to the uncertainty of the health effects re-
sults. For example, the uncertainty associated with
the current state of the published scientific litera-
ture could potentially have two contradictory influ-
ences 011 the results of this analysis. First, to the
extent that the published literature may collectively
overstate the effects of pollution, our analysis will
overstate the benefits of CAAA-related pollution
reduction. This overestimation is possible because
scientific journals tend to publish research report-
ing significant associations between pollution and
disease more often than research that fails to find
such associations. On the other hand, our analysis
may underestimate overall health benefits of the
CAAA because, as the state of the science evolves,
current pollutant/health effect associations may be
found to be stronger than previously thought, and
new associations may be identified. For example, in
recent years, studies have shown the potential health
benefits from reductions in ambient PM to be much
greater than previously believed. To the extent that
the present analysis does not include health effects
whose link to air pollution has not been subject to
adequate scientific inquiry, this analysis may under-
state CAAA-related health benefits.
Our method of identifying appropriate C-R func-
tions for use in the benefits analysis may also intro-
duce uncertainty. We evaluate studies using the nine
selection criteria summarized in Table 5-2 and de-
scribed 111 detail in Appendix D. These criteria in-
clude consideration of whether the study was peer-
reviewed, the study design and location, and charac-
teristics of the study population, among others. The
selection of C-R functions for the benefits analysis
is guided by the goal of achieving a balance between
comprehensiveness and scientific defensibility. How-
ever, to the extent that this selection process may
lead to the exclusion of valid studies, the process in-
troduces uncertainty into the analysis. The overall
effect of this uncertainty is expected to be minor,
given the emphasis of the selection process on scien-
tific validity. Appendix D lists the studies selected
for each category of health effects, and presents the
associated C-R functions for each criteria pollutant.
Once the C-R functions have been selected, un-
certainty may also enter the analysis due to both
within-study and across-study variation in C-R func-
tions for individual health effects. Within-study
variation refers to the uncertainty and error that may
surround a given study's estimate of a C-R function.
Health effects studies provide both "best estimates"
of the relationship between air quality changes and
health effects and a measure of the statistical uncer-
tainty of the relationship. We use statistical simula-
tion modeling techniques to evaluate the overall
uncertainty of the results given the uncertainties as-
sociated with individual studies. Across-study varia-
tion refers to the fact that different published stud-
ies of the same pollutant/health effect relationship
typically do not report identical findings; in some
instances the differences are substantial. These dif-
ferences can exist even between equally reputable
studies and may result in health effect estimates that
vary considerably.
Across-study variation can result from two pos-
sible causes. One possibility is that studies report
different estimates of the single true relationship
between a given pollutant and a health effect due to
differences in study design, random chance, or other
factors. For example, a hypothetical study conducted
in New York and one conducted in Seattle may re-
port different C-R functions for the relationship
between PM and mortality in part because of differ-
ences between these two study populations (e.g.,
demographics, activity patterns). Alternatively,
study results may differ because they are in fact esti-
mating different relationships; that is, the same re-
duction in PM in New7 York and Seattle may result
in different reductions in premature mortality. This
may result from a number of factors, such as differ-
ences in the relative sensitivity of these two popula-
tions to PM pollution and differences in the compo-
sition of PM in these two locations.5 In either case,
where we identify multiple studies that are appro-
5 PM is a mix of particles of varying size and chemical
properties. The composition oi PM can vary considerably irorn
one region to another depending on the sources oi pardculate
emissions in each region.
55
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 5-2
Summary of Considerations Used in Selecting C-R Functions
Consideration
Comments
Peer reviewed Peer reviewed research is preferred to research that has not undergone the peer review
research process.
Study type Among studies that consider chronic exposure (e.g., over a year or longer) prospective
cohort studies are preferred over cross-sectional studies (a.k.a. "ecological studies")
because they control for important confounding variables that cannot be controlled for in
cross-sectional studies. If the chronic effects of a pollutant are considered more important
than its acute effects, prospective cohort studies may also be preferable to longitudinal time
series studies because the latter type of study is typically designed to detect the effects of
short-term (e.g. daily) exposures, rather than chronic exposures.
Study period Studies examining a relatively longer period of time (and therefore having more data) are
preferred, because they have greater statistical power to detect effects. More recent
studies are also preferred because of possible changes in pollution mixes, medical care,
and life style overtime.
Study population
Studies examining a relatively large sample are preferred. Studies of narrow population
groups are generally disfavored, although this does not exclude the possibility of studying
populations that are potentially more sensitive to pollutants (e.g., asthmatics, children,
elderly). However, there are tradeoffs to comprehensiveness of study population.
Selecting a C-R function from a study that considered all ages will avoid omitting the
benefits associated with any population age category. However, if the age distribution of a
study population from an "all population" study is different from the age distribution in the
assessment population, and if pollutant effects vary by age, then bias can be introduced
into the benefits analysis.
Study location U.S. studies are more desirable than non-U.S. studies because of potential differences in
pollution characteristics, exposure patterns, medical care system, and life style.
Pollutants Models with more pollutants are generally preferred to models with fewer pollutants, though
included in careful attention must be paid to potential collinearity between pollutants. Because PM has
model been acknowledged to be an important and pervasive pollutant, models that include some
measure of PM are highly preferred to those that do not.
Measure of PM PM25 and PM-io are preferred to other measures of particulate matter, such as total
suspended particulate matter (TSP), coefficient of haze (COH), or black smoke (BS) based
on evidence that PM2.s and PM-|0 are more directly correlated with adverse health effects
than are these other measures of PM.
Economically Some health effects, such as forced expiratory volume and other technical measurements
valuable health of lung function, are difficult to value in monetary terms. These health effects are not
effects quantified in this analysis.
Non-overlapping
endpoints
Although the benefits associated with each individual health endpoint may be analyzed
separately, care must be exercised in selecting health endpoints to include in the overall
benefits analysis because of the possibility of double counting of benefits. Including
emergency room visits in a benefits analysis that already considers hospital admissions, for
example, will result in double counting of some benefits if the category "hospital
admissions" includes emergency room visits.
priate for estimating a given health effect, we use
the multiple C-R estimates, applied to the entire U.S.,
to derive a range of possible results for that health
effect.
Whether this analysis estimates the C-R relation-
ship between a pollutant and a given health endpoint
using a single function from a single study or using
multiple C-R functions from several studies, each
C-R relationship is applied throughout the U.S. to
generate health benefit estimates. However, to the
extent that pollutant/health effect relationships are
region-specific, applying a location-specific C-R func-
tion at all locations in the U.S. may result in overes-
timates of health effect changes in some locations
and underestimates of health effect changes in other
locations. It is not possible, however, to know the
extent or direction of the overall effect on health
benefit estimates introduced by application of a single
C-R function to the entire U.S. This may be a sig-
56
-------
Chapters: Human Health Effects of Criteria Pollutants
nificant uncertainty in the analysis, but the current
state of the scientific literature does not allow for a
region-specific estimation of health benefits.
This section discusses the estimation of one of
the most serious health impacts of air pollution: pre-
mature mortality associated with PM exposure. This
section consists of three parts. It begins with a dis-
cussion of the uncertainties surrounding the PM/
mortality relationship. Then, it presents specific
factors to consider when selecting a PM mortality
C-R function. It ends with a brief discussion of the
advantages and disadvantages of the study we selected
for the PM mortality analysis: Pope et al., 1995.
in the PM
Health researchers have consistently linked air
pollution, especially PM, with excess mortality. A
substantial body of published scientific literature
recognizes a correlation between elevated PM con-
centrations and increased mortality rates. However,
there is much about this relationship that is still un-
certain.6 These uncertainties include:
• Causality. For this analysis, we assume a
causal relationship between exposure to el-
evated PM and premature mortality, based
011 the evidence of a correlation between PM
and mortality reported in the scientific lit-
erature. This assumption is necessary be-
cause the epidemiological studies on which
this analysis relies, by design, can not defini-
tively prove causation.
• Other Pollutants. PM concentrations are
correlated with the concentrations of other
criteria pollutants, such as ozone and CO,
and it is unclear how much each pollutant
may influence elevated mortality rates. Re-
cent studies have explored whether ozone
and CO may have mortality effects indepen-
dent of PM, but we do not view the evidence
as sufficient to include such effects in the
overall CAAA-related health benefits esti-
mate.7 As a result, we use the reported PM/
mortality relationship as a proxy for the
mortality effects of the air pollutant mixture.
Shape of the C-R Function. The shape of
the true PM mortality C-R function is un-
certain, but this analysis assumes the C-R
function to have a log-linear form (as derived
from the literature) throughout the relevant
range of exposures.8 If this is not the cor-
rect form of the C-R function, or if certain
scenarios (e.g., 2010 Pre-CAAA) predict con-
centrations well above the range of values
for which the C-R function was fitted,
avoided mortality may be mis-estimated.
Regional Differences. As discussed earlier,
significant variability exists in the results of
different PM studies. This variability may
reflect regionally-specific C-R functions re-
sulting from regional differences in factors
such as the physical and chemical composi-
tion of PM. If true regional differences ex-
ist, applying these C-R functions to regions
other than the study location would result
in mis-estimation of effects in these regions.
Exposure/Mortality Lags. It is currently
unknown whether there is a time lag — a
delay between changes in PM exposures and
changes in mortality rates — in the chronic
PM/mortality relationship. The existence
of such a lag could be important for the valu-
ation of benefits, if one were to assume that
lagged incidences of premature mortality
should be discounted over the period be-
tween when the fatal increment of exposure
is experienced and premature mortality ac-
tually occurs. Although there is no specific
scientific evidence of the existence or struc-
ture of a PM effects lag, current scientific
literature on adverse health effects such as
those associated with PM (e.g., smoking-re-
lated disease) leaves us skeptical that all inci-
' Appendix D discusses the evidence linking both ozone
and CO with mortality. Tt also describes and presents the re-
sults of an illustrative analysis estimating CAAA-related reduc-
tions in ozone-related mortality using currently available stud-
5 The morbidity studies used in this analysis may also be
subject to many of the uncertainties listed in this section.
8 C-R functions for other health effects may be assumed to
be linear or log-linear. See Appendix D for more details.
57
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
dences of premature mortality associated
with a given incremental change in PM ex-
posure would occur in the same year as the
exposure reduction. This same literature
implies that lags of up to a fewf years are plau-
sible, and we chose to assume a five-year lag-
structure, with 25 percent of deaths occur-
ring in the first year, another 25 percent in
the second year, and 16.7 percent in each of
the remaining three years.
Cumulative Effects. We attribute the PM/
mortality relationship used in this study
(Pope et al., 1995) primarily to PM-associ-
ated cumulative damage to the cardiopulmo-
nary system, since the short-term mortality
estimates reported in time-series studies ac-
count for only a minor fraction of total ex-
cess mortality. However, the relative roles
of exposure duration and exposure level re-
main unknown at this time.
of a PM C-R
Function
In addition to the study selection criteria listed
in Table 5-2, we consider three additional factors
when selecting a PM mortality function. The first
focuses on the PM indicator (i.e., PM10 or PM23), the
second focuses on whether the stud}' measured short-
term or long-term PM exposure, and the third fo-
cuses 011 whether the study used a cohort or eco-
logic design.
Current research suggests that particle size mat-
ters \vhen estimating the health impacts of PM. Par-
ticulate matter is a heterogeneous mixture that in-
cludes particles of varying sizes. Fine PM is gener-
ally viewfed as having a more harmful impact than
coarse PM, especially for coarse particles larger than
10um in aerodynamic diameter, although it is not
clear to what extent this may differ by the type of
health effect or the exposed population. While one
cannot necessarily assume that coarse PM has no
adverse impact on health, we prefer the use of PM0 5
as the best currently available measure of the impact
of PM on mortality.9
9 Due to the relative abundance of studies using PM1C, how-
ever, and the reasonably good correlation between PM, and
PM the 812 prospective analysis also uses PM.,. studies to esti-
mate the impact of PM on non-mortality health effects.
Two types of exposure studies (short-term and
long-term) have been used to estimate a PM/mortal-
ity relationship. Short-term exposure studies attempt
to relate short-term (often day-to-day) changes in PM
concentrations and changes in daily mortality rates
up to several days after a period of elevated PM con-
centrations. Long-term exposure studies examine
the potential relationship between longer-term (e.g.,
annual) changes in exposure to PM and annual mor-
tality rates. Researchers have found significant cor-
relations using both types of studies; however, for
this analysis, we rely exclusively on long-term stud-
ies to quantify PM mortality effects, though the
short-term studies provide additional scientific evi-
dence supporting the PM/mortahty relationship.
Because short-term studies focus only on the
acute effects associated with daily peak exposures,
they are unable to evaluate the degree to which ob-
served excess mortality is premature,10 and they may
underestimate the C-R coefficient because they do
not account for the cumulative mortality effects of
long-term exposures (i.e., exposures over many years
rather than a few days). Long-term studies, on the
other hand, are able to discern changes in mortality
rates due to long-term exposure to elevated air pol-
lution concentrations, and are not limited to mea-
suring mortalities that occur within a few days of a
high-pollution event (though they may not predict
cases of premature mortality that were only has-
tened by a few days). Consequently, the use of C-R
functions derived from long-term studies is likely to
result in a more complete assessment of the effect of
air pollution on mortality risk. However, to the
extent that long-term studies fail to capture acute
mortality effects related to peak exposures, the use
of long-term mortality studies may underestimate
CAAA-related avoided mortality benefits.
Among long-term PM studies, \ve prefer studies
using a prospective cohort design to those using an
ecologic or population-level design. Prospective
10 This can be important in cost-benefit analysis if benefits
are estimated in terms of life-years lost. In short-term studies
evaluating peak pollution events, it is likely that many of the
''excess mortality" cases represented individuals who were al-
ready suiiering impaired health, and ior whom the high-pollu-
tion event represented an exacerbation oi an already serious
condition. Based on the episodic studies only, however, it is
unknown how many of the victims would have otherwise lived
only a few more days or weeks, or how many would have re-
covered to enjoy many years of a healthy life in the absence of
the high-pollution event.
58
-------
Chapters: Human Health Effects of Criteria Pollutants
cohort studies follow individuals forward in time for
a specified period, periodically evaluating each
individual's exposure and health status. Population-
level ecological studies assess the relationship between
population-wide health information (such as counts
for daily mortality) and ambient levels of air pollu-
tion. Prospective cohort studies are preferred be-
cause they are better at controlling a source of un-
certainty known as "confounding." Confounding
is the mis-estimation of an association that results if
a study does not control for factors that are corre-
lated with both the outcome of interest (e.g., mor-
tality) and the exposure of interest (e.g., PM expo-
sure). For example, smoking is associated with mor-
tality. If populations in high PM areas tend to smoke
more than populations in low PM areas, and a PM
exposure study does not include smoking as a factor
in its model, then the mortality effects of smoking
may be erroneously attributed to PM, leading to an
overestimate of the risk from PM. Prospective co-
hort studies are better at controlling for confound-
ing than ecologic studies because the former follow
a group of individuals forward in time and can gather
individual-specific information on important risk
factors such as smoking. However, it is always pos-
sible, even in well-designed studies, that a relevant
risk factor (e.g., climate, the presence of other pol-
lutants) may not have been adequately considered
or controlled for. As a result, it is possible that dif-
ferences in mortality rates ascribed to differences in
average PM levels may be due, in part, to some other
factor or factors (e.g., differences among communi-
ties in diet, exercise, ethnicity, climate, industrial
effluents, etc.) that have not been adequately ad-
dressed 111 the exposure models.
The Pope Study
Three recent studies have examined the relation-
ship between mortality and long-term exposure to
PM: Pope et al. (1995), Dockery et al. (1993), and
Abbey et al. (1991). Of these three studies, we pre-
fer using the Pope et al. study as the basis for devel-
oping the primary PM mortality estimates in this
analysis. Pope et al. studied the largest cohort, had
the broadest geographic scope, and effectively con-
trolled for potentially significant sources of con-
founding.
Pope et al. examined a much larger population
(over 295,000) and many more locations (50 metro-
politan areas) than either the Dockery study or the
Abbey study. The Dockery study covered a cohort
of over 8,000 individuals in six U.S. cities, and the
Abbey study covered a cohort of 6,000 people in
California. In particular, the cohort in the Abbey
study was considered substantially too small and too
young to enable the detection of small increases in
mortality risk. The study was therefore omitted
from consideration in this analysis. Even though
Pope et al. (1995) reports a smaller premature mor-
tality response to elevated PM than Dockery et al.
(1993), the results of the Pope study are nevertheless
consistent with those of the Dockery study.
Pope et al., (1995) is unique in that it followed a
largely white and middle class population. The use
of this study population reduces the potential for
confounding because it decreases the likelihood that
differences in premature mortality across locations
were attributable to differences in socioeconomic
status or related factors rather than PM. However,
the demographics of the study population may also
produce a downward bias in die PM mortality coef-
ficient, because short-term studies indicate that the
effects of PM tend to be significantly greater among
groups of lower socioeconomic status.
Although it is the strongest of the PM cohort
studies, Pope et al. does have some limitations. For
example, Pope et al. did not consider the migration
of cohort members across study cities, which would
cause exposures to be more similar across individu-
als than those indicated by assigning city-specific
annual average pollution levels to each member of
the cohort. As intercity migration increases among
cohort members, the exposure experienced by mi-
grating individuals will tend toward an intercity
mean. If this migration is significant and is ignored,
approximating true differences in exposure levels
by differences in city-specific annual average PM lev-
els will exaggerate changes in exposure, resulting in
a downward bias of the PM coefficient. This occurs
because a given difference in mortality rates is being
associated with a larger difference in PM levels than
that actually experienced by individuals in the study
cohort. When the relationship between elevated PM
exposure and premature mortality derived from the
59
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pope et al. study is applied in the present analysis,
the effect of the potential mis-specification of expo-
sure due to migration in the underlying study is to
underestimate PM-related mortality reduction ben-
efits attributable to the CAAA.
Also, Pope et al. only included PM when esti-
mating a C-R function. Because PM concentrations
are correlated with the concentrations of other cri-
teria air pollutants (e.g., ozone), and because these
other pollutants may be correlated with premature
mortality (see Appendix D), the PM risk estimate
may be overestimated because it includes the mor-
tality impacts of these confounders. However, in
an effort to avoid overstating benefits, and because
the evidence associating mortality with PM expo-
sure is stronger than for other pollutants, the 812
Prospective analysis uses PM as a surrogate for PM
and related criteria pollutants.
Although we use the Pope study exclusively to
derive our primary estimates of avoided mortality,
die C-R function based on Dockery et al. (1993) may
provide a reasonable alternative estimate. While the
Dockery et al. study used a smaller sample of indi-
viduals from fewer cities than the study by Pope et
al., it features improved exposure estimates, a slightly
broader study population (adults aged 25 and older),
and a follow-up period nearly twice as long as that
of Pope et al. We present an alternative estimate of
the premature adult mortality associated with long-
term PM exposure based on Dockery et al. (1993) in
Chapter 8 and in Appendix D. We emphasize, how-
ever, that the estimate based on Pope et al. (1995) is
our primary estimate of the effect of the 1990 Amend-
ments on this important health effect.
Results
This section presents a summary of the differ-
ences in health effects resulting from improvements
in air quality between the Pre-CAAA and Post-
CAAA scenarios. Table 5-3 summarizes the CAAA-
related avoided health effects in 2010 for each study
included in the analysis. The mean estimate is pre-
sented as the Primary Central estimate, the 5th per-
ccntilc observation from the statistical uncertainty
modeling is presented as the Primary Low estimate,
and the 95th pcrcentile observation is presented as
the Primary High estimate of the number of avoided
cases of each endpomt.11 To provide context for these
results, Table 5-3 also expresses the mean reduction
in incidence for each adverse health effect as a per-
centage of the baseline incidence of that effect (ex-
trapolated to the appropriate future year) for the
population considered (e.g., adults over 30 years of
age). In general, because the differences in air qual-
ity between the Pre- and Post-CAAA scenarios are
expected to increase from 1990 to 2010 and because
population is also expected to increase during that
time, the health benefits attributable to the CAAA
are expected to increase consistently from 1990 to
2010. .More detailed results are presented in Appen-
dix D.
Table 5-3 summarizes the avoided mortality due
to reductions in PM exposure in 2010 between the
Pre- and Post-CAAA scenarios. As this table shows,
our Primary Central estimate implies that PM re-
ductions due to the CAAA in 2010 will result in
23,000 avoided deaths, with a Primary Low and Pri-
mary High bound on this estimate of 14,000 and
32,000 avoided deaths, respectively. The Primary
Central estimate of 23,000 avoided deaths represents
roughly one percent of the projected annual non-
accidental mortality of adults aged 30 and older in
the year 2010. Additionally, Table 5-4 summarizes
the distribution of avoided mortality for 2010 by
age cohort, along with the expected remaining life-
span (i.e., the life years lost) for the average person
111 each age cohort. The majority of the estimated
deaths occur in people over the age of 65 (due to
their higher baseline mortality rates), and this group
has a shorter life expectancy relative to other age
groups. The life years lost estimates might be higher
if data were available for PM-relatcd mortality in
the under 30 age group.
1' The Primary Low, Primary Central and Primary High
health benefit estimates represent points on a distribution of
estimated incidence changes for each health effect. This distri-
bution reflects the uncertainty associated with the coefficient of
the C-R function for each health endpoint. More information
about C-R Junction uncertainty and the uncertainty modeling
that generates the results distributions is presented in Appendix
D.
60
-------
Chapters: Human Health Effects of Criteria Pollutants
Table 5-3
Change in Incidence of Adverse Health Effects Associated with Criteria Pollutants in 2010
(Pre-CAAA minus Post-CAAA) - 48 State U.S. Population (avoided cases per year)
Endpoint
Mortality
ages 30 and older
Chronic Illness
chronic bronchitis
chronic asthma
Hospitalization
respiratory
admissions
cardiovascular
admissions
emergency room
visits for asthma
Minor Illness
acute bronchitis
upper respiratory
Pollutant
PM
PM
03
PM, CO, NO2,
S02, 03
PM, CO, NO2,
SO2, O3
PM, O3
PM
PM
5th %
14,000
5,000
1,800
13,000
10,000
430
0
280,000
2010
mean
23,000
20,000
7,200
22,000
42,000
4,800
47,000
950,000
95th %
32,000
34,000
12,000
34,000
100,000
14,000
94,000
1,600,000
% of Baseline
Incidences for
the mean
estimates a
2010
1.00%
3.14%
3.83%
0.62%
0.86%
0.55%
5.06%
0.86%
symptoms
lower respiratory PM
symptoms
respiratory illness NO2
moderate or worse PM
asthmac
asthma attacks0 O3, PM
chest tightness, SO2
shortness of breath,
or wheeze
shortness of breath PM
work loss days PM
minor restricted O3, PM
activity days / any of
19 respiratory
symptomsd
restricted activity PM
daysc
240,000
76,000
80,000
920,000
290
26,000
3,600,000
25,000,000
520,000
330,000
400,000
1,700,000
110,000
91,000
4,100,000
31,000,000
10,000,000 12,000,000
770,000
550,000
720,000
2,500,000
520,000
150,000
4,600,000
37,000,000
13,000,000
3.57%
10.44%
0.24%
1.04%
0.003%
1.69%
0.94%
2.15%
1.00%
a The baseline incidence generally is the same as that used in the C-R function for a particular health effect. However, there are a few
exceptions. To calculate the baseline incidence rate for respiratory-related hospital admissions, we used admissions for persons of all
ages for International Classification of Disease (ICD) codes 460-519; for cardiovascular admissions, we used admissions for persons of
all ages for ICD codes 390-429; for emergency room visits for asthma, we used the estimated ER visit rate for persons of all ages; for
chronic bronchitis we used the incidence rate for individuals 27 and older; for the pooled estimate of minor restricted activity days and
any-of-19 respiratory symptoms, we used the incidence rate for minor restricted activity days.
b Percentage is calculated as the ratio of avoided mortality to the projected baseline annual non-accidental mortality for adults aged 30
and over. Non-accidental mortality was approximately 95% of total mortality for this subpopulation in 2010.
0 These health endpoints overlap with the "any-of-19 respiratory symptoms" category. As a result, although we present estimates for
each endpoint individually, these results are not aggregated into tine total benefits estimates.
d Minor restricted activity days and any-of-19 respiratory symptoms have overlapping definitions and are pooled.
61
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
We report non-fatal health effects estimates in a
similar manner to estimates of premature mortali-
ties: as a range of estimates for each quantified health
cndpoiiit, with the range dependent on the quanti-
fied uncertainties in the underlying concentration-
response functions. The range of results for 2010
only is characterized in Table 5-3 with 5th percen-
tile, mean, and 95th percentile estimates which cor-
respond to the Primary Low, Primary Central, and
Primary High estimates, respectively. All estimates
are expressed as new cases avoided in 2010, with the
following exceptions. Hospital admissions reflect
admissions for a range of respiratoty and cardiovas-
cular diseases, and these results, along with emer-
gency room visits for asthma, do not necessarily rep-
resent the avoidance of new cases of disease (i.e., air
pollution may simply exacerbate an existing condi-
tion, resulting in an emergency room visit or hospi-
tal admission). Further, each admission is only
counted once, regardless of the length of stay in the
hospital. "Shortness of breath" is expressed in terms
of symptom days: that is, one "case" represents one
child experiencing shortness of breath for one day.
Likewise, "Restricted Activity Days" and "Work
Loss Days" are expressed in person-days.
Of
This section discusses the health effects associ-
ated with non-criteria air pollutants regulated by the
Clean Air Act Amendments of 1990. It first dis-
cusses the effects of pollutants known as "air tox-
ics", and then summarizes the effects associated with
stratospheric ozone depleting substances.
of Air Toxics
In addition to addressing the control of criteria
pollutants, the Clean Air Act Amendments re-
vamped regulations for air toxics — defined as non-
criteria pollutants which can cause adverse effects to
human health and to ecological resources — under
section 112 of the Act. Among other changes, the
1990 Amendments establish a list of air toxics to be
regulated, require EPA to establish air toxic emis-
sions standards based on maximum achievable con-
trol technology (MACT standards), and include a
provision that requires EPA to establish more strin-
gent air toxic standards if MACT controls do not
sufficiently protect the public health against residual
risks. Control of air toxics is expected to result both
from these changes and from incidental control due
to changes in criteria pollutant programs.
Table 5-4
Mortality Distribution by Age in Primary Analysis (2010 only), Based on Pope et al. (1995)a
Age Group
Proportion of Premature Mortality by Age
Life Expectancy (years)
Infants
not estimated
1-29
30-34
35-44
45-54
55-64
65-74
75-84
85+
not estimated
1%
4%
6%
12%
24%
30%
24%
-
48
38
29
21
14
9
6
' Results based on PM-related mortality incidence estimates for the 48 state U.S. population.
1 Percentages may not sum to 100 percent due to rounding.
62
-------
Chapters: Human Health Effects of Criteria Pollutants
For several decades, the primary focus of risk
assessments and control programs designed to reduce
air toxics has been cancer. According to present EPA
criteria, over 100 air toxics are known or suspected
carcinogens. EPA's 1990 Cancer Risk study indi-
cated that as many as 1,000 to 3,000 cancers annu-
ally may be attributable to the air toxics for which
assessments were available (virtually all of this esti-
mate came from assessments of about a dozen well-
studied pollutants).12 We note, however, that the
results of this analysis are based, in part, on conser-
vative, upper-bound estimates of chemical specific
risk factors.
In addition to cancer, inhalation of air toxics
compounds can cause a wide variety of health ef-
fects, including neurotoxicity, respiratory problems,
and adverse reproductive and developmental effects.
However, there has been considerably less work
done to assess the magnitude of non-cancer effects
from air toxics.
Air toxics can also cause adverse health effects
via non-inhalation exposure routes. Persistent
bioaccumulating pollutants, such as mercury and
dioxins, can be deposited into water or soil and sub-
sequently taken up by living organisms. The pol-
lutants can biomagnify through the food chain and
exist 111 high concentrations when consumed by
humans in foods such as fish or beef. The resulting
exposures can cause adverse effects in humans.
Finally, there are a host of other potential eco-
logical and welfare effects associated with air toxics,
for which very little exists in the way of quantita-
tive analysis. Toxic effects of these pollutants have
the potential to disrupt both terrestrial and aquatic
ecosystems and contribute to adverse welfare effects
such as fish consumption advisories in the Great
hakes.13
*2 These pollutants included PIC (products of incomplete
combustion), 1,3-butadieiie, hexavalent chromium, benzene,
formaldehyde, chloroform, asbestos, arsenic, ethylene
dibromide, dioxiri, gasoline vapors, and ethylene dichloride. See
U.S. EPA, Cancer Risk from Outdoor Exposure to Air 'Toxics.
EPA-450/1-90-0041. Prepared by EPA/OAR/OAQPS.
1J U.S. EPA, Office of Air Quality Planning and Standards.
"Deposition of Air Pollutants to the Great Waters, First Report
to Congress," May 1994. EPA-453/R-93-055.
Unfortunately, the effects of air toxics emissions
reductions could not be quantified for the present
study. Unlike criteria pollutants, monitoring data
for air toxics are relatively scarce, and the data that
do exist cover only a handful of pollutants. Emis-
sions inventories are very limited and inconsistent,
and air quality modeling has only been performed
for a few source categories. In addition, the scien-
tific literature on the effects of air toxics is generally
much weaker than that available for criteria pollut-
ants. Appendix I presents a list of research needs
identified by the Project Team which, if met, would
enable at least a partial assessment of air toxics ben-
efits in future section 812 prospective studies.
for Provisions to
Protect
We estimate benefits of stratospheric ozone pro-
tection programs by relying on analyses conducted
to support a scries of regulatory support documents
for these provisions. The series of basic steps to ar-
rive at physical effects estimates — from emissions
estimation, atmospheric modeling, exposure assess-
ment, and dose-response characterization — is simi-
lar to that used to estimate effects of criteria pollut-
ants, but the details of each modeling step are vastly
different. The emissions and atmospheric modeling
yields estimates of changes in ultraviolet-b (UV-b)
radiation, and the exposure and dose-response analy-
ses then yield estimates of the effects of changes in
UV-b radiation, including human health, welfare,
and ecological effects. Appendix G provides a de-
tailed description of the methodology and sources
used to generate these estimates. Several of the ben-
efits can be identified but cannot yet be reliably quan-
tified, and so are described qualitatively.
The quantified physical effects estimates of sec-
tions 604 and 606 of Title VI, the provisions that
provide the primary controls on production and re-
lease of CFCs and HCFCs generate about 98 per-
cent of the monetized quantified benefits estimate.
The quantified health benefits include the follow-
ing: reduced incidences of mortality and morbidity
associated with skin cancer (melanoma and
nonmelanoma); and reduced incidences of cataracts
63
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
and then: associated pain and suffering.14 Using the
change in UV radiation dose, we estimate the num-
ber of additional cases of skin cancer (melanoma and
nonmelanoma) and cataracts. Because the baseline
levels of all of these UV-related health effects tend
to be higher for older people and for those with
lighter skins, EPA's method for projecting future
incremental skin cancers and cataracts incorporates
these factors in its benefits estimates.15 We present
a brief summary of these benefits in Table 5-5, and
the analysis is described in detail in Appendix G.
To calculate the number of deaths from mela-
noma, the model uses a dose response function simi-
lar to the C-R functions for criteria pollutants. For
nonmelanoma, the model estimates the number of
deaths by assuming that a fixed percentage of the
total nonmelanoma cases will result in death.16 We
estimate that from 1990 to 2165 sections 604 and
606 will result in 6.3 million avoided deaths from
skin cancer, 27.5 million avoided cataract cases, and
299.0 million cases of non-fatal skin cancers (mela-
noma and nonmelanoma). The unquantified effects
of sections 604 and 606 include avoided pain and
suffering from skin cancer and human health and
environmental benefits outside the United States.
Table 5-5
Major Health Benefits of Provisions to Protect Stratospheric Ozone
(CAAA Sections 604, 606, And 609)
Health Effects- Quantified
Estimate
Basis for Estimate
Melanoma and nonmelanoma
skin cancer
(fatal)
6.3 million lives saved from skin
cancer in the U.S. between 1990
and 2165
Dose-response function based on UV
exposure and demographics of
exposed populations.1
Melanoma and nonmelanoma
skin cancer
(non-fatal)
299 million avoided cases of non-
fatal skin cancers in the U.S.
between 1990 and 2165
Dose-response function based on UV
exposure and demographics of
exposed populations.1
Cataracts
27.5 million avoided cases in the
U.S. between 1990 and 2165
Dose-response function uses a
multivariate logistic risk function based
on demographic characteristics and
medical history. 1
Health Effects- Unquantified
Skin cancer: reduced pain and suffering
Reduced morbidity effects of increased UV. For example,
reduced actinic keratosis (pre-cancerous lesions resulting from excessive sun exposure)
reduced immune system suppression.
Notes:
1 For more detail see EPA's Regulatory Impact Analysis: Protection of Stratospheric Ozone (1988).
2 Note that the ecological effects, unlike the health effects, do not reflect the accelerated reduction and
phaseout schedule of section 606.
Benefits due to the section 606 methyl bromide phaseout are not included in the benefits total because
annual incidence estimates are not currently available.
4 Quantitative estimates presented in Appendix G also
include reduced crop damage associated with UV-b radiation
and tropospheric ozone; reduced damage to fish harvests associ-
ated with LJV-b radiation; and reduced polymer degradation
from UV-b radiation. The derivation of these effects is described
in more detail in Chapter 7.
13 The dose-response equation is (fractional change in inci-
dence) = (fractional change in UV-b dose + 1)° -1, where b (the
biological amplification factor) equals the percent change in in-
cidence associated with a one percent change in dose. More
iniormation about the origins oi the models can be iouiid in
Appendix G.
16 Scotio, Fears, and Fraumeni, U.S. Department oi Health
and Human Sendees, I\:IH, "Incidence of Nonmelanoma Skin
Cancer in the United States," 1981, pages 2, 7, and 13.
64
-------
Chapters: Human Health Effects of Criteria Pollutants
in the
As discussed above, and in greater detail in Ap-
pendix D, a number of important assumptions and
uncertainties in the physical effects analysis may in-
fluence the estimate of monetary benefits presented
in this study. Several of these key uncertainties, their
potential directional bias (i.e., overestimation or
underestimation), and the potential significance of
each of these uncertainties for the overall net ben-
efit results of the analysis are summarized in Table
5-6. As shown in this table, the decisions made to
overcome the problems of limited data, the inad-
equacy of the currently available scientific literature,
and other constraints do not clearly bias the overall
results of this analysis in one particular direction.
Table 5-6
Key Uncertainties Associated with Human Health Effects Modeling (continued)
Potential Source of Error
Direction of Potential
Bias for Net Benefits
Estimate
Likely Significance Relative to
Key Uncertainties in Net Benefit Estimate*
Application of C-R
relationships only to those
subpopulations matching the
original study population.
Underestimate. Potentially major. The C-R functions for several
health endpoints (including PM-related premature
mortality) were applied only to subgroups of the U.S.
population (e.g., adults over 30) and thus may
underestimate the whole population benefits of
reductions in pollutant exposures. In addition, the
demographics of the study population in the Pope et
al. study (largely white and middle class) may result
in an underestimate of PM-related mortality, because
the effects of PM tend to be significantly greater
among groups of lower socioeconomic status.
No quantification of health
effects associated with
exposure to air toxics.
Underestimate
Potentially major. According to EPA criteria, over
100 air toxics are known or suspected carcinogens,
and many air toxics are also associated with adverse
health effects such as neurotoxicity, reproductive
toxicity, and developmental toxicity. Unfortunately,
current data and methods are insufficient to develop
(and value) quantitative estimates of the health
effects of these pollutants.
Use of long-term global
warming estimates in Title VI
analysis that show more
severe warming than is now
generally anticipated.
Overestimate (for Title
VI estimate only)
Potentially major. Global warming can accelerate
the pace of stratospheric ozone recovery; if warming
is less severe than anticipated at the time the Title VI
analyses were conducted, the modeled pace of
ozone recovery may be overestimated, suggesting
benefits of the program could be delayed, perhaps
by many years. The magnitude of estimated Title VI
benefits suggests that the impact of delaying
benefits could be major.
The quantitative analysis of
Title VI (see next section)
does not account for
potential increases in
averting behavior (i.e.,
people's efforts to protect
themselves from UV-b
radiation).
Unable to determine
based on current
information.
Potentially major. Murdoch and Thayer (1990)
estimate that the cost-of-illness estimates for
nonmelanoma skin cancer cases between 2000 and
2050 may be almost twice the estimated cost of
averting behavior (application of sunscreen). Our
Title VI analysis relies on epidemiological studies,
which incorporate averting behavior as currently
practiced. Omission of future increases in averting
behavior, however, may overstate the benefits of
reduced emissions of ozone-depleting chemicals.
Benefits could be understated if individuals alter their
behaviors in ways that could increase exposure or
risk (e.g., sunbathing more frequently). A recent
European study by Autier et al. (1999) found that the
use of high sun protection factor (SPF) sun screen is
associated with increased frequency and duration of
sun exposure.
65
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 5-6
Key Uncertainties Associated with Human Health Effects Modeling (continued)
Potential Source of Error
Direction of Potential
Bias for Net Benefits
Estimate
Likely Significance Relative to
Key Uncertainties in Net Benefit Estimate*
Analysis assumes a causal
relationship between PM
exposure and premature
mortality based on strong
epidemiological evidence of
a PM/mortality association.
However, epidemiological
evidence alone cannot
establish this causal link.
Unable to determine
based on current
information.
Potentially major. A basic underpinning of this
analysis, this assumption is critical to the estimation
of health benefits. However, the assumption of
causality is suggested by the epidemiologic evidence
and is consistent with current practice in the
development of a best estimate of air pollution-
related health benefits. At this time, we can identify
no basis to support a conclusion that such an
assumption results in a known or suspected
overestimation bias.
Across-study variance /
application of regionally
derived C-R estimates to
entire U.S.
Unable to determine
based on current
information.
Potentially major. The differences in the expected
changes in health effects calculated using different
underlying studies can be large. If differences reflect
real regional variation in the PM/mortality
relationship, applying individual C-R functions
throughout the U.S. could result in considerable
uncertainty in health effect estimates.
Estimate of non-melanoma
skin cancer mortality
resulting from reductions in
stratospheric ozone is
calculated indirectly, by
assuming the mortality rate
is a fixed percentage of non-
melanoma incidence.
Unable to determine
based on current
information.
Potentially major. New data on the death rate for
non-melanoma skin cancer may significantly
influence the Title VI mortality estimate. Some
preliminary estimates suggest that this estimate may
need to be adjusted downward.
The baseline incidence
estimate of chronic bronchitis
based on Abbey et al. (1995)
excluded 47 percent of the
cases reported in that study
because those reported
"cases" experienced a
reversal of symptoms during
the study period. These
"reversals" may constitute
acute bronchitis cases that
are not included in the acute
bronchitis analysis (based on
Dockery et al., 1996).
Underestimate.
Probably minor. The relative contribution of acute
bronchitis cases to the overall benefits estimate is
small compared to other health benefits such as
avoided mortality and avoided chronic bronchitis.
CAAA fugitive dust controls
implemented in PM non-
attainment areas would
reduce lead exposures by
reducing the re-entrainment
of lead particles emitted prior
to 1990. This analysis does
not estimate these benefits.
Underestimate
Probably minor. While the health and economic
benefits of reducing lead exposure can be
substantial (e.g., see section 812 Retrospective
Study Report to Congress), most additional fugitive
dust controls implemented under the Post-CAAA
scenario (e.g., unpaved road dust suppression,
agricultural tilling controls, etc) tend to be applied in
relatively low population areas.
Exclusion of C-R functions
from short-term exposure
studies in PM mortality
calculations.
Underestimate
Probably minor. Long-term PM exposure studies
may be able to capture some of the impact of short-
term peak exposure on mortality; however the extent
of overlap between the two study types is unclear.
66
-------
Chapters: Human Health Effects of Criteria Pollutants
Table 5-6
Key Uncertainties Associated with Human Health Effects Modeling (continued)
Potential Source of Error
Direction of Potential
Bias for Net Benefits
Estimate
Likely Significance Relative to
Key Uncertainties in Net Benefit Estimate*
Age-specific C-R functions
for PM related premature
mortality not reported by
Popeetal. (1995).
Estimation of the degree of
life-shortening associated
with PM-related mortality
used a single C-R function
for all applicable age groups.
Unable to determine
based on current
information.
Unknown, possibly major when using a value of life
years approach. Varying the estimate of degree of
prematurity has no effect on the aggregate benefit
estimate when a value of statistical life approach is
used, since all incidences of premature mortality are
valued equally. Under the alternative approach
based on valuing individual life-years, the influence
of alternative values for numbers of average life-
years lost may be significant.
Assumption that PM-related
mortality occurs over a
period of five-years following
the critical PM exposure.
Analysis assumes that 25
percent of deaths occur in
year one, 25 percent in year
two, and 16.7 percent in
each of the remaining three
years.
Unable to determine
based on current
information.
Probably minor. If the analysis underestimates the
lag period, benefits will be overestimated, and vice-
versa. However, available epidemiological studies
do not provide evidence of the existence or potential
magnitude of a lag between exposure and incidence.
Thus, an underestimate of the lag seems unlikely. If
the assumed lag structure is an overestimate, even if
benefits are fully discounted from the future year of
death, application of reasonable discount rates over
this period would not significantly alter the monetized
benefit estimate.
Extrapolation of criteria
pollutant concentrations to
populations distant from
monitors.
Unable to determine
based on current
information.
Probably minor. Extrapolation method is most
accurate in areas where monitor density is high.
Monitor density tends to be highest in areas with
high criteria pollutant exposures; thus most of this
uncertainty affects low exposure areas where
benefits are likely to be low. In addition, an
enhanced extrapolation method incorporating
modeling results is used for areas far (> 50 km) from
a monitor.
Exposure analysis in areas
beyond 50 km is based on a
new technique that relies on
the direct use of air quality
modeling results in
combination with adjusted
monitor data.
Unable to determine
based on current
information.
Probably minor. The new technique is used for less
than 10 percent of the country for PM exposure, and
less than 15 percent for ozone. The approach we
use should be more accurate than the alternative
approach of linear interpolation over long distances.
The new method nonetheless requires further testing
against monitor data to access its accuracy.
Pope et al. (1995) study did
not include pollutants other
than PM.
Unable to determine
based on current
information.
Probably minor. If ozone and other criteria pollutants
correlated with PM contribute to mortality, that effect
may be captured in the PM estimate. Thus, PM is
essentially used as a surrogate for a mix of
pollutants. This uncertainty does make it difficult to
disaggregate avoided mortality benefits by pollutant,
however other studies (besides Pope) suggest that
PM is the dominant factor in premature mortality.
* The classification of each potential source of error reflects the best judgement of the section 812 Project Team. The
Project Team assigns a classification of "potentially major" if a plausible alternative assumption or approach could influence
the overall monetary benefit estimate by approximately five percent or more; if an alternative assumption or approach is
likely to change the total benefit estimate by less than five percent, the Project Team assigns a classification of "probably
67
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[Thispage left blank intentionally.]
68
-------
Economic Valuation
of Human Health
Effects
The reduced incidence of physical effects is a
valuable measure of health benefits for individual
endpomts; however, to compare or aggregate ben-
efits across endpoints, the benefits must be mon-
etized. Assigning a dollar value to avoided incidences
of each effect permits us to sum monetized benefits
realized as a result of the CAAA, and compare them
with the associated costs.
In the 812 prospective analysis, we have two
broad categories of benefits, health and welfare ben-
efits. Human health effects include mortality and
morbidity endpoints, which are presented in this
chapter. Welfare effects include agricultural and eco-
logical benefits, visibility, and worker productivity,
which are covered in the following chapter. We
obtain valuation estimates from the economic lit-
erature, and report them in "dollars per case reduced
for health effects" and "dollars per unit of avoided
damage for welfare effects".1 Similar to estimates of
physical effects provided by health studies, we re-
port each of the monetary values of benefits applied
111 this analysis in terms of a central estimate and a
probability distribution around that value. The sta-
tistical form of the probability distribution varies
by endpoint. For example, we use a Weibull distri-
bution to describe the estimated dollar value of an
avoided premature mortality, while we assume the
estimate for the value of a reduced case of acute bron-
chitis is uniformly distributed between a minimum
and maximum value.
Although human health benefits of the 1990
Amendments are attributed to reduced emissions of
criteria pollutants (Titles I through V) and reduced
emission of ozone depleting substances (Title VI),
this chapter focuses only on the valuation of human
health effects attributed to the reduction of criteria
: The literature reviews and process lor developing valua-
tion estimates are described ill detail ill Appendix I and in refer-
enced supporting reports.
pollutants. The chapter begins with an brief review
of the economic concepts behind valuing human
health effects in a cost-benefit context and a sum-
mary of the unit values applied to health endpoints.
We follow with a discussion of how we derive valu-
ation estimates for specific health effects. We then
present the results of this analysis. Wre conclude the
chapter with a review of the uncertainties associated
with benefits valuation.
Our analysis indicates that the benefit of avoided
premature mortality risk reduction dominates the
overall net benefit estimate. This is, in part, due to
the high monetary value assigned to the avoidance
of premature mortality relative to the unit value of
other health endpoints. Because of the critical im-
portance of this endpoint in the study's results, this
chapter pays particular attention to the major chal-
lenges to valuing mortality risk reductions and the
limitations of the estimate we apply in this analysis.
There are also significant reductions in short term
and chronic health effects and a substantial number
of health (and welfare) benefits that we could not
quantify or monetize.
Valuation of
Estimates
In an environmental benefit-cost analysis, the
dollar value of an environmental benefit (e.g., a
health-related improvement due to environmental
quality) enjoyed by an individual is the dollar amount
such that the person would be indifferent between
experiencing the benefit and possessing the money.
In general, the dollar amount required to compen-
sate a person for exposure to an adverse effect is
roughly the same as the dollar amount a person is
willing to pay to avoid the effect. Thus, economists
speak of "willingness-to-pay" (WTP) as the appro-
priate measure of the value of avoiding an adverse
69
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
effect. For example, the value of an avoided respira-
tory symptom would be a person's WTP to avoid
that symptom.
For most goods, WTP can be observed by ex-
amining actual market transactions. For example, if
a gallon of bottled drinking water sells for one dol-
lar, it can be observed that at least some persons are
willing to pay one dollar for such water. For goods
that are not exchanged m the market, such as most
environmental "goods," valuation is not so straight-
forward. Nevertheless, a value may be inferred from
observed behavior, such as through estimation of
the WTP for mortality risk reductions based on
observed sales and prices of products that result m
similar effects or risk reductions, (e.g., non-toxic
cleaners or bike helmets). Alternatively, surveys may
be used m an attempt to directly elicit WTP for an
environmental improvement.
Wherever possible in this analysis, we use esti-
mates of mean Wl'P. In cases where WTP estimates
are not available, we use die cost of treating or miti-
gating the effect as an alternative estimate. For ex-
ample, for the valuation of hospital admissions we
use the avoided medical costs as an estimate of the
value of avoiding the health effects causing the ad-
mission. These costs of illness (COI) estimates gen-
erally understate the true value of avoiding a health
effect. They tend to reflect the direct expenditures
related to treatment and not the utility an individual
derives from improved health status or avoided
health effect. As noted above, we use a range of
values for most environmental effects, to support
the primary central estimate of net benefits. Table
6-1 summaries the mean unit value estimates that
we use m this analysis. We present the full range of
values m Appendix H, including those used to de-
rive the primary low and primary high estimates, as
well as values used to generate an alternative value
for avoiding premature mortality.
of Mortality
Some forms of air pollution increase the prob-
ability that individuals will die prematurely. Wre use
concentration-response functions for mortality that
express the increase in mortality risk as cases of "ex-
Table 6-1
Health Effects Unit Valuation (1990 dollars)
Endpoint
Mortality
Chronic Bronchitis
Chronic Asthma
Pollutant
PM-io
PM10
03
Valuation
$4,800,000
$260,000
$25,000
(mean est.)
per case
per case
per case
Hospital Admissions
All Respiratory
All Cardiovasular
Emergency Room Visits for Asthma
SO2, NO2, PM-io & O3
SO2, NO2, & CO PM10 &
03
PM-IO&OS
$6,900
$9,500
$194
per case
per case
per case
Respiratory Illness and Symptoms
Acute Bronchitis
Asthma Attack or Moderate or
Worse Asthma Day
Acute Respiratory Symptoms
Upper Respiratory Symptoms
Lower Respiratory Symptoms
Shortness of Breath, Chest
Tightness, or Wheeze
Work Loss Days
Mild Restricted Activity Days
PM10
PM-IO&OS
SO2, NO2, PM-i, &O3
PM-i
PM-io
PM-io &SO2
PM10
PM-IO&OS
$45
$32
$18
$19
$12
$5.30
$83
$38
per case
per case
per case
per case
per case
per day
per day
per day
70
-------
Chapter 6: Economic Valuation of Human Health Effects
cess premature mortality" per time period (e.g., per
year). The benefit, however, is the avoidance of small
increases in the risk of mortality. By summing indi-
viduals' WTP to avoid small increases in risk over
enough individuals, we can infer the value of a sta-
tistical premature death avoided.2 For expository
purposes, we express this valuation as "dollars per
mortality avoided," or "value of a statistical life"
(VSL), even though the actual valuation is of small
changes in mortality risk experience by a large num-
ber of people. The economic benefits associated with
avoiding premature mortality were the largest cat-
egory of monetised benefits in the section 812 CAA
retrospective analysis (U.S. EPA 1997) and continue
to be the largest source of monetized benefits for
this prospective analysis. .Mortality benefits, how-
ever, are also the largest contributor to the range of
uncertainty in monetized benefits. For a more de-
tailed discussion of the factors affecting the valua-
tion of premature mortality sec Appendix H.
The health science literature on air pollution
indicates that several human characteristics affect die
degree to which mortality risk affects an individual.
For example, some age groups appear to be more
susceptible to air pollution than others (e.g., die eld-
erly and children). Health status prior to exposure
also affects susceptibility. At risk individuals include
those who have suffered strokes or are suffering from
cardiovascular disease and angina (Rowiatt, et al.
1998). An ideal economic benefits estimate of mor-
tality risk reduction would reflect these human char-
acteristics, in addition to an individual's willingness
to pay (WTP) to improve one's own chances of sur-
vival plus WTP to improve other individuals' sur-
vival rates.3 The ideal measure would also take into
account the specific nature of the risk reduction com-
modity that is provided to individuals, as well as the
context in which risk is reduced. To measure this
value, it is important to assess how reductions in air
pollution reduce the risk of dying from the time that
reductions take effect onward, and how individuals
2 Because people are valuing small decreases ill the risk of
premature mortality, it is expected deaths that are inferred. For
example, suppose that a given reduction in pollution confers on
each exposed individual a decrease in mortal risk of 1/100,000.
Then among 100,000 such individuals, one fewer individual can
be expected to die prematurely . If each individual's WTP for
that risk reduction is $50, then the implied value of a statistical
premature death avoided is $50 x 100,000 = $5 million.
•' For a more detailed discussion oi altruistic values related
lo the value of Hie, see Jones-Lee (1992).
value these changes. Each individual's survival curve,
or the probability of surviving beyond a given age,
should shift as a result of an environmental quality
improvement. For example, changing the current
probability of survival for an individual also shifts
future probabilities of that individual's survival. This
probability shift will differ across individuals because
survival curves are dependent on such characteris-
tics as age, health state, and the current age to which
the individual is likely to survive
Although a survival curve approach provides a
theoretically preferred method for valuing the eco-
nomic benefits of reduced risk of premature mortal-
ity associated with reducing air pollution, the ap-
proach requires a great deal of data to implement.
The economic valuation literature does not yet in-
clude good estimates of the value of this risk reduc-
tion commodity. As a result, in this study we value
avoided premature mortality risk using the value of
statistical life approach, supplemented by an alter-
native valuation based on a value of statistical life
years lost approach. We provide a review7 of the
relevant literature and a more detailed discussion of
our selected approach in Appendix H.
As in the retrospective, we use a mortality risk
valuation estimate which is based on an analysis of
26 policy-relevant value-of-life studies (see Table 6-
2). Five of the 26 studies are contingent valuation
(CV) studies, which directly solicit WTP informa-
tion from subjects; the rest are wage-risk studies,
which base WTP estimates on estimates of the addi-
tional compensation demanded in the labor market
for riskier jobs. We used the best estimate from each
of the 26 studies to construct a distribution of mor-
tality risk valuation estimates for the section 812
study. A Weibull distribution, with a mean of $4.8
million and standard deviation of S3.24 million, pro-
vided the best fit to the 26 estimates. There is con-
siderable uncertainty associated with this approach.
We discuss this issue in detail later in this chapter
and in Appendix H.
In addition, we developed alternative calculations
based on a life-years lost approach. To employ the
value of statistical life-year (VSLY) approach, we first
estimated the age distribution of those lives projected
to be saved by reducing air pollution. Based on life
expectancy tables, we calculate the life-years saved
71
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 6-2
Summary of Mortality Valuation Estimates (millions of $1990)
Type of Valuation
Study Estimate (millions 1990$)
Kneisner and Leeth (1991) (US)
Labor Market
Smith and Gilbert (1984)
Labor Market
Dillingham(1985)
Labor Market
Butler (1983)
Labor Market
Miller and Guria (1991)
Cont. Value
Moore and Viscusi (1988a)
Labor Market
Viscusi, Magat, and Huber(1991b) Cont. Value
Gegaxetal. (1985)
Cont. Value
Marin and Psacharopoulos (1982)
Labor Market
Kneisner and Leeth (1991)
(Australia)
Labor Market
Gerking, de Haan, and Schulze
(1988)
Cont. Value
Cousineau, Lacroix, and Girard
(1988)
Labor Market
Jones-Lee (1989)
Cont. Value
Dillingham(1985)
Labor Market
Viscusi (1978, 1979)
Labor Market
R.S. Smith (1976)
Labor Market
V.K. Smith (1976)
Labor Market
Olson (1981)
Labor Market
Viscusi (1981)
Labor Market
R.S. Smith (1974)
Labor Market
Moore and Viscusi (1988a)
Labor Market
Kneisner and Leeth (1991) (Japan) Labor Market
Herzog and Schlottman (1987)
Labor Market
Leigh and Folson (1984)
Labor Market
Leigh (1987)
Labor Market
Garen(1988)
Labor Market
Source: Viscusi, 1992 and EPA analysis.
0.6
0.7
0.9
1.1
1.2
2.5
2.7
3.3
2.8
3.3
3.4
3.6
3.8
3.9
4.1
4.6
4.7
from each statistical life saved within each age and
gender cohort. To value these statistical life-years,
we hypothesized a conceptual model which depicted
the relationship between the value of life and the
value of life-years. As noted in Chapter 5, the aver-
age number of life-years saved across all age groups
for which data were available is 14 for PM-related
mortality. The average for PM, in particular, differs
from the 35-year expected remaining lifespan derived
from existing wage-risk studies.4 Using the same
distribution of value of life estimates used above (i.e.
'' See, for example, Moore and Viscusi (1988) or Viscusi
the Weibull distribution with a
mean estimate of $4.8 million),
we estimated a distribution for the
value of a life-year and combined
it with the total number of esti-
mated life-years lost. The details
of these calculations are presented
in Appendix H.
Of
Chronic
The best available estimate of
WTP to avoid a case of chronic
bronchitis (CB) comes from
Viscusi et al. (1991). The Viscusi
study, however, describes to the
respondents a severe case of CB.
We employ an estimate of WTP
to avoid a pollution-related case
of CB that is based on adjusting
the WTP to avoid a severe case,
as estimated by Viscusi et al.
(1991), to account for the likeli-
hood that an average case of pol-
lution-related CB is not as severe.
Wre use the mean of a distri-
bution of WTP estimates as the
central tendency estimate of WTP
to avoid a pollution-related case
of chronic bronchitis in this
analysis. The distribution incor-
porates uncertainty from three
^^^^~^^^^~ sources: (1) the WTP to avoid a
case of severe CB, as described by
Viscusi ct al., 1991; (2) the seventy level of an aver-
age pollution-related case of CB (relative to that of
the case described by Viscusi ct al., 1991); and (3) the
elasticity of WTP with respect to severity of the ill-
ness. Based on assumptions about the distributions
of each of these three uncertain components, we
derive a distribution of WTP to avoid a pollution-
related case of CB by statistical uncertainty analysis
techniques.5 The expected value of this distribution,
3 The statistical uncertainty analysis technique we used,
which is also known as simulation modeling, is a probabilistic
approach to characterizing the uncertainty or the distribution
of potential values around a central estimate.
5.2
6.5
7.2
7.3
7.6
9.1
9.7
10.4
13.5
72
-------
Chapter 6: Economic Valuation of Human Health Effects
which is about $260,000, is taken as the central ten-
dency estimate of WTP to avoid a pollution-related
case of CB. We describe the three underlying distri-
butions, and the generation of the resulting distri-
bution of WTP, in Appendix H.
Chronic Asthma
relationship, could be comprised of just one symp-
tom or several. At the high end of the range, the
valuation represents an aggregate of WTP estimates
for several individual symptoms. The low end rep-
resents the value of avoiding a single mild symptom.
Minor Restricted Activity
The valuation of this health endpoint requires
an estimate which reflects an individual's desire to
avoid the effects of chronic asthma throughout his
or her lifetime. We derive this valuation estimate
from two studies that solicit values from individuals
diagnosed as asthmatics. Blumenschein and
Johannesson (1998) generate an estimate of monthly
WTP, while O'Conor and Blomquist (1997) gener-
ate an annual WTP estimate. In order to extend
monthly and annual WTP estimates over an
individual's lifetime, we adjusted the reported esti-
mates to reflect the average life-years remaining and
age distribution of the adult U.S. population, given
that chronic asthma is not expected to affect the av-
erage life expectancy. The mean estimate of WTP
to avoid a case of chronic asthma resulting from this
method is approximately $25,000.
In general, the values we assign to the respira-
tory-related ailments in Table 6-1 are a combination
of WTP estimates for individual symptoms compris-
ing each ailment. For example, a willingness to pay
estimate to avoid the combination of specific upper
respirator}- symptoms defined in the concentration-
response relationship measured by Pope et al. (1991)
is not available. While that study defines upper res-
piratory symptoms as one suite of ailments (runny
or stuffy nose; wet cough; and burning, aching, or
red eyes), the valuation literature reports individual
WTP estimates for three closely matching symptoms
(head/sinus congestion, cough, and eye irritation).
We therefore use these available WTP estimates and
a benefits transfer procedure to estimate the value
of avoiding those symptoms defined in the concen-
tration-response study.
To capture the uncertainty associated with the
valuation of respiratory-related ailments, we incor-
porated a range of values reflecting the fact that an
ailment, as defined in the concentration-response
An individual suffering from a single severe pol-
lution-related symptom or combination of symp-
toms may experience a Minor Restricted Activity
Day (MRAD). Krupnick and Kopp (1988) argue that
mild symptoms will not be sufficient to result in a
MRAD, so that WTP to avoid a MRAD should ex-
ceed WTP to avoid any single mild symptom. On
the other hand, WTP to avoid a MRAD should not
exceed the WTP to avoid a work loss day (which
results when the individual experiences more severe
symptoms). No studies report an estimate for WTP
to avoid a day of minor restricted activity. There-
fore, we derive for this analysis a value from WTP
estimates for avoiding combinations of symptoms
which may result in a day of minor restricted activ-
ity ($38 per day). The uncertainty range associated
with this value extends from the highest value for a
single symptom to the value for a work loss day.
Furthermore, a distributional form is used which
reflects our expectations that the actual value is likely
to be closer to the central estimate than either ex-
treme.
The valuation of this benefits category reflects
the value of reduced incidences of hospital admis-
sions due to respiratory or cardiovascular conditions.
We use avoided hospital admissions as a measure as
opposed to the number of avoided cases of respira-
tory or cardiovascular conditions, because of the
availability of C-R relationships for the hospital ad-
missions endpoint. Hospital admissions reflect a class
of health effects linked to air pollution which are
acute in nature but more severe than die symptom-
day measures discussed above.
As described in Chapter 5, our approach to esti-
mating the number of incidences for this category
involves reliance on several concentration-response
(C-R) functions. "Each concentration response func-
73
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
tion provides an alternative definition of either res-
piratory effects or cardiovascular effects, and may
be based on different pollutants. For valuation of
the incidences, the current literature provides well-
developed and detailed cost estimates of hospitaliza-
tioti by health effect or illness. Using illness-specific
estimates of avoided medical costs and avoided costs
of lost work-time, developed by Rlixhauser (1993),
we construct cost of illness (COT) estimates that are
specific to the suite of health effects defined by each
C-R function. For example, we use twelve distinct
C-R functions to quantify the expected change in
respiratory admissions.6 Consequently in this analy-
sis, we develop twelve separate COI estimates, each
reflecting the unique composition of health effects
considered in the individual studies.
The use of COI estimates suggests we likely un-
derstand the \X/TP to avoid these effects. The valu-
ation of any given health effect, such as hospitaliza-
tion, should reflect the value of avoiding associated
pain and suffering and lost leisure time, in addition
to medical costs and lost work time. While the prob-
ability distributions in this analysis characterize a
range of potential costs associated with hospitaliza-
tion, they do not account for the omission of fac-
tors from the COI estimates such as pain and suffer-
ing. Consequently, the valuations for these end-
points most likely understate the true social values
for avoiding hospital admissions due to respiratory
or cardiovascular conditions.
Provisions
We develop monetary estimates of the health
benefits due to stratospheric ozone provisions based
on estimated incidences presented in a series of ex-
isting regulatory support analyses. To ensure con-
sistency with the valuation strategy of this analysis,
however, we adjust certain parameters used in the
existing regulatory analyses of Title VI provisions.
Specifically, we re-evaluate the physical effects change
projected in the RIAs using the discount rate and
the value of statistical life adopted throughout the
rest of our present study. The net effect of these
changes is to reduce the estimates of benefits from
those found in the regulatory source support docu-
L For more detailed discussion of the various health effects
considered by each C-R function and methodology for estimat-
ing the number of avoided hospital admissions, see Appendix
D.
ments. The most important change is the discount
rate. Because the benefits of stratospheric ozone
protection accrue over several hundred years, the
discount rate chosen can have an especially large in-
fluence on the benefits estimate. The central esti-
mate employed in this analysis is five percent; the
rate used in the source documents is two percent.
The value of statistical life (VSL) estimate is also
an important factor in the calculations, because the
vast majority of benefits of stratospheric ozone pro-
tection result from avoided fatal skin cancer cases.
To reflect the uncertainty of the VSL estimates, we
employ the same statistical uncertainty aggregation
approach used in the criteria pollutant analysis, us-
ing a Weibull distribution of VSL estimates as an
input. Appendix G describes the details of these and
other changes made to ensure consistency between
our stratospheric ozone provision benefits analysis
and our criteria pollutant analyses.
Of
We combine the number of reduced incidences
of our health endpoints with our estimated values
of avoiding the health effect to generate total annual
monetized human health benefits in 2000 and 2010.
We attribute to Titles 1 through V of the CAAA
total annual human health benefits of S68 billion in
2000 and $110 billion in 2010. We summarize the
Post-CAAA 2010 monetized benefit in Table 6-3.
The table provides our central estimate, in addition
to the 5th and 95th percentile estimates for each ben-
efit category.
There are two aspects of our results that war-
rant discussion. The first is the valuation of prema-
ture mortality due to PM exposure. The second is
our strategy to avoid double-counting when aggre-
gating health benefits. As discussed in Chapter 5,
premature mortality is attributed to PM exposure
and our primary estimate reflects a lag between PM
exposure and premature mortality. While this lag
does not alter the number of estimated incidences, it
does alter the monetization of benefits. Because we
value the "event" rather than the present risk, in this
analysis we assume that the value of avoided future
premature mortality should be discounted. There-
fore, the type of lag structure employed plays a di-
rect role in the valuation of this endpoint.
74
-------
Chapter 6: Economic Valuation of Human Health Effects
Table 6-3
Results of Human Health Benefits Valuation, 2010
Monetary Benefits
(in millions 1990$)
5th %ile Mean 95th %ile
Mortality
Ages 30+
Chronic Illness
Chronic Bronchitis
Chronic Asthma
Hospitalization
All Respiratory
Total Cardiovascular
Asthma-Related ER Visits
$ 14,000
$360
40
$75
93
0.1
$ 100,000
$ 5,600
180
$ 130
390
1
$ 250,000
$18,000
300
$200
960
3
Minor Illness
Acute Bronchitis
URS
LRS
Respiratory Illness
Mod/Worse Asthma1
Asthma Attacks1
Chest tightness, Shortness of
Breath, or Wheeze
Shortness of Breath
Work Loss Days
MRAD/Any-of-19
Total Benefits in 20102
$0
4
2
1
2
20
0
0
300
680
$2
19
6
6
13
55
0.6
0.5
340
1,200
$ 110,000
Note:
1 Moderate to worse asthma and asthma attacks are endpoints included in the
definition of MRAD/Any-of-19 respiratory effects. Although valuation estimates are
presented for these categories, the values are not included in total benefits to avoid
the potential for double-counting.
2 Summing 5th and 95th percentile values would yield a misleading estimate of the
5th and 95th percentile estimate of total health benefits. For example, the likelihood
that the 5th percentile estimates for each endpoint would simultaneously be drawn
during the statistical uncertainty analysis is much less than 5 percent. As a result,
we present only the total mean.
The primary analysis reflects a five-year lag struc-
ture. Under this scenario, 50 percent of the esti-
mated cases of avoided mortality occur within the
first two years. The remaining 50 percent are then
distributed across the next three years. Our valua-
tion of avoided premature mortality applies a five
percent discount rate to the lagged estimates over
the periods 2000 to 2005 and 2010 to 2015. We dis-
count over the period between the initial PM expo-
sure change (either 2000 or 2010) and timing of the
incidence.
Many of the monetized health
benefit categories include overlapping
health endpoints, creating the poten-
tial for double-counting. In an effort
to avoid overstating the benefits, we
do not aggregate all of the quantified
health effects. For example, asthma
attacks and moderate to worse asthma
are considered components of the
endpoint, "Any of 19 Respiratory
Symptoms". Consequently, we
present the results but do not include
them 111 our reported total benefits
figures. In other cases, there are end-
points included in our aggregation of
benefit that appear to have overlap-
ping health effects. For those ben-
efit categories that describe similar
health effects, it is important to keep
in mind that estimated incidences are
based on unique portions of the popu-
lation.
Valuation
Uncertainties
We addressed many valuation
uncertainties explicitly and quantita-
tively by expressing values as distri-
butions (see Appendix II for a com-
plete description of distributions
employed), using a computerized sta-
tistical technique to apply the valua-
tions to physical effects (see Chapters
5 and 8) with the mean of each valu-
^^^^_^^^^ ation distribution providing the foun-
dation for the primary central esti-
mate of total net benefits. This approach does not,
of course, guarantee that all uncertainties have been
adequately characterized, nor that the valuation es-
timates are unbiased. It is possible that the actual
WTP to avoid an air pollution-related impact is out-
side of the range of estimates used in this analysis.
Nevertheless, we assume that the distributions em-
ployed are reasonable approximations of the ranges
of uncertainty, and that there is no compelling rea-
son to believe that the mean values employed arc
systematically biased (except for the cost of illness
values, which probably underestimate WTP). There
are, however, a limited number of health endpoints
$5
39
12
15
29
100
3
1.2
380
1,800
75
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
for which a different valuation approach may yield
results significantly different from out primary cen-
tral benefit estimate. For example, using a value of
statistical life year approach in lieu of the value of
statistical life method for valuing avoided premature
mortality yields a mean estimate for this benefit
which is approximately 45 percent lower than our
primary central estimate. For those few endpoints
where reasonable alternative valuation paradigms
yield significantly different results from our preferred
approach, see our discussion in Chapter 8.
The potential for biases as introduced by ben-
efits transfer methodology is applicable to all ben-
efits categories and, as noted in Table 6-4, the direc-
tion of its bias is unknown. Because changes in
mortality risk are the single most important compo-
nent of aggregate benefits, mortality risk valuation
is also the dominant component of the quantified
uncertainty. This category accounts for over 90
percent of total annual estimates under the Post-
CAAA scenario. The second largest benefits cat-
egory, reduced risk of chronic bronchitis, valued at
approximately $5.6 billion per year in 2010, accounts
for roughly five percent of the total estimated ben-
efits. Consequently, any uncertainty concerning
mortality risk valuation beyond that addressed by
the quantitative uncertainty assessment (i.e., that
related to the Weibull distribution with a mean value
of $4.8 million) deserves note.
Transfer
One issue that merits special attention is the
uncertainties and possible biases related to the "ben-
efits transfer" from the 26 valuation source studies
to valuation of reductions in PM-related mortality
rates. Given the limitations of the current litera-
ture, we address this source of uncertainty qualita-
tively in this section. Although each of the mortal-
ity risk valuation source studies (see Table 6-2) esti-
mate the average WTP for a given reduction in mor-
tality risk, the degree of reduction in risk being val-
ued varies across studies and is not necessarily the
same as the degree of mortality risk reduction esti-
mated 111 this analysis. The transferability of esti-
mates of the value of a statistical life from the 26
studies to the section 812 benefit analysis rests on
the assumption that, within a reasonable range, WTP
for reductions in mortality risk is linear in risk re-
duction. For example, suppose a study estimates that
the average WTP for a reduction in mortality risk
of 1/100,000 is $50, but that the actual mortality
risk reduction resulting from a given pollutant re-
duction is 1/10,000. If WTP for reductions in mor-
tality risk is linear in risk reduction, then a WTP of
S50 for a reduction of 1/100,000 implies a WTP of
S500 for a risk reduction of 1/10,000 (which is ten
times the risk reduction valued in the study). Un-
der the assumption of linearity, the estimate of the
value of a statistical life does not depend on the par-
ticular amount of risk reduction being valued. This
assumption has been shown to be reasonable pro-
vided the change in the risk being valued is within
the range of risks evaluated in the underlying stud-
ies (Rowlatt et al. 1998).
Although the particular amount of mortality risk
reduction being valued in a study may not affect the
transferabilitv of the WTP estimate from the study
Table 6-4
Valuation of CAAA Benefits: Potential Sources and Likely Direction of Bias
Likely Direction of Bias in WTP
Factor Estimates Used in this Study
Benefits Category
Premature Mortality
Age
Degree of Risk Aversion
Income
Voluntary vs. Involuntary
Catastrophic vs. Protracted Death
Discounting over a latency period
Uncertain, perhaps overestimate
Underestimate
Uncertain
Underestimate
Uncertain, perhaps underestimate
Uncertain, perhaps underestimate
Chronic Bronchitis
Severity-level Uncertain
Elasticity of WTP with respect to Uncertain
severity
All other benefit endpoints Benefits Transfer
Uncertain
76
-------
Chapter 6: Economic Valuation of Human Health Effects
to the benefit analysis, the characteristics of the study
subjects and the nature of the mortality risk being
valued in the study could be important. Certain char-
acteristics of both the population affected and the
mortality risk facing that population are believed to
affect the average WTP to reduce risk. The appro-
priateness of the mean of the WTP estimates from
the 26 studies for valuing the mortality-related ben-
efits of reductions in pollutant concentrations there-
fore depends not only on the quality of the studies
(i.e., how well they measure what they are trying to
measure), but also 011 (1) the extent to which the
subjects 111 the studies are similar to the population
affected by changes in air pollution and (2) the ex-
tent to which the risks being valued are similar.
The substantial majority of the 26 studies relied
upon are wage-risk (or labor market) studies. Com-
pared with the subjects in these wage-risk studies,
the population most affected by air pollution-related
mortality risk changes is likely to be, on average,
older and probably more risk averse. Some evidence
suggests that approximately 85 percent of those iden-
tified in short-term ("episodic") studies who die pre-
maturely from PM-related causes are over 65.7 The
average age of subjects in wage-risk studies, in con-
trast, would be well under 65, and probably closer
to 40 years of age.
The direction of bias resulting from the age dif-
ference is unclear. We could argue that, because an
older person has fewer expected years left to lose,
his or her WTP to reduce mortality risk would be
less than that of a younger person. This hypothesis
is supported by one empirical study, Jones-Lee et al.
(1985), which found WTP to avoid mortality risk at
age 65 to be about 90 percent of what it is at age 40.
On the other hand, there is reason to believe that
those over 65 are, in general, more risk averse than
the general population. This would imply that older
populations are likely to select occupations that are
relatively less risky than workers represented in
wage-risk studies or the general population. Al-
though the list of 26 studies used here excludes stud-
ies that consider only much-higher-than-average oc-
cupational risks, there is nevertheless likely to be
some selection bias in the remaining studies, because
these studies are likely to be based on samples of
7 See Schwartz and Dockery (1992), Ostro et al. (1995),
and Chestnut (1995).
\vorkers who are, on average, more risk-loving than
the general population. In contrast, older people as
a group exhibit more risk-averse behavior.
There is substantial evidence that the income
elasticity of WTP for health risk reductions is posi-
tive (although there is uncertainty about the exact
value of this elasticity). This implies that individu-
als with higher incomes and/or greater wealth should
be willing to pay more to reduce risk, all else equal,
than individuals with lower incomes or wealth. The
comparison between the income, both actual and
potential, or wealth of the workers in the wage-risk
studies versus that of the population of individuals
most likely to be affected by changes in pollution
concentrations, however, is unclear. One could ar-
gue that because the elderly are relatively wealthy,
the affected population is also wealthier, 011 aver-
age, than are the wage-risk study subjects, who tend
to be middle-aged (on average) blue-collar workers.
On the other hand, the wforkcrs in the wage-risk
studies \vill have potentially more years remaining
in which to acquire streams of income from future
earnings. On net, the potential income comparison
is unclear.
Although there may be several ways in which
job-related mortality risks differ from air pollution-
related mortality risks, the most important differ-
ence may be that job-related risks are incurred vol-
untarily, or generally assumed to be, whereas air
pollution-related risks are incurred involuntarily.
There is some evidence8 that people will pay more
to reduce involuntarily incurred risks than risks in-
curred voluntarily. If this is the case, WTP estimates
based on wage-risk studies may understate WTP to
reduce involuntarily incurred air pollution-related
mortality risks.
Another important difference related to the na-
ture of the risk may be that some workplace mortal-
ity risks tend to involve sudden, catastrophic events,
whereas air pollution-related risks tend to involve
longer periods of disease and suffering prior to death.
Some evidence suggests that WTP to avoid a risk of
a protracted death involving prolonged suffering and
loss of dignity and personal control is greater than
the WTP to avoid a risk (of identical magnitude) of
sudden death. To the extent that the mortality risks
8 See, lot example, Violette and Chestnut, 1983.
77
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
addressed in this assessment are associated with longer
periods of illness or greater pain and suffering than
are the risks addressed in the valuation literature,
the WTP measurements employed in the present
analysis would reflect a downward bias.
Economic assessment of WTP for lagged mor-
tality effects also introduces uncertainty. For lack
of a more refined technique, our analysis relies on
the simplifying assumption that lagged mortality
risks can be valued at the tune of the occurrence of
death, rather than at the time of exposure. In subse-
quent development of the annual and present value
estimates, we therefore discount the dollar benefits
estimate as if the full benefit accrues only in the year
of death. There are several reasons to believe that
this approach underestimates willingness to pay.
Most importantly, while death may occur after a lag
period, morbidity effects may appear at any time
prior to death, including immediately upon expo-
sure. It is not clear that other dose-response assess-
ments capture the full range of morbidity effects,
direct and indirect, that might be associated with a
latent fatal exposure. Other potentially important
factors include the use of a financial discount rate,
which may or may not accurately represent the rate
at which individuals might discount delayed health
benefits and the effect of knowledge of a fatal expo-
sure 011 valuation of a delayed effect, in other words
whether the valuation is affected by a prior diagno-
sis of a fatal condition.
We summarize the potential sources of bias in-
troduced by relying on wage-risk studies to derive
an estimate of the WTP to reduce air pollution-re-
lated mortality risk in Table 6-4; the overall effect of
these multiple biases is addressed in Table 6-5.
Among these potential biases, it is disparities in age
and income between the subjects of the wage-risk
studies and those affected by air pollution which have
thus far motivated specific suggestions for quantita-
tive adjustment;9 however, the appropriateness and
the proper magnitude of such potential adjustments
remain unclear given presently available information.
These uncertainties are particularly acute given the
possibility that age and income biases might offset
each other m the case of pollution-related mortality
risk aversion. Furthermore, the other potential bi-
ases discussed above, and summarized in Table 6-4,
add additional uncertainty regarding the transferabil-
ity of WTP estimates from wage-risk studies to en-
vironmental policy and program assessments.
Chestnut, 1995; lEc, 1992.
78
-------
Chapter 6: Economic Valuation of Human Health Effects
Table 6-5
Key Uncertainties Associated with Valuation of Health Benefits
Direction of Potential
Potential Source of Error Bias for Net Benefits
Likely Significance Relative to Key
Uncertainties on Net Benefits Estimate1
Benefits transfer for mortality
risk valuation, including
differences in age, income,
degree of risk aversion, the
nature of the risk, and
treatment of latency between
mortality risks presented by PM
and the risks evaluated in the
available economic studies.
Unable to determine based
on currently available
information
Potentially major. The mortality valuation
step is clearly a critical element in the net
benefits estimate, so any uncertainties can
have a large effect. As discussed in the text,
however, information on the combined effect
of these known biases is relatively sparse,
and it is therefore difficult to assess the
overall effect of multiple biases that work in
opposite directions.
Benefits transfer for chronic
bronchitis, including
adjustments made to better
match the severity of the risks
modeled in the available
economic studies.
Unable to determine based
on currently available
information
Probably minor. Benefits of avoided chronic
bronchitis account for about five percent of
total benefits, limiting the effect on net
benefits to a maximum of about seven
percent. Steps taken in the study to adjust
for severity using the best available empirical
information likely limit the effect to much less
than this maximum value.
Inability to value some
quantifiable morbidity
endpoints, such as impaired
lung function.
Underestimate
Probably minor. Reductions in lung function
are a well-established effect, based on
clinical evaluations of the impact of air
pollutants on human health, and the effect
would be pervasive, affecting virtually every
exposed individual. There is therefore a
potential for a major impact on benefits
estimates. The lack of a clear symptomatic
presentation of the effect, however, could
limit individual WTP to avoid lung function
decrements.
Note: 1 The classification of each potential source of error reflects the best judgement of the section 812 Project Team. The
Project Team assigns a classification of "potentially major" if a plausible alternative assumption or approach could influence
the overall monetary benefit estimate by approximately five percent or more; if an alternative assumption or approach is likely
to change the total benefit estimate by less than five percent, the Project Team assigns a classification of "probably minor."
79
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[Thispage left blank intentionally.]
80
-------
Ecological and
Other Welfare
Effects
EPA's traditional focus in environmental ben-
efits assessment has been on quantifying beneficial
impacts of environmental regulation on human
health. As we have learned more about the effects
of anthropogenic stressors on ecological systems,
however, pursuit of environmental programs tar-
geted 011 reductions of damage to the environment
have become more common. The CAAA Title IV
provisions, collectively referred to as the Acid Ram
Program, are a good example. These provisions are
in place largely as the result of a major research ef-
fort to better understand and quantify the effects of
sulfur and nitrogen oxides on natural systems sus-
ceptible to acid ram. Although the benefits of this
program include improvements in human health, the
initial impetus was protection of ecological resources.
We have designed this first section 812 prospec-
tive analysis to be responsive to the increased focus
on the importance of ecological resources by devot-
ing a great deal of effort to characterizing and, where
possible, quantifying and monetizing the impacts of
air pollutants on natural systems. This increased fo-
cus is also partly a result of the outcome of EPA's
retrospective analysis, in which we identified an in-
creased understanding of and focus on ecological ef-
fects as one of the important research directions for
the first prospective and subsequent analyses. This
chapter presents the results of these efforts.
This chapter consists of four sections. First, we
provide an overview of our approach to estimating
the effects of air pollution on ecological systems.
Second, we provide a characterization of these ef-
fects in qualitative terms. The second section con-
cludes with a summary of the process for selecting
specific impacts which can be quantified and mon-
etized using currently available methods. Third, we
present the results of our quantitative and economic
analyses. Finally, we discuss major uncertainties of
the ecological and other welfare effects analyses.
Of
Our analysis of ecological effects involves three
major steps:
* First, we identify and characterize ecologi-
cal effects from air pollution.
* Second, we develop and implement selection
criteria for more in-depth assessment of eco-
logical impacts.
• Third, we perform quantitative and qualita-
tive analyses to characterize a portion of the
benefits of the 1990 CAAA provisions.
The first step involves taking a broad view of
pollutants controlled under the CAAA and their
documented effects on ecological systems, both as
individual pollutants and, to the extent possible, as
one component in multiplc-strcssor effects on eco-
systems and their components. We organize our
analysis in terms of major pollutant classes and by
the level of biological organization at which impacts
are measured (e.g., regional ecosystem, local ecosys-
tem, community, population, individual, etc.).
After completing the first step on a broad level,
the second step involves narrowing the scope of sub-
sequent analyses. While it is desirable to focus ef-
fort 011 those impacts that are of greatest importance,
in practice the state of the science in ecological as-
sessment largely dictates the subsequent focus of the
analysis. There exist only a handful of comprehen-
sive ecological assessments from which to draw con-
clusions about those effects that are most important
either ecologically or in economic terms, and those
studies are potentially controversial in their meth-
ods and conclusions, in part because of the incom-
plete understanding of many of these effects. As a
result, the categories of effects ultimately chosen for
assessment here are necessarily limited by available
81
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
methods and data. As scientific understanding and
impact assessment methods grow more comprehen-
sive, however, we expect that the focus of subsequent
analyses will be on those effects whose avoidance
would have the greatest potential ecological and/or
economic value.
The third step involves implementing a wide
range of analyses to more exhaustively characterize
specific effects of air pollution on ecological systems.
We provide quantitative estimates of the benefits of
the 1990 CAAA for the following effects:
* eutrophication of estuaries associated with
airborne nitrogen deposition;
* acidification of freshwater bodies associated
with airborne nitrogen and sulfur deposition;
and
* reduced forest growth associated with ozone
exposure.
In addition, in tins chapter we present the meth-
ods and results for quantitative analysis of other
welfare effects, including reduced agricultural yields
associated with ozone exposure, the impact of am-
bient participate matter on visibility, the effects of
ozone on farm worker productivity, and the effects
of stratospheric ozone on crop and fisheries yields.
These effects have been identified as important cat-
egories of benefits in many previous analyses, includ-
ing the section 812 retrospective analysis. As a re-
sult, these effects were not considered in the same
three step process used for other service flows.
We attempted to conduct quantitative analyses
of two other benefits categories: the accumulation
of toxics in freshwater fisheries associated with air-
borne toxics deposition; and aesthetic degradation
of forests associated with ozone and airborne toxics
exposure. However, we found that, while some
quantitative methods exist to evaluate these benefits,
key links are missing in the analytic process. This in
turn prevents development of defensible benefits es-
timates which can be reasonably associated with the
air quality and air pollutant deposition patterns de-
veloped from our Post-CAAA and Pre-CAAA sce-
narios. See Appendix E for more detailed discus-
sion of these service flows. In addition, in assessing
nitrogen deposition impacts to estuarine systems, we
relied on a displaced cost approach with results that
we chose to omit from the primary benefits estimate
because of uncertainties in the methodology. These
results are nonetheless reported in this chapter, but
are used for the purposes of sensitivity testing only.
Because the breadth and complexity of air pol-
lutant-ecosystem interactions do not allow for com-
prehensive quantitative analysis of all the ecological
benefits of the CAAA, we stress the importance of
continued consideration of those impacts not val-
ued in this report in policy decision-making and in
further technical research. Judging from the geo-
graphic breadth and magnitude of the relatively
modest subset of impacts that we find sufficiently
well-understood to quantify and monetize, it is ap-
parent that the economic benefits of the CAAA's
reduction of air pollution impacts on ecosystems are
substantial.
Of
of Air Pollution on
Systems
The purpose of this section is to provide an over-
view of potential interactions between air pollutants
and the natural environment. We identify major
single pollutant-environment interactions, as well as
the synergistic impacts of ecosystem exposure to
multiple air pollutants. Although a wride variety of
complex air pollution-environment interactions are
described or hypothesized in the literature, for the
purposes of this analysis we focus on major aspects
of ecosystem-pollutant interactions. Wre do this by
limiting our review according to the following crite-
ria:
Pollutants regulated by the CAAA.
• Known interactions between pollutants and
natural systems as documented in
peer-reviewed literature.
* Pollutants present in the atmosphere in suf-
ficient amounts after 1990 to cause signifi-
cant damages to natural systems.
Our understanding of air pollution effects on
ecosystems has progressed considerably during re-
cent decades. Previously, air pollution was regarded
primarily as a local phenomenon and concern was
associated with the vicinity of industrial facilities,
82
-------
Chapter 7: Ecological and Other Welfare Effects
power plants or urban areas. The pollutants of con-
cern were gaseous (e.g., sulfur dioxide and ozone) or
heavy metals (e.g., lead) and the observed effects were
visible stress- specific symptoms of injury (e.g., fo-
liar chlorosis). The most typical approach to docu-
menting the effects of specific pollutants was a
dose-response experiment, where the objective was
to develop a regression equation describing the rela-
tionship between exposure and some easily measured
effect (e.g., growth, yield or mortality). As analytic
methods unproved and ecology progressed, a broader
range of effects of air pollutants was identified and
understanding of the mechanisms of effect improved.
Observations made on various temporal scales (e.g.,
long-term studies) and spatial scales (e.g., watershed
studies) led to the recognition that air pollution can
affect all organizational levels of biological systems.
Our current understanding of ecosystem impacts
can be organized by the pollutants of concern and
by the level of biological organization at which im-
pacts are directly measured. We attempt to address
both dimensions of categorization in this overview.
In Table 7-1 we summarize the major pollutants of
concern, and the documented acute and long-term
ecological impacts associated with them.
The summary in Table 7-1 is a highly condensed
version of the results of our characterization of eco-
logical impacts. In addition to the pollutant-specific
effects outlined in the table, it is important to iden-
tify the level of biological organization and types of
ecosystems that are susceptible to these types of ef-
fects. Tables 7-2 through 7-4 provide more detail on
pollutant-specific impacts at a range of levels of bio-
logical organization. It is important to note that the
interactions listed are intended to illustrate the range
of possible adverse effects. For a more complete re-
view of air-pollutant-induced effects on ecosystems,
see Appendix R.
of Mercury
Table 7-2 summarizes the effects of mercury and
ozone on ecological systems. To illustrate the na-
ture of our review of effects, consider the second
row in Table 7-2. This row summarizes the effects
of the air pollutants mercury and ozone at the "indi-
vidual" level of biological organization. As indicated
in die table, in a general sense air pollutants can in-
duce a direct physiological response in individuals
(analogous to that experienced by humans exposed
to pollutants), or an indirect effect either through
impacts on the individual's surroundings or by weak-
ening the individual and making it more susceptible
to other stressors. "Mercury has several direct effects
to fauna, including effects to the central nervous
system and the liver, while the documented direct
effects of ozone tend to be to a variety of plant func-
tions. Indirect effects of mercury are not well un-
derstood, but the indirect effects of ozone may serve
to compound the direct effects to plants by also
making the plants more susceptible to drought or
heat stress, for example. This type of cataloging of
Table 7-1
Classes of Pollutants and Ecological Effects
Pollutant
Class
Acidic
Deposition
Nitrogen
Deposition
Hazardous Air
Pollutants
(HAPs)
Ozone
Major Pollutants and
Precursor Emissions
Sulfuric acid, nitric acid
Precursor emissions: Sulfur
dioxide, nitrogen oxides
Nitrogen compounds (e.g.,
nitrogen oxides)
Mercury, dioxins
Tropospheric ozone
Precursor emissions: Nitrogen
Oxides and Volatile Organic
Compounds (VOCs)
Acute Effects
Direct toxic effects to
plant leaves and
aquatic organisms.
Direct toxic effects to
animals.
Direct toxic effects to
plant leaves.
Long-term Effects
Progressive deterioration of soil
quality. Chronic acidification of
surface waters.
Saturation of terrestrial ecosystems
with nitrogen. Progressive nitrogen
enrichment of coastal estuaries.
Conservation of mercury and dioxins
in biogeochemical cycles and
accumulation in the food chain.
Alterations of ecosystem wide
patterns of energy flow and nutrient
cycling.
83
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 7-2
Interactions of Mercury and Ozone with Natural Systems At Various Levels of Organization
Examples of Interactions
Spatial Scale
Molecular and
cellular
Individual
Population
Community
Local Ecosystem
(e.g. .landscape
element)
Regional Ecosystem
(e.g., watershed)
Type of Interaction
Chemical and
biochemical processes
Direct physiological
response.
Indirect effects:
Response to altered
environmental factors or
alterations of the
individual's ability to cope
with other kinds of stress.
Change of population
characteristics like
productivity or mortality
rates.
Changes of community
structure and competitive
patterns
Changes in nutrient
cycle, hydrological cycle,
and energy flow of lakes,
wetlands, forests,
grasslands, etc.
Biogeochemical cycles
within a watershed.
Region-wide alterations
of biodiversity.
Mercury in
streams and lakes
Mercury enters the body of
vertebrates and binds to
sulfhydril groups (i.e.
proteins).
Neurological effects in
vertebrates. Behavioral
abnormalities. Damages to
the liver.
Few interactions known.
Damages through
increased sensitivity to
other environmental stress
factors could occur, for
example, through
impairment of immune
response.
Reduced reproductive
success offish and bird
species. Increased
mortality rates, especially in
earlier life stages.
Loss of species diversity of
benthic invertebrates.
Not well understood.
Not well understood.
Ozone
Oxidation of enzymes of plants.
Disruption of the membrane
potential.
Direct injuries include visible
foliar damage, premature needle
senescence, reduced
photosynthesis, altered carbon
allocation, and reduction of
growth rates and reproductive
success.
Increased sensitivity to biotic
and abiotic stress factors like
pathogens and frost. Disruption
of plant-symbiont relationship
(mychorrhiza), and symbionts.
Reduced biological productivity.
Selection for less sensitive
individuals. Possibly
microevolution for ozone
resistance.
Alteration of competitive
patterns. Selective advantage
for ozone-resistant species.
Loss of ozone sensitive species
and individuals. Reduction in
productivity.
Alterations of ecosystem-wide
patterns of energy flow and
nutrient cycling.
Region-wide loss of sensitive
species.
effects, while limited in its direct usefulness in a cost-
benefit framework, nonetheless does convey the wide
range of documented effects of air pollutants on eco-
logical resources. These tables and the accompany-
ing text, found in Appendix E, also provide a frame-
work for determining the extent to which impor-
tant factors may not be well characterized by quan-
titative analysis, setting the stage for prioritization
of research needs.
Table 7-3 provides a summary of the effects of
nitrogen deposition 011 natural systems. These im-
pacts are manifest in both terrestial and coastal es-
tuarine systems. In both types of systems, nitrogen
can be a growth-enhancing nutrient. As shown in
the rows characterizing individual and population
level impacts, the effects on many varieties of plants
are beneficial. This growth can have other harmful
effects, however. For example, excessive growth of
84
-------
Chapter 7: Ecological and Other Welfare Effects
Table 7-3
Interactions Between Nitrogen Deposition and Natural Systems
At Various Levels of Organization
Examples of Interactions
Eutrophication and
Nitrogen Saturation of
Eutrophication of Coastal
Spatial Scale
Molecular and
cellular
Individual
Population
Community
Local Ecosystem
(e.g., landscape
element)
Regional Ecosystem
(e.g., watershed)
Type of Interaction
Chemical and
biochemical processes
Direct physiological
response.
Indirect effects:
Response to altered
environmental factors or
alterations of the
individual's ability to
cope with other kinds of
stress.
Change of population
characteristics like
productivity or mortality
rates.
Changes of community
structure and
competitive patterns
Changes in nutrient
cycle, hydrological cycle,
and energy flow of lakes,
wetlands, forests,
grasslands, etc.
Biogeochemical cycles
within a watershed.
Region-wide alterations
of biodiversity.
Terrestrial Landscapes
Assimilation of nitrogen by
plants and microorganisms
Increases in leaf- size of
terrestrial plants.
Decreased resistance to
biotic and abiotic stress
factors like pathogens and
frost. Disruption of plant-
symbiont relationships with
mycorrhiza fungi.
Increase in biological
productivity and growth
rates of some species.
Alteration of competitive
patterns. Selective
advantage for fast growing
species and individuals
that efficiently use
additional nitrogen. Loss
of species adapted to
nitrogen-poor
environments.
Magnification of the
biogeochemical nitrogen
cycle. Progressive
saturation of
microorganisms, soils, and
plants with nitrogen.
Leaching of nitrogen from
terrestrial sites to streams
and lakes. Acidification of
aquatic bodies.
Eutrophication of estuaries.
Estuaries
Assimilation of nitrogen by
plants and microorganisms.
Increase in growth of marine
plants.
Injuries to marine fauna through
oxygen depletion of the
environment. Loss of physical
habitat due to loss of sea-grass
beds. Injury through increased
shading. Toxic blooms of
plankton.
Increase in biological
productivity. Increase of growth
rates (esp. of algae and marine
plants).
Excessive algal growth.
Changes in species
composition. Decrease in sea-
grass beds.
Magnification of the nitrogen
cycle. Depletion of oxygen,
increased shading through
algal growth.
Additional input of nitrogen
from nitrogen-saturated
terrestrial sites within the
watershed.
marine organisms can lead to eutrophy, a state where
the enhanced surface growth of plants shields bot-
tom growing plants from sunlight, causing them to
die and, in extreme cases, lead to low dissolved oxy-
gen, or anoxic, conditions that impair a wide range
of species and ecological functions. These effects
are described in the table in the rows characterizing
effects at the community and ecosystem levels. For
this reason, isolated analysis of the effects of nitro-
gen on individuals or populations may provide mis-
leading results; by the same token, analyses which
ignore the beneficial effects of nitrogen in certain
types of systems may lead to similarly misleading-
results. These complex linkages across biological
levels of organization suggest that, when feasible, a
systems level approach to ecological assessments is
preferable to isolated analyses of effects at lower or-
ders of organization.
of Acid
Table 7-4 provides a summary of the effects of
acid deposition on forest and freshwater systems.
The direct effects of acid deposition in lakes and
streams include effects on fish species, as charaterized
in the row describing individual-level effects. These
85
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 7-4
Interactions Between Acid Deposition and Natural Systems At Various Levels of Organization
Examples of Interactions
Acidification of Streams
Spatial Scale
Molecular and
cellular
Individual
Population
Community
Local Ecosystem
(e.g., landscape
element)
Regional Ecosystem
(e.g., watershed)
Type of Interaction
Chemical and
biochemical processes
Direct physiological
response
Indirect effects:
Response to altered
environmental factors or
alterations of the
individual's ability to
cope with other kinds of
stress.
Change of population
characteristics like
productivity or mortality
rates.
Changes of community
structure and
competitive patterns
Changes in nutrient
cycle, hydrological cycle,
and energy flow of lakes,
wetlands, forests,
grasslands, etc.
Biogeochemical cycles
within a watershed.
Region-wide alterations
of biodiversity.
Acidification of Forests
Damages to epidermal
layers and cells of plants
through deposition of
acids.
Increased loss of nutrients
via foliar leaching.
Cation depletion in the soil
causes nutrient
deficiencies in plants.
Concentrations of
aluminum ions in soils can
reach phytotoxic levels.
Increased sensitivity to
other stress factors like
pathogens and frost.
Decrease of biological
productivity of sensitive
organisms. Selection for
less sensitive individuals.
Microevolution of
resistance.
Alteration of competitive
patterns. Selective
advantage for acid-
resistant species. Loss of
acid sensitive species and
individuals. Decrease in
productivity. Decrease of
species richness and
diversity.
Progressive depletion of
nutrient cations in the soil.
Increase in the
concentration of mobile
aluminum ions in the soil.
Leaching of sulfate, nitrate,
aluminum, and calcium to
streams and lakes.
Acidification of aquatic
bodies.
and Lakes
Impairment of ion interactions
offish at the cellular level.
Decreases in pH and increase
in aluminum ions causes
pathological changes in gill
structure offish.
Aluminum ions in the water
column can be toxic to many
aquatic organisms through
impairment of gill regulation.
Acidification can indirectly
affect submerged plant species,
because it reduces the
availability of dissolved carbon
dioxide (CO2).
Decrease of biological
productivity of sensitive
organisms. Selection for less
sensitive individuals.
Microevolution of resistance.
Alteration of competitive
patterns. Selective advantage
for acid-resistant species. Loss
of acid sensitive species and
individuals. Decrease in
productivity. Decrease of
species richness and diversity.
Measurable declines of
decomposition of some forms
of organic matter, potentially
resulting in decreased rates of
nutrient cycling.
Additional acidification of
aquatic systems through
processes in terrestrial sites
within the watershed.
effects are not as straightforward as they might ap-
pear, however, because it is not only the acidity (pH)
of the water itself that causes the effect but the in-
creased leaching of metals, particularly aluminum,
which takes place in acidic (low pH) environments
that contributes substantially to the effects on fish.
These effects will vary widely from place to place
according to the mineral content of the soil near the
lake and the lakebed sediment, as well as the natural
resistance of the lake in absorbing acid deposition
(i.e., its buffering capacity). Other important effects
characterized in the table include the ability of acid
deposition to deplete cation concentrations in ter-
restrial ecosystems; increase the concentration of
aluminum in soils; and leach nutrients, sulfates, and
metals to surrounding streams and lakes. Effects of
note at the individual level include foliar damage to
trees.
86
-------
Chapter 7: Ecological and Other Welfare Effects
A few general points emerge from our review of
ecological effects:
• Air pollutants have indirect effects that are
at least as important as direct toxic effects
on living organisms. Indirect effects include
those in which the pollutant alters the physi-
cal or chemical environment (e.g., soil prop-
erties), the plant's ability to compete for lim-
ited resources (e.g., water, light), or the
plant's ability to withstand pests or patho-
gens. Examples are excessive availability of
nitrogen, depletion of nutrient cations 111 the
soil by acid deposition, mobilization of toxic
elements such as aluminum, and changes in
winter hardiness. As is tine for other com-
plex interactions, indirect effects are more
difficult to observe than direct toxic relation-
ships between air pollutants and biota, and
there may be a variety of interactions that
have not yet been detected.
* There is a group of pollutants that tend to
be conserved in the landscape after they have
been deposited to ecosystems. These con-
served pollutants are transformed through
biotic and abiotic processes within ecosys-
tems, and accumulate in biogeochemical
cycles. These pollutants include, but are not
limited to, hydrogen ions (H+), sulfur (S)
and nitrogen (N) containing substances, and
mercury (Ilg). Chronic deposition of these
pollutants can result in progressive increases
in concentrations and cause injuries due to
cumulative effects. Indirect, cumulative
damages caused by chronic exposure (i.e.,
long-term, moderate concentrations) to these
pollutants may increase in magnitude over
time frames of decades or centuries with very
subtle annual increments of change. Ex-
amples are N-saturation of terrestrial ecosys-
tems, cation depletion of terrestrial ecosys-
tems, acidification of streams and lakes, and
accumulation of mercury in aquatic food
webs.
• Damages to ecosystems are most likely
caused by a combination of environmental
stress factors. These include anthropogenic
factors such as air pollution and other envi-
ronmental stress factors such as low tempera-
ture, excess or limited water, and limited
availability of nutrients. The specific com-
binations of factors differ among regions and
ecosystems where declines have been ob-
served. Accurately predicting the impacts
of multiple stress factors is an extremely dif-
ficult task, but this is an area of very active
research among ecologists.
• Pollutant-environment interactions are com-
plicated by the fact that biotic and abiotic
factors in ecosystems change dramatically
over time. Besides oscillations on a daily-
basis, and changes in a seasonal rhythm, there
are long-range successional developments
over time periods of years, decades, or even
centuries. These temporal variations occur
111 polluted and pristine ecosystems, and no
single point in time or space can be defined
as representative of the entire system.
to
Based on this broad overview of effects, we iden-
tify a set of pollutant-environment interactions
which are amenable to more detailed quantification
and monetization. We evaluate the long list of ef-
fects and seek categories where a defensible link ex-
ists between changes in air pollution emissions and
the quality or quantity of the ecological service flow,
and where economic models are available to mon-
etize these changes. The use of these criteria greatly
constrains the range of impacts that can be treated
quantitatively. While the previous section identi-
fies many pollutant-ecosystem interactions, only a
handful are understood and have been modeled to
an extent sufficient to reliably quantify their impact.
The theoretical basis of economic benefits as-
sessment is that ecosystems provide services to man-
kind, and that those services have economic value.
The application of this theory requires the isolation
of service flows that have market values or are oth-
erwise amenable to available methods for determin-
ing value in the absence of formal markets. Avail-
able methods do not exist to comprehensively value
all service flows for any particular ecosystem or ag-
gregation of ecosystems. Generally, we are limited
87
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
to those service flows that are either sources of ma-
terial inputs or associated with natural amenities that
involve active recreation. Impacts to these service
flows that can be valued tend to manifest themselves
immediately and can be readily measured and assessed
in terms of the established cause and effect relation-
ships.
Based on the constraints of economic valuation
methods and data, we select from the host of ecosys-
tem impacts identified in the previous section a set
of service flows as candidate endpoints for analysis.
The list of service flows establishes the potential
scope of economic analysis for ecological effects fea-
sible in the context of the present study. Table 7-5
presents the service flow impacts that we quantita-
tively estimate in this analysis plus those effects that
currently cannot be quantified for each of the four
ecological pollutant categories discussed in Table 7-1.
From the list of effects in Table 7-5, we further
limited the quantitative and qualitative analyses con-
ducted to reflect the available model coverage. The
results are summarized in Table 7-6. The relatively
short list of effects in Tables 7-5 and 7-6 demonstrates
that, of the great number of known impacts of air
pollution, only a subset can be assessed quantita-
tively. Note that for one category of effects, nitro-
gen deposition impacts to estuarine systems, we re-
lied on a displaced cost approach (described below)
Table 7-5
Ecological Effects of Air Pollutants
Pollutant
Quantified Effects
Unquantified Effects
Acidic Deposition
Impacts to recreational
freshwater fishing
Impacts to commercial forests
(e.g., timber, non-timber forest products)
Impacts to commercial freshwater fishing
Watershed damages (water filtration
flood control)
Impacts to recreation in terrestrial
ecosystems (e.g. forest aesthetics,
nature study)
Reduced existence value and option
values for nonacidified ecosystems (e.g.
biodiversity values)
Nitrogen
Deposition
Additional costs of alternative or
displaced nitrogen input controls
for eastern estuaries
Impacts to commercial fishing,
agriculture, and forests
Watershed damages (water filteration,
flood control)
Impacts to recreation in estuarine
ecosystems (e.g. Recreational fishing,
aesthetics, nature study)
Reduced existence value and option
values for non-eutrophied ecosystems
(e.g. biodiversity values)
Tropospheric
Ozone Exposure
Reduced commercial timber
yields and reduced tons of carbon
sequestered
Impacts to recreation in terrestrial
ecosystems (e.g. forest aesthetics,
nature study)
Reduced existence value and option
values for ozone-impacted ecosystems
Hazardous Air
Pollutant (HAPS)
Deposition
No service flows quantified
Impacts to commercial and recreational
fishing from toxification of fisheries
Reduced existence value and option
values for non-toxified ecosystems (e.g.
biodiversity values)
-------
Chapter 7: Ecological and Other Welfare Effects
Table 7-6
Summary of Endpoints Selected for Quantitative Analysis
Endpoint
Analysis
Geographic Scope
Lake acidification impacts on
recreational fishing
Quantification of improved fishing
with monetization of recreational
value
Case study of New York State
Estuarine eutrophication
impacts on recreational and
commercial fishing
Quantification of improved fishing
with monetization of displaced
costs of alternative
eutrophication control methods
Case studies of Chesapeake Bay,
Long Island Sound, and Tampa
Bay (with illustrative extensions to
East Coast estuaries)
Ozone impacts on commercial
timber sales
Quantification of improved timber
growth with monetization of
commercial timber revenues
National assessment
Ozone impacts on carbon
sequestration in commercial
timber
Quantification of improved
carbon sequestration
National assessment
that we chose to omit from the primary benefits es-
timate because of uncertainties in the methodology.
These results are nonetheless reported in this chap-
ter, but are used for the purposes of sensitivity test-
ing only. In the next section we discuss die meth-
ods, results, and caveats of the analyses of these se-
lected endpoints.
Results
In this section we summarize the methods used
for, and results obtained from, our quantitative and
economic analyses of selected service flows. We first
review the methods for each analysis, and then
present a summary of key quantitative results. For
a more detailed description of methods and results,
see Appendix E.
Eutrophication
Atmospherically derived nitrogen makes up a
sizable fraction of total nitrogen inputs in estuaries
in the eastern United States. Airborne nitrogen depo-
sition accounts for a significant fraction of the total
nitrogen loads to coastal estuaries, particularly on
the East and Gulf coasts. For example, the most
recent estimates for the Chesapeake Bay indicate air-
borne deposition accounts for over 40 percent of the
total nitrogen load to the estuary; in Galveston Bay,
the share is almost 50 percent. When nitrogen en-
ters estuaries it can cause eutrophication, or an in-
creased nutrient load that, in excess, changes the
ecosystem's structure and function and affects eco-
logical service flows. Many state governments and
multi-state regional authorities have expressed in-
creasing concern about the control of airborne ni-
trogen deposition as an important source of nitro-
gen loading.
Our analysis of the effects of nitrogen deposi-
tion followed two tracks. We first attempted to quan-
tify the service flows affected by and the damages
associated with eutrophication, and derive dose-re -
spoiisc relationships and valuation strategics for each
of the key service flow categories (for example, rec-
reational fishing). The derivation of dose-response
relationships between atmospheric nitrogen loading
and ecological effects, however, is complicated by
the dynamic nature of ecological systems. In addi-
tion to being characterized by non-linear, "thresh-
old" type responses, estuarine ecosystems are simul-
taneously influenced by a variety of stressors (both
anthropogenic and natural). This makes it difficult
to quantify the nature and magnitude of ecological
changes expected to result from a change in a single
stressor such as nutrient loading. Further, if the state
of the ecosystem has changed (as from oligotrophic1
to eutrophic) the removal of the initial stressor does
not necessarily mean a rapid return to the prior state.
This complicates the quantitative benefits assessment
of controlling nitrogen deposition through the
CAAA.
1 Oligotrophy refers to a state of relatively low nutrient
enrichment and low productivity of aquatic ecosystems. In con-
trast, eutrophy refers to a state of relatively high nutrient load-
ing and higher prod uc li vi t y, soine Limes lea ding lo
overenrichmeiit and reduction in ecological service flows due
Lo waLer quality decline.
89
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Our second track relies on a displaced cost ap-
proach to benefit estimation. To reduce excess nu-
trient loads (including nitrogen) to local estuaries,
many coastal communities are pursuing a range of
abatement options. These options include waste-
water and stormwater discharge point source con-
trols as well as urban non-point and agricultural
non-point source controls for runoff from the land.
If atmospheric nitrogen depostion is reduced, the
need for these types of expenditure to control other
sources of nitrogen loading is also lessened, and the
displaced control expenditures represent a benefit
to society.
Displaced or avoided cost approaches are not
always justified. In order to establish that the costs
would truly be avoided, and to ensure that the avoid-
ance of that cost represents a real benefit to society,
we need to show that realistic and enforceable nitro-
gen reduction goals exist for each evaluated estuary.
Without specific targets or reduction goals, it is not
possible to suggest that there are specific control
expenditures to be displaced. Therefore, we choose
case study estuaries that most closely meet this crite-
rion: Chesapeake Bay, Long Island Sound, and
Tampa Bay. These areas have established nitrogen
reduction programs that rely primarily on reductions
of effluent from point sources as well as reductions
in non-point source discharges. Information on the
reduction goal and potential abatement options for
meeting those goals allows us to estimate the por-
tion of the goal that can be met by the CAAA, as
well as the associated cost savings.2
The benefits valuation derived using the dis-
placed-costs approach should be interpreted cau-
tiously for two reasons. First, it is an estimation of
capital costs that serve more purposes than mitigat-
ing nitrogen inputs into the estuaries of concern.
Water treatment works are intended to provide waste
water treatment for a variety of pollutants and may
be required even in the absence of deposition of air-
borne nitrogen. Second, the nitrogen loading tar-
gets for the estuaries are not concrete, strictly en-
forced limits, based on certain knowledge of the ca-
pacity of the estuaries to accept nitrogen inputs.
z With increasing populations, controls of alternative
sources (e.g., automobile and utility emissions) may be needed
simply to meet the original target or goal. If the CAA amend-
ments are necessary just to achieve the target reductions, then
we are actually measuring alternative costs and not avoided costs.
Instead, the targets may change over time as knowl-
edge of the effects of nitrogen to these estuaries
change. For these reasons, and because of the un-
certainty about the ability of local and regional enti-
ties to enforce the nitrogen reduction targets, we
calculate estimates of displaced costs for these three
estuaries but do not include them in the primary
benefits estimate for the CAAA.
Our approach involves three basic steps. First,
we estimate the total loading of nitrogen to each of
the three target estuaries. We use nitrogen deposi-
tion estimates from the RADM model, generated
for each 80 km x 80 km grid cell in the eastern U.S.
We then estimate the ultimate fate of deposited ni-
trogen through a GIS-based model of nitrogen "pass-
through." The pass-through is the share of nitrogen
deposited that is ultimately transported to the cstua-
rinc waters rather than retained by the land. Pass-
through factors vary by land use, from about 20 per-
cent (for forests and wetlands) to 100 percent (for
open water). We estimate the nitrogen loading for
each scenario, and the within-year, cross-scenario
differences are the reduced nitrogen deposition at-
tributed to the CAAA. We present these estimates
in the second column of Table 7-7.
Second, we estimate the marginal costs of alter-
native abatement actions which could be imple-
mented in the three case study estuaries. We develop
our displaced-cost estimate by assuming that deci-
sion makers will choose to forego the most costly
nitrogen abatement projects first. That is, we as-
sume that reduced deposition and the resulting load-
ings reduction will eliminate the need for additional
point or non-point source controls at the high end
of the marginal cost curve. We summarize those re-
sults 111 the third and fourth columns of Table 7-7.
Third, we multiply the reduced nitrogen load-
ing attributed to the CAAA by the marginal cost
estimate to arrive at a range of estimates of displaced
cost, ensuring that the reduction in airborne nitro-
gen is less than or equal to the potential tonnage
reduction achieved by the displaced, high marginal
cost abatement strategies. We present our results in
the last column of Table 7-7. Our estimates suggest
that the displaced cost is substantial for the large
Chesapeake Bay and Long Island Sound estuaries,
and more modest for Tampa Bay. The Chesapeake
90
-------
Chapter 7: Ecological and Other Welfare Effects
Table 7-7
Estimated Displaced Costs for Three Estuaries
Estuary
Long Island
Sound
Chesapeake
Bay
Tampa Bay
Reduced N Deposition in
2010(millions of pounds)
12.8
58.1
1.8
Low Marginal
Cost($/lb/yr.)
$2
$6
$6
High Marginal Cost
($/lb/yr.)
$8
$22
$38
Estimated Annual
Displaced Costs in
2010 ($millions)
$26-$ 100
Central Estimate: $63
$350-$1,300
Central Estimate: $820
$11 -$68
Central Estimate: $40
Bay and Long Island Sound watersheds together ac-
count for about 40 percent of the total estuarine
watershed area on the East (Atlantic) coast that is
sensitive to nitrogen deposition, while Tampa Bay
accounts for about two percent of the sensitive wa-
tershed area for the Gulf coast.
of Freshwater
During the 1970s and 1980s, "acid rain" came to
be known to the public as a phenomenon that in-
jures trees, forests, and water bodies throughout
Europe and in some areas of the United States and
Canada. One of the goals of the CAAA was to ad-
dress the problem of acidification of terrestrial and
aquatic ecosystems caused by acidic deposition. To
assess this effect we conducted a quantitative analy-
sis of benefits derived from a reduction in acidifica-
tion of aquatic bodies as they relate to recreational
fishing in die Adiroiidacks region of New York State.
As discussed earlier in this chapter, acidification
of water bodies is a complex process. Airborne ac-
ids, 111 the form of sulfur and nitrogen compounds,
are deposited to water bodies and surrounding drain-
age areas, with the potential to change the pH of the
water body. Many water bodies are relatively resis-
tant and can absorb a great deal of deposition before
pH changes substantially. This buffering capacity is
referred to as acid neutralizing capacity (ANC).
Once pH begins to be affected, a scries of interac-
tions occur, the most important of which is the leach-
ing of aluminum from sediments and surrounding-
soil and the suspension of this metal in the water
column. While acidic pH presents a direct stress to
aquatic organisms, it is the combined effect of pH
and aluminum exposure that presents the greatest
risk. Lakes in the Adirondacks region of New York
State are particularly susceptible to acidification be-
cause they have low baseline ANC, relative to wa-
ter bodies in other areas of the country.
Because of these complex physical and chemical
interactions, acidification stress is typically evaluated
by application of a model that simulates these pro-
cesses, and requires data on individual lake chemis-
try and sediment composition. We relied on the
scenario-specific atmospheric deposition data (both
sulfur and nitrogen) from the RADM air quality
model (see Chapter 4 and Appendix C) as an input
to EPA's Model of Acidification of Groundwatcr in
Catchments (MAGIC). MAGIC generates several
measures of the impact of sulfur and nitrogen depo-
sition on lake acidity, including ANC and pH.3 We
used the pH outputs to classify lakes where recre-
ational fishing might be impaired, and those estimates
were used in an economic model of recreational fish-
ing behavior in New York State to develop economic
estimates of the impact of acid rain on recreational
fishing resources in that state.
We summarize the results of our analysis of eco-
nomic benefits of avoided Adirondacks acidification
attributable to the CAAA in 2010 in Table 7-8. The
range of annual benefits from the CAAA are f 12
million to |49 million using the low-end assump-
tions on the threshold of effect (pH 5.0), and f 82 to
$88 million for the high-end assumptions on the ef-
fects threshold (pH 5.4). Higher pH (or, less acidic)
threshold assumptions lead to greater damage esti-
mates, because more lakes cross the less acidic thresh-
old. We calculate our benefits results by comparing
3 For more iniormaiion on EPA's MAGIC model see Cosbv
el al. (1985a); as referenced in Appendix E.
91
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 7-8
Annual Economic Impact of Acidification in 2010 (Millions of 1990 Dollars)
Year Scenario
1990 Base Year
2010 Post-CAAA
Pre-CAAA
Range of CAAA Benefits in 2010
Range of Economic Impact
Low Estimate Central Estimate High Estimate
$61
$24 to $61
$73
$12 to $49
$320
$261 to $281
$349 to $363
$50 $82 to $88
the suite of Post-CAAA 2010 estimates of total dam-
ages to the corresponding suite of estimates using
Pre-CAAA deposition. The impact of nitrogen satu-
ration in the surrounding terrestrial environment is
reflected in the range of estimates presented in Table
7-8. If surrounding soils are saturated, less deposited
nitrogen will remain on the land and more nitrogen
will enter the water bodies, increasing the stress on
the aquatic ecosystem. This phenomenon is reflected
by the higher damage estimates for saturated versus
non-saturated scenarios, other factors equal, although
our model shows no effect of saturation in the 2010
Pre-CAAA low estimate. The results we present
are in line with those generated from previous analy-
ses that find annual benefits to the Adirondacks of
halving utility emissions to be approximately in the
millions to tens of millions of dollars.4
Timber Growth
with
The third category of effects we quantify is im-
proved commercial timber growth through the re-
duction of tropospheric ozone concentrations attrib-
utable to the CAAA. There is substantial scientific
evidence to suggest that elevated ozone concentra-
tions in the troposphere disrupt ecosystems by dam-
aging and slowing the growth of vegetation. In this
analysis, we examine one aspect of these impacts,
reduced commercial timber growth. Much of the lit-
erature on the effects of ozone on tree growth is
based on laboratory exposures of seedlings or leaf-
scale experiments in the field. Estimates from those
studies have been used in previous analyses, making-
use of professional judgment as an interpretive tool,
but always with strong caveats about the potential
applicability of the seedling and leaf-scale results to
tree growth and, in particular, the rate of accumula-
tion of wood mass that is important for commercial
timber production.5 In an attempt to overcome these
issues, we sought to find a concentration-response
relationship that would provide a more defensible
and broadly applicable basis for estimating effects
on tree growth.
Our analysis makes use of the Net Photosyn-
thesis and Evapo-Transpiration model II (PiiET II),
a biological model of timber stand productivity to
estimate the impacts of ozone on timber yields. The
PnET II model was designed to estimate the com-
bined effects of several stressors on die rate of net
primary productivity (NPP), a measure of the rate
of photosynthesis. NPP in a tree does not necessar-
ily all go towards accumulation of wood mass; some
may be allocated to root growth, leaf growth, or
other tree functions. The PnET II model provides a
means to measure both NPP and wood mass growth,
as well as die effect on trees of several stressors com-
bined. One important stressor to acknowledge in
an analysis of the effects of ozone on trees is drought
stress. Ozone has the effect of reducing water loss in
trees by stimulating the closing of stomata through
which water is transpired. As a result, in drought
stress conditions, ozone can have beneficial effects
on tree growth. The PnET II model reflects the
impact of this factor in combination with other di-
rect effects of ozone on tree function.
We used the PnET II model to provide estimates
of timber stand responses to ozone exposure under
each of the scenarios examined in this analysis. We
aggregated tree growth results by region, with sepa-
rate estimates for hardwoods and softwoods, and used
them as inputs to the Timber Assessment Market
^ For alternative estimates see, lor example, Eriglin el al.
(1991), Mullen and Menz (1985), and Morey and Shaw (1990), as
referenced in Appendix E.
:> See de Steiger el al. (1990) ior an example oi the genera-
tion of tree growth dose-response estimates based on professional
judgement.
92
-------
Chapter 7: Ecological and Other Welfare Effects
O
Model (TAMM), an economic model of the for-
est sector maintained by the United States For-
est Service. There are three stages to the eco-
nomic estimation. First, forest growth rate in-
formation generated by PnET II is provided to
the Aggregate Timber Land Assessment System
(ATLAS), the forest inventory tracking com-
ponent of TAMM. Growth rate information is
provided for each of the forest production re-
gions defined by TAMM.6 Second, ATLAS
generates an estimate of forest inventories in
each major region, which in turn serves as in-
put to the market component of TAMM. i|
Third, TAMM estimates the future harvests and
market responses in each region.
Our analysis suggests that there is a signifi-
cant and measurable difference in timber har-
vests attributable to ozone exposure under the Post-
CAAA and Pre-CAAA scenarios. At the outset of
our modeling period, the early 1990s, virtually no
change is measured in forest harvest volumes. This
result occurs because increases in growth rates do
not substantively affect timber volume over a short
period of time. By the end of our modeling period,
nearing 2010, increased growth rates over the previ-
ous decade(s) begin to affect overall forest yields of
harvestable timber. This is observed in Figure 7-1 as
an increasing annual benefit estimate over the mod-
eling period. The shape of the benefits time-series
reveals a production spike in the 2007 to 2008 pe-
riod. This spike is due to a large anticipated harvest
of Southeast U.S. timber due to forest maturity dur-
ing this period. The spike would occur even in the
absence of the CAAA, but is elevated by the CAAA
due to increased growth rates projected under the
Post-CAAA scenario. Although this change is small
in percentage terms relative to total economic sur-
plus generated by the timber sector, it contributes
to a large portion of the commercial timber benefits
estimate over the 1990-2010 period.
We calculate the cumulative value of annual ben-
efits based on the discounted stream of the annual
differences in consumer and producer surplus from
6 TAMM includes Canadian as well as U.S. timber produc-
tion regions because of the important influence of Canadian tim-
ber supply on the U.S. market. This analysis reflects modeling
of Canadian timber regions and their impact on U.S. produc-
tion, but we did not simulate changes in ozone in Canadian re-
gions.
Figure 7-1
Annual Economic Welfare Benefit of Mitigating Ozone
Impacts on Commercial Timber: Difference Between
the Pre-CAAA and Post-CAAA Scenarios
$1,150T
« $950-
$750
*- $550
o
V)
commercial timber harvests under the Post-CAAA
and Pre-CAAA ozone exposure scenarios from 1990
to 2010. Discounting annual benefits to 1990 using
a five percent discount rate, the total cumulative
benefits estimate is approximately $1.9 billion. These
estimates are incorporated into the primary central
estimate by developing a range of annual estimates
for the year 2000, based on model results for the
period 1998 to 2002, and the year 2010, based on
model results for the period 2005 to 2010. The aver-
aging of results across several years to generate our
target year results avoids the potential problem of a
particular year's results (such as for 2010)
mischaracterizing the full time series of estimates
when we later calculate the net present value of ef-
fects.
Reduced Carbon Sequestration
Associated with Reduced Timber
Growth
Forest ecosystems help mitigate increasing atmo-
spheric concentrations of carbon dioxide by seques-
tering carbon from the atmosphere. These ecosys-
tems convert atmospheric carbon into biological
structures (e.g., wood) or substances needed in the
tree's physiological processes. As described above,
however, ozone reduces the growth of forests,
thereby limiting the amount of carbon that is se-
questered. Sequestered carbon can help mitigate glo-
bal climate change that has been linked to anthro-
pogenic emissions of carbon and other greenhouse
gases.
93
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
We used the timber inventory output of the
TAMM/ATLAS modeling system (described above),
in combination with a forest carbon model
(FORCARB), to estimate changes in carbon storage
in each of four ecosystem components: trees, forest
understory, forest floor, and soil. The estimates from
FORCARB, however, do not account for 'leakages"
of carbon back to the atmosphere as wood or wood
products decay and decompose over time. To esti-
mate the amount of carbon that is sequestered over
the long-term, we used a second model,
HARVCARB, to estimate the life-cycle of harvested
forest timber and thereby adjust the forest carbon
sequestration estimates of FORCARB.
The results of these calculations yield estimates
of long-term increases in carbon storage as a result
of the CAAA provisions of 8 million metric tons of
carbon per year by the year 2000, and 29 million
metric tons of carbon per year by the year 2010.
Because of the great uncertainties in assessing the
mitigating effect of carbon sequestration on global
climate change, and the economic value of avoiding
climate change, we do not attempt to monetize this
category of benefit.
Otter of
There were two additional categories of ecologi-
cal effects for which we considered developing eco-
nomic estimates; however, we abandoned the exer-
cise when key portions of the analysis proved to be
excessively problematic. Aesthetic degradation of
forests, the first of these additional categories, was
supported by a benefits transfer of contingent valu-
ation studies of individual willingness to pay to avoid
foliar damage. This category of effects, however,
proved too difficult to link to the specific air quality
scenarios we evaluated. In other words, available
scientific methods and data on the visual appearance
of forest stands and their impact on perceived forest
aesthetics make it difficult to precisely describe
changes in forest aesthetics. Evaluation of the sec-
ond additional effect category, toxification of fresh-
water fisheries, was limited by the lack of toxic depo-
sition and exposure data as well as by the limitations
of available economic estimates of the impacts of
toxics on recreational and commercial fishery re-
sources. (See Appendix E for a more detailed dis-
cussion of these service flows). These and many other
ecological benefit categories could not be quantified
given current data and methods and are thus not re-
flected in our overall benefits estimates.
Valuation of Other
As discussed earlier in this chapter, tropospheric
ozone affects the growth of a wide range of plant
species, including agricultural crops. Our agricul-
tural benefits analysis relies on crop-yield loss C-R
functions derived from the National Crop Loss As-
sessment Network (NCLAN) research and a national
economic model of the agricultural sector (AGSIM).
The NCLAN-denved relationships use a sum of
hourly ozone concentration at or above 0.06 ppm
(SUM06) as a measure of ozone exposure for the .May
to September ozone season; these exposure estimates
arc derived from the ozone air quality modeling re-
sults discussed in Chapter 4. Where the C-R func-
tions require a longer time period of ozone concen-
trations, for example, for winter crops or when the
growing or harvest season for summer crops extends
beyond the end of September, we rely on 1990 moni-
tor data to estimate ozone exposure, conservatively
using the same estimates for both Pre-CAAA and
Post-CAAA scenarios. The NCLAN functions
cover the following crops: corn, cotton, peanuts,
sorghum, soybeans, and winter wheat.
The AGSIM agricultural sector model takes the
yield loss information, incorporates agricultural
price, farm policy, and other data for each year, and
then estimates production levels for each crop and
the economic benefits to consumers and producers
associated with these production levels. The crop
coverage in the AGSIM model includes a wider range
of crops than the NCLAN data inputs, adding bar-
ley, oats, hay, rice, and cottonseed. The broader
crop coverage ensures that the model addresses price
and production quantity effects on potential substi-
tute crops that might be related to the effects in the
six NCLAN crops. We estimate economic effects
using a range of C-R outcomes for several crops, to
reflect the variation in ozone sensitivity among the
various crop cultivars. Our central estimate is the
expected value of the range of results that emerge
from the economic model.
94
-------
Chapter 7: Ecological and Other Welfare Effects
Our results indicate significant beneficial effects
of ozone reductions in the agricultural sector. Our
Primary Central estimate of the benefit in 2000 is
$450 million; the annual benefit rises to $550 mil-
lion in 2010. Our estimated uncertainty around the
Primary Central estimates, however, is very broad.
For example, in 2010, the Primary Low estimate is
$7.1 million, and the Primary High is SI,100 mil-
lion. The uncertainty range reflects variation in the
ozone response of crop cultivars and uncertainty
about the suitability of alternative crop cultivars for
the soil types and climate conditions in various agri-
cultural regions. See Appendix F for more details
on the methods and results of the C-R functions and
economic modeling for agricultural effects.
Visibility
As outlined in Chapter 4, air pollution impairs
visibility in both residential and recreational settings.
An individual's willingness to pay to avoid reduc-
tions 111 visibility differs in these two settings. Im-
pairments in residential visibility are experienced
throughout an individual's daily life and activities.
Visibility in recreational settings, on the other hand,
is experienced by visitors to areas with notable vis-
tas. For the purposes of this report, we interpret
recreational settings applicable for this category of
effects to include National Parks throughout the
nation. Other recreational settings may also be ap-
plicable, for example National Forests, state parks,
or even hiking trails or roadside areas, but a lack of
suitable economic valuation literature to identify
these other areas, as well as a lack of visitation data
m some cases, prevents us from generating estimates
for those recreational vista areas.
We derive a residential visibility valuation func-
tion from the Chestnut and Dennis (1997) published
estimates for the Eastern U.S. These estimates are
based on original research conducted by McClelland
et al. (1990) in two Eastern cities (Atlanta and Chi-
cago). Because of technical concerns about the
study's methodology, however, we calculate a ben-
efits estimate but omit the results from the primary
benefits estimates.' For recreational visibility, we
' The two technical concerns involve the method of adjust-
ing the contingent valuation survey results for non-response, and
the failure to include adjustments for the "warm glow" effect,
or the tendency of respondents to indicate higher willingness to
pay for an environmental good because of a strong desire to
improve the environment in general.
derive values from die the Chestnut and Rowe (1989)
study of WTP for visibility in three park regions in
the Western, Southwestern, and Eastern U.S.8 In
both cases, the valuation function takes the follow-
ing form:
HHWTP = B * ln(VR1/VR2)
where:
HHWTP = annual WTP per household for
visibility changes
VR1 = the starting annual average visual
range
VR2 = the annual average visual range after
the change in air quality
B = the estimated visibility coefficient.
The form of this valuation function is designed
to reflect the way individuals perceive and express
value for changes in visibility. In general terms, ex-
pressed WTP for visibility changes varies with the
percentage change in visual range, a measure that is
closely related to, though not exactly analogous to,
the Deciview index used in Chapter 4. We use a
central B coefficient for residential visibility of 141,
as reported in Chestnut and Dennis (1997). For rec-
reational visibility, the coefficients vary based on the
region of study and whether the household is within
or outside of the National Park region studied. Iii-
region coefficients are higher than those for out-of-
region households. The in-region estimates for Cali-
fornia, the Southwest, and Southeast are $105, $137,
and $65, respectively; the corresponding out-of-re-
gion estimates are S73, $110, and $40, respectively.
The derivation and application of these valuation
functions are described in more detail in Appendix
H. The results of this procedure suggest visibility is
an important category of CAAA benefits; the Pri-
mary Central estimate for 2010, for example, indi-
cates annual recreational visibility benefits of $2.9
billion.
Productivity
We base the valuation of worker productivity
on a study that measures the decline in worker pro-
3 The visibility valuation function, and the sources of esti-
mates for the coefficients for the functions, were originally de-
veloped as part of the National Acid Precipitation Assessment
Program (NAPAP), and were subjected to peer-review as part
of that program.
95
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
ductivity among outdoor farm workers exposed to
ozone (Crocker and Horst, 1981). In our analysis,
we estimate the value of reduced productivity at f 1
per 10 percent increase in ozone concentration. This
estimate reflects valuing reduced productivity in
terms of the reduction in percentage of daily income
incurred by the average worker engaged in strenu-
ous outdoor labor.
Provisions
The quantified benefits of stratospheric ozone
protection provisions are dominated by the reduced
health effects expected from reductions in UV-b ra-
diation; the derivation of health benefits of these
provisions is discussed in Chapter 5. We summarize
other categories of benefits associated with reduced
UV-b radiation in Table 7-9. The quantified ben-
efits include: reduced crop damage; and reduced poly-
mer degradation. To estimate crop damage, we ap-
ply the results of existing studies on the relationship
between crops and UV-b radiation to the changes in
UV-b radiation predicted by the emissions and at-
mospheric models.9 The polymer damage function
is based on a study by Horst (1986). The estimated
total cumulative benefits associated with these eco-
logical and other welfare effects are about 2 percent
of the total cumulative benefits of the Title VI pro-
visions.
'* Sources oi dose-response relationship for crops and UV-
b: Teramura and Murali (1986) and Rowe and Adams (1987).
Source of dose-response relationship for crops and tropospheric
ozone: Rowe and Adams (1987).
Table 7-9
Quantified and Unquantified Ecological and Welfare Effects of Title VI Provisions
Ecological Effects- Quantified
Estimate
Basis for Estimate
American crop harvests
American crop harvests
Avoided 7.5 percent decrease Dose-response sources: Teramura and Murali
from UV-b radiation by 2075 (1986), Rowe and Adams (1987)
Avoided decrease from
tropospheric ozone
Estimate of increase in tropospheric ozone:
Whitten and Gery (1986). Dose-response
source: Rowe and Adams (1987)
Polymers
Avoided damage to materials
from UV-b radiation
Source of UV-b/stabilizer relationship: Horst
(1986)
Ecological Effects- Unquantified
Ecological effects of UV. For example, benefits relating to the following:
• recreational fishing
• forests
• marine ecosystem and fish harvests
• avoided sea level rise, including avoided beach erosion, loss of coastal wetlands, salinity of estuaries
and aquifers
• other crops
• other plant species
• fish harvests
Ecological benefits of reduced tropospheric ozone relating to the overall marine ecosystem, forests, man-made
materials, crops, other plant species, and fish harvests
Benefits to people and the environment outside the U.S.
Notes:
1) For more detail see EPA's Regulatory Impact Analysis: Protection of Stratospheric Ozone (1988).
2) Note that the ecological effects, unlike the health effects, do not reflect the accelerated reduction and phaseout schedule
of section 606.
3) Benefits due to the section 606 methyl bromide phaseout are not included in the benefits total because the EPA provides
neither annual incidence estimates nor a monetary value.
96
-------
Chapter 7: Ecological and Other Welfare Effects
of
Results
Although the effects of air pollutants on eco-
logical systems are likely to be widespread, many
effects may be poorly understood and lack quantita-
tive effects characterization methods and support-
ing data. In addition, many of our quantitative re-
sults reflect an incomplete geographic scope of analy-
sis; for example, we generated monetized acidifica-
tion results only for the Adirondacks region of New
York State. As a result, the quantitative results we
generate for the purposes of estimating the benefits
of the CAAA reflect only a small portion of the over-
all impacts of air pollution on ecological systems or
ecological service flows.
Despite these limitations, it is important to rec-
ognize the magnitude of the monetized ecological
benefits that we could estimate and reflect those re-
sults in the overall estimates of benefits generated
in die larger analysis. Table 7-10 provides a tabular
summary of the results documented earlier in this
chapter. It is not possible to indicate the degree to
which ecological benefits are underestimated, but
considering the magnitude of benefits estimated for
the select endpoints considered in our analysis, it is
reasonable to conclude that a comprehensive ben-
efits assessment would yield substantially greater total
benefits estimates.
In Table 7-11 we provide a summary of benefits
estimates for other welfare effects, including reduced
agricultural yields, impaired visibility, and decreased
Table 7-10
Summary of Evaluated Ecological Benefits (millions 1990$)
Description
of Effect
Freshwater
acidification
Reduced tree
growth - Lost
commercial
timber
Air
Pollutant
Sulfur and
nitrogen
oxides
Ozone
Geographic
Scale of
Economic
Estimate
Regional
(Adirondacks)
National
TOTAL MONETIZED
ECONOMIC BENEFIT
Range of
Annual Impact
Estimates in
2010
$12 to $88
$190 to $1000
$200 to $1,1 00
Primary
Central
Estimate
for 2010
$50
$600
$650
Primary Central
Cumulative
Impact Estimate
1990-2010
$260
$1,900
$2,200
Key Limitations
- Captures only
recreational fishing
impact
- Incomplete
geographic coverage
leads to underestimate
of benefits
- Uncertainties in
stand-level response to
ozone exposure
- Uncertainty in future
timber markets
- Partial estimate that
omits major
unquantifiable benefits
categories; see text
Note: Estimates reflect only those benefits categories for which quantitative economic analysis was supported. A
comprehensive total economic benefit estimate would likely greatly exceed the estimates in the table. Range of
estimates for timber assessment is based on variation in annual point estimates for 2005 through 2010.
Table 7-11
Summary of Other Welfare Benefits (millions 1990$)
Description
of Effect
Air
Pollutant
Geographic
Scale of
Economic
Estimate
Primary Central
Annual Estimate
2000 2010
Primary Central
Cumulative
Estimate
1990-2010
Key Limitations
Reduced Ozone
Agricultural
Yields
National $450 $550 $3,900 - Covers only major grain crops
- Omits effects on fruits and
vegetables
Impaired
Recreational
Visibility
Reduced
Worker
Productivity
Particulate National
Matter
Ozone National
$2,000
$460
$2,900
$710
$19,000
$4,400
- National Parks only
- Omits residential visibility
benefits
- Reflects effects on workers
engaged in strenuous outdoor
employment
Note: Estimates reflect only those benefits categories for which quantitative economic analysis was supported.
97
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
worker productivity. These estimates add substan-
tially to the total non-health benefits of the CAAA.
In particular, our estimates for the annual value of
avoiding visibility impairments is $2,900 million by
2010, even through this estimate does not reflect the
value of residential visibility improvements.
Uncertainty
Because of the limitations in the available meth-
ods and data, the benefits assessment in this report
does not represent a comprehensive estimate of the
economic benefits of the CAAA. Moreover, the
potential magnitude of long-term economic impacts
of ecological damages mitigated by the CAAA sug-
gests that great care must be taken to consider those
ecosystem impacts that are not quantified here. Sig-
nificant future analytical work and basic ecological
and economic research must be performed to build
a sufficient base of knowledge and data to support
an adequate assessment of ecological benefits. For
the current analysis, this incomplete coverage of
effects represents the greatest source of uncertainty
in the ecological assessment. This and other key
uncertainties are summarized in Table 7-12.
Because the chronic ecological effects of air pol-
lutants may be poorly understood, difficult to ob-
serve, or difficult to discern from other influences
011 dynamic ecosystems, our analysis focuses on acute
or readily observable impacts. Disruptions that may
seem inconsequential in the short-term, however, can
have hidden, long-term effects through a series of
interrelationships that can be difficult or impossible
to observe, quantify, and model. This factor sug-
gests that many of our qualitative and quantitative
results may underestimate the overall, long-term ef-
fects of pollutants on ecological systems and re-
sources.
Table 7-12
Key Uncertainties Associated with Ecological Effects Estimation
Potential Source of Error
Direction of
Potential
Bias for Net
Benefits
Estimate
Likely Significance Relative to Key Uncertainties in Net
Benefit Estimate*
Incomplete coverage of
ecological effects
identified in existing
literature, including the
inability to adequately
discern the role of air
pollution in multiple
stressor effects on
ecosystems.
Underestimate Potentially major. The extent of unquantified and unmonetized
benefits is largely unknown, but the available evidence suggests
the impact of air pollutants on ecological systems may be
widespread and significant. At the same time, it is possible that a
complete quantification of effects might yield economic valuation
results that remain small in comparison to the total magnitude of
health benefits.
Omission of the effects of
nitrogen deposition as a
nutrient with beneficial
effects.
Incomplete assessment of
long-term bioaccumulative
and persistent effects of
air pollutants.
The PnET II modeling of
the effects of ozone on
timber yields relies on a
simplified mechanism of
response (i.e., changes in
net primary productivity).
Overestimate Probably minor. Although nitrogen does have beneficial effects
as a nutrient in a wide range of ecological systems, nitrogen in
excess also has significant and in some cases persistent
detrimental effects that are also not adequately reflected in the
analysis.
Underestimate Potentially major. Little is currently known about the longer-term
effects associated with the accumulation of toxins in ecosystems,
but what is known suggests the potential for major impacts.
Future research into the potential for threshold effects is
necessary to establish the ultimate significance of this factor.
Overestimate Probably minor. Existing evidence suggests that the growth
changes PnET II projects are relatively large, however none of
the currently available points of conparison fully address such
issues as the impact of stand-level competition, and the net
primary productivity results are within the range of results of other
studies of environmental and anthropogenic stressors.
The classification of each potential source of error reflects the best judgement of the section 812 Project Team. The Project
Team assigns a classification of "potentially major" if a plausible alternative assumption or approach could influence the
overall monetary benefit estimate by approximately five percent or more; if an alternative assumption or approach is likely to
change the total benefit estimate by less than five percent, the Project Team assigns a classification of "probably minor."
98
-------
Comparison of
Costs and Benefits
In this chapter we present our summary of the
primary estimates of monetized benefits of the
CAAA from 1990 to 2010, compare the benefits es-
timates with the corresponding costs, and explore
some of the major sources of uncertainty in die ben-
efits estimates. We also present the results of our
calculations using alternative assumptions for sev-
eral key input variables.
of the
CAAA
Iii this section we provide an overview of the
three types of analyses conducted to estimate ben-
efits, present the annual estimates of monetized ben-
efits for the human health, ecological, and welfare
analyses, and then present an aggregate measure of
benefits from all titles of the CAAA for the full study
period.
Overview of
Our primary estimates of the monetized eco-
nomic benefits for the 1990 to 2010 period derive
from three distinct analyses: (1) the analysis of
changes in human health effects associated with re-
duced exposures to criteria pollutants and the subse-
quent valuation of these changes, summarized and
described in Chapters 5 and 6; (2) the analysis of
monetized ecological and other welfare benefits (e.g.,
visibility), described in Chapter 7; and (3) the analy-
sis of the benefits of stratospheric ozone protection
provisions, summarized briefly in Chapters 5, 6, and
7 and described in detail in Appendix G.
We measure the benefits and present the results
from each of these analyses in slightly different ways.
For the first two analyses, we generate annual esti-
mates of benefits that result from changes in expo-
sures in two target years of the study, 2000 and 2010.
These estimates can be directly compared to the es-
timates of costs incurred in the target years, because
for the most part the annual benefits accrue in the
same year as the costs are incurred. There is one
exception, however: we model the effect of particu-
late matter on premature mortality to occur over a
period of five years from the time of exposure. In
this case, we have accounted for the incidence of
premature mortality over the assumed lag period,
and discounted the valuation of this effect back to
the target year.
The annual estimates provide an indication of
the trend in benefits accrued over the 20-year study
period. To generate a cumulative measure of ben-
efits over the full 20-year period, we must make an
assumption about the level of benefits that would
be realized in the years between the target years. Wre
interpolate these values, assuming a linear trend in
both costs and benefits over the 1990 to 2000 and
2000 to 2010 periods (assuming benefits and costs in
the starting year, 1990, are zero). In one portion of
the ecological benefits analysis, acidification, we gen-
erate only a single annual estimate for the target year
2010. In that case, we assume a linear trend in an-
nual benefits over the full 20-year study period.
The third analysis, assessing changes in strato-
spheric ozone and the resulting health effects, is dif-
ferent from the criteria pollutant analyses. The long-
term nature of the program, and the significant lag
effects associated with the processes of ozone deple-
tion over decades-long time scales, make it difficult
to generate a meaningful estimate for any single tar-
get year. As a result, we could not generate an an-
nual benefit estimate that could be reliably linked to
emissions reductions in a single year and, by exten-
sion, compared to the costs incurred to achieve that
year's allocation of reductions in stratospheric ozone
depleting substances. Instead, we generate an annu-
alized equivalent of the cumulative present value of
99
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
benefits and costs of the Title VI program. These
annualized equivalents cannot be ascribed to any
particular target year.
These fundamental differences in the measure-
ment of benefits affect our presentation of benefits
estimates in this chapter. Although we generate and
report an annual estimate of costs and benefits of
Title VT provisions, we encourage the reader to in-
terpret aggregations of these annual estimates with
those from other titles of the CAAA with caution.
In particular, we discourage the use of these CAAA
Title-specific benefit-cost ratios as the sole, or even
primary, basis for comparing the relative economic
value of Title VI versus other CAAA titles. The
comparative benefit-cost ratios are too sensitive to
important, highly uncertain analytical assumptions
such as the discount rate.
Summary of for
Human Welfare
As discussed above, we generate annual estimates
for the human health and welfare effects based on
exposure analysis conducted for each of the two tar-
get years of the analysis, 2000 and 2010. The range
of estimates we generate for the monetized benefits
of human health effects incorporates both the quan-
tified uncertainty associated with each of the health
effect estimates and the quantified uncertainty asso-
ciated with the corresponding economic valuation
strategy. Quantitative estimates of uncertainties in
earlier steps of the analysis (i.e., emissions and air
quality changes) could not be developed adequately
and are therefore not applied in the present study.
As a result, the range of estimates for monetized ben-
efits presented in this chapter is more narrow than
would be expected with a complete accounting of
the uncertainties in all analytical components. The
characterization of the uncertainty surrounding eco-
nomic valuation is discussed in detail in Appendix
IT. The characterization of the uncertainty surround-
ing specific health effect estimates is discussed in
Appendix D. Below, we discuss the combined ef-
fect of these two categories of uncertainty and our
techniques for aggregating uncertainty across end-
points and analyses.
We assume that for each endpoint-pollutant com-
bination there are distributions for both the con-
centration-response function and the valuation co-
efficients. We combine these distributions by using
a computerized, statistical aggregation technique to
estimate the mean of the monetized benefit estimate
for each endpoint-pollutant combination and to char-
acterize the uncertainty surrounding each estimate.1
In the first step of our procedure, we employ
statistical analysis to generate mean estimates and
quantified uncertainty measures for the C-R func-
tion for each endpoint-pollutant combination. For
many health and welfare effects, only a single study
is available to use as the basis for the C-R function.
In this case, the best estimate of the mean of the
distribution of C-R coefficients is the reported esti-
mate in the study. The uncertainty surrounding the
estimate of the mean C-R coefficient is character-
ized by the standard error of the reported estimate.
This yields a normal distribution, centered at the
reported estimate of the mean. If multiple studies
are considered for a given C-R function, a normal
distribution is derived for each study, centered at
the mean estimate reported in the study. On each
iteration of the aggregation procedure, a C-R coeffi-
cient is selected from an aggregate distribution of C-
R estimates for that endpoint. The aggregate distri-
bution of C-R coefficients is determined by a vari-
ance-weighted aggregate distribution of values.
In the second step, we estimate incidence for each
exposure analysis unit (i.e., 8 km by 8 km cell in a
grid pattern) in the 48 contiguous states, and aggre-
gate the results into an estimate of the change in
national incidence of the health or welfare effects.
Through repeated iterations from the distribution
of mean C-R coefficients, we generate a distribution
of the estimated change in incidence for each health
and welfare effect due to the change in air quality
between the Post-CAAA and Pre-CAAA scenarios.
Finally, in the third step we use computerized
statistical aggregation methods once again to charac-
* The statistical aggregation technique applied is commonly
referred to as simulation modeling. The technique involves many
re-calculations of results, using different combinations of input
parameters each time. For each calculation, values from each
input parameter's statistical distribution are selected at random
to ensure that the calculation does not always result in extreme
values, or rely solely on low end or solely on high end input
parameters. The aggregate distribution more accurately reflects
a reasonable likelihood of the joint occurrence oi multiple in-
put parameters.
100
-------
Chapters: Comparison of Costs and Benefits
terize the overall uncertainly surrounding monetized
benefits. For each distinct health and welfare effect,
the aggregation procedure selects an estimated inci-
dence change from the distribution of changes for
that endpoint, selects a unit value from the corre-
sponding distribution of economic valuation unit
values, and multiplies the two to generate a mon-
etized benefit estimate. We then repeat the process
many times to generate a distribution of estimated
monetized benefits for each endpoint-pollutant com-
bination. Combining the results for the individual
endpoints using the aggregation procedure yields a
distribution of total estimated monetized benefits for
each target year (2000 and 2010) .2 We present the
results of this analysis of health effects in Table 6-3
in Chapter 6.
The ecological and welfare results are not cur-
rently amenable to the same type of uncertainty
analysis. The modeling procedures for estimating
the effects of sulfur and nitrogen deposition in acidi-
fying lakes, the effects of ozone in reducing timber
and agricultural production, and the effects of par-
ticular matter on visibility are all subject to uncer-
tainty and require substantial resources simply to
develop single estimates. We describe key uncer-
tainties in Chapter 7 and they are reflected in the
ranges of values we present at the end of that chap-
ter. The sources of uncertainty in these estimates,
however, cannot as easily be dis-
aggregated among physical ef-
fects modeling and valuation
components. The endpoints of
the ranges we present reflect rea-
sonable alternative choices in
key input variables, but the
ranges cannot currently be inter-
preted as points on a statistical
distribution of results. For these
ecological effects, the central es-
timate is the midpoint of the
ranges of values. We then inter-
pret the endpoints of the range
of estimates as the upper and
lower bounds of a uniform dis-
tribution of values. The uni-
form distribution is used when we aggregate the eco-
logical and other welfare effects analyses with the
analyses of human health.
Annual Benefits Estimates
We present the results of our aggregation of pri-
mary annual benefits estimates for Titles I through
V in Figure 8-1 below. The figure provides a charac-
terization of both the primary central estimate and
the range of values generated by the aggregation
procedure described above, for each of the two tar-
get years of the analysis (2000 and 2010). The Pri-
mary High estimate corresponds to the 95th percen-
tile value from the aggregation, and the Primary Low
estimate corresponds to the 5th percentile value. The
total benefits estimates are substantial; the Primary
Central estimate in 2010 is $110 billion.
Table 8-1 shows the detailed breakdown of ben-
efits estimates for one of the two target years, 2010.
As shown in the table, $100 billion of the $110 bil-
lion total benefit estimate in 2010, or roughly 90
percent, is attributable to reductions in premature
mortality associated with reductions in ambient par-
ticulate matter and associated criteria pollutants. The
remaining benefits are divided among two broad
categories of benefits: avoided morbidity, the larg-
est component of which is avoided chronic bron-
Figure 8-1
Central, Low, and High Primary Benefits Results for
Target Years (in billions of 1990 dollars) - Titles I through V
300-
« 250-
O
i= 200-
OQ
0)
c
0)
CD
3
o
150-
100-
50-
Benefits
Benefits
I^High 160
•« Central 71
^High 270
•^Central 110
•< Low 26
2000
2010
2 This procedure implicitly assumes independence between
the specific aggregation simulation draws from the distribution
of health and economic valuation estimates.
101
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 8-1
Criteria Pollutant Health and Welfare Benefits in 2010
Monetary Benefits (in millions 1990$)*
Primary
Benefits Category
Mortality
Ages 30+
Chronic Illness
Chronic Bronchitis
Chronic Asthma
Hospitalization
All Respiratory
Total Cardiovascular
Asthma-Related ER Visits
Minor Illness
Acute Bronchitis
URS
LRS
Respiratory Illness
Mod/Worse Asthma1
Asthma Attacks1
Chest Tightness, Shortness of
Breath, or Wheeze
Shortness of Breath
Work Loss Days
MRAD/Any-of-19
Welfare
Decreased Worker
Productivity
Visibility - Recreational
Agriculture (Net Surplus)
Acidification
Commercial Timber
Aggregate Range of Benefits2
Primary Low
14,000
360
40
76
93
0.1
0.0
4.2
2.2
0.9
1.9
20
0.0
0.0
300
680
710
2,500
7.1
12
180
26,000
Central
100,000
5,600
180
130
390
1.0
2.1
19
6.2
6.3
13
55
0.6
0.5
340
1,200
710
2,900
550
50
600
110,000
Primary High
250,000
18,000
300
200
960
2.8
5.2
39
12
15
29
100
3.1
1.2
380
1,800
710
3,300
1,100
76
1,000
270,000
Note:
* The estimates reflect air quality results for the entire population in the US.
1 Moderate to worse asthma, asthma attacks, and shortness of breath are endpoints
included in the definition of MRAD/Any of 19 respiratory effects. Although valuation
estimates are presented for these categories, the values are not included in total benefits to
avoid the potential for double-counting.
2 The Aggregate Range reflects the 5th, mean, and 95th percentile of the estimated credible
range of monetary benefits based on quantified uncertainty, as discussed in the text.
102
-------
Chapters: Comparison of Costs and Benefits
Table 8-2
Present Value of Monetized Benefits for 48 State Population
Present Value (millions 1990$, discounted to 1990 at 5 percent)
Primary Low
Primary Central
Primary High
Titles I through V (1990 through 2010)
$160,000
$690,000
$1,600,000
Title VI (1990 through 2165)
$100,000
$530,000
$900,000
chitis, comprises about 60 percent of the non-mor-
tality benefits; and avoided ecological and other wel-
fare effects, the largest component of which is im-
proved recreational visibility, comprises about 40
percent. Note that, because of the aggregation pro-
cedure used, and because we round all intermediate
results to two significant digits for presentation pur-
poses, the columns of Table 8-1 may not sum to the
total estimate presented in the last row.3
Aggregate
As discussed earlier in this chapter, we linearly
interpolate benefit estimates between 1990 and 2000
and between 2000 and 2010 and then aggregate the
resulting annual estimates across the entire 1990 to
2010 period of the study to yield a present discounted
value of total aggregate benefits for the period. In
this section we discuss issues involved in each stage
of aggregation, as well as the results of the aggrega-
tion.
As noted earlier, air quality modeling was car-
ried out only for the two target years (2000 and 2010).
The resulting annual benefit estimates provide a tem-
poral trend of monetized benefits across the period
resulting from the annual changes in air quality.
They do not, however, characterize the uncertainty
associated with the yearly estimates for intervening
years. In an attempt to capture uncertainty associ-
ated with these estimates, we relied on the ratios of
the 5th percentile to the mean and the 95th percen-
tile to the mean in the two target years. In general,
these ratios were fairly constant across the target
•' The sum of benefits across endpoiiits at a given percentile
level does not result in the total monetized beneiits estimate at
the same percentile level in fable 8-1. For example, if the fifth
percentile benefits of the endpoints shown in Table 8-1 were
added, the resulting total would be substantially less than $30
billion, the fifth percentile value of the distribution of aggregate
monetized benefits reported in Table 8-1. This is because the
various health and welfare effects are treated as stochastically
independent, so that the probability that the aggregate monetized
benefit is less than or equal to lie sum of the separate five per-
centile values is substantially less than live percent.
years, for a given cndpoint. The ratios were inter-
polated between the target years, yielding ratios for
the intervening years. Multiplying the ratios for each
intervening year by the central estimate generated
for that year provided estimates of the 5th and 95th
percentiles, which we use to characterize uncertainty
about the Primary Central estimate.
In Table 8-2 we present the cumulative mon-
etized benefits aggregated from 1990 to 2010. We
present the mean estimate from the aggregation pro-
cedure, along with the Primary Low7 (i.e., 5th per-
centile of the distribution) and Primary High (i.e.,
95th percentile of the distribution) estimates, for all
provisions of Titles I through V and, then, separately
for Title VI. Aggregating the stream of monetized
benefits across years involved discounting the stream
of monetized benefits estimated for each year to the
1990 present value (using a five percent discount rate).
Aggregate of Title ¥1 Provisions
As described in summary form in Chapters 5, 6,
and 7 and in detail in Appendix G, expected human
health benefits from Title VI provisions are substan-
tial. The analysis we conducted is based largely on
existing results from EPA Regulatory Impact Analy-
ses for individual rules promulgated under Title VI.
To the extent possible, we adjusted existing estimates
to reflect both the central estimates and uncertainty
characterizations used in the criteria pollutant analy-
sts. We made major adjustments for both the value
of statistical life (VSL) and the discount rate. We
adjusted the VSL estimate to reflect the Weibull dis-
tribution of VSL used in our analysis for other pro-
visions. As discussed in the appendix, the choice of
the discount rate for estimated benefits which ac-
crue over decades to century-long time spans pre-
sents special problems. Although we argue that a
two percent discount rate is more appropriate for
such long-term discounting, for consistency in this
chapter we present estimates using the five percent
discount rate used throughout the rest of this study.
103
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The results of the benefits calculations in Ap-
pendix G indicate a cumulative central benefit esti-
mate of $530 billion for Title VI (see Appendix G
for details). Using the same aggregation techniques
for the valuation analysis described above, but only
for the mortality valuation step, we generate a 90
percent confidence interval around this central esti-
mate to derive Primary Low and Primary High esti-
mates of $100 billion to $900 billion, respectively.
We present these estimates in Table 8-2 above. The
annual human health benefits from Title VI provi-
sions steadily increase until about 2045, then decrease
until 2165, the last year in the analysis. About 93
percent of the benefits accrue from 2015 to 2165.
These benefit estimates only partially reflect poten-
tial averting behaviors, such as remaining indoors
or increasing use of sun screens or hats, which may
mitigate the effects of the UV-b exposure increases
estimated under the Prc-CAAA scenario.
Of
Table 8-3 presents summary quantitative results
for the prospective assessment, with costs disaggre-
gated by Title and benefits disaggregated by major
category. We present annual,
primary estimate results for
each of the two target years of
the analysis, with all dollar fig-
ures expressed as inflation-ad-
justed 1990 dollars. The final
columns provide net present
value estimates for costs and
benefits from 1990 to 2010 or,
in the case of stratospheric
ozone protection provisions,
1990 to 2165, discounted to
1990 at five percent. The re-
sults indicate that the Primary
Central estimate of benefits
clearly exceeds the costs of the
CAAA, for each of the two
target years and for the cumu-
lative estimates of present
value over the 1990 to 2010 pe-
riod.
gating benefits by CAAA Title or even by pollut-
ant. As the table indicates, a very high percentage
of the benefits is attributable to reduced premature
mortality associated with reductions in ambient par-
ticulate matter and associated criteria pollutants. The
CAAA achieves ambient PM reductions through a
wide range of provisions controlling emissions of
both gaseous precursors of PM that form particles
in the atmosphere (sulfur and nitrogen oxides as well
as, to a lesser extent, organic constituents) and di-
rectly emitted PM (i.e., dust particles). Because the
effects of these constituents on ambient PM are non-
linear, and because some precursor pollutants inter-
act with each other in ways which influence the to-
tal concentration of participates in the atmosphere,
separating the effects of individual pollutants on the
change in ambient PM would require many itera-
tions of our air quality modeling system. These dif-
ficulties in separating the effects of individual emis-
sions reductions on the benefits estimates also high-
light the need for an integrated air quality modeling
system that can more readily analyze multiple sce-
narios within reasonable time and resource con-
straints. A tool of this nature could allow us to more
reliably and cost-effectively estimate incremental
contributions to ambient PM and ozone concentra-
tion reductions.
Table 8-3
Summary of Quantified Primary Central Estimate Benefits and Costs
(Estimates in million 1990$)
Cost or Benefit
Category
Annual Estimates
2000
2010
Present Value
Costs:
Title I
Title II
Title III
Title IV
Title V
Total Costs, Title I-V
Title VI
$8,600
$7,400
$780
$2,300
$300
$19,000
$1,
$14,500
$9,000
$840
$2,000
$300
$27,000
400*
$85,000
$65,000
$6,600
$18,000
$2,500
$180,000
$27,000*
Monetized Benefits:
Avoided Mortality
Avoided Morbidity
Ecological and
Welfare Effects
Total Benefits, Title I-V
Stratospheric Ozone
$63,000
$5,100
$3,000
$71,000
$25
$100,000
$7,900
$4,800
$110,000
,000*
$610,000
$49,000
$29,000
$690,000
$530,000*
The estimates in Table 8-3
reflect the difficulty we en-
countered in reliably disaggre-
* Annual estimates for Title VI stratospheric ozone protection provisions are annualized
equivalents of the net present value of costs over 1990 to 2075 (for costs) or 1990 to 2165
(for benefits). The difference in time scales for costs and benefits reflects the persistence of
ozone depleting substances in the atmosphere, the slow processes of ozone formation and
depletion, and the accumulation of physical effects in response to elevated UV-b radiation
levels.
104
-------
Chapters: Comparison of Costs and Benefits
Table 8-4 provides the results of our compari-
son of primary benefits estimates to primary cost
estimates. In the top half of the table we show both
annual and present value estimates for Titles I
through V, present value estimates for Title VI, and
a total present value for all titles. The "monetized
benefits" indicate both the Primary Central estimate
(the mean) from our statistical aggregation model-
ing analysis and the Primary Low and Primary High
estimates (5th and 95th percentile values, respec-
tively). In the bottom half of the table we present
two alternative methods for comparing benefits to
costs. "Net benefits" are the Primary Central esti-
mates of monetized benefits less the Primary Cen-
tral estimates of costs. The table also notes the ben-
efit/cost ratios implied by the benefit ranges.
The conclusion we draw from Table 8-4 is that,
given the particular data, models and assumptions
we believe are most appropriate at this time, our
analysis indicates that the benefits of the CAAA sub-
stantially exceed its costs. Furthermore, the results
of the uncertainty analysis imply that it is extremely
unlikely that the monetized benefits of the CAAA
over the 1990 to 2010 period could be less than its
costs. Looking at Titles I through V, the central
benefits estimate exceeds costs by a factor of four to
one, whether we are looking at annual or present
value measures, and the high estimate exceeds costs
by more than twice that factor (a ratio of nine or ten
to one). Using the Primary Low estimate of ben-
efits, the annual estimates of benefits in 2000 and
2010 are slightly less than the annual costs for that
year. The data also suggest that costs for criteria
Table 8-4
Summary Comparison of Benefits and Costs (Estimates in millions 1990$)
Titles I through V
Annual Estimates
2000 2010
Present Value
Estimate
1990-2010
Title VI
Present Value
Estimate
1990-2165
All Titles
Total Present
Value
Monetized Direct Costs:
Low3
Central
High3
$19,000
$27,000
Not Estimated
$180,000
Not Estimated
$27,000
$210,000
Monetized Direct Benefits:
Lowb
Central
Highb
$16,000
$71,000
$160,000
$26,000
$110,000
$270,000
$160,000
$690,000
$1,600,000
$100,000
$530,000
$900,000
$260,000
$1,200,000
$2,500,000
Net Benefits:
Low
Central
High
($3,000)
$52,000
$140,000
($1,000)
$93,000
$240,000
($20,000)
$510,000
$1,400,000
$73,000
$500,000
$870,000
$50,000
$1,000,000
$2,300,000
Benefit/Cost Ratio:
Lowc
Central
Highc
less than 1/1
4/1
more than 8/1
less than 1/1
4/1
more than 10/1
less than 1/1
4/1
more than 9/1
less than 4/1
20/1
more than 33/1
1/1
6/1
12/1
"The cost estimates for this analysis are based on assumptions about future changes in factors such as
consumption patterns, input costs, and technological innovation. We recognize that these assumptions introduce
significant uncertainty into the cost results; however the degree of uncertainty or bias associated with many of the
key factors cannot be reliably quantified. Thus, we are unable to present specific low and high cost estimates.
b Low and high benefits estimates are based on primary results and correspond to 5th and 95th percentile results
from statistical uncertainty analysis, incorporating uncertainties in physical effects and valuation steps of benefits
analysis. Other significant sources of uncertainty not reflected include the value of unqualified or unmonetized
benefits that are not captured in the primary estimates and uncertainties in emissions and air quality modeling.
0 The low benefit/cost ratio reflects the ratio of the low benefits estimate to the central costs estimate, while the high
ratio reflects the ratio of the high benefits estimate to the central costs estimate. Because we were unable to reliably
quantify the uncertainty in cost estimates, we present the low estimate as "less than X," and the high estimate as
"more than Y", where X and Y are the low and high benefit/cost ratios, respectively.
105
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
pollutant programs grow somewhat more rapidly
than benefits from 1990 to 2000, but that benefits
grow more rapidly from 2000 to 2010.
The estimates for Title VI indicate that benefits
well exceed costs, even at the low benefits estimate.
This conclusion holds despite the relatively high dis-
count rate used for the estimates in Table 8-4 (5 per-
cent) a value that most analysts would consider too
high for the long time period over which benefits of
this program are discounted (175 years).4 The total
estimates for all titles of the CAAA also indicate ben-
efits in excess of costs for the full range of primary
benefits.
The approach to premature mortality valuation
used 111 our primary estimates is a method that al-
lows us to aggregate the benefits of reducing mortal-
ity risks with other monetized benefits of the CAAA.
One of the great advantages of the benefit-cost para-
digm is that a wide range of quantifiable benefits can
be compared to costs to evaluate the economic effi-
ciency of particular actions. Some analysts suggest,
however, that presentation of the results of a cost-
benefit analysis may mask the key assumptions that
are made to quantify all benefits in monetary terms.
Another evaluative paradigm, cost-effectiveness
analysis, is sometimes suggested as further evidence
of whether the benefits of a regulatory program jus-
tify its costs. Cost-effectiveness analysis involves es-
timation of the costs per unit of benefit (e.g., lives
saved). This type of analysis is most useful for com-
paring programs that have similar goals, for example,
alternative medical interventions or treatments that
can save a life or cure a disease. They arc less readily
applicable to programs with multiple categories of
benefits, such as the CAAA, because the cost-effec-
tiveness calculation is based on quantity of a single
benefit category. In other words, we cannot readily
convert reductions in new cases of chronic bronchi-
tis, reduced hospital admissions, improvements in
visibility, and increased commercial timber and crop
yields to a single metric such as "lives saved." For
4 The primary central benefit-cost ratio for Tide VT using
a 3 percent discount rate is 44 to 1, higher than any of those
presented in Table 8-4 (see 'Table 8-6 below). In addition, the
ratio using a 2 percent discount rale, the rate used in the under-
lying RIA.S, is 75 to 1. See Appendix G ior more detail on the
sensitivity oi Tide VI benefits to the choice oi discount rale.
these reasons, we prefer to present our results in
terms of monetary benefits.
Despite the risks of oversimplification of ben-
efits, cautiously interpreted cost-effectiveness calcu-
lations may provide further evidence of whether the
costs incurred to implement the CAAA are a rea-
sonable investment for the nation. The most com-
mon cost-effectiveness metric, costs per life saved,
can be readily calculated from the information pre-
sented in this report. For example, we estimate the
total annual direct costs of implementation of Titles
I through V in 2010 to be approximately $27 bil-
lion. In exchange for this expenditure, in the year
2010 we avoid 23,000 cases of premature mortality
and gam estimated non-mortality benefits of about
$20 billion. We can generate a net cost per life saved
by subtracting from costs the total non-mortality
benefits, and then dividing by lives saved. For Titles
I through V, we estimate a net cost per life saved of
approximately $300,000 ($27 billion minus $20 bil-
lion divided by 23,000).3 Although we are also con-
cerned about many of the uncertain assumptions
required to generate cost per life-year saved estimates,
we include an estimate for illustrative purposes. For
the year 2010, the net cost per life-year saved esti-
mate implied by the primary central case results is
$23,000 per life-year ($7 billion divided by 310,000
life-years saved).6
Of
We can obtain additional insights into key as-
sumptions and findings of the present study through
further analysis of potentially important variables
and inputs. The estimated uncertainty ranges for
each endpomt category summarized in Table 8-1
reflect the measured uncertainty associated with two
aspects of the analysis: avoided physical effects (both
health and welfare benefits) and economic valuation
of benefits. In addition, in Chapter 3 we conduct
quantitative sensitivity analyses of key components
of die direct cost estimates. For many other aspects
of our analysis, however, including emissions esti-
= The illustrative calculations presented here do not reflect
discounting of the physical incidence of mortality.
6 Because of Agency concerns regarding discounting of
physical effects, the ratio presented here reflects undiscounted
life-years saved. II iuture years were discounted, the implicit
cost per liie-year saved would be significantly higher.
106
-------
Chapters: Comparison of Costs and Benefits
mates, air quality modeling, and unquantified cat-
egories of benefits, we are unable to conduct quanti-
tative analysis of uncertainty. Instead, we have at-
tempted throughout this report to identify and char-
acterize major sources of uncertainty — we present
the results of these efforts at the end of Chapters 2
through 7. In this section, we provide a summary
evaluation of the relative importance of key sources
of uncertainty.
Table 8-5 below provides a summary of both
quantified and unquantified sources of uncertainty
and our estimates of the impact of these sources of
uncertainty on the primary central estimates of ben-
efits and costs. The table covers seven major catego-
ries of uncertainties: measurement uncertainties in
physical effects and valuation components of the
benefits analysis; measurement uncertainties in esti-
mation of direct costs; alternative assumptions for
PM-rclatcd mortality valuation; alternative assump-
tions for PM-relatcd mortality risk; unquantified
sources of error in emissions and air quality model-
ing; and omissions of key benefits categories. The
table entries cover quantitative analyses of uncer-
tainty, characterization of unquantified uncertainty,
and the potential effect of alternative modeling para-
digms for costs and benefits. Additional treatment
of alternative paradigms is necessary because reason-
able people may disagree with our methodological
choices regarding these issues, and these choices
might be considered to significantly influence the
results of the study.
Of
As discussed previously in this chapter, we have
conducted quantitative uncertainty analysis of our
benefits estimates to reflect measurement error in
two key steps of the analysis: estimation of physical
effects and economic valuation. We present the re-
sults of our analysis in Figure 8-1 and Table 8-1 above.
The procedure used to generate these estimates is
well-suited to analysis of uncertainties where the
probability of alternative outcomes can be quantita-
tively characterized in an objective manner. For
example, most studies that estimate concentration-
response relationships report an estimate of the sta-
tistical uncertainty around the central estimate. Be-
cause many estimates are available for the value of
statistical life, wre can use the discrete distribution of
the best available estimates as a basis for quantita-
tively characterizing the probability of alternative
values. It is important to recognize, however, that
this procedure reflects only a portion of the range of
possible sources of uncertainty in our benefits esti-
mates. Other, nonqualified sources of uncertainty
must also be factored into conclusions about the ra-
tio of benefits to costs.
As part of our analysis of key contributors to
uncertainty in benefits estimates, wre also conducted
a sensitivity analysis to determine the physical ef-
fects estimation and economic valuation variables
with the greatest contribution to the quantified mea-
surement uncertainty range. We present the results
of this sensitivity analysis in Figure 8-2. In this sen-
Figure 8-2
Analysis of Contribution of Key Parameters to Quantified Uncertainty
i-H
| $250-
s $200 -
1 $150~
=
M $100-
o
| $0^
^ 951h %ile
<
i
Mean
> (
5th %ile
1
T
> O ffi fl 9 9
ii ii
H | £ | ^ | .§ | | .§ | | | g | |
^ 1 il il o|| i|i §11 l|
0
1,8
^ i— I 0)
P ^ T3
"^ O 'o
CH CH
0
107
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 8-5
Summary of Key Sources of Uncertainty and Their Impact on Costs and Benefits
Source of
Uncertainty
Measurement
error and
uncertainty in the
physical effects
and economic
valuation steps
Measurement
error and
uncertainty in
direct cost inputs
Value of
statistical life-
based estimates
do not reflect
age at death
Basis of estimate
of avoided
mortality from
PM exposure
Uncertainties in
Title VI health
benefits analysis
Uncertainties in
emissions and
air quality steps
Omission of
potentially
important
benefits
categories from
primary estimate
Description of Alternative Parameter
Inputs
Use a range of input assumptions to
reflect statistical measurement
uncertainty in concentration-response
functions, modeling of physical effects,
and estimation of economic values.
Most important input parameters are
value of statistical life and estimated
relationship between particulate matter
and premature mortality (see Chapters
5, 6, and 7).
Use alternative assumptions for key
input parameters for six of the highest
cost provisions. Conduct sensitivity
tests for each provision separately (see
Chapter 3, pages 30 to 32). As
discussed in Chapter 3 and in this
chapter, aggregation of provision-
specific results would be inappropriate.
Use estimates of the incremental
number of life-years lost from exposure
to ambient PM and a value of statistical
life-year as opposed to measuring
number of lives lost and a value of
statistical life (see Chapters 5 and 6).
The Dockery et al. study provides an
alternative estimate of the long-term
relationship between chronic PM
exposure and mortality (see Chapter 5).
Major uncertainties include: estimating
fatal cancer cases resulting from UV-b
exposure; not accounting for future
averting behavior; and not accounting
for future improvements in the early
detection and treatment of melanoma
(see Table 5-6).
Major uncertainties include:
underestimation of PM2 .5 emissions;
omission of changes in primary and
organic PM in eastern U.S.; emissions
estimation uncertainties in the western
U.S.; scarcity of PM25 monitors; and
lack of a fully integrated air quality and
emissions modeling system (see
Tables 2-5 and 4-7).
Non-quantified categories of impacts
summarized in Chapters 5 and 7.
Quantified but omitted categories
include household soiling, nitrogen
deposition, and residential visibility (see
Chapter 7).
Impact on Annual
Costs
None
High estimates for
some provisions are
$1 billion higher
than primary
estimate. Low
estimates are as
much as $2 billion
below primary
estimate
None
None
None
Uncertainties in
emissions estimates
affects some costs,
but net effect is
minor.
None
Estimates in 2010
Benefits
For Titles I through V,
effect of the use of
alternative input
assumptions ranges
from a $84 billion
decrease (5th
percentile) to a $160
billion increase (95th
percentile).
None
Decrease by $47
billion
Increase by $100 to
$150 billion
Not quantified, but net
effect is probably that
benefits estimates are
too high.
Not quantified, but net
effect is probably that
benefits estimates are
too low.
Increase by at least $8
billion, (does not
reflect unquantified
categories)
108
-------
Chapters: Comparison of Costs and Benefits
sitivity analysis, we hold constant all inputs to the
probabilistic uncertainty analysis except one — for
example, the economic valuation of mortality. We
allow that one variable to vary across the estimated
range of that variable's uncertainty. The sensitivity
analysis isolates the effect of this single source of un-
certainty on the total measured uncertainty in esti-
mated aggregate benefits. The first uncertainty bar
represents the range associated with the total mon-
etized benefits of the Clean Air Act, based on analy-
sis of quantifiable components of uncertainty, as
reported above. This range captures the multiple
measurement uncertainties in the quantified benefits
estimation. The rest of the uncertainty bars repre-
sent die quantified measurement uncertainty ranges
generated by single variables. As shown in Figure 8-
2, the most important contributors to aggregate quan-
tified measurement uncertainty are mortality valua-
tion and incidence, followed by chronic bronchitis
valuation and incidence.
Error Uncertainty in
Direct
As noted in Chapter 3, explicit and implicit as-
sumptions about changes in consumption patterns,
input costs, and technological innovation are cru-
cial to estimating the direct compliance costs of the
CAAA. For many of the factors contributing to
uncertainty, the degree and, in some cases, the di-
rection of the bias are unknown or cannot be deter-
mined. Uncertainties and sensitivities can be identi-
fied, however, and in many cases the potential mea-
surement errors can be quantitatively characterized.
We designed our sensitivity analyses of key input
parameters to provide a sense of the relative impor-
tance of various input parameters and assumptions
necessary to generate estimates of direct costs. The
sensitivity tests use ranges of input parameters that
include all reasonable alternative estimates that we
could identity.
The results indicate that the sensitivity of our
primary central cost estimates is not uniform across
provisions. Low and high estimates may vary by as
much as a factor of two. Unlike our quantitative
analysis of benefits, we do not assign probabilities
to the likelihood of alternative input parameters. In
our judgement, assignment of probabilities to these
alternative outcomes would be a largely subjective
task; we knowf of no objective means to develop these
probabilities. As a result, it would be inappropriate
simply to add up the array of low and the array of
high estimates to arrive at an overall range of uncer-
tainty around the central estimates, because it is un-
likely that a plausible scenario could be constructed
where all the estimates are concurrently either at the
high or low end of their individual plausible ranges.
A better interpretation of these results is that uncer-
tainty in key input parameters can have a significant
effect on the overall uncertainty of our estimates of
direct compliance costs and ultimately the net ben-
efits calculation.'
PM
on Life-years
The primary analytical results we present ear-
lier in this chapter assign the same economic value
to incidences of premature mortality regardless of
the age and health status of those affected. Although
this has been the traditional practice for benefit-cost
studies conducted within EPA, some argue this may
not be the most appropriate method for valuation
of premature mortality caused by PM exposure.
Some short-term PM exposure studies suggest that a
significantly disproportionate share of PM-related
premature mortality occurs among persons 65 years
of age or older. Combining standard life expectancy
tables with the limited available data on age-specific
incidence allows rough approximations of the num-
ber of life-years lost by those who die prematurely
as a result of exposure to PM or, alternatively, the
changes in lite expectancy of those who are exposed
to PM.
The ability to estimate, however roughly,
changes in age-specific life expectancy raises the is-
sue of whether available measures of the economic
value of mortality risk reduction can, and should,
be adapted to measure the value of specific numbers
•' Although the analysis conducted here is a direct cost analy-
sis, other sources of uncertainty would also need to be consid-
ered for a social cost: analysis. For example, forecasts of key
economic variables (e.g., interest rates), specification of produc-
tion functions, and the reliability of key supply and demand
elasticities are all important factors in social cost modeling that
contribute to measurement uncertainty. In addition, most cur-
rent, social cost, analyses assume that markets are currently oper-
ating under optimally efficient, conditions. Emerging literature
suggests that a full accounting of the social costs and efficiency
impacts of environmental regulations could also include an as-
sessment of the incremental costs that reflect existing market
distortions, such as those imposed by the current tax code. Our
assessment of uncertainties in direct cost estimates do not re-
flect these considerations.
109
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
of life-years saved.8 As stated in our retrospective
analysis, we have 011 occasion performed sensitivity
calculations that adjust mortality values for those
over age 65. Nonetheless, as discussed in Appendix
H, the current state of knowledge and available ana-
lytical tools do not conclusively support using a life-
years lost approach or any other approach which
assigns different risk reduction values to people of
different ages or circumstances. While we prefer an
approach which makes no valuation distinctions
based on age or other characteristics of the affected
population, we present alternative results based on
a VSLY approach below. The method used to de-
velop life years lost estimates is described briefly in
Chapter 5 and Appendix D. The method used to
develop VSLY estimates is described in Appendix
H.
The fourth row of Table 8-5 summarizes the ef-
fect of using a VSLY approach on results for 2010.
The results indicate that the choice of valuation meth-
odology significantly affects the estimate of the mon-
etized value of reductions in air pollution-related pre-
mature mortality. However, the downward adjust-
ment which would result from applying a VSLY ap-
proach in lieu of a VSL approach does not change
the basic conclusion of this study, since the central
estimate of monetized benefits of the CAAA still
substantially exceeds the costs of compliance.
We emphasize that the results of the VSLY ap-
proach to valuing avoided mortality benefits repre-
sent a crude estimate of the value of changes in agc-
specific life expectancy. These results should be in-
terpreted cautiously, due to the several significant
assumptions required to generate a monetized esti-
mate of life years lost from the relative risks reported
in the Pope et al., 1995 study and the available eco-
nomic literature. These assumptions include, but
are not limited to: extrapolation of die age distribu-
tion of the U.S. population in future years; assump-
tions about the age-specificity of the relative risk
reported by Pope et al., 1995; assumptions about the
life expectancy of different age groups, adjustment
8 This issue was extensively discussed during the Science
Advisory Board Council review of drafts of the retrospective
study. The Council suggested it would be reasonable and ap-
propriate to show PM mortality benefit estimates based on value
of statistical life-years (VSL\) saved as well as the value of statis-
tical life (VSL) approach traditionally applied by the Agency to
all incidences oi premature mortality. Consistent with SAB
Council review advice ior the present study, we apply the same
approach in lids analysis.
of the life years lost estimates by an appropriate lag
period (if any); assumptions about the age-specific-
ity of the lag period (if any); derivation of VSLY
estimates from VSL estimates; assumptions about the
variation in VSLY with age; and selection of an ap-
propriate rate at which to discount the lagged esti-
mates of life years lost. Changes in any of these
assumptions could significantly affect the VSLY ben-
efit estimate. For example, if we were to assume no
lag period for PM-related mortality effects instead
of the five-year lag structure described in Chapter 5,
VSLY benefit estimates would increase from $53
billion to |61 billion. The specific assumptions we
used in generating these results are discussed in Ap-
pendix H.
PM
the Dockery Study
As described in Chapter 5, we chose to use the
results of die Pope et al. (1995) study to estimate the
magnitude of the effect of ambient PM exposure on
the incidence of premature mortality. Alternative
estimates do exist in the literature, however. Al-
though we chose the Pope study because of its cov-
erage of the largest number of cities and other tech-
nical advantages, the Dockery et al. (1993) study
provides a credible and reasonable alternative to the
Pope study. The Dockery study used a smaller
sample of individuals in fewer U.S. cities than the
Pope study, but it features improved exposure esti-
mates, a slightly broader study population (includ-
ing adults aged 25 to 30), and a follow-up period
nearly twice as long as that used in the Pope study.
Use of the Dockery study in place of the Pope
study would substantially increase the benefits esti-
mate. As shown in the fifth row of Table 8-5, we
estimate that using the Dockery study estimates
would increase the annual central benefits estimate
by $100 to $150 billion, more than doubling the to-
tal annual benefits for Titles 1 through V and, in
turn, doubling the estimated benefit-cost ratio.
in Title VI
As discussed in Chapter 5 and Appendix G,
health benefits such as avoided mortality from mela-
noma and non-melanoma skin cancers constitute the
majority of monetized benefits resulting from Title
110
-------
Chapters: Comparison of Costs and Benefits
VI regulations on stratospheric ozone-depleting
chemicals. Estimates of avoided mortality from skin
cancer due to reduced UV-b exposure between 1990
and 2165 represent over 90 percent of the total health
benefits of Title VI. As a result, uncertainties re-
lated to avoided mortality estimation under Title VI
represent key uncertainties for our overall CAAA
benefits estimate. Three main areas of uncertainty
are important for our avoided mortality estimates
for Title VI: dose-response relationships; predicting
averting behavior; and predicting future medical
advancements.
Because the literature on the relationship be-
tween exposure to ultraviolet rays and melanoma
and non-melanoma mortality is not as well devel-
oped as that for other health effects, the dose-response
functions for both of these endpoints are character-
ized by significant uncertainty. The association of
UV-b exposure with melanoma is controversial, al-
though studies suggest that sunlight exposure is a
major environmental risk factor for melanoma. If
one assumes that a causal relationship exists between
UV-b rays and melanoma, uncertainty still remains
about three aspects of the nature of the dose-response
relationship. Specifically, the relative contribution
of different wavelengths of light to melanoma de-
velopment, the critical exposure period (e.g., acute,
intermittent, or chronic), and the existence (and
length) of a latency period between UV exposure
and disease are all unclear. The effect of the first
two uncertainties on our results cannot be deter-
mined from available information. If a significant
latency period exists, then the third uncertainty may
indicate that our analysis, which does not include a
latency period, overestimates avoided melanoma
mortality benefits. Because limited data on non-
melanoma mortality precluded the development of
a dose-response function for this endpoint in the
current analysis, our estimate of non-melanoma skin
cancer mortality resulting from UV-b exposure is
calculated indirectly, by assuming the mortality rate
is a fixed percentage of non-melanoma incidence.
New data on the death rate for non-melanoma skin
cancer may significantly influence this mortality es-
timate.
Our analysis of avoided mortality also does not
incorporate adjustments for future increases in avert-
ing behavior (i.e., efforts by individuals to protect
themselves from UV-b radiation ). Our estimates
rely on epidemiological studies that incorporate
averting behavior as currently practiced. However,
if people would react to increased skin cancer risk in
the future by applying sun screen more frequently,
spending more time indoors or otherwise reducing
their UV-b exposure, then our estimate of avoided
mortality would significantly overestimate Title VI
benefits. It is not certain, though, that individuals
will pursue such behavior, and studies show that
those engaging in averting behavior may also alter
their behavior in ways that may increase exposure
or risk, counteracting the benefits of averting be-
havior. For example, a recent study of young Euro-
peans by Autier et al. (1999) found that the use of
high sun protection factor (SPF) sun screen is associ-
ated with increased frequency and duration of sun
exposure.
Finally, our analysis does not adjust estimates of
future mortality for possible advances in medical
technology that could lead to earlier detection and
more effective treatment of melanomas. Such ad-
vancements could significantly reduce the expected
future melanoma mortality, and by not adjusting for
such developments, we may be overestimating
avoided melanoma mortality. However, future re-
search may also identify additional adverse human
health outcomes associated with UV exposure that
we have not considered in this analysis, resulting in
an underestimate of Title
benefits.
in
and Air Quality
The emissions estimates presented in this analy-
sis are a critical component of the overall analysis.
As the starting point for both costs and benefits, they
provide a consistent basis for evaluating the economic
efficiency of the CAAA. Characterizing emissions
can be very difficult, however, particularly for those
source categories where emissions monitoring data
are sparse or nonexistent. In general, all our emis-
sions estimates are affected by three major sources
of uncertainty: estimation of the base-year inven-
tory, prediction of the growth in pollution-generat-
ing activity, and assumptions about future-year con-
trols.
Base-year emissions were estimated using emis-
sions factors that express the relationship between a
particular human/industrial activity and the level of
111
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
emissions. The accuracy of base-year emissions esti-
mates vanes from pollutant to pollutant, depending
largely on how directly the selected activity and
emissions correlate. We likely estimated 1990 SO,
emissions with the greatest precision. Sulfur diox-
ide emissions are generated during combustion of
sulfur-containing fuel and are directly related to fuel
sulfur content. In addition, we were able to verify
these estimates through comparison with Continu-
ous Emission Monitoring (CEM) data. As a result,
we were able to accurately estimate SO, emissions
using emissions factors based on data on fuel usage
and fuel sulfur content. Nitrogen oxides are also a
product of fuel combustion, allowing us to estimate
emissions of this pollutant using the same general
technique used to estimate SO, emissions. However,
the processes involved in the formation of NO
during combustion are more complicated than those
involved in the formation of SO,; thus, our NOx
emissions estimates arc more variable and less cer-
tain than SO, estimates.
Volatile organic compounds, like SO, and NOx,
are products of fuel combustion; however, these
compounds are also a product of evaporation. To
estimate evaporative emissions of this pollutant we
used emissions factors that relate changes in emis-
sions to changes in temperature. Because future
meteorological conditions are difficult to predict,
the uncertainty associated with forecasting tempera-
ture influences the uncertainty in our VOC emis-
sions estimates. The likely significance of this un-
certainty, in terms of its impact on the overall mon-
etary benefit present in this analysis, is probably
minor.
Of particular importance, however, are uncer-
tainties that affect the estimation of future year emis-
sions of particulate matter and secondarily formed
PM precursors. In this analysis we estimated primary
PM,g emissions based on unit emissions that may
not accurately reflect the composition and mobility
of particles. The ratio of crustal to carbonaceous
particulate material, for example, likely is high as a
result of overestimation of the fraction of crustal
material, primarily composed of fugitive dust, and
underestimation of the fraction of carbonaceous
material. Because the CAAA have a greater impact
on emissions sources that generate carbonaceous par-
ticles (mobile sources) than on sources that mainly
emit crustal material (area sources), we likely under-
estimate the impact of the CAAA on reducing PM,
thereby reducing monetary benefits estimates. The
uncertainty associated with estimating the partition
of PM emissions components could conceivably
have a major impact on the net benefit estimate.
Compared to secondary PM25 precursor emissions,
however, changes in primary PM emissions have a
relatively small impact on PM,5 related benefits.
Our future-year control assumptions are also a
source of uncertainty. Despite our efforts to mini-
mize this uncertainty, w7hether each of the Post-
CAAA controls will be adopted, whether Post-
CAAA control programs will be more or less effec-
tive than estimated, and whether unanticipated tech-
nological shifts will reduce future-year emissions are
all unknown. For example, the Post-CAAA scenario
includes implementation of a region-wide NO con-
trol strategy designed to regulate the regional trans-
port of ozone. However, the control program as-
sumed under the Post-CAAA scenario may not re-
flect the NOx controls that are actually implemented
in a regional ozone transport rule.
In addition to potential inaccuracies in the emis-
sions inventories used as air quality modeling inputs,
there are at least three sources of air quality model-
ing uncertainty that may have a major effect on the
precision and accuracy of our projected changes in
air quality. First, we estimate changes in PM con-
centrations in the eastern U.S. based exclusively on
changes in die concentrations of sulfate and nitrate
particles. By not accounting for changes in organic
and primary particulate fractions, we likely under-
estimate the impact of the CAAA on PM concen-
trations. Second, by using separate air quality mod-
els for individual pollutants and different geographic
regions, as opposed to a single integrated model, we
were unable to fully capture the interaction among
air pollutants or reflect transport of pollutants or
precursors across the boundaries of the models cov-
ering the western and eastern states. Third, the lack
of a well-developed modeling network for PM
means we must estimate monitored concentrations
of this pollutant based on PM10 monitor estimates.
The direction and magnitude of bias these limita-
tions impose 011 net benefits estimate presented in
this analysis can not be determined based on current
information.
Some model-related uncertainties, however, may
be mitigated because this analysis uses the air qual-
112
-------
Chapters: Comparison of Costs and Benefits
ity modeling results in a relative, not absolute, sense.
We focus on the change in air quality between the
Pre- and Post-CAAA scenarios and not on the am-
bient concentrations projected by the individual
models themselves. Therefore, uncertainties that
affect a model's ability to accurately predict the rela-
tive change in concentration of a pollutant from one
scenario to another are more important in the con-
text of this study than those that affect only the ab-
solute model results. In addition, as summarized in
the previous chapters, most of the uncertainties in
emissions estimation and air quality modeling con-
tribute to a conservative bias in our benefits results.
When faced with alternative approaches to emissions
and air quality modeling, we made explicit attempts
to choose parameters, assumptions and modeling-
strategies that would tend to understate benefits.
of Potentially
As described in Chapters 5 through 7 above, and
in more detail in Appendix H, the primary estimate
reflects application of a strict set of criteria for inclu-
sion of monetized benefits categories. For example,
estimates of the value of improved visibility in U.S.
residential areas indicate a positive value for this ser-
vice flow, but the best available residential visibility
estimates rely on an unpublished study of values in
the eastern U.S. Although our physical effects analy-
sis indicates significant visibility improvements in
all regions of the U.S., our application of the results
of the economic valuation literature reflect a con-
servative approach to valuation of improved visibil-
ity in the U.S. While we believe our conservative
inclusion criteria for the primary benefits reflects
the greater uncertainty in measuring some economic
values, we also believe that the statutory language of
section 812 clearly warns against the practice of as-
suming a default value of zero for demonstrated cat-
egories of benefits. Therefore, the last row of Table
8-5 presents the effect of using a somewhat more
inclusive set of criteria for accepting benefits trans-
fer-based economic values. In this alternative case,
we included estimates for improved residential vis-
ibility, displaced costs from reduced airborne nitro-
gen loadings to estuaries, and reduced expenditures
for household soiling (which are not included in any
form in the primary estimate).
In addition to these quantified but omitted cat-
egories of benefits, there is a wide range of benefits
of the CAAA that we can identify but cannot quan-
tify. We present summaries of unquaiitified health
effects in Chapter 5 (Tables 5-1 and 5-5) and
unquaiitified ecological and welfare effects in Chap-
ter 7 (Tables 7-5 and 7-9). Two of the most impor-
tant omissions, in our judgement, are the lack of any
quantified estimates for the health benefits of air
toxics control and the omission of the systemic and
long-term ecological effects of mercury and other
persistent air pollutants. The importance of these
two categories of effects are discussed in Chapters 5
and 7, respectively.
In some instances, the choice of discount rate
can have an important effect on the results of a ben-
efit-cost analysis; for example, when the distribution
of costs and benefits throughout the time period are
very different from one another. In this assessment,
the discount rate affects annualized costs (i.e., amor-
tized capital expenditures), and the discounting of
all costs and benefits to 1990. Table 8-6 summarizes
the effect of alternative discount rates on the Pri-
mary Central estimate results of this analysis. The
estimates we present show that altering the discount
rate has only a small effect on annual cost and ben-
efit estimates. In part, this is due to limitations in
our ability to conclusively identify costs as annual-
ized capital expenditures or annual operating costs
in the underlying estimates. As described in Chap-
ter 3, about $3 billion (or roughly 10 percent) of the
2010 estimate is annualized capital costs. Varying
the discount rate, which we also use to represent the
cost of capital, affects only this component of costs.
The benefits estimates that employ a discount rate
include the mortality estimate, where it is used as
part of our valuation of the lag effect of PM mortal-
ity, and the chronic asthma value, where we use a
discount rate to develop a lump-sum value for avoid-
ance of incidence from an annual payment value in
the underlying literature.
Not surprisingly, the effect of discount rates on
the net present value benefit calculations is greater.
Nonetheless, the estimates we present in Table 8-6
show that varying the discount rate assumption also
does not change our overall conclusion that the ben-
efits of the CAAA exceed its costs.
113
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table 8-6
Effect of Alternative Discount Rates on Primary Central Estimates
(Estimates in million 1990$)
Discount Rate Assumption
3% 5% 7%
Annual Costs in 2010:
Titles I through V $26,600 $26,800 $26,900
Annual Benefits:
Titles I through V $110,000 $110,000 $107,000
Present Value of Costs:
Titles I through V $230,000 $180,000 $140,000
Title VI $43,000 $27,000 $20,000
Present Value of Benefits:
Titles I through V $890,000 $690,000 $520,000
Title VI $1,900,000 $530,000 $240,000
Cumulative Net Benefits:
Titles I through V $650,000 $510,000 $380,000
Title VI $1,860,000 $500,000 $220,000
Benefit/Cost Ratio:
Titles I through V 4/1 4/1 4/1
Title VI 44/1 20/1 12/1
114
-------
Emissions Analysis
The prospective analysis examines the emissions
of eight air pollutants: volatile organic compound
(VOC), nitrogen oxides (NO,), sulfur dioxide (SO2),
carbon monoxide (CO), participate matter of ten
microns or less (PM10), participate matter with an
aerodynamic diameter of 2.5 microns or less (PM25),
ammonia (NH3), and mercury (Hg). Changes in
emissions of these pollutants were projected based on
two emissions control scenarios: a) pre-1990 Clean
Air Act Amendments ("Pre-CAAA") scenario
assuming that no additional controls would be
implemented beyond those that were in place when
the CAAA were passed; and b) "Post-CAAA"
scenario incorporating the effects of controls
authorized by the 1990 Amendments. Comparison of
the resulting projections revealed the predicted impact
of the CAAA on emissions. In addition, these
estimates provided the basis for the subsequent cost
estimation and air quality modeling steps of the
prospective analysis.
This appendix summarizes the Pre- and Post-
CAAA emissions estimates for each of the major
sources of VOC, NOX, SO2, CO, PM10, PM25, NH3,
and Hg and describes the methodology for the
Agency's projections. EPA based its Pre- and Post-
CAAA emissions estimates on projections from 1990
base year emissions estimates. For all of the
pollutants, except particulate matter and mercury,1 the
Agency selected emissions levels taken from Version
3 of the National Particulates Inventory (NPI) to
serve as the baseline. For both PM10 and PM25,
however, EPA updated NPI estimates to reflect the
emissions from the National Emission Trends (NET)
inventory. Once EPA finalized the base year levels,
the Agency projected 1990 emissions to 2000 and
2010 under both the Pre- and Post- CAAA scenarios.
At the time the emissions data base selection was
made, the NPI was the most comprehensive source of
emission estimates of all criteria pollutant emissions.
Other available data sets, such as that available from
the Ozone Transport Assessment Group (OTAG),
were considered, but not selected because they were
limited to ozone precursor emissions.
EPA estimated future emissions for all major
source categories: industrial point sources, utilities,
nonroad engines/vehicles, motor vehicles, and area
sources. To make these projections, for all but utility
sources, the Agency relied on emissions analysis that
incorporated growth forecasts and future year control
assumptions about rule effectiveness and control
efficiency. In this analysis EPA projected growth
largely based on anticipated changes in economic
activity, and treated the rule effectiveness and the rate
of control efficiency as the key differences between
the Pre- and Post-CAAA scenarios.
EPA used the Integrated Planning Model (IPM)
to estimate utility emissions. With this optimization
model EPA forecasts emissions for the 48 contiguous
States and the District of Columbia. All existing
electric power generation units are covered in the
model, as well as independent power producers and
other cogeneration facilities that sell wholesale power,
if they were included in the North American Electric
Reliability Council (NERC) data base for reliability
planning. The model considers future capacity
additions by both utilities and independent power
producers. In addition, this model is capable of
producing baseline air emission forecasts and
estimates of air emissions levels under various control
scenarios at the national, and NERC region and
subregional, level. A full explanation of the IPM
model and the assumptions EPA used for this
prospective analysis may be found in Analyzing
Electric Power Generation under the CAAA. EPA,
July 1996.2
This appendix first provides an overview of Pre-
and Post-CAAA scenario development. It discusses
A separate methodology was used to estimate mercury
emissions. Mercury emissions are discussed independently
beginning on page A-48 of this appendix.
This document was updated in March 1998 to describe
model refinements made for IPM Version 7.1 and the latest base
case forecasts. (EPA, Office of Air and Radiation, Analyzing
Electric Power Generation under the CAAA. March 1998.)
A-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
key factors influencing the 2000 and 2010 estimates,
such as the timing of modeling decisions with respect
to control programs and available data. In the
following section, there is a comparison of the
prospective analysis base year inventory and emissions
projections with other existing data. Subsequent
sections address the influence of Title 1 progress
requirements and the major sources of uncertainty,
and provide detailed breakdowns of emissions
projections for the five major source categories
responsible for air pollution: industrial point, utility,
nonroad, motor vehicle, and area sources. For each of
these source categories this appendix provides: (1) an
overview of the approach used to make the emissions
estimates; (2) a discussion of how base year emissions
levels were determined; (3) an explanation of how
growth projections were determined; (4) an outline of
the assumptions made for the control scenarios; and
(5) a summary of the emissions estimates for both the
Pre- and Post-CAAA scenarios.
Scenario Development
EPA projected emissions by adjusting 1990 base
year emissions to reflect projected economic activity
levels in 2000 and 2010, and applying future year
control assumptions. The resulting estimates
depended largely upon three factors: how the base
year inventory was selected, what indicators were used
to forecast growth, and what specific regulatory
programs were incorporated in the Pre- and Post-
CAAA scenarios. These three factors are addressed in
Tables A-l through A-3. Table A-l highlights the
approach EPA used to establish the base year
inventory. The indicators the Agency relied on to
forecast growth and predict future activity levels,
along with the analytical approach EPA used to
project emissions, are shown in Table A-2. The Pre-
and Post-CAAA regulatory scenarios are summarized
in Table A-3.
Of the factors that influence EPA's emissions
projections for 2000 and 2010, the most significant is
the suite of air pollution regulations and programs the
Agency incorporated in the Pre- and Post-CAAA
scenarios. For the Pre-CAAA scenario, air pollution
controls are frozen at their 1990 levels; only standards
and initiatives implemented prior to the CAAAs are
included. The Post-CAAA scenario, in addition to the
measures contained in the Pre-CAAA scenario,
incorporates emission controls associated with the
1990 Amendments. Due to the necessity of
developing emissions scenarios early in the
prospective analysis process, the exact provisions of
some regulatory programs could not be foreseen. For
example, decisions about how to translate the OTAG
recommendations into regional NOX control
requirements had not been made, so estimates were
made on affected sources, geographic coverage, and
control levels.
EPA included in the Post-CAAA scenario:
• Title I VOC and NOX reasonably available
control technology (RACT) and reasonable
further progress (RFP) requirements for
ozone nonattainment areas (NAAs);
• Title II motor vehicle and nonroad
engine/vehicle provisions;
• Title III 2- and 4-year maximum achievable
control technology (MACT) standards;
• Title IV SO2 and NOX emissions programs
for utilities;
• Title V permitting system for primary sources
of air pollution; and
• Title VI emissions limits for chemicals that
deplete stratospheric ozone.
This scenario also assumes the implementation of
a region-wide NOX cap and trade system for the entire
OTAG domain3 and a similarly designed trading
program for the Ozone Transport Region (OTR) that
is consistent with Phase II of the Ozone Transport
Commission (OTC) NOX Memorandum of
Understanding (MOU). For motor vehicles, emission
reductions associated with a 49-State low emission
vehicle (LEV) program were also included in the Post-
CAAA scenario. A more detailed outline of the
controls included in both the Pre- and Post-CAAA
scenarios is provided in Table A-3.
The NOS control program incorporated in the Post-CAAA scenario
may not reflect the NOK controls that are actually implemented in a
regional ozone transport rule.
A-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-1
Base Year Inventory - Summary of Approach
Sector
Analysis Approach/Data Sources
Industrial Point
Sources
Utilities
1985 National Acid Precipitation Assessment Program (NAPAP) emissions
inventory grown to 1990 based on historical Bureau of Economic Analysis (BEA)
earnings data.
PM10 emissions based on total suspended particulate (TSP) emissions and
particle-size multipliers.
The 1990 utility emission estimates are from the 1990 NPI.
Nonroad Nonroad Engines/Vehicles (VOC, NOX, CO, PM10): 1991 Office of Mobile Sources
(QMS) Nonroad Inventory.
Nonroad Engines/Vehicles (SO2) and Aircraft, Commercial Marine Vessels,
Railroads: 1985 NAPAP grown to 1990 based on historical BEA earnings data.
Motor Vehicles
Federal Highway Administration (FHWA) travel data, MOBIL5a/PART5 emission
factors.
Area Sources 1985 NAPAP inventory grown to 1990 based on historical BEA earnings data and
State Energy Data Systems (SEDS) fuel use data; emission factor changes for
selected categories.
A-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-2
Analysis Approach by Major Sector
Sector
Growth Forecast
Analysis Approach
Industrial 1995 BEA Gross State Product (GSP)
Point Projections by State/Industry.
VOC, NOX — Emission Reduction and Cost
Analysis Model (ERCAM): applies BEA
growth projection to base year emissions and
applies future year controls as selected by the
user.
PM10, SO2, CO — While no formal model
exists, the same basic approach applied in
ERCAM was used for these pollutants.
Utilities Projections of heat input by unit based
on National Electric Reliability Council
(NERC) data, price and demand
forecasts, and technology
assumptions.
SO2, NOX — Integrated Planning Model (IPM).
VOC, PM10, CO — Base year emissions rates
or AP-42 emission factors applied to IPM
projected heat input by unit.
Nonroad 1995 BEA GSP and Population
Projections by State/Industry.
VOC, NOX — ERCAM.
PM10, SO2, CO — ERCAM approach (no
formal model).
Motor MOBILE Fuel Consumption Model
Vehicles (FCM) National Vehicle Miles
Traveled (VMT) Projection Scaled to
Metropolitan/REST-of-State Areas by
Population.
NOX, VOC, CO — ERCAM: applies MOBIL5a
emission factors to projected VMT by month
and county/vehicle type/roadway
classification.
PM10, SO2 — PART5 emission factors applied
to projected VMT.
Area 1995 BEA GSP and Population
Projections by State/Industry, and
USDA Agricultural Projections.
VOC, NOX —ERCAM.
PM10, SO2, CO — ERCAM approach (no
formal model).
A-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-3
Projection Scenario Summary by Major Sector
Sector
Pre-CAAA
Post-CAAA*
Industrial
Point
RACT held at 1990 levels.
VOC and NOX RACT for all NAAs (except NOX waivers).
New control technique guidelines (CTGs) (VOC).
0.15 pounds per million British thermal unit (Ibs/MMBtu)
OTAG-wide NOx cap on fuel combustors of 250 MMBtu
per hour or above(NOx).
OTAG Level 2 NOX controls across OTAG States (NOX).
2- and 4- year MACT standards (VOC).
Ozone Rate-of-Progress (3 percent per year)
requirements (further reductions in VOC).
Utilities 250 ton prevention of
Significant Deterioration
(PSD) and New Source
Performance Standards
(NSPS) held at 1990 levels.
RACT and New Source
Review (NSR) held at 1990
levels.
Title IV SO2 emission allowance program (SO2).
Title IV Phase I and Phase II emission limits for all
boiler types (NOX).
250 ton PSD and NSPS.
RACT and NSR for all non-waived (NOX waiver) NAAs
(NOX).
Phase II of the Ozone Transport Commission (OTC)
NOx memorandum of understanding (MOD) (NOX).
0.15 Ibs/MMBtu OTAG-wide seasonal NOX cap with
banking/trading (NOX).
Nonroad Controls (engine standards)
held at 1990 levels.
Federal Phase I and II compression ignition (Cl) engine
standards (NOX, PM).
Federal Phase I and II spark ignition (SI) engine
standards (VOC, CO, NOX).
Federal locomotive standards (NOX, PM).
Federal commercial marine vessel standards (NOX).
Federal recreational marine vessel standards (VOC,
NOX).
Motor Federal Motor Vehicle Control
Vehicles Program (FMVCP) — engine
standards set prior to 1990.
Phase 1 Reid vapor pressure
(RVP) limits.
I/M programs in place by
1990.
Tier 1 tailpipe standards (Title II) (VOC, NOX).
49-State LEV program (Title I) (VOC, NOX, CO).
Phase 2 RVP limits (Title II) (VOC).
I/M programs for ozone and CO NAAs (Title I) (VOC,
NOX, CO).
Federal reformulated gasoline for ozone NAAs (Title I)
(VOC, NOX, CO).
California LEV (California only) (Title I) (VOC, NOX,
CO).
California reformulated gasoline (California only) (Title
I) (VOC, NOX, CO).
Diesel fuel sulfur content limits (Title II) (SO2, PM).
Oxygenated fuel in CO NAAs (Title I) (CO).
A-5
-------
Table A-3
Projection Scenario Summary by Major Sector
Sector Pre-CAAA Post-CAAA*
Area Controls held at 1990 levels. VOC and NOX RACT requirements.
New CTGs (VOC).
2- and 4- year MACT standards (VOC).
Ozone Rate-of-Progress (3 percent per year)
requirements (further reductions in VOC).
PM NAA controls (PM).
Onboard vapor recovery (vehicle refueling) (VOC).
Stage II vapor recovery systems (VOC).
*Also includes all Pre-CAAA measures.
A-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Decisions about which control programs to
include in the Post-CAAA scenario and how their
emission reductions were modeled, were made during
the summer of 1996. While some adjustments to the
emission projection methods were made since that
time in response to review comments, opportunities
were not available to revise the emission projections
once air quality modeling was initiated using these data
sets. The result is that there are some differences
between the emissions modeling practices being used
currently in EPA regulatory analyses and those
reflected in the Section 812 Prospective. As examples,
the VOC emission reductions of 7 and 10 year MACT
standards are not included in the Prospective, nor are
CO, NO^ SO2, and PM emission benefits of any
MACT standards. Similarly, the NOX State
Implementation Plan (SIP) Call affects 22 States, not
37, and OTAG Level 2 controls are not applied to all
non-electrical generating units (EGUs) NOX sources,
only to cement kilns and internal combustion engines.
Comparison of the
Base Year Inventory and
Emissions Projections
with Other Existing Data
Comparison of the emissions estimates in the
prospective analysis to historical emissions estimates
drawn from EPA's National Air Pollutant and
Emissions Trends reports (hereafter referred to as
Trends) can provide a check on the reasonableness of
our emissions inventories. In addition, comparison of
emissions projections from the prospective analysis
with those of the Grand Canyon Visibility Transport
Commission (GCVTQ study of western regional haze
(Radian, 1995; Science & Policy Associates, 1995;
Argonne, 1995) provides an initial test of the
sensitivity of emissions projections to baseline
inventories and growth assumptions. Analysis of PM
emissions and comparison of estimated PM data with
observed PM data presented in the 1997 National Air
Quality and Emissions Trends Report (EPA, 1998a)
also helps evaluate the prospective study's emissions
estimation methods.
Post-CAAA Emissions Estimates and
EPA Trends Data
EPA publishes annual National Air Pollutant
Emission Trends reports that contain estimates of
historical trends in emissions of VOC, NO,, SO2, CO,
PM10, and lead (Pb). Comparison of the Trends (EPA,
1997a) and Post-CAAA estimates reveals that from
1990 to 1995, VOC, NOX, SO2, CO, and PM10
emissions figures from both are similar.
Figures A-l through A-5 display the Pre-CAAA,
Post-CAAA, and Trends, emissions estimates for
VOC, NO,, SO2, CO, and PM10 respectively.
Although the Post-CAAA and Trends emissions trends
are comparable for all five of these pollutants, there
are several instances where there are differences
between the estimates from these two different
sources. VOC emissions are highly variable from year
to year. To illustrate this fluctuation, annual Trends
estimates, and five year Trends increments, are
provided in Figure A-l for the years 1985 to 1996.
Although the 1990 Trends estimate is lower than the
prospective analysis' base year inventory, the general
emissions trend projected under the Post-CAAA
scenario is similar to that represented by the historical
EPA estimates .4
4VOC emissions estimates not only fluctuate from year to
year, but also from Trends report to Trends report. Had the Trends
report published in October of 1996, a year earlier, been used as
the data source for Figure A-l, the 1990 Trends VOC estimate
would have been higher, not lower, than the prospective analysis'
base year inventory.
A-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure A-1. Comparison of Pre-CAAA, Post-CAAA, and Trends VOC
Emissions Estimates
• Pre-CAAA
Post-CAAA
- Trends
Year
As shown in Figure A-2, the 1995 Trends estimate
for NOX emissions exceeds the Post-CAAA NO
projection. The primary influence on NOX emissions
in the mid-1990s was the requirement to install RACT
on major stationary source NOX emitters in certain
ozone nonattainment areas, and the Northeast OTR.
Because the EPA Trends report is still in the process of
incorporating the State's 1996 periodic emission
inventories in the NET data base, it is believed that
the Trends values in Figure A-2 do not capture all of
the NOX emission reductions that have occurred to
date. When these State data are incorporated in
Trends, it is expected that the 1990 to 1996 NOX
emissions trend line will more closely parallel the Post-
CAAA estimates.
Comparison of SO2 Post-CAAA and Trends
estimates based upon the profiles plotted in Figure A-
3 shows that the 1995 Trends estimates are somewhat
lower than the corresponding prospective projection
line. This is because the Trends profile reflects the
sudden reduction in SO2 emissions that resulted from
the implementation of the acid rain SO2 trading
program in the early 1990's. While this reduction is
incorporated in the Post-CAAA scenario, it is not
captured until the year 2000 emissions projections;
plotting the corresponding trend line has the effect of
distributing these early reductions over the entire
decade in constant annual increments. Between 1995
and 2000 the actual rate of SO2 reduction will almost
certainly slow so that by the year 2000 Post-CAAA
and Trends estimates will be much more comparable
than the 1995 levels depicted in Figure A-3.
Post-CAAA and Trends emissions estimates for
CO (Fig. A-4) and PM10 (Fig. A-5) respectively are
similar for the years 1990 and 1995. For PM10, the
values used to develop the Pre- and Post-CAAA, and
Trends profiles were adjusted to eliminate the influence
of wind erosion, a natural source of PM10 that can
cause significant fluctuation in emission estimates
from year to year. Even though this source is
A-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
controlled for, there is still significant variability in the
yearly historical PM10 emissions estimates. To capture
this variability in the Trends estimates, instead of
providing estimates at five year increments, annual
Trends emissions levels are displayed in Figure A-5.
For the years since the adoption of the 1990 CAAA,
both the Trends estimates and the Post-CAAA scenario
projections show PM10 emissions remaining at the
same relative level. The drastic drop in the Trends line
from 1989 to 1990 is the result of a change in
methodology used to calculate PM10 emissions.
Figures A-l through A-5 contain Trends estimates
for 1980 and 1985 in addition to the 1990 and 1995
values primarily discussed above. This information is
included to provide a broader picture of actual
emissions levels over the last 15 years and to show
how the general historical trends in emissions
compare to the projected future trends under both the
Pre- and Post-CAAA scenarios. Close comparison of
pre-1990 Trends estimates with 1990 to present Trends
estimates and prospective analysis projections,
however, has the potential to be misleading.
Beginning in 1990 there was a significant change in the
methodology used to estimate Trends emissions. The
1980 and 1985 figures presented here are intended
only for general comparison.
A-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
O
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure A-4. Comparison of Pre-CAAA, Post-CAAA, and Trends
CO Emissions Estimates
w .2 60
l i-
o 40 +
20
0
H—I-
C&
Year
H—h
- Pre-CAAA
Post-CAAA
• Trends
Figure A-5. Comparison of Pre-CAAA, Post-CAAA, and Trends
PMio Emissions Estimates
Tn
Emissions in Short Tons
(Millions)
-^ -^ NJ NJ C
D Ol O Ol O Ol C
*-+-+-*^
Q Q Q
Year
— •— Pre-CAAA
Post-CAAA
—A — Trends
A-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Prospective Analysis and GCVTC
Emissions Estimates
The GCVTC conducted, for the Western States,
an air pollution analysis that projected emissions for
selected pollutants from 1990 base year levels for the
year 2000 and every tenth subsequent year up to 2040.
Comparison of the 2000 and 2010 GCVTC study and
prospective analysis Post-CAAA scenario projections
for the Western States indicate that although their
estimates for NOX, SO2, and PM25 are somewhat
different, these differences arise from the use of
different baseline inventories in the two studies, or
from specific regional reductions not incorporated in
the prospective study scenarios (Table A-4).
NOX and SO2 base year figures in the prospective
analysis are approximately 10-15 percent higher than
the corresponding estimates in the GCVTC study.
The difference is most likely the result of the separate
inventories that are relied upon by the two analyses to
develop their respective base year emissions levels.
Version 3 of the NPI, the primary source of base year
emissions data for the prospective analysis, is largely
derived from 1985 emissions figures that are adjusted
to 1990 levels using BEA growth projections. This
inventory does not capture the effect of new controls
and technology change on emissions between 1985
and 1990. The GCVTC base year estimates for NOX
and SO2, based primarily on State provided point
source emissions figures from one of the years 1990
to 1992, however, incorporate these effects. As a
result, 1990 emissions estimates in the GCVTC study
are lower than those in the prospective analysis.
Due to the difference in the two studies' base year
NOX estimates, their projected absolute levels of NOX
emissions also differ. Both studies, however, estimate
that NOX pollution will decrease at a similar rate from
1990 levels. The prospective analysis Post-CAAA
scenario shows a 16 percent drop by the year 2010,
while the GCVTC estimates a 17 percent reduction.
SO2 projections for the two studies are not
characterized by similar percentage changes in
emissions. In fact, under the Post-CAAA scenario the
prospective analysis estimates an increase in SO2
emission from 1990 to 2010 of about 15 percent,
while the GCVTC study shows roughly an 11 percent
decrease over this same time period. The reason for
this disparity is that only the GCVTC emissions
forecasts take into account specific modernization
plans for the Kennecott-Utah Copper Corporation
which are predicted to lower future SO2 emissions in
the West by approximately 30 thousand tons per year
(tpy), as well as, a regional electric utility cut of
roughly 80 thousand tpy. Together, these anticipated
reductions account for the bulk of the difference
between the two studies future year estimates.
Emissions figures for PM25 are the source of the
largest disparity between the prospective analysis and
the GCVTC study. In general, emissions estimates are
roughly 40 percent lower in the former than in the
latter. This is due to a difference in the inventories
used to develop the 1990 base year PM25 estimates.
While the more recent prospective analysis relied on
NPI data that was updated to incorporate NET
estimates that reflected revisions to PM25 emissions
factors for fugitive dust, the GCVTC study was
conducted prior to the lowering of these factors. As
a result, the PM25 estimates in the GCVTC study are
considerably higher than those in the prospective
analysis. The percent change in PM25 emissions from
1990 to 2010, however, is similar in the two studies
(approximately 10 percent in prospective analysis and
approximately 13 percent in GCVTC study).
A-12
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-4
Comparison
of Emissions: Prospective
1990
Post-CAA
NOX
SO2
PM2.5
3,303,536
1,245,439
1,701,869
NOTE: The figures in this table
Colorado, Idaho, Nevada, New
GCVTC
3,058,221
1,094,928
2,412,177
Analysis and GCVTC Stud
2000
Post-CAA
2,938,833
1,326,546
1,759,434
represent the total annual emissions (tons
Mexico, Oregon, Utah, and Wyoming.
2010
GCVTC Post-CAA
2,596,409 2,784,580
944,689 1,434,470
2,535,829 1,864,656
per year) for the Western States: Arizona
GCVTC
2,532,855
970,762
2,730,304
, California,
Prospective Analysis PM2S Emissions
Estimates and Observed Data
The 1997 National Air Quality and Emissions
Trends Report provides a summary of PM25
concentration speciation data. This report shows
the relative contribution of the major PM emissions
source components (crustal material, carbonaceous
particles, nitrate, and sulfate) to ambient PM25
concentrations in urban and nonurban areas
throughout the U.S. Comparison of primary PM25
emissions estimates generated for this analysis with
the observed concentration data presented in the 1997
report indicates that the ratio in the prospective study
of crustal material to primary carbonaceous particles
is high. At least part of this apparent overestimation
Figure A-6
1990 Primary PM2.5 Emissions by EPA Region (tons/year)
Region 1
Crustal - Fugitive Dust Souro
Crustal - Industrial Sources
Other Primary
Elemental Carbon
Organic Carbon
o
100,000 450,000 850,000
A-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
of crustal material and underestimation of
carbonaceous participates, however, is due to the fact
that much of the emitted crustal material quickly
settles and does not have a quantifiable impact on
ambient air quality. In this analysis, we apply a factor
of 0.2 to crustal emissions to estimate the fraction of
crustal PM25 that makes its way into the "mixed layer"
of the atmosphere and influences pollutant
concentrations. Figure A-6 displays the breakout of
primary PM25 into its adjusted crustal and
carbonaceous (elemental carbon (EC) and organic
carbon (OQ) components. The figure divides crustal
material into two subcategories based on the source of
the material (fugitive dust or industrial sources) and
also shows the portion of primary PM25 that is neither
crustal nor carbonaceous. The ratios, one for each
EPA Region, of adjusted crustal material to primary
carbonaceous particles presented in Figure A-6 are in
line with the observed PM25 concentration data
presented in the 1997 report.
Industrial Point Sources
This section addresses industrial sector emitters ~
boilers and processes ~ that are large enough to be
included in the 1990 emissions data base as individual
point sources. In most cases, these are facilities that
emit more than 100 tons per year of at least one
criteria air pollutant. For industrial point source VOC
and NOX emitters, trend analysis using the Emission
Reduction and Cost Analysis Model (ERCAM) was
conducted to project emissions for the years 2000 and
2010 (Pechan, 1994a; 1994b). The same procedures
employed in the VOC and NOX projections were also
used in developing CO, SO2, and PM10 estimates.
Overview of Approach
In order to estimate the combined effects of
activity growth and CAA controls on industrial point
sources, the base year 1990 point source inventory was
projected to 2000 and 2010 using growth factors from
the Bureau of Economic Analysis (BEA), and CAA
control assumptions. In its guidance for projecting
emissions by combining growth and control effects,
EPA identifies the following two options:
1. Aggregating all base year emissions and
control information at the county level and
performing all projections on that basis; or
2. Retaining source-specific information in the
base year inventory and performing
projections on a source-by-source basis
(EPA, 199 la).
The second of these two approaches was selected, and
future year emissions were projected by multiplying
source-specific base year emissions by a
corresponding growth factor and control factor. The
decision to follow this option was based on the need
to use source-level emissions estimates as the input
for the air quality modeling phase of the prospective
analysis.
The growth factors used in this analysis for
projecting industrial point source emissions are from
1995 BEA industry-level Gross State Product (GSP)
and population projections by State (BEA, 1995).
ERCAM was used to model VOC and NOX emissions
under each of the control scenarios. The basic
approach for projecting emissions in ERCAM is as
follows:
EMISr = EMISgo * GFACY *
1 - (REr * CEY)
- (RE9
where:
EMISy
EMIS90
GFACY
REY
CEY
RE
Emissions in projection
yeary
1990 base year emissions
Growth factor for
projection year y
Future year rule
effectiveness (RE)
Future year control
efficiency
Base year (1990) RE
Base year (1990) control
efficiency
A-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
In cases where the future year control level (REY
* CEY) is less stringent than the base year control
level, 1990 base year control levels are retained (i.e.,
REY and CEY equal 1990 levels in the emission
projection algorithm).
Base Year Emissions
The base year 1990 point source emission
inventory for the prospective analysis is Version 3 of
the NPI, originally developed in 1994 as a component
of the Office of Policy, Planning and Evaluation
(OPPE)'s "National Particulate Matter Study"
(Pechan, 1994c; Pechan, 1995a). The NPI is a 1990
air emissions inventory for the United States
(excluding Alaska and Hawaii). This data base
contains plant and process level emissions for each of
the criteria pollutants examined in this analysis.
Industrial point source emissions in the NPI were
estimated using emission estimates from the 1985
National Acid Precipitation Assessment Program
(NAPAP) Inventory projected to 1990 using BEA
industrial sector earnings. Emission estimates for
1985 were projected to 1990 based on the State-level
growth in earnings by industry (2-digit Standard
Industrial Classification (SIC) code). Each record in
the point source inventory was matched with the BEA
earnings data based on the State and the 2-digit SIC
code.
The industrial sector 1990 emission estimation
procedures do not account for technological
improvements since 1985 that may have lowered
emissions per unit of production/output, nor do they
account for emission controls that were added during
this period. As a result, the base year emissions
estimates, if biased, may overstate industrial point
source emissions for 1990. In the Western States,
with the incorporation of the GCVTC estimates, the
1990 emissions baseline for this region of the country
may be less biased. The reason for this is that the
GCVTC inventory is based upon State-supplied
emissions information covering one of the years from
1990 to 1992. These more recent State reports reflect
the effects of technological improvements and
emission controls of the latter half of the 1980's that
the NPI baseline inventory does not capture.
The NAPAP Inventory does not contain PM10
and PM25 emissions estimates. As a result, the annual
PM10 and PM25 emissions figures in the NPI were
calculated from 1985 total suspended particulate
(TSP) emissions. These 1985 TSP estimates were
projected to 1990 using BEA data and emissions
estimates from each point source in the NAPAP
Inventory (excluding steam electric utilities). What
portion of 1990 TSP emissions was PM10 and what
portion was PM2 5 was then determined. In order to
make this determination, however, the 1990 TSP
estimates first had to be adjusted to eliminate the
effect of particulate controls, because in order to
estimate particle size distribution using EPA's
Compilation of Emission Factors (EPA, 1995),
uncontrolled source data were required. Once this
adjustment was made, PM10 and PM2 5 emissions were
calculated by applying a Source Classification Code
(SCC)-specific particle size distribution factor to the
1990 "uncontrolled" TSP emissions estimates. Then,
the effects of primary and secondary controls on the
two pollutants were estimated and base year PM10 and
PM25 emissions were calculated.
Growth Projections
The base year 1990 point source emission
inventory was projected to 2000 and 2010 to
determine the effects of Pre-CAAA and Post-CAAA
controls on future year emission levels. Point source
emissions growth is based on 1995 BEA industry GSP
and population projections by State (BEA, 1995).
EPA guidance for projecting emissions (EPA, 1991a)
lists the following economic variables (in order of
preference) for projecting emissions:
• Product output;
• Value added;
• Earnings; and
• Employment.
In the absence of product output projections, EPA
guidance recommends value added projections. Value
A-15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
added is the difference between the value of a firm's
output and the inputs it purchases from other firms.
BEA GSP projections represent a measure of value
added, and are a fuller measure of growth than BEA's
earnings projections because earnings represents only
one component of GSP. GSP measures reflect the
value added to revenue from selling a product minus
the amounts paid for inputs from other firms. By
incorporating inputs to production, GSP reflects
future changes in production processes, efficiency, and
technological changes. A comparison of BEA's 1995
GSP projections and BEA's 1990 earnings projections
indicates that GSP growth factors are slightly higher
than the earnings data. This is more often true for
capital-intensive industries (e.g., manufacturing) than
for labor-intensive industries (e.g., services).
Components of GSP include payments to capital.
This is an important distinction to make because it
implicitly reflects the effect of factor substitution in
production. As discussed in EPA's projections
guidance, factor substitution should be included in
growth projections, making value added data
preferable to earnings data for projecting emissions.
For reasons mentioned above, the 1995 BEA
industry GSP and population projections by State
(BEA, 1995) were selected as the best available growth
factors for projecting 1990 emissions to 2000 and
2010 for the prospective analysis. BEA's GSP
estimates are broken down by industry sector (2-digit
SIC codes) and State. For each record in the industrial
point source component of the NPI, a link was
established between the State code, the SIC code field,
and the BEA GSP growth factors. Then projected
future year emissions for each point source record
were calculated by multiplying the 1990 emissions by
the corresponding BEA growth factor.
BEA GSP growth factors were used to project
industrial point source emissions for the prospective
analysis because BEA data provide growth by industry
on a State-level in a form that provides a
straightforward link to the industrial point source
component of the NPI (the SIC code field). GSP
growth factors also comply with EPA's guidance for
projecting emissions, since they represent a measure
of value added. In the development of the BEA GSP
projections, BEA ensures consistency with national
projections of population from the Bureau of the
Census, of the labor force from the Bureau of Labor
Statistics (BLS), of the unemployment rate from the
Congressional Budget Office, and of mining output
from the Department of Energy (DOE). It is
important to note, however, that BEA's projections
are based on the assumption that past economic
relationships will continue and that no major policy
changes will occur. The growth factors used in this
analysis therefore do not explicitly reflect potential
future changes in economic conditions or
technologies except those that may be reflected in
historical industry trends.
Control Scenarios
The Pre-CAAA scenario represents expected
point source emissions after the application of BEA
GSP growth factors, with 1990 levels of control
efficiency retained. The Post-CAAA scenario
incorporates control efficiencies based on measures
mandated by the CAAA. The control assumptions
associated with each of the two scenarios are
described separately below.
Pre-CAAA Scenario
The Pre-CAAA scenario assumes the
continuation of 1990 control efficiencies for all
emitters. Point source emissions of VOC, NOX, CO,
SO2, PM10, PM25, and NH3 under the Pre-CAAA
scenario were projected to 2000 and 2010 by applying
BEA GSP growth factors to the base year 1990 point
source emission inventory based on the State and SIC
code data fields.
Post-CAAA Scenario
The Post-CAAA scenario represents point source
emissions after the application of BEA GSP growth
factors, and the effects of controls implemented under
the CAAA. CAAA provisions affecting industrial
point sources include:
A-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
• Title III 2-year and 4-year MACT standards
(VOC only);
• VOC and NOX RACT requirements in ozone
NAAs;
• New control technique guidelines (CTGs);
• A 0.15 Ibs/MMBtu NOX cap on fuel
combustors of 250 MMBtu per hour
or above across the OTAG 37-State
region; and
• Ozone rate-of-progress (HOP) requirements.
The control assumptions used to estimate emissions
from industrial point sources under the Post-CAAA
scenario differ by pollutant. For each of the
pollutants, the corresponding controls assumed to
effect emissions in 2000 and 2010 are briefly described
in Table A-5.
Table A-5
Industrial Point Source Control Assumptions For The Post-CAAA Scenario
Pollutant
Point Source Control Measures
VOC Point source control measures for VOC include RACT, new CTGs, and Title III MACT
controls. Title III MACT controls are generally as stringent, or more stringent, than RACT
controls and are thus the dominant control option for many source categories. An 80
percent RE is assumed for all control measures.
NOX Industrial point source NOX controls include NOX RACT, OTAG level 2 NOX controls, and a
0.15 Ibs/MMBtu cap on fuel combustors of 250 MMBtu per hour and above across the
OTAG 37-State region. Major stationary source NOX emitters in marginal and above NAAs
and in the northeast Ozone Transport Region (OTR) are required to install RACT-level
controls under the ozone nonattainment-related provisions of Title I. RACT control levels
are specified by each State and are based on an assumed rule effectiveness (RE) of 80
percent.
CO No new CO controls were modeled for the Post-CAAA scenario, although some CO NAAs
may have adopted controls for specific point sources within NAAs.
SO2 SO2 nonattainment provisions of the CAAA do not specify any mandatory controls for SO2
emitters, though an emission cap of 5.6 million tons of SO2 per year was set by the CAAA
for industrial sources.
PM10 and Possible control initiatives for particulates under the CAA would result from the Title I
PM25 provisions related to PM10 nonattainment. Because the controls are specific to each area,
the CAAA PM10 emissions for industrial point sources were assumed to be equivalent to the
Pre-CAAA emissions. Point source PM25 emissions were also assumed to be unaffected
by CAAA provisions.
NH,
Point source NH3 emissions were assumed to be unaffected by CAAA provisions.
A-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Emission Summary
National point source emission projections for
2000 and 2010 for each of the pollutants are shown in
Table A-6. VOC emissions from industrial point
sources in these two years are primarily affected by
new National Emission Standards for Hazardous Air
Pollutants (NESHAPs) under Title III and new CTGs
for achieving further VOC emission reductions in
ozone NAAs. Source categories with significant VOC
emission reductions in this sector include chemical
and allied product manufacturing, petroleum
refineries, solvent utilization, and petroleum storage.
NOX emission reductions from industrial fuel
combustors result mainly from the implementation of
RACT for major stationary sources in ozone NAAs
and addition of further NOX controls for large fuel
combustors (larger than 250 MMBtu per hour)
throughout the 37 OTAG States. The OTAG
stationary source NOX strategy included in this
analysis assumes that large fuel combustors meet a
0.15 Ibs/MMBtu NOX emission limit. With these and
other standards, CAA NOX emission benefits in 2010
for this sector are projected to be more than one
million tons.
Industrial point source emission projections for
the other criteria pollutants (CO, SO2, PM10, PM25,
and NH3) show no appreciable effect of the CAAA on
future year emissions. Stationary source CO emitters
could be subjected to further control requirements as
part of individual area CO State Implementation Plans
(SIPs), but this is unlikely. Similarly for PM10, there
may be industrial source PM10 emission reductions
observed through application of best available control
measures in some PM10 attainment plans, but these
potential reductions are not captured in this analysis.
A-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-6
Industrial Point Source Emission
(thousand tons per year)
Pollutant/Source Category
VOC
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
Miscellaneous
TOTAL
NOX
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
TOTAL
CO
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
TOTAL
Summaries
1990
126.1
10.3
1,066.2
72.5
238.2
327.0
1,126.2
490.1
9.7
0.3
3,466.6
1,955.8
103.9
275.4
81.0
99.9
308.0
2.5
2.4
20.7
2,849.7
529.1
96.8
1,923.4
2,106.3
436.3
754.0
2.5
54.8
96.7
5,999.7
by Pollutant For 1990,
2000
Pre-
CAAA
135.9
12.3
1,101.4
85.8
243.1
372.5
1,256.7
540.9
9.9
0.4
3,758.9
2,181.8
122.9
281.5
103.0
104.1
350.9
2.9
2.6
23.5
3,173.3
586.8
118.4
1,957.2
2,418.1
475.1
947.1
2.8
51.5
106.8
6,663.9
2000
Post-
2000, and
2010
CAAA Pre-CAAA
134.9
12.3
878.9
83.1
160.2
365.3
1,036.0
405.7
9.9
0.4
3,086.7
1,213.5
77.4
263.0
98.9
104.1
277.6
2.9
2.6
20.4
2,060.4
586.8
118.4
1,957.2
2,418.1
475.1
947.1
2.8
51.5
106.8
6,663.9
153.7
14.4
1,266.6
90.3
269.5
418.9
1,394.1
617.0
11.1
0.6
4,236.2
2,464.5
141.1
322.2
111.3
111.1
394.6
3.2
3.0
26.9
3,577.9
656.7
140.1
2,233.0
2,486.1
545.9
1,134.1
3.2
58.7
118.7
7,376.6
2010*
2010
Post-CAAA
157.7
14.9
1,007.4
87.2
166.7
417.3
1,120.0
467.5
11.7
0.6
3,450.9
1,255.9
84.1
300.0
104.7
108.1
302.8
3.1
3.0
22.9
2,184.5
656.7
140.1
2,233.0
2,486.1
545.9
1,134.1
3.2
58.7
118.7
7,376.6
A-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pollutant/Source Category
S02
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
TOTAL
PM10
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
Miscellaneous
TOTAL
PM2.5
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
Miscellaneous
TOTAL
1990
2,482.2
202.4
440.1
664.7
434.8
393.6
0.8
4.6
21.0
4,644.2
221.1
16.6
62.5
137.9
28.9
374.3
2.1
64.4
8.0
10.7
926.4
162.0
8.2
42.7
96.3
19.5
224.3
1.8
26.5
6.7
1.6
589.5
2000
Pre-
CAAA
2,861.9
243.7
486.6
808.9
449.5
456.4
0.9
5.4
22.8
5,336.2
245.2
19.8
65.4
167.2
31.7
427.3
2.6
73.8
8.7
13.7
1,055.3
178.1
9.7
45.7
116.7
21.3
257.9
2.2
30.5
7.2
2.1
671.5
2000
Post-
CAAA
2,861.9
243.7
486.6
808.9
449.5
456.4
0.9
5.4
22.8
5,336.2
245.2
19.8
65.4
167.2
31.7
427.3
2.6
73.8
8.7
13.7
1,055.3
178.1
9.7
45.7
116.7
21.3
257.9
2.2
30.5
7.2
2.1
671.5
2010
Pre-CAAA
3,262.0
282.9
546.4
857.1
489.3
522.6
1.0
6.4
26.1
5,993.9
275.8
23.0
74.2
179.4
36.0
485.5
3.0
83.6
9.7
16.9
1,186.9
200.0
11.1
52.0
124.4
24.1
294.7
2.5
34.4
8.0
2.6
754.0
2010
Post-CAAA
3,262.0
282.9
546.4
857.1
489.3
522.6
1.0
6.4
26.1
5,993.9
275.8
23.0
74.2
179.4
36.0
485.5
3.0
83.6
9.7
16.9
1,186.9
200.0
11.1
52.0
124.4
24.1
294.7
2.5
34.4
8.0
2.6
754.0
A-20
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pollutant/Source Category
NH3
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
TOTAL
1990
10.0
0.3
182.6
5.9
42.8
2.1
243.6
2000
Pre-
CAAA
10.3
0.3
202.6
6.8
48.7
2.1
270.8
2000
Post-
CAAA
10.3
0.3
202.6
6.8
48.7
2.1
270.8
2010 2010
Pre-CAAA Post-CAAA
11.3
0.3
231.6
7.0
55.1
2.1
307.5
11.3
0.3
231.6
7.0
55.1
2.1
307.5
The totals reflect emissions for the 48 contiguous States, excluding Alaska and Hawaii. Totals may not add due to rounding.
Utilities
EPA used the IPM to estimate future year heat
input, SO2 and NOX emissions, fuel type, and optimal
control techniques for each current and planned
electric utility unit.5 The IPM modeling inputs,
outputs, and key assumptions are discussed in more
detail in EPA 1997b. This section focuses on the
steps used to supplement these projections by adding
emissions of VOC, CO, PM10, PM25, and NH3, as well
as adding data elements needed for air quality
modeling (location and stack parameters).
Overview of Approach
For the prospective analysis, EPA matched each
unit in the IPM file to the 1990 NPI (Pechan, 1994c;
Pechan, 1995a) based on the Office of the Regulatory
Information System (ORIS) plant and boiler code.
For units that were matched, stack parameters and
location coordinates were taken directly from the NPI.
VOC, CO, PM10, and PM25 emissions were calculated
using AP-42 emission rates (standard EPA emission
factors that are developed from stack tests and
engineering calculations) and control efficiencies as
reported in the NPI. NH3 emissions were calculated
for ammonia slippage where boilers were forecast to
install selective catalytic reduction (SCR) as the control
technique to reduce NOX emissions.
Base Year Emissions
The base year emission inventory used in the air
quality modeling portion of the Section 812
prospective analysis is Version 3 of the NPI. The
utility portion of this inventory, which covers fossil-
fuel fired steam electric generating boilers, was
developed from DOE Form EIA-767 (the Steam-
Electric Plant Operation and Design Report) fuel use
data combined with AP-42 emission factors and
emission limits from EIA-767. The NPI also includes
gas turbines and internal combustion engines to the
extent that these were included in the 1985 NAPAP.
The IPM uses a different modeling set as input
for emission projections. This set is consistent with
the NPI in that boilers included in the NPI are also in
the IPM modeling set. The IPM data set, however,
also includes combustion turbines (more than those
included in the NPI) and non-utility generators. Some
of these units may be included in the NPI as part of
the industrial point source inventory or area source
fuel combustion (and would thus be double-counted
in the projections) while others may be missing from
the NPI data set. It should be noted that in
projection year 2000, under the Pre-CAAA scenario,
these units account for just over 1 percent of total
utility NOX emissions and only 0.1 percent of total
utility SO2 emissions. Therefore, the potential
exclusion of the units from the 1990 base year data set
and the potential double-counting in the projection
year is expected to have minimal effect on air quality,
benefits, and cost modeling.
5 The IPM was constructed and is maintained by ICF, an
EPA contractor.
A-21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Control Scenarios
The Pre-CAAA and Post-CAAA scenario NOX
and SO2 emissions were modeled using IPM (EPA,
1997b). Heat input by unit under each projection year
and scenario was also derived from IPM results.
Default emission factors were applied to the unit-
specific heat input to calculate VOC, CO, PM10, and
PM25 emissions. For these pollutants, the Pre-CAAA
and Post-CAAA emission rates were assumed to be
the same. Any differences in emissions between these
scenarios are due to shifts in operation between units
or fuel changes (including ash content of the coals as
well as switching to natural gas) predicted by IPM.
Differences in NH3 emissions between the scenarios
is a fraction of added SCR controls to reduce NOX
under the Post-CAAA scenario.
Emission Summary
Table A-7 is a summary of utility emissions by
unit and fuel type. Oil- and gas-fired units have been
grouped together, because information on the division
of fuel for boilers burning oil and gas was not
contained in the unit-level file developed from IPM.
Utility SO2 and NOX emissions are affected most by
the CAAA. Differences between Pre-CAAA and
Post-CAAA NOX emissions result from a combination
of Title IV - Acid Rain regulations, and nonattainment
provisions that require NOX RACT controls for major
stationary sources in ozone NAAs. The anticipated
effect of the OTC MOU and a regional NOX trading
program on stationary source NOX emissions in the
eastern portion of the United States also influence the
Post-CAAA utility NOX emissions estimates (OTC,
1994). In 2010, the difference between Pre-CAAA
and Post-CAAA utility NOX is about 5 million tons.
SO2 emission reductions attributable to the CAA
is the result of the Title IV - Acid Rain control
program. Through this program, annual emissions of
SO2 are to be reduced by 10 million tons from 1980
levels through a market-based allowance system.
Differences between Pre-CAAA and Post-CAAA SO2
estimates for coal-fired units reflect flue gas
desulfurization installations and some switching to
lower sulfur coal in the Post-CAAA case. SO2
emissions from oil/gas units are actually lower in the
Pre-CAAA scenario because a higher percentage of
units would have been expected to burn coal with less
stringent air pollution emission limits. The 2010 Pre-
CAAA estimates show no significant oil use by
utilities.
A-22
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-7
Utility Emission Summary*
(thousand tons per year)
Pollutant/Source Category
VOC
NOX
CO
SO2
PM10
PM,5
NH3
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
Coal
Gas/Oil/Other
Internal Combustion
TOTAL
2000 2000
1990 Pre-CAAA Post-CAAA
27.1
7.8
1.9
36.8
6,689.5
679.1
57.1
7,425.7
232.6
81.9
14.7
329.2
15,221.9
611.9
30.7
15,897.5
268.4
10.6
4.1
283.1
99.2
5.9
3.7
108.8
0.0
0.0
0.0
0.0
* The totals reflect emissions for the 48 contiguous States,
23.6
1.8
5.6
31.0
7,895.7
324.1
97.3
8,317.1
191.8
48.0
50.7
290.5
16,111.3
44.1
0.0
16,155.4
244.7
1.5
6.1
252.3
82.9
1.3
6.0
90.2
0.0
0.0
0.0
0.0
excluding Alaska
23.1
1.9
6.1
31.1
3,779.0
216.0
82.1
4,077.1
188.7
49.3
55.4
293.4
10,315.0
175.5
0.0
10,490.5
245.1
2.5
6.6
254.2
82.3
2.4
6.6
91.3
33.3
0.0
0.0
33.3
and Hawaii.
2010
Pre-CAAA
26.3
1.5
21.2
49.0
8,700.3
220.0
134.4
9,054.7
215.3
44.8
193.6
453.7
17,696.0
0.0
0.0
17,696.0
281.1
1.0
23.4
305.5
97.0
1.0
23.3
121.3
0.0
0.0
0.0
0.0
2010
Post-CAAA
24.7
1.7
23.5
49.9
3,610.0
72.6
83.7
3,766.3
202.5
45.9
214.8
463.2
9,776.6
84.2
0.0
9,860.8
249.0
1.6
26.0
276.6
83.7
1.6
25.8
111.1
221.9
0.0
0.0
221.9
A-23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Nonroad Engines/Vehicles
The nonroad engines/vehicles sector includes all
transportation sources that are not counted as
highway vehicles. Thus, this sector includes marine
vessels, railroads, aircraft, and nonroad internal
combustion engines and vehicles. Nonroad engines
are significant emitters of NOX, PM10, and VOC.
Diesel engines account for most of the NOX and PM10
emissions, while gasoline engines emit most of the
VOC. This section contains summaries of 1990
emissions from major source categories in the
nonroad engine/vehicle sector. The growth factors
and control efficiencies used to project emissions to
2000 and 2010 under the two control scenarios are
also described.
Overview of Approach
Nonroad VOC and NOX emissions were
projected using ERCAM; similar modeling techniques
were used for the remaining pollutants. The algorithm
for projecting nonroad emissions is:
EMISY = EMIS90 * GFACr * [l - CEY * PEY]
where:
EMISY
EMIS90
GFAC
CEY
PEY
emissions in projection
year y
1990 emissions
growth factor for
projection year y
control efficiency for
projection year y
penetration rate for
projection year y
The control efficiency is a function of the
percentage reduction or decrease in emission rate
expected through new engine standards and the
fraction of emissions covered through fleet turnover.
The penetration rate accounts for the fraction of
emissions from affected engine types (generally
resulting from horsepower (hp) cutoffs) in a broad
engine category (e.g., construction).
Growth factors applied are based on the 1995
BEA GSP projections by State and industry and
population projections. The 1990 base year emissions
are from Version 3 of the NPI. Under the Pre-CAAA
scenario, no changes in engine standards are modeled
(future year emission rates are assumed to be
equivalent to 1990 rates). Under the Post-CAAA
scenario, Federal nonroad engine standards are
incorporated. All modeling is at the county and SCC
level to retain necessary details for cost and air quality
modeling.
Base Year Emissions
The 1990 emission estimates for nonroad vehicles
are from the 1990 NPI. The emissions in the NPI
from sources in the nonroad engines/vehicles sector
are based on one of the following sources: (1) a
nonroad emission inventory compiled by EPA's OMS
(EPA, 1991b); or (2) the 1985 NAPAP Area Source
Emissions Inventory. EPA's OMS inventories
provided the majority of criteria pollutant emissions
for the base year 1990 inventory, accounting for nearly
90 percent of VOC emissions and nearly 60 percent
of NOX emissions. The remaining emissions for the
nonroad engines/vehicles sector are based on the
NAPAP emissions inventory.
Growth Projections
The 1990 estimates from the NPI for the nonroad
engine/vehicle sector are projected to 2000 and 2010
to estimate the impact of CAAA controls on future
year emission levels. For each major nonroad
engine/vehicle category a growth surrogate is
identified for estimating future emissions. Growth
surrogates for nonroad engine/vehicle categories
include 1995 BEA projections of population for
recreational and lawn and garden equipment
categories, and an appropriate GSP by SIC code
estimate for all other categories. SIC codes are
assigned to area source categories according to an
assignment made for other EPA projects, such as the
ozone and PM NAAQS cost analyses. This
assignment of nonroad engine/vehicle categories to
BEA indicators is shown in Table A-8.
A-24
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The 1995 BEA GSP and population projections
by State and industry were selected as the best
available growth factors for projecting 1990 emissions
to 2000 and 2010 for the prospective analysis (BEA,
1995). EPA's projection guidance recommends that
area source emissions be projected using surrogate
growth indicators such as BEA, or using local
studies/surveys (EPA, 199la). BEA provided a
consistent set of projections by SIC code that could be
easily applied to the 1990 nonroad engine/vehicle
sector across all geographic regions.
Control Scenarios
Emissions from engines used in nonroad
equipment are a significant source of NOX, VOC, and
PM emissions. In some areas of the country,
emissions from nonroad engines represent a third of
the total mobile source NOX and VOC emissions and
over two-thirds of the mobile source PM emissions.
Pre-CAAA Scenario
The Pre-CAAA scenario incorporates the growth
factors described above, and assumes that future year
emission rates from nonroad engines remain the same
as 1990 levels.
A-25
-------
Table A-8
BEA Growth Forecasts by Major Source Category:
Nonroad Engines/Vehicles
Major Category
BEA Growth Category*
Annual Growth (% per year):
1990-2000
1990-2010
Nonroad Internal Combustion Engines and Vehicles:
Airport Service Equipment
Construction Equipment
Farm Equipment
Industrial Equipment
Lawn & Garden Equipment
Light Commercial Equipment
Logging Equipment
Recreational Marine Vessels
Recreational Vehicles
Aircraft:
Military
Commercial
Civil
Railroads
Commercial Marine Vessels
Transportation by air (SIC 45)
Construction (SIC 15, 16, and 17)
Farm (SIC 01)
Total Manufacturing
Population
Total Manufacturing
Agricultural Services, Forestry, Fisheries (SIC 07, 08, 09)
Population
Population
Federal, military
Transportation by air (SIC 45)
Transportation by air (SIC 45)
Railroad Transportation (SIC 40): Earnings
Water Transportation (SIC 44)
5.8%
0.8%
2.4%
1.9%
1.1%
1.9%
7.8%
1.1%
1.1%
-1.2%
5.8%
5.8%
-1.5%
-0.5%
5.5%
1.0%
2.4%
1.9%
1.0%
1.9%
7.4%
1.0%
1.0%
-0.4%
5.5%
5.5%
-0.9%
-0.2%
*BEA growth category refers to GSP projections for each industry, unless otherwise specified.
A-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Post-CAAA Scenario
The CAAA specifically directed EPA to study the
contribution of nonroad engines to urban air
pollution, and to regulate them, if warranted. In 1991,
EPA released a study that documented higher than
expected emission levels across a broad spectrum of
nonroad engines and equipment (EPA, 1991b). In
response, EPA initiated several regulatory programs
for nonroad engines. The impact of these programs
is incorporated in the Post-CAAA scenario.
Emission Summary
A summary of projected emissions by engine
classification is shown in Table A-9. Future year VOC
and NOX emissions are the pollutants most affected by
the CAAA, as most of the new engine standards focus
on controlling these ozone precursors. CO, SO2, and
PM10 emissions under the Pre-CAAA scenario are
nearly equal to emissions under the Post-CAAA
scenario since CAAA controls only affect NOX and
VOC emissions from the nonroad engine/vehicle
sector. Effects in 2000 are modest because the new
engine standards do not affect emissions until the
mid-to-late 1990s. More dramatic differences are seen
in 2010.
Gasoline-powered engines are the most significant
nonroad VOC emitter, so most of the VOC emissions
difference in this sector is the result of small spark
ignition (SI) engine standards. Lawnmowers, for
example, are affected by these new standards.
NOX emission benefits shown in Table A-9 for
non-road engines result principally from compression
ignition (CI) (diesel engine) standards. In 2010,
difference between the Post-CAAA and Pre-CAAA
nonroad diesel engine NOX emissions is almost 0.5
million tons. These engines are primarily used in
construction equipment. Other off-highway NOX
sources with lower emissions levels under the Post-
CAAA scenario include railroads (diesel locomotives)
and marine vessels. In contrast, a small NOX
disbenefit is associated with non-road gasoline engines
in the Post-CAAA scenario; this is because the small
SI engine standard for hydrocarbons (HCs) is
expected to increase NOX emissions.
A-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-9
Nonroad National Emission Projections by Source Category*
(thousand tons per year)
Pollutant/Source Category
VOC
NOX
CO
SO2
PM10
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
1990
1,596.8
185.0
191.9
36.0
44.2
2,053.9
176.0
1,438.4
139.7
183.7
898.0
2,835.8
12,047.2
781.9
960.9
58.0
121.8
13,969.8
3.2
16.7
8.0
147.5
66.6
242.1
42.1
185.6
40.4
24.2
44.0
336.3
2000
Pre-CAAA
1,810.8
225.3
242.6
33.6
37.2
2,349.5
205.4
1,751.4
194.8
169.8
759.4
3,080.9
13,973.8
948.6
1,423.6
54.1
102.8
16,503.0
3.6
19.0
10.9
139.8
56.3
229.6
47.0
227.3
38.1
22.7
37.3
372.5
2000
Post-CAAA
1,549.3
225.3
242.6
33.6
37.2
2,088.0
220.5
1,603.0
194.8
169.8
759.4
2,947.5
13,417.5
948.6
1,423.7
54.1
102.8
15,946.6
3.6
19.0
10.9
139.8
56.3
229.6
47.0
227.3
38.1
22.7
37.3
372.5
2010
Pre-CAAA
2,004.0
261.6
300.3
34.4
35.5
2,635.8
237.0
2,032.5
249.2
173.2
725.9
3,417.8
15,735.4
1,097.3
1,855.9
55.5
98.2
18,842.3
4.1
22.4
13.8
142.3
53.8
236.5
51.6
262.6
41.3
23.3
35.7
414.4
2010
Post-CAAA
1,257.4
261.6
300.3
34.4
35.4
1,889.2
269.8
1,546.9
249.2
161.5
513.6
2,740.9
15,020.6
1,097.3
1,855.9
55.5
98.2
18,127.5
4.1
22.4
13.8
142.3
53.8
236.5
51.6
187.1
41.3
23.3
33.1
336.3
A-28
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pollutant/Source Category
1990
2000
Pre-CAAA
2000
Post-CAAA
2010
Pre-CAAA
2010
Post-CAAA
PM2.5
NH3
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
Nonroad Gasoline
Nonroad Diesel
Aircraft
Marine Vessels
Railroads
TOTAL
35.0
170.8
28.5
17.8
40.5
292.6
0.0
0.0
0.0
1.1
1.8
2.9
* The totals reflect emissions for the 48 contiguous States,
39.2
209.1
26.9
16.6
34.3
326.1
0.0
0.0
0.0
1.1
1.5
2.6
excluding Alaska
39.2
209.1
26.9
16.6
34.3
326.1
0.0
0.0
0.0
1.1
1.5
2.6
and Hawaii.
42.9
241.6
29.1
17.0
32.8
363.5
0.0
0.0
0.0
1.1
1.4
2.5
42.9
172.1
29.1
17.0
30.4
291.7
0.0
0.0
0.0
1.1
1.4
2.5
Motor Vehicles
Motor vehicles are a significant contributor of
VOC, NOX, and CO emissions. In 1990, motor
vehicles contributed 30 percent of total VOC, 33
percent of total NOX, and 66 percent of total CO
emissions. The CAA includes provisions to reduce
motor vehicles emissions in both Title I and Title II.
Overview of Approach
The general procedure for calculating historic and
projection year motor vehicle emissions is to multiply
activity, in the form of VMT by pollutant specific
emission factor estimates. ERCAM (Pechan, 1996)
was used to project motor vehicle emissions for VOC,
NOx, and CO. Emission factors for these pollutants
were generated using the EPA's motor vehicle
emission factor model MOBILESa (EPA, 1993a).
PM10, PM25, and SO2 emission factors were generated
using another EPA motor vehicle emission factor
model, PARTS (EPA, 1994). Emission factors for all
pollutants are modeled using common assumptions
about ambient temperatures and vehicle speeds at the
State level. Control programs (I/M, reformulated
gasoline) are specified at the county level. Temporally,
emissions are calculated by month and summed to
develop annual emission estimates.
Base Year Emissions
Base year emissions are from Version 3 of the
NPI. The NPI VMT, by county/SCC (i.e., vehicle
type/functional roadway class), are based on data
from the Federal Highway Administration (FHWA)
Highway Performance Monitoring System (HPMS).
The HPMS area wide data base contains State-level
VMT estimates for rural and small urban areas, as well
as separate VMT estimates for each large urban area
within the State. VMT estimates for each of these
categories are by functional roadway class. Two
procedures were performed to convert this VMT data
into a county/SCC level format. First, each State's
rural, small urban, and large urban VMT by functional
roadway class were distributed to the county level
based on population data. Second, the resulting
county/functional roadway class VMT were allocated
to the vehicle type level based on HPMS and other
FHWA data. The resulting VMT estimates are
county-level estimates segregated by vehicle type and
roadway class.
The 1990 emission estimates were calculated by
applying 1990 control-specific emission factors to the
A-29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
VMT estimates. The 1990 emission factors were
generated using historical temperatures, gasoline
volatility Reid vapor pressure (RVP) data, and
inspection and maintenance (I/M) information.
Emissions estimates are calculated at the
county/vehicle type/roadway type level, allowing for
county differences in I/M programs.
Growth Projections
Vehicle Miles Traveled
The general procedure used to project motor
vehicle emissions was to grow 1990 activity (VMT) to
the future year (2000 or 2010) and then to apply future
year emission factors. Estimates of national growth in
VMT from the MOBILE4.1 Fuel Consumption
Model (FCM) (EPA, 1991c; Wolcott and Kahlbaum,
1991) were used as the basis for VMT projections.
Primary MOBILE4.1 FCM inputs were vehicle
registrations, VMT, and fuel economy for each vehicle
class. MOBILE4.1 FCM outputs included estimates
of fleet fuel consumption, VMT, on-road fuel
economy, and vehicle registrations. All are national
values. Historical vehicle stock information is
available from R.L. Polk and Department of
Transportation (DOT). The MOBILE4.1 FCM relies
primarily on the R.L. Polk data to estimate historical
stocks of cars and light trucks in 1990, and uses DOT
and American Association of Automobile
Manufacturers statistics to estimate truck stocks by
weight class.
Modeled Motor Vehicle Emission
Rates
The tunnel study portion of the South Coast Air
Quality Study (SCAQS) (Ingalls et al., 1989) showed
that there were wide discrepancies between measured
and modeled motor vehicle emissions in an
experiment performed in 1987 at a tunnel near Los
Angeles. Running VOC emission factors were from
1.4 to 6.9 times the emission factors calculated from
the California Air Resources Board (CARB) computer
program output, measured CO emission rates were
from 1.1 to 3.6 times modeled emission factors, and
measured NOX emission rates were 0.6 to 1.4 times
the modeled values. Since that time there have been
many research studies performed to attempt to
identify the reasons for the observed discrepancies
and to modify the two models developed by
regulatory agencies (EPA's MOBILE emission factor
model and CARB's emission factors model
(EMFAQ) to perform better in estimating real world
emission rates.
It is difficult to estimate how present uncertainties
in estimating motor vehicle emissions might affect the
estimated difference between Pre-CAAA and Post-
CAAA emissions. The difference between the
scenarios will widen if the excess emissions are
successfully reduced by Post-CAAA measures.
However, if the excess emissions are irreducible, or not
influenced by new Post-CAAA initiatives, then the
relative emissions difference between the cases would
be expected to remain the same as is estimated in this
analysis.
Control Scenarios
This section describes the control assumptions
for the Pre-CAAA and Post-CAAA scenarios. Table
A-10 summarizes the geographic applicability of all
controls modeled.
Pre-CAAA Scenario
The Pre-CAAA scenario applies estimated
increases in activity levels with emission factors
reflecting control programs in place prior to the
passage of the 1990 Amendments. The motor vehicle
controls applied under this scenario include the
Federal Vehicle Motor Control Program (FMVCP)
(tailpipe standards), Phase I gasoline volatility (RVP
limits), and current 1990 I/M programs.
PM10 emission factors representing gasoline
vehicles do not vary between the Pre-CAAA and
Post-CAAA control scenarios. Pre-CAAA PM10
emission factors for diesel vehicles were generated by
freezing emission rates at 1993 levels, since the
PARTS model does not calculate emission factors
without application of CAA tailpipe standards. A
composite diesel emission factor by diesel vehicle type
was then calculated by applying by-model-year
emission rates by yearly travel fractions.
A-30
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Post-CAAA Scenario
The Post-CAAA control scenario incorporates the
likely effects of controls mandated under the CAAA.
Motor vehicle controls applied under this scenario
include CAA tailpipe standards and evaporative
standards, Phase II RVP limits, reformulated gasoline,
oxygenated fuel, I/M (none, basic, low enhanced,
OTR low enhanced, and high enhanced), and low
emission vehicles. Of the above mentioned controls,
only I/M and reformulated gasoline affect particulate
and SO2 emission factors. Each of the CAA controls
and their applicability are summarized in Table A-10.
Emission Summary
Table A-ll summarizes national emissions by
vehicle type. Comparison of Pre- and Post-CAAA
scenarios shows motor vehicle VOC emissions
reductions of 28 percent in 2000 and 46 percent in
2010 as a larger fraction of the vehicle fleet meets low
emission vehicle (LEV) program emission standards.
CAA tailpipe standards, reformulated gasoline, and
I/M requirements also contribute to declines in motor
vehicle VOC emissions.
For motor vehicle-emitted NOX, again the
differences between Post-CAAA and Pre-CAAA
scenarios are most pronounced in 2010 (38 percent,
compared with a 14 percent difference in 2000) as the
49 State LEV program becomes more effective with
fleet turnover.
The Post-CAAA scenario also shows that there
are expected to be significant CO benefits achieved
through the Nonattainment (Title I) and Motor
Vehicle Provisions (Title II) of the 1990 Amendments.
The most important new provisions and programs
expected to be providing these benefits, in order of
estimated importance, include: enhanced vehicle
emission inspections, wintertime oxygenated fuel use,
and LEV program adoption.
SO2 motor vehicle emissions decrease under the
Post-CAAA scenario as a result of sulfur limits for
diesel fuel. Section 211 of the CAAA limited the
sulfur content of motor vehicle diesel fuel to 0.05
percent (by weight) beginning October 1, 1993.
Motor vehicle PM10 and PM25 emission changes
result from CAA tailpipe standards. Almost all of the
motor vehicle-emitted PM changes occur because of
PM exhaust emission standards for heavy-duty diesel
trucks pDDTs).
A-31
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-10
Applicability of Mobile Source Control Programs
Control Measure
Applicability
Pre-CAAA Scenario
FMVCP
Phase I RVP
I/M
Post-CAAA Scenario
Phase II RVP
CAA Tailpipe Standards
CAA Evaporative Controls
Heavy Duty NOX Standard
Federal Reformulated Gasoline
Oxygenated Fuel
Basic I/M
Low Enhanced I/M
High Enhanced I/M
National LEV
California LEV
National
National (standard varies by region)
Programs in place in 1990
National (standard varies by region)
National
National
National
Nine areas required to adopt this program under the CAA plus areas
All CO NAAs
All moderate ozone NAAs, moderate CO NAAs, and areas with I/M in
All areas previously required to implement high enhanced I/M who are
Serious and above ozone NAAs, in metropolitan areas in the OTR with
Nationally, with the exception of California
California
A-32
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table A-1 1
National Highway Vehicle Emissions by Vehicle Type*
(thousand tons)
voc
NOX
CO
SO2
Vehicle
Type
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
1990
4,207.2
954.0
803.5
466.9
11.6
2.8
313.9
50.5
6,810.5
3,406.1
775.2
557.8
333.1
35.9
7.4
2,318.4
11.6
7,445.6
40,073.4
8,458.6
6,533.4
5,895.0
29.1
5.7
1,300.1
185.6
62,480.9
143.0
37.6
20.5
10.8
12.7
2.8
340.1
0.3
567.7
2000
Pre-CAAA
3,823.6
1,089.0
711.4
291.1
1.4
1.0
367.5
42.5
6,327.6
3,633.6
1,045.5
646.6
333.2
4.0
2.5
2,135.1
14.1
7,814.6
36,759.0
10,566.9
6,867.5
2,921.9
3.8
2.3
1,827.2
220.8
59,169.6
151.6
50.1
25.6
11.7
1.1
0.8
390.3
0.4
631.6
2000
Post-CAAA
2,692.9
793.5
550.4
228.8
1.4
1.0
227.2
37.8
4,533.1
2,873.8
815.8
565.1
324.3
4.0
2.5
2,051.7
14.1
6,651.3
26,920.2
7,947.5
5,349.4
2,766.0
3.8
2.3
1,826.8
209.5
45,025.5
151.6
50.1
25.6
11.7
0.3
0.2
97.6
0.4
337.4
2010
Pre-CAAA
4,311.9
1,341.0
867.1
279.8
<0.1
0.4
469.6
49.9
7,319.7
4,161.6
1,326.1
821.6
392.6
<0.1
1.0
2,359.1
16.5
9,078.6
40,304.6
13,054.8
8,457.5
1,807.1
<0.1
1.0
2,397.1
259.3
66,281.6
178.0
64.0
32.6
14.1
<0.1
0.3
480.3
0.5
769.6
2010
Post-CAAA
2,263.6
760.6
583.0
142.2
<0.1
0.4
151.8
44.3
3,945.9
2,402.1
783.8
631.6
296.3
<0.1
1.0
1,443.0
16.5
5,574.4
23,711.9
7,986.4
6,102.3
1,639.3
<0.1
1.0
2,208.3
245.9
41,895.1
178.0
64.0
32.6
14.1
<0.1
0.1
120.1
0.5
409.2
A-33
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Vehicle
Type
PM10
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
PM2.5
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
NH3
LDGV
LDGT1
LDGT2
HDGV
LDDV
LDDT
HDDV
MC
Total
1990
2000 2000
Pre-CAAA Post-CAAA
63.1
15.2
16.9
10.6
8.8
1.7
238.2
0.4
354.7
38.1
9.8
11.1
7.0
7.8
1.5
215.7
0.2
291.0
165.6
23.6
8.6
0.4
<0.1
<0.1
0.2
<0.1
198.5
66.1
18.4
10.7
8.0
0.6
0.6
186.9
0.4
291.7
36.9
11.2
6.7
5.3
0.5
0.5
164.4
0.2
225.7
262.8
58.5
25.8
1.6
<0.1
<0.1
0.4
<0.1
349.2
The totals reflect emissions for the 48 contiguous States,
66.1
18.4
10.7
8.0
0.6
0.4
152.9
0.4
257.6
36.9
11.2
6.7
5.3
0.5
0.3
134.4
0.2
195.6
262.8
58.5
25.8
1.6
<0.1
<0.1
0.4
<0.1
349.2
2010
Pre-CAAA
75.3
23.0
11.5
6.6
<0.1
0.2
179.8
0.5
297.0
43.4
13.8
7.0
4.3
<0.1
0.2
164.7
0.3
233.7
314.0
80.6
36.8
2.9
<0.1
<0.1
0.5
0.1
434.9
excluding Alaska and Hawaii.
2010
Post-CAAA
75.3
23.0
11.5
6.6
<0.1
0.1
88.4
0.5
205.4
43.4
13.8
7.0
4.3
<0.1
0.1
73.9
0.2
142.7
314.0
80.6
36.8
2.9
<0.1
<0.1
0.5
0.1
434.9
Totals may not add due to
rounding.
A-34
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Area Sources
This section discusses the base year 1990 area
source inventory, and the development of the future
year emission estimates for area sources. Area sources
include small stationary sources not included in the
point source data base (e.g., dry cleaners, graphic arts,
industrial fuel combustion, gasoline marketing) and
solvent use (e.g., consumer solvents, architectural
coatings). The growth factors and control efficiencies
used to project the base year 1990 area source
inventory to 2000 and 2010 under the two control
scenarios are also described, and alternative growth
indicators for area sources are identified. ERCAM
was used to project area source VOC and NOX
emissions under the two control scenarios. The
approach used in ERCAM was also used to project
controlled area source CO, SO2, PM25, and PM10
emissions.
Overview of Approach
The base year 1990 area source inventory was
projected to 2000 and 2010 to estimate the combined
effects of growth and CAA controls on area sources.
In order to project emissions, a surrogate activity
indicator (e.g., population, gasoline consumption) was
identified for each area source category. In its
guidance for projecting emissions for area sources,
EPA identifies preferred growth indicators for each
area source category (EPA, 1991a). Pechan chose a
growth indicator for each area source category based
on EPA's guidance, the availability of projection data
in the relevant years of analysis, and the
appropriateness of the measure for projecting
emissions. Emissions were then projected using
growth factors calculated based on projections for
each activity indicator. The growth rates represent an
increase or decrease in the basic activity that causes
emissions.
Area source emissions for VOC and NOX under
each control scenario were projected using ERCAM;
similar modeling techniques were used for the other
criteria pollutants. The algorithm for projecting area
source emissions is:
EMSr = EMIS90
GFACV
where:
1 - (CEY * REJ
1 - (CE90 * RE90)
EMISY
EMIS90
GFACY
CEY
REY
CE9
REQ
emissions in projection
yeary
1990 emissions
growth factor for
projection year y
control efficiency in
projection year y
rule effectiveness (RE) for
the control in projection
yeary
1990 control efficiency
1990 RE
In cases where the control level for the projection
year control strategy is less than the control level in
1990,1990 control levels are retained in the projection
year. All computations and reporting are at the
county/SCC level for air quality and cost modeling.
Under the Pre-CAAA scenario, future year
control levels are assumed to be equivalent to 1990
levels. The Post-CAAA scenario applies control levels
to model the effects of the Title I nonattainment
provisions, Federal rules, and, in the case of VOC,
Title III MACT standards.
Base Year Emissions
The base year 1990 area source emission
inventory for the prospective analysis is Version 3 of
the NPI (Pechan, 1995a; Pechan, 1995b). This
inventory contains county/SCC level emissions for
area source categories. Most non-fugitive dust area
source emissions estimates in the NPI originate with
the 1985 NAPAP Area Source Emission Inventory.
Exceptions to this are solvent emissions, prescribed
burning, forest wildfires, fugitive dust, and residential
wood combustion.
The general method for estimating 1990 area
source emissions using the 1985 NAPAP Inventory
was to apply growth factors to the NAPAP Inventory
values. BEA historical earnings data, population, fuel
A-35
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
use from State Energy Data System (SEDS) (DOE,
1991), and other category-specific indicators were used
to project the 1985 NAPAP to 1990. SEDS data were
used as an indicator of emissions growth for the area
source fuel combustion categories and for the gasoline
marketing categories (EPA, 1993b). Particle size
multipliers were applied to estimate PM10 emissions
from TSP estimates (EPA, 1995).
Solvent emissions were estimated from a national
solvent material balance using solvent data from
various marketing surveys (EPA, 1993b). Emissions
are allocated to the county-level based on employment
and population data.
; emission estimates were based on
a 1989 United States Department of Agriculture
(USDA) Forest Service Inventory of PM and air toxics
(USDA, 1989). This inventory of prescribed burning
contained State-level emissions, which were allocated
to the county level using the State-to-county
distribution of emissions in the 1985 NAPAP
Inventory.
Wildfire emissions were taken from estimates
developed for the GCVTC for the 11 GCVTC States
(Western U.S.) (Radian, 1995). The wildfire data in
the GCVTC Inventory represent a detailed survey of
forest fires in the study area. For non-GCVTC States,
emissions are based on the 1985 NAPAP Inventory
values.
PM10 emissions for fugitive dust sources were taken
from the Trends inventory (EPA, 1997a) for
agricultural tilling, agricultural burning, construction
activity, paved roads, unpaved roads, prescribed
burning, and wind erosion. Emissions from beef
cattle feedlots were developed for the NPI. In
general, the Trends Inventory emission estimates are
available at the State level, with the exception of
construction activity emission estimates, which are at
the EPA-region level. These were disaggregated to
the county level based on Census of Agriculture data,
land use, and construction earnings data. Paved and
unpaved road emissions are estimated using the EPA's
OMS PARTS emission factor model combined with
paved and unpaved road VMT estimates based on
FFIWA data. PART 5 reentrained road dust emission
factors depend on the average weight, speed, and
number of wheels of the vehicles traveling on paved
and unpaved roadways, the silt content of roadway
surface material, and precipitation data. The activity
factor for calculating reentrained road dust emissions
is VMT.
Residential wood combustion emissions estimated for
EPA's NET effort were used in Version 3 of the NPI
(EPA, 1993c). Residential wood combustion
emissions include those from traditional masonry
fireplaces, freestanding fireplaces, wood stoves, and
furnaces.
For the States of California and Oregon, 1990
criteria pollutant emissions from the GCVTC
inventory were incorporated for all area source
categories. The data for these two States are based on
State-compiled inventories that are presumably based
on more recent and detailed data than the emissions
in the NAPAP Inventory.
Growth Projections
The base year 1990 area source inventory was
projected to 2000 and 2010 to determine the effects of
CAA controls on future emission levels. Growth in
pollution generating activity to future years was
estimated using the BEA industry GSP and
population projections for most area source
categories. Exceptions include activity indicators for
agricultural tilling and burning and managed (or
prescribed) burning. For example, the USDA has
developed baseline projections of farm acres planted
(USDA, 1998). These data, combined with historical
data back to 1990, for eight major crop types shows
an average annual growth of only 0.38 percent per
year from 1990 to 2007. The BEA GSP projections
for farm result in an annual average growth of 2.0
percent per year. Projections of acres planted
represent better predictors of future activity than GSP
for agricultural tilling, so they were used in this
projection.
EPA's projection guidance states that area source
projections can be made using local studies or surveys,
A-36
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
or through surrogate growth indicators such as BEA.
Because this is a national projection, BEA was chosen
as a consistent data set which could be used across all
regions. Emissions must be allocated to the grid cell
in order to perform air quality modeling; these
projections will not reflect changes in the spatial
patterns of emissions between counties or grid cells,
since State-level growth surrogates are used.
Essentially, BEA GSP projections incorporate inputs
to production, and therefore, reflect future changes in
technology, processes, and efficiency. Ideally,
projections from States and Metropolitan Planning
Organizations would be a more reliable estimate of
growth in population, land use, and employment, but
these were not available in a consistent format for the
entire contiguous United States.
Control Scenarios
The Pre-CAAA scenario for area source emissions
assumes that future year control levels are equal to
those in 1990 with the exception of applying the new
source performance standards (NSPS) for residential
woodstoves. The residential woodstove NSPS affects
emissions of both PM and VOC in all areas. The
Post-CAAA scenario applies future year controls to
model the impact of the 1990 Amendments on
projected emissions.
Changes in agricultural practices are likely to
influence future fugitive dust emissions from activities
like agricultural tilling. In recent years, agricultural
practices such as conservation tillage have been
instituted to provide protection against surface soil
erosion, primarily from water and runoff losses.
These practices have also affected wind erosion losses.
The primary attributes of conservation tillage practices
are: (1) reducing the number of passes by farm
vehicles; and (2) maintaining a higher amount of crop
residue in the soil. With respect to particle emissions
from tilling operations, the reduced number of vehicle
passes through the field is the most important
parameter. The emission rate (using current EPA
estimation methods) is primarily related to the acres
tilled (and therefore, the number of vehicle passes)
and the soil silt content. Assuming the soil silt
content remains the same, reducing the number of
vehicle passes produces a proportional reduction in
emissions. The increased crop residue provided by
conservation tillage acts to help shelter the soil
particles from wind erosion, which reduces soil
depletion and reduces vertical fluxes of particles to the
atmosphere. The increased residue has little, if any,
effect on emissions.
Projections of conservation tillage practices are
that the amount of conservation tillage in 2000 will be
26 percent of total acres tilled. The 26 percent figure
is the level achieved in 1990. The 2010 projection
assumes that conservation tillage increases to 50
percent by 2010. Because the trend toward
conservation tillage appears to result from the 1985
Farm Bill conservation compliance program,
economic influences, and improved
efficiency, the same assumptions are used for
estimating Pre- and Post-CAAA PM10 and PM25
emissions for this category.
Under the Post-CAAA scenario, controls are
implemented in PM NAAs. The controls modeled
depend on the severity of PM nonattainment and the
level of emissions from source categories for which
controls are available. The Post-CAAA projection for
NOX incorporates controls for industrial fuel
combustion emissions to model the effects of
lowering the RACT source size cutoff in ozone
NAAs. Low NOX burners were selected as the
representative NOX control. CAA controls affecting
VOC include controls for Title I (RACT, new CTGs,
stage II vapor recovery, and Federal consumer solvent
controls), Title III MACT standards, and onboard
vapor recovery systems. The same control level is
applied in 2000 and 2010. Future year control levels
for SO2 and CO area source emitters (generally fuel
combustion and fires) were assumed to be equivalent
to 1990 levels under both the Pre-CAAA and Post-
CAAA scenario.
Emission Summary
Table A-12 is a summary of emission projections
by year and scenario at the Tier 2 source category
level.
A-37
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The VOC emission differences between Post-
CAAA and Pre-CAAA scenarios shown in Table A-12
for 2000 and 2010 are largely expected to result from
Federal measures and ozone NAA-specific
requirements to reduce ozone precursor emissions.
Area source categories with the biggest differences
between Post-CAAA and Pre-CAAA VOC emissions
include commercial and consumer solvent use, surface
coating (paints), small graphic arts shops, and
hazardous waste treatment, storage, and disposal
facilities (TSDFs). Note that while many hazardous
waste TSDFs are large enough emitters to be classified
as point sources, they are represented in the NPI as
area sources. Service station VOC emissions are
reduced in 2000 primarily by Stage II vapor recovery
systems installed in NAAs, with further reductions in
refueling emissions expected by 2010 as onboard
vapor recovery systems are installed in new cars and
light trucks.
Only modest NOX emission benefits are expected
from CAA mandates for area sources, as most CAA
initiatives focus on major stationary sources. Some
NOX reducing measures, however, do affect the area
source category; some controls are required in serious,
severe, and extreme ozone NAAs emitting 25 tons per
year, or less. In addition, some sources emitting 25 to
100 tons of NOX per year are represented in the area
source emissions file. Area source PM10 emitters with
differences between the Post-CAAA and Pre-CAAA
scenarios are fugitive dust sources.
A-38
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableA-12
Area Source Emission Summary
(thousand tons)
Pollutant/Source Category
VOC
Fuel Comb. Industrial
Fuel Comb. Other
Chemical & Allied Product Mfg
Petroleum & Related Industries
Other Industrial Processes
Solvent Utilization
Storage & Transport
Waste Disposal & Recycling
Natural Sources
Miscellaneous
TOTAL
NOX
Fuel Comb. Industrial
Fuel Comb. Other
Petroleum & Related Industries
Other Industrial Processes
Waste Disposal & Recycling
Miscellaneous
TOTAL
CO
Fuel Comb. Industrial
Fuel Comb. Other
Petroleum & Related Industries
Other Industrial Processes
Waste Disposal & Recycling
Miscellaneous
TOTAL
SO2
Fuel Comb. Industrial
Fuel Comb. Other
Petroleum & Related Industries
Other Industrial Processes
Waste Disposal & Recycling
Miscellaneous
TOTAL
by Pollutant For
1990
17.8
686.0
449.2
450.2
84.4
4,701.0
1,220.3
2,154.6
13.8
568.6
10,345.9
1,269.7
611.2
19.4
4.3
60.0
224.1
2,188.8
192.6
3,759.1
3.9
2.0
1,401.9
6,246.0
11,605.5
626.9
390.6
1.4
1.7
14.8
6.2
1,041.5
1990,2000,
2000
Pre-CAAA
22.4
623.6
517.0
454.8
94.8
5,459.2
1,537.8
2,596.5
13.8
621.6
11,941.7
1,615.6
702.7
15.8
5.2
67.3
228.1
2,634.8
244.3
4,674.4
3.2
2.7
1,542.9
6,477.6
12,945.0
803.6
463.8
1.2
2.2
17.1
6.3
1,294.0
and 2010*
2000
Post-CAAA
22.4
623.6
366.3
198.5
93.1
4,290.7
1,328.8
471.3
13.8
584.9
7,993.5
1,600.2
702.7
15.8
5.2
67.3
228.1
2,619.4
244.3
4,674.4
3.2
2.7
1,542.9
6,477.6
12,945.0
803.6
463.8
1.2
2.2
17.1
6.3
1,294.0
2010
Pre-CAAA
26.6
518.2
578.5
494.5
108.0
6,146.6
1,744.6
3,030.2
13.8
656.8
13,317.7
1,917.8
797.7
15.0
6.0
74.5
230.7
3,041.7
289.5
5,485.9
3.0
3.1
1,672.8
6,625.6
14,080.0
948.4
541.5
1.1
2.5
19.6
6.3
1,519.4
2010
Post-CAAA
26.6
518.2
408.5
207.7
106.2
4,780.6
1,298.1
524.2
13.8
656.8
8,540.6
1,900.1
797.7
15.0
6.0
74.5
230.7
3,023.9
289.5
5,485.9
3.0
3.1
1,672.8
6,625.6
14,080.0
948.4
541.5
1.1
2.5
19.6
6.3
1,519.4
A-39
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pollutant/Source Category
PM10
Fuel Comb. Industrial
Fuel Comb. Other
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Waste Disposal & Recycling
Natural Sources
Miscellaneous
TOTAL
PM,5
Fuel Comb. Industrial
Fuel Comb. Other
Metals Processing
Petroleum & Related Industries
Other Industrial Processes
Waste Disposal & Recycling
Natural Sources
Miscellaneous
TOTAL
NH3
Fuel Comb. Industrial
Fuel Comb. Other
Other Industrial Processes
Waste Disposal & Recycling
Miscellaneous
TOTAL
1990
29.1
510.7
<0.1
1.6
34.6
218.2
2,092.4
23,501.7
26,388.4
14.8
495.8
<0.1
1.6
22.6
190.7
313.9
4,769.0
5,808.0
7.3
7.7
35.5
81.8
3,593.8
3,726.1
2000
Pre-CAAA
37.1
470.1
<0.1
1.3
37.9
240.7
2,092.4
23,915.2
26,794.7
18.8
453.0
<0.1
1.3
25.1
210.1
313.9
5,016.9
6,038.7
9.1
8.8
42.5
100.7
4,650.6
4,811.6
2000
Post-CAAA
37.1
469.6
<0.1
1.3
37.9
240.7
2,092.4
23,262.0
26,141.0
18.8
452.2
<0.1
1.3
25.1
210.1
313.9
4,911.3
5,932.7
9.1
8.8
42.5
100.7
4,650.6
4,811.6
2010
Pre-CAAA
43.8
398.8
<0.1
1.2
44.2
261.6
2,092.4
23,947.9
26,790.0
22.2
379.3
<0.1
1.2
29.3
228.1
313.9
5,296.1
6,270.2
10.7
10.0
49.7
119.2
5,542.2
5,731.8
2010
Post-CAAA
43.8
398.3
<0.1
1.2
44.2
261.6
2,092.4
23,189.6
26,031.3
22.2
378.3
<0.1
1.2
29.3
228.1
313.9
5,173.0
6,146.0
10.7
10.0
49.7
119.2
5,542.2
5,731.8
The totals reflect emissions for the 48 contiguous States, excluding Alaska and Hawaii. Totals may not add due to rounding.
A-40
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Reasonable Further Progress
Requirements
The Post-CAAA scenario incorporates the effect
that Title I ozone controls will have on the emissions
of VOC and NOX, both ozone precursors. The ozone
nonattainment provisions in this section of the 1990
CAAA state that areas not in compliance with the
ozone national ambient air quality standard (NAAQS)
must reduce precursor emissions. NAAs classified as
moderate, serious, severe, or extreme all are directed
to meet the required rate of progress (ROP), and by
1996, cut annual VOC emissions by 15 percent from
1990 levels. In addition, serious, severe, and extreme
ozone NAAs must continue reducing precursor
emissions after the 1996 ROP deadline. Areas falling
into these three categories are required to make
reasonable further progress (RFP) towards attainment.
To satisfy this regulation NAAs must cut VOC levels
by 3 percent annually until they comply with the
ozone NAAQS.6
For areas to comply with ROP requirements VOC
reductions are mandated. For those NAAs that must
make additional cuts and satisfy RFP regulations it is
possible for NOX reductions to be substituted for
VOC cuts. This trading of one ozone precursor for
another is acceptable as long as: 1) a NAA has not
been given a NOX waiver, and 2) the substitution of
NOX for VOC does not result in a greater reduction of
NOx than is necessary for an area to comply with the
ozone NAAQS.
The ROP and RFP requirements are designed to
establish a minimum standard for reducing ozone
precursor emissions. In many cases, nonattainment
areas satisfy these two regulations simply by
complying with other ozone provisions of the CAAA.
Reduction of VOC and NO,, below 1990 baseline
levels, made in order to meet other standards, are
6 The prospective analysis does not model for attainment.
For the purposes of this analysis NAAs are assumed to make
reasonable further progress until their respective attainment
deadlines, as outlined in the CAAA, are reached.
credited towards ROP/RFP requirements. These
credited reductions, although captured by the Post-
CAAA scenario, are not a direct result of ROP/RFP
standards. To accurately capture the influence of
ROP/RFP requirements in the Post-CAAA scenario
it was necessary to predict which NAAs would have
to make emissions cuts solely for the purpose of
satisfying these progress requirements, which
precursor(s) would be cut, what the size of the cuts
would be, and which sources would be forced to make
these cuts.
For the purpose of the prospective analysis it was
assumed that NAAs working to satisfy ROP and RFP
requirements would, whenever possible, first take
credit for all available NOX reductions and then for all
available VOC reductions; any remaining shortfall,
corresponding to cuts that NAAs would have to make
specifically to meet ROP or RFP requirements, would
be made up through additional VOC emission
reductions. The size of the total shortfall for all
NAAs, thus, is a measure of the impact of these Title
I progress requirements on the emission of ozone
precursors. This shortfall is captured by the Post-
CAAA scenario.
To estimate the VOC shortfall, a separate daily
VOC target and daily NOX target was calculated for
the years 2000 and 2010 for every NAA subject to
Title I progress requirements. These target figures
represent the daily maximum allowable emissions
levels that NAAs cannot exceed if they are to comply
with ROP/RFP standards. The NOX target was set
according to how much NOX credit it was assumed
will be counted towards ROP/RFP requirements, and
the VOC target was set based on the assumption that
the remainder of the emissions cuts needed to satisfy
ROP/RFP requirements will come through
reductions in VOC. The difference between the VOC
target and the expected level of daily VOC emissions
in the absence of ROP/RFP requirements,
represented by ozone season daily (OSD) emissions
estimates, equals the shortfall.
Table A-13 shows, for each NAA subject to
ROP/RFP requirements, both the OSD and target
VOC and NO,, emissions levels for 2000. Table A-14
A-41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
contains the same information for projection year
2010. In addition, these tables indicate whether or not
an area has a NOX waiver (these areas cannot be given
NOX credit towards RFP), and reveal how RFP
reductions were divided between VOC and NOX in
order to calculate the target values for both ozone
precursors. For example, looking at Table A-13
shows that for Philadelphia in the year 2000, NOX
reductions of 18 percent from 1990 base year levels
are credited towards satisfying RFP provisions. After
accounting for this credit, the maximum daily VOC
emissions that is allowable for Philadelphia, if this area
is to comply with progress requirements, is 1,376.55
tons. Since this VOC target value is greater than the
predicted level of daily emissions (VOC OSD) of
1,090.49 tons, there is no shortfall. For Philadelphia
in the year 2010 (Table A-14), however, an additional
9 percent reduction from 1990 VOC emission levels
is necessary in order to satisfy RFP requirements.
Once again there is no shortfall.
After calculating the shortfall for each NAA, EPA
estimated how these additional VOC reductions
would be achieved. The Agency compiled a list of the
available area and point source controls and assumed
that each NAA would adopt the most cost-effective
measure, followed by the second most cost-effective
option, and so on until the area satisfied its ROP/RFP
requirements. Table A-15 displays the control option
identified by EPA and lists them in the approximate
order the Agency believed they would be selected by
NAAs attempting to eliminate shortfall emissions. In
general, it is assumed that nonroad controls would be
the last to be selected, along with AIM coatings.
Furthermore, since most areas in the analysis already
have reformulated gasoline and I/M programs in the
baseline, these were not considered as potential
discretionary measures. It should be noted that
episodic bans are not creditable towards ROP
shortfalls. These measures were chosen to be
illustrative of what individual areas might select to
meet ROP shortfalls, however, individual areas may
select measures that differ from those modeled here.
Some of the measures may be politically infeasible for
some areas, however, the cost thresholds used to
estimate the costs of these measures are consistent
with those used in the ozone NAAQS analysis.
Shortfalls are the greatest and most difficult to
eliminate in severe ozone NAAs with NOX waivers.
These areas have to meet significant ozone precursor
reduction requirements solely by cutting VOC
emissions. In Chicago and Milwaukee-Racine, both
wavier areas, the VOC shortfall is so large that it
cannot be eliminated even if all of the identified
controls are implemented. As a result, for these two
NAAs it is assumed that unidentified controls are
adopted to reduce the remaining shortfall. In all other
areas, however, the required reduction can be achieved
with the identified controls.
A-42
-------
TableA-13
2000 Rate of Progress Analysis
Att.
Class.
Ser.
Mod.
Sev.
Ser.
Ser.
Ser.
Sev.
Mod.
Mod.
Mod.
Ser.
Mod.
Ser.
Sev.
Mod.
Mod.
Mod.
Ext.
Mod.
Mod.
Sev.
Mod.
Mod.
Mod.
Sev.
Sev.
Mod.
Mod.
Mod.
Ser.
Ser.
Mod.
Att.
Date
1999
1996
2005
1999
1999
1999
2007
1996
1996
1996
1999
1996
1999
2007
1996
1996
1996
2010
1996
1996
2007
1996
1996
1996
2007
2005
1996
1996
1996
1999
1999
1996
Ozone Nonattainment Area
Atlanta
Atlantic City
Baltimore
Baton Rouge
Beaumont-Port Arthur
Boston-Lawrence-Worcester-E.MA
Chicago-Gary-Lake County
Cincinnati-Hamilton
Cleveland-Akron-Lorain
Dallas-Fort Worth
El Paso
Grand Rapids
Greater Connecticut
Houston-Galveston-Brazoria
Kewaunee Co Wi
Knox & Lincoln Cos ME
Lewiston-Auburn ME
Los Angeles-South Coast
Louisville
Manitowoc Co WI
Milwaukee-Racine
Monterey Bay
Muskegon
Nashville
NewYork-N New Jersey-Long Is
Philadelphia-Wilmingtn-Trenton
Phoenix
Pittsburgh-Beaver Valley
Portland ME
Portsmouth-Dover-Rochester
Providence
Readinq PA
2000 OSD (tons per day)1
NO, VOC Target Selection2
420.93
46.11
335.42
449.70
245.20
560.20
1,056.70
378.70
369.12
528.83
113.35
129.61
211.67
1,094.58
3.23
9.26
23.56
1,010.32
264.00
13.69
272.22
78.60
34.25
167.12
1,280.26
678.53
404.47
534.53
53.72
37.55
92.51
48.66
518.94
37.97
318.33
203.77
340.66
822.65
1,240.89
305.34
521.14
694.53
85.38
182.72
316.27
1,426.65
4.86
9.95
34.08
972.91
219.66
19.72
327.09
64.16
46.67
231.71
1,994.96
1,090.49
377.43
407.05
70.05
53.54
173.78
60.53
1%NOX/8%VOC
ROP-15%VOC
18%NOX
NOX waiver
NOX waiver
9% NOX
NOX waiver
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
0%NOX/9%VOC
ROP-15%VOC
9%NOX
NOX waiver
ROP-15VOC
ROP-15%VOC
ROP-15%VOC
1%NCyi7%VOC
ROP-15%VOC
ROP-15%VOC
NOX waiver
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
18%NOX
18%NOX
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
9%NOX
5%NOX/4%VOC
ROP-15%VOC
NOX
Target
426.28
0.00
381.53
0.00
0.00
602.06
0.00
0.00
0.00
0.00
0.00
0.00
224.49
0.00
0.00
0.00
0.00
1,019.06
0.00
0.00
0.00
0.00
0.00
0.00
1,553.41
690.79
0.00
0.00
0.00
44.26
91.77
0.00
VOC
Target
476.35
45.71
376.47
415.65
450.66
918.01
1,202.25
341.18
573.07
673.97
69.02
175.66
370.11
2,268.31
4.56
10.37
35.98
939.08
215.97
17.20
293.24
79.63
44.32
205.60
2,407.97
1,376.55
347.91
399.80
73.33
58.70
180.51
61.14
Shortfall
42.59
0.00
0.00
0.00
0.00
0.00
38.64
0.00
0.00
20.56
16.36
7.06
0.00
0.00
0.30
0.00
0.00
33.83
3.69
2.52
33.85
0.00
2.35
26.11
0.00
0.00
29.52
7.25
0.00
0.00
0.00
0.00
A-43
-------
Att.
Class.
Mod.
Sev.
Mod.
Ser.
Ser.
Mod.
Mod.
Sev.
Ser.
Mod.
Sev.
Ser.
NOTES:
Att.
Date
1996
2005
1996
1999
1999
1996
1996
2007
1999
1996
2005
1999
2000 OSD (tons per day)1
Ozone Nonattainment Area NOy VOC
Richmond-Petersburg 141.62 179.97
Sacramento Metro 164.60 158.01
Salt Lake City 178.25 182.75
San Diego 228.55 192.90
San Joaquin Valley 499.25 470.50
Santa Barbara-Santa Maria-Lomp 63.77 82.75
Sheyboygan 38.02 24.49
Southeast Desert Modified 355.88 227.71
Springfield/Pittsfield-W. MA 112.38 155.51
St. Louis 476.98 465.64
Ventura Co CA 80.45 65.69
Washington DC 449.80 402.76
1OSD = ozone season daily
2The target selection column indicates the percentage reduction of
in VOC is then needed to satisfy Title I progress requirements.
Target Selection2
ROP-15%VOC
0%NOX/18%VOC
ROP-15%VOC
0%NOX/9%VOC
5%NOX/4%VOC
ROP-15%VOC
ROP-15%VOC
3%NOX/15%VOC
0%NOX/9%VOC
ROP-15%VOC
13%NOX/5%VOC
9%NOX
NOX
Target
0.00
0.00
0.00
0.00
505.63
0.00
0.00
358.76
0.00
0.00
79.55
499.12
NOX NAA's are credited towards RFP and
VOC
Target Shortfall
201.70
155.08
150.80
189.71
532.41
83.10
22.44
219.32
152.42
549.14
70.52
477.03
what percentage
0.00
2.93
31.95
3.19
0.00
0.00
2.05
8.39
3.09
0.00
0.00
0.00
reduction
A-44
-------
TableA-14
2010 Rate of Progress Analysis
Att.
Class.
Ser.
Mod.
Sev.
Ser.
Ser.
Ser.
Sev.
Mod.
Mod.
Mod.
Ser.
Mod.
Ser.
Sev.
Mod.
Mod.
Mod.
Ext.
Mod.
Mod.
Sev.
Mod.
Mod.
Mod.
Sev.
Sev.
Att.
Date
1999
1996
2005
1999
1999
1999
2007
1996
1996
1996
1999
1996
1999
2007
1996
1996
1996
2010
1996
1996
2007
1996
1996
1996
2007
2005
Ozone Nonattainment Area
Atlanta
Atlantic City
Baltimore
Baton Rouge
Beaumont-Port Arthur
Boston-Lawrence-Worcester-E. MA
Chicago-Gary-Lake County
Cincinnati-Hamilton
Cleveland-Akron-Lorain
Dallas-Fort Worth
El Paso
Grand Rapids
Greater Connecticut
Houston-Galveston-Brazoria
Kewaunee Co Wi
Knox & Lincoln Cos ME
Lewiston-Auburn ME
Los Angeles-South Coast
Louisville
Manitowoc Co WI
Milwaukee-Racine
Monterey Bay
Muskegon
Nashville
New York-N New Jersey-Long Is
Philadelphia-Wilminqtn-Trenton
2010 OSD (tons per day)1
NO, VOC Target Selection2
336.10
39.46
278.64
407.49
232.32
478.53
990.61
288.20
306.32
472.66
113.85
92.07
192.62
1,002.13
2.62
8.51
22.20
950.39
273.40
12.10
246.74
71.93
26.99
143.58
1,148.95
631.39
492.40
33.21
293.56
206.65
377.63
775.66
1,236.73
283.74
485.90
687.15
84.33
183.11
292.53
1,530.07
4.77
8.98
32.03
847.66
216.86
19.37
321.89
61.76
46.86
230.00
1,842.53
1,070.05
9%NOX
ROP-15%VOC
27%NOX
NOX waiver
NOX waiver
9%NOX
NOX waiver
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
0%NOX/9%VOC
ROP-15%VOC
9%NOX
NOX waiver
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
5%NOX/31%VOC
ROP-15%VOC
ROP-15%VOC
NOX waiver
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
36%NOX
18%NOV/9%VOC
NOX
Target
391.84
0.00
311.86
0.00
0.00
602.06
0.00
0.00
0.00
0.00
0.00
0.00
224.49
0.00
0.00
0.00
0.00
964.60
0.00
0.00
0.00
0.00
0.00
0.00
1,166.20
632.97
VOC
Target
541.99
45.71
376.47
415.65
450.66
918.01
840.15
341.18
573.07
673.97
69.02
175.66
370.11
1,606.75
4.56
10.37
35.98
670.95
215.97
17.20
204.72
79.63
44.32
205.60
2,407.97
1,194.41
Shortfall
0.00
0.00
0.00
0.00
0.00
0.00
396.58
0.00
0.00
13.18
15.31
7.45
0.00
0.00
0.21
0.00
0.00
176.71
0.89
2.17
117.17
0.00
2.54
24.40
0.00
0.00
A-45
-------
Att.
Class.
Mod.
Mod.
Mod.
Ser.
Ser.
Mod.
Mod.
Sev.
Mod.
Ser.
Ser.
Mod.
Mod.
Sev.
Ser.
Mod.
Sev.
Ser.
Notes:
Att.
Date
1996
1996
1996
1999
1999
1996
1996
2005
1996
1999
1999
1996
1996
2007
1999
1996
2005
1999
Ozone Nonattainment Area
Phoenix
Pittsburgh-Beaver Valley
Portland ME
Portsmouth-Dover-Rochester
Providence
Reading PA
Richmond-Petersburg
Sacramento Metro
Salt Lake City
San Diego
San Joaquin Valley
Santa Barbara-Santa Maria-Lomp
Sheyboygan
Southeast Desert Modified
Springfield/Pittsfield-W. MA
St. Louis
Ventura Co CA
Washington DC
1OSD = ozone season daily
2010
NO,
395.64
368.47
51.54
33.08
75.73
46.61
137.12
152.59
179.22
213.59
466.55
59.33
34.84
329.75
97.63
381.48
75.89
393.57
2The target selection column indicates the percentage reduction
OSD (tons per day)1
VOC
347.52
358.67
66.88
52.49
166.61
55.33
179.35
135.99
189.83
174.04
448.37
81.53
24.69
213.87
147.45
439.47
62.33
355.35
Target Selection2
ROP-15%VOC
ROP-15%VOC
ROP-15%VOC
9%NOX
9%NOX
ROP-15%VOC
ROP-15%VOC
1%NOX/26%VOC
ROP-15%VOC
5%NOX/4%VOC
9%NOX
ROP-15%VOC
ROP-15%VOC
9%NOX/27%VOC
7%NOX/2%VOC
ROP-15%VOC
13%NOX/14%VO
9%NOX
of NOX NAA's are credited towards
NOX
Target
0.00
0.00
0.00
44.26
87.91
0.00
0.00
152.62
0.00
213.42
484.34
0.00
0.00
332.55
97.87
0.00
76.34
499.12
RFP and what
VOC
Target
347.91
399.80
73.33
58.70
193.15
61.14
201.70
120.24
150.80
202.24
566.96
83.10
22.44
172.34
166.51
549.14
63.11
477.03
percentage
Shortfall
0.00
0.00
0.00
0.00
0.00
0.00
0.00
15.75
39.03
0.00
0.00
0.00
2.25
41.53
0.00
0.00
0.00
0.00
reduction in
VOC is then needed to satisfy Title 1 progress requirements.
A-46
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableA-15
Discretionary Control Measures Modeled For ROP/RFP
Source Category
Measure
Reduction
Area Source
Adhesives - industrial
Metal product surface
coating
Cutback asphalt
Wood product surface
coating
Wood furniture surface
coating
Degreasing
Open burning
Automobile refinishing
Bulk Terminals
POTWs
Bakeries
Petroleum dry cleaning
Perchloroethylene dry
cleaning
Livestock
Miscellaneous surface
coating
Aerosols
Incineration
Synthetic fiber manufacture
Misc. industrial processes
Consumer solvents
AIM coatings
Lawn & garden
Recreational vehicles
Industrial equipment
Recreational marine
Point Source
Open burning
Industrial surface coating
Metal product surface
coating
Wood product surface
coating
Point sources
Reformulation
VOC content limits & improved transfer efficiency
Switch to emulsified asphalts (100% RE)
Reformulation
Reformulation
Solvent Limits
Seasonal ban
CARS Best Available Retrofit Control Technology
(BARCT) limits
Leak Detection and Repair (LDAR)
Covers/Adsorption
Afterburner
Recovery dryers
Recovery dryers
Recovery system
Reformulation
South Coast Air Quality Management District
(SCAQMD) Standards - Reformulation
Seasonal ban
Adsorber
Process change/incineration
Additional reformulation
Additional reformulation
Episodic Ban
Episodic Ban
Episodic Ban
Episodic Ban
Seasonal Ban
Add-on Control Levels
Reformulation
Reformulation
Rule effectiveness improvements (80% RE to
90% RE)
63%
30% (50% in 2010 for difficult
areas)
100%
43%
43%
63% (80% in difficult areas)
80%
47%
90% (80% RE)
50%
33%
44%
70% (80%RE)
50%
30%
50% (60% in 2010)
80%
78% (80%RE)
50%
40% (50% in 2010)
40% (50% in 2010)
50%
50%
50%
50%
80%
90%
88%
85%
'"Episodic" and "seasonal ban" controls are not creditable toward the ROP shortfall.
A-47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Mercury Emission Estimates
EPA, as part of this analysis, also estimates the
effect of CAAA provisions on airborne mercury (Hg)
emissions for five separate Hg emissions sources:
medical waste incinerators (MWI), municipal waste
combustors (MWCs), electric utility plants, hazardous
waste combustors, and chlor-alkali plants.7 While the
Integrated Planning Model (IPM) was used to generate
Pre- and Post-CAAA electric utility Hg emissions
estimates, data from previously conducted analyses
were relied on to estimate Hg emissions from the
other sources. The following section provides a
description of the methods and data sources used to
develop 1990 base-year Hg emissions and 2000 and
2010 mercury emission projections for the Pre- and
Post-CAAA scenarios.
Medical Waste Incinerators (MWI)
During the Maximum Achievable Control
Technology (MACT) development process for MWI,
1990 emission estimates for hazardous air pollutants
(HAPs), including mercury (Hg), were developed by
back-casting 1995 emission estimates (EPA, 1996b).
The back-casting was performed by adding to the
1995 database MWIs that had shut down during the
1990-1995 time period. These MWI were shut down
as a result of economic considerations prompted by
the adoption of strict regulations in six states
(California, New Mexico, New York, Oregon,
Washington, and Wisconsin). The number of MWIs
in the remaining states were assumed to be the same
in 1990 as in 1995.
Based on the back-casting analysis described
above, EPA estimated Hg emissions of 16 tons per
year (tpy) in 1995 and 50 tpy in 1990 (Cocca, 1997).
Therefore, 50 tpy is selected as the base year estimate
for this analysis. To estimate the Pre-CAAA scenario
forecasts, it is assumed that no other states would
have adopted MWI regulations between 1995 and
2010. The annual growth rates estimated by the
Bureau of Economic Analysis (BEA) for health
services are 2.3 percent from 1990 to 2000 and also
2.3 percent from 1990 to 2010 (Pechan, 1998). The
Pre-CAAA forecasts were estimated by multiplying 16
tpy in 1995 by a 1.12 growth factor for 2000 and a
1.41 growth factor for 2010. This yielded 17.9 tpy for
2000 and 22.6 tpy for 2010 (see Table A-16).
EPA estimated a 93 to 95 percent reduction of
Hg nationally for existing units in the Emission
Guidelines (which incorporate the MACT standards
for MWI; EPA, 1996c). A 45 to 74 percent reduction
was estimated to occur within 5 years of New Source
Performance Standard (NSPS) promulgation for new
sources (these sources were estimated to produce 0.2
tpy without the standard by 2002). The combined
effects of the NSPS/EG are approximately a 93
percent reduction (from 1995) in the years 2000 and
2010. This emission reduction estimate was used to
calculate the Post-CAAA emissions for MWI (see
Table A-16). No other CAAA regulatory efforts are
known that would further impact emissions from
MWI.
Municipal Waste Combustors (MWCs)
Methods and sources of information used to
estimate Hg emissions for MWC are given in a
supporting document for the NSPS/EG for MWC
promulgated in 1995 (EPA, 1996c). EPA estimated
that there was 54 tpy emitted by MWC in 1990.
Between 1990 and 1995, several factors contributed to
a decline in Hg emissions from MWC: addition of air
pollution controls on existing facilities; retirement of
some units; a decrease in the Hg content of municipal
waste being burned. EPA estimated that there were
29 tpy of Hg emitted by MWC in 1995 (EPA, 1996c).
For the purposes of this analysis, it is assumed that
these reductions occurred due to influences unrelated
to the CAAA. It is further assumed that no additional
reductions would have occurred between 1995 and
2000 (or 2010) without promulgation of the
NSPS/EG.
Together, these sources account for 75 to 80 percent of national
anthropogenic airborne Hg emissions.
A-48
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
EPA further estimated that following full
implementation of the NSPS/EG in 2000, national
Hg emissions would be reduced to 4.4 tpy. However,
the NSPS/EG was amended in 1997. As part of the
amendments, MWC units with capacities less than 250
tons/day and cement kilns burning municipal waste
were exempted from the NSPS/EG. Due to this
exemption, EPA estimated that 87 percent of the
national MWC capacity was now covered by the
NSPS/EG (small units were to be covered in a later
rule-making; EPA 1997c). Therefore, to estimate
Post-CAAA 2000 emissions for MWC, the original 4.4
tpy estimate was divided by 0.87 to account for the
exempted units. After adjusting for growth, the 5.5
tpy shown in Table A-16 was estimated for 2000.
Between 1995 and 2000, EPA (1997d) estimated
that municipal solid waste combustion would increase
by 7.5 percent. Therefore, the Pre-CAAA 2000 Hg
emissions were estimated to be 31.2 tpy (29 tpy x
1.075). Between 2000 and 2010, growth in municipal
waste combustion is estimated to be an additional 8.3
percent (EPA, 1996f). This growth factor was applied
to both Pre- and Post-CAAA 2000 emission estimates
to yield the year 2010 estimates.
Electric Utility Generation
In a report to Congress which details
anthropogenic mercury emissions, the EPA estimated
that total 1990 mercury emissions from utility boilers
was 51.3 tpy (EPA, 1996g). The Integrated Planning
Model (IPM) was used to estimate emissions resulting
from electric power generation in 2000 and 2010 using
different control scenarios to reflect the Pre and Post-
CAAA scenarios. The results generated by the IPM
are presented in Table A-16 (EPA, 1996g). The
scenarios modeled did not include any assumptions
about the effects of any MACT standard to be
promulgated by EPA in the future. Any differences in
emissions between the control scenarios are due to
shifts in operation between units or fuel changes
predicted by IPM (including ash content of the coals
as well as switching to natural gas). The modeled
emission estimates are based on information available
at the time, and if the same analysis were performed
today, with currently available inputs, the results could
be different.
Hazardous Waste Combustion
Pechan-Avanti received preliminary draft data
from Industrial Economics on the benefits of the
MACT Standard for hazardous waste combustors
(Yates, 1999). No information was available on
growth for this source category, although the analyses
conducted to date indicate that there might be a slight
contraction (e.g., five percent) of the category between
1990 and 2000 (or 2010; Yates, 1999). Therefore, it
was assumed that there would not be any growth from
the 1990 baseline. Also, it was assumed that emission
reductions would not occur until after 2000, since the
standard will not be fully-implemented until 2002.
Chlor-alkali Plants
EPA documents estimate mercury emissions from
mercury-cell chlor-alkali plants to be 9.8 tpy (EPA,
1998b; EPA, 1998c) in 1990, and 7.1 tpy in 1994
(EPA, 1996h; EPA, 1996i). The number of facilities
which use the mercury-cell process has been declining
since reaching a peak of 35 in 1970. By 1995, there
were 14 facilities, one of which has converted to a
membrane process (no mercury emissions), and
another of which has ceased operation, leaving 12
facilities in 1999. In the past, closures or process
changes have been due to economics rather than
regulations. There are no existing CAAA programs
which influence mercury emissions from this source.
A MACT standard is currently in the development
process, and will not be promulgated for a year or
more (Rosano, 1999;Dungan, 1999).
The MACT standard, when issued, will focus on
both point source and fugitive emissions. There are
generally three point sources at mercury-cell chlor-
alkali plants which emit mercury: hydrogen vents, end-
box vents, and mercury recovery vents. Point source
controls being considered for the MACT standard are
a combination of a cooling device and either a
molecular sieve, or a carbon adsorption system. The
level of control is yet to be determined, pending
A-49
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
further investigation into the typical concentrations of
mercury found at the vents. The MACT standard for
fugitive mercury emissions from the cellhouse has not
been identified at this point, and investigations and
meetings are planned for the near future in order to
quantify typical fugitive emissions (Rosario, 1999).
Assuming that no additional plants will close or
convert between the time of this writing and the end
of 2000, and that mercury emissions from the 12
remaining sources have remained constant since 1994,
and will continue to remain constant, 2000 emissions
for both the Pre and Post-CAAA scenario are
estimated to be 6.0 tpy. This is based on the 7.1 tpy
reported for 1994, minus the 1994 emissions for the
two facilities which no longer have the potential to
emit mercury (Georgia-Pacific in Bellingham, WA
which ceased operations in 1999 and is reported to
have emitted 0.65 tpy in 1994, and LCP Chemicals in
Reigelwood, NC which converted in 1999 and is
reported to have emitted 0.55 tpy).
Future mercury emissions from mercury-cell
chlor-alkali plants are assumed to continue to decline,
following the trend of plant closing and conversions.
Assuming that the decline in the number of mercury-
cell facilities has been linear since 1970, and will
remain linear in the future, a linear regression was
used to estimate the number of facilities in 2010.
Assuming that there will be 12 facilities in 2000, the
linear regression (r2 = 0.98) indicates a decline of
approximately 0.75 facilities per year, which, when
extrapolated to future years, results in an estimate that
there will be approximately four mercury-cell facilities
still in operation in 2010. Assuming that emissions
will decline at a rate proportional to that of the
number of facilities, emissions are estimated to decline
from 6.0 tpy in 2000 to 2.0 tpy in 2010. This estimate
was used as the Pre-CAAA scenario estimate for 2010
(see Table A-16).
In order to account for the possible effect of an
un-promulgated MACT standard on Hg-cell chlor-
alkali plants operating in 2010, the limited information
available was reviewed. This information is principally
made up of a brief summary of test data submitted by
the operating facilities. At some point in the future
after more testing has been performed, the MACT
standard and estimates of the resulting reductions will
be made publicly available (Rosario, 1999).
Currently, there are facilities which control
mercury emissions, some of which control the
emissions to near or below what the MACT limit may
eventually be. Emissions from these sources are
unlikely to be reduced as a result of the MACT
standard being promulgated. However, emissions
from the plants required to upgrade or add control
systems as a result of the MACT standard being
promulgated, will be reduced in future years. Based
on the limited data available, and assuming that the
test data available are typical of what would be
reported by the facilities for which no valid test data
are available, the overall reduction is estimated to be
approximately 35 percent. This estimate is reflected
in the Post-CAAA emission estimate for 2010 in
Table A-16. The difference between 2010 emissions
with and without the CAAA will be greater if the
number of mercury-cell chlor-alkali plants remains
steady at present levels rather than declining at a linear
rate, as is assumed here.
A-50
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableA-16
Airborne Mercury Emission Estimates
2000 Emissions (tons)
1990
Source Category
Medical Waste Incin.
Municipal Waste Comb.
Electric Utility Generation
Hazardous Waste Comb.
Chlor-Alkali Plants
Total CAAA Benefits (Reductions)
Emissions
(tons)
50
54
51.3
6.6
9.8
Pre-
CAAA
17.9
31.2
63.0
6.6
6.0
Post-
CAAA
1.3
5.5
61.1
6.6
6.0
Diff.
16.6
25.7
1.9
0
0
44.2
2010 Emissions (tons)
Pre-
CAAA
22.6
33.8
68.5
6.6
2.0
Post-
CAAA
1.6
6.0
65.4
3.0
1.3
Diff.
21.0
27.8
3.1
3.6
0.7
56.2
Uncertainties in the Emission
Estimates
This discussion is organized according to the
three major sources of uncertainty in the emissions
inventory and emission projections: the base year
emission estimates, economic growth forecasts, and
future year control assumptions.
Base Year Emission Estimates
Of the pollutants covered in this analysis, the
most certain emission estimates are those for SO2.
SO2 is generated during combustion of any sulfur-
containing fuel and is emitted by industrial processes
that consume sulfur-containing raw materials.
Because sulfur emissions are directly related to the fuel
sulfur content, as long as fuel usage and fuel sulfur
content are measured, SO2 emissions, prior to the
imposition of controls, can be precisely estimated
within a narrow range. Electric utilities emit about 70
percent of the SO2 in the United States. Under
existing utility industry regulations, fuel consumption
and sulfur content of fuels are regularly reported to
DOE. Recent comparisons of Continuous Emission
Monitoring (CEM) data for SO2 with estimates based
on SO2 emission factors and fuel consumption for a
sample of plants showed that the two techniques
produced emission estimates within an average of 8
percent at a State level. The difference is due, in part,
to higher fuel consumption numbers reported by
CEM systems, as a result of the missing data
substitution requirements of the acid rain program
(Schott, 1996).
As part of the GCVTC emission inventory (for 11
Western States), uncertainty estimates were developed
for key source sectors, representing over 70 percent of
the emissions (Balentine and Dickson, 1995). SO2
sources examined included copper smelters and motor
vehicles. The uncertainty estimate calculated for SO2
emissions from copper smelting is ± 50 percent.
Diesel and gasoline vehicle emissions have uncertainty
estimates of a factor of +_ 1.5. Most of this
uncertainty is due to the variability in the sulfur
content of the fuels.
After SO2, the next most certain emission
estimates are probably the NOX values. Like SO2,
NOX is a product of fuel combustion. However, there
are two NOX sources in fossil-fuel combustion
(Seinfeld, 1986). The first is the oxidation of
atmospheric molecular nitrogen at the high
temperatures of combustion. NOX formed by this
route is referred to as thermal NOX. The second
source is the oxidation of nitrogen-containing
compounds in the fuel. NOX formed by this path is
called fuel NOX. Since NOX formation is somewhat
more complicated than SO2, emission estimates are
more variable, and uncertain, as well.
A comparison of NOX emissions based on CEM
data and NOY emissions based on AP-42 emission
A-51
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
factors for a sample of utilities in Louisiana resulted in
a difference of 22 percent between the two methods.
The difference is attributable to improved emission
factors resulting from the use of GEM data, rather
than one-time stack tests or AP-42 emission factors
(Schott, 1996).
The level of uncertainty in primary PM10 emission
estimates varies widely by source category. The largest
component of the 1990 PM10 emission estimates is
fugitive dust sources, which include paved and
unpaved roads, construction activity, agricultural
tilling, and windblown dust. The GCVTC study
estimated the uncertainty for unpaved road emissions
to be a factor of ± 4.0. The estimated uncertainty for
PM25 emissions from paved road dust is a factor of +_
1.8.
PM10 emission estimates for large point sources
such as utility boilers would be expected to be less
uncertain than the fugitive dust source estimates,
because these stacks are typically controlled using
baghouses or electrostatic precipitators, with frequent
stack tests to ensure compliance with regulations.
VOC emissions estimates are uncertain because
organics are emitted both as a product of fuel
combustion and through evaporation. Evaporative
emissions are difficult to quantify because of the
associated measurement problems. The GCVTC
study estimated the VOC emissions uncertainty for
motor vehicles to be a factor of +_ 1.5.
Estimates of emissions from solvents and other
evaporative VOC sources are probably even more
uncertain than the motor vehicle VOC emission
estimates. Emission estimates for such sources
typically assume that all of the organic content of
solvent ultimately evaporates. However, usage
patterns determine what time of year these solvents
are released to the atmosphere, and emissions that
occur outside the ozone season may not influence
ozone levels. Solvent emission estimates used in this
study are based on a national material balance.
Solvent emission estimates made by State and local air
pollution control agencies for SIPs typically use per
capita emission factors to estimate solvent emissions.
This will produce different emission estimates than
used in this study.
Growth Forecasts
The 2000 and 2010 emission estimates in this
analysis are influenced by the projected changes in
pollution-generating activity. Inherent uncertainties
and data inadequacies/limitations exist in forecasting
growth for any future period. As a result, it was
necessary in this analysis to use indicators of growth
that may not directly correlate with changes in the
factors that influence emissions. In the previous
chapters of this report, alternative growth forecasts
were presented for major sectors, and the implications
of these alternative forecasts were noted.
The best indicator of pollution-generating activity
is fuel use or some other measure of input/output
that most directly relates to emissions. The key BEA
indicator used in this analysis, GSP, is closely linked
with the pollution-generating activity associated with
many manufacturing industry processes (iron and
steel, petroleum refining, etc.). However, a good
portion of industrial sector emissions are from boilers
and furnaces, whose activity is related to production,
but not as closely as product output. Activities such
as fuel switching may produce different emission
patterns than those reflected in the results of this
study. The modeling methods applied in this study
would only capture such effects for electric utilities,
but not for the industrial sector.
While it is expected that there will be energy
efficiency improvements in the 1990 to 2010 forecast
horizon, potential energy efficiency improvements
have not been incorporated in the growth factors.
The U.S. Department of Energy currently estimates
that energy intensity — the amount of energy used for
each dollar of output in the economy - will decline by
1 percent per year through the time horizon of this
study. If these potential energy efficiency improve-
ments had been incorporated in the 2000 and 2010
emission projections, then both the Pre-CAAA and
Post-CAAA emission estimates would be lower than
those presented in this report.
A-52
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
In general, emissions from the point, area, and
nonroad engine/vehicle sectors are projected to 2000
and 2010 in this analysis based on BEA GSP by State
and industry, and population projections by State.
Source categories were matched with surrogate activity
indicators that represent proxies for emission growth.
The uncertainty of the growth forecasts used in this
analysis is attributable to two factors: the uncertainty
of the projections data used, and the use of surrogate
activity levels to estimate future emission levels.
Throughout this analysis, efforts were made to
identify potential sources of growth surrogates and to
evaluate the impacts of alternative growth factors on
emission projections. The impact of alternative
growth factors on the emission projection results of
this study vary by source category and pollutant. For
example, point source emissions from chemical
manufacturing would increase at an average annual
rate from 0.9 (BEA GSP) to 2.6 (E-GAS/WEFA)
percent per year between 1990 and 2010 depending on
the activity factor used as a surrogate for emissions
growth. In the nonroad vehicle sector, emissions
from aircraft are projected to grow from 2 to 5
percent per year for the 1990-2010 time period,
depending on whether GSP or landing and takeoff
operations (LTO) data are used as the surrogate
growth indicator. Growth projections for the railroad
industry can range from 0.3 percent to 4.4 percent
depending on whether the growth variable is ton-
miles, fuel use, GSP, or earnings. In this analysis,
BEA earnings data were used to represent growth in
emissions for this industry because it was possible to
differentiate growth rates at the State level, and
because the data were available for the relevant years
of this analysis. In future years, industry analysts
predict lower prices per ton-mile in response to
increased competition for rail traffic. To the extent
that future predictions of lower rail transport prices
occur as railroad transport increases, miles traveled
may be a more accurate activity level surrogate for
emissions than earnings. The outlook for the railroad
industry is uncertain, and emissions may be over- or
understated for the 2000 and 2010 scenarios
depending on future industry conditions.
Each of the available variables for projecting
emissions has advantages and disadvantages with
respect to this analysis. The Agency chose growth
surrogates for this analysis based on EPA guidance;
the availability of data for 1990, 2000, and 2010;
geographic detail of projections data; coverage relative
to the detail of the base year inventory; and the
appropriateness of using the variable as a measure of
emissions growth. For this analysis, BEA provided a
consistent data set that could be applied across source
categories and across States.
Future Year Control Assumptions
The uncertainties associated with future year
control assumptions can be grouped in three types:
(1) will the control programs be adopted; (2) will
control programs be as effective as estimated in this
analysis; and (3) will technological shifts produce
enough changes in emission patterns to affect future
year results?
On the first and second issues, there have been
eight years of progress in implementing the CAAA
provisions, and emission trends estimates have shown
that significant emission reductions have occurred in
this period (EPA, 1997a). Relative to expectations
when the CAA was passed, SO2 emission reductions
have occurred at a faster rate than originally
anticipated, while some of the VOC and NOX
emissions have been less than originally anticipated, as
many vehicle emission inspection programs have been
delayed. By 2000, though, any short run perturbations
may have a negligible effect on overall emission
benefits. Future implementation depends on
decisions that EPA makes about Federal rules, such as
commercial/consumer solvent rules and MACT
standards.
Also on the second issue, concerns about the
ability of regulations to achieve expected reductions as
implemented have resulted in some new programs and
techniques for assuring that new programs are
effective. Rule effectiveness discounting is applied to
stationary source controls (other than those for SO2)
in this analysis to account for control equipment
malfunctions and downtime, unrecognized control
A-53
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
responsibility, and gross noncompliance. In addition,
much more continuous emission monitoring is now
required for major SO2 and NOX sources to ensure
that emission limits are met, so many point source
emission reductions should be verifiable. Verifying
area source emission reductions is much more
difficult. Experiments using remote sensing, tracers,
and other real-world measurement tools are being
performed to better assess the effectiveness of motor
vehicle, nonroad engine, and solvent emission control
initiatives.
On the third issue, any major technological
improvements to create lower-polluting systems by
2010 could influence the emission forecasts, and
would be expected to produce more emission benefits
than have been estimated in this study. In-depth
analyses of two sectors, petroleum refining and motor
vehicles, as part of the Section 812 prospective
analysis found no major technology changes that
would significantly alter emission estimates in these
two important sectors to be likely before 2010. For
the motor vehicle industry, research has been focused
on battery-powered electric vehicles since the
California LEV program requirement for zero
emission vehicles (ZEVs) was announced. However,
these electric vehicles will be unlikely to capture more
than 10 percent of LDV sales by 2010.
significant programs associated with the CAAA that
impact VOC emissions from the solvent cleaning
source category are being implemented through local
regulations as part of ozone attainment plans. Chief
among these is the revised SCAQMD Rule 1171
(SCAQMD, 1996). In the revision of this rule,
SCAQMD requires the use of low-VOC solvents (e.g.,
aqueous) for all regulated sources (e.g., those who are
not regulated under Rule 1122, which has equipment
instead of solvent requirements). SCAQMD is still
working on issues surrounding applicability of Rules
1171 and 1122, but has initially estimated VOC
emission reductions of 46 percent associated with the
revisions to Rule 1171 (SCAQMD, 1996).
The refinery sector study identified the significant
post-1990 technological trends in this industry to be
(1) continued investment in downstream processing of
petroleum; (2) increasing refinery capacity utilization;
(3) continued improvement in refining process control
and catalyst use; and (4) steady capital investment in
stationary source controls. None of these trends is
expected to significantly change refinery emission
rates in 2010 apart from the further investment in
stationary source controls (to meet MACT standards,
for instance) and these are accounted for in the post-
CAAA scenario emission estimates.
Air pollution control regulations will be
technology forcing in that many VOC containing
solvents will be re-formulated to low-VOC solvents or
replaced with water-based substitutes. The most
A-54
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
Argonne, 1995. Grand Canyon Visibility Transport Commission, "Development of Emission Control and New
Technology Options for the Grand Canyon Visibility Transport Region, Volume I, Technology Costs,
Performance, and Applicability," Denver, CO, prepared by Argonne National Laboratory, October,
1995.
Balentine and Dickson, 1995. Balentine, Howard W., and Ronald J. Dickson, "Development of Uncertainty
Estimates for the Grand Canyon Visibility Transport Commission Emissions Inventory," in The
Emission Inventory: Programs & Progress, proceedings of a Specialty Conference sponsored
by the Air & Waste Management Association, Research Triangle Park, NC, October 11-13,1995.
BEA, 1995. Bureau of Economic Analysis, "Regional State Projections of Economic Activity and Population
to 2045," U.S. Department of Commerce, Washington, DC, July 1995.
Cocca, 1997: P. Cocca, U.S. EPA Office of Water, e-mail to S. Roe, Pechan-Avanti Group, April 21,1997.
DOE, 1991. U.S. Department of Energy, Energy Information Administration, "State Energy Data Report —
Consumption Estimates 1960-1989," DOE/EIA-0214(89), Washington, DC, May 1991.
DOT, 1992. U.S. Department of Transportation, Federal Highway Administration, "Highway Statistics 1991,"
Washington, DC, FHWA-PL-92-025,1992.
Dungan, 1999: A. Dungan, The Chlorine Institute, personal communication with E. Albright, The Pechan-Avanti
Group, August 11 and 12,1999.
EPA, 1991a. U.S. Environmental Protection Agency, "Procedures for Preparing Emissions Projections," Office
of Air Quality Planning and Standards, Research Triangle Park, NC, EPA-450/4-91-019, July 1991.
EPA, 1991b. U.S. Environmental Protection Agency, "Nonroad Engine and Vehicle Emission Study," Office
of Air and Radiation, Washington, DC, November 1991.
EPA, 1991c. U.S. Environmental Protection Agency, "MOBILE4 Fuel Consumption Model," draft output
provided by the Office of Mobile Sources, Ann Arbor, MI, August 12,1991.
EPA, 1993a. U.S. Environmental Protection Agency, Office of Mobile Sources, "Users Guide to MOBILESa,"
March 1993.
EPA, 1993b. U.S. Environmental Protection Agency, "Regional Interim Emission Inventories (1987-1991),
Volume I: Development Methodologies," EPA-454/R-93-021a, Research Triangle Park, NC, May 1993.
EPA, 1993c. U.S. Environmental Protection Agency, "National Air Pollution Trends, 1900-1992," EPA-454/R-
93-032, Office of Air Quality Planning and Standards, Research Triangle Park, NC, October 1993.
A-55
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
EPA, 1994. U.S. Environmental Protection Agency, Office of Mobile Sources, "Draft User's Guide to PARTS:
A Program for Calculating Particle Emissions From Motor Vehicles," EPA-AA-AQAB-94-2, Ann
Arbor, MI, July 1994.
EPA, 1995. U.S. Environmental Protection Agency, "Compilation of Air Pollutant Emission Factors, Volume
I: Stationary Point and Area Sources," Fifth Edition (AP-42), Office of Air Quality Planning and
Standards, January 1995.
EPA, 1996a. U.S. Environmental Protection Agency, "National Air Quality and Emissions Trends Report,
1995," Office of Air Quality Planning and Standards, EPA-454/R-96-005, October 1996.
EPA, 1997a. U.S. Environmental Protection Agency, "National Air Pollutant Emission Trends, 1900-1996,"
EPA-454/R-97-011, Office of Air Quality Planning and Standards, Research Triangle Park, NC,
December 1997.
EPA, 1997b. U.S. Environmental Protection Agency, "Air Emissions Estimates from Electric Power Generation
for the CAAA Section 812 Prospective Study," Office of Air and Radiation, February 1997.
EPA, 1998a. U.S. Environmental Protection Agency, "National Air Quality and Emissions Trends Report,1997,"
EPA-454/R-98-016, Office of Air Quality Planning and Standards, Research Triangle Park NC,
December 1998.
EPA, 1996b: National Dioxin Emissions from Medical Waste Incinerators, U.S. EPA, Emission Standards Division,
Office of Air Quality Planning and Standards, Docket #A-91-61, Item IV-A-007, June 1996.
EPA, 1996c: National Mercury Emissions Estimates for Municipal Waste Combustors, U.S. EPA, Emission Standards
Division, Office of Air Quality Planning and Standards, October 1996.
EPA, 1996c: FACT SHEET, Existing Hospital/Medical/Infectious Waste Incinerators — (formerly known as medical waste
incinerators orMWI), PromulgatedSubpart Ce Emission Guidelines, U.S. EPA, downloaded from the EPA TTN
web site, August 1997.
EPA, 1996d: FACT SHEET, New Hospital!Medical!Infectious Waste Incinerators — (formerly known as medical waste
incinerators orMWI), Promulgated Subpart EC Nen> Source Performance Standards, U.S. EPA, downloaded from
the EPA TTN web site, August 1997.
EPA, 1996e: FACT SHEET, Amended Air Emission Regulations for Municipal Waste Combustion (MWC) Units, U.S.
EPA, August 15, 1997.
EPA, 1996f: Characterisation of Municipal Solid Waste in the United States: 1996 Update, U.S. EPA, Office of Solid
Waste and Emergency Response, EPA 530-R-97-015, May 1997.
EPA, 1996g: Air Emissions Estimates from Electric Power Generation for the CAAA Section 812 Prospective Study, U.S.
EPA, Office of Air and Radiation, February 1997.
A-56
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
EPA, 1996h: Mercury Study Report to Congress, Volume II: An Inventory of Anthropogenic Mercury Emissions in the United
States, U.S. EPA, Office of Air Quality Planning & Standards and Office of Research and Development,
EPA-452/R-97-004, December 1997.
EPA, 1996i: Locating and Estimating Air Emissions from Sources of Mercury and Mercury Compounds, U.S. EPA, Office
of Air Quality Planning and Standards, EPA-454/R-97-012, December 1997.
EPA, 1998b: 1990 Emissions Inventory of Forty Section 112(k) Pollutants, Supporting Data for EPA's Proposed Section
112(k) Regulatory Strategy, U.S. EPA, Office of Air Quality Planning & Standards, January 1998.
EPA, 1998c: Emissions Inventory of Section 112(c)(6) Pollutants: Polycyclic Organic Matter (POM), 2,3,7,8-
Tetrachlorodiben^p-p-doixin (TCDD)/ 2,3,7,8-Tetrachlorodiben^pfuran (TCDF), Poly chlorinated Eiphenyl Compounds
(PCBs), Hexachloroben^ene, Mercury, and Alkylated Eead, U.S. EPA, Office of Air Quality Planning &
Standards, April 1998.
Ingalls et al., 1989. Ingalls, Melvin N., Lawrence R. Smith, and Raymond E. Kirksey, "Measurement of On-Road
Vehicle Emission Factors in the South Coast Air Basin: Volume I - Regulated Emissions," Southwest
Research Institute, San Antonio, TX, June 1989.
OTC, 1994. Ozone Transport Commission, "Memorandum of Understanding Among the States of the OTC
on Development of a Regional Strategy Concerning the Control of Stationary Source NOX Emissions,"
September 27, 1994.
Pechan, 1994a. E.H. Pechan & Associates, Inc., "The Emission Reduction and Cost Analysis Model for NOX
(ERCAM-NOx) - Final Report," prepared for U.S. Environmental Protection Agency, Ozone/CO
Programs Branch, May 1994.
Pechan, 1994b. E.H. Pechan & Associates, Inc., "Enhancements to the Emission Reduction and Cost Analysis
Model for VOC (ERCAM-VOQ - Draft Final," prepared for U.S. Environmental Protection Agency,
Office of Air Quality Planning and Standards, Ambient Standards Branch, March 31, 1994.
Pechan, 1994c. E.H. Pechan & Associates, Inc., "Emissions Inventory for the National Particulate Matter Study
- Draft Final," prepared for U.S. Environmental Protection Agency, Office of Policy Planning and
Evaluation/Office of Policy Analysis, July 1994.
Pechan, 1995a. E.H. Pechan & Associates, Inc., "Regional Particulate Strategies - Draft Report," prepared for
U.S. Environmental Protection Agency, Office of Policy Planning and Evaluation. September 29,1995.
Pechan, 1995b. E.H. Pechan & Associates, Inc., Letter to William Kuykendal, U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards, "Updates to Fugitive Emission Components of
the National Particulate Inventory," January 29,1996.
Pechan, 1996. E.H. Pechan & Associates, Inc., "The Emission Reduction and Cost Analysis Model for NOX
(ERCAM-NOx) Revised Documentation," prepared for U.S. Environmental Protection Agency, Ozone
Policy and Strategies Group, September 1996.
A-57
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pechan, 1998: Emission Projections for the Clean Air Act Section 812 Prospective Analysis, prepared by E.H. Pechan &
Associates, Inc., prepared for Industrial Economics, Inc., June 1998.
Rosario, 1999:1. Rosario, U.S. EPA, Office of Air Quality Planning and Standards, personal communications
with E. Albright, The Pechan-Avanti Group, on August 9, 11, and 13,1999.
Radian, 1995. Grand Canyon Visibility Transport Commission, "An Emissions Inventory for Assessing Regional
Haze on the Colorado Plateau," Denver, CO, prepared by Radian Corporation, Sacramento, CA, January
23, 1995.
SCAQMD, 1996. South Coast Air Quality Management District, "Draft Staff Report for Proposed Amendment
to Rule 1171 - Solvent Cleaning Operations," June 14,1996.
Schott, 1996. Schottjim, "Lots of Data, How Do We Use It? Strengths and Inaccuracies of Utility Acid Rain
Electronic Data Reports," Entergy Corporation, paper presented at Air and Waste Management
Association Conference - The Emission Inventory: Key to Planning, Permits, Compliance, and
Reporting, New Orleans, LA, September 1996.
Science & Policy, 1995. Grand Canyon Visibility Transport Commission, "Options for Improving Western
Vistas Volume One, I. Overview and II. Origins, Organization, Process, and Technical Approach,"
Denver, CO, prepared by Science & Policy Associates, Inc., November 4,1995.
Seinfeld, 1986. Seinfeld, John H., Atmospheric Chemistry and Physics of Air Pollution, John Wiley & Sons, 1986.
USD A, 1989. U.S. Department of Agriculture, Forest Service, "An Inventory of Participate Matter and Air Toxic
Emissions from Prescribed Fires in the United States for 1989," Seattle WA, 1989.
USD A, 1998. U.S. Department of Agriculture, "USD A - Agricultural Baseline Projections to 2007," World
Agricultural Outlook Board, Office of the Chief Economist, Staff Report No. WOAB-98-1, 1998.
Wolcott and Kahlbaum, 1991. Wolcott, Mark A., and D.F. Kahlbaum, "The MOBILE4 Fuel Consumption
Model," EPA-AA-TEB-EF-91-X, April 1991.
Yates, 1999: A. Yates, Industrial Economics, Inc., personal communication with S. Roe, The Pechan-Avanti
Group, August 13,1999.
A-58
-------
Direct Costs
Introduction
In this appendix, we present the estimation of
direct compliance costs associated with the Clean Air
Act Amendment programs under Title I through V
that control the following criteria pollutants:
Volatile organic compounds (VOCs)
Oxides of Nitrogen (NO,,)
Carbon monoxide (CO)
Sulfur dioxide (SO2)
P articulate matter with an
aerodynamic diameter of 10 microns
or less (PM10)
Particulate matter with an
aerodynamic diameter of 2.5 microns
or less (PM25)
The first section of the appendix provides a general
overview of our methodology for estimating direct
compliance costs and the models used in the analysis.1
The following section presents costs first by emission
sources and then by CAAA title. Cost by emission
source reviews the specific costing approach (i.e.,
source-specific cost equations or operating cost
estimates), sources of data, and emission control
scenarios applied to five regulated sectors and ozone
nonattainment areas. Costs are also presented by
CAAA title, where the cost components (i.e., the
emission sources and provision) are identified for
Titles I through V. In the following section, we
discusses several additional issues related to fully
accounting for the broader economic consequences of
reallocating resources to the production and use of
pollution abatement equipment (i.e., estimating social
costs versus direct compliance costs). We conclude
with a discussion of analytic limitations and
characterizations of the potential impact of several key
uncertainties of cost estimates.
Summary of Methods
We use two modeling approaches to calculate cost
estimates under Post-CAAA control scenarios in the
projection years, 2000 and 2010. The control
assumptions (i.e. emissions scenarios) used as inputs
in the models are consistent with the assumptions
used in the analysis of both emissions projections and
benefits. The cost data used as parameters in these
models includes results and information from EPA
regulatory impact assessments (RIAs), background
information documents (BIDs), regulatory support
documents, and Federal Register notices.
ERCAM Model
We use ERCAM to estimate the costs associated
with regulating particulate matter (PM), volatile
organic compounds (VOCs), and non-utility oxides of
nitrogen (NO^.2 The model is essentially a cost-
accounting tool that provides a structure for
modifying and updating changes in inputs while
maintaining consistency with the emission and cost
analyses. Cost scenarios and assumptions are
developed for source categories (e.g., point, area,
nonroad, and motor vehicle sources) and in response
to specific provisions and emission targets. The
model estimates costs based on inputs such as cost
per ton, source-specific cost equations, incremental
production, and operating cost estimates. For this
analysis, we collected data and inputs from
information presented in regulatory impact
assessments (RIAs), background information
documents (BIDs), regulatory support documents,
and Federal Register notices.
1 This appendix is a condensed version of more detailed
reports completed under EPA's direction. For more details see
Pechan, 1998.
2 This model was developed by E. H. Pechan & Associates,
Inc. to facilitate EPA's analysis of emissions control.
B-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
IPM Model
We rely on a utility planning model, Integrated
Planning Model (IPM), to estimate the costs of NOX
and SO2 controls for electric utilities. IPM is a linear
program/optimization model that can estimate costs
and emissions based on key constraints and
parameters. One of the significant advantages to this
model is that it provides the analysis with flexibility in
the level of detail for characterizing constraints and
economic assumptions. In this analysis, the model
estimates compliance costs based on assessing the
optimal mix of pollution control strategies subject to
a series of specified constraints. Key inputs to the
model include targeted emissions reductions (on a
seasonal and annual basis), characteristics of control
technology, and economic parameters. The
characteristics of control technology examines
operational costs and constraints associated with the
performance of existing and new utility generating
units. Examples of inputs for existing units include
plant capacity, fuel usage rates, fixed and variable
O&M costs. For new utility generating units, inputs
are generally associated with unit characteristics such
as capacity and costs of capital. Economic
assumptions include the projected electric industry
growth, changes in seasonal and regional demand, and
forecasts of fuel prices.
Additional Methods
We estimate non-utility SO2 emission control
costs for point sources by applying source-specific
cost equations for flue gas desulfurization
(FGD)/scrubber technology to affected sources in
2000 and 2010. While we do not explicitly model CO
attainment costs, we include in the analysis the costs
of programs designed to reduce CO emissions, such
as oxygenated fuels and a cold temperature CO motor
vehicle emission standard.
(O&M) costs.3 They do not represent actual cash flow
in a given year, but rather are an estimate of average
annual burden over the period during which firms will
incur costs (i.e., equipment life). In annualizing costs,
we convert total capital investment, plus O&M and
other re-occurring costs, to a uniform series of per-
year expenditures over a given time period. The
discounted sum of these annual expenditures is equal
to the net present value of total costs incurred over
the time period of this analysis.4
CAAA Costs
We estimate costs of implementing the Clean Air
Act Amendments under two Post-CAAA scenarios,
2000 and 2010. The estimates, therefore, represent
differences in costs between pre- and post-scenarios
in each of the two years. The cost estimates for
implementing Tides I through V of the Clean Air Act
Amendments are $19 billion under the Post-CAAA
2000 scenario and $27 billion under the Post-CAAA
2010 scenario. All costs are in 1990 dollars. This
appendix presents the costs first by source and then
by title.
This section summarizes our costing methods and
results for the following CAAA regulated sectors:
Industrial point sources
Electric utilities
Nonroad engines and vehicles
Motor vehicles
Area sources
Ozone nonattainment areas
Compliance with the CAAA provisions for motor
vehicles is the single largest cost component: $9
billion for the Post-CAAA 2000 scenario, and $12
billion for Post-CAAA 2010. The costs of compliance
Annualization of Costs
The costs presented in this analysis are total
annualized costs (TAQ in 2000 and 2010. Annualized
costs include both capital costs, such as costs of
control equipment, and operation and maintenance
3 For a few VOC source categories, we estimate that capital
investment will not be necessary; for these sources, compliance
costs reflect O&M costs only.
4 We re-calculate the control cost estimates from regulatory
documents that use a seven or ten percent discount rate so that the
costs will be consistent with the five percent discount rate
assumption used in this analysis. We also calculate cost using three
percent and seven percent discount rates, as sensitivity tests: for
detail see the discussion of uncertainty later in this appendix.
B-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
for industrial point sources, utilities, and area sources
are somewhat smaller; they range from $3 to $5 billion
dollars each. Table B-l summarizes the cost estimates
by year and emissions source.
Table B-1
Summary of Cost Estimates by Emissions Source
Annual Cost (million 1990 dollars)
Sector/Pollutant
Total Non-utility Point
Non-utility Point/VOC
Non-utility Point/NOx
Non-utility Point/Non-
VOC MACT1
Utilitv/SO, and NOV
Non-Road Engines/Vehicles
Motor Vehicles
Total Area Sources
Area/VOC
Area/NOx
Area/PM
Progress Requirements
Permits2
TOTAL
Post-CAAA 2000
$ 2,900
900
1,700
310
$3,100
$100
$9,100
$ 2,900
920
16
1,900
$1,200
$300
$ 19,400
Post-CAAA 2010
$ 3,400
960
2,100
320
$ 4,600
$400
$ 12,300
$ 3,300
1,000
18
2,200
$ 2,500
$300
$ 26,800
Notes:
Costs reflect estimates of annualized costs from final rules. Source categories are not modeled in ERCAM-VOC
because the National Emission Standards for Hazardous Air Pollutants (NESHAPs) are associated with non-VOC HAP
emission reductions, and are therefore not included in the Post-CAAA 2000 and 2010 inventories.
These costs include costs only for State-implemented permitting programs. We exclude the costs of Federally-
implemented programs since all Title V permit programs will be State-run in 2005.
Industrial Point Sources
Industrial point sources are non-utility sources
that are large enough to be included in the 1990
emissions database as individual sources of emissions.
To determine the level of air pollution controls
necessary for reducing emissions under the 2000 and
2010 Post-CAAA scenarios, we apply the following
CAAA controls to point source emission inventory:
• Title III 2-year and 4-year MACT
standards for VOCs
• Title I CTGs for controlling VOCs
• Title I VOC and NOX RACT
requirements in ozone NAAs
• A 0.15 Ibs/MMBtu NOX cap on fuel
combustors of 250 MMBtu per hour
or above in the OTAG 37-State
region
• Ozone
NAA
requirements
rate-of-progress
To estimate the quantity and type of VOC
controls, we apply point source Title I RACT and
B-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
CTGs requirements in areas according to ozone
nonattainment classification. The Clean Air Act
requires VOC controls in moderate and above ozone
nonattainment areas (NAA) and throughout the
ozone transport region (OTR). Existing controls are
taken into consideration in our determination of
which CAA-mandated controls are necessary to limit
projected emissions. We use a threshold of ten
percent efficiency for this determination. We calculate
costs for new control if the existing control is less
efficient than the model control by more than ten
percent (i.e. emissions changes of less than ten percent
are assumed to be de minimus and are not included in
the cost estimate).
To estimate the quantity and type of NOX
controls, we apply these controls to the point source
inventory on a year-round basis. The ozone
nonattainment provisions of Title I require installation
of RACT-level controls for major stationary sources
of NOX located in marginal and above NAAs and the
northeast OTR. We determine affected source sizes
according to ozone nonattainment classifications. The
analysis applies the 0.15 Ibs/MMBtu NOX limit to
industrial boilers at or above 250 MMBtu per hour in
the Ozone Transport Assessment Group (OTAG)
region to approximate the effects of NOX initiatives
under consideration. We also account for Title I
requirements that include the application of Level 2
controls in the OTAG region.
Cost Approach
We use ERCAM-VOC and ERCAM-NOX models
for generating cost estimates. Model inputs include
costs per ton and incremental cost estimates derived
from RIAs and from control measure information
provided by EPA, States, industry, and other
agencies.5 Using the projected 2000 and 2010 emission
inventories, we also estimate costs by applying cost
equations to the following individual source
categories:
• Adipic and nitric acid manufacturing
plants
Cement manufacturing
Gas turbines
Glass manufacturing
Industrial boilers
Internal combustion engines
Iron and steel mills
Medical waste incinerators (MWTs)
Municipal waste combustors
(MWCs)
• Process heaters
For some source categories, capital and O&M
cost estimates are available in the literature for two or
more source sizes typical to that category. For these
cases, we apply size-specific cost equations. Operating
characteristics and source size, both of which
influence the ease of retrofit, reduction performance,
and control costs, are major factors in determining
costs of controls. Although site specific
characteristics can affect the overall cost, this type of
information is not available in the emission inventory.
Therefore we model costs based on a "typical" set of
controls.
For source categories with insufficient data, we
estimate annual costs for controls using average cost
per ton values from the ACTs, instead of size specific
cost equations. These values do not account for
economies of scale or variations in capacity factor,
which generally impact the cost per ton of pollutant
reduced.
Recovery Factor
ERCAM-VOC and ERCAM-NOX cost equations
use a five percent discount rate and a 15-year
equipment life, or a capital recovery factor (CRF) of
0.096. To calculate the capital recovery factor for
converting capital charges to equivalent annual costs,
we use the following formula:
CRF = [i * (1 +t)"]/[(1 + if - 1]
where i = pre-tax marginal annual rate of
return (discount rate), and
n = equipment economic life (in years).
5 The Agency bases cost effectiveness values for rules that
have not yet been proposed on engineering judgement and
technology transfer from other categories.
B-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
To obtain annual costs, we use the following
algorithm:
CRC = CKF * Capital Costs,
where CRC = capital recovery cost (or annualized
capital cost).
Cost Results
Table B-2 summarizes estimated point source
VOC control costs. we estimate costs to be
approximately $901 million in 2000; of that total, $421
million will be Title I VOC controls costs and $480
million will result from the Title III MACT Standards.
In 2010, the total annual cost of point source VOC
controls is approximately $962 million: $440 million in
Title I controls and $521 million in Title III controls.
Table B-3 summarizes the point source NOX
control costs under the 2000 and 2010 Post-CAAA
scenarios. OTAG region costs under the 2000 Post-
CAAA scenario total $1.6 billion, increasing to $2.1
billion by 2010. Point source NOX control costs in the
rest of the nation are $21 million under the 2000 Post-
CAAA scenario and $22 million under the 2010 Post-
CAAA scenario. Nationwide, ICI boilers bear the
majority of point source NOX control costs, which
account for seventy-nine percent of the total costs in
2010.
Table B-2
Point Source VOC Cost Summary
Source Category
Annual Costs (million 1990 dollars)1
Post-CAAA 2000
Post-CAAA 2010
National Rules
Marine vessel loading: petroleum liquids $20 $30
TSDFs Less than 0.1 Less than 0.1
New CTGs (moderate)
Printing - lithographic (0.7) (0.7)
SOCMI distillation 0.1 0.1
SOCMI reactor 1.9 2.2
Non-CTG and Group III CTG RACT (moderate and above)
Automobile surface coating 210 220
Bakeries 0.9 1.1
Beverage can surface coating 47 47
Carbon black manufacture 1.2 1.3
Charcoal manufacturing 0.0 0.0
Cold cleaning 17 18
Fabric printing 22 23
Flatwood surface coating 20 21
Leather products 1 1.1
Metal surface coating 51 57
Organic acids manufacture 1.7 2.0
Paint and varnish manufacture 2.5 2.8
Paper surface coating 5.5 5.5
Plastic parts surface coating 5.1 5.3
Rubber tire manufacture 1.4 1.4
SOCMI processes - pharmaceutical 3.7 4.1
Whiskey fermentation - aging 0.2 0.2
B-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Annual Costs (million 1990 dollars)1
Source Category
CTG RACT (marginal and above)
Cellulose acetate manufacture
Dry cleaning - Stoddard solvents
In-line degreasing
Open top degreasing
Printing - letterpress
Terephthalic acid manufacture
Vegetable oil manufacture
Total Title I Costs
NESHAP
Benzene NESHAP
2-Year MACT (national):
Dry Cleaning - PCE
SOCMI HON:
Chemical manufacture
SOCMI - process vents
SOCMI fugitives
SOCMI processes
VOL storage
4-Year MACT (national)
Aerospace industry
Coke Oven Batteries
Gasoline distribution - Stage I
Halogenated solvent cleaning
Marine vessel loading: petroleum liquids2
Petroleum refineries: other sources not distinctly listed
Polymers and Resins Group I
Polymers and Resins Group II
Polymers and Resins Group IV
Printing and Publishing
Shipbuilding and ship repair
Wood furniture surface coating
Total Title III Costs
Total Point Source VOC Control Costs (Title I and Title III)
Post-CAAA 2000
1.5
0.1
(0.3)
(1.0)
0.5
2.3
Less than 0.1
$420
$0.2
2.2
12.0
2.1
(3.9)
22
1.5
3.5
21
12
(8.5)
17
40
110
4.3
5.3
200
0.4
37
$480
$900
Post-CAAA 2010
1.6
0.1
(0.3)
(1.2)
0.5
2.5
Less than 0.1
$440
$0.2
2.7
13.0
2.4
(4.5)
26
1.7
4.7
21
13
(9.1)
20
45
130
5.0
6.7
210
0.5
38
$520
$960
Notes:
1 Control costs reflect growth projections and CAAA control assumptions relative to a 1990 baseline.
2 The costs for the joint MACT/RACT rule for marine vessel loading are allocated between Title I and Title III based on the 58
percent/42 percent distribution in the addendum to the final rule (EPA, 1995b).
B-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-3
Point Source NOX Summary
Annual Costs (million 1990 dollars)
Source Category
RACT (outside of OTAG Region)
Adipicand Nitric Acid Manufacturing
Cement Manufacturing
Gas Turbines
Glass Manufacturing
ICI Boilers
Internal Combustion Engines
Iron & Steel Mills
Waste Combustors
Process Heaters
Subtotal (RACT outside of OTAG Region)
RACT/OTAG Level 2 (OTAG Region)
Adipicand Nitric Acid Manufacturing
Cement Manufacturing
Gas Turbines
Glass Manufacturing
ICI Boilers
Internal Combustion Engines
Iron & Steel Mills
Waste Combustors
Process Heaters
Subtotal (RACT+OTAG Level 2+0.15 Cap)1
Total Point Source NOX Control Costs
Post-CAAA 2000
$<0.1
1.7
0.7
3.1
14
0.8
<0.1
0.1
0.8
$21
$31
97
18
38
1,200
190
2.5
10
21
$1,600
$1,700
Post-CAAA 2010
$<0.1
1.7
0.7
3.2
15
0.8
<0.1
0.1
0.8
$22
$35
110
28
41
1,700
190
2.4
12
23
$2,100
$2,100
Notes:
1 The 0.15 Ibs/MMBtu cap on fuel combustors of 250 MMBtu per hour and above is only applied under the 2010 Post-CAAA
scenario.
* Totals may not add due to rounding.
B-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Utility Sources
The electric power industry is comprised of
entities that generate and sell electricity under two
types of conditions: (i) under firm contracts to electric
utilities; (ii) directly to consumers as electric utilities.
These entities include businesses, governmental
agencies, and cooperative organizations. In this
analysis, we include only independent power
producers and cogeneration units in the contiguous
United States that report to the North American
Electricity Reliability Council (NERC).6 We exclude
a large number of electric utilities that simply
distribute power since those facilities are unlikely to
directly face CAAA regulations.
Scenarios
Our assumptions of electricity demand are based
on NERC's 1994 generation forecast with a slight
downward adjustment to reflect expected changes in
demand due to the Administration's Climate Change
Action Plan. In general, we expect that the industry
will respond to CAAA regulations by adjusting the
mix of fuel types for future generation capacity (i.e.,
increasing electricity generation by combined cycles
and decreasing use of combustion turbines), rather
than significantly altering production levels.
Consequently, modeled differences in total generation
capacity for Pre- and Post-CAAA scenarios are also
relatively small and demand for electricity under both
scenarios is essentially the same.7
The predominant emitters of air pollutants by the
electric power industry are generation units that use
fossil fuels. This includes coal-fired steam, oil/gas-
fired steam, oil/gas combustion turbine, and natural
gas combined cycle units. Under the Pre-CAAA
We do not include trust territories, Alaska, and Hawaii in this
analysis. Trust territories are not directly covered by the CAA.
With respect to Alaska and Hawaii, these States generate such
small amounts of power that excluding them does should not have
a significant effect on the results of this analysis.
7Demand is 3.0 trillion kilowatt hours (kWh) in 2000 and 3.6
trillion kWh in 2010.
regulatory scenario, we fix standards at prevailing 1990
levels. We assume that existing controls of carbon
monoxide and particulate matter remain constant in
both Pre- and Post-CAAA scenarios. The Post-
CAAA regulatory scenario reflect standards that target
these generation units and their emissions of SOX and
NOX.8
In the Pre-CAAA scenario developed for utility
SOX emissions, we assume existing units satisfied State
Implementation Plan (SIP) requirements which
specify unit-specific permits for individual boilers or
plants. Typically, these permits restrict sulfur-content
levels of coal or fuel oil that are burned. In addition,
new coal-fired units must continue to meet the New
Source Performance Standards (NSPS) set in 1978. In
the Post-CAAA scenario, units subject to compliance
with Title IV Acid Rain Allowance Trading program
are existing units that burn fossil fuels and are over 25
megawatts (MW) and all new units that burn fossil
fuels (regardless of size). Lastly, compliance with the
trading program is phased in by 2000.
Under the Pre-CAAA scenario, we do not model
NOX controls on existing sources. New sources must
meet either existing New Source Performance
Standards (NSPS) or Best Available Control
Technology (BACT) standards, whichever is lower.
In the Post-CAAA scenario, existing sources of NOX
emissions are regulated: (i) under Title I, where
existing units comply with RACT requirements in
ozone transport regions (OTR) and non-attainment
areas (NAA), and (ii) under Tide IV, where coal-fired
units must meet with phased requirements by 2000.9
New sources must meet the most stringent standard
among the following, NSPS requirements of Title I,
8Under both regulatory scenarios, we do not account for the
costs of regulating air toxics. The Amendments mandate the
Agency to evaluate the human health impact of utilities' air toxics
emissions. In the case where harmful effects are determined, the
Agency is required to promulgate regulation of their emissions.
The Agency, however, has not reached any conclusions on air
toxic emissions from power plants.
The OTR consists of New England, New York,
Pennsylvania, New Jersey, Delaware, Maryland, District of
Columbia, and sections of northern Virginia.
B-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
BACT requirements of Tide I, or Title IV
requirements. We summarize the Post-CAAA
scenarios for the control of these two pollutants are
summarized in Table B-4.
Table B-4
Differences in the Control of Utility NOX and SOX for the
Pre-CAAA and the Post-CAAA Regulatory Scenarios
Pollutant
Pre-CAAA
Post-CAAA
SOV
Existing units: Comply with State
Implementation Plan (SIP) requirements
prevailing in 1990 to ensure compliance
with the National Ambient Air Quality
Standard.
Existing units: Comply with the Acid Rain
Allowance Program under Title IV of the CAAA
1990 with phased-in requirements. Phase I
covers the largest 110 coal-fired power plants in
1995. All other units above 25 megawatts are
covered in Phase II beginning in 2000.
New units: Comply with New Source
Performance Standards (NSPS) set in
1978 and BACT fixed at 1990 levels
applied through the New Source Review
(NSR) process.
New units: Comply with the NSPS set in 1978,
BACT/LAER (Lowest Achievable Emission
Requirements), and the Acid Rain Allowance
Trading Program under Title IV of the CAAA
1990.
NOV
Existing Units: No federal standards,
except NSPS for new units built after
passage of the law.
Existing units: Meet Reasonably Available Control
Technology (RACT) in 1995 in the OTR and all
non-attainment areas per Title I. States can file
waivers from RACT requirements. Coal-fired
units comply with Title IV NOX requirements that
are phased in overtime, or RACT, whichever is
more stringent. Group 1/Phase I coal-fired units
comply in 1996. Group 1/Phase II and Group 2
coal-fired units comply in 2000. Collective action
by the 37 eastern States in the Ozone Transport
Assessment Group (OTAG) will lead to additional
requirements (known as "Level 2 controls") under
Title I for reducing NOX emissions during the
summer months (May - September).
New units: Units using fossil fuels comply
with the NSPS for each generation
technology and fuel. Application of BACT
in the NSR process at levels existing in
1990.
New units: Comply with Title I NSPS and
BACT/LAER and Title IV standards for coal-fired
units, whichever is more stringent. Units subject
to OTAG Level 2 controls for reducing NOX during
summer months.
B-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Compliance Actions
In order to comply with the Title IV SOX
Allowance Trading program under the Post-CAAA
scenario, the electric power industry must install
continuous emissions monitoring systems. In
addition to the monitoring system, they may be
required to adopt at least one of the following four
types of action:
• Improve the performance of existing
scrubber units and scrubbers that facilities
will build on new units under the NSPS of
the Pre-CAAA Scenario
• Add scrubbers on existing units
• Switch to lower sulfur coals
• Switch over from coal-fired to gas-fired units
We assume in the Section 812 cost analysis that
the electric power industry faces four NOX regulatory
programs. These programs require the industry to:
• Place RACT controls on existing generation
units in States without EPA waivers
• Build new generation units to meet BACT
requirements
• Comply with Title IV NOX rules for new and
existing coal-fired units
• Comply with NOX Cap-and-Trade program
for reducing emissions during the summer
months in the eastern United States
Cost Approach
We configured the IPM to forecast the operation
of the electric power industry from 2000 to 2010. The
baseline case, used in EPA's Clean Air Power Initiative
(CAPI), includes the set of CAAA controls that the
Agency promulgated or States established through
their permit decisions by the middle of 1996. The
baseline case also includes RACT and BACT decisions
under the New Source Review program, Phase I and
Phase II of the Title IV SOX Allowance Trading
Program, and Phase I NOX control requirements
applied to all tangentially-fired and wall-fired boilers
that use coal.
In simplest terms, we sets up the Pre-CAAA
scenario for the electric power industry by removing
the CAAA controls from the CAPI base case and
running the IPM model to forecast emission levels
and costs of producing electric power. We fix
standards under the Pre-CAAA 2000 and Pre-CAAA
2010 scenarios at 1990 levels. To establish the Post-
CAAA scenario, we add further NOX controls to the
CAPI base case, which focuses on the emissions and
costs of producing electric power under the CAAA
Title IV SOX Allowance Trading program. The Post-
CAAA scenario reflects a NOX cap-and-trade program
that EPA presented at OTAG meetings and was
considered, at the time the utility analysis was initiated
(1995-1996), to be a plausible outcome of the OTAG
process. The NOX control program incorporated in
the Post-CAAA scenario may not reflect the NOX
controls that are actually implemented in a regional
ozone transport rule.
Cost Results
Cost results are presented in Table B-5 below.
Based on the Section 812 cost analysis for the electric
power industry, we estimate that the annual national
costs of the CAAA will be roughly $3.1 billion in 2000
and $4.6 billion in 2010.
Table B-5
Electric Power Industry Costs from Post-
CAAA Controls for SOX and NOX
Annual Control Costs
(millions 1990 dollars)
Pollutant
sox
NOX
Total
Post-CAAA
2000
$1,900
$1,200
$3,100
Post-CAAA
2010
$1,600
$2,900
$4,600
B-10
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Non-Road Engines/Vehicles
Nonroad sources are mobile (non-highway)
emission sources. They include the following: lawn
and garden equipment, construction equipment,
agricultural equipment, industrial equipment, aircraft
and airport service vehicles, logging equipment,
recreational vehicles, locomotives, and marine vessels.
We use ERCAM to estimate future emissions from
non-road engines/vehicles. This model incorporates
Federal regulatory programs for controlling NOX,
PM10, VOC, and CO emissions from nonroad engines
and equipment under the Post-CAAA scenario.
Cost Approach
To develop cost estimates for nonroad control
measures, we apply cost-effectiveness values from
several sources (e.g., draft or final rules and the
Section 812 emission projections analysis (Pechan,
1997a)). The analysis includes costs for control inputs
applied to the following nonroad source categories:
small SI (gasoline) engines, CI (diesel) engines,
locomotives, and marine vessels.
We calculate TACs in each implementation year to
calculate the net present value (NPV) of both costs
and benefits over the estimated period of fleet
turnover.10 Because we base the benefits analysis on
projected emission reductions in 2000 and 2010,
rather than the discounted stream of benefits, the
inputs to this cost analysis represent the annualized
cost per ton of reduction, not the NPV cost-
effectiveness. The exception is the input used for the
Federal locomotives rule; because TAG in each
implementation year are not available, we use the
average annualized cost per ton across the entire
implementation period in both 2000 and 2010.
Cost Results
Table B-6 summarizes the cost estimates for each
nonroad engine/vehicle control measure modeled in
this analysis for 2000 and 2010. Total nonroad
engine/vehicle costs, under Post-CAAA scenarios,
are $104 million and nearly $400 million in 2000 and
2010, respectively. Estimated SI engine costs are $56
million under the 2000 Post-CAAA scenario and $104
million under the 2010 Post-CAAA scenario.
Reducing VOC emissions from lawn and garden
equipment contributes to the majority of SI engine
costs. CI engine control costs are $22 million in the
2000 Post-CAAA scenario, and $32 million in 2010.
NOX emission reductions from construction
equipment account for a significant proportion of
total CI engine control costs. Locomotive and
commercial marine vessel benefits are not realized
until after 2000; costs under the 2000 Post-CAAA
scenario are therefore zero.11
10 We were unable to apply the per engine costs of
modifying equipment/vehicles to meet EPA standards because
engine populations were not available for all areas.
11 NOX standards for locomotive and commercial marine
vessels do not take effect until 2000. For the purpose of this
analysis, costs are small enough in 2000 that they are omitted.
B-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-6
Cost Estimates for Nonroad Engine/Vehicle CAAA Programs
Annual Cost (million 1990 dollars)
Engine/Vehicle Category
SI Engines:
Construction Equipment
Industrial Equipment
Lawn and Garden Equipment
Farm Equipment
Commercial Equipment
Logging Equipment
Cl Engines:
Construction Equipment
Industrial Equipment
Farm Equipment
Logging Equipment
Airport Service Equipment
Other
Locomotives
Marine Vessels:
Recreational
Commercial
Total Nonroad Engine/Vehicle Control Costs
Post-CAAA 2000
$1.7
0.5
41
0.2
12
0.9
$12
3.6
2.9
0.1
2.8
0.2
$01
$27
O1
$104
Post-CAAA 2010
$3.1
1.2
74
0.4
23
2.1
$ 17
5.2
4.4
0.2
4.6
0.2
$35
$230
1
$400
Note:
Costs in 2000 are zero because program emissions reductions are not realized until after 2000. See text and
Pechan (1998 and 1997a) for further explanation.
Motor Vehicles
Motor vehicle emissions account for almost thirty
percent of 1990 anthropogenic VOC emissions and
thirty-two percent of NOX emissions. To determine
the costs of controlling VOC and NOX, we first
project motor vehicle emissions with ERCAM-VOC
and ERCAM-NOX (Pechan, 1998). Then we use the
emissions projections to estimate future year motor
vehicle program costs for each of the modeled control
assumptions.12
Cost Approach
We convert all motor vehicle-related control
costs into one of three forms: cost per new vehicle,
cost per registered vehicle, or cost per mile traveled.
We calculate separate costs for each vehicle type (i.e.,
LDGV, light-duty gasoline truck (LDGT) 1, LDGT2).
Motor vehicle calculations required projections of
vehicle miles traveled (VMT), vehicle registrations, or
vehicle sales estimates. We applied the following
equations:
Cost per new vehicle — projected vehicle sales * production
cost ($1 new vehicle)
Cost per registered vehicle = projected vehicle registrations *
12 See the emission projection report for a discussion of the QQ$I pgf vehicle \
emission projection methodology and the control assumptions
(Pechan, 1998).
B-12
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Cost per mile traveled = projected vehicle miles traveled
(VMT) * cost per mile ($/mile)
Sources of Data
The 1990 NPI Inventory provides the 1990
VMT data, which we project in the same manner as
it our emissions (Pechan, 1998).13 National
registrations from the MOBILE4 FCM are the
source of vehicle registration data for 1990 (EPA,
199 Id). The source of motorcycle registrations is
Highway Statistics (FHWA, 1991). National sales data
is based on projected sales compiled by Data
Resources Incorporated (DRI). This information
was also used by EPA in the onboard vapor
recovery RIA (DRI, 1993; EPA, 1993d).14
Cost Categories
The CAAA motor vehicle provisions generate
costs in the following categories: emissions
standards, fuel requirements, emissions inspections,
and low emission vehicle programs. The following
section describes the methodology we use calculate
costs for each category of provisions.15
Emission Standards:
• Tier 1 Certification Standards and
Evaporative Controls. We calculate costs
for tailpipe standards and evaporative
controls with per-vehicle production costs
applied to projected sales.
• Heavy-Duty Vehicle 2g/bhp-hr
Equivalent NOX Standard. We calculate
the cost of complying with the 2004 model
year emission standards by estimating the
baseline package of emission control
technology for meeting 1998 model year
standards (EPA, 1997f). We use the 1994
model year sales of different size classes of
diesel trucks to establish sales fractions,
assumed to represent future sales as well. We
multiply these sales fractions by the year 2009
per vehicle cost increases for light, medium,
and HDVs to compute a sales-weighted per
vehicle cost increase.
• Onboard Vapor Recovery. To estimate the
costs of onboard vapor recovery, we use
expected increases in vehicle price (also
referred to as retail price equivalent) and
average lifetime operating cost (net present
value) (EPA, 1993f).
• Cold Temperature CO Standard. The cost
of the cold temperature CO standard to the
consumer includes the cost to the
manufacturer, the manufacturer's and dealer's
overhead and profits, and the increase or
decrease in maintenance and fuel costs. We
do not include fuel economy improvements
in the analysis. We base cost estimates on
retail price increases of $19 per LDV, $32 per
LDT1, and $48.50 per LDT2 (Pechan, 1998).
• Onboard Diagnostic (OBD) Systems.
With OBD now appearing on all 1996 model
year cars and light-trucks, Federal OBD costs
are approximately $65 to $100 per vehicle.16
Fuels:
13 EPA uses MOBILE4 FCM national projections, scaled to
metropolitan statistical areas (MSAs) according to population
projections, to project VMT and vehicle registrations.
14 EPA assumes that motorcycle sales from Highway Statistics
(FHWA, 1991) increase at the same rate as light-duty gasoline
vehicle (LDGV) sales.
15 For more details on these standards see Appendix A.
• Gasoline Volatility Limits. In order to
calculate the costs of lowering the Reid
vapor pressure (RVP) from 10.5 to 9.0 in
Class C areas, we apply the cost estimate of
16 We apply the per vehicle cost estimate of $65. However,
there is evidence that OBD costs are more likely in the range of
$65 to $100 per vehicle (EPA, 1993d).
B-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
0.225 cents per gallon in the five month
ozone season (Wysor, 1988).
• Federal Reformulated Gasoline. We base
reformulated gasoline costs on an incremental
refiner's cost increase and a monetized fuel
economy disbenefit. An estimate of 3.9 cents
per gallon for Phase I and 5.1 cents per gallon
for Phase II was used (EPA, 1993g). Phase II
reformulated gasoline modifications occur
only in the summer. As a result, we consider
Phase II costs to only five months out of the
year. The Phase I benefits will occur year-
round and are primarily due to the oxygenate
(affecting the aromatic content) and the
reduction of fuel benzene content.
• California Reformulated Gasoline. We use
the estimates from the California Air
Resources Board (GARB) to determine the
increase in per gallon fuel costs to consumers
(GARB 1990; GARB, 1991).
• Oxygenated Fuels. We base oxygenated
fuel costs on an incremental cost of 3.8 cents
per gallon (EPA, 1993g).
• California Reformulated Diesel. We base
reformulated diesel costs on an incremental
per gallon increase of six cents (Green, 1994).
• Diesel Fuel Sulfur Limits. We use an
average value of 2.1 cents per gallon as an
estimate of the incremental cost of reducing
the sulfur content of conventional diesel fuel
(EPA, 1990). The cost estimate do not
include a fuel economy penalty for low sulfur
diesel fuel because we estimate that energy
content is essentially the same as that of
conventional fuel (less than 1% lower).
Emissions Inspection Programs:
• Basic I/M. We use the RIA on enhanced
I/M for deriving this program's basic costs.
Total per vehicle costs include the inspection
fee, average repair cost, and the fuel economy
benefit. The average per vehicle cost is
approximately $5.70. We apply this estimate
to LDGVs, LDGTls, and LDGT2s in areas
where basic I/M is required. Basic I/M costs
are evenly apportioned among VOC, NO^
and CO. No additional costs are attributed
to areas that face I/M program requirements,
but already have a program in place (EPA,
1992c).
• Low Enhanced I/M. Costs for low
enhanced I/M and OTR low enhanced I/M
are not well defined. Therefore, we equate
low enhanced I/M costs equivalent to those
of basic I/M. The average cost per vehicle
for this program is approximately $5.70. This
per vehicle cost applies to all registered
LDGV, LDGT1, and LDGT2 (EPA, 1992c).
• Enhanced I/M. Estimates of enhanced
I/M costs are subject to change as States
make decisions about their program designs.
I/M program costs may be higher or lower
according to each State's selected program
designs such as centralized testing and caps
on the costs of required repairs.17 The
estimated per-vehicle cost is $15.70. We base
this figure on a test fee ($18), an average
repair cost ($14.20 per vehicle), and an
average fuel economy benefit ($16.50 per
vehicle) (EPA, 1992c).18 We estimate annual
costs by applying the per vehicle costs to an
area's projected vehicle registrations.
17To date, only four States have implemented Enhanced I/M
programs. Preliminary cost estimates indicate that opportunity
costs to vehicle owners in the form of travel and wait time do not
play as significant role as originally anticipated. (Harrington and
McConnell, 1999.)
18Test fee and the relationship between test sites and States
vary. In some cases the test fee represents a payment to the state
or local government. In other cases, the fee covers the direct
costs of the testing program. It is not clear, however, how the
test fee could be apportioned between the two possibilities. To
the extent the fee represents a transfer payment, we may be
overestimating the direct cost and social cost of the program.
B-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Low Emission Vehicles:
• California Low Emission Vehicle
Program. We base costs for the California
LEV on the incremental production cost of
vehicles meeting each of the LEV standards
(Pechan, 1998). The overall incremental
production cost for a vehicle type reflects the
projected fraction of sales of each type of
LEV for each projection year.
• National Low Emission Vehicle Program.
We calculate costs for the National LEV
program by multiplying the incremental
production cost of vehicles meeting each of
the LEV standards by the estimated new
vehicle sales volumes (Pechan, 1998).
Additional Programs:
• Clean Fuel Fleet Program (CFFP).
CAAA mandated implementation of CFFP
beginning in 1998 for ozone NAAs
designated serious and above. We estimate
that the model year 1998 fleet demand for
clean-fuel vehicles under the CFFP will be
approximately 47,000 LDVs and 12,000
HDVs (Oge, 1997). However, we do not
include these costs in the analysis.19
vehicle provisions are listed in 1990 dollars by vehicle
type. Phase II RVP and Phase II Federal reformulated
gasoline limits generate costs only in the ozone
season, while oxygenated fuel provisions result in CO
season (winter time) costs. All other fuel programs
listed in Table B-7 generate year round costs.
• Transportation Conformity. Under the
transportation conformity rule, the
Metropolitan Planning Organizations (MPOs)
must perform regional transportation and
emissions modeling and document the
regional air quality impacts of transportation
plans and programs. We expect these
requirements will generate the primary costs
of this rule.
Table B-7 summarizes the motor vehicle unit
costs used in this analysis. Costs of individual motor
19 EPA cannot requke manufacturers to produce CFVs and
areas covered by the CAAA can opt out of the program.
B-15
-------
Table B-7
Motor Vehicle Unit Costs by Provision
Provision
Emission Standards:
Tier 1 Tailpipe Standards: VOC
Tier 1 Tailpipe Standards: NOX
Cold Temperature CO Standard
Evaporative Controls (New Evaporative Emissions
Test Procedure)
On-Board Vapor Recovery System
On-Board Diagnostics
Heavy Duty Engine Standard (2 gram equivalent)
Low Emission Vehicles:
TLEV
LEV
ULEV
ZEV
Fuels:
Phase II RVP Limits
Federal Reformulated Gasoline: Phase I
Phase II
Oxygenated Fuels
Low-Sulfur Diesel Fuel Requirements (0.05% sulfur)
California Phase II Reformulated Gasoline
California Reformulated Diesel
Inspection/Maintenance Programs:
Basic
Low Enhanced
High Enhanced
Notes:
Year
Dollars
1990
1990
1989
1993
1992
1993
1995
1996
1990
1993
1993
1993
1990
1991
1992
1992
1992
1 LDGV = light duty gasoline vehicle; LDGT = light duty gasoline truck;
LDDV = light duty diesel vehicle; LDDT = light duty
diesel truck
Cost Unit
Sales
Sales
Sales
Sales
Sales
Sales
Sales
Cents/gallon
Cents/gallon
Cents/gallon
Cents/gallon
Cents/gallon
Cents/gallon
Registrations
Registrations
Registrations
MC = motorcycle;
LDGV
36.8
115.0
15-23
1.0
4.54
65.0
53.0
95.0
145.0
5,000.0
0.2
3.9
1.2
3.8
12.3
5.7
5.7
15.7
LDGT1
33.7
80.6
15-48
8.0
4.48
65.0
53.0
95.0
145.0
5,000.0
0.2
3.9
1.2
3.8
12.3
5.7
5.7
15.7
Cost Estimate by Vehicle Type:1
LDGT2 MC HDGV HDDV LDDV LDDT
11.7
45.3 16.0 78.0
42-55
8.0 (13.0)
4.48
65.0
140.0
0.2 0.2 0.2
3.9 3.9 3.9
1.2 1.2 1.2
3.8 3.8 3.8
2.1 2.1 2.1
12.3 12.3 12.3
6.0 6.0 6.0
5.7
5.7
15.7
HDGV = heavy duty gasoline vehicle; HDDV = heavy duty diesel vehicle;
B-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Cost Results
Table B-8 summarizes the motor vehicle costs for
2000 and 2010. The total cost post-CAA is $9 billion
in 2000 and $12 billion in 2010.
Table B-8
Cost Estimates of Motor Vehicle Program
Annual Cost (million 1990 dollars)
Program
Title I
California LEV
National LEV
Basic I/M
Low/OTR Enhanced I/M
High Enhanced I/M
Title II
Onboard Vapor Recovery*
Stage II Vapor Recovery*
Phase II RVP
Tailpipe/Extended Useful Life - VOC
Tailpipe/Extended Useful Life - NOX
Evaporative/Running Losses
Onboard Diagnostics
Cold Temperature CO Standard
Federal Reformulated Gasoline
California Reformulated Gasoline
Oxygenated Fuels
2 gram NOX Heavy Duty Standard
Low Sulfur Diesel Fuel
California Reformulated Diesel**
Total Motor Vehicle Control Costs
Post-CAAA 2000
$320
180
57
82
1,100
$63
71
280
504
1,500
42
880
380
720
2,000
160
0
570
170
$ 9,070
Post-CAAA 2010
$ 1,100
1,060
69
99
1,400
$69
86
340
550
1,700
46
960
410
860
2,400
204
69
740
230
$ 12,300
Notes:
* The benefits of onboard vapor recovery and stage II vapor recovery are accounted for under area sources. The cost
for onboard vapor recovery systems is estimated assuming phase-in for light duty gasoline vehicles and light duty
trucks. Heavy duty trucks are not affected.
** The analysis does not account for the benefits (emission reductions).
B-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Area Sources
Area sources comprise small stationary sources
not listed in the point source database (e.g., dry
cleaners, graphic arts, industrial fuel combustion,
gasoline marketing) and solvent use (e.g., consumer
solvents, architectural coatings). Area sources of
NOX emissions include industrial fuel combustion
units in the industrial, commercial/institutional, and
residential sectors. The following are VOC sources:
pharmaceutical manufacturing, wood furniture surface
coating, aerospace manufacturing (surface coating),
ship building and ship repair (surface coating),
halogenated solvent cleaning, dry cleaning -
perchloroethylene (PCE), and petroleum refinery
fugitives. Area sources of PM10 are paved roads,
unpaved roads, construction, cattle feedlots,
agricultural tilling, and agricultural burning.
Cost Results
Table B-9 summarizes area source control
measure costs for NOX and PM under the 2000 and
2010 Post-CAAA scenarios. The costs associated with
applying NOX point source fuel combustion controls
to smaller sources are approximately $16 million in
2000 and $18 million in 2010. Control measures
applied to reduce PM emissions from area sources are
estimated to cost $1.9 billion under the 2000 Post-
CAAA scenario and $2.2 billion under the 2010 Post-
CAAA scenarios. Controlling fugitive dust emissions
from construction activity generates the majority of
PM control costs.
Cost Approach
To assess the costs of reducing emissions from
area sources, we use annualized costs per ton
reduced.20 We estimate total annual costs under each
of the Post-CAAA scenarios by applying annualized
costs per ton from a variety of regulatory documents
to corresponding emission reductions. The annual
cost formula is:
Annual Cost = AnnuaR^ed Cost Per Ton * Emission
Reduction.21
ERCAM-NOX and ERCAM-VOC incorporate
separate cost equations for each area source category.
20 Source-specific data are not available for area and nonroad
sources.
21 In the present analysis, we annualize total capital costs with
a five percent discount rate. In some cases, we re-calculate the
annualized cost per ton reported in the source material if that
estimate was based on a discount rate other than five percent.
B-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-9
Cost Summary of Area Source NOX and PM Controls
Annual Costs (million 1990 dollars)
Source Category
Post-CAAA 2000
Post-CAAA2010
Area Source NOX Controls
Industrial Fuel Combustion - Coal
Industrial Fuel Combustion - Oil
Industrial Fuel Combustion - Natural Gas
Total Area Source NOX
Area Source PM Controls
Agricultural Burning
Agricultural Tilling
Beef Cattle Feedlots
Construction
Paved Roads
Unpaved Roads
Total Area Source PM
$6.6
0.8
8.5
$16
$37
4.1
1.4
1,500
350
1.1
$1,900
$7.5
0.8
9.9
$18
$39
3.6
1.7
1,800
440
0.8
$ 2,200
Reasonable Further Progress
Requirements
Title I of the CAAA includes provisions that
require ozone nonattainment areas to make steady
progress toward compliance with NAAQS. NAAs
classified as moderate, serious, severe, or extreme
must demonstrate that they are working to lower
ambient ozone concentrations at a reasonable rate of
progress (ROP) and, by 1996, reduce annual VOC
emission by fifteen percent from 1990 levels. In
addition to satisfying ROP requirements, areas
classified as having an ozone nonattainment problem
that is serious or worse must continue to cut
emissions and make reasonable further progress (RFP)
toward attainment. To meet RFP standards, after
1996, NAAs have to reduce precursor emissions by
three percent per year until they each reach their
respective compliance deadlines. While ROP
requirements mandate VOC cuts to comply with RFP
standards it is often possible to substitute NOX for
VOC. (Refer to Appendix A for more discussion of
ROP/RFP requirements.)
Title I progress requirements establish minimum
emissions reduction standards for ozone NAAs. In
many cases, the areas subject to ROP/RFP
regulations satisfy these requirements simply by
complying with other existing emissions standards.
VOC and NOX reductions made to meet other
regulations are credited towards ROP/RFP
requirements. For the purposes of the prospective
analysis, we assume that where possible, credit is given
for all available NOX cuts and that any remaining
emission reduction needed to satisfy Title I progress
requirements come from VOC. In the majority of
cases, credited VOC cuts account for this remaining
reduction. For NAAs that are not able to fulfill the
remainder of their ROP/RFP obligations with
credited VOC emissions reductions, there is a
shortfall. This shortfall represents the quantity of
VOC that ozone NAAs must reduce through control
efforts beyond those mandated by other clean air
provisions.
Tables B-10 and B-ll show, for the years 2000
and 2010 respectively, which NAAs are assumed to
B-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
satisfy, and not satisfy, their Title I progress
requirement. Failure to meet the requirement is
indicated by a shortfall. The shortfall is measured by
the amount ozone season daily (OSD) level exceeds
the maximum allowable daily VOC emission. The
OSD level of VOC emission represents the predicted
daily emission in the absence of RFP/ROP
requirements, and VOC target presents the maximum
allowable daily VOC emissions.
B-20
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-10
2000 Rate of Progress Analysis
Ozone Nonattainment Area
VOC OSD1 VOC Target2 Shortfall
Atlantic City
Baltimore
Baton Rouge
Beaumont-Port Arthur
Chicago-Gary-Lake County
Cincinnati-Hamilton
Cleveland-Akron-Lorain
Dallas-Fort Worth
El Paso
Grand Rapids
Houston-Galveston-Brazoria
Lewiston-Auburn ME
Los Angeles-South Coast
Louisville
Milwaukee-Racine
Muskegon
Nashville
NewYork-N New Jersey-Long Is
Philadelphia-Wilmington-Trenton
Phoenix
Pittsburgh-Beaver Valley
Portland ME
Portsmouth-Dover-Rochester
Providence
Reading PA
Richmond-Petersburg
Sacramento Metro
St. Louis
Monterey Bay
Salt Lake City
San Diego
Santa Barbara-Santa Maria-Lomp
Sheyboygan
Washington DC
Knox & Lincoln Cos ME
Kewaunee Co Wi
Manitowoc Co WI
San Joaquin Valley
Ventura Co CA
Southeast Desert Modified
Boston-Lawrence-Worcester-E. MA
Springfield/Pittsfield-W. MA
Greater Connecticut
37.97
318.33
203.77
340.66
1,240.89
305.34
521.14
694.53
85.38
182.72
1,426.65
34.08
972.91
219.66
327.09
46.67
231.71
1,994.96
1,090.49
377.43
407.05
70.05
53.54
173.78
60.53
179.97
158.01
465.64
64.16
182.75
192.90
82.75
24.49
402.76
9.95
4.86
19.72
470.50
65.69
227.71
822.65
155.51
316.27
45.71
376.47
415.65
450.66
1,202.25
341.18
573.07
673.97
69.02
175.66
2,268.31
35.98
939.08
215.97
293.24
44.32
205.60
2,407.97
1,376.55
347.91
399.80
73.33
58.70
180.51
61.14
201.70
155.08
549.14
79.63
150.80
189.71
83.10
22.44
477.03
10.37
4.56
17.20
532.41
70.52
219.32
918.01
152.42
370.11
0.00
0.00
0.00
0.00
38.64
0.00
0.00
20.56
16.36
7.06
0.00
0.00
33.83
3.69
33.85
2.35
26.11
0.00
0.00
29.52
7.25
0.00
0.00
0.00
0.00
0.00
2.93
0.00
0.00
31.95
3.19
0.00
2.05
0.00
0.00
0.30
2.52
0.00
0.00
8.39
0.00
3.09
0.00
Notes:
The VOC OSD (ozone season daily) values are the estimated daily emissions in the absence of
ROP/RFP requirements. These estimates do, however, incorporate the effect of the VOC
reductions that are credited towards Title I progress requirements.
The VOC target represents the maximum allowable daily VOC emission for NAAs to comply with
ROP/RFP requirements. The VOC target is calculated based upon the assumption that all available
NOx cuts are credited towards ROP/RFP requirements and that all necessary remaining reductions
come from VOC.
B-21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-11
2010 Rate of Progress Analysis
Ozone Nonattainment Area
VOC OSD1
VOC
Target2 Shortfall
Atlanta
Atlantic City
Baltimore
Baton Rouge
Beaumont-Port Arthur
Chicago-Gary-Lake County
Cincinnati-Hamilton
Cleveland-Akron-Lorain
Dallas-Fort Worth
El Paso
Grand Rapids
Houston-Galveston-Brazoria
Lewiston-Auburn ME
Los Angeles-South Coast
Louisville
Milwaukee-Racine
Muskegon
Nashville
NewYork-N New Jersey-Long Is
Philadelphia-Wilmington-Trenton
Phoenix
Pittsburgh-Beaver Valley
Portland ME
Portsmouth-Dover-Rochester
Providence
Reading PA
Richmond-Petersburg
Sacramento Metro
St. Louis
Monterey Bay
Salt Lake City
San Diego
Santa Barbara-Santa Maria-Lomp
Sheyboygan
Washington DC
Knox & Lincoln Cos ME
Kewaunee Co Wi
Manitowoc Co WI
San Joaquin Valley
Ventura Co CA
Southeast Desert Modified
Boston-Lawrence-Worcester-E. MA
Springfield/Pittsfield-W. MA
Greater Connecticut
492.40
33.21
293.56
206.65
377.63
1,236.73
283.74
485.90
687.15
84.33
183.11
1,530.07
32.03
847.66
216.86
321.89
46.86
230.00
1,842.53
1,070.05
347.52
358.67
66.88
52.49
166.61
55.33
179.35
135.99
439.47
61.76
189.83
174.04
81.53
24.69
355.35
8.98
4.77
19.37
448.37
62.33
213.87
775.66
147.45
292.53
541.99
45.71
376.47
415.65
450.66
840.15
341.18
573.07
673.97
69.02
175.66
1,606.75
35.98
670.95
215.97
204.72
44.32
205.60
2,407.97
1,194.41
347.91
399.80
73.33
58.70
193.15
61.14
201.70
120.24
549.14
79.63
150.80
202.24
83.10
22.44
477.03
10.37
4.56
17.20
566.96
63.11
172.34
918.01
166.51
370.11
0.00
0.00
0.00
0.00
0.00
396.58
0.00
0.00
13.18
15.31
7.45
0.00
0.00
176.71
0.89
117.17
2.54
24.40
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
15.75
0.00
0.00
39.03
0.00
0.00
2.25
0.00
0.00
0.21
2.17
0.00
0.00
41.53
0.00
0.00
0.00
Notes:
The VOC OSD (ozone season daily) values are the estimated daily emissions in the absence of
ROP/RFP requirements. These estimates do, however, incorporate the effect of the VOC reductions
that are credited towards Title I progress requirements.
The VOC target represents the maximum allowable daily VOC emission for NAAs to comply with
ROP/RFP requirements. The VOC target is calculated based upon the assumption that all available
NOx cuts are credited towards ROP/RFP requirements and that all necessary remaining reductions
come from VOC.
B-22
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Cost Results
We base the RFP cost estimate on the assumption
that ozone nonattainment areas (NAAs) will take
credit for NOX reductions for meeting progress
requirements. Additional area-specific analysis would
be necessary to determine the extent to which areas
find NOX reductions beneficial in meeting attainment
and progress requirement targets. Trading of NOX for
VOC to meet RFP requirements may result in
distributions of VOC and NOX emission reductions
that differ from those used in this analysis. In part as
a response to these uncertainties, we adopt a
conservative strategy for estimating the costs of RFP
reductions, using the relatively high cost per ton
reduced value of $10,000 for all required reductions.
We calculate these annual figures by multiplying the
aggregate daily shortfall by 365, and then multiplying
this number by the estimated cost of each ton of
reduction, $10,000. Based on this calculation, the
annual estimated cost of Title I progress requirements
is $1,150 million in Post-CAAA 2000 and $2,460
million in Post-CAAA 2010.
Since the time we conducted our initial cost
analysis, control measures for several nonattainment
areas (NAA) have been identified that suggest controls
may be much less. For example, the dollar per ton
estimate associated with control measures selected in
Chicago is $3,500. We incorporate this information in
our sensitivity analysis. In the sensitivity test analysis,
we calculate overall costs by applying the cost per ton
of reduction associated with each identified control.
Where the required reduction cannot be achieved
through implementation of all of the identified
controls, we assume unidentified controls will be used
to eliminate the remaining shortfall. We apply the
$10,000 per ton reduced estimate for these
unidentified controls (see Appendix B for more
details). Results of the sensitivity analysis suggest that
our conservative approach of applying $10,000 per
ton reduced to all VOC shortfalls may overstate cost
by as much as several billion dollars in 2010.
Costs by Title
Examining CAAA costs by title, in addition to
reviewing them by source, is useful for understanding
the cost components. Table B-12 summarizes the cost
estimates generated in this analysis by year and Title.
As shown in the table, the cost estimate under the
Post-CAAA 2000 scenario is $19 billion, increasing to
nearly $27 billion under the Post-CAAA 2010
scenario. All costs are in 1990 dollars.
B-23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableB-12
Summary of Cost Estimates by CAAA Title
Annual Cost (million 1990 dollars)
Title
Title I -
Title II
Title III
Title IV
Title V
TOTAL
Sector/Pollutant
Provisions for Attainment and
Non-utility Point/VOC
Non-utility Point/NOx
Utility/SO2 and NOX
Area/VOC
Area/NOx
Area/PM
Motor Vehicle
Progress Requirements
- Provisions Relating to Mobile
Motor Vehicle
Nonroad
- Hazardous Air Pollutants
Point/VOC
Area/VOC
Non-VOC MACT1
- Acid Deposition Control
Utility/SO2 and NOX
- Permits
Post-CAAA 2000
Maintenance of NAAQS
$420
1,700
790
920
16
1,900
1,800
1,200
Sources
$7,300
104
$480
130
170
$2,300
$3002
$ 19,400
Post-CAAA 2010
$440
2,200
2,500
1,040
18
2,200
3,700
2,500
$ 8,700
400
$520
150
170
$ 2,040
$3002
$ 26,800
Notes:
1 Costs reflect estimate of annualized costs from final rules. We do not use ERCAM-VOC to model source categories,
because the National Emission Standards for Hazardous Air Pollutants (NESHAPs) are associated with non-VOC HAP
emission reductions. Consequently, they are not included in the Post-CAAA 2000 and 2010 inventories.
2 Includes costs only for State-implemented permitting programs, excluding the costs of Federally-implemented programs,
since all Title V permit programs will be State-run in 2005.
Joint Rules
Assigning costs to a CAAA Title is difficult in the
case of "joint rules" issued under more than one Title.
For example, the marine vessel rule incorporates
controls to reduce VOC emissions through reasonably
available control technology (RACT) standards and
hazardous air pollutant (HAP) emissions through
maximum achievable control technology (MACT)
standards. In general, we assign the costs for joint
rules to the CAAA title based on the implementation
dates and the year by which emission reductions are
expected to occur. In some cases, we assign joint
rules costs to the more stringent rule (in terms of
sources covered or reduction required). Examples of
source categories with overlapping Title I and Title III
control measures include the following: aerospace,
surface coating, petroleum refineries, shipbuilding,
synthetic organic chemical manufacturing industry
(SOCMI) categories, printing, and wood furniture. In
cost accounting, we generally allocate the costs for
these source categories to Title III, rather than Title I.
Title I
Title I, Provisions for Attainment and
Maintenance of National Ambient Air Quality
Standards (NAAQS), includes national VOC rules and
any controls that NAAs will likely apply to meet
Federal standards for ozone and PM. For Title I,
B-24
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
determining which rules are CAAA-related and which
are associated with other legislation is sometimes
difficult. For example, EPA actually promulgated the
hazardous waste transport, storage, and disposal
facilities (TSDF) rule under the authority of the
Resource Conservation and Recovery Act (RCRA).
We attribute, however, the costs and associated VOC
emission reductions of the Phase I and Phase II
RCRA rule to Title I, because the rule is consistent
with CAAA programs that promote attaining and
maintaining the NAAQS.
The costs associated with Title I consist of point
and area source costs for VOC, NOX, and PM control
measures. Total Title I costs are $8.6 billion in 2000
and $14.5 billion in 2010.22 The following two tables
summarize the cost analysis results for provisions
promulgated under Title I of the CAAA. Table B-13
presents costs specifically associated with Title I
national point and area VOC rules. Table B-14 costs
by national Title I provisions regulating sectors
ranging from motor vehicles to utilities. To preserve
consistency with the assumptions made in the
emission projections analysis, we simulate attainment
of the ozone and PM NAAQS as they were prior to
the 1997 revisions.
22 Note that provisions included in other CAAA Tides, as well
as die decisions diat individual States make about how best to meet
progress requirements and attainment targets, affect Tide I costs.
B-25
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableB-13
Title I National Rules, Point and Area Source VOC Control Costs
Annual Costs by Sector (million 1990 dollars)
Post-CAAA 2000
Post-CAAA2010
Title I National Rules
Point
Area
Total
Point
Area
Total
Consumer Products
AIM Coatings
Automobile Refinish Coatings
Hazardous Waste TSDFs
Municipal Landfills
Marine Vessel Loading
TOTAL
$0.0
0.0
0.0
<0.1
0.0
24
$24
$81
24
6
300
160
0
$570
$81
24
6
300
160
24
$590
$0.0
0.0
0.0
<0.1
0.0
28
$28
$88
27
7
350
170
0
$650
$88
27
7
350
170
28
$680
Notes:
because the NESHAPs are associated with non-VOC HAP emission reductions.
The Off-site Waste Treatment NESHAP was not modeled in this analysis. We assume that the Title I modeling of the
RCRA Phase I and Phase 2 rules for hazardous waste TSDFs will capture any future MACT reductions and costs.
EPA estimated that the Medical Waste Incineration guideline would cost between $59 million per year to $120 million per
year, depending on the extent to which affected facilities switch to less expensive methods of treatment and disposal. The
cost above represents the midpoint of this range.
TableB-14
Summary of Costs for Title I
Annual Cost (million 1990 dollars)
Provision
Post-CAAA 2000
Post-CAAA 2010
Area Specific:
California LEV
National LEV
Basic I/M
Low/OTR Enhanced I/M
High Enhanced I/M
RACT:
VOC RACT
Non-utility NOX RACT
Utility NOV RACT/Best Available Control
$320
180
57
82
1,100
620
37
140
$ 1,100
1,060
69
99
1,400
660
40
530
Technology (BACT)
New CTG
OTR:
Utility Cap-and-Trade Program
NOX Stationary (Non-utility)
PM NAA Controls
Progress Requirements
TOTAL
130
640
1,600
1,900
$ 1,200
$ 8,050
150
1,200
2,100
2,200
$2,500
$ 13,900
B-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Title II
Title II provisions include Federal motor vehicle
and nonroad engine/vehicle rules, in addition to
regulations requiring fuel reformulations. Table B-15
summarizes the results of our cost analysis for
provisions promulgated under Title II of the CAAA.
Table B-15
Summary of Title II Motor Vehicle and Nonroad
Engine/Vehicle Program Costs
Annual Cost (million 1990 dollars)
Provision Post-CAAA 2000 Post-CAAA2010
Motor Vehicles/Fuels:
Motor Vehicle Emission Standards:
Tailpipe/Extended Useful Life - VOC
Tailpipe/Extended Useful Life - NOX
2 gram NOX Heavy Duty Standard1
Onboard Vapor Recovery
Cold Temperature CO Standard
Onboard Diagnostics
Evaporative/Running Losses
Fuels:
Phase II RVP
Federal Reformulated Gasoline
California Reformulated Gasoline
Oxygenated Fuels
California Reformulated Diesel
Low Sulfur Diesel Fuel
Stage II Vapor Recovery
Motor Vehicle Total
Nonroad Engines/Vehicles:
Phase I Cl engine standards
Phase I and II SI engine standards
Federal locomotive standards1
Federal commercial marine vessel standards1
Federal recreational marine vessel standards
Nonroad Engine Vehicle Total
Total Title II Costs
$504
1,500
0
63
370
880
42
280
720
2,000
170
170
570
71
$ 7,300
$22
56
0
0
27
$104
$ 7,400
$550
1,700
69
69
410
960
46
330
860
2,400
204
230
740
86
$ 8,700
$32
104
35
1
230
$400
$9,100
Notes:
Columns may not sum to totals due to rounding.
1 Costs under the 2000 Post-CAAA scenario are zero because emission reductions are not realized until after 2000.
B-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableB-16
Title III, MACT Standards, Point and Area Source VOC Control Costs
Annual Costs by Sector
(in million 1990 dollars)
Source Category
Post-CAAA 2000
Post-CAAA2010
Benzene NESHAP
2-Year MACT:
Dry Cleaning-Perchloroethylene
SOCMI HON
4-Year MACT:
Aerospace Industry (surface coating)
Chromium Electroplating1
Coke Ovens
Commercial Sterilizers1
Gasoline Distribution-Stage I
Halogenated Solvent Degreasing
Industrial Process Cooling Towers1
Magnetic Tape1
Marine Vessels
Medical Waste Incineration1'3
Municipal Waste Combustors1
Off-Site Waste Treatment2
Petroleum Refineries-Other
Sources Not Distinctly Listed
Printing/Publishing
Polymers & Resins Group I
Polymers & Resins Group II
Polymers & Resins Group IV
Secondary Lead Smelters1
Shipbuilding and Ship Repair
Wood Furniture (surface coating)
TOTAL
Notes:
' Onoto rafla/H1 eietimoteie r\f anni lohveirJ /^r*ctc frr»m final nil
$0.2
28
26
4
17
21
7
12
(37)
14
0.8
17
89
43
-
160
200
110
4.3
5.3
2.0
8.5
49
$780
QC _Qr»i irr*a r^atannriao or a nrvr mnrlal
$0.2
31
29
5.3
17
21
7
13
(42)
14
0.8
20
89
43
-
180
207
128
5.0
6.7
2.0
11
50
$840
orl in CG>rAM_\/nr ha,~=ii=a
the NESHAPs are associated with non-VOC HAP emission reductions.
The Off-site Waste Treatment NESHAP was not modeled in this analysis. We assume that the Title I modeling of the
RCRA Phase I and Phase 2 rules for hazardous waste TSDFs will capture any future MACT reductions and costs.
EPA estimated that the Medical Waste Incineration guideline would cost between $59 million per year to $120 million per
year, depending on the extent to which affected facilities switch to less expensive methods of treatment and disposal. The
cost above represents the midpoint of this range.
B-28
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Title III
Tide III of the CAAA requires the promulgation
of MACT regulations to control HAP emissions from
specific source categories. Total Title III costs
represent the TACs for individual two- and four-year
MACT standards. Not all Title III regulations are
modeled by ERCAM-VOC because Title III
regulations target HAP emissions which are not
included in the Section 812 base year inventory. To
provide a complete cost accounting, we use the annual
cost estimates from the final rules for MACT
standards that are expected to reduce non-VOC HAP
emissions. The cost for these MACT categories is
$173 million. The cost estimates for Title III are
summarized in Table B-16. Costs do not differ
significantly under the Post-CAAA 2000 and the Post-
CAAA 2010 scenarios because we do not derive the
costs from ERCAM-VOC using future year emission
estimates.23
Title IV
Title IV of the CAAA is the Acid Deposition
Control Program. Title IV controls include SO2 and
NOX controls at electric utilities and are summarized
in Table B-17, below. The model for SO2 controls
incorporates EPA's program for SO2 allowance
trading. The annual national costs of Title IV of the
CAAA are $2.3 billion in 2000 and $2.0 billion in
2010. The decline in annual costs from 2000 to 2010
results primarily from the increase in use of Western
coal, the cost of which we project will decrease over
time. Cost reductions also occur following the
increased use of gas-fired combined-cycle units to
generate electricity without SOX emissions. As
employed technology becomes more efficient, we
expect that generation costs will decrease.
Table B-17
Annual Costs of Title IV
Annual Costs (million 1990 dollars)
Post-CAAA 2000 Post-CAAA 2010
Title
Title IV
$2,300
$2,000
Note: These estimates reflect the base case used in the
Clean Air Power Initiative. See EPA, "Analyzing Electric
Power Generation Under the CAA" (1996) for detail on
scenario development. For more information on its
application to the cost analysis, see EPA, 1997a.
Title V
23We do not estimate the costs associated with seven- and ten-
year standards due to the lack of adequate data regarding the
implementation of these standards.
Title V of the CAAA establishes requirements for
a new operating permits program. Using costs from
final regulations rather than models, we estimate that
Title V will cost $300 million. Consequently, we base
this on the estimated cost of State-developed
programs, excluding Federally-implemented State
programs. The States are expected to implement all
Title V permit programs by 2005. We estimate each
source's permit fees and administrative costs in the
first five-year implementation period, including the
explicit cost to the permitted sources (industry), State
and local permitting agencies, and EPA. The $300
million cost estimate may be an overestimate, since
many States already have operating permit systems
with fee provisions in place, and we do not
incorporate existing state programs into the baseline
in the RIA documents (EPA, 1992a and EPA, 1995).
Social Costs
In an ideal setting, a cost-benefit analysis would
not only identify but also quantify and monetize an
exhaustive list of associated social costs due to a
regulatory option. This would include assessing how
regulatory action targeting a specific industry or set of
facilities can alter the level of production and
consumption in the directly affected market and
related markets. For example, regulation of emissions
from the electric utility industry that results in higher
electricity rates would have both supply-side and
demand-side responses. In secondary markets, the
increased electricity rates affect production costs for
various industries and initiate behavioral changes (e.g.,
using alternative fuels as a substitute to electric
B-29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
power). With each affected market, there are also
associated externalities that should be included in
estimating social costs. Returning to the utilities
example, the externalities associated with electric
power generation versus nuclear power generation can
be very different. The mix of externalities could
change as consumers substitute nuclear power for
electric power. It is frequently difficult to accurately
characterize one or all of these dimensions of market
responses and estimate the resulting social costs.
There are three generally practiced approaches to
estimating regulatory costs: (i) direct compliance cost,
(ii) partial equilibrium modeling, and (iii) general
equilibrium modeling. The direct compliance cost
approach is the most straightforward of the three.
Direct compliance cost estimates are calculated
differently than economic welfare impact estimates
that result from partial or general equilibrium. This
technique develops ex ante estimates of increased
production costs, and may in some cases (such as in
the case of the IPM model for utilities in our analysis)
measure supply-side response to a regulatory action by
modeling changes in supply price and quantity. In
general, the technique does not account for how
demand and consumption levels may change in
response to higher production costs and prices.
Instead, this approach measures how an industry's or
firm's marginal cost curve shifts due to the additional
production costs associated with pollution abatement
controls.
The direct compliance approach, however, is a
reasonable estimate of incremental expenditures. In
certain instances, this method may be a conservative
approximation of primary social costs because it
overstates direct costs by not reflecting efficiency
enhancing demand-side responses. There are two
major difference between direct and social costs that
influence the results. First, direct cost methods may
overstate the actual compliance costs that are
associated with demand and consumption level
changes in response to higher production costs. By
not accounting for market responses, total direct costs
reflect the incremental costs per unit of output
multiplied by the higher, pre-regulation quantity
produced. Second, a direct cost approach assumes
firms incur the full costs of pollution abatement
activities. The marginal cost curves of firms, however,
do not necessarily increase by the full amount of the
pollution abatement technology. For example, firms
can adopt cost-saving activities that help to offset the
new costs.24 (Morgenstern et al., 1998).
Capturing consumer and producer behavioral
responses to regulatory action requires either partial or
general equilibrium modeling. These more
complicated approaches estimate social costs by
accounting for a wider range of consequences
associated with altered resource allocation due to the
use of pollution abatement equipment. A partial
equilibrium analysis requires modeling both supply
and demand functions in affected markets. Therefore,
measures of social cost reflect behavioral responses by
both producers and consumers in one or more
markets. The variation between results from a direct
cost approach and a partial equilibrium approach will
generally depend on the extent price and quantity
demanded change. Moreover, the estimates of a
partial equilibrium model can overstate or understate
total, economy-wide social costs depending on the
type of existing market distortions and the extent to
which there are spillover effects from the targeted
sector to other economic sectors.
The partial equilibrium approach is particularly
appropriate when social costs are predominantly
incurred in a limited set of directly affected markets,
and has minimal effects on other sectors. In cases
where the regulatory action is known to have an
impact on a broad range of sectors in the U.S.
economy, the general equilibrium model can be a
more appropriate approach to estimating social costs.
Like the partial equilibrium model, the general
equilibrium model estimates social costs by accounting
for direct compliance costs and producer and
consumer dynamics. The general equilibrium model
can capture large and small first-order effects that
occur in multiple sectors of the economy.
It is difficult to determine the relationship
between general equilibrium estimates and direct cost
estimates. Relative to the general equilibrium
estimates, direct cost estimates are likely to overstate
costs in the primary markets because they do not
reflect consumer and producer responses. General
equilibrium estimates, however, have a broader basis
from which to estimate social costs and reflect net
social welfare changes across the economy's economic
sectors. There still remain significant barriers to
24Morgenstern et al. (1998) estimates the multiplier of
abatement expenditures to total costs can be as low as 0.8.
B-30
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
assessing the potential magnitude and direction of
actual total welfare changes, as experienced by the
economy as a whole, to those estimated by a general
equilibrium model.25 Without insight into the
accuracy of general equilibrium model estimates, it is
difficult to characterize how direct cost estimates
relate to general equilibrium cost estimates.
In the 812 retrospective analysis (EPA, 1997), we
opted for the general equilibrium approach,
recognizing that the Clean Air Act has a pervasive
impact on the U.S. economy. Moreover, the
retrospective nature of the analysis provided us with
fairly well-developed historical data sets of goods and
service flows throughout the economy. These data
sets facilitated the construction of extensive
expenditure data from which we modeled producer
and consumer behavior and estimated net social costs.
In the retrospective, our central estimate of total
annualized costs, from 1970 to 1990, was $523 billion.
In comparison, the aggregate welfare effects were
estimated between $493 and $621 billion.26
Although a general equilibrium approach
represents a theoretically preferable, and potentially
more accurate, method for measuring social costs, as
described in Chapter 3 we adopted a direct compliance
cost approach for the prospective analysis. We
selected the simpler direct cost modeling method for
three reasons:
• First, we believe that the direct cost approach
provides a good first approximation of the
economic impact of the CAAA on the U.S.
economy. For example, recent research
suggests that ex ante analyses of regulatory
costs are far more likely to overstate than
understate costs.27 In addition, the direct cost
approach, because it does not reflect
adjustments to prices and quantities that
might be adopted to mitigate the effects of
regulation, likely overstates the producer
surplus loss to the entity that incurs the
pollution control cost expenditure. Under
these conditions, direct cost may actually
overestimate the social costs of a particular
economic sector. It is also possible that the
direct cost estimates understate the effects of
long-term changes in productivity and the
ripple effects of regulation on other
economic sectors.
• Second, we believe that the precision in
estimating social costs that might be gained
through a general equilibrium approach could
be compromised by the difficulty and
uncertainty associated with projecting future
economic and technological change. The
general equilibrium approach could provide
many insights that the direct cost approach
cannot, but as a tool for estimating social
costs it is very data-intensive and introduces
a significant level of additional uncertainty as
a result.
• Third, undertaking a general equilibrium
modeling exercise remains a very resource-
intensive task. In light of our concerns about
the potential gains in precision or accuracy of
our social costs estimates for the purposes of
comparing costs to benefits, we concluded
that more detailed modeling would not be the
most cost-effective use of the project
resources.
25Harriiigton et al. (1999): "The general equilibrium effects of
environmental regulations are likely to be important, but are
likewise impossible to examine empirically. Computable general
equilibrium models have not been tested against real-world
outcomes and may be untestable."
26 Estimates are presented in 1990 dollars. The retrospective
states, "In general the estimated second order macroeconomic
effects were small relative to the size of the U.S. economy." The
rate of long term GNP growth between the control and no-control
scenarios amounted to roughly one-twentieth of one percent less
growth. It is important to note that although the difference is
small, the direct compliance cost method does represent an
underestimation.
Limitations and Uncertainties
Several factors contribute uncertainty to the cost
estimates.
• Emissions Projections. We base total cost
estimates for individual CAAA provisions on
projected emission reductions in 2000 and
2010. As a result, the quality of the cost
27 See, for example, Harrington et al (1999) for a comparative
analysis of ex ante and ex post regulatory cost estimates.
B-31
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
analysis results is dependent, in part, on the
quality of the emission projection estimates.
• Evolving Rules. We estimate many of the
costs based on assumptions regarding how
stringent evolving or draft rules may be when
finalized. Costs are likely to change as these
rules are amended or finalized.
• Facility response to regulation. Facilities
may respond to regulations in a manner
different from our model assumptions and
thereby affect cost estimates. The cost
estimates for individual CAAA provisions will
ultimately depend on the mix of compliance
options facilities choose to meet each rule's
requirements. In addition, we do not
quantify the effect of economic incentive
provisions, which provide greater flexibility to
facilities affected by the rules.28
• SIPs for meeting ozone NAAQS. It is
difficult for us to predict how States will
design control plans for meeting ozone
NAAQS attainment requirements.
• Technology assumptions. We develop
costs based on data for today's technologies.
To the extent that control technologies
improve over time and lower cost control
alternatives become available, we may
overstate costs.
• Discount rate. Discount rates affect costs.
In some instances of this cost analysis, we use
data sources that do not explicitly list the
applied discount rate assumptions.
In this section, we first discuss the impact of the
above listed key limitations. We then identify cost
inputs and conduct quantitative uncertainty analyses
for those factors. Table B-18 summarizes the
limitations and the likely effect on the cost analysis
results.
28 For example, the cost savings associated with a cap-and-
trade program, such as South Coast Air Quality Management's
Regional Clean Air Incentive Market (RECLAIM), is not reflected
in the prospective cost assessment.
B-32
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableB-18
Potential Effects of Uncertainty on Cost Estimates
Description
Potential Effect on Cost
Estimates
Innovations in future emission control technology
Emission projections:
Growth factors/activity indicators
RACT controls for individual States
Inclusion of economic incentive provisions1
Use of costs for rules that are currently in draft form (not yet finalized)
Uncertainty of final State strategies for meeting Reasonable Further
Progress (RFP) RFP requirements
Inclusion of 7- and 10-year MACT standards
Revisions to Title V cost estimate to reflect current State permit
program costs
Decrease
Unknown
Unknown
Decrease
Unknown
Unknown
Increase
Decrease
Note:
1 Examples include banking, trading, and emissions averaging provisions.
Emission Projections
The selection of activity indicators for individual
source categories can have significant impacts on
projected emissions, and in turn, on the cost estimates
in this analysis. In addition, we select RACT controls
based on representative controls, yet the controls for
individual States/facilities may differ from these
representative controls.
Draft Rules
EPA is currently revising several promulgated
rules in response to public comments, legal actions, or
other factors. The cost data used in this analysis
reflect the latest available estimates, yet these costs are
subject to change as the Agency modifies existing
rules. For example, while we were developing CAAA
costs, the Agency was also proposing a rule to limit
summer season NOX emissions in a group of OTAG-
participating States. Cost estimates for the regional
OTAG NOX strategy will most likely be different than
those for the Ozone Transport Rulemaking due to
uncertainty about the final form of the rulemaking.
In an effort to maintain consistency between
emission and cost data, we do not update its costs to
reflect modification to drafted and existing rules. For
example, because EPA revised the ozone and PM
NAAQS after projecting emissions, we continued to
use earlier cost estimates that are consistent with the
prior NAAQS control assumptions. Another example
is that of estimating cost for the proposed
compression ignition (CI) engine Phase II rule which
was not proposed by the time we completed our
emissions projection for the nonroad sector.
Although costs are now available for the Phase II rule,
to maintain consistency with the benefits analysis we
include only Phase I costs.29
In general, rule amendments such as exemptions
for particular types of sources or opportunities for
sources to postpone compliance dates are likely to
ease the regulatory burden on regulated sources, and
therefore, will result in lower total costs than those
estimated in this analysis.
Economic Incentive Provisions
EPA created economic incentive provisions in
several rules to provide flexibility for affected facilities
that comply with the rules. These provisions include
banking, trading, and emissions-averaging provisions.
Flexible compliance provisions lower the cost of
compliance. For example, the emissions-averaging
program grants flexibility to facilities affected by the
marine vessels rule, the petroleum refinery NESHAP,
and the gasoline distribution NESHAP; these
facilities can choose which sources to control, as long
as they achieve the required overall emissions
reduction. In many of the cost analyses, we do not
29 However, this implies CI nonroad engine rule costs are
understated.
B-33
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
attempt to quantify the effect that economic incentive
provisions will have on the overall costs of a particular
rule. In these cases, to the extent that affected sources
use economic incentive provisions to minimize
compliance costs, costs may be overstated.
Reasonable Further Progress (RFP)
and Attainment Costs
Considerable uncertainty surrounds the
development of States' control plans for meeting
ozone NAAQS attainment requirements. We develop
the RFP cost estimate by assuming that ozone
nonattainment areas (NAAs) will take credit for NOX
reductions for meeting progress requirements.
Additional area-specific analysis would be necessary to
determine the extent to which areas find NOX
reductions beneficial in meeting attainment and
progress requirement targets. Trading of NOX for
VOC to meet RFP requirements may result in
distributions of VOC and NOX emission reductions
that differ from those used in this analysis.
Future Year Control Cost Assumptions
The regulatory documents which provide cost
inputs to ERCAM and the IPM contain the most
recent data available, given existing technological
development. Between 2000 and 2010, additional
control technologies will allow sources to comply with
CAAA requirements at lower costs. For example, we
anticipates technological improvements for complying
with the multiple tiers of proposed emission standards
during the phase-in of nonroad engine controls; these
improvements will likely lead to reduced costs. In
addition, the costs for certain control equipment may
decrease over time as demand increases. The trend in
cost of selective catalytic reduction (SCR) costs
illustrate this. Costs have decreased over the past
three years as more facilities begin to apply the
technology. We also believes that even in the absence
of new emission standards, manufacturers will
eventually upgrade engines to improve performance or
to control emissions more cost-effectively; firms will
institute technologies such as turbocharging,
aftercooling, and variable-valve timing, all of which
improve engine performance.
Discount Rate Assumptions
We apply a rate of five percent to both the
discount rate and cost of capital. In some cases, we
base costs on analyses that apply alternative discount
rate assumptions (usually seven percent). Whenever
possible we recalculate total annualized costs (TAG)
for these rules in an effort to maintain consistency.
We use TACs, in turn, to calculate cost per ton
estimates that are applied to cost-equations. For some
source categories, there was insufficient data available
to identify the discount rate assumptions used in the
TAG estimate and the relevant cost per ton estimates.
For example, the national municipal landfills rule
applies only to facilities that emit above 50 megagrams
of non-methane organic compounds (NMOC).
However, the cost analysis for the proposed rule used
a different emissions cutoff (150 megagrams of
NMOC). Because we did not revise the cost analysis
to reflect the new cutoff, it is impossible to replicate
the calculations used to estimate the TAG in the final
rule. Additional research would be necessary to
calculate costs for the national municipal landfills rule
under alternative discount rate assumptions.
Source-Specific Cost Equations
We estimate the costs of Title III control
measures for point and area source emitters using an
average annual cost per ton value. For future analyses,
assuming sufficient data are available, it may be
possible to develop source-specific control equations
using a similar approach to that used for point source
NOX emitters. The point source inventory generally
includes larger, inventoried point sources, and the area
source inventory includes emissions for smaller
emission points. For this reason, we try to determine
whether sufficient cost data by plant size are available
to model costs specific to smaller plants, rather than
using an overall cost effectiveness value across all
plant sizes. If costs are available for sources by size,
we apply the cost estimates for larger sources to point
sources and apply cost estimates for smaller model
plants to area sources. We do not, however, use this
approach in all cases due to insufficient data.
B-34
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Sensitivity Analyses to Quantify Key
Uncertainties
We develop cost estimates based on a variety of
studies and assumptions regarding future behavioral
responses to provisions of the CAAA. These
assumptions (i.e., changes in consumption patterns,
input costs, and technological innovation) introduce
some uncertainty to the cost projections. In order to
characterize the potential importance of these
uncertainties with respect to several provisions, we
conduct sensitivity tests on selected Post-CAAA 2010
cost estimates. They are:
• Progress Requirements
• PM10 Nonattainment Area Controls
• Non-utility Stationary Source NOX
Costs
• California Reformulated Gasoline
• Low Emission Vehicle Costs
• NOX Tailpipe/Useful Life Standards
These provisions represent the most significant
contributors to total costs. Collectively, they
constitute nearly half of the total 2010 Post-CAAA
estimated costs. In addition, we examine the impact
of alternative discount rates on the cost assessments.
We summarize the results of the sensitivity analyses in
Table B-19.
A significant portion of the cost of attaining and
maintaining the one-hour average ozone NAAQS is
attributable to rate of progress (ROP) and rate of
further progress (RFP) compliance expenses. The
costs associated with reducing VOC (and NCQ
emissions in order to satisfy these progress
requirements are particularly difficult to estimate
because cost-effective control measures have not been
identified that will readily enable some ozone NAAs
to make the required precursor emissions cuts. The
estimated costs of unidentified VOC controls is one
source of uncertainty that affects the overall cost of
ROP/RFP requirements.
In the prospective, we assume that the cost of
Title I progress requirements is equal to the cost of
eliminating the VOC shortfall. We expect NAAs will
reduce VOC emissions using identified control
measures. The cost-effectiveness of each of these
measures is known, and we assume that the control
technique yielding the greatest reduction per dollar is
the first to be implemented, followed by the second
most cost-effective option, and so on until further
VOC cuts are no longer necessary to satisfy
ROP/RFP requirements.
We estimate that it will be possible to sufficiently
lower VOC emissions through implementation of
identified VOC controls for every NAA with a
shortfall, except Chicago and Milwaukee. These two
exceptions, however, have NOx waivers and cannot
credit NOX cuts towards RFP requirements. As a
result, they will have to significantly lower VOC
emissions; the necessary reduction is so sizable in
both areas that neither will be able to make the
required cuts, even if it adopts all of the identified
control measures. Thus, in order to satisfy ROP/RFP
requirements, Chicago and Milwaukee will have to
implement unidentified VOC emissions control
techniques. The estimated costs associated with these
measures are a source of uncertainty potentially
influencing the overall cost of Title I progress
requirements. We conduct a sensitivity analysis to
help characterize the influence of this uncertainty on
the 2010 progress requirements cost estimate.
We base the three scenarios of the sensitivity
analysis upon different assumptions regarding the
cost-effectiveness of VOC shortfall controls. For the
lower estimate, after applying identified controls (the
approximate cost per ton of reduction is known for
these measures), the remaining shortfall is eliminated
through the implementation of unidentified controls.
In the lower estimate scenario, we estimate the
marginal cost of these unidentified measures is equal
to the weighted average of the cost per ton estimate
from the recently revise ozone NAAQS RIA, which
is $10,000, and the average dollar per ton cost for
identified measures.30 The central estimate is identical
to the lower estimate with one exception, unidentified
controls are assumed to cost $10,000 per ton of
For example, in Chicago sixty percent of the required
reduction of VOC emissions will come from identified measures
at an average cost of $3,500 per ton. The remaining forty percent
will thus come through unidentified controls. This means that,
according to the lower estimate, the approximate cost per ton of
reduction through the implementation of unidentified controls is
$6,000 [$3,500(.60) + $10,000(.40)].
B-35
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
reduction in the central scenario. A flat $10,000 per
every ton of shortfall VOC emissions reduced, from
both identified and unidentified measures, is assumed
for the upper estimate. Our sensitivity analysis shows
that ROP/RFP costs range from $0.61 billion to $2.5
billion, with a central estimate of $1.1 billion. We
provide a more detailed breakdown of these costs in
Table B-20. It is important to note, that this
sensitivity analysis is the only case in which the
primary estimate of our cost analysis differs from the
central case in a sensitivity test.
B-36
-------
TableB-19
Factors Affecting Cost of Major CAAA Provisions
Provision
Progress
Requirements
Factors Affecting
Cost
Cost for unidentified
measures is most
uncertain.
Conduct
Sensitivity
Analysis?
Yes
Strategy for Sensitivity Analysis
Continue to examine costs of identified measures in other specific areas.
Lower Bound: Assume average per ton cost of identified measures (e.g.,
Potential Effect of
Uncertainty on Post-
CAAA 2010 Cost
Estimates1
Central Estimate:
$1.1
$3,500 in Chicago) for all reductions, including unidentified measures.
Central Estimate: Use cost figure for identified measures for that fraction of
reductions (e.g., $3,500 for 60 percent in Chicago) and assume $10,000 per ton
cost for unidentified measures. This central estimate reflects more recent cost
per ton information than was applied to our primary cost estimate.
Upper Bound: Assume $10,000 per ton cost for all reductions, including
identified measures. Our cost analysis adopts a conservative approach and
applies this cost per ton value to our primary cost analysis.
Range:
($0.06-$2.5)
Impact of revised ozone
standard.
No Emissions projections in the 812 Prospective do not include revisions to the
ozone NAAQS.
no estimate
California
Reformulated
Gasoline
Incremental fuel costs
show wide range and
are most uncertain.
Yes Lower Bound: Assume 7.3 cents per gallon cost from CARS.
Central Estimate: Current analysis assumes 12.3 cents per gallon cost from
CARS.
Upper Bound: Assume 17.3 cents per gallon cost from CARS.
Central Estimate:
$2.5
Gasoline sale
quantities are
important, but less
uncertain.
Yes Gasoline sales are a function of vehicle miles traveled (VMT).
Apply alternative VMT projection for California: California Motor Vehicle Stock,
Travel, and Fuel Forecast, California Department of Transportation, November
1997.
Alternative VMT would impact emission scenario.
Range:
($1.4-$3.5)
B-37
-------
Provision
Factors Affecting
Cost
Conduct
Sensitivity
Analysis?
Strategy for Sensitivity Analysis
Potential Effect of
Uncertainty on Post-
CAAA 2010 Cost
Estimates1
PM NAA Controls Base year emissions
and growth.
No Emissions projections and growth estimates are underlying assumptions of the
cost analysis.2
no estimate
Area specific plans may
differ from the "model
plan" applied.
Yes Strategy for sensitivity analysis includes:
1) Apply area-specific control measures where available.
2) Use "model plan" when area plans are unknown.
Cost per ton estimates
and effectiveness of
individual measures.
Yes Apply upper and lower bound cost estimates for model plan controls based on
the SCAQMP, the MRI study of agricultural operations, and the PM NAAQS
study:
Agricultural Tilling: Low$154/ton (1997 SCAQMP)
High $5,900/ton (midpoint of range from MRI study)
Construction: Low $1,900/ton (50% below value used)
High $5,700/ton (50% above value used)
Paved Roads: Low$50/ton (1997 SCAQMP)
High $1,350/ton (50% above value used)
Unpaved Roads: Low$560/ton (1997 SCAQMP)
High $2,700/ton for rural roads (PM NAAQS)
Central Estimate:
$2.2
Range:
($0.9-$3.3)
B-38
-------
Provision
Factors Affecting
Cost
Conduct
Sensitivity
Analysis?
Strategy for Sensitivity Analysis
Potential Effect of
Uncertainty on Post-
CAAA 2010 Cost
Estimates1
LEV Costs
Will 49-State LEV
occur?
No Recently agreed to by the 23 automobile manufacturers that sell cars in the US
and are regulated by EPA. Four States in the Northeast (MA, ME, NY, VT) have
opted not to join the NLEV program.
no estimate
Per vehicle costs.
Projected vehicle sales.
Yes Current analysis uses CARB's per vehicle cost estimates. These estimates are
the lowest and most fully documented, and differ from other industry estimates
by a factor often.
Lower Bound: 50% below study per vehicle cost estimates.
Central Estimate: Use current study (CARS adjusted for national sales volume)
per vehicle cost estimates.
Upper Bound: Use unadjusted CARS per vehicle cost estimates.
Yes Vehicle sales data were obtained from EPA's onboard vapor recovery RIA.
Apply alternative sales projection: DOE's Annual Energy Outlook 1998 - NEMS
Transportation Demand Model.
Central Estimate:
$2.2
Range:
($1.08-$2.5)
(national and CA LEV
combined)
B-39
-------
Provision
Factors Affecting
Cost
Conduct
Sensitivity
Analysis?
Strategy for Sensitivity Analysis
Potential Effect of
Uncertainty on Post-
CAAA 2010 Cost
Estimates1
Non-Utility
Stationary Source
NOx Costs
Unit-level cost
equations and cost per
ton.
Inventory data
elements (e.g.,
capacity, operating
rate) used in cost
calculations.
Yes ICI boilers account for 79 percent of the total point source NOx control cost
estimate in 2010. Apply+50 percent range. Other available data are 3-4 years
old and would not reflect the fact that the control technology is being
manufactured and applied by more sources.
No
Inventory data elements are well-defined for each point source category.
Central Estimate:
$2.2
Range:
($1.1 -$3.2)
no estimate
Cap applied to 37
States, proposed NOx
budgets affect only 22
States.
No Current estimates overstate costs for fuel combustors in the 15 States not
affected by the NOx cap. NOx SIP call RIA provides estimates for 22-state
program.
no estimate
Banking not accounted
for.
No
None
no estimate
NOx
Tailpipe/Useful
Life Standards
Per vehicle costs date
to 1991 FR Notice.
Yes Lower Bound: No alternative estimates. Scale down medium estimate by 50 Central Esimate:
percent. $1.7
Central Estimate: Use current $115 estimate from EPA.
Upper Bound: No alternative estimates. Scale up medium estimate by 50 Range:
percent. ($0.83-$2.5)
Vehicle Sales
Yes
Same as LEV projected vehicle sales. See above.
High Enhanced
I/M
Per vehicle costs are
most uncertain.
No Alternative per vehicle cost estimates are similar to the costs currently used in
the model.
no estimate
B-40
-------
Factors Affecting
Provision Cost
Vehicle registrations
Conduct
Sensitivity
Analysis?
No 1990 vehicle regis
Strategy for Sensitivity Analysis
itrations are well-documented, and the method used to proie
Potential Effect of
Uncertainty on Post-
CAAA 2010 Cost
Estimates1
iCt
are important, but less
uncertain.
future registrations based on population projections is sound.
no estimate
Discount Rate
Economic
Growth Case
Study
Vary the discount rate.
Macroeconomic growth
projections may affect
cost drivers.
Yes Current cost estimates use a five percent discount rate. Vary the cost estimates
using two alternative discount rates, three percent and seven percent.
No The current methodology for calculating PM emissions relies on activity level
projections more than macroeconomic growth rates. For example, the model
uses the USDA agricultural baseline projections of farm acres planted to
calculate PM emissions from agricultural production, the largest source of PM.
The PM sources influenced by macroeconomic growth rates contribute only
about five percent of total PM emissions.
Central Estimate:
$3.0
Range:
($2.8 -$3.2)
no estimate
Notes:
1 Estimates are in billion 1990 dollars.
B-41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
An additional source of uncertainty associated
with estimating the cost of ROP/RFP requirements
stems from the fact that the impact of the revised
ozone NAAQS is not incorporated into the Post-
CAAA scenario. We do not, however, conduct a
sensitivity test designed to characterize the influence
of the stricter NAAQS. In this instance, developing
a cost range would not be very meaningful since
benefits, as well as costs, would be affected by the
changed NAAQS.
Table B-20
Rate-of-Progress Cost Sensitivity Summary
Annual Post-CAAA2010 Costs (million 1990 dollars)
Ozone Nonattainment
Area
Chicago-Gary-Lake County
Dallas-Fort Worth
El Paso
Grand Rapids
Los Angeles-South Coast
Louisville
Milwaukee-Racine
Muskegon
Nashville
Salt Lake City
Sheyboygan
Kewaunee Co. Wl
Monitowoc Co. Wl
Total
Low Estimate
$430
0.4
8.5
0.2
0.8
0.1
120
0.6
15
28
0.5
0.2
1.1
$607
Central Estimate
$810
0.4
8.5
0.2
0.8
0.1
210
0.6
15
28
0.5
0.2
1.1
$1,080
High Estimate
$1,400
61
76
27
44
30
430
9.3
130
180
8.2
1.7
11.3
$ 2,500
PM10 nonattainment area controls account for
roughly seven percent of total annual costs in 2010.
Two sources of uncertainty with respect to our
estimate are: (i) how well the model plan mirrors
actual application of controls in nonattainment areas,
and (ii) how representative the cost per ton estimates
used in the model are of actual control measure costs.
We develop a sensitivity analysis that incorporates
both factors. We analyze how well the model replicate
selected SIP controls by applying the prospective cost
equations to areas that have already implemented
controls.31 The second part of the sensitivity analysis
assesses the impact of higher and lower cost per ton
estimates on the original set of control measures. In
cases where there were no alternative estimates, we
adjust values up and down by fifty percent.32
31 We use a survey of SIPs designed in 1991 and implemented
in 1995.
32 The sensitivity analysis does not reflect point source
controls.
B-42
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Our analysis highlights two primarily differences
between model and actual SIP designs. First, the
model plan probably overstates the application of
fugitive dust controls in practice. Second, over half of
the areas that have adopted emission measures
emphasized point source controls, at lower cost than
reductions that could be achieved through other
measures. Testing the uncertainty of cost per ton
values resulted in a range between $0.9 and $3.3 billion
in PM10 nonattainment costs. Total area source PM
control costs had a low estimate of $1.5 billion and a
high estimate of $3.3 billion. We summarize these
results in Table B-21.
Table B-21
Area Source PM Control Cost Sensitivity Analysis, Year 2000
(in million 1990 dollars)
Source Type
Agricultural Burning
Agricultural Tilling
Beef Cattle Feedlots
Construction Activity
Paved Roads
Unpaved Roads
Total
Central
Estimate
$39
3.9
1.7
1,700
440
0.8
$2,200
Model Plan Sensitivity
Analysis
$23
1.5
0.3
1,070
380
0.5
$1,500
Low
Estimate
$20
0
0.9
870
13
0.2
$907
High
Estimate
$58
20
2.6
2,600
650
1.2
$3,300
Note:
1 Examples include banking, trading, and emissions averaging provisions.
* Costs are in millions of 1990 dollars.
Another significant portion of total CAAA cost is
incurred by non-utility stationary NOX sources. This
category contributes approximately eight percent of
total costs in 2010. Its costs reflect unit-level costs for
combustion processes at industrial, commercial, and
institutional facilities. We identify the accuracy of
future control costs assessment (nationwide and for
the subset of OTAG states subject to the NOX SIP
Call) as a source of uncertainty that may affect cost
projections. Our evaluation of this uncertainty
involves applying alternative unit cost estimates. For
the sensitivity analysis, we estimate upper and lower
estimates by varying prospective costs up and down
by fifty percent.33 As a result, the central estimate for
2010 is $2.1 billion, and costs range from $1.1 billion
to $3.2 billion.
33 After reviewing alternative cost studies sponsored by
STAPPA/ALAPCO and the OTAG stationary source committee
EPA found the studies relied on the same sources used by the 812
Project Team. Consequently, the agency opted for the above
mentioned approach to varying cost inputs.
B-43
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
By 2010, we estimate California's reformulated
gasoline program will cost $2.5 billion annually. The
program is a significant factor in the our cost analysis,
because it represents eight percent of total annual
CAAA costs and ten percent of national gasoline sales.
We identify two primary sources of uncertainty,
projected car sales and projected gasoline
consumption levels. The sensitivity analysis varies
costs by applying the high and low cost per gallon
estimates developed by GARB. The test indicates that
cost may range between $1.4 and $3.5 billion.
Moreover, we integrate into the sensitivity test
alternative projections of car sales as a proxy for
consumption levels.34 California's Department of
Transportation calculated VMT projection
approximately five percent lower than the projection
we use in the prospective. This lower VMT estimate
estimaes costs of $2.27 billion in 2010 (compared with
the central estimate of $2.45 billion).
Our estimate of low emission vehicle (LEV) costs
are also subject to similar sources of uncertainty. We
rely on assumptions regarding the types of
implemented emission control technology, its
associated costs (estimated as costs per vehicle),
projected vehicle sales, and the extent to which LEV
will be adopted around the country. We conduct
sensitivity analyses for costs of both the California
LEV program and a 49-State LEV program. The
analyses reflect uncertainty with respect to cost per
vehicle and vehicle sales. To test the uncertainty
related to car sales, we use an alternative set of
projections from the Department of Energy. The low
estimate reflects per vehicle costs that are fifty percent
below baseline costs. We use CARB's high estimates,
which were unadjusted for national sales volume, to
develop a high estimate. Our sensitivity analysis result
are a low estimate of $0.55 billion and a high estimate
of $1.34 billion for California's LEV costs in 2010.
For a 49-State LEV program, the estimates range
between $0.53 billion to $1.34 billion. The central
estimates are $1.1 billion for the California LEV
program and $1.06 billion for the 49-State LEV
program. We summarize results in Table B-22.
34 EPA used VMT projections by California Department of
Transportation.
B-44
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-22
Results of Sensitivity Analysis of LEV Costs
Post-CAAA 2010 Annual Cost (million dollars)
Program
California LEV
49-State LEV
Central
$1,100
1,060
Low
$550
530
High
$1,100
1,300
Alternative VMT
$870
820
TOTAL $2,200 $1,080
Note: Columns may not sum to totals due to rounding
$ 2,500
$ 1,700
Costs associated with NOx Tailpipe/Useful Life
Standards are a sizable portion of both Title II and
total CAAA costs. By 2010, we estimate these costs
will contribute nineteen percent of the Title II motor
vehicle costs. As a share of total CAAA costs, it is six
percent. Key sources of uncertainty are the same as
those associated with LEV, per vehicle costs and
projected car sales. Similarly, the sensitivity analyses:
(i) scaled per vehicle costs up and down by fifty
percent; and (ii) used alternative sales projections
generated by the Department of Energy. The
variation of cost inputs produced a cost range of $0.83
billion to $2.48 billion for 2010. Application of sales
projections by the Department of Energy resulted in
costs slightly lower than the central estimate, $1.25
billion compared with $1.65 billion respectively.
Unlike the other sensitivity analyses, the discount
rate affects cost estimates for multiple provisions. As
noted, we calculate total annualized cost estimates
using a 5 percent discount rate. However, variations
in the discount rate can potentially have a significant
effect on the overall cost estimate, because the
discount rate is also used as an estimate of the real
cost of capital to finance pollution control equipment.
Our sensitivity analysis of annualized focuses on
source categories where available information is
available to distinguish capital from operating and
maintenance expenses. Source sectors and pollutants
include non-utility VOC and NOX and area source
VOC, NOx, and PM.
We present the discount rate sensitivity analysis
results in Table B-23. We estimate total annualized
costs using three discount rates, three percent, five
percent, and seven percent. As a result, costs
estimates vary from two percent to fifteen percent.
The results of the analysis do not assess how the
discount rate would impact a large fraction of the total
estimated costs in the Post-CAAA scenarios.
Excluded costs include motor vehicles and PM10 area
source categories in nonattainment areas. Most of the
capital costs associated with motor vehicle provisions
are in the form of research and development. PM10
area source costs are generally calculated using a cost
per ton estimate, which does not have sufficient
available data for identifying the discount rate.
B-45
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table B-23
Discount Rate Sensitivity Analysis for 2010 Cost Estimates
Capital Cost (million 1990 dollars)
Discount Rate
Percent Difference
Sector
Non-Utility
Area
Total
Pollutant
VOC
NOX
VOC
NOX
PM
3%
$480
1,400
508
17
430
$ 2,800
5%
$501
1,500
540
18
440
$ 3,000
7%
$530
1,600
570
20
440
$3,100
Rate
11%
12%
11%
15%
2%
11%
The sensitivity analyses assess the potential effect
of uncertainty on components of the total CAAA
costs. We project costs for progress requirements,
PM nonattainment area controls, non-utility NOX
sources, and utility emissions, based on modeling
future emission controls. Accurately identifying the
set of controls that will be adopted introduces a key
source of uncertainty. The analyses indicate that there
may be considerable variability in the cost estimates.
However, it is important to note that for most of
these scenarios, the high estimates are most likely
representative of upper bounds. There are two
sources of uncertainty associated with the motor
vehicle provisions. The first source is projecting
future car sales. The second is the accuracy of per
vehicle costs. The high and low estimates relative to
the central estimate do not present as wide a range.
B-46
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
GARB, 1990: California Air Resources Board, "Reformulated Gasoline: Proposed Phase I
Specifications - Technical Support Document," August 13, 1990.
GARB, 1991: California Air Resources Board, "California Phase 2 Reformulated Gasoline Specifications -
Volume 1 - Proposed Regulations for California Phase 2 Reformulated Gasoline - Staff Report," October
4,1991.
DRI, 1993: DRI/McGraw-Hill, "Review of the U.S. Economy: Summer 1993 - The 25-Year Outlook:
Dynamics Within the Trend," Summer 1993.
EPA, 1990: U.S. Environmental Protection Agency, "Regulatory Impact Analysis: Control of Sulfur and
Aromatic Contents of On-Highway Diesel Fuel," NTIS #PB93-207660,June 1990.
EPA, 1991d: U.S. Environmental Protection Agency, "MOBILE4.1 Fuel Consumption Model," Draft output
provided by the Office of Mobile Sources, Ann Arbor, MI, August 12,1991.
EPA, 1992: U.S. Environmental Protection Agency, "Regulatory Impact Analysis/Regulatory Flexibility Act:
Operating Permits Program Part 70," EPA-450/2-91-011,June 1992.
EPA, 1992c: U.S. Environmental Protection Agency, "I/M Costs, Benefits, and Impacts Analysis," (Draft),
February 1992.
EPA, 1993d: U.S. Environmental Protection Agency, "Regulatory Impact Analysis — On-Board Diagnostics,"
Office of Air and Radiation, Office of Mobile Sources, January 27,1993.
EPA, 1993f: U.S. Environmental Protection Agency, "Final Regulatory Impact Analysis: Onboard Refueling
Emission Regulations for Light Duty Vehicles and Trucks and Heavy Duty Vehicles," Office of Mobile
Sources, Regulation Development and Support Division, December 1, 1993.
EPA, 1993g: U.S. Environmental Protection Agency, "Final Regulatory Impact Analysis for Reformulated
Gasoline," Office of Mobile Sources, Ann Arbor, MI, December 13, 1993.
EPA, 1995: U.S. Environmental Protection Agency, "Regulatory Impact Analysis for Part 71, Federal Operating
Permits Rules," May 1995.
EPA, 1997a: U.S. Environmental Protection Agency, "Air Emissions Estimates from Electric Power Generation
for the CAAA Section 812 Prospective Study," prepared by Office of Air and Radiation, February 1997.
EPA, 1997f: U.S. Environmental Protection Agency, "Final Regulatory Impact Analysis: Control of Emissions
of Air Pollution from Highway Heavy-Duty Engines," Office of Mobile Sources, EPA-420-R-97-011,
September 1997.
EPA, 1996: U.S. Environmental Protection Agency, "EPA's Clean Air Power Initiative," Oct. 1996.
Obtained from http://www.epa.gov/capi/capifs3.html, September 1998.
B-47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
FHWA, 1991: Federal Highway Administration, U.S. Department of Transportation, "Highway Statistics 1991,"
1991.
Green, 1994: K. Green, Northeast States for Coordinated Air Use Management (NESCAUM), letter to Bruce
Carhart, Ozone Transport Commission, Washington, DC, June 2,1994.
Harrington, Winston and Virginia McConnell, "Coase and Car Repair: Who Should Be Responsible for Emissions
of Vehicles in Use?" Discussion Paper 99-22, Resources for the Future, February 1999.
Hazilla, Michael and Raymond J. Kopp. "Social Cost of Environmental Quality Regulations: A General
Equilibrium Analysis." Journal of Political Economy, 98 (1990): 853-873.
Oge, 1997: Margo T. Oge, "Clean Fuel Fleet Implementation," memorandum from Margo T. Oge, Director,
Office of Mobile Sources to EPA Regional Directors, U.S. Environmental Protection Agency,
Washington, DC, May 22,1997.
Pechan, 1997a: E.H. Pechan & Associates, Inc., "Emission Projections for the Clean Air Act Section 812
Prospective Analysis," External Review Draft, February 12, 1997.
Pechan, 1997b: E.H. Pechan & Associates, Inc., "The Emission Reduction and Cost Analysis Model for NOX
(ERCAM-NO,)," Revised Documentation, September 1997.
Pechan, 1997c: E.H. Pechan & Associates, Inc., "Analysis of Costs and Benefits of a National Low Emission
Vehicle Program," Springfield, VA, prepared for Industrial Economics, Inc., Cambridge, MA, December
1997.
Pechan, 1998: E.H.Pechan & Associates, Inc. "Clean Air Act Section 812 Prospective Cost Analysis - Draft
Report," Springfield, VA, prepared for Industrial Economics, Inc., Cambridge, MA, September 1998.
Wysor, 1988: T. Wysor, Office of Mobile Sources, U.S. Environmental Protection Agency, Ann Arbor, MI,
personal communication, April 4,1988.
B-48
-------
Quality Modeling
X
0)
Introduction
Section 812 of the 1990 Clean Air Act
Amendments (CAAA) requires the U.S. Environ-
mental Protection Agency (EPA) to perform periodic,
comprehensive, prospective analyses of the costs and
benefits associated with programs implemented
pursuant to the Clean Air Act (CAA). Such analysis
requires the estimation of future-year emissions levels
and associated air-quality-related values for scenarios
reflecting compliance with the CAA, as well as for
scenarios for which the effects of programs associated
with the CAA are not accounted for in establishing the
future-year estimates. This report summarizes the
results of an air quality modeling and analysis study
designed to estimate the effects of the CAAA on
future air quality. The Section 812 prospective study
includes analysis of following criteria pollutants:
ozone, particulate matter (PM), sulfur dioxide (SO2),
nitrogen oxide (NO), nitrogen dioxide (NO2), and
carbon monoxide (CO). Future-year estimates of
these atmospheric constituents were obtained through
the application of air quality modeling tools and
techniques, as described in this report.
An integral component of the modeling analysis
was the estimation of future-year emission levels
associated with the two CAAA scenarios and two
future years. Scenarios that incorporate the emission
reductions associated with CAAA are referred to as
Post-CAAA while those that incorporate growth but
reflect 1990 regulations are referred to as Pre-CAAA.
The two future years considered in the analysis are
2000 and 2010. The emissions estimates (Pechan,
1998) provide the basis for the estimation of ozone,
PM, and other criteria pollutant concentrations
associated with each scenario and future year.
Future-year estimates of ozone concentrations
were obtained through the combined application of
the Urban Airshed Model (UAM) and the variable-
grid UAM (UAM-V), yielding urban- and/or regional-
scale estimates of ozone concentrations for each
scenario and future year for the entire U.S. (48
contiguous states).
Concentrations of primary and secondary PM for
the future-year scenarios (including PM10, with a
diameter of less than 10 micrometers, and PM25, with
a diameter of less than 2.5 micrometers) were
estimated through the combined application of the
Regional Acid Deposition Model/Regional Particulate
Model (RADM/RPM) and the Regulatory Modeling
System for Aerosols and Acid Deposition
(REMSAD). RADM/RPM was used for the eastern
U.S., while REMSAD was applied to the analysis of
PM within the western U.S.
An emissions-based, linear "roll-back" technique
was used to estimate future-year concentrations for
the other pollutants considered as part of this analysis
- SO2, NO, NO2, and CO.
Following application of the modeling techniques,
site-specific estimates of future-year air quality were
obtained by adjusting observational data (correspond-
ing to a base year of 1990) in accordance with the
changes in air quality predicted by the modeling
systems. Statistical quantities or "profiles" describing
the predicted concentration distributions for each
monitoring site were then calculated. The resulting
statistical concentration distributions provide the basis
for the examination and quantification of the effects
of changes in air quality on health, agriculture, and the
economy (i.e., physical effects and economic valuation
modeling) resulting from compliance with the CAAA.
The remainder of this report summarizes the
methods and results of the section 812 prospective air
quality modeling analysis. An overview of the
modeling/analysis methodology is provided in section
2. The methods and results for ozone are presented
in section 3. The methods and results for PM are
C-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
provided in section 4. The linear-rollback modeling
for the other criteria pollutants is summarized in
section 5. A discussion of the attributes and limita-
tions of the modeling analysis methodologies is
provided in section 6. Finally, recommendations for
further research are given in section 7.
Overview of the Section 812
Prospective Modeling Analysis
The air quality modeling component of the
section 812 prospective analysis included the
application of a variety of air quality modeling tools
and techniques, as well as the combined use of
observational data and modeling results to estimate
future-year concentrations of several criteria
pollutants. An overview of the modeling approach is
provided in this section of the report.
The overall objective of the modeling exercise was
to provide base- and future-year estimates of ozone,
PM, SO2, NO, NO2, and CO for the subsequent
analysis of the effects of the CAAA on health,
agriculture, and the economy within the continental
U.S. Although the CAAA applies to the entire nation,
due to geographical considerations, the modeling
domain includes the contiguous 48 states. The
modeling was performed for a base year (1990) and
for four future-year scenarios. The future-year
scenarios include Post-CAAA and Pre-CAAA
scenarios (the former incorporating emission changes
associated with measures and programs pursuant to
the CAAA) for the years 2000 and 2010. These years
were selected to accommodate implementation
schedules and time for effectiveness periods
associated with many of the CAAA measures and
programs.
Air Quality Models and Databases
To the extent possible, the section 812
prospective modeling analysis utilized existing
modeling databases (from State Implementation Plan
or other regional-scale modeling efforts). To
accommodate the geographical extent and resolution
required for this study, these included the input
databases corresponding to both urban- and regional-
scale applications of several different modeling
systems. The lack of an existing comprehensive, fully
tested, integrated modeling system (and associated
databases) for use in this study precluded the
integrated analysis of the various pollutants. This,
however, may be an area for improving future
prospective analyses.
The UAM and UAM-V modeling systems were
applied to the analysis of the effects of the CAAA on
ozone air quality. Specifically, the UAM-V modeling
system was applied for the regional-scale analysis of
ozone concentrations within both the eastern and
western portions of the U.S. (separately). The analysis
of the eastern U.S. relied upon the use of modeling
databases developed as part of the Ozone Transport
Assessment Group (OTAG) regional-scale modeling
analysis. This modeling system was also applied for
the western U.S., but at a relatively coarse resolution.
To enhance the analysis for selected urban areas in the
western U.S., the regional-scale modeling results were
supplemented with higher-resolution modeling results
for Los Angeles, Phoenix, and the San Francisco Bay
Area. The results for both Los Angeles and Phoenix
were obtained using the UAM modeling system, while
those for the San Francisco Bay Area were obtained
using the UAM-V modeling system.
The RADM/RPM and REMSAD modeling
systems were used to estimate PM concentrations for
the eastern and western U.S., respectively. Again,
many of the inputs for application of these models
were developed as part of other studies and adapted
for use in the section 812 prospective modeling
analysis.
As noted earlier, an emissions-based, linear "roll-
back" technique was used to estimate future-year
concentrations for SO2, NO, NO2, and CO. This
approach was used for all areas of the continental U.S.
All of the modeling applications relied on the use
of detailed estimates of emissions for the base year
and each of the future-year scenarios. These are
described by (Pechan, 1998). Modeling emission
C-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
inventories were prepared using the Emissions
Preprocessing System (EPS2.5) (SAI, 1992).
Methodology for the Combined Use of
Observations and Air Quality Modeling
Results
The 812 prospective modeling analysis included
several steps. First, concentration estimates for each
pollutant of interest, corresponding to a base year of
1990, were prepared based on 1990 emissions and
application of the appropriate modeling tool(s). For
each scenario, the remaining steps consisted of (1)
preparation of future-year, model-ready emission
inventory estimates, (2) application of the appropriate
modeling technique to estimate the change in air
quality from the base year of 1990, (3) adjustment of
the 1990 observed data to reflect the change as
predicted by the modeling system, and (4) calculation
of statistical quantities or "profiles" describing the
predicted pollutant concentration distribution for each
monitoring site.
Conceptually, the methodology for estimating
future-year ozone air quality using both observations
and modeling results is rather simple. The modeling
results are used to calculate adjustment factors for
each monitoring site that is located within the
modeling domain. This is done on a grid-cell by grid-
cell basis (i.e., the adjustment factor for a monitoring
site is based on the simulation results for the grid cell
in which it is located). The adjustment factor
represents the ratio of the future-year-scenario
concentrations to the base-year concentrations and is
calculated using appropriately matched values for
several different concentration levels (i.e., the changes
in concentration are dependent upon concentration
level). The observed concentrations for each
monitoring site are then modified using the site-
specific (or grid-cell-specific) adjustment factors. The
resulting values represent the estimated future-year
concentrations.
This approach to estimating future air quality
differs from that for a typical air quality model
application (e.g., for ozone attainment demonstration
purposes) in that the modeling results are used in a
relative sense, rather than an absolute sense. This may
enhance the reliability of the future-year concentration
estimates, especially in the event that the uncertainty
inherent in the absolute concentration values is greater
than that associated with the response of the modeling
system to changes in emissions.
Although the ratios are calculated using modeling
results for a limited number of simulation days, it is
assumed, using this methodology, that the ratios can
be used to represent longer time periods.
Consequently, all observations contained within the
dataset (a few exceptions are discussed later in this
document) are adjusted using the model-derived
ratios. Thus, by applying the model-derived ratios to
observed values representing longer periods, this
approach also permits the estimation of seasonal and
annual concentration distributions - a requirement for
this study. Following the calculation of various la-
bour rolling averages for each monitoring site,
statistical quantities, or "profiles", describing the
ozone distribution for each monitor are then
calculated.
The future-year air quality profile estimation
methodology, as applied to the analysis of results for
the section 812 prospective analysis, is described in
detail in the remaining sections of this document. A
flowchart illustrating the methodology is provided in
Figure C-l. The procedure makes use of the statistical
functions and data handling capabilities of the
Statistical Analysis Software (SAS) package.
C-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-1
Schematic diagram of the future-year concentration estimation methodology.
1990
(Base Case)
city
\
AQ Model
Predictions
Cone
-city3
j»oV \ \^. \/\
\ \ \ ^yV ^^-
countyl
Cone
Cone
\\ \ \ \ \~x
-|post-CAAA|->
Ratio to Base Case
S\\\N\—TpcrtCAAAl-fr \\S\SS\
\\ \ \ \ \\ \\ \ \ \ \\
Ratio to Base Case
Adjustments
Factors
cityl
AQ
Observations
i city2
cityS
AQ
Results
countyl
Concentration
distributions
Concentration
distributions
Concentration
distributions
[NOTE: Figure illustrates how model results and observations are used to produce air quality profiles (concentration distributions)
for the benefits analysis. The figure shows model runs at the top; four sets of "ratios" of model results in space in the middle; and
frequency distributions of pollutant monitor concentrations and the space-dependent scaling of these by the ratios of the model
predictions on the bottom.]
Estimating the Effects of the
CAAA on Ozone Air Quality
Future-year ozone concentrations corresponding
to the Post-CAAA and Pre-CAAA scenarios were
estimated through application of the UAM and UAM-
V modeling systems. This section of the report
contains an overview of the modeling systems and, for
each geographical domain, a description of the
application procedures and results. The calculation
of ozone air quality profiles using the combined
modeling results from the regional- and urban-scale
modeling applications is also described.
For ease of reading, all figures follow the text in
this section.
C-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Overview of The UAM and UAM-V
Photochemical Modeling
Systems
UAM
The UAM is a three-dimensional photochemical
grid model that calculates concentrations of pollutants
by simulating the physical and chemical processes that
occur in the atmosphere. It is formulated based on
the atmospheric diffusion or species continuity
equation. This equation represents a mass balance
that includes all of the relevant emissions, transport,
diffusion, chemical reaction, and removal processes in
mathematical terms. The UAM incorporates the
Carbon Bond IV chemical mechanism, which groups
pollutant species to limit the number of chemical
reactions, while permitting reasonable accuracy in
simulating ozone and its precursors.
The major factors that affect photochemical air
quality include:
• spatial distribution of emissions of volatile
organic compounds (VOC) and NOX, both
natural and anthropogenic,
• composition of the emitted VOC and NO^
• spatial and temporal variations in the wind
fields,
• dynamics of the boundary layer, including
stability and the level of mixing,
• chemical reactions involving VOC, NOX, and
other important species,
• diurnal variations of solar insolation and
temperature,
• loss of ozone and ozone precursors by dry
and wet deposition, and
• ambient background concentration of VOC,
NOx, and other species in, immediately
upwind, and above the region of study.
The UAM simulates all of these processes. The
species continuity equation is solved using the
following fractional steps: emissions are injected;
horizontal advection/diffusion is calculated; vertical
advection/diffusion and deposition are calculated; and
chemical transformations are performed for reactive
pollutants. The UAM performs these four calculations
during each time step. The maximum time step is a
function of the grid size and the maximum wind
velocity and diffusion coefficient. The typical time
step is 10-15 minutes for coarse (10-20 km) grids and
a few minutes for fine (1-2 km) grids.
Because it accounts for spatial and temporal
variations as well as differences in the reactivity of
emissions, the UAM is ideal for evaluating the air-
quality effects of emission control scenarios. This is
achieved by first replicating a historical ozone episode
to establish a base-case simulation. Model inputs are
prepared from observed meteorological, emissions,
and air quality data for the episode days using
prognostic meteorological modeling and/or diagnostic
and interpolative modeling techniques. The model is
then applied with these inputs, and the results are
evaluated to determine model performance. Once the
model results have been evaluated and determined to
perform within prescribed levels, the same base-case
meteorological inputs are combined with modified or
projected emission inventories to simulate possible
alternative/future emission scenarios.
The current UAM modeling system was released
by the EPA in 1990 and is fully documented in the
UAM user's guide (SAI, 1990). Features of the
modeling system include a mixing-height-based
vertical coordinate system and flux- and process-
analysis capabilities to facilitate the comprehensive
assessment of model performance and the
interpretation of simulation results.
UAM-V
The UAM-V modeling system represents an
extension of the UAM. Like UAM, the UAM-V
incorporates the Carbon Bond IV chemical
mechanism. Other features of the UAM-V modeling
system include:
C-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
• Variable vertical grid structure: The structure of
vertical layers can be arbitrarily defined. This
allows for higher resolution near the surface
and facilitates matching with output from
prognostic meteorological models.
• Three-dimensional meteorological inputs: The
meteorological inputs for UAM-V vary
spatially and temporally. These are usually
calculated using a prognostic meteorological
model.
• Variable grid resolution for chemical kinetic
calculations: A chemical aggregation scheme
can be employed, allowing chemistry
calculations to be performed on a variable
grid while advection/diffusion and emissions
injections are performed on a fixed grid.
• Two-way nested grid: Finer grids can be
imbedded in coarser grids for more detailed
representation of advection/diffusion,
chemistry, and emissions. Several levels of
nesting can be accommodated.
• Updated chemical mechanism: The original
Carbon Bond IV chemical mechanism has
been updated to include the XO2/RO2
reaction, along with new temperature effects
for PAN reactions. Aqueous-phase
chemistry is also an option.
• Dry deposition algorithm: The dry deposition
algorithm is similar to that used by the
Regional Acid Deposition Model (RADM).
• True mass balance: Concentrations are advected
and diffused in the model using units of mass
per unit volume rather than parts per million.
This maintains true mass balance in the
advection and diffusion calculations.
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
period, the exceedances occurred mostly within the
southeastern U.S. During the 1995 period, high ozone
concentrations were observed in several regions
including the Lake Michigan area, the Northeast
Corridor, and the Southeast. These periods were
chosen to be representative of regional-scale ozone
transport events for the southeastern and eastern U.S.
respectively. In both cases, the extent of the high
ozone concentrations is attributable to persistent,
regional-scale ozone conducive meteorological
conditions. The simulation periods include two and
three initialization (or start-up) days, respectively.
These are included to reduce the effects of
uncertainties in the initial conditions on the simulation
results.
Input Preparation
The UAM-V modeling system requires a variety
of input files that contain information pertaining to
the modeling domain and simulation period. These
include gridded, day-specific emissions estimates and
meteorological fields; initial and boundary conditions;
and land-use information.
Separate emission inventories were prepared for
the base-year and each of the future-year scenarios.
All other inputs were specified for the base-year
model application (1990) and remained unchanged for
each future-year modeling scenario.
Modeling Emission Inventories
The UAM-V requires detailed emission
inventories containing temporally allocated emissions
for each grid cell in the modeling domain and for all
primary pollutant species represented by the chemical
mechanism. An extended version of EPA's UAM
Emissions Preprocessor System, Version 2.0, or EPS
2.0 (SAI, 1992) called EPS 2.5e was used to process
the inventories. In addition to the capabilities of EPS
2.0, this system has been enhanced to facilitate
regional-scale model applications of participate matter
and toxic species, as well as ozone precursors.
Each inventory includes weekday/weekend area
source emissions, typical summer day utility emissions,
weekday/weekend non-utility point source emissions,
and day-specific biogenic emissions. The on-road
motor-vehicle emissions were based on typical
summer weekday/weekend estimates.
Anthropogenic input emissions inventory data
were provided by Pechan (1998). These included area
and point source emissions data from the National
Particulates Inventory (by county and for specific
point sources), county-level vehicle miles traveled
(VMT) estimates, and mobile-source emission factors
for VOC, NOx, and CO. Area source emissions
include emissions from a variety of sources such as
commercial and residential fuel combustion, non-
point-source industrial emissions, solvent utilization,
construction equipment, off-highway vehicles,
gasoline distribution, furniture refinishing, and lawn
mowers. Day-specific, model-ready biogenic emission
inventories were obtained from the OTAG database.
Preparation of the emission inventory data is
described in detail by Pechan (1998). A brief
description of the emissions processing is provided in
this section.
Preliminary processing of the data prior to the
application of the EPS 2.5e system was necessary.
This consisted of generating the on-road mobile
emissions and reformatting all data into Atmospheric
Information Retrieval System (AIRS) Mobile-Source
Subsystem (AMS) and Facility Subsystem (AFS) work-
file formats. On-road mobile emissions were
generated using the inputs provided by EPA and the
MOBILESa model. The outputs from MOBILESa
include future-year emissions of the ozone precursor
pollutants VOC, NOX, and CO. MOBILESa accesses
a matrix of emissions factors that are based on
temperature, speed, and other site-specific parameters.
Estimates of VMT were multiplied by emission
factors to generate on-road motor vehicle emission
estimates. The VMT estimates were provided at
county level and were broken down into 12 different
urban and rural roadway classifications.
C-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
All anthropogenic emission inputs to UAM-V
were preprocessed through the EPS 2.5e emissions
processing system. Photochemical grid models such
as the UAM-V require detailed emission inventories,
containing hourly emissions for each grid cell in the
modeling domain for each species being simulated.
The core EPS system is a series of FORTRAN
modules that incorporate spatial, temporal, and
chemical resolution into an emissions inventory used
for photochemical modeling. Point, area, and mobile
source emission data were processed separately to
facilitate both data tracking for quality control and the
use of the data in evaluating the effects of alternative
control strategies on simulated air pollutant
concentrations. The mobile source component was
further broken down into rural and urban motor
vehicle emissions based on the roadway classifications.
The model-ready components (including biogenic)
were then merged to generate the final model inputs.
The UAM-V requires hourly estimates of
emissions for each grid cell to accurately simulate
hourly concentrations of ozone. Accordingly, annual
average or peak ozone season daily emission rates
must be adjusted to reflect the conditions of the
ozone episode being modeled, including seasonal
adjustments for activity levels (if base year emissions
are reported as annual averages), adjustments for the
day of the week, and hourly temperature and activity
adjustments for each hour of the episode day. EPA
has developed a default set of temporal allocation
factors (TAF) for each source category and these have
been incorporated into EPS 2.5e. TAF were applied
to all model inputs. For the eastern U.S. domain, the
available typical peak ozone season day NOX and VOC
emissions were adjusted for day of week and hourly
allocation.
The Carbon-Bond IV chemical mechanism
employed by the UAM-V modeling system, groups or
"lumps" pollutants to limit the number of reactions
and species to a reasonable level while permitting
reasonable accuracy in predicting air quality. Ozone
precursor hydrocarbon emissions were aggregated into
the carbon-bond species required by the UAM-V
using speciation profiles developed by the EPA (1991)
and the default assignments provided with EPS. The
chemical speciation scheme for VOCs includes eight
categories: olefins, paraffins, toluene, xylene,
formaldehyde, higher aldehydes, ethenes, and
isoprene. For this study, the default NOXspeciation of
90 percent NO and 10 percent NO2 by weight,
included in EPS 2.5e, was assumed for all point and
area sources.
For the UAM-V model to accurately simulate
observed air quality concentrations for the selected
grid, it must be supplied with emissions data that have
the same degree of spatial resolution (i.e., by grid cell).
The effort required to implement this resolution
varies depending on the type of source. For point
sources, geographical coordinates for each source,
typically reported to within a fraction of a kilometer,
are used for direct assignment of emissions to the
appropriate grid cells. By contrast, spatial resolution
of emissions reported as county totals, as is usually the
case for area sources and motor vehicles, requires
substantially more effort. The most commonly
employed approach for apportioning county-level
emissions to grid cells is to use a surrogate indicator
for spatial distribution of emission levels or activity
(e.g., population, type of land use, or location of major
links such as interstate roadways or airport runways).
A spatial allocation surrogate is a quantity whose area!
distribution is either known or has been estimated and
is assumed to be similar to the areal distribution of
emissions from some source category whose spatial
distribution is not well known. County-level
emissions are spatially allocated to the grid cells of the
modeling domain. Surrogate data input used to create
the spatial allocation factors included U.S. Geological
Survey (USGS) land-use data, 1990 census data, and
digitized county boundaries.
Emissions totals by component for VOC, NOX,
and CO for the base- and future-year scenarios are
provided in Table C-l. This table shows increases in
VOC and NOX under the Pre-CAAA scenarios and
substantial decreases under the Post-CAAA scenarios.
The decreases in VOC are primarily attributable to
reductions in area- source and motor-vehicle
emissions. The decreases in NOX are due to decreases
in motor-vehicle and utility and non-utility point-
source emissions.
C-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Air Quality, Meteorological, and
Land-Use Inputs
The air quality, meteorological, and land-use
inputs for application of the UAM-V modeling system
for this study were identical to those used for the
OTAG modeling exercise. The initial and boundary
concentrations for the OTAG simulations were
represented by "clean" air values for all species. The
individual species concentrations vary slightly with
elevation (or height above the ground) and with time
of day, and are approximately 0.1 parts per billion
(ppb) of NOX, 5 parts per billion carbon (ppbC) of
reactive hydrocarbons (RHC), and 100 ppb of CO.
The boundary concentrations for ozone range from
about 31 to 34 ppb. Further detail on the OTAG
initial and boundary concentrations has been
presented in OTAG publications (e.g., Deuel et. al.,
1996).
Other input data required by the UAM-V model
for simulating the ozone episodes (including the
meteorological and land-use inputs) were obtained
directly from the OTAG datasets without
modification. Model options were the same in the
current application as in the OTAG application,
except that the plume-in-grid (P-i-G) treatment (a
detailed treatment of the chemistry and geometry of
plumes from elevated point sources) was not
employed. This exception was made in order to
reduce the demands on computer resources.
C-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-1
Emission Totals by Component for each Scenario for the OTAG Domain (tpd)
voc
Area
Onroad Mobile
Point
Utility
Total
Base 1990
33,417
17,518
8,247
87
59,270
2000 Pre-
CAAA
38,517
15,102
9,027
81
62,727
2000 Post-
CAAA
27,982
10,074
7,317
82
45,454
2010 Pre-
CAAA
43,113
17,400
10,194
111
70,818
2010 Post-
CAAA
8,638
8,552
8,204
113
45,507
NOx
Area
Onroad Mobile
Point
Utility
Total
Base 1990
12,109
17,915
6,647
17,637
54,307
2000 Pre-
CAAA
13,858
17,463
7,345
20,668
59,335
2000 Post-
CAAA
13,351
14,923
4,444
8,254
40,972
2010 Pre-
CAAA
15,770
20,222
8,365
22,670
67,026
2010 Post-
CAAA
13,741
12,616
4,681
5,182
36,220
CO
Area
Onroad Mobile
Point
Utility
Total
Base 1990
46,606
147,842
13,766
710
208,924
2000 Pre-
CAAA
53,087
112,656
15,463
784
181,990
2000 Post-
CAAA
51,544
84,569
15,463
811
152,388
2010 Pre-
CAAA
58,952
124,385
17,192
1,225
201,753
2010 Post-
CAAA
57,100
78,396
17,192
1,327
154,015
C-10
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
UAM-V Simulation Results for the
Eastern U.S.
Model Performance
The assessment of model performance is an
important component of a modeling analysis and is
used to ensure that the modeling system, including the
inputs, is able to replicate the observed concentration
levels associated with the historical modeling episode
period. The evaluation of model performance is
typically achieved through the comparison of
simulated concentrations with observed data. In this
case, the observed data correspond to the actual
episode period and the emissions reflect emission
levels for that same period/year. For the OTAG
modeling component, model performance considered
the base-case applications for 1993 and 1995.
Model performance for the OTAG episodes is
documented in the OTAG modeling report (OTAG,
1997). In general, the observed ozone concentration
levels were represented in the simulations, with some
over- and under-estimation of the maximum values.
Scatter plots comparing the simulated and observed
concentrations for key modeling days for each episode
(28 July 1993 and 15 July 1995) show generally good
agreement between the simulated and observed values,
with some tendency for over- and underestimation on
all days, distributed among the concentration levels
(scatter along the axis). These are typical of the
comparisons for the other simulation days.
Since the simulation results corresponding to all
concentration levels will be used to adjust the
observed data for Section 812 modeling analysis a
comparison of the mean values was also performed.
Plots comparing the mean values for each simulation
day of the 1993 and 1995 simulation periods in both
cases show that the mean simulated values are slightly
higher than the mean observed values, but the day-to-
day tendencies are similar.
For the 1993 simulation period, the mean
unsigned relative error (or normalized bias) ranges
from approximately -15 percent to 1 percent. The
corresponding values for the 1995 simulation period
are -12 to 9 percent. These are all within the EPA
recommended range (for urban-scale modeling) of
+ 15 percent. For both simulation periods, the mean
relative error (or gross error) is less than 25 percent
for each simulation day. The EPA recommended
range is less than 35 percent.
The good agreement between the simulated and
observed ozone concentrations, suggests that the
OTAG modeling system (including the
meteorological, air quality, and geographical inputs)
provides an appropriate basis for the Section 812
prospective modeling.
UAM-V Modeling Results
The UAM-V simulation results for the Pre- and
Post-CAAA scenarios were used in this study to
calculate factors for adjustment of observed data and
estimation of future-year concentration levels. These
were calculated by comparing the simulated
concentrations corresponding to each future-
year/scenario simulation with those for the base-year
simulation (1990). Examples of this comparison are
illustrated using isopleth plots for maximum ozone
concentration in Figures C-2 and C-3.1 These isopleth
plots correspond to 1995 simulation period and
depict differences in maximum ozone concentration
for 15 July between the 1990 baseline and the 2010
Pre- and 2010 Post-CAAA scenarios, respectively.
The differences are calculated as scenario minus base,
so that negative values indicate lower concentrations
for the future-year scenario. These plots indicate that
for 2010 the Pre-CAAA simulation results are
characterized by increases in ozone, while the Post-
CAAA results show decreases in ozone. Similar
results were found for both future years and for both
episodes modeled (SAI, 1999). The increases occur
over the mid- and southern sections of the domain
while the decreases are more widespread. Both the
'For many of the figures in this appendix the Pre-CAAA
scenario and Post-CAAA Scenario are referred to as Pre-CAAA90
and post CAAA90, respectively.
C-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
increases and decreases are larger and more
widespread for 2010.
It is also useful to directly compare the Pre- and
Post-CAAA simulation results for each future year.
This gives a direct indication of the effects of the
CAAA on the simulated ozone concentrations. For
example, Figure C-4 illustrates the differences in
maximum simulated ozone concentration between the
Pre- and Post-CAAA simulations for 2010 for the 15
July 1995 simulated ozone episode. The differences
are calculated as Post-CAAA minus Pre-CAAA, so
that negative values indicate lower concentrations for
the Post-CAAA scenario. In general the results of
these comparisons indicate that except for isolated
increases (single grid cells), the simulated daily
maximum ozone concentrations for the Post-CAAA
scenario are lower than the corresponding Pre-CAAA
values for both future years. The spatial extent of the
decreases is greater for 2010.
Regional-Scale Modeling of the
Western U.S.
Application of the UAM-V modeling system to
the western U.S. utilized inputs from the regional-scale
application of REMSAD (as described in the next
section of this report). The objective of this
application was to provide regional-scale ozone
concentration estimates for those areas that are
neither included in the OTAG domain nor in the
urban-scale analyses. The application procedures and
modeling results are summarized in this section.
UAM-V Application Procedures for
the Western U.S.
Modeling Domain
The modeling domain used to obtain results for
the western U.S. is identical to that used for
application of the REMSAD modeling system (as
described in the following section of this report) for
the PM-related analysis of the CAAA. The modeling
domain encompasses the contiguous 48 states,
extending from 126 degrees west longitude to 66
degrees west longitude, and from 24 degrees north
latitude to 52 degrees north latitude. A grid cell size
of 2/3 longitude by 1/2 latitude (approximately 56 by
56 km) results in a 90 by 55 grid (4,950 cells) for each
vertical layer. Eight vertical layers were used. Note
that although the domain includes the entire
contiguous 48 states, results using this domain
configuration were only used to estimate ozone
concentrations for the western states.
Simulation Period
For the western U.S., the simulation period
included 1-10 July 1990. As noted earlier, this
simulation period was selected to accommodate use of
the REMSAD inputs and, therefore, represents the
summertime simulation period for PM modeling of
the western U.S. This period is characterized by high
ozone concentrations (in excess of the 1-hour ozone
NAAQS) in the Los Angeles area on all days, and in
the San Joaquin Valley on 9 and 10 July. Relatively
high concentrations were also observed in the San
Francisco Bay Area, the Sacramento Valley, the San
Diego area, and Denver. Throughout the remainder
of the domain, concentrations typically did not exceed
100 ppb. The simulation period includes three
initialization (or start-up) day that were included to
limit the influence of the initial conditions on the
simulation results.
Input Preparation
Preparation of the model-ready emission
inventories for this application utilized the same data
and followed the same procedures outlined in the
previous section of this report. Emissions totals for
the base- and future-year scenarios are provided in
Table C-2 for VOC, NOX, VOC, and CO.
The meteorological, air quality, and land-use
related inputs were identical to those used for the
application of the REMSAD modeling system to the
western U.S. The reader is referred to Section 4 of
this report for a description of these inputs and input
preparation procedures.
C-12
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-2
Emission Totals by Component for each Scenario for the Entire U.S. (tpd).
voc
Area
Onroad Mobile
Point
Utility
Total
Base 1990
33,972
18,659
9,503
96
62,229
2000 Pre-
CAAA
39,154
16,454
10,298
85
65,991
2000 Post-
CAAA
27,620
10,683
8,457
85
46,845
2010 Pre-
CAAA
43,708
18,776
11,606
134
74,224
2010 Post-
CAAA
28,575
8,804
9,454
137
46,970
NOx
Area
Onroad Mobile
Point
Utility
Total
Base 1990
13,766
20,399
7,964
20,188
62,316
2000 Pre-
CAAA
15,659
20,660
8,694
22,787
67,800
2000 Post-
CAAA
15,252
17,421
5,645
11,170
49,487
2010 Pre-
CAAA
17,697
24,142
9,803
24,808
76,450
2010 Post-
CAAA
15,794
14,696
5,985
10,319
46,793
CO
Area
Onroad Mobile
Point
Utility
Total
UAM-V Simulation
Western U.S.
Base 1990
70,069
171,181
16,478
861
258,589
Results for
2000 Pre-
CAAA
80,679
142,346
18,257
796
242,078
2000 Post-
CAAA
79,155
103,332
18,257
804
201,547
the although typical
2010 Pre-
CAAA
90,198
153,706
20,210
1,243
265,357
2010 Post-
CAAA
88,240
92,058
20,210
1,269
201,777
model performance crite
applicable for the grid resolution and do:
Model Performance
Model performance for ozone was assessed for
the entire western region and for five subregions.
Model performance was evaluated through graphical
comparison of the simulated and observed regional
and subregional maximum ozone concentration
patterns and values. Quantitative measures of model
performance were calculated on a subregional basis,
used for this analysis. Overall, the results indicate that
ozone concentrations in the western U.S. are
somewhat underestimated, relative to the observed
values. On a subregional basis, the results vary from
day to day and can be characterized as follows:
• Southern California Coast: Gross concentra-
tion gradients are directionally represented in
the simulation results, with lower values
along the coast and higher ozone inland.
However, the resolution is not sufficient to
C-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
resolve even the higher values within the Los
Angeles Basin. The maximum simulated
value on any day is 83 ppb, while the
maximum observed value exceeds 100 ppb (at
a number of the sites within the region) on
any given day of the simulation period.
Results corresponding to Los Angeles were
not used in subsequent portions of the
analysis.
• Northern California/Southern Oregon/Cent-
ral and Western Nevada: Simulated ozone
concentrations tend to be lower than
observed in the Sacramento Valley and San
Joaquin Valley as well as (toward the end of
the simulation period) the eastern portion of
the San Francisco Bay Area. Daily maximum
simulated ozone concentrations in the San
Joaquin Valley range from approximately 60
to 86 ppb. Observed values greater than or
equal to 100 ppb were recorded during 7-10
July. Representation of the observed
concentration pattern improves throughout
the simulation period. Concentrations at
monitors in Oregon and Nevada are generally
well represented. Results corresponding to
northern California were not used in the
subsequent analysis.
• Pacific Northwest/Eastern Washington: Low
observed ozone concentrations are slightly to
moderately overestimated through 7 July and
underestimated (in some cases just slightly)
for 8-10 July. For days with ozone
concentrations greater than 40 ppb, the
normalized bias ranges from approximately -
12 to 20 percent. The normalized gross error
is less than about 38 percent.
• Four Corners States: Maximum ozone
concentration in Phoenix, Las Vegas, and Salt
Lake City are reasonably well represented in
the simulation results. Concentrations for the
Denver are consistently underestimated.
Those for Albuquerque and El Paso are well
represented for certain of the days and
underestimated for others. There are also a
few isolated sites for which maximum ozone
is reasonably well simulated (the observed
concentrations are low). The normalized bias
ranges from approximately -5 to 14 percent.
The normalized gross error is less than 40
percent. Results for Phoenix were not used
in the subsequent analysis.
• Montana/Idaho/Wyoming/ Western Portion
of Dakotas: Day-to-day differences in
concentrations are not well represented,
however, the simulated values are generally
consistent with the limited observations. The
normalized bias ranges from zero to
approximately 30 percent. The normalized
error is greater than 35 percent for four of
the simulation days.
• Note on the Eastern U.S.: Maximum simula-
ted ozone concentrations range from
approximately 160 to 250 ppb during the
simulation period. Simulated peaks occur
over Baton Rouge, Houston, St. Louis, and
Atlanta with some high values along the NE
corridor. These values have not yet been
compared with observations, but simulated
ozone concentrations are much higher in the
east than in the west.
In general, the coarse resolution limits the ability
of the modeling system to resolve peak concentrations
and, in some cases, concentration gradients (such as
those that occur along the coast of California). Based
on these results, it was decided that the western ozone
modeling results could be used to characterize the
regional-scale concentration changes but would be
supplemented with higher resolution modeling for
Los Angeles, the San Francisco Bay Area (and
portions of northern California), and Phoenix.
Simulated and observed concentrations for two of
the modeling days (4 and 8 July 1990) were compared
by SAI (1999). The differences between the simulated
and observed values are typically larger than those for
the OTAG simulations (possibly due to the coarser
grid resolution) and represent both under- and
overestimation of the maximum observed
C-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
concentrations. Underestimation of the higher
concentrations is prevalent for nearly all of the
simulation days.
In addition, a comparison of mean simulated and
observed values by SAI showed that, while the highest
values are underestimated, the mean simulated values
are slightly greater than the observed means (SAI,
1999).
The model performance evaluation for the
western ozone modeling application suggests that the
modeling results can be used for the regional-scale
analysis. Although the peak concentrations tend to be
underestimated, there is not a uniform bias in the
representation of the daily maxima. In addition, the
mean values are fairly well characterized.
UAM-V Modeling Results
The UAM-V simulation results corresponding to
the Pre- and Post-CAAA scenarios for 2010 are
compared to the base-year values in Figures C-5 and
C-6, respectively. The isopleth plots depict the
differences in maximum ozone concentration for 8
July between the base (1990) simulation and the 2010
Pre- and Post-CAAA simulations, respectively. The
differences are calculated as scenario minus base, so
that negative values indicate lower concentrations for
the future-year scenario. Similar results were found
for both future years modeled (SAI, 1999) and
indicate increases in daily maximum ozone for large
portions of the western U.S. with smaller areas of
decrease (e.g., over California) for the Pre-CAAA
scenario. For the Post-CAAA scenarios, the plots
indicate widespread decreases with small areas of
increase.
A comparison of the Pre- and Post-CAAA
simulation results for 2010 is provided in Figure C-7.
The differences are calculated as Post-CAAA minus
Pre-CAAA, so that negative values indicate lower
concentrations for the Post-CAAA scenario. This
comparison indicates lower ozone concentrations for
the Post-CAAA scenario compared to the Pre-CAAA
scenario over most of the western U.S., with some
increases in the San Francisco Bay Area, Los Angeles,
and Seattle. The simulation results suggest that NOX
reductions within these areas are disbeneficial with
respect to ozone air quality. This is likely attributable
to the reduced ozone titration that occurs in the
simulation when NOX emissions are reduced. This
phenomenon is most frequently apparent in area
where NOX emissions are large relative to VOC
emissions (VOC-limited areas).
Urban-Scale Modeling of the San
Francisco Bay Area
High-resolution, urban-scale modeling of the San
Francisco Bay Area (northern California) was intended
to provide an improved basis (compared to the
regional-scale application of UAM-V for the western
U.S.) for the estimation of future-year ozone profiles
for the Bay Area and portions of northern California.
With the exception of the emission inventories, all
inputs for this application were obtained from the Bay
Area Air Quality Management District (BAAQMD),
and used by permission. The application procedures
and modeling results are summarized in this section.
UAM-V Application Procedures for
the San Francisco Bay Area
Modeling Domain
The modeling domain for this application of the
UAM-V modeling system includes the San Francisco
Bay Area, the Monterrey Bay Area, Sacramento, and
a portion of the San Joaquin Valley. The location and
geographical extent of the domain is illustrated in
Figure C-8. The domain consists of 102 by 102
horizontal grid cells with a grid spacing of 4 km. It
also includes 16 vertical layers.
Simulation Period
The simulation period for the application to
northern California is 3-6 August 1990. This episode
period occurred during the San Joaquin Valley Air
Quality Study and was characterized by moderate to
C-15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
high ozone concentrations in the San Francisco Bay
Area on 5 and 6 August, and in the Sacramento area
and the San Joaquin Valley on 4, 5, and 6 August. The
observed peak in the Bay Area was 120 ppb, while that
for the other two more inland areas reached 150 ppb.
The simulation period includes one initialization (or
start-up) day that was included to limit the influence
of the initial conditions on the simulation results.
Input Preparation
Preparation of the model-ready emission
inventories for this application utilized the same data
and followed the same procedures outlined in the
previous section of this report. Emissions totals for
the base- and future-year scenarios are provided in
Table C-3 for VOC, NOX, and CO. This table
indicates a downward trend in emissions (between
1990 and 2000) followed by an upward trend (between
2000 and 2010) for the Pre-CAAA scenario. The
increases are attributable to area- source and motor-
vehicle emissions (i.e., increases in population and
vehicle miles traveled). Emissions for both future
years are lower than the base-year for the Post-CAAA
scenario. The decreases are primarily due to a
reduction in motor-vehicle emissions.
The meteorological, air quality, and land-use
related inputs were prepared by the BAAQMD for use
in their SIP modeling analysis. Documentation of the
input preparation procedures and resulting inputs is
available on-line (BAAQMD, 1998). Initial and
boundary conditions for the future-year applications
were estimated based on the corresponding emission
reductions for VOC and NOX; for ozone the square
root of the product of the VOC and NOX reduction
factors was used.
C-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-3
Emission Totals by Component for each Scenario for the San Francisco Bay Area (tpd)
voc
Area
Onroad Mobile
Point
Utility
Total
Base 1990
783
900
111
1
1,795
2000 Pre-
CAAA
795
572
110
0
1,477
2000 Post-
CAAA
644
223
110
0
977
2010Pre-
CAAA
886
680
110
1
1,677
2010 Post-
CAAA
698
97
93
1
889
NOx
Area
Onroad Mobile
Point
Utility
Total
Base 1990
381
850
202
46
1,479
2000 Pre-
CAAA
407
827
197
4
1,435
2000 Post-
CAAA
392
545
140
4
1,081
2010Pre-
CAAA
458
1,014
197
2
1,671
2010 Post-
CAAA
396
337
140
2
874
CO
Area
Onroad Mobile
Point
Utility
Total
UAM-V Simulation
Base 1990
1,846
7,414
123
46
9,429
Results for
2000 Pre-
CAAA
2,113
5,652
119
9
7,893
the values
2000 Post-
CAAA
2,098
2,526
119
10
4,753
2010Pre-
CAAA
2,411
6,630
119
26
9,186
2010 Post-
CAAA
2,394
1,444
119
26
3,983
with a tendency for underestimation of th
San Francisco Bay Area
Model Performance
Model performance was evaluated by the
BAAQMD as part of their SIP modeling analysis and
the inputs (with the exception of the modeling
emission inventories) were used directly for the 812
prospective modeling analysis. Sscatter plots
comparing the maximum simulated and observed
ozone concentrations for both simulation periods are
available in (SAI, 1999). These comparisons indicate
good agreement between the simulated and observed
observed ozone concentrations. Mean values are
underestimated by about 10 to 15 percent on all days,
which is within the current EPA range for acceptable
urban-scale model performance (SAI, 1999).
Since good model performance is achieved,
results of the model performance evaluation for
ozone suggest that the UAM-V modeling platform for
northern California (including the meteorological, air
quality, and geographical inputs) provides an
appropriate basis for the Section 812 prospective
modeling.
C-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
UAM-V Modeling Results
Comparison of the UAM-V simulation results for
the Pre- and Post-CAAA scenarios with the base-year
values indicates both increases and decreases in the
simulated concentrations for the Bay Area , both of
which are greater in magnitude and more widespread
for the Post-CAAA scenario and for 2010. Isopleth
plots for the Bay Area are available in (SAI, 1999).
A comparison of the Pre- and Post-CAAA
simulation results for 2010 is provided in Figure C-9.
The differences are calculated as Post-CAAA minus
Pre-CAAA, so that negative values indicate lower
concentrations for the Post-CAAA scenario. This
comparison indicates that the CAAA results in higher
daily maximum ozone in the Bay Area but lower
ozone throughout the remainder of the domain.
Similar results were obtained for 2000 (SAI, 1999).
These results are qualitatively consistent with the
regional-scale modeling results presented in the
previous section of this report. However, the extent
of the increases is more limited and the decreases are
greater in the refined modeling. Note the increases
occur in areas where the base-year ozone
concentrations are low to very low.
Urban-Scale Modeling of the Los
Angeles Area
High-resolution, urban-scale modeling of the Los
Angeles area was intended to provide an improved
basis (compared to the regional-scale application of
UAM-V for the western U.S.) for the estimation of
future-year ozone profiles for this area. With the
exception of the emission inventories, all inputs for
this application were obtained from the South Coast
Air Quality Management District (SCAQMD), and
used by permission. As noted earlier, modeling of this
area was performed using the UAM modeling system.
The model formulation is similar to that for the
UAM-V modeling system, but lacks certain features
that make UAM-V suitable for regional-scale
applications. The application procedures and
modeling results are summarized in this section.
UAM Application Procedures for the
Los Angeles Area
Modeling Domain
Application of the UAM-IV for the Los Angeles
area was based on modeling performed by SCAQMD,
as reported in the 1994 Air Quality Management Plan
(SCAQMD, 1994). The modeling domain for this
application is a 65 by 40 array of 5 km resolution grid
cells. The domain also contains 5 vertical layers. The
domain encompasses the South Coast Air Basin
(SoCAB) (from Los Angeles to beyond Riverside) and
a portion of the Mojave Desert. The location and
geographical extent of the domain is illustrated in
Figure C-8.
Simulation Period
Two simulation periods were included in the
modeling analysis for Los Angeles: 23-25 June 1987
and 26-28 August 1987. Both of these episodes
occurred during the 1987 Southern California Air
Quality Study (SCAQS). In both cases, the simulation
period includes one initialization, or start-up, day (in
order to reduce the influence of the somewhat
uncertain initial concentrations on model results).
Input Preparation
Preparation of the model-ready emission
inventories for this application utilized the same data
and followed the same procedures outlined in a
previous section of this report. Emissions totals for
the base- and future-year scenarios are provided in
Table C-4 for VOC, NO,, and CO. The Post-CAAA
scenarios are characterized by lower emissions than
the base year and the Pre-CAAA scenarios. The
differences are largely attributable to changes in the
motor-vehicle emissions.
The meteorological, air quality, and land-use
related inputs were prepared by the SCAQMD for use
in their SIP modeling analysis. The reader is referred
to SCAQMD (1994 and 1996) for detailed
information on the input preparation procedures and
resulting inputs. Initial and boundary conditions for
C-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
the future-year applications were estimated based on
the corresponding emission reductions for VOC and
NOX; for ozone the square root of the product of the
VOC and NCX reduction factors was used.
Table C-4
Emission Totals by Component for each Scenario for Los Angeles (tpd)
VOC
Area
Onroad Mobile
Point
Low Level
Elevated
Base 1990
758
1,179
197
1
Total 2,135
2000 Pre-
CAAA
770
999
196
3
1,968
2000 Post-
CAAA
607
410
196
3
1,216
2010Pre-
CAAA
871
1,168
196
2
2,237
2010Post-
CAAA
700
213
158
2
1,073
NOx
Area
Onroad Mobile
Point
Low Level
Elevated
Base 1990
450
993
216
19
Total 1,678
2000 Pre-
CAAA
467
1,280
186
19
1,953
2000 Post-
CAAA
453
879
139
18
1,489
2010Pre-
CAAA
529
1,573
186
12
2,300
2010Post-
CAAA
463
626
139
8
1,236
CO
Area
Onroad Mobile
Point
Low Level
Elevated
Base 1990
1,142
9,046
208
2
Total 10,398
2000 Pre-
CAAA
1,302
10,043
197
43
11,586
2000 Post-
CAAA
1,286
5,046
197
44
6,573
2010Pre-
CAAA
1,515
11,278
197
34
13,024
2010Post-
CAAA
1,495
3,728
197
35
5,456
DAM Simulation Results for the Los
Angeles Area
Model Performance
Model performance was evaluated by SCAQMD
as part of their SIP modeling analysis and the inputs
(with the exception of the modeling emission
inventories) were used directly for the 812 prospective
modeling analysis. Comparisons of maximum
simulated and observed concentrations for each of the
simulation periods are available in (SAI, 1999). They
indicate a tendency for underestimation of the high
observed ozone concentrations. This underestimation
C-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
also shows up in the comparison of the mean values
(SAI, 1999).
While the urban-scale results are much better than
those obtained with the coarser-resolution grid, both
the maximum and mean values are underestimated.
For the primary episode days, the normalized bias
exceeds the EPA recommended range, while the
normalized gross error is within approximately 35
percent. While this does not preclude the use of these
results for the 812 study, it should be noted that the
simulated changes in ozone between the base- and
future-year scenarios may be influenced by the lack of
good model performance. Use of the simulation
results in the relative sense (through the calculation of
adjustment factors) should reduce the uncertainty,
compared to use of the absolute values.
DAM Modeling Results
Comparison of the UAM-V simulation results for
the Pre- and Post-CAAA scenarios with the base-year
values shows large reductions in daily maximum
ozone for all four future-year scenarios. Some
increases in ozone are simulated for 2010 for the Pre-
CAAA scenario; overall, the extent and magnitude of
the reductions is greater for the Post-CAAA scenario
for both years. Isopleth plots for the Los Angeles
area are available in (SAI, 1999).
A comparison of the Pre- and Post-CAAA
simulation results for 2010 is provided in Figure C-10.
The differences are calculated as Post-CAAA minus
Pre-CAAA, so that negative values indicate lower
concentrations for the Post-CAAA scenario. This
comparison indicates lower maximum ozone
concentrations under the Post-CAAA scenario for
both years. Small increases occur over the urban area;
these are smaller in extent and magnitude than for the
regional modeling application. Similar results were
found for 2000 (SAI, 1999).
Urban-Scale Modeling of the Maricopa
County (Phoenix) Area
High-resolution, urban-scale modeling of
Maricopa County, Arizona (which includes the
Phoenix urban area) was intended to provide an
improved basis (compared to the regional-scale
application of UAM-V for the western U.S.) for the
estimation of future-year ozone profiles for this area.
With the exception of the emission inventories, all
inputs for this application were obtained from the
Maricopa Association of Governments (MAG), and
used by permission. As noted earlier, modeling of this
area was performed using the UAM modeling system.
DAM Application Procedures for the
Phoenix Area
Modeling Domain
The modeling domain for the application of the
UAM modeling system to the Phoenix area
encompasses the urbanized portion of Maricopa
County, Arizona; this domain was based on that used
for a previous application of UAM for the area
(Douglas et al., 1994). The domain consists of a 44 by
33 array of 2 km grid cells and 5 vertical layers. The
location and geographical extent of the domain is
illustrated in Figure C-8.
Simulation Period
Two ozone episodes were also simulated for the
Phoenix area: 9-10 August 1992 and 13-14 June 1993.
Exceedances of the 1-hour NAAQS for ozone were
recorded during both episodes. Each period also
includes one initialization day.
Input Preparation
Preparation of the model-ready emission
inventories for this application utilized the same data
and followed the same procedures outlined in a
previous section of this report. Emissions totals for
the base- and future-year scenarios are provided in
C-20
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-5 for VOC, NOX, and CO. The changes in
emissions are characterized by both increases and
decreases, reflecting an expected growth in population
that is offset by fleet turnover and other emission
reduction measures.
The meteorological, air quality, and land-use
related inputs were prepared by Douglas et al. (1994).
The reader is referred to this technical report for
detailed information on the input preparation
procedures and resulting inputs. Initial and boundary
conditions for the future-year applications were
estimated based on the corresponding emission
reductions for VOC and NOX; for ozone the square
root of the product of the VOC and NOX reduction
factors was used.
DAM Simulation Results for the
Phoenix Area
Model Performance
Model performance was evaluated by Maricopa
Association of Governments (MAG) as part of their
SIP modeling analysis and the inputs (with the
exception of the modeling emission inventories) were
used directly for the 812 prospective modeling
analysis. Comparison of hourly simulated and
observed ozone concentrations indicates good
agreement between the simulated and observed values.
Mean values are well represented as well. Plots of
these comparisons are available in (SAI, 1999). For
the primary modeling days, the normalized bias and
error statistics indicate very good model performance.
The values are less than 5 percent (bias) and 20
percent (error) respectively (SAI, 1999).
The model performance results indicate that the
UAM modeling system (including the meteorological,
air quality, and land-use input) are appropriate for use
in the Section 812 prospective analysis.
UAM Modeling Results
Comparison of the UAM-V simulation results for
the Pre- and Post-CAAA scenarios with the base-year
values indicates both increases and decreases for the
Pre-CAAA scenario simulations and large decreases
for the Post-CAAA scenario simulations. Isopleth
plots for the Phoenix area are available in (SAI, 1999).
A comparison of the Pre- and Post-CAAA
simulation results for each future year indicates that
the CAAA measures reduce daily maximum ozone
concentrations within the Phoenix domain by
approximately 10 to 20 ppb (more or less in some
areas) for both future years. Isopleth plots for the
Phoenix area are available in (SAI, 1999).
C-21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-5
Emission Totals
by Component for each Scenario for Phoenix (tpd)
voc
Area
Onroad Mobile
Point
Low Level
Elevated
Total
Base 1990
241
184
2
0
426
2000 Pre-
CAAA
310
215
2
0
527
2000 Post-
CAAA
201
141
1
0
344
2010Pre-
CAAA
369
231
2
0
602
2010Post-
CAAA
250
85
2
0
337
NOx
Area
Onroad Mobile
Point
Low Level
Elevated
Total
Base 1990
213
151
1
0
371
2000 Pre-
CAAA
270
214
1
0
485
2000 Post-
CAAA
259
174
1
0
434
2010Pre-
CAAA
326
264
2
0
592
2010Post-
CAAA
290
147
2
0
438
CO
Area
Onroad Mobile
Point
Low Level
Elevated
Total
Base 1990
684
1,186
0
1
1,871
2000 Pre-
CAAA
820
2,001
0
1
2,822
2000 Post-
CAAA
795
1,538
0
1
2,334
2010Pre-
CAAA
941
1,883
0
1
2,825
2010Post-
CAAA
907
1,002
0
1
^^^Qlt^
C-22
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Calculation of Ozone Air Quality
Profiles
The overall objective of the photochemical
modeling exercise was to provide estimates of future-
year ozone air quality for 2000 and 2010 (for
assessment of the effects of the CAAA). This was
accomplished using an approach that combines
observed data and air quality modeling results to
estimate future-year concentrations. The method-
ology is designed to provide site-specific, seasonal and
annual ozone concentration distributions. The
statistical concentration distributions are estimated
(based on the results of air quality modeling) for
specific future-year scenarios and, in turn, provide the
basis for examination and quantification of the effects
of changes in air quality on health, agriculture, etc.
(i.e., physical effects and economic valuation
modeling). Through comparison with corresponding
results for a baseline simulation (in this case without
the CAAA measures and programs), the effects of the
CAAA can be assessed. The future-year air quality
profile estimation methodology, as applied to the
analysis of the CAAA, is described in this section.
Overview of the Methodology
Conceptually, the methodology for estimating
future-year ozone air quality using both observations
and UAM-V simulation results is rather simple. The
UAM-V simulation results are used to calculate
adjustment factors for selected ozone monitoring sites
within the modeling domain. This is done on a grid-
cell by grid-cell basis (i.e., the adjustment factor for a
monitoring site is based on the simulation ozone
concentrations for the grid cell in which it is located).
The adjustment factor represents the ratio of the
future-year-scenario to the base-year concentrations
and is calculated (using the appropriately matched
values) for several different concentration levels (i.e.,
the changes in concentration are dependent upon
relative concentration level). The observed ozone
concentrations for each monitoring site are then
modified using the site-specific (or grid-cell-specific)
adjustment factors. The resulting values represent the
estimated future-year ozone concentrations for the
modeled scenario.
As noted earlier in this report, the overall
approach to estimate future air quality differs from
that for a typical air quality model application (e.g., for
attainment demonstration purposes) in that the
modeling results are used in a relative sense, rather
than an absolute sense. This may enhance the
reliability of the future-year concentration estimates,
especially in the event that the uncertainty inherent in
the absolute concentration values is greater than that
associated with the response of the modeling system
to changes in emissions. This approach also permits
the estimation of seasonal and annual concentration
distributions, a requirement for this study.
Although the ratios are calculated using modeling
results for a limited number of simulation days, it is
assumed, using this methodology, that the ratios can
be used to represent longer time periods.
Consequently, all observations contained within the
dataset are adjusted using the model-derived ratios.
Following adjustment of the observed data, statistical
quantities, or "profiles", describing the ozone
distribution for each monitoring site are then
calculated.
Description of the Observation Dataset
One of the unique aspects of this approach to
evaluating future ozone air quality is the use of
observed ozone concentrations to supplement model
results. As such, one of the earliest tasks was the
creation of a dataset containing the observed hourly
ozone concentrations for all monitoring sites located
within the modeling domains for the months of May
through September 1990.
Hourly ozone concentrations for 1990 were
extracted from the Aerometric Information Retrieval
System (AIRS) and input into a single AMP350-
format datafile. From the information contained in
this file, two SAS datasets were created: a
concentration dataset and a monitor information
dataset). The concentration dataset contains the
C-23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
hourly concentrations for each monitor, with each
record in the dataset representing a single monitor-
day. The monitor information dataset contains
monitor-specific information such as land-use and
location.
In creating the concentration dataset, some data
handling issues arose and were addressed in the
following manner:
• In some instances, multiple ozone monitoring
devices were operated at the same location.
Even though these different devices have the
same AIRS state-county-site identification
code (ID), they are differentiated by a
parameter occurrence code (POC). The
AIRS state-county-site ID was concatenated
with the POC to form a unique identifier for
each monitor. A POC greater than 5 typically
indicates that a device was being calibrated;
information/data for these monitors
was/were not included in either the monitor
or the concentration dataset.
• In the AIRS database, ozone concentrations
are reported using the default unit of the
reporting agency. Thus, multiple units were
present in the AMP350 file. For ease of
analysis, all of the concentrations were
converted to a single unit, ppm.
• Missing ozone concentrations in the AIRS
AMP350 report are indicated by a blank in
the decimal field. In the concentration
dataset for this study, the SAS missing value
code was used to indicate missing data.
• For each monitor a method detection limit
(MDL) was provided. The MDL indicates a
threshold below which reported ozone
concentrations do not accurately reflect the
sample distribution. For most monitors the
MDL is 0.005 ppm. Because this value is low
relative to typical ambient concentration
levels, observed values below the MDL were
not reset to the MDL and instead were left
unchanged.
Only monitors with "complete" data were used in
the analysis. For the ozone data, a monitor record
was considered to be complete if data were available
for 50 percent of the days in the peak ozone season
(May-September). Each of these days in turn had to
have at least 12 hourly observations. There were 842
ozone monitors with complete data.
Calculation of Percentile-Based
Adjustment Factors
For each future-year modeling scenario, grid-cell-
specific adjustment factors were calculated using the
hourly simulated ozone concentrations contained in
the UAM-V or UAM xymap2 output files. Individual
monitoring sites were mapped onto the gridded model
output (to determine the grid cell in which each
monitor was located) and the concentrations for the
corresponding grid cells were used to calculate a set of
ten adjustment factors for each future-year modeling
scenario. The adjustment factors were specified to be
the ratio of the percentile concentrations for the
future- and base-year simulations, where the percentile
concentrations were calculated using data for each
hour of each simulation day:
Adjustment Factor, =—
x:thPercentile Concentration
future year
x, th Percentile Concentrat ionbase year
{xi}= {5, 15, 25, 35, 45, 55, 65, 75, 85, 95}
For calculation of the percentile concentrations,
the empirical distribution function with averaging was
employed. Because the concentrations for the lower
percentiles can be rather small, a threshold value of 1
ppb was set to keep the adjustment factors reasonable.
In other words, all concentrations below 1 ppb were
reset to 1 ppb. This percentile-based approach was
selected due to the limitations of using a single
adjustment to represent the change in the modeled
ozone concentrations in moving from the base- to the
future-year scenarios. Finally, if either the base-year
The UAM-V xymap file contains hourly, gridded, surface-
layer ozone concentrations.
C-24
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
or future year percentile concentration was set equal
to 1 ppb, the adjustment factor was set equal to 1.
A SAS dataset containing the monitor-level
adjustment factors was created for each future-year
modeling scenario considered in this study.
Use of Adjustment Factors to Modify
Observed Concentrations
Using the calculated adjustment factors for each
future-year scenario and the observed monitor- and
pseudo-site-level observations, a dataset containing
modified observed hourly ozone concentrations for
each of the two scenarios was created. Because each
monitor has ten adjustment factors per scenario, it
was first necessary to rank order the observed
concentrations into 10 decile-based groups with ties
being assigned to the higher group. Once each of the
observed concentrations was identified with a
particular decile group, the appropriate adjustment
factor was selected and applied:
AdjustedCom, — ObsCom, * Adj.Factork[ollsCma]
In this equation, {ObsConci} is the set of
observed hourly ozone concentrations (in ppm) for a
given monitor or pseudo-site. The k[ObsConc] is the
number of the decile group to which ObsConc,belongs.
AdjFactorkpbsCmdj is then the appropriate adjustment
factor for the decile group to which ObsConc, belongs.
The resulting set of adjusted hourly concentrations,
{AdjustedCom^, represents the future-year estimates of
the hourly ozone concentrations.
Calculation of Ozone Profiles
Datasets containing the ozone air quality
"profiles" were compiled for the base 1990, 2000 Pre-
CAAA, 2000 Post-CAAA, 2010 Pre-CAAA, and 2010
Post-CAAA simulations. The profiles used data for
the period May through September. The databases
contained the number, the arithmetic mean, the
median, the (seasonal) second highest, and the 2.5 to
97.5 percentiles (in increments of five) of the hourly
concentrations. The profiles are reported at the
monitor level and include 842 locations.
The histograms in Figures C-lla through 12b
illustrate the distribution of ratios for the 95th
percentile monitor-level ozone concentrations
corresponding to the 2000 and 2010 simulations,
respectively. In this figure, ratios greater than one
indicate that the future-year/scenario concentration is
greater than the base-year (1990) value, whereas ratios
less than one indicate a lower value for the future-
year.
The 2000 Pre-CAAA ratios (Figure C-lla)
indicate that the 95th percentile ozone concentrations
corresponding to this scenario are higher in some
areas and lower in other areas than the base-year
(1990) values. The ratios generally range from
approximately 0.8 to 1.2, but also include some lower
values. In contrast, the ratios corresponding to the
2000 Post-CAAA simulation (Figure C-llb) are
generally less than one. In this case, the ratios range
from approximately 0.75 to 1.1 with only a very small
number of values greater than one. There are also
some lower values.
Figure C-12a and 12b displays the distribution of
ratios of the future-year-scenario to base-year 95th
percentile concentrations for 2010. Compared to the
histogram plots for 2000, the shift in distribution is
such that the ratios are higher for the Pre-CAAA
scenario and lower for the Post-CAAA scenario. That
is, compared to 2000, concentrations for 2010 are
higher relative to the base year under the Pre-CAAA
scenario and lower relative to the base year under the
Post-CAAA scenario.
For both future years, the ratios indicate that the
Post-CAAA concentrations (95th percentile level) are
lower than the corresponding Pre-CAAA values (with
a few exceptions). This is illustrated in Figures C-13a
and C- 13b. The smaller ratios for 2010 (Figure C-
13b) reflect the larger differences between the Pre-
and Post-CAAA scenarios for this year.
C-25
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 Ozone (ppb)
Time: 0-2400 July 15, 1995
-94
-89
Deg. Longitude
-84 -79
+ MAXIMUM = 31.8 ppb
- MINIMUM = -23.0 ppb
-74
-69
,1111111 i | i i i I I I: i: i:: If I I fcj; :f t t r» j > IM
10
20
30
40
50
60
1-026
Figure C-2: Difference in daily maximum simulated ozone concentration (ppb) for
the 15 July 1995 OTAG episode day: 2010 pre-CAAA90 minus base 1990.
C-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 Ozone (ppb)
Time: 0-2400 July 15, 1995
-99
-94
-89
Deg. Longitude
-84 -79
+ MAXIMUM =14.1 ppb
- MINIMUM = -47.6 ppb
-74
-69
Figure C-3. Difference in daily maximum simulated ozone concentration (ppb)
for the 15 July 1995 OTAG episode day: 2010 post-CAAA90 minus base 1990.
C-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 Ozone (ppb)
Time: 0-2400 July 15, 1995
-99
-94
-89
Deg. Longitude
-84 -79
+ MAXIMUM = 11.8 ppt>
- MINIMUM = -64.0 ppb
-69
Figure C-4. Difference in daily maximum simulated ozone concentration (ppb)
for the 15 July 1995 OTAG episode day: 2010 post-CAAA90 minus pre-CAAA90.
C-28
-------
LEVEL 1 Ozone (ppb)
Time 0-2400 July 8, 1990
-126
-118
-110
Deg. Longitude
102 -94
+ MAXIMUM - 29 6 ppb
MINIMUM - -23 -1 ppb
-86
-78
-70
I I I I I I ! I I I I I I I I I I I I I I I I | I I I I I I I I i I I | I I I I I I I I I I I {! I I I I I I I I I I | I I I I I I I I I I I I I I I I I I I I I I I I I I I I l_
Figure C-5. Difference in daily maximum simulated ozone concentration (ppb)
for the 8 July 1990 western U.S. simulation day: 2010 pre-CAAA90 minus base 1990.
C-29
-------
LEVEL 1 Ozone (ppb)
Time: 0-2400 July 8, 1990
-126
-118
-110
Deg. Longitude
102 -94
+ MAXIMUM
- MINIMUM
-86
-78
9_2 ppb
-35.3 ppb
-70
1 49
.'•"''•, 7 v' i— jL
-------
LEVEL 1 Ozorio (ppb)
Tune 0-2400 July 8 1990
126
-118
-110
Deg. Longitude
102 -94
4 MAXIMUM VA 7 ppb
-- MINIMUM -•!•'! 5 ppb
-86
-78
-70
10-
nn i i * i i i i i i i i i i i i i i i i i i i i i i i i ili i i i i I i i i i i i i i i i i i i I i i i i i i i i i I i j_i j_i i i i i I i i t i i M i i i i i i rl 24
0 10 20 30 40 50 60 70 80 9(7
Figure C-7. Difference in daily maximum simulated ozone concentration (ppb)
for the 8 July 1990 western U.S. simulation day: 2010 post-CAAA90 minus pre-CAAA90.
C-31
-------
-126
-118
-110
Deg. Longitude
-102
-94
-86
-78
-70
50
40
30
20
10
: I: I:':;lsisilfiTil' I]liyMiiiiiil a I.:'I.;!• j ;1 j I |jirl:.:;l\n;;i;; 1: I; I,. | :;l':.
49
44
39 jg
0}
o>
Q
34
29
n r°( i I I i ii i i I I i i ii iii I i n i ill ii I i i i i i i i i i I i li i I I i i i I ii i i i ii i I I i i i I ii i ill i i i i i i i i i I i i i i i I I I il 2 4
0 10 20 30 40 50 60 70 80 9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 Ozone (ppb)
Time: 0-1800 August 6, 1990
+ MAXIMUM = 24.6 ppb
- MINIMUM = -16.5 ppb
. _ _
A vfvf •
IIIIMii1lLlirillJlklllMI>lhMllllllllJill!l,llllkLII
-------
LEVEL 1 Ozone (ppb)
Time: 0-2400 August 28, 1987
+ MAXIMUM = 7.4 ppb
- MINIMUM = -57.6 ppb
rrnTTTTTTTTTTTnTTTTTTTTTTTT^ r~1 '1 ft TTTf TT~r
10 -
,9 '
/ <;:: .-ecr'1
t !l I I I I I 1 J_LJ_L ' I: L::i I -» ' » ' « ' ' ' '
' &
- at-.,,'
.?
^
..
o
?0-
JLL4 4-JJUl-L-t-L-L-l-l i \ J-LLlJLLJ-L. \1
40 50 60
Figure C-10. Difference in daily maximum simulated ozone concentration (ppb)
for the 28 July 1987 simulated day for Los Angeles: 2010 post-CAAA90 minus pre-CAAA90.
C-34
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-11a. Distribution of Monitor-Level Ratios
for 95th Percentiles Ozone Concentration:
2000 Pre-CAAA /1990 Base-Year
^ 60
I
o>
o
c
d>
d>
^c
re
50 -
40 -
30 -
20 -
10 -
0
median: 1.013
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-11b. Distribution of Monitor-Level Ratios
for 95th Percentiles Ozone Concentration:
2000 Post-CAAA/1990 Base-Year
-- 60
e 50-
0)
a.
> 40 H
o
c
§ 30-
£ 20-
Qi
| 10 H
"55
* 0
median: 0.910
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-35
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-12a. Distribution of Monitor-Level Ratios
for 95th Percentiles Ozone Concentration:
2010 Pre-CAAA/1990 Base-Year
0)
e 50-
40-
30-
it 20-
o>
a>
0
median: 1.051
ft*
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-12b. Distribution of Monitor-Level Ratios
for 95th Percentile Ozone Concentration:
2010 Post-CAAA/1990 Base-Year
0)
£ 50 H
o>
a.
> 40 H
o
§ 30 H
o-
0)
£ 20-
I 10 H
0
r-t-rn
median: 0.883
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-36
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-13a. Distribution of Monitor-Level Ratios
for 95th Percentiles Ozone Concentration:
2000 Post-CAAA / 2000 Pre-CAAA
c- 60
"c
S 50 -
o>
Q.
O
C
d>
30 -
20 -\
o>
| 10 -\
"55
median: 0.900
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-13b. Distribution of Monitor-Level Ratios
for 95th Percentiles Ozone Concentration:
2010 Post-CAAA / 2010 Pre-CAAA
^ 60
8 50-
o>
Q.
O
C
d>
d>
30-
20 -
median: 0.838
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-37
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Estimating the Effects of the CAAA
on Participate Matter
Future-year concentrations of PM10 and PM25
corresponding to the Post-CAAA and Pre-CAAA
scenarios were estimated through application of the
RADM/RPM and REMSAD modeling systems. The
former was used for the eastern U.S., while the latter
was applied for the western U.S. Details of both
RADM/RPM and REMSAD modeling are presented
in this section. Included is an overview of each
modeling system, and a description of the application
procedures and modeling results. The calculation of
PM air quality profiles using the combined modeling
results from both models is also described.
For ease of reading, all figures follow the text in
this section.
Overview of the RADM/RPM Modeling
System
RADM was developed over a ten-year period,
1984-1993, under the auspices of the National Acid
Precipitation Assessment Program (NAPAP) to help
address policy and technical issues associated with acid
deposition. More recently, EPA created the Regional
Particulate Model, expanding the Agency's
atmospheric modeling capabilities. Functioning
together, RADM and RPM help predict PM
concentrations by generating estimates of secondary
paticulates that comprise a significant portion of total
PM.
RADM, a three-dimensional Eulerian grid-based
model, is designed to provide a scientific basis for
predicting regional air pollution concentrations and
levels of acid deposition resulting from changes in
precursor emissions. The concentration of a specific
pollutant in a grid cell at a specified time is determined
by the following factors:
• the emissions rate;
• the transport of that species by wind into and
out of the grid in three dimensions;
• movement of the atmosphere via turbulent
motion;
• chemical reactions that either produce or
deplete the chemical;
• the change in concentration due to vertical
transport by clouds;
• aqueous chemical transformation and
scavenging; and
• removal by deposition.3
RPM is an extension of RADM. Like RADM,
RPM is a three-dimensional Eulerian air quality
model. Functioning in tandem with RADM, RPM
predicts the chemistry, transport, and dynamics of the
secondary aerosols of sulfate, nitrate, ammonium, and
organics.4 For this study, however, RPM organic
aerosol estimates were not included in the final
analysis because the model significantly
underestimates organics and the reason for this
systematic underestimation has not yet been
characterized. The model's predictions of secondary
sulfate, nitrate, and ammonium concentrations were
used to develop particulate matter concentration
estimates.
Application of RADM/RPM for the
Eastern U.S.
In this analysis, the RADM/RPM modeling
system was used to estimate future year nitrate and
sulfate concentrations, two major components of
secondary PM. These model results were then used to
generate adjustment factors, which in turn aided
development of PM predictions for the eastern half
of the United States. A summary of the model's
application and results follows.
Modeling Domain
The domain of application for both RADM and
RPM is eastern North America, from the Rocky
A more detailed description of RADM is provided in R.
Dennis, 1995.
4A more detailed description of the structure and basic
features of RPM is given in F.S. Binkowski and U. Shankar, 1995.
C-38
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Mountains eastward to Newfoundland, Canada and
the Florida Keys. This expansive model area that
includes part of Southern Canada allows
RADM/RPM to accurately reflect the several-day
residence times of sulfur and nitrogen species in the
atmosphere and the resulting transport distances of
1,000 kilometers (km) or more that may be covered
during that time. The 2,800 by 3,040 km model
domain is divided into 80-km grid cells. Nested
within this domain are a set of finer resolution 20-km
grid cells, covering the geographic region extending
eastward from central Illinois to the Atlantic Ocean
and southward from Sudbury, Ontario to the upper
section of North Carolina (Figure C-14). The model
also consists of vertical layers that, in total, stretch 16
km above ground level.
Simulation Periods
RADM/RPM model runs were conducted for 30
five-day periods. The 30 periods, which represent
dominant transport regimes spanning four years, were
randomly selected to develop annual averages. Later,
to develop warm season (May through September)
and cold season (October through April) averages,
they were divided into these two seasonal groups.
Annual warm and cold season averages were
developed using a weighting scheme based on the
frequency of occurrence of transport regimes. To
avoid the influence of the model starting up and
adjusting to a new set of conditions associated with
each period, only results from the last three days of
each period were used to estimate PM levels.
Model Inputs
RADM
Detailed emissions and meteorological data are
required to run RADM. The emissions inventory for
the model must account for both the timing and
location of emissions. Accurate model predictions
also depend on a host of meteorological inputs, most
notably temperature, wind speed, and wind direction.
Separate emissions inventories were used as input
in this analysis for each of the emissions scenarios:
1990 base year, 2000 Pre-CAAA, 2000 Post-CAAA,
2010 Pre-CAAA, and 2010 Post-CAAA.5 These
scenarios and their accompanying inventories,
described in more detail in Appendix A, incorporate
emissions from all five major source categories:
industrial point sources, utilities, nonroad
engines/vehicles, motor vehicles, and area sources.
This inventory for each scenario contains hourly, day-
specific emissions figures for every source category;
area and mobile source data are provided at the
county level, while utility and industrial point source
emissions are given at the source classification code
level.
Biogenic emissions were also included in the
RADM input. This inventory was developed from
version two of EPA's Biogenic Emissions Inventory
System (BIES-2). BEIS-2 estimates biogenic
emissions based on a variety of factors including
biomass and emissions factors.
The meteorological inputs for RADM were
derived using output from the Pennsylvania State
University/National Center for Atmospheric Research
(PSU/NCAR) mesoscale model (MM4). Using MM4
results, EPA generated essential grid-specific RADM
input, including wind flow patterns, temperatures, and
water vapor concentrations.
RPM
RPM requires inputs similar to those described
for RADM. This model uses a subset of RADM
emissions data and the RADM meteorological fields.
Additional RPM inputs include atmospheric water
data generated by RADM and RADM-predicted levels
of oxidants, nitric acid, and ammonia.
5See Pechan, June 1998 for a detailed description of the
emissions scenarios developed for this analysis.
C-39
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
RADM/RPM Simulation Results
Model Performance
The assessment of model performance for
particulate models is a difficult task due to a relative
lack of data and information regarding the spatial
distribution, composition, and size fractionation of
airborne particulates. Development and evaluation of
particulate measurement and modeling techniques are
active areas of research. As a result, there are
currently no standard approaches or model
performance criteria for the evaluation of regional-
scale particulate models.
Development of RADM began in the mid-1980's.
The evolution of this model, along with its application
and performance evaluation have all been documented
extensively by NAPAP.6 RADM continues to
undergo periodic peer review, evaluations, and
improvements.7 In addition to the present study and
the section 812 retrospective analysis, RADM has
been used in other Agency studies of acid deposition8
and in assessments of deposition of nitrogen to
coastal estuaries.9
RPM was evaluated by comparing the model's
1990 base year seasonal nitrate and sulfate estimates
with observed data measured by EPA's Clean Air Act
Status and Trends Network (CASTNet). CASTNet is
a network of monitors distributed throughout the
Eastern U.S. that measures dry deposition of
atmospheric sulfur and nitrogen compounds. RPM
predictions for particulate sulfate and CASTNet data
are provided in Table C-6. Examination of these
ambient concentrations shows that RPM predicts the
significant seasonal differences in sulfate production,
although the model overestimates the annual average
sulfate concentration by approximately 20 percent.
Table C-7 displays RPM and CASTNet seasonal
average nitrate concentrations and ratios showing the
fraction of total nitrate that is in particulate form.
Comparison of the values in this table indicate that
RPM accurately captures the ratio of particulate to
total nitrate, but underestimates overall nitrate levels
in the colder months and overestimates them during
the warmer months. Averaged over the entire year,
however, RPM results and CASTNet data are similar.
Table C-6
Comparison of CASTNet and RPM
Average Concentration of SO4
Season
CASTNet
S04
(M9/m3)
RPM
S04
Warm
7.8
9.1
Cold
3.7
3.6
Annual
5.4
6.6
6Chang, J. et al. 1987, Chang,], et al. 1990, and Dennis, R. et
al. 1990.
7Dennis, R. et al. 1993, McHenry J. and Dennis, R. 1994, and
External Review Panel 1994.
8U.S. EPA, 1995.
9Dennis, R. 1997 and EPA 1997.
C-40
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-7
Comparison of CASTNet and RPM
Average Concentrations and Fractions of NO3
Season
Autumn
Winter
Spring
Summer
Annual
CASTNet
NO3
(,ug/m3)
1.39
1.67
0.85
0.42
1.06
RPM
N03
(Mg/m3)
1.25
1.01
1.07
0.57
1.06
CASTNet
NO3/t-NO3
(ratio)
0.42
0.44
0.24
0.14
0.31
RPM
NO3/t-NO3
(ratio)
0.42
0.44
0.24
0.10
0.27
RADM/RPM Modeling Results
RADM/RPM generated estimates of nitrate and
sulfate concentrations for the years 2000 and 2010
under both the Pre- and Post-CAAA scenarios. These
two constituents are major components of secondary
PM. As described in more detail later in this
appendix, these RADM/RPM results were used to
project 1990 observed nitrate and sulfate
concentrations to future year levels. From these
future year estimates, monitor-level PM10 and PM25
concentrations were calculated for 2000 and 2010.
Comparison of 1990 base year PM levels with
future year Pre- and Post-CAAA estimates shows that
under the Pre-CAAA scenario, concentrations of PM10
and PM25 are generally expected to increase from base
year levels. Under the Post-CAAA scenario, both
PM10 and PM25 concentrations are predicted to
decrease throughout much of the U.S. in both 2000
and 2010, with greater decreases expected in 2010.
The histograms in Figures C-21 through C-24 show
the relationship between base year and future year PM
estimates. In these figures, ratios greater than one
indicate that the future year concentration is greater
than the 1990 base year value, while ratios less than
one indicate a lower value for the future. Figures C-25
and C-26 show the relationship between Pre- and
Post-CAAA PM estimates. All of these histograms
present data for the entire U.S., including
RADM/RPM data for the East and REMSAD data
for the West (see below).
Overview of the REMSAD Modeling
System
The Regulatory Modeling System for Aerosols
and Deposition (REMSAD) programs have been
developed to support a better understanding of the
distributions, sources, and removal processes relevant
to fine particles and other airborne pollutants,
including soluble acidic components and toxics.
Consideration of the different processes that affect
primary and secondary (i.e., formed by atmospheric
processes) particulate matter at the regional scale in
different places is fundamental to advancing this
understanding and to assessing the effects of
proposed pollution control measures. These same
control measures will, in most cases, affect ozone,
particulate matter and deposition of pollutants to the
surface.
The REMSAD system was initially focused on
atmospheric aerosols and the deposition of toxic
pollutants such as mercury from the air to the surface.
Any modeling system for aerosols and deposition
must be built on the foundation of an atmospheric
transport and dispersion model. Many atmospheric
dispersion models have been developed since the late
1970s and applied for various purposes. Urban and
regional air quality models are generally based on the
Eulerian approach. The REMSAD system is built on
the foundation of the UAM-V regional air quality
model, which includes a number of advantageous
capabilities. The REMSAD aerosol and toxics
C-41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
deposition module (ATDM) is capable of "nesting" a
finer-scale subgrid within a coarser overall grid, which
permits high resolution over receptor regions without
an intolerable computing burden. The modeling
system may thus be applied at scales ranging from a
single metropolitan region to a continent containing
multiple urban areas.
The REMSAD system consists of a
meteorological data preprocessor, the core aerosol and
toxic deposition model (ATDM), and postprocessing
programs. The ATDM is a three-dimensional grid
model designed to calculate the concentrations of
both inert and chemically reactive pollutants by
simulating the physical and chemical processes in the
atmosphere that affect pollutant concentrations. The
basis for the model is the atmospheric diffusion or
species continuity equation. This equation represents
a mass balance in which all of the relevant emissions,
transport, diffusion, chemical reactions, and removal
processes are expressed in mathematical terms. The
model is usually exercised over a multi-day period,
typically a full year.
Fine particles (or aerosols) are currently thought
to pose one of the greatest problems for human health
impacts from air pollution. The major factors that
affect aerosol air quality include:
• The spatial and temporal distribution of toxic
and particulate emissions including sulfur
dioxide (SO2), oxides of nitrogen (NO).,
volatile organic compounds (VOC), and
ammonium (NH3) (both anthropogenic and
nonanthropogenic),
• The size composition of the emitted PM,
• The spatial and temporal variations in the
wind fields,
• The dynamics of the boundary layer,
including stability and the level of mixing,
• The chemical reactions involving PM, SO2,
NOX and other important precursor species,
• The diurnal variations of solar insulation and
temperature,
• The loss of primary and secondary aerosols
and toxics by dry and wet deposition, and
• The ambient air quality immediately upwind
and above the region of study.
The ATDM module simulates these processes
when it is used to simulate aerosol distribution and
toxic deposition. The model solves the species
continuity equation using the method of fractional
steps, in which the individual terms in the equation are
solved separately in the following order: emissions are
injected; horizontal advection/diffusion is solved;
vertical advection/diffusion and deposition is solved;
and chemical transformations are performed for
reactive pollutants. The model performs this four-
step solution procedure during one half of each
advective (driving) time step, and then reverses the
order for the following half time step. The maximum
advective time step for stability is a function of the
grid size and the maximum wind velocity or horizontal
diffusion coefficient. Vertical diffusion is solved on
fractions of the advective time step to keep their
individual numerical schemes stable. A typical driving
time step for coarse (50-80 km) grid spacing is 10-15
minutes, whereas time steps for fine grid spacing
(10—30 km) are on the order of a few minutes.
Model inputs are prepared for meteorological and
emissions data for the simulation days. Once the
model results have been evaluated and determined to
perform within prescribed levels, a projected emission
inventory can be used to simulate possible policy-
driven emission scenarios.
REMSAD provides gridded averaged surface and
multi-layer instantaneous concentrations, and surface
deposition output for all species and grids simulated.
The averaged surface concentrations and depositions
are intended for comparison with measurements and
ambient standards. The instantaneous concentration
output is primarily used to restart the model, and to
examine model results in the upper levels.
Concentrations of particulates are passed as input to
a module that estimates atmospheric visibility. Wet
and dry acidic deposition fluxes are calculated hourly
and may be accumulated for any desired interval.
The particulate matter species modeled by
REMSAD include a primary coarse fraction
C-42
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
(corresponding to participates in the 2.5 to 10 micron
size range), a primary fine fraction (corresponding to
participates less than 2.5 microns in diameter), and
several secondary particulates (e.g., sulfates, nitrates,
and organics). The sum of the primary fine fraction
and all of the secondaries is taken to be roughly
representative of PM25. Table C-8 lists the simulated
species written to the REMSAD output files.
A number of issues are particularly important to
a successful application of REMSAD for evaluating
the atmospheric transport and deposition of
pollutants. These include the meteorology, accuracy
and representativeness of the emission inventory,
resolution, structure and extent of the modeling grid,
and the treatment of urban areas in both the source
and receptor areas of the computational grid.
Accurate representation of the input meteorological
fields is necessary both spatially and temporally in
order to adequately capture the complex effects of
terrain on airflow and hence transport and deposition
of pollutants. In addition the meteorology must be
sufficiently resolved in order for the model to
accurately diagnose the appropriate cloud
characteristics required by the various
parameterizations of the cloud processes treated by
the model. The required input fields include
temporally varying three dimensional gridded wind
fields, temperature, humidity and vertical exchange
coefficients in addition to the surface pressure and
precipitation rates.
Version 4.0 of the REMSAD modeling system
(with simplified ozone chemistry) was employed for
this study.
Table C-8
REMSAD Output File Species
REMSAD Species1 Gas/Aerosol
Description
NO
Nitric oxide
NO9
Nitrogen dioxide
SO,
Sulfur dioxide
CO
Carbon monoxide
NH,
Ammonia
VOC
Volatile organic compounds
HNO,
Nitric acid
PNO,
Particulate nitrate
GSO4
Particulate sulfate (gas phase production)
ASO4
Particulate sulfate (aqueous phase production)
NH4N
Ammonium nitrate
NH4S
Ammonium sulfate
SOA
Secondary organic aerosols
POA
Primary organic aerosols
PEC
Primary elemental carbon
Pmfine
Primary fine PM (<2.5 microns)
Pmcoarse
Primary coarse PM (2.5 to 10 microns)
Sulfate=GSO4+ASO4+NH4S
Nitrate=PNO3+NH4N
Total PM2.5 surrogate=sulfate+nitrate+SOA+POA +Pmfine
1 These are names used in the model and, for the aerosols, are not necessarily the correct molecular formula (the
integers are subscripted only when the formula correctly reflects the species).
2 Note that (for consistency with the REMSAD User's Guide) we are using the terminology "coarse PM" to mean PM in
the size range of 2.5 to 10 microns, which is not in agreement with general use, which defines coarse PM to be
particles with size greater than 2.5 microns.
C-43
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Application ofREMSAD for the
Western U.S.
For this study, the REMSAD modeling system
was applied for the analysis of PM and visibility in
western U.S. Although the modeling domain includes
the entire U.S. (contiguous 48 states), only the results
for the western U.S. were used to calculate the future-
year PM concentration profiles. However, the results
for the entire domain are presented here. The
application procedures and modeling results are
summarized in this section.
Modeling Domain
The REMSAD modeling domain encompasses
the contiguous 48 states. The domain extends from
126 degrees west longitude to 66 degrees west
longitude, and from 24 degrees north latitude to 52
degrees north latitude. A grid cell size of 2/3
longitude by 1/2 latitude (approximately 56 by 56 km)
was used across the grid, resulting in a 90 by 55 grid
(4,950 cells) for each vertical layer. Eight vertical
layers were used for the PM modeling and the first
layer results were used to estimate future air quality for
the surface monitoring sites. Although REMSAD
covers the entire U.S., in this analysis only results for
their 11 westernmost states are used.
Simulation Periods
Four simulation periods or episodes were
modeled. These episodes correspond to the four
seasons of the year and consist of the first ten days of
the months of May (spring), July (summer), October
(fall), and December (winter).
Model Inputs
The REMSAD modeling system also requires a
variety of input files that contain information
pertaining to the modeling domain and simulation
period. These include gridded, day-specific emissions
estimates and meteorological fields; initial and
boundary conditions; and land-use information.
Separate emission inventories were prepared for
the base-year and each of the future-year scenarios.
All other inputs were specified for the base-year
model application (1990) and remained unchanged for
each future-year modeling scenario.
Modeling Emission Inventories
The data and methodologies used to prepare the
REMSAD modeling emission inventories for this
study were consistent with those used for the
photochemical modeling, but included primary
particulates and other species as required for the
particulate chemistry. Similar to UAM/UAM-V,
REMSAD, requires detailed emission inventories,
containing temporally allocated emissions for each
grid cell in the modeling domain for each species
being simulated. EPS 2.5e was used for the emissions
processing. Note that this system has been
specifically designed to accommodate regional-scale
model applications of particulate matter and toxic
species as well as ozone precursors.
The emissions scenarios for this study included
1990 base, 2000 Pre-CAAA, 2000 Post-CAAA, 2010
Pre-CAAA, and 2010 Post-CAAA scenarios. Each
inventory includes typical season weekday area source
emissions, typical summer or winter day utility
emissions (as appropriate), typical season weekday
non-utility point source emissions, and typical season
daybiogenic emissions.
The anthropogenic input emissions inventory data
were provided by Pechan (1998). These included area
and point source emissions data from the National
Particulates Inventory (by county and for specific
point sources); county-level vehicle miles traveled
(VMT) estimates; mobile-source emission factors for
VOC, NOx, and CO; and PM emission estimates for
mobile sources. Note that road dust and other
primary particulates are included in the area-source
emissions file.
Seasonal biogenic emission estimates for the
domain were prepared using version 2 of the EPA's
UAM Biogenic Emissions Inventory System (BEIS-2).
C-44
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
BEIS-2 (which estimates biogenic emissions based on
various biomass, emission, and environmental factors)
utilizes land-use information to determine the
distribution of biogenic emissions.
Preliminary processing of the data prior to the
application of the EPS 2.5e system was necessary.
This consisted of generating the on-road mobile
emissions and reformatting all data into AMS and AFS
workfile format. Particulate matter pollutants from
on-road mobile emissions were provided at county
level and were broken down into 12 different urban
and rural roadway classifications. To take advantage
of the temporal information provided in the utility
emissions data, seasonal AFS workfiles were generated
separately for the summer and winter months.
All anthropogenic emissions inputs to the various
models were preprocessed through the EPS 2.5e
system. Point, area, and on-road mobile source
emission data were processed separately to facilitate
both data tracking for quality control and the use of
the data in evaluating the effects of alternative control
strategies on simulated air pollutant concentrations.
Temporal and spatial allocation were performed as
described in Section III.
Primary particulate and secondary particulate
precursor emissions are basically derived from
particulate matter species, i.e., PM10, PM25, and NH3.
Therefore a chemical speciation scheme that differs
from that for VOC speciation is applied. Table C-9
provides the chemical speciation applied for
REMSAD.
Table C-9
Chemical Speciation Schemes Applied for
REMSAD
VOC: VOC
NH3. NH3
NO,: NOX, NO, NO2
X A' ' Z.
PMC: POA, PEC, GSO4, PNO3, PMcoars
PM: POA, PEC, GSO4, PNO3, Pmfine
Emission inputs to the REMSAD for selected
species, by component, are provided in Table C-10.
The purpose of the tables is to quantify the
contribution of each source category to total
emissions. The species shown include primary
participates and other species that are important to
secondary particulate formation. VOC, NO^ and SO2
emissions are estimated to increase under the Pre-
CAAA scenario and to decrease under the Post-
CAAA scenario. For SO2, the decreases come from
the utility sector and are offset by increases in the
other components. NH3 emissions increase for both
scenarios and are slightly higher under the Post-
CAAA scenario for both years, presumably due to
increased use of natural gas fuel. PM10 and PM25
emissions (primary particulates) are similar for all
scenarios.
The primary chemical process for PM
applications in REMSAD is sulfate formation. In-
cloud processes can account for the majority of
atmospheric sulfate formation, especially in the
wintertime when gas-phase chemistry is slow. The
two most important pathways for in-cloud sulfate
formation are the reactions of aqueous SO2 with
ozone and hydrogen peroxide (H2O2). At cloud pH
below 4.5 (most common in the eastern U.S.), the
ozone reaction is slow and the H2O2 reaction
dominates. Since the H2O2 is often present at
ambient concentrations below those of SO2,
formation of sulfate can be limited by the availability
of H2O2, and thus can be quite nonlinear. The
formation of H2O2 is tied to the overall atmospheric
photochemical system, and responds to changes in
ambient levels of VOC and NOX. Because of this link,
emission changes for VOC and NOX may have effects
on ambient sulfate levels. In short, the emissions of
ozone and PM precursors (i.e., NOX and VOC) will
affect the oxidizing capacity of the troposphere which
is represented primarily by the concentrations of
radicals and hydrogen peroxide, and thus affect the
rate of oxidation of the NOX and SO2 to nitrate and
sulfate.
In REMSAD, there is no relationship between
VOC emissions and the production of secondary
organic aerosol (SOA).
C-45
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-10
Emission Totals
by Component for each Scenario for the Entire U.S. (tpd)
voc
Area
Onroad Mobile
Point
Utility
Total
Base 1990
33,972
18,659
9,503
96
62,229
2000 Pre-
CAAA
39,154
16,454
10,298
85
65,991
2000 Post-
CAAA
27,620
10,683
8,457
85
46,845
2010Pre-
CAAA
43,708
18,776
11,606
134
74,224
2010Post-CAAA
28,575
8,804
9,454
137
46,970
NOx
Area
Onroad Mobile
Point
Utility
Total
Base 1990
13,766
20,399
7,964
20,188
62,316
2000 Pre-
CAAA
15,659
20,660
8,694
22,787
67,800
2000 Post-
CAAA
15,252
17,421
5,645
11,170
49,487
2010Pre-
CAAA
17,697
24,142
9,803
24,808
76,450
2010Post-CAAA
15,794
14,696
5,985
10,319
46,793
SO2
Area
Onroad Mobile
Point
Utility
Total
Base 1990
3,517
1,555
12,808
43,380
61,260
2000 Pre-
CAAA
4,174
1,730
14,620
44,261
64,786
2000 Post-
CAAA
4,174
924
14,620
28,742
48,460
2010Pre-
CAAA
4,811
2,109
16,422
48,482
71,823
2010Post-CAAA
4,811
1,121
16,422
27,016
49,369
NH3
Area
Onroad Mobile
Point
Utility
Total
Base 1990
10,230
544
667
-
11,441
2000 Pre-
CAAA
13,189
957
742
-
14,888
2000 Post-
CAAA
13,189
957
742
91
14,979
2010Pre-
CAAA
15,710
1,191
842
-
17,744
2010Post-CAAA
15,710
1,194
1,015
608
18,527
PM10
Area
Onroad Mobile
Point
Utility
Total
Base 1990
73,221
972
2,549
764
77,507
2000 Pre-
CAAA
74,431
799
2,891
691
78,812
2000 Post-
CAAA
72,640
706
2,891
697
76,933
2010Pre-
CAAA
74,532
814
3,252
837
79,435
2010Post-CAAA
72,240
563
3,252
758
76,813
PM2.5
Area
Onroad Mobile
Point
Utility
Total
Base 1990
16,717
797
1,625
288
19,427
2000 Pre-
CAAA
17,438
618
1,840
247
20,143
2000 Post-
CAAA
17,147
536
1,840
250
19,773
2010Pre-
CAAA
18,174
640
2,066
332
21,212
2010Post-CAAA
17,637
391
2,066
305
20,398
C-46
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Air Quality, Meteorological, and
Land-Use Inputs
Initial species concentrations and lateral boundary
conditions were specified to approximate background
concentrations of the species; for the lateral
boundaries the concentrations varied (decreased
parabolically) with height. The background
concentrations are listed in Table C-ll.
Table C-11
Background Species Concentration used
for REMSAD Initial and Boundary
Conditions.
Species
NO
NO2
SO2
NH3
VOC
NHO3
PNO3
GSO4
ASO4
NH4N
NH4S
SOA
POA
PEC
PMFINE
PMCOARS
Concentration
(PPd)
0.0
0.1
0.7
0.5
20.0
0.01
0.01
0.1
0.0
0.01
0.1
1
1
5
1
1
Meteorological inputs were derived based on
output from the Pennsylvania State University/
National Center for Atmospheric Research
(PSU/NCAR) mesoscale model (MM4). Gndded
fields of horizontal wind components, temperature,
water-vapor concentration, vertical exchange
coefficient, precipitation, and pressure were prepared
for input to REMSAD. Land-use information was
obtained from the USGS database (at 18 km
resolution).
REMSAD Simulation Results
Model Performance
The assessment of model performance for
participate models is a difficult task due to a relative
lack of data and information regarding the spatial
distribution, composition, and size fractionation of
airborne particulates. Development and evaluation of
particulate measurement and modeling techniques are
active areas of research. As a result, there are
currently no standard approaches or model
performance criteria for the evaluation of regional-
scale particulate models. For this study, model
performance for REMSAD was examined by
comparing the simulated values of selected species
with available data. This comparison is intended to
provide an indication as to whether the simulated
values represent the concentration levels and the range
of concentrations indicated by the available
observations.
Summaries of model performance were prepared
by comparing the simulated values of PM with
observed values representing seasonal averages.
Comparisons were performed for the entire domain
(entire U.S.), the western U.S., and the eastern U.S.
Only the western U.S. results are presented here.
Data from both the AIRS and IMPROVE PM
monitoring networks were included in the evaluation.
REMSAD-derived sulfate and nitrate concentrations
were also compared to a small number of IMPROVE
measurements.
C-47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Scatter plots for PM10 are provided in Figures C-
15 through C-18. For PM10, there is a tendency for
underestimation of the seasonal averages in the
western U.S., in particular for the fall and winter
simulation periods. Similar plots for PM25, sulfate,
and nitrate are available in (SAI, 1999) and show
generally good agreement for these species.
These plots indicate that model performance
varies throughout the western U.S. and throughout
the year. A closer look at the comparison between the
simulated and observed values indicates that the
agreement is generally better for the IMPROVE sites
and that most of the large underestimation occurs for
the AIRS sites. The IMPROVE sites tend to be
located in rural areas, while the AIRS sites tend to be
located in urban areas. There are numerous possible
explanations for the differences. One possibility is
that one or more components of the urban emissions
may not be accurately represented in the inventory. A
second possibility is that the grid resolution
(approximately 56 km) is not sufficient to resolve the
urban-scale processes influencing particulate
formation and transport. It is encouraging that
generally good agreement is achieved for the limited
number of sulfate and nitrate measurements. Overall,
the model performance results suggest that the
REMSAD modeling system (including the
meteorological, air quality, and geographical inputs)
provides a reasonable basis for the Section 812
prospective modeling.
REMSAD Modeling Results
The REMSAD simulation results for the Pre- and
Post-CAAA scenarios were used in this study to
calculate factors for adjustment of observed data and
estimation of future-year concentration levels. These
were calculated by comparing the simulated
concentrations corresponding to each future-
year/scenario simulation with those for the base-year
simulation (1990). These comparisons indicate that
for both future years and both size categories, the Pre-
CAAA simulation results are characterized by
increases in PM, while the Post-CAAA results show
both increases and decreases. Focusing on the
western U.S., the increases occur over the larger urban
areas and are likely attributable to increases in areas-
source emissions of precursors.
While there are increases in primary particulate
emissions for some portions of the west, most of the
increases are attributable to secondary particles.
Isopleth maps for these comparisons are available in
(SAI, 1999).
Figures C-19 and C-20 illustrate the differences in
seasonal average simulated PM concentration between
the Pre- and Post-CAAA simulations for 2010 for the
summer period. The differences are calculated as
Post-CAAA minus Pre-CAAA, so that negative values
indicate lower concentrations for the Post-CAAA
scenario. The simulated values for the Post-CAAA
scenario are lower than the corresponding Pre-CAAA
values for both years. The magnitude and spatial
extent of the decreases is greater for 2010 than for
2000 (not shown).
Calculation ofPM Air Quality Profiles
The calculation of PM profiles for 2000 and 2010
(for assessment of the effects of the CAAA) include
the use of REMSAD results for the western U.S. and
RADM/RPM results for the eastern U.S. As for
ozone, this was accomplished using an approach that
combines observed data and air quality modeling
results to estimate the future-year concentrations.
While the overall approach is similar to that for ozone
(as described in Section III), there are some
differences. The future-year air quality profile
estimation methodology for PM, as applied to the
analysis of the CAAA, is described in this section.
Overview of the Methodology
The methodology for calculation of the
adjustment factors differed slightly for the
RADM/RPM and REMSAD applications. For
RADM/RPM the modeling results were used to
calculate adjustment factors for several PM
component species; for REMSAD adjustment factors
C-48
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
for PM10 and PM2 5 were computed directly from the
model output. The adjustment factors for each
monitoring site were calculated (using the
appropriately matched values) for several different
concentration levels (i.e., the changes in concentration
are dependent upon concentration level). The species
concentrations for each monitoring site (estimated
using the observations) were then modified using the
site-specific (or grid-cell-specific) adjustment factors.
For RADM/RPM, PM concentrations were then
recalculated using the resulting component values.
For both models, the ratios were calculated on a
seasonal basis and were used accordingly to adjust the
observed values. Following adjustment of the
observed data, statistical quantities, or "profiles",
describing the PM distribution for each monitoring
site were then calculated.
Description of the Observation
Dataset
One of the first tasks in calculating the future-year
PM profiles was the creation of a dataset containing
the observed concentrations for all monitoring sites
located within the modeling domain for the year 1990.
The starting point for this analysis is a database
retrieved from the EPA Aerometric Information
System (AIRS) of measured ambient concentrations of
TSP, PM10, and PM25 for the year 1990. Due to the
limited number of measurements (usually taken once
every six days), data for 1989 and 1991 were also used
to supplement the 1990 database. Cross-estimation
was performed when one of measurements was
missing (i.e., PM10 or PM25). The PM component
species (that make up secondary PM) were estimated
based on a methodology developed by Langstaff and
Woolfolk (1995) for the Section 812 retrospective
modeling analysis. Size fractionation (PM10 fraction of
TSP and PM25 fraction of PM10) and apportionment
of secondary PM species relied on a review of
previous studies to provide general relationships used
to estimate these components of particulate matter.
The relationships used for this study depend only on
broad geographic region (East, Central, West), time of
year (quarter for PM10 and season for PM25), and
whether the monitor is located in an urban or rural
setting.
The geographical regions used throughout this
analysis are presented in Table C-12. In addition to
secondary composition fractions, ratios relating PM2 5
to PM10 were employed. The literature review
conducted for establishing secondary particulate
matter concentrations for the 1990 data and the
sources of ratios and apportionment factors used in
the equations below is discussed in some detail by
Langstaff and Woofolk (1995).
It should be noted that there is considerable
variability in the size and species composition of
particulate matter, not only between different
locations, but also from day to day in the same
location. The average size fractions and speciation
factors used for this study represent a rather sweeping
simplification of the actual physical phenomena that
are being modeled. However, this may be justified in
the context of this study, due to data limitations and
the fact that the results are aggregated to the annual
level.
As mentioned earlier, cross-estimation of TSP,
PM10, and PM25 was used to estimate values not
present in the original AIRS database. The results of
a linear regression of TSP versus PM10 by region,
quarter, and land-use were used to fill in either PM10
or TSP, if the other was missing. After this, the
results of a linear regression of PM25 versus PM10 by
region, season, and land use were then used to fill in
PM25 values where missing. With both estimated and
observed TSP, PM10, and PM25, the coarse PM
concentration was calculated as well as the PM
concentration greater than 10 microns.
PM>10 =TSP-PM10
PMC = PM10 - PM2 5
(1)
C-49
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableC-12
Geographical Regions of the U.S.
Central
Oklahoma
Missouri
Kansas
Nebraska
Iowa
South Dakota
North Dakota
Minnesota
Wisconsin
Illinois
irse PM were partitioned
rticulate concentrations.
East
Indiana
Kentucky
Ohio
Michigan
Virginia
West Virginia
Pennsylvania
New York
Maryland
New Jersey
Connecticut
Rhode Island
Massachusetts
Vermont
New Hampshire
Maine
Delaware
Washington, DC
Florida
Georgia
Alabama
Mississippi
Louisiana
Arkansas
Tennessee
North Carolina
South Carolina
into O =
As shown P =
West
Nevada
Utah
Colorado
New Mexico
Arizona
Texas
California
Oregon
Washington
Idaho
Wyoming
Montana
organic concentration
other particulate concentratii
below, each equation illustrates how the secondary
particulate concentrations are calculated from coarse
and fine PM.
S
N
O
P
=[PM2.5*rs2.5]
=[PM2.5*rn2.5] + [PMc*rnC]
=[PM2.5*ro2.5]
=[PM
2.5rp2.5
(3)
(4)
(5)
(6)
where
S = sulfate concentration
N = nitrate concentration
PM25 = PM less than or equal to 2.5 microns in
size
PMC = PM between 2.5 and 10 microns in size
(coarse PM)
rx2.s = rati° of <=2.5 micron sulfate (x=s),
nitrate (x=n), organic (x=o), and other
particulate (x=P) to PM25
rxC = ratio of 2.5-10 micron sulfate (x=s),
nitrate (x=n), organic (x=o), and other
particulate (x=P) to coarse PM
Note that r^ was based on a review of available
data/literature and depends on geographic region,
C-50
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
time of year, and land-use characteristics of the
monitoring site location.
The observed and estimated species concentra-
tions were then input into a single AMP350-format
datafile. From the information contained in this file,
two SAS datasets were created: a concentration dataset
and a monitor information dataset. The concentration
dataset contains the daily concentrations for each
monitor, with each record in the dataset representing
a single monitor-day. The monitor information
dataset contains monitor-specific information such as
land-use and location.
Because PM monitors are typically operated on a
one-in-six day monitoring schedule, calculating
percentiles for the PM profiles using data for a single
year can be very sensitive to the method used in the
percentile calculation. This is especially true when a
monitor record only needs to be 50 percent complete
(i.e., contain at least 30 values) for a profile to be
generated. To minimize dependence on the form of
the percentile equation, the 1990 PM data were
supplemented with that from the years 1989 and 1991.
In using multiple years worth of monitoring data, it
was discovered that the identifier (ID) corresponding
to a monitor in a given physical location could change
from one year to the next. Also, a monitor could have
moved to a nearby location and been assigned a
different ID. It was also possible that the monitor ID
for a PM10 monitor might be different from that of a
TSP or PM25 monitor despite the fact that their
physical separation is zero. Because much of the
profile work is dependent upon the monitor ID, this
led to a vast increase in the reported number of
operating monitors.
To accommodate these possibilities, monitors
with different monitor ID's were considered the same
monitor if their physical separation was less than or
equal to 1 km. Monitoring data from the two
monitors were combined. If data existed for both of
the monitors on the same day, the daily data from the
monitor with the higher ID was removed.
For particulate data, a monitor record was
considered to be complete if data were available for 50
percent of the 24-hour observations for a given year
(assuming a one-in-six day monitoring schedule).
Although three years worth of data were used for the
PM analysis, these data were considered to represent
one year with respect to the completeness
requirement. There were 2048 PM monitors with
complete data.
Calculation of Percentile-Based
Adjustment Factors
For each future-year modeling scenario, grid-cell-
species-season-specific adjustment factors were
calculated using the speciated, daily-simulated
concentrations from RADM/RPM and REMSAD.
Because the species and seasons differed between the
two models, the exact calculation of adjustment
factors also differed. Nevertheless, the overall
approach was the same. Individual monitoring sites
were mapped onto the gridded output (to determine
the grid cell in which each monitor was located) and
the concentrations for the corresponding grid cells
were used to calculate a set of adjustment factors for
each species, season, and future-year modeling
scenario. The adjustment factors were specified to be
the ratio of the percentile concentrations for the
future- and base-year simulations of a given species-
season, where the percentile concentrations were
calculated using data for the selected species and
season concentrations:
Adjustment
Factor
xth Percentile
Concentration
future year, species, season
i,species,season
xth Percentile
Concentration^
ase vear.svecies.season
{*,.}= {10, 30, 50, 70, 90}
For calculation of the percentile concentrations,
the empirical distribution function with averaging was
employed. Because the concentrations for the lower
percentiles can be rather small, a threshold value of
0.01 microgram/m3 was set to keep the adjustment
factors reasonable. In other words, all concentrations
below 0.01 microgram/m3 were reset to 0.01
C-51
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
microgram/m3. If either the base year or the future
ear percentile concentration was set to the minimum
value, the adjustment factor was set equal to one.
This percentile-based approach was selected due to
the limitations of using a single adjustment to
represent the change in the modeled PM species
concentrations in moving from the base- to the
future-year scenarios.
For RADM, adjustment factors were calculated
for the sum of sulfate, nitrate, and ammonium.
These were calculated for the entire year (i.e., only one
"season"). For REMSAD, the adjustment factors
were calculated for PM10 and PM25. These were
calculated on a seasonal basis.
A SAS dataset containing the monitor-level
adjustment factors was created for each future-year
modeling scenario considered in this study for this
year.
Use of Adjustment Factors to
Modify Observed Concentrations
Using the calculated adjustment factors for each
future-year scenario and the monitor-level
observations, a dataset containing modified PM10 and
PM2 5 concentrations for each of the four future-year
scenarios was created. Because each monitor has five
adjustment factors per scenario, species, and season,
it was first necessary to rank order the observed
concentrations into five quintile-based groups (with
ties being assigned to the higher group) with respect
to the species and season definitions mentioned
previously. Thus for RADM, the quintiles were
calculated for the daily sum of the observed sulfate,
nitrate, and ammonium concentrations over the entire
year (ignoring that the data are actually for the years
1989, 1990, and 1991). For REMSAD the quintiles
were calculated for the observed daily PM10 and PM2 5
over each of the four seasons. Once each of the
observed concentrations was identified with a
particular quintile group, the appropriate adjustment
factor was selected and applied to calculate the future-
year-scenario PM10 and PM25.
For RADM, the adjustment factor was applied
using the following equations:
AdjNitrateSulfate, = ObsNitrateSulfate,
AdjOrganicst
AdjP,
— ObsOrganicst * 1
= ObsP, * 1
For example in the first equation,
{ObsNitrateSulfate^ is the set of observed daily sums of
the nitrate and sulfate concentrations (in
micrograms/m3) for a given monitor. The
k\ObsNitrateSulfate^ subscript is the number of the
quintile group to which ObsNitrateSulfate1 belongs.
Adj.Factork[0bsNitrateSa,fatei]iNUrateM.fate is then the appropriate
adjustment factor for ObsNitrateSutfatet. The resulting
set of adjusted daily sums of nitrate and sulfate
concentrations, {AdjNitrateSulfate^, represents the
future year estimates of the daily sum of nitrate and
sulfate concentrations. In this case, P represents other
participate components. For those monitors within
the RADM domain, PM10 and PM25 concentrations
were calculated by summing each of the above
components.
For monitors within the REMSAD domain, the
procedure for calculating the future-year PM10 and
PM25 is more direct. Future-year concentrations of
these two PM species are calculated using the
observed/estimated PM10 and PM25 concentrations
and the appropriate adjustment factors:
AdjPMW, = ObsPM10t * AdjFactork[0h!mm]iPM10iSea!m
AdjPM2.5t = ObsPM2.5t *
In the first equation, {ObsPMW} is the set of
observed daily PM10 concentrations (in
micrograms/m3) for a given monitor. The
k[ObsPM10J subscript is the number of the quintile
group (based on season) to which ObsPM10t belongs.
AdjFactork[0bsPUWl]tPMW/xasm is then the appropriate
adjustment factor for ObsPM10t. The resulting set of
PM10 and PM25 concentrations, {ObsPMW} and
{ObsPM2Ji}, therefore represents the future-year
estimates of these PM species.
C-52
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Calculation of PM Profiles
PM10 and PM2 5 air quality profile databases were
compiled for all simulations performed as part of the
Section 812 prospective analysis. For each of the
particulate species, these data bases contained the
number, the arithmetic mean, the median, the annual
second highest, and the 2.5 to 97.5 percentiles (in
increments of five) of the daily (as available)
concentrations. The profiles are reported at the
monitor level and include 2048 site locations.
The histograms in Figures C-21a through C-24b
illustrate the distribution of ratios for the annual
average monitor-level PM10 and PM2 5 concentrations
corresponding to the 2000 and 2010 simulations. In
these figures, ratios greater than one indicate that the
future-year/scenario concentration is greater than the
base-year (1990) value, whereas ratios less than one
indicate a lower value for the future-year.
The 2000 Pre-CAAA ratios for PM10 (Figure C-
2 la) indicate that the annual average PM10
concentrations corresponding to this scenario are
higher in some areas and lower in other areas than the
base-year (1990) values. The ratios generally range
from approximately 0.95 to 1.1, but also include some
higher values. In contrast, the ratios corresponding to
the 2000 Post-CAAA simulation (Figure C-21b) are
generally less than or equal to one, with most sites
being assigned a ratio consistent with a small decrease
in annual average PM10 concentration. There are also
some lower values.
generally higher than (or equal to ) the base-year
(1990) values. The ratios generally range from
approximately 0.975 to 1.15. In contrast, the ratios
corresponding to the 2000 Post-CAAA simulation
(Figure C-23b) are generally less than one. In this
case, the ratios range from approximately 0.925 to
1.075.
For 2010, the PM25 ratios (Figures C-24a and C-
24b), indicate increases for the Pre-CAAA scenario
and mostly decreases for the Post-CAAA scenario.
Again, compared to 2000, concentrations for 2010 are
higher relative to the base year under the Pre-CAAA
scenario and similar to or slightly lower relative to the
base year under the Post-CAAA scenario.
For both future years (2000 and 2010), the ratios
indicate that the Post-CAAA concentrations (annual
average) are lower than the corresponding Pre-CAAA
values. This is illustrated in Figures C-25a through C-
26b. The smaller ratios for 2010 reflect larger
differences between the Pre- and Post-CAAA
scenarios.
Figure C-22a and C-22b display the distribution of
ratios of the future-year-scenario to base-year annual
average PM10 concentrations for 2010. Compared to
the histogram plots for 2000, the ratios are higher for
the Pre-CAAA scenario but similar for the Post-
CAAA scenario. There is some indication that, by
2010, increases due to growth are limiting the
effectiveness of the CAAA measures.
The 2000 Pre-CAAA ratios for PM25 (Figure C-
23a) indicate that the annual average PM25
concentrations corresponding to this scenario are
C-53
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-14
80-km RADM Domain
Note: Nested 20-km grid estimates were not used to generate final results, but were used in evaluating
the reasonableness of results.
C-54
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
1
8
I
:
I
f
1
i
-
i
-
i
-
OprmiM • IWD
00<
1 1
II 1
....
H
'• i
H
1.1
•: '
l.i
",
» 1C ED » 4D -5C BD TO DO BO J<
i i i i i i * i * i * i i ' i ' •
X-
: / :
ff
- /*' -
/
.^
' >X '
;.. " x
**
-
~ **
^. •* j&L'i"' '«•_ v
j JSU^^S^f^
~/^^^^F ' ]
h IO ZQ X> 40 3C GO TO BO 94 1
CiB3EM>-ELi f H1O lui/n-.T-l
•*-i Vuuimffc . 1 VH>
TOO a LD £0 JO rt M «i •» iD >» iflft,,,
DO
H
«
»* - TO
•W _' • , i
a
SD 5 -<1
«D | Jll
5
9(1 1 "
=D «
m "
A
• .
- / '-
"*
/
- ^ -
.' .*" *+*
-
* tv - .^
• ^ *Y •/ *
_
*
**A& -
^Wv^ .*
i^™* B*"-^-" "rf- . *
•ii- .
' * ' J"JMlJ |JT *^W*
: •••aMpt •
•••'. " r*. ...
4 LO*o50wW'W1'*SM*il
O&SCEVID PUlO I.ut/m3f
-.11
HI)
."
.„
-II
III
I.I
ill
..
v.
Figure C-15. Comparison of simulated and Figure C-16. Comparison of simulated and
observed seasonal PM10 concentration (ug/m3) for observed seasonal PM10 concentration (ug/m3) for
REMSAD for the western U.S.: spring 1990 REMSAD for the western U.S.: summer 1990
P*U . lWt» !,_ ..„ .Oift
.11
1 ' ' ' .X
'-
-j» „
- Ad
/ ' W
- : - s «
i. i 00 i
j an
_. i
XWI a
3 OU
s
- *c
•* ---ii' V * \
i.| *** - *i
1 'II
:••, »»*'•' "- W
><*HFd
KtdfXfm^CJ^' -
id- \Q| - i*
jX"ff^W^T^ V , i , '.
"* " OBSEBi's£ PH?» (ue/mar W " '^ ^
LD BO » 40 3O ftO TO flfl H d(
i i | • | i | i | i | | i i | i .
/' m
/
jf
-
- *"* * '"""
^ -•;'
*
•
-
• • •* \ * '_
/ AS^"\ ^^** i
V^» ^f .. •^ ^ rf * •*• **
^^tfitrfrf^l^llVv ^l^fc • ~ ^r fc . r *i
jjjrtprttfSJSBjt^fJ-"-
^^^^ i ". i .* i . i .1 "| ,
ID BO 3C -HO 3D DO 7X1 BD BO Ji
',.
....
1,
...
•-
-..
1
'
••••
1-
a
otifenvcD rm a inf..-m?i
Figure C-17. Comparison of simulated and
observed seasonal PM10 concentration (ug/m3) for
REMSAD for the western U.S.: fall 1990
Figure C-18. Comparison of simulated and
observed seasonal PM10 concentration (ug/m3) for
REMSAD for the western U.S.: winter 1990
C-55
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 PM10 (ug/m3) + MAXIMUM =
Time: 100 Jul 1, 1990-100 Jul 11 1990 - MINIMUM =
Deg. Longitude
126 -118 -110 -102
0 0 ug/m3 (30,5)
385 ug/m3 (13,20)
-94
50
40
30
20
10
j i i M M i n I | i ij i i it T i rr i i i r in rn i i | i M i i i i n i i JT 11 i_
j.i i i 1 i i .1.1 i i i i.j.ljj j j.i..1.1.i I.LLLIXJ. i LLLJ j.j.i i i i i t i I 11'
49
44
OJ
-d
39
ti
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 PM25 (ug/m3) + MAXIMUM =
Time 100 Jul 1, 1990-100 Jul 11, 1990 - MINIMUM -
Deg. Longitude
126 -118 -110 -102
50
40
30
20
10
0
i i i
0
0.0 ug/m3 (30,5)
120 ug/'m3 (13.20)
-94
] i
I I I I I | I I I I I I M I i I I I I I I I I I I If I | I I I l_
1 I
I i I 1 I 1
I 1 1 i 1 j I j 1 1 I I 1 I i I 1 I i I I if
49
44
o>
T)
3
39 g
Q
34
29
10
20
30
40
50
24
Figure C-20. Difference in seasonal average PM25 concentration
(ug/m3) for the summer REMSAD simulation period (1-10 July 1990) for
2010: post-CAAA90 minus pre-CAAA90
C-57
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-21a. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios
F
60 -
> 50-
0
= — 40 -
o> §
£ £ 30 -
"55
* 10 -
o -
c
»Mio Concentration: 2000 Pre-CAAA
for Annual Average
/ 1990 Base-Year
median: 1.017
i i 1 —
i —
50 0 55 0 60 0 65 0 70 0 75 0 80 0 85 0 90 0 95 1 00 1
Ratio
05 1 10 1 15 1 20 1 25 1 30
Figure C-21b. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PMio Concentration: 2000 Post-CAAA /1990 Base-Year
60
> 50
o
I C-40
O" c
0 Q)
i £ 30
2s
S 20
10 -
median: 0.976
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-58
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-22a. Distribution of Combined RADM/RPM-and
REMSAD-Derived Monitor-Level Ratios for Annual Average
P
60 -i
> 50-
0
§ ~ 40-
o- c
g> o>
ul £ 30-
5 i
~~ 20-
0)
* 10-
0-
0
MIO Concentration: 2010 Pre-CAAA/ 1990 Base-Year
median: 1.040
i— I
.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1
Ratio
.05 1.10 1.15 1.20 1.25 1.30
Figure C-22b. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
Concentration: 2010 Post-CAAA/1990 Base-Year
60
> 50 H
o
§ ~ 40-I
0) 0)
> -°:
13
30 -
20 -
10 -
0
median: 0.981
T-Thn
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-59
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-23a. Distribution of Combined RADM/RPM-and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PM2.s Concentration: 2000 Pre-CAAA/1990 Base-Year
60
50-
= ~ 40-1
o> §
it e 30-
5 £
;§~20H
a)
* 10-1
median: 1.033
ft*
01 r^
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-23b. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PM2.s Concentration: 2000 Post-CAAA /1990 Base-Year
60
> 50
o
I ~ 40
O d)
i ^ 30
0) 0>
> 3
5 20
10 -
median: 0.968
T—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—hrl—I—I—I—I—I—F=l—I I I—i—i—T
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-60
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-24a. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PM2.5 Concentration: 2010 Pre-CAAA /1990 Base-Year
60
& 5(H
0)
3 53- 40
^ §
u- H 30
o a>
> 3
ts 20
0)
tf 10 H
median: 1.070
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-24b. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PM2.s Concentration: 2010 Post-CAAA/1990 Base-Year
60
> 50
o
|- 4°
f S
i ^ so
0) CD
> Q-
5 ^ 20
10 -
median: 0.976
T—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—f—\—I—I—I—I—I—I I I I—I—I—F
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-61
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-25a. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
Concentration: 2000 Post-CAAA / 2000 Pre-CAAA
60
>
o
50 -
40 -
30 -
20 -
10 -
0
median: 0.964
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-25b. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PMio Concentration: 2010 Post-CAAA / 2010 Pre-CAAA
60
o
c
d>
CT
£
LL
d>
d>
o:
50-
— 40 -
+rf -rw
d>
^ 301
o>
Q.
" 20 1
10-
-n-n-
median: 0.946
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-62
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-26a. Distribution of Combined RADM/RPM- and
REMSAD-Derived Monitor-Level Ratios for Annual Average
PM2.s Concentration: 2000 Post-CAAA / 2000 Pre-CAAA
60
o
c
d>
3
o-
0)
0)
_>
^5
JS
o>
o:
50-
~ 40-
£ 30 -
o>
Q.
"" 20 H
10-
0
median: 0.946
fl
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
Figure C-26b. Description of Combined RADM/RPM- and REMSAD-
Derived Monitor-Level Ratios for Annual Average PM2.5
Concentration: 2010 Post-CAAA / 2010 Pre-CAAA
60-
> 50-
o
I?401
g> a)
u. a 301
Q) 0
"55
median: 0.919
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30
Ratio
C-63
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Estimating the Effects of the
CAAA on Visibility
Light traveling through the atmosphere is
"absorbed" and "scattered" by gases and suspended
particles. These distortion processes contribute to
total atmospheric light extinction which, in turn,
causes visibility degradation. To characterize and
ultimately quantify the effect of changes in emissions
on visibility, an understanding of the concentrations
and types of gaseous participate constituents in the air
is necessary.
The influence of gaseous absorption on light
extinction is almost negligible. Gaseous scattering has
a larger impact, although this impact is generally not as
significant as either particulate absorption or
scattering. Together the influence of all four of these
light distortion processes is expressed quantitatively as
the light extinction coefficient, b^,. In this analysis
RADM/RPM and REMSAD are both used to
calculate bext.
RADM/RPM and Visibility
RADM/RPM estimates bext in the eastern U.S.,
for each emissions scenario (1990 base year, 2000 Pre-
CAAA, 2000 Post-CAAA, 2010 Pre-CAAA, and 2010
Post-CAAA), by combining the influences of particle
scattering and absorption and incorporating the effect
of scattering caused by water. The fine particles
estimated by RADM/RPM (including their associated
water) are secondary particulates: sulfates, nitrates,
associated ammonium, and organics. Absorption by
carbon particles is not included in the model's
calculations, nor is extinction resulting from primary
particles. By not including these latter influences,
RADM/RPM may underestimate the effects of air
pollution on visibility.
RADM/RPM, along with generating atmospheric
light extinction values, calculates "visual range" and
deciview (dV), both measures that quantify visibility.
The former, VR, is related to the light extinction
coefficient by the following equation:
VR(meters) = 3.912/bext,
where bext is in inverse meters. The latter measure of
visibility, dV, and the related DeciView Haze Index
are improved indicators of the clarity of the
atmosphere. This index more accurately captures the
relationship between air pollution and human's
perception of visibility than does VR or bext (Pitchford
and Malm, 1994). A deciview is defined by the
equation:
dV = lOln (bext/10) ,
where bext is expressed in inverse megameters.
The DeciView Haze Index has a value of
approximately zero when the light extinction
coefficient is equal to the scattering coefficient for
particle-free air. A roughly 10 percent increase in bext
translates to a one unit change in dV. Since the
apparent change in visibility is related to a percent
change in bext, equal changes in dV correspond to
approximately equally perceptible changes in visibility.
Research indicates that, for most observers, a "just
noticeable change" in visibility corresponds to an
increase or decrease of about one to two dV units.
An increase in the deciview level translates to
degradation of visibility, while a decrease represents
and improvement.
RADM/RPM Modeling Results
For this analysis, under the 1990 base year and
future year emissions scenarios, the annual mean
daylight hour bext, VR, and dV were estimated for each
RADM/RPM grid cell. A summary of 1990 and 2010
deciview levels for selected cities, metropolitan areas,
and national parks is provided in Table C-13. These
deciview estimates show that under the Pre-CAAA
scenario visibility degradation is expected throughout
much of the eastern U.S. Comparison of 1990 base
year and 2010 Post-CAAA estimates, however,
indicates that with the implementation of CAAA
related measures, a perceptible improvement in
visibility can be expected.
C-64
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableC-13
Comparison of Visibility in Selected Eastern Cities, Metropolitan Areas, and
National Parks
Mean Annual Deciview
Area Name
Acadia NP
Atlanta Metro Area
Boston Metro Area
Chicago Metro Area
Columbus
Detroit Metro Area
Everglades NP
Great Smoky Mtns. NP
Indianapolis
Little Rock
Milwaukee Metro Area
Minn. -St. Paul Metro Area
Nashville
New York City Metro Area
Pittsburgh Metro Area
St. Louis Metro Area
Shenandoah NP
Syracuse
Washington, DC Metro Area
State
ME
GA
MA
IL
OH
Ml
FL
TN
IN
AR
Wl
MN
TN
NY/NJ
PA
MO
VA
NY
DC/VA/MD
1990
Base Year
11.1
20.9
13.2
17.5
16.5
16.0
7.6
20.4
20.1
15.0
15.6
10.1
20.4
15.2
15.8
16.5
16.5
12.4
17.5
2010
Pre-CAAA
12.0
22.8
14.0
19.1
17.7
18.5
9.2
22.3
21.1
17.2
18.4
12.4
21.5
18.0
16.9
17.8
17.8
13.2
19.2
2010
Post-CAAA
10.4
20.0
11.9
17.0
15.1
15.3
6.9
19.6
19.0
15.1
15.3
10.3
19.0
13.9
14.2
16.0
15.2
11.5
16.3
*For cities, metro areas, or national parks not contained by a single RADM/RPM grid cell, the visibility measure
presented in this table is a weighted average of the mean annual deciview level from each of the grid cells that
together completely contain the selected area. Weighting is based upon the spatial distribution of an area over the
various grid cells.
C-65
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
REMSAD and Visibility
REMSAD was used to estimate the effect of
changes in emissions on visibility for the western U.S.
This model calculates light extinction coefficients
based upon estimates of the gridded ground-level
concentrations of the following species — sulfate
(NH4S+GSO4+ASO4), nitrate (NH4N+PNO3),
NO2, SOA, POA, PEC, PMfine and PMcoarse (refer
to Table C-8 for a description of these species
abbreviations). The contribution from each of these
species is adjusted based on the extinction efficiency
of each and, in the case of sulfate, nitrate and SOA, an
adjustment dependent on the relative humidity. The
total extinction coefficient is then given by:
CAAA scenario visibility is expected to remain
unchanged between 1990 and 2010 throughout much
of the West and actually improve in coastal Oregon
and along the western Idaho border. In the larger
urban areas, however, perceptible visibility
degradation is predicted. Visibility improvement in
and around western cities, especially in California, is
predicted under the Post-CAAA scenario. Figure C-
29 captures these changes and shows that in 2010
improvements in visibility are not expected to be
restricted to just the larger urban areas; compared to
1990 base year estimates, Post-CAAA deciview levels
are also predicted to be lower throughout much of
Washington, Oregon, and Nevada and in sizeable
sections of Arizona, Idaho, Utah, and Wyoming.
bext = 10.+ 0.17*NO2 + 4,4(RH)*sulfate + f;o3(RH)*Nitrate
+ fsoa(RH)*SOA +6.2*POA + 10.5*PEC + PMFINE +
0.6*PMCOARSE
where the constant value of 10.0 is the contribution to
the scattering coefficient for particle-free air (Rayleigh
scattering). REMSAD generated bext values are then
converted to deciviews.
REMSAD Modeling Results
Visibility estimates and change in visibility were
calculated for each of the future-year scenarios for use
in the effects analysis. Figure C-27 illustrates 1990
base year deciview levels for the western U.S. This
map shows that visibility is poorer in the region of
California extending from San Francisco southward to
Los Angeles, the Pacific Northwest, and larger
metropolitan areas such as Denver, CO; Albuquerque,
NM; and Phoenix, AZ. Most noticeable is the
comparatively high deciview level in the Los Angeles
region.
Figures C-28 and C-29 illustrate the difference
between 2010 Pre-CAAA and 1990 base year
estimates and the difference between 2010 Post-
CAAA and 1990 base year estimates, respectively.
The first of these maps shows that under the Pre-
C-66
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 DV ( )
Summer , 1990
-126.00
+ MAXIMUM = 34.6 (5,51)
- MINIMUM = 4.7 (16,32)
Deg. Longitude
-119.33 -112.67 -106.00
- 48.0
24.0
Figure C-27. Seasonal Average Deciview for the summer REMSAD
simulation period (1-10 July 1990): base 1990 (western U.S. only)
C-67
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
LEVEL 1 DV ( )
Summer , 1990
-126.OO
+ MAXIMUM = 3.6 (15,18)
- MINIMUM = -0.4 (4,37)
Deg. Longitude
-119.33 -112.67 -1O6.OO
•9 - 3 - ^
- 48.0
J-J24.0
Figure C-28. Difference in seasonal average Deciview for the summer REMSAD
simulation period (1-10 July 1990): 2010 pre-CAAA90 minus base 1990 (western
United States only)
C-68
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Acid Deposition
The acid deposition modeling efforts for this
analysis focused on estimating the change in ambient
concentrations of selected pollutants as a result of
changes in emissions. The need to focus on relative
changes, rather than absolute predictions, is especially
acute when estimating air quality outcomes for
pollutants subject to long-range transport, chemical
transformation, and atmospheric deposition. The
complexity of the relationships between emissions, air
concentrations, and deposition is well-described in the
following paragraph from the RADM report
document developed by Robin Dennis of U.S. EPA's
National Exposure Research Laboratory:
"Sulfur, nitrogen, and oxidant species in the
atmosphere can be transported hundreds to thousands
of kilometers by meteorological forces. During
transport the primary emissions, S02, NOX, and
volatile organic compounds (VOC) are oxidised in
the air or in ckud-water to form new, secondary
compounds, which are acidic, particularly sulfate and
nitric acid, or which add to or subtract from the
ambient levels of oxidants, such as o^pne. The
oxidi^ers, such as the hydroxyl radical, hydrogen
peroxide and o^one are produced by reactions of
VOC and NOX. The sulfur and nitrogen pollutants
are deposited to the earth through either wet or dry
deposition creating a bad of pollutants to the earth's
surface... However, the atmosphere is partly cleansed
of oxidants through a number of physical processes
including deposition (e.g., o^pne is removed by wet and
dry deposition). Dry deposition occurs when particles
settle out of the air onto the earth or when gaseous or
fine particle species directly impact land, plants, or
water or when plant stomata take up gaseous species,
such as S02. In wet deposition, pollutants are
removed from the atmosphere by either rain or snow.
In addition, fine particles or secondary aerosols formed
by the gas- and aqueous-phase transformation
processes scatter or absorb visible light and thus
contribute to impairment of visibility. "^
"' Dennis, R. RADM Report (1995), p. 1.
The complexity and nonlinearity of the
relationships between localized emissions of
precursors, such as SO2 and VOCs, and subsequent
regional scale air quality and deposition effects are so
substantial that advanced modeling is required to
accurately estimate the broad-scale impact of changes
in emissions on acid deposition. For this analysis,
EPA used the Regional Acid Deposition Model
(RADM) to estimate acid deposition in the eastern
United States.
Overview of the RADM Modeling
System
RADM, a three-dimensional Eulerian grid-based
model also used in the PM analysis, estimated nitrogen
and sulfur deposition for the 1990 base year and each
of the future year emissions scenarios. Estimates,
expressed in kg/ha, were developed for 2000 and
2010 and calculated for each 80-km RADM grid cell.
It is important to note, however, that ammonia
deposition, a significant contributor to total nitrogen
deposition, was held constant for each of the model
runs. This was because livestock farming and other
activities that drive ammonia formation and
deposition were essentially unaffected by the CAAA-
related control programs. A more detailed description
of RADM, its domain, and its inputs is provided
earlier in this appendix..
RADM Modeling Results
Figures C-30 and C-31 show the 1990 base-year
deposition estimates for sulfur and nitrogen
respectively. Predictions for both pollutants under the
Pre- and Post-CAAA scenarios are displayed in
Figures C-32 through C-35. Comparison of the three
maps showing sulfur deposition and comparison of
the three maps showing nitrogen deposition reveals
that for both pollutants annual deposition under the
Pre-CAAA scenario is expected to increase between
1990 and 2010. Year 2010 Post-CAAA sulfur and
nitrogen deposition projections, however, are not only
lower than 2010 Pre-CAAA projections, but also
below 1990 base year levels. Together, these maps
indicate that between 1990 and 2010 average annual
C-69
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
acid deposition is expected to decrease as a result of
the Clean Air Act Amendments.
Noticeable in each of the figures, especially those
mapping nitrogen, is an area of high deposition along
the Virginia-North Carolina border. This "hot spot"
is above Person County, NC, a region with one large
and one very large utility plant.11 Emissions from
these plants, particularly NOX, likely are the source of
the high deposition in this area. Person County
exhibits the highest base year and future year Pre- and
Post-CAAA acid deposition estimates in the entire
RADM domain.
Comparison of 2010 Pre- and Post-CAAA
emissions in Person County shows that NOX
emissions are expected to be lower in 2010 as a result
of the CAAA. This change in emissions, however,
translates to a change in acid deposition that is not
captured by the maps provided in this section. 2010
Post-CAAA nitrogen and sulfur deposition estimates
for this county are 27.2 and 78.0 kg/ha respectively.
These figures represent a decrease in nitrogen
deposition of 14.0 kg/ha and a decrease in sulfur
deposition of 4.5 kg/ha from 2010 Pre-CAAA levels.
Compared to the base year, the 2010 Post-CAAA
nitrogen deposition estimate for Person County is 4.1
kg/ha lower than 1990 levels, the 2010 Post-CAAA
sulfur deposition prediction, however, is 12.9 kg/ha
higher.
"Under the 2010 Post-CAAA scenario the Mayo (large) and
Roxboro (very large) utility plants are predicted to emit 9,400 and 30,100
tons of NOS per year respectively.
C-70
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-29
Annual Sulfur Deposition
1990 Base Case Scenario
0 - 5 kg/ha
5-10 kg/ha
10-15 kg/ha
15-20 kg/ha
20- 100 kg/ha
C-71
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-30
Annual Nitrogen Deposition
1990 Base Case Scenario
0 - 5 kg/ha
5-10 kg/ha
10-15 kg/ha
15 - 20 kg/ha
20- 100 kg/ha
C-72
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-31
Annual Nitrogen Deposition
2000 Pre CAAA Scenario
100 0 100 200 300400 Miles-*'
0 - 5 kg/ha
5-10 kg/ha
10-15 kg/ha
15-20 kg/ha
20- 100 kg/ha
H
+
C-73
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-32
Annual Sulfur Deposition
2010 Post CAAA Scenario
100 0 100200300400500 Miles
0 - 5 kg/ha
5-10 kg/ha
10-15 kg/ha
15-20 kg/ha
20-100 kg/ha
C-74
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-33
Annual Nitrogen Deposition
2010 Pre CAAA Scenario
1DD LI 1DD2DD30D4DD Miles
0 - 5 kg/ha
5-10 kg/ha
10-15 kg/ha
15-20 kg/ha
20-100 kg/ha
C-75
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-34
Annual Nitrogen Deposition
2010 Post CAAA Scenario
0 - 5 kg/ha
5-10 kg/ha
10-15 kg/ha
15-20 kg/ha
20- 100 kg/ha
H
+
100 0 100200 Miles
C-76
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Estimating the Effects of the
CAAA on Sulfur Dioxide, Oxides
of Nitrogen, and Carbon
Monoxide
Future-year Pre- and Post-CAAA ambient SO2,
NO, NO2, and CO concentrations were estimated by
adjusting 1990 concentrations using future-year to
base-year emissions ratios. The methodology for
calculating and applying these ratios is described
below. The resulting future-year concentration also
are discussed in this section; histograms are used to
illustrate the relationship between Post- and Pre-
CAAA emissions estimates.
Methodology for Estimating Future-
Year SO2, NO, NO2, and CO
Concentrations
emissions for each grid cell was prepared. This was
done for each season for the 1990 base year and 2000
and 2010 Pre- and Post-CAAA scenarios.
Once the emissions inventory was prepared,
emission-based ratios for SO2, NO, NO2, and CO
were generated. For each RADM grid cell, adjustment
factors were calculated comparing future-year (2000
and 2010) emissions under each projection scenarios
to base-year (1990) emissions. Separate sets of ratios
were developed for each season.
Following the calculation of emission-based
ratios, future-year concentrations were then estimated
by applying these ratios to observed 1990 base-year
monitor concentrations. For REMSAD grid cells
without 1990 monitor concentration data
interpolation was used to estimate base-year
concentrations. Adjustment factors for the grid cell
were then applied to the interpolated values.
To estimate future-year SO2, NO, NO2, and CO
concentrations, adjustment factors were calculated
using grid cell specific REMSAD emissions data
(Douglas et al., 1999). REMSAD's domain
encompasses the 48 contiguous states and is divided
into 4,950 grid cells, each measuring approximately 56
km by 56 km. As part of the model's input, gridded
emission inventories containing seasonal Pre- and
Post-CAAA SO2, NO, NO2, and CO emissions
estimates were prepared. These same emissions
estimates used as REMSAD input in other parts of
this prospective analysis, were also used to calculate
SO2, NO, NO2, and CO adjustment factors.
Before emission-based ratios (adjustment factors)
were calculated, two separate inventories maintained
individually for REMSAD modeling purposes, one
containing elevated point source emissions data and
the other containing emissions data for low-level
sources, were combined. Each stack corresponding to
an elevated point source was assigned to a grid cell
based on location. Emissions from elevated point
sources were then added to the low-level emissions
corresponding to the grid cell in which the stack is
located. In this manner, a file containing total
C-77
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-35. Distribution of Monitor-Level Ratios of Summer
SO2 Emissions: 2010 Post-CAAA / 201 0 Pre-CAAA
60 -I
f 50-
(5
£ 40-
o
c
§ 30-
cr
^ 20-
0)
1 10-
ct
Q-\
median: 0.892
I —
0 40 0 45 0 50 0 55 0 60 0 65 0 70 0 75 0 80 0 85 0 90 0 95 1 00 1 05 1 10 1 15 1 20
Ratio
Note: 2.4 percent of the distribution of ratios is less than 0.40.
Figure C-36. Distribution of Monitor-Level Ratios of Summer
NO Emissions: 2010 Post-CAAA / 201 0 Pre-CAAA
60-
53"
§ 50~
(5
B 40-
o
c
§ 30-
cr
£ 20-
0)
^ 10-
0)
£
median: 0.666
I — I — i i — i i — 1 — 1
— i r i— i n L-I n
~K 1 1 -T-TTl l-n-l 1 1 1 r-l 1 „
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20
Ratio
Note: 3.3 percent of the distribution of ratios is less than 0.40.
C-78
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure C-37. Distribution of Monitor-Level Ratios of Summer
60 -
§ 50-
JU-
o
§ 30 -
cr
i 20 -
0)
« 10-
0)
£
o -I
0
NO2 Emissions: 201 0 Post-CAAA / 201 0 Pre-CAAA
median: 0.575
-H^_ ~hh .-r-LrTL-, r-.
hi— rTn fTli-i n
40 0 45 0 50 0 55 0 60 0 65 0 70 0 75 0 80 0 85 0 90 0 95 1 00 1 05 1 10 115 12
Ratio
D
Note: 2.7 percent of the distribution of ratios is less than 0.40.
Figure C-38. Distribution of Monitor-Level Ratios of Summer
CO Emissions: 2010 Post-CAAA / 2010 Pre-CAAA
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20
Ratio
Note: 15.7 percent of the distribution of ratios is less than 0.40.
C-79
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Emission-Based Ratios for SO2, NO,
NO2, and CO
Emission-based ratios were calculated for each
grid cell, regardless of whether or not the cell
contained a monitoring site. The figures included in
this section however, represent the distribution of
ratios for actual monitoring site locations. These
distributions reveal the relationship between future-
year and base-year concentrations. A ratios greater
that one indicates an increase in ambient
concentration from the base-year, while a ratio less
than one indicates a decrease.
Our results indicate that compared to the base-
year, future-year concentrations of SO2, NO, NO2,
and CO tend to increase under the Pre-CAAA
scenario, while Post-CAAA concentrations for all four
pollutants except SO2 tend to decrease. For example,
the median 2010 Pre-CAAA emission-based ratio for
SO2 is roughly 1.35, indicating an increase in median
2010 Pre-CAAA SO2 concentration of approximately
35 percent from the 1990 base-year. The median
ratios for NO, NO2, and CO are roughly 1.13, 1.17,
and 1.05 respectively. Under the Post-CAAA scenario
we estimate that in 2010 NO, NO2, and CO
concentrations will tend to be approximately 25 and
30 percent below base-year levels. The median 2010
Post-CAAA emission-based ratios for these three
pollutants are roughly 0.74, 0.70, and 0.76 respectively.
We estimate that SO2, concentrations, however, will
increase in many areas of the U.S. The median
adjustment ratio for this pollutant is approximately
1.21.
Comparison of the Pre- and Post-
CAAA Ratios
estimate that future-year concentrations of SO2, NO,
NO2, and CO are lower under the Post-CAAA
scenario than under the Pre-CAAA scenario.
Figures C-35 through C-38 show the distribution
of 2010 Post-CAAA to 2010 Pre-CAAA ratios for
summertime SO2, NO, NO2, and CO respectively.
These figures illustrate the regional variation in the
influence of the 1990 Amendments on ambient
concentrations of these pollutants. Although we
estimate concentrations in some areas will increase
under the Post-CAAA scenario relative to Pre-CAAA
estimates, the median summertime 2010 Post- to Pre-
CAAA ratios for SO2, NO, NO2, and CO are 0.90,
0.67, 0.58, and 0.72 respectively. These values, each
less than one, indicate that the central tendency for
summertime 2010 Post-CAAA concentration
estimates of these four pollutants is to be lower than
2010 Pre-CAAA estimates.
Table C-14 displays the median values of the
distribution of Post- to Pre-CAAA ratios for the
summer months described above and the remaining
three seasons. Just as for the summer; spring,
autumn, and winter median values are less than one.
Averaged over all four seasons, we estimate a median
reduction in SO2, NO, NO2, and CO concentrations
of approximately 9, 33, 40, and 25 percent
respectively. RACT requirements, tailpipe emissions
standards, and NOX emissions trading account for the
bulk of the reduction in NO and NO2 concentrations.
Title I nonattainment area controls and Title II motor
vehicle provisions are responsible for much of the
change in CO concentrations, while regulation of
utility and motor vehicle emissions account for
majority of the decrease in SO2 concentrations.
Comparison of Pre- and Post-CAAA emission-
based adjustment factors also helps illustrate the effect
of the 1990 Amendments on ambient pollution
concentrations. The ratio of 2010 Post-CAAA
adjustment factors to 2010 Pre-CAAA adjustment
factors shows the impact of the 1990 Amendments
on ambient concentrations relative to the baseline
scenario. Ratios less than one indicate that we
C-80
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table C-14
Median Values of the Distribution of Ratios of 2010 Post-CAAA/Pre-CAAA
Adjustment Factors
Spring
Summer
Autumn
Winter
SO2
0.904
0.892
0.916
0.924
NO
0.669
0.666
0.677
0.686
NO2
0.598
0.575
0.614
0.626
CO
0.790
0.720
0.756
0.692
Table C-15
Background Concentrations used to Prepare the SO2, NO, NO2, and CO Profiles
Pollutant
SO2
NO
NO2
CO
Background Concentration
0
0
0
0.2 ppm
Attributes and Limitations of the
Modeling Analysis Methodology
The Section 812 prospective modeling analysis
utilized a set of modeling tools and input databases
that for the most part had been developed, tested, and
evaluated as part other modeling studies (e.g., OTAG,
SIP modeling analyses, etc.). This provided a cost-
effective means of conducting a national-scale
modeling exercise. The models used for the study are
among the most widely used and evaluated tools for
ozone and PM modeling, and have been used for
previous regulatory applications. The modeling was
performed in manner that is consistent with
established practice and EPA guidelines regarding air
quality model applications.
Although appropriate techniques were used for
the analysis of each pollutant, use of separate
models/techniques for the analysis of ozone, PM, and
the other criteria pollutants does not allow a fully
integrated analysis of the effects of each.
Consequently, the results do not reflect all potential
interactions between pollutants (e.g., ozone and PM).
Ongoing research involving the development and
testing of integrated modeling tools (by EPA and
other organizations) may provide the opportunity for
fully integrated future Section 812 prospective
modeling efforts.
Analysis of the effects on the national scale (the
CAAA applies to the entire nation) required the use of
several different domains with varying grid resolution
as well as the use of relatively coarse resolution for
many areas of the country for the grid-based modeling
effort. The use of relatively coarse grid resolution (12
km and greater) is a potentially important source of
uncertainty with respect to the modeling results.
Previous studies have found that the response of the
UAM-V modeling system to emission reductions is
affected by grid resolution (Douglas et al., 1996).
Thus, the use of grid-cell specific adjustment factors
to modify site-specific data may introduce some
uncertainty into the future-year estimates.
There are always uncertainties associated with the
use of modeling results to estimate future-year air
quality. These derive from inaccuracies in the model
inputs and/or model formulation and were
manifested in this study in the evaluation of model
performance. While good model performance was
C-81
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
achieved for most model applications, ozone
concentrations were underestimated within the Los
Angeles domain and PM concentrations were
underestimated during the fall and winter simulation
periods in the REMSAD application. RADM/RPM,
used as part of the PM and visibility analyses, showed
a tendency to overestimate annual average sulfate
concentrations and warm season nitrate
concentrations. Annual average nitrate predictions
generated by RADM/RPM, however, matched air
quality monitor data.
The acid deposition estimates included in the
present analysis are limited to the eastern states within
the RADM domain. Deposition in the western U.S.
was not modeled for this study. Although acid
deposition is a problem primarily for the eastern U.S.,
deposition does occur in states west of the RADM
domain. The magnitude of the benefits of reducing
acid deposition in these western states is likely to be
small, however, relative to the overall benefits
associated with the Clean Air Act Amendments.
The approach used in this study to estimate future
air quality (the combined use of observed data and
modeling results) may, compared to a more standard
air quality model application (e.g., a model application
for attainment demonstration purposes), tend to
minimize the effects of many of the uncertainties
mentioned in this section. The reason for this is that
the modeling results are used in a relative sense, rather
than an absolute sense. This may enhance the
reliability of the future-year concentration estimates,
especially in the event that the uncertainty inherent in
the absolute concentration values is greater than that
associated with the response of the modeling system
to changes in emissions.
The ratios for adjusting the observed data are
calculated using modeling results for a limited number
of simulation days and it is assumed, using this
methodology, that the ratios can be used to represent
longer time periods. This approach permits the
estimation of seasonal and annual concentration
distributions. Nevertheless, the use of the model-
based ratios in adjusting data for an entire season or
year may result in some over- or underestimation of
the benefits of the simulated control measures,
depending upon whether the simulation results for the
modeled days are sufficiently representative of the
meteorological and air quality conditions that occurred
during 1990.
Finally, there are numerous ways in which the
adjustment factors could be calculated and applied.
The approach used in this study was designed to make
the best use of the information and level of detail
present in both the observations and the modeling
results (e.g., use of decile and quintile based ratios for
ozone and PM, respectively). The specific
assumptions employed in the application of the
methodology, however, may affect the resulting air
quality profiles and should be carefully considered in
the subsequent use and interpretation of the results.
Conclusions and
Recommendations for Further
Research
The results from the air quality modeling
component of the Section 812 prospective analysis
indicate that for both future years (2000 and 2010),
the measures and programs associated with the CAAA
are expected to result in lower concentrations of
ozone, PM, and the other criteria pollutants compared
to a future-year scenario without such programs. The
degree of improvement in air quality varies among the
criteria pollutants and the various portions of the
country included in the modeling analysis. The results
also differ between the two future years, such that the
improvements are greater and more widespread for
2010.
The modeling analysis relied on a set of modeling
tools and input databases that (for the most part) had
been developed, tested, and evaluated as part other
modeling studies. It also made use of several of the
most widely used and comprehensively tested tools
for ozone and PM modeling. The modeling was
performed in a manner that is consistent with
established practice and EPA guidelines regarding air
quality model applications. However, as noted in the
C-82
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
previous section of this report, there are several
features of the modeling analysis that could be
improved upon, especially considering recent advances
in the development of integrated modeling tools and
techniques. Recommendations for future air quality
modeling efforts to support the Section 812
prospective analyses include:
• Selection of modeling episode periods using
an integrated episode selection procedure
(e.g., Deuel and Douglas, 1998) such that the
modeling periods are representative of the
historical meteorological and air quality
conditions and can be used to represent
seasonal and annual ozone, PM, and visibility
metrics
• Reconfiguration of the modeling domain(s)
such that a consistent use of high-resolution
grids over urban areas with complex
meteorological or emissions-based features
are accommodated.
• Review and update of the input data and
input preparation techniques to include, for
example, updated (more recent) emissions
estimates (anthropogenic and biogenic),
higher-resolution meteorological inputs,
enhanced estimates of future land-use
patterns (reflecting growth of urban areas,
changes in the interstate transportation
networks, etc.).
• Use of an integrated modeling tool for the
simultaneous analysis of the effects of
emissions changes on ozone, PM, and other
pollutants (several tools, including
MODELS-3 and UAM-VPM, are currently
undergoing development and testing). A
comprehensive evaluation of model
performance will be required.
• Continued review and enhancement (as
appropriate) of the methodology for the
combined use of observed data and modeling
results.
C-83
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
BAAQMD. 1998. "Bay Area - North Central Coast Photochemical Modeling Investigation of Ozone
Formation and Transfer" available electronically at:
http://sparc2.baaqmd.gov/MontereyPri/MontereyPri.html.
Binkowski, F.S. and U. Shankar. 1995. "The Regional Particulate Matter Model 1. Model description and
preliminary results." /. of Geophysical Research. 100: 26,191-209.
Chang, J.S., R.A. Brost, I.S.A. Isaksen, S. Madromch, P. Middleton, W.R. Stockwell, and CJ. Walcek. 1987.
"A Three-Dimensional Eulerian Acid Deposition Model: Physical Concepts and Formulation." /. of
Geophysical Research. 92: 14,681-700.
Chang, J.S., P.B. Middleton, W.R. Stockwell, CJ. Walcek, J.E. Pleim, H.H. Lansford, F.S. Binkowski, S.
Madronich, N.L. Seaman, and D.R. Stauffer. 1990. "The Regional Acid Deposition Model and
Engineering Model". NAPAP SOS/T Report 4. In: Acidic Deposition: State of Science and Technology,
Volume 1. National Acid Precipitation Assessment Program, Washington D.C., December.
Dennis, R.L., W.R. Barchet, T.L. Clark, and S.K. Seilkop. 1990. "Evaluation of Regional Acid Deposition
Models (Part I)." NAPAP SOS/T Report 5. In: Acidic Deposition: State of Science and Technology, Volume
1. National Acid Precipitation Assessment Program, Washington, D.C., September.
Dennis, R.L., J.N. McHenry, W.R. Barchet, F.S. Binkowski, and D.W. Byun. 1993. "Correcting RADM's
sulfate underprediction: Discovery and correction of model errors and testing the corrections
through comparisons against field data." Atmospheric Environment. 27A(6): 975-997.
Dennis, R. 1995. "Estimation of Regional Air Quality and Deposition Changes Under Alternative 812
Emissions Scenarios Predicted by the Regional Acid Deposition Model, RADM." Draft Report for
U.S. Environmental Protection Agency. ORD/NERL. October.
Dennis, R.L. 1997. "Using the Regional Acid Deposition Model to Determine the Nitrogen Deposition
Airshed of the Chesapeake Bay Watershed." Atmospheric Deposition to the Great Lakes and Coastal
Waters. Ed. J.E. Baker, Society of Environmental Toxicology and Chemistry, Pennsacola, FL: 393-
413.
Deuel, H. P, N. K. Lolk, and S. G. Douglas. 1996. "Preparation of Air Quality and Other Inputs for
Application of the UAM-V Model for the 1-15 July 1988 OTAG Modeling Episode." Systems
Applications International, Inc., San Rafael, California (SYSAPP-96/51).
Deuel, H. P, and S. G. Douglas. 1998. "Episode Selection for the Integrated Analysis of Ozone, Visibility,
and Acid Deposition for the Southern Appalachian Mountains." Systems Applications International,
Inc., San Rafael, California (SYSAPP-98/07rl).
Douglas, S. G., J. L. Fieber, A. B. Hudischewskyj, and J. L. Haney. 1994. "Photochemical Modeling of the
Maricopa County Ozone Nonattainment Area." Systems Applications International, Inc., San
Rafael, California (SYSAPP-94/079).
C-84
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Douglas, S. G., N. K. Lolk, and J. L. Haney. 1996. "Investigation of the Effects of Horizontal Grid
Resolution on UAM-V Simulation Results for Three Urban Areas." Systems Application
International, Inc., San Rafael, California (SYSAPP-96/65).
Douglas, S.G., R. Iwamiya, G. Mansell, H. Tunggal, and C. van Landingham. 1999. "Air Quality Modeling
to Support the Section 812 Prospective Analysis". ICF Kaiser Consulting Group, Systems
Applications International, Inc., San Rafael, California (SYSAPP-99/09).
External Review Panel. 1994. "Report on RADM Evaluation for the Eulerian Model Evaluation Field Study
Program." Ed. Anton Eliassen. Norwegian Meteorological Institute, Norway.
Langstaff, J. E., S. Eberly, and K. A. McAuliffe. 1994. "Retrospective Analysis of SO2, NOX, and CO Air
Quality in the United States." Systems Application International, Inc., San Rafael, California.
Langstaff,]. E., and M. E. Woolfolk. 1995 "Retrospective Analysis of PM Air Quality in the United States."
Systems Application International, Inc., San Rafael, California.
McHenry,J.N. and R.L. Dennis. 1994. "The Relative Importance of Oxidation Pathways and Clouds to
Atmospheric Ambient Sulfate Production as Predicted by the Regional Acid Deposition Model."
Journal of AppM Meteorology. 33(7): 890-905.
OTAG. 1997. OTAG Regional and Urban Scale Modeling Workgroup. Modeling Report. Version 1.1."
Pechan. 1998. "Emission Projections for the Clean Air Act Section 812 Prospective Analysis." E. H.
Pechan and Associates.
Pitchford, M.L. and W.C. Malm. 1994. "Development and applications of a standard visual index."
Presented at the Conference on Visibility and Fine Particulates, Vienna, Austria. Atmospheric
Environment. (28): 1055-63.
SAL 1990. "User's Guide for the Urban Airshed Model, Volume I: User's Manual for UAM (CB-IV). U.S.
Environmental Protection Agency (SP-450/4-90-007b; NTIS No. PB91-13235).
SAL 1992. "User's Guide for the Urban Airshed Model, Volume IV: User's Manual for the Emissions
Preprocessor System 2.0". U.S. Environmental Protection Agency, (EPA-450/4-90-007D), May 27,
1992.
SCAQD. 1994. "Ozone Modeling - Performance Evaluation." Draft Technical Report V-B. South Coast
Air Quality Management District, Diamond Bar, California.
SCAQD. 1996. "1997 Air Quality Management Plan." South Coast Air Quality Management District,
Diamond Bar, California.
U.S. Environmental Protection Agency. 1991. "VOC/PM speciation Data Base Management System
(SPECIATE), Version 4.1." Office of Air Quality Planning and Standards, Research Triangle Park,
North Carolina (EPA-450/4-91-027).
C-85
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
U.S. Environmental Protection Agency. 1995. "Acid Deposition Standard Feasibility Study Report to
Congress." Office of Air and Radiation, Acid Rain Division, Washington, D.C. (EPA-430/R-95-
OOla).
U.S. Environmental Protection Agency. 1997. "Deposition of Air Pollutants to the Great Waters." Office of
Air Quality Planning and Standards, Research Triangle Park, NC (EPA-453/R-97-011).
C-86
-------
Human Health Effects
of Criteria Pollutants
Introduction
In response to the mandate of section 812 of the
Clean Air Act Amendments of 1990 (CAAA), EPA
identified and estimated the quantifiable health
benefits Americans should enjoy in 2000 and 2010 due
to improved air quality resulting from the CAAA. The
results suggest that the CAAA will result in significant
reductions in mortality, respiratory illness, heart
disease, and other adverse health effects, in addition to
those reported in EPA's (1997) retrospective analysis
of the Clean Air Act. In that analysis, the Agency
found that significant health benefits accrued between
1970 and 1990, especially as a consequence of the
reductions in ambient particulate matter (PA/1).
This appendix presents an overview of EPA's
approach for modeling the human health effects of
the CAAA. It outlines the principles used to guide the
human health benefits analysis, describes methods
used to quantify criteria air pollutant exposure
nationwide, and discusses issues that arise in using
health effect information. Following this overview,
the appendix presents the modeling results, reported
as estimates of avoided incidences of adverse health
effects.
Health Effects Modeling
Approach
In the section 812 retrospective analysis, EPA
(1997) developed an approach for quantifying the
effects of reduced pollutant exposure in the 48
continental states and the District of Columbia, with
particular focus on those effect categories for which
monetary benefits could be estimated. The study
design adopted for this analysis follows a similar
approach, using a sequence of linked analytical models
to estimate benefits. The first step is an analysis of the
likely implementation activities undertaken in
response to the CAAA. These forecasted activities
provided a basis for modeling criteria pollutant
emissions under the two scenarios considered (the
Pre-CAAA scenario and the Post-CAAA scenario), as
documented in Appendix A. The emissions estimates
were input into the air quality models (Appendix C),
and ambient pollutant concentrations estimated by the
air quality models were input into the health benefits
model, the focus of this appendix.
The health benefits model relies on two inputs:
(1) forecasted changes in pollutant exposures across
the study period, and (2) concentration-response (C-
R) functions that quantify the relationship between
the forecasted changes in exposure and expected
changes in specific health effects .We discuss the
inputs used for the 48 continental states and the
District of Columbia below.1
Quantifying Changes in Pollutant
Exposures
Quantifying changes in pollutant exposures in this
analysis relies on two inputs: (1) forecasts of ambient
pollution levels at the available air quality monitors in
the 48 contiguous states, and (2) extrapolations from
the available air quality monitors (which are not
uniformly distributed across the U.S.) to a population
grid system of eight km by eight km cells that covers
the 48 contiguous states and the District of Columbia.
1 These inputs could also be used to estimate exposure in the
border regions of Mexico and Canada that might have improved
air quality in the Post-CAAA scenario.
D-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Forecasting 2000 and 2010
Pollution Levels at Ambient Air
Quality Monitors
When quantifying adverse human health effects,
the section 812 prospective analysis estimated 2000
and 2010 ambient concentrations for both the Pre-
CAAA and Post-CAAA scenarios for the following
pollutants and air quality parameters:
• Particulate matter, less than 10 microns in
diameter (PM10)
• Particulate matter, less than 2.5 microns in
diameter (PM25)
• Ozone (O3)
• Nitrogen dioxide (NO2)
• Sulfur dioxide (SO2)
• Carbon monoxide (CO)
The sixth criteria pollutant, lead (Pb), is not
included in this analysis since airborne emissions of
lead were virtually eliminated by pre-1990 CAA
programs. The methods used to estimate the
concentrations of these pollutants at monitors are
described in Appendix C.
Extrapolating Forecasts at Air
Quality Monitors to Population Grid
Cells
The next step is to extend forecasts for a limited
number of air quality monitors to estimate population
exposure at all locations in the continental United
States, using the Criteria Air Pollutant Modeling
System (CAPMS). CAPMS divides the United States
into eight kilometer by eight kilometer grid cells and
estimates the changes in incidence of adverse health
effects associated with given changes in air quality in
each grid cell. The national incidence change (or the
changes within individual states or counties) is then
calculated as the sum of grid-cell-specific changes. To
calculate changes in population exposure in a grid cell,
CAPMS requires data on the population in the grid-
cell and the change in air quality.
First, grid-cell-specific population counts for 1990
are derived from U.S. Census Bureau block level
population data (Wessex, 1994). Future year grid-cell
population estimates are then extrapolated from 1990
grid-cell population levels using the ratio of future-
year and 1990 state-level population estimates
provided by the U.S. Bureau of Economic Analysis
(1995). CAPMS assumes that all grid cell populations
in a state grow at the same rate as the state population
as a whole (where a grid cell is defined as being in a
state if the grid cell centroid is in the state).
Second, CAPMS requires estimates of two air
quality regimes at CAPMS grid cell centers: baseline
(in this case, 1990) air quality levels and regulatory
alternative air quality levels in future years (in this case,
2000 and 2010). Air quality inputs to CAPMS for pre-
and Post-CAAA scenarios must use the averaging
time required by the C-R functions being used.2 For
example, a C-R function relating mortality to annual
median PM25 concentrations requires that annual
median PM25 concentrations be available at CAPMS
grid cell centers. Although the input PM25 data must
be in the form of daily averages, the monitors need
not be at CAPMS grid cell centers. Given any set of
location-specific air quality data, CAPMS interpolates
the corresponding air quality values at each CAPMS
grid cell center.
To reduce computational time when estimating
the change in health effects associated with daily
pollution levels, CAPMS approximates a year's (or
season's) worth of daily pollutant concentrations at
each monitor by 20 "bins" of pollutant
concentrations. Each bin represents five percent of
the daily pollutant concentrations in the period of
interest, and is set at the midpoint of the percentile
range it represents. For n = 20 and a year's worth of
observations, the first bin represents the first (lowest)
five percent of the distribution of 365 pollutant
concentrations at the given location, and is set at the
2.5th percentile value; the second bin represents the
next five percent of the distribution of daily values,
2The development of C-R functions is discussed later in this
appendix.
D-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
and is set at the 7.5th percentile value, and so on.
Each of the twenty bins therefore represents 18.25
(=365/20) days. Interpolation of air quality levels at
CAPMS grid cell centers is based on these input
location-specific bins, so that the annual incidence
changes in each grid cell are calculated for twenty
pollutant concentrations (the 20 bins of air quality)
rather than for 365 pollutant concentrations. The
resulting incidence change is then multiplied by 18.25
to reconstruct an entire years' worth of incidence
change in the CAPMS grid cell.
As shown in Figure 1, actual ambient pollution
data is only available from limited monitor sites. Data
must be extrapolated to unmonitored locations, in
order to estimate the impact of air pollution on the
health and welfare effects considered in this analysis.
The available air monitoring data were extrapolated
from all available monitor locations to a grid of eight
km by eight km population grid-cells throughout the
contiguous 48 states, using a Voronoi neighbor
averaging (VNA) spatial interpolation procedure.3
The VNA procedure interpolates air quality
estimates from the set of surrounding air quality
monitors to the center of each population grid-cell.
The VNA procedure is a generalization of planar
interpolation. Rather than arbitrarily limiting the
selection of monitors, VNA identifies the set of
monitors that best "surrounds" the center of each
grid-cell by identifying which monitor is closest
(considering both angular direction and horizontal
distance) in each direction from the grid-cell center.
Each selected monitor will likely be the closest
monitor for multiple directions. The set of monitors
found using this approach forms a polygon around the
grid-cell center.
For each grid cell, CAPMS calculates the distance
to each member of a set of monitors surrounding that
grid cell. Monitors close to the grid cell are assumed
to yield a more accurate air quality description of that
grid cell, and are given a larger weight when calculating
the average air quality for that grid cell. Conversely,
monitors that are further away receive less weight.
After determining the final set of surrounding
monitors, the grid cell's air quality level is calculated as
an inverse, distance-weighted average of the air quality
levels at the selected monitors.
Air quality estimates generated using this VNA
method are likely to be most uncertain at population
grid cell locations far removed from the nearest
monitor. For example, if a grid cell encompasses a
relatively unpolluted rural area, but the nearest (albeit
distant) monitors are measuring air quality in
industrialized urban areas, the VNA method described
above will overestimate the pollution level for that
grid cell. As a result, this monitor-based VNA
extrapolation method is used only at grid cells located
within 50 kilometers of an air pollution monitor.
At distances greater than 50 kilometers from a
monitor, additional information is needed to improve
the estimates of air quality in unmonitored areas. A
modified VNA method incorporating both monitor
data and air quality modeling predictions is employed
at these grid cell locations. In addition to the
distance-weighted averaging of monitor
concentrations, this modified extrapolation method
incorporates a spatial adjustment factor that reflects
the ratio of model-derived air quality predictions at
the target and source locations. The addition of the
modeling results helps account for differences in
geography, meteorology, land use and other factors
affecting air pollution levels between the target and
source areas. Additional details on both VNA
extrapolation methods can be found in Abt Associates
(1999).
3For locations within 50 kilometers of a monitor, the
interpolation method is the same as that used by Abt Associates
(1998) for the NOX SIP call analysis; previously termed the
"convex polygon" method, it is more accurately described as
Voronoi neighbor averaging (VNA) spatial interpolation, which
will be used throughout this document.
D-3
-------
841 Ozone Monitors
761 SQ2 Monitors
Figure D-l
Location of Air Quality Monitors
Section 812 Analysis
2048PM Monitors
348 NO2 Monitors
494 CO Monitors
D-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Quantifying Human Health Effects of
Exposure
To calculate point estimates of the changes in
incidence of a given selection of adverse health and
welfare effects associated with a given set of air quality
changes, CAPMS performs the following steps for
each grid cell: (1) Interpolation of the air quality in the
baseline scenario and in the control scenario at each
CAPMS grid cell center for each pollutant. (2)
Calculation of the changes in air quality from baseline
to control scenario in the CAPMS grid cell. The
changes in air quality are calculated as the differences
between the baseline bins and the corresponding
control scenario bins. (3) Identification of the selected
C-R functions being used, and the required baseline
incidence rates and the relevant grid cell population.
(4) Calculation of the change in incidence of each
adverse health effect for which a C-R function has
been identified. The resulting annual incidence change
for each grid cell is then summed with those of the
other grid cells, to calculate the estimated change in
incidence nationwide.
Types of Health Studies
Research on the health effects of air pollution
strongly suggests that reductions in the incidence of
adverse health effects are a significant benefit of air
pollution control. The available human health studies
that could serve as the basis of the section 812
prospective assessment can be categorized into
chamber studies and epidemiology studies. Chamber
studies involve examination of human responses to
controlled conditions in a laboratory setting, while
epidemiological studies investigate the association
between exposure to ambient air pollution and
observed health effects in a study population. The
relative advantages of reliance on each type of research
are described below.
Chamber Studies
Chamber studies of air pollution involve exposing
human subjects to various levels of air pollution in a
carefully controlled and monitored laboratory
situation. The physical condition of the subjects is
measured before, during and after the exposure.
These measurements can include general biomedical
information (e.g., pulse rate and blood pressure),
physiological effects specifically induced by the
pollutant (e.g., altered lung function), the onset of
symptoms (e.g., wheezing or chest pain), or the ability
of the individual to perform specific physical or
cognitive tasks (e.g., maximum sustainable speed on a
treadmill). These studies often involve exposing the
individuals to pollutants while exercising, which
increases respiration and the amount of pollution
introduced into the lungs.
One advantage of chamber studies is that they can
potentially establish cause-effect relationships between
pollutants and certain human health effects. In
addition, repeated experiments altering the pollutant
level, exercise regime, and type of participants can
potentially identify effect thresholds, the impact of
recovery (rest) periods, and the differences in
response among population groups. While cost
considerations tend to limit the number of
participants and experimental variants examined in a
single study, chamber studies can follow rigorous
laboratory scientific protocols, such as the use of
placebos (clean air) to establish a baseline level of
effects and precise measurement of certain health
effects of concern.
There are drawbacks to using chamber studies as
the basis for a comprehensive benefits analysis.
Chamber studies are most appropriate for examining
acute symptoms caused by exposure to a pollutant for
a few hours. While this permits examination of some
important health effects from air pollution, such as
broncho-constriction in asthmatic individuals caused
by sulfur dioxide, it precludes studying effects caused
by long term exposure. Another drawback is that
health effects measured in some well-designed
chamber studies are selected on the basis of the ability
to measure precisely an effect, for example forced
expiratory volume, rather than a larger symptom.
Some of these measurable but relatively minor health
effects, such as reduced lung function, have an unclear
impact on future medical condition and lifestyle,
although some research discussed later has addressed
this question.
Ethical considerations in experiments involving
humans also impose important limits on the potential
scope of chamber research. Chronic effects cannot be
D-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
investigated because people cannot be kept in
controlled conditions for an extended period of time,
and because chronic effects are generally irreversible.
Participation is generally restricted to healthy subjects,
or at least excludes people with existing health
conditions that compromise their safe inclusion in the
study. This can result in a lack of direct evidence
about populations of most concern, such as people
who already have serious respiratory diseases. Ethical
considerations also limit experimental pollutant
concentrations to relatively modest exposure levels,
and confine studies to examining only mild health
effects that are believed to do no permanent damage.
Obviously, for ethical reasons, evidence from chamber
studies cannot be obtained on the possible
relationship between pollution and mortality, heart
attack, strokes, or cancer.
The information derived from chamber research
raises some questions as to how well it applies to the
general population and their activity patterns and
pollution exposures. (1) The dose-response functions
developed from chamber research are specific to the
population participating in the study. Chamber
studies typically study a small population — certainly
much smaller than those typically evaluated in
epidemiological studies (discussed below) — so there
are concerns that the results may not apply to the
much larger and likely more diverse general
population. (2) Chamber studies evaluate only a
certain number of activity patterns (e.g., exercise), and
cannot replicate the diverse pattern of activity seen in
the course of a day. (3) Chamber studies cannot easily
replicate the varied pollution levels to which people
are exposed during the course of their day at work, on
the freeway, at home, and other places.
As discussed below in the section on health
effects study selection, the generalizability of results is
an important factor in this analysis. Studies that use a
large, diverse group of subjects are easier to apply to
the general population than studies using smaller,
narrowly defined group of subjects. This does not,
however, rule out studies that focus on asthmatics,
children, or the elderly, since these groups may be
particularly sensitive to air pollution. Similarly, studies
that use exposure regimes and exercise levels similar to
what large groups of the population experience are
easier to apply in a benefits model than are less
representative studies.
Epidemiological Studies
Epidemiological studies present the results of a
statistical analysis of the relationship between ambient
pollution exposure and adverse health effects. The
data for these studies includes ambient air quality
monitoring data and adverse health effects data such
as mortality incidence (e.g., National Center for
Health Statistics, 1994), hospital admissions (e.g.,
Graves and Gillum, 1997), questionnaires (e.g., Adams
and Marano, 1995), and diaries that are kept by study
participants over a period of time (e.g., Ostro et al.,
1991).
At least to some extent, these estimated
relationships implicitly take into account complex real-
world human activity patterns (including actions to
avoid air pollution), spatial and temporal differences
in air pollution distributions, and possible synergistic
effects of multiple pollutants. Epidemiological studies
typically involve a large number of people and may
not suffer as much from the extrapolation problems
common to chamber studies, which often have a
limited number of subjects. In addition, observable
health endpoints are measured, unlike chamber
studies, which often monitor endpoints that do not
result in observable health effects (e.g. forced
expiratory volume).
Two types of epidemiological studies are
considered for inclusion in this analysis: individual-
level cohort studies and population-level ecological
studies. Cohort-based studies track individuals over
a certain period of time, with periodic evaluation of
the individuals' exposure and health status. Cohort
studies can either follow a group of initially disease-
free individuals forward in time (a prospective cohort)
or gather historical data on exposure and disease for
a given group (a retrospective cohort). Studies about
relatively rare events such as cancer incidence or
mortality can require tracking the individuals over a
long period of time, while more common events (e.g.,
respiratory symptoms) occur with sufficient frequency
to evaluate the relationship over a shorter time period.
An important feature of cohort studies is that
information is collected about each individual that
may include other variables that could be correlated
with both the exposure and the disease outcome, such
as smoking or income. If investigators do not identify
and control for these variables, called confounders, in
D-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
a study, they may produce a spurious association
between air pollution and adverse health effects.
A second type of study used in this analysis is a
population-level ecological study. These studies assess
the relationship between population-wide health
information (such as counts for daily mortality,
hospital admissions, or emergency room visits) and
ambient levels of air pollution. There are two types of
such studies: cross-sectional and time-series studies.
Using data at a point in time from a variety of
locations, cross-sectional studies examine the
relationship between pollution exposure and adverse
health effects, while trying to control confounding
variables. Cross-sectional studies are not as desirable
as prospective cohort studies, in part, because of their
failure to control for important covariates such as
smoking.4 Rather than look at variety of locations at
one point in time, a time series analysis studies a single
location and typically examines the relationship
between daily changes in ambient pollution level and
daily changes in adverse health effects. An important
advantage of the time-series design is that it allows the
population to serve as its own control with regard to
certain factors such as race and gender, and is thus
similar to a cohort study (Schwartz, 1997, p. 372).
Other factors that change over time can also affect
health (tobacco, alcohol and illicit drug use, access to
health care, employment, and nutrition). However,
since such potential confounding factors are unlikely
to vary from day to day in the same manner as air
pollution levels, these factors are unlikely to affect the
magnitude of the association between air pollution
and daily variations in human health responses.
Drawbacks to epidemiological studies include
difficulties associated with adequately characterizing
exposure to individuals (that tends to lead to a
downward bias in the estimated pollution-health effect
relationship), and uncontrolled confounding variables,
that can potentially lead to spurious conclusions. In
particular, air pollutants are often highly correlated, so
it is difficult to determine which may be associated
with an adverse effect. In addition, epidemiological
studies, by design, are unable to definitively prove a
causal relationship between an exposure and a given
4Criticisms of cross-sectional studies are considered in Evans
et al. (1984), Lipfert and Wyzga (1995), and others.
health effect; they can only identify associations or
correlations between exposure and the health
outcome. However, given the major advantage of
epidemiological studies — relatively severe health
effects may be observed in a large, more
heterogeneous population — epidemiological studies
are used as the basis for determining the majority of
health effects and C-R functions in this analysis.
Chamber studies are used if there are health effects
observed in chamber studies not observed in
epidemiological studies, such as shortness of breath in
young asthmatics induced by SO2 (e.g., Linn et al.,
1987).
Selection of C-R Functions
This section describes the methods used to derive
the C-R functions used in this analysis to quantify the
effect of CO, NO2, SO2, O3, and PM on people's
health. It discusses the general issues that arise with
the choice and use of C-R functions, and the issues
specific to C-R functions for mortality and morbidity.
C-R Function General Issues
Derivation of C-R Functions
For expository simplicity, the following discussion
focuses on PM C-R functions, although it applies to
all of the health effects and pollutants considered in
the 812 prospective analysis. In what follows, the
health effect estimated is simply denoted as y, and is
estimated at a single location (population cell), where
a change in PM air quality (APM) corresponds to a
change in the health endpoint (Ay). The calculation of
Ay depends on a C-R function, derived typically from
an epidemiological study.
There are a variety of epidemiological studies in
the science literature, making it important to
understand the nuances of each study before
developing a C-R function. Different epidemiological
studies may have estimated the relationship between
PM and a particular health endpoint in different
locations. The C-R functions estimated by these
studies may differ from each other in several ways.
They may have different functional forms; they may
have measured PM concentrations in different ways;
D-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
they may have characterized the health endpoint, y, in
slightly different ways; or they may have considered
different types of populations. Some studies have
assumed that the relationship between y and PM is
best described by a linear form, where the relationship
between y and PM is estimated by a linear regression
in which y is the dependent variable and PM is one of
several independent variables, while other studies have
assumed that the relationship is best described by a
log-linear form (i.e., the relationship between the
natural logarithm of y and PM is estimated by a linear
regression).5 Some studies of the relationship between
ambient PM concentrations and mortality have
excluded accidental deaths from their mortality
counts; others have included all deaths. One study
may have measured daily (24-hour) average PM
concentrations, while another study may have used
two-day averages. Finally, one study may have
considered changes in the health endpoint only among
members of a particular subgroup of the population
(e.g., individuals 65 and older), while other studies may
have considered the entire population in the study
location.
Estimating the relationship between PM and a
health endpoint, y, consists of two steps: (1) choosing
a functional form of the relationship, and (2)
estimating the values of the parameters in the function
assumed. The two most common functional forms in
the epidemiological literature on health effects are the
log-linear and the linear relationship. The log-linear
relationship is of the form:
(1)
or, equivalently,
(2)
where the parameter B is the incidence of y when the
concentration of PM is zero, the parameter [3 is the
coefficient of PM, ln(y) is the natural logarithm of y,
and a = ln(B).6 If the functional form of the C-R
relationship is log-linear, the relationship between
APM (= PMbasdme - PMafteIchmg?!) and Ay is:
Ay = y - ya
(3)
where y is the baseline incidence of the health effect
(i.e., the incidence before the change in PM). For a
log-linear C-R function, the relative risk (RR)
associated with the change in PM is:
J after change
RR
APM
y
(4)
Epidemiological studies often report a relative risk
for a given APM, rather than the C-R coefficient, [3.
The coefficient can be derived from the reported
relative risk and APM, however, by solving for [3 in
equation (4):
ln(M)
APM
(5)
The linear relationship is of the form:
y = a + j8 • PM , (6)
5The log-linear form used in the epidemiological literature on
ozone- and PM-related health effects is often referred to as
"Poisson regression" because the underlying dependent variable is
a count (e.g., number of deaths), believed to be Poisson
distributed. The model may be estimated by regression techniques
but is often estimated by maximum likelihood techniques. The
form of the model, however, is still log-linear.
Other covariates besides pollution clearly affect mortality.
The parameter B might be thought of as containing these other
covariates, for example, evaluated at their means. That is, B =
B0exp{(3jXj + ... + |3nxn}, where B0 is the incidence of y when all
covariates in the model are zero, and xb ... , x,, are the other
covariates evaluated at their mean values. The parameter B drops
out of the model, however, when changes in y are calculated, and
is therefore not important.
D-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
where a incorporates all the other independent
variables in the regression (evaluated at their mean
values, for example) times their respective coefficients.
When the C-R function is linear, the relationship
between a relative risk and the coefficient, [3, is not
quite as straightforward as it is when the function is
log-linear. Studies using linear functions usually report
the coefficient directly.
If the functional form of the C-R relationship is
linear, the relationship between APM and Ay is simply:
Aj; = /3 • APM .
(7)
If the C-R function is linear, equation (7) may be
used to estimate Ay associated with APM, assuming
the measurement of APM is consistent with the PM
measurement used in the health effects study from
which the C-R function was derived. If the function
is log-linear, the baseline incidence for y and an
appropriate measure for APM may be used in
equation (3).
A few epidemiological studies, estimating the
relationship between certain morbidity endpoints and
air pollution, have used functional forms other than
linear or log-linear forms. Of these, logistic
regressions are the most common. The details of the
models used in these studies are given in the papers
reporting the methods and results of the studies.
Thresholds
When conducting chamber and epidemiological
studies, C-R functions may be estimated with and
without explicit thresholds. Air pollution levels below
the threshold are assumed to have no associated
adverse health effects. When a threshold is not
assumed, as is often the case in epidemiological work,
any exposure level is assumed to pose a non-zero risk
of response to at least one segment of the population.
Thresholds may also be incorporated by a policy
analyst using a C-R function derived from the original
study, even if the original study did not assume a
threshold. A threshold may be set at any point,
although some points may be considered more
obvious candidates than others. One possible
assumption is that there is a threshold at the non-
anthropogenic background level of the pollutant.
Another possibility is there is a threshold at the lowest
observed level in the study that estimated the C-R
function. Another might be a relevant standard for
the pollutant.
One method to conduct policy analysis assuming
a threshold model is to simply truncate the C-R
function at the threshold (i.e., to exclude any physical
effect changes associated with, say, PM levels below
the designated threshold). This method uses the
original C-R function, but calculates the change in PM
as [max(T, baseline PM) - max(T, regulatory
alternative PM)], where T denotes the designated
threshold. An alternative method is to replace the
original C-R function with a "hockey stick" model
that best approximates the original function that was
estimated using actual data. A typical hockey stick C-
R function is horizontal up to a designated threshold
PM level, T, and is linear with a positive slope for PM
concentrations greater than T. This is just the
following variation on equation (2) above:
ln(j;) = a for PM < T , (8)
= a + /3-PM forPM > T, wherefi>0. (9)
The specification of such a 'hockey stick' model,
while theoretically preferable to a simple truncation
model, requires re-analysis of the underlying data from
the original health effect study. Such primary re-
analysis is beyond the scope of the section 812
analysis. Alternatively, if a simple truncation model is
used, application of the resulting C-R function would
likely result in a significant underestimate of the health
effects avoided by reductions in pollutant exposures
above the assumed threshold.
The possible existence of an effect threshold is a
very important scientific question and issue for policy
analyses such as the section 812 analysis. However,
there is currently no scientific basis for selecting a
particular threshold for the effects considered in this
D-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
analysis, if a threshold is defined as a level
characterized by an absence of observable effects.
Therefore, this analysis assumes there are no
thresholds for modeling health effects. However, the
potential impact of alternative threshold assumptions
for PM-related premature mortality is explored as a
key sensitivity analysis.
Pooling Studies
When only a single study has estimated the C-R
relationship between a pollutant and a given health
endpoint, the estimation of a population cell-specific
incidence change is straightforward. For some
endpoints, however, C-R functions have been
estimated by several studies, often in several locations.
In this case, if the input components (e.g., functional
forms, pollutant averaging times, study populations)
are all the same (or very similar), a pooled, "central
tendency" C-R function can be derived from the
multiple study-specific C-R functions.
One potential method of pooled analysis is simply
averaging the coefficients from all the studies. This
has the advantage of simplicity, but the disadvantage
of not taking into account the measured uncertainty of
each of the estimates. Estimates with great
uncertainty surrounding them are given the same
weight as estimates with very little uncertainty.
An alternative approach to pooling the estimates
from different studies is to give more weight to
estimates from studies with little reported uncertainty
than to estimates with a great deal of uncertainty. The
exact way in which weights are assigned to estimates
of PM coefficients from different studies in a pooled
analysis depends on the underlying assumption about
how the different estimates are related to each other.
If, for example, there is actually a distribution of true
effect coefficients, or [3's, that differ by location
(referred to as the random effects model), the
different coefficients reported by different studies may
be estimates of different underlying coefficients, rather
than just different estimates of the same coefficient.
In contrast to the fixed effects model (which assumes
that there is only one [3 everywhere), the random-
effects model allows the possibility that different
studies are estimating different parameters.
A third approach to pooling studies is to apply
subjective weights to the studies, rather than
conducting a random effects pooling analysis. If the
analyst is aware of specific strengths and weaknesses
of the studies involved, this prior information may be
used as input to the calculation of weights which
reflect the relative reliability of the estimates from the
studies.
In some cases, studies reported several estimates
of the C-R coefficient, each corresponding to a
particular year or particular study area. For example,
Ostro and Rothschild (1989b) report six separate
regression coefficients that correspond to regression
models run for six separate years. This analysis
combined the individual estimates using a meta-
analysis on the six years of data.
Pollution Exposure Measure
The study on which an acute exposure C-R
function is based may have used pollution
concentrations averaged over several days. Those
studies that use multi-day averages are in effect using
a smoothed data set, comparing each day's adverse
health effects to recent average exposure rather than
simply exposure on the same day. This does not have
much effect on the estimated adverse health effects,
especially when the C-R function has a linear or nearly
linear functional form. For example, if the functional
form were linear and based on a five-day pollution
average, then the estimated effects over the course of
the year would be essentially the same between using
daily pollution observations in the C-R function or a
two-day average. This is analogous to summing up
five numbers (6,4,8,4,8=30) or taking their average
and multiplying by five (6*5=30); in each case the
answer is 30. This analysis uses daily pollution levels
in cases where there are multi-day averaging times.
D-10
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Regional Variation
Whether the C-R relationship between a pollutant
and a given health endpoint is estimated by a single
function from a single study or by a pooled function
of C-R functions from several studies, that same C-R
relationship is applied everywhere in the current
benefits analysis. Although the C-R relationship may
in fact vary somewhat from one location to another
(for example, due to differences in population
susceptibilities or differences in the composition of
PM), location-specific C-R functions are available only
for those locations in which studies were conducted.
A single function applied everywhere may result in
overestimates of incidence changes in some locations
and underestimates of incidence changes in other
locations. It is not possible, however, to know the
extent or direction of the overall bias in the total
incidence change introduced by application of a single
C-R function everywhere.
PM Size and Composition
Current research suggests that particle size, and
perhaps particle composition, matters when estimating
the health impacts of PM. Particulate matter is a
heterogeneous mix that varies over time and space,
and may include solid or liquid compounds, including
organic aerosols, sulfates, nitrates, metals, elemental
carbon, and other material. Fine PM is generally
viewed as having a more harmful impact than coarse
PM, although it is not clear to what extent this may
differ by the type of health effect or the exposed
population. While one cannot necessarily assume that
coarse PM has no adverse impact on health, it seems
reasonable to prefer the use of PM2 5 as a proxy for
the impact of PM. Due to the relative abundance of
studies using PM10, however, and the reasonably good
correlation between PM25 and PM10 in urban areas, in
many cases this analysis also uses PM10 studies to
estimate the impact of PM. Similarly, at this stage of
knowledge, it is not clear what composition
distinctions to make, if any, when estimating the
impact of PM. The C-R functions used in this analysis
relate adverse health effects to an undifferentiated
mass of particles (e.g., PM10); they do not relate effects
to individual PM components.
Baseline Incidence Rate
Some C-R functions (those expressed as a change
relative to baseline conditions) require baseline
incidence data associated with ambient levels of
pollutants. Baseline incidence data necessary for the
calculation of risk and benefits were obtained from
national sources whenever possible, because these
data are most applicable to a national assessment of
benefits. County-specific estimates of baseline
mortality incidences used in this analysis were
obtained from the National Center for Health
Statistics (1994). The National Center for Health
Statistics also provided much of the information on
national incidence rates. However, for some studies,
the only available baseline incidence information
comes from the studies themselves; in these cases, the
baseline incidence in the study population is assumed
to represent the baseline incidence nationally.
Population
Many studies focus on a particular age cohort.
The age group chosen is often a matter of
convenience (e.g., extensive Medicare data may be
available for the elderly population) and not because
the effects are necessarily restricted to the specific age
group, even though their incidence may vary
considerably over an individual's life span.
Nevertheless, to avoid overestimating the benefits of
reduced pollution levels, this analysis applies the given
C-R relationships only to those age groups
corresponding to the cohorts studied. Likewise, some
studies were performed on individuals with specific
occupations, activity patterns, or medical conditions
because these traits relate to the likelihood of effect,
such as in the estimation of worker productivity. In
these cases, application of dose-response functions
has been restricted to populations of individuals with
these same characteristics. As discussed in more detail
below, however, by assuming that the C-R
relationships should only be applied to those
subpopulations matching the original study
population, the present analysis may be significantly
underestimating the whole population benefits of
reductions in pollutant exposures.
D-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
C-R Function Selection Criteria
A number of considerations arose in selecting and
applying concentration-response (C-R) functions for
the section 812 prospective assessment. These
considerations are summarized in Table D-l below.
Because concentration-response functions are the
means of relating changes in pollutant levels to
changes in health endpoints, they are a critical
component of a benefits analysis. While a study may
be superior with regard to one consideration (e.g.,
number of pollutants considered), it may be inferior
with regard to another consideration (e.g., number of
observations). The selection of C-R functions for the
benefits analysis was guided by the goal of achieving
a balance between comprehensiveness and scientific
defensibility. The issues considered are discussed
below in some detail.
D-12
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table D-1
Summary of Considerations Used in Selecting C-R Functions
Consideration
Comments
Peer reviewed
research
Peer reviewed research is preferred to research that has not undergone the peer review
process.
Study type Among studies that consider chronic exposure (e.g., over a year or longer) prospective
cohort studies are preferred over cross-sectional studies (a.k.a. "ecological studies")
because they control for important confounding variables that cannot be controlled for in
cross-sectional studies. If the chronic effects of a pollutant are considered more
important than its acute effects, prospective cohort studies may also be preferable to
longitudinal time series studies because the latter type of study is typically designed to
detect the effects of short-term (e.g. daily) exposures, rather than chronic exposures.
Study period Studies examining a relatively longer period of time (and therefore having more data)
are preferred, because they have greater statistical power to detect effects. More
recent studies are also preferred because of possible changes in pollution mixes,
medical care, and life style overtime.
Study population Studies examining a relatively large sample are preferred. Studies of narrow population
groups are generally disfavored, although this does not exclude the possibility of
studying populations that are potentially more sensitive to pollutants (e.g., asthmatics,
children, elderly). However, there are tradeoffs to comprehensiveness of study
population. Selecting a C-R function from a study that considered all ages will avoid
omitting the benefits associated with any population age category. However, if the age
distribution of a study population from an "all population" study is different from the age
distribution in the assessment population, and if pollutant effects vary by age, then bias
can be introduced into the benefits analysis.
Study location U.S. studies are more desirable than non-U.S. studies because of potential differences
in pollution characteristics, exposure patterns, medical care system, and life style.
Pollutants included in
model
Models with more pollutants are generally preferred to models with fewer pollutants,
though careful attention must be paid to potential collinearity between pollutants.
Because PM has been acknowledged to be an important and pervasive pollutant,
models that include some measure of PM are highly preferred to those that do not.
Measure of PM PM25 and PM10 are preferred to other measures of particulate matter, such as total
suspended particulate matter (TSP), coefficient of haze (COH), or black smoke (BS)
based on evidence that PM25 and PM10 are more directly correlated with adverse health
effects than are these more general measures of PM.
Economically
valuable health
effects
Some health effects, such as forced expiratory volume and other technical
measurements of lung functioning, are difficult to value in monetary terms. These health
effects are not quantified in this analysis.
Non-overlapping
endpoints
Although the benefits associated with each individual health endpoint may be analyzed
separately, care must be exercised in selecting health endpoints to include in the overall
benefits analysis because of the possibility of double counting of benefits. Including
emergency room visits in a benefits analysis that already considers hospital admissions,
for example, will result in double counting of some benefits if the category "hospital
admissions" includes emergency room visits.
D-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Peer-Review of Research
Whenever possible, peer-reviewed research rather
than unpublished information has been used.
Research that has been reviewed by the EPA's own
peer review processes, such as review by the Clean Air
Scientific Advisory Committee (CASAC) of the
Science Advisory Board (SAB), has been used
whenever possible. Research reviewed by other public
scientific peer review processes, such as the National
Academy of Science, the National Acidic Precipitation
Assessment Program, and the Health Effects Institute
is also included in this category.
Studies published (or accepted for publication) in
peer reviewed journals but not reviewed by CASAC
have also been considered for use in the section 812
prospective assessment, and have been used if they are
determined to be the most appropriate available
studies. Indications that EPA intends to submit
research to the CASAC (such as inclusion in a draft
Criteria Document or Staff Paper) are considered
further evidence that specific journal-published
research is acceptable for use in this analysis.
Air pollution health research is a very active field
of scientific inquiry, and new results are being
produced regularly. Many research findings are first
released in university working papers, dissertations,
government reports, non-reviewed journals and
conference proceedings. Some research is often
published in abstract form in journals, which does not
require peer review. In order to use the most recent
research findings and be as comprehensive as possible,
unpublished research was examined for possible use.
Study Type and Quality
Epidemiological studies of the relationship
between air pollutants and health endpoints can
generally be categorized as (1) "ecological" cross-
sectional, (2) prospective cohort, or (3) longitudinal
time series studies. The first two types of study are
concerned with longer exposure periods, such as a
year or over several years, while the third type is
concerned with short-term exposures over one or
more days. Among studies that consider longer
exposure periods, or chronic exposure, prospective
cohort studies are preferable to "ecological" cross-
sectional studies, because they control for important
confounding variables which cannot be controlled for
in "ecological" cross-sectional studies. If the effects
of chronic exposures are considered more significant
than acute effects, prospective cohort studies may also
be preferable to longitudinal time series studies
because the latter type of study is typically designed to
detect the effects only of daily exposures, rather than
chronic exposures.
Studies that control for a broad range of likely
confounders can offer a more robust conclusion
about an individual pollutant, even if the statistical
confidence interval is larger due to the inclusion of
more variables in the analysis. For example, a study
that considers only air pollution, omitting other
variables associated with a health outcome, could
incorrectly conclude that a reduction in air pollution
is exclusively responsible for a reduction in the health
outcome. Potential confounders include weather-
related variables (e.g., temperature) and ambient
pollutants other than those being studied.
Study Population
Many of the studies relevant to quantifying the
benefits of air pollution reductions have focused on
subpopulations that may or may not be representative
of the general population. Extrapolating results from
studies on nonrepresentative subpopulations to the
general population introduces uncertainties into the
analysis, but the magnitude of the uncertainty and its
direction are often unknown. Because of these
uncertainties, benefit analyses often limit the
application of the C-R functions only to those
subpopulations with the characteristics of the study
population. While this approach has merit in
minimizing uncertainty, it can result in a severe
underestimate of benefits if similar effects are likely to
occur in the rest of the population. For these reasons,
studies that examine broad, representative populations
may be preferable to studies with narrower scope,
because they allow application of the functions to
larger numbers of persons. There are, however,
tradeoffs to comprehensiveness of study population.
D-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Selecting a C-R function from a study that considered
all ages will avoid omitting the benefits associated with
any population age category. However, if the age
distribution of the study population from an "all
population" study is different from the age
distribution in the assessment population, and if
pollutant effects vary by age, then bias can be
introduced into the benefits analysis.
Study Period
Studies examining a relatively longer period of
time are preferable because they have more data and
therefore have greater statistical power to detect
effects. In addition, more recent studies are preferable
to older studies because of possible changes in
pollution mixes, medical care, and life style over time.
This latter issue is effectively a benefits transfer issue.
Differences across time between the study period and
the assessment period introduce uncertainties into the
benefits analysis, because it is not known to what
extent the C-R relationship estimated during the study
period will be the same during the assessment period.
Study Location
Studies conducted in locations that are different
from the assessment location are generally less
desirable because of the introduction of possible
benefits transfer problems. The characteristics of a
population (e.g., the proportion of the population that
is particularly susceptible to pollution, or the behavior
patterns of the population) and/or the pollution mix
to which it is exposed may differ notably between the
study location and the assessment location. As with
differences in time periods, these differences in
location introduce uncertainties into the benefits
analysis, because it is not known to what extent the C-
R relationship estimated in the study location is the
same in the assessment location. For that reason,
studies conducted in the United States or Canada are
preferable for this benefits analysis to studies
conducted, for example, in Europe or in developing
countries. In addition, studies that include a wide
range of areas are preferred. Studies focusing on a
single city are not as desirable as studies that focus on
multiple cities.
The preference for studies that focus on a range
of areas, in the U.S. and Canada, is driven by a
concern that there may be significant regional
variation in the estimated C-R functions. There has
not, however, been enough research to establish
regional specific values.
Pollutants Included in the Model
In many cases, several pollutants in a "pollutant
mix" are correlated with each other — that is, their
concentrations tend to change together. Although
there may be an association between an adverse health
effect and this mix, it may not be clear which pollutant
is causally related to the health effect — or whether
more than one pollutant is causally related. Using
separate regressions (from single pollutant models) for
each pollutant may overstate the effect of each
pollutant alone. Models that consider pollutants
simultaneously are therefore preferred, though careful
attention must be paid to potential collinearity
between pollutants. Because PM has been
acknowledged to be an important pollutant, models
that include some measure of PM are highly preferred
to those that do not.
Measure of Particulate Matter
Different epidemiological studies examining the
health effects associated with particulate matter (PM)
have used different measures of PM. Some have used
PM10 while others have used PM2 5. The number of
studies using PM25 as the indicator of PM is
substantially more limited than the number using
PM10 because of the relative sparseness of PM25
monitor data. A number of studies have used total
suspended particulate matter (TSP), British Smoke,
coefficient of haze (COPT) and other measures of
particulate matter. There is some evidence that the
relationship between fine particulates (PM25) and
health effects may be stronger than that between other
measures of PM and health effects. If this is true,
then studies that use measures of PM that more
closely approximate the fine fraction of PM (such as
PM10) are preferable to those that use other measures.
D-15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Economically Valuable Health Effects
A number of the health endpoints examined in
the literature are difficult to value in monetary terms.
These effects include forced expiratory volume and
other technical measurements of lung functioning. It
is not clear how to assign an economic value to such
effects, as their impact on future medical condition
and lifestyle are not well understood. One method to
value these "clinical" measures is to estimate their
association with adverse health effects that are valued.
Ostro et al. (1989a) reanalyzed data from four
controlled ozone exposure studies, and found a
statistically significant relationship between forced
expiratory volume in one second (FEVj) and the
probability that an individual will report a mild,
moderate or severe respiratory symptom. In this case,
one could estimate ozone benefits by first calculating
the change in FEVj associated with a given change in
ozone concentration, converting this to a change in
respiratory symptoms, and then valuing the respiratory
symptom change. In a separate study, Neas and
Schwartz (1998) found that certain measures of
reduced pulmonary functioning are significant
predictors of mortality. This result, however, would
be difficult to use to calculate air pollution benefits,
because they looked at the relationship between
declines in lung function and mortality, and they did
not estimate the impact of air pollution on this
decline; separate work would be required to estimate
the impact of air pollution on lung function.
The main concern when translating a clinical
measure such as FEVj to an economically valuable
one such as acute respiratory symptoms is that
epidemiological work may already be available from
which one can directly estimate a C-R function. To
estimate acute respiratory symptoms directly (from an
epidemiological study) and indirectly through the
clinical measure, would double-count the effect.
Another concern is that using the indirect method
adds a layer of uncertainty because one must first
translate the estimated clinical measure to the
estimated economically valuable measure.
Non-Overlapping Health Effects
Several endpoints reported in the health effects
literature overlap with each other. For example, the
literature reports relationships for hospital admissions
for single respiratory ailments (e.g. pneumonia or
chronic obstructive pulmonary disease) as well as for
all respiratory ailments combined. Similarly, several
studies quantify the occurrence of respiratory
symptoms where the definitions of symptoms are not
unique (e.g., shortness of breath or upper respiratory
symptoms). Measures of restricted activity provide a
final example of overlapping health endpoints.
Estimates are available for pollution-related restricted
activity days, mild restricted activity days, and activity
restriction resulting in work loss. While the benefits
analysis estimates the benefits associated with
individual endpoints, it takes care in deciding which
endpoints to include in an estimate of total benefits,
in order to avoid double-counting of benefits from
overlapping endpoints.
Mortality
Health researchers have consistently linked air
pollution with excess mortality. Prospective cohort
and cross-sectional studies have found a relationship
between mortality over the course of a year or more
with pollution levels measured over the course of a
year or several years. In addition, a number of so-
called "short-term" mortality studies have linked daily
variations in mortality with daily pollution levels.
The EPA Clean Air Council (U.S. EPA, 1999, p.
11) recommends using the prospective cohort study
by Pope et al. (1995), rather than short-term mortality
studies. Although short-term studies lend substantial
support to the hypothesis that there is a relationship
between PM and mortality, they focus only on the
acute effects associated with daily peak exposures. In
contrast, the Pope et al. study was designed to capture
the effect of exposure over many years, however it
may be less able to capture the short-term impact of
peak exposures. This creates an overlap of unknown
size between the mortality estimates based on short-
term studies and Pope et al. Capturing the chronic
impact, however, is judged more important than
D-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
missing the impact of an unknown number of deaths
occurring shortly after short-term peak exposures.
For this reason, the Pope et al. study is preferred. A
second prospective cohort study by Dockery et al.
(1993) is also used to estimate the impact of PM on
mortality. However, the Dockery et al. study used a
smaller sample of individuals from fewer cities than
the study by Pope et al., and is therefore presented
only as an illustrative calculation that is consistent with
Pope et al. (1995); the Pope et al. estimate is used in
the primary analysis.7
The total mortality effect estimated by the Pope et
al. (1995) and the Dockery et al (1993) studies does
not necessarily occur in the same year as the estimated
exposure. However, the exact relationship between
the time of exposure and mortality is not well known.
In the primary analysis, we assume that mortality
occurs over a five year period, with 25 percent of the
deaths occurring in the first year, 25 percent in the
second year, and 16.7 percent in each of the third,
fourth, and fifth years. We also perform an analysis of
the sensitivity of benefits valuation to the lag
structure by considering a range of assumptions about
the timing of mortality (see Appendix H). It is
important to keep in mind that changes in the lag
assumptions do not change the total number of
estimated deaths, but rather the timing of those
deaths.
At least some evidence has been found linking
each of the criteria pollutants with mortality. This
raises concerns that the mortality-related benefits of
air pollution reduction may be overstated if separate
pollutant-specific estimates, some of which may have
been obtained from models excluding the other
pollutants, are aggregated. In addition, there may be
important interactions between pollutants and their
effect on mortality.
7The Pope et al., 1995 study estimated a C-R coefficient using
median PM concentration data; however, mean pollutant
concentration is the measure of central tendency commonly used
in other health studies. We will explore the possibility of re-
estimating the PM mortality C-R function using mean
concentration data in future 812 prospective analyses.
The Pope et al. (1995) study included only PM, so
it is unclear to what extent it may be including the
impacts of ozone or other gaseous pollutants. Because
of concern about overstating of benefits and because
the evidence associating mortality with exposure to
participate matter is currently stronger than for other
pollutants, only the benefits of PM-related mortality
avoided are included in the total benefits estimate.
The benefits associated with ozone reductions are
estimated but are not included in the estimate of total
benefits. The relationship between CO and mortality
is briefly considered, but the evidence reviewed does
not point to a clear link between the two.
Statistical Lives Saved Versus
Statistical Life-Years Saved
Considerable attention has been paid to using life-
years lost as an alternative to lives lost as a measure of
pollution-related premature mortality. This analysis
uses both approaches to estimating pollution-related
premature mortality.
The actual number of years any particular
individual is going to live cannot be known. Instead,
one estimates the expected, or statistical average,
number of 'life-years lost". The number of life-years
lost may be expressed as the average number of life-
years lost for all of the people who are exposed (the ex
ante measure), or as the average number of life-years
lost for the people who died from exposure (the ex
post measure).
The ex ante estimate of life-years lost depends on
the individual having been exposed to a pollutant, not
on the individual having died prematurely from that
exposure. Suppose, for example, that a 25 year old
has a life expectancy of 50 more years in the absence
of PM exposure and only 48 more years in the
presence of exposure. The exposed 25 year old
would, on average, have her life expectancy shortened
by two years. That is, two expected life-years would
be lost by every exposed individual.
The ex post estimate of life-years lost depends on
the individual actually having died from exposure to
pollution. When an individual dies of exposure to
D-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
PM, he is said to have lost the number of years he
would have been expected to live; this can be
calculated from age- and gender-specific life
expectancy tables. Suppose that the life expectancy of
25 year olds is 75 — a 25 year old can expect to live 50
more years. A 25 year old who dies from exposure to
PM has therefore lost 50 expected years of life. This
is the life-years lost that can be expected by every
affected 25 year old (i.e., every 25 year old who actually
dies from exposure to PM).
Estimates of the total life-years lost by a
population exposed to PM depend on several factors,
including the age distribution and the size of the
exposed population, the magnitude of the PM change,
the relative risk assumed to be associated with the
change in PM, and the length of exposure. A
population chronically exposed to a given increase in
PM will lose more life-years than a population
exposed to the same increase in PM for only a year or
two.8 A population that is generally older will lose
fewer life-years, all else equal, than one that is
generally younger, because older individuals have
fewer (expected) years of life left to lose. And a
population exposed to a greater increase in PM will
lose more life-years than one exposed to a smaller
increase in PM. Finally, the life-years lost by the
population will increase as the relative risk associated
with the increase in PM increases.
Life-years lost are usually reported as averages
over a population of individuals. The population over
which the average is calculated, however, can make a
crucial difference in the reported average life-years
lost. The average life-years lost per exposed individual
(the ex ante estimate) is just the total life-years lost by
the population of exposed individuals who died
divided by the number of exposed individuals.
Although those individuals who do die prematurely
from exposure to PM may lose several expected years
8 Even in the absence of cumulative effects of exposure,
exposure of a population for many years will result in a greater
total number of pollution-related deaths than exposure for only a
year or two, because the same relative risk is applied repeatedly,
year after year, to the population, rather than for only a year or
two.
of life, most exposed individuals do not actually die
from exposure to PM and therefore lose zero life-
years. The average life-years lost per exposed
individual in a population, alternatively referred to as
the average decrease in life expectancy of the exposed
population, is therefore heavily weighted towards
zero.
The ex ante and ex post measures of life-years lost
take the same total number of life-years lost by the
exposed population and divide them by different
denominators. The ex ante measure divides the total
life-years lost by the total population exposed; the ex
post measure divides the same total life-years lost by
only a small subset of the total population exposed,
namely, those who died from PM. The average per
exposed individual is therefore much smaller than the
average per affected individual. Because both types of
average may be reported, and both are valid
measurements, it is important to understand that,
although the numbers will be very dissimilar, they are
consistent with each other and are simply different
measures of the estimated mortality impact of PM.
To illustrate the different measures of life-years
lost and the effects of various input assumptions on
these measures, death rates from the 1992 U.S.
Statistical Abstract were used to follow a cohort of
100,000 U.S. males from birth to age 90 in a "dirty"
scenario and a "clean" scenario, under various
assumptions. Death rates were available for ages less
than 1, ages 1-4, and for ten-year age groups
thereafter. The ten-year age groups were divided into
five-year age groups, applying the death rate for the
ten-year group to each of the corresponding five-year
age groups. Ex ante and ex post measures of life-years
lost among those individuals who survive to the 25-29
year old category (96,947 individuals) were first
calculated under the assumptions in the World Health
Organization (WHO) 1996 report. These
assumptions were that the relative risk of mortality in
the "dirty" scenario versus the "clean" scenario is 1.1;
that exposure does not begin until age 25; that the
effect of exposure effects observed throughout the
fifteen year exposure period can be summed and
attributed (for mathematical convenience) to the 15th
year of exposure; that individuals at the beginning of
D-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
each age grouping either survive to the next age
grouping or live zero more years; and that all
individuals age 85 live exactly five more years. Under
these assumptions, the expected life-years lost per
exposed individual in the 25-29 year old cohort is 1.32
years, while the expected life-years lost per affected
individual (i.e., for each of the 7,646 expected PM-
related deaths) is 16.44 years.
Ozone and Mortality
The literature investigating the relationship
between ozone and mortality has been rapidly
evolving over the last several years. Of the 31 time-
series epidemiology studies identified in the literature
that report quantitative results on a possible
association between daily ozone concentrations and
daily mortality, 25 were published or presented since
1995. These studies were conducted in various urban
areas throughout the world: sixteen in the United
States or Canada, nine in Europe, two in Australia,
and four in Latin America. Seventeen of the studies
report a statistically significant relationship between
ozone and mortality, with the more recent studies
tending to find statistical significance more often than
the earlier studies.
While the growing body of epidemiological
studies suggests that there may be a positive
relationship between ozone and premature mortality,
there is still substantial uncertainty about this
relationship. Because the evidence linking premature
mortality and particulate matter is currently stronger
than the evidence linking premature mortality and
ozone, it is important that models of the relationship
between ozone and mortality include a measure of
particulate matter as well. Because of the lack of
monitoring data on fine participates or its
components, however, the measure of particulate
matter used in most studies was generally either PM10
or TSP or, in some cases, Black Smoke. If a
component of PM, such as PM2 5 or sulfates, is more
highly correlated with ozone than with PM10 or TSP,
and if this component is also related to premature
mortality, then the apparent ozone effects on mortality
could be at least partially spurious.
Even if there is a true relationship between ozone
and premature mortality, after taking particulate
matter into account, there would be a potential
problem of double counting in this analysis if the
ozone effects on premature mortality were added to
the PM effects estimated by Pope et al., 1995, because,
as noted above, the Pope study does not include
ozone in its model. Because of this, the potential
ozone-mortality relationship is not included in the
primary analysis. Instead the benefits associated with
ozone reductions are estimated in a sensitivity analysis.
The results of this sensitivity analysis should be
reviewed with the appropriate caution, however, in
view of the above-noted uncertainties surrounding a
potential ozone-mortality relationship.
To synthesize the results of multiple studies on
the relationship between ozone and premature
mortality, a modified meta-analysis method was used.
Because of differences in the averaging times used in
the underlying studies (some use daily average ozone
levels, while others use 1-hour daily maximum values),
the meta-analysis approach was applied to the
predicted mortality incidence changes estimated by
each of the studies rather than to the coefficients of
ozone in the C-R functions.
A study was included in the meta-analysis if it (1)
is in or has been accepted by a peer-reviewed
publication; (2) reports quantitative results for daily
mortality and ozone (rather than for other measures
such as total oxidants); (3) considers the entire
population (rather than only a subset of the
population) in the study location; (4) considers the
whole year (rather than only a season or seasons); (5)
considers all non-accidental or total mortality; (6)
considers only one location (rather than a pooling of
results across multiple locations); and (7) reports
results from a copollutant model, that includes PM or
some proxy for PM in the model with ozone, as well
as some measure of temperature and season. The
selection of a single result from among multiple ozone
results reported in the same study was facilitated by
the following three additional selection criteria: (8) PM
(PM10 or PM25) is preferable to other measures of
particulate matter; (9) more pollutants in the model is
preferable to fewer pollutants; and (10) Poisson
D-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
regression is preferred to other specifications.9 Nine
studies were chosen using these criteria. To minimize
benefits transfer problems, the meta-analysis was
limited to the four of these nine studies that were
conducted in the United States. Table D-2a briefly
describes the four studies included in the meta-
analysis.
Almost all the models in the literature used Poisson
regression. This final criterion was therefore included to impose
consistency, if there was no other means by which to select a
model from among several models in a study.
D-20
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table D-2a
Studies and Results Selected for Meta-Analysis of the Relationship between Daily Mortality
and Exposure to Ambient Ozone in the United States
Study
Ito and Thurston
(1996)a
Kinney et al. (1995)
Study Location/
Duration
Cook County,
Illinois
1985-1990
Los Angeles
County
1985-1990
The following studies were
Moolgavkar et al.
(1995)
Sametetal. (1997)
Philadelphia
1973-1988
Philadelphia
1974-1988
Copollutants
in model
PM10
PM10
used to generate a
TSP, SO2
TSP, SO2, NO2,
Lagged CO
O3 Concentration
Measure
(PPb)
average of same day
and previous day 1-hr
maxima
daily 1-hr max
single distribution for
daily avg
2-day avg
Relative Risk and
95% Cl for a 25 ppb
Increase in O3
1.016
(1.004 — 1.029)
1.000
(0.989 — 1.010)
Philadelphia:
1.015
(1.004 — 1.026)
1.024
(1.008 — 1.039)
a Relative risks derived from the ozone coefficient and standard error from the copollutant model were provided by personal
communication with Dr. Kazuhiko Ito.
Carbon Monoxide and Mortality
Research work presents some evidence that CO
may be significantly linked to mortality, although it is
not clear to what extent CO may have an effect
independent of PM. Burnett et al. (1998) studied
mortality in association with CO, NO2, O3, SO2,
coefficient of haze, TSP, sulfates and estimated PM25
and PM10 from 1980-1994 in metropolitan Toronto.
In models that included the day of the week, weather,
CO and one of the other pollutants, they found that
daily average CO and all of the PM measures
contributed a significant fraction of the daily number
of non-accidental deaths. The measure for coefficient
of haze had the strongest impact on the relative risk
for CO. The relative risk associated with a 1.4 ppm
change (i.e., 95th to the 5th CO percentile) was 1.070 in
the single pollutant model; with the addition of COH,
it fell to 1.043 (Burnett et al., 1998, Table 2).
Nevertheless, the impact of CO is still quite large, and
it is reported to occur in all seasons, age, and disease
groupings. The model with the best fit included CO
and TSP. With both CO and TSP in the model and
using the mean levels of the pollutants reported for
Toronto, CO contributed, on average, 4.7% of daily
non-accidental deaths and TSP contributed 1%
(Burnett et al., 1998, p. 689).
A review of three articles suggests that Burnett et
al.'s results may not be consistent with other
published results (Table D-2b).10 In a model with CO
and PM10, Kinney et al. (1995, Figure 3) reported a
relative risk of 1.05 for a 10 ppm CO increase (with a
95% confidence interval of 0.98-1.12). This is not
statistically significant at the usual significance level of
5%, and the implied relative risk (1.007) for a 1.4 ppm
change is about six times smaller than that reported by
in Burnett et al's two-pollutant model.11 Saldiva
(1995, Table 4) reported a positive and significant CO
regression coefficient in a model with just CO.
Estimated at the mean, this suggests a relative risk of
1.039 per 1.4 ppm of CO, or about half the size of
that reported in Burnett et al.'s single pollutant model
(RR = 1.070).12 Saldiva et al. also reported a model
with CO along with all of the other measured
10A fourth study, by Gwynn, Burnett, and Thurston, cited as
being submitted for publication, was not considered here.
"The underlying coefficient equals the logarithm of the
relative risk divided by the change in pollution.
12The regression coefficient, |3, = 1.69 (Saldiva et al., 1995,
Table 4) and the mean mortality rate per day = 62.6 (1995, Table
1). Estimated mortality after reducing CO by 1.4 ppm = 60.23
deaths per day. The relative risk = (62.9/60.23) = 1.039.
D-21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
pollutants: PM10, SO2, NOX, and O3. In this model,
the PM10 coefficient remained significant and
unchanged from its single-pollutant model value, but
the CO coefficient dropped substantially and became
insignificant (1995, Table 4). Touloumi et al. (1996,
Table 4) estimated a single pollutant model with a
reported relative risk of 1.05 for a 7.6 mg/m3 rise in
CO. Assuming a conversion of 1 ppm = 1.145
mg/m3 (U.S. EPA, 1991, Table 3-1), this suggests a
relative risk (1.015) that is about five times smaller
than the relative risk (1.070) in Burnett et al.'s single
pollutant model value.
In 1991, the EPA (1991, p. 1-12) concluded that
the results of CO epidemiological work "is suggestive,
but not conclusive evidence" that CO may lead to
sudden death in persons with coronary artery disease.
Since that time, studies by Morris et al. (1995) and
Schwartz and Morris (1995) reported that ambient CO
concentrations increase the likelihood of
hospitalization for cardiovascular disease. It is not
unlikely that a certain fraction of these admittances
will die, and thus indirectly one might estimate the
impact of CO on mortality. However, there does not
appear to be a study from which one may develop
with confidence a C-R function to directly estimate
CO-related mortality.13 The results from Burnett et al.
(1998) suggest that CO may have an effect on
mortality independent of other pollutants, but it is
premature to base an estimate of CO-related mortality
with the relative risk published in their study.
13This difficulty may be related in part to the highly variable
CO concentrations that are typically found in an urban area.
D-22
-------
Table D-2b
Selected Studies and Results for Carbon Monoxide and Mortality
Study
Burnett et
al. (1998)
Kinney et al.
(1995)
Saldiva et
al. (1995)
Touloumi et
al. (1996)
Location and
Period
Toronto, Canada
1980-1994
Los Angeles
County
1985-1990
Sao Paulo, Brazil
1990 to 1991
Athens, Greece
1987-1991
Population
All ages,
metropolita
n Toronto
All ages
Elderly
(+65 years)
All ages
Endpoint
non-
accidental
mortality
non-
accidental
mortality
mortality
from
natural
causes
total
mortality
Pollutants
CO,NO2,
SO2, O3,
SO4, TSP,
COM,
PM10,
PM25
CO, O3,
PM10
CO, O3,
PM10, S02,
NOX
CO, SO2,
black
smoke
Main Findings
Significant CO effect found in all two
pollutant models. Controlling for CO,
significant effect found for SO4, TSP, COM,
PM10, and PM25.
In single pollutant models, CO significant,
and PM10 and O3 are marginally significant.
In model with CO and PM10, both CO and
PM10 are not significant.
CO significant in single pollutant model.
CO not significant in model with all other
pollutants.
CO, SO2, and black smoke significant in
single pollutant models.
Comment
Association with cardiac-
related mortality is stronger,
but CO is also significantly
related to non-cardiac
mortality. PM10 and PM25
estimated from SO4, TSP,
and COM.
Magnitude of single pollutant
CO relationship drops
modestly with inclusion of
PM10.
Deaths during a one month
summertime heat wave were
excluded from analysis
D-23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Post-Neonatal Mortality
In a recent study of four million infants in 86 U.S.
metropolitan areas, Woodruff et al. (1997) linked PM10
exposure in the first two months of an infant's life
with the probability of dying between the ages of 28
days and 364 days. In addition to the work by
Woodruff et al., recent work in Mexico City (Loomis
et al., 1999), the Czech Republic (Bobak and Leon,
1992), Sao Paulo (Pereira et al., 1998; Saldiva et al.,
1994), and Beijing (Wang et al., 1997) provides
additional evidence that particulate levels are
significantly related to infant or child mortality, low
birth weight or intrauterine mortality (Table D-3).
Conceptually, neonatal or child mortality could be
added to the premature mortality predicted by Pope et
al. (1995), because the Pope function covers only the
population over 30 years old. Predicted neonatal
mortality could not be added to the premature
mortality predicted by the daily (short-term exposure)
mortality studies, however, because these studies
cover all ages. The EPA Clean Air Council recently
advised the Agency not to include post-neonatal
mortality in this analysis because the study is of a new
endpoint and the results have not been replicated in
other studies (U.S. EPA, 1999, p. 12). The estimated
avoided incidences of neonatal mortality are estimated
and presented as a sensitivity analysis, but are not
included in the aggregate benefits analysis results.
D-24
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table D-3
Studies and Results Selected for Adverse Effects in
Study
Bobakand
Leon
(1992)
Loomis et
al. (1999)
Pereira et
al. (1998)
Ritz and
Yu(1999)
Saldiva et
al. (1994)
Wang et al.
(1997)
Woodruff
etal.
(1997)
Xu et al.
(1995a)
Location and
Period
45 of 86
administrative
districts in the
Czech Republic
1986-1988
southwestern
Mexico City
1/93-7/95
Sao Paulo,
Brazil
1/91-12/92
Los Angeles,
CA
1989-1993
Sao Paulo,
Brazil
5/90-4/91
Beijing, China
1988-1991
86 metropolitan
areas in the
U.S.
1989-1991
Beijing, China
1988
Population
neonates (0-
1 month);
post-
neonates (1-
12 months)
infants <1
year old
fetuses over
28 weeks of
pregnancy
age
gestational
age 37-44
weeks
children <5
gestational
age 37-44
weeks
post-
neonates (1-
12 months)
25,370
pregnant
women
Endpoint
all-cause
mortality;
respiratory
mortality
all cause
mortality
intrauterine
mortality
low birth
weight
respiratory
mortality
low birth
weight
all-cause
mortality;
respiratory
mortality
p re-term
delivery
Fetuses, Infants, and Young Children
Pollutants
TSP, SO2,
NOX
PM25, 03,
NO2, SO2
PM10, 03,
NO2, SO2,
CO
CO
PM10, 03,
NOX, S02,
CO
TSP, SO2
PM10
TSP, SO2
Main Findings
Controlling for SO2 and NO,, TSP
linked to all-cause and respiratory
post-neonatal mortality; weaker,
insignificant effect found for
neonatal. Controlling for TSP and
SO2, NOX marginally significant for
all-cause and respiratory post-
neonatal mortality; no effect for
neonatal mortality. No effect found
for SO2.
PM25 and NO2 significant in single
pollutant models. PM25 and NO2
both not significant in two pollutant
model; PM25 coefficient changed
little from single pollutant; NO2
coefficient dropped substantially. O3
not significant. SO2 not analyzed
since ambient levels were negligible.
In single pollutant models, NO2, SO2,
and CO significantly related to
intrauterine mortality. PM10 and O3
not significant. Considering all
pollutants simultaneously, NO2 is the
only significant pollutant.
Average CO exposure in the last
trimester associated with low birth
weight.
NOX significantly related to
respiratory mortality. No effect found
for the other pollutants.
TSP and SO2 exposure in the final
trimester significantly related to low
birth weight. Both pollutants highly
correlated (r=0.92).
PM10 exposure in the first two months
of life significant for all-cause
mortality. PM10 significant for
respiratory mortality in average birth-
weight infants, but not low birth-
weight infants.
TSP and SO2 exposure significant for
pre-term delivery.
D-25
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Chronic Illness
There are a limited number of studies that have
estimated the impact of air pollution on chronic
bronchitis (Table D-4). An important hindrance is the
lack of health data and the associated air pollution
levels over a number of years. Schwartz (1993) and
Abbey et al. (1995; 1993) provide evidence that PM
exposure over a number of years gives rise to the
development of chronic bronchitis in the U.S., and a
recent study by McDonnell et al. (1999) provides
evidence that ozone exposure is linked to the
development of asthma in adults. These results are
consistent with research that has found chronic
exposure to pollutants leads to declining pulmonary
functioning (Abbey et al., 1998; Ackermann-Liebrich
etal, 1997; Betels et al, 1991).
Schwartz (1993) examined survey data collected
from 3,874 adults ranging in age from 30 to 74, and
living in 53 urban areas in the U.S. The survey was
conducted between 1974 and 1975, as part of the
National Health and Nutrition Examination Survey,
and is representative of the non-institutionalized U.S.
population. Schwartz (1993, Table 3) reported
chronic bronchitis prevalence rates in the study
population by age, race, and gender. Non-white males
under 52 years old had the lowest rate (1.7%) and
white males 52 years and older had the highest rate
(9.3%). The study examined the relationship between
the prevalence of reported chronic bronchitis and
annual levels of TSP, collected in the year prior to the
survey.
Abbey et al. (1995; 1993) are part of a series of
studies of an ongoing prospective cohort tracking
research project that began in 1977. These two
studies on the development of chronic respiratory
illness are based on a ten year follow-up examination
of adult Seventh-Day Adventists living in California.
Abbey et al. (1993) examined 3,914 adults, and
estimated the relationship between annual mean
ambient TSP, ozone and SO2 and the presence of
certain chronic respiratory symptoms (including
airway obstructive disease (AOD), chronic bronchitis,
and asthma) that were not present at the beginning of
the study. TSP was significantly linked to new cases
of AOD and chronic bronchitis, but not to asthma or
the severity of asthma. Ozone was not linked to the
incidence of new cases of any endpoint, but ozone
was linked to the severity of asthma. No effect was
found for SO2. Abbey et al. (1995) examined the
relationship between estimated PM25 (annual mean
from 1966 to 1977), PM10 (annual mean from 1973 to
1977) and TSP (annual mean from 1973 to 1977) and
the same chronic respiratory symptoms in a sample
population of 1,868 Californian Seventh-Day
Adventists. In this single-pollutant study, there was a
statistically significant PM25 relationship with
development of chronic bronchitis, but not for AOD
or asthma; PM10 was significantly associated with
chronic bronchitis and AOD; and TSP was
significantly associated with all cases of all three
chronic symptoms.
The McDonnell et al. (1999) study used the same
cohort of Seventh-Day Adventists, and examined the
association between air pollution and the onset of
asthma in adults between 1977 and 1992. Males who
did not report doctor-diagnosed asthma in 1977, but
reported it in 1987 or 1992, had significantly higher
ozone exposures, controlling for other covariates; no
significant effect was found between ozone exposure
and asthma in females. No significant effect was
reported for females or males due to exposure to PM,
NO2, SO2, or SO4.
We estimate the changes in the new cases of
chronic bronchitis using the studies by Schwartz
(1993), Abbey et al. (1993), and Abbey et al. (1995);
also, we estimate the onset of asthma in adult males
using the work by McDonnell et al. (1999). The
Schwartz study is somewhat older and uses a cross-
sectional design; however, it is based on a national
sample, unlike the Abbey et al. studies which are based
on a sample of California residents who were non-
smokers. We first pool the estimates from the two
studies by Abbey et al. — since they are based on the
same sample population and simply use different
measures of PM - and then pool this estimate with
that from Schwartz.
The Abbey et al. (1995; 1993) studies are based on
the incidence of new cases of chronic bronchitis,
D-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
however, Schwartz (1993) is based on tine prevalence of
chronic bronchitis, not its incidence. To use Schwartz's
study and still estimate the change in incidence, there
are at least two possible approaches. The first is to
simply assume that it is appropriate to use the baseline
incidence of chronic bronchitis in a C-R function with
the estimated coefficient from Schwartz's study, to
directly estimate the change in incidence. The second
is to estimate the percentage change in the prevalence
rate for chronic bronchitis using the estimated
coefficient from Schwartz's study in a C-R function,
and then to assume that this percentage change applies
to a baseline incidence rate obtained from another
source. (That is, if the prevalence declines by 25
percent with a given decrease in PM, then baseline
incidence drops by 25 percent with the same drop in
PM). This analysis uses the latter approach, and
estimates the change in incidence by first estimating
the percentage change in prevalence.
D-27
-------
Table D-4
Summary of Selected Studies
Study
Abbey et al.
(1993)
Abbey et al.
(1995)
Chapman et
al. (1985)
McDonnell
etal. (1999)
Portney and
Mullahy
(1990)
Location and
Period
California
initial survey: 1977
final survey: 1987
California
initial survey: 1977
final survey: 1987
4 Utah
communities
1976
California
initial survey: 1977
final survey: 1992
Nationwide sample
from the 1979 U.S.
National Health
Interview Survey
for Chronic Illness
Population
3,914
Seventh
Day
Adventists
1,868
Seventh
Day
Adventists
5,623
young
adults
3,091
Seventh
Day
Adventists
1,318
persons
age 17-93
Endpoint
AOD; chronic
bronchitis;
asthma
AOD; chronic
bronchitis;
asthma
persistent
cough and
phlegm
asthma
sinusitis, hay
fever, AOD
Pollutants Main Findings
TSP, O3, TSP linked to new cases of AOD and
SO2 chronic bronchitis, but not to asthma or
the severity of asthma. O3 not linked to
the incidence of new cases of any
endpoint, but O3 was linked only to the
severity of asthma. No effect found for
SO2.
PM25 PM25 related to new cases of chronic
bronchitis, but not to new cases of
AOD or asthma.
SO2, SO4, Persistent cough and phlegm is higher
NO3, TSP in the community with higher SO2, SO4,
and TSP concentrations.
O3, PM10,, Single pollutant models: O3
SO4, SO2, significantly linked to new asthma
NO2 cases in males, but not in females;
other pollutants not significantly linked
to new asthma cases in males or
females. Two pollutant models
estimated for ozone with another
pollutant; little impact found on size of
ozone coefficient.
O3, TSP Controlling for TSP, O3 significantly
related to the initiation (or
exacerbation) of sinusitis and hay
fever; no effect on AOD. TSP not
significantly related to any endpoint,
although it is marginally significant for
AOD.
Comment
Emphysema, chronic
bronchitis, and asthma
comprise AOD.
PM25 estimated from visibility
data.
Average pollution level from
1973-1992 used. Prior to
1987, PM10 estimated from
TSP.
D-28
-------
Study
Schwartz
(1993)
Xu et al.
(1993)
Zemp et al.
(1999)
Location and
Period
Nationwide sample
from the National
Health and
Nutrition
Examination
Survey
1974-1975
Beijing, China
Survey conducted
August-September
1986
Eight sites in
Switzerland
1991
Population
6,138
individuals
ages 30-74
1,576
never
smokers
9,651
individuals
ages 18-60
Endpoint Pollutants
chronic TSP
bronchitis;
asthma;
shortness of
breath
(dyspnea);
respiratory
illness
chronic TSP; SO2
bronchitis;
asthma
chronic TSP, PM10,
phlegm, NO2, O3
chronic cough,
breathlessness
, asthma,
dyspnea on
exertion
Main Findings
TSP significantly related to the
prevalence of chronic bronchitis, and
marginally significant for respiratory
illness. No effect on asthma or
dyspnea.
Chronic bronchitis significantly higher
in the community with the highest TSP
level. TSP not linked to the prevalence
of asthma.
Single pollutant models: PM10 and NO2
significantly associated with chronic
phlegm, chronic cough or phlegm,
breathlessness and dyspnea. Similar
though less significant associations
found for TSP. No significant effect
found for O,.
Comment
Respiratory illness defined as
a significant condition, coded
by an examining physician as
ICD8 code (460-51 9)
D-29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Hospital Admissions
There is a wealth of epidemiological information
on the relationship between air pollution and hospital
admissions for various respiratory and cardiovascular
diseases; in addition, some studies have examined the
relationship between air pollution and emergency
room (ER) visits. Because most emergency room
visits do not result in an admission to the hospital —
the majority of people going to the ER are treated and
return home — we treat hospital admissions and ER
visits separately, taking account of the fraction of ER
visits that do get admitted to the hospital, as discussed
below.
Hospital admissions require the patient to be
examined by a physician, and on average may
represent more serious incidents than ER visits
(Lipfert, 1993, p. 230). The two main groups of
hospital admissions estimated in this analysis are
respiratory admissions and cardiovascular admissions.
There is not much evidence linking air pollution with
other types of hospital admissions. The only types of
ER visits that have been linked to air pollution in the
U.S. or Canada are asthma-related visits.
To estimate the number of hospital admissions
for respiratory illness, we pool the incidence estimates
from a variety of U.S. and Canadian studies, using a
random effects weighting procedure. These studies
differ from each other in two important ways: (1)
Some studies considered people of all ages while
others considered only people ages 65 and older; and
(2) The International Classification of Diseases - 9th
revision (ICD-9) codes included in studies of
respiratory hospital admissions and air pollution vary
substantially.
The broadest classification used (for example, in
Schwartz, 1996) includes ICD-9 codes 460-519.
Other studies, however, considered only subsets of
the broader classification. For example, Burnett et al.
(1997b) consider ICD-9 codes 466,480-486, 490-494,
and 496. The correct set of ICD codes for this study
is difficult to determine. If the broadest category
(460-519) is too broad, including respiratory illnesses
that are not linked to air pollution, we would expect
the estimated pollutant coefficients to be biased
downward; however, they would be used in
combination with a larger baseline incidence in
estimating changes in respiratory hospital admissions
associated with changes in pollutant concentrations.
If the broadest category is correct (i.e., if all the
respiratory illnesses included are actually associated
with air pollution), then studies using only subsets of
ICD codes within that category would presumably
understate the change in respiratory hospital
admissions. It is likely, however, that all the studies
have included the most important respiratory illnesses,
and that the impact of differences in the definition of
"all respiratory illnesses" may be less than that of
other study design characteristics. We therefore treat
each study equally, at least initially, in the pooling
process, assuming that each study gives a reasonable
estimate of the impact of air pollution on respiratory
hospital admissions.
There are several steps in our estimation process:
• Develop study-specific estimates of
respiratory admissions incidence change;
• Develop C-R functions for each pollutant in
a model from a given study: e.g., Burnett et
al. (1997b) included PM25_10, O3, NO2, and
SO2 in their final model for respiratory
admissions (ICD-9 codes 464-466, 480-486,
490-494, 496);
• Estimate the change in incidence associated
with the change in each air pollutant
considered in the model, and aggregate these
incidence changes across the pollutants in the
model: e.g., for Burnett et al. (1997b) we sum
the incidence changes associated with PM25_
10, 03, N02, and SO2;
• If a study estimated separate models for non-
overlapping respiratory illness categories, sum
the estimated incidence changes across these
non-overlapping categories: e.g., Delfino et
al. (1994) estimated two separate models: one
for asthma (ICD code 493) and one for all
respiratory non-asthma (ICD codes 462-466,
D-30
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
480-487, 490-492, 494, and 496); we
estimated and summed incidences for these
two categories.
Aggregate estimates across non-overlapping age
categories:
• Seven studies estimated C-R functions for
respiratory admissions for people ages 65 and
older. One study, Sheppard et al. (1999),
estimated a C-R function for asthma only for
people under 65. Using a Monte Carlo
procedure, we aggregate the results from the
Sheppard study with those from each of the
over-65 respiratory admissions studies.
Pool estimates of respiratory hospital admissions
changes:
• Four studies estimated C-R functions for
respiratory admissions for people of all ages.
With the seven "all ages" estimates developed
in step 2, there are eleven separate estimates
of the change in respiratory hospital
admissions associated with a change in air
pollutant concentrations. Using Monte Carlo
procedures, the results of these eleven studies
are pooled.
Table D-5 summarizes the studies used in
estimating respiratory admissions; Table D-6 provides
more detailed information on these studies, and other
studies that were not chosen for this analysis.
Similar issues of definition arise for cardiovascular
hospital admissions. The broadest classification we
have seen in the epidemiological literature includes
ICD codes 390-429 (see, for example, Schwartz,
1999). Some studies, however, use a much more
narrow definition, including only subsets of the larger
group of ICD codes. We use a similar procedure for
cardiovascular admissions as we used for respiratory
hospital admissions. Table D-7 summarizes the
studies used in estimating cardiovascular admissions;
Table D-8 provides more detailed information on
these studies, and other studies that were not chosen
this analysis.
Because we are estimating ER visits as well as
hospital admissions for asthma, we must avoid
counting twice the ER visits for asthma that are
subsequently admitted to the hospital. To avoid
double-counting, the baseline incidence rate for
emergency room visits is adjusted by subtracting the
percentage of patients that are admitted into the
hospital. Three studies provide some information to
do this: Richards et al. (1981, p. 350) reported that
13% of children's ER visits ended up as hospital
admissions; Lipfert (1993, p. 230) reported that ER
visits (for all causes) are two to five times more
frequent than hospital admissions; Smith et al. (1997,
p. 789) reported 445,000 asthma-related hospital
admissions in 1987 and 1.2 million asthma ER visits.
The study by Smith et al. seems the most relevant
since it is a national study and looks at all age groups.
Assuming that air-pollution related hospital
admissions first pass through the ER, the reported
incidence rates suggest that 37%
(=445,000/1,200,000) of ER visits are subsequently
admitted to the hospital, or that ER visits for asthma
occur 2.7 times as frequently as hospital admissions
for asthma. The baseline incidence of asthma ER
visits is therefore taken to be 2.7 times the baseline
incidence of hospital admissions for asthma. To
avoid double-counting, however, only 63% of the
resulting change in asthma ER visits associated with a
given change in pollutant concentrations is counted in
the ER visit incidence change.
Table D-9 summarizes the studies used in
estimating ER visits for asthma; Tables D-10 and D-
11 provide more detailed information on these studies
and other ER studies that were not used in the
analysis.
D-31
-------
Table D-5
Studies Used to Develop Respiratory Admissions Estimates
Location
Toronto, Canada
Toronto, Canada
Toronto, Canada
Minneapolis-St.
Paul, MN
Minneapolis-St.
Paul, MN
Birmingham, AL
Detroit, Ml
Spokane, WA
New Haven, CT
Tacoma, WA
Seattle, WA
Study
Burnett et al. (1997b)
Burnett et al. (1999)
Thurston et al. (1994)
Moolgavkar et al. (1997)
Schwartz (1994c)
Schwartz (1994a)
Schwartz (1994b)
Schwartz (1996)
Schwartz (1995)
Schwartz (1995)
Sheppard et al. (1999)
Endpoints Estimated a
(ICD code)
all respiratory (464-466, 480-486,
490-494, 496)
asthma (493); respiratory infection
(464, 466, 480-487, 494); non-
asthma COPD (490-492, 496)
all respiratory (466, 480-482, 485,
490-493)
pneumonia (480-487); COPD
(490-496)
pneumonia (480-486); COPD (490-
496)
pneumonia (480-487); COPD
(490-496)
pneumonia (480-486); non-asthma
COPD (491 -492, 494-496)
all respiratory (460-519)
all respiratory (460-519)
all respiratory (460-519)
asthma (493)
Pollutants Used in Final Model
PM25.10, O3, NO2, SO2
O3, CO, PM2 wo (asthma); O3, NO2, PM25
(respiratory infection); O3, CO, PM2510
(COPD).
03, PM25
O3, SO2, NO2, PM10 (pneumonia); O3, CO,
PM10(COPD)
O3, PM10 (pneumonia); PM10 (COPD)
PMn
03, PM10
PM10
03, PM10
03, PM10
CO, PM25
Study
Population
all ages
all ages
all ages
>64
>64
>64
>64
>64
>64
>64
<65
a Monetized benefits of non-overlapping endpoints within each study are aggregated. Monetized benefits for asthma among people age <65 (Sheppard et al., 1999) are aggregated with
the benefits in studies of people age >64.
D-32
-------
Table D-6
Summary of Hospital Admissions Studies - Respiratory Illnesses
Study
Anderson et
al. (1997)
Burnett et
al. (1995)
Burnett et
al. (1997b)
Burnett et
al. (1997a)
Location and
Period
Barcelona, Paris
Amsterdam,
Rotterdam,
Milano
Period varies by
city from 5-13
years
southern
Ontario, Canada
1/83-12/88
Toronto,
Canada
summers in
1992-1994
16 Canadian
cities
3/81-12/91
Population Endpoint
all ages; COPD
>64 (490-492,
496)
<65; >64 all
respiratory
(466, 480-
486, 490-
494, 496)
all ages all
respiratory
(464-466,
480-486,
490-494,
496)
<65; >64 all
respiratory
(466, 480-
486, 490-
494, 496)
Pollutants a
NO2, BS
(black
smoke),
TSP, SO2,
03
S04, 03
O3, CO, NO2,
SO2, COM
(coefficient
of haze), H+,
S04, PM25,
PM25.10,
PM10,
O3, CO, SO2,
NO2, COM
Main Findings
COPD: Single pollutant models: meta-analysis
of city specific results found significant effect
for BS, NO2, O3, and SO2 in the all age group;
similar results reported for ages >64.
Strongest effect found for O3. TSP not
significant in meta-analysis. For a given
pollutant, results varied considerably by city.
All respiratory: SO4 significantly related to
respiratory admissions in ages <65 and >64.
O3 significant impact from May-September; no
effect the rest of the year.
All respiratory: COM and O3 linked to
respiratory admissions; other PM measures
less strongly linked. Two pollutant models:
CO, NO2, and SO2 not significant, controlling
for COM; O3 significant, controlling for COM.
Four pollutant models: COM and O3
significant; no effect for NO2 and SO2; other
PM measures not significant, controlling for
O3, NO2, and SO2.
All respiratory: Multiple pollutant models: O,
significantly related to admissions, controlling
for CO and COM; significant effect also
reported for CO and COM; no significant effect
found for NO2 and SO2 after controlling for O3
and CO. Montreal and Vancouver decreased
the size of the effect of O3 substantially. O3
significant with and without these cites in the
model.
Comment
Study not used to estimate
incidence: study outside
U. S. and Canada.
May-September results
also discussed in Burnett
etal. (1994). Study not
used to estimate incidence:
no study specific
conversion available
between SO4 and PM2S or
PM10.
Four pollutant model
(PM2S_m O3, NO2, and SO2)
used to estimate all
respiratory incidence.
Study not used to estimate
incidence: no study specific
conversion available
between COM and PM2S or
PM10.
D-33
-------
Study
Burnett et
al. (1999)
Delfino et
al. (1994)
Lipfert and
Hammerstro
m(1992)
Location and
Period
Toronto,
Canada
1980-1994
Montreal,
Canada
May-October in
1984-1988.
July-August
subset used to
examine all
respiratory
admissions.
Southern
Ontario, Canada
January-
February and
July-August in
4/79-3/85
Population Endpoint
all ages asthma
(493);
respiratory
infection
(464, 466,
480-487,
494);
COPD
(490-492,
496)
all ages asthma
(493); all
respiratory
(462-466,
480-487,
490-494,
496); all
respiratory
non-asthma
all ages all
respiratory
(466, 480-
482, 485,
490-493)
Pollutants a
O3, CO, NO2,
SO2, PM25,
PM25.10,
PM10,
03, S04,
PM10
03, S04,
N04, S04,
COM, TSP
Main Findings
Multiple pollutant models estimated, where
pollutants for best fitting model chosen using
stepwise regression based on AIC criterion.
Asthma: O3, CO, PM25_10 significantly related
to asthma admissions; other pollutants not
chosen in stepwise regression. Respiratory
infection: O,, NO,, and PM,K chosen in
stepwise regression COPD: O3 and PM25_10
chosen in stepwise regression.
Asthma: Two pollutant model: marqinallv
significant effect for PM10, controlling for O3.
No effect for O, and SO,,. All respiratory and
all respiratory non-asthma: PM10 suggestive
but not significant, after controlling for
temperature. Significant link between all
respiratory non-asthma and SO4. No effect for
o3.
All respiratory: SO2, SO4, and O3 found to be
significant predictors of respiratory admissions
in July-August.
Comment
PM25, PM2WO, and PM10
estimated from TSP, COM,
and SO4 data. Multiple
pollutant models used to
estimate incidence of:
asthma (O3, CO, PM2 wo/),
respiratory infection (O3,
NO2, PM25;, and COPD
(03, CO, PM25.10/
SO4 and PM10were both
estimated from COM and
other variables. Study not
used to estimate incidence
Study not used to estimate
incidence: estimated
coefficients not reported.
D-34
-------
Study
Moolgavkar
etal. (1997)
Morgan et
al. (1998)
Pantazopol
ou et al.
(1995)
Ponce de
Leon et al.
(1996)
Ponka and
Virtanen
(1994)
Location and
Period
Minneapolis-St.
Paul, MN;
Birmingham, AL
1/86-12/91
Sydney,
Australia
1/90-12/94
Athens, Greece
1988
London,
England
4/87-2/92
Helsinki, Finland
1/87-12/89
Population Endpoint
>64 pneumonia
(480-487);
COPD
(490-496);
all
respiratory
(480-487,
490-496)
1-14; 15- asthma;
64; >64 COPD
(490-
492,494
496)
all ages all
respiratory
(not defined
by ICD
code)
0-14; 15- all
64; >64 respiratory
(460-519)
<65; >64 COPD
(491-492)
Pollutants a
O3, CO, SO2,
N02, PM10
O3, NO2,
bscat
(measure of
light
scattering)
BS, CO, NO2
BS, SO2, O3,
NO2
NO2, SO2,
O3, TSP
Main Findings
Pneumonia: Four pollutant model: O,
significant (NO2, SO2, and PM10 not significant)
in Minneapolis-St. Paul; no significant effect
found for any pollutant in Birmingham. COPD:
No significant effect found in Birmingham or
Minneapolis-St. Paul for any pollutant. AN
respiratory: Single pollutant models: O3, NO2,
and PM10 significant in Minneapolis-St. Paul.
Multiple pollutant models (results presented in
graph): O3 significant, controlling for other
pollutants; PM10 significant controlling for O3,
but not significant controlling for O3, SO2, and
NO2 together. No significant effect found in
Birmingham for admissions with O3, CO, or
PM10; NO2 and SO2 data not available for
Birmingham.
Asthma: Sinqle pollutant models: NO,
significant for ages 1-14 but not other age
groups. O3 and bscat not significant for any
age group. Three pollutant model: NO2
remains significantly related to asthma
admission in ages 1-14. COPD: No pollutant
significantly related to COPD admissions.
All respiratory: Single-pollutant models: BS,
CO, NO2 significantly related to respiratory
admissions in the wintertime. No significant
effect found any pollutant in the summer.
All respiratory: O, significantly related to
admissions in ages >14. No significant effect
found for SO2, O3, and NO2.
COPD: Single pollutant models: SO2 linked to
(491-492) admissions in ages <65; NO2 linked
to admissions in ages >64; no significant
effect seen for O3 and TSP.
Comment
Multiple pollutant models
used to estimate
pneumonia incidence (O3,
SO2, NO2, PM10) and
COPD incidence (O3, CO,
PM10) in Minneapolis-St.
Paul. No model estimated
for Birmingham:
coefficients and standard
errors not reported.
Study not used to
estimate incidence: study
outside U.S. and Canada.
Study not used to estimate
incidence: study outside
U. S. and Canada.
Study not used to estimate
incidence: study outside
U. S. and Canada.
Study not used to
estimate incidence: study
outside U.S. and Canada.
D-35
-------
Study
Schwartz
(1994a)
Schwartz
(1994b)
Schwartz
(1996)
Schwartz
(1994c)
Schwartz
(1995)
Location and
Period
Birmingham, AL
1/86-12/89
Detroit, Ml
1/86-12/89
Spokane, WA
1/88-12/90
Minneapolis-St.
Paul, MN
1/86-12/89
New Haven, CT;
Tacoma, WA
1/88-12/90
Population Endpoint Pollutants a
>64 pneumonia PM10, O3
(480-487);
COPD
(490-496)
>64 asthma PM10, O3
(493);
pneumonia
(480-486);
non-asthma
COPD
(491-492,
494-496)
>64 pneumonia PM10, O3
(480-487);
COPD
(490-496);
all
respiratory
(460-519)
>64 pneumonia PM10, O3
(480-486);
COPD
(490-496)
>64 all PM10, O3,
respiratory SO2
(460-519)
Main Findings
Pneumonia: PM1P siqnificant and O, not
significant in single pollutant models. COPD:
PM10 significant and O3 not significant in single
pollutant models.
Asthma: admissions not associated with either
pollutant; coefficients and standard errors not
reported. Pneumonia: Two pollutant model:
PM10 and O3 both significant for pneumonia.
Non-asthma COPD: Two pollutant model:
PM10 and O3 both significant.
Pneumonia: PM1P marqinallv siqnificant and
O3 not significant for pneumonia in single
pollutant models. COPD: PM1P siqnificant and
O3 not significant in single pollutant models.
All respiratory: Sinqle pollutant models: PM1P
and O3 both significant. Two pollutant model
not estimated because of limited overlap
between PM10 and O3 data.
Pneumonia: Two pollutant model: PM1P
significantly related to pneumonia; O3 weakly
linked to pneumonia. COPD: Sinqle pollutant
models: PM10 significant and O3 not
significant.
All respiratory: Sinqle pollutant models: PM1P,
O3, SO2 significant, except O3 in New Haven.
Two pollutant model results varied by city: O3
significant (3 of 4 models) and stable
coefficient estimates PM10 significant (3 of 4
models), but less stable estimates. SO2
significant (1of 4 models).
Comment
Single pollutant models
(PMW) used to estimate
pneumonia incidence and
COPD incidence.
Two pollutant models (PM10
and O3) used to estimate
pneumonia incidence and
non-asthma COPD
incidence.
Single pollutant model
(PMW) used to estimate all-
respiratory incidence.
Two pollutant model (PM10,
O3) used to estimate
pneumonia incidence;
single pollutant model
(PM10) used to estimate
COPD incidence.
Two pollutant model (PM10,
O3) used to estimate all
respiratory incidence.
D-36
-------
Study
Sheppard et
al. (1999)
Spix et al.
(1998)
Sunyer et
al. (1997)
Tenias et al.
(1998)
Thurston et
al. (1994)
Location and
Period Population
Seattle, WA <65
1/87-12/94
London, 15-64; >64
Amsterdam,
Rotterdam,
Paris, Milano
Period varies by
city from 5-13
years
Barcelona, 0-14; 15-64
Helsinki,
London, Paris
Period varies by
city from 3-6
years
Valencia, Spain >14
1/93-12/95
Toronto, all ages
Canada
six weeks in July
and August
1986-1988
Endpoint
asthma
(493)
all
respiratory
(460-519)
asthma
asthma
asthma
(493); all
respiratory
(466, 480-
482, 485,
490-493)
Pollutants a Main Findings
CO, SO,, O,, Asthma: Sinqle pollutant models: each
PM25, PM25_ pollutant significantly related to asthma,
10, PM10 except SO2. Multiple pollutant models: PM25
and CO reported to have best fit in models
without O3. Both PM25 and CO are significant
when included together in a model.
NO,, SO,, All respiratory: Sinqle pollutant models: O,
O3, TSP, BS significantly related to admissions in ages 15-
64 and >64. BS significantly related to
admissions in ages 15-64. SO2 significantly
related to admissions in ages >64.
NO2, SO2, Asthma: Two pollutant models: NO2 significant
O3, BS in ages 15-64, controlling for BS; NO2 had no
effect on ages 0-14. SO2 significant in ages 0-
14, controlling for either BS or NO2; SO2 had
no effect on ages 15-64.
NO2, SO2, Asthma: Two pollutant models: O3 and NO2
O3, BS both significant. No significant effect found for
SO2 and BS.
H+, SO,,, O,, Asthma: Sinqle pollutant models: O,, H+, SO,,,
PM25, PM25_ O3, and TSP linked to all respiratory
10, PM10, TSP admissions; PM25, PM2 wo, PM10 not
significant. Two pollutant models: O3
significant, but PM measures no longer
significant. Best fitting PM measure is H+. AN
respiratory: Single pollutant models: O3 and
various measures of PM linked to all
respiratory admissions. Two pollutant models:
with O3 and PM together, O3 still significant,
but PM often not significant (only H+
significant).
Comment
In most years, O3 data was
available only from April
through October. O3
reported to have the best
fit, but authors did not
consider O3 further
because of limited data.
Two pollutant model (CO,
PM2S) used to estimate
asthma incidence.
Study not used to estimate
incidence: study outside
U. S. and Canada.
Study not used to estimate
incidence: study outside
U. S. and Canada.
Study not used to estimate
incidence: study outside
U. S. and Canada.
Two pollutant model (O3,
PM2S) used to estimate all
respiratory incidence.
D-37
-------
Study
Thurston et
al. (1992)
Vigotti et al.
(1996)
Location and
Period
Buffalo, NY;
New York City
June-August in
1988-1989
Milan, Italy
1/89-12/89
Population Endpoint
all ages all
respiratory
(466, 480-
486, 490-
493)
15-64;>64 all
respiratory
(460-519)
Pollutants a Main Findings
H+, SO,,, O, All respiratory: Three pollutant model: H+, SO,,,
and O3 are all significant. This result is found
in both Buffalo and New York City.
TSP, SO, All respiratory: Sinqle pollutant models: TSP
and SO2 linked to admissions.
Comment
Study not used to estimate
incidence: no study specific
conversion available
between study pollutants
(7-f and SO4) and PM25 or
PM10.
Study not used to estimate
incidence: study outside
U. S. and Canada.
' Not all pollutants considered in a study are necessarily included in the model used to develop C-R functions.
D-38
-------
Table D-7
Studies Used to Develop Cardiovascular Admissions Estimates
Location
Study
Endpoints Estimated
(ICD code)
Pollutants Used in Final Model
Study
Population
Toronto, Canada
Burnett et al. (1997b)
cardiac (410-414, 427-428)
O, PM,
all ages
Toronto, Canada
Burnett et al. (1999)
ischemic heart disease (410-414);
dysrhythmias (427); congestive heart
failure (428)
NO2, SO2 (ischemic heart disease); PM25, CO,
O3 (dysrhythmias); CO, NO2 (heart failure
incidence)
all ages
Detroit, Ml
Schwartz and Morris
(1995)
ischemic heart disease (410-414);
congestive heart failure (428)
CO, PM,
>64
Eight U.S. counties
1/88-12/90
Schwartz (1999)
cardiovascular disease (390-429)
CO, PM.,
>64
Tucson, AZ
1/88-12/90
Schwartz (1999)
cardiovascular disease (390-429)
CO, PM.,
>64
D-39
-------
Table D-8
Summary of Hospital Admissions Studies - Cardiovascular Illnesses
Study
Burnett et al.
(1995)
Burnett et al.
(1997b)
Burnett et al.
(1997c)
Burnett et al.
(1999)
Location and
Period
southern and
central
Ontario,
Canada
1/83-12/88
Toronto,
Canada
summers
1992-1994
10 Canadian
cities
1/81-12/91
Toronto,
Canada
1980-1994
Endpoint
Population (ICD code)
all ages cardiac
(410,413,
427-428)
all ages cardiac
(410-414,427-
428)
>64 congestive
heart failure
(428)
all ages ischemic
heart disease
(410-414);
dysrhythmias
(427);
congestive
heart failure
(428)
Pollutants
S04, 03
O3, CO,
NO2, SO2,
COM
(coefficient
of haze),
H+, S04,
PMZ5,
PM2.WO,
PM10,
O3, CO,
NO2, SO2,
COM
O3, CO,
NO2, SO2,
PM2.5,
PM2.WO,
PM10,
Main Findings
Cardiac: Two pollutant model: SO4
significantly related to cardiac admissions;
O3 not significant, in any season or over
the whole year.
Cardiac: COM and O, linked to cardiac
admissions; other PM measures less
strongly linked. Two pollutant models: CO,
NO2, and SO2 not significant, controlling for
COM. O3 significant, controlling for COM.
Four pollutant models: COM and O3
significant; no effect for NO2 and SO2.
Other PM measures not significant,
controlling for O3, NO2, and SO2.
Congestive heart failure: Single pollutant
models: CO, NO2, SO2, COM are
significant; no effect for O3. CO and NO2
have the best fit. Two pollutant models:
Controlling for NO2, CO significant, with
only small reduction in coefficient size;
NO2 insignificant in this model.
Multiple pollutant model, where pollutants
for best fitting model chosen using
stepwise regression based on AIC
criterion. Ischemic heart disease: NO2 and
SO2, chosen by stepwise regression.
Other pollutants not chosen.
Dvsrhythmias: polluO3, CO, and PM25
chosen by stepwise regression. Other
pollutants not chosen. Conqestive heart
failure: NO2 and CO chosen by stepwise
regression procedure, other pollutants not
chosen in stepwise regression.
Comment
Study not used to
estimate incidence: no
study specific conversion
available between SO4
and PM2Sor PMW.
Two pollutant model (O3,
PM25_10) used to estimate
cardiac incidence.
Study not used to
estimate incidence:
limited endpoint.
PM25, PM25.10, and PM10
estimated from TSP,
COM, and sulfate (SO4)
data. Multiple pollutant
models used to estimate
ischemic heart disease
(NO2, SO2), dysrhythmias
(PM2S, CO, O3), and
congestive heart failure
incidence (CO, NO2).
D-40
-------
Study
Morgan et al.
(1998)
Morris et al.
(1995)
Morris and
Naumova (1998)
Pantazopolou et
al. (1995)
Schwartz and
Morris (1995)
Location and
Period
Sydney,
Australia
1/90-12/94
seven U.S.
cities
1/86-12/89
Chicago, IL
1/86-12/89
Athens,
Greece
1988
Detroit, Ml
1/86-12/89
Endpoint
Population (ICD code)
0-64; >64 heart disease
(410,413,427-
428)
>64 congestive
heart failure
(428)
>64 congestive
heart failure
(428)
all ages cardiac
(not defined by
ICD code)
>64 ischemic heart
disease (410-
414);
dysrhythmias
(427);
congestive
heart failure
(428)
Pollutants
O3, NO2,
bscat
(measure
of light
scattering)
O3, CO,
NO2, SO2
O3, CO,
NO2, SO2,
PM10
BS (black
smoke),
CO, NO2
O3, CO,
S02, PM10
Main Findings
Single pollutant models: bscat significant
for ages >64; NO2 significant in ages 0-64
and >64. Three pollutant model: NO2
significant in ages >64; O3 and bscat not
significant.
Single pollutant models: CO, NO2, and SO2
significant in single pollutant models. Four
pollutant model: CO is significant in five of
the seven cities; NO2 is significant in one
city; SO2 and O3 are not significant in any
cities.
Single pollutant models: CO, NO2, SO2,
and PM10 significant. Five pollutant model:
CO significant (Cl for RR=1.03-1.12); PM10
borderline significant (Cl for RR=0.99-
1.06); other pollutants not significant.
Single pollutant models: BS, CO, NO2
significantly related to cardiac admissions
in the winter. No significant effect found
for any pollutant in the summer.
Ischemic heart disease: Two pollutant
models: PM10and CO both significant; no
effect seen for SO2 and O3. Dvsrhythmias:
Air pollutants did not have a significant
effect. Congestive heart failure: Single
pollutant models: PM10 and CO significant;
SO2 and O3 not significant. Two pollutant
models: PM10 significant, controlling for CO
and SO2. Controlling for PM10, CO
significant.
Comment
Results from three
pollutant model for ages
0-64 not presented. Study
not used to estimate
incidence: study outside
U.S. and Canada.
Study not used to
estimate incidence: no PM
measure used in the
study, plus limited
endpoint.
Study not used to
estimate incidence:
limited endpoint.
Study not used to
estimate incidence: study
outside U. S. and Canada.
Two pollutant models
(PM10, CO) used to
estimate ischemic heart
disease and congestive
heart failure incidence.
D-41
-------
Location and
Study Period Population
Schwartz (1999) Eight U.S. >64
counties
1/88-12/90
Schwartz (1997) Tucson, AZ >64
1/88-12/90
Yang et al. Reno/Sparks, all ages
(1998) NV
1/89-12/94
Endpoint
(ICD code) Pollutants
cardiovascular CO, PM10
disease (390-
429)
cardiovascular O3, CO,
disease (390- SO2, NO2,
429) PM10
cardiovascular CO
illness (390-
459)
Main Findings
Two pollutant model: CO and PM10 both
significant.
In a model with the two pollutants, CO and
PM10 were both significant. No effect seen
for O3, SO2, and NO2.
Reported significant relationship between
CO and admissions.
Comment
Two pollutant model
(PM10, CO) used to
estimate incidence of
cardiovascular
admissions.
Two pollutant model
(PM10, CO) used to
estimate incidence of
cardiovascular
admissions.
Study not used to
estimate incidence: no PM
measure used in the
study.
Table D-9
Studies Used to Develop Asthma Emergency Room Visits
Location Study
central and northern NJ Cody et al. (1992)
central and northern NJ Weisel et al. (1995)
Seattle, WA Schwartz et al. (1993)
St. John, New Stieb et al. (1996)
Brunswick, Canada
Endpoints Estimated
asthma
asthma
asthma
asthma
Pollutants Used in Final
Model
03
03
PM10
03
Study Population
all ages
all ages
<65
all ages
D-42
-------
Table D-10
Summary of Selected Studies for Emergency Room Visits - Asthma
Study
Atkinson et
al. (1999)
Bates et al.
(1990)
Buchdahl et
al. (1996)
Castellsagu
e et al.
(1995)
Cody et al.
(1992)
Goldstein
and
Weinstein
(1986)
Location and
Period Population Endpoint
London, England 0-14; 15- asthma
1/92-12/94 64; >64; all
ages
Vancouver, 1-14; 15- asthma
Canada 60; >60
7/84-10/86
London, England <17 acute
3/92-2/93 wheezing
Barcelona, Spain >14 asthma
January-March and
July-September in
1985-1989
central and all ages asthma
northern NJ
May-August in
1988-1989
New York City alleges asthma
1/69-2/72
Pollutants
NO2, BS
(black
smoke),
PM10, S02,
CO, O3
NO2, SO2,
S04, 03
NO2, SO2,
03
NO2, BS,
S02, 03
PM10, S02,
03
SO2
and Acute Wheezing
Main Findings
Single pollutant models: PM10 and NO2
significantly related to asthma visits in all age
groups. SO2 significant in ages 0-14. BS is
significant for ages 15-64. No effect seen for
O3. Two pollutant results only for ages 0-14:
NO2 and SO2 significant; other pollutants not
significant.
SO4 correlated with asthma in all age groups
with some variation by season. SO2
correlated with asthma visits in ages 15 and
up. No effect found for NO2 and O3.
SO2 significantly related to acute wheezing.
O3 has a significant, U-shaped result,
suggesting that the optimal level of ozone is
not zero. NO2 not significant.
Single pollutant models: NO2 significant in
both July-September and January-March. BS
linked to asthma ER visits in July-September.
No significant effect found for SO2 and O3.
Two pollutant model: O3 linked to asthma
visits; SO2 not significant. No significant effect
seen for PM10; PM10 considered in separate
analysis, because of limited (every sixth day)
sampling.
No significant correlation found between SO2
and asthma ER visits.
Comment
Study not used to
estimate incidence: study
outside U. S. and Canada.
Study not used to
estimate incidence:
correlations only
presented.
Study not used to
estimate incidence: study
outside U. S. and Canada.
Study not used to
estimate incidence: study
outside U. S. and Canada.
Single pollutant model
(O3) used to estimate
incidence of asthma
visits.
Study not used to
estimate incidence: only
SO2 in the analysis.
D-43
-------
Study
Lipsett et al.
(1997)
Richards et
al. (1981)
Romieu et
al. (1995)
Rosas et al.
(1998)
Schwartz et
al. (1993)
Stieb et al.
(1996)
Weisel et al.
(1995)
Location and
Period
Santa Clara
County, CA
November-January
in 1988-1992
Los Angeles, CA
8/79-1/80
Mexico City,
Mexico
1/90-6/90
Mexico City,
Mexico
1991
Seattle, WA
9/89-9/90
St. John, New
Brunswick, Canada
May-September in
1984-1992
central and
northern NJ
May-August in
1986-1990
Population
all ages
children
(median
age =6)
<16
<15; 16-59;
>59
<65; >64
0-15; >15;
all ages
all ages
Endpoint
asthma
asthma and
bronchiolitis
(92%
asthma
only)
asthma
asthma
asthma
asthma
asthma
Pollutants
PM10, COM
(coefficient
of haze),
03, N02
COM, HC
(hydrocar-
bons), NO,
NO2, O3,
SO2, SO4,
TSP
S02, 03
03, S02,
N02, PM10,
TSP
S02, PM10,
03
NO2, TSP,
SO2, SO4,
03
03
Main Findings
Single pollutant models: NO2, PM10, and COM
significant; O3 not significant. Two pollutant
models: PM10 and COM linked to ER visits
controlling for NO2; NO2 not significant. PM10
reported to provide a slightly better fit than
COM
COM, HC, NO, and NO2 have positive and
significant correlation with ER visits; O3 and
SO2 have negative significant correlation; SO4
and TSP have insignificant correlation.
Two pollutant model: O3 significant and SO2
marginally significant.
Little effect found for air pollutants. Strong
effect found for aeroallergens, such as grass
pollen.
Single pollutant models: PM10 linked to ER
visits in ages <65, with no effect in ages >64.
No effect for SO2 and O3 on ER visits in either
age group.
O3 linked to ER visits in ages >15, especially
when O3 levels exceed 75 ppb; O3 not
significant in ages 0-15. No significant effect
seen for the other pollutants.
O3 linked to ER visits.
Comment
PM10 estimated from COM
observations. Study not
used to estimate
incidence: results depend
on temperature
interaction that we cannot
model.
13% of reported visits
subsequently admitted to
the hospital. Study not
used to estimate
incidence: correlations
only presented.
Study not used to
estimate incidence: study
outside U. S. and Canada.
Study not used to
estimate incidence: study
outside U. S. and Canada.
O3 only available May-
September. Single
pollutant model (PMW)
used to estimate
incidence of asthma
visits.
TSP and SO4 gathered
every sixth day. Single
pollutant model (O3) used
to estimate incidence of
asthma visits.
Single pollutant model
used.
D-44
-------
Location and
Study Period Population Endpoint Pollutants Main Findings Comment
White etal. Atlanta, GA 1-16 asthma and O3, SO2, O3 linked to ER visits when O3 levels PM10 estimated from
(1994) 6/90-8/90 restrictive PM10 exceeded 110 ppb. No significant effect visibility levels. Study not
airway reported for SO2 or PM10. used to estimate
disease incidence: limited study
data.
D-45
-------
Table D-1 1
Summary of Selected Studies for Emergency Room Visits -All-Cause, All-Respiratory, Chronic Obstructive Pulmonary Disease
(COPD), and Bronchitis
Study
Atkinson et al.
(1999)
Bates et al.
(1990)
Cody et al.
(1992)
Delfino et al.
(1997)
Delfino et al.
(1998)
Samet et al.
(1981)
Location and
Period
London,
England
1/92-12/94
Vancouver,
Canada
7/84-10/86
central and
northern New
Jersey
May-August
1988-1989
Montreal,
Canada
June-
September
1992-1993
Montreal,
Canada
June-August
1989-1990
Steubenville,
Ohio
March-April
and October-
November
1974-1977
Endpoint
Population (ICD code)
0-14; 15-64; all
>64; all respiratory
ages (not defined
by ICD code)
1-14; 15-60; all
>60 respiratory
(466, 480-
486, 491-
493, 496)
all ages bronchitis
(466, 490,
491,496)
<2; 2-64; all
>64 respiratory
(not defined
by ICD code)
>64 all
respiratory
(not defined
by ICD code)
all ages all
respiratory
(not defined
by ICD code)
Pollutants
NO2, BS
(black
smoke),
PM10, S02,
CO, O3
NO2, SO2,
S04, 03
PM10,
SO2, O3
03, PM10,
PM2.5,
S04, H+
PM25, O3
NO2, TSP,
SO2, CO,
03
Main Findings
Single pollutant models: PM10 significant in
ages 0-14 and 15-64. BS and SO2
significant in ages 0-14. CO and NO2
significant in ages >64. O3 not significant.
SO2 correlated with respiratory visits in all
age groups. SO4 correlated in ages >14.
NO2 correlated in ages 15-60. O3 not
significant.
No significant effect seen for PM10, O3, or
SO2 on bronchitis admissions.
Single pollutant models: H+ and SO4
significant in ages <2; no effect in ages 2-
64 for any pollutants; O3, PM10, PM25, and
SO4 significant in ages >64. Two pollutant
model: O3 significant and PM25 not
significant in ages >64.
Two pollutant model: O3 significant; PM25
has consistent link but not significant.
Single pollutant models: TSP and SO2
significant; NO2,, CO, or O3 were not
significant.
Comment
Study not used to estimate
incidence: study outside U.S.
and Canada.
Results varied somewhat by
season. Study not used to
estimate incidence:
correlations only presented.
PM10 sampled every sixth
day, so limited dataset. PM10
considered in separate
analysis. Study not used to
estimate incidence.
Limited number of results
presented for two pollutant
models. Study not used to
estimate incidence: all
respiratory not defined by
ICD code.
PM25 measured every sixth
day, with rest of daily
observations estimated from
visibility and other data.
Study not used to estimate
incidence: all respiratory not
defined by ICD code.
Study not used to estimate
incidence: all respiratory not
defined by ICD code.
D-46
-------
Location and Endpoint
Study Period Population (ICD code) Pollutants Main Findings Comment
Pantazopolou
etal. (1995)
Athens,
Greece
1988
all ages all outpatient
visits
BS, CO,
NO2
Single-pollutant models: NO2 significant in
the winter. No effects found for any
pollutant in the summer.
Study not used to estimate
incidence: study outside U.S.
and Canada.
Sunyeretal. Barcelona, >14 COPD (not SO2, BS SO2 correlated with ER visits in the summer Study not used to estimate
(1993) Spain defined by and winter. BS significant in the winter only incidence: study outside U.S.
1985-1989 ICD code) and Canada.
Xu et al. Beijing, China all ages all causes SO2, TSP SO2 and TSP both linked to ER visits. Study not used to estimate
(1995b) 1990 incidence: study outside U.S.
and Canada.
D-47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Minor Illness
In addition to chronic illnesses and hospital
admissions, there is a considerable body of scientific
research that has estimated significant relationships
between elevated air pollution levels and other
morbidity health effects. Chamber study research has
established relationships between specific air pollution
chemicals and symptoms such as coughing, pain on
deep inspiration, wheezing, eye irritation and
headaches. In addition, epidemiological research has
found air pollution relationships with acute infectious
diseases (e.g., bronchitis, sinusitis) and a variety of
"symptom-day" categories. Some "symptom-days"
studies examine excess incidences of days with
identified symptoms such as wheezing, coughing, or
other specific upper or lower respiratory symptoms.
Other studies estimate relationships for days with a
more general descriptions of days with adverse health
impacts, such as "respiratory restricted activity days"
or work loss days.
A major challenge in preparing an analysis of the
minor morbidity effects is identifying a set of effect
estimates that reflects the full range of identified
adverse health effects but avoids double counting.
From the definitions of the specific health effects
examined in each research project, it is possible to
identify a set of effects that are non-overlapping, and
can be ultimately treated as additive in the monetary
benefits analysis. This section primarily focuses on
the set of effect relationships that have been identified
that make up a non-overlapping set. Table D-12
summarizes the studies used in estimating minor
illnesses; Tables D-13 and D-14 provide more detailed
information on these studies and other studies that
were not used in the analysis.
Acute Bronchitis
Dockery et al. (1996) examined the relationship
between PM and other pollutants on the reported
rates of asthma, persistent wheeze, chronic cough, and
bronchitis, in a study of 13,369 children ages 8-12
living in 24 communities in U.S. and Canada. Health
data were collected in 1988-1991. Single-pollutant
models were used in the analysis. Annual levels of
sulfates and particle acidity were significantly related
to bronchitis, and PM25 and PM10 were marginally
significant. Earlier work, based on six U.S. cities, by
Dockery et al. (1989) found acute bronchitis and
chronic cough significantly related to PM15. Because
it is based on a larger sample, the Dockery et al. (1996)
study is used to develop a C-R function linking PM2 5
with acute bronchitis.
Upper Respiratory Symptoms (URS)
Using logistic regression, Pope et al. (1991)
estimated the impact of PM10 on the incidence of a
variety of minor symptoms in 55 subjects (34 "school-
based" and 21 "patient-based") living in the Utah
Valley from December 1989 through March 1990.
The children in the Pope et al. study were asked to
record respiratory symptoms in a daily diary, and the
daily occurrences of upper respiratory symptoms
(URS) and lower respiratory symptoms (LRS), as
defined above, were related to daily PM10
concentrations. Pope et al. describe URS as consisting
of one or more of the following symptoms: runny or
stuffy nose; wet cough; and burning, aching, or red
eyes. Levels of ozone, NO2, and SO2 were reported
low during this period, and were not included in the
analysis. The sample in this study is relatively small
and is most representative of the asthmatic
population, rather than the general population. The
school-based subjects (ranging in age from 9 to 11)
were chosen based on "a positive response to one or
more of three questions: ever wheezed without a cold,
wheezed for 3 days or more out of the week for a
month or longer, and/or had a doctor say the 'child
has asthma' (Pope et al., 1991, p. 669)." The patient-
based subjects (ranging in age from 8 to 72) were
receiving treatment for asthma and were referred by
local physicians. Regression results for the school-
based sample (Pope et al., 1991, Table 5) show PM10
significantly associated with both upper and lower
respiratory symptoms. The patient-based sample did
not find a significant PM10 effect. The results from
the school-based sample are used here.
D-48
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Lower Respiratory Symptoms (LRS)
Schwartz et al. (1994) used logistic regression to
link lower respiratory symptoms in children with SO2,
NO2, ozone, PM10, PM25, sulfate and H+ (hydrogen
ion). Children were selected for the study if they were
exposed to indoor sources of air pollution: gas stoves
and parental smoking. The study enrolled 1,844
children into a year-long study that was conducted in
different years (1984 to 1988) in six cities. The
students were in grades two through five at the time
of enrollment in 1984. By the completion of the final
study, the cohort would then be in the eighth grade
(ages 13-14); this suggests an age range of 7 to 14.
Respiratory Illness
Several epidemiological studies report that NO2
exposure increases risk of respiratory illness in
children. The results of many of the studies are not
statistically significant. In addition, many of the
studies do not provide ambient NO2 measurements,
having focused on the presence or absence of gas
stoves as surrogates for exposure. However, there are
data available from a well-designed study with
adequate ambient exposure measurements. Based on
work by Melia et al. (1980; 1982), Hasselblad et al.
(1992) examined data from 103 children in homes
where gas stoves were present and where bedroom
NO2 measurements were taken. A significant increase
in respiratory illness was found to be a function of
bedroom NO2 levels, independent of social class, age,
gender, or the presence of a smoker in the house.
Hasselblad et al. used a multiple logistic model fitted
to the Melia data with a linear slope for NO2 and
separate intercepts for boys and girls. This analysis
uses the average slope of these two estimates.
Work Loss Days (WLD)
Ostro (1987) estimated the impact of PM on the
incidence of work-loss days (WLD) in a national
sample of the adult working population, ages 18 to 65,
living in metropolitan areas. Separate coefficients
were developed for each year in the analysis (1976-
1981); we then combined these coefficients for use in
this analysis using a weighted average based on the
inverse of the variances.
Minor Restricted Activity Days (MRAD) /
Any of 19 Respiratory Symptoms
Two studies by Ostro and Rothschild (1989b) and
Krupnick et al. (1990) cover a variety of minor
respiratory symptoms. To avoid double counting, we
treat these two studies as alternative measures of the
same health effect, and pool the incidence estimates.
Ostro and Rothschild (1989b) estimated the
impact of ozone and PM25 on the incidence of minor
restricted activity days (MRAD) in a national sample
of the adult working population, ages 18 to 65, living
in metropolitan areas. We developed separate
coefficients for each year in the analysis (1976-1981),
which were then combined for use in this analysis.
The coefficient used in this analysis is a weighted
average of the coefficients using the inverse of the
variance as the weight.
Krupnick et al. (1990) estimated the impact of
coefficient of haze (COH, a measure of particulate
matter concentrations), ozone and SO2 on the
incidence of any of 19 respiratory symptoms or
conditions.14 They used a logistic regression model
that takes into account whether a respondent was well
or not the previous day. A key difference between
this and the usual logistic model is that the model they
used includes a lagged value of the dependent variable.
Moderate or Worse Asthma
This health endpoint comes from Ostro et al.
(1991), a study in which asthmatics, ages 18 to 70,
were asked to record daily a subjective rating of their
overall asthma status each day (0=none, l=mild,
2=moderate, 3=severe, 4=incapacitating). Ostro et al.
14Krupnick et al. (1990) list 13 specific "symptoms or
conditions": head cold, chest cold, sinus trouble, croup, cough
with phlegm, sore throat, asthma, hay fever, doctor-diagnosed ear
infection, flu, pneumonia, bronchitis, and bronchiolitis. The other
six symptoms or conditions are not specified.
D-49
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
then examined the relationship between moderate (or
worse) asthma and H+, sulfate, SO2, PM25, estimated
PM25, PM10, nitrate, and nitric acid. The published
results used in the prospective analysis are from a
single-pollutant linear regression model where the log
of the pollutant is used.
Asthma "attacks" associated with ozone are
estimated using the study by Whittemore and Korn
(1980). Symptoms in asthmatic children associated
with SO2 are from Linn et al. (1987; 1988; 1990) and
Roger etal. (1985).
Shortness of Breath
Using a logistic regression estimation, Ostro et al.
(1995) estimated the impact of PM10, ozone, NO2, and
SO2 on the incidence of coughing, shortness of
breath, and wheezing in 83 African-American
asthmatic children ages 7-12 living in Los Angeles
from August through September 1992. Regression
results show both PM10 and ozone significantly linked
to shortness of breath; the beginning of an asthma
episode was also significantly linked to ozone. Results
for single-pollutant models only were presented in the
published paper.
Restricted Activity Days (RADs)
Ostro (1987) used a log-linear regression to
estimate the impact of PM25 on the incidence of
restricted activity days (RAD) in a national sample of
the adult population, ages 18 to 65, living in
metropolitan areas. Separate coefficients were
developed for each year in the analysis (1976-1981);
these coefficients were pooled. The coefficient used
in the concentration-response function used here is a
weighted average of the coefficients in Ostro (1987,
Table III) using the inverse of the variance as the
weight.
D-50
-------
TableD-12
Studies Used to Develop Minor Illness
Endpoints Estimated
acute bronchitis
upper respiratory symptoms
lower respiratory symptoms
respiratory illness
any of 19 respiratory symptoms
moderate or worse asthma
asthma attacks
Estimates
Study Population Age
8-12
9-11
7-14
6-7
18-65
all ages (asthmatics)
all ages (asthmatics)
chest tightness, shortness of breath, or wheeze all ages (asthmatics)
shortness of breath
work loss days
minor restricted activity days
restricted activity days
7-12 (African-American asthmatics)
18-65
18-65
18-65
Study
Dockery et al. (1996)
Popeetal. (1991)
Schwartz et al. (1994)
Hasselblad et al. (1992)
Krupnick et al. (1990)
Ostroetal. (1991)
Whittemore and Korn (1980)
Linnetal. (1987; 1988; 1990)
and Roger et al. (1985)
Ostroetal. (1995)
Ostro(1987)
Ostro and Rothschild (1989b)
Ostro(1987)
Pollutants Used in Final
Model
PM25
PM10
PM25
NO2
Ps, PMio
PM25
03, PM10
SO2
PM10
PM25
PM25, O3
PM25
D-51
-------
TableD-13
Summary of Selected Studies for Minor Illness
Study
Dockery et
al. (1996)
Dockery et
al. (1989)
Hasselblad
etal. (1992)
Hoek and
Brunekreef
(1995)
Krupnick et
al. (1990)
Ostro
(1987)
Location and
Period
24 communities
in U.S. and
Canada
1988-1991
Six U.S. cities
1980-1981
Meta-analysis of
11 studies from
the U.S. and
Europe
Two rural towns
in the
Netherlands.
Spring-Summer
1989
Three
communities in
Los Angeles
County,
California
9/78-3/79
Nationwide
sample from U.S.
Health Interview
Survey
1976-1981
Population
13,369
children
ages 8-12
5,422
children
ages 10-12
children
ages 5-12
300 children
ages 7-11
570 adults
and 756
children
Adults ages
18-65
Endpoint
asthma, persistent
wheeze, chronic
cough, bronchitis
bronchitis, chest
illness, cough,
wheeze, asthma
lower respiratory tract
illness
symptoms including:
cough, phlegm,
wheeze, runny nose,
throat pain,
headache, eye
irritation, physician
visit
any of 19 respiratory
symptoms including
cough with phlegm
work-loss days
restricted activity
days (RADs),
respiratory-related
RADs
Pollutants
particle
acidity,
S04, PM21,
PM10,
HNO2,
HNO3, O3
SO2, NO2,
03, PM25,
PM15, TSP,
S04
NO2
SO2, NO2,
03, PM25,
PM10,
PM25.10,
SO4, NO3
O3, COM
(coefficient
of haze),
SO2, NO2
PM25
Main Findings Comment
Annual level of sulfates and particle Study examined annual
acidity related to bronchitis. HNO2 pollution exposures, and
and HNO3 linked to asthma. SO2 the authors did not rule out
linked to chronic phlegm. that acute (daily)
exposures could be related
to asthma attacks and
other acute episodes.
Annual level of PM15 significantly
related to bronchitis and chronic
cough. Annual O3 significantly
related to asthma.
Annual NO2 change of 30 ug/m3
associated with lower respiratory
tract illness.
Daily pollutant levels not associated
with any of the symptoms studied.
In single pollutant models, daily O3,
COM, and SO2 related to respiratory
symptoms in adults. O3 significant
controlling for other pollutants.
Results more variable for COM and
SO2, perhaps due to collinearity.
NO2 had no significant effect. No
effect seen in children for any
pollutant.
Two-week average PM25 levels PM25 estimated from
significantly linked to work-loss days, visibility data.
RADs, and respiratory-related RADs.
Some year-to-year variability in
results.
D-52
-------
Study
Ostro et al.
(1993)
Ostro and
Rothschild
(1989b)
Peters et al.
(1999)
Pope and
Dockery
(1992)
Pope et al.
(1991)
Location and
Period
Three
communities in
Los Angeles
County,
California
9/78-3/79
Nationwide
sample from U.S.
Health Interview
Survey
1976-1981
Twelve
communities in
southern
California
1994
Utah Valley
12/90-3/91
Utah Valley
12/89-3/90
Population
321 non-
smoking
adults
Adults ages
18-65
3,676 fourth,
seventh,
tenth grade
students
79 children
ages 10-12
34 children
ages 9-11,
and 21
asthmatics
ages 8-72
Endpoint
lower respiratory
symptoms, upper
respiratory
symptoms, eye
irritation
respiratory-related
RADs, minor RADs.
asthma, wheeze,
bronchitis, cough
upper respiratory
symptoms, lower
respiratory
symptoms, cough
upper respiratory
symptoms, lower
respiratory
symptoms, took
asthma medication
Pollutants
O3, COM,
SO4, SO2,
NO2
03, PM25
SO2, NO2,
03, PM25,
PM10,
PM25.10,
SO4, NO3,
NH4,
gaseous
acids
PM10
PM10
Main Findings
In single pollutant model, daily O3
linked to lower and upper respiratory
symptoms. SO4 linked to lower
respiratory symptoms. No significant
effects seen for COM, SO2, and NO2
Controlling for PM25, two-week
average O3 has highly variable
association with respiratory-related
and minor RADs. Controlling for O3,
two-week average PM25 significantly
linked to both health endpoints in
most years.
Wheeze in males, linked to annual
average NO2 and acid in 1994
(similar link for exposure averaged
over 1 986-1 990). Peak ozone
reported associated with decreased
asthma prevalence in females. No
other reported effects.
PM10 linked to daily reported
incidences of upper and lower
respiratory symptoms and cough.
Effect seen in symptomatic sample.
Only cough in symptomatic sample
linked to PM10.
PM10 significantly linked to upper and
lower respiratory symptoms in
sample of 34 children. PM10 linked
only to increased asthma medication
use in the asthmatic sample.
Comment
PM25 estimated from
visibility data.
Of the 79 children in the
sample, 39 were
symptomatic, and the other
40 were asymptomatic.
D-53
-------
Study
Schwartz et
al. (1994)
Schwartz
and Zeger
(1990)
von Mutius
etal. (1995)
Location and
Period
Six U.S. cities
April-August in
one year
between 1984
and 1988 (year
varies by city)
Los Angeles, CA
1961-1964
Leipzig,
Germany
10/91-7/92
Population
1,844
children
110 student
nurses
1,500
children
ages 9-11
Endpoint
upper respiratory
symptoms, lower
respiratory
symptoms, cough
cough, phlegm, sore
throat, headache,
chest discomfort, eye
irritation
upper respiratory
symptoms
Pollutants
SO2, NO2,
O3, PM25,
PM10, S04,
H+
CO, SO2,
N02, Ox
S02, NOX,
PM
Main Findings
In single pollutant models SO2, NO2,
PM25, and PM10 significantly linked to
cough. In two-pollutant models, PM10
has most consistent effect; other
pollutants not significant, controlling
for PM10. In single pollutant models,
SO2, O3, PM25, PM10, SO4, and H+
linked to lower respiratory symptoms.
No effect seen for upper respiratory
symptoms.
NO2 linked to sore throat, phlegm,
and eye irritation. Oxidants (Ox)
linked to chest discomfort and eye
irritation. CO linked to headache.
In single pollutant models, SO2, NOX,
and PM linked to upper respiratory
symptoms in winter (high pollution
season). In the summer, only NOX
linked to respiratory symptoms.
Comment
Results presented as a mix
of single pollutant and dual
pollutant models. Stepwise
selection used to pick
significant covariates.
PM measured by beta-
absorption. The limited
modeling results presented
for models with more than
one pollutant were similar
to single pollutant results.
D-54
-------
TableD-14
Summary of Selected Studies for Asthmatics
Study
Forsberg et
al. (1993)
Gielen et
al. (1997)
Hiltermann
etal.
(1998)
Linn et al.
(1987;
1988;
1990) and
Roger et al.
(1985)
Location and
Period
Pitea, Sweden
about sixty days
Amsterdam,
Netherlands
summer
1995
Leiden
University,
Netherlands
7/3/95-10/6/95
Chamber
studies.
Population
31 persons
ages 9-71
61 children
ages 7-13
60 adults
ages 18-55
Exercising,
young
asthmatics
Endpoint
shortness of
breath,
wheeze, cough,
phlegm
upper
respiratory
symptoms,
lower
respiratory
symptoms,
medication use
symptoms
include:
shortness of
breath, cough,
phlegm,
wheeze, runny
nose, throat
pain,
headache, eye
irritation,
physician visit
chest tightness,
shortness of
breath, or
wheeze
Pollutants Main Findings
BS (black Controlling for other pollutants, daily levels
smoke), of BS linked to shortness of breath. No link
SO2, NO2 between pollutants and wheeze, cough, and
phlegm.
O3, PM10, In single pollutant model, daily levels of BS
BS significantly linked to lower and upper
respiratory symptoms and medication use.
PM10 linked to lower respiratory symptoms
and medication use. O3 linked to upper
respiratory symptoms.
O3, PM10, In single pollutant models, daily levels of O3,
BS, NO2, PM10, BS, and NO2 linked to shortness of
SO2 breath. Some significant negative
associations reported for nasal symptoms
and levels of PM10, BS, and NO2. No
significant effect reported for SO2.
SO2 SO2 exposure linked to moderate symptoms
in these studies of moderately exercising
young asthmatics.
Comment
Black smoke is an indirect
measure of PM.
Results in the model highly
dependent on the lag
length used. The five-day
mean black smoke and
PM10 yielded significant
results, but current, one
and two day lags did not.
Current day O3 significant.
D-55
-------
Study
Neukirch et
al. (1998)
Ostro et al.
(1991)
Ostro et al.
(1995)
Peters et
al. (1996)
Roemer et
al. (1998)
Romieu et
al. (1996)
Location and
Period
Paris, France
11/92-5/93
Denver, CO
12/87-2/88
Los Angeles, CA
8/92-11/92
Three cities in
East Germany
and the Czech
Republic
9/90-6/92
28 locations in
Europe
winter 1993-
1994
Mexico City
Population
40 persons
(mean age
of sample
was 46)
207
persons
ages 18-70
83 children
ages 7-12
155 children
ages 7-15
and 102
adults ages
32-80
2,010
children
ages 6-12
71 children
ages 5-7
Endpoint
asthma,
wheeze,
shortness of
breath, cough,
respiratory
infection
severity of
asthma
symptoms,
cough, wheeze,
shortness of
breath, chest
tightness
cough,
shortness of
breath, wheeze
symptom score
based on a
variety of
respiratory
symptoms
symptoms
include:
shortness of
breath, cough,
phlegm,
wheeze, runny
nose, sore
throat,
headache, eye
irritation
cough, phlegm,
difficulty
breathing,
wheezing,
lower
respiratory
illness
Pollutants
PM13, BS,
NO2, SO2
SO2, PM25,
SO4, NO3,
H+, nitric
acid
03, N02,
S02, PM10
TSP, SO2,
S04,
particle
acidity
PM10, BS,
NO2, SO2
PM10,
PM25, 03,
NO2, SO2
Main Findings
In single pollutant models, daily levels of
PM13, BS, NO2, and SO2were each
significantly associated with asthma
attacks, wheeze, cough, respiratory
infections, and shortness of breath.
Daily levels of H+ linked to cough, asthma,
and shortness of breath. PM25 linked to
asthma. SO4 linked to shortness of breath.
No effects seen for other pollutants.
In single pollutant models, daily levels of O3
and PM10 linked only to shortness of breath.
No effect seen for NO2 and SO2.
Daily SO2 linked to the respiratory symptom
score. No link between the other pollutants
and the symptom score.
Daily pollutant levels not related to adverse
health symptoms.
Controlling for PM25, daily levels of O3
linked to cough and lower respiratory
illness. Controlling for O3, PM25 linked to
cough, phlegm, and lower respiratory
symptoms.
Comment
PM13 used rather than the
more common PM10.
Some PM2 5 estimated.
Exclusion of estimated
data removes significant
link to asthma. Only single
pollutant models reported.
PM10 also linked adverse
symptoms. Published
results focused on O3 and
PM25. Results for NO2 and
SO2 not reported.
D-56
-------
Location and
Study Period Population Endpoint Pollutants Main Findings Comment
Whittemore Six communities 443 children asthma Ox, TSP In a two pollutant model, daily levels of both Respirable PM, NO2, SO2
and Korn in southern CA and adults TSP and Ox were significantly related to were highly correlated with
(1980) Three 34-week reported asthma attacks. TSP and excluded from the
periods 1972- analysis.
1975
D-57
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
C-R Functions Linking Air
Pollution and Adverse Health
Effects
After selecting studies appropriate for the present
analysis, the published information was used to derive
a C-R function for estimating nationwide benefits for
each health effect considered. In general, these
functions combine air quality changes, the affected
population and information regarding the expected
per person change in incidence per unit change in
pollutant level. The following tables present the
functions used in this analysis, information needed to
apply these functions, and references for information.
Carbon Monoxide
Four C-R relationships are available for estimating
hospital admissions related to ambient CO levels.
These are summarized in Table D-15.
D-58
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableD-15
Summary of C-R Functions for Carbon Monoxide
Health
End point
C-R Function
Source of C-R Function
hospital
admissions -
asthma
A.asthmaadmissions =- [y0 -(e /JACO -1)]- pop,
where:
y0 = daily hospital admission rate for asthma per person = 4.75 E-6
p = CO coefficient = 0.0332
AGO = change in daily average CO concentration (ppm)
pop = population of all ages
Op = standard error of p = 0.00861
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model:
PM2.5.io, 03
hospital
admissions -
obstructive
lung disease
Aobs. lung disease admissions =-[y0 -(e /5ACO -1)]-pop,
where:
y0 = daily hospital admission rate for obstructive lung disease per
person = 5.76 E-6
p = CO coefficient = 0.0250
AGO = change in daily average CO concentration (ppm)
pop = population of all ages
Op = standard error of p = 0.0165
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model:
PM2.5.io, 03
hospital
admissions -
COPD
ACOPDadmissions =-[y0 -(e^ACO -1)]-pop,
where:
y0 = daily hospital admission rate for COPD per person = 3.75 E-5
p = CO coefficient = 0.0573
A CO = change in daily average CO concentration (ppm)
pop = population age 65 and older
Op = standard error of p = 0.0329
Study: Moolgavkar (1997)
Location: Minneapolis-St. Paul
Other pollutants in model: O3,
PM,n
hospital
admissions-
asthma
Aasf/imaadmissions =-[y0 -(e MCO-1)]-pop,
where:
y0 = daily hospital admission rate for asthma per person = 4.52 E-6
p = CO coefficient = 0.0528
A CO = change in daily average CO concentration (ppm)
pop = population of ages < 65
Op = standard error of p = 0.0185
Study: Sheppard (1999)
Location: Seattle, WA
Other pollutants in model:
PM2.5
hospital
admissions -
dysrhythmias
Adysrhythmiasadmissions =-[/0 -(e /5ACO -1
where:
y0 = daily hospital admission rate for dysrhythmias per person = 6.46
E-6
p = CO coefficient = 0.0573
AGO = change in daily average CO concentration (ppm)
pop = population of all ages
Op = standard error of p = 0.0229
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model:
D-59
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Health
End point
C-R Function
Source of C-R Function
hospital
admissions -
congestive
heart failure
Acongestive heart failure admissions =-[/0 -(e /5ACO -1)]-pop,
where:
Vo
P
AGO
pop
OR
= daily hospital admission rate for congestive heart failure per
person = 9.33 E-6
= CO coefficient = 0.0340
= change in daily average CO concentration (ppm)
= population of all ages
= standard error of p = 0.0163
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model: NO2
hospital
admissions -
ischemic heart
disease
kischemic heart disease admissions = -[y0 -(e /5ACO -1)]-pop,
where:
Vo
P
AGO
pop
OR
= daily hospital admission rate for ischemic heart disease per
person 65 and older = 9.96 E-5
= CO coefficient= 0.000467
= change in daily one-hour maximum CO concentration (ppm)
= population age 65 and older
= standard error of p = 0.000435
Study: Schwartz and Morris
(1995)
Location: Detroit, Ml
Other pollutants in model:
PM,n
hospital
admissions -
congestive
heart failure
Acongestiveheart failureadmissions = -[y0 -(e /5ACO -1)]- pop,
where:
Vo
P
AGO
pop
OR
= daily hospital admission rate for congestive heart failure per
person 65 and older = 5.82 E-5
= CO coefficient = 0.0170
= change in daily one-hour maximum CO concentration (ppm)
= population age 65 and older
= standard error of p = 0.00468
Study: Schwartz and Morris
(1995)
Location: Detroit, Ml
Other pollutants in model:
PM,n
hospital
admissions -
cardiovascular
where:
Vo
P
AGO
pop
OR
Acardiovascularadmissions = -[y0 -(e /5ACO -1)]- pop,
= daily hospital admission rate for cardiovascular disease per
person 65 and older = 2.23 E-4
= CO coefficient = 0.0127
= change in daily one-hour maximum CO concentration
= population age 65 and older
= standard error of p = 0.00255
Study: Schwartz (1999)
Location: eight U.S. counties
Other pollutants in model:
PM10
hospital
admissions -
cardiovascular
where:
Vo
P
AGO
pop
OR
Acardiovascular admissions = -[y0 -(e /5ACO -1)]- pop,
= daily hospital admission rate for cardiovascular disease per
person 65 and older = 2.23 E-4
= CO coefficient = 0.0139
= change in daily one-hour maximum CO concentration
= population age 65 and older
= standard error of p = 0.00715
Study: Schwartz (1997)
Location: Tucson, AZ
Other pollutants in model:
PM10
D-60
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Nitrogen Dioxide
Nitrogen dioxide (NO2) is the primary focus of
health studies on the nitrogen oxides and serves as the
basis for this analysis. Table D-16 summarizes the C-
R functions that are used to quantify the relationship
between NO2 and adverse health effects.
D-61
-------
TableD-16
Summary of C-R Functions for Nitrogen Dioxide
Health Endpoint
C-R Function
Source of C-R Function
hospital admissions -
all respiratory
A/A// respiratory =-[/„ .
-1)]-pop,
where:
Vo
P
ANO2
pop
: daily hospital admission rate for all respiratory per person = 2.58 E-5
' NO2 coefficient = 0.00378
: change in daily 12-hour average NO2 concentration (ppb)
: population of all ages
•• standard error of p = 0.00221
Study: Burnett et al. (1997b)
Location: Toronto, Canada
Other pollutants in model: PM25.10, O3, SO2
hospital admissions -
respiratory infection
ARespiratorylnfectionAdmissions=-[ya-(e /J
hospital admissions -
pneumonia
Apneumoniaadmissions =-[y0
/JAW°2
where
Vo
P
= daily hospital admission rate for pneumonia per person = 5.30 E-5
= NO2 coefficient = 0.00169
A NO2 = change in daily average NO2 concentration (ppb)
pop = population age 65 and older
aB = standard error of p = 0.00125
Study: Burnett etal. (1999)
Location: Toronto, Canada
where:
y0 = daily hospital admission rate for respiratory infection per person = 1 .56 E-5
p = NO2 coefficient = 0.001 72
ANO2 = change in daily average NO2 concentration (ppb)
pop = population of all ages
Op = standard error of p = 0.000521
Other pollutants in model: PM25, O3
Study: Moolgavkar et al. (1997)
Location: Minneapolis, MN
Other pollutants in model: O3, SO2, PM10
hospital admissions -
congestive heart failure
where:
Vo
P
ANO2
pop
OR
ACongestive Heart Failure Admissions =-[ya-(e /MW°2 -1)]-pop,
= daily hospital admission rate for congestive heart failure per person = 9.33 E-6
= NO2 coefficient = 0.00264
= change in daily average NO2 concentration (ppb)
= population of all ages
= standard error of p = 0.000769
Study: Burnett etal. (1999)
Location: Toronto, Canada
Other pollutants in model: CO
D-62
-------
Health Endpoint
C-R Function
Source of C-R Function
hospital admissions -
ischemic heart disease
A.lschemicHeartDiseaseAdmissions =-[/0 '(
where:
y0 = daily hospital admission rate for ischemic heart disease per person = 2.23 E-5
p = NO2 coefficient = 0.00318
ANO2 = change in daily average NO2 concentration (ppb)
pop = population of all ages
Op = standard error of p = 0.000521
Study: Burnett etal. (1999)
Location: Toronto, Canada
Other pollutants in model: SO2
respiratory symptoms
Aresp. symptoms =\ N02 ^AAA.^gender.Y
\_ i ~t~"
where:
a = constant = -0.536
p = NO2 coefficient = 0.0275
Y = gender coefficient (used for males only) = -0.0295
ANO2 = change in annual NO2 concentration (ppb)
pop = children ages 6-7
OB = standard error of p = 0.0132
1
1 + e
-a-N02:p,sl_cAAA-p-gender-Y
•pop,
Study: Hasselblad et al. (1992)
Location: Middlesborough, England
Other pollutants in model: none
Comments: The NO2 coefficient was reported
by Hasselblad et al. The constant and the
gender coefficient were obtained via personal
communication with V. Hasselblad 2/28/95 by
Abt Associates. The equation is based on
study results by Melia et al. (1980).
D-63
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Ozone
The health effects literature includes studies of
the relationships between ozone and a variety of
health effects. Table D-17 summarizes the ozone C-R
functions used in this analysis.
D-64
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableD-17
Summary of C-R Functions for Ozone
Health
Endpoint
C-R Function
Source of C-R Function
mortality
^Mortality = -[/„ •
- 1)]- pop,
where:
Vo
P
AO3
pop
OR
: non-accidental deaths per person of any age
= ozone coefficient = 0.000634
: change in daily one-hour maximum ozone concentration (ppb)
: population all ages
= standard error of p = 0.000251
Study: Ito and Thurston (1996)
Location: Chicago, IL
Other pollutants in model: PM10
mortality
where:
Vo
P
AO3
pop
OR
^.Mortality = -[/„ • (e^A°3 -1)]• pop,
• non-accidental deaths per person of any age
: ozone coefficient = 0
: change in daily 1-hour maximum ozone concentration (ppb)
: population all ages
= standard error of p = 0.000214
Study: Kinneyetal. (1995)
Location: Los Angeles, CA
Other pollutants in model: PM10
mortality
A Mortality = -[/„ •
- 1)] • pop,
where:
Vo
P
AO3
pop
• non-accidental deaths per person of any age
= ozone coefficient = 0.00061 1
: change in daily average ozone concentration (ppb)
: population all ages
= standard error of p = 0.00021 6
Study: Moolgavkar et al. (1995)
Location: Philadelphia, PA
Other pollutants in model: SO2,
TSP
mortality
^.Mortality = -[/„ • (e^A°3 -1)]- pop,
Study: Samet et al. (1997)
Location: Philadelphia, PA
adult onset
asthma
where:
Vo
P
AO3
pop
°P
= non-accidental deaths per person
= ozone coefficient = 0.000936
of any age
Other pollutants in model: CO,
NO2, SO2, TSP
= change in daily average ozone concentration (ppb)
= population all ages
= standard error of p = 0.000312
r Yo
AChronic Asthrns — —\ — — ^Q—
L( -Yol-e
1
/j y0 -pop,
+ /0 J
Study: McDonnell et al. (1999)
Location: California
Other pollutants in model: none
where:
y0 = annual asthma incidence rate per person = 0.00219
p = estimated O3 coefficient = 0.0277
AO3 = change in annual average 8-hour O3 concentration
pop = population of non-asthmatic males ages 27 and older
OR = standard error of p = 0.0135
hospital
admissions -
all respiratory
\Allrespiratory =-[y0 .(e-"A°3 -1)]-pop,
where:
y0 = daily hospital admission rate for all respiratory per person = 2.58
E-5
p = O3 coefficient = 0.00498
AO3 = change in daily 12-hour average O3 concentration (ppb)
pop = population of all ages
Op = standard error of p = 0.001 06
Study: Burnett et al. (1997b)
Location: Toronto, Canada
Other pollutants in model:
PM2.5.10, N02, S02
D-65
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Health
Endpoint
C-R Function
Source of C-R Function
hospital
admissions -
asthma
where:
y0
p
AO3
pop
AAsthma Admissions =-[y0 -(e I>'A°3 -\)\pop,
= daily hospital admission rate for asthma per person = 4.75 E-6
= ozone coefficient = 0.00250
= change in daily average ozone concentration (ppb)
= population of all ages
= standard error of p = 0.000718
Study: Burnett etal. (1999)
Location: Toronto, Canada
Other pollutants in model: CO,
PM2.5.10
hospital
admissions-
obstructive
lung disease
^Obstructive Lung Disease Admissions =-[/0 '(e ftA°3 -1)]'P°P.
where:
y0 = daily hospital admission rate for obstructive lung disease per
person = 5.76 E-6
p = ozone coefficient = 0.00303
AO3 = change in daily average ozone concentration (ppb)
pop = population of all ages
OB = standard error of p = 0.00110
Study: Burnett etal. (1999)
Location: Toronto, Canada
Other pollutants in model: CO,
PM,,.,n
hospital
admissions -
respiratory
infection
ARespiratory Infection Admissions =- [ya • (e /M°3 - 1)] • pop,
where:
y0 = daily hospital admission rate for respiratory infection per person =
Study: Burnett etal. (1999)
Location: Toronto, Canada
Other pollutants in model:
PM2.5, N02
1.56E-5
p = ozone coefficient = 0.00198
AO3 = change in daily average ozone concentration (ppb)
pop = population of all ages
OB = standard error of p = 0.000520
hospital
admissions-
all respiratory
where:
p
AO3
pop
OB
Aa// respiratory admissions =/?• A03 -pop,
ozone coefficient = 1 .68 E-8
change in daily one-hour maximum ozone concentration (ppb)
population all ages
standard error of p = 9.71 E-9 .
Study: Thurston et al. (1994)
Location: Toronto, Canada
Other pollutants in model: PM25
hospital
admissions-
pneumonia
where:
y0
p
AO3
pop
Apneumonia admissions =-[/0 '(e /JA°3 -1)]'P°P>
= daily hospital admission rate for pneumonia per person = 5.30 E-5
= O3 coefficient = 0.00370
= change in daily average O3 concentration (ppb)
= population age 65 and older
= standard error of p = 0.001 03
Study: Moolgavkar et al. (1997)
Location: Minneapolis, MN
Other pollutants in model: SO2,
N02, PM10
hospital
admissions-
COPD
where:
y0
p
AO3
pop
o
ACOPD admissions =-[ya •(e"'3403 -1)]-pop,
daily hospital admission rate for COPD per person = 3.75 E-5
O3 coefficient = 0.00274
change in daily average O3 concentration (ppb)
population age 65 and older
standard error of p = 0.001 70
Study: Moolgavkar et al. (1997)
Location: Minneapolis, MN
Other pollutants in model: CO,
PM,n
D-66
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Health
Endpoint
C-R Function
Source of C-R Function
hospital
admissions -
pneumonia
where:
Vo
P
AO3
pop
OR
Apneumoniaadmissions =-[/0 '(e /JA°3 -1)]'P°P.
= daily hospital admission rate for pneumonia per person = 5.30 E-5
= O3 coefficient = 0.00280
= change in daily average O3 concentration (ppb)
= population age 65 and older
= standard error of p = 0.00172
Study: Schwartz (1994c)
Location: Minneapolis, MN
Other pollutants in model: PM10
hospital
admissions-
pneumonia
where:
Vo
P
AO3
pop
OR
A.pneumoniaadmissions =-[y0 -(e /JA°3 -1)]-pop,
= daily hospital admission rate for pneumonia per person = 5.18 E-5
= O3 coefficient = 0.00521
= change in daily average O3 concentration (ppb)
= population age 65 and older
= standard error of p = 0.0013
Study: Schwartz (1994b)
Location: Detroit, Ml
Other pollutants in model: PM10
hospital
admissions-
COPD
A.COPD admissions =-[ya
-1)]-pop,
where:
y0
p
AO3
pop
= daily hospital admission rate for COPD per person = 3.05 E-5
= O3 coefficient = 0.00549
= change in daily average O3 concentration
= population age 65 and older
= standard error of p = 0.00205
Study: Schwartz (1994b)
Location: Detroit, Ml
Other pollutants in model: PM10
hospital
admissions -
all respiratory
where:
Vo
Aall respiratory admissions = -[ya -(e /JA°3 -1)]-pop,
= daily hospital admissions for all respiratory per person 65 and
Study: Schwartz (1 995)
Location: New Haven, CT
Other pollutants in model: PM10
older = 1.187E-4
p = ozone coefficient = 0.00265
AO3 = change in daily average ozone concentration (ppb)
pop = population age 65 and older
aB = standard error of p = 0.00140
hospital
admissions-
all respiratory
Aallrespiratory related admissions = -\y0 • (e /M°3 -1)]- pop,
where:
Vo
P
AO3
pop
OR
= daily hospital admissions for all respiratory conditions per person
65 and older = 1.187E-4
= ozone coefficient = 0.00715
= change in daily average ozone concentration (ppb)
= population age 65 and older
= standard error of p = 0.00257
Study: Schwartz (1995)
Location: Tacoma, WA
Other pollutants in model: PM10
hospital
admissions-
cardiac
Acardiac =- [y0
- 1)]- pop,
where:
y0
p
AO3
pop
= daily hospital admission rate for cardiac per person = 3.81 E-5
= O3 coefficient = 0.00531
= change in daily 1 2-hour average O3 concentration (ppb)
= population of all ages
= standard error of p = 0.001 42
Study: Burnett et al. (1997b)
Location: Toronto, Canada
Other pollutants in model:
PM,,.,n
D-67
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Health
Endpoint
C-R Function
Source of C-R Function
hospital
admissions -
dysrhythmias
where:
Vo
P
AO3
pop
ADysrhythmias Ad missions =-[y0 '(e /JA°3 -1)]'P°P.
= daily hospital admission rate for dysrhythmias per person = 6.46
E-6
= ozone coefficient = 0.00168
= change in daily average ozone concentration (ppb)
= population of all ages
= standard error of p = 0.00103
Study: Burnett etal. (1999)
Location: Toronto, Canada
Other pollutants in model:
PM25, CO
emergency
room visits -
asthma
Aasthma related ER visits =
BasePop
• AO3 • pop,
where:
p = ozone coefficient = 0.0203
BasePop = baseline population in northern New Jersey = 4,436,976
AO3 = change in daily five-hour average ozone concentration (ppb)
pop = population all ages
Op = standard error of p = 0.00717
Study: Cody etal. (1992)
Location: Northern NJ
Other pollutants in model: none
Comment: 63 % of estimate
used to avoid double-counting
hospital admissions for
asthma.
emergency
room visits -
asthma
^asthma related ER visits =-
BasePop
• A03 -pop,
where:
p = ozone coefficient = 0.0443
BasePop = baseline population in northern New Jersey = 4,436,976
AO3 = change in daily five-hour average ozone concentration (ppb)
pop = population all ages
OR = standard error of p = 0.00723
Study: Weisel et al. (1995)
Location: Northern, NJ
Other pollutants in model: none
Comment: 63 % of estimate
used to avoid double-counting
hospital admissions for
asthma.
emergency
room visits -
asthma
\asthmarelated ER visits =
BasePop
• AO3 • pop,
where:
p = ozone coefficient = 0.0035
BasePop = baseline population in Saint John, New Brunswick = 125,000
AO3 = change in the daily one-hour maximum ozone concentration
(Ppb)
pop = population all ages
a. = standard error of p = 0.0018
Study: Stieb et al. (1996)
Location: New Brunswick,
Canada
Other pollutants in model: none
Comment: 63 % of estimate
used to avoid double-counting
hospital admissions for
asthma.
presence of
any of 19
acute
respiratory
symptoms
where:
P*
AO3
pop
OR
= first derivative of the stationary probability = 0.000137
= change in daily one-hour maximum ozone concentration (ppb)
= population aged 18-65 years old
= standard error of p* = 0.0000697
Study: Krupnicket al. (1990)
Location: Glendora-Covina-
Azusa, CA
Other pollutants in model: SO2,
COM
self-reported
asthma
attacks
Aasthmaattacks = -
P°P,
where:
Vo
P
AO3
pop
= daily incidence of asthma attacks = 0.027
= ozone coefficient = 0.00184
= change in daily one-hour maximum ozone concentration (ppb)
= population of asthmatics of all ages = 5.61% of the population
of all ages
= standard error of p = 0.000714
Study: Whittemore and Korn
(1980)
Location: Los Angeles, CA
Other pollutants in model: TSP
D-68
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Health C-R Function Source of C-R Function
Endpoint
respiratory AMR/AD =-IV •(e"/JA°3 -1)1- pop Study: Ostro and Rothschild
and L ° J (1989b)
nonrespiratory where: Location: U.S.
conditions y0 = daily MRAD incidence per person = 0.02137 Other pollutants in model: PM25
resulting in a p = inverse-variance weighted PM25 coefficient = 0.00220 Comments: An inverse-
minor AO3 = change in two-week average of the daily one-hour maximum variance weighting used to
restricted ozone concentrations (ppb) estimate the coefficient, based
activity day pop = adult population aged 18 to 65 on Ostro and Rothschild.
(MRAD) OB = standard error of p = 0.000658
D-69
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Particulate Matter
The C-R functions used to quantify expected
changes in health effects associated with reduced
exposure to particulate matter are summarized in
Table D-18. The measures of particulate matter used
in this analysis are PM2 5 and PM10, with a preference
for PM25. Other measures of PM, however, have
been used, including total suspended particulates
(TSP) and coefficient of haze.
D-70
-------
TableD-18
Summary of C-R Functions for Particulate Matter
Health
End point
C-R Function
Source of C-R Function
mortality
ANonaccidental Mortality = -\y0
-1)]-pop,
where:
Vo
P
APM25
pop
: county-level annual non-accidental deaths of persons ages 30+ per person
= PM25 coefficient = 0.006408
: change in annual median PM25 concentration
: population ages 30 and older
= standard error of p = 0.001509
Study: Pope et al. (1995)
Location: 50 U.S. cities
Other pollutants in model: none
mortality
kNonaccidental Mortality = -[/„ . (
-1)]-pop,
where:
Vo
P
APM2.5
pop
county-level annual non-accidental deaths of persons ages 25+ per person
PM25 coefficient = 0.0124
change in annual mean PM25 concentration
population ages 25 and older
standard error of p = 0.00423
Study: Dockery et al. (1993)
Location: six U.S. cities
Other pollutants in model: none
neonatal
mortality
Alnfant Mortality = -
pop,
where:
Vo
P
APM10
pop
= county annual postneonatal infant deaths per infant 0-1 years old
PM10 coefficient = 0.00392
change in annual average PM10 concentration
population infants ages 0-1
standard error of p = 0.001 22
Study: Woodruff et al. (1997)
Location: 86 U.S. metropolitan areas
Other pollutants in model: none
chronic
bronchitis
where:
Vo
P
APM10
pop
^Chronic Bronchitis = -\y0 • (e^A™10 -1)]- pop,
= annual bronchitis incidence rate per person = 0.00378
= estimated PM10 logistic regression coefficient = 0.00932
= change in annual average PM10 concentration
= population ages 27 and older "without chronic bronchitis"
= standard error of p = 0.00475
Study: Abbey etal. (1993)
Location: California
Other pollutants in model: none
Comments: Abbey et al. used TSP to measure
PM. The TSP coefficient is applied to changes
in PM10.
D-71
-------
Health
End point
C-R Function
Source of C-R Function
chronic
bronchitis
where:
Vo
P
APM2.5
pop
AChronic Bronchitis = -\y0 • (e /JA™25 -1)]• pop,
• annual bronchitis incidence rate per person = 0.00378
: estimated PM25 logistic regression coefficient = 0.09132
= change in annual average PM25 concentration
= population ages 27 and older "without chronic bronchitis"
= standard error of p = 0.00680
Study: Abbey etal. (1995)
Location: California
Other pollutants in model: none
chronic
bronchitis
where:
Vo
P°
APM10
pop
Study: Schwartz (1993)
Location: 53 U.S. urban areas
Other pollutants in model: none
= national chronic bronchitis prevalence rate for individuals 18 and older= 0.0535
= annual bronchitis incidence rate per person = 0.00378
= estimated PM10 logistic regression coefficient = 0.0123
= change in annual average PM10 concentration
= population ages 30 and older "without chronic bronchitis"
= standard error of p = 0.00434
hospital
admissions-all
respiratory
where:
Vo
P
APM2.5.10
pop
A/A// respiratory =- [y0 . (e -^"2 =-"> -l)].pop,
daily hospital admission rate for all respiratory per person = 2.58 E-5
PM25.10 coefficient = 0.00147
change in daily average PM25.10 concentration
population of all ages
standard error of p = 0.001 79
Study: Burnett et al. (1997b)
Location: Toronto, Canada
Other pollutants in model: O3, NO2, SO2
hospital
admissions -
asthma
^.Asthma Admissions =- [y0
-\)\pop,
where:
Vo
P
APM2.5.10
pop
= daily hospital admission rate for asthma per person = 4.75 E-6
= PM25.10 coefficient = 0.00321
= change in daily average PM25.10 concentration
= population of all ages
= standard error of p = 0.00106
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model: CO, O3
D-72
-------
Health
End point
C-R Function
Source of C-R Function
hospital
admissions -
obstructive lung
disease
A.OLD Admissions =- [y0 -(e
-/J.APM25_10
where:
Vo
P
A PM2.5.10
pop
daily hospital admission rate for obstructive lung disease per person = 5.76 E-6
PM25.10 coefficient = 0.00310
change in daily average PM25_10 concentration
population of all ages
standard error of p = 0.00163
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model: CO, O3
hospital
admissions -
respiratory
infection
where:
Vo
P
APM25
pop
^Respiratory Infection Admissions =-[y0 -(e /JA™25 -1)]-pop,
= daily hospital admission rate for respiratory infection per person = 1.56 E-5
= PM25 coefficient = 0.00328
: change in daily average PM25 concentration
: population of all ages
= standard error of p = 0.000735
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model: NO2, O3
hospital
admissions-all
respiratory
Aa// respiratory admissions =/?• APM25 -pop,
where:
P
APM25
pop
: ozone coefficient = 1.81 E-8
: change in daily average PM25
= population all ages
= standard error of p = 1.79 E-8 .
Study: Thurston et al. (1994)
Location: Toronto, Canada
Other pollutants in model: O3
hospital
admissions -
pneumonia
where:
Vo
P
APM10
pop
Apneumoniaadmissions =-[y0-(e 'JA™1° -1)]-pop,
= daily hospital admission rate for pneumonia per person = 5.30 E-5
= PM10 coefficient = 0.000498
= change in daily average PM10 concentration
: population age 65 and older
= standard error of p = 0.000505
Study: Moolgavkar et al. (1997)
Location: Minneapolis, MN
Other pollutants in model: O3, SO2, NO2
hospital
admissions -
COPD
where:
Vo
P
APM10
pop
ACOPDadmissions=-[ya .(
= daily hospital admission rate for COPD per person = 3.75 E-5
= PM10 coefficient = 0.000877
: change in daily average PM10 concentration
= population age 65 and older
= standard error of p = 0.000777
Study: Moolgavkar et al. (1997)
Location: Minneapolis, MN
Other pollutants in model: CO, O3
D-73
-------
Health
End point
C-R Function
Source of C-R Function
hospital
admissions -
pneumonia
where:
Vo
P
APM10
pop
OR
Apneumonia admissions =-[/0 '(e /JA™10 -1)]-pop,
daily hospital admission rate for pneumonia per person = 5.18 E-5
PM10 coefficient = 0.00157
change in daily average PM10 concentration
population age 65 and older
standard error of p = 0.000677
Study: Schwartz (1994c)
Location: Minneapolis, MN
Other pollutants in model: O3
hospital
admissions -
COPD
ACOPD admissions =-[ya
-1)]-pop,
where:
Vo
P
APM10
pop
daily hospital admission rate for COPD per person = 3.75 E-5
PM10 coefficient = 0.00451
change in daily average PM10 concentration
population age 65 and older
standard error of p = 0.00138
Study: Schwartz (1994c)
Location: Minneapolis, MN
Other pollutants in model: none
hospital
admissions -
pneumonia
where:
Vo
P
APM10
pop
OR
Apneumonia admissions =-[/0 '(e /JA™10 -1)]-pop,
daily hospital admission rate for pneumonia per person = 5.30 E-5
PM10 coefficient = 0.00174
change in daily average PM10 concentration
population age 65 and older
standard error of p = 0.000536
Study: Schwartz (1994a)
Location: Birmingham, AL
Other pollutants in model: none
hospital
admissions -
COPD
ACOPD admissions =-[ya
-1)]-pop,
where:
Vo
P
APM10
pop
daily hospital admission rate for COPD per person = 3.75 E-5
PM10 coefficient = 0.00239
change in daily average PM10 concentration
population age 65 and older
standard error of p = 0.000838
Study: Schwartz (1994a)
Location: Birmingham, AL
Other pollutants in model: none
hospital
admissions -
pneumonia
where:
Vo
P
APM10
pop
OR
Apneumonia admissions =-[/0 '(e /JA™10 -1)]-pop,
daily hospital admission rate for pneumonia per person = 5.18 E-5
PM10 coefficient = 0.00115
change in daily average PM10 concentration
population age 65 and older
standard error of p = 0.00039
Study: Schwartz (1994b)
Location: Detroit, Ml
Other pollutants in model: O3
D-74
-------
Health
End point
hospital
admissions -
COPD
hospital
admissions -all
respiratory
where:
Vo
P
APM10
pop
°P
where:
Vo
P
C-R Function
ACOPDadmissions=-[ya -(e~'
= daily hospital admission rate for COPD per person
= PM10 coefficient = 0.00202
= change in daily average PM10 concentration
= population age 65 and older
= standard error of p = 0.00059
Aa// respiratory admissions =- [ya •
= daily hospital admission rate for all respiratory per
= PM10 coefficient = 0.00163
*"•"<• -1)]. pop,
= 3.05 E-5
(e-^""-1)]-pop,
person 65 or older = 1 .187 E-4
Source of C-R
Study: Schwartz (1994b)
Location: Detroit, Ml
Other pollutants in model:
Study: Schwartz (1996)
Location: Spokane, WA
Other pollutants in model:
Function
03
none
A PM10 = change in daily average PM10 concentration
pop = population age 65 and older
Op = standard error of p = 0.000470
hospital
admissions-all
respiratory
where:
Vo
P
APM10
pop
OR
&.all respiratory admissions = -[y0 -(e ^A™1° -1)]-pop,
= daily hospital admissions for all respiratory per person 65 and older = 1.187 E-4
= PM10 coefficient = 0.00172
: change in daily average PM10 concentration
= population age 65 and older
= standard error of p = 0.000930
Study: Schwartz (1995)
Location: New Haven, CT
Other pollutants in model: O3
hospital
admissions-all
respiratory
where:
Vo
P
APM10
pop
\allrespiratory related admissions = -[y0-(e /M™10 -1)]-pop,
= daily hospital admissions for all respiratory conditions per person 65 and older = 1.187 E-4
= PM10 coefficient = 0.00227
= change in daily average PM10 concentration
= population age 65 and older
= standard error of p = 0.00145
Study: Schwartz (1995)
Location: Tacoma, WA
Other pollutants in model: O3
hospital
admissions -
asthma
where:
Vo
P
APM25
pop
OR
AAsthmaAdmissions =-[y0-(e /M™25 -1)]-pop,
= daily hospital admission rate for asthma per person = 4.52 E-6
= PM25 coefficient = 0.0027
: change in daily average PM25 concentration
= population of ages < 65
= standard error of p = 0.000948
Study: Sheppard et al. (1999)
Location: Seattle, WA
Other pollutants in model: CO
D-75
-------
Health
End point
C-R Function
Source of C-R Function
hospital
admissions -
cardiac
Acardiac =-[/„
-1)]-pop,
where:
Vo
P
APM2.5.10
pop
OR
= daily hospital admission rate for cardiac per person = 3.81 E-5
= PM25.10 coefficient = 0.00704
= change in daily average PM25_10 concentration
= population of all ages
= standard error of p = 0.00215
Study: Burnett et al. (1997b)
Location: Toronto, Canada
Other pollutants in model: O3
hospital
admissions -
dysrhythmias
where:
Vo
P
APM25
pop
OR
ADysrhythmiasAdmissions =-[y0 -(e /JA™25 -1)]-pop,
= daily hospital admission rate for dysrhythmias per person = 6.46 E-6
= PM25 coefficient = 0.00136
: change in daily average PM25 concentration
: population of all ages
= standard error of p = 0.000910
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model: CO, O3
hospital
admissions -
ischemic heart
disease
where:
Vo
P
APM10
pop
Alschemic Heart Disease Admissions = -\y0 -(e ^A™1° -1)]-pop,
= daily hospital admission rate for ischemic heart disease per person 65 and older = 9.96 E-5
= PM10 coefficient= 0.000496
= change in daily average PM10 concentration
= population age 65 and older
= standard error of p = 0.000220
Study: Schwartz and Morris (1995)
Location: Detroit, Ml
Other pollutants in model: CO
hospital
admissions -
congestive heart
failure
ACongestiveHeart Failure Admissions = -[y0 -(e f'APM"> -1)]-pop,
where:
y0 = daily hospital admission rate for congestive heart failure per person 65 and older = 5.82 E-5
p = PM10 coefficient = 0.000741
APM10 = change in daily average PM10 concentration
pop = population age 65 and older
OR = standard error of p = 0.000311
Study: Schwartz and Morris (1995)
Location: Detroit, Ml
Other pollutants in model: CO
hospital
admissions -
cardiovascular
where:
Vo
P
APM10
pop
OR
^cardiovascularadmissions = -\y0 -(e 'JA™1° -1)]-pop,
= daily hospital admission rate for cardiovascular disease per person 65 and older = 2.23 E-4
= PM10 coefficient = 0.000737
: change in daily average PM10 concentration
: population age 65 and older
= standard error of p = 0.000170
Study: Schwartz (1999)
Location: eight U.S. counties
Other pollutants in model: CO
D-76
-------
Health
End point
C-R Function
Source of C-R Function
hospital
admissions -
cardiovascular
where:
Vo
P
APM10
pop
Acardiovascularadmissions = -[ya-(e 'JA™1° -1)]-pop,
= daily hospital admission rate for cardiovascular disease per person 65 and older = 2.23 E-4
= PM10 coefficient = 0.00102
= change in daily average PM10 concentration
= population age 65 and older
= standard error of p = 0.000423
Study: Schwartz (1997)
Location: Tucson, AZ
Other pollutants in model: CO
emergency room
visits
A asthma visits = -[/„ • (e^A™10 -1)]- pop,
where:
Vo
P
APM10
pop
= daily ER visits for asthma per person under 65 years old = 7.69 E-6
= PM10 coefficient (Schwartz et al., 1993, p. 829) = 0.00367
: change in daily average PM10 concentration
= population ages 0-64
= standard error of p (Schwartz et al., 1993, p. 829) = 0.00126
Study: Schwartz (1993)
Location: Seattle, WA
Other pollutants in model: none
acute bronchitis
AAcute Bronchitis = -
where:
Vo
P
APM25
pop
Study: Dockery et al. (1996)
Location: 24 U.S. and Canadian cities
Other pollutants in model: none
annual bronchitis incidence rate per person = 0.044
estimated PM25 logistic regression coefficient = 0.0272
change in annual average PM25 concentration
population ages 8-12
standard error of p = 0.0171
lower respiratory
symptoms (LRS)
defined as
cough, chest
pain, phlegm,
and wheeze
ALowerRespiratory Symptoms = -
where:
Vo
P
APM25
pop
Study: Schwartz, et al. (1994)
Location: six U.S. cities
Other pollutants in model: none
= daily lower respiratory symptom incidence rate per person = 0.0012
: estimated PM25 logistic regression coefficient = 0.01823
: change in daily average PM25 concentration
= population ages 7-14
= standard error of p = 0.00586
D-77
-------
Health
End point
Shortness of
breath, days
URS, defined as
runny or stuffy
nose, wet
cough, burning,
aching, or red
eyes
presence of any
of 19 acute
respiratory
symptoms
moderate or
worse asthma
status
where:
Vo
P
APM10
pop
°P
where:
Vo
P
APM10
pop
°P
where:
P*
APM10
pop
°P
where:
P
PM2.5
pop
°u
C-R Function
Aonorrncssortirearn- .ea>M,,-e >o
= daily shortness of breath incidence rate per person = 0.056
= estimated PM10 logistic regression coefficient = 0.00841
= change in daily average PM10 concentration
= asthmatic African-American population ages 7 to 12
= standard error of p = 0.00363
[ /o
Source of C-R Function
Study: Ostro et al. (1995)
pop, Location: Los Angeles, CA
Other pollutants in model: none
] Study: Pope etal. (1991)
AUppcrRcspiratory Symptoms - \ &PM,,-IS yo r (JU(-'i Luodiiun. uidii vdiiey
Ll ~ y°' y° 1 Other pollutants in model: none
= daily upper respiratory symptom incidence rate per person = 0.3419
= estimated PM10 logistic regression coefficient = 0.0036
= change in daily average PM10 concentration
= asthmatic population ages 9 to 1 1 = 6.91 % of population ages 9 to 1 1
= standard error of p = 0.0015
AARD2 = p"PMw-APMw-pop,
= first derivative of the stationary probability = 0.000461
= change in daily average PM10 concentration
= population ages 18-65
= standard error of p* = 0.000239
(PM2.5,b,lar,,
= estimated PM25 coefficient for year i = 0.0006
= change in daily average PM25 concentration
= asthmatic population of all ages = 5.61% of the population of a
= standard error of p = 0.0003
Study: Krupnicket al. (1990)
Location: Glendora-Covina-Azusa, CA
Other pollutants in model: SO2, O3
Comments: COM used in estimation of model.
The estimated COM coefficient is used with
PM10 data.
Study: Ostro et al. (1991)
•P°P> Location: Denver
Other pollutants in model: none
Comments: The estimated coefficient is applied
to populations of all ages.
II ages
D-78
-------
Health
End point
C-R Function
Source of C-R Function
asthma attacks
Aasthmaattacks = -
-/o I POP,
where:
Vo
P
APM10
pop
OR
daily incidence of asthma attacks = 0.027
PM10 coefficient = 0.00144
change in daily PM10 concentration
= population of asthmatics of all ages = 5.61% of the population of all ages
standard error of p = 0.000556
Study: Whittemore and Korn (1980)
Location: Los Angeles, CA
Other pollutants in model: O3
Restricted
Activity Days
(RADs)
where:
Vo
P
APM25
pop
OR
= daily RAD incidence rate per person = 0.0177
: inverse-variance weighted PM25 coefficient = 0.00475
= change in daily average PM2 5 concentration
= adult population ages 18 to 65
= standard error of p = 0.00029
Study: Ostro(1987)
Location: U.S. metropolitan areas
Other pollutants in model: none
Comments: An inverse-variance weighting used
to estimate the coefficient, based on Ostro
(1987, Table III)].
respiratory and
nonrespiratory
conditions
resulting in a
minor restricted
activity day
(MRAD)
where:
Vo
P
APM25
pop
OR
: daily MRAD daily incidence rate per person = 0.02137
: inverse-variance weighted PM25 coefficient = 0.00741
= change in daily average PM25 concentration
= adult population ages 18 to 65
= standard error of p = 0.0007
Study: Ostro and Rothschild (1989b)
Location: U.S.
Other pollutants in model: O3
Comments: An inverse-variance weighting used
to estimate the coefficient, based on Ostro and
Rothschild (1989b, Table 4)
work loss days
(WLDs)
where:
Vo
P
APM25
pop
OR
AI/W.D = Ay • pop =- [y0 • (e-"A™25 - 1)]- pop,
: daily work-loss-day incidence rate per person = 0.00648
= inverse-variance weighted PM25 coefficient = 0.0046
= change in daily average PM25 concentration
: population ages 18 to 65
= standard error of p = 0.00036
Study: Ostro (1987)
Location: U.S. metropolitan areas
Other pollutants in model: none
Comments: An inverse-variance weighting used
to estimate the coefficient, based on Ostro
(1987, Table III).
D-79
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Sulfur Dioxide
The C-R functions used to estimate the impact of
sulfur dioxide are summarized in Table D-19.
D-80
-------
TableD-19
Summary of C-R Functions for Sulfur Dioxide
Health Endpoint
Concentration-Response Function
Source of C-R Function
hospital admissions
- all respiratory
where:
Vo
P
ASO2
pop
\Allrespiratory =- [y0 • (e /iAS°2 - 1)]- pop,
= daily hospital admission rate for all respiratory per person = 2.58 E-5
= SO2 coefficient = 0.00446
= change in daily one-hour maximum SO2 concentration (ppb)
= population of all ages
= standard error of p = 0.00293
Study: Burnett et al. (1997b)
Location: Toronto, Canada
Other pollutants in model: PM25.10, NO2, O3
hospital admissions
- pneumonia
\pneumoniaadmissions =-[y0 -(
/JAS°2
Study: Moolgavkar et al. (1997)
Location: Minneapolis, MN
where:
y0 = daily hospital admission rate for pneumonia per person = 5.30 E-5
p =SO2 coefficient = 0.00143
A SO2 = change in daily average SO2 concentration (ppb)
pop = population age 65 and older
Op = standard error of p = 0.00290
Other pollutants in model: O3, NO2 PM10
hospital admissions
- ischemic heart
disease
where:
Vo
P
ASO2
pop
\lschemicHeart Disease Admissions =-[y0 -(e f>'AS°2 -1)]-pop,
= daily hospital admission rate for ischemic heart disease per person = 2.23 E-5
= SO2 coefficient = 0.00177
= change in daily average SO2 concentration
= population of all ages
= standard error of p = 0.000854
Study: Burnett et al. (1999)
Location: Toronto, Canada
Other pollutants in model: NO2
chest tightness,
shortness of breath,
or wheeze
Asympfoms =
1
+ e"
•pop,
.where:
a = constant = -5.65
P
Y = status coefficient (used for moderate asthmatics only) = 1.10
SO2 = peak five minute SO2 concentration (ppb) in an hour = hourly SO2
concentration (ppb) multiplied by 2.5 peak to mean ratio of 2.5
= exercising asthmatics = population of asthmatics of all ages (5.61% of the
population of all ages (Adams et al., 1995 Table 57)) of whom 1.7% are
exercising. Moderate asthmatics compose one third of exercising asthmatics;
mild asthmatics compose the other two thirds (U.S. EPA, 1997, p. D-39).
= standard error of p = 0.00247
Study: Linn et al. (1987; 1988; 1990) and Roger et al. (1985)
Location: Chamber study
Other pollutants in model: none
Comments: The results of four chamber studies were
combined to develop this C-R function. Moderate asthmatics
compose one third of exercising asthmatics; mild asthmatics
compose the other two thirds.
pop
D-81
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Modeling Results
This section presents the results of the health
effects modeling resulting from improvements in air
quality between the Pre-CAAA and Post-CAAA
scenarios for the years 2000 and 2010. Tables D-20
and D-21 summarize the health effects for each study
included in the analysis, presenting the mean, as well
as the estimated credible interval (5th and 95th
percentiles) of the number of avoided cases of each
endpoint. Table D-20 presents these results for the
subpopulation living within 50 kilometers of an air
quality monitor. Table D-21 presents results for the
entire population of the 48 contiguous states. Table
D-22 summarizes the life-years lost by age group;
Tables D-23 and D-24 present illustrative calculations
of the impact of air pollution on mortality; and Figure
D-2 presents the results of using alternative effect
thresholds in the calculation of mortality.
The ranges of estimates presented in Tables D-
20 and D-21 reflect the measured uncertainty inherent
in the estimated C-R coefficients used in calculating
the avoided incidence for each endpoint. These
ranges are only a partial measure of the total
uncertainty associated with the estimation of the
avoided incidence of each health effect. There are
other potentially important sources of uncertainty in
this benefits analysis that would likely lead to a wider
uncertainty range. For example, some of the analytical
components are point estimates that do not
incorporate information about the uncertainty
inherent in the estimates, such as the emissions and air
quality estimates. A complete depiction of the
uncertainty of the estimates would include the
uncertainty in these important analytical components.
Incorporating quantitative uncertainty estimates into
each of these components is not feasible for this
current analysis. Therefore, the range of estimates
presented herein is only a partial reflection of the total
uncertainty range.
Uncertainty
The stated goal of this study is to provide a
comprehensive estimate of the benefits of the Clean
Air Act Amendments of 1990. To achieve this goal,
information with very different levels of confidence
must be used. The analysis presents information on
the plausible range of estimates through the use of
two approaches. The first approach is to reflect the
measured uncertainty in estimating the avoided
incidence of health effects by using an estimated
probability distribution for each C-R coefficient used
in the analysis. The second approach is to present
analysis using different key assumptions. The
threshold choice, the time between PM exposure and
mortality, the choice of studies, and whether to
estimate mortality using statistical life years or
statistical lives lost are important assumptions that are
examined in this analysis.
To capture the variation in the C-R function
coefficient estimates used to estimate the avoided
incidence of health effects, this analysis uses a Monte
Carlo procedure to generate distributions of estimated
effects by randomly sampling the distribution of
coefficients (given by the mean coefficient and
standard deviation reported in the literature) and then
evaluating the C-R function with the randomly
selected coefficient. This yielded an estimate of
avoided incidence for the given effect and was
repeated many times to generate distributions of
avoided incidence. Both the mean estimates and the
5th and 95th percentile estimates of the resulting
distributions of avoided incidence estimates are
presented here for each health effect.
The second type of uncertainty considered here
addresses the fact that different published results
reported in the scientific literature typically do not
report identical findings; in some instances the
differences are substantial. For this analysis, some
health endpoints used more than one concentration-
response function, each representing a different study.
The alternative concentration-response functions
provide differing measures of the effect air quality
reductions have on changes in particular health
endpoints. This between-study variability is captured
by considering the range of estimates for a given
endpoint, and can be used to derive a range of
possible results. For example, concentration-response
functions for developing chronic bronchitis from
D-82
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
three different studies are used to estimate the range
of avoided cases of chronic bronchitis.
Another important source of uncertainty that is
considered as an alternative analysis is the estimation
of statistical life years lost. Table D-22 presents the
percentage of lives lost for each age group considered
and the average number of life years lost. The
majority of the estimated deaths occur in people over
the age of 65 (due to their higher baseline mortality
rates), and this group has a short life expectancy
relative to other age groups.
Sensitivity Analyses
One particularly important uncertainty is the
impact that alternative threshold assumptions have on
both the estimates of specific health effects and
ultimately on monetary benefits. The available
evidence has failed to identify thresholds - or safe
levels of air pollution - for any of the effects
associated with criteria pollutants, so this analysis
assumes that there are no effective thresholds and that
air pollution has effects down to zero ambient levels.
Nevertheless, thresholds may exist and their potential
impact on the overall benefits analysis could be
substantial. Any of the health effects estimated in this
analysis could have a threshold; however a threshold
for PM-related mortality would have the greatest
impact on the overall benefits analysis. Figure D-2
shows the effect of incorporating a range of possible
thresholds, using 2010 PM levels and the Pope et al.
(1995) study.
Pope et al. (1995) did not explicitly include a
threshold in their analysis. However, if the true
mortality C-R relationship has a threshold, then Pope
et al.'s slope coefficient would likely have been
underestimated for that portion of the C-R
relationship above the threshold. This would likely
lead to an underestimate of the incidences of avoided
cases above any assumed threshold level. It is difficult
to determine the size of the underestimate without
data on a likely threshold and without re-analyzing the
Pope et al. data.
The quantitative results of several other
sensitivity analyses are also presented. As discussed
above, there is information suggesting a possible
relationship between ozone and premature mortality,
and between PM and infant mortality. However,
there is considerable uncertainty about these
relationships at this time, so quantitative estimates of
these effects are not included in the aggregated results.
The possible magnitude of these health effects are
explored as sensitivity analyses, reported in Tables D-
23 (for the population within 50 kilometers of a
monitor) and D-24 (for the entire population of the
48 contiguous states). In addition, the results of an
alternative estimate of the premature adult mortality
associated with long-term PM exposure based on
Dockery et al. (1993) are also presented in Tables D-
23 and D-24. The Dockery et al. study used a smaller
sample of individuals from fewer cities than the study
by Pope et al., although it features improved exposure
estimates, a slightly broader study population, and a
follow-up period nearly twice as long as that of Pope
et al. The results based on Dockery are presented
only as sensitivity calculation for this important health
effect; the Pope et al. (1995) estimate is used in the
primary analysis.
Finally, this study includes a sensitivity analysis
illustrating the effect of alternative assumptions about
a potential lag between PM exposure and premature
mortality on monetized benefit estimates. As
discussed earlier, a change in the assumed lag period
will have no effect on the total estimate of avoided
mortality presented in Table D-21; it will only affect
the distribution of those avoided deaths through time.
Changes to this distribution will, however, affect
monetized benefit estimates if the values of the
avoided future deaths are discounted. Therefore,
although we discuss the various lag scenarios here,
the results of this sensitivity analysis are presented in
the valuation appendix, Appendix H.
Before describing the lag scenarios, we
emphasize that no scientific evidence currently exists
to support the assumption of a significant lag (i.e.,
several years or more) between PM exposure and
premature mortality. The prospective cohort study
design of long-term epidemiological studies of PM
D-83
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
exposure (including Pope et al., 1995) provides no
information about whether a lag exists or whether a
particular length of exposure is required to elicit an
effect. Further, we have identified no studies
specifically designed to test for such a PM/mortality
lag. However, we have incorporated a lag into our
primary analysis and conducted this sensitivity analysis
for two reasons. First, other similar types of
exposures, such as cigarette smoking, do show
evidence of a lag. Studies of reductions in cigarette
smoking suggest that the benefits of smoking
cessation occur over a several year period. Second,
differences between the relative risk estimates of
short-term studies of PM exposure and those of long-
term (i.e., cohort) studies may suggest the presence of
a lag for some portion of the overall mortality effect
of PM exposure.
Short-term epidemiological studies linking daily
measures of PM exposures with daily mortality rates
show statistically significant increases in mortality
within days of increased PM exposure. However, the
appropriate lag period for the portion of deaths that
do not occur immediately is unclear. Some interpret
the analysis by Brunekreef (1997) as indicative of a
much longer mortality lag of 15 years; however, it
appears that Brunekreef simply employed assumptions
consistent with the cohort design of the Dockery et
al., 1993 study, which examined the relative risk for a
cohort aged 25 to 74 over a 15-year period. The
selection of such a follow-up period by Dockery et al.
was not based on biological or epidemiological
evidence of a 15-year lag, and Brunekreef cites no
evidence supporting a lag of this length. Therefore,
we do not find the Brunekreef (1997) study to be
convincing evidence of a fifteen-year lag.
Table D-25 compares the distribution of
avoided mortality benefits assumed in the primary
analysis with each of the sensitivity analysis scenarios.
In the primary analysis, we apply the same lag
structure used as a sensitivity analysis in the draft RIA
for the proposed Tier 2 motor vehicle emission
standards (U.S. EPA, 1999). Under this scenario, the
avoided mortality occurs over a five year period, with
fifty percent of the avoided mortality occurring within
the first two years (i.e. 25 percent per year), and the
remainder of avoided deaths distributed evenly across
the last three years (approximately 16.7 percent per
year). As mentioned above, the appropriate length of
the lag period is highly uncertain, and so is the
distribution of deaths over that period, although
evidence from short-term studies suggests weighting
the distribution toward the first couple of years
following exposure. The assumptions of the Tier 2
lag structure reflect the best judgment of the Agency
on this issue; however, they do not represent any
known lag structure for PM mortality.
We evaluate three lag scenarios for the
sensitivity analysis. The first scenario assumes no lag;
that is, all avoided mortality occurs in the same year as
exposure (Year 1 in Table D-25). The second
distributes avoided mortality evenly across eight years;
this scenario is based on the eight-year cohort follow-
up period of the Pope et al., 1995 study. The third
scenario is based on the Dockery et al., 1993 follow-
up period of 15 years, with avoided mortality
distributed evenly across that period. As discussed
earlier, we find the 15-year lag to be an extremely
conservative assumption.
The effect of these different lag assumptions on
our estimate of monetary benefits depends on the
discount rate used. Given the discount rate used in
the primary analysis, five percent, the no lag scenario
would increase the primary mortality reduction
benefits estimate by nine percent; the Pope-based lag
estimate would decrease the estimate by eight percent;
and the Dockery-based lag estimate would decrease
the estimate by 21 percent. The actual monetary
benefit estimates generated by this sensitivity analysis
are presented in Appendix H.
D-84
-------
Table D-20
Change in Incidence of Adverse Health Effects Associated with Criteria Pollutants (Pre-CAAA minus Post-CAAA) - 48 State U.S.
Population within 50 km of a Monitor (avoided cases per year)
Endpoint
Pollutant
5th %
2000
mean
95th %
5th %
2010
mean
% of Baseline Incidences for the
mean estimates a
95th %
2000
2010
Mortality
ages 30 and older
PM
7,900
13,000
18,000
13,000
20,000
28,000
0.67%b
0.95%b
Chronic Illness
chronic bronchitis
chronic asthma
PM
03
2,300
910
11,000
4,900
20,000
8,300
4,300
1,100
18,000
5,500
30,000
8,700
2.05%
3.46%
3.09%
3.59%
Hospitalization
respiratory
admissions
cardiovascular
admissions
emergency room
visits for asthma
PM, CO, NO2,
SO2, O3
PM, CO, NO2,
SO2, O3
PM, O3
7,300
5,400
200
12,000
19,000
1,300
17,000
41,000
6,600
12,000
9,400
310
19,000
38,000
2,000
29,000
90,000
10,000
0.52%
0.59%
0.19%
0.76%
1.10%
0.27%
Minor Illness
acute bronchitis
upper respiratory
symptoms
lower respiratory
symptoms
respiratory illness
moderate or
worse asthmad
asthma attacksd
PM
PM
PM
NO2
PM
O3, PM
0°
170,000
130,000
24,000
32,000
520,000
25,000
570,000
270,000
110,000
210,000
960,000
49,000
970,000
420,000
180,000
370,000
1 ,400,000
0°
260,000
210,000
63,000
48,000
800,000
40,000
870,000
440,000
270,000
310,000
1,500,000
79,000
1 ,500,000
670,000
450,000
570,000
2,100,000
3.19%
0.61%
2.19%
7.62%
0.15%
0.74%
4.71%
0.86%
3.30%
17.29%
0.21%
1.06%
D-85
-------
Endpoint Pollutant
chest tightness, SO2
shortness of
breath, or wheeze
shortness of PM
breath
work loss days PM
minor restricted O3, PM
activity days / any
of 1 9 respiratory
symptoms6
restricted activity PM
daysd
5th %
200
16,000
1 ,900,000
14,000,000
5,500,000
2000
mean
80,000
57,000
2,200,000
17,000,000
6,200,000
95th %
370,000
95,000
2,500,000
20,000,000
6,800,000
5th %
270
25,000
3,100,000
22,000,000
9,000,000
2010
mean
100,000
88,000
3,500,000
27,000,000
10,000,000
% of Baseline Incidences for the
mean estimates a
95th % 2000
470,000 0.003%
150,000 1.25%
4,000,000 0.60%
32,000,000 1.47%
11,000,000 0.61%
2010
0.004%
1 .79%
0.87%
2.16%
0.91%
aThe baseline incidence generally is the same as that used in the C-R function for a particular health effect. However, there are a few exceptions. To calculate the baseline incidence
rate for respiratory-related hospital admissions, we used admissions for persons of all ages for ICD codes 460-519; for cardiovascular admissions, we used admissions for persons of
all ages for ICD codes 390-429; for emergency room visits for asthma, we used the estimated ER visit rate for persons of all ages; for chronic bronchitis we used the incidence rate for
individuals 27 and older; for the pooled estimate of minor restricted activity days and any-of-19 respiratory symptoms, we used the incidence rate for minor restricted activity days.
b Calculated as the ratio of avoided mortality to the projected baseline annual non-accidental mortality for adults aged 30 and over. Non-accidental mortality was approximately 95% of
total mortality for this subpopulation in 2010.
0 Monte Carlo modeling returned a negative value for the fifth percentile estimate of this endpoint. However, we believe the negative result represents an artifact of the statistical methods
employed in the uncertainty analysis, since none of the studies used in the health benefits analysis suggest a negative correlation between criteria air pollutant exposure and this health
endpoint. We therefore truncate this value at zero for presentation. The full distribution of estimates, including negative values, is used in all aggregations of benefits estimates presented
in this document.
d These health endpoints overlap with the "any-of-19 respiratory symptoms" category. As a result, although we present estimates for each endpoint individually, these results are not
aggregated into the total benefits estimates.
e Minor restricted activity days and any-of-19 respiratory symptoms have overlapping definitions and are pooled.
D-86
-------
Table D-21
Change in Incidence of Adverse Health Effects Associated with Criteria Pollutants (Pre-CAAA
Population (avoided cases per year)
Endpoint
Pollutant
5th %
2000
mean
95th %
5th %
2010
mean
minus Post-CAAA) - 48 State U.S.
% of Baseline Incidences for
the mean estimates a
95th % 2000
2010
Mortality
ages 30 and older
PM
8,800
14,000
19,000
14,000
23,000
32,000 0.66%b
1 .00%b
Chronic Illness
chronic bronchitis
chronic asthma
PM
03
3,100
1,300
13,000
5,600
22,000
9,600
5,000
1,800
20,000
7,200
34,000 2.21%
12,000 3.22%
3.14%
3.83%
Hospitalization
respiratory
admissions
cardiovascular
admissions
emergency room
visits for asthma
PM, CO, NO2,
SO2, O3
PM, CO, NO2,
SO2, O3
PM, O3
8,100
5,800
260
13,000
22,000
3,100
20,000
48,000
8,900
13,000
10,000
430
22,000
42,000
4,800
34,000 0.40%
100,000 0.49%
14,000 0.39%
0.62%
0.86%
0.55%
Minor Illness
acute bronchitis
upper respiratory
symptoms
lower respiratory
symptoms
respiratory illness
moderate or
worse asthmad
asthma attacksd
PM
PM
PM
NO2
PM
O3, PM
0°
180,000
150,000
31,000
52,000
590,000
29,000
620,000
320,000
130,000
260,000
1,100,000
59,000
1,000,000
480,000
220,000
460,000
1,600,000
0°
280,000
240,000
76,000
80,000
920,000
47,000
950,000
520,000
330,000
400,000
1 ,700,000
94,000 3.39%
1,600,000 0.61%
770,000 2.38%
550,000 4.46%
720,000 0.17%
2,500,000 0.73%
5.06%
0.86%
3.57%
10.44%
0.24%
1.04%
D-87
-------
Endpoint Pollutant
chest tightness, SO2
shortness of
breath, or wheeze
shortness of PM
breath
work loss days PM
minor restricted O3, PM
activity days / any
of 1 9 respiratory
symptoms6
restricted activity PM
daysd
5th %
220
16,000
2,200,000
16,000,000
6,400,000
2000
mean
88,000
59,000
2,500,000
19,000,000
7,200,000
95th %
410,000
98,000
2,900,000
23,000,000
7,900,000
5th %
290
26,000
3,600,000
25,000,000
10,000,000
2010
mean
110,000
91 ,000
4,100,000
31,000,000
12,000,000
% of Baseline Incidences for
the mean estimates a
95th % 2000
520,000 0.002%
150,000 1.19%
4,600,000 0.62%
37,000,000 1 .43%
13,000,000 0.65%
2010
0.003%
1.69%
0.94%
2.15%
1.00%
aThe baseline incidence generally is the same as that used in the C-R function for a particular health effect. However, there are a few exceptions. To calculate the baseline incidence
rate for respiratory-related hospital admissions, we used admissions for persons of all ages for ICD codes 460-519; for cardiovascular admissions, we used admissions for persons of
all ages for ICD codes 390-429; for emergency room visits for asthma, we used the estimated ER visit rate for persons of all ages; for chronic bronchitis we used the incidence rate for
individuals 27 and older; for the pooled estimate of minor restricted activity days and any-of-19 respiratory symptoms, we used the incidence rate for minor restricted activity days.
b Calculated as the ratio of avoided mortality to the projected baseline annual non-accidental mortality for adults aged 30 and over. Non-accidental mortality was approximately 95% of
total mortality for this subpopulation in 2010.
0 Monte Carlo modeling returned a negative value for the fifth percentile estimate of this endpoint. However, we believe the negative result represents an artifact of the statistical methods
employed in the uncertainty analysis, since none of the studies used in the health benefits analysis suggest a negative correlation between criteria air pollutant exposure and this health
endpoint. We therefore truncate this value at zero for presentation. The full distribution of estimates, including negative values, is used in all aggregations of benefits estimates presented
in this document.
d These health endpoints overlap with the "any-of-19 respiratory symptoms" category. As a result, although we present estimates for each endpoint individually, these results are not
aggregated into the total benefits estimates.
e Minor restricted activity days and any-of-19 respiratory symptoms have overlapping definitions and are pooled.
D-88
-------
Table D-22
Mortality Distribution by Age in Primary Analysis, Based on Pope et al. (1995)
Age Group Proportion of Premature Mortality by Agea Life Expectancy (years)
Infants not estimated
1-29 not estimated
30-34 1 % 48
35-44 4% 38
45-54 6% 29
55-64 12% 21
65-74 24% 14
75-84 30% 9
85+ 24% 6
' Percentages sum to 101 percent due to rounding.
D-89
-------
Table D-23
Illustrative Estimates of the Impact of Criteria Pollutants on Mortality - 48 State U.S. Population within 50 km of a Monitor (cases per
year)
Endpoint
ages 30 and older
(Popeetal., 1995)
ages 25 and older
(Dockeryetal., 1993)*
all ages *
post-neonatal *
Pollutant
PM
PM
03
PM
5th %
7,900
12,000
81
39
The Dockery et al. (1993), ozone mortality, and post-neonatal
Table D-24
Illustrative Estimates
Endpoint
ages 30 and older
(Popeetal., 1995)
ages 25 and older
(Dockeryetal., 1993)*
all ages *
post-neonatal *
of the Impact
Pollutant
PM
PM
03
PM
of Criteria
5th %
8,800
15,987
Of
45
2000
mean
13,000
29,000
1,100
81
95th %
18,000
46,000
2,200
120
5th %
13,000
20,000
130
59
mortality estimates are not aggregated into total
Pollutants on
2000
mean
14,000
34,860
1,400
88
Mortality - 48
95th %
19,000
54,677
2,800
130
State U.
5th %
14,000
26,000
Of
69
2010
mean
20,000
47,000
1,600
120
benefits estimates
S. Population
2010
mean
23,000
56,000
2,200
130
% of Baseline Incidences for
the mean estimates
95th %
28,000
73,000
3,400
180
(cases per
2000
0.67%
1 .35%
0.06%
0.93%
year)
2010
0.95%
2.19%
0.08%
1.39%
% of Baseline Incidences for
the mean estimates
95th %
32,000
88,000
4,600
200
2000
0.66%
1 .60%
0.07%
1 .02%
2010
1 .00%
2.39%
0.09%
1.38%
*The Dockery et al. (1993), ozone mortality, and post-neonatal mortality estimates are not aggregated into total benefits estimates.
fMonte Carlo modeling returned a negative value for the fifth percentile estimate of this endpoint. However, we believe the negative result represents an artifact of the statistical methods
employed in the uncertainty analysis, since none of the studies used in the health benefits analysis suggest a negative correlation between criteria air pollutant exposure and this health
endpoint. We therefore truncate this value at zero for presentation. The full distribution of estimates, including negative values, is used in all aggregations of benefits estimates presented
in this document.
D-90
-------
Table D-25
Comparison of Alternative Lag Assumptions for Premature Mortality Associated with PM Exposure
Year
Tier II SA Lag
(Primary Estimate)
No Lag
Lag Distributed
Evenly Over the
Period Covered by
Pope et al., 1995
Lag Distributed
Evenly Over the
Period Covered by
Dockery et al., 1993
Percent of Avoided Mortality By Year
12 3 4 5 6 7 8 9 10 11 12 13 14 15
25 25 16.67 16.67 16.67 0 0 00000000
100 00 0 00000000000
12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 0000000
6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67 6.67
Totals may not sum to 100 percent due to rounding.
D-91
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure D-2
Long-Term Mortality Based on Pope (1995): National Avoided Incidence Estimates (2010)
at Different Assumed Effect Thresholds, Based on a 50 Km Maximum Distance
25,000
8 8 8 8
Assumed Effect Threshold (Annual Mean PM2.5 (ug/m3))
D-92
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
Abbey, D.E., RJ. Burchette, S.F. Knutsen, W.F. McDonnell, M.D. Lebowitz and P.L. Ennght. 1998. Long-
term particulate and other air pollutants and lung function in nonsmokers. American Journal of
Respiratory and Critical Care Medicine. 158(1): 289-298.
Abbey, D.E., B.E. Ostro, F. Petersen and RJ. Burchette. 1995. Chronic Respiratory Symptoms Associated
with Estimated Long-Term Ambient Concentrations of Fine Particulates Less Than 2.5 Microns in
Aerodynamic Diameter (PM2.5) and Other Air Pollutants. J Expo Anal Environ Epidemiol. 5(2):
137-159.
Abbey, D.E., F. Petersen, P.K. Mills and W.L. Beeson. 1993. Long-Term Ambient Concentrations of Total
Suspended Particulates, Ozone, and Sulfur Dioxide and Respiratory Symptoms in a Nonsmoking
Population. Archives of Environmental Health. 48(1): 33-46.
Abt Associates Inc. 1998. Air Quality Estimation for the NOx SIP Call RIA. Prepared for U.S. EPA,
Office of Air Quality Planning and Standards, under contract no. 68-D-98-001. Research Triangle
Park, NC. September.
Abt Associates Inc. 1999. Memorandum from Leland Deck to Jim Neumann, Industrial Economics, Inc.,
"Prospective 812 Temporal and Spatial Air Quality Modeling Procedures," May 27, 1999.
Ackermann-Liebrich, U., P. Leuenberger, J. Schwartz, C. Schindler, C. Monn, C. Bolognini, J.P. Bongard, O.
Brandli, G. Domenighetti, S. Elsasser, L. Grize, W. Karrer, R. Keller, H. Keller-Wossidlo, N. Kunzli,
B.W. Martin, T.C. Medici, A.P. Perruchoud, M.H. Schom, J.M. Tschopp, B. Villiger, B. Wuthnch,
J.P. Zellweger and E. Zemp. 1997. Lung function and long term exposure to air pollutants in
Switzerland. Study on Air Pollution and Lung Diseases in Adults (SAPALDIA) Team. Am J Respir
Cnt Care Med. 155(1): 122-129.
Adams, P.P. and M.A. Marano. 1995. Current Estimates from the National Health Interview Survey, 1994.
Vital Health Statistics, Series 10, No. 193. National Center for Health Statistics. Hyattsville, MD.
Anderson, H.R., C. Spix, S. Medina, J.P. Schouten, J. Castellsague, G. Rossi, D. Zmirou, G. Touloumi, B.
Wojtyniak, A. Ponka, L. Bacharova, J. Schwartz and K. Katsouyanni. 1997. Air pollution and daily
admissions for chronic obstructive pulmonary disease in 6 European cities: Results from the
APHEA project. European Respiratory Journal. 10(5): 1064-1071.
Atkinson, R.W., H.R. Anderson, D.P. Strachan, J.M. Bland, S.A. Bremner and A. Ponce de Leon. 1999.
Short-term associations between outdoor air pollution and visits to accident and emergency
departments in London for respiratory complaints. Eur Respir J. 13(2): 257-65.
Bates, D.V., M. Baker-Anderson and R. Sizto. 1990. Asthma attack periodicity: a study of hospital
emergency visits in Vancouver. Environ Res. 51(1): 51-70.
Bobak, M. and D.A. Leon. 1992. Air pollution and infant mortality in the Czech Republic, 1986-88. Lancet.
340(8826): 1010-4.
D-93
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Buchdahl, R., A. Parker, T. Stebbings and A. Babiker. 1996. Association between air pollution and acute
childhood wheezy episodes: prospective observational study. Bmj. 312(7032): 661-5.
Burnett, R.T., J.R. Brook, W.T. Yung, R.E. Dales and D. Krewski. 1997a. Association between ozone and
hospitalization for respiratory diseases in 16 Canadian cities. Environmental Research. 72(1): 24-31.
Burnett, R.T., S. Cakmak, J.R. Brook and D. Krewski. 1997b. The role of particulate size and chemistry in
the association between summertime ambient air pollution and hospitalization for cardiorespiratory
diseases. Environ Health Perspect. 105(6): 614-20.
Burnett, R.T., S. Cakmak, M.E. Raizenne, D. Stieb, R. Vincent, D. Krewski, J.R. Brook, O. Philips and H.
Ozkaynak. 1998. The association between ambient carbon monoxide levels and daily mortality in
Toronto Canada. Journal of the Air & Waste Management Association. 48(8): 689-700.
Burnett, R.T., R. Dales, D. Krewski, R. Vincent, T. Dann and J.R. Brook. 1995. Associations between
ambient particulate sulfate and admissions to Ontario hospitals for cardiac and respiratory diseases.
AmJ Epidemiol. 142(1): 15-22.
Burnett, R.T., R.E. Dales, J.R. Brook, M.E. Raizenne and D. Krewski. 1997c. Association between ambient
carbon monoxide levels and hospitalizations for congestive heart failure in the elderly in 10 Canadian
cities. Epidemiology. 8(2): 162-167.
Burnett, R.T., R.E. Dales, M.E. Raizenne, D. Krewski, P.W. Summers, G.R. Roberts, M. Raadyoung, T.
Dann andj. Brook. 1994. Effects of Low Ambient Levels of Ozone and Sulfates On the Frequency
of Respiratory Admissions to Ontario Hospitals. Environ Res. 65(2): 172-194.
Burnett, R.T., M. Smith-Doiron, D. Stieb, S. Cakmak and J.R. Brook. 1999. Effects of particulate and
gaseous air pollution on cardiorespiratory hospitalizations. Archives Environmental Health. 54(2):
130-139.
Castellsague, J., J. Sunyer, M. Saez and J.M. Anto. 1995. Short-term association between air pollution and
emergency room visits for asthma in Barcelona. Thorax. 50(10): 1051-6.
Chapman, R.S., D.C. Calafiore and V Hasselblad. 1985. Prevalence of persistent cough and phlegm in
young adults in relation to long-term ambient sulfur oxide exposure. Am Rev Respir Dis. 132(2):
261-7.
Cody, R.P., C.P. Weisel, G. Birnbaum and PJ. Lioy. 1992. The effect of ozone associated with summertime
photochemical smog on the frequency of asthma visits to hospital emergency departments. Environ
Res. 58(2): 184-94.
Delfino, R.J., M.R. Becklake and J.A. Hanley. 1994. The relationship of urgent hospital admissions for
respiratory illnesses to photochemical air pollution levels in Montreal. Environ Res. 67(1): 1-19.
D-94
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Delfmo, R.J., A.M. Murphy-Moulton and M.R. Becklake. 1998. Emergency room visits for respiratory
illnesses among the elderly in Montreal: association with low level ozone exposure. Environ Res.
76(2): 67-77.
Delfmo, R.J., A.M. MurphyMoulton, R.T. Burnett, J.R. Brook and M.R. Becklake. 1997. Effects of air
pollution on emergency room visits for respiratory illnesses in Montreal, Quebec. Am J Respir Grit
CareMed. 155(2): 568-576.
Betels, R., D.P. Tashkin, J.W. Sayre, S.N. Rokaw, FJ. Massey, A.H. Coulson and D.H. Wegman. 1991. The
Ucla Population Studies of Cord .10. a Cohort Study of Changes in Respiratory Function Associated
With Chronic Exposure to Sox, Nox, and Hydrocarbons. American Journal of Public Health. 81(3):
350-359.
Dockery, D.W., J. Cunningham, A.I. Damokosh, L.M. Neas, J.D. Spengler, P. Koutrakis, J.H. Ware, M.
Raizenne and F.E. Speizer. 1996. Health Effects of Acid Aerosols On North American Children -
Respiratory Symptoms. Environmental Health Perspectives. 104(5): 500-505.
Dockery, D.W., C.A. Pope, X.P. Xu, J.D. Spengler, J.H. Ware, M.E. Fay, B.C. Ferris and F.E. Speizer. 1993.
An association between air pollution and mortality in six U.S. cities. N Engl J Med. 329(24): 1753-
1759.
Dockery, D.W., F.E. Speizer, D.O. Strain, J.H. Ware, J.D. Spengler and B.C. Ferris, Jr. 1989. Effects of
Inhalable Particles on Respiratory Health of Children. Am Rev Respir Dis. 139: 587-594.
Evans, J.S., T. Tosteson and P.L. Kinney. 1984. Cross-Sectional Mortality Studies and Air Pollution Risk
Assessment. Environment International. 10: 55-83.
Forsberg, B., N. Stjernberg, M. Falk, B. Lundback and S. Wall. 1993. Air pollution levels, meteorological
conditions and asthma symptoms. Eur Respir J. 6(8): 1109-15.
Gielen, M.H., S.C. vanderZee, J.H. vanWijnen, CJ. vanSteen and B. Brunekreef. 1997. Acute effects of
summer air pollution on respiratory health of asthmatic children. Am J Respir Crit Care Med.
155(6): 2105-2108.
Goldstein, I.F. and A.L. Weinstein. 1986. Air pollution and asthma: effects of exposures to short-term
sulfur dioxide peaks. Environ Res. 40(2): 332-45.
Graves, EJ. and B.S. Gillum. 1997. Detailed Diagnoses and Procedures, National Hospital Dsicharge
Survey, 1994. Vital Health Statistics, Series 13, No. 127. National Center for Health Statistics.
Hyattsville, MD. March.
Hasselblad, V., D.M. Eddy and DJ. Kotchmar. 1992. Synthesis of Environmental Evidence - Nitrogen
Dioxide Epidemiology Studies. J Air Waste Manage Assoc. 42(5): 662-671.
Hiltermann, T.J.N., J. Stolk, S.C. vanderZee, B. Brunekreef, C.R. deBruijne, P.H. Fischer, C.B. Ameling, PJ.
Sterk, P.S. Hiemstra and L. vanBree. 1998. Asthma severity and susceptibility to air pollution. Eur
Respir J. 11(3): 686-693.
D-95
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Hoek, G. and B. Brunekreef. 1995. Effect of Photochemical Air Pollution On Acute Respiratory Symptoms
in Children. Am J Respir Grit Care Med. 151(1): 27-32.
Ito, K. and G.D. Thurston. 1996. Daily Pm(10)/Mortality Associations - an Investigation of At-Risk
Subpopulations. Journal of Exposure Analysis and Environmental Epidemiology. 6(1): 79-95.
Kinney, P.L., K. Ito and G.D. Thurston. 1995. A Sensitivity Analysis of Mortality Pm-10 Associations in
Los Angeles. Inhalation Toxicology. 7(1): 59-69.
Krupnick, A.J., W. Harrington and B. Ostro. 1990. Ambient Ozone and Acute Health Effects - Evidence
From Daily Data. Journal of Environmental Economics and Management. 18(1): 1-18.
Linn, W.S., E.L. Avol, R.C. Peng, D.A. Shamoo and J.D. Hackney. 1987. Replicated dose-response study of
sulfur dioxide effects in normal, atopic, and asthmatic volunteers. Am Rev Respir Dis. 136(5):
1127-34.
Linn, W.S., E.L. Avol, D.A. Shamoo, R.C. Peng, C.E. Spier, M.N. Smith and J.D. Hackney. 1988. Effect of
metaproterenol sulfate on mild asthmatics' response to sulfur dioxide exposure and exercise. Arch
Environ Health. 43(6): 399-406.
Linn, W.S., D.A. Shamoo, R.C. Peng, K.W. Clark, E.L. Avol and J.D. Hackney. 1990. Responses to sulfur
dioxide and exercise by medication-dependent asthmatics: effect of varying medication levels. Arch
Environ Health. 45(1): 24-30.
Lipfert, F.W. 1993. A Critical Review of Studies of the Association Between Demands For Hospital
Services and Air Pollution. Environmental Health Perspectives. 101(S2): 229-268.
Lipfert, F.W. and T. Hammerstrom. 1992. Temporal Patterns in Air Pollution and Hospital Admissions.
Environmental Research. 59(2): 374-399.
Lipfert, F.W. and R.E. Wyzga. 1995. Air Pollution and Mortality - Issues and Uncertainties. J Air Waste
ManagAssoc. 45(12): 949-966.
Lipsett, M., S. Hurley and B. Ostro. 1997. Air pollution and emergency room visits for asthma in Santa
Clara County, California. Environmental Health Perspectives. 105(2): 216-222.
Loomis, D., M. Castillejos, D.R. Gold, W. McDonnell and V.H. Borja-Aburto. 1999. Air pollution and
infant mortality in Mexico City. Epidemiology. 10(2): 118-23.
McDonnell, W.F., D.E. Abbey, N. Nishino and M.D. Lebowitz. 1999. Long-term ambient ozone
concentration and the incidence of asthma in nonsmoking adults: the ahsmog study [In Process
Citation]. Environ Res. 80(2 Pt 1): 110-21.
Melia, R.J., C.D. Florey, S. Chinn, B.D. Goldstein, A.G.F. Brooks, H.H.John, D. Clark, I.E. Craighead and
X. Webster. 1980. The relation between indoor air pollution from nitrogen dioxide and respiratory
illness in schoolchildren. Clinical Respiratory Physiology. 16: 7P-8P.
D-96
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Melia, R.J., C.d.V. Florey, R.W. Morns, B.D. Goldstein, H.H. John, D. Clark, I.E. Craighead and J.C.
Mackinlay. 1982. Childhood respiratory illness and the home environment. II. Association between
respiratory illness and nitrogen dioxide, temperature and relative humidity. Int J Epidemiol. 11(2):
164-9.
Moolgavkar, S.H., E.G. Luebeck and E.L. Anderson. 1997. Air pollution and hospital admissions for
respiratory causes in Minneapolis St. Paul and Birmingham. Epidemiology. 8(4): 364-370.
Moolgavkar, S.H., E.G. Luebeck, T.A. Hall and E.L. Anderson. 1995. Air Pollution and Daily Mortality in
Philadelphia. Epidemiology. 6(5): 476-484.
Morgan, G., S. Corbett and J. Wlodarczyk. 1998. Air pollution and hospital admissions in Sydney, Australia,
1990 to 1994. Am J Public Health. 88(12): 1761-6.
Morris, R.D. and E.N. Naumova. 1998. Carbon monoxide and hospital admissions for congestive heart
failure: evidence of an increased effect at low temperatures. Environ Health Perspect. 106(10): 649-
53.
Morris, R.D., E.N. Naumova and R.L. Munasinghe. 1995. Ambient Air Pollution and Hospitalization for
Congestive Heart Failure Among Elderly People in Seven Large U.S. Cities. American Journal of
Public Health. 85(10): 1361-1365.
National Center for Health Statistics. 1994. Vital Statistics of the United States, 1990, vol II, Mortality, Part
B. Public Health Service: Washington, DC.
Neas, L.M. and J. Schwartz. 1998. Pulmonary function levels as predictors of mortality in a national sample
of US adults. American Journal of Epidemiology. 147(11): 1011-1018.
Neukirch, F., C. Segala, Y. Le Moullec, M. Korobaeff and M. Aubier. 1998. Short-term effects of low-level
winter pollution on respiratory health of asthmatic adults. Arch Environ Health. 53(5): 320-8.
Ostro, B.D. 1987. Air Pollution and Morbidity Revisited: A Specification Test. Journal of Environmental
Economics and Management. 14: 87-98.
Ostro, B.D., MJ. Lipsett and N.P. Jewell. 1989a. Predicting Respiratory Morbidity From Pulmonary
Function Tests - a Reanalysis of Ozone Chamber Studies. Japca. 39(10): 1313-1318.
Ostro, B.D., MJ. Lipsett, J.K. Mann, H. Braxtonowens and M.C. White. 1995. Air Pollution and Asthma
Exacerbations Among African-American Children in Los Angeles. Inhalation Toxicology. 7(5):
711-722.
Ostro, B.D., MJ. Lipsett, J.K. Mann, A. Krupnick and W. Harrington. 1993. Air Pollution and Respiratory
Morbidity Among Adults in Southern California. AmJ Epidemiol. 137(7): 691-700.
Ostro, B.D., MJ. Lipsett, M.B. Wiener and J.C. Seiner. 1991. Asthmatic Responses to Airborne Acid
Aerosols. AmJ Public Health. 81(6): 694-702.
Ostro, B.D. and S. Rothschild. 1989b. Air Pollution and Acute Respiratory Morbidity - an Observational
Study of Multiple Pollutants. Environ Res. 50(2): 238-247.
D-97
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pantazopoulou, A., K. Katsouyanni, J. Kourea-Kremastinou and D. Trichopoulos. 1995. Short-term effects
of air pollution on hospital emergency outpatient visits and admissions in the greater Athens, Greece
area. Environ Res. 69(1): 31-6.
Pereira, L.A.A., D. Loomis, G.M.S. Conceicao, A.L.F. Braga, R.M. Areas, H.S. Kishi, R.M. Singer, G.M.
Bohm and P.H.N. Saldiva. 1998. Association between air pollution and intrauterine mortality in Sao
Paulo, Brazil. Environmental Health Perspectives. 106(6): 325-329.
Peters, A., I.F. Goldstein, U. Beyer, K. Franke, J. Heinrich, D.W. Dockery, J.D. Spengler and H.E.
Wichmann. 1996. Acute Health Effects of Exposure to High Levels of Air Pollution in Eastern
Europe. American Journal of Epidemiology. 144(6): 570-581.
Peters, J.M., E. Avol, W. Navidi, SJ. London, WJ. Gauderman, F. Lurmann, W.S. Linn, H. Margolis, E.
Rappaport, H. Gong and D.C. Thomas. 1999. A Study of Twelve Southern California Communities
with Differing Levels and Types of Air Pollution. I. prevalence of respiratory morbidity. Am J
Respir Cnt Care Med. 159(3): 760-767.
Ponce de Leon, A., H.R. Anderson, J.M. Bland, D.P. Strachan and J. Bower. 1996. Effects of air pollution
on daily hospital admissions for respiratory disease in London between 1987-88 and 1991-92. J
Epidemiol Community Health. 50 Suppl 1: s63-70.
Ponka, A. and M. Virtanen. 1994. Chronic bronchitis, emphysema, and low-level air pollution in Helsinki,
1987-1989. Environ Res. 65(2): 207-17.
Pope, C.A. and D.W. Dockery. 1992. Acute Health Effects of PM10 Pollution On Symptomatic and
Asymptomatic Children. American Review of Respiratory Disease. 145(5): 1123-1128.
Pope, C.A., D.W. Dockery, J.D. Spengler and M.E. Raizenne. 1991. Respiratory Health and PmlO Pollution
- a Daily Time Series Analysis. American Review of Respiratory Disease. 144(3): 668-674.
Pope, C.A., MJ. Thun, M.M. Namboodin, D.W. Dockery, J.S. Evans, F.E. Speizer and C.W. Heath. 1995.
Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J Respir
Cnt Care Med. 151(3): 669-674.
Portney, P.R. and J. Mullahy. 1990. Urban Air Quality and Chronic Respiratory Disease. Regional Science
and Urban Economics. 20(3): 407-418.
Richards, W., S.P. Azen, J. Weiss, S. Stocking and J. Church. 1981. Los Angeles air pollution and asthma in
children. Ann Allergy. 47(5 Pt 1): 348-54.
Ritz, B. and F. Yu. 1999. The effect of ambient carbon monoxide on low birth weight among children born
in southern California between 1989 and 1993. Environ Health Perspect. 107(1): 17-25.
Roemer, W., G. Hoek, B. Brunekreef, J. Haluszka, A. Kalandidi and J. Pekkanen. 1998. Daily variations in
air pollution and respiratory health in a multicentre study: the PEACE project. Pollution Effects on
Asthmatic Children in Europe. Eur Respir J. 12(6): 1354-61.
Roger, L.J., H.R. Kehrl, M. Hazucha and D.H. Horstman. 1985. Bronchoconstriction in asthmatics exposed
to sulfur dioxide during repeated exercise. J Appl Physiol. 59(3): 784-91.
D-98
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Romieu, I., F. Meneses, S. RuizJJ. SienraJ. Huerta, M.C. White and R.A. Etzel. 1996. Effects of air
pollution on the respiratory health of asthmatic children living in Mexico City. Am J Respir Grit
CareMed. 154(2 Pt 1): 300-7.
Romieu, I., F. Meneses, JJ. Sienra-Monge,J. Huerta, S. Ruiz Velasco, M.C. White, R.A. Etzel and M.
Hernandez-Avila. 1995. Effects of urban air pollutants on emergency visits for childhood asthma in
Mexico City. AmJ Epidemiol. 141(6): 546-53.
Rosas, I., H.A. McCartney, R.W. Payne, C. Calderon, J. Lacey, R. Chapela and S. Ruiz-Velazco. 1998.
Analysis of the relationships between environmental factors (aeroallergens, air pollution, and
weather) and asthma emergency admissions to a hospital in Mexico City. Allergy. 53(4): 394-401.
Saldiva, P.H.N., A. Lichtenfels, P.S.O. Paiva, LA. Barone, M.A. Martins, E. Massad, J.C.R. Pereira, V.P.
Xavier, J.M. Singer and G.M. Bohm. 1994. Association Between Air Pollution and Mortality Due to
Respiratory Diseases in Children in Sao Paulo, Brazil - a Preliminary Report. Environ Res. 65(2):
218-225.
Saldiva, P.H.N., C.A. Pope, J. Schwartz, D.W. Dockery, AJ. Lichtenfels, J.M. Salge, I. Barone and G.M.
Bohm. 1995. Air Pollution and Mortality in Elderly People - a Time-Series Study in Sao Paulo,
Brazil. Arch Environ Health. 50(2): 159-163.
Samet,J.M., Y. Bishop, F.E. Speizer,J.D. Spengler and E.G. Ferris, Jr. 1981. The relationship between air
pollution and emergency room visits in an industrial community. J Air Pollut Control Assoc. 31(3):
236-40.
Samet, J.M., S.L. Zeger, J.E. Kelsall, J. Xu and L.S. Kalkstein. 1997. Air Pollution, Weather, and Mortality in
Philadelphia 1973-1988. Health Effects Institute. Cambridge, MA. March.
Schwartz, J. 1993. Particulate Air Pollution and Chronic Respiratory Disease. Environ Res. 62:7-13.
Schwartz, J. 1994a. Air Pollution and Hospital Admissions For the Elderly in Birmingham, Alabama.
American Journal of Epidemiology. 139(6): 589-598.
Schwartz, J. 1994b. Air Pollution and Hospital Admissions For the Elderly in Detroit, Michigan. American
Journal of Respiratory and Critical Care Medicine. 150(3): 648-655.
Schwartz, J. 1994c. PM(10) Ozone, and Hospital Admissions For the Elderly in Minneapolis St Paul,
Minnesota. Archives of Environmental Health. 49(5): 366-374.
Schwartz, J. 1995. Short term fluctuations in air pollution and hospital admissions of the elderly for
respiratory disease. Thorax. 50(5): 531-538.
Schwartz, J. 1996. Air pollution and hospital admissions for respiratory disease. Epidemiology. 7(1): 20-28.
Schwartz, J. 1997. Air pollution and hospital admissions for cardiovascular disease in Tucson.
Epidemiology. 8(4): 371-377.
D-99
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Schwartz,}. 1999. Air pollution and hospital admissions for heart disease in eight U.S. counties.
Epidemiology. 10(1): 17-22.
Schwartz,]., D.W. Dockery, L.M. Neas, D. WypijJ.H. WareJ.D. Spengler, P. Koutrakis, F.E. Speizer and
E.G. Ferris. 1994. Acute Effects of Summer Air Pollution On Respiratory Symptom Reporting in
Children. Am J Respir Grit Care Med. 150(5): 1234-1242.
Schwartz,}, and R. Morris. 1995. Air Pollution and Hospital Admissions For Cardiovascular Disease in
Detroit, Michigan. American Journal of Epidemiology. 142(1): 23-35.
Schwartz,]., D. Slater, T.V. Larson, W.E. Pierson and J.Q. Koenig. 1993. Participate air pollution and
hospital emergency room visits for asthma in Seattle. Am Rev Respir Dis. 147(4): 826-31.
Schwartz,], and S. Zeger. 1990. Passive Smoking, Air Pollution, and Acute Respiratory Symptoms in a
Diary Study of Student Nurses. American Review of Respiratory Disease. 141(1): 62-67.
Sheppard, L., D. Levy, G. Norris, T.V. Larson and J.Q. Koenig. 1999. Effects of ambient air pollution on
nonelderly asthma hospital admissions in Seattle, Washington, 1987-1994. Epidemiology. 10(1): 23-
30.
Smith, D.H., D.C. Malone, K.A. Lawson, LJ. Okamoto, C. Battista and W.B. Saunders. 1997. A national
estimate of the economic costs of asthma. Am J Respir Crit Care Med. 156(3 Pt 1): 787-93.
Spix, C., H.R. Anderson,]. Schwartz, M.A. Vigotti, A. LeTertre, J.M. Vonk, G. Touloumi, F. Balducci, T.
Piekarski, L. Bacharova, A. Tobias, A. Ponka and K. Katsouyanni. 1998. Short-term effects of air
pollution on hospital admissions of respiratory diseases in Europe: A quantitative summary of
APHEA study results. Archives of Environmental Health. 53(1): 54-64.
Stieb, D.M., R.T. Burnett, R.C. Beveridge and J.R. Brook. 1996. Association between ozone and asthma
emergency department visits in Saint John, New Brunswick, Canada. Environmental Health
Perspectives. 104(12): 1354-1360.
Sunyer, J., M. Saez, C. Murillo, J. Castellsague, F. Martinez and J.M. An to. 1993. Air pollution and
emergency room admissions for chronic obstructive pulmonary disease: a 5-year study. Am J
Epidemiol. 137(7): 701-5.
Sunyer, J., C. Spix, P. Quenel, A. PoncedeLeon, A. Ponka, T. Barumandzadeh, G. Touloumi, L. Bacharova,
B. Wojtyniak, J. Vonk, L. Bisanti, J. Schwartz and K. Katsouyanni. 1997. Urban air pollution and
emergency admissions for asthma in four European cities: the APHEA Project. Thorax. 52(9): 760-
765.
Tenias, J.M., F. Ballester and M.L. Rivera. 1998. Association between hospital emergency visits for asthma
and air pollution in Valencia, Spain. Occup Environ Med. 55(8): 541-7.
Thurston, G.D., K. Ito, C.G. Hayes, D.V Bates and M. Lippmann. 1994. Respiratory Hospital Admissions
and Summertime Haze Air Pollution in Toronto, Ontario - Consideration of the Role of Acid
Aerosols. Environ Res. 65(2): 271-290.
D-100
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Thurston, G.D., K. Ito, P.L. Kinney and M. Lippmann. 1992. A Multi-Year Study of Air Pollution and
Respiratory Hospital Admissions in 3 New-York State Metropolitan Areas - Results For 1988 and
1989 Summers. Journal of Exposure Analysis and Environmental Epidemiology. 2(4): 429-450.
Touloumi, G., E. Samoli and K. Katsouyanni. 1996. Daily Mortality and Winter Type Air Pollution in
Athens, Greece - a Time Series Analysis Within the Aphea Project. Journal of Epidemiology and
Community Health. 50(S1): S47-S51.
U.S. Bureau of Econmic Analysis. 1995. BEA Regional Projections to 2045: Volume 1, States. U.S.
Departmnet of Commerce. Washington, DC. July.
U.S. EPA. 1991. Air Quality Criteria for Carbon Monoxide. U.S. EPA, Office of Research and
Development. Washington, DC. EPA/600/8-90/045F. December.
U.S. EPA. 1997. The Benefits and Costs of the Clean Air Act: 1970 to 1990. U.S. EPA, Office of Air and
Radiation, Office of Policy, Planning and Evaluation. Washington, DC. October.
U.S. EPA. 1999. An SAB Advisory: The Clean Air Act Section 812 Prospective Study Health and Ecological
Initial Studies. Prepared by the Health and Ecological Effects Subcommittee (HEES) of the
Advisory Council on the Clean Air Compliance Analysis, Science Advisory Board, U.S.
Environmental Protection Agency. Washington, DC. EPA-SAB-Council-ADV-99-005. February.
Vigotti, M.A., G. Rossi, L. Bisanti, A. Zanobetti and J. Schwartz. 1996. Short term effects of urban air
pollution on respiratory health in Milan, Italy, 1980-89. J Epidemiol Community Health. 50 Suppl 1:
s71-5.
von Mutius, E., D.L. Sherrill, C. Fritzsch, F.D. Martinez and M.D. Lebowitz. 1995. Air pollution and upper
respiratory symptoms in children from East Germany. Eur Respir J. 8(5): 723-8.
Wang, X., H. Ding, L. Ryan and X. Xu. 1997. Association between air pollution and low birth weight: a
community-based study. Environ Health Perspect. 105(5): 514-20.
Weisel, C.P., R.P. Cody and PJ. Lioy. 1995. Relationship between summertime ambient ozone levels and
emergency department visits for asthma in central New Jersey. Environ Health Perspect. 103 Suppl
2: 97-102.
Wessex,!. 1994. PRO/FILER, US Demographics and ZIPS: Summary Tape File 1A. 1.1. (CD-ROM).
Winnetka, Illinois.
White, M.C., R.A. Etzel, W.D. Wilcox and C. Lloyd. 1994. Exacerbations of childhood asthma and ozone
pollution in Atlanta. Environ Res. 65(1): 56-68.
Whittemore, A.S. and E.L. Korn. 1980. Asthma and Air Pollution in the Los Angeles Area. Am J Public
Health. 70: 687-696.
Woodruff, T.J., J. Grille and K.C. Schoendorf. 1997. The relationship between selected causes of
postneonatal infant mortality and particulate air pollution in the United States. Environmental
Health Perspectives. 105(6): 608-612.
D-101
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
World Health Organization (WHO). 1996. Final Consultation on Updating and Revision of the Air Quality
Guidelines for Europe. Bilthoven, The Netherlands 28-31 October, 1996 ICP EHH 018 VD 96
2.11.
Xu, X., H. Ding and X. Wang. 1995a. Acute effects of total suspended particles and sulfur dioxides on
preterm delivery: a community-based cohort study. Arch Environ Health. 50(6): 407-15.
Xu, X., B. Li and H. Huang. 1995b. Air pollution and unscheduled hospital outpatient and emergency room
visits. Environ Health Perspect. 103(3): 286-9.
Xu, X. and L. Wang. 1993. Association of indoor and outdoor particulate level with chronic respiratory
illness. Am Rev Respir Dis. 148(6 Pt 1): 1516-22.
Yang, W., B.L. Jennison and S.T. Omaye. 1998. Cardiovascular disease hospitalization and ambient levels of
carbon monoxide. J Toxicol Environ Health. 55(3): 185-96.
Zemp, E., S. Elsasser, C. Schindler, N. Kunzli, A.P. Perruchoud, G. Domenighetti, T. Medici, U.
Ackermann-Liebrich, P. Leuenberger, C. Monn, G. Bolognini, J.P. Bongard, O. Brandli, W. Karrer,
R. Keller, M.H. Schoni,J.M. Tschopp, B. Villiger and J.P. Zellweger. 1999. Long-term ambient air
pollution and respiratory symptoms in adults (SAPALDIA study). The SAPALDIA Team. Am J
Respir Cnt Care Med. 159(4 Pt 1): 1257-66.
D-102
-------
Ecological Effects
Introduction
This appendix characterizes the ecological benefits
of the Clean Air Act Amendments. Although EPA's
focus on a clean environment has long included
protection of both ecosystem health and human
health, many past analyses, particularly economic
analyses, have focused on human health benefits of
pollution control. Ecological benefits, by comparison,
have not always been as well-represented, for a variety
of reasons:
• Ecological impacts may be complex and non-
linear, involving relationships at various levels
of biological organization. Important
ecological effects such as population decline
of a keystone species can ripple through a
food web and alter community structure and
ecosystem function.
• Ecological systems, like human bodies,
possess a wide range of adaptive capacities
that can mitigate or mask effects and make
them difficult to detect. What differentiates,
and further complicates the measurement of
ecological effects is the lack of sufficient
baseline data on natural ecosystem structure
and function though successional stages.
• Prevention of ecological effects may be
viewed by the public and decision-makers as
a lower priority than the protection of human
health.
Nonetheless, within the last few decades air
pollution started to receive attention for not only
affecting human health but also its dramatic injuries to
ecosystems. Increased public awareness and research
results have led to the development of air pollution
research as an important branch of applied biological
sciences. Numerous scientific studies have revealed
adverse effects of air pollution on natural systems that
have, in turn, led to increasingly heightened levels of
public concern and subsequent environmental
statutory developments. Public policy concerning the
regulation of air pollution to mitigate these impacts
requires accurate appraisals of the effectiveness of
regulatory options, but not until quite recently has it
become possible to reliably quantify at least some of
the ecological and economic benefits of ecosystem
impacts linked to air pollution.
This analysis attempts to incrementally expand the
base of quantitative and qualitative information that
can be used to assess affects to ecosystems associated
with air pollution. There are two major goals of the
analysis: to provide a broad overall characterization of
the range of effects of air pollutants on ecosystem
structure, function, and health; and to extend existing
methods and data to characterize the potential
magnitude of economic benefits derived from the
1990 Clean Air Act Amendments (CAAA). The
economic analysis is focused on a relatively small
subset of effects for which ecologists' and economists'
understanding of and ability to model an effect is
sufficient to develop a quantitative characterization.
In most cases, we rely on published, peer-reviewed
literature to establish the validity of the methods and
data applied.
The remainder of this appendix is comprised of
seven major sections. We first provide a broad
overview of the ecological impacts of the air
pollutants regulated by the CAAA, and then outline
the rationale for choosing a subset of these effects for
quantitative and economic analysis. Following this
largely qualitative characterization of effects, we
describe the methods, data, and results used to
quantitatively assess benefits of the Clean Air Act
Amendments for the following categories of effects:
E-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
• Eutrophication of estuaries associated with
airborne nitrogen deposition;
• Acidification of freshwater bodies associated
with airborne nitrogen and sulfur deposition;
• Reduced tree growth associated with ozone
exposure;
• Accumulation of toxics in freshwater fisheries
associated with airborne toxics deposition;
• Aesthetic degradation of forests associated
with ozone and airborne toxics exposure;
• Other less well-understood effects of air
pollution on ecosystem health.
The concluding section includes a summary of those
economic estimates that are used in the larger 812
analysis, a summary of major limitations, and
recommendations for future research.1
Because the breadth and complexity of air
pollutant-ecosystem interactions does not allow for
comprehensive quantitative analysis of all the
ecological benefits of the CAAA we stress the
importance of continued consideration of those
impacts not valued in this report in policy decision-
making and in further technical research. Judging
from the geographic breadth and magnitude of the
relatively modest subset of impacts that we find
sufficiently well-understood to quantify and monetize,
it is evident that the economic benefits of the CAAA's
reduction of air pollution impacts on ecosystems are
of a large magnitude.
1 More detailed documentation of the ecological benefits of
the CAAA can be found in a series of memoranda and other work
products prepared as a part of EPA's benefits analysis and research
effort. These memoranda provide comprehensive descriptions of
the ecological impacts avoided by the CAAA, methods used to
characterize those damages, data sources, and ecological and
economic benefits assessments. The more detailed documentation
can be obtained through the EPA contacts identified in the
Acknowledgements section of the overall report.
Ecological Overview of the
Impacts of Air Pollutants
Regulated by the CAAA
The purpose of this section is to provide an
overview of potential interactions between air
pollutants and the natural environment. We identify
major single pollutant-environment interactions, as
well as the synergistic impacts of ecosystem exposure
to multiple air pollutants. Although a wide variety of
complex air pollution-environment interactions are
described or hypothesized in the literature, for the
purposes of this analysis we focus on major aspects of
ecosystem-pollutant interactions. We do this by
limiting our review according to the following criteria:
• Pollutants regulated by the CAAA;
• Known interactions between pollutants and
natural systems as documented in
peer-reviewed literature; and,
• Pollutants present in the atmosphere in
sufficient amounts after 1970 to cause
significant damages to natural systems.
Our understanding of air pollution effects on
ecosystems has progressed considerably during the
past decades. Previously, air pollution was regarded
primarily as a local phenomenon and concern was
associated with the vicinity of industrial facilities,
power plants or urban areas. The pollutants of
concern were gaseous (e.g., sulfur dioxide and ozone)
or heavy metals (e.g., lead) and the observed effects
were visible stress- specific symptoms of injury (e.g.,
foliar chlorosis). The most typical approach to
document the effects of specific pollutants was a
dose-response experiment, where the objective was to
develop a regression equation describing the
relationship between exposure and some easily
measured effect (e.g., growth, yield or mortality). As
analytic methods improved and ecology progressed, a
broader range of effects of air pollutants were
identified and understanding of the mechanisms of
effect improved. Observations made on various
temporal scales (e.g., long-term studies) and spatial
scales (e.g., watershed studies) lead to the recognition
E-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
that air pollution can affect all organizational levels of
biological systems.
In this analysis, we attempt to broadly describe
the impacts of air pollutants at all levels of
organization, though we are constrained to a review of
the most significant impacts on a national scale. For
a comprehensive review of the ecological impacts of
air pollutants regulated under the CAAA, see Overview
of Ecological Impacts of Air Pollutants Regulated by the 1990
Clean Air Act Amendments (EPA, 1998a).
Effects of Atmospheric Pollutants on
Natural Systems
Ecosystem impacts can be organized by the
pollutants of concern and by the level of biological
organization at which impacts are directly measured.
We attempt to address both dimensions of
categorization in this overview. In Table E-l we
summarize the major pollutants of concern, and the
documented acute and long-term ecological impacts
associated with them. We follow with a description of
each of the major pollutant classes and conclude with
a summary of pollutant impacts at each level of
biological organization.
Table E-1
Classes of Pollutants And Ecological Effects
Pollutant Class
Acidic deposition
Nitrogen Deposition
Hazardous Air
Pollutants (HAPs)
Ozone
Major Pollutants and
Precursors
Sulfuric acid, nitric acid
Precursors: Sulfur dioxide,
nitrogen oxides
Nitrogen compounds (e.g.,
nitrogen oxides)
Mercury, dioxins
Tropospheric ozone
Precursors: Nitrogen oxides
Acute Effects
Direct toxic effects to plant
leaves and aquatic
organisms.
Direct toxic effects to
animals.
Direct toxic effects to plant
leaves.
Long-term Effects
Progressive deterioration of soil
quality. Chronic acidification of
surface waters.
Saturation of terrestrial ecosystems
with nitrogen. Progressive nitrogen
enrichment of coastal estuaries.
Conservation of mercury and dioxins
in biogeochemical cycles and
accumulation in the food chain.
Alterations of ecosystem wide
patterns of energy flow and nutrient
cycling.
and Volatile Organic
Compounds (VOCs)
E-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Acidic Deposition
Acidification is perhaps the best-studied effect of
atmospheric pollutant deposition to natural
environments. Acidification of ecosystems has been
shown to cause direct toxic effects to sensitive
organisms as well as long-term changes in ecosystem
functions. Acidification can affect all levels of
biological organization in both terrestrial and aquatic
ecosystems. Adverse effects seen in terrestrial
ecosystems can include acute toxic interactions of
acids with terrestrial plants or, more importantly,
chronic acidification of terrestrial ecosystems leading
to nutrient deficiencies in soils, aluminum
mobilization, and concomitant decreases in health and
biological productivity of forests (Smith, 1990;SOS/T
18, 1989). Similar to terrestrial ecosystems, adverse
acidification-induced effects on surface waters may
include elevated mortality rates of sensitive species,
changes in the composition of communities, and
changes in ecosystem-level interactions like nutrient
cycling and energy flows. In the United States,
acidification-related injuries to aquatic ecosystems may
be more significant than injuries of terrestrial sites
(EPA, 1995a;NAPAP 1991).
The predominant causes of acidic precipitation are
sulfuric and nitric acid (H2SO4 and HNO3). These
strong mineral acids are formed from sulfur dioxide
(SO2) and nitrogen oxides (NCQ in the atmosphere
(i.e., they are secondary pollutants). Sulfur compounds
are emitted from anthropogenic sources in the form
of sulfur dioxide and, to a much lesser extent, primary
sulfates, principally from coal and residual-oil
combustion and a few industrial processes. Since the
late 1960s electric utilities have been the major
contributor to SO2 emissions (NAPAP, 1991 p. 178).
The combustion of fuels is the principal
anthropogenic source of emissions of NOX. Such
combustion occurs in internal combustion engines,
residential and commercial furnaces, industrial boilers,
electric utility boilers, engines, and other miscellaneous
sources. Because a large portion of anthropogenic
NOX emissions come from transportation sources (i.e.,
non-point source pollution), NOX sources are on
average more dispersed compared with anthropogenic
sources of SO2 (NAPAP, 1991, p. 189).
In the atmosphere, SO2 and NOX are converted to
sulfates and nitrates, transported over long distances,
and deposited over large areas downwind of point
sources or in the vicinity of urban areas. Deposition
occurs via three main pathways: 1) precipitation or
wet deposition, where material is dissolved in rain or
snow; 2) dry deposition, involving direct deposition of
gases and particles (aerosols) to any surface; and 3)
cloud-water deposition, involving material dissolved
in cloud droplets that is deposited when cloud or fog
droplets are intercepted by vegetation (NAPAP, 1991,
p. 181).
Initially, it was thought that SO2 emissions were
the only significant contributor to acidic deposition.
Subsequently, emissions of SO2 declined substantially
between 1970 and 1988 due to a variety of factors,
including emissions controls mandated by the Clean
Air Act and changes in industrial processes such as
the switch of electric utility plants to coal with lower
sulfur content (NAPAP, 1991, p. 198). During this
period, the role of nitrogen deposition as a
contributor to aquatic acidification became apparent.
While initial evidence suggested that most deposited
nitrogen would be taken up by biota, more recent
research has indicated that nitrogen may be leaching
from terrestrial systems and causing aquatic
acidification.
Comprehensive research on the ecological
impacts of acidification is found in the publications of
the National Acid Precipitation Assessment Program
and EPA's Add Deposition Standard Feasibility Study
Report to Congress (1995a). In this analysis we rely
heavily upon the extensive research conducted under
these two programs.
Nitrogen Deposition
Atmospheric nitrogen deposition to terrestrial and
aquatic ecosystems can cause deleterious ecological
effects ranging from eutrophication to acidification (as
discussed above). Deposition of nitrogen can
stimulate nitrogen-uptake by plants and
microorganisms and increase biological productivity
and growth. Chronic deposition of nitrogen may
adversely affect biogeochemical cycles of watersheds
E-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
by progressively saturating terrestrial portions with
nitrogen. Nitrogen saturation is a gradually occurring
process, during which watersheds undergo progressive
changes in their nitrogen cycle. This process can lead
to increases in the amount of nitrogen leached into
lower-elevation terrestrial ecosystems, wetlands and,
most notably, surface waters (Stoddard, 1994; Aber et
al., 1989).
Among the most pernicious effects of nitrogen
leaching from terrestrial ecosystems can be
acidification of fresh water bodies (as previously
discussed) and eutrophication of estuaries
(Richardson, 1996; Vollweider et al., 1990). Similar to
terrestrial ecosystems, nitrogen enrichment of coastal
estuaries can have a fertilizing effect, stimulating
productivity of algae, marine plants (Vitusek and
Howarth, 1991) and aquatic animals, including fish
and shell fish. If eutrophication is excessive, however,
it is likely to result in serious damages to estuarine
ecosystems. Specifically, massive algae blooms can
develop, leading to declining oxygen levels, habitat
loss, and declines in fish and shellfish populations.
Nitrogen loading to estuaries is a major and
growing problem. A 1996 inventory of estuarine
water quality performed by coastal states and
encompassing 72 percent of estuaries in the U.S.
shows that nutrient enrichment pollutes 6,254 square
miles (22 percent) of the surveyed waters, and
contributes to 57 percent of all the reported water
quality problems. At a recent meeting of National
Estuary Program directors, eleven out of twenty-eight
directors ranked nutrient overloading as a high priority
issue for their programs, and seven additional
directors ranked it as a mid-level priority (EPA,
1997b). Eighty-six percent of East Coast estuaries are
considered susceptible to nitrogen enrichment (EPA,
1997c), and many coastal communities are finding that
the nutrient loading problem is already so severe that
they must add advanced wastewater treatment to
existing plants, add infrastructure to promote water
reuse, and impose stricter controls on all development
and agricultural practices (EPA, 1997a).
Atmospherically derived nitrogen makes up a
sizable fraction of total nitrogen inputs to estuaries in
the Northeast pinga et al. 1991; Jaworski et al. 1997;
Paerl 1997; Paerl et al. 1990; McMahon and
Woodside, 1996; Rendell et al. 1993; Valiela et al.
1997). Atmospheric nitrogen is deposited to waters
and watersheds in wet (rain, snow, and fog) and dry
(aerosols and gases) forms. Approximately 10 to 50
percent of total nitrogen load to coastal waters is
derived from direct and indirect atmospheric
deposition. Estuaries on the eastern seaboard, and
those downwind of urban areas tend to have a larger
percentage of total nitrogen coming from atmospheric
deposition.
Hazardous Air Pollutant Deposition
Hazardous air pollutants (HAPs), are a general
category of toxic substances covered under a single
title of the Clean Air Act. Title III lists 189 HAPs,
though only five are responsible for the majority of
currently documented ecosystem impacts. These
HAPs are mercury, polychlorinated biphenyls (PCBs),
chlordane, dioxins, and dichlorodiphenyl-
trichloroethane (DDT). The use of three of these
compounds (PCBs, chlordane, and DDT) was
effectively illegal in the United States prior to 1990
(EPA 1992), and there are currently no plans for
additional CAAA regulations of these compounds
(Federal Register Unified Agenda 1998). Emissions of
the remaining two toxins, mercury and dioxins,
continue to cause ecosystem impacts.
Mercury (Hg) is a toxic element found
ubiquitously throughout the environment. Unlike
many HAPs, much of the mercury released to the
environment comes from natural sources.
Anthropogenic sources can also release mercury to the
environment. Estimates of the percentage of mercury
emissions attributable to anthropogenic sources range
from 10 to 80 percent (Mason et al. 1994, Hudson et
al. 1995, Stein et al. 1996), although most estimates
cluster between 40 and 75 percent (EPA 1997d).
About 80 percent of all anthropogenic mercury
loadings to the environment are from air emissions.
Global atmospheric concentrations of mercury have
approximately tripled since pre-industrial times
(Mason et al. 1994). Atmospheric deposition of
E-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
mercury has increased by a factor of about 3.7 (Swain
et al. 1992), and the concentration of mercury in
sediments in remote lakes has increased by a factor of
2.3 (Lucotte et al. 1995). These findings suggest that
approximately 57 to 73 percent of atmospherically
deposited mercury is anthropogenic in origin.
Atmospheric deposition of mercury and its
subsequent movement in ecosystems may result in the
concentration of mercury within organisms
("bioaccumulation") and its subsequent transfer
throughout the food chain. As a consequence,
mercury tends to accumulate along the hierarchical
organization of food webs, with increasing
concentrations found in animals at higher levels of the
food chain ("biomagnification"). Fish, birds and
mammals are among the group of organisms most
threatened by mercury contamination of the
environment. Symptoms may range from behavioral
abnormalities to reduced reproductive success and
death (EPA, 1997d). In 1996, mercury levels in fish
were high enough that 11 states had mercury-based
statewide fish consumption advisories. Twenty-eight
more had at least one water body under advisory
because of mercury concerns. These observations
suggest that atmospheric mercury deposition may
contribute significantly to mercury levels in freshwater
ecosystems nationally.
Mercury is a neurotoxin that, at sufficient levels,
can cause neurologic damage and death in both
animals and humans. Adverse effects on wildlife
include neurotoxicity, reproductive, and
developmental effects (EPA 1997d). While fish are
unlikely to experience toxic effects from mercury
poisoning in the absence of point discharges,
piscivorous predators and predators who eat
piscivores accumulate more mercury and may suffer
from mercury poisoning. However, the only species
for which there is currently strong evidence of
poisoning from atmospheric mercury are the common
loon and possibly the Florida panther. It is unclear
whether other piscivorous species, such as kingfishers,
mink, and river otters, have suffered adverse health
effects as a consequence of atmospheric mercury
deposition (EPA 1997d).
Mercury is likely to persist at levels of concern in
ecosystems for some time. The majority of
atmospherically-released mercury is deposited to
terrestrial environments, where it is largely
sequestered. However, as mercury accumulates in
soils, some amount (less than 30 percent of that which
is deposited within a watershed) will be slowly released
to freshwater bodies and oceans. Modeling efforts by
Swain et al. (1992, reviewed in Mason et al. 1994)
suggest that the retention of mercury by some lakes is
essentially complete. Studies by Mason et al. (1994)
predicted that elimination of anthropogenic mercury
presently in the oceans and in the atmosphere would
take 15 to 20 years after the complete termination of
all anthropogenic emissions. Because of mercury's
persistence in terrestrial and aquatic environments, it
will probably take some time for reductions in
mercury emissions to be notable in ecosystems (Swain
et al. 1992, reviewed in Mason et al. 1994).
The other HAPs of concern are polychlorinated
dibenzo-p-dioxins (PCDDs), a group of 75
organochlorine compounds that are sometimes
referred to as dioxins. The most toxic member of this
group is 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD).
Because TCDD is the most toxic dioxin, the toxicity
of a dioxin mixture is often expressed as the toxic
equivalency (TEQ) of some amount of TCDD.
Polychlorinated dibenzofurans (PCDFs) are close
chemical relatives of PCDDs. Both classes of
compounds are produced by the same processes, and
both are ubiquitous in the environment (WHO 1989).
TEQ estimates are often given jointly for dioxins and
furans.
Dioxins and furans, unlike mercury, are not
natural to the environment. They are formed during
the combustion of wastes and fossil fuels, and as a
consequence of fires or spills that involve particular
chemicals like benzenes or PCBs. They are also
formed as by-products in both the manufacturing of
other chemicals and in pulp and paper mill bleaching
processes (WHO 1989, EPA 1992a). EPA estimates
that combustion sources emit over ten times as many
TEQs as did all other categories combined.
E-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Dioxins accumulate in the fat of animals and
bioaccumulate through food chains. Fish are among
the most sensitive vertebrates to the effects of TCDD,
especially during early life stages. Fish are exposed to
dioxins primarily through their food (EPA 1993), but
some studies have reported that they can also absorb
the trace amounts of dioxins present in water. EPA
(1992) reported bioconcentration factors (BCFs)2 for
dioxins in fish of 5,000 to 9,000, and EPA (1993)
estimated that bioaccumulation factors (BAFs) for
lake trout can be on the order of 500,000 to
1,200,000.3 Toxic effects on young fish include
decreased feeding, weight loss, and fin necrosis;
however, with a few exceptions, TCDD levels in the
environment are generally too low to result in toxicity
to juvenile or adult fish (EPA 1993, Walker and
Peterson 1994).
The risk that dioxins pose to other wildlife is
difficult to assess because both laboratory and field
studies in this area are limited (EPA 1993, Giesey et al.
1994). One study (White et al. 1994) found that wood
duck eggs from a contaminated area had levels of
PCDDs and PCDFs 50 times higher than levels in
control eggs. The contaminated nests were
significantly less successful than control nests, and
contaminated ducklings also suffered from teratogenic
effects.
TCDD is an extremely stable chemical and is
unlikely to be significantly degraded by chemical or
biological hydrolysis under normal environmental
conditions. Its half-life in soils may be on the order of
a decade or more, and it may be even more persistent
2Bioconcentration factors (BCFs) are calculated based on
laboratory experiments. BCFs represent the ratio between the
chemical's concentration in the organism to its concentration in the
water, but unlike bioaccumulation factors (BAFs), they measure
only how much of a chemical an organism accumulates as a
consequence of its exposure to contaminated water. BCFs do not
measure contaminant uptake as a function of exposure to
contaminated food or sediments (EPA 1993).
Because dioxins have such low solubility, accurately
measuring their concentrations in water is extremely difficult. For
this reason, any reported BCFs and BAFs should be examined
carefully.
in aquatic sediments (Webster and Commoner 1990).
For example, Johnson et al. (1996) found that, though
TCDD levels in fish and sediments from an Arkansas
river declined significantly during the 12 years
following the initial pollution of the river, fish from
some locations continued to have levels of TCDD
that exceeded Food and Drug Administration (FDA)
guidelines. TCDD is subject to photochemical
degradation, but since the penetration of light into
soils and many natural water bodies is limited, this
degradation is not likely to be environmentally
significant (WHO 1989, Zook and Rappe 1990).
Because of dioxins' toxicity and persistence, their
presence in freshwater ecosystems is likely to be an
issue of concern for decades.
Tropospheric Ozone
Ozone pollution is widespread in the eastern
United States, in southern California, and in the
vicinity of most major cities. Many of the observed
effects of ozone on vegetation are related to direct
toxic or harmful interactions with essential
physiological functions of plants and subsequent
reductions in biomass production (reduced growth).
Damages to plants are commonly manifested as stress
specific symptoms such as necrotic spots of plant
leaves, acceleration of leaf aging, and reduced
photosynthesis. Ozone damages at the community
and ecosystem-level vary widely depending upon
numerous factors, including concentration and
temporal variation of tropospheric ozone, species
composition, soil properties and climatic factors. In
most instances, responses to chronic or recurrent
exposure are subtle and not observable for many
years. These injuries can cause stand-level forest
decline in sensitive ecosystems (EPA, 1996; McBride
et al., 1985; Miller et al., 1982).
Species that are particularly sensitive to ozone can
be found among all groups of plants. Although many
visible injuries have occurred in conifer species (e.g.,
Ponderosa and Jeffrey pine), a variety of deciduous
trees and shrubs are also sensitive to ozone. Black
cherry, many poplars (genus Populus) and many fruit
trees including almond (Prunus amygdalis Batsch), peach
(Prunuspersica), and plum (Prunus domestica) trees are all
E-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
affected by elevated levels of ozone (EPA 1996a). In
annual species, effects of ozone on production occur
through changes in allocation of carbohydrates and
can result in reduced seed production. Many annual
plant species, including commercial crops, are among
the most sensitive species. The National Crop Loss
Assessment Network (NCLAN 1988), a multi-year
program of the EPA, established that ambient ozone
levels cause physical damages to crop plants and
statistically significant reductions in agricultural yields.
Ecosystems with known damages that are
attributed to ozone include the San Bernardino
Mountains of Southern California, the Sierra Nevada
Mountains, and sites in the vicinity of urban areas
throughout the country. According to EPA (1996a),
the San Bernardino Mountain range is by far the most
severely ozone-impacted ecosystem. This
mixed-conifer forest ecosystem has been exposed to
chronically elevated ozone levels over a period of 50
or more years. This exposure has resulted in major
changes of ecosystem characteristics, including species
composition, nutrient cycling and energy flow. The
first indications of ozone damages to the ecosystem
were observed on the more sensitive members of the
forest community: individual Ponderosa and Jeffrey
pines. Direct injuries included visible foliar damage,
premature needle senescence, reduced photosynthesis,
altered carbon allocation, and reduction of growth
rates and reproductive success. Changes in the energy
available to trees (i.e., changes in carbohydrate
production and allocation) influenced interactions
with predators, pathogens and symbionts.
Subsequently, the accumulation of weakened trees
resulted in heavy bark beetle attack that significantly
elevated mortality rates and extensive salvage logging
during the 1960s and 1970s (Miller and McBnde,
1998). Alterations in the composition and population
density of the fungal microflora weakened soil
microbial organisms and slowed the rate of
decomposition, leading to the accumulation of a thick
needle layer under stands with the most severe needle
injuries and defoliation. Reduced production of seeds
and fruits also affected the amount of food available
to small vertebrates in the ecosystem, thereby affecting
the local food chain (EPA 1996a). Similarly, ozone
concentrations capable of causing injury to the Sierra
Nevada Mountains have been occurring for many
years, but injury to sensitive trees has never reached
the same proportions as in the San Bernardino Forest.
Significant differences in both the forest stand
composition (e.g., the presence of fewer conifers and
more hardwoods), and site dynamics have probably
played an important role in determining the different
ecosystem responses.
In each of these areas, ozone may act
synergistically with other stress factors to induce
further damages to vegetation. In the eastern United
States, for example, regionally elevated levels of
tropospheric ozone co-occur with high deposition
rates of nitrogen, sulfur and acids. These multiple
stress factors may have acted synergistically in injuring
many high elevation forests throughout the eastern
United States.
Multiple Stresses and Patterns of
Exposure
Although air pollutants can be grouped into
classes according to their effects, as described above,
it is recognized that one pollutant (or one class of
pollutants) does not solely impact most ecosystems.
Many environmental damages are the result of the
combined action of multiple stress factors, including
several types of air pollution and other anthropogenic
or natural stress factors.
The recognition of interactions between several
types of pollutants and between pollutants and other
kinds of stress has introduced a new level of
complexity in air pollution research. In many cases,
various stress factors act synergistically to induce
damages to ecosystems. These multiple stress factors
can include: (1) various kinds of air pollutants (e.g.,
acidic deposition, ozone and nitrogen deposition); (2)
other anthropogenic stress factors such as harvesting,
overfishing or habitat disruption (e.g., the disruption
of ecosystems by roads or urban areas); (3)
environmental factors including availability of water,
nutrients, light, or temperature (including heat and
frost); and (4) biological factors such as animals
feeding on plants, pathogens, and the status of
micro-organisms in the soil (Taylor et al., 1994;
E-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Winner, 1994; Smith, 1990). In the eastern US, for
example, elevated deposition rates of nitrogen, sulfur
and acids co-occur with regionally elevated levels of
tropospheric ozone and climatic stress factors.
In addition, air pollutants have indirect effects
that are at least as important as direct toxic effects on
living organisms. Indirect effects include those in
which the pollutants (s) alter the physical or chemical
environment (e.g., soil properties) the plant's ability to
compete for limited resources (e.g., water, light), or its
ability to withstand pests or pathogens. Examples are
excessive availability of nitrogen, soil depletion caused
by acidic deposition, and changes in the ability to
adapt to cold temperatures induced by acidic
deposition (Taylor et al., 1994; NAPAP, 1991).
Unfortunately, few mechanisms of interactions
between various stress factors are known, and
interpretations of scientific findings are usually
associated with a high degree of uncertainty.
The situation is further complicated by the fact
that the specific temporal and spatial patterns of
pollutant exposures play a significant role in the
response of organisms and ecosystems to air
pollution. Temporal patterns include timing, duration
and patterns of recurrent exposure to a specific
pollutant or pollutants. For example, plant response
to peak concentrations of ozone during daylight can
be more severe (compared to exposure to the same
level of ozone at night) because uptake of ozone is
often higher during the day (EPA, 1996a). Spatial
patterns include proximity of ecosystems to various
pollution sources and the identification of specific
source-receptor relationships between ecosystems and
pollution sources (Taylor et al., 1994). For example,
the deposition of most pollutants occurs after
long-range atmospheric transport, with deposition
rates depending upon climate, land-use and geology.
Summary of Ecological Impacts from Air
Pollutants Regulated by the CAAA
We summarize major examples of air pollution
interactions with various levels of biological
organization in Table E-2 through E-4. We organize
these interactions according to classes of pollutants
and injuries they cause, the various levels of biological
systems, and types of affected ecosystems. It is
important to note that interactions listed are intended
to illustrate the range of possible adverse effects.
These effects are examples for a wide variety of
interactions but do not cover all aspects of air
pollution-environment interactions.
E-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-2
Interactions Between Acid Deposition and Natural Systems
At Various Levels of Organization
Examples of Interactions
Spatial
Scale
Molecular
and cellular
Individual
Population
Community
Local
Ecosystem
(e.g.,
landscape
element)
Regional
Ecosystem
(e.g.,
watershed)
Type of Interaction
Chemical and
biochemical processes
Direct physiological
response
Indirect effects: Death
due to ionoregulatory
failure. Acidification can
indirectly affect response
to altered environmental
factors or alterations of
the individual's ability to
cope with other kinds of
stress.
Change of population
characteristics like
productivity or mortality
rates.
Changes of community
structure and competitive
patterns
Changes in nutrient
cycle, hydrological cycle,
and energy flow of lakes,
wetlands, forests,
grasslands, etc.
Biogeochemical cycles
within a watershed.
Region-wide alterations
of biodiversity.
Acidification of Forests
Damages to epidermal layers and
cells of plants through deposition
of acids.
Increased loss of nutrients via
foliar leaching.
Cation depletion in the soil
causes nutrient deficiencies in
plants. Concentrations of
aluminum ions in soils can reach
phytotoxic levels. Increased
sensitivity to other stress factors
like pathogens and frost.
Decrease of biological
productivity of sensitive
organisms. Selection for less
sensitive individuals.
Microevolution of resistance.
Alteration of competitive patterns.
Selective advantage for acid-
resistant species. Loss of acid
sensitive species and individuals.
Decrease in productivity.
Decrease of species richness and
diversity.
Progressive depletion of nutrient
cations in the soil. Increase in the
concentration of mobile aluminum
ions in the soil.
Leaching of sulfate, nitrate,
aluminum, and calcium to
streams and lakes. Acidification
of aquatic bodies.
Acidification of Streams
and Lakes
Decreases in pH and increases
in aluminum ions cause
pathological changes in
structure of gill tissue in fish.
Hydrogen and aluminum ions in
the water column impair
regulation of body ions.
Aluminum ions in the water
column can be toxic to many
aquatic organisms through
impairment of gill regulation.
Acidification can indirectly
affect submerged plant species,
because it reduces the
availability of dissolved carbon
dioxide (CO,).
Decrease of biological
productivity of sensitive
organisms. Selection for less
sensitive individuals.
Microevolution of resistance.
Alteration of competitive
patterns. Selective advantage
for acid-resistant species. Loss
of acid sensitive species and
individuals. Decrease in
productivity. Decrease of
species richness and diversity.
Measurable declines of
decomposition of some forms
of organic matter, potentially
resulting in decreased rates of
nutrient cycling.
Additional acidification of
aquatic systems through
processes in terrestrial sites
within the watershed.
E-10
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-3
Interactions Between Nitrogen Deposition and Natural Systems
At Various Levels of Organization
Examples of Interactions
Eutrophication and Nitrogen
Spatial
Scale
Molecular
and cellular
Individual
Population
Community
Local
Ecosystem
(e.g.,
landscape
element)
Regional
Ecosystem
(e.g.,
watershed)
Type of Interaction
Chemical and biochemical
processes
Direct physiological
response.
Indirect effects: Response
to altered environmental
factors or alterations of the
individual's ability to cope
with other kinds of stress.
Change of population
characteristics like
productivity or mortality
rates.
Changes of community
structure and competitive
patterns
Changes in nutrient cycle,
hydrological cycle, and
energy flow of lakes,
wetlands, forests,
grasslands, etc.
Biogeochemical cycles
within a watershed. Region-
wide alterations of
biodiversity.
Saturation of Terrestrial
Landscapes
Assimilation of nitrogen by
plants and microorganisms
Increases in leaf- size of
terrestrial plants.
Decreased resistance to biotic
and abiotic stress factors like
pathogens and frost. Disruption
of plant-symbiont relationships
with mycorrhiza fungi.
Increase in biological
productivity and growth rates of
some species.
Alteration of competitive
patterns. Selective advantage
for fast growing species and
individuals that efficiently use
additional nitrogen. Loss of
species adapted to nitrogen-
poor environments.
Magnification of the
biogeochemical nitrogen cycle.
Progressive saturation of
microorganisms, soils, and
plants with nitrogen.
Leaching of nitrogen from
terrestrial sites to streams and
lakes. Acidification of aquatic
bodies. Eutrophication of
estuaries.
Eutrophication of Coastal
Estuaries
Assimilation of nitrogen by
plants and microorganisms.
Increase in growth of marine
plants.
Injuries to marine fauna through
oxygen depletion of the
environment. Loss of physical
habitat due to loss of sea-grass
beds. Injury through increased
shading. Toxic blooms of
plankton.
Increase in biological
productivity. Increase of growth
rates (esp. of algae and marine
plants).
Excessive algal growth.
Changes in species
composition. Decrease in sea-
grass beds.
Magnification of the nitrogen
cycle. Depletion of oxygen,
increased shading through
algal growth.
Additional input of nitrogen
from nitrogen-saturated
terrestrial sites within the
watershed.
Note: See Overview of Ecological Impacts of Air Pollutants Regulated by the 1990 Clean Air Act Amendments (I EC 1998) for
sources.
E-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-4
Interactions of Mercury and Ozone with Natural Systems
At Various Levels of Organization
Examples of Interactions
Spatial
Scale
Type of Interaction
Mercury in streams
and lakes
Ozone
Molecular
and cellular
Chemical and
biochemical processes
Mercury enters the body of
vertebrates and binds to sulfhydril
groups (i.e. proteins).
Oxidation of enzymes of
plants. Disruption of the
membrane potential.
Individual
Direct physiological
response.
Neurological effects in vertebrates.
Behavioral abnormalities. Damages
to the liver.
Direct injuries include visible
foliar damage, premature
needle senescence, reduced
photosynthesis, altered
carbon allocation, and
reduction of growth rates and
reproductive success.
Indirect effects:
Response to altered
environmental factors or
alterations of the
individual's ability to
cope with other kinds of
stress.
Few interactions known.
Damages through increased
sensitivity to other environmental
stress factors could occur, for
example, through impairment of
immune response.
Increased sensitivity to biotic
and abiotic stress factors like
pathogens and frost.
Disruption of plant-symbiont
relationship (mychorrhizae),
and symbionts.
Population Change of population
characteristics like
productivity or mortality
rates.
Reduced reproductive success offish
and bird species. Increased mortality
rates, especially in earlier life stages.
Reduced biological
productivity. Selection for
less sensitive individuals.
Possibly microevolution for
ozone resistance.
Community
Changes of community
structure and
competitive patterns
Loss of species diversity of benthic
invertebrates.
Alteration of competitive
patterns. Selective
advantage for ozone-
resistant species. Loss of
ozone sensitive species and
individuals. Reduction in
productivity.
Local
Ecosystem
(e.g.Jandsc
ape
element)
Regional
Ecosystem
(e.g.,
watershed)
Changes in nutrient Not well understood.
cycle, hydrological
cycle, and energy flow of
lakes, wetlands, forests,
grasslands, etc.
Biogeochemical cycles Not well understood.
within a watershed.
Region-wide alterations
of biodiversity.
Alterations of ecosystem-
wide patterns of energy flow
and nutrient cycling.
Region-wide loss of sensitive
species.
Note: See Overview of Ecological Impacts of Air Pollutants Regulated by the 1990 Clean Air Act Amendments (I EC 1998) for
sources.
E-12
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Predicting ecological impacts of air pollution at
the regional scale or for the United States as a whole
would require an understanding of interactions at
many temporal and spatial scales, where there is
currently a general lack of data. Furthermore, there is
limited transferability of existing information between
various spatial and temporal scales and between
geographic regions. However, we can reach several
general conclusions, based on the existing literature.
• Although ambient concentrations of most air
pollutants significantly decreased after the
Clean Air Act of 1970, some pollutants still
occur in concentrations high enough to
directly injure living organisms. These direct
injuries can be observed, for example, in areas
with high ambient levels of tropospheric
ozone or in some high-elevation ecosystems
that are exposed to high levels of acid
deposition (EPA, 1996a;NAPAP, 1991).
• Air pollutants have indirect effects that are at
least as important as direct toxic effects on
living organisms. Indirect effects include
those in which the pollutant alters the
physical or chemical environment (e.g., soil
properties), the plant's ability to compete for
limited resources (e.g., water, light), or the
plant's ability to withstand pests or
pathogens. Examples are excessive
availability of nitrogen, depletion of nutrient
cations in the soil by acid deposition,
mobilization of toxic elements such as
aluminum, and changes in winter hardiness
(Taylor et al., 1994). As is true for other
complex interactions, indirect effects are
more difficult to observe than direct toxic
relationships between air pollutants and biota,
and there may be a variety of interactions that
have not yet been detected.
• There is a group of pollutants that tend to be
conserved in the landscape after they have
been deposited to ecosystems. These
conserved pollutants are transformed through
biotic and abiotic processes within
ecosystems, and accumulate in
biogeochemical cycles. These pollutants
include, but are not limited to, hydrogen ions
(H+), sulfur (S) and nitrogen (N) containing
substances, and mercury (Hg). Chronic
deposition of these pollutants, can result in
progressive increases in concentrations and
cause injuries due to cumulative effects.
Indirect, cumulative damages caused by
chronic exposure (i.e., long-term, moderate
concentrations) to these pollutants may
increase in magnitude over time frames of
decades or centuries with very subtle annual
increments of change. Examples are
N-saturation of terrestrial ecosystems, cation
depletion of terrestrial ecosystems,
acidification of streams and lakes, and
accumulation of mercury in aquatic food
webs (Pitelka 1994; Taylor et al. 1994; Likens
etal. 1996; EPA 1997e).
Damages to ecosystems are most likely
caused by a combination of environmental
stress factors with every interactive stress or
else have a mechanistic model that
incorporates interactions among pollutants.
Unfortunately neither approach is yet
possible. These include anthropogenic
factors such as air pollution and other
environmental stress factors such as low
temperature, excess or limited water, and
limited availability of nutrients. The specific
combinations of factors differ among regions
and ecosystems where declines have been
observed (Taylor et al., 1994; Winner, 1994;
Smith, 1990). To accurately predict the
impacts of multiple acting stress factors we
would have to build a catalogue of research
results that defines the response of every
plant species to every air pollutant, with every
interactive stress or else have a mechanistic
model that incorporates interactions among
pollutants. Unfortunately neither approach is
yet possible.
E-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
• Pollutant-environment interactions are
complicated by the fact that biotic and abiotic
factors in ecosystems change dramatically
over time. Besides oscillations on a daily
basis, and changes in a seasonal rhythm, long
range successional changes occur over time
periods of years, decades, or even centuries.
These temporal variations occur in polluted
and pristine ecosystems, and no single point
in time or space can be defined as
representative of the entire system.
Long-term impacts of air pollution are often
manifested in interactions at the regional scale. The
history of lead pollution may provide a useful
illustration of impacts of air pollution long after
deposition rates have declined significantly due to
environmental regulations. Historically, scientists
were concerned about lead deposition because of its
high affinity to soil organic matter and its
accumulation in the litter layers of soils. Starting
around 1960, lead accumulated in forest soils in the
northeastern United States as a result of human
activities. Following a significant decline of
combustion of leaded gasoline between 1970 and
1988, deposition rates dropped, and decreases in lead
levels in soils and rivers have been observed
throughout the United States. Apparently forest
floors have responded rapidly to the decline of lead
input, and instead of accumulating lead, forest soils
are now slowly releasing lead to the underlying mineral
horizon. It has been estimated that sometime in the
middle of the next century, forests will begin to release
anthropogenic lead deposited after 1960 to rivers and
streams (Miller and Friedland, 1994), where it may
cause unforeseen damages to aquatic ecosystems.
There is evidence that current air pollution is an
important environmental stress factor over large areas
of the United States and other countries, even if
effects have not yet been fully documented. Actions
taken now to reduce air emissions may have
consequences far into the future and may affect
ecosystems in ways that are not yet known. Because
it is not yet possible to predict what long-term,
continuous exposure to multiple pollutants might do
to ecosystem structure and function, it may be
concluded that the ecological benefits of air pollution
control lie in the prevention of long-term damages to
resources and the potential for increased recovery
rates, as well as the more traditional prevention of
acute injuries. Because it is not yet possible to predict
what long-term, continuous exposure to multiple
pollutants might do to ecosystem structure and
function, it may be prudent to focus on the
prevention of possible long-term damages to
resources and preserve the potential for increased
recovery rates, as well as preventing more traditional
acute injures to ecosystems.
Methodological Overview
In this section we describe the methods for
characterizing the economic benefits of reducing
several classes of ecological impacts through the
regulations of the CAAA. As indicated in the
previous section, it is not possible to characterize and
quantify all impacts associated with air pollution.
Instead, we select those impacts amenable to
quantitative analysis, using two criteria:
Criterion #1:
flow
ml must be an identi
'£ service
Criterion #2: A defensible link must exist between
changes in air pollution emissions and the quality or
quantity of the ecological service flow, and quantitative
models must be available to monetise these changes
The use of these criteria greatly constrains the
range of impacts that can be treated in this analysis.
While the previous section identifies many pollutant-
ecosystem interactions, only a handful are understood
and have been modeled to an extent sufficient to
reliably quantify their impact. We attempt to present
both reliable quantitative information regarding the
benefits of the CAAA while demonstrating the
potential magnitude of the ecological benefits of the
CAAA if all impacts were valued. A more detailed
description of the choice of endpoints is found in
E-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Methods for Selecting Moneti^able Benefits Derived from
Ecological Resources as a Result of Air Quality Improvements
Attributable to the 1990 Clean Air Act Amendments,
1990-2010 (lEc 1998b) and Characterising Economic
benefits of Reducing Impacts to Ecosystem Integrity (lEc
1998c).
Using Service Flow Endpoints for
Valuation
from environmental regulations in an analysis of
national scope.
Based on the constraints of economic valuation
methods and data, we select from the host of
ecosystem impacts identified in the previous section
a set of service flows as candidate endpoints for
analysis. These endpoints are listed in Table E-5.
The theoretical basis of economic benefits
assessment is that ecosystems provide services to
humankind, and that those services have economic
value. The application of this theory requires the
isolation of service flows that have market values or
are otherwise amenable to available methods for
determining value in the absence of formal markets.
Freeman (1997) provides one possible grouping of
ecological service flows:
• Sources of material inputs to the economy,
including fossil fuels, wood products,
minerals, water, and animals;
• Life support services, including breathable air
and a livable climate;
• Amenities that provide opportunities for
active recreation and passive enjoyment of
nature, including nonuse values; and
• Processing of waste products that are
generated by economic activity and
discharged into the environment.
Available methods do not exist to
comprehensively value each of these service flows for
all ecosystems. Generally, we are limited to those
service flows that either are sources of material inputs
or natural amenities that involve active recreation.
Impacts to these service flows that can be valued tend
to manifest themselves immediately and can be readily
measured and assessed in terms of the proven cause
and effect relationships. The result is that we can
value only a small subset of the ecosystem benefits
E-15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-5
Ecological Impacts with Identifiable Human Service Flows
Pollutant Class
Ecosystem Effect
Service Flow Impacted
Acidification (H2SO4,
HNO3)
High-elevation forest acidification resulting in
dieback
Freshwater acidification resulting in aquatic
organism (e.g. fish) population decline
Changes in biological diversity and species mix in
terrestrial and aquatic systems
Forest aesthetics
Recreational fishing
Existence value for maintenance of
biological diversity
Nitrogen Saturation
and Eutrophication
(NOX)
Freshwater acidification resulting in aquatic
organism (e.g. fish) population decline
Estuarine eutrophication causing oxygen depletion
and changes in nutrient cycling
Changes in biological diversity and species mix in
terrestrial and aquatic systems
Recreational fishing
Recreational and commercial
fishing
Existence value for maintenance of
biological diversity
Toxics Deposition
(Mercury, Dioxin)
Terrestrial bioaccumulation of mercury and dioxin
Aquatic bioaccumulation of mercury and dioxin
Changes in biological diversity and species mix in
terrestrial and aquatic systems
Hunting, wildlife aesthetics
Recreational and commercial
fishing
Existence value for maintenance of
biological diversity
Tropospheric Ozone
Terrestrial plant foliar damage causing lower
productivity and reduced competitiveness
Commercial timber productivity,
forest aesthetics, existence value
Multiple Pollutant
Stress
Ecosystem deterioration resulting in visual effects,
habitat loss, and changes in biological diversity and
species mix caused by synergistic action of several
pollutants
Ecosystem aesthetics, ecosystem
existence value
Defensible Links and Quantitative
Modeling Requirements
The second criterion for endpoint selection is
satisfied when complete data and model coverage is
available to describe the impacts of air pollutants. We
briefly describe the types of data and models necessary
to accomplish quantitative benefits assessment, then
identify those endpoints that we can pursue in this
analysis.
In order to determine changes in ecological
service flows, defensible links between pollution
emissions and service flow changes must be
quantitatively modeled. Described generally, five steps
are necessary to complete a quantitative analysis;
emissions characterization; environmental fate and
transport assessment; exposure characterization;
ecosystem effects characterization; and economic
behavior models.
Emissions characterization requires models that
project the level of air pollutants entering the
atmosphere over the period of time in question for
both factual and counterfactual scenarios under
consideration in the analysis. In our analyses the
factual scenario is the level of emissions in the United
States generated between 1990 and 2010, as regulated
by the CAAA (Post-CAAA). The counterfactual
scenario is the level of emissions during the same
E-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
period without the regulations promulgated under the
CAAA (Pre-CAAA)4.
The geographic transport and deposition of air
pollutants are estimated using models that consider
multiple chemical and meteorologic factors. The
section 812 prospective analysis of the CAAA uses
three models, detailed in Appendix C and Air Quality
Modeling to Support the Section 812 Prospective Analysis
(prepared for EPA by Systems Applications
International, Inc., 1999)5.
In cases where the presence of a pollutant in a
geographic region, as estimated by dispersion models,
is not an adequate measure of the exposure of biota to
the pollutant, an exposure model is required. These
models must take biotic and abiotic ecosystems
processes into account.
Once the exposure of the biota in question is
estimated, the physiological effect of that exposure
must be estimated. Dose-response functions that
describe the effects of varying levels of pollutants to
specific organisms are derived from laboratory, field,
and modeling experiments. The intensive nature of
this research and the necessity of studying each species
individually causes this link to be weak in most
quantitative ecological assessments.
4For all analyses in this report, emissions under each scenario
are based upon EPA's National Emissions Inventory (NEI) with
modeling provided by the Emissions Reduction and Cost Analysis
Model (ERCAM). See Appendix A for details.
5The three regional-scale air quality modeling systems applied
include the regulatory Modeling System for Aerosols and
Deposition (REMSAD), the Regional Acid Deposition Model
(RADM), and the Urban Airshed Model (UAM-IV and UAM-V).
In addition, this prospective ecological benefits assessment uses
results from the Regional Lagrangian Model of Air Pollution
(RELMAP) to estimate mercury and dioxin deposition.
E-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-6
Model Coverage for Candidate Endpoints for Quantitative Assessment
Pollutant
Acidification
(H2S04,
HN03)
Nitrogen
Saturation and
Eutrophication
(NOY)
Toxics
Deposition
(Hg, Dioxin)
Multiple
Pollutant
Stress
Endpoint
Forest
aesthetics
Recreational
fishing
Biological
diversity
existence value
Recreational
and commercial
fisheries
Biological
diversity
existence value
Forest
aesthetics
Hunting, wildlife
aesthetics
Recreational
and commercial
fishing
Biological
diversity
existence value
Ecosystem
aesthetics,
ecosystem
existence
value.
Emissions
Model
NEI,
ERCAM
NEI,
ERCAM
NEI,
ERCAM
NEI,
ERCAM
NEI,
ERCAM
None
Available
None
Available
None
Available
None
Available
NEI,
ERCAM
Transport
and
Deposition
RADM
RADM
RADM
RADM
RADM
RELMAP,
ISC3
RELMAP,
ISC3
RELMAP,
ISC3
RELMAP
Exposure
Model
Not
Required
MAGIC
(region
specific)
MAGIC
(region
specific)
Estuary-
specific
models
available
Estuary-
specific
models
available
None
Available
None
Available
IEM-2M (site
specific)
None
Available
None
Available
Dose-
response
Functions
Multiple
available
Multiple
available
Multiple
available
None
Available
Multiple
available
Multiple
available
Multiple
available
Multiple
available, or
consumption
advisory limits
can be used
Multiple
available
None
Available
Economic
Model
Only site-
specific
models
available
Only site-
specific
models
available
Only site-
specific
models
available
Only site-
specific
models
available
None Available
Only site-
specific
models
available
Only site-
specific
models
available
Only site-
specific
models
available
None Available
None Available
NEI: National Emissions Inventory; ERCAM: Emissions Reduction and Cost Analysis Model; RADM: Regional Acid Deposition
Model;REMSAD: Regulatory Modeling System for Aerosols and Deposition; RELMAP: Regional Lagrangian Model of Air Pollution;
UAM: Urban Airshed Model; TAMM: Timber Assessment Market Model, developed and maintained by the U.S. Forest Service.
E-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
When pollutant doses are sufficiently high to
cause physiologic responses in biota, ecological service
flows may be affected. In order to monetize these
impacts, a model of the economic behavior associated
with the service flow must be developed. Economic
models are specific to the service flow and the
consuming population, and not all service flows have
adequate economic models that describe their value.
For example, recreational fishing models account for
the preferences and geographic distribution of anglers
as well as the site characteristics of the fisheries.
These data are site specific, making the model
specification fairly non-transferable.
Table E-6 describes the extent to which models
are available to estimate the full chain of defensible
links for the ecological endpoints identified in Table
E-5. Each column must have an identified model in
order to complete the required modeling steps to
quantify changes in that endpoint. In cases where
defensible links are not quantified, opportunities exist
for qualitative analysis.
Table E-7 summarizes the quantitative and
qualitative analyses that we propose based on the
available model coverage. Geographic scope plays an
important role in determining the level of analysis,
such as a national assessment, a case study or a
qualitative description that is possible given existing
models. This exhibit demonstrates that, of the great
number of known impacts of air pollution, only a
subset can be assessed. In the next section we discuss
the methods, results, and caveats of the analyses of
these selected endpoints.
Table E-7
Summary of Endpoints Selected for Quantitative Analysis
Endpoint
Analysis
Geographic Scope
Lake acidification impacts on
recreational fisheries
Quantification of improved fisheries
with monetization of recreational
value
Case study of New York State
Estuarine eutrophication impacts on
recreational and commercial fisheries
Quantification of improved fisheries
with monetization of avoided costs
of alternative eutrophication control
methods
Illustrative calculations for case
studies of Chesapeake Bay,
Long Island Sound, and Tampa
Bay (with extensions to East
Coast estuaries)
Ozone impacts on commercial timber
sales
Quantification of improved timber
growth with monetization of
commercial timber revenues
National assessment
Ozone impacts on carbon sequestration Quantification
in commercial timber sequestration
of improved carbon National assessment
Toxicity impacts on recreational fishing
Qualitative analysis of improved
recreational fisheries
Qualitative regional case studies
of New York and Tennessee
E-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Extending Future Analyses
By focusing on the readily measured impacts
identified in Table E-7, it is possible to lose sight of
ecosystem-level changes that may threaten ecosystem
integrity in ways that alter or increase the risk of
changing ecosystem structure and function. The
isolation of service flows may often imply an
oversimplified cause and effect relationship between
pollution and the provision of the service flow, when
more often the service flow is affected by complex
non-linear relationships that govern ecosystem
structure and function. Economic analyses that focus
on a narrow class of acute service-flow impacts will
not cover larger ecosystem-wide impacts that may
ultimately prove most relevant to environmental
policy decision making. This analytical weakness
becomes apparent when impacts to ecological
functions such as nutrient cycling and biological
diversity are assessed.
Issues on which to focus future analytic work in
this field include:
• Major linkages of cause and effect between
air pollution and subtle deterioration in
ecosystem integrity are difficult to quantify;
• Degradation of ecosystem integrity most
often does not cause immediate measurable
declines in ecosystem service flows that are
monetarily valued by society;
• The time-frame required for many ecological
impacts to manifest themselves is such that
the present value of these impacts discounts
to negligible sums; and,
• Uncertainties associated with the scale of
complex ecological impacts are too great to
allow for reliable estimation of the economic
implications.
Because of the weaknesses in the available
methods and data, the benefits assessment in this
appendix does not represent a comprehensive estimate
of the economic benefits of the CAAA. Moreover,
the potential magnitude of long-term economic
impacts of ecological damages mitigated by the CAAA
suggests great care must be taken to consider those
ecosystem impacts that are not quantified here.
Significant future analytical work must be performed
to build a sufficient base of knowledge and data to
allow the expansion of this benefits assessment.
Eutrophication of Estuaries
This analysis considers the economic benefits of
reduced nitrogen deposition and the effects on
selected eastern estuaries attributable to the 1990
Clean Air Act Amendments (CAAA). Note that these
estimates were not included in the primary benefits of
the CAAA; these are presented here as an alternative
calculation only. We present a description of how
nitrogen deposition affects estuarine ecosystems, an
explanation of the effects on ecological service flows,
and an assessment of the benefits of reducing
nitrogen deposition in the context of several case
studies using avoided-damage and displaced-cost
approaches as alternative estimates of benefits. A
more comprehensive description of this analysis is
found in Benefits Assessment of Decreased Nitrogen
Deposition to Estuaries in the United States Attributable to
the 1990 Clean Air Act Amendments, 1990-2010 (lEc,
1999a).
Impacts of Nitrogen Deposition on
Estuaries
Atmospherically derived nitrogen makes up a
sizable fraction of total nitrogen inputs in estuaries in
the eastern United States. When atmospheric nitrogen
enters estuaries it can cause eutrophication, or an
increased nutrient load that, in excess, changes the
ecosystem's structure and function and affects the
provision of ecological service flows. The ecological
effects and their associated service flows are listed in
Table E-8.
E-20
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-8
Service Flows Affected by Changes in Estuarine Ecosystems
Ecosystem Changes
Service Flow Impacts
Deterioration of breeding grounds for fisheries
Commercial fishing yields, species mix
Recreational fishing catch rate, species mix
Loss of habitat for aquatic and avian biota
Existence value of a healthy estuarine ecosystem
Wildlife viewing
Aesthetics
Derivation of dose-response relationships
between atmospheric nitrogen loading and ecological
effects is complicated by the dynamic nature of
ecological systems. In addition to being characterized
by non-linear, "threshold" type responses, esturine
ecosystems are simultaneously influenced by a variety
of stressors (both anthropogenic and
non-anthropogenic). This makes it difficult to
quantify the nature and magnitude of ecological
changes expected to result from a change in a single
stressor such as nutrient loading. Further, if the state
of the ecosystem has changed (as from oligotrophic6
to eutrophic) the removal of the initial stressor does
not necessarily mean a rapid return to the prior state.
This complicates the quantitative benefits assessment
of controlling nitrogen deposition through the CAAA.
Economic Benefits of Decreasing
Atmospheric Deposition of Nitrogen
EPA's analysis begins with a geographic
information system (GIS) approach to estimate the
total volume of nitrogen inputs that the CAAA would
reduce to three major estuaries, the Chesapeake Bay,
Long Island Sound, and Tampa Bay. Unfortunately,
resource limitations prevented us from examining
more than three estuaries at this point. The three
estuaries were chosen for several reasons. First, each
of these areas maintains an active research center,
either under Clean-Water Act or National Estuary
Program provisions, and has identified airborne
6Oligotrophy refers to a state of relatively low nutrient
enrichment and low productivity of aquatic ecosystems. In
contrast, eutrophy refers to a state of relatively high nutrient
loading and higher productivity, sometimes leading to
overenrichment and reduction in ecological service flows via water
quality decline.
nitrogen as a major factor in its efforts to limit
eutrophication. Second, each has conducted some
research into the level of damages associated with
nitrogen eutrophication, although in the final analysis
only the Chesapeake Bay Program's research was
sufficient to establish some measure of avoided
damages. Third, each of these estuaries has in place a
binding commitment to meet its nitrogen reduction
targets, a necessary condition for applying the avoided
costs approach. In these estuaries, failure to reduce
airborne nitrogen deposition (as would be the case if
no CAAA were in place) would imply that additional
nitrogen reductions would be necessary from other
sources, such as point or nonpoint surface water
discharges. Implementation of the CAAA therefore
effectively avoids the imposition of costs to achieve
those nitrogen discharge reductions.
For each of the three estuaries selected, we then
conduct two types of analyses. First, we assess the
avoided nitrogen deposition loadings to the watershed
and avoided costs from reducing nitrogen deposition,
using submerged aquatic vegetation as a key
biophysical indicator. Second, we estimate the
avoided cost of implementing planned alternatives to
the CAAA for reducing nitrogen deposition.
GIS-Based Deposition and Loadings
Estimates
The first step toward calculating
deposition-related nitrogen loadings to the three
estuaries is to estimate the total deposition of nitrogen
to the watersheds. Table E-9 presents our estimates
of the quantity of nitrogen deposited to the
Chesapeake Bay, Long Island Sound, and Tampa Bay
E-21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-9
Total Nitrogen Deposition Based on GIS Analysis
(millions of Ibs.)
Watershed
Chesapeake Bay
Long Island Sound
Tampa Bay
Scenario 1: 1990
345.1
78.3
8.1
Scenario 2:
2010 without CAAA
452.4
93.7
11.3
Scenario 3:
2010 with CAAA
258.1
56.8
7.0
Difference
194.3
36.9
4.3
watersheds. We present data for three different
scenarios. The first scenario is our estimate of the
quantity of nitrogen deposited in 1990, prior to the
introduction of the CAAA. Scenario 2 is our estimate
of the quantity of nitrogen deposited in 2010 without
the CAAA, and Scenario 3 is the quantity deposited
with the CAAA. The difference between Scenarios 2
and 3 represents the potential future impacts of the
CAAA on nitrogen deposition.7
As the exhibit indicates, the CAAA are likely to
have a significant impact on the quantity of nitrogen
deposited to each of the three watersheds. For the
Chesapeake Bay watershed, nitrogen deposition is
expected to be nearly 195 million pounds less in 2010
(43 percent) than it would have been without the
CAAA. For the Long Island Sound and Tampa Bay
watersheds, this figure is approximately 37 million
pounds (39 percent) and four million pounds (38
percent), respectively.
We also estimate the prevalence of major
categories of land use in each of the three watersheds.
Land use is a critical component of our analysis
because the quantity of nitrogen runoff that eventually
reaches the estuary varies according to the type of land
that receives the atmospheric deposition. For
example, the fate of atmosphericly deposited nitrogen
will differ if the nitrogen falls on forest versus urban
land, because forest land generally retains a greater
percentage of nitrogen than urban land. Our analysis
uses distinct nitrogen "pass-through" figures for each
category of land use. Pass-through represents the
percentage of atmospherically deposited nitrogen that
is ultimately transported to surface water rather than
retained by the land.
Table E-10 presents the land use prevalence and
the pass-through factors that we use for each of the
three watersheds in our analysis.8 As the exhibit
indicates, forests (53 percent) and agricultural lands
(32 percent) represent the majority of the land use in
the Chesapeake Bay watershed. In the Long Island
Sound watershed, forests (67 percent) again dominate
land use; however, urban lands account for as much
territory as agricultural lands (11 percent). For the
Tampa Bay watershed, agricultural lands constitute the
largest land use (33 percent), while rangelands (19
percent), urban land (15 percent), and wetlands (10
percent) represent a much greater proportion of the
land use than in the other two watersheds.
These data are derived from lEc's spatial analysis of the
watersheds and RADM nitrogen deposition modeling. The
RADM modeling is described in Appendix C of this report.
8 Pass-through estimates were derived from EPA's analysis of
the relevant literature - see lEc (1999).
E-22
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-10
Land Use Prevalence and Pass-Through Figures
Watershed
Forest
Agricultural
Urban
Water
Wetlands
Other*
Chesapeake Bay
Land Use
N Pass-Through
53%
20%
32%
30%
6%
50%
7%
100%
1%
20%
1%
30%
Long Island Sound
Land Use
N Pass-Through
67%
20%
11%
30%
11%
50%
9%
100%
2%
20%
0%
30%
Tampa Bay
Land Use
N Pass-Through
* "Other" areas in Tampa Bay
5%
20%
33%
30%
include rangeland (19%) and barren
15%
50%
land (3%).
15%
100%
10%
20%
22%
30%
We use the pass-through figures and land use
prevalence in conjunction with deposition quantities
to estimate nitrogen loadings to each estuary. Table
E-ll displays our nitrogen loadings estimates for the
three watersheds under the three scenarios. As the
exhibit indicates, loadings from atmospheric
deposition decrease significantly due to the CAAA.
For Chesapeake Bay, for example, we estimate that
nitrogen loadings with the CAAA will be
approximately 79 million pounds in 2010,
approximately 58 million pounds less than our
estimate for loadings in 2010 without implementation
of the CAAA. For the Long Island Sound and Tampa
Bay, the difference between the two scenarios is
approximately 13 million pounds and 1.8 million
pounds, respectively.
Table E-11
Nitrogen Loadings from Atmospheric Deposition
(millions of Ibs.)
Watershed
Chesapeake Bay
Long Island Sound
Tampa Bay
Scenario 1:
1990
105.2
26.7
3.4
Scenario 2:
2010 without CAAA
137.5
31.9
4.7
Scenarios: 2010
W/CAAA
79.4
19.1
2.9
Difference
58.1
12.8
1.8
Displaced Costs from Reducing
Atmospheric Deposition to Estuaries
It is possible to use a displaced cost approach to
determine the benefits associated with reduced
nitrogen emissions. To reduce excess nutrient loads
(including nitrogen) to local estuaries, many coastal
communities are pursuing costly abatement options.
These options include point source controls as well as
urban non-point and agricultural non-point source
controls. We estimate the marginal costs of
abatement associated with these controls as
implemented in the three case study estuaries. To the
extent that nitrogen deposition can be controlled
more cost effectively than point-source discharges, the
control expenditures displaced by the CAAA
represent a benefit to society.
Ideally, a nitrogen management program would
result in the least expensive abatement possible,
E-23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
thereby minimizing the resources society expends on
nitrogen control. The lowest marginal cost pollution
reduction is exploited first, and pursued to its limit
before the next least costly alternative is exploited, and
so on until the required nitrogen reduction is met
(represented by the dark columns in Figure E-l).
With the CAAA, a portion of the resources
society committed to or would have committed to
reducing a quantity of waterborne nitrogen may be
unnecessary. Following the cost minimizing strategy,
society will forego the most expensive control cost
option first, pursue it to its limit before the next most
expensive alternative is foregone, and so on until the
nitrogen reduction benefits from the CAAA are
exhausted (represented by the lightly shaded columns
in Figure E-2). The level of nitrogen reduction
remaining will therefore be accomplished at the lowest
cost (represented by the dark columns in Figure E-2).
E-24
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
10 -i
S/t/y
8 -
7 -
'•8 (>]
Marginal Cos
Figure E-l. Nitrogen Reduction Without CAAA
Required reductions from waterborne sources
/-
^
V —
\
Target (allowed)
N levels at 20 10
Projected N levels
at 2010
N Level (t/yr)
10
9
uction
s
u
•a
•a
c3
S
3 -
2 -
Figure E-2. Nitrogen Reduction With CAAA
Reductions from CAAA
Required reductions
from waterborne
t
Target (allowed)
N levels at 2010
Projected N levels
at 2010
N Level (t/yr)
E-25
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
We develop our avoided-cost estimate by
assuming that decision makers will choose to forego
the most costly nitrogen abatement projects first.
That is, we assume that reduced deposition and the
resulting loadings reduction will eliminate the need for
additional point or non-point source controls at the
high end of the marginal cost curve.
To estimate the economic benefits of reduced
nitrogen deposition from the CAAA, we require site
specific information from the watershed level. A
justifiable avoided cost analysis relies upon the
existence of realistic and enforceable nitrogen
reduction goals for each estuary. Without specific
targets or reduction goals, it is not possible to suggest
that there are any control costs to be avoided. As
described earlier, we have chosen case study estuaries
that fit this criterion. These areas have established
nitrogen reduction programs that rely primarily on
reductions of effluent from point sources as well as
reductions in non-point source discharges.
Information on the reduction goal and potential
abatement options for meeting those goals allow us to
estimate the portion of the goal that can be met by the
CAAA, as well as the associated cost savings.9 We
summarize those results in Table E-12.
Next, we need to know the annual quantity of
atmospheric nitrogen deposited on the watershed.
Last, we need to understand details about the different
nitrogen reduction programs that could be
implemented in the watershed. This includes the
quantity of nitrogen reduced through a particular
control option (e.g., agriculture best management
plans[BMPs]), and the unit cost of reducing that
nitrogen (i.e. dollars per pound or ton of nitrogen
reduced).
The benefits valuation derived using the avoid-
costs approach should be interpreted cautiously for
two reasons. First, it is an estimation of capital costs
'With increasing populations, controls of alternative sources
(e.g., automobile and utility emissions) may be needed simply to
meet the original target or goal; if the CAAA amendments are
necessary just to achieve the target reductions, then we are actually
measuring alternative costs and not displaced costs.
that serve more purposes than mitigating nitrogen
inputs into the estuaries of concern. Water treatment
works are intended to provide waste water treatment
for a variety of pollutants and may be required even in
the absence of air deposition of nitrogen. Second, the
nitrogen loading targets for the estuaries are not
concrete, strictly enforced limits, based on certain
knowledge of the capacity of the estuaries to accept
nitrogen inputs. Instead, the targets may change over
time as knowledge of the effects of nitrogen to these
estuaries change. For these reasons, we do not
include these estimates in the primary benefits
estimates for the CAAA.
E-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableE-12
Estimated Avoided Costs For Three Estuaries
Estuary Reduced N Deposition in
2010(millions of pounds)
Long Island
Sound
Chesapeake
Bay
Tampa Bay
12.8
58.1
1.8
Lower-BoundMarginal Upper-Bound
Cost($/lb/yr.) Marginal Cost
($/lb/yr.)
$2
$6
$6
$8
$22
$38
Estimated Annual
Avoided Costs in 2010
($millions)
$25.6-$102.4
$349-$1,278
$11 -$68
Results for Case Study Estuaries
• Under the Chesapeake Bay Agreement, the
signatories (EPA, Maryland, Virginia,
Pennsylvania, and the District of Columbia)
have agreed to reduce nutrient loadings to the
Bay by 40 percent by the year 2000, relative to
1985 levels. This goal translates to a nitrogen
reduction of about 186 million pounds per
year. A great deal of progress toward this
goal has already been made, although success
differs across sub-regions in the watershed
(Chesapeake Bay Program 1997). Nitrogen
loadings reductions achieved thus far are the
result of both point and non-point source
controls. Since 1985, 33 of the 315 major
municipal treatment plants in the region have
upgraded to biological nutrient removal
(BNR), an advanced treatment technology
specifically focused on nitrogen removal.
These upgrades have reduced annual loadings
by about 13 million pounds (as of 1996).
Approximately 60 additional facilities are
expected to implement BNR in the future. In
addition, agricultural and urban best
management practices have reduced
non-point source loadings by about 16
million pounds per year, with additional
implementation of BMPs planned for coming
years (Chesapeake Bay Program, 1997).
Because both point and non-point source
controls play a role in anticipated future
nitrogen reductions in Chesapeake Bay, we
develop marginal and avoided cost estimates
that incorporate both types of control. As
reflected in Exhibit 15, we estimate avoided
cost benefits for Chesapeake Bay ranging
from about $349 million to $1.3 billion.
Long Island Sound has established a goal of
reducing nitrogen loadings by approximately
48 million pounds by 2015. Point source
controls are anticipated to be the primary
source of these reductions. Numerous
sewage treatment plant upgrades are slated
for the region, many of which are currently
under construction. We use data from the
Connecticut Department of Environmental
Protection and New York Department of
Environmental Conservation. The marginal
cost figures yield an estimate of avoided costs
that ranges from about $26 to $102 million
per year.
In 1996, the Tampa Bay Estuary Program
(TBEP) adopted a five-year nutrient
management goal that caps annual nitrogen
loadings at 1992-1994 levels. Nitrogen
loading to Tampa Bay is expected to increase
seven percent by the year 2010 as a result of
population growth and related commercial
and residential development. To offset this
growth and maintain current nitrogen levels,
the TBEP has asked local governments,
agencies, and industries to reduce total
nitrogen loadings to the Bay by
approximately 84 tons (168,000 pounds) per
year by the year 2000. The result of this
planning effort is the Nitrogen Management
Action Plan (TBEP, 1998). This plan lists
the projects undertaken or planned by
industry, local governments, and agencies
E-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
that will contribute to meeting the nitrogen
management goal for 2000 and beyond.
These projects, which together surpass the
nitrogen reduction goal for Tampa Bay, are a
combination of point and non-point source
control measures. The non-point source
control projects include urban stormwater
retention ponds, wetlands restoration, and
land acquisition. The point source projects
focus on advanced treatment technologies
such as BNR at (publically owned treatment
works (POTWs). For example, one project,
proposed for implementation after the year
2000, will involve additional treatment of
effluent from a POTW prior to reuse in the
regional water supply. We estimate annual
avoided costs for Tampa Bay ranging from
about $11 million to $68 million.
These three estuaries represent only a portion of
the total estuarine area affected by nitrogen deposition
in the United States. The Chesapeake Bay and Long
Island Sound account for roughly 20 to 25 percent of
the East Coast estuarine watershed area addressed by
the National Estuary Program, and Tampa Bay is a
small fraction of the total Gulf Coast estuarine
watershed area. As a result, our estimates reflect only
a partial analysis of the national impact of nitrogen
deposition.
Results for Total East Coast
Estuarine Area
To extrapolate the results of this analysis to all
East Coast estuaries, we assume all estuaries along the
Atlantic Coast have binding nitrogen budgets. We
then use the same geographic information system
(GIS) approach we used in our analysis of Chesapeake
Bay, Long Island Sound, and Tampa Bay to estimate
total nitrogen deposition and the associated loadings
to estuaries located along the Atlantic Coast. This
approach allows us to estimate nitrogen loadings in
the year 2010 with and without CAAA emissions
controls. Since watershed specific nitrogen control
program information is not available for each
watershed, we extrapolate BMP cost and nitrogen
reduction data from the three case studies across all
East Coast watersheds.10
Although we simplistically assume that each
estuary has a nitrogen budget, the total East Coast
displaced cost analysis does not include all estuaries
along the Atlantic Coast, since some estuaries are not
sensitive to nitrogen loadings. Certain estuaries are
able to process large amounts of nutrients (nitrogen
and phosphorus) without problems, while others are
unable to process even low amounts of nutrients.
Rather than rely on nitrogen levels to determine which
estuaries to include in the displaced cost analysis, we
use a measure of eutrophic susceptibility developed by
Bncker et al. (1999-DRAFT), to determine how
sensitive estuaries are to nitrogen loadings. Using
their eutrophic susceptibility classification (low,
medium, and high) we then exclude estuaries with low
eutrophic susceptibility from the displaced cost
analysis.11
The total displaced costs for the East Coast is
simply the sum of the displaced costs for each estuary
along the Atlantic Coast classified as moderately or
highly susceptible to eutrophication (see Table E-
13).12 The lower-bound estimate of $261.5 million
represents a point source control strategy. The upper-
bound estimate of $2,766 million represents a strategy
The control cost and nitrogen reduction data is available in the
lEc memorandum to EPA dated August 26, 1999.
11 This index is based on a variety of factors influencing
sensitivity of estuaries to eutrophication, including water surface
area of the estuary, estuary volume, freshwater inflow, tidal
cycles, and vertical stratification.
12 In the North Atlantic Region, Blue Hill Bay, Casco Bay,
Englishman Bay, Kennebec/Androscoggin River, Merrimack
River, Musconguc Bay, Narraguagus Bay, Penobscot Bay, and
Saco Bay have a low susceptibility to eutrophication and are
excluded from the displaced cost analysis. All estuaries in the
Mid Atlantic Region are moderately to highly sensitive to
eutrophication. In the South Atlantic Region, New River,
Ossabaw Sound, Savannah River, and St. Helena Sound have a
low susceptibility to eutrophication and are not included in the
displaced cost analysis.
E-28
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableE-13
Avoided Cost for Atlantic Coast
Watershed
North Atlantic
Cape Cod Bay
Great Bay
Massachusetts Bay
Sheepscot Bay
Sub-Total
Mid Atlantic
Barnegat Bay
Buzzards Bay
Chincoteaque Bay
Delaware Bay
Delaware Inland Bays1
Gardiners Bay
Great South Bay
Hudson River/Raritan Bay
Narragansett Bay
New Jersey Inland Bays
Sub-Total2
South Atlantic
Albemarle Sound
Altamaha River
Biscayne Bay3
Bogue Sound
Broad River3
Cape Fear River
Charleston Harbor
Indian River
North/South Santee Rivers
Pamlico Sounds
St. Andrew/St. Simons Sounds3
St. Catherines/Sapelo Sounds
St. Johns River
St. Marys River/Cumberland Sound
Winyah Bay1
Sub-Total
Total East Coast
1 CAAA N reductions met with agriculture,
Reduction through
CAAA, Millions of
Pounds (2010)
0.51
0.57
1.19
0.13
13.02
0.48
0.78
0.33
12.11
0.18
0.64
1.30
11.15
1.65
1.12
29.74
16.29
7.24
0.78
0.30
0.46
6.59
10.58
0.76
0.32
9.58
1.18
0.28
4.72
0.53
11.63
81.07
123.82
forestry, and urban BMPs
Avoided Cost, 100
Percent from Point
Source ($1000/yr.)
1,282
1,446
3,015
340
6,083
1,202
1,976
824
30,640
459
1,624
3,294
28,197
4,176
2,825
75,219
41,223
18,315
1,973
756
1,160
16,674
26,764
1,926
821
24,225
2,974
721
11,940
1,338
29,415
180,225
261,527
(point source reductions not
Avoided Cost, from
Nonpoint Source First,
then Point Source on
Difference ($1000/yr.)
7,915
19,130
66,887
2,397
96,329
19,688
14,429
3,690
384,002
4,592
13,929
78,682
460,659
65,841
28,314
1,073,826
272,772
182,591
12,136
9,426
7,137
171,805
306,698
36,045
3,671
191,289
18,297
6,525
152,909
8,886
215,612
1,595,799
2,765,954
required).
2 Excluding Long Island Sound and Chesapeake Bay.
3 CAAA N reductions met with agriculture
BMPs (further reductions not required).
E-29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
of nonpoint source BMPs, with further nitrogen
reductions from point sources, if required.13
This estimate is based on an extrapolation of
nitrogen abatement costs from representative
watersheds. Although this is a broad assumption, it
does provide a gross estimation of the range and
magnitude of the CAAA benefits for the Atlantic
Coast as a whole and its component watersheds.
Due to our general assumptions, a high degree of
uncertainty is associated with this range. First, we do
not know enough about the nature of the nitrogen
budgets for each estuary and if those budgets would
be binding. If nitrogen budgets are not binding, these
regions may have little incentive to reduce nitrogen
loadings. Furthermore, because of the lack of
watershed specific cost information, we rely on
available abatement cost data from the Chesapeake
Bay and Long Island Sound to represent point source
and nonpoint source unit abatement costs for all
watersheds along the Atlantic Coast. The tightness of
the range and accuracy of the displaced cost analysis is
dependant on an accurate understanding and
representation of the nitrogen abatement costs
associated with the different point source control
options and nonpoint source BMPs in each individual
watershed. Lastly, we use marginal cost figures that
represent averages of the costs associated with certain
control measures. For example, our agriculture BMP
cost figure is a simple average of six different
agriculture BMPs, each with a different level of
nitrogen reduction and a different cost per pound of
nitrogen reduced. While beyond the scope of this
effort, a more refined analysis would use marginal cost
estimates based on the precise mix of agriculture
BMPs to be implemented in each watershed and the
cost of each, as determined by local factors within the
watershed.
13 In our analysis of the total displaced costs for the
Atlantic Coast, several watersheds meet their CAAA nitrogen
reduction levels without relying on point source controls.
Avoided Damages to Estuarine
Ecosystems
Theoretically, a modeling system that describes
water quality changes, including fish population
dynamics as a function of nitrogen input, would
provide an assessment of the avoided damages from
mitigating nitrogen deposition (Figure E-3). Because
of current modeling and data constraints, however,
the only means to quantify the damages of
eutrophication from nitrogen deposition is through
the use of specific biophysical indicators of estuarine
health. Changes in an indicator, such as aerial extent
of seagrass beds, can measure habitat damage. Based
on changes in habitat, the change in ecosystem service
flows associated with that habitat can be estimated.
From an economic perspective, this approach is
useful in cases where habitat is closely related to the
provision of ecological service flows, such as
commercial and recreational fishing yields. Using
seagrass beds as an indicator, we describe the potential
for avoiding estuarine damages through the CAAA.
Submerged aquatic vegetation (SAV) is a group of
angiosperms (plants that bear seeds, as opposed to
algae, which reproduce via cell division) that form
extensive meadows, providing habitat, breeding
grounds, and nursery for a variety of organisms
including fish, birds, shellfish, and invertebrates
(Jacobs et al. 1981; Bell et al. 1989; Howard et al.
1989; Burkholder et al. 1992; Orth et al. 1994). SAV
meadows give considerable three-dimensional
structure to the seabed that provides small organisms
with a place to hide from predators, acts as a sediment
trap, and functions as a breakwater offering natural
shoreline protection (Vermaat et al. 1998). Along with
aerial extent, the density of SAV may be important in
defining the health of the SAV community.
Though a universal nitrogen-SAV relationship has
not been derived, field data show that increased
nitrogen loading has been accompanied by extensive
decline in SAV in a variety of estuaries (Valiela et al.
1997; Burkholder et al., 1992; Coastlines, 1994;
Vermaat, 1998). In the Chesapeake, SAV acreage
declined from more than 76,000 to about 40,000 acres
E-30
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure E-3
Estuary Models and Ecological Impacts of Concern
RADM
Wateished
Model
>•
Water Quality Model
Dissolved
Oxygen
Submodel
SAV
Submodel
Changes in
Dissolved
Oxygen
Fishing Benefit
ofCAA
between 1870 and 1950, concurrent with increased
nutrient loading (Coastlines, 1994). From 1950 to
1980, when nutrient loading took a sharp upward turn,
the decline continued to a low point of about 21,000
acres. This decline is quite likely due to shading by
excessive algal growth and increased growth of
epiphytic plants that also respond positively to
increased nitrogen availability (Coastlines, 1994).
Empirical evidence from Tampa Bay and the
Chesapeake Bay show that SAV populations will likely
benefit from reduced nitrogen inputs to these
estuaries. In Tampa Bay, there is a notable correlation
between SAV and nitrogen inputs, as described in
Figure E-4.14
In Chesapeake Bay, it is not possible to identify a
statistically significant relationship between these two
variables in isolation because of the large and
heterogenous nature of the Chesapeake Bay.
Embayments vary greatly in physical characteristics
including depth and salinity, complicating the
relationship. Nonetheless, the general trend has been
an increase in seagrass acreage and a decrease in
nitrogen loading. Figure E-15 shows the change in
SAV acreage over the past two decades.
Because the relationships we describe from
existing data are not sufficiently robust to allow
projections of SAV coverage as a function of nitrogen
deposition in our scenarios, we utilize alternative
methods to provide quantitative benefits estimates in
the following section.
14 Two factor analysis of variance (ANOVA) analysis
confirms this correlation; see EPA 1999 for more details.
E-31
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure E-4
Nitrogen Load vs. Seagrass Acreage in Tampa Bay
500 1000 1500
Annual Nitrogen Load (tons/yr)
2000
20000
Figure E-5
Chesapeake Bay SAV
1978-1996
D Upper Bay
• Middle Bay
D Lower Bay
Year
E-32
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Caveats and Uncertainties
Though it is difficult to directly predict the nature
and magnitude of the ecosystem impacts of
eutrophication, we conclude that continued nitrogen
inputs at current levels will result in deleterious effects.
The major caveats and uncertainties associated with
these analyses follow.
• Our nitrogen loading estimates are derived
using a highly simplified approach that takes
into account total deposition and the nitrogen
retention characteristics of different land
uses. Some factors suggest that we may
overstate loadings because we do not
consider the effects of nitrogen travel
through varied distances and heterogeneous
geography, such as rivers and streams. For
example, an additional 20 to 75 percent of
nitrogen is retained during transport in rivers
and streams (Hinga, et al., 1991).
Other factors suggest that we may understate
loadings. Most significantly, the Geographic
Information Retrieval Analysis System
(GIRAS) land use data in our GIS analysis
were compiled in the early 1980s. It is likely
that the current amount of urbanized land is
greater than these data indicate.
Furthermore, continued development of
forest and other open land suggests that land
uses will change significantly in the period
between now and 2010. Because nitrogen
removal in urban land is low, more refined
land use data for future years would likely
lead to greater estimates of nitrogen loadings.
We compare our nitrogen loading estimates
with those of estuary programs and published
literature for 1990 to verify whether our
approach generates results that correspond
with existing estimates. For the Chesapeake
our estimate of 150 million tons falls in the
middle of existing estimates ranging from 58
to 159 million tons (EPA 1997e; Patwardhan
and Domgian 1997; Fisher et al. 1998; Tyler
1998) for Long Island Sound our estimate of
27 million tons corresponds with the existing
estimate of 26 million tons (Stacey 1998), and
for Tampa Bay our estimate of 3 millions
tons is close to the existing estimate of 2
million tons (TBEP 1998; Zarbuck et al.
1994).
• We base our estimates of avoided costs on
simplified assumptions regarding the control
measures that would be eliminated as a result
of reducing atmospheric nitrogen. The mix
of nitrogen controls that could be displaced
will be influenced by state regulations
affecting treatment plants and non-point
sources as well as by pollution reduction
goals for different sub-basins in each
watershed. For example, water quality
objectives for pollutants other than nitrogen
may require controls that we assume could be
eliminated.
• Similarly, we use marginal cost figures that
represent averages of the costs associated
with control measures. For example, we
apply generic non-point source control cost
estimates based on a mix of agriculture,
forestry, and urban best management
practices. While beyond the scope of this
effort, a more refined analysis would use
marginal cost estimates based on the precise
mix of best management practices to be
implemented in each watershed and the cost
of each, as determined by local factors such
as levels of nitrogen in soils, evolving
agricultural practices, and changing
development patterns.
Acidification of Freshwater
Fisheries
During the 1970s and 1980s, "acid rain" came to
be known to the public as a phenomenon that injures
trees, forests, and water bodies throughout Europe
and in some areas of the United States and Canada.
One of the goals of the CAAA was to address the
problem of acidification of terrestrial and aquatic
E-33
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
ecosystems caused by acidic deposition. In this
section we evaluate economic benefits accruing to
society as a consequence of reductions in emissions of
sulfur and nitrogen oxides mandated by the CAAA.
In particular, we focus on a quantitative analysis of
benefits derived from a reduction in acidification of
aquatic bodies as they relate to recreational fishing in
the Adirondacks region of New York State. Our
analysis indicates that by mitigating acidification with
the regulations promulgated under the CAAA,
cumulative benefits between 1990 and 2010 can be
accrued in the range of $67 to $465 million. Using the
results from the acidification estimates, it is also plain
to see that the CAAA may be preventing further
ecological impacts from acidification that are not
quantifiable in economic terms using available
methods. A more comprehensive description of this
analysis is found in Economic Benefits Assessment of
Decreased Acidification of Fresh Water Lakes and Streams in
the United States Attributable to the 1990 Clean Air Act
Amendments, 1990-2010 (lEc, 1999b).
Acidification of Surface Waters and
Ecological Impacts
Acidification of surface waters is frequently
described using two measures. One measure of
acidification is pH, which is based on the hydrogen
ion (H+) concentration found in surface waters.15 The
pH of a water sample can range from 1 to 14 on a
logarithmic scale, with pure water having a pH of
seven. The term acidic usually refers to a pH below
seven, indicating high concentrations of hydrogen ions
(H+). Rain water that is unaffected by anthropogenic
factors (natural rain) is weakly acidic (pH 5.0 - 6.0),
due to the presence of natural weak acids. With
addition of acids from human activities, however, the
pH of rain can range from 3.5 to 5.0 (NAPAP, 1991,
p. 15). Most freshwater lakes and streams have a pH
between 6.5 and 8, indicating that surface waters can
be naturally acidic. Only a small percentage of aquatic
ecosystems are naturally acidic with pHs below 6.5,
and concerns about anthropogenic acidification focus
15The pH of a water sample is equal to the negative log of its
hydrogen ion concentration: pH= 4og[H+].
on the effects that may occur with decreases in pH
below pH 6.5 (EPA, 1995ap. 9).
The second commonly used measure of
acidification is Acid Neutralizing Capacity (ANC),
which describes a water body's ability to neutralize
acids added to the water column. Surface waters with
higher ANC are generally more resistant to
acidification and empirically tend to have higher pH
levels. ANC is measured in micro-equivalents per liter
(ueq/L). Surface waters with an ANC of less than 200
ueq/L are considered to have a low capacity for
neutralizing acids. Water bodies with an ANC of 50
ueq/L or less have a very low capacity for neutralizing
acids, and water bodies with an ANC of 0 ueq/L or
less have no ability to neutralize acids and are acidic.
These water bodies have no ability to neutralize acids
and tend to be the most sensitive for long-term pH
depressions below 6.0, which can produce the most
severe effects on aquatic life (EPA, 1995a p. 9,
NAPAP 1991 p.15).
Acidic deposition can lead to two kinds of
acidification processes, depending on the duration of
acidifying events. First, chronic acidification describes
a situation in which acidic deposition leads to
long-term changes in soil and water characteristics,
causing chronically toxic environmental effects.
Second, episodic acidification is a phenomenon in
which surface waters experience short-term (hours to
weeks) decreases in pH, usually during extreme
hydrological events such as storm discharge or
snowmelt (EPA, 1995a, p.9, NAPAP, 1991 p. 18). For
acid-sensitive fish species in some lakes or streams,
for example, episodic events can cause complete
spawning or recruitment failures (EPA, 1995a p.10).
The most comprehensive survey of surface waters
comes from the National Surface Waters Survey
(NSWS), which was conducted as part of the National
Acid Precipitation Program (NAPAP). Based on the
results from the NSWS surveys of lakes throughout
the United States, an estimated 4.2 percent (1,180) of
the NSWS lakes were acidic, defined as having ANC
less than 0 ueq/1 with pH levels in the range of 5.0 to
5.5. Nearly all of these lakes were in eastern portions
of the United States, located in six "high-interest
E-34
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
regions". The six areas identified are New England,
the Adirondacks, the mid-Atlantic Coastal Plain, the
mid-Atlantic Highlands, Florida, and the upper
Midwest. The NSWS found that acidic deposition is
the dominant source of acid anions in about 75
percent of the acidic lakes and 50 percent of the acidic
streams in their sample. Figure E-6 depicts the
geographic ranges of known acidification.
The effects of acidity on aquatic organisms are
determined by a number of different water quality
variables, the most important of which are pH,
inorganic aluminum, and calcium. The combined
effects of these variables can adversely affect the
physiology of individual organisms as well as
population-level parameters. The direct effects may
be classified as involving either recruitment failure or
reduced adult survival. The outcomes on these
impacts are declining acid-sensitive fish populations
and a consequent decline in species richness (SOS/T
13p.l3-126).
Modeling Acidification
Figure E-7 shows the stages of modeling the
ecological and economic impacts of acidification. This
analysis uses RADM deposition data for the year 2010,
and an extended deposition scenario that we develop
for the year 2040, to demonstrate the possible lagged
effects of the CAAA. We use these data in an
acidification model that generates an estimate of the
acidity of lakes in the Adirondacks. Lake acidity is
input to an economic model that estimates the costs
to anglers of diminishing lake water quality, and
consequently declining fish populations.
We use the same emissions and deposition data
for the period 1990 to 2010 as in each of our other
endpoint analyses. The primary difference is that we
extend the deposition scenarios to 2040 in order to
accommodate for the lagged physical effects of acid
deposition. This lag is a function of multiple
watershed and water body characteristics influencing
recovery from prolonged acidic deposition. Results of
various efforts to model freshwater acidification
showed that recovery of acidified water bodies can
take over 50 years, even with substantial (up to 70
percent) reductions in sulfate deposition (Jenkins et al.
1990; Wright et al. 1994), while watershed soils may
require 150-200 years for full recovery (Cosby et al.
1985).
Because appropriate data are lacking to simulate
the emission and deposition of acidic pollutants
between the years 2010 and 2040, we use two
deposition scenarios for the period 2010-2040:
• Constant deposition from 2010 to 2040 at
levels projected for 2010 under the
regulations of the CAAA; and,
• Constant deposition from 2010 to 2040 at
levels projected for 2010 without the
regulations of the CAAA.
E-35
-------
Figure E-6
Percentage of Acidic Surface Waters in the National Surface Water Survey Regions
West
New
England
Mid-Atlantic
Coastal Plain
Southec stern
Highla ids
10%-20%
>20%
Note: The National Surface Water Survey was conducted between 1984 and 1986.
Source: National Acid Precipitation Assessment Program (NAPAP). 1990 Integrated Assessment Report.
November 1991, p. 27.
E-57
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure E-7
Acidification of Freshwater Ecosystems
SO 2 and NO x
Emissions Model
Deposition Scenario I:
Full Implementation
of the 1990CAAA
MAGIC
Model
Scenario I:
% of Acidified Waterbodies
Economic
Model(s)
Scenario I:
Economic Impact
of Lake Acidification
Deposition Scenario II:
No Implementation
of the 1990CAAA
Scenario II:
% of Acidified Waterbodies
•
Critical pH
Threshold for Fish
Scenario II:
Economic Impact
of Lake Acidification
Scenario II - Scenario I =
Economic Benefit of CAAA
E-37
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Using these deposition scenarios, we estimate the
extent of acidification in the Adirondacks. To do this,
we use the Model of Acidification of Groundwater in
Catchments (MAGIC). MAGIC is a lumped
parameter model that was originally developed to
project the long-term effects (i.e., decades to
centuries) to surface water caused by acidic deposition
(Cosby et al, 1985a, 1985b). NAPAP and the Acid
Rain Program of EPA use MAGIC extensively in
analysis of acidification in the Eastern United States
(Church et al, 1989; Church et al, 1992; EPA, 1995),
the results of which have been rigorously
peer-reviewed and used in previous policy analyses.
The data that MAGIC produces describe the extent of
acidification (i.e., pH and ANC levels) that will occur
in a sample of sensitive lakes in the Adirondacks as a
function of acidic deposition levels.
As mentioned earlier, capacities of watersheds to
retain deposited sulfur or nitrogen containing
compounds are among the most important factors
influencing surface water acidification. Because
increasing stages of nitrogen saturation are likely to
lead to decreasing nitrogen-retention capacities, it is
necessary to consider these effects in our modeling
approach. MAGIC, however, currently does not
explicitly represent detailed cycling or processes
affecting the rate of nitrogen uptake and release
because processes that control the transition of a
watershed to a state of nitrogen saturation leading to
surface water acidification are not well understood
(Van Sickle and Church, 1995). Instead, we use a
sensitivity analysis of two boundary conditions,
representing a nitrogen-saturated watershed and a
watershed where terrestrial ecosystems continue to
utilize the majority of deposited nitrogen.
Our means to assess the economic impact of
acidification is to measure the change in social welfare
that results from reducing the quality of available lakes
for recreational fishing. In order to accomplish this,
we must determine the water chemistry parameters at
which declines in recreational fish populations are
perceptible by anglers. As mentioned earlier in this
memorandum, the toxicity of low levels of pH to fish
species depends on a variety of factors, including the
concentration of inorganic aluminum and calcium in
the water column as well as specific sensitivities of
locally occurring fish species. The level of pH is not a
precise indicator of the habitability of a lake for fish.
But because the full complement of relevant water
chemistry variables for lakes in the region is not
available, we are forced to work with pH as a proxy
for habitability. In order to accommodate for the
variable effects of pH due to other water chemistry
variables, we derive a range of pH values where
negative effects on fish species are empirically known
to occur. We summarize these results in Table E-14.
Table E-14
Summary of pH-Based Effects Threshold
pH Effects Threshold
(Low End)
pH Effects Threshold
(High End)
Range for All Fish Species
4.2
5.8
Range of Mean Values for All Species
4.8
5.3
Range of recreationally important
species (weighted average)
4.6
5.4
E-38
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Our review of the empirical effects literature
demonstrates the difficulty in discerning a single pH
threshold that could ever adequately characterize the
ability of a water body to sustain recreational fishing.
The most rigorous use of the available data might
employ species-specific thresholds and apply these
thresholds to individual lakes in the economic
modeling domain. Unfortunately, information on the
prevalent recreationally important species present in
the sample of lakes modeled by MAGIC, as well as in
the larger domain of lakes to which these results
would be extrapolated, is not currently available. We
therefore adopt a range of estimates of a pH threshold
for acidification of a lake of 5.0 to 5.4. The range is
consistent with a reasonable approximation of effects
noticeable to anglers. Knowledge of effects with
certainty would imply a more conservative assumption
of a lower pH, perhaps consistent with the low end
weighted average for recreationally important species,
reported in the last row of Table E-14. We therefore
report acidification results for an extreme low end
threshold estimate of pH 4.6, but do not interpret
those results as providing useful central estimates for
a study specifically concerned with recreational fishing.
The final step in our analysis is to use an
economic model that monetizes the impacts to
recreational fishing under each of the acidification
scenarios. This involves the selection of an economic
model that appropriately covers the effects of
acidification of multiple sites over the geographic area
that is impacted, and the proper integration of the
water quality information. The ideal for this
application is a random utility model (RUM) that
allows for the substitution among sites and fisheries as
water quality parameters change, an essential feature
when estimating recreational benefits.
Very few models of this type exist, and fewer
cover a region of high acidification that is impacted by
the CAAA. Efforts by Englin et al (1991), Mullen and
Menz (1985), and Morey and Shaw (1990) advance
this line of inquiry by relating regional acid deposition
to recreational fishing damages, but Montgomery and
Needelman (1997) are the first to use direct water
quality measures in conjunction with a random utility
model16. The estimation proceeds in three steps.
First, a site-choice model determines the impact of
water quality and other lake characteristics on the
choice of a fishing site among the set of all potential
sites. The model estimates the value of the available
set of lakes to each New York resident. In the second
step, a model predicts whether a New York resident
will choose to fish on a particular day. Third, based
on the results of the site-choice and fishing decision
models, it is possible to estimate the change in
economic welfare caused by altering water quality of
the lakes available to New York residents.
We use the Montgomery and Needelman model
by inputting MAGIC acidification estimates and
simulating the impact on anglers. We do not re-
estimate the econometric model's parameters for this
application due to resource constraints, though it is
important to note that the model was originally
estimated to describe angler response to acidity at pH
6.0, while we assume that anglers respond at a lower
pH level. Data from MAGIC are input to the model
in the form of a percentage estimate of the lakes in the
Adirondacks that fall below a chosen pH level. The
effect of acidification has a negative impact on the
utility of anglers that might wish to use that resource.
By simulating the effect on anglers' utility of acidifying
a percentage of lakes within a region, the model can
compute the economic impact of a specific level of
acidification. Subtracting the economic value of
fishing at our baseline level of acidification from that
which would occur if the CAAA were promulgated
provides an estimate of the benefits accrued to
recreational fishermen from reducing acidification in
the region.
Acidification Results
We summarize the results of our acidification
projections from MAGIC in Exhibit 19. Each
scenario described in these tables provides an estimate
of the percentage of lakes in the Adirondacks likely to
suffer from acidification given the deposition and
16The Montgomery and Needelman model applies a
technique developed by Morey, Rowe, and Watson (1993).
E-39
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
nitrogen saturation parameters assigned to that
scenario. The exhibits present the level of
acidification expected given a specific threshold level
of acidity (pH 4.6, 5.0, or 5.4), covering uncertainty
associated with the impacts of acidity on a range of
aquatic species.
Table E-15 shows that in the year 2010, the
CAAA can be expected to reduce the number of lakes
whose pH falls below 4.6 by zero percent, below 5.0
by one to four percent, and the percentage of lakes
falling below a pH of 5.4 by five percent. These
results are obtained by subtracting the acidification
estimates for the year 2010 without the CAAA from
the acidification estimate for 2010 with the CAAA.
Note that we only compare scenarios with consistent
nitrogen saturation parameters. One can not reliably
compare the impacts of acidification on a
nitrogen-saturated watershed in one deposition
scenario with a non-saturated watershed in another
deposition scenario.
Table E-15
Acidification Results - 2010
Year
1990
(Base Year)
2010
Status of CAAA Level of N
Saturation
No CAAA
Regulations
Promulgated
With CAAA No Saturation
Saturated
Without CAAA No Saturation
Saturated
Range of Benefits from CAAA in
2010
Percentage of Lakes Acidified at
Selected pH Levels
pH4.6
0%
0%
0%
0%
0%
0%
pHS.O
5%
2%
5%
6%
6%
1%-4o/o
pH5.4
20%
18%
17%
23%
22%
5%
Acid deposition between 1990 and 2010 also
contributes to lagged acidification impacts after 2010.
MAGIC estimates that a significant amount of
acidification between 2010 and 2040 is avoided by the
CAA. This is an area that requires further research in
order to fully quantify these impacts.
Economic Results
As acidification of the Adirondacks is reduced by
the CAAA, economic benefits accrue to society. In
annual terms, the economic benefits of the CAAA in
2010 are summarized in Table E-16. Following the
presentation used for the acidification data, the range
of annual benefits from the CAAA is $12 million to
$49 million using an effects threshold of pH 5.0, and
$82 to $88 million for an effects threshold of pH 5.4.
These results correspond to previous analyses (Englin
et al, 1991; Mullen and Menz, 1985; Morey and Shaw,
1990) that find annual benefits to the Adirondacks of
halving utility emissions in the millions to tens of
millions of dollars. We do not provide an economic
assessment of acidity in 2040, as the behavioral model
is not sufficiently robust to estimate economic
impacts 50 years into the future.
E-40
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableE-16
Annual Economic Impact of Acidification in 2010
(Millions of 1990 Dollars)
Year
1990
(Base Year)
2010
Range of CAAA
Status of
CAAA
No CAAA
Regulations
Promulgated
With CAAA
Without CAAA
Benefits in 2010
Level of N
Saturation
No Saturation
Saturated
No Saturation
Saturation
Economic Impact of Acidification
at Each pH Threshold
pH5.0
$61
$24
$61
$73
$73
$12-$49
pH5.4
$320
$281
$261
$363
$349
$82-$88
We calculate the cumulative economic benefits
from the CAAA by summing the difference between
the discounted annual economic impact of
acidification with and without the CAAA for every
year from 1990 to 2010.17 We perform this
calculation for the minimum and maximum values for
each of the three pH thresholds for survival of
recreational fish species, assuming a straight line
increase in the level of acidification between 1990 and
the 2010 level for each scenario. We present the
results as a cumulative net present value calculated
with 1990 as the base year (i.e. costs in 2010 are
discounted over 20 years). As indicated in Table E-17,
the range of cumulative potential benefits from the
CAAA between 1990 and 2010 is from $67 to $465
million.
"The formula is £ (
Where: c = economic impact in the baseline, or
counterfactual case;
f = economic impact with the CAAA, or the
factual case;
t = the year, where 1990 is year 1;
r = the social discount rate, in this case we use 5%.
E-41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableE-17
Cumulative Economic Benefits of Acidification from 1990 to 2010
(Millions of 1990 Dollars)
Economic Impact of Acidification at Each pH Threshold
pHS.O
pH5.4
CAAA Benefit Minimum
$67
$433
CAAA Benefit Maximum
$271
$465
Avoided Cost of Liming
An additional factor that must be considered in
light of the context of economic damages from
acidification is the possibility of mitigating these
damages by local means. In the case of the
Adirondacks, acidic lakes are systematically limed in
order to raise pH and improve habitability for
recreational fish species, which can be stocked after
liming. This alternative is costly for local resource
managers, and difficult to conduct in most
Adirondack lakes with limited access, but it does serve
to locally mitigate the damages caused by acid
deposition. Naturally, damaged aquatic ecosystems
can not be entirely replaced by liming and restocking,
but impacts to recreational fishing may be minimized.
Currently a limited number of lakes are limed in the
Adirondacks. In this section we examine the economic
implications of this practice and demonstrate that
liming will be necessary both with and without the
CAAA.
support brook trout regardless of the pH of the water;
or, liming of the water would be too expensive, due to
its remote location.19
It is not possible to predict the number of lakes
that the New York Department of Environmental
Conservation (NYSDEC) would choose to lime and
restock in 2010 based on the available data, but we can
estimate the potential costs and impacts if the
program remains constant from 1990 to 2010, or if it
grows at NYSDEC's proposed rate of two additional
lakes per year. In 1990 approximately 25 percent of
the region's lakes suffered a pH below 6.0, according
to MAGIC, and NYSDEC limed and monitored 32
lakes and restocked 30. MAGIC estimates that the
same percentage of lakes will maintain a pH below 6.0
in 2010, both with and without the CAAA. Table E-
18 presents the cumulative costs associated with
liming lakes in the region from 1990 to 2010.
The goal of liming in the Adirondacks is to
mitigate the effects of acidification by the addition of
acid neutralizing products in selected waters to
maintain and/or restore brook trout populations.
Waters may be considered for liming and re-stocking,
if the pH drops below 6.O.18 The present liming
program is limited in scope due to policy constraints,
environmental regulations, and factors affecting the
economic feasibility. With a few exceptions, lakes are
typically not limed, if any of the following applies: the
water is considered naturally acidic; the flushing rate is
greater than two times a year; the water will not
18Personal communications with Larry Straight, Rick Costanza
(NYSDEC, Region 5), and Bill Gordon (NYSDEC, Region 6).
19NYSDEC, 1990 and personal communications with Larry
Straight, Rick Costanza (both at NYSDEC, Region 5), and Bill
Gordon (NYSDEC, Region 6).
E-42
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableE-18
Cumulative Cost of Ph Stabilization from 1990 to 2010
(Millions of 1990 Dollars)
Number
of Lakes
Program Remains 32
Constant
Program Grows by 72
Two Lakes per Year
Cost of
Liming
$0.11
$0.16
Cost of Monitoring
$0.07
$0.10
Cost of
Stocking
$0.23
$0.35
Total Cost
$0.40
$0.61
Under the current plan to lime lakes with a pH
below 6.0, this practice will continue under both
scenarios of our analysis - with and without the
CAAA. Therefore, liming costs will be incurred
regardless of regulatory efforts. What we can not
determine is the impact that liming may have on our
avoided damages analysis. It is possible that liming
lakes with the greatest recreational potential will offset
the majority of economic impacts of acidification,
though due to the structure of our economic modeling
approach it is not possible to test this hypothesis at
present. On the other hand, it is important to note
that liming is feasible only on lakes where access is
very easy, and therefore is limited in the scope of its
impact. Furthermore, as previously stated, liming is a
stop-gap measure that is both temporary and not a
complete substitute for restoring natural ecosystem
conditions.
Caveats and Uncertainties
The impacts of acid deposition in the eastern
United States include both terrestrial and aquatic
ecosystem damages. Many of these effects are
difficult to measure, and most are impossible to
monetize given current methods. The result is that
our analysis treats a very narrow definition of the
impact of acidification. A far more broad definition
would include costs associated with damaging the
integrity of terrestrial and aquatic ecosystems, many of
which are not quantifiable at this time. Nevertheless,
our case study of the Adirondacks region
demonstrates that the CAAA is generating substantial
economic benefits in just the narrow scope of
recreational fishing. Our analysis states that by
mitigating the impacts to recreational fisheries from
acidification with the regulations promulgated under
the CAAA, benefits can be accrued in the hundreds of
millions of dollars.
The limitations that affect these estimates are
caused by data and computational constraints at each
stage of the simulation process. We detail each of
these limitations below, and indicate the directional
bias these limitations may create in our final benefits
estimates.
Emissions, Deposition, and
Acidification Estimates
Each of the models that contribute to the
acidification estimates (i.e. the emissions model,
RADM, and MAGIC) has been rigorously tested. For
example, MAGIC estimates of acidification have been
tested extensively including the following procedures:
individual process formulations in the model have
been tested against laboratory experiments with soils;
model hindcasts of historical lake chemistries in the
Adirondacks have been made and compared with
values inferred from lake sediment records; and,
predictions of the effects from whole-watershed
manipulations have been compared to observed
effects. Nevertheless, it is well documented that
MAGIC estimates suffer from unquantified
uncertainty, parameterization, and validation problems
(EPA, 1995).
It is beyond the scope of this report to dissect
MAGIC, RADM, or the emissions projections to
identify factors that might affect their results.
Furthermore, it is not possible to quantify the
cumulative uncertainty that propagates in the linking
E-43
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
of these models to provide acidification estimates. It
is sufficient to note that the acidification estimates
generated by MAGIC should be treated with proper
caution, applying sensitivity analysis to any further
modeling work that uses these data as input. Several
limitations are detailed below.
• We consider the potential effects of nitrogen
saturation on lake acidification by performing
sensitivity analysis of two boundary
conditions, representing a nitrogen-saturated
watershed and a watershed with complete
nitrogen uptake. While this approach allows
us to estimate the range of effects nitrogen
saturation may have on acidification of
surface waters, it does, however, not account
for the fact that nitrogen saturation is a
continuous process leading to increased
leaching of nitrogen compounds as a
watershed progresses through the various
stages of nitrogen saturation (see for example:
Stoddard, 1994).
• The sample of lakes simulated by MAGIC
must be extrapolated to the entire population
of lakes in the region. In order to simulate
the complex hydrological, biological, and
chemical dynamics of lakes, intensive data
collection is required, forcing the developers
of MAGIC to limit the number of simulated
lakes to only 33. This sample represents lakes
with an ANC of less than 400
microequivalents per liter (jjEq/L). Lakes
with greater ANC are believed not to be
vulnerable to acidification from acid
deposition. The results from MAGIC
therefore are only applicable to those lakes
with ANC less than 400 uEq/L, but we have
no assurance that the sample of 33 lakes is
representative of the distribution of lake
ANC levels below 400 uEq/L in the total
population. In addition, we are forced to use
pH 7 as a proxy for ANC 400 uEq/L where
ANC data is not available. Though pH and
ANC are correlated, there is significant
variance in this relationship.
Ecological Factors
In the ecological assessment two major limitations
arise. It is not possible to address either of these
limitations with sensitivity analyses, so it is important
to keep in mind that results may be biased by these
factors.
• Acidic episodes may significantly affect fish
populations. They are, however, not
considered in our analysis due to significant
limitations of our ability to model episodic
events.
• It is well documented that pH is not the only
factor that determines fish survival, although
we do use it as the single indicator for
ecological health. This overlooks the
importance of other components of water
chemistry such as aluminum and calcium
concentrations. This is necessary because
there is insufficient data for our geographic
region to develop a sufficiently sophisticated
ecological-economic model that would
consider all these variables. Because we test
several pH thresholds at which anglers might
perceive declining fish populations, this
simplification of ecosystem dynamics should
not bias our final economic estimates.
Economic Estimates
The economic model is also subject to some
uncertainty that may bias our monetary results. Again,
it is not possible to address all of these limitations
with sensitivity analyses, so it is important to keep in
mind that results may be biased by these factors.
• This analysis includes only day trips to sites
within three hours of the angler's home. We
take no account of people who would come
and spend several days fishing at a site. By
excluding these people we likely understate
the costs of increased acidity.
E-44
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
This model treats every day as a potential
fishing day, and assumes that each fishing
occasion is independent of all others.
Neither assumption is realistic. The
implication is that our measure of the
seasonal costs are likely biased upward.
Montgomery and Needelman do not
compute a confidence interval for the value
estimates of lake acidification in their study.
This could result in an overestimate of the
economic impact of acidification as we can
not determine that the value estimates are
significantly different from zero.
Even though this study offers a much more
comprehensive set of alternative fishing sites
than most recreational fishing studies, we
were unable to account for rivers and
streams, or for lakes and ponds in nearby
states. To the extent that these alternative
sites are substitutes for New York lakes, our
welfare measures may overstate the costs of
acidification.
We assume that anglers perceive the effects
of acidification at a pH threshold lower than
that at which the random utility model was
estimated. This is a conservative approach
which potentially underestimates the total
impact of acidification, but overestimates the
benefits of the CAAA because the difference
between the percentage of lakes that are
acidified in our baseline and CAAA scenarios
is larger at lower threshold pH levels.
Timber Production Impacts
from Tropospheric Ozone
The purpose of this section is to evaluate the
prospective benefits of improved commercial timber
growth through the reduction of tropospheric ozone
concentrations attributable to the CAAA.
Tropospheric ozone (O3) is a secondary pollutant that
is created in the atmosphere by a photochemical
reaction among nitrogen oxides (NOx) and volatile
organic compounds (VOCs). Documented scientific
evidence suggests that elevated ozone concentrations
in the troposphere disrupt ecosystems by damaging
and slowing the growth of vegetation. We examine
one aspect of these impacts in this analysis, reduced
commercial timber growth, and find the cumulative
impacts from 1990 to 2010 to be $1.87 billion.
Ecological Effects of Ozone
In terms of forest productivity and ecosystem
diversity, ozone may be the pollutant with the greatest
potential for regional-scale forest impacts (NAPAP,
1991). Studies have demonstrated repeatedly that
ozone concentrations commonly observed in polluted
areas can have substantial impacts on plant function
(see U.S.EPA 1996; De Steiguer 1990; Pye 1988 for
summaries).
Like carbon dioxide (CO2) and other gaseous
substances, ozone enters plant tissues primarily
through apertures in leaves in a process called
stomatal uptake. To a lesser extent, ozone can also
diffuse directly through surface layers to the plant's
interior (Winner and Atkinson 1986). Once ozone
reaches the interior of plant cells, as a highly reactive
substance, it inhibits or damages essential cellular
components and functions, including enzyme
activities, lipids, and cellular membranes, disrupting
the plant's osmotic (i.e., water) balance and energy
utilization patterns (U.S.EPA 1996; Tingey and Taylor
1982). Damage to plants is commonly manifested as
stress specific symptoms such as chlorotic or necrotic
spots, increased leaf senescence (accelerated leaf
aging) and reduced photosynthesis. All these factors
reduce a plants' capacity to form carbohydrates
(U.S.EPA 1996), which are the primary form of
energy storage and transport in plants. Reduction of
carbohydrate production and disruption of carbon
allocation patterns in turn can impact the growth rates
of trees, shrubs, herbaceous vegetation and crops.
In this section we focus on the economic impacts
of reducing commercial timber growth on the U.S.
economy. Timber supply is a direct ecological service
flow affected by tropospheric ozone and is therefore
E-45
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
an ideal quantitative example of the benefits of
controlling tropospheric ozone in the U.S.
Nevertheless, it is important to note that this benefit
represents only a small portion of the overall
ecological benefits of reducing the impacts of
tropospheric ozone on ecosystems across the nation.
Modeling Timber Impacts from Ozone
In this section we describe our methods for
quantifying the impacts of tropospheric ozone on
commercial timber production. The assessment of the
benefits of regulating tropospheric ozone involves
three major steps:
• Estimation of ambient ozone concentrations
under a regulatory and a non-regulatory
scenario;
• Estimation of the growth changes from
ozone exposure on commercial forests;
• Estimation of the economic impact of
changes in commercial timber growth.
We describe the completion of each of these
steps, the models we use and their input data. Upon
completion of these steps it is possible to compare the
tropospheric ozone concentrations, ecological effects,
and resultant economic impacts over the period
1990-2010 both with and without the CAAA. The net
difference between ecosystem effects and economic
impacts with and without the CAAA represents the
benefits accrued to society from the implementation
of the CAAA.
Step 1: Estimating Ambient Ozone
Concentrations
In order to simulate the impacts of ozone on
commercial forest productivity we must estimate the
ambient ozone concentrations at which forests are
exposed both with and without the regulations of the
CAAA. We accomplish this using historical ambient
ozone data for 1990, and projected ozone data for the
years 2000 and 2010. We use historical hourly ozone
concentrations from EPA's Aerometric Information
Retrieval System (AIRS).20 AIRS is a comprehensive
database that contains ambient air quality monitor
data for the contiguous U.S. To estimate future year
concentrations of ozone we use the Urban Airshed
Model (UAM-V), a three-dimensional photochemical
grid model that calculates concentration of pollutants
by simulating the physical and chemical processes in
the atmosphere.
Step 2: Ecological Effects
We use the PnET-II model to estimate the
impacts on timber growth of elevated ambient ozone.
The model assesses the average change in productivity
for softwood and hardwood forests in each of nine
timber growing regions defined by the U.S. Forest
Service (see following section). The strength of PnET
II is that it provides a means to use a geographically
transferable method to assess forest stand-level
estimates of the impacts of ambient ozone on
productivity. For the purposes of a national
assessment, this provides a significant advantage over
alternative existing methods based on plant-level
models (e.g. the tree grow model (TREGRO)) or
expert opinion surveys (e.g. . Pye et al. 1998;
deSteiguer 1990). The disadvantages of the model
include: the use of photosynthetic rates as an indicator
of ozone impacts rather than a mechanistic measure
of respiratory change; potential bias created when
scaling net primary productivity (NPP) changes in
plants to the forest stand level; and the use of an
ozone measure that may be overly sensitive to changes
in ambient concentrations (D4021). We discuss the
major facets of the model's construction below.
The PnET-II model is a monthly time step,
canopy- to stand-level model of forest carbon and
water balances based on several generalized
relationships (e.g. maximum net photosynthesis as a
function of foliar nitrogen content). Carbon and
water balances are linked in that potential
evapotranspiration is determined as a function of leaf
gas exchange rates and the atmospheric vapor
20 See "www.epa.gov/airs",
21 The D40 measure represents the cumulative ozone dose
above a threshold concentration of 40 ppb.
E-46
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
pressure deficit (i.e. humidity). Actual
evapotranspiration is determined from a comparison
of potential evapotranspiration with available soil
water, which is affected by precipitation, snow melt,
direct evaporation from canopy surfaces, soil water
holding capacity and a fast flow fraction that
represents macropore flow to below the rooting zone.
The model simulates a multi-layered forest canopy
that includes gradients in available light, specific leaf
weight and hence, leaf level carbon gain. Annual,
whole-canopy carbon gain is allocated to leaves, wood
and roots after calculations for growth and
maintenance respiration costs. The model has been
successfully validated for forest production and water
balances at a number of temperate and boreal forest
sites. For a full description of model algorithms,
inputs, assumptions, sensitivity analyses and validation
exercises, see Aber and Federer (1992), Aber et
al,.(1995) and Ollmger et al. (1998).
PnET-II uses an algorithm to allow prediction of
ozone effects on forest growth that relates
ozone-induced reductions in net photosynthesis to
cumulative ozone uptake (Ollinger et al. 1997).
Uptake is determined for ozone concentrations above
40 ppb and is affected by ozone exposure levels and
leaf gas exchange rates. Application at sites located
across the northeastern US show an interesting
interaction between ozone and water availability
whereby the occurrence of drought stress reduced
ozone damage via reductions in stomatal conductance,
and hence, ozone uptake.
Step 3: Economic Impacts
To monetize the ecological effects of elevated
ambient ozone on commercial timber production it is
necessary to estimate the market changes that result
from reduced timber growth rates. We use the USDA
Forest Service Timber Assessment Market Model
(TAMM) to analyze the changes in timber inventories
that would result under each of our ozone exposure
scenarios, and the consequent changes in harvests,
prices and regional demand for timber. Using the
inventory and market computations, TAMM estimates
the overall economic welfare impact of changes in
forest growth rates, in terms of changes in consumer
and producer surplus. Previous peer-reviewed EPA
analyses of changes in timber productivity (U.S.EPA
1997) use this same model.
There are three stages to the economic estimation.
First forest growth rate information generated by
PnET-II is provided to the forest inventory tracking
component of TAMM, called Aggregate Timber Land
Assessment System (ATLAS). Growth rate
information is provided for each of the forest
production regions defined by TAMM. (We do not
simulate changes in the Canadian regions for this
analysis.) Second, ATLAS generates an estimate of
forest inventories in each major region, which in turn
serves as input to the market component of TAMM.
In the third stage, TAMM estimates the future
harvests and market responses in each region. A
detailed description of TAMM's structure is found in
Adams andHaynes (1996).
Ecological Results
P-Net II partitions NPP of forest trees according
to tissue type. Changes in NPP for wood tissue result
in changes in tree growth rates. On the whole, P-Net
II estimates that commercial timber growth rates are
improved as a result of reduced tropospheric ozone
exposure attributable to the CAAA. The
improvement in growth rates by the year 2010 ranged
from negative 0.56 percent to 10.91 percent. Table E-
19 summarizes the estimated changes in growth rates,
by region, for the entire U.S.
E-47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableE-19
Difference in
Region
PN W-E
PN W-W
S. West
N. Rocky
S. Rocky
S. Central
S. East
N. Central
N. East
Commercial Timber Growth
Rates With and Without The CAAA
Difference in 2000
Softwoods
1.68%
1.17%
0.84%
2.67%
4.77%
4.54%
5.40%
1.80%
4.27%
Hardwoods
1.58%
0.42%
1.77%
0.40%
2.25%
4.80%
5.65%
5.74%
6.68%
Difference
Softwoods
2.11%
-0.56%
-0.14%
4.46%
4.14%
7.93%
10.38%
4.36%
9.58%
in 2010
Hardwoods
1.25%
1.13%
1.59%
2.05%
3.88%
8.41%
10.91%
9.22%
11.49%
It is important to note that the difference in
growth rates gradually grows from zero percent in
1990 to the values presented for 2000, and then 2010.
In other words, the difference in growth rate
estimated for 2010 is not experienced over the entire
1990-2010 modeling period.
Economic Impacts
TAMM estimates that there is a measurable
difference in timber harvests attributable to ozone
exposure under our two scenarios. At the outset of
our modeling period, early 1990s, virtually no change
is measured in forest harvest volumes. This is an
expected result because increases in growth rates
should not substantively affect timber volume over so
short a period of time. By the end of our modeling
period, late 2000s, increased growth rates over the
previous decade (s) begin to affect overall forest yields
in the form of harvestable timber. This is observed in
Figure E-8 as an increasing annual benefit estimate
over the modeling period.
The shape of the benefits time-series reveals a
production shift in one region of the United States as
a result of increased timber availability. This shift
produces a spike in economic surplus for a period of
three years. Although this change is small in
percentage terms relative to total economic surplus
generated by the timber sector it contributes to a large
portion of the benefits estimate over the 1990-2010
period.
The cumulative value of annual benefits is
calculated as the sum of the annual differences in
consumer and producer surplus from commercial
timber harvests under our CAAA and no CAAA
ozone exposure scenarios from 1990 to 2010. We
discount annual benefits to 1990 dollars using a five
percent discount rate. The total cumulative benefits
estimate is $1.87 billion.
Caveats and Uncertainties
In interpreting results from this analysis, several
points should be considered. First, large-scale
analyses of complex ecosystem processes are typically
conducted with simulation models because it is
impossible to conduct large-scale manipulation
experiments that would provide similar predictive
capabilities. This brings with it inherent uncertainties
in that there may be little or no data with which to
validate model predictions. In the case of ozone
effects on forest production, the absence of
controlled, whole-forest fumigation studies across the
range of climatic, vegetation and pollution conditions
experienced across the U.S. makes it presently
impossible to validate all model predictions. In this
E-48
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure E-8
Annual Economic Welfare Benefit of Mitigating Ozone Impacts on
Commercial Timber: Difference Between the Pre-CAAA and Post-CAAA
Scenarios
Year
analysis, we have combined established empirical
relationships between ozone exposure and plant
physiological function in a peer-reviewed model that
is based on sound forest growth processes. As such,
the resulting model predictions should be viewed as a
set of refined hypotheses, but nevertheless,
hypotheses that have not been thoroughly tested.
Second, while ozone has repeatedly been
identified as an important environmental stress agent
affecting forest vegetation, it is not the only such
factor to which forests are currently exposed at
regional to global scales. Human activities have
profoundly affected global cycles of carbon, nitrogen
and a number of other elements in ways that may be
at least as important as ozone. Because a number of
changes (e.g. elevated CO2 and increased atmospheric
nitrogen deposition) have significant potential to cause
large-scale fertilization effects, growth predictions that
include ozone effects alone should be viewed as
incomplete.
Ozone Modeling
Because it is not possible to model ozone
levels throughout the country during the
months of October through April during
future years, it is necessary to employ another
method to obtain estimates for ozone levels
during these months. We assume that ozone
levels during these months for 2000 and 2010
will be identical to the levels during the same
months of 1990. Thus, any differences in
timber production under the two scenarios of
CAAA promulgation and no CAAA
promulgation will be driven solely by ozone
differences during the warmer part of the
year that comprises the majority of the
growing season.
It is important to note that ozone monitoring
is not complete, with coverage especially low
in forested regions of the United States.
E-49
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Only two percent of ozone monitors are in
forested areas (U.S.EPA 1996). We work
with the best possible estimates of
tropospheric ozone concentrations but
identify this as a significant area of
uncertainty in this analysis.
Ecological Modeling
Preliminary model results revealed an
interesting and unexpected interaction
between ozone, drought stress and carbon
allocation. On moist, productive sites, ozone
resulted primarily in decreased wood growth
because the simulated trees can afford to lose
wood without reducing more important
tissues which are given higher allocation
priority in the model (leaf and root). On
progressively colder or drier sites, ozone
exposure causes reductions in all plant tissues
because plants are already stressed enough
that additional reductions in carbon gain must
come from all plant pools. Complex
interactions among ozone, drought and this
carbon allocation dynamic produced
unexpectedly variable results, which, in some
cases caused an increase in growth in
response to ozone. Although these are
interesting and biologically feasible
interactions, in the absence of any real data in
this area, it is impossible to determine the
extent to which they actually occur.
Economic Modeling
There are two important caveats to the
economic modeling. First, we generalize
changes in growth rates for entire forest types
across potentially heterogeneous regions.
TAMM is capable of modeling timber growth
and harvest with greater precision, breaking
down forests into many species and
age-classes and by county. We do not
anticipate that increasing the precision of
growth rate data on a national scale would
substantially alter our results.
The second caveat is economic benefits may
be underestimated by using so short a
modeling period. It is evident from the data
we present that improved growth takes years
to affect actual harvests. Therefore, the
complete benefits of improved growth during
1990 to 2010 will not be accrued until after
2010. By not including these years in our
analysis we can not fully account for the
commercial timber benefits of ozone
mitigation over the period of our analysis.
Carbon Sequestration Effects
It is possible to extend the analysis of timber
growth rates to account for the differences in
temporary and long-term carbon sequestration under
each of our ozone scenarios. This is accomplished by
linking two USDA Forest Service Models to
TAMM/ATLAS to generate estimates of carbon
sequestered in standing forest, and carbon sequestered
in commercial forest products. We briefly summarize
those steps here.
TAMM/ATLAS provides an estimate of the
standing timber stock and the commercial timber
harvests that will occur under each of our ozone
exposure scenarios. Using this information we
estimate the respective volumes of carbon sequestered
in each scenario using the forest carbon model
(FORCARB) and harvested carbon model
(HARVCARB).
FORCARB contains a set of stand level carbon
budgets that relate the timber growth and yield output
from ATLAS to trends in total ecosystem carbon over
the course of stand development. These include
carbon sequestered in trees, woody debris, understory
vegetation, and the forest floor. Using these data,
FORCARB estimates the total carbon sequestered in
commercial forests at any point in time. This
information provides a useful baseline for the rate of
forest carbon sequestration that can be expected
under different ozone exposure scenarios. For a
complete description of FORCARB and its
application see Turner et al. (1993) and Turner et al.
(1995).
E-50
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The age of natural forests and the management
regime of commercial forests largely determine the
fate of forest carbon. In natural forests, carbon
sequestration is temporary, with sequestered carbon
eventually returning to the nutrient cycle.
Alternatively, harvested timber is transformed into
commercial products that alter the life cycle of
sequestered carbon. Using the HARVCARB model,
we use harvest information from TAMM to track the
lifecycle of timber. The ultimate fate of this
sequestered carbon depends on the efficiency of
timber conversion (i.e. how much timber becomes a
product), and the durability of that product.
HARVCARB estimates the long-term carbon
sequestration resulting from timber harvests under
each of our scenarios. A full description of
HARVCARB is found in Row and Phelps (1990).
Forest ecosystems help mitigate increasing
anthropogenic carbon dioxide emissions by
sequestering carbon from the atmosphere, converting
atmospheric carbon into biological structures or
substances needed in physiological processes. Some air
pollutants, however, may adversely affect the potential
of forests to sequester carbon by slowing down the
rate of biomass accumulation of sensitive forest tree
species. This may affect the global carbon cycle and
may contribute to anthropogenically induced changes
in the earth's climatic conditions.
Using output from TAMM/ATLAS, timber
inventories can be converted into estimates of carbon
sequestered in commercial forests by a forest carbon
model (FORCARB). FORCARB estimates the
carbon storage in each of four ecosystem components:
trees; forest understory, forest floor, and soil. The
model uses forest carbon storage and flux estimates
based on ecological analyses of each of the forest
ecosystem components. The details of these studies
and their synthesis into the FORCARB model can be
found in Birdsey (1992a, 1992b) and Heath and
Birdsey (1993). Heath and Birdsey (1995) provide a
technical description of integrated simulations using
TAMM/ATLAS and FORCARB. Of the carbon
sequestered in forests, some portion is subsequently
harvested as timber and processed into wood
products, paper, and biomass fuel. We use a harvest
carbon model (HARVCARB) to estimate the life-cycle
of harvested forest timber and thereby adjust the
forest carbon sequestration estimates of FORCARB.
HARVCARB relies on a range of assumptions
approximately 50 percent of harvested wood
ultimately becomes a wood or paper product, the
remainder becomes waste from the production
process. Of the final wood and paper products, a
small percentage become durable products or are
landfilled and decompose at a rate of less than one
percent a year (Row and Phelps, 1990). Wood that is
either manufactured into a durable product (e.g.
permanent building construction material, furniture)
or materials that are landfilled (e.g. paper) contribute
to long-term carbon sequestration. The remainder of
the harvested wood mass (e.g. biomass fuel, non-
durables that are not landfilled) is re-released to the
environment and therefore is not included in the
volume of carbon estimated to be sequestered in
forests.
We find that forest carbon sequestration increases
with improved air quality under the CAAA. This
result corresponds with the intuition that forests tend
to grow faster when tropospheric ozone exposure is
reduced. Carbon flux, or annual forest carbon
sequestration minus forest harvest losses (excluding
long-term carbon sequestration in forest products) is
also greater under the CAAA than under our No-
CAAA air quality scenario. We summarize our results
in Table E-20.
E-51
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-20
Differences in Carbon Flux (millions of metric tons/year)
1990-1999
2000-2010
Forest Flux
8
28
Land Use Change
Cumulative Fate of
Removals
> 1
> 1
TOTAL FLUX
29
Forest carbon flux attributable to the CAAA
represents approximately four to sixteen percent of
anticipated total carbon flux in U.S. forests between
1990-2010.
In the event of a binding international carbon
mitigation agreement, the implication of this result is
that substantial costs of carbon mitigation may be
avoided by improved forest growth attributable to the
CAAA. Though it is not possible to evaluate the
monetary value of the avoided cost at this time due to
uncertainty regarding the actual cost of carbon
mitigation, it will be possible to estimate the value
using the data in this analysis once reliable carbon
mitigation costs become available.
Caveats and Uncertainties
Additional caveats and uncertainties associated
with the estimation of carbon sequestration in U.S.
commercial forests include the following:
• FORCARB estimates are based on a
synthesis of a variety of empirical studies of
the four ecosystem components (soil, forest
floor, understory, and trees). The total error
of the composite of these studies is not
treated explicitly as a modeling output.
• FORCARB also estimates the carbon storage
and flux for a variety of forest types based on
a synthesis of empirical studies. The error
associated with extrapolating these data
across a variety of forest ecosystem types is
not explicitly treated.
HARVCARB utilizes data on the life span of
durable wood products that is over 50 years
old, originally compiled by the Internal
Revenue Service for purposes of calculating
depreciation of these products. Though the
authors of HARVCARB state that this data
continues to be reliable, changes in
construction, product and their uses most
likely biases these data. No estimate is made
of the magnitude or direction of this bias.
Aesthetic Degradation of Forests
The purpose of this section is to evaluate the
prospective benefits of forest aesthetic improvements
associated with improved air quality attributable to the
CAAA. In order to assess these benefits, we first
evaluate the known changes in visible injuries over
time. Available scientific methods and data on the
visual appearance of forest stands and their impact on
perceived forest aesthetics, however, make it difficult
to precisely describe changes in forest aesthetics.
Nevertheless, it is possible to describe a range of
visual impacts that may be caused by air pollutants
and their potential effect on forest aesthetics. Second,
we assess the economic value associated with such
aesthetic changes. The focus of much of this work
tends to be site-specific, describes the aesthetic
impacts of a number of causal factors, and utilizes a
variety of experimental methods making it difficult to
generalize results. We conclude that air quality
improvements attributable to the CAAA should result
in improved forest health, possibly providing aesthetic
E-52
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
value to society in the range of billions of dollars. A
more detailed description of this analysis is found in
Characterising the Forest Aesthetics benefits Attributable to
the 1990 Clean Air Act Amendments, 1990-2010 (TEC,
1999c).
Forest Aesthetic Effects from Air
Pollutants
Air pollution can cause a wide variety of visual
injuries to forest stands, ranging in severity from
subtle injuries (e.g., minor leaf discoloration) to severe
forest decline (e.g., extensive defoliation and death of
trees). The severity of symptoms depends on many
factors including the atmospheric concentration of air
pollutants, the sensitivity of tree species to air
pollution and the presence of other environmental
stress factors (Fox and Mickler, 1995; Eagar and
Adams, 1992; Olson et al, 1992; Smith, 1990).
Many CAAA-regulated air pollutants are
associated with visual symptoms, including, but not
limited to, tropospheric ozone, sulfur dioxide and
hydrogen fluoride, the three major pollutants known
to have caused significant visual injuries to forest trees
in the past (NAPAP, 1987). Other air pollutants
known to potentially cause visual injuries to plants are
strong mineral acids, precursors of which are also
regulated by the CAAA (NAPAP, 1987). In addition,
there are a variety of other air pollutants potentially
affecting the visual appearance of plants, including
heavy metals such as lead and mercury (EPA, 1997d;
Gawel et al., 1996; Smith, 1990; NAPAP, 1987);
nitrogen oxides; ammonia; peroxyacetyl nitrate;
chlorides; and ethylene (Smith, 1990; NAPAP, 1987;
Jacobson and Hill, 1970). However, very limited
information is presently available on visual damages
caused by these pollutants. Tables E-21 and E-22
summarize the known visual impacts of air pollutants
on forests and their geographic extent.
As a consequence of complex natural forest
dynamics, lack of extensive long-term monitoring
networks, and difficulties in establishing cause and
effect relationships, it is not possible to quantify the
extent of visual forest injuries caused by air pollutants
or changes that may have occurred since the
implementation of the CAAA. In addition,
mechanisms that induce threats to forests may operate
on such a long-term scale that benefits in the visual
appearance of forests may not be exhibited during the
period of our analysis.
E-53
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-21
Typical Impacts of Specific Pollutants on the Visual Quality of Forests
Geographic Extent
Direct/Indirect Injuries
Major Types of Visual Injuries
Ozone
Area or regional effects
Direct injuries Foliar injuries (e.g., pigmented stipple), increased needle/leaf abscission, premature
senescence of leaves. Pattern, size, location, and shape of foliar injuries to indicator
species can be specific for ozone.
Indirect Injuries Increased susceptibility to visual injuries that may result from other adverse
environmental factors, such as insect attacks. For example, increased needle/leaf
abscission, elevated mortality rates, and/or changes in species composition.
Acidic Deposition
Area or regional effects
Indirect Injuries Increased susceptibility to visual injuries that may result from other adverse
environmental factors, such as climatic factors. For example, increased needle/leaf
abscission, elevated mortality rates, and/or changes in species composition.
Acidic deposition can also cause direct foliar injuries. Acids are, however, more likely
to indirectly affect the visual appearance of forest trees, unless exposure levels are
very high.
Sulfur Dioxide
Point source pollution
Direct Injuries Foliar injuries including leaf/needle discoloration and necrosis. Pattern, size, location,
and shape of foliar injuries to indicator species can be specific for sulfur dioxide. At
high concentrations, elevated mortality rates of sensitive species and changes in
species composition may occur.
Sulfur dioxide may also cause indirect injuries. Indirect injuries, however, are not well
documented.
Hydrogen Fluoride
Point source pollution
Direct Injuries Foliar injuries including leaf/needle discoloration and necrosis. Pattern, size, location,
and shape of foliar injuries to indicator species can be specific for sulfur dioxide. At
high concentrations, elevated mortality rates of sensitive species and changes in
species composition may occur.
Hydrogen fluoride may also cause indirect injuries. Indirect injuries, however, are not
well documented.
E-54
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-22
Forests Affected by Regional Pollution
Affected Forest Type
/ Species
Mixed Conifer Forest /
Ponderosa and Jeffrey
Pines
Region
San Bernardino
Mountains,
California
Sierra Nevada,
California
Major Air
Pollutants
Ozone and nitrogen
containing
substances
Ozone
Documented Visual
Injuries
Foliar injuries include
chlorotic mottle, tip
necrosis, premature
senescence of
needles, and
increased needle
abscission. Elevated
mortality rates and
changes in species
composition have
occurred.
Foliar injuries include
chlorotic mottle, tip
necrosis, premature
senescence of
needles, and
increased needle
abscission.
Suspected Mechanisms of Injury
Direct ozone-induced foliar injuries. Heavy bark
beetle attacks facilitated by drought, ozone, and
nitrogen containing air pollutants.
Ponderosa and Jeffrey pine have shown air
pollution-related symptoms of decline probably
since the mid 1950s.
Direct ozone-induced foliar injuries.
The Sierra Nevada contains the largest forest area
in the world with documented damage from a non-
point source pollutant but ozone exposure and
injuries are not as severe as in the San Bernardino
Mountains. Visible ozone-induced foliar injuries
Sources
EPA1996a; Miller,
1992; Stolteetal.,
1992; NAPAP,
1991; Miller and
McBride, 1998
EPA, 1996a;
Peterson and
Arbaugh, 1992;
NAPAP, 1991;
Miller and Millecan,
1971
were first documented in the early 1970s.
E-55
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Affected Forest Type
/ Species
Spruce-Fir Forest/
Red Spruce
Region
High elevation
areas in the
northern
Appalachians.
High elevation
areas in the
southern
Appalachians.
Major Air
Pollutants
Acidic deposition
(esp. acidic cloud
water), and ozone
Acidic deposition
(esp. acidic cloud
water) and ozone
Documented Visual
Injuries
Foliar dieback, bud
injury, foliar loss,
Elevated mortality
rates.
Crown thinning and
pockets of high red
spruce mortality have
been detected on a
few mountain sites.
Ozone-induced foliar
injury.
Suspected Mechanisms of Injury
Acidic deposition increases the susceptibility of red
spruce to winter injury (freezing).
A dramatic increase in the frequency of winter injury
in red spruce stands occurred in the late 1950s and
1 960s, coincident with a significant increase in the
emissions of precursors of acidic deposition.
Acidic deposition leads to nutrient imbalances
through accelerated foliar leaching and soil
acidification. Soil acidification is characterized by a
loss of soil nutrient cations and occurrence of toxic
aluminum levels. Also: direct foliar injuries caused
by ozone.
Sources
EPA, 1995a;
Johnson et al.,
1992; DeHayes,
1992;
NAPAP, 1991
EPA, 1995a;
Johnson et al.,
1992; Johnson and
Fernandez, 1992;
Cook and Zedaker,
1992; NAPAP,
1991
Eastern Hardwood
Forest / Sugar Maple
Northeastern US
and Canada
Acidic deposition
and ozone
Crown thinning,
branch dieback,
elevated mortality
rates
Acidic deposition leads to nutrient imbalances
through accelerated foliar leaching and soil
acidification. Soil acidification is characterized by a
loss of soil nutrient cations and occurrence of toxic
aluminum levels.
During 1980s sugar maple declined in many stands
in the northeastern US and Canada. Involvement of
acidic deposition in sugar maple decline has not
been demonstrated but cannot be ruled out.
USFS, 1995b;
EPA, 1995a;
NAPAP, 1991
E-56
-------
Figure E-9
U.S. Major Forest Types Affected by Air Pollution-Induced Visual Injuries
White Mountains (NH)
Green Mountains (VT)
Adirondacks (NY)
Sierra
Navada
(CA)
San Bernadino
Mountains (CA)
Mt. Rogers (VA)
i
Mt. Mitchell (NC)
Great Smoky
Mountains
Eastern Spruce/Fir Forest
J Eastern Hardwood Forest
Sierra and Los Angeles Basin Ecosystem
Note: Only areas affected by non-point pollution are shown. Scientific certainty varies with location. Direct ozone-induced
injuries also occur in several other locations not indicated (e.g., Southern Forests, Berraug et al, 1995).
Sources: NAPAP, 1991 and White and Cogbill, 1992.
E-57
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Despite limitations in detecting trends in forest
health and associated causal agents, it is possible to
identify areas in the US that contain forests known or
suspected to experience visual injuries. Forests
affected by high concentrations of air pollutants in the
vicinity of point sources may provide useful case
studies because cause and effect relationships may be
easier to establish and visual injuries can be severe
enough to cause significant aesthetic impacts. In
particular, point sources can lead to well-defined
concentration gradients in the prevailing downwind
direction causing corresponding gradients of visual
injuries (Smith, 1990; NAPAP, 1987).
In contrast, concentrations of regionally
distributed air pollutants (e. g., ozone and acidic
deposition), can be fairly uniform over large
geographic areas. Visual symptoms can be more
intense in the vicinity of urban areas or industrial sites
but may not be limited to these regions (NAPAP,
1987) making it more difficult to establish cause-and
effect relationships. Despite difficulties in establishing
cause-and-effect relationships, all identified forest
ecosystems likely to have experienced air
pollution-induced visual injuries in recent history are
affected by regionally distributed air pollutants.
Economic Value of Changes
in Forest Aesthetics
Though studies that attempt to estimate the value
of changing aesthetics are limited in number and
scope, they do suggest that people value forest
aesthetics and change outdoor recreational behavior
according to the quality of forest health in recreational
areas. The shear volume of forest-based recreation in
the United States suggests that improvements in forest
aesthetics could result in substantial benefits. For
example, the United States Forest Service reports that
recreation visitor days to national forests have
increased over the last ten years from 250 million to
over 350 million. With the potential magnitude of
aggregated individual preferences in mind, we review
several studies that relate individual preference for
forests with respect to overall appearance and attempt
to extend these analyses to those regions where forests
are most affected by air pollution.
Peterson et al. (1987) used the contingent
valuation (CV) method and a hedonic property
valuation model to estimate willingness to pay to
avoid ozone-induced forest damage in the Los
Angeles area. This contingent valuation survey
involved two samples: one made up of recreationalists
in the greater Los Angeles area, and the second made
up of individuals who owned property within the
boundaries of the San Bernardino and Angeles
National Forests. Each group was shown a set of
photographs depicting varying degrees of vegetative
damage. Mean WTP by recreationalists and residents
were found to be approximately $43 and $137 per
household per year, respectively. The hedonic analysis
revealed a significant and positive WTP to avoid
homes located in forested areas exhibiting ozone
damage. Using these two methods, total damages
resulting from the current levels of ozone induced
forest injury were estimated to be between $31 and
$161 million per year. The study authors rejected a
significant percentage of responses as "protest" or
"inconsistent" bids (40 percent), which would indicate
that many respondents may not have understood or
accepted the scenario and the commodity being
valued. Apart from this, the study also does not
address a series of concerns related to the application
of CV to assess nonuse values. First, the survey
instrument did not include reminders of budget
constraints or substitute goods and services. Second,
the survey did not clearly define the commodity. The
WTP scenario did not clearly indicate how forest
damages were to be mitigated.
Walsh et al. (1990) interviewed 200 individuals
representing the general population of Colorado and
were shown three color photographs representing
three levels of forest quality. The mid-level quality
was said to represent the present state of the forest
(100 to 125 live trees measuring more than six inches
in diameter at breast height (dbh) per acre).
Respondents were asked their WTP to prevent the
lowest state (zero to 50 live trees measuring more than
six inches dbh per acre) and attain the highest state
(125 to 175 trees per acre in this size class). All
respondents were informed beforehand that the
damage being valued was due to pine beetle and
spruce budworm infestations. Mean WTP per
respondent was estimated to be $47. An evaluation of
the Walsh et al (1990) study reveals several notable
E-58
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
strengths. The survey included reminders of budget
constraints, and the authors ensured that respondents
were familiar with the commodity being valued and
were accustomed to paying for access to recreation
sites with good forest quality. Only five percent of the
responses were rejected as "protest" or "large" bids.
Weaknesses of the study include a small sample size
(198), inconsistency between results solicited using
different question formats (iterative bidding vs. direct
question), and potential biases attributable to framing
the question as one of the most important issues
affecting Colorado residents and the possibility of a
"warm glow" affect concerning payment for a social
cause.
Holmes et al (1992) used a CV survey to
determine WTP to protect threatened spruce-fir
forests in Southern Appalachia from insect and air
pollution damage. In this study, residents within 500
miles of Asheville, NC were surveyed about their
willingness to pay to eliminate damages to regional
spruce-fir forests. The authors used two survey
formats, discrete choice and payment cards. The
mean willingness to pay for protecting the spruce-fir
forests was $20.86 using the payment card method,
and $99.57 using the discrete choice method. The
study ensured that the sample had adequate
knowledge of the commodity being valued, and the
overall sample size was large. Unfortunately, several
weaknesses arise from the fact that the sample was
divided into two groups in order to test different
survey formats. The study used a small sample size
for each of the tested methods (232 and 236,
respectively). The number of protest bids was small
(7 to 10 percent), indicating that the respondents
understood the function of the survey, but the final
results generated by the two different methods were
substantially different. This study was later revised in
Holmes and Kramer (1996), where the results were
published as mean willingness to pay of $36.22 for
forest users, and $10.37 for nonusers.
Extending Economic Estimates to a
Broader Area
These studies provide an incomplete picture of
the total benefits that could be obtained by eliminating
visual damages to forests associated with air pollution
in the country. As an illustrative calculation, we
extend the range of valuation estimates provided in
Peterson (1987); Walsh et al. (1990); and Holmes and
Kramer (1996) to the major regions of affected
landscape in the United States. We do not estimate
aesthetic value as a function of forest damage from
varying levels of air pollution, but rather provide an
estimate of the values placed on avoiding damages
characteristically experienced during the 1980s in the
United States.
In Table E-23 we present the results from the
three studies. We base our calculations of benefits on
the value per household of avoiding forest damages
multiplied by the number of households in the study
region.
In Table E-24 we present the results of an
illustrative calculation that extends the "market" for
this commodity to a broader group of households.
The annual value of avoiding the forest damages is the
product of the range of household values in Table E-
23 and the total number of households in the states
most affected by air pollution.
E-59
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-23
Summary of Monetized Estimates of the Annual Value of Forest Quality Changes
Study
Peterson et
Aesthetic
Change Valued
Ozone damage to
Value of
Change per
Household
(Current
Dollars)
$6.31-$32.70N
Value of
Change per
Household
(1990 Dollars)1
$7.26-$37.62
Total Annual Value of
Change for Region
(Current Dollars)
$27-$140 million
Total Annual
Value of
Change for
Region (1990
Dollars)1
$31-$161million
al. (1987) San Bernardino
and Angeles
National Forests
Walsh et al. Visual damage to
(1990) Colorado's Front
Range
$47
$61.68
$55.7 million
$73.09 million
Holmes and Visual damage to $10.81 $10.37
Kramer spruce-fir forests nonusers nonusers
(1996) in southern
Appalachia $36.22 users $34.76 users
NA
NA
Note: i.) Values adjusted using all item Consumer Price Index, Economic Report of the President, 1998. Years for current dollar
estimates: Peterson et al, 1987; Walsh et al, 1983; Holmes et al, 1991.
ii) Based on 4.3 million households in Los Angeles, Orange, and San Bernardino counties.
iii) Assumes 2.5 million households in North Carolina and 1.8 million in Tennessee.
Table E-24
Illustrative Value of Avoiding Forest Damage in the United States
(1990 Dollars)
Affected States
System Included
Sierra Nevada CA
and Los
Angeles Basin
Eastern Spruce ME, VT, NH,
Fir and MA, NY, PA,
Selected WV, TN,
Eastern KY, NC, VA
Hardwood
Notes: i.) Household data from 1990
discount rate.
Value per
Household
$7.26-$37.62
$7.26-$37.62
Census; ii) Total Value
Estimated Total
Households1 Annual Value"
10. 4 million $75.5 million -$391.2
million
23.2 million $168 million - $872.8
million
Cumulative
Value
(1 990-201 Of1
$1.02 billion -
$5.27 billion
$2.27 billion -
$11. 75 billion
= Households x Value per Household; iii)Assumes a 5 percent real
The results of existing work in this area suggest
that improvements in air pollution controls result in
positive changes in the aesthetic quality of forest
stands. Pollutant control provisions of the 1970 CAA
and the 1977 CAAA, for example, may have resulted
in a significant decrease or elimination of forests
visually affected in the vicinity of emission sources.
Further reductions in air pollution emissions
mandated by the 1990 CAAA should result in
additional improvements in forest health and
associated economic benefits derived from improved
forest aesthetics.
Our illustrative calculation of the regional effects
of improving the aesthetic quality of forest stands
(Table E-24) likely overstates the extent of market for
this commodity. Estimates presented in Table E-23,
however, based on a more conservative application of
the extent of market for this commodity, provide a
better basis to estimating the order of magnitude of
E-60
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
this category of effects of air pollution on ecosystem
health. Considering only the Peterson et al. and Walsh
et al. studies, conducted in two areas that have been
shown in previous assessments to be affected by
accumulated air pollution damages, estimates of the
total annual value of improvements in the aesthetic
quality of forests are in the $100 million to $250
million range.
Caveats and Uncertainties
To quantitatively assess the effects of air pollution
emission reductions on forest aesthetic benefits,
considerable amounts of high-quality data are
required. These data include extensive long-term
monitoring networks producing consistent and
comparable information over time frames as long as
several decades. In addition, injuries captured by
monitoring networks have to be linked to the causal
agent(s), a task that is currently associated with high
factors of uncertainty. Only rarely, if ever, is air
pollution the only factor negatively affecting forest
health. Typically, a variety of adverse environmental
factors act synergistically to induce injuries,
considerably limiting our ability to detect air pollution
as one of the factors causing injury and to
quantitatively assess the amount of injuries attributable
to air pollutants.
There are caveats to the use of benefits transfer in
this context. The application of this method is
intended to provide an order of magnitude estimate of
the benefits associated with avoided aesthetic damages
to forests in the United States. More sophisticated
estimation methods will be required if a truly accurate
estimate of value, especially the marginal value of
incremental changes, is to be derived. Following is a
summary of the caveats to using this approach.
• The impacts that we value are not equivalent
to those avoided through the implementation
of the CAAA, they are historical effects. A
comprehensive assessment of forest
aesthetics-related benefits associated with
improvements in air quality is limited by
significant factors of uncertainty occurring in
both the natural science component of the
assessment and the economic analysis.
Factors of uncertainty in natural sciences
include difficulties detecting trends in forest
health in general, attributing changes in forest
health to specific factors such as air pollution,
and establishing valid dose-response
relationships of forest exposure to air
pollutants and resulting visual injuries.
The types of aesthetic deterioration in the
original studies are not necessarily the same
as those experienced in other regions. The
nature of forest aesthetic deterioration will
vary (e.g. the yellowing of conifer needles vs.
gypsy moth defoliation of hardwoods) as will
the intensity.
We do not fully assess the range of potential
substitutes for the aesthetic health of regional
forests to each household. Having ready
substitutes could lower the value a specific
household might place on aesthetic quality of
regional forests.
The distinction between marginal values for
forest health and average value is not made.
As marginal values for changes in forest
health diverge from the assumed average
value in this analysis, the estimates develop
bias.
We assume that differences in average
regional income do not affect estimates.
Toxification of Freshwater
Fisheries
The purpose of this section is to assess, from
1990 through 2010, the ecological benefits likely to
accrue as a consequence of reductions in the
emissions of hazardous air pollutants (HAPs), as
mandated by the CAAA. Title III of the CAAA lists
189 chemicals considered to be HAPs. Ideally, a
comprehensive economic analysis of the ecological
benefits of CAAA-mandated reductions in HAP
emissions would include analyses for all service flows
potentially affected by the emissions of HAPs.
However, a broad quantitative analysis of all these
benefits is not yet scientifically possible. What is
E-61
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
possible is a qualitative analysis of the likely benefits of
reduced HAP emissions for recreational fishing. A
more detailed description of this analysis is found in
Economic Benefits of Decreased Air Toxics Deposition
Attributable to the 1990 Clean Air Act Amendments,
1990-2010 (lEc 1998d).
Impacts of Toxic Air Emissions
Five HAPs, mercury, PCBs, chlordane, dioxins,
and DDT were responsible for nearly 95 percent of
the fishing advisories extant in 1995 (EPA 1996b).
The use of three of these compounds (PCBs,
chlordane, and DDT) was effectively illegal in the
United States prior to 1990 (EPA 1992a), and there
are currently no plans for additional CAAA
regulations of these compounds (Federal Register
Unified Agenda 1998). The remaining two HAPs,
mercury and dioxins, are therefore the focus of this
analysis.
Because the ecosystem responses to toxic
contamination are poorly understood, and observable
service flow impacts are difficult to model, we use
fishing advisories as a measure of the extent of toxic
contamination. In addition, we can characterize the
economic impact of HAPs emissions based on altered
fishing behavior caused by toxic contamination of
freshwater fisheries. It is important to note that
fishing advisories alone do not provide a
comprehensive view of impacts of toxic
contamination on ecosystems, and more expansive
measures should be examined in future research.
Fishing advisories are issued by state and tribal
agencies when the levels of toxins in the tissue of fish
exceed limits established by both state and federal
authorities. Fishing advisories generally take one of
four forms:
• Advisory for any consumption by the general
population;
• Advisory for pregnant women, nursing
mothers, and children;
• Advisory for limitation on consumption
based on size of fish and frequency of
consumption; and
• Advisory for limitation on consumption for
specific sub-populations.
According to the U.S. Fish and Wildlife Service
(1998), the total number of advisories in the U.S. in
1997 was 2,299, increasing five percent from 1996.
The number of water bodies under advisory
represents 16.5 percent of the nation's total lake acres
and 8.2 percent of total river miles. In addition, 100
percent of the Great Lakes waters and their
connecting waters and a large portion of the nation's
coastal waters are also under advisory. The total
number of advisories in the U.S. has steadily increased
for mercury and dioxin.
Mercury is responsible for approximately 75
percent of all fish consumption advisories in effect in
1995 (EPA 1996b). Mercury from point sources, as
opposed to mercury deposited from the atmosphere,
may be responsible for many of these advisories. The
lakes and streams with advisories are concentrated in
the northern portions of Minnesota and Wisconsin, as
well as in Florida, Missouri, Indiana, Ohio, North
Carolina and New England (EPA 1997a, 1997d).
Judging by fish consumption advisories, fish mercury
levels do not appear to be a widespread problem in
the remainder of the United States, and EPA (1997d)
found that the typical consumer eating purchased fish
is not at risk of methylmercury poisoning.
Approximately three percent of fish consumption
advisories in effect in 1995 were due to the presence
of dioxins (EPA 1996b), and in 1996, 18 states had
one or more water bodies under advisement because
of dioxin levels in fish (EPA 1997a). Dioxins from
point sources may be responsible for many of these
advisories.
Several limitations to the fish advisory data exist.
First, many lakes, rivers and streams have not been
analyzed for toxicity, and it is possible that advisories
eventually will be issued for these water bodies. Table
E-25 summarizes the sampling intensity for toxicity
through 1997. Second, current levels of toxics in
watersheds may result in future toxification of healthy
water bodies, even in the absence of additional future
HAP deposition. Therefore, the current set of fish
advisories underestimates the magnitude of
toxification from air deposition to date. Third, the
protocol for fishing advisory issuance may vary from
E-62
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-25
Summary of National Data on Toxicity Sampling for Fishing Advisories
Water Body
Percentage of Water
Bodies Assessed for
Contaminants
Percentage of Assessed Water Bodies
Under Advisory
Lakes
(acres)
11.36
78.61
Rivers and Streams
(miles)
2.41
29.58
Source: EPA1997a
state to state, removing any consistent basis on which
to judge the levels and causes of fisheries' toxicity for
each state.
Illustration of Economic Cost to Anglers
The economic welfare implication of water quality
changes to recreational fishing are well studied. Most
literature in this field focuses on the impacts of
deteriorating water quality in a specific fishery. More
recently, economic models are appearing that address
the social welfare cost of water quality deterioration in
multiple fisheries within a region. Such an approach
accounts for choices made by fishermen concerning
travel to, and the attributes of (e.g., fish advisories),
multiple fisheries. Random utility models (RUM)
provide the computational method for these regional
analyses.
Montgomery and Needelman (1997) were the first
to use direct water quality measures in conjunction
with a RUM approach to analyze the economic
impacts of toxification on regional anglers. Using data
from the New York Department of Environmental
Conservation (NYDEC), Montgomery and
Needelman identify 23 water bodies with toxicity
advisories among 2,561 lakes and ponds in the state.
Using water quality data and geographic location of
both water bodies and anglers, the authors estimate
the economic cost of the toxification within the state.
The results are presented in Table E-26.
Table E-26
Estimates of the Welfare Cost of Toxification in New York State
(1990 Dollars)
Compensating
Level of Toxicity Variation per Trip
Toxic Contamination $1.23
Site Closed Due to Toxic $1 .69
Contamination
Compensating Compensating
Variation per Capita Variation per Capita
per Day per Season
$0.37 $51.51
$0.50 $70.92
Source: Montgomery and Needelman 1997
E-63
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
The results from Montgomery and Needelman
indicate that the economic welfare implication of
existing toxic contamination is substantial for New
York State, as described below:
$0.37/person/day x 17,990,000 people x 140
fishing days/season = $931,882,000/season.
In perpetuity,22 the value of eliminating toxicity in
New York State, using a five percent discount rate, is
calculated to be $18,637,640,000.
Clearly, the results using these assumptions are
very large. Applications of this model for purposes of
estimating the effects of air toxics deposition on
recreational fishing requires further investigation of
the assumptions in this model.
Jakus et al. (1997) conducted a similar RUM
analysis of toxification of reservoirs in Tennessee.
Data from the Tennessee Valley Authority showed
fishing advisories for two of 14 reservoirs in central
Tennessee, and six of 14 reservoirs in the eastern
portion of the state. Again, using water quality data
and the geographic locations of both water bodies and
anglers, Jakus et al. (1997) estimated the economic
impact of the fish consumption advisories. Anglers
living in central Tennessee suffered a $17.92 per trip
per season loss from the advisories, and anglers in
eastern Tennessee suffered a $38.27 loss (1990
dollars). Therefore, considering an angler population
of 146,450 individuals, the impact of this level of
toxification into perpetuity, using a five percent
discount rate, is approximately $65.96 million.
These results indicate that fish advisories impose
substantial economic cost on anglers in the United
States. Measuring the marginal changes in toxification
that would occur in the absence of the CAAA is not
possible, but it is plausible to state that continued
HAP emissions impose a cost on society if they result
22 A perpetuity is a stream of benefits, accrued over an infinite
time horizon. A simplified formula for calculating a perpetuity of
equal benefits accrued annually, in which the first payment is
received at the end of year one, the second payment is received at
the end of year two, etc., is: (Nominal Value of Benefit) /
(Discount Rate)
in the issuance of additional fish advisories. Any
efforts to minimize these emissions, including the
CAAA, may generate corresponding benefits.
If air deposition of toxics results in statewide
fishing advisories (e.g., Connecticut, Washington D.C.,
Illinois, Maine, Massachusetts, Missouri, New
Hampshire, New Jersey, New York, North Carolina,
Ohio, Vermont), substitution away from recreational
fishing for other activities may begin to occur. No
models are available to estimate the economic impact
of a large-scale substitution away from recreational
fishing. The RUM approach does not adequately
capture the magnitude of ubiquitous toxification
because the models measure only the choice to
participate in the activity and not the welfare
implications of participation in alternative activities,
nor do they account for the industries that provide
supplies and services to anglers in the region.
However, the economic cost of statewide advisories
could be substantial.
Although Montgomery and Needelman (1997)
and Jakus et al. (1997) examined only two areas of the
country - New York State and part of Tennessee -
their work demonstrates that HAP emissions have a
measurable economic cost when the consequence of
these emissions is the issuance of fish advisories for
recreational fisheries. While it is not possible to
measure the differences in HAP deposition and the
marginal ecological impacts that will result from the
CAAA, it is clear that continued emissions of HAPs
will result in further toxification of aquatic resources,
and reductions in HAP emissions may provide
economic benefits.
The toxification of freshwater ecosystems in the
United States by mercury and dioxins is a problem,
and emissions of mercury and dioxins to the
atmosphere contribute significantly to the problem.
Quantifying the magnitude of ecosystem effects of air
toxics deposition is not yet possible, but it is clear that
the deposition of air toxics to some ecosystems, such
as freshwater recreational fisheries, can result in
measurable economic costs.
E-64
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Caveats and Uncertainties
Because of limitations in the currently available
data and models, a comprehensive quantitative
analysis of the ecological benefits of reduced mercury
and dioxin emissions for recreational fishing is not
possible. However, such an analysis may be possible
in the foreseeable future.
• The potential for mercury and dioxins to
persist for long periods of time in the
environment is a confounding factor in this
analysis. Because these pollutants can persist
in aquatic ecosystems for decades, even
though the CAAA may reduce their
emissions, it is possible that the status of
toxified ecosystems may not be significantly
affected during the time frame of the analysis
(i.e., through 2010).
• In addition, the persistent nature of
toxification presents challenges with respect
to how benefits are discounted over time. In
those cases where recovery from toxification
will take a number of years, the benefits
accrued by society will be diminished in terms
of their present value. In other words, if all
air emissions ceased, many fish consumption
advisories would remain in place until the
fisheries recovered. If this recovery period
were to extend for several decades, the
present value of economic benefits from the
eventual retraction of advisories could be
reduced dramatically. In a cost-benefit
decision analysis, these benefits might not
justify the costs of HAP regulations. In this
case, an inter-generational benefits
assessment, where discounting is not applied,
would be required.
• The global nature of mercury pollution is
another confounding factor Because a
significant portion of mercury deposited
within the U.S. comes from the global pool,
a decrease in U.S. emissions may be offset by
increases in emissions in other countries. If
this should occur, it might be difficult to
detect or predict actual changes in the toxicity
of U.S. aquatic ecosystems, despite
reductions in U.S. emissions.
To quantitatively assess the effects of
mercury and dioxin emission reductions on
recreational fishing, more and better data and
models are required. The most pressing
research needs in this area are a model that
can predict the national fate and transport of
dioxin, and models that can, on a national
scale, convert mercury and dioxin deposition
quantities to amounts of the contaminants in
fish. Data to verify these models is also
highly desirable.
Even if it were currently possible to perform
the analysis discussed here, it would likely
capture only a fraction of all the benefits
attributable to CAAA-mandated HAP
emissions reductions. The analysis focused
entirely on two HAPs and on one endpoint.
Neither the potential benefits of reductions
in the emissions of other HAPs nor other
endpoints were considered.
Conclusions and Implications
Our analysis has identified four major categories
of air pollutants that affect ecological structure and
function: sulfur compounds, nitrogen compounds,
tropospheric ozone, and hazardous air pollutants.
Each of these pollutants is scientifically documented
as a cause of ecosystem degradation due to acute and
chronic exposure. Sulfur and nitrogen compounds
contribute to episodic and chronic acidification of
aquatic and terrestrial ecosystems, while the chronic
deposition of nitrogen compounds alone may cause
harmful eutrophication to terrestrial and aquatic
ecosystems. Tropospheric ozone disrupts the normal
functioning of plants, leading to acute, visible damages
to terrestrial ecosystems, and chronic exposure at
levels that do not produce acute damages may result
in reduced growth rates and eventually alter ecosystem
nutrient cycling. Finally, hazardous air pollutants
deposited across the landscape are accumulating in
aquatic organisms and subsequently entering both
E-65
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
aquatic and terrestrial foodchains. Though the
ecological impacts are not fully understood, the long-
term effects of introducing hazardous air pollutants to
ecosystems may be slow to manifest and irreversible
in nature.
Ecological effects can occur at different levels of
biological organization. Most effects that are currently
quantifiable are understood at the individual or
population level, perhaps because of the feasiblity of
conducting controlled experiments at this level. For
example, research on the effects of ozone on timber
began with experimental research on the response of
seedlings and leaves of mature trees to elevated levels
of ozone. Only recently have modeling efforts begun
to consider interactions of factors at the community
level, taking into account the dynamics of competitive
relationships among tree and plant species.
Experimental research continues to progress toward
a better understanding of the full range of ecological
impacts including effects at the ecosystem level.
Continued consideration of these higher-order effects
of pollutants on ecological systems is necessary for a
more complete understanding of the benefits of
pollution control.
Because the chronic ecological effects of air
pollutants may be poorly understood, difficult to
observe, or difficult to discern from other influences
on dynamic ecosystems, our analysis focuses on acute
or readily observable impacts. Disruptions that may
seem inconsequential in the short-term, however, can
have hidden, long-term effects through a series of
interrelationships that can be difficult or impossible to
observe, quantify, and model. This factor suggests
that many of our qualitative and quantitative results
may underestimate the overall, long-term effects of
pollutants on ecological systems and resources.
Summary of Quantitative Results
Although the effects of air pollutants on
ecological systems are likely to be widespread, many
effects may be poorly understood and lack quantitative
effects characterization methods and supporting data.
In addition, many of our quantitative results reflect an
incomplete geographic scope of analysis; for example,
we generated monetized acidification results only for
the Adirondacks region of New York State. As a
result, quantitative results we generate for the
purposes of estimating the benefits of the CAAA
reflect only a small portion of the overall impacts of
air pollution on ecological systems. Our quantitative
overview of effects nevertheless suggests that the
overall impacts of air pollution are far greater than
those quantified.
E-66
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table E-27
Summary of Monetized Ecological Benefits (millions 1990$)
Geographic Range of Primary
Scale of Annual Impact Central
Description Air Economic Estimates in Estimate
of Effect Pollutant Estimate 2010 for 2010
Freshwater Sulfur and Regional $12 to $88 $50
acidification nitrogen (Adirondacks)
oxides
Reduced tree Ozone National $190 to $1000 $600
growth - Lost
commercial
timber
TOTAL MONETIZED ECONOMIC BENEFIT $200 to $1,100 $650
Primary
Central
Cumulative
Impact
Estimate
1990-2010
$260
$1,900
$2,200
Key Limitations
- Captures only
recreational
fishing impact
- Incomplete
geographic
coverage leads to
underestimate of
benefits
- Uncertainties in
stand-level
response to
ozone exposure
- Uncertainty in
future timber
markets
- Partial estimate
that omits major
unquantifiable
benefits
categories; see
text
Note: Estimates reflect only those benefits categories for which quantitative economic analysis was supported. A comprehensive
total economic benefit estimate would likely greatly exceed the estimates in the table. Range of estimates for timber assessment is
based on variation in annual point estimates for 2005 through 2010.
E-67
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Despite these limitations, it is important to
recognize the magnitude of the monetized ecological
benefits that we could estimate and reflect those
results in the overall estimates of benefits generated
in the larger analysis. Table E-27 provides a tabular
summary of the results documented earlier in this
appendix. It is not possible to indicate the degree to
which ecological benefits are underestimated, but
considering the magnitude of benefits estimated for
the select endpoints considered in our analysis, it is
reasonable to conclude that a comprehensive benefits
assessment would yield substantially greater total
benefits estimates.
Recommendations for Future Research
Previous sections of this appendix have discussed
several areas for future research related to the
individual research and analytic efforts conducted.
From a broader perspective, there are three key
research needs to improve benefits assessments of this
type:
• Exemplary assessments that incorporate a
greater emphasis on ecosystem structure and
function rather than specific service flows;
• Assessments with broader geographic
coverage of impacts categories assessed in
this report; and
• More sophisticated treatment of uncertainty
and complexity, including careful
consideration of the irreversibility of
ecosystem impacts.
Assessing Changes in Ecosystem
Structure and Funtion
A major limitation of our quantitative analysis is
that by focusing on individual acute and chronic
impacts it is possible to lose sight of ecosystem-level
changes to structure and function. These ecosystem-
level changes could eventually lead to large-scale
impacts far greater in degree and geographic extent.
Determining the appropriate ecological level of
analysis is crucial to properly account for ecological
benefits that may accrue from environmental
regulations. While quantifying the decrease in impacts
on species attributable to air pollutant control is
analytically tractable, the impact of pollutant
reductions on ecosystem structure and function may
be a more appropriate measure that can be further
explored in future analyses.
Changes in ecosystem structure and function may
not be obvious to the lay person, and the ultimate
effects of such changes in ecosystems are sometimes
unpredictable in scale and nature. Ecosystems
affected by humankind may respond in a
discontinuous manner around critical thresholds that
are boundaries between locally stable equilibria.
Complexity in ecosystems prevents analysts from
using linear methods to "add up" the discrete
ecological effects of pollution. Understanding the
complex cause and effect relationships between
pollution and deterioration of ecosystem structure and
function is fundamental to making adequate policy
decisions that will protect ecological resources. The
isolation of service flows may often imply an
oversimplified cause and effect relationship between
pollution and the provision of the service flow, when
more often the service flow is affected by complex
non-linear relationships that govern ecosystem
structure and function. The result is that ecosystem
impacts may not be adequately assessed by analyses
that focus on specific service flows.
One potentially fruitful approach to assessing
impacts on the ecosystem scale would be to more
adequately model a wide range of ecosystem functions
that do not necessarily contribute to human welfare.
Assessments at the watershed scale might provide an
appropriate level of detail to more adequately
characterize some of these intermediate service flows.
This type of research effort would require close
cooperation between air pollution specialists,
ecologists, and economists to be most useful within
the context of benefit-cost analyses such as this one.
Broader Geographic Scale
Several of the ecological analyses conducted to
support the first prospective section 812 report are
limited by their partial geographic coverage. For
example, while nitrogen deposition is an important
contributor to eutrophication in a wide range of
Eastern and Gulf Coast estuaries, resource, time, and
data availability constraints, as well as limitations in
our ability to reasonably apply an avoided cost
E-68
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
approach, prevented EPA from conducting a national
economic assessment for this category of impacts. In
this and many other effects categories, extension of
the methods applied here to new geographic areas
could greatly enhance the comprehensiveness of the
physical effects and economic impact estimates.
Alternative Treatment of Uncertainty
At present a variety of economic schools of
thought are converging on quantitative analysis of
environmental impacts that integrate uncertainty,
irreversibility and ecological complexity. Efforts
within the field of "ecological economics" to develop
structured appraisals of uncertainty associated with
environmental management and procedural rationale
for decision making have yielded a variety of
theoretical proposals. Drepper and Mansson (1993)
argue that most aspects of uncertainty are compressed
into the discount rate for policy analysis, resulting in
the inappropriate use of a constant positive discount
rate for environmental existence values. These
existence values, they argue, may be more
appropriately assigned negative discount rates.
Faucheux and Munda (1997) advance a similar
criticism of the unified discount rate and posit that a
differentiated discount rate be applied to multiple
aspects of a policy decision according to the implied
uncertainty of each aspect. This quantitative approach
evolves into a multi-criteria decision framework that
departs from conventional cost-benefit analysis.
Alternatively, Hinterberger and Wegner (1997)
abandon quantitative analysis as a futile exercise due to
ecosystem complexity in favor of simply applying the
precautionary principal of reducing any and all
environmental impacts that have uncertain outcomes.
In the resource economics literature, discussion of
alternatives to cost-benefit analysis when the
magnitude of benefits or costs are uncertain have
focused on the concept of quasi-option value (see
Freeman 1993 for a summary). The term was coined
by Arrow and Fisher (1974) to describe the potential
welfare gain of altering the timing of
development/preservation decisions under uncertainty
and when at least one of the choices involves an
inrreversible commitment of resources (either spent or
preserved). While much of the quasi-option value
literature suggests that adopting this type of
framework would lead to greater environmental
protection, Freeman (1993) argues that it is also
possible that the information gained by some
incremental development of ecological resources
might be the only way to reduce uncertainty and gain
information about the magnitude of the trade-offs
involved in preventing ecological exposures. It is
nonetheless important to recognize that option and
quasi-option value should not be considered as
additional components of willingness-to-pay, but
rather a value of altering decision making practices
(e.g., the value of moving from a benefit-cost
framework based on expected value to a framework
that better considers the value of information gained
over time and the irreversibility of certain effects).
The main implication of this body of work is that
cost-benefit analysis may well underestimate the value
of both the costs and benefits of uncertain,
irreversible environmental outcomes from public
policy. From the cost perspective, regulating a
pollutant that may have no environmental
consequence may cause economic losses that reduce
unknown investment and growth opportunities in the
future. From the benefits perspective, the value of
preserving ecosystem integrity may include the
mitigation of irreversible damage to a variety of
service flows previously not associated with simplified
dose-response relationships between pollution and
ecosystems. Applications of these principles in
economic assessments, including more rigorous
assessments of option and quasi-option value,
probabilistic analysis of multiple scenarios, and value
of information approaches have the potential to
greatly increase the utility of uncertain ecological
assessment results for the purposes of making
environmental policies.
E-69
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
Aber, J.D., Nadelhoffer, K.J., Steudler, P., and Melillo, J.M., 1989. Nitrogen Saturation in Northern Forest
Ecosystems. BioScience, 39(6): 378-386.
Arrow, K.J. 1968. Optimal capital policy and irreversible investment. (In ed. J.N. Wolfe) Value, Capital and
Growth" Aldine: Chicago.
Arrow, K.J. and A.C. Fisher. 1974. Environmental preservation, uncertainty, and irreversibility. Quarterly
Journal of Economics 88: 312-319.
Asman, W.A.H., and S.E. Larsen. 1996. Atmospheric Processes, (Ch. 2) Eutrophication in Coastal Marine
Ecosystems Coastal and Estuarine Studies, 52: 21-50. American Geophysical Union.
Ayers, H., Hager,}., and Little, C.E., 1997. An Appalachian Tragedy. Air Pollution and Tree Death in the Eastern
Forests of North America. (Eds. Ayers, H., Hager,}., and Little, C.E.) Sierra Club Books:, San
Francisco.
Baden, S.P., L.O. Loo, L. Pihl, and R. Rosenberg. 1990. Effects of eutrophication on benthic communities
including fish: Swedish West Coast. Ambio 19(3): 113-123.
Baggetta, A.M. 1998. List of Dioxin-Producing Industry Sectors Gets Tentative OK From Peer Review
Panel. EnvironmentalReporter 29(7):377'-37'8. }une 12.Bartell, S.M., R.H. Gardner, and R.V O'Neill.
1992. Ecological Risk Estimation. Lewis Publishers: Chelsea, MI.
Bell, J.D., and D.A. Pollard. 1989. Ecology offish assemblages and fisheries associated with seagrasses.
Aquatic Plant Studies Vol. 2: 565-609. (Eds. A.W.D. Larkum, A}. McComb, and S.A. Shepherd)
Elsevier: Amsterdam.
Birdsey, R.A., 1992a. Carbon storage and accumulation in United States forest ecosystems. Gen. Tech. Report WO-59.
Washington, D.C.: U.S. Department of Agriculture, Forest Service.
Birdsey, R.A., 1992b. Changes in forest carbon storage from increasingforest area and timber growth. In: Forest and
Global Change. Volume One: Opportunities for Increasing Forest Cover. (R.N. Sampson and D.
Hair, eds.) Washington, DC 1992. p.23-39.
Birdsey, R.A., and L.S. Heath, 1995. Carbon changes in U.S. forests. In: Productivity of America's Forests and
Climate Change (Joyce, L.A., ed.), U.S. Department of Agriculture, Forest Service, General Technical
Report RM-271.
Binkley, D., T.D. Droessler, and}. Miller, 1992. Pollution impacts at the stand and ecosystem level. (In eds.
R.K.Olson, D. Binkley, and M. Boehm) The Response of Western Forests to Air Pollution,. Ecological
Studies 97. Springer-Verlag: New York, p. 235-258.
Black, F, and M. Scholes. 1973. The Pricing of Options and Corporate Liabilities. Journal of Political Economy,
81: 637-659.
Boesch, D.F. 1997. "The Cambridge Consensus" Forum on Land-Based Pollution and Toxic
Dinoflagellates in Chesapeake Bay. W.H. Bell and D.A. Nemazie, Rapporteurs.
E-70
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Boesch, D.F., D.M. Anderson, R.A. Homer, S.E. Shumway, P.A. Tester, and T.E. Whitledge. 1997.
"Harmful algal blooms in coastal waters: Options for prevention, control and mitigation" An
assessment conducted for the National Fish and Wildlife Foundation and The National Oceanic and
Atmospheric Administration Coastal Ocean Program.
Bonsdorff, E., E.M. Blomquist, J. Mattila, and A. Norkko. 1997. Coastal eutrophication: Causes,
consequences, and perspectives in the archipelago areas of the northern Baltic Sea. Estuanne Coastal
& Shelf Science 44(Suppl A): 63-72.
Boring, L.R., W.T. Swank, J.B. Waide, and G.S. Henderson. 1988. Sources, fates, and impacts of nitrogen
inputs to terrestrial ecosystems: review and synthesis. Eiogeochemistry 6: 119-125.
Boynton, W.R., J.H. Garber, R. Summers, and W.M. Kemp. 1995. Inputs, transformations, transport of
nitrogen and phosphorus in Chesapeake Bay and selected tributaries. Estuaries 18: 285-314.
Brown, T.C., 1987. Production and cost of scenic beauty: Examples for a ponderosa pine forest. Forest
Science 33(2): 394-410.
Brown, T.C.and T.C. Daniel, 1984. Modeling Forest scenic beauty: Concepts and applications to ponderosa
pine. USDA Forest Service Research Paper RM-256. Rocky Mountain Forest and Range
Experiment Station, Fort Collins, Colorado.
Bruck, R.I., Robarge, W.P., and McDaniel, A., 1989. Forest decline in the boreal montane ecosystems of the
southern Appalachian Mountains. Water, Air, and Soil Pollution. 48: 161-180.
Buhyoff, GJ. and J.D. Wellman, 1980. The specification of a non-linear psychophysical function for visual
landscape dimensions./. Leisure Res 12(3): 257-272.
Buhyoff, G.J., J.D. Wellman, and T.C. Daniel, 1982. Predicting scenic quality for mountain pine beetle and
western spruce budworm damaged forests. Forest Science 28(4): 827-838.
Buhyoff, G.J., R.B. Hull IV,J.N. Lien, and H.K. Cordell, 1986. Prediction of scenic quality for southern pine
stands. Forest Science 32(3): 769-778.
Buyoff, G. J., and W. A. Leuschner. 1978. Estimating psychological disutility from damaged forest stands.
Forest Science, 28(3).
Buhyoff, G.J., W.A. Leuschner, and J.D. Wellman, 1979. Aesthetic impacts of southern pine beetle damage.
Journal of Environmental Management 8:261-267.
Buhyoff, G. J., and J. D. Wellman, 1980. The specification of a non-linear psychological function for visual
landscape dimensions. Journal of Eeisure Research, 12(3).
Burkholder, J.M., H.B. Glasgow Jr., C.W. Hobbs. 1995. Fish kills linked to a toxic ambush-predator
dinoflagellate: distribution and environmental conditions. Marine Ecology Progress Series 124: 43-61.
Burkholder, J.M., K.M. Mason, H.B. Glasgow, Jr. 1992. Water-column nitrate enrichment promotes decline
of eelgrass Zostera marina, evidence from seasonal mesocosm experiments. Marine Ecology Progress
TO-81: 163-178.
E-71
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Burkholder, J.M., G.B. Glasgow, and J.E. Cooke. 1994. Comparative effects of water column nitrate
enrichment on eelgrass Zostera marina, shoalgrass Halodule wrightii, and widgeongrass Ruppia
Marine Ecology Progress Series 105:121-138.
Bytnerowicz, A., and N.E. Grulke. 1992 . Physiological effects of air pollutants on western trees. (In eds.
R.K. Olson, D. Binkley, and M. Boehm) The Response of Western Forests to Air pollution. Ecological
Studies 97. Springer-Verlag: New York, p. 183-234.
Camacho, Rodolfo, Chesapeake Bay Program Nutrient Reduction Strategy Reevaluation, Financial Cost Effectiveness of
Point and Nonpoint Source Nutrient Reduction Technologies in the Chesapeake Basin, December 1992.
Canela, M.C. and W.F. Jardim. 1997. The Fate of Hg° in Natural Waters. /. Bra??. Chem. Soc. 8(4): 421-426.
Carroll, G. 1998. Are our coastal waters turning deadly. National Wildlife April/May.42-46.
Chesapeake Bay Program, 1997 Nutrient Reduction Reevaluation Summary Report, obtained online at
www.chesapeakebay.net/bayprogram/pubs/97rpt.
Chichilnisky, G and G. Heal. 1998. Economic Returns form the Biosphere. Nature 392: 629-30.
Church, M.R., K.W. Thornton, P.W. Shaffer, D.L. Stevens, B.P. Rochelle, G.R. Holdren, M.G.Johnson, JJ.
Lee, R.S. Turner, D.L. Cassell, D.A. Lammers, W.G. Campbell, C.I. Liff, C.C Brandt, L.H. Liegel,
G.D. Bishop, D.C. Mortenson, S.M. Pierson, and D.D. Schmoyer. 1989. Direct/'Delayed'Response
Project: Future Effects ofEong-term Sulfur Deposition on Surface Water Chemistry in the Northeast and Southern
Blue Ridge Province. EPA/600/3-89/026a-d, U.S. Environmental Protection Agency, Washington,
DC.
Church, M.R., P.W. Shaffer, K.W. Thornton, D.L. Cassell, C.I. Liff, M.G. Johnson, D.A. Lammers,JJ. Lee,
G.R. Holdren, J.S. Kern, L.H. Liegel, S.M. Pierson, D.L. Stevens, B.P. Rochelle, and R.S. Turner.
1992. Direct/Delayed Response Project: Future Effects ofEong-term Sulfur Deposition on Stream Chemistry in the
Mid-Appalachian Region of the Eastern United States. EPA/600/R-92/186, U.S. Environmental
Protection Agency, Washington, DC.
Coastlines. 1994. Seagrasses as a primary indicator of water quality, obtained online,
http://www.epa.gov/docs/OWOW/estuaries/coastlines/fall94/seagrasses.html(June 1998).
Coggins, J.S. and C.A. Ramezani. 1998. An Arbitrage-Free Approach to Quasi-Option Value. Journal of
Environmental Economics and Management35: 103-125.
Connecticut Department of Environmental Protection. 1998. Nitrogen Removal Program: Eong Island Sound.
Connelly, N.A., B.A. Knuth and T.L. Brown. 1996. Sportfish Consumption Patterns of Lake Ontario Anglers
and the Relationship to Health Advisories. North American Journal of Fisheries Management 16: 90-101.
Cook, E.R., and Zedaker, S.M., 1992. The dendroecology of Red Spruce decline. (In eds. C. Eagar, and M.B.
Adams) Ecology and Decline of Red Spruce in the Eastern United States, Ecological Study 96. Springer-
Verlag: New York, p.192-234.
E-72
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Correll, D.L., and D. Ford. 1982. Comparison of precipitation and land runoff as sources of estuarine
nitrogen. Estuarine Coastal and Shelf Science 15: 45-56.
Cosby, B.J., G. Hornberger, J. Galloway. 1985a. Time scales of catchment acidification: a quantitative model
for estimating freshwater acidification. Environmental Science and Technology 19 (1144-1149).
Cosby, B.J., G.M. Hornberger, J.N. Galloway, and R.F. Wright. 1985b. Modeling the effects of acid
deposition: Assessment of a lumped parameter model of soil water and streamwater chemisty. Water
Resources Research 21: 51-63.
Costa, J.E., B.L. Howes, A.E. Giblin, and I. Valiela. 1992. Monitoring nitrogen and indicators of nitrogen
loading to support management action in Buzzards Bay. p. 499-431; (In eds. DH McKenzie, DE
Hyatt, and VJ McDonald) Ecological Indicators. Elsevier Applied Science:: New York.
Costanza, R., L. Wainger, C. Folke, K. Maler. 1993. Modeling Complex Ecological Economic Systems.
BioScience 43: 545-555.
Crocker, T. D., 1985. On the value of the condition of a forest stock. Land Economics, 61(3).
Daily, G. 1997. Nature's Services: Societal Dependence on Natural Ecosystems. Island Press, Washington, DC.
Daily, G., P. Matson and P. Vitousek. 1997. Ecosystem services supplied by soil. (In ed. G. Daily) Nature's
Services: Societal Dependence on Natural Ecosystems. Island Press: Washington, DC.
Dame, R.F. 1993. Bivalve filter feeders in estuarine and coastal ecosystem processes, Vol. G33. Springer Verlag: Berlin.
Dame, R.F., J.D. Spurrier, and T.G. Wolaver. 1989. Carbon, nitrogen, and phosphorous processing by an
intertidal oyster reef. Marine Ecology Progress Series 54:249-256.
Dame, R.F. R.G. Zingmark, and E. Haskin. 1984. Oyster reefs as processors of estuarine materials. Journal
of Experimental Marine Biology and Ecology 83: 239-247.
Dame, R.F., R.G. Zingmark, L.H. Stevenson and D. Nelson. 1980. Filter feeder coupling between the
estuarine water column and benthic subsystems (In ed. V.C. Kennedy) Estuarine Perspectives.
Academic Press: New York, pp. 521-526.
Day, J.W., A.S. Hall, W.M. Kemp, and A. Yanez-Arancibia. 1989. Estuarine Ecology. Wiley-Interscience: New
York.
DeHayes, D.H., 1992. Winter injury and development of cold tolerance of Red Spruce. (In eds. C. Eagar and
M.B. Adams) Ecology and Decline of Red Spruce in the Eastern United States, Ecological Studies 96.
Springer Verlag: New York, p. 295-337.
Delaware Bays NEP. 1996 Personal communication with National Oceanic and Atmospheric Association.
As cited in Valigura et al.
De Steiguer, J., J. Pye, C. Love. 1990. Air Pollution Damage to U.S. Forests. Journal of Forestry, Aug 90, p. 17-
22, 1990.
E-73
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Desvousges, W.H., V. Kerry Smith, and A. Fisher. 1987. Option Price Estimates for Water Quality
Improvements: A Contingent Valuation Study for the Monongahela River. Journal of "Environmental
Economics and Management 14: 248-267.
Diana, S.C., C. A. Bisogni, K. L. Gall. 1993. Understanding Anglers Practices Related to Health Advisories
for Sport-Caught Fish. Journal of Nutrition Education 25(6): 320-328.
Dixit, A.K and R.S. Pindyck. 1994. Investment Under Uncertainty. Princeton University Press: Princeton, NJ.
Doering, P.H. 1989. On the contribution of the benthos to pelagic production. Journal of Marine "Research 47:
371-383.
Drepper, F.R. and B.A. Mansson. 1993. Intertemporal valuation in an unpredictable environment. Ecological
Economics 7: 43-67.
Duarte, C.M. 1995. Submerged aquatic vegetation in relation to different nutrient regimes. Ophelia 41:87-112.
Eagar, C., and M.B. Adams, 1992. Ecology and Decline of "Red Spruce in the Eastern United States. Ecological
Studies 96. Springer Verlag: New York, 1992.
Elks, R.D., Director of Water Resources, Greenville Utilities Commission, Greenville, NC.
Englin, J.E., T.A. Cameron, R.E. Mendelsohn, G.A. Parsons and S.A. Shankle. 1991. Valuation of Damages to
Recreational Trout Fishing in the Upper Northeast Due to Acidic Deposition, Prepared for the National Acidic
Precipitation Assessment Program, Washington, D.C. by Pacific Northwest Laboratory. PNL-7683.
Engstrom, D.R. and E.B. Swain. 1997. Recent Declines in Atmospheric Mercury Deposition in the Upper
Midwest. Environ. Sci. Technol. 31:960-967.
Faucheux, S., G. Froger, G. Munda. 1997. "Toward an integration of uncertainty, irreversibility, and
complexity in environmental decision making." In eds. J. van den Bergh, J. van der Straaten, Economy
and Ecosystems in Change, Cheltenham, UK: Edward Elgar.
Federal Register. 1998. Unified Agenda. 63(80), Book 3. April 27.
FDA (U.S. Food and Drug Administration). 1994. Action Levels for Poisonous or Deleterious Substance in
Human Food and Animal Feed. Industry Activities Staff Booklet. Obtained online,
http://vm.cfsan.fda.gov/~lrd/fdaact.html July 30, 1998.
Fisher, D., J. Ceraso, T. Mathew, and M. Oppenheimer. 1988. Polluted Coastal Waters: The Role of Acid "Rain.
Environmental Defense Fund: New York.
Fisher, D.J., and M. Oppenheimer. 1991. Atmospheric Nitrogen Deposition and the Chesapeake Bay
Estuary. Ambio 20(3-4):102-108.
Fitzgerald, W.F. 1995. Is Mercury Increasing in the Atmosphere? The Need for an Atmospheric Mercury
Network (AMNET). Water, Air, and Soil "Pollution 80:245-254.
E-74
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Flowers, P.J., HJ. Vaux, P.D. Gardner, and T. J. Mills, 1985. Changes in recreation values after fire in the northern
rocky mountains. Research Note PSW-373, USD A.
Frankel, O., A Brown, and J. Burdon. 1995. The Conservation of Plant biodiversity. Cambridge University Press:
Cambridge.
Freeman, M., 1997. On Valuing the Services and Functions of Ecosystems. (In eds. Simpson, R.D. and N.L.
Christensen, Jr.) Ecosystem Function & Human Activities. Chapman & Hall: New York.
Freeman, M., 1993. The Measurement of Environmental and Resources Values: Theory andMethods Resources for the
Future: Washington, DC.
Fogel, M.L., and H.W. Paerl. 1994. Isotopic tracers of nitrogen from atmospheric deposition to coastal
waters. Chemical Geology 107:233-236.
Folke, C., C. Holling, and C. Perrings. 1994. biological Diversity, Ecosystems andHuman Welfare. Beijer Institute,
Stockholm.
Fox, S., and R.A. Mickler, 1995. Impact of Air Pollutants on Southern Pine Forests Ecological Studies 118.
Springer Verlag: New York.
Gawel, J.E., B.A. Ahner, AJ. Fnedland & F.M.M. Morel, 1996. Role for heavy metals in forest decline
indicated by phytochelatin measurements. Nature 381, p. 64-65.
Gray, J.S. 1992. Eutrophication in the sea. (In eds. G. Colombo, I. Ferrari, VU Ceccherelli, and R Rossi)
Eutrophication and "Population Dynamics. Olsen and Olsen, Fredensburg, Denmark, pp. 3-15.
Greening, H., Science Director. 1998. Tampa Bay National Estuary Program, personal communication.
Hammitt, W. E. et al, 1994. Identifying and predicting visual preferences of southern appalachian forest
recreation vistas. Eandscape and Urban "Planning, 29 (2): 171-183.
Hanifen, J.G., W.S. Perret, R.P. Allemand, and T.L. Romaire. 1998. "Louisiana's fishery-independent data:
potential impacts of hypoxia" On-line Hypoxia Conference Proceedings.
Http://pelican.gmpo.gov/gulfweb/hypoxia/.hypoxia.html (June 1998).
Hansson S., and L.G. Rudstam. 1990. Eutrophication and Baltic fish communities. Ambio 19(3): 123-125.
Heath, L.S., and R.A. Birdsey, 1993. Carbon trends of productive temperate forests of the conterminous United States.
Water, Air, and Soil Pollution. In Press.
Heimlich, R. Forthcoming. Wetlands and Agriculture: Private Interests and Public Benefits. Office of Policy,
Economics and Statistics Administration, U.S. Department of Commerce.
Henry, C. 1974. Investment decisions under uncertainty: The irreversibility "effect". American Economic Review
64: 1006-1012.
Heywood, V, Baste, and KA Gardner. 1995. Introduction, pp.1-19. Global Biodiversity Assessment UNEP',
Cambridge University Press: Cambridge.
E-75
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Hinga, K.R., A.A. Keller, and C.A. Oviatt. 1991. Atmospheric deposition and nitrogen inputs to coastal
waters. Ambio 20(6): 256-260.
Hinga, K.R., H. Jeon, and N.F. Lewis. 1995. Marine Eutrophication Review NOAA Coastal Ocean Program
Decision Analysis Series No. 4. NOAA Coastal Ocean Office, Silver Spring, MD. Parts 1 & 2.
Hinterberger, F., G. Wegner. 1997. Limited knowledge and the precautionary principle: On the feasibility of
environmental economics. (In eds. J. van den Bergh, J. van der Straaten) Economy and Ecosystems in
Change, Cheltenham, UK: Edward Elgar.
Hollenhorst, S. J., S. M. Brock, W. A. Freimund, and M. J. Twery, 1993. Predicting the effects of gypsy moth
on near-view aesthetic preferences and recreation appeal. Forest Science, 39(1): 28-40.
Holmes, T., R. Kramer, M. Haefele. 1992. Economic Valuation of Spruce-Fir Decline in the Southern
Appalachian Mounains: A Comparison of Value Elicitation Methods. Presented at the Forestry and
the Environment: Economic Perspectives Conference, March 9-11, 1992 in Jasper, Alberta, Canada.
Howard, R.K, GJ. Edgar, and P.A. Hutchings. 1989. Fauna! assemblages of seagrass beds. (In eds. A.W.D.
Larkum, AJ. McComb, and S.A. Shepherd) biology ofSeagrasses. Aquatic Plant Studies Vol. 2..
Elsevier, Amsterdam, pp. 536-564.
Howarth, R.W. 1988. Nutrient limitation of net primary production in marine ecosystems. Annual Reviews in
Efl?%19:89-110.
Howell, P., and D. Simpson. 1994. Abundance of marine resources in relation to dissolved oxygen in Long
Island Sound. Estuaries 17(2): 394-402.
Hudson, R.J.M., S.A. Gherini, W.F. Fitzgerald, and D.B. Porcella. 1995. Anthropogenic Influences on the
Global Mercury Cycle: A Model-Based Analysis. Water, Air, and Soil Pollution 80:265-272.
Hull,]. 1997. Options, Futures, and other Derivative Securities. Third Ed., Prentice Hall, Englewood Cliffs, NJ.
HydroQual. 1996. "Water Quality Modeling Analysis of Hypoxia in Long Island Sound Using LIS 3.0" Job
Number NENGOO35. HydroQual Inc. Mahwah, NJ.
Industrial Economics (lEc). 1998a. Overview of Ecologicallmpacts ofAirPollutants Regulated by the 1990 Clean Air
Act Amendments. Prepared by Industrial Economics, Inc. for B. Heninger, EPA Office of Policy.
Industrial Economics (lEc). 1998b. Methods for Selecting Moneti^able Benefits Derived from Ecological Re sources as a
Result of Air Quality Improvements Attributable to the 1990 Clean Air Act Amendments, 1990-2010. Prepared
by Industrial Economics, Inc. for B. Heninger, EPA Office of Policy.
Industrial Economics (lEc). 1998d. Economic Benefits of Decreased Air Toxics Deposition Attributable to the 1990
Clean Air Act Amendments, 1990-2010.
Industrial Economics (lEc). 1999a. Benefits Assessment of Decreased Nitrogen Deposition to Estuaries in the United
States Attributable to the 1990 Clean Air Act Amendments, 1990-2010. Prepared by Industrial Economics,
Inc. for B. Heninger, EPA Office of Policy.
E-76
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Industrial Economics (lEc). 1999b. Economic Benefits Assessment of Decreased Acidification of Fresh Water Lakes and
Streams in the United States Attributable to the 1990 Clean Air Act Amendments, 1990-2010. Prepared by
Industrial Economics, Inc. for B. Heninger, EPA Office of Policy.
Industrial Economics (lEc). 1999c. Characterising the Forest Aesthetics Benefits Attributable to the 1990 Clean Air
Act Amendments, 1990-2010. Prepared by Industrial Economics, Inc. for B. Heninger, EPA Office of
Policy.
Industrial Economics (IEc), 1999d. Prospective Carbon Sequestration Benefits of the 1990 Clean Air Act Amendments
(CAAA), 1990-2010. Prepared by Industrial Economics Inc. for B. Heninger, EPA Office of Policy.
Industrial Economics (IEc), 1999e. Characterising the Commercial Timber Benefits from Tropospheric O^pne Reduction
Attributable to the 1990 Clean Air Act Amendments, 1990-2010. Prepared by Industrial Economics Inc.
for B. Heninger, EPA Office of Policy.
IRIS. 1997. Integrated Risk Information System. U.S. Environmental Protection Agency, Washington, DC.
http://www.epa.gov/ins/ July 30,1998Jacobs, R.P.W.M., D. Hartog, B.F. Braster, and F.C.
Carriere. 1981. Grazing of the seagrass Zostera no/tiibj birds at Terschelling (Dutch Wadden Sea).
Aquatic Botanylfr. 241-259.
Jacobson, L.L. and A.C. Hill, 1970. Recognition of air pollution injury to vegetation: A pictorial atlas. Air Pollution
Control Association: Pittsburgh.
Jakus, P.M., M. Downing, M. Bevelhimer, J.M. Fly. 1997. Do Sportfish Advisories Affect Reservoir Angler's
Site Choice? Agricultural and Resource Economics Review 26 (2).
Jaworski N.A., R.W. Howarth, and LJ. Hetling. 1997. Atmospheric deposition of nitrogen oxides onto the
landscape contributes to coastal eutrophication in the northeast United States. Environmental Science
and Technology 31: 1995-2004.
Jenkins, A., P.G. Whitehead, BJ. Cosby and H.J.B. Birks. 1990. Modeling long-term acidification: a
comparison with diatom reconstructions and the implications of reversibility. Phil. Trans. R. Soc.
Eondon 5327(433-440).
Johansson, J.O.R. 1997. Seagrass in Tampa Bay: Historic Trends and future expectations. (In ed. S.F. Treat)
Tampa Bay Area Scientific Information Symposium (Tampa B-A-ST-S) 3: Applying Our Knowledge, Tampa Bay
Regional Planning Council.
Johansson, J.O.R. and R.R. Lewis III. 1992. Recent improvements in Hillsboro Bay, a highly impacted
subdivision of Tampa Bay, Florida, USA. Science of the To tal Environment Supplement 1992. Elsevier
Science BV, Amsterdam.
Johnson, A.H., S.B. McLaughlin, M.B. Adams, E.R. Cook, D.H. DeHayes, C. Eagar, I.J., Fernandez, D.W.
Johnson, RJ. Kohut, V.A. Mohnen, N.S. Nicholas, D.R. Peart, G.A. Schier, and P.S. White. 1992.
Synthesis and conclusions from epidemiological and mechanistic studies of Red Spruce decline. (In
eds. C. Eagar and M.B. Adams) Ecology and Decline of Red Spruce in the Eastern United States. Ecological
Studies 96. Springer Verlag New York, p. 385-411.
E-77
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Johnson, D.W., and G.E Taylor, Jr., 1989. Role of air pollution in forest decline in eastern North America..
Water, Air, and Soil Pollution. 48: 21-43.
Johnson, D.W. and Fernandez, I.J., 1992. Soil mediated effects of atmospheric deposition on eastern U.S.
spruce-fir forests. (In eds. C. Eagar and M.B. Adams) Ecology and Decline of Red Spruce in the Eastern
United States, Ecological Studies 96. Springer Verlag: New York, p. 235-270.
Johnson, J.D., A.H. Chappelka,F.P. Hain, and A.S. Heagle, 1995. Interactive effects of air pollutants with
abiotic and biotic factors on southern pine forests. (In eds. S. Fox, and R.A. Mickler) Impact of Air
Pollutants on Southern Pine Forests, Ecological Studies 118. Springer Verlag: New York, p. 281-314.
Johnson, J.E., W. D. Heckathorn, Jr., and A.L. Thompson. 1996. Dispersal and Persistence of 2,3,7,8-
Tetrachlorodibenzo-^dioxin (TCDD) in a Contaminated Aquatic Ecosystem, Bayou Meto,
Arkansas. Trans. Amer. Fisheries Soc. 125:450-457.
Jordan, T.E., D.L. Correll, J. Miklas, and D.E. Weller. 1991. Nutrients and chlorophyll at the interface of a
watershed and an estuary. Limnology and Oceanography 36(2): 251-267.
Karr, J.R. and D.R. Dudley. 1981. Ecological perspective on water quality goals. Environmental Management 5:
55-68.
Kauffman,J. 1980. Effect of a Mercury-Induced Consumption Ban on Angling Pressure. Fisheries 5(1): 10-
12.
Kemp, WM, R.R. Twilley, W.R. Boynton, and J.C. Means. 1983. The decline of submersed vascular plants in
upper Chesapeake Bay: Summary of results concerning possible causes. Marine Technology Society
'17:78-89.
Kerr, S.R. and R.A. Ryder. 1992. Effects of cultural eutrophication on coastal marine fisheries: a
comparative approach. Science of the Total Environment Suppl. 599-614.
Kittel, T.G.F., N.A. Rosenbloom, T.H. Painter, D.S. Schimel, H.H. Fisher, A. Grimsdell, VEMAP
Participants, C. Daly, and E.R. Hunt, Jr. 1996. The VEMAP Phase I Database: An Integrated Input
Dataset for Ecosystem and Vegetation Modeling for the Conterminous United States. CD ROM and
World Wide Web (URL=http://www.cgd.ucar.edu/vemap/).
Leuscher, W.A. and R. L. Young., 1978. Estimating the southern pine beetle's impact on reservoir campsites.
Forest Science, 24(4).
Levin, L. 1997. Risk Approaches for Water-Borne Exposure to Atmospherically Deposited Trace
Substances. The Environmental Professional 19:43-47.
Levin, S.A., M.A. Harwell, J.R. Kelly, and K.D. Kimball. 1989. Eco toxicology: Problems and Approaches, Springer
Verlag, N.Y.
Leffler, M. 1997. Harmful algal blooms on the move. Maryland Marine Notes 15(4).
Linker, L. 1997. Using RADM and water quality models of Chesapeake Bay. Atmospheric Deposition of Pollutants
to the Great Waters, SETAC Publication.
E-78
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Limburg, R.H., S.A. Levin, and C.C. Harwell. 1989. Ecology and environmental impact assessment: lessons
learned from the Hudson River (USA) and other estuarine experiences.
'22:255-280.
Long Island Sound Study Program. 1994. The Comprehensive Conservation and Management Plan, The
Long Island Sound Study.
Long Island Sound Study. 1993. Hypoxia and Nutrient Enrichment — Assessment of Conditions and
Management Recommendations.
Lucotte, M., A. Mucci, C. Hillaire-Marcel, P. Pichet, and A. Grondin. 1995. Anthropogenic Mercury
Enrichment in Remote Lakes of Northern Quebec (Canada). Water, Air and Soil Pollution 80:467-476.
Mason, R.P., W.F. Fitzgerald, and F.M.M. Morel. 1994. The Biogeochemical Cycling of Elemental Mercury:
Anthropogenic Influences. Geochim. et Cosmochim. Acta 58(15): 3191-3198.
MacDonald, H.F. and K.J. Boyle. 1997. Effect of Statewide Sport Fish Consumption Advisory on Open-
Water Fishing in Maine. North American Journal of Fisheries Management 17: 687-695.
Massachusetts Bays National Estuaries Program. 1996. Comprehensive Conservation and Management Plan,
and Program Fact Sheet No. 6.
McBnde, J.R., P.R. Miller, and R.D. Laven. 1985. Effects of Oxidant Air Pollutants on Forest Succession in
the Mixed Conifer Forest Type of Southern California. In: Air Pollutants Effects On Forest Ecosystems,
Symposium Proceedings, St. P., p. 157-167.
McMahon, G., M.D. Woodside. 1996. Nutrient mass balance for the Albemarle-Pamlico drainage basin,
North Carolina and Virginia, \99Q.Journal of the American Water Resources Association 33(3): 573-590.
McLaughlin, S.B., Andersen, C.P., Hanson, P.J., Tjoelker, M.G., and Roy, W.K. 1991. Increased dark
respiration and calcium deficiency of red spruce in relation to acid deposition at high elevation
southern Appalachian mountain sites. Can. J. for Res. 21: 1234-1244.
Michaels, A.F., D. Olson, J.L. Sarmiento, J.W. Ammerman, K. Fanning, R. Jahnke, A.H. Knap, F. Lipschultz,
andJ.M. Prospero. 1996. Eiogeochemistty 35(1): 181-226.
Miller, P.R., 1973. Oxidant-induced community change in a mixed conifer forest. Advances in Chemistry Series
122: 101-117.
Miller, P.R., 1983. O^pne effects in the San Bernardino National Forest. In: Proceedings: Air pollution and the
Productivity of the Forest. Izaak Walton League and Penn State University, pp 161-197.
Miller, P.R., 1992. Mixed Conifer Forests of the San Bernardino Mountains. (In eds.R.K.Olson, D. Binkley,
and M. Boehm) The Response of Western Forests to Air Pollution Ecological Studies 97. Springer-Verlag:
New York, p. 461-500.
Miller, E.K., and AJ. Friedland. 1994. Lead Migration in Forest Soils: Response to Changing Atmospheric
Inputs. Environmental Science & Technology 28(4):662-669.
E-79
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Miller, P.R., and J.R. McBnde, 1975. Effects of air pollutants on forests. (In eds. J.B. Mudd, and T.T.
Kozlowski) Responses of Plants to Air Pollution. Academic Press: New York, p.196-235.
Miller, P.R., and J.R. McBride, 1998. Air pollution impacts in the montane forests of southern California: The San
Bernardino case study. (DRAFT). Publication expected in: Ecological Study Series 134. (P.R. Miller,
and J.R. McBride, eds.) Springer Verlag: New York.
Miller, P.R., O.C. Taylor, R.G. Wilhour, 1982. OxidantAir Pollution Effects on a Western Coniferous Forest
Ecosystem. Corvallis, OR: U.S. Environmental Protection Agency, Environmental Research
Laboratory; EPA-600/D-82-276.
Montgomery, M. and M. Needelman. 1997. The Welfare Effects of Toxic Contamination in Freshwater Fish.
Land Economics 13(2): 211-23.
Morey, E.R. and W.D. Shaw. 1990. An Economic Model to Assess the Impact of Acid Rain: A
Characteristics Approach to Estimating the Demand for and Benefits from Recreational Fishing.
Advances in Micro-Economics 5: 195-216.
Mullen, J.K., and F.C. Menz. 1985. The Effects of Acidification Damages on the Economic Value of the
Adirondack Fishery to New York Anglers. American Agricultural Economics Association, Feb: 112-119.
Musser, W. N., R. Ziemer, and F. C. White, 1982. Trade-offs between nonmarket and market land use: Crop
production, forestry and outdoor recreation. Research Bulletin 270, The University of Georgia College of
Agricultural Experiment Stations.
Nabhan, G., S. Buchmann. 1997. "Services Provided by Pollinators." In ed. G. Daily. Nature's Services: Societal
Dependence on Natural Ecosystems. Island Press, Washington, DC.
NAPAP, 1987. Diagnosing Injuries to Eastern Forest Trees. National Acid Precipitation Assessment Program.
Forest Response Program. National Vegetation Survey.
NAPAP, 1991. National Acid Precipitation Assessment Program. 1990 Integrated Assessment Report. National Acid
Precipitation Program. Office of the Director, Washington DC.
National Research Council. 1993. Managing Wastewater in coastal Urban Areas: Committee on Wasteivater
Management for Coastal Urban Areas, National Academy Press: Washington, DC.
Naylor, R., P. Ehrlich. 1997. Natural Pest Control Services and Agriculture. (In ed. G. Daily) Nature's Services:
Societal Dependence on Natural Ecosystems. Island Press, Washington, DC.
NCLAN. 1988. Assessment of Crop Loss from Air Pollutants. (Eds. Walter W. Heck, O. Clifton Taylor and
David T. Tmgey) Elsevier Science Publishing Co.: New York,. Pp. 1-5. (ERL,GB 639).
NOAA, U.S. Department of Commerce. 1993. Natural Resource Damage Assessments Under the Oil
Pollution Control Act of 1990, 57 FR 23067, June 1, 1992.
NOAA. 1998. Reporting on the State of Our Coasts, http://www.enn.com/enn-news
archive/1998/02/021398/coastrpt.asp (Feb, 1998).
E-80
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
NESCAUM (Northeast States and Eastern Canadian Provinces). 1998. Mercury Study: A Framework for Action.
New York Statewide Angler Survey 1996. Report 2: Angler Preferences, Satisfaction, and Opinion on Management
Issues. New York State, Department of Environmental Conservation. Division of Fish, Wildlife and
Marine Resources. Albany, New York. April, 1998.
Newell, R.I.E. 1988. Ecological changes in Chesapeake Bay: Are they a result of over-harvesting the
American Oyster, Crassostrea virginicd? In Understanding the Estuary: Advances in Chesapeake Bay Research
Conference Proceedings, Chesapeake Bay Consortium, Baltimore, MD.
Nixon, S.W. 1990. Marine eutrophication: a growing international problem. Ambio 19(3): 101.
Nixon, S.W. 1995. Eutrophication: a definition, social causes, and future concerns. Ophelia 41: 199-220.
Nixon, S.W., L.S. Granger, D.I. Taylor, P.W. Johnson, and B.A. Buckley. 1994. Subtidal volume fluxes,
nutrient inputs and the brown tide — an alternate hypothesis. Estuarine, Coastal and Shelf Science 39:
303-312.
Nixon, S.W., C.A. Oviatt, J. Frithsen, and B. Sullivan. 1986. Nutrients and the productivity of estuarine and
coastal marine ecosystems. Journal of the Umnological Society of South Africa 12(1/2): 43-71.
Norton, S.B, DJ. Rodier, J.H. Gentille, W.H. van der Schalie, W.P. Wood, and M.W. Slimack. 1992. A
framework for ecological risk assessment at the EPA. Environmental Toxicology and Chemistry 11: 1663-
1672.
Norwood, R. and P. Stacey. 1998. Connecticut Department of Environmental Protection, Personal
Communication.
Nriagu, J.O. and J.M. Pacyna. 1988. Quantitative Assessment of Worldwide Contamination of Air, Water,
and Soils by Trace Metals. Nature 333:134-139.
Ollinger, S.B., Aber. J.D., and Reich, P.B. 1997. Simulating ozone effects on forest productivity: interactions
among leaf-, canopy-, and stand-level processes. Ecological Applications 7(4): 1237-1251.
Olson, R.K., D. Bmkley, and M.Boehm, 1992. The Response of Western Forests to Air Pollution (R.K. Olson, D.
Bonkley, and M. B ohm, eds.). Ecological Studies 97. Springer Verlag: New York.
Orth, RJ. and K.A. Moore. 1983. Chesapeake Bay: An unprecedented decline in submerged aquatic
vegetation. Science 22:51-53.
Orth, R.J., M. Luckenback, and K.A. Moore. 1994. Seed dispersal in a marine macrophyte: implications for
colonization and restoration. Ecology 75(7): 1927-1939.
Orth, R.J., R.A. Batiuk, andJ.F. Nowack. 1994. Trends in the distribution, abundance, and habitat quality of submerged
aquatic vegetation in Chesapeake Bay audits tidal tributaries: 1971 -1991. USEPA for Chesapeake Bay
Program. Annapolis, Maryland. EPA 903-R-95-009.
E-81
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Oviatt, C.A., P.H. Doering, B.L. Nowicki, and A. Zoppini. 1993. Net system production in coastal waters as
a function of eutrophication, seasonally, and benthic macrofaunal abundance. Estuaries 16(2): 247-254.
Oviatt, C.A., A.A. Keller, P.A. Sampou, and L.L. Beatty. 1986. Patterns of productivity during eutrophication:
a eutrophication experiment. Marine Ecology Progress Series 28: 69-80.
Patwardhan, A.S. and A.S. Donigian, Jr. 1997. Assessment of nitrogen loads to aquatic systems. EPA Project
Summary USEPA. EPA/600/SR-95/173.
Paerl, H.A. 1993. Emerging role of atmospheric nitrogen deposition in coastal eutrophication:
biogeochemical and trophic perspectives. Canadian Journal of Fisheries and Aquatic S cience 50: 2254-
2269.
Paerl, H.A. and J.L. Pinkney. 1997. Hypoxia, anoxia, and fish kills in relation to nutrient loading in the Neuse
River Estuary: Why was 1995 a 'bad' year? Unpublished.
Paerl, H.W. 1997. Coastal eutrophication and harmful algal blooms: importance of atmospheric deposition
and groundwater as 'new' nitrogen and other nutrient sources, limnology and Oceanography 42(5, part
2): 1154-1165.
Paerl, H.W., C. Aguilar, and M.L. Fogel. 1997. Atmospheric nitrogen deposition in estuarine and coastal
waters: biogeochemical and water quality impacts. Chapter 22 (Id ed. J.E. Baker) Atmospheric
Deposition of Contaminants to the Great Lakes and Coastal Waters, SET AC 15th Annual Meeting
Proceedings 30 Oct.-3 Nov. 1994. SET AC Press, Pensacola, FLA.
Paerl, H.W. and M.L. Fogel. 1994. Isotopic characterization of atmospheric nitrogen inputs as sources of
enhanced primary production in coastal Atlantic Ocean waters. Marine Biology 119: 635-645.
Paerl, H.W.,J. Rudek, and M.A. Malin. 1990. Simulation ofphytoplankton production in coastal waters by
natural rainfall inputs: nutritional and trophic implications. Marine Biology 107: 247-254.
Pearl, H.W. 1988. Nuisancephytoplankton blooms in coastal, estuarine, and inland waters. Limnology and
Oceanography 33(4, part 2): 823-847.
Paquet, J., and L. Belanger, 1997. Public acceptability thresholds of clear cutting to maintain visual quality of
boreal balsam fir landscapes. Forest Science, 43(1): 46-55.
Peart, D.R., N.S. Nicholas, S.M. Zedaker, M.M. Miller-Weeks, and T.G. Siccama, 1992. Condition and recent
trends in high-elevation Red Spruce populations. (In eds. C. Eagar, and M.B. Adams) Ecology and
Decline of Red Spruce in the Eastern United States Ecological Study 96.p.125-191.
Peine,J.D.,J.C. Randolph, andJJ. Presswood,Jr. 1995. Evaluating the Effectiveness of Air Quality
Management within the Class I Area of Great Smokey Mountains National Park,.
Management 19(4): 515-526.
Pelley, J. 1998. What is causing toxic algal blooms? Environmental Science and Technology 26A-30-A.
E-82
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Peterson D.L., and MJ. Arbaugh, 1992. Coniferous forests of the Colorado front Range. Part B: Ponderosa
Pine second-growth stands. (In eds. R.K.Olson, D. Binkley, and M. Boehm) The Response of Western
"Forests to Air pollution Ecological Studies 97. Springer-Verlag: New York, p. 433-460.
Peterson, D. G., et al., 1987. Improving Accuracy and Reducing Cost of Environmental Benefit Assessments. Draft
report to the U.S. Environmental Protection Agency, by Energy and Resource Consultants, Boulder,
CO.
Petersen, G., J. Munthe, and R. Bloxam. 1995. Numerical Modeling of Regional Transport, Chemical
Transformations and Deposition Fluxes of Airborne Mercury Species. (In eds. Baeyerns, W., R.
Ebinghaus, O. Vasilev). Global and Regional Mercury Cycles: Sources, Fluxes and Mass Balances. NATO
Advanced Science Institute (ASI) Series, sub-series 2. Environment, Vol. 21. Kluwer Academic
Publishers: Dordrect, The Netherlands, p. 191-217.
Pimentel, D., C. Wilson, C. McCullum, R. Huang, P. Dwen, J. Flack, Q. Tran, T. Saltman, and B. Cliff.
Economic and Environmental Benefits of Biodiversity. BioScience'M (11): 747-57.
Porcella, D.B., J.W. Huckabee, and B. Wheatley, eds. 1995. Mercury as a Global Pollutant: Proceedings of
the Third International Conference held in Whistler, British Columbia, July 10-14,1994. Reprinted
from Water, Air, and Soil Pollution 80(1-4). Kluwer Academic Publishers: Dordrecht, The
Netherlands.
Postel, S. and S. Carpenter. 1997. Freshwater Ecosystem Services. (In ed. G. Daily) Nature's Services: Societal
Dependence on Natural Ecosystems. Island Press, Washington, DC.
Price, KS, D.A. Flemer, J.L. Taft, G.B. Mackiernan, W. Nehlsen, R.B. Biggs, N.H. Burger, D.A. Blaylock.
1985. Nutrient enrichment of Chesapeake Bay and its impact on the habitat of striped bass: A
speculative hypothesis. Transactions of the American Fisheries Society 114 (1): 97-106.
Prospero, J.M., and D.L. Savoie. 1989. Effects of continental sources on nitrate concentrations over the
Pacific Ocean. Nature 339: 687-689.
Pye, J.M., Impact of ozone on the growth and yield of trees: A review. Journal of Environmental Quality 17:347-
360,1988.
Rahel, F.J., and Magnuson, J.J., 1983. Low pH and the Absence of Fish Species in Naturally acidic Wisconsin
lakes: Inference for Cultural Acidification. Can.]. Fish. Aquat. Sci. 40: 3-9.
Randall, A., 1984. Benefit estimation for scenic and visibility services. (In eds.G.L. Peterson and A. Randall)
Valuation of Wildland Resource Benefits. Westview Press: Boulder, CO.
Regier, H.A., P. Tuunainen, Z. Russek, and L.E. Persson. 1988. Rehabilitative redevelopment of the fish and
fisheries of the Baltic Sea and the Great Lakes. Ambio 17: 121-130.
Rendell, A.R., CJ. Ottley, T.D. Jickells, and R.M. Harrison. 1993. The atmospheric input of nitrogen species
to the North Sea. Tellus 45B: 53-63.
E-83
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Research Triangle Institute (RTI). 1996. Risk Assessment Support to the Development of Technical Standards for
Emissions from Combustion Units Burning Hazardous Wastes: Background Information Document. Prepared for
Industrial Economics, Inc. EPA Contract # 68-W3-0028.
Ribe, R.G., 1990. A general model for understanding the perception of scenic beauty in northern hardwood
forests. Landscape Journal, 9(2): 86-101.
Richardson, K, 1996. Conclusion, Research, and Eutrophication Control. (In eds. BB Jorgensen and K
Richardson) Eutrophication in Coastal Marine Ecosystems Coastal and Estuarine Studies Vol. IV, American
Geophysical Union, pp. 243-269.
Richardson, K. and B.B. Jorgensen. 1996. "Eutrophication: definition, history and effects (In eds. BB
Jorgensen and K Richardson) Eutrophication in Coastal Marine Ecosystems Coastal and Estuarine Studies
Vol. IV, American Geophysical Union, pp 1-20.
Ridker, R.G. and J.A. Henning, 1967. Determinants of residential property values with special reference to
air pollution. Review of Economics and Statistics, 49: 246-257.
Rosseland, B.O. and Staurnes, M., 1994. Physiological Mechanisms for Toxic Effects and Resistance to
Acidic Water: An Ecophysiological Approach. (In eds. Steinberg and R.F. Wright) Acidification of
Freshwater Ecosystems: Implications for the Future. Chapter 16. C.E.W. John Wiley & Sons Ltd., 1994.
Rosenberg, R. 1985. Eutrophication - future marine coastal nuisance? Maine Pollution Bulletin 16:227-231.
Row, C. and R.B. Phelps. 1990. Tracing the Flow of Carbon through U.S. Forest Product Sector. Prepared
for the 19th World Congress, International Union of Forestry, Research Organizations, Montreal,
Canada, August 11, 1990.
Rowe, R., C. Lang, L. Chestnut, D. Latimer, D. Rae, S. Bernow, D. White. 1995. The New York Electricity
Externality Study, Vol. 1: Introduction and Methods. Prepared by Hagler Bailly Consulting, Inc. for the
Empire State Electric Energy Research Corporation (ESEERCO), Proj. No. EP91-50.
Rowe, C. and R.B. Phelps, 1990. Tracing the Flow of Carbon through U.S. Forest Product Sector. Presented at the
19th World Congress, International Union of Forestry Research Organizations, Montreal, Canada.
Ruddell, E. J., and J. H. Gramann, 1989. The psychological utility of visual penetration in near-view forest
scenic-beauty models. Environment and Behavior, 21(4):393-412.
Rudis, V.A., J.H. Gramann, EJ. Rudell, and J.M. Westphal, 1988. Forest inventory and management-based
visual preferences models of southern pine stands. Forest Science 34(4): 846-863.
Ryther, J.H., and W.M. Dunstan. 1971. Nitrogen, phosphorus and eutrophication in the coastal marine
environment. Science 171: 1008-1112.
Sarasota Bay National Estuarine Program notes cite Camp, Dresser, and McKee, Inc. (1992). Point-/non-
point-source pollution-loading assessment. Phase 1. Final Report to Sarasota Bay National Estuary
Program.
E-84
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Schier, G.A., and K.F. Jensen, 1992. Atmospheric deposition effects on foliar injury and foliar leaching in
Red Spruce. (In eds. C. Eagar and M.B. Adams) Ecology and Decline of Red Spruce in the Eastern United
States Ecological Studies 96. Springer Verlag: New York, p. 271-294.
Schroeder, H.W. and T.C. Daniel, 1981. Progress in predicting the perceived scenic beauty of forest
landscapes. Forest Science 21 (I}: 71-80.
Short, F.T., G.E. Jones, and D.M. Burdick. 1991. Seagrass decline: problems and solutions. (In ed. HS
Bolton) Coastal Wetlands Proceedings of Coastal Zone '91 Conference. American Society of Civil Engineers,
New York, p. 439-453.
Short, F.T., D.M. Burdick, and J.A. Kaldy III, 1995. Mesocosm experiments quantify the effects of
eutrophication on eelgrass, Zostera marina. Eimnology and Oceanography 40(4):740-749.
Short, F.T. and D.M. Burdick. 1996. Quantifying eelgrass habitat loss in relation to housing development and
nitrogen loading in Waquoit Bay, Massachusetts. Estuaries 19:730-739.
Shuyler, L. 1995. Cost Analysis for Nonpoint Source Control Strategies in the Chesapeake Basin.
Unpublished.
Sigal, L.L. and G.W. Suter, 1987. Evaluation of Methods for Determining Adverse Impacts of Air Pollution
on Terrestrial Ecosystems. Environmental Management 11: 675-694.
Smayda, TJ. 1990. Novel and nuisance phytoplankton blooms in the sea: evidence for a global epidemic (In
eds. E. Graneli) Toxic Marine Phytoplankton Elsevier Science Publishers: Amsterdam, Netherlands, p.
29-40.
Smith, W.H., 1990. Air Pollution and Forests: Interaction EetmenAir Contaminants and Forest Ecosystems. 2nd
edition, Springer Verlag: New York.
Smith, V.K. 1987. Uncertainty, Benefit-Cost Analysis, and the Treatment of OptionVahie. Joarna/of
EnvironmentalEmnomics and Management 14: 283-292.
SOS/T 9, 1990. Current Status of Surface Water Acid-Ease Chemistry. Baker, L.A., Kaufmann, P.R., Ross-Todd,
B.M., and Beauchamp, JJ. National Acid Precipitation Assessment Program. State of Science and
Technology Report 9.
SOS/T 10, 1990. Watershed and Eake Processes Affecting Chronic Surface Water Acid-Ease Chemistry. Turner, R.S.,
Cook, R.B, Miegrot, H.V.Johnson, D.W, Elwood, J.W, Bncker, O.W, Lindberg, S.E,
Hornberger, G.M. National Acid Precipitation Assessment Program. State of Science and
Technology Report 10.
SOS/T 12, 1990. Episodic Acidification of Surface Waters Due to Acidic Deposition. Wigington, P.J., Davies, T.D.,
Tranter, M., and Eshleman, K.N. National Acid Precipitation Assessment Program. State-of-
Science/Technology Report 12.
E-85
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
SOS/T 13, 1990. Biological Effects of Changes in Surface Water Acid-Base Chemistry. Baker, J.P., Bernard, D.P.,
Christensen, S.W., Sale, M.J., Freda,]., Heltcher, K., Marmorek, D., Rowe, L., Scanlon, P., Suter, G.,
Warren-Hicks, W., and Welbourne, P. National Acid Precipitation Assessment Program. State of
Science and Technology Report 13.
SOS/T 16, 1990. State of Science/Technology Report 16. In: Summaries of National Acid Precipitation
Assessment Program State-of-Science/Technology Report 16.
SOS/T 18, 1990. Response of Vegetation to Atmospheric Deposition and Air Pollution. Shriner, D.S., W.W. Heck, S.B.
McLaughlin, D.W. Johnson, J.D. Joslin, and C.E. Peterson. National Acid Precipitation Assessment
Program. State-of-Science/Technology Report 18.
SOS/T 27, 1990. Methods for valuing acidic deposition and air pollution effects. National Acid Precipitation
Assessment Program. State-of-Science / Technology Report 27, Part B.
Sorensen, J.A., G.E. Glass, K.W. Schmidt, J.K. Huber, and G.R. Rapp, Jr. 1990. Airborne Mercury
Deposition and Watershed Characteristics in Relation to Mercury Concentrations in Water,
Sediments, Plankton, and Fish of Eighty Northern Minnesota Lakes. Environ. Set. Technol.
24(11):1716-1731.
Soule, M. 1991. Conservation: tactics for a constant crisis. Science 253, pp.744-749.
Stacey, Paul E., Report on Nitrogen Loads to Long Island Sound, Connecticut Department of Environmental Protection,
Draft, April 1998.
Stoddard, J.L., 1994. Long-Term Changes in Watershed Retention of Nitrogen. (In ed. Lawrence A. Baker)
Environmental Chemistry of Lakes and Reservoirs, American Chemical Society, Washington, DC, pp. 223-
284.
Stein, E.D., Y. Cohen, and A.M. Winer. 1996. Environmental Distribution and Transformation of Mercury
Compounds. Critical Reviews in Environmental Science and Technology 26(l):l-43.
Stolte, K.W., D.M. Duriscoe, E.R. Cook, and S.P. Cline. Methods of assessing responses of trees, stands and
ecosystems to air pollution. (In eds. R.K.Olson, D. Binkley, and M. Boehm) The Response of Western
Forests to Air Pollution. Ecological Studies 97. Springer-Verlag: New York, p.259-332.
Suchanek, T.H., PJ. Richerson, LJ. Holts, B.A. Lamphere, C.E. Woodmansee, D.G. Slotton, EJ. Harner,
and L.A. Woodward. 1995. Impacts of Mercury on Benthic Invertebrate Populations and
Communities within the Aquatic Ecosystem of Clear Lake, California. Water, Air and Soil Pollution
80:951-960.
Swain, E.B., D.R. Engstrom, M.E. Brigman, T.A. Henning, and P.L. Brezonik. 1992. Increasing Rates of
Atmospheric Mercury Deposition in Midcontinental North America. Science 257:784-787.
Systems Applications International, Inc. 1999. Air Quality Modeling to Support the Section 812 Prospective Analysis,
Prepared for EPA.
Tampa Bay Nitrogen Management Consortium. 1990. Tampa Bay Nitrogen Management Consortium 1995-1999
Action Plan.
E-86
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Tampa Bay National Estuary Program. 1994. Estimates of total nitrogen, total phosphorus, and total
suspended solids loadings to Tampa Bay, Florida," Technical Publication #04-94. Prepared by P.E.
Hans Zarbock, A. Janicki, D. Wade, D. Heimbuch, and H. Wilson. Coastal Environmental, Inc. St.
Petersburg, Florida, May, 1994.
Tampa Bay National Estuary Program. 1995. Submerged aquatic vegetation distribution in tributaries of
Tampa Bay," Technical Publication #08-94 Prepared by King Engineering Associates, Inc. 5010 W.
Kennedy Blvd
Tampa Bay National Estuary Program. 1996. Charting the Course: The Comprehensive Conservation and Management
Tampa Bay Estuary Program. 1998. Tampa Bay Nitrogen Management Consortium 1995-1999 Action Plan,
1998.Tar-Pamlico NSW Implementation Strategy, as revised Feb. 13, 1992.
Tar-Pamlico Association and Hydroqual. 1995. Application of a Coupled Hydrodynamic/Water
Column/Sediment Model for the Tar-Pamlico River, North Carolina.
Taylor, G.E. Jr., D.W. Johnson, and C.P. Andersen, 1994. Air pollution and forest ecosystems: A regional to
global perspective. EcologicalApplications, 4(4): 662-689.
Tedesco, M. 1998. Long Island Sound Study Program, Stamford, CT, Personal Communication.
Tingey, D.T., and G.E. Taylor. 1982. Variation in plant response to ozone: a conceptual model of
physiological events (In eds.Unsworth, M.H., Omrod, D.P.) Effects of Gaseous Air Pollution in
Agriculture and Horticulture. Butterworth Scientific: London, UK, pp. 113-138.
Tomasko, D.A., D.M. Alderson, P. Clark, J. Culter, L.K. Dixon, R. Edwards, E. Estevez, M.G. Heyl, S.
Lowrey, Y.P. Sheng, and J. Stevely. 1992. Technical synthesis of Sarasota Bay, p. 14.1-14.16 (In eds.
P. Roat, C. Ciccolella, H. Smithy, and D. Tomasko), Sarasota Bay: Framework for Action. Sarasota
Bay National Estuary Program.
Tomasko, D.A., CJ. Dawes, and M.O. Hall. 1996. The effects of anthropogenic nutrient enrichment on the
turtle grass (Thalassia testudinum) in Sarasota Bay, Florida. "Estuaries 19(2B): 448-456.
Turner, D.P.,JJ. Lee, GJ. Koerper, andJ.R. Barker. 1993. TheForest Sector'Carbon Budget of'the United States:
Carbon fools and Flux Under Alternative Policy Options. EPA/600/3-93/093, EPA Environmental
Research Laboratory, Corvallis, OR.
Turner, D.P., GJ. Koerper, M. Harmon,J. Lee. 1995. Carbon Sequestration by Forests in the United States.
Current Status and Projections to the Year 2040. Tellus 47(B): 232-239.
Turner, K., C. Perrings, C. Folke. 1997. Ecological Economics: Paradigm or Perspective. (In eds. J. van den
Bergh, J. van der Straaten) Economy and Ecosystems in Change, Edward Elgar: Cheltenham, UK.
Twilley R.R., W.M. Kemp, K.W. Staver, J.C. Stevenson, and W.R. Boynton. 1985. Nutrient enrichment of
estuarine submersed vascular plant communities 1. Algal growth and effects on production of plants
and associated communities. Marine Ecology Progress Series 23:179-191.
E-87
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Tyler, M. 1988. Contributions of Atmospheric Nitrate Deposition to Nitrate Loading in the Chesapeake Bay, Report No.
RP-1052, Versar, Inc.
United States Environmental Protection Agency. 1984. Ambient Water Quality Criteria for 2,3,7,8-
Tetrachlorodi%en%p-p-dioxin. EPA 440/5-84-007.
United States Environmental Protection Agency. 1990. Air Pollution Emission Standards and Guidelines for
Municipal Waste Combustors: Revision and Update of Economic Impact Analysis and Regulatory Impact Analysis.
EPA-450/3-91-003.
United States Environmental Protection Agency. 1992a. National Study of Chemical Residues in Fish. Vol.s
I and II. EPA 823-R-92-008a, EPA 823-R-92-008b.
United States Environmental Protection Agency. 1992b. Report on the Ecological Risk Assessment Guidelines
Strategic Planning Workshop. Risk Assessment Forum Washington, B.C., EPA/630/R-92/002.
United States Environmental Protection Agency. 1993. Interim Report on Data and Methods for Assessment of
2,3,7,8-Tetrachlorodiben^p-p-dioxin to Aquatic Eife and Associated Wildlife. EPA-600/R-055.
United States Environmental Protection Agency. 1994a. Workshop on the Use of Available Data and Methods for
Assessing the Ecological Risks of 2,3,7.8-Tetrachlorodiben^p-p-dioxin to Aquatic'Eife and Associated Wildlife.
EPA-630-R-94-002.
United States Environmental Protection Agency. 1994b. Medical Waste Incinerators - Background Information for
Proposed Standards and Guidelines: Analysis of Economic Impacts for Existing Sources. EPA-453/R-94-048a.
United States Environmental Protection Agency. 1995a. Acid Deposition Standard 'Feasibility Study Report to
Congress. EPA/430-R-95-001a.
United States Environmental Protection Agency. 1995b. National Primary Drinking Water Regulations:
Contaminant Specific Fact Sheets. EPA 811-F-95-003-T.
United States Environmental Protection Agency. 1996a. Air Quality Criteria for O^pne and Related Photochemical
Oxidants, National Center for Environmental Assessment. Office of Research and Development. U.S.
Environmental Protection Agency, NC. Vol. II, 1996.
United States Environmental Protection Agency. 1996b. Update: National Listing of Fish and Wildlife
Consumption Advisories. Fact Sheet EPA 823-F-96-006. June.
United States Environmental Protection Agency. 1997a. Listing of Fish and Wildlife Advisories - 1997.
Version 3.0 EPA 823-C-98-001.Obtained online, http://www.epa.gov/OST/fishadvice/ July 30,
1998.
United States Environmental Protection Agency. 1997b. Issue fact sheet on nutrients Report of Annual Meeting
of National Estuary Program Directors, March 1997. National Estuary Program, Washington, D.C.
United States Environmental Protection Agency. 1997c. Deposition of Air Pollutants to the Great Waters Second
art to Congress. Washington, D.C.
E-88
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
United States Environmental Protection Agency EPA, 1997d. Mercury Study Report to Congress. Office of Air
Quality Planning and Standards and Office of Research and Development, EPA -452/R-97-003.
United States Environmental Protection Agency. 1997e. Benefits of reducing deposition of atmospheric nitrogen in
estuarine and coastal waters.
United States Environmental Protection Agency. 1997f. Long Island Sound Study Proposal for Phase III Actions for
Hypoxia Management. EPA 840/R/97/001.
United States Environmental Protection Agency. 1998e. The Inventory of Sources ofDioxin in the United States.
External Review Draft. EPA/600/P-98/002a.
United States Environmental Protection Agency. 1998f, The Costs of Water Pollution Control in the Chesapeake
• Basin, Office of Water, September 30, 1998.
U.S. Fish and Wildlife Service. 1998. Database of Fishing Advisories.
United States Forest Service. 1992. Forest Health Monitoring. Summary Report. New England/Mid-Atlantic 1991.
United States Department of Agriculture Forest Service. NE\NA-INF-115-92.
United States Forest Service. 1993a. Forest Health Monitoring. Summary Report. New England'/Mid-Atlantic 1992.
United States Department of Agriculture Forest Service. NE\NA-INF-115-R93.
United States Forest Service. 1993b. Forest Health Monitoring. Northeastern Area Forest Health Report
United States Department of Agriculture Forest Service. NA-TP-03-93.
United States Forest Service. 1995a. Forest Health Monitoring. ForestHealth Assessment for the Northeastern Area
1993. United States Department of Agriculture Forest Service Northeastern Area and Northeastern
Forest Research Station. NA-TP-01-95.
United States Forest Service. 1995b. Forest Health Monitoring. Forest Health Highlights. Northeastern States
United States Department of Agriculture Forest Service.
United States Forest Service. 1996. Forest Health Monitoring. 1996 Summary Report. Northern Forest Health
Monitoring. New England/Mid-Atlantic/Lake States (DRAFT). United States Department of Agriculture
Forest Service. NE\NA-INF-115-R96.
United States Forest Service. 1997a. Forest Health Monitoring. Summary Report. Northern Forest Health Monitoring.
New England, Mid-Atlantic/Lake States 1995. United States Department of Agriculture Forest Service.
NE/NA-INF-115R-97.
United States Forest Service. 1997b. Forest Health Monitoring. Forest Health Highlights for New York and New
England States. Northeastern States. United States Department of Agriculture Forest Service.
United States Forest Service. 1998. Forest Health Monitoring. Field Methods Guide (National 1998).
United States Forest Service. National Forest Health Monitoring Program. Research Triangle Park, NC.
E-89
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
United States Forest Service. 1998. http://www.fs.fed.us/news/roads/19980121 qa.html.
Unsworth, R. E., et al., 1992. Approaches to Environmental Benefits Assessment to Support the Clean Air Act Section
812 Analysis. lEc report to the EPA Office of Policy Analysis and Review.
Valiela, I., G. Collins,]. Kremer, K. Lajtha, M. Geist, B. Seely, J. Brawley, and C.H. Sham. 1997. Nitrogen
loading from coastal watersheds to receiving estuaries: new method and its application. Ecological
Applications 1(2): 358-380.
Valiela, I.,J. Costa, K. Forman, J.M. Teal, B.L. Howes, and D. Aubrey. 1990. Transport of groundwater-
borne nutrients from watersheds and their effects on coastal waters. Biogeochemistry 10:177-97.
Valigura, R.A., W.T. Luke, R.S. Artz, and B.B. Hicks. 1996. Atmospheric nutrient input to coastal areas:
Reducing the uncertainties. NOAA Coastal Ocean Program Decision Analysis Series No. 9.
Van Sickle,}, and M.R. Church. 1995. Nitrogen Bounding Study: Methodsfor Estimating the Relative Effects of Sulfur
and Nitrogen Deposition on Surface Water Chemistry. U.S. Environmental Protection Agency.
Vaux, H.J., P.D. Gardner, and TJ. Mills, 1984. Methods for assessing the impact of fire on forest recreation. General
Technical Report PSW-79, USD A, Berkeley, CA.
Velicer, C.M. and B.A. Knuth. 1994. Communicating Contaminant Risks from Sport-Caught Fish: The
Importance of Target Audience Assessment. Risk Analysis 14(5): 833-841.
Vermaat et al. 1998. The capacity of seagrasses to survive increased turbidity and siltation: the significance of
growth form and light use. Ambio 26(8): 499-504.
Vitusek P.M., and R.W. Howarth, 1991. Nitrogen Limitation on Land and in The Sea: How Can It Occur?
Biogeochemistry 13: 87-115.
Vollenweider, R.A., R. Marchetti, and R. Viviani, 1990. Marine Coastal Eutrophication: The Response of Marine
Transitional Systems to Human Impact: Problems and Perspectives for Restoration. Elsevier: New York.
Walsh R. G., R. D. Bjonback, R. S. Aiken, and Donald H. Rosenthal, 1990. Estimating the public benefits of
protecting forest quality. Journal of Forest Management,?)®: 175-189.
Walsh, R. G., F. A. Ward, and J. P. Olienyk, 1989. Recreational demand for trees in national forests. Journal of
'/, 28: 255-268.
Watras, C.J., K.A. Morrison, and R.C. Back. 1996. Mass Balance Studies of Mercury and Methyl Mercury in
Small Temperate/Boreal Lakes of the Northern Hemisphere. (In eds. Baeyerns, W., R. Ebinghaus,
O. Vasilev) Global and'Regional'Mercury Cycles: Sources, Fluxes and Mass Balances. NATO Advanced
Science Institute (ASI) Series, sub-series 2. Environment, Vol. 21. Kluwer Academic Publishers:
Dordrect, The Netherlands, p.329-358.
White, D.H., and J.T. Seginak. 1994. Dioxins and Furans Linked to Reproductive Impairment in Wood
Ducks. /. WildLManage. 58(1):100-106.
E-90
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
White, P.S., and C.V. Cogbill, 1992. Spruce-fir forests of eastern North America. (In eds. C. Eagar and M.B.
Adams) Ecology and Decline of Red Spruce in the Eastern United States Ecological Studies 96. Springer
Verlag: New York, p. 3-39.
Williams, S.L., and M.H. Ruckleshaus. 1993. Effects of nitrogen availability and herbivory on eelgrass (Zostera
marina) and epiphytes. Ecology 74:904-918.
Winner, W.E., 1994. Mechanistic Analysis of Plant Responses to Air Pollution. Ecological Applications 4(4):
651-661.
Winner, W.E., and CJ. Atkinson. 1986. Absorption of air pollution by plants, and consequences for growth.
Trends in Ecology and Evolution 1:15-18.
World Health Organization (WHO). 1989. Environmental Health Criteria 88: Polychlorinated Dibenzo-
iioxins and Dibenzofurans. International Programme on Chemical Safety. Geneva.
World Health Organization (WHO). 1990. Environmental Health Criteria 101: Methylmercury.
International Programme on Chemical Safety. Geneva.
World Health Organization (WHO). 1998. WHO Experts Re-evaluate Health Risks from Dioxins. Press
Release WHO/45. June 3.
Wright, R.F., BJ. Cosby, R.C. Ferrier, A. Jenkins, AJ. Bulger, R. Harriman. 1994. Changes in acidification
of lochs in Galloway, southwestern Scotland, 1979-1988: The MAGIC model used to evaluate the
role of afforestation, calculate critical loads, and predict fish status. Journal of Hydrology 161 (257-285).
Zacaroli, A. 1998. Utilities Told to Monitor Mercury Levels in Coal Under EPA Requirement. Environmental
Reporter 28(48):2629-2630. April 10.
Zarbock, P.A., A. Janicki, D. Wade, D. Higmuch, and H. Wilson. 1994. Estimates of total nitrogen, total
phosphorus, and total suspended solids loadings to Tampa Bay, Florida. Prepared by Coastal
Environmental Inc., for the Tampa Bay Estuary Program.
Zimmerman, R., J. Nance, andj. Williams. 1998. On-line Hypoxia Conference Proceedings
http://pelican.gmpo.gov/ gulfweb/hypoxia/hypoxia.html (June, 1998).
E-91
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[This page left blank intentionally.]
E-92
-------
Effects of Criteria
Pollutants on
Agriculture
Introduction
One potential impact of air pollutants on
economic welfare is their effect on agricultural crops,
including annual and perennial species. Pollutants
may affect processes within individual plants that
control or alter growth and reproduction, thereby
potentially increasing or decreasing yields of
agricultural crops. Possible physiological effects of
pollutants include: decreased photosynthesis; changes
in carbohydrate allocation; increased foliar leaching;
decreased nutrient uptake; increased sensitivity to
climatic stress, pests, and pathogens; decreased
competitive ability; and decreased reproductive
efficiency. These physiological effects, in conjunction
with environmental factors and intraspecies
differences in susceptibility, may affect crop yields.
Air pollutants that might damage plants include
SO2, NOX, peroxyacetyl nitrate (PAN), and volatile
organic compounds (VOCs). These pollutants may
have direct effects on crops, or they may damage
crops indirectly by contributing to tropospheric
(ground-level) ozone and/or acid deposition, both of
which damage plants. Tropospheric ozone is formed
by photochemical reactions involving VOCs and NOX,
while SO2 and NOX cause acidic deposition.
While all of these air pollutants may inflict
incremental stresses on crop plants, in most cases air
pollutants other than ozone are not a significant
danger to crops. Based primarily on EPA's National
Acid Precipitation Assessment Program (NAPAP),1
this analysis considers ozone to be the primary
pollutant affecting agricultural production.
This analysis estimates the economic value of the
difference in agricultural production between 1990
and 2010 that is projected to result from passage of
the 1990 CAA Amendments (CAAA). The analysis is
restricted to a subset of agricultural commodities, and
excludes those commodity crops for which ozone
response data are not available. Fruits, vegetables,
ornamentals, and specialty crops are also excluded
from this analysis for a variety of reasons, mostly
related to the absence of a national level benefits
model (for vegetables and specialty crops) and
difficulties in quantifying the physical impacts of air
quality changes and their associated effect on welfare
(for ornamentals). To estimate the economic value of
ozone reductions under the CAAA, agricultural
production levels expected from post-CAAA scenario
ozone conditions are first compared with those
expected to be associated with ozone levels projected
under the pre-CAAA scenario. Estimated changes in
economic welfare are then calculated based on a
comparison of estimated economic benefits associated
with each level of production.
Ozone Concentration Data
For this analysis, the SUM06 index - a cumulative
index of ozone concentrations over a specified
threshold (0.06 ppm) — was selected to conform with
the recent EPA ozone NAAQS benefits analysis.2
The SUM06 index is one of several cumulative
statistics that emphasize peak concentrations (in this
case by use of a threshold), and may correlate more
closely to crop damage than do unweighted indices.3
Slimier et al., 1990; NAPAP, 1991.
2 Abt Associates, 1998.
3 Lefohn et al., 1988.
F-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Because crop production data are available at the
county level, the lowest level of aggregation that could
be used for ozone indices is also the county level.
Therefore, monitor level data needed to be aggregated
to a county level.
Three main steps are used in the process of
estimating the county-level SUM06 values:
(1) 1990 hourly ozone concentrations obtained
for all available monitors from EPA's AIRS
system.4
(2) For each county centroid, the 1990 hourly
data from the closest set of monitors are
temporally- and spatially-adjusted using
UAM-V modeling data (as described in
Appendix C), and the SUM06 is calculated
for each monitor for each month.
(3) A distance-weighted average SUM06 is
estimated for each month from the
temporally- and spatially-adjusted monthly
values.
One difference between the agricultural analysis
and the health analysis is the treatment of distance
extrapolation. The health effects results in this 812
analyses are calculated first for the population living
within 50 km of monitors, and then for the whole
country by extrapolating the air quality modeling
results to provide universal coverage. The air quality
modeling results near to monitors are believed to be
more certain than the modeling for more remote
areas. The less certain air quality modeling results is a
very important issue for the agricultural analysis, as the
majority of the commodity crops are grown in
locations some distance from ozone monitors.
Because only a small portion of cropland is within
50km of an ozone monitor, the agricultural analysis is
not conducted for the within 50km of a monitor
locations. The agricultural analysis is only conducted
using the full national extrapolation of ozone
modeling results.
Calculation of the SUM06 Statistic
The hourly ozone concentrations are screened to
identify those that equal or exceed 0.06 ppm, and
these values are summed to obtain a "raw" monthly
SUM06 index:
day 30 7:59 PM
2\ ^ ozone. , for all ozone. > 0.06 ppm
j=day\ i=S:OOAM
In this analysis, the SUM06 statistic was calculated
on a monthly rather than a daily basis, reflecting the
same hours of the day as if daily statistics had been
individually calculated. Although a completeness
criterion had been used to select monitors, there were
still missing data for some included monitors.
Therefore, this "raw" statistic was adjusted by a
completeness ratio, the proportion of hours with
available data to total hours in the period (either 12 in
a day or 360 in a 30-day month), in order to address
missing data as follows:
raw statistic *
maximum hours per month
actual hours in month
The assumption implicit in using a completeness ratio
is that the distribution of hourly ozone values for the
missing data is the same as the distribution for the
available data.
October to April Ozone
Concentration Data
Agricultural crop seasons extend the May to
September period used in the health analysis, and the
SUM06 index is cumulative, requiring data for the
entire agricultural season. To address the need for
SUM06 indices in months between October and
April, 1990 monitoring data from AIRS were used —
4The analysis reflects the application of a 50 percent
completion criterion, ensuring that included monitors have at least
12 hours of data for at least half the days in the modeling season.
F-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
no temporal- or spatial-adjustments were made to
reflect potential ozone conditions in future years
outside of the modeling season.5
Yield Change Estimates
There are several steps involved in generating
yield change estimates. The first is the selection of
relevant ozone exposure-response functions
(minimum and maximum) for each crop in the
analysis. Ozone data at the county level are
transformed into an index suitable for use in the
selected function(s) to estimate county level predicted
yield losses for both the post-CAAA and pre-CAAA
scenarios. In the next step, the proportion of each
county to the national production of each crop is
calculated to permit national aggregation of estimated
yield losses. Finally, the post-CAAA scenario
percentage relative yield loss (PRYL) is compared to
the minimum and maximum PRYL for the pre-CAAA
scenario. Each step is discussed in more detail below.
Exposure-Response Functions
Yield impacts resulting from changes in from
ozone concentrations are estimated using exposure-
response functions that are specific to each crop being
analyzed. This analysis was restricted to estimating
changes in yields for those commodity crops for
which consistent exposure-response functions are
available and that are included in national agricultural
sector models. Consistent with EPA's ozone NAAQS
benefits analysis, we used National Crop Loss
Assessment Network (NCLAN)-based exposure-
response functions that were derived using a Weibull
distribution for available data, and a 12-hour SUM06
ozone index.
Minimum/Maximum Exposure-
Response Functions
Experimental data to evaluate the response of
crops to ozone has been collected for a limited
number of crops under the NCLAN program. The
objective of this program was to employ a consistent
experimental methodology to provide comparable
results across crops. The crops included in the
NCLAN experiments are corn, cotton, peanuts,
sorghum, soybeans, winter wheat, potatoes, lettuce,
kidney beans, tomatoes, and hay. For many crops, the
NCLAN program evaluated the effects of ozone on
several different cultivars. Although not necessarily
representative of the full range of variability in crop
response, the results for different cultivars do permit
identification of a range of responsiveness. The most
tolerant and responsive response functions are used to
represent minimum and maximum impacts, within the
limits of available data.
Use of cumulative exposure-response functions is
relatively recent, and few experiments have been
designed or reported in terms of the SUM06 index.
Because the NCLAN program used a consistent
protocol and developed a database of experimental
conditions and results for all of its studies, U.S. EPA's
Environmental Research Laboratory (ERL) was able
to use original data from NCLAN studies to develop
SUM06 exposure response functions for most
NCLAN crops6 (Lee and Hogsett, 1996). In
addition, the agricultural model used in this analysis
does not reflect non-commodity crops such as lettuce,
tomatoes, potatoes, alfalfa, tobacco, turnips, and
kidney beans. Table F-l presents the exposure-
response functions used in this analysis. Finally, one
commodity crop, spring wheat, was excluded because
the NCLAN exposure-response function was only
developed for winter wheat.
Estimated responsiveness of a given crop to
ozone varies within the NCLAN data. This range of
response is partially explained by the program's
5AIRS data for all U.S. monitors were screened using the 50
percent completeness criterion for each month. All hourly data
was converted to parts per million and rounded to the nearest
0.0001 ppm.
or hay.
6Data were not sufficient to develop functions for tomatoes
F-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
evaluation of several cultivars for some crops; ozone
sensitivity varies across cultivars. In addition, the
conditions for different experiments varied due to
variations in location, year, and additional treatments
included in some experiments. No one exposure-
response function can be assumed to be representative
of all cultivars in use, or of all environmental
conditions for crop production. To develop a range
of benefits estimates that reflects this variation in
responsiveness, a minimum responsiveness and a
maximum responsiveness function were selected for
each crop. In actuality, a number of different
cultivars are planted by producers, and so actual ozone
response will be a weighted average of the
responsiveness of each cultivar to its ozone condition
and its proportion of total acreage. It is important to
note that these values do not necessarily bound the
analysis, since the number of cultivars evaluated by
NCLAN is small relative to the number grown for
many crops.
For the crops used in this study, ERL conducted
an analysis to identify the ozone concentration
required to reduce yields by 10 percent for each crop
cultivar using its 12-hour SUM06 exposure-response
function. For each crop, the function demonstrating
the lowest ozone concentration at a 10 percent yield
loss represents the maximum response, and the
function with the highest concentration at 10 percent
yield loss represents the minimum response. Table F-
1 reports the minimum and maximum exposure-
response functions for each crop. Two crops, peanuts
and sorghum, did not have multiple NCLAN
experiments on which to base a comparison of the
responsiveness of different cultivars or the variation
in response with different experimental conditions.
In this analysis, the maximum and minimum yield
change results are used to bound a uniform
distribution of possible yield change, recognizing that
this distribution reflects only known potential yield
losses. Each percentile change in yield, including the
minimum and the maximum, is used to estimate a
distribution of possible changes in economic welfare
(see below).
Table F-1
Ozone Exposure-Response Functions for Selected Crops (SUM06)
Ozone Index
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
SUM06
Source: Lee and
Quantity
Max
Max
Max
Max
Max
Max
Min
Min
Min
Min
Min
Min
Hogsett(1996)
Crop
Cotton
Field Corn
Grain Sorghum
Peanut
Soybean
Winter Wheat
Cotton
Field Corn
Grain Sorghum
Peanut
Soybean
Winter Wheat
Median
Experimental
Function Duration (Days)
1-exp(-(index/78)A1.311)
1-exp(-(index/92.4)A2.816)
1-exp(-(index/177.8)A2.329)
1-exp(-(index/99.8)A2.219)
1-exp(-(index/131.4)A1)
1-exp(-(index/27.2)A1.0)
1-exp(-(index/116.8)A1.523)
1-exp(-(index/94.2)A4.307)
same as max (see above)
same as max (see above)
1-exp(-(index/299.7)A1.547)
1-exp(-(index/72.1)A2.353)
119
83
85
112
104
58
119
83
85
112
104
58
Median
Duration
(Months)
4
3
3
4
3
2
4
3
3
4
3
2
F-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Calculation of Ozone Indices
The SUM06 index is cumulative, and so is
sensitive both to the duration over which it is
calculated and to the specific month(s) within a
growing season that are included in it. For each crop
included in NCLAN ozone exposure-response
experiments, the period of ozone exposure reflected
only a portion of the crop's growing season. The
duration of the NCLAN experiments was provided by
ERL, and reflects the duration of the function that
provides the median responsiveness to ozone
exposure. Because cropping seasons vary across the
U.S., the ozone index used to calculate county-level
changes in yield due to ozone must reflect the local
season for each crop. To calculate the SUM06 index
for the appropriate growing season, state-level data on
planting and harvesting dates was used in this
analysis.7 To calculate the cumulative SUM06 index,
the experimental duration for each crop was anchored
on that crop's harvest date in each state in order to
most closely approximate the relevant period of
exposure for yield analysis. The harvest date was
assumed to be the first day in the month of harvest, so
that the SUM06 index includes the months up to but
not including the harvest month. Because the baseline
and regulatory ozone data were developed as monthly
SUM06 values, for the first month of the duration
period the proportion of remaining days to days in the
month were used to adjust the monthly SUM06 value.
The SUM06 index was calculated using the county
level ozone data developed in the prior section,
summed for the number of months of NCLAN
experimental duration, with the exposure period
anchored on the usual harvest month for each crop.8
USDA, 1984. Some states did not have explicit growing
seasons reported for certain crops due to the low production in
these states. In these cases a proxy state growing season was used.
In most of these cases the proxy growing season was taken from
a state with an adjoining boundary within the same geographic
region. Peanut emergence and harvest dates were taken from the
U.S. EPA Pesticide Root Zone Model-2 (PRZM) data, US EPA
1993.
8 This analysis required "rounding" some months: if a harvest
date was specified to be from the 15th to the end of a month, the
W126 index was calculated using that month's data; if the harvest
The form of the exposure response functions is
an exponential function based on a Weibull
distribution of the original NCLAN data, estimated to
predict a yield loss relative to conditions of "clean air"
(charcoal filtered/zero ozone) , or a zero SUM06
value. The resulting equation is in the form of:
where:
SUM06 =
B,C
predicted relative yield loss
(PRYL), expressed as a decimal
value (i.e., not multiplied by 100 to
report as a percent loss), and
relative to a zero SUM06 (or clean
air) condition
cumulative SUM06 ozone statistic
at a specified level of spatial
representation, in ppm
statistically estimated parameters,
unitless
Calculation of County Weights
Because the benefits analysis did not require a
regional level of disaggregation and to minimize
computational burdens the economic analysis was
conducted at a national level. Ozone data and
estimated yield responses, however, were developed at
a county level. To conduct a national analysis, the
county level yield change estimates were weighted to
develop a single national percent relative yield loss for
each crop relative to the post-CAAA scenario, for
both the minimum and the maximum yield responses.
Weights based on 1997 crop production data9 were
used to represent all years in this analysis (1990 to
2010). Because weather and other conditions may
change the proportion of counties' production to the
total national production in each year, weights based
date was specified to be from the first to the 14th of a month, the
W126 index was calculated using the prior month's data as the
final month in the exposure period.
9 USDA1998a.
F-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
on a single year may bias the estimates to some extent.
The weights were calculated by dividing the
production level of a crop in a county10 by the sum of
all states' reported production for that crop.11 These
county weights were applied to the percent relative
yield loss results for each county, as discussed below,
to develop a national level yield change estimate.
Calculation of Percent Change in Yield
There is an issue associated with applying the yield
loss functions to analysis of alternative regulatory
profiles. The functions provide a predicted yield loss
relative to "clean" air, while policy analysis needs to
compare policy options with a baseline, non-zero
ozone condition. Therefore, the yield change resulting
from the Clean Air Act Amendments is evaluated as
the yield loss relative to clean air under the CAAA
scenario being evaluated compared to the yield loss
under baseline (no-CAAA) conditions.
The change in yields, relative to "clean air" is
calculated as:
To create the national percent change in yield for
each crop, the results of this equation are multiplied
by the county level weights and summed for each
scenario (maximum and minimum) and for each year.
Table F-2 presents the resulting percent yield changes
that were used as inputs to the economic model.
PR VT — PR VT
rix i LJ Post_CAAA rix i LJ Pr e_CAAA
and, if yield under clean conditions is 100 percent of
possible yield, then baseline yield in this context is 1
minus baseline yield loss. Thus the change in yield
under clean air conditions can be divided by the
baseline yield, and the change in yields relative to the
baseline can be given as:
PRYLPost_CAAA -PRYL Pre_CAAA II-PRYLPost_CAAA
USDA, 1995.
11 The national total does not include USDA areas designated
"other counties". These areas are groups of counties that for one
reason or another (disclosure rules, low amount of production,
etc.) are not individually listed. Because we did not have ozone
values for these groups, we did not use their production levels in
the calculation of the total national production.
F-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table F-2
Relative Percent Yield Change
Corn Cotton Peanuts Sorgham Soybeans
Winter Wheat
2000
2010
Minimum Response
Maximum Response
Minimum Response
Maximum Resoonse
0.01%
0.05%
0.01%
0.10%
1.66%
3.79%
2.84%
6.58%
0.61%
0.61%
1.36%
1.36%
0.01%
0.01%
0.02%
0.02%
0.26%
2.75%
0.42%
4.38%
0.20%
5.07%
0.39%
9.11%
Economic Impact Estimates
To estimate the economic benefits of controls on
ozone precursor pollutants implemented pursuant to
the 1990 CAAA Amendments, we evaluated the
changes in yields resulting from additional, post-1990
controls in terms of their effect on agricultural
markets. To do this, yield changes can be
incorporated into an economic model capable of
estimating the associated changes in economic
surpluses within the agricultural economy, preferably
one that reflects changes in producers' production
decisions and demand substitution between crops.
This type of dynamic analysis is needed because even
small changes in yield or price expectations can cause
large shifts in the acreage allocated to specific crops,
and the degree to which alternative crops will be
substituted (particularly for feed uses).
The modeling approach used in this analysis is to
use an econometric model of the agricultural sector,
which estimates demand and supply under different
production technologies and policy conditions. The
AGricultural Simulation Model (AGSIM©) has been
used extensively to evaluate air pollution impacts, as
well as a number of other environmental policy
analyses. The version of AGSIM© used in this
analysis reflects production conditions and projections
for three discrete periods: 1990, 2000, and 2010.
Projections of the 2000 and 2010 baseline are
essentially those reported by USDA/ERS (USDA
1998b). A few endogenous variables in AGSIM©
were not included in the USDA baseline. In those
cases, the 1997 Food and Agricultural Policy Research
Institute (FAPRI) baseline was used (FAPRI 1997) ,12
The AGSIM© baseline production and price data
serve as the post-CAAA scenario baseline. Percent
relative yield losses (PRYLs) between the post-CAAA
and pre-CAAA scenarios are the relevant input
parameter for this analysis, from which AGSIM©
calculates new yield per planted acre values. Based on
these values (as well as on lagged price data, ending
stocks from the previous year, and other variables),
AGSIM© predicts acreage, production, supply, and
price parameters for each crop for each year, as well as
calculating yield per harvested acre. From these
results and the demand relationships embedded in the
model, AGSIM© calculates the utilization of each
crop (i.e., exports, feed use, other domestic use, etc.),
as well as the change in consumer surplus, net crop
income, deficiency payments and other government
support payments. Net surplus is calculated as net
crop income plus consumer surplus, less government
payments.
Table F-3 presents the net changes in economic
surpluses in nominal terms for our two target years,
2000 and 2010. The positive net surpluses are a result
of the increase in yields associated with lower ozone
levels than those predicted to occur under the pre-
CAAA scenario. The annual value of the estimated
agricultural benefits of the CAAA in 2010 ranges
between $7.5 million in the minimum response case to
approximately $1.1 billion in the maximum response
case, with a median response of $550 million. It
12 Documentation for this version of AGSIM can be found
in Abt Associates, 1998.
F-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
should be reiterated that this range represents the
impacts that would occur if all of the acreage planted
to a given crop had an ozone response function
similar to either the minimum available response
function or the maximum available response function.
The available response functions do not necessarily
bracket the true range of potential crop responses, and
it is unrealistic to anticipate that all acreage will be
planted in cultivars with a uniform response to ozone
exposure. These considerations notwithstanding,
these values do indicate the likely magnitude of
agricultural benefits associated with post-CAAA of
ozone precursors under the CAAA, but not the
precise value of those benefits.
Table F-3
Change in Net Crop Income, Consumer Surplus and Net Surplus
Under the Post-CAAA Scenario (millions of 1990$)
Change in Net Crop Income
Change in Consumer Surplus
Change in Net Surplus
Minimum
Maximum
Minimum
Maximum
Minimum
Maximum
1990
2000
2010
$0
-$320
-$736
$0
-$1,901
-$4.555
$0
$367
$743
not he includ
$0
$2,763
$5.643
ed in the analvsis d
$0
$46
$7.5
ue to eitl
$0
$862
$1.088
ner exnosure-
Conclusions
Agricultural benefits associated with post-CAAA
levels of ozone precursors under the Clean Air Act are
likely to be fairly large. Because it is possible that
over time producers have adopted more ozone-
resistant cultivars, it may be appropriate to consider
the lower end of the range of predicted benefits to be
more indicative of the likely total benefits for those
crops included in the analysis. The estimates
developed in this analysis, however, do not represent
all of the likely benefits accruing to agriculture, in that
many high-value and/or ozone sensitive crops could
response data limitations or agricultural sector
modeling limitations. The second consideration
implies that benefits will likely be larger than
estimated. The minimum case may be the most
appropriate starting point, however, due to the first
consideration: the current crop mix may be biased
toward higher ozone responsiveness. Therefore, we
anticipate that cumulative net present value
agricultural benefits from the Clean Air Act
Amendments over the period 1990 to 2010 are on the
order of $4 billion dollars.
F-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
Abt Associates. 1998. Agricultural Benefits Using AGSIM for the NOx SIP Call. Prepared for U.S.
EPA/OAQPS. September.
FAPRI. 1997. U.S. Agricultural Outlook. FAPRI Staff Report No. 1-97. Obtained via
http://www.fapri.missouri.edu.
Lefohn, Allen S. et. al. 1988. A comparison of indices that describe the relationship between exposure to
ozone and reduction in the yield of agricultural crops. Atmospheric Environment 22: 1229-1240.
Lee, E.H., W.E. Hogsett. 1996. Methodology for calculating inputs for ozone secondary standard benefits
analysis: Part II. Prepared for the U.S. EPA, OAQPS. March.
National Acid Precipitation Assessment Program (NAPAP). 1991. 1990 Integrated assessment report.
National Acid Precipitation Assessment Program, 722 Jackson Place NW, Washington, D.C. 20503.
Shnner, D.S., W.W. Heck, S.B. McLaughlin, D.W. Johnson, P.M. Irving, J.D. Joslin and C.E. Peterson.
1990. Response of vegetation to atmospheric deposition and air pollution. NAPAP SOS/T Report
18, In: Acidic Deposition: State of Science and Technology, Volume III, National Acid Precipitation
Assessment Program, 722 Jackson Place NW, Washington, D.C. 20503.
USDA 1984. Usual Planting and Harvesting Dates for U.S. Field Crops. Statistical Reporting Service
Agricultural Handbook No. 628.
USDA. 1998a. Crops County Datasets. National Agricultural Statistics Service. Electronic files obtained via
http://usda.mannlib.cornell.edu/data-sets/crops/9X100/Fl.
USDA 1998b. USDA Agricultural Projections to 2007. World Agricultural Outlook Board, Office of the
Chief Economist. Prepared by the Interagency Agricultural Projections Committee. Staff Report
No. WA OB-98-1. Obtained via http://www.econ.ag.gov/briefing/ baseline.
U.S. EPA. 1993. PRZM-2. A model for predicting pesticide fate in the crop root and unsaturated soil
zones. User's manual for Release 2.0. EPA/600/R-93/046.
F-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[This page left blank intentionally.]
F-10
-------
Stratospheric Ozone
Assessment
Introduction
This appendix describes the assessment of costs
and benefits of Title VI of the Clean Air Act
Amendments (CAAA). Provisions under Title VI
limit emissions of stratospheric ozone-depleting
chemicals. Cost and benefit estimates of Title VI
provisions are derived from a secondary analysis
that updates key economic valuation components of
previous analyses conducted by EPA's Office of Air
and Radiation, Stratospheric Protection Division.
We chose to perform a secondary analysis for
several reasons: (1) prior analyses suggest that the
benefits of these programs far exceed the costs,
suggesting a potentially low value of information
from re-estimating costs and benefits in new
primary analyses; (2) prior analyses were extensively
peer-reviewed; (3) the costs and benefits of these
provisions are largely separable from provisions
implemented under other titles of the Clean Air Act
Amendments; and (4) developing new primary
estimates would involve considerable time and
expense. We therefore provide a new assessment of
the valuation of benefits to ensure consistency with
other portions of the prospective analysis, but have
not re-assessed the Agency's previous estimates of
stratospheric ozone depleter emissions,
stratospheric ozone loss, changes in exposure to
UV-b radiation, changes in physical effects, or the
costs of Title VI provisions.
Unlike estimates for other Titles of the CAAA,
we present estimates for Title VI as net present
values of the streams of annual costs and benefits.
The rationale for this type of presentation of costs
and benefits relates to the long-term nature of the
mechanisms of stratospheric ozone depletion and
measures taken to avoid depletion. Stratospheric
ozone is a global resource, and its formation and
depletion are governed by long-term processes that
may take place over decadal or longer time scales.
Attempting to parse the incremental effects of an
annual reduction in emissions of ozone depleting
substances and estimate its impact at the unit
emissions level is an extremely difficult, if not
impossible, task and was not attempted for this
exercise.
For the same reasons, We conduct a longer
time-scale of analysis than is used in the remainder
of the study. We estimate cost over the period 1990
through 2075, and benefits are estimated over the
period 1990 to 2175. The difference in time scales
for costs and benefits reflects the persistence of
ozone depleting substances in the atmosphere, the
slow processes of ozone formation and depletion,
and lags in the manifestation of physical effects in
response to exposure to elevated levels of UV-b
radiation. The full benefits of emissions reductions
achieved during 1990 to 2075 accrue across many
decades and several generations, requiring an
extended time scale for benefits analysis.
In the next section of the appendix, we provide
a brief history of Title VI and its amendments.
Next we summarize the general approaches used to
estimate the costs and benefits in previous analyses,
and we describe our strategy for modifying several
analytical parameters to ensure consistency with the
assessments of other titles of the Clean Air Act.
Finally, we present the adjusted cost and benefits
from 1990 to 2165 and discuss the uncertainty
associated with the analyses.
History of Stratospheric Ozone
Protection and the CAAA
The protection of stratospheric ozone is an
international effort, resulting in several multinational
agreements. The United States has both participated
in the development of these international
G-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
agreements and has created its own reduction and
phaseout schedules for ozone-depleting substances
used within its borders. These reduction and
phaseout schedules, codified under Title VI of the
Clean Air Act Amendments, are often stricter than
the international agreements that preceded them.
Below we briefly describe the history of the
international agreements and their relationship to
the Clean Air Act Amendments.
The United States, the European Economic
Community, and 23 other countries signed the
"Montreal Protocol on Substances that Deplete the
Ozone Layer" (1987 Montreal Protocol) on
September 16, 1987. Thirty-four countries then
ratified this protocol. The agreement prohibits the
use of chlorofluorocarbons (CFCs) beyond 1986
usage levels starting in mid-1989 and establishes a
schedule for reducing the production of CFCs (i.e.,
a 20 percent reduction in CFC production in 1993
and a 50 percent reduction in 1998). The protocol
also forbids the production of halons beyond 1986
production levels starting in 1992. On August 12,
1988, the U.S. Environmental Protection Agency
(EPA) published final regulations to protect
stratospheric ozone and comply with the
requirements of the 1987 Montreal Protocol.1
After ratification of the Montreal Protocol
scientists determined that the loss of stratospheric
ozone was greater than they had originally thought
and that man-made compounds containing bromine
and chlorine were responsible for this loss.
Consequently, in June 1990 the countries that had
signed the Montreal Protocol met in London to
develop an accelerated CFC reduction schedule (i.e.,
a decrease in CFC production to 50 percent of 1986
production levels by 1995 and 15 percent of 1986
levels by 1997). According to this London
Agreement, production of CFCs, halons, and
carbon tetrachloride will cease by 2000 and methyl
chloroform (MCF) production would end by 2005.2
In November 1990 President George Bush
signed the Clean Air Act Amendments (CAAA),
which include Title VI. This title consists of six
major sections, of which the most important are
sections 604 and 606. Section 604 accelerates the
London Agreement's reduction schedules for CFCs,
halons, and carbon tetrachloride and shortens the
time allowed for methyl chloroform phaseout by
three years. Section 606 allows Congress to
accelerate the reduction and phaseout schedules of
section 604 if necessary to protect human health
and the environment. Together, sections 604 and
606 generate nearly all of the costs and benefits of
Title VI. The remaining sections include the
following: section 608 (which requires the EPA to
establish standards regarding the use and disposal of
ozone-depleting substances during the service,
repair, or disposal of appliances and industrial
refrigerators); section 609 (which requires the EPA
to regulate the servicing of motor vehicle air
conditioners); and section 611 (which stipulates that
the EPA must establish labeling requirements for
containers of ozone-depleting substances and for
products containing these substances). This analysis
does not examine the costs and benefits of section
605, which institutes the reduction and phaseout
schedules for hydrochlorofluorocarbons (HCFCs),
because the schedules of section 606 supersede
section 605's timetables. Table G-l provides a
description of the principal sections of the CAAA's
Title VI.
1 ICF, Incorporated, Regulatory Impact Analysis: Compliance
with Section 604 of the Ckan Air Act for the Phaseout of O^pne
Depleting Chemicals, Prepared for Global Change Division, Office
of Air and Radiation, U.S. Environmental Protection Agency,
Washington, D.C.July 1, 1992, page ES-1.
2Ibid, ES-1 and ES-2.
G-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-1
Six Major Sections of Title VI
Section
Description
Section 604 - Class I Phaseout
Institutes the reduction and phaseout schedules for Class I substances
(methyl chloroform, halons, chlorofluorocarbons, carbon tetrachloride).
Section 605 - Class II Phaseout
Institutes the reduction and phaseout schedules for Class II substances
(hydrochlorofluorocarbons).
Section 606 - Accelerated Schedule Permits Congress to accelerate the phaseout schedule of Class I and II
substances if necessary for the protection of human health and the
environment.
Also, if countries modify the Montreal Protocol to accelerate phaseout
schedules of Class I and II chemicals, Congress can amend the Clean Air
Act to reflect these modifications.
Section 608 - National Recycling and
Emission Reduction Program
Requires the EPA to establish standards regarding the use and disposal of
Class I and II substances during the service, repair, or disposal of
appliances and industrial refrigeration units.
Section 609 - Servicing of Motor
Vehicle Air Conditioners
Requires the EPA to regulate the servicing of motor vehicle air
conditioners.
Section 611 - Labeling
Stipulates that the EPA establish labeling requirements for containers of
Class I and II substances and for products containing these substances.
In November 1992 the parties to the Montreal
Protocol met in Copenhagen to establish an
agreement that incorporates new scientific
information on stratospheric ozone depletion.3 For
example, data from the National Aeronautics and
Space Administration (NASA) indicated that ozone
depletion was progressing more rapidly than
expected. In addition, ozone depletion was
extending further south in the United States than
anticipated and lasting longer (late fall to late May).4
In response to new data, the parties to the Montreal
Protocol agreed to several changes in the reduction
and phaseout schedules of ozone-depleting
chemicals. First, they agreed to a 1999 phaseout
U.S. Environmental Protection Agency,
http://www.epa.gov/ttn/oarpg/t6/fact_sheets/66.txt, March
25, 1998.
4 Ibid, ES-3 and ES-4.
deadline for hydrobromofluorocarbons (HBFCs),
chemicals not regulated under the earlier London
Agreement. Second, they called for a freeze on
production of methyl bromide by stipulating that
the chemical should not exceed 1991 levels starting
in 1995.5 Third, these countries decided to
accelerate the reduction schedule for the production
and consumption of hydrochlorofluorocarbons
(HCFCs). Lastly, they hastened the phaseout of
halons by agreeing to 1994 as the production and
consumption phaseout deadline.6
5 U.S. Environmental Protection Agency,
http://www.epa.gov/spdpublc/mbr/harmoniz.html, March 26,
1999.
U.S. Environmental Protection Agency,
http://www.epa.gov/ttn/oarpg/t6/fact_sheets/66.txt, March
25, 1998.
G-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Under the Clean Air Act Amendments' section
606, the EPA also responded to the new scientific
data by periodically accelerating the reduction
schedules for MCF, CFCs, halons, carbon
tetrachloride, and HCFs and by establishing earlier
phaseout deadlines. In addition, the EPA targeted
HBFCs and methyl bromide, chemicals not
explicitly addressed by the 1990 Clean Air Act
Amendments.
The most recent changes established under
section 606 involve methyl bromide. In 1993 the
EPA called for a freeze on production at 1991 levels
starting in 1994 and established a phaseout deadline
of 2001. In 1995 and 1997 the parties to the
Montreal Protocol met in Vienna and Montreal,
respectively, to address issues such as the phaseout
of methyl bromide. In 1998 Congress directed the
EPA to match the 1997 Montreal Amendment's
reduction and phaseout schedule for this chemical.
In 1998, Congress amended the Clean Air Act to
establish a new methyl bromide reduction schedule,
which establishes 2005 as the phaseout deadline,
and helps to address the needs of American farmers
who are currently waiting for methyl bromide
alternatives that are in the research and
development stage.7 Table G-2 presents the most
significant sections of the new and old phaseout schedules.
U.S. Environmental Protection Agency,
http://www.epa.gov/spdpublc/nibr/harrnoniz.htnil,
November 9, 1999.
G-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-2
Phaseout Scenario in Clean Air Act Section 604 and
Phaseout Scenario in Amendments Added under Clean Air Act Section 606
Year Section 604 (Original Schedule)
Section 606 (Revisions to Original Schedule)
1991 • Methyl chloroform (MCF) production decreases to 1989 levels
1993 • MCF production decreases 10% from 1989 levels
1994 • MCF production decreases 15% from 1989 levels -»
1995 • MCF production decreases 30% from 1989 levels -»
1996 • MCF production decreases 50% from 1989 levels -»
2000 • MCF production decreases 80% from 1989 levels
2002 • 100% phaseout of MCF
MCF production decreases from 1989 levels by 50%
MCF production decreases from 1989 levels by 70%
100% phaseout of MCF
1989 • Chlorofluorocarbon (CFC) production decreases to 1986 levels
1991 • CFC production decreases 15% from 1986 levels
1992 • CFC production decreases 20% from 1986 levels
1993 • CFC production decreases 25% from 1986 levels
1994 • CFC production decreases 35% from 1986 levels -»
1995 • CFC production decreases 50% from 1986 levels -»
1996 • CFC production decreases 60% from 1986 levels -»
1997 • CFC production decreases 85% from 1986 levels
2000 • 100% phaseout of CFC production
CFC production decreases to 75% from 1986 levels
CFC production decreases to 75% from 1986 levels
100% phaseout of CFC production
G-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Year Section 604 (Original Schedule)
Section 606 (Revisions to Original Schedule)
1991 • Halon production decreases 15% from 1986 levels
1992 • Halon production decreases 20% from 1986 levels
1993 • Halon production decreases 25% from 1986 levels
1994 • Halon production decreases 35% from 1986 levels
1995 • Halon production decreases 50% from 1986 levels
1996 • Halon production decreases 60% from 1986 levels
1997 • Halon production decreases 85% from 1986 levels
2000 • 100% phaseout of halon production
100% phaseout of halons
1991
1992
1993
1994
1995
1996
2000
Carbon tetrachloride production decreases to 1989 levels
Carbon tetrachloride production decreases 10% from 1989 levels
Carbon tetrachloride production decreases 20% from 1989 levels
Carbon tetrachloride production decreases 30% from 1989 levels
Carbon tetrachloride production decreases 85% from 1989 levels
100% phaseout of carbon tetrachloride
-» • Carbon tetrachloride production decreases 50% from 1989 levels
-» • Carbon tetrachloride production decreases 85% from 1989 levels
-» • Carbon tetrachloride production decreases 100% phaseout of carbon
tetrachloride
G-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Year Section 604 (Original Schedule)
Section 606 (Revisions to Original Schedule)
1994
1999
2001
2003
2005
• Freeze on production and consumption of methyl bromide at 1991 levels
• 25% phaseout of methyl bromide
• 50% phaseout of methyl bromide
• 75% phaseout of methyl bromide
• 100% phaseout of methyl bromide
(quarantine and preshipment uses exempt; critical agriculture uses allocated after 2005)
1996
100% phaseout of hydrobromofluorocarbons (HBFCs)
2003
2010
(2010-
2020)
2015
2020
2030
Freeze on HCFC production
Prohibition of HCFC production after
January 1, 2030
Production and consumption of HCFC-141b banned
Production and consumption of HCFC-142b and HCFC-22 decreases to 1989 levels
Production and consumption of HCFC-142b and HCFC-22 permissible only for servicing
equipment manufactured prior to 2010
Production and consumption of the remaining HCFCs decreases to 1989 levels
100% phasout of HCFC-142b and HCFC-22
100% phaseout of the final category of HCFCs
G-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Cost and Benefit Approaches
To estimate the costs and benefits of Title VI,
we rely primarily on past EPA regulatory impact
assessments (RIAs), including the following:
• Addendum to the 1992 Phaseout Regulatory
Impact Analysis: Accelerating the Phaseout of
CFCs, Halons, Methyl Chloroform, Carbon
tetrachloride, andHCFCs (1993);
• Regulatory Impact Analysis: Compliance nith
Section 604 of the Clean Air Act for the Phaseout
ofO^pne Depleting Chemicals (1992);
• Regulatory Impact Analysis: The National
Recycling and Emission Reduction Program
(Section 608 of the Clean Air Amendments of
1990) (1993);
• Section 609 of the 1990 Clean Air Act:
Cost-Benefit Anajysis and Regulatory flexibility
Analysis (1991);
• Draft: Regulatory Impact Analysis of the Proposed
Rule Requiring Labeling of Products Containing
or Manufactured with O^pne Depleting Substances
(1991).
The major difference between this analysis and the
RIA analyses involves the parameters used to value
the costs and benefits. To ensure consistency with
the larger Section 812 analysis, we adjust the
discount rate in the costs calculations, and we adjust
the value of statistical life, GDP per capita growth
rates, and the discount rate in the benefits
calculations.
Cost Approach in RIAs
Existing regulatory impact assessments (RIAs)
for individual provisions of Title VI provide the
basis for the social costs of phasing out CFCs,
halons, methyl chloroform, and HCFCs.8 These
social costs are the additional quantities of
resources necessary to produce equivalent quantities
of goods and services for consumers. To generate
social cost estimates, the RIAs calculate the costs of
replacing ozone-depleting chemicals (ODSs) with
alternative technologies and materials, as well as the
costs of recycling and storing unused ODSs. The
estimates also include costs of training, labeling, and
administration. The total cost estimate of Title VI
comprises the costs of sections 604 and 606 and the
incremental costs of the remaining sections (i.e., the
cost estimates for sections 608, 609, and 611 do not
include the costs of actions already required by
sections 604 and 606). Table G-3 indicates the
major costs associated with each section. We
present summaries of the cost methodologies in this
appendix; for more details, see the RIAs of the
individual provisions of Title VI.
8 Only a small percentage of carbon tetrachloride (less than
three percent of the total produced in the U.S.) is subject to the
CAAA, and the costs of phasing out carbon tetrachloride are
likely insignificant compared to the costs of phasing out CFCs,
halons, and MCFs. Consequently, the RIAs do not quantify the
costs of phasing out carbon tetrachloride. In addition, this
analysis does not assess the cost of methyl bromide because Part
2: The Cost and Cost-Effectiveness of the Proposed Phaseout of Methyl
Bromide does not provide the corresponding benefits.
G-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-3
Scope of Title VI Cost Estimates
Section
608
Costs
604 & 606 « Capital investment and variable costs associated with switching to alternative technologies
and non-ozone-depleting substances;
Possible elimination of products containing ozone-depleting substances (ODSs) if firms are
unable to develop cost-effective manufacturing alternatives;
• Costs of recycling ODSs;
Additional costs of switching from methyl bromide to other substances (e.g. purchases of
more costly substitutes, incremental labor needs, and crop and throughput issues).
Purchase of ODS recovery equipment;
Training and certification of technicians;
Filtration of refrigerants to remove impurities;
Leak repair requirements;
Storage of unused chlorofluorocarbons (CFCs);
Paperwork.
609 « Training and certification of mobile air conditioner (MAC) service technicians;
Recycling costs, including fees for off-site recycling or labor and capital costs for on-site
recycling.
611 « Development and application of new labels;
Administrative activities associated with compliance;
Possible accelerated substitution of less harmful substances relative to the mandates of the
rule codifying the phaseout of ODSs (thereby resulting in additional costs beyond those
resulting from the phaseout rule).
Costs: Sections 604 and 606
To generate the costs of section 604, the
Addendum to the 1992 Phaseout Regulatory Impact
Analysis: Accelerating the Phaseout of CFCs, Halons,
Methyl Chloroform, Carbon Tetrachloride, and HCFCs
(1993) and the Regulatory Impact Analysis: Compliance
with Section 604 of the Clean Air Act for the Phaseout of
O^pne Depleting Chemicals (1992) use engineering
analyses that involve several steps. First, the RIAs
identify potential technical responses to section 604
from 1990 to 2075. For example, they examine data
on the CFG reduction technologies available to each
CFC-using industry (i.e., mobile air conditioners,
household refrigeration, foam blowing, solvent
cleaning, sterilization, rigid polyurethane foams,
chillers, and process-12 refineries).9 For halon the
ICF, Incorporated, Addendum to the 1992 Phaseout
Regulatory Impact Analysis: Accelerating the Phaseout of CFCs, Halons,
Methyl Chloroform, Carbon Tetrachloride, and HCFCs, Prepared for
the Stratospheric Protection Division, Office of Air and
RIAs analyze 74 categories of fire extinguishing
applications, the primary users of the substance.10
With these data, the RIAs identify three potential
technical responses: use of chemical substitutes for
CFC and halon use in new and existing equipment,
implementation of engineering controls that reduce
use of ODSs through recycling, and use of product
substitutes for ODS-containing products.
Second, the RIAs assess the costs of the
responses. For halons and CFCs the RIAs examine
the date at which an action is first available for
adoption, the time necessary for the entire industry
to evaluate the action, and the time required for
Radiation, U.S. Environmental Protection Agency, September
10, 1993, page 6.
10ICF, Incorporated, Regulatory Impact Analysis: Compliance
with Section 604 of the Clean Air Act for the Phaseout of O^pne
Depleting Chemicals, Prepared for Global Change Division, Office
of Air and Radiation, U.S. Environmental Protection Agency,
July 1, 1992; page 4-7.
G-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
firms to adopt the action if it is cost-effective. They
also study reduction potential (decrease in ODS use
for all firms that have decided to take the action)
and applicability to new and/or existing equipment.
This CFG and halon cost approach is similar to the
approach used in the U.S. Environmental
Protection Agency's Regulatory Impact Analysis:
Protection of Stratospheric O^pne (1988); the main
difference is the expansion of the methods used to
model the lifetimes of equipment using CFC and
halons.11 By contrast, for MCFs the RIAs do not
focus on the lifetime of the equipment because
firms can retrofit MCF-consuming end uses with
MCF alternatives.12 Using the relevant data for
MCFs, halons, and CFCs, the RIAs estimate feasible
schedules for implementation of reduction measures
and estimate the following costs:
• Variable costs (e.g., materials, energy, labor,
and operating expenses);
11 Specifically, the cost model improves upon the 1988
model in the following ways: tracking the size, age, and turnover
of CFC-consuming equipment over time; simulating the
lifecycle of the equipment in terms of manufacturing, operation,
servicing, and disposal; estimating CFC use, CFC emissions,
energy use, and costs for each point in the lifecycle; calculating
CFCs potentially available through recycling from older
equipment; and simulating the impact of CFC reduction
measures by assessing the degree of emissions reductions, total
costs, and total changes in energy use associated with
implementing groups of controls over time for each end use.
(ICF Incorporated, Regulatory Impact Analysis: Compliance with
Section 604 of the Clean Air Act for the Phaseout ofQ^one-Depleting
Chemicals, Prepared for the Global Change Division, Office of
Air and Radiation, U.S. Environmental Protection Agency,
1992, pages 4-2, 4-8, 4-9, and 4-16.)
12 The EPA's Costs and Benefits of Phasing Out Production of
CFCs and Hahns in the United States (1989) and the EPA's
Regulatory Impact Analysis: Protection of Stratospheric O^pne (1988)
provide the basis of the cost methodology for the MCF
phaseout. Note that the MCF model calculates energy costs as
a part of operating costs, rather than as a separate component,
because MCF end uses are not energy intensive. (ICF
Incorporated, Regulatory Impact Analysis: Compliance with Section
604 of the Clean Air Act for the Phaseout ofO^pne-Depleting Chemicals,
Prepared for the Global Change Division, Office of Air and
Radiation, U.S. Environmental Protection Agency, 1992, pages
ES-8, 4-27.)
• Capital costs;
• One-time fixed costs (e.g., research and
development or training); and
• Changes in energy efficiency.
For the last step of the cost analyses, the model
selects technologies that minimize production cost
increases and achieve the necessary ODS
reductions. The final set of control plans must
satisfy the following requirements: 1) the plans
contain components that industry has already
implemented or intends to implement in the near
future; 2) they jointly ensure that the amount of
CFC production over time does not exceed the
maximum stipulated by the phaseout schedule; and
3) they collectively prevent CFC use after the
phaseout deadline from exceeding feasible recycling.
Under section 606 the EPA accelerates the
reduction and phaseout schedules of ODSs; for
these substances the cost methodology is the same
as the section 604 methodology for CFCs, halons,
methyl chloroform, and carbon tetrachloride. For
HCFCs the EPA calculates the costs of phaseout by
multiplying the quantity of replaced HCFCs by the
difference in price between the HCFC compound
and its substitute. This analysis assumes that the
replacement compounds will cost between 10 and
50 percent more than the HCFCs. Together, the
costs of sections 604 and 606 are $55.9 billion (1990
dollars) with a two percent discount rate; these costs
comprise nearly all of the costs of Title VI.
Costs: Sections 608
For section 608, the Regulatory Impact Analysis:
The National Recycling and Emission Reduction Program
(Section 608 of the Clean Air Amendments of 1990)
(1993) uses the section 604 cost model to forecast
the timing of emissions controls, resulting prices,
and recycling levels from 1994 to 2015. In
particular, the model assumes that recovery
efficiency is 95 percent and predicts that all users of
chillers, industrial processes, cold storage, retail
food, and refrigerated transport will either recover
G-10
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
or recycle CFCs at service and disposal and that
after the phaseout all household appliance users will
recover CFCs at disposal.13 With a two percent
discount rate the cost estimate is $1.2 billion (1990
dollars).
Costs: Section 609
Section 604 prompts service establishments to
recycle CFC-12 in mobile air conditioners (MACs).
Therefore, the section 604 cost estimate accounts
for most costs associated with the recycling of CFC-
12 from MACs. The costs of section 609 that are
not included in the section 604 cost estimate are the
costs of training and certifying MAC service
technicians. The analysis of the social costs of
Section 609 examines these costs from 1992 to 2008
under two scenarios: a lower bound scenario in
which small and large shops recover CFC-12 and
pay an off-site recycler to purify and return the used
refrigerant and an upper bound scenario in which all
shops recycle CFC-12 on-site.14 The central cost
estimate for section 609 with a two percent discount
rate is $14.3 million (1990 dollars).
Costs: Section 611
For section 611, the Draft: Regulatory Impact
Analysis of the Proposed Rule Requiring Labeling of
Products Containing or Manufactured with O^pne Depleting
Substances (1991) evaluates two response options:
companies label all products or they
reformulate/redesign the products to eliminate the
use of Class I ozone-depleting substances (fully
halogenated CFCs, three halons, methyl chloroform,
and carbon tetrachloride).15 The RIA then assesses
three associated costs: costs of implementing
substitutes (for MCF-containing products) more
rapidly than predicted under the phaseout schedule,
administrative activity costs, and costs of labeling.16
From 1994 to 2000 the costs of section 611 are
$252 million with a two percent discount rate.
Benefits Approach in RIAs
The RIAs' Tide VI benefits analyses
necessarily differ from the benefits analyses for
other parts of our CAAA-analysis because, unlike
most of the effects of criteria air pollutants, the
effects of Tide VI are global in scale and occur over
several hundred years. The delay in effects occurs
for several reasons. First, emissions often emanate
from products that leak the ozone-depleting
chemicals over a significant period of time. Second,
ozone-depleting chemicals rise into die stratosphere
and affect the ozone layer at a slow rate. Third,
ozone-depleting substances can persist in the
stratosphere for many years. Fourth, natural
processes that replace stratospheric ozone are slow.
To reflect the long time period during which
stratospheric ozone depletion occurs, this analysis
assumes that the benefits accrue from 1990 to 2165.
Figure G-l is a simplified illustration of the
relationships between the sets of data used in the
existing benefits analyses. First, the EPA estimates
change in ozone-depleting substance emissions.
With these data the EPA calculates the extent of
stratospheric ozone depletion and global warming.
Then the EPA calculates the effects of stratospheric
ozone on UV-b radiation, which in turn affects
13ICF, Incorporated, Regulatory Impact Analysis: The National
Recycling andEmission Reduction Program (Section 608 of the Clean Air
nts of 1990)., Prepared for the Stratospheric Protection
Division, U.S. Environmental Protection Agency, March 25,
1993, page 4-2.
14ICF, Incorporated, Section 609 of the 1990 Chan Air Act:
RefrigerantRecycling for Mobile Air Conditioners: Cost-Benefit Analysis
and Regulatory Flexibility Analysis., Prepared for the Division of
Global Change, U.S. Environmental Protection Agency, May 24,
1991, pages 4 and 8.
15 The costs analysis does not account for the accelerated
reduction and phaseout schedule of section 606. ICF,
Incorporated, Draft: Regulatory Impact Analysis of the Proposed Ra/e
Requiring Labeling, of Products Containing, or Manufactured with Oyone
i & & j & j ^
Depleting Substances., Prepared for Global Change Division, Office
of Air and Radiation, U.S. Environmental Protection Agency,
November 1991, pages 1, 4, and 6.
16 Ibid, page 37.
G-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Fig u re G -1
SCHEMATIC OF COST AND
BENEFIT ANALYSES OF TITLE VI
Emissions Estimates
Cost estim ates
Present
value
Stratospheric
ozone m odeling
Global warming
estimates*
I
UV-b
radiation
Tropospheric
ozone
Hum an health
effects
Ecological
effects
Som e benefits
addressed
qualitatively
Monetization of
som e benefits
Present
value
Monetization of
som e benefits
I
Present
value
Several ozone-depleting substances are also greenhouse gases with high radiative forcing
potential relative to carbon dioxide. We do not assess the impact ofglobal warming in this analysis.
G-12
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
estimates of tropospheric ozone.17 Using UV-b
radiation, tropospheric ozone, and global warming
data as inputs, the EPA estimates human health and
environmental effects.18 Lastly, the EPA monetizes
the benefits of improved human and environmental
health where possible.19 In this analysis, our
assessment of benefits is slightly different from that
of the previous RIAs because we attribute benefits
to effects of reduced stratospheric ozone and not to
global warming. We present the benefits estimate as
a net present value, rather than an annualized value,
because annualization incorrectly imputes benefits
of later phaseouts to earlier years. For example,
annualization of the benefits of phasing out HCFC-
22 by 2020 attributes benefits to years prior to 2030,
when neither the costs nor the benefits of that
phaseout have yet occurred. Consequently, an
annualized estimate will overstate benefits at the
beginning of the time span and understate them
later.
Emissions Modeling
The methodology for predicting global use of
ozone-depleting chemicals and the resulting
emissions is similar to the methodology used in
Regulatory Impact Analysis: Compliance with Section 604
17 The relationships between stratospheric ozone depletion
and global warming and between stratospheric ozone depletion
and tropospheric ozone are incompletely understood at this
time. The Regulatory Impact Analysis: Compliance with Section 604 of
the Clean Air Act for the Phaseout ofO^pne Depleting Chemicals (1992)
and the Addendum to the 1992 Phaseout Regulatory Impact Analysis:
Accelerating the Phaseout of CFCs, Halons, Methyl Chloroform, Carbon
Tetrachloride, andHCFCs (1993) base their tropospheric ozone
estimates on the "Effects of Increased UV Radiation on Urban
Ozone" (Whitten and Gery, 1986).
18 Note that the Regulatory Impact Analysis: Compliance with
Section 604 of the Clean Air Act for the Phaseout of O^one Depleting
Chemicals (1992) and the Addendum to the 1992 Phaseout Regulatory
Impact Analysis: Accekrating the Phaseout of CFCs, Halons, Methyl
Chloroform, Carbon Tetrachloride, and HCFCs (1993) do not
provide data on the models used to estimate UV-b radiation.
19 The benefits estimates used in the RIAs' benefit/cost
comparison sections do not reflect economic impacts (e.g.,
profit increases that occur if alternative technologies are more
efficient than ODS-using technologies).
of the Ck an Air Act for the Phaseout of Q^pne-Depleting
Chemicals (1992). The main difference in the baseline
scenario, which assumes no Title VI controls,
involves methyl bromide. In this analysis we
assume the following as a baseline: 1) in 1990
facilities worldwide produce 63 million kilograms of
methyl bromide and facilities in the U.S. produce
29.1 million kilograms; 2) methyl bromide
production grows at 5.5 percent annually until 2025
and zero percent thereafter; 3) 50 percent of methyl
bromide production generates emissions; 4) humans
generate about 25 percent of total methyl bromide
emissions20; and 5) bromine is 40 times as effective
as chlorine at destroying ozone. Also, based on
NASA's new data regarding the extent of ozone
depletion in the Northern Hemisphere, the model
assumes that the weighted average ozone depletion
was 3.38 percent in 1989 relative to 1979 .21
The control scenario used in this prospective
analysis is based on the "CAA phaseout scenario"
established in the 1992 RIA and the "President's
schedule" outlined w. Addendum to the 1992 Phaseout
Regulatory Impact Analysis: Accelerating the Phaseout of
CFCs, Halons, Methyl Chloroform, Carbon Tetrachloride,
and HCFCs (1993). The phaseout schedule,
presented in Table G-2, summarizes the emission
reductions incorporated in this study's control
scenario. The emissions model forecasts global use
and emissions of CFCs, MCFs, carbon
tetrachloride, HCFCs, and halons under the control
scenario in two major steps.22 For sections
20 Sources of methyl bromide include anthropogenic and
natural sources. Natural sources include the ocean, plants, and
soil.
21 Stolarski, Watson, Testimony to the Senate Commerce,
Science, and Transportation Subcommittee on Science,
Technology, and Space, April 16, 1991; ICF, Incorporated,
Addendum to the 1992 Phaseout Regulatory Impact Analysis:
Accelerating the Phaseout of CFCs, Halons, Methyl Chloroform, Carbon
Tetrachloride, and HCFCs, Prepared for the Stratospheric
Protection Division, Office of Air and Radiation, U.S.
Environmental Protection Agency, September 10, 1993, page 9.
22 See ICF, Incorporated, Addendum to the 1992 Phaseout
Regulatory Impact Analysis: Accelerating the Phaseout of CFCs, Halons,
Methyl Chloroform, Carbon Tetrachloride, and HCFCs, Prepared for
G-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
reflecting the accelerated reduction and phaseout
schedule, the model first applies the chemical
demand growth rates developed for the Regulatory
Impact Analysis: Protection of Stratospheric O^pne (1988),
with the assumption that all nations will comply
with the Copenhagen Amendments. For the other
sections (i.e., section 611, section 609, and the
ecological components of sections 604 and 606), the
model assumes that all nations comply with the
Montreal Protocol and the 1990 London
Agreements.23 Second, the model applies these
growth rates to the release rates for the chemicals.24
This calculation generates the total emissions
estimate for each chemical from 1985 to 2165.25
Stratospheric Ozone Depletion
Modeling and Global Warming
Using the total emissions estimates of ODSs,
the atmospheric lifetimes of the chemicals, and
conversion factors, the EPA calculates stratospheric
chlorine and bromine concentrations. Lifetimes
indicate the length of time that the chlorine and
bromine associated with a specific chemical will
likely remain in the atmosphere. Conversion factors
relate the emissions to stratospheric ozone
changes.26 To translate changes in stratospheric
chlorine and bromine concentrations to changes in
the Stratospheric Protection Division, Office of Air and
Radiation, U.S. Environmental Protection Agency, September
10, 1993, page 9.
23 Ibid, ICF (1993a), 9; Ibid IGF (1992), 3-6 and 3-12.
24 Ibid, ICF (1993a), 1; Ibid ICF (1992), 3-12.
25 ICF Incorporated's Addendum to the 1992 Phaseout
Regulatory Impact Analysis: Acceleratingthe Phaseout of CFCs, Halons,
Methyl Chloroform, Carbon Tetrachloride, andHCFCs (1993a) does
not present emissions estimates.
26 The sources of the ozone depleting potential estimates
include Fisher et al. (1990a) and the 1987 Montreal Protocol.
(ICF Incorporated, Regulatory Impact Analysis: Compliance with
Section 604 of the Clean Air Act for the Phaseout ofO^pne-Depleting
Chemicals, Prepared for the Global Change Division, Office of
Air and Radiation, U.S. Environmental Protection Agency,
1992, page 3-16.)
total column ozone, the EPA modifies Connell's
parameterized version of a one-dimensional
atmospheric chemistry model.27 The EPA calibrates
this model to incorporate the effects of atmospheric
processes (e.g., heterogeneous chemical reactions)
by applying an adjustment factor to the
stratospheric ozone content; this calibration ensures
that the model's global results are consistent with
historical ozone trends for northern hemisphere
middle and high latitudes.28 The EPA's model
assumes that increases in stratospheric chlorine are
the primary causes of the observed ozone change
and that the annual average change in UV-b
predicted by the modeling framework equals the
annual average UV-b change inferred from
observed ozone trends and radiation models.
In conjunction with the calibrated version of
Connell's parameterized model, a second model
incorporates the ability of CFCs, halons, MCF,
carbon tetrachloride, and HCFCs to act as
greenhouse gases (substances that contribute to the
warming of the earth's atmosphere by absorbing
infrared radiation emitted from the earth's
surface).29 This second model, adapted from the
27 Connell, Peter, "A Parameterized Numerical Fit to Total
Column Ozone Changes Calculated by the LLNL 1-D Model
of the Troposphere and Stratosphere," Lawrence Livermore
National Laboratory, Livermore, CA, 1986. (For a description
of the stratospheric ozone model, see U.S. Environmental
Protection Agency, Assessing the Risks of Trace Gases that Can
Modify the Stratosphere., 1987; U.S. Environmental Protection
Agency, Future Concentrations of Stratospheric Chlorine andBromine,
EPA 400/1-88-005, August 1988; and U.S. Environmental
Protection Agency, Regulatory Impact Analysis: Protection of
Stratospheric O^one, August 1, 1988.)
28 Rodriguez, J.M., M.K.W. Ko, and N.D. Sze, " Antarctic
Chlorine Chemistry: Possible Global Implications," Geophysical
Research letters, 15, 1988, pages 257-260.
29 CFC substitutes may indirectly influence global warming
by affecting the energy efficiency of CFC-using capital stock
(e.g., insulating foam and refrigerators). As chemicals that are
more or less energy efficient replace CFCs, total energy demand
could diminish or increase, causing changes in the emissions of
energy-related greenhouse gases. (ICF Incorporated, Regulatory
Impact Analysis: Compliance with Section 604 of the Clean Air Act for
the Phaseout of O^pne-Depleting Chemicals, Prepared for the Global
G-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Goddard Institute of Space Sciences' climate model,
uses estimates for emissions of controlled
chemicals, substitute chemicals, and energy-related
greenhouse gases to calculate changes in global
temperature over time.30 The model adjusts column
ozone and temperature so that they are consistent
with consensus ozone-depleting potential and global
warming potential estimates.31 The model also
reflects 1) radiative and chemical feedback from
water vapor, 2) ocean absorption, 3) atmospheric
circulation effects, and 4) chemical interactions
between substances.32'33
Estimates from stratospheric ozone modeling
may be under- or overestimates, depending on
heterogeneous reactions in the aerosol layer, ozone
depletion in the Arctic, the linearity of the
Change Division, Office of Air and Radiation, U.S.
Environmental Protection Agency, 1992, pages 3-17 and 3-18.)
30 U.S. Environmental Protection Agency, Regulatory Impact
Analysis: Protection of Stratospheric O^pne, August 1, 1988; ICF
Incorporated, Regulatory Impact Analysis: Compliance with Section
604 of the Clean Air Act for the Phaseout of O^pne-Depleting Chemicals,
Prepared for the Global Change Division, Office of Air and
Radiation, U.S. Environmental Protection Agency, 1992, page
3-18.
31 National Aeronautics and Space Administration
Conference Publication 3023, ^!» Assessment Model for Atmospheric
Composition, 1988; ICF Incorporated, Regulatory Impact Analysis:
The National Recycling and Emission Reduction Program (Section 608
of the Clean Air Act Amendments of 1990), Prepared for the
Stratospheric Protection Division, U.S. Environmental
Protection Agency, 1993, page 5-2.
32 Radiative forcing constants and lifetimes form the basis
of the global warming potential estimates, calculated with an
infinite time horizon (Lashof and Ahuja, 1990). Fisher et al.
provide data on direct radiative forcing constants (Fisher et al.,
1990b).
33 ICF Incorporated, Regulatory Impact Analysis: The National
Recycling and Emission Redaction Program (Section 608 of the Clean Air
Act Amendments of 1990), Prepared for the Stratospheric
Protection Division, U.S. Environmental Protection Agency,
1993, page 5-2; ICF, Incorporated, Addendum to the 1992
Phaseout Regulatory Impact Analysis: Accelerating the Phaseout of CFCs,
Halons, Methyl Chloroform, Carbon Tetrachloride, and HCFCs,
Prepared for the Stratospheric Protection Division, Office of
Air and Radiation, U.S. Environmental Protection Agency,
September 10, 1993, page 9.
atmospheric response, and other factors.34 The
EPA estimates that the accelerated reduction and
phaseout schedules of section 604 and 606 will
result in 7 percent less ozone depletion from
baseline levels in 2005 and 47.01 percent less ozone
depletion in 2075.35
Physical Effects
For the physical effects that scientists have
modeled with dose-response functions, we use data
on UV-b radiation and tropospheric ozone to
calculate benefits. We include benefits that scientists
have identified but not yet quantified in a qualitative
discussion. Below we present the benefits
methodology for each section.
Physical Effects: Sections 604
and 606
Table G-4 presents the quantified and
unquantified physical effects estimates of sections
604 and 606, which generate about 98 percent of the
benefits. The quantified benefits include the
following: reduced incidences of mortality and
morbidity associated with skin cancer (melanoma
and nonmelanoma); reduced incidences of cataract
morbidity and the associated pain and suffering;
reduced crop damage associated with UV-b
radiation and tropospheric ozone; and reduced
polymer degradation from UV-b radiation.
34 ICF Incorporated, Regulatory Impact Analysis: Compliance
with Section 604 of the Clean Air Act for the Phaseout of O^pne-
Depleting Chemicals, Prepared for the Global Change Division,
Office of Air and Radiation, U.S. Environmental Protection
Agency, 1992, page 5-8.
35 ICF, Incorporated, Addendum to the 1992 Phaseout
Regulatory Impact Analysis: Accelerating the Phaseout of CFCs, Halons,
Methyl Chloroform, Carbon Tetrachloride, and HCFCs, Prepared for
the Stratospheric Protection Division, Office of Air and
Radiation, U.S. Environmental Protection Agency, September
10, 1993, page 14.
G-15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-4
Benefits of Section 604, 606, and 609
Health Effects- Quantified
Estimate
Basis for Estimate
Melanoma and nonmelanoma
skin cancer
(fatal)
6.3 million lives saved from skin
cancer in the U.S. between 1990
and 2165
Dose-response function based on UV
exposure and demographics of exposed
populations.1
Melanoma and nonmelanoma
skin cancer
(non-fatal)
299 million avoided cases of non-
fatal skin cancers in the U.S.
between 1990 and 2165
Dose-response function based on UV
exposure and demographics of exposed
populations.1
Cataracts
27.5 million avoided cases in the
U.S. between 1990 and 2165
Dose-response function uses a multivariate
logistic risk function based on demographic
characteristics and medical history. 1
Ecological Effects- Quantified
Estimate
Basis for Estimate
American crop harvests
Avoided 7.5 percent decrease from
UV-b radiation by 2075
Dose-response sources: Teramura and Murali
(1986), Rowe and Adams (1987)
American crops
Avoided decrease from
tropospheric ozone
Estimate of increase in tropospheric ozone:
Whitten and Gery (1986). Dose-response
source: Rowe and Adams (1987)
Polymers
Avoided damage to materials from
UV-b radiation
Source of UV-b/stabilizer relationship: Horst
(1986)
Health Effects- Unquantified
Skin cancer: reduced pain and suffering
Reduced morbidity effects of increased UV. For example,
• reduced actinic keratosis (pre-cancerous lesions resulting from excessive sun exposure)
• reduced immune system suppression.
Ecological Effects- Unquantified
Ecological effects of UV. For example, benefits relating to the following:
• recreational fishing
• forests
• overall marine ecosystem
• avoided sea level rise, including avoided beach erosion, loss of coastal wetlands, salinity of estuaries and aquifers
• other crops
• other plant species
• fish harvests
Ecological benefits of reduced tropospheric ozone relating to the overall marine ecosystem, forests, man-made materials,
crops, other plant species, and fish harvests
Benefits to people and the environment outside the U.S.
Effects, both ecological and human health, associated with global warming.
Notes:
1) For more detail see EPA's Regulatory Impact Analysis: Protection of Stratospheric Ozone (1988).
2) Note that the ecological effects, unlike the health effects, do not reflect the accelerated reduction and phaseout
schedule of section 606.
3) Benefits due to the section 606 methyl bromide phaseout are not included in the benefits total because the EPA
provides neither annual incidence estimates nor a monetary value. The EPA does provide, however, a total
estimate of 2,800 avoided skin cancer fatalities in the U.S.
G-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Using the change in UV radiation exposure due to
current and future ozone depletion, we estimate the
number of additional cases of skin cancer
(melanoma and non-melanoma) and cataracts. With
the exception of non-melanoma mortality, which is
estimated as a fixed percentage of non-melanoma
incidence, we use dose-response functions to
develop future incremental skin cancer estimates.
We employ nearly identical approaches in
developing the three dose-response functions for
non-fatal non-melanoma (i.e., basal cell and
squamous cell carcinoma), non-fatal melanoma, and
fatal melanoma.
The first step uses results from studies that
have identified key groups of wavelengths ("action
spectra") within the UV spectrum that are
associated with specific types of health effects (e.g.,
DNA damage).36 Once the appropriate action
spectrum for a health effect is determined, the next
step involves estimating the amount of UV dose
received at various latitudes across the U.S. in the
years prior to ozone depletion. The third step
involves obtaining nationwide skin cancer incidence
and mortality data for each health effect.37 These
data are then combined with the estimated variation
in UV doses across latitudes in a cross-sectional
analysis of the relationship between skin cancer
incidence or mortality and differences in UV
exposure.38 This statistical analysis uses an equation
of the form: (fractional change in incidence) =
(fractional change in UV dose + l)b -1, where b (the
biological amplification factor) equals the percent
change in incidence associated with a one percent
change in dose. The dose-response function for
cataracts is developed similarly.39
The health benefits model uses these dose-
response functions to project incremental cases of
non-fatal non-melanoma, fatal and non-fatal
melanoma, and cataracts that will occur due to
future increases in UV exposure caused by
stratospheric ozone depletion. In essence, future
incremental health effects are estimated by
multiplying the baseline level of each health effect
by the percentage change in UV exposure for
different latitudes in the U.S. times the appropriate
dose-response factor. Because the baseline levels of
all of these UV-related health effects tend to be
higher for older people and for those with lighter
skins, our method for projecting future incremental
skin cancers and cataracts incorporates this and
other relevant factors in its benefits estimates.
Estimates of non-melanoma fatalities are not
calculated from a dose-response function. Instead,
the model assumes that the number of non-
melanoma deaths will be a fixed percentage of the
36 Sedow, R.B., "The Wavelengths of Sunlight Effective in
Producing Skin Cancer: A Theoretical Analysis," Proceedings of
die National Academy of Sciences, 71(9):3363-3366, 1974.
Non-fatal basal and squamous cell non-melanoma
incidence rates were obtained from Scotto, J., T. Fears, and
Fraumeni, "Incidence of Nonmelanoma Skin cancer in the
United States," U.S. Department of Health and Human
Services, (NIH) 82-2433, Befhesda, MD, 1981. Non-fatal
melanoma incidence rates were obtained from National Cancer
Institute SEER Report, 1984. Fatal melanoma incidence rates
were obtained from Pitcher, H.M., "Examination of the
Empirical Relationship Between Melanoma Death Rates in the
United States 1950-1979 and Satellite-Based Estimates of
Exposure to Ultraviolet Radiation," U.S. EPA, Washington, DC,
March 17, 1987, draft.
Non-fatal Non-Melanoma: Scotto, J, and T. Fears,
"Estimating Increases in Skin Cancer Morbidity Due to
Increases in Ultraviolet Radiation Exposure," Cancer
Investigation, 1(2), 119-126,1983. Non-fatal Melanoma: Scotto,
J, and T. Fears, "The Association of Solar Ultraviolet and Skin
Melanoma Incidence Among Caucasians in the United States,"
Cancer Investigation, 5(4), 275-283,1987. Melanoma mortality:
Pitcher, H.M., and J.D. Longstreth, "Melanoma Mortality and
Exposure to Ultraviolet Radiation: An Empirical Relationship,
Environment International, vol. 17, 7-21, 1991.
Cataract prevalence data were obtained from Leske and
Sperduto, "The Epidemiology of Senile Cataracts: A Review,"
American Journal of Epidemiology, Vol. 118, No.2, 152-165,
1983. For information on the dose-response relationship, see
Hiller, R., R. Sperduto, and F. Ederer, "Epidemiological
Associations with Cataract in the 1971-1972 National Health
and Nutrition Survey," American Journal of Epidemiology, Vol.
18, No. 2, 239-249,1983.
G-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
total non-melanoma cases.40 We estimate that from
1990 to 2165 sections 604 and 606 will result in 6.3
million avoided deaths from skin cancer, 27.5
million avoided cataract cases, and 299.0 million
cases of non-fatal skin cancers (melanoma and non-
melanoma) .
Although the evidence linking UV-b and
melanoma is controversial, studies suggest that
exposure to sunlight is a major environmental risk
factor for melanoma. However, uncertainty exists
about three aspects of this relationship: the
appropriate action spectrum (i.e., the relative
contribution of different wavelengths of light to
overall risk), the appropriate dose metric (acute,
intermittent, or chronic), and the importance of age
at exposure. Although UV-b was initially thought
to be solely responsible for melanoma, studies by
Setlow et al. (1993) and Ley (1997) have shown that
UV-a as well as UV-b is a significant factor in the
induction of melanoma. The uncertainty
surrounding the dose-metric stems from the fact
that chronic, cumulative, low-level exposures to
sunlight are not associated with development of
melanoma. Instead, melanoma risk is higher among
those intermittently exposed to sunlight and that
melanoma occurs most frequently on body parts
that are intermittently exposed. Therefore, current
thinking suggests that intermittent, intense bursts of
UV exposure (i.e., sunburns) are an important factor
in the development of melanoma. Epidemiological
studies exploring this hypothesis have confirmed
such an association, though the strength of these
findings may be weakened by recall bias (Berwick
1998). Finally, melanoma may exhibit a significant
latency period; studies such as Holman and
Armstrong (1984) have found that severe early life
exposures to UV are an important risk factor for
melanoma in adults. However, the most recent
study of this effect (Autier and Dore, 1998) found
that childhood exposures are important only in
addition to severe adult exposures.
The effect of the uncertainties in the first two
aspects of the UV/melanoma relationship (action
spectrum and dose metric) on the melanoma
mortality estimates is difficult to determine based on
current information. If melanoma mortality exhibits
a latency period, our results may be overestimated,
because the analysis did not specifically model a
latency period.
To estimate crop damage, we apply earlier
studies on the relationship between crops, UV-b
radiation, and tropospheric ozone to the changes in
UV-b radiation and tropospheric ozone predicted
by the emissions models.41 We estimate that the
avoided increase in damage to American crop
harvests from UV-b radiation by 2075 will equal
about 7.5 percent. To calculate the benefits of
avoided photodegradation of all UV-b sensitive
polymers, we use the Horst et al. study (1986) on the
relationship between UV-b radiation and the
increase in polymer stabilizers needed to mitigate
rigid PVC pipe damage.42
The unquantified effects of sections 604 and
606 include the following: avoided pain and
suffering from skin cancer, ecological effects of UV-
b radiation and tropospheric ozone, human health
and environmental benefits outside the United
States, and changes in pulmonary and respiratory
Non-melanoma mortality estimates are based on
the assumption that one percent of non-melanoma incidence
results in mortality.
41 Sources of dose-response relationship for crops and UV-
b: Teramura and Murali (1986) and Rowe and Adams (1987).
Source of dose-response relationship for crops and tropospheric
ozone: Rowe and Adams (1987). Source of increased
tropospheric ozone estimates: Whitten and Gery (1986). Our
benefits analysis does not include assessing the effects of
tropospheric ozone on forests. Although there are C-R
functions available that would allow an assessment, we could
not use them because of we do not have the necessary measure
of tropospheric ozone changes.
42 Although the Regulatory Impact Analysis: Compliance with
Section 604 of the Clean Air Act for the Phaseout of O^one-Depleting
Chemicals provides monetized benefits estimates of reduced
crop damage from tropospheric ozone and reduced polymer
damage from UV-b, it does not provide a quantified estimate in
non-monetary terms.
G-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
functions from increased tropospheric ozone.43 The
EPA also lists avoided actinic keratosis (pre-
cancerous lesions from excessive sun exposure) as
another unquantifred benefit. Using the first
National Health and Nutrition Examination
(NHANES I), Engles et d. linked increased UV
exposure to increased incidence of actinic
keratosis.44 This study, however, did not provide
sufficient quantitative information relating the
incidence of actinic keratosis to levels of UV
radiation. In addition, several researchers provide
data suggesting that avoided increases in infection
intensity may constitute an unquantified benefit. For
example, Perna et al. associated UV-b exposure with
the reactivation of Herpes virus infections.45 Other
studies have linked UV exposures to reductions in
the ability of animals to control infections with
Leishmania sp. (Giannini and DeFabqj; malaria
(Taylor and Eagles). The yeast Candida (Denkins et
al. and Chung et a/);, the bacterium staphylococcus aureus
(Chung et al).46 Valerie et al. also showed that UV
43ICF Incorporated's Regulatory Impact Analysis: Compliance
with Section 604 of the Clean Air Act for the Phaseout of O^pne-
Depleting Chemicals (1992) does not present the information on
the studies linking tropospheric ozone with pulmonary and
respiratory effects.
44 Engles, A., M.L. Johnson, and S. Haynes,"Health Effects
of Sunlight Exposure in the United States: Results from the
First National Health and Nutrition Examination Survey, 1971-
1974," Archives of Dermatology, Vol. 4, January 1988, pages 72-79.
45 Perm, J.J., M.L. Mannix, J.E. Rooney, A.L. Notkins, and
S.E. Straus, "Reactivation of Latent Herpes Simplex Virus
Infections by Ultraviolet Light: A Human Model," Journal of the
American Academy of Dermatology, 17, 1987, pages 473-478.
46 Chung, H.T., D.C. Lee, S.Y. Im, and R.A Daynes,
"UVR-Exposed Animals Exhibit and Enhanced Susceptibility
to Bacterial and Fungal Infections," Journal of Investigative
Dermatology, Vol. 90, No. 4, April 1988, page 52; Denkins, Y.,
I.J., Fidler, and Kripke, M.L, "Exposure of Mice to UV-B
Radiation Suppresses Delayed Hypersensitivity to Candida
albicans" fhotobiology and Photochemistry, 1989; Giannini, S.H., and
E.G. DeFabo, "Abrogation of Skin Lesions in Cutaneous
Leishmaniasis by Ultraviolet B Irradiation," Leishmaniasis: The
First Centenary (1885-1985) Neiv Strategies for Control, Heart, D.T.
(ed.), NATO ASI Series A: Life Sciences , London, Plenum
Pub., Cos.; Taylor, D.W. and D.A Eagles, "Assessing the
Effects of Ultraviolet Radiation on Malarial Immunity,"
Prepared for Sabotka and Company under EPA contract
number 68-01-7288, subcontract number 132.914.
irradiation of cells grown in vitro and exposed to
sunlight for as little as 10 to 30 minutes can activate
the human immunodeficiency virus type 1 (HIV-
l).47 Scientists, however, have not yet provided a
quantitative relationship between the impact of UV-
b-induced immunosuppression and human
disease.48
Physical Effects: Sections 608. 609.
and 611
For sections 608, 609, and 611 we base the
quantified benefits estimates on the methodology
used for sections 604 and 606, but do not provide
the quantified estimates cited in the RIAs. For
section 608 we use the same emissions,
stratospheric ozone, and UV-b radiation
methodologies used for sections 604 and 606; the
quantified benefits of section 608, however,
comprise only benefits from reduced incidences of
skin cancer morbidity and mortality. For section 609
the benefits estimate is simply a percentage of the
benefits of section 604; in fact, we avoid double
counting by omitting 609 benefits from the
calculation of the total Title VI benefits estimate.
For section 611 we calculate the benefits estimate
with a benefit per kilogram ratio obtained from data
in the Regulatory Impact Analysis: Compliance with Section
604 of the Clean Air Act for the Phaseout of O^pne
Depleting Chemicals (1992). We apply this ratio to the
emissions reduction caused by firms that accelerate
the use of MCF substitutes to avoid labeling.
The unquantified benefits estimates of sections
608, 609, and 611 are the same as the unquantified
benefits of sections 604 and 606, with one
exception. The analysis of section 611 includes two
additional benefits: an increase in available
information regarding ozone-depleting substances
and enhanced implementation and enforcement of
EPA's refrigerant recycling program. Quantified
47Valerie, K., A. Delers, C. Bruck, C. Thiriart, H.
Rosenberg, C. Debouck, and M. Rosenberg, "Activation of
Human Immunodeficiency Virus Type 1 by DNA Damage in
Human Cells," Nature, Vol. 333, May 5, 1988, pages 78-81.
48 Ibid, ICF (1992), 6-26.
G-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
and unquantified benefits of 608 and 611 are
summarized in Table G-5 and G-6 respectively.
Table G-5
Benefits of Section 608
Quantified Health Effects
Skin cancer: fatal and nonfatal
Health Effects-Unquantified
Skin cancer: reduced pain and suffering
Cataracts: reduced morbidity, pain and suffering
Reduced morbidity effects of increased UV. For example,
• reduced actinic keratosis (pre-cancerous lesions resulting from excessive sun exposure)
• reduced immune system suppression.
Ecological Effects- Unquantified
Ecological effects of UV. For example, benefits relating to the following:
• recreational fishing
forests
overall marine ecosystem
• avoided sea level rise - which, in turn, leads to:
decreased beach erosion
decreased loss of coastal wetlands
decreases in the salinity of estuaries and aquifers
• other crops
• other plant species
Other ecological benefits of reduced tropospheric ozone relating to
• the overall marine ecosystem
forests
man-made materials (e.g., degradation of elastomers, textile fibers and dyes, certain paints)
• other crops
• other plant species
Benefits to people and the environment outside the U.S.
Effects, both ecological and human health, associated with global warming.
G-20
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-6
Benefits of Section 611
Health Effects- Quantified
Skin cancer: fatal and nonfatal
Cataracts: reduced morbidity, pain and suffering
Ecological Effects- Quantified
Crops: reduced damage associated with increased UV radiation
Crops: reduced damage associated with increased tropospheric ozone
Polymers: reduced degradation from UV-b radiation
Health Effects- Unquantified
Skin cancer: reduced pain and suffering
Reduced morbidity effects of increased UV. For example,
reduced actinic keratosis (pre-cancerous lesions resulting from excessive sun exposure)
• reduced immune system suppression
Ecological Effects- Unquantified
Ecological effects of UV. For example, benefits relating to the following:
• recreational fishing
• forests
overall marine ecosystem
avoided sea level rise - which, in turn, leads to:
decreased beach erosion
decreased loss of coastal wetlands
decreases in the salinity of estuaries and aquifers
crops in general
• other plant species
• fish harvests
Ecological benefits of reduced tropospheric ozone relating to
• the overall marine ecosystem
• forests
• man-made materials (e.g., degradation of elastomers, textile fibers and dyes, certain paints)
• crops in general
• other plant species
• fish harvests
Benefits to people and the environment outside the U.S.
Enhanced implementation and enforcement of EPA's refrigerant recycling program
Increase in available information regarding ozone-depleting substances (ODSs); consumers who wish to
buy products that do not contain ODSs will be better able to express their preferences through their
purchasing power.
Effects, both ecological and human health, associated with global warming.
G-21
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Valuation
To calculate monetary values of the quantified
benefits, we multiply the physical effects estimates
by the appropriate physical effects value. For the
health benefits, we use $15,000 for the avoided cost
of cataracts, $15,000 for the avoided cost of
melanoma skin cancer, and $5,000 for the avoided
cost of nonmelanoma skin cancer.49 This analysis
employs a value of statistical life of $4.8 million
(1990 dollars), which is the value used to calculate
the criteria pollutant mortality benefits estimate
presented in Chapter 6, Table 6-3. To calculate the
monetary benefits of increased crop yields, the
model multiplies the change in crop yields by crop
values from the Department of Agriculture.50 To
calculate the monetary benefits related to fish, we
apply $739 per ton (1990 dollars) to the increase in
fish harvests.51 We define the polymer benefits as
the avoided loss in consumer surplus associated
with increased polymer prices. We assume that the
cost is proportional to the increase in price
following the addition of stabilizers and that the
price of polymer stabilizers will increase by 1.86
percent for each 25 percent increase in stabilizer.52
With a two percent discount rate, the benefits of
sections 608, 609, and 611 are $671 million, $296
million, and $831 million (1990 dollars),
respectively. We do not separate these values into
their components. The total monetized health
benefits for section 604 and 606 with a two percent
discount rate are $4.2 trillion and the total
monetized ecological benefits are $92.5 billion; thus,
the total benefits of sections 604 and 606 are about
$4.3 trillion. Table G-7 is a tabular summary of the
monetary values of the benefits from sections 604
and 606, which generate about 98 percent of the
monetized benefits.53
49 Wasson, John and Steve Abseck, "Memorandum:
Further Detail on the Costs and Benefits of Phasing Out Ozone
Depleting Substances, EPA Contract No. 68-D4-0103, WA-
205," Prepared for Jim DeMocker, U.S. Environmental
Protection Agency, October 9, 1995.
50 The RIAs do not provide the specific crop values used.
51 The U.S. Department of Commerce provides the fish
values. (Ibid, ICF (1992), 6-29.)
52 Ibid, ICF (1992), 6-41.
53 The dollar year was not available for some cost of illness
estimates in Table G-7. Because these estimates come from a
1988 RIA, we are thus underestimating the monetized benefits
of the health effects associated with these unadjusted values.
G-22
-------
Table G-7
Sections 604 and 606: Valuation of Total Benefits from 1990 to 2165, With a Two Percent Discount Rate
Quantified Effects
Valuation
(1990 dollars)
Source Data
Health Effects
Mortalities from skin cancer (melanoma and
nonmelanoma) in the U.S. (1990-2165)
$3,900 billion Value of statistical life: $4.8 million (1990 dollars). (See Appendix H fora description of
source data.)
Cataract cases in the U.S. (1990-2165)
$72 billion Avoided cost of cataracts: $15,000 (dollar year not provided) Costs include increased
medical costs, increased work loss, increased costs for chores, other indirect social and
economic costs, and willingness to pay to avoid cataracts. Data from literature review,
contacts with health providers, and cataract patient survey. (Source: Rowe etal. 1987)
Nonfatal skin cancer cases (melanoma and
nonmelanoma) in the U.S. (1990-2165)
$220 billion Cost of melanoma skin cancer: $15,000 per case (dollar year not provided); costs of
nonmelanoma skin cancer: $5,000 per case (dollar year not provided). Estimates include
increased medical costs and decreased productivity but do not include costs of caregiving
and chores performed by others. Data from Skin Cancer Focus Group. (See ICF's August
1988 RIA for details.)
Total Health Benefits
$4,200 billion
Ecological Effects
Decrease in American crop harvests from UV-b
radiation by 2075
$49 billion
Crop values from Department of Agriculture.
• Decrease in American crops from tropospheric
ozone by 2075
$28 billion
Crop values from Department of Agriculture.
Damage to polymers from UV-b radiation by 2075
$6 billion Costs are proportional to the increase in polymer prices following the addition of
stabilizers. Price increase of 1.86% expected for a 25% increase in stabilizer.
Total Environmental Benefits
$84 billion
Total Benefits
$4,300 billion
Note : 1) The RIAs do not provide specific crop values
G-23
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Adjustments to Estimates From
Existing Analyses
To ensure consistency with assumptions in
other portions of the current study, we adjust
certain parameters used in existing regulatory impact
assessments of Title VI provisions. We vary several
parameters and compare the resulting net present
values of Title VTs costs and benefits. The Title VI
analysis generates net present values, rather than
annualized values, because annualization incorrectly
imputes benefits of later phaseouts to earlier years.
For example, annualization of the benefits of
phasing out HCFC-22 by 2020 attributes benefits to
years prior to 2030, when neither the costs nor the
benefits of that phaseout have yet occurred.
Consequently, the annualized estimate overstates
benefits at the beginning of the time span and
understates them later.
Discount Rate
Because the benefits occur over several hundred
years, the chosen discount rate can have an
especially large effect on the benefits estimate. In
this analysis we use a five percent discount rate for
our primary estimate. This is consistent with the
retrospective analysis of the Clean Air Act and the
other analyses conducted for the present study.55
We also preform sensitivity tests using discount
rates of three percent and seven percent. Finally,
for consistency with cost and benefit estimates that
we cannot adjust, we calculate aggregate benefits
and costs using a discount rate of two percent.
Table G-8 describes the values we use for the
following parameters: discount rate and value of
statistical life. We are able to adjust key parameters
in the benefits analyses of sections 604, 606, and
609 and the cost analyses of sections 604 and 606.
We cannot adjust parameters for other sections,
however, because we lack annual cost and benefit
data from these sections.54 Moreover, for section
604 and the accelerated phaseout schedule of
section 606, we are unable to modify the parameters
for the analysis of ecological benefits. Nevertheless,
the benefits from sections 604 and 606 constitute
the majority of Title VI benefits (approximately 98
percent at a two percent discount rate) and only
about one percent of the benefits of these sections
result from ecological benefits. In addition, sections
604 and 606 account for about 97 percent of the
costs (at a two percent discount rate).
54 We calculate the benefits of section 609 as a percentage
of the benefits of section 604, so we do not need annual data for
section 609.
55 U.S. Environmental Protection Agency, The Benefits and
Costs of the Clean Air Act, 1970 to 1990, October 1997.
G-24
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-8
Adjustments to Key Parameters of Existing Analyses That Support the 812 Title VI Estimates
Parameter
Discount Rate
(Used for
Costs and
Benefits)
Value of
Statistical Life
(Benefits
Only)
Assumption for
Section 812
Prospective
Central Case: 5%
Sensitivity tests:
3% and 7%
Distribution of
values from $0.6
million to $13.5
million with
expected value of
$4.8 million
Adjustments to Title VI Analyses
Sections 604 and
606
Used Section 812
parameters, plus
2%.
Used Section 812
parameters.
Section 608
2% (No
adjustment
possible.)
$3 to $12 million
Section 609
2% (No
adjustment
possible for
costs.)
Used Section
812
parameters.
Section 611
2% (No
adjustment
possible.)
$3 to $12 million
Value of Statistical Life
The value of statistical life (VSL) is essential for
measuring the monetized benefits associated with a
reduced number of skin cancer mortality cases. We
use a $4.8 million central estimate of VSL, based on
analysis described in Appendix H. To reflect the
uncertainty of the VSL estimates, we employ a
Monte Carlo approach using a Weibull distribution
of VSL estimates as an input. This distribution is
the same as that used in the analysis of criteria
pollutants.56
56 The Weibull distribution has the following parameters:
a location of $0.0, a scale of $5.32 million and a shape of
1.509588.
Cost and Benefit Results
With Adjusted Parameters
Both cost and benefit estimates are sensitive to
the discount rate. As mentioned earlier, the
discount rate has a particularly significant effect on
the benefits estimate because the benefits occur
over several hundred years (1990 to 2165). These
benefits result from actions taken to reduce ozone-
depleting chemical emissions from 1990 to 2075,
the time period over which costa are incurred. In
this section we first present the net present value of
the costs and benefits using the central discount rate
of five percent. We then discuss the results of the
sensitivity tests using discount rates of three and
seven percent. Lastly, we show the results using a
two percent discount rate, which is consistent with
the discount rates used in existing RIAs and which
allows us to compare the costs and benefits of all
the major sections of Title VI, including provisions
where discount rate adjustments are not possible.
The adjusted primary benefit estimate (using a
five percent discount rate) for Title VI is $530
billion and the cost estimate is $30 billion. The
benefits range from $240 billion with a seven
percent discount rate and $1,900 billion with a three
percent discount rate. The costs range from $20
G-25
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
billion to $40 billion, with the same respective
discount rates. The benefits of Title VI greatly
exceed the costs for all discount rate assumptions;
in fact, the benefits are about 20 times greater than
the costs at a five percent discount rate.57 (See Table
G-9) Even the seven percent discount rate
sensitivity test yields total benefits that are 12 times
greater than the costs.
Five Percent Discount Rate
With a five percent discount rate, the expected
human health benefits from sections 604 and 606
are approximately $400 billion. Table G-10 shows
the results of the statistical simulation modeling
analysis; the 5th and 95th percentile values are $100
billion to $900 billion, respectively. The annual
human health benefits from sections 604 and 606,
calculated with a five percent discount rate, steadily
increase until about 2045; they then decrease until
2165, the last year in the analysis. (See Figure G-2.)
About 93 percent of the benefits occur from 2015
to 2165.
The costs of sections 604 and 606 of Title VI
are approximately $26 billion; these sections
generate approximately 97 percent of the costs. The
human health benefits for sections 604 and 606 are
almost 17 times greater than the costs of these
sections.
57 Note that we do not include the costs of the methyl
bromide phaseout of section 606 because existing RIAs do not
provide benefits estimates for this phaseout. The costs,
calculated with a three percent discount rate, are about $1.7
billion.
G-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-9
Costs and Benefits of Sections 604 and 6061
Discount
Rate
2%
3%
5%
7%
Notes:
1.
2.
3.
Benefits (Trillions) Costs (Trillions) Benefits/Cost Ratio
1990 Dollars 1990 Dollars
$4.24 $0.06 76
$1.81 $0.04 44
$0.44 $0.03 17
$0.14 $0.02 8
The cost and benefits estimates associated with a two percent discount rate are the estimates for sections 604,
606, 608, 609, and 61 1 . For the other discount rates the estimates represent the costs and the human health
benefits for sections 604 and 606. These two sections generate the majority of the Title VI costs and benefits
(approximately 98 percent of the benefits and 97 percent of the costs in the two percent discount rate
calculations).
We do not include the costs of the methyl bromide phaseout of section 606 because existing RIAs do not
provide benefit estimates for this phaseout.
In general, the costs occur from 1990 to 2075, while the benefits occur from 1990 to 21 65. (Tables G-11 and G-
12 provide the specific time frame for each section of Title VI.)
Three Percent and Seven Percent
Sensitivity Tests
The expected human health benefits from
sections 604 and 606 are approximately $1,800
billion at a three percent discount rate and $100
billion at a seven percent discount rate. With a three
percent discount rate, the range of expected human
health benefits is $100 billion to $7,800 billion, with
90 percent of these expected benefits between $400
billion and $4,000 billion. By contrast, at a seven
percent discount rate the range of expected human
health benefits is $14 billion to $700 billion, with 90
percent of the expected human health benefits
between $33 billion and $300 billion.
The annual benefits from sections 604 and 606,
calculated with a three percent discount rate, steadily
increase until about 2062; they then decrease till
2165, the last year in the analysis. About 99 percent
of the benefits occur from 2015 to 2165. At a seven
percent discount rate the annual benefits from
sections 604 and 606 steadily increase until about
2038 and then decrease till 2165. About 92 percent
of the benefits calculated with a seven percent
discount rate occur from 2015 to 2165.
The costs of sections 604 and 606 of Title VI
are approximately $41 billion at a three percent
discount rate and approximately $18 billion at a
seven percent discount rate; these sections account
for approximately 97 percent of the total costs. The
costs for sections 604 and 606 are about 44 times
smaller than the human health benefits for these
sections at a three percent discount rate and about
8 times smaller at a seven percent discount rate.
G-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-10
Human Health Benefits for Sections 604 and 606
Billions of 1990 dollars
Range:
$40 to $2,600
Mean:
$400
Median:
$400
Standard Deviation:
$300
Percentiles:
5%
$100
50%
$400
95%
$900
10,000 Trials
.0381
Forecast: G4
Frequency Chart
$1.0
trillions
1 Outlier
- 375
Percent of Benefits from 1990 to 2014
3.64%
Percent of Benefits from 2015 to 2165
93.36%
Note: Estimates calculated with a five percent discount rate.
G-28
-------
$9,000
Figure G-2
Annual Human Health Benefits From Sections 604 and 606 (Discounted at 5%)
Year
G-29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Two Percent Discount Rate
Using benefits estimates from RIAs for sections
608 and 611 and re-calculating the benefits of
sections 604, 606, and 609 with a two percent
discount rate and a VSL of $4.8 million yields a total
benefits estimate of $4.3 trillion (1990 dollars) for
Title VI.58 (See Table G-ll). Of this estimate, $4.2
trillion (98 percent) results from the human health
benefits of sections 604 and 606. The range of
expected benefits from human health improvements
is $0.3 trillion to $20.8 trillion, with 90 percent of
the expected benefits between $0.9 trillion and $9.4
trillion. The annual human health benefits from
sections 604 and 606, calculated with a two percent
discount rate, steadily increase from 1990 until
about 2077; they then decrease till 2165. About 99
percent of the benefits occur from 2015 to 2165.
To estimate the costs of Title VI with a two
percent discount rate, we use the estimates from the
RIAs that analyze sections 608, 609, and 611 and
we re-calculate the costs of sections 604 and 606
using the two percent discount rate.59 Total present
value Title VI costs with a two percent discount rate
are approximately $57 billion. The cost estimate is
about 76 times smaller than the comparable benefits
estimate. Table G-12 lists the cost components.
Undiscounted Benefits
The annual undiscounted benefits from sections
604 and 606 steadily increase until about 2110; they
then decrease in steps until 2165. The steps appear
to be related to the application of 10-year cohort
survival rates for persons born after 2075. See
Figure G-3 for a graphical depiction of the annual
benefits.
58 The RIAs for sections 608 and 611 present a lower
bound benefits estimate that incorporates a $3 million VSL and
an upper bound benefits estimate that incorporates a $12
million VSL. To obtain a point estimate, we weight the lower
bound by 80 percent and the upper bound by 20 percent. (The
benefits of section 608 do not include the HCFC phaseout
benefits because the phaseout is too complex to model.) In
addition, note that the benefits of section 609 are a subset of the
benefits of sections 604 and 606, so we do not add them
separately to the benefits of the other sections to obtain the
total benefits.
59 The estimate for section 611 is an average of the lower
bound, which assumes that only companies using ODSs as
solvents will label products, and an upper bound, which
assumes that 10 times as many companies use solvent-cleaned
products and will need to label their products. Also, the costs
for section 608 do not include the costs of the HCFC phaseout.
G-30
-------
Table G-11
Summary of Benefits of Title VI with a Two Percent Discount Rate and $4.8 Million VSL
Section
Benefits
(Millions)
1990 Dollars
Notes
Years During
Which Benefits
Accrue
604 & 606 Class I Phaseout
- Reduced Mortality, Cataracts, and
Non-Fatal Cancers
- Methyl Bromide Reductions
- Ecological Benefits
$4,338,000
$4,243,000
$84,000
No information currently available.
1990 to 2165
1994 to 2160
1989 to 2075
605
Class II Phaseout
No information currently available.
608 National Recycling and
Emission Reduction Program
$670 1) Only health effects are monetized.
2) The RIA for section 608 does not include benefits of HCFC phaseout
because this phaseout is too complex and predicting baseline innovation
is too difficult.
3) The benefits estimate listed is the weighted average of the benefits
calculated with $3 million and $12 million for the VSL. (The $3
million estimate has a weight of 0.8 and the $12 million estimate has a
weight of 0.2)
4) Benefits reflect the accelerated phaseout schedule.
1994 to 2165
609 Servicing of Motor Vehicle Air
Conditioners
$300 The benefits of section 609 are a subset of the benefits of sections 604 and 1991 to 2075
606; we calculate section 609 benefits as 0.00682% of the benefits of 604
and 606 combined. (This percentage is the total benefits of 609 in 1989
dollars divided by the total 604 benefits in 1989 dollars.)
611 Labeling
$830 1) The benefits do not reflect the accelerated phase-out schedule.
2) The benefits estimate listed is the weighted average of the benefits
calculated with $3 million and $12 million for the value of a statistical
life. (The $3 million estimate has a weight of 0.8 and the $12 estimate
has a weight of 0.2)
1989 to 2075
TOTAL
$4,339,000
Notes: 1) All benefits expressed in 1990 dollars using the implicit GDP deflator from the 1998 Economic Report of the President.
2) The benefits of section 611 and the ecological benefits of section 604 do not reflect the accelerated reduction and phaseout schedule of section 606.
G-31
-------
TableG-12
Summary of Costs for Title VI by Section with a Two Percent Discount Rate
604 &
606
605
608
609
611
Section
Class I Phaseout
Class II Phaseout
National Recycling and
Emission Reduction
Program
Servicing of Motor
Vehicle Air
Conditioners
Labeling
Cost
Estimate
(Millions,
1990 Dollars)
$56,000
$1,200
$14
$250
Notes
Does not include cost of methyl bromide reductions.
No information currently available.
1 ) We convert costs from 1 991 to 1 990 dollars using the GDP deflator from the 1998 Economic
Report of the President.
2) The RIA for Sect. 608 does not include costs of HCFC phaseout.
3) At a 4% discount rate the costs are $1,074.38 million, and at a 7% discount rate the costs are
$853.91.
1 ) We convert from 1 991 to 1 990 dollars using the GDP deflator from the 1998 Economic Report of
the President.
2) To avoid counting the same costs for both section 604 and 609, we include only the operator
training and equipment certification costs of section 609 here.
1) The costs are probably in 1990 dollars, but this is unclear.
2) Most costs are one-time costs.
3) The cost estimate is an average of the lower bound, which assumes that only firms using ODSs
as solvents will label products, and an upper bound, which assumes that 10 times as many firms
will need to label their products because they incorporate solvent-cleaned products.
Years
During
Which
Costs
Accrue
1990 to
2075
1994 to
2015
1992 to
2008
1994 to
2000
TOTAL
$57,000
Notes: 1) The costs listed above for sections 604 & 606 do not include methyl bromide costs, which equal $1.7 billion with a 3% discount rate, because the RIA did not present the
corresponding benefits.
2) The costs of sections 609 and 611 do not reflect the accelerated reduction and phaseout schedules of section 606.
G-32
-------
Figure G-3
ANNUAL UNDISCOUNTED HUMAN HEALTH BENEFITS OF SECTIONS 604 AND 606
$600,000
$500,000 -
$400,000 -
$300,000 --
i)
m
$200,000 - -
$100,000 --
Year
G-33
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Limitations And Uncertainties
An analysis of programs that protect
stratospheric ozone necessarily considers impacts
over hundreds of years, introducing a wide range of
uncertainty in the estimates of costs and benefits.
There are clearly limitations and uncertainties in
predicting the advancement of both medical
treatment and ODS substitution technologies over
time, modeling averting behavior of individuals in
response to changes in UV-b radiation levels,
anticipating the results of new information that
might alter the modeling of stratospheric ozone
depletion and formation, and forecasting economic
parameters such as the growth in GDP and the
valuation of health risk reduction.
We have not attempted to characterize the
impact of all the uncertainties and limitations
inherent in this type of analysis. For more detail, we
refer the reader to the source material in EPA's
RIAs for Title VI provisions and the descriptions of
the cost and benefit modeling approaches found
there. As part of this analysis, however, we conduct
selected quantitative sensitivity tests and literature
reviews to characterize several major uncertainties in
the cost and benefits analyses presented here. The
discussion that follows includes characterization of
the following: issues in long-term discounting;
limitations in the cost modeling; and limitations in
the benefits modeling, including the modeling of
averting behavior.
Long-term Discounting
As demonstrated above, the discount rate can
have an important effect on the estimation of costs
and benefits that accrue over a long period of time.
Long-term discounting may present special
problems that are worth exploring in some detail
through sensitivity tests of alternative discount rate
assumptions. For example, some economic
literature suggests that accounting for
intergenerational transfers in a manner different
from intragenerational transfers may be
appropriate.60 One possible rationale for treating
long-term, intergenerational transfers differently is
that an individual's rate of time preference (which
presumably applies only for his or her lifetime, or
intragenerationally) may differ from his or her
bequest motive for future generations. At least one
empirical study suggests that individuals may
implicitly apply lower discount rates for programs
where the benefits accrue later in time.61 In addition,
people may attribute the same level of importance
to all events that occur in the far-distant future,
regardless of the relative position of these events in
time. According to Weitzman, analysts should apply
the lowest possible nonnegative rate to events in the
far-distant future.62
Although some of the arguments for using an
alternative discounting procedure for long-term
benefits and costs are persuasive, implementation of
an alternative procedure is not straightforward.
There appears to be little guidance in the existing
economic literature on the key issues of what
discount rate to use for long-term versus short-term
discounting as well as when to alter the discount
rate. Recently drafted EPA guidance on the
conduct of economic analyses, however, suggests
that longer-term discount rates might be
60 Arrow, K.J., W.R. Cline, KG. Maler, M. Munasinghe, R.
Squitieri, and J.E. Stiglitz, "Intertemporal Equity, Discounting,
and Economic Efficiency," Climate Change 1995: Economic and
Social Dimensions of Climate Change. Edited by J.P. Bruce, H. Lee,
and E.F. Haites, Cambridge: Cambridge University Press, 1996;
Lind, Robert C., "Intergenerational Equity, Discounting, and the
Role of Cost-Benefit Analysis in Evaluating Global Climate
Policy," Integrative Assessment oj"Mitigation, Impacts, and Adaptation
to Climate, Edited by N. Nakicenovic, et aL, Laxenburg, Austria:
International Institute of Applied Systems Analysis, 1994;
Schelling, Thomas C., "Intergenerational Discounting," Energy
Policy. 23(4/5), 1995, pages 395-401; Solow, Robert, "An Almost
Practical Step Toward Sustainability," Paper presented at the
Fortieth Anniversary of Resources for the Future, October 8,
1992, in Washington, D.C.
61 Cropper, Maureen L., Sema K. Aydede, and Paul R.
Portney, "Discounting Human Lives." American Journal of
lies, December 1991: 1410-1414.
62 Weitzman, Martin L., "Why the Far-Distant Future
Should Be Discounted at its Lowest Possible Rate," Journal of
Environmental Economics and Management, Volume 36, 201-208
(1998).
G-34
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
approximated through Ramsey Rule discounting,
using effective annual discount rates of from 0.5 to
3.0 percent.63 Throughout our presentation of Title
VI cost and benefit results, we use a five percent
discount rate for our primary estimate. We also
calculate alternative estimates using three and seven
percent discount rates. These discount rates
maintain consistency with other analyses of the
prospective.
Costs
Major uncertainties in the cost estimates result
where it is difficult to predict the pace and nature of
innovation in key industries. To the extent the
models used do not quantify transition costs in the
long term, the uncertainty in the cost estimates
increases. In addition, predicting the responses of
manufacturers to the different sections of Title VI
is difficult. Table G-13 lists the primary causes of
uncertainty in the cost estimates.
63 We use the discount rate of 0.5 percent as a low
nonnegative discount rate for events in the far-distant future.
(Frank Arnold et a/.'s Draft Final Report: Discounting in
Environmental Policy Evaluation supports long-term discount rates
from 0.5 percent to 3.0 percent.)
G-35
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
TableG-13
Major Limitations of Existing Cost Analyses for Title VI
Sections 604*, 606, and 609
Limitations
Effects on Cost Estimates
The assumption of little or no technical innovation in ODS-
related industries over the time span of the analysis may be
inaccurate.
Inclusion of innovation may decrease the cost estimate.
The model does not quantify transition costs, such as
temporary layoffs, administrative costs, and the costs of
unknown environmental hazards created by the use of
alternatives to CFCs.
Inclusion of transition costs may increase the cost
estimate.
Section 608
The assumptions of the capital and operating costs of
recovery devices and the time necessary to perform
recycling operations are largely hypothetical.
A better understanding of the capital and operating costs,
as well as the time necessary for recycling, may either
increase or decrease cost estimates.
The extent of compliance with recycling rates mandated by
the Venting Prohibition is very uncertain.1 Therefore, the
baseline assumptions regarding the percentages of CFCs
that are actually recycled are also very uncertain.
A better ability to forecast recycling rates may either
decrease or increase cost estimates.
The baseline recycling value assumes no innovation in
recycling technologies through 2017, which may be
inaccurate.
Including innovation in recycling technologies may
decrease the cost estimates.
Section 611
Manufacturers' responses to the labeling requirement may
include labeling, reformulating products, ceasing production,
or petitioning for an exemption to the labeling requirement.
Predicting the frequency of these responses is difficult.
A better ability to forecast manufacturers' responses may
either decrease or increase cost estimates.
Note: *We do not include the costs of the methyl bromide phaseout in the total cost estimate because the RIAs do not provide
the benefits of this phaseout. The cost estimates for the methyl bromide phaseout are uncertain, because the model
assumes that the demand for output manufactured with methyl bromide is perfectly inelastic and that the methyl bromide
production industry is perfectly competitive. While these assumptions may be unrealistic, they allowed the analysis to
focus on consumer impacts and ignore effects on output markets.
1 The Venting Prohibition is essentially a recovery and recycling requirement. For more detail see ICF (1993).
G-36
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Benefits
Several factors contribute to uncertainty in the
benefit estimates. (See Table G-14 for a list of
major limitations.) For example, scientists have an
incomplete understanding of the processes that
govern ozone depletion and affect exposure to UV-
b radiation. In addition, the dose-response
coefficients relating UV-b exposure to melanoma
skin cancer and cataracts are difficult to estimate.
Scientists have not yet developed quantified dose-
response relationships for some benefits, such as
reduced damage to the immune system from UV-b
radiation. As a result, the benefit estimates may
either overestimate or underestimate the true
benefits of Title VI provisions.
Data limitations also impede attempts to
monetize certain benefits. For example, there are
well established concentration-response functions
that would allow us to measure the effects of
tropospheric ozone on forests. We are, however,
unable to use the CR functions because we do not
have the necessary measured changes in ozone. As
a result, we are also unable to monetize these
benefits.
2050 may be almost twice the estimated cost of
averting behavior (application of sunscreen).65 To
estimate benefits, the Title VI analysis relies on
epidemiological studies, which incorporate averting
behavior as currently practiced. Omission of future
increases in averting behavior may nonetheless
overstate the benefits of reduced emissions of
ozone-depleting chemicals.66 The analysis may
understate the benefits, however, if individuals alter
their behaviors in ways that could increase exposure
or risk (e.g., sunbathing more frequently and/or for
longer periods) ,67
Another difficulty involves the long term nature
of the study. Predicting invention, research and
development, producer and consumer responses to
price changes, and technological change for the next
century and a half is highly speculative. Predicting
major natural events that influence the effects of
stratospheric ozone depletion is also difficult.64 Our
inability to forecast with accuracy may cause the
benefit estimate to be too high or too low.
Lastly, the quantitative analysis of Title VI does
not account for potential increases in averting
behavior (e.g., people's efforts to protect themselves
from UV-b radiation). Murdoch and Thayer (1990)
estimate that the cost-of-illness estimates for
nonmelanoma skin cancer cases between 2000 and
64 For example, volcanic eruptions increase dust levels,
which may affect risks from stratospheric ozone depletion.
65 Murdoch, James C. and Mark A. Thayer, "The Benefits
of Reducing the Incidence of Nonmelanoma Skin Cancers: A
Defensive Expenditures Approach," Journal of Environmental
Economics and Management, 1990, pages 107-119.
Although Dr. Marianne Berwick, an epidemiologist at
Memorial Sloan-Kettering Cancer Center in New York, issued
a study indicating that sunscreens are ineffective in preventing
melanoma, many dermatologists contest this assertion ("Studies
Doubt Sunscreens Stop a Cancer," The New York Times,
February 2, 1998, page 19; Berwick , Marianne, Sunscreens and
Skin Cancer: The Epidemiologcal Evidence, February 17, 1998; Boyd,
Christopher, "Sunscreen Research Bums Up Skin Specialists:
Doctors Fear Controversial Report Will Confuse Public,"
Orlando Sentinel, March 1, 1998, page A4).
/., 1999.
G-37
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table G-14
Major Limitations of Existing Benefits Analyses for Title VI
Limitations
Effects on Benefits Estimates
Scientists have an incomplete understanding of
the chemical and physical processes that cause ozone
depletion,
• the relationship between ozone depletion and exposure to
ultraviolet radiation (UV-b), and
the dose-response coefficients relating UV-b exposure to
melanoma skin cancer and cataracts.
A better understanding may either increase or decrease
the benefits estimate.
Scientists have not yet developed quantified dose-response
relationships for some benefits, such as reduced damage to
the immune system from UV-b radiation.
Additional dose-response relationships will increase the
benefits estimate.
The long term nature of the studies introduces a significant
degree of uncertainty. For example, predicting innovation,
research and development, producer and consumer responses
to price changes, technological change, and major natural
events for the next 100 to 150 years is difficult.
A better ability to forecast future events may either
decrease or increase benefits estimates.
Although truncation of benefits and cost streams is necessary
for the analysis, it does influence the size of the benefit and
cost estimates.
The current method of truncating benefit and costs
streams results in a greater underestimation of benefits
than costs.
The RIAs do not account for averting behavior (i.e., people's
efforts to protect themselves from UV-b radiation) or behavior
increasing exposure or risk (e.g., increased sunbathing).
Inclusion of averting behavior may decrease the benefits
estimate, while inclusion of enhancing behavior may
increase the benefits estimate.
Not all RIAs for Title VI include comprehensively monetized
benefits, due, in part, to key data gaps (e.g., accepted
concentration response functions for ozone effects on forests).
More comprehensive monetization will increase the
benefits estimate.
Section 608
The extent of compliance with recycling rates mandated by the
Venting Prohibition is very uncertain. Therefore, the baseline
assumptions regarding the percentages of CFCs that are
actually recycled are also very uncertain.
A better ability to forecast recycling rates may either
decrease or increase benefits estimates.
The baseline recycling value assumes no innovation in
recycling technologies through 2017, which may be inaccurate.
Including innovation in recycling technologies may
increase the benefits estimate.
Section 611
Manufacturers' and consumers' responses to labeling rules are
difficult to predict.
An enhanced ability to forecast their responses may either
increase or decrease the benefits estimate.
Benefits attributed to labeling regulations may actually result
from other circumstances as well.
Consequently, the benefits resulting from this rule may be
less than the estimate included in the RIA.
Although some sectors may reduce the use of MCF as a result
of the labeling rule, other sectors may increase their use of this
substance.
A better ability to predict people's actions may decrease
the benefits estimate.
G-38
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
Alberini, A., M. Cropper, T. Fu, A. Krupnick,J. Liu, D. Shaw, and W. Harrington. 1997. "Valuing Health
Effects of Air Pollution in Developing Countries: The Case of Taiwan."
Economics and Management. Vol. 34. No. 2. October, pages 107-126.
Arnold, Frank, Frances G. Sussman, and Leland B. Deck. 1997. Draft Final Report: Disci
Policy Evaluation.
Arrow, K.J., W.R. Cline, K.G. Maler, M. Munasinghe, R. Squitieri, and J.E. Stiglitz. 1996. "Intertemporal
Equity, Discounting, and Economic Efficiency." Climate Change 1995: Economic and Social Dimensions of
Climate Change. Edited by J.P. Bruce, H. Lee, and E.F. Haites. Cambridge: Cambridge University
Press.
Autier, Philippe, Jean-Francois Dore, Sylvie Negrier, Daniele Lienard, Renato Panizzon, Ferdy J. Lejeune,
David Guggisberg, and Alexander M. M. Eggermont. 1999. "Sunscreen Use and Duration of Sun
Exposure: a Double-Blind, Randomized Trial." Journal of the National Cancer Institute. Vol. 91. No. 15.
August 4. pages 1304-1309.
Autier, P., and J.F. Dore. 1998. "Influence of Sun Exposures During Childhood and During Adulthood on
Melanoma Risk." Int] Cancer. Vol. 77. pages 533-537.
Berwick, Marianne. 1998. Sunscreens and Skin Cancer: The Epidemiological Evidence.
Berwick, M. 1998. "Epidemiology: Current Trends, Risk Factors, and Environmental Concerns." In Cutaneous
Melanoma (eds) Batch, CM, Houghton, AN, Sober, AJ and Soong, S-J, St. Louis: Quality Medical
Publishing, Inc.
Boyd, Christopher. 1998. "Sunscreen Research Burns Up Skin Specialists: Doctors Fear Controversial Report
Will Confuse Public." Orlando Sentinel. March 1. page A4.
Bureau of Economic Analysis. 1998. http://www.bea.doc.gov/bea/dn/0898nip3/tab2a.htm..
Census Bureau. 1998. http://www.census.gov/population/projections/nation/npaltsrs.txt.
Chung, H.T., D.C. Lee, S.Y. Im, and R.A. Daynes. 1988. "UVR-Exposed Animals Exhibit and Enhanced
Susceptibility to Bacterial and Fungal Infections." Journal of Investigative Dermatology. Vol. 90. No. 4.
April, page 52.
Congressional Budget Office, Congress of the United States. 1998. The Economic and Budget Outlook: Fiscal
Years 1999 to 2008. January.
Connell, Peter. 1986. "A Parameterized Numerical Fit to Total Column Ozone Changes Calculated by the
LLNL 1-D Model of the Troposphere and Stratosphere." Lawrence Livermore National Laboratory.
Livermore, CA.
G-39
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Council of Economic Advisers. 1998. Table B-3, "Quantity and Price Indexes for Gross Domestic Product,
and Percent Changes, 1959-97." Economic Report of the President.
Council of Economic Advisers. 1997. Table B-32, "Population by Age Group, 1929-96." Economic Report of the
Cropper, Maureen L., Sema K. Aydede, and Paul R. Portney. 1991. "Discounting Human Lives." American
Journal of Agricultural Economics.
Denkins, Y., I.J., Fidler, and Kripke, M.L. 1989. "Exposure of Mice to UV-B Radiation Suppresses Delayed
Hypersensitivity to Candida albicans" Photobiology and Photochemistry.
Engles, A., M.L.Johnson, and S. Haynes. 1988. "Health Effects of Sunlight Exposure in the United States:
Results from the First National Health and Nutrition Examination Survey, 1971-1974." Archives of
Dermatology. Vol. 4. January, pages 72-79.
Fisher, D. A., C. H. Hales, D. L. Filkin, M. K.W. Ko, N. D. Sze, P. S. Connell, D. J. Wuebbles, I. S. A.
Isaksen, and F. Stordal. 1990a. "Model Calculations of the Relative Effects of CFCs and their
Replacements on Stratospheric Ozone," Nature, 344, pages 508-512.
Fisher, D. A., C. H. Hales, W. C. Wang, K. W. Ko, and N. D. Sze. 1990b. "Model Calculations of the Relative
Effects of CFCs and their Replacements on Global Warming." Nature, 344. pages 513-516.
Giannini, S.H., and E.G. DeFabo. 1987. "Abrogation of Skin Lesions in Cutaneous Leishmaniasis by
Ultraviolet B Irradiation." Eeishmaniasis: The First Centenary (1885-1985) New Strategies for Control,
Heart, D.T. (ed.). NATO ASI Series A: Life Sciences, London, Plenum Pub., Cos.
Horst, R., K. Brown, R. Black, and M. Kianka, Mathtech, Incorporated. 1986. The Economic Impact of Increased
UV-B Radiation on Polymer Materials: A Case Study of Rigid PVC. June.
Hunter, J.R., S.E. Kaupp, and J.H. Taylor. 1982. "Assessment of effects of radiation on marine fish larvae."
The Role of Solar Ultraviolet Radiation in Marine Ecosystems. Edited by J. Calkins.
ICF Incorporated. 1993a. Addendum to the 1992 Phaseout Regulatory Impact Analysis: Accelerating the Phaseout of
CFCs, Halons, Methyl Chloroform, Carbon Tetrachloride, andHCFCs. Prepared for the Stratospheric
Protection Division, Office of Air and Radiation, U.S. Environmental Protection Agency.
ICF Incorporated. 1993. Regulatory Impact Analysis: The National Recycling and Emission Reduction Program (Section
608 of the Clean Air Act Amendments of 1990). Prepared for the Stratospheric Protection Division, U.S.
Environmental Protection Agency.
ICF Incorporated. 1992. Regulatory Impact Analysis: Compliance mth Section 604 of the Clean Air Act for the Phaseout
of 0 ^one-Depleting Chemicals. Prepared for the Global Change Division, Office of Air and Radiation,
U.S. Environmental Protection Agency.
G-40
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
ICF Incorporated. 1991. Draft: Regulatory Impact Analysis of the Rule Requiring Labeling of Products Containing or
Manufactured with O^pne Depleting Substances. Prepared for the Global Change Division, Office of Air
and Radiation, U.S. Environmental Protection Agency.
ICF Incorporated. 1991. Section 609 of the 1990 Clean Air Act: Refrigerant Recycling for Mobile Air Conditioners:
Cost Benefit Analysis and Regulatory Flexibility Analysis. Prepared for the Division of Global Change, U.S.
Environmental Protection Agency.
Johannesson, M. and P. Johansson. 1997. "Quality of Life and the WTP for an Increased Life Expectancy at
an Advanced Age." Journal of Public Economics. Vol. 65. pages 219-228.
Jones-Lee, M.W., M. Hammerton, and P.R. Philips. 1985. "The Value of Safety: Results of a National Sample
Survey." The Economic Journal. Vol. 95. March, pages 49-72.
Lashof, D. A. and D.R. Ahuja. 1990. "Relative Global Warming Potentials of Greenhouse Gas Emissions."
Nature, 344. pages 529-531.
Ley, R.D. 1997. "Ultraviolet Radiation A-induced Precursors of Cutaneous Melanoma in Mondelphis
Domestica." Cancer Research. Vol. 57. pages 3682-3684.
Lind, Robert C. 1994. "Intergenerational Equity, Discounting, and the Role of Cost-Benefit Analysis in
Evaluating Global Climate Policy." Integrative Assessment of Mitigation, Impacts, and Adaptation to Climate.
Edited by N. Nakicenovic, W.D. Nordhaus, R. Richels, and F.L. Toth. Laxenburg, Austria:
International Institute of Applied Systems Analysis.
Loehman, E. and V De. 1982. "Application of Stochastic Choice Modeling to Policy Analysis of Public
Goods: A Case Study of Air Quality Improvements." The Review a/Economics and Statistics. Vol. 64.
pages 474-480.
"Montreal Protocol on Substances that Deplete the Ozone Layer." 1987. 26 ILM 1550. September 16.
Murdoch, James C. and Mark A. Thayer. 1990. "The Benefits of Reducing the Incidence of Nonmelanoma
Skin Cancers: A Defensive Expenditures Approach." Journalof'EnvironmentalEconomics and
Management, pages 107-119.
National Aeronautics and Space Administration Conference Publication 3023. 1988. An Assessment Modelfor
Atmospheric Composition.
Perna, J.J., M.L. Mannix, J.E. Rooney, A.L. Notkins, and S.E. Straus. 1987. "Reactivation of Latent Herpes
Simplex Virus Infections by Ultraviolet Light: A Human Model." Journal of the American Academy of
y. 17. pages 473-478.
Rodriguez, J.M., M.K.W. Ko, and N.D. Sze. 1988. " Antarctic Chlorine Chemistry: Possible Global
Implications." Geophysical Research Letters, 15. pages 257-260.
Rowe, R.D. and R. M. Adams. 1987. Analysis of Economic Impacts of Lower Crop Yields Due to Stratospheric O^pne
Depletion, Draft Report. Prepared for the U.S. Environmental Protection Agency.
G-41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Rowe, R.D., T.N. Neithercut, and W.D. Schulze. 1987. Economic Assessment of the Impacts of Cataracts, Draft
Report. Prepared for the U.S. Environmental Protection Agency.
Schelling, Thomas C. 1995. "Intergenerational Discounting." Energy Policy. 23(4/5).
Scotto, Fears, and Fraumeni, U.S. Department of Health and Human Services, NIH. 1981. "Incidence of
Nonmelanoma Skin Cancer in the United States." pages 2, 7, and 13.
Setlow, R.B., Grist, E., Thompson, K., and Woodhead, A.D. 1993. "Wavelengths Effective in the Induction
of Malignant Melanoma." Proceedings of the National Academy of Sciences USA. Vol. 71. pages 3363-3366.
Solow, Robert. 1992. "An Almost Practical Step Toward Sustainability." Paper presented at the Fortieth
Anniversary of Resources for the Future. October 8. Washington, D.C.
Stolarski, Watson. 1991. Testimony to the Senate Commerce, Science, and Transportation Subcommittee on
Science, Technology, and Space. April 16.
"Studies Doubt Sunscreens Stop a Cancer." 1998. The New York Times, February 2. page 19.
Taylor, D.W. and D.A. Eagles. (No date.) "Assessing the Effects of Ultraviolet Radiation on Malarial
Immunity." Prepared for Sabotka and Company under EPA contract number 68-01-7288,
subcontract number 132.914.
Teramura, A.H. and N.S. Murali. 1986. "Intraspecific Differences in Growth and Yield of Soybean Exposed
to Ultraviolet-B Radiation Under Greenhouse and Field Conditions." Environmental and Experimental
botany.
Unsworth, Robert, Jim Neumann, and W. Eric Browne, Industrial Economics, Inc. 1990. "Review of
Existing Value of Life Estimates: Valuation Document." Prepared for Jim DeMocker, Office of
Policy Analysis and Review, Office of Air and Radiation, U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency. 1999. http://www.epa.gov/spdpublc/mbr/harmoniz.html. March
26.
U.S. Environmental Protection Agency. 1998. http://www.epa.gov/ttn/oarpg/t6/fact_sheets/66.txt. March
25.
U.S. Environmental Protection Agency. 1997. Benefits and Costs of the Clean Air Act, 1970 to 1990. October.
U.S. Environmental Protection Agency. 1993. Part 2: The Costs and Cost-Effectiveness of the Proposed Phaseout of
U.S. Environmental Protection Agency. 1989. Costs and Benefits of Phasing Out Production ofCFCs andHalons in
the United States. November 3.
U.S. Environmental Protection Agency. 1988. Regulatory Impact Analysis: Protection of Stratospheric O^pne. August
1.
G-42
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
U.S. Environmental Protection Agency. 1988. Future Concentrations of Stratospheric Chlorine and Bromine, EPA
400/1-88-005. August.
U.S. Environmental Protection Agency. 1987. Assessing the Risks of Trace Gases that Can Modify the Stratosphere.
Valerie, K., A.Delers, C. Bruck, C. Thinart, H. Rosenberg, C. Debouck, and M. Rosenberg. 1988. "Activation
of Human Immunodeficiency Virus Type 1 by DNA Damage in Human Cells." Nature. Vol. 333.
May 5. pages 78-81.
Viscusi, W.K. and W. Evans. 1990. "Utility Functions that Depend on Health Status: Estimates and
Economic Implications." American Economic Review. Vol. 80. No. 2. pages 353-374.
Wasson, John and Steve Abseck. 1995. "Memorandum: Further Detail on the Costs and Benefits of Phasing
Out Ozone Depleting Substances, EPA Contract No. 68-D4-0103, WA-205." Prepared for Jim
DeMocker, U.S. Environmental Protection Agency. October 9
Weitzman, Martin L. 1998. "Why the Far-Distant Future Should Be Discounted at its Lowest Possible Rate."
Journal of Environmental Economics and Management. Vol. 36. pages 201-208.
Whitten, G.Z. and M. Gery. 1986. "Effects of Increased UV Radiation on Urban Ozone," EPA Report
600/9-8 6016. Presented at EPA workshop on global atmospheric change and EPA planning. Edited
by H.Jeffries.
G-43
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[This page left blank intentionally.]
G-44
-------
Valuation of Human Health
and Welfare Effects of
Criteria Pollutants
This appendix describes the derivations of the
economic valuations for health and welfare endpoints
considered in the benefits analysis. It includes three
primary sections. First, we introduce the method for
monetizing improvements in health and welfare.
Second, we summarize dollar estimates used to value
benefits and outline the derivation of each estimate.
Valuation estimates were obtained from the literature
and reported in dollars per case avoided for health
effects, and dollars per unit of avoided damage for
welfare effects. Economic valuations are characterized
in terms of a central (point) estimate as well as a
probability distribution which reflects the uncertainty
around the central estimate. Third, we present the
results of the economic benefits analysis. All dollar
values are in 1990 dollars. This third section
concludes with an exploration of the uncertainties in
valuing the benefits attributable to the CAAA.
Methods Used to Value Health
And Welfare Effects
The general approach to benefits analysis involves
a three-step process— (i) identification of potential
physical effects (i.e., individual health and welfare
endpoints); (ii) quantification of significant endpoints;
and (iii) monetization of benefits. The first two steps,
identification and quantification of physical effects, are
described in Appendix D, Human Health and Welfare
Effects of Criteria Pollutants. The third step is
detailed in this appendix. Monetization of benefits
attributed to the CAAA involves applying dollar
estimates obtained from economic literature to
individual health and welfare endpoints relevant for
the 812 prospective analysis. As context to
understanding the methodology for transferring
estimated values of physical effects, this section
provides a brief discussion of the theoretical economic
foundation of, and general approach to, valuing the
benefits of improved air quality.
Economists equate the dollar value of a benefit to
the level of well-being an individual enjoys from the
provision or consumption of a particular good or
composite good (i.e., bundle or mix of goods). A
fundamental assumption in economic theory is that
individuals can trade between different consumption
levels of these goods, services, or money, and
maintain the same level of welfare. Typically, this
willingness to trade-off between goods is measured as
willingness to pay (WTP) or willingness to accept
compensation (WTA). These measures are essentially
dollar equivalents to changes in the level of
consumption of a good or service so that the
individual maintains the same level of well-being. In
other words, the individual is indifferent between his
or her current bundle of goods and the alternative
bundle of goods.
While WTP and WTA represent an individual's
own assessment of the dollar value of better health,
they are not necessarily equivalent measures.1 WTP,
in the case of health, is the largest amount of money
a person would pay to obtain an improvement (or
avoid a decline) in health. When faced with two
1 The measures differ for several reasons. For example the
measures have different points of reference from which to
evaluate changes in welfare. WTP's reference point is the level of
utility without the improvement. WTA's reference point is the
level of utility with the improvement. Moreover, the measures have
different upper bound constraints. WTP measures what a person
would pay to obtain better health and is bound by the person's
wealth and income. WTA, on the other hand, measures what a
person must be paid to forego better health. WTA does not have
an upper bound, but it must be at least as large as WTP.
Economists, however, do not expect significant differences
between WTP and WTA when the dollar amounts are small
relative to the individual's wealth and income.
H-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
options, to either (1) pay a certain dollar amount to
enjoy the health improvement or (2) abstain from
paying the dollar amount and not experience the
health improvement, the individual feels either choice
provides the same degree of well-being. Alternatively,
willingness to accept compensation (WTA) is the
smallest amount of money a person would voluntarily
accept as compensation to forego an improvement, or
endure a decline, in health. The individual feels that to
accept the payment and not experience the health
improvement or refuse the compensation and
experience improved health will provide the same
degree of well-being. In practice, WTP is generally
used to value benefits because it is often easier to
measure and quantify.2 In this report, we refer to all
valuation estimates as WTP values, even though the
underlying economic valuation literature is based on
studies which elicited expressions of WTP and/or
WTA.3
In the context of cost-benefit analysis, WTP is
useful for estimating the monetary value of non-
market, public goods. A major characteristic of public
goods is that they are nonrival (i.e., one person's
consumption of the good does not reduce the amount
available to others). In the case of health-related
improvements due to environmental quality, the
benefits are also nonexclusive. Benefits are not (and
to some extent, cannot be) regulated. As a result, the
benefits are actually reductions in the probabilities or
risk of enduring certain health
2It is worth noting that the appropriateness of either WTP or
WTA also depends on property rights. In the case of a policy
aimed at reducing existing pollution levels, a WTP measure
implicitly assumes that the property rights rest with the polluting
firm. Alternatively, WTA measures implicitly assume that the
property rights rest with the public. (Carson and Mitchell, 1993.)
In some cases (e.g., hospital admissions), neither WTA nor
WTP estimates are available. In those cases, cost of illness (COI)
estimates are applied in lieu of WTP values. COI estimates
understate the true welfare change since important value
components (e.g., pain and suffering associated with the health
effect) are not reflected in the out-of-pocket costs for the hospital
stay.
problems. In theory, the total social value associated
with the decrease in risk is
N
2_i (number of units of risk reduction \ * (WTP per unit risk reduction ) (1)
where (number of units of risk reduction^ is the
number of units of risk reduction conferred on the ith
exposed individual as a result of the pollution
reduction, (WTP per unit risk reduction^ is the ith
individual's willingness to pay for a unit risk reduction,
and N is the number of exposed individuals. The
units are in terms of cases reduced per unit of time
(usually one year).
Using mortality risk as an example, suppose that
a given reduction in PM concentrations results in
lowering the risk of death by 1/10,000 per year. Then
for every 10,000 individuals, one less death would be
expected if ambient PM concentrations are reduced.
If an individual's WTP for this 1/10,000 decrease in
mortality risk is $500 (assuming, for now, that all
individuals' WTPs are the same), then the value of a
statistical life is 10,000 x $500, or $5 million.
While the estimation of WTP for a market good
(i.e., the estimation of a demand schedule) is not a
simple matter, the estimation of WTP for a
nonmarket good, such as a decrease in the risk of
having a particular health problem, is substantially
more difficult. Estimation of WTP for decreases in
very specific health risks (e.g., WTP to decrease the
risk of a day of coughing or WTP to decrease the risk
of admission to the hospital for a respiratory illness)
is further complicated by several factors, such as
wealth, income, age, pre-existing health impairments,
or other personal characteristics. There are many
policy contexts where distinguishing among WTP
estimates based on categorical differences (e.g.,
distinguishing between WTP of a low-income group
and a high-income group) is controversial. Given the
consideration of these influencing factors and the
limitations on information available for developing
WTP estimates, EPA sought to develop the most
appropriate and accurate estimates possible.
Derivations of the dollar value estimates for this study
are discussed below.
H-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Valuation of Specific Health
Endpoints
Since the Section 812 CAA retrospective analysis
(U.S. EPA 1997), there have been significant
advances made in economic valuation methodologies
for both mortality and morbidity effects. Much of the
literature presents emerging new approaches for
characterizing the effects of potentially important
determinants of WTP, such as age, income, risk
perception, and current health status. Despite this
progress, many of the more recent studies test
techniques that are in the development stage and use
data from work reviewed and incorporated in the
Section 812 retrospective analysis. This section
reviews the sources and methodology used to derive
WTP estimates for premature mortality and a variety
of morbidity effects valued in the present study. In
addition, there are brief discussions of more recent
advances relevant to particular endpoints.
Valuation of Premature Mortality
Avoided
The economic benefits associated with premature
mortality were the largest category of monetized
benefits in the Section 812 CAA retrospective analysis
(U.S. EPA 1997).4 In addition, EPA identified
valuation of mortality benefits as the largest
contributor to the range of uncertainty in monetized
benefits. Because of the uncertainty in estimates of
the value of premature mortality avoidance, it is
important to adequately characterize and understand
the various types of economic approaches available
for mortality valuation. Such an assessment also
requires an understanding of how alternative valuation
4As noted in the methods section, it is actually reductions in
mortality risk that are valued in a monetized benefit analysis.
Individual WTPs for small reductions in mortality risk are summed
over enough individuals to infer the value of a statisticalV£e saved.
This is different from the value of a particular, identified life saved.
The "value of a premature death avoided," then, should be
understood as shorthand for "the value of a statistical premature
death avoided."
approaches reflect that some individuals may be more
susceptible to air pollution-induced mortality.
The health science literature on air pollution
indicates that several human characteristics affect the
degree to which mortality risk affects an individual.
For example, some age groups are more susceptible to
air pollution than others (e.g., the elderly and
children). Health status prior to exposure also affects
susceptibility — at risk individuals include those who
have suffered strokes or are suffering from
cardiovascular disease and angina (Rowlatt, et al.
1998).
To reflect the full range of knowledge of air
pollution-induced mortality, an ideal estimate of
mortality risk reduction benefits would be an ex ante
willingness to pay (WTP) to improve one's own
chances of survival plus WTP to improve other
individuals' survival rates.5 The measure would take
into account the specific nature of the risk reduction
commodity that is provided to individuals, as well as
the context in which risk is reduced. To measure this
value, it is important to assess how reductions in air
pollution reduce the risk of dying from the time that
reductions take effect onward, and how individuals
value these changes. Each individual's survival curve,
or the probability of surviving beyond a given age,
should shift as a result of an environmental quality
improvement. That is, changing the current
probability of survival for an individual also shifts
future probabilities of that individual's survival. This
probability shift will differ across individuals because
survival curves are dependent on such characteristics
as age, health state, and the current age to which the
individual is likely to survive. For example, Figure H-
1 illustrates how a risk reduction may change a
survival curve for a given population. In this figure,
the solid line shows a survival curve for white males,
from California 1980 life tables (adapted from Selvin,
1996), up to age 80. The dashed line shows that the
probability of survival beyond a given age increases
with a reduction in mortality risk.
5 For a more detailed discussion of altruistic values related to
the value of life, see Jones-Lee (1992).
H-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
While the change in a survival curve represents a
cumulative effect of a change in risk over time, the
annual change in risk of death represents a static effect
of a risk reduction. As discussed in Appendix D in
greater detail, the instantaneous risk of death at a
specific age is often used to illustrate the effects of
changes in risk. The annual risk of death is related to
the probability of survival in that it represents the rate
at which the survival probability changes at any given
age, divided by the probability of surviving beyond
that age. Figure H-2 shows how a constant risk
reduction reduces annual risk of death across various
age cohorts. The baseline risk of death increases with
each cohort (solid line). As a result, the reduction in
risk (in this hypothetical example a constant 25
percent reduction) lowers each cohorts' risk level at a
different rate. The elderly experience a greater
reduction in risk than younger cohorts as can be seen
by the increasing difference between the solid and
dashed line. It is important to note that this example
shows the effect of a uniform risk reduction, and air
pollution controls may have risk reduction effects that
vary across age cohorts.
An alternative way to view the age-dependent
effect of risk reduction is to consider changes in the
cumulative effect of risk as measured by changes in
remaining life expectancy. Remaining life expectancy
is measured as the average number of additional years
expected to be lived by those individuals alive at a
given age, and derives from the area under the survival
curve at any given age. The age-dependent effects of
a hypothetical change in risk are portrayed in Figure
H-3. Consider the effect of risk reduction on two
cohorts, aged 10 years apart. When each cohort was
at age 40 both had the same life expectancy shown in
Figure H-3 as point A'. Given a risk reduction in the
future that occurs when one cohort is at age 60 and
the other at age 70, the life expectancy of the 60 year
old increases by the amount A'B', and the life
Figure H-1
Hypothetical Survival Curve Shift
H-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Figure H-2
Change in 1990 Annual Risk of Death by 25 Percent
Figure H-3
Increase in 1990 Remaining Life Expectancy
60
Age
H-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
expectancy of the 70 year old increases by the amount
A"B". The change in life expectancy is greater for
the younger cohort than the older cohort because
these measures represent a cumulative accrual of
increased life expectancy (i.e., the younger cohort will
benefit from the lower risk environment for more
years).
Because the risk reduction results in various
changes in risk levels, individual values for risk
reduction are likely to vary as well. Some individuals
having a greater change in risk, and hence life
expectancy, may have different values for the change
than those individuals experiencing a smaller change
in risk. Note that future generations may hold values
for health as well. Cropper and Sussman (1990)
develop theoretical models formalizing these concepts
when investigating how an individual's values for
reduction of a future risk to oneself and to future
generations should be discounted to the present.
While these theoretical models reflect the types of
values necessary to estimate the impact of the CAAA,
they are difficult to implement. First, they require an
estimate of individuals' survival curves. In order to
develop these survival probabilities, it is necessary to
characterize the dose/response relationship for the
regulated pollutants and know how this information
varies with age and health states over time. Second, it
is necessary to estimate values for risk reductions,
considering the key dimensions in which risk and
valuation of risk reduction may vary (e.g., with age and
health state).
Mortality Valuation Methodologies
This section summarizes alternative approaches to
mortality risk valuation, and outlines the approach
used to measure the economic value of these types of
benefits for air pollution reductions associated with
the CAAA. The first part provides background on the
methods that individuals have developed to estimate
the value of risk reduction benefits, including
commonly-applied approaches to valuation as well as
approaches that are beginning to be established in the
risk valuation literature. The second part discusses the
appropriateness of using these methodologies for
assessing the economic value of mortality benefits
associated with air pollution reduction. The Agency
has concluded that recent advances in the literature
show promise in incorporating several of the factors
that are likely to influence value, but problems with
the methodological approaches and lack of data
needed to reliably to appropriately estimate values
with the newer models leads us to adopt a value of
statistical life approach for the primary estimate of air
pollution-related mortality benefits.
Commonly Applied Approaches
The preferred approach researchers have taken to
estimate values for avoiding premature mortality is
based on individual WTP for risk reduction.
Although some cost-benefit analyses have based
values on avoided lost earnings (i.e., the human capital
approach), the WTP approach is preferred because it
more closely conforms to economic theory.6 The
common WTP measures of the value of life-saving
programs include the value of statistical life (VSL) and
the value of a statistical life year (VSLY). Newer
approaches to estimate values incorporate changes in
life expectancy, risk of dying, life-days per person, and
age-specific preferences. This section describes these
approaches and discusses issues that arise in their
application to estimate the value of mortality risk
reduction benefits.
The most commonly applied approaches for
mortality valuation are the value of statistical life and
value of statistical life year. Both of these approaches
In a recent article by Ireland and Gilbert (1998), the authors
evaluate value of life estimates used in tort recovery cases. The
article discusses the concept that for an individual there can be
finite utility (or determined value) to life and at the same time no
monetary equivalent. The authors do, however, build on this
argument to demonstrate that existing value of life estimates are in
fact lower bounds to the true value. By "lower bound," the
authors refer to a value representative of a specific individual, not
of a statistical life. In citing a reasonable value of life range, they
use a range similar to that of the 812 retrospective analysis,
although the authors do not cite the source of this range. Ireland
and Gilbert write, "A decedent has lost something of immense
value, for which estimates in the $4-$6 million range is clearly a
low market value estimate".
H-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
directly address the value of premature death and
health impairment. The VSL method measures the
value of a given reduction in risk and an individual's
WTP to reduce that risk, relying on wage and
occupational risk tradeoff data or the results of
contingent valuation surveys. Individual WTP
amounts for small reductions in mortality risk are
"standardized" to reflect reduction of population risk
of one statistical life saved. The result of applying this
method is not the value of an identifiable life, but
instead the value of reducing fatal risks in a population
(Viscusi 1992).
Viscusi (1992) summarizes the value of life
literature, including almost forty studies providing
VSL estimates relevant for policy application. For the
section 812 retrospective analysis, EPA identified 26
studies from that review that reflect the application of
the most sound and defensible methodological
elements (see Table H-l). Five of the 26 studies are
contingent valuation (CV) studies, which directly
solicit WTP information from subjects; the rest are
wage-risk studies, which base WTP on estimates of
the additional compensation demanded in the labor
market for riskier jobs. Using a Weibull distribution
to describe the distribution of the mean mortality risk
valuation estimates from these studies, the mean
estimate of the distribution is $4.8 million with a
standard deviation of $3.2 million (1990$).
Since EPA's retrospective analysis, Desvousges et
al. (1998) has conducted a meta-analysis of twenty-
nine mortality studies presented in Viscusi (1993) and
Fisher, Chestnut, and Violette (1989).7 Desvousges et
al.'s meta-analysis yields $3.3. million (1990 dollars) as
a value of statistical life, with a 90 percent confidence
interval between $0.4 and $6.3 million.8 Their
estimate, $3.3 million, falls well within the range
generated by EPA's uncertainty analysis of VSL
estimates. The selection of studies accounts for much
of the difference between their analysis and EPA's.
The Desvousges et al. analysis includes thirteen studies
that EPA did not use and EPA includes ten studies
omitted by Desvousges et al.
1 In addition to the Viscusi (1993) study, the 812 retrospective
examined two other studies, Miller et al. (1990) and the Fisher,
Chestnut, and Violette (1989). We opted to not use the Miller et
al. study given our concerns regarding the appropriateness of the
selection of studies for valuing reductions in environment-related
mortality risk and concerns about the adjustments made to the
underlying data. The Fisher, Chestnut, and Violette (1989) study
was not used because the data was not as current or
comprehensive as the data in the Viscusi study.
8 Desvousges et al. do not adjust the value of statistical life to
account for age differences. They do note that a single estimate
for the value of statistical life may not be a good representation of
the differences between willingness-to-pay of the elderly and
young, healthy workers. They state that Moore and Viscusi (1988)
demonstrate that willingness-to-pay is higher for people with more
life years to lose while Desvousges et al. (1996) and Johnson et al.
(1998) indicate that willingness-to-pay is lower for people with
limited abilities to engage in activities and care for themselves.
H-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table H-1
Summary of Mortality Valuation Estimates
_ . . Type of
Study Estimate
Kneisner and Leeth (1991) (US)
Smith and Gilbert (1984)
Dillingham (1985)
Butler (1983)
Miller and Guria (1991)
Moore and Viscusi (1988a)
Viscusi, Magat, and Huber(1991b)
Gegaxet al. (1985)
Marin and Psacharopoulos (1982)
Kneisner and Leeth (1991) (Australia)
Gerking, de Haan, and Schulze (1988)
Cousineau, Lacroix, and Girard (1988)
Jones-Lee (1989)
Dillingham (1985)
Viscusi (1978, 1979)
R.S. Smith (1976)
V.K. Smith (1976)
Olson (1981)
Viscusi (1981)
R.S. Smith (1974)
Moore and Viscusi (1988a)
Kneisner and Leeth (1991) (Japan)
Herzog and Schlottman (1987)
Leigh and Folson (1984)
Leigh (1987)
Garen (1988)
Labor Market
Labor Market
Labor Market
Labor Market
Cont. Value
Labor Market
Cont. Value
Cont. Value
Labor Market
Labor Market
Cont. Value
Labor Market
Cont. Value
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Valuation (millions 1990$)
0.6
0.7
0.9
1.1
1.2
2.5
2.7
3.3
2.8
3.3
3.4
3.6
3.8
3.9
4.1
4.6
4.7
5.2
6.5
7.2
7.3
7.6
9.1
9.7
10.4
13.5
SOURCE: Viscusi, 1992 and EPA analysis.
When applying VSL estimates to estimate
mortality benefits, it is important to determine the
differences between the nature of air pollution risk
and risks faced by persons whose risk-dollar tradeoff
decisions have been addressed in the literature. First,
several studies indicate that the value people place on
mortality risk reduction may depend on the nature of
the risk (e.g., Fisher et al. 1989; Beggs 1984). Current
VSL estimates do not account for a number of the
important factors that affect risk perception. For
example, premature mortality risks from air pollution
are experienced on an involuntary basis and are
generally uncompensated, while job-related risks are
assumed by individuals who presumably have some
choice as to occupation and are compensated for
taking a riskier job. Second, the demographics of the
population at risk from air pollution, particularly in
H-8
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
terms of age, income, and health state, may differ
from the demographics of individuals surveyed in the
literature. For a more detailed discussion of how
these factors can affect the economic valuation of
premature mortality, and specifically estimates derived
from the VSL approach, see the discussion, "Benefits
Transfer and VSL," presented in the section titled,
"Uncertainties in the Valuation Estimates."
The VSLY method values life-years that would be
lost if an individual were to die prematurely. Most
commonly, VSLY estimates are an annualized
equivalent of VSL estimates (Moore and Viscusi 1988,
French and Mauskopf 1992). A VSLY estimate may
imply a stream of constant values per year. The
annualized VSLY estimate depends on three factors:
the underlying VSL estimate; a discount rate; and the
number of remaining life years implied by the
underlying VSL estimate.
We develop an estimate of the value of a statistical
life-year lost (VSLY) based on an approach suggested
by Moore and Viscusi (1988). They assume that the
willingness to pay to save a statistical life is the value
of a single year of life times the expected number of
years of life remaining for an individual. They also
suggest that the typical respondent in a mortality risk
study may have a life expectancy of an additional 35
years. Using the 35-year life expectancy and VSL
estimate of $4.8 million, their approach yields an
estimate of $137,000 per life-year lost or saved. In the
prospective analysis, we also assume that an individual
discounts future additional years. This implies that the
value of each life-year lost must be greater than the
non-discounted value. Assuming a five percent
discount rate and adopting the above outlined
approach, the implied value of each life year lost used
in the prospective analysis is $293,000 (in 1990
dollars).
Critics note several disadvantages to using this
type of VSLY method, most notably that the value of
avoiding premature death depends on more than just
lifespan. With the VSLY approach, the benefit
attributed to avoiding a premature death depends
directly on how premature it is — resulting in smaller
values for older people, who have shorter life
expectancies, and larger values for younger people.
While this approach attempts to derive age-
adjusted values of expected life remaining using VSL
estimates, it does not address potential differences in
the value of a statistical life due to differences in the
average age of the affected population or the average
age at which an effect is experienced. Studies have
shown that simple progressive declines in value as
estimated with the VSLY method may be an
oversimplification; in many cases, values for health
peak several times throughout a lifetime (e.g., after
having children, after retirement). In addition, in
many cases, data restrictions limit researchers' ability
to estimate VSLY because it is difficult to obtain
estimates of age-specific risks and the number of life-
years lost.
Life Quality Adjustments
Another way to make adjustments to account for
heterogeneity in value of life estimates is an approach
that incorporates health status by applying a VSLY
estimate (generated from the VSL literature) to an
estimate of quality-adjusted life years (QALY). The
resulting value estimates measure improvements in
health based on individuals' attitudes toward
symptoms or different levels of pain or physical
impairment (Tolley et al. 1994). This approach utilizes
survey techniques to rate different health conditions
and adjust the number of life years lost to represent
lost quality-adjusted life years. As a result, this
approach aims to develop a value for a single QALY
that is the same regardless of individual characteristics.
In other words, the approach tries to standardize the
measure of mortality risk reduction that emerges from
a health effects analysis, making valuation more
straightforward.
The Life Quality Adjustment approach may
implicitly incorporate morbidity impacts to assess
values for various causes of death, and is often used in
health economics to assess the cost effectiveness of
medical spending programs, to value morbidity
avoidance, and to value mortality avoidance. Using a
QALY rating system, health quality ranges from 0 to
H-9
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
1, where 1 may represent full health, 0 death, and
some number in between (e.g., 0.8) an impaired
condition. If an individual lives with a health quality
index of 0.8, then the implied value of avoiding a year
with this condition and having full health in its place
would be 0.2 x VSLY. By the same token, the value
of gaining an additional life year in this condition is 80
percent of the value of gaining a year in full health
(i.e., 0.8 x VSLY) and represents an annual value for
mortality risk avoidance for a person with the
condition.
Tolley, et al. (1994) estimate values for a variety of
health conditions using numerous techniques,
including, in some cases, valuation of quality-adjusted
life years. For example, when estimating values for
acute and chronic symptoms using QALYs, the
authors calculate low, medium and high value
estimates based on a range of VSL estimates.
Specifically, the authors use the following three VSLY
estimates (1991$) for QALY valuation:
Low Estimate —
.000 VSLY: Derived
from Miller, Calhoun and Arthur (1990) -
VSL of $1.95 million, two percent discount
rate.
• Medium Estimate = $120.000 VSLY:
Derived from Miller, Calhoun and Arthur
(1990) - VSL of $1.95 million, six percent
discount rate.
• High Estimate = $175.000 VSLY: Derived
from Moore and Viscusi (1988) - VSL of
$6.0 million, 0 percent discount rate.
The authors multiply the VSLY estimate by the
estimate of QALYs to calculate a value for each
symptom. It is not clear from the analysis discussion
which symptom values represent the application of
this approach.
Cutler and Richardson (1998) apply a VSL
estimate to an estimate of QALYs to measure the
value of health improvements between 1970 and 1990
for ten health conditions. To do this, the authors use
an VSLY estimate of $100,000, derived as the
intermediate value of results reported in studies by
Viscusi (1993) and Tolley et al. (1994). In addition,
the authors estimate QALYs using information on
disease prevalence in the US from 1970 to 1990,
weighted by a factor that represents how quality of life
for a given condition has changed over time (e.g.,
more buildings have ramps and elevators for
individuals who have mobility problems, thus raising
quality of life over time).
Murray and Lopez (1996) modify the above
theoretical approach by deriving an estimate of
disability-adjusted life years (DALYs). DALY
estimates consider the years of life lost and years lived
with disability, adjusted for the severity of the
disability. The approach to estimate DALYs is similar
to that used to estimate QALYs in that both
incorporate judgments about the value of time spent
in different health states. However, DALY and
QALY estimation methods differ in that the methods
to estimate DALYs are elicited from preferences for
particular value choices using a specific standardized
set of value choices.
The Life Quality Adjustment approach scales
WTP values (VSL estimates) using a measure of life
years that reflect heterogeneity in quality of health
(QALYs). In many cases, the applied VSLY estimates
do not reflect consistent use of VSL estimates or
discount rates. In addition, in each of these valuation
analyses health economists have constructed a scale or
index that ranks health outcomes in terms of how
adverse individuals believe them to be. Often, the
extreme points on the scale are "perfect health" and
"immediate death," but some applications allow for
health outcomes that might be viewed as worse than
death. These ranking methods do not yield estimates
of WTP, and therefore are not linked to utility theory.
It is not clear that the ranking of health outcomes
obtained by these indices would match the ranking
obtained by knowing individuals' WTP for various
health effects. As discussed by Johansson (1995),
these scales or indices rely on much more restrictive
assumptions about the nature of individual
preferences than are normally made in WTP studies.
H-10
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Longevity
Several recent efforts estimate values for an
identifiable life by estimating the WTP for own life
extension. Johannesson and Johansson (1996,1997)
estimate the "WTP to increase one's life expectancy by
one additional year (i.e., extending men's life
expectancy from age 75 to 76, and women's from age
80 to 81, conditional on reaching age 75 or 80).
Johannesson, Johansson, and Lofgren (1997) estimate
the value of an immediate small reduction in mortality
risk (a "blip" or one year of fatal risk prevention).
While this methodology represents a utility-theory
based value, the value estimate for a single year of
longevity does not exactly correspond to what is
needed for an assessment of air pollution benefits.
Johannesson and Johansson (1996, 1997) estimate a
value for a single year of life extension near the end of
one's lifetime - values at this age are likely to be low
because of a low expectation of quality of life at this
advanced age. It is likely that mortality values will vary
within an individual's lifetime and with probability of
survival. In addition, mortality associated with
pollutant exposure will likely yield a longevity loss
greater than one year (e.g., mortality associated with
particulate matter yields an average longevity loss of
approximately 14 life years among those who are
afflicted). Moreover, because of the hypothetical
nature of the contingent valuation method, it is
unclear whether respondents accept the scenarios
presented and whether enough context was provided
to understand the risk and the budget implications of
the scenario and the response.
Cost Effectiveness
Garber and Phelps (1997) present a methodology
for valuing a discounted life year that is determined
by income and risk aversion in a life-cycle model. To
calculate the optimal cost effectiveness cut-off for
medical intervention, the authors assume values of a
utility function, health production function, income,
discount rate, and baseline mortality to derive a value
equivalent to W^TP for a discounted life year. In this
model, utility is a function of income (less medical
expenditures), and future income is a function of
survival and medical expenditures. As a result, the
authors use mortality rates to calculate expected
income. Changes in these mortality rates result in
changes in survival probabilities, and hence income.
The model estimates an individual's willingness to
trade income from one period to another; the
discounted change in income is equivalent to W/TP for
a change in risk.
Although this methodology is based on a life-
cycle model using survival probabilities, it is simplistic
in its assumptions and is based on assumed
preferences, rather than on revealed preferences or
those stated by an individual. In effect, the model
estimates values based largely on one empirical input:
individual income. For example, the VSL for a 40
year-old cannot exceed $250,000 because that amount
exceeds the discounted expected income. The largest
value of discounted life-year obtained by the authors
is approximately $37,000.
Valuation Strategy Chosen for this
Analysis
To estimate the economic value of mortality
benefits associated with air pollution reductions,
economic theorists prefer estimates that reflect ex ante
values of reducing the risk of mortality across the
population (i.e., for individuals having different health
states and other characteristics such as income level
and risk perception). This requires an estimate of an
individual WTP for a reduction in an involuntary risk
that will change individuals' survival probabilities for
a lifetime. Developing a valuation estimate based on
this theoretically ideal approach, however, is currently
subject to significant data and methodological
problems. Moreover, many of the valuation methods
that are frequently presented as an alternative to the
VSL approach rely on VSL estimates and calculate
values that depend on lifespan data, which may be
difficult to measure given the current health data
limitations. Consequently, EPA's current interpreta-
tion of the state-of-the-art in premature mortality
valuation leads to adoption of the VSL approach for
development of the primary benefit estimate.
H-11
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
As discussed above, several different approaches
for estimating a mortality-related value have been
developed. Each, however, has either methodological
inconsistencies with the preferred utility-based
approach, or does not provide a value estimate for a
commodity comparable to that provided by reduced
air pollution. We summarize the potential problems
of these alternatives below and in Table H-2:
• Life Quality Adjustment: This approach
relies on VSL estimates applied to survey
estimates of life-years (i.e., QALYs or
DALYs) for the economic valuation.
Currently, no generally accepted estimate or
range of estimates of VSLY have been
established, instead these values derive from
various VSL studies and reflect numerous
discount rates. In addition, the life years
estimates require data sets that can account
for the health states or utilities specific to a
wide variety of health effects associated with
air pollution. In many cases, these estimates
are not available or are based on health
professionals' perceptions of various health
outcomes, and not necessarily based in
economic utility theory.
• Longevity: The longevity valuation
approach of Johannesson and Johansson
(1996 and 1997) provides an estimate of the
value for an identifiable one-year life
extension. While the contingent valuation
approach used may be consistent with utility
theory, the commodity valued does not
represent the commodity gained through
improvement of ambient air quality.
• Cost Effectiveness: While the approach
taken by Garber and Phelps relies on survival
probabilities throughout an individual's
lifetime, the methodology is based on a utility
function that makes specific assumptions
about individual preferences to measure WTP
rather than eliciting value from either a
revealed or stated preference approach.
Moreover, this approach measures a WTP
that is constrained by income. Where
individual risks are small (perhaps one in ten
thousand) relative to certain loss of life,
individual WTP may also be small relative to
income.
H-12
-------
Exhibit H-2
Summary of Alternative Methods for Assessing the Value of Reduced Mortality Risk
Method
Description
Strengths
Weaknesses
References
Value of Statistical
Life (VSL) - hedonic
wage studies
Uses wage and risk data to
estimate WTP to avoid risk in the
workplace
- Revealed preference
-Well-established approach: more
than 60 primary studies
-Workplace risk context; working-age
subjects and voluntary risk
- VSL may imply ex post risk
Summaries by Viscusi
(1992) and others;
many primary studies
VSL - contingent
valuation studies
Uses survey responses to
estimate WTP to avoid risks
- Flexible approach; some studies
use environmental risk context
- Good data on WTP by
respondent
- Risk information not well-understood
by subjects; questions may be unfamiliar
- VSL may imply ex post risk
Summaries by Viscusi
(1992) and others
VSL - consumer
market studies
Value of Statistical
Life Year (VSLY)
Quality Adjusted Life
Year (QALY)
WTP for change in
survival curve
WTP for change in
longevity
Cost-Effectiveness
Note: WTP = willinaness to
Uses consumer expense and risk
data (e.g., smoke detectors) to
estimate WTP to avoid risks
Annual equivalent of VSL
estimates
Applies quality adjustment to life-
extension data, uses cost-
effectiveness data to value
Reflects WTP for change in risk,
potentially incorporates age-
specific nature of risk reduction
Uses stated preference approach
to generate WTP for longevity or
longer life expectancy
Develops a standard of
comparison to measure the
efficiency of various treatments in
achieving a given health outcome
oav
- Revealed preference
- Flexible approach
- Provides financially accurate
adjustment for age at death
-Widely used in public health
literature that assess different
private medical interventions
- Theoretically preferred approach
that most accurately reflects
nature of risk reductions from air
pollution control
- Life expectancy is familiar term to
most individuals
-Widely used in public health
contexts
- Major difficulties estimating both risk
and expense variables
- VSL may imply ex post risk
- Adjustment may not reflect how
individuals consider life-years; assumes
they have equal value for all remaining
life-years
- Lack of data on health state indices
and life quality adjustments that are
applicable to an air pollution context
- Almost no current literature
- Lack of available data due to the
severe methodological difficulties in
presenting complex risk data to subjects
and eliciting reliable values
- Life expectancy is a simplifed term that
does not incorporate age-specific risk
information
-Methodological and data problems in
attempting to adapt to air pollution
context
- Public health context may be for
private goods (i.e., treatment)
- Dollar values do not necessarily reflect
patient preferences
Summaries by Viscusi
(1992) and others
Viscusi and Moore
(1988); French and
Mauskopf(1992)
Tolley(1994); Cutler
and Richardson
(1998)
Cropper and Sussman
(1990)
Johannesson and
Johansson (1997);
Health Canada (1998)
Garber and Phelps
(1997)
H-13
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Valuation of Hospital Admissions
Avoided
The valuation of this benefits category reflects the
value of reduced incidences of hospital admissions
due to respiratory or cardiovascular conditions. We
measure avoided hospital admissions as opposed to
the number of avoided cases of respiratory or
cardiovascular conditions, because of the availability
of C-R relationships for the hospital admissions
endpoint. Hospital admissions reflect a class of health
effects linked to air pollution which are acute in nature
but more severe than the symptom-day measures
discussed below.
As described in Chapter 5, our approach to
estimating the number of incidences for this category
involves reliance on several concentration-response
(C-R) functions. Each concentration-response
function provides an alternative definition of either
respiratory effects or cardiovascular effects, and
defines alternative relationships between a single
health affect and different pollutants. For the
valuation of these incidences, the current literature
provides well-developed and detailed cost estimates of
hospitalization by health effect or illness. Using
illness-specific estimates of avoided medical costs and
avoided costs of lost work-time, developed by
Elixhauser (1993), we construct cost of illness (COT)
estimates that are specific to the suite of health effects
defined by each C-R function. For example, we use
twelve distinct C-R functions to quantify the expected
change in respiratory admissions.9 Consequently in
this analysis, we develop twelve separate COI
estimates, each reflecting the unique composition of
health effects considered in the individual studies.
Because each epidemiology study defines a health
effect by a group of ICD codes, we construct COI
estimates for each study by aggregating estimates that
are specific to an ICD code. These estimates use the
following information reported by Elixhauser (1993):
average hospital costs, average length of stay, and
baseline incidences.10 We use this ICD code
information to develop valuation estimates that have
two components, hospital charges and lost earnings
due to the hospital stay. Our estimate of lost earnings
due to time spent in the hospital is based on valuing
the average length of hospital stay at a daily rate of
$83. This daily rate is the median weekly wage divided
by five work days and is based on U.S. Department of
Commerce figures (1992). After developing values for
each relevant ICD code (i.e., hospital costs plus lost
earnings), we weight these values based on their
prevalence in the baseline. The final COI estimate,
specific to each study, is the sum of the weighted
value of ICD code-specific estimates.
We use a Monte Carlo approach to combines the
valuation and physical effects modeling to generate a
benefits estimate for hospital admissions. This
approach also allows us to account for the variability
in costs due to alternative definitions of respiratory
and cardiovascular conditions that result in a hospital
admission. The Monte Carlo process for integrating
the C-R function and its COI value involves first
randomly selecting an estimated change in incidences
from the suite of applicable C-R functions. For
example, we use five epidemiology studies for the
endpoint hospital admissions due to cardiovascular
effects, and develop COI estimates specific to each
study. The Monte Carlo modeling then selects the
COI estimate specifically developed for that C-R
function. These values are multiplied to generate a
single benefits estimate for reduced hospital
admissions. This process is repeated so that the value
from each iteration is collected to generate a
distribution that characterizes the range and
probability of possible benefits estimates. The
primary benefit estimates of avoided cardiovascular-
related hospital admissions reflect the central value of
this distribution.
The use of COI estimates suggests we are likely to
significantly underestimate the WTP to avoid hospital
For more detailed discussion of the various health effects
considered by each C-R function and methodology for estimating
the number of avoided hospital admissions, see Appendix D.
'"Potential illnesses associated with respiratory
cardiovascular admissions were identified by ICD-9 code.
and
H-14
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
admission. The valuation of any given health effect,
such as hospitalization, should reflect the value of
avoiding associated pain and suffering and lost leisure
time, in addition to medical costs and lost work time.
While the probability distributions in this analysis
characterize a range of potential costs associated with
hospitalization, they do not account for the omission
of factors from the COI estimates, such as pain and
suffering. Consequently, the valuations for these
endpoints most likely understate the true social values
for avoiding hospital admissions due to respiratory or
cardiovascular conditions.
Valuation of Chronic Bronchitis
Avoided
In this analysis, chronic bronchitis is one of the
two monetized morbidity endpoints whose effects
may be expected to last from the initial onset of the
illness throughout the rest of the individual's life.
WTP to avoid chronic bronchitis therefore
incorporates the present discounted value of a
potentially long stream of costs (e.g., medical
expenditures and lost earnings) and reduced health-
state utility.11
Two studies, Viscusi et al. (1991) and Krupnick
and Cropper (1992) provide estimates of WTP to
avoid a case of chronic bronchitis. While alternative
estimates exist, many are derived from these two
primary studies.12 The study by Viscusi et al. uses a
sample that is larger and more representative of the
general population, while the Krupnick and Cropper
study solicits values only from individuals who have a
relative with the disease. As a result, the valuation of
"The severity of cases of chronic bronchitis valued in some
studies approaches that of chronic obstructive pulmonary disease.
To maintain consistency with the existing literature, we do not
treat those cases separately in this analysis.
12For examples of alternative estimates see Desvousges et al.
(1998) and Tolley et al. (1994). Both studies present estimates of
avoiding one year of chronic bronchitis that are based on adjusting
values from either Viscusi et al. (1991) or Krupnick and Cropper
(1992).
chronic bronchitis is based on the distribution of
WTP responses from Viscusi et al. (1991).
Both the Viscusi et al. and the Krupnick and
Cropper studies estimate the WTP to avoid a severe
case of chronic bronchitis (CB). The incidence of
pollution-related chronic bronchitis, however, is based
on three studies which consider only new incidences
of the illness and the resulting severity is unknown.13
In response to the uncertainty regarding how the
severity of a new case may progress, the prospective
analysis adjusts Viscusi et a/.'s WTP estimates
downward. This adjustment reflects the decrease in
severity of a case of pollution-related CB relative to
the case in the Viscusi study and the elasticity of WTP
with respect to severity. The elasticity of WTP to
avoid CB is a marginal value and not unit elastic (i.e.,
not equal to one). Consequently, WTP adjustments
are made in one percent increments. At each step, the
WTP specific to a given CB severity level (sev), is
adjusted to derive the WTP to avoid a case with a one
percent lower level of severity by calculating (
0.99*jw).14 In this analysis, we derive an estimate of
WTP for a case of chronic bronchitis that represents
a 50 percent reduction in the severity described in the
Viscusi study. The iterative procedure continues until
the severity is half of the of the Viscusi value.
With the downward adjustment to Viscusi et a/.'s
WTP estimate, calculating the WTP to avoid a case of
13 The three studies are Abbey etal (1993), Abbey etal (1995)
and Schwartz (1993). For more discussion of estimating the
number of avoided cases of chronic bronchitis see Appendix D,
Human Health Effect of Criteria Pollutants. Incidences are
predicted separately for each year during the period 1990-2010.
It is important that only new cases of chronic bronchitis are
considered in this analysis because WTP estimates reflect lifetime
expenditures and lower utility associated with the illness. If the
total prevalence of chronic bronchitis, rather than the incidence of
only new chronic bronchitis were predicted each year, valuation
estimates reflecting lifetime losses could be repeatedly applied to
the same individual for many years, resulting in a severe
overestimation of the value of avoiding pollution-related chronic
bronchitis.
14 Note that the elasticity changes at each iteration because
the elasticity of WTP with respect to severity is a function of
severity.
H-15
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
pollution-related chronic bronchitis has three
components, each introducing some uncertainty. The
components are (1) WTP to avoid a case of severe
CB, (2) the severity level of an average pollution-
related case of CB relative to that of the severe case,
and (3) the elasticity of WTP with respect to severity.
Based on assumptions about the distributions of each
component's value, a distribution of WTP to avoid a
pollution-related case of CB is derived by Monte Carlo
methods. Each of the three underlying distributions
is described briefly below.
The distribution of WTP to avoid a severe case of
CB is based on the distribution of WTP responses in
the Viscusi study. Viscusi ef al. derived an implicit
WTP to avoid a statistical case of chronic bronchitis
from respondents' WTP for a specified reduction in
risk. The mean response implied a WTP of about $1
million (1990 dollars); the median response implied a
WTP of about $530,000 (1990 dollars).15 Viscusi et al.
report the mean and median of their distribution of
WTP responses and the decile points. The
distribution of reliable WTP responses from the
Viscusi study can therefore be approximated by a
discrete distribution, assigning equal probability to
each of the first nine decile points (or one-ninth
probability to each decile). This method omits five
percent of the responses from each end of the
distribution (i.e., the extreme tails which are
considered unreliable). Our present study uses this
trimmed distribution of Viscusi et a/.'s WTP
responses, for which the mean is $720,000 (1990
dollars), as the distribution of WTPs to avoid a severe
case of CB.
The distribution of the severity level of an average
case of pollution-related CB is based on the severity
levels used in Krupnick and Cropper's study, which
estimates the relationship between severity level and
the natural log of WTP. The distribution is triangular
with a mean of 6.5 and endpoints at 1.0 and 12,
although the most severe case of CB in that study is
assigned a severity level of 13.16
The elasticity of WTP to avoid a case of CB with
respect to the severity of the case equals a constant
times the severity level. This constant, estimated in
Krupnick and Cropper's study of the relationship
between severity and the natural log of WTP, is
normally distributed with mean of 0.18 and standard
deviation of 0.0669.
Using distributions of the three WTP
components described above, the Monte Carlo
analysis generates a distribution with a mean of
$260,000 for WTP to avoid a pollution-related case of
CB. Consistent with economic theory, the COI
estimates generated by Cropper and Krupnick (1990)
are lower than the mean WTP estimate (i.e., COI does
not reflects the desire to avoid pain and suffering).17
These COI estimates are approximately $86,000 for a
30 year old, $84,000 for a 40 year old, $76,000 for a 50
year old, and $43,000 for a 60 year old (in 1990
dollars). The prospective's WTP estimate is 3 to 6
times greater than the full COI estimate for 30 year
olds and 60 year olds, respectively.
Valuation of Chronic Asthma Avoided
Chronic asthma is the other morbidity endpoint
that is valued as a health condition lasting throughout
an individual's lifetime. The number of new cases of
chronic asthma is based on a study by McDonnell et
al. (1999), and specifically examines the effects of
ozone as a potential cause of the illness among adult
males (i.e., ages 27 and older). Similar to the valuation
of chronic bronchitis, WTP to avoid chronic asthma
15There is an indication in the Viscusi paper that the dollar
values in the paper are in 1987 dollars. Under this assumption, the
dollar values were converted to 1990 dollars.
16The Krupnick and Cropper study bases its most severe case
of CB (i.e., severity level equal to 13) on that used in the Viscusi
study.
17 Using a 5 percent discount rate and assuming that 1) lost
earnings continue until age 65, 2) medical expenditures are
incurred until death, and 3) life expectancy is unchanged by
chronic bronchitis, Cropper and Desvousges calculate several
estimates of the present value of the stream of medical
expenditures and lost earnings associated with an average case of
chronic bronchitis.
H-16
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
is presented as the net present value of what would
potentially be a stream of costs and lower well-being
incurred over a lifetime.
Estimates of WTP to avoid asthma are provided
in two studies, one by Blumenschein and Johannes son
(1998) and one by O'Conor and Blomquist (1997).
Both studies use the contingent valuation method to
solicit annual WTP estimates from individuals who
have been diagnosed as asthmatics. Each study,
however, applies a different valuation approach.
Blumenschein and Johannesson solicit WTP values by
asking dichotomous choice and open-ended bidding
game questions. They report an average monthly
WTP of $162 which amounts to an annual value of
approximately $1,900 (1990 dollars). Alternatively,
O'Conor and Blomquist apply a risk-risk tradeoff
approach similar to that used in the chronic bronchitis
studies. They calculate $1,200 (1990 dollars) as the
average annual WTP to avoid asthma.
To maintain consistency between the health
effects modeling and the valuation, the WTP estimates
were adjusted to account for two factors. As
mentioned earlier, valuation of chronic morbidity
endpoints should approximate the costs and lowered
health-state utility that are incurred over an
individual's lifetime. We assume that the health
condition does not affect the average life expectancy
of an individual (i.e., does not cause premature
mortality). Recognizing that the average life
expectancy will vary with different age groups and that
each age group does not represent an equal portion of
the population, the present discounted stream of WTP
is calculated for seven different age cohorts (between
the ages 27 and 85). In turn, the net present value for
each age group is weighted by that age category's
representative share of the total population. This
calculation was performed for the mean WTP
estimates presented in the two studies. The central
estimate of WTP to avoid a case of chronic asthma
among adult males, approximately $25,000, is the
average of the present discounted value from the two
studies. The analysis characterizes the uncertainty
around this estimate by applying upper and lower
values based on the present discounted value derived
from each study, $19,000 derived from O'Conor and
Blomquist study and $29,000 from the Blumenschein
and Johannesson study.
Valuation of Other Morbidity Endpoints
Avoided
The valuation of a specific short-term morbidity
endpoint is generally solicited by representing the
illness as a cluster of acute symptoms. For each
symptom, the WTP is calculated. These values, in
turn, are aggregated to arrive at the WTP to avoid a
specific short term condition. For example, the
endpoint lower respiratory symptoms (LRS) is
represented by two or more of the following
symptoms: runny or stuffy nose; coughing; and eye
irritation. The WTP to avoid one day of LRS is the
sum of values associated with these symptoms. The
primary advantage of this approach is that is provides
some flexibility in constructing estimates to represent
a variety of health effects.
At the time of the Section 812 retrospective
analysis there were only a small number of available
studies on which to base estimates (two or three
studies, for some endpoints; only one study for
others). Since the retrospective analysis, much of the
literature suggests there are developing approaches
that may eventually lead to the refinement of estimates
and the overcoming of some limitations to the current
approach to constructing values. For example there
is extensive progress in developing valuation
techniques that reflect an individual's current health
state and more accurately account for a symptoms's
attributes (i.e., duration and severity).
There are several aspects of the short-term
morbidity valuation estimates worth noting. First,
estimates of WTP may be understated for at least two
reasons. If exposure to pollution has any cumulative
or lagged effects, then a given reduction in pollution
concentrations in one year may confer benefits not
only in that year but in future years as well. Benefits
achieved in later years are not included. In addition,
the possible effects of altruism are not considered in
any of the economic value derivations. Individuals'
WTP for reductions in health risks for others are
implicitly assumed to be zero. The second point
H-17
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
worth noting is that the total benefit attributed to the
reduction of particular pollutant's concentration is
determined largely by the benefit associated with its
corresponding reduction in mortality risk. This is
largely due to the dollar value associated with mortality
which is significantly greater than any other valuation
estimate. More detailed explanations for valuation of
specific morbidity endpoints are given in Table H-3.
The table summarizes the sources and derivation of
the economic values used in the analysis.
Valuation of Welfare Effects
Economic valuations for welfare effects quantified
in the analysis (i.e., visibility and worker productivity)
are documented in Table H-3.18 Worker productivity,
unlike the avoidance of work loss days or restricted
activity days, reflects productivity benefits due to
improvements in work conditions (i.e., reduced
ambient ozone) rather than health improvements (i.e.,
reduced risk of hospitalization). It is measured in
terms of the reduction in daily income of the average
worker engaged in strenuous outdoor labor and
estimated at $1 per ten percent increase in ozone
concentration. (Crocker and Horst, 1981). We discuss
the derivation of the visibility valuation further below.
18 In valuing welfare effects, the retrospective analysis
included the benefits of reduced household soiling. This valuation
was based on 1972 data that projected expenditure patterns from
1972 to 1985 (Manuel et al., 1982). While this study was
appropriate for the twenty year time period of the retrospective
(1970 to 1990), it is of questionable applicability for the current
study. Since the original study, there have been alternative
estimates of benefits due to reduced soiling. These estimates,
however, continue to be based on the original study and its
underlying data (e.g., Desvousges et al., 1998). Consequently, these
valuation coefficients do not reflect more recent information on air
pollution composition and potentially significant changes in
patterns of household expenditure and allocation. Progress in the
valuation of this category's benefits is further limited by the
challenges of developing dose-response functions that accurately
assess the level and rate of materials damage and soiling. Recent
literature does suggest there is progress in refining approaches,
although it has not quite advanced to the level necessary for
credible quantification or monetization of benefits associated with
reduced materials damage and soiling.
Visibility Valuation
Since the late 1970s, a number of contingent
valuation (CV) studies of visibility changes have been
published in the economics literature. These studies
often classify visibility benefits as either residential or
recreational. CV studies of residential visibility
generally survey individuals in urban and suburban
settings. The valuation is also applicable to
households in rural areas. Residential values relate to
the impact of visibility changes on an individual's daily
life (e.g., at home, at work, and while engaged in
routine recreational activities). Benefits of recreational
visibility relate to the impact of visibility changes
manifested at parks and wilderness areas that are
expected to be experienced by its visitors.
Recreational visibility benefits may, however, reflect
the value an individual places on visibility
improvements regardless of whether or not the
person plans to visit the park.19
The reported estimates, expressed as household
willingness to pay (WTP) for a hypothesized
improvement in visibility, have a wide range of values.
For examples, studies of visibility values from western
cities have reported somewhat lower values than those
from eastern cities. This difference raises the question
of how visibility benefits should be evaluated with
respect to location (e.g., eastern U.S. versus western
U.S.), commodity definition (e.g., changes in
recreational areas versus residential areas), and units of
measurement (e.g., visual range, light extinction, and
deciview). While the differing values reported in the
literature may appear to imply that visibility is valued
differently in the eastern and western U.S., other
evidence suggests that eastern and western visibility
are not fundamentally different commodities. For
example, NAPAP data indicates that California's
South Coast Air Basin, which encompasses Los
Angeles and extends northward to the vicinity of San
Francisco, has median baseline visibility more
characteristic of the eastern U.S. than of other areas of
the west (NAPAP 1991; lEc 1992, 1993a). These
19This type of valuation is typically labeled "existence value."
For more discussion see Chestnut and Rowe, 1990.
H-18
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
results suggest that the valuation of marginal visibility
changes is dependent on baseline conditions and
proximity to the commodity being valued (e.g.,
improved visibility in a region with an abundance of
National Parks such as the Pacific Northwest).
Returning to the NAPAP example, the similarity in
values may reflect the similarities between baseline
visibility in eastern and western coastal zones (i.e.,
coastal areas typically have higher humidity, while
areas of the west tend to have lower humidity and
hence a greater baseline visibility).
For the purposes of this report, we interpret
recreational settings applicable for this category of
effects to include National Parks throughout the
nation. Other recreational settings may also be
applicable, for example National Forests, state parks,
or even hiking trails or roadside areas with scenic
vistas. In those cases, a lack of suitable economic
valuation literature to identify these other areas and/or
a lack of visitation data prevents us from generating
estimates for those recreational vista areas. Moreover,
we develop estimates of recreational visibility changes
that account for the tendency of individuals to value
visibility changes based on proximity to the National
Park.
We estimate visibility benefits based on a derived
visibility valuation function. In both cases, residential
and recreational visibility, the valuation function takes
the following form:
HHWTP = B* ln(VR1/VR2)
where:
HHWTP = annual WTP per household for
visibility changes
VR1 = the starting annual average visual
range
VR2 = the annual average visual range after
the change in air quality
B = the estimated visibility coefficient.
The form of this valuation function is designed to
reflect the way individuals perceive and express value
for changes in visibility. In other words, the expressed
WTP for visibility changes varies with the percentage
change in visual range, a measure that is closely related
to, though not exactly analogous to, the Deci View
index used in Chapter 4.
We develop estimates of the visibility coefficients
for residential and recreational visibility from two
studies.20 We use figures reported in Chestnut and
Dennis (1997) for the valuation of residential visibility.
This study publishes estimates of visibility benefits
for the Eastern U.S that are based on original research
conducted in two Eastern cities (Atlanta and Chicago)
by McClelland et al. (1990). We use a central B
coefficient for residential visibility of $141, as reported
in Chestnut and Dennis (1997). For the valuation of
recreational visibility benefits, we use a study by
Chestnut and Rowe (1990). This study reports WTP
estimates of recreational visibility in three park
regions, the Western, Southwestern, and Eastern U.S.
For recreational visibility, the coefficients vary based
on the study region and whether the household is
within or outside of the National Park region of
concern. "In-region" coefficients are higher than
those for "out-of-region" households. The "in-
region" estimates for California, the Southwest, and
Southeast are $105, $137, and $65, respectively; the
corresponding "out-of-region" estimates are $73,
$110, and $40, respectively.
Our valuation of visibility changes is largely based
on unpublished, but peer-reviewed work. For
example, we use the secondary analysis of Chestnut
and Dennis (1997) to value residential visibility
benefits. This article is published in the Journal of Air
and Waste Management Association, but relies on the
unpublished results reported by McClelland et al.
(1990). The source of our recreational visibility
estimates, Chestnut and Rowe (1990), is also
unpublished. Both studies were originally developed
as part of the National Acid Precipitation Assessment
Program (NAPAP) and, therefore, have been subject
to peer-review as part of that program. Moreover,
these two studies are frequently cited and
20The unit of measure for the visibility coefficients is dollars.
However, these coefficients are scaled by the small incremental
changes in visibility to generate our WTP estimates.
H-19
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
recommended for use in published analyses of
visibility valuation.21
Concerns about the method used in the
McClelland et al. study, however, suggest their results
may not incorporate two potentially important
adjustments. First, their study does not account for
the "warm glow" effect, in which respondents may
provide higher willingness to pay estimates simply
because they favor "good causes" such as
environmental improvement. Second, while the study
accounts for non-response bias, it may not employ the
best available methods. The effect of both these
factors is to suggest an overestimate of WTP. As a
result, we exclude residential visibility estimates from
the overall primary benefits estimate.
!1For example see Desvousges et al. (1998).
H-20
-------
Table H-3
Unit Values Used for Economic Valuation of Health and Welfare Endpoints
Health or Welfare
Endpoint
Estimated Value Per Incidence (1990$)
Central Estimate Uncertainty Distribution
Derivation of Estimates
Mortality
$4.8 million per
statistical life
Weibull distribution,
mean = $4.8 million
std. dev. = 3,240,000
Central Estimate: Value is the mean of value-of-statistical-life estimates
from 26 studies (5 contingent valuation and 21 labor market studies).
$293,000 per
statistical life-year
Weibull distribution,
mean = $293,000
std. dev. = 198,000
Uncertainty: Best-fit distribution to the 26 sample means. The Weibull
distribution prevents selection of negative WTP values.
Central Estimate: Value is the mean of the distribution of the value of a
statistical life-year, derived from the distribution of the value of a
statistical life (see below).
Uncertainty: Assuming the discount rate is five percent, and assuming
an expected 35 years remaining to the average worker in the wage-risk
studies (see above), the value of a statistical life-year is just a constant,
0.061, multiplied by the value of a statistical life. The distribution of the
value of a life-year is derived from the distribution of the value of a
statistical life. Because the VSL is expressed as a Weibull distribution,
as indicated above, the value of a statistical life-year is also expressed
as a Weibull distribution, with mean equal to 0.061 multiplied by the
mean of the original Weibull distribution (0.061 x $4.8 million =
$293,000) and standard deviation equal to 0.061 multiplied by the
standard deviation of the original distribution (0.061 x $3.24 =
$198,000).
H-21
-------
Health or Welfare
Endpoint
Estimated Value Per Incidence (1990$)
Central Estimate Uncertainty Distribution
Derivation of Estimates
Chronic Bronchitis (CB)
$260,000
A Monte Carlo-generated
distribution, based on three
underlying distributions, as
described more fully under
"Derivation of Estimates"
and in the text.
Central Estimate: Value is the mean of a Monte Carlo distribution of
WTP to avoid a case of pollution-related CB. WTP to avoid a case of
pollution-related CB is derived by adjusting WTP (as described in
Viscusi et al., 1991) to avoid a severe case of CB for the difference in
severity and taking into account the elasticity of WTP with respect to
severity of CB. The mean of the resulting distribution is $260,000.
Uncertainty: The distribution of WTP to avoid a case of pollution-related
CB was generated by Monte Carlo methods, drawing from each of three
distributions: (1) WTP to avoid a severe case of CB is assigned a 1/9
probability of being each of the first nine deciles of the distribution of
WTP responses in Viscusi et al., 1991; (2) the severity of a pollution-
related case of CB (relative to the case described in the Viscusi study)
is assumed to have a triangular distribution, centered at severity level
6.5 with endpoints at 1.0 and 12.0 (see text for further explanation); and
(3) the constant in the elasticity of WTP with respect to severity is
normally distributed with mean = 0.18 and standard deviation = 0.0669
(from Krupnick and Cropper, 1992). See text for further explanation.
Chronic Asthma
$25,000 Triangular distribution,
centered at $25,000 on the
interval [$19,000, $30,000]
Central Estimate: Based on results reported in two studies
(Blumenschein and Johannesson, 1998 and O'Conor and Blumquist,
1997). Assumes a 5% discount rate and reflects adjustments for age
distribution among adults (ages 27 and older) and projected life years
remaining.
Uncertainty: Reflects the range in central estimate values reported in
the two studies.
H-22
-------
Health or Welfare
Endpoint
Estimated Value Per Incidence (1990$)
Central Estimate Uncertainty Distribution
Derivation of Estimates
Hospital Admissions
1. All Respiratory
-ICD codes: 460-519
variable—
function of the
analysis
See Derivation of Estimates
Central Estimate: Central estimate is the result of the analysis. The
analysis uses 12 distinct C-R functions. A COI estimate is constructed
for each. The COI estimates are based on ICD-9 code level information
(e.g., average hospital care costs, average length of hospital stay, and
weighted share of total respiratory illnesses) reported in Elixhauser
(1993).
Uncertainty: Probability distribution is a result of the analysis and
reflects: (1) uncertainty range of C-R function outcome; and (2) variation
in study-specific COI estimates.
2. All Cardiovascular
- ICD codes: 390-429
variable—
function of the
analysis
See Derivation of Estimates
Central Estimate: Central estimate is the result of the analysis. The
analysis uses five distinct C-R functions. A COI estimate is constructed
for each. The COI estimates are based on ICD-9 code level information
(e.g., average hospital care costs, average length of hospital stay, and
weighted share of total respiratory illnesses) reported in Elixhauser
(1993).
Uncertainty: Probability distribution is a result of the analysis and
reflects: (1) uncertainty range of C-R function outcome; and (2) variation
in study-specific COI estimates.
3. Emergency room visits for $194
asthma
Triangular distribution,
centered at $194 on the
interval [$144, $269]
Central Estimate: COI estimate based on data reported by Smith et al.
(1997).
Uncertainty: Based on reported 95% confidence intervals for annual
estimates of the number and costs of ER visits.
H-23
-------
Health or Welfare
Endpoint
Estimated Value Per Incidence (1990$)
Central Estimate Uncertainty Distribution
Derivation of Estimates
Respiratory Ailments Not Requiring Hospitalization
1. Upper Resp. Symptoms
(URS)
(defined as one or more
of the
following:
runny or
stuffy nose,
wet cough,
burning,
aching, or
red eyes)
$19 Continuous uniform
distribution over the interval
[$7, $33]
Central Estimate: Combinations of the 3 symptoms for which WTP
estimates are available that closely match those listed by Pope et al.
result in 7 different "symptom clusters," each describing a "type" of
URS. A dollar value was derived for each type of URS, using lEc mid-
range estimates of WTP to avoid each symptom in the cluster and
assuming additivity of WTPs. The dollar value for URS is the average
of the dollar values for the 7 different types of URS.
Uncertainty: Assumed to be a continuous uniform distribution across
the range of values described by the 7 URS types.
2. Lower Resp. Symptoms
(LRS)
(defined in the study as
two or more of the
following: cough, chest
pain, phlegm, and
wheeze.)
$12 Continuous uniform
distribution over the interval
[$5, $19]
Central Estimate: Combinations of the 4 symptoms for which WTP
estimates are available that closely match those listed by Schwartz et
al. result in 11 different "symptom clusters," each describing a "type" of
LRS. A $ value was derived for each type of LRS, using lEc mid-range
estimates of WTP to avoid each symptom in the cluster and assuming
additivity of WTPs. The $ value for LRS is the average of the $ values
for the 11 different types of LRS.
Uncertainty: Taken to be a continuous uniform distribution across the
range of values described by the 11 LRS types.
3. Acute Bronchitis
$45 Continuous uniform
distribution over the interval
[$13, $77]
Central Estimate: Average of low and high values recommended by
IEC for use in section 812 analysis (Neumann et al., 1994).
Uncertainty: Continuous distribution between low and high values
(Neumann et al., 1994) assigns equal likelihood of occurrence of any
value within the range.
H-24
-------
Health or Welfare
Endpoint
Estimated Value Per Incidence (1990$)
Central Estimate Uncertainty Distribution
Derivation of Estimates
4. Acute Respiratory
Symptoms and Illnesses
- Presence of any of 19
acute respiratory
symptoms
-Any Resp. Symptom
- Respiratory Illness
$18 1. URS, probability = 40%
LRS, probability = 40%
URS+LRS, prob. = 20%
2. If URS, use URS $ dist.
If LRS, use LRS $ dist.
If URS+LRS, randomly
select one value each from
URS and LRS $
distributions; sum the two
Central Estimate: Assuming that respiratory illness and symptoms can
be characterized as some combination of URS and LRS, namely: URS
with 40% probability, LRS with 40% probability, and both URS and LRS
with 20% probability. The $ value for these endpoints is the weighted
average (using the weights 0.40, 0.40, and 0.20) of the $ values derived
for URS, LRS, and URS + LRS.
Uncertainty: Based on variability assumed for central estimate, and
URS and LRS uncertainty distributions presented previously.
5. Asthma Attack
$32 Continuous uniform
distribution over the interval
[$12, $54]
Central Estimate: Mean of average WTP estimates for the four severity
definitions of a "bad asthma day." Source: Rowe and Chestnut (1986), a
study which surveyed asthmatics to estimate WTP for avoidance of a
"bad asthma day," as defined by the subjects.
Uncertainty: Based on the range of values estimated for each of the
four severity definitions.
6. Moderate or worse
asthma
$32 Continuous uniform
distribution over the interval
[12, 54]
Central Estimate: Reflects the mean WTP to avoid a "bad asthma day"
as reported by Rowe and Chestnut (1986).
Uncertainty: Taken to be a continuous uniform distribution across the
range of values obtained from the study.
7. Shortness of breath,
chest tightness or
wheeze
$5.30 Continuous uniform
distribution over the interval
[$0, $10.60]
Central Estimate: From Ostro et al., 1995. This is the mean of the
median estimates from two studies of WTP to avoid a day of shortness
of breath: Dickie et al., 1991 ($0.00), and Loehman et al., 1979
($10.60).
Uncertainty: Taken to be a continuous uniform distribution across the
range of values obtained from the two studies.
H-25
-------
Health or Welfare
Endpoint
Estimated Value Per Incidence (1990$)
Central Estimate Uncertainty Distribution
Derivation of Estimates
Restricted Activity and Work Loss Days
1. WLDs
$83
none available
Central Estimate: Median weekly wage for 1990 divided by 5 (U.S.
Department of Commerce, 1992)
Uncertainty: Insufficient information to derive an uncertainty estimate.
2. MRADs
$38 Triangular distribution
centered at $38 on the
interval [$16, $61]
Central Estimate: Median WTP estimate to avoid 1 MRRAD — minor
respiratory restricted activity day - from Tolley et al. (1986)
(recommended by lEc as the mid-range estimate).
Uncertainty: Range is based on assumption that value should exceed
WTP for a single mild symptom (the highest estimate for a single
symptom—for eye irritation—is $16.00) and be less than that for a WLD.
The triangular distribution acknowledges that the actual value is likely to
be closer to the point estimate than either extreme.
Welfare Effects
1. Visibility
Residential Visibility
"in-region"
"out-of-region"
Valuation function:
HHWTP= B * ln(VR1/VR2)
where:
HHWTP = annual WTP per household
B = estimated visibility coefficient
VR1 = starting annual average visual range
VR2 = the annual average visual range after the
change in air quality
Central Estimate: Estimated WTP for valuation of visibility changes
depend upon two factors: (i) visibility coefficient, B, and (ii) incremental
change in visual range. Visibility coefficients applied in the primary
analysis vary by category of visibility change and region.
Recreational visibility valuation is based on Chestnut and Rowe (1990).
For "in region" recreational visibility, the coefficients are $105, $137,
$65, for California, the Southwest, and the Southeast, respectively. For
"out-of-region" recreational visibility, the coefficients are $73, $110, $40,
for California, the Southwest, and the Southeast, respectively.
2. Worker Productivity Change in daily none available
wages: $1 per
worker per 10%
change in O3
Central Estimate: Based on elasticity of income with respect to O3
concentration derived from study of California citrus workers (Crocker
and Horst, 1981 and U.S. EPA, 1994). Elasticity applied to the average
daily income for workers engaged in strenuous outdoor labor, $73 (U.S.
1990 Census).
Note: All WTP estimates converted to 1990 dollars using the Consumer Price Index (CPI); COI estimates converted using the CPI-Medical.
H-26
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Results of Valuation of Health
and Welfare Effects
We estimate total human health and welfare
benefits by combining the economic valuations
described in this Appendix with the health and welfare
effects results presented in Appendix D for projection
years 2000 and 2010. The valuation results reflect the
annual estimates of benefits for the 48 contiguous
States, or "all U.S. population," which provides a more
accurate depiction of the trend of economic benefits
over the 20-year study period 22 For our Primary
Central estimate we attribute to Titles I through V of
the CAAA total annual human health benefits of $68
billion in 2000 and $118 billion in 2010.
As noted in Appendix D, we also include
alternative estimates for some health and welfare
impacts, which form the basis of several alternative
benefit estimates. For each of the health effects
estimates, we quantify statistical uncertainty. The
range of estimated health and welfare effects, along
with the uncertain economic unit valuations, are
combined to estimate a range of possible results. We
use the Monte Carlo method presented in Chapter 8
to combine the health and economic information.
Both tables show the mean estimate results, as well as
the measured credible range (upper and lower five
percentiles of the results distribution), of economic
benefits for each of the quantified health and welfare
categories. We summarize our primary estimates of
2000 and 2010 monetized benefits in Table H-3 and
22In Appendix D, we present physical effect estimates for
affected population in the contiguous 48 States and for affected
populations within 50 kilometers of a monitor. We present those
results as a sensitivity test that characterizes the possible magnitude
of human health effects. For the purpose of assessing the total
benefit of the CAAA, the results affecting populations in 48 states
provide a better characterization of the total direct benefits than do
the "monitored area only" results. The results of only monitored
areas does not account for the benefits of air quality improvements
affecting approximately 25 percent of the population. The "all
U.S. population" results, however, rely on uncertain extrapolations
of pollution concentrations, and subsequent exposures, from
distant monitoring sites to provide coverage for the 25 percent or
so of the population living far from air quality monitors.
Table H-5, respectively. The tables provide our
Primary Central estimate, in addition to our Primary
Low estimate, 5th percentile values, and our Primary
High estimate, 95th percentile estimates, for each
benefit category.
We also apply the Monte Carlo method when
generating aggregate monetized benefit results. The
Monte Carlo method used in the analysis assumes that
each health and welfare endpoint is independent of
the others. We adopt this approach in response to the
very low probability that the aggregate benefits will
equal the sum of the fifth percentile benefits from
each of the ten endpoints. Consequently, the upper
and lower fifth percentiles of the estimated benefits
from the individual endpoints does not equal the
estimated totals for the Primary High and Primary
Low estimates.
There are two additional aspects of our results
that warrant discussion. The first is the valuation of
premature mortality due to PM exposure. The second
is our strategy to avoid double-counting when
aggregating health benefits. As discussed in Chapter
5, premature mortality is estimated based on PM
exposure. Our primary estimates reflect a lag between
PM exposure and the timing of premature mortality.
While this lag does not alter the number of estimated
incidences, it does alter the monetization of benefits.
Because we value the "event" rather than the present
change in risk, the value of avoided future premature
mortality should be discounted. Therefore, the type
of lag structure employed plays a direct role in the
valuation of this endpoint.
The primary analysis reflects a five-year lag
structure. Under this scenario, 50 percent of the
estimated cases of avoided mortality occur within the
first two years. The remaining 50 percent are then
distributed across the next three years. Our valuation
of avoided premature mortality applies a five percent
discount rate to the lagged estimates over the periods
2000 to 2005 and 2010 to 2015. We discount over the
period between the initial PM exposure change (either
2000 or 2010) and timing of the projected incidences.
H-27
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Many of the monetized health benefit categories
include overlapping health endpoints, creating the
potential for double-counting. In an effort to avoid
overstating the benefits, we do not aggregate all of the
quantified health effects. For example, "asthma
attacks" and "moderate to worse asthma", are all
considered components of the endpoint, "Any of 19
Respiratory Symptoms". Consequently, we present
the results but do not include them in our reported
total benefits figures. In other cases, there are
endpoints included in our aggregation of benefits that
appear to have overlapping health effects. For those
benefit categories that describe similar health effects,
it is important to keep in mind that estimated
incidences are based on unique portions of the
population.
H-28
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table H-4
Primary Estimates of Health and Welfare Benefits Due to Criteria Pollutants in 2000
Monetary Benefits (in millions 1990$)
Benefits Category
5th %ile
Mean
95th %ile
Mortality
Ages 30+
Chronic Illness
Chronic Bronchitis
Chronic Asthma
Hospitalization
All Respiratory
Total Cardiovascular
Asthma-Related ER Visits
Minor Illness
Acute Bronchitis
URS
LRS
Respiratory Illness
Mod/Worse Asthma1
Asthma Attacks1
Chest tightness, Shortness of
Breath, or Wheeze
Shortness of Breath
Work Loss Days
MRAD/Any-of-19
$ 8,600
$220
29
$46
53
0.1
$0
2.8
1.4
0.4
1.2
13
0
0
180
420
$63,000
$3,600
140
$78
200
0.6
$ 1.3
12
3.9
2.5
8.5
35
0.5
0.3
210
760
$150,000
$11,000
240
$ 120
430
1.8
$3.3
26
7.2
6.1
19
66
2.4
0.7
240
1,100
Welfare
Decreased Worker Productivity $ 460 $ 460 $ 460
Visibility
Recreational 1,700 2,000 2,300
Agriculture 46 450 860
Total Benefits2 $71,000
Note:
1 Moderate to worse asthma and asthma attacks are endpoints included in the definition of MRAD/Any of 19 respiratory effects.
Although valuation estimates are presented for these categories, the values are not included in total benefits to avoid the potential
for double-counting.
2 Summing 5th and 95th percentile values would yield a misleading estimate of the 5th and 95th percentile estimate of total health
benefits. For example, the likelihood that the 5th percentile estimates for each endpoint would simultaneously be drawn from a
Monte Carlo procedure is much less than 5 percent. As a result, we present only the total mean.
H-29
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Table H-5
Primary Estimates of Health and Welfare Benefits Due to Criteria Pollutants in 2010
Monetary Benefits (in millions 1990$)
Benefits Category 5th %ile Mean 95th %ile
Mortality
Ages 30+
Chronic Illness
Chronic Bronchitis
Chronic Asthma
Hospitalization
All Respiratory
Total Cardiovascular
Asthma-Related ER Visits
Minor Illness
Acute Bronchitis
URS
LRS
Respiratory Illness
Mod/Worse Asthma1
Asthma Attacks1
Chest tightness, Shortness of
Breath, or Wheeze
Shortness of Breath
Work Loss Days
MRAD/Any-of-19
$ 14,000
$360
40
$76
93
0.1
$0
4.2
2.2
0.9
1.9
20
0
0
300
680
$ 100,000
$5,600
180
$ 130
390
1
$2.1
19
6.2
6.3
13
55
0.6
0.5
340
1,200
$250,000
$18,000
300
$200
960
2.8
$5.2
39
12
15
29
100
3.1
1.2
380
1,800
Welfare
Decreased Worker Productivity $710 $710 $710
Visibility
Recreational 2,500 2,900 3,300
Agriculture 7_1 550 1,100
Total Benefits2 $110,000
Note:
1 Moderate to worse asthma, asthma attacks, and shortness of breath are endpoints included in the definition of MRAD/Any of 19
respiratory effects. Although valuation estimates are presented for these categories, the values are not included in total
benefits to avoid the potential for double-counting.
2 Summing 5th and 95th percentile values would yield a misleading estimate of the 5th and 95th percentile estimate of total health
benefits. For example, the likelihood that the 5th percentile estimates for each endpoint would simultaneously be drawn from a
Monte Carlo procedure is much less than 5 percent. As a result, we present only the total mean.
H-30
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Uncertainties in the Valuation
Estimates
The uncertainly ranges for the results on the
present value of the aggregate measured monetary
benefits reported in Table H-4 and Table H-5 reflect
two important sources of measured uncertainty:
• Uncertainty about the avoided incidence of
health and welfare effects deriving from
the concentration-response functions,
including both selection of scientific
studies and statistical uncertainty from the
original studies;
• Uncertainty about the economic value of
each quantified health and welfare effect.
These aggregate uncertainty results incorporate many
decisions about analytical procedures and specific
assumptions discussed in the Appendices to this
report.
In order to provide a more complete
understanding of the economic benefit results, we
conduct sensitivity analyses which examine several
additional important aspects of the main analysis. We
begin with an analysis of the sources of the measured
aggregate uncertainty, identifying which of the
measured uncertainty components of incidence and
valuation for individual health effects categories drive
the overall uncertainty results. We then follow with
an examination of several issues involving the
estimated economic benefits of mortality. In the third
section, we provide some insight into the potential
effects of income growth on the valuation of health
effects.
Relative Importance of Different
Components of Uncertainty
The estimated uncertainty ranges in our primary
results tables, Table H-4 and Table H-5, reflect the
measured uncertainty associated with both avoided
incidence and economic valuation. A better
understanding of the relative influence of individual
Figure H-4
Analysis of Contribution of Key Parameters to
Quantified Uncertainty
0\ 3>JUU
| $250-
a $200-
G«
i $150-
a
M $100-
o
o $50-
| $oJ
H
i
(
i
• 95th %ile
i
Mean
) (
5th %ile
i i
•
> 33 a e e e
i
i
i
+2 -&•§ ^o o-S^o-S^fe1 -^o fe1
rt 1 | 1 || 'o -o '^ 'o -o | ^ os -S 2 | ^ 2 H
-30 s.3 53 ^c_3bc3^^=g '« ^ Q.^
A
Comp
Contribution to Variance from the Mean
H-31
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
variables on the overall uncertainty in the analysis can
be gained by isolating the individual effects of
important variables on the range of estimated total
benefits. This can be accomplished by holding all the
inputs to the Monte Carlo uncertainty analysis
constant (at their mean values), and allowing only one
variable — for example, the economic valuation of
mortality — to vary across the range of that variable's
quantified uncertainty. The sensitivity analysis then
isolates how this single source of variability
contributes to the variation in the primary estimates of
total benefits. The results are summarized in Figure
H-4. The nine individual uncertainty factors that
contribute the most to the overall uncertainty are
shown in Figure H-4, ordered by the relative
significance of their contribution to overall
uncertainty. Each of the additional sources of
quantified uncertainty in the overall analysis not
shown contribute a smaller amount of uncertainty to
the estimates of monetized benefits than the sources
that are shown.
Economic Benefits Associated with
Reducing Premature Mortality
Because the economic benefits associated with
premature mortality are the largest source of
monetized benefits in the analysis, and because the
uncertainties in both the incidence and value of
premature mortality are the most important sources of
uncertainty in the overall analysis, it is useful to
examine the mortality benefits estimation in greater
detail. We begin with a discussion of the uncertainties
and possible biases related to the "benefits transfer"
approach employed to develop our VSL estimate. We
then discuss an alternative method for the valuation of
reduced premature mortality, value of statistical life
year (VSLY). We conclude this section with a
sensitivity test that compare the benefit estimates
using a VSL approach and a VSLY approach. Given
the lag structure employed in estimating reduced
premature mortality, we also provide alternative
calculations for the valuation of this benefits category
using two additional discount rates, three and seven
percent.
Benefits Tranfer and VSL
The analytical procedure used in the main analysis
to estimate the monetary benefits of avoided
premature mortality assumes that the appropriate
economic value for each incidence is a value from the
currently accepted range of the value of a statistical
life. As discussed above, the estimated value per
predicted incidence of excess premature mortality is
modeled as a Weibull distribution, with a mean value
of $4.8 million and a standard deviation of $3.2
million. This estimate is based on 26 studies of the
value of mortal risks.
There is considerable uncertainty as to whether
the 26 studies on the value of a statistical life provide
adequate estimates of the value of a statistical life
saved by air pollution reduction. Although there is
considerable variation in the analytical designs and
data used in the 26 underlying studies, the majority of
the studies involve the value of risks to a middle-aged
working population. Most of the studies examine
differences in wages of risky occupations, using a
wage-hedonic approach. Certain characteristics of
both the population affected and the mortality risk
facing that population are believed to affect the
average willingness to pay (WTP) to reduce the risk.
The appropriateness of a distribution of WTP
estimates from the 26 studies for valuing the
mortality-related benefits of reductions in air pollution
concentrations therefore depends not only on the
quality of the studies (i.e., how well they measure what
they are trying to measure), but also on (1) the extent
to which the risks being valued are similar, and (2) the
extent to which the subjects in the studies are similar
to the population affected by changes in pollution
concentrations. As discussed below, there are
possible sources of both upward and downward bias
in the estimates provided by the 26 studies when
applied to the population and risk being considered in
this analysis.
Although there may be several ways in which job-
related mortality risks differ from air pollution-related
mortality risks, the most important difference may be
that job-related risks are incurred voluntarily whereas
air pollution-related risks are incurred involuntarily.
H-32
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
There is some evidence (see, for example, Violette and
Chestnut, 1983) that people will pay more to reduce
involuntarily incurred risks than risks incurred
voluntarily. If this is the case, WTP estimates based
on wage-risk studies may be downward biased
estimates of WTP to reduce involuntarily incurred air
pollution-related mortality risks.
Another possible difference related to the nature
of the risk may be that some workplace mortality risks
tend to involve sudden, catastrophic events (e.g.,
workplace accidents), whereas air pollution-related
risks tend to involve longer periods of disease and
suffering prior to death. Several studies indicate that
the value people place on mortality risk reduction may
depend on the nature of the risk (e.g., Fisher et al.
1989; Beggs 1984). Some evidence suggests that WTP
to avoid a risk of a protracted death involving
prolonged suffering and loss of dignity and personal
control is greater than the WTP to avoid a risk (of
identical magnitude) of sudden death. Some
workplace risks, such as risks from exposure to toxic
chemicals, may be more similar to pollution-related
risks. It is not clear, however, what proportion of the
workplace risks in the wage-risk studies were related to
workplace accidents and what proportion were risks
from exposure to toxic chemicals. To the extent that
the mortality risks addressed in this assessment are
associated with longer periods of illness or greater
pain and suffering than are the risks addressed in the
valuation literature, the WTP measurements employed
in the present analysis would reflect a downward bias.
If the individuals who die prematurely from air
pollution are consistently older than the population in
the valuation studies, the mortality valuations based on
middle-aged people may provide a biased estimate of
the willingness to pay of older individuals to reduce
mortal risk. There is some evidence to suggest that
the people who die prematurely from exposure to
ambient particulate matter tend to be older than the
populations in the valuation studies. In the general
U.S. population far more older people die than
younger people; 88 percent of the deaths are among
people over 64 years old. It is difficult to establish the
proportion of the pollution-related deaths that are
among the older population because it is impossible to
isolate individual cases where one can say with even
reasonable certainty that a specific individual died
because of air pollution.
There is considerable uncertainty whether older
people will have a greater willingness to pay to avoid
risks than younger people. There is reason to believe
that those over 65 are, in general, more risk averse
than the general population, while workers in
wage-risk studies are likely to be less risk averse than
the general population. More risk averse people
would have a greater willingness to pay to avoid risk
than less risk averse people. Although the list of
recommended studies excludes studies that consider
only much-higher-than-average occupational risks,
there is nevertheless likely to be some selection bias in
the remaining studies — that is, these studies are likely
to be based on samples of workers who are, on
average, more risk-loving than the general population.
In contrast, older people as a group exhibit more risk
averse behavior.
The direction of bias resulting from the age
difference is unclear, particularly because age is
confounded by risk aversion (relative to the general
population). It could be argued that, because an older
person has fewer expected years left to lose, his WTP
to reduce mortality risk would be less than that of a
younger person. This hypothesis is supported by one
empirical study, Jones-Lee et al. (1985), that found the
value of a statistical life at age 65 to be about 90
percent of what it is at age 40. Citing the evidence
provided by Jones-Lee et al. (1985), a recent sulfate-
related health benefits study conducted for EPA (U.S.
EPA, 1995) assumes that the value of a statistical life
for those 65 and over is 75 percent of what it is for
those under 65. In addition, it might be argued that
because the elderly have greater average wealth than
those younger, the affected population is also
wealthier, on average, than wage-risk study subjects,
who tend to be blue collar workers. It is possible,
however, that among the elderly it is largely the poor
elderly who are most vulnerable to pollution-related
mortality risk (e.g., because of generally poorer health
care). If this is the case, the average wealth of those
affected by a pollution reduction relative to that of
subjects in wage-risk studies is uncertain. In addition,
H-33
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
the workers in the wage-risk studies will have
potentially more years remaining in which to acquire
streams of income from future earnings.
There is substantial evidence that the income
elasticity of WTP for health risk reductions is positive
(see, for example, Alberini et al., 1994; Mitchell and
Carson, 1986; Loehman and Vo Hu De, 1982;
Gerking et al., 1988; and Jones-Lee et al., 1985),
although there is uncertainty about the exact value of
this elasticity. Individuals with higher incomes (or
greater wealth) should be willing to pay more to
reduce risk, all else equal, than individuals with lower
incomes or wealth. This does not imply that
individuals with higher incomes are willing to pay
proportionally higher values. While many analyses
assume income elasticity of willingness to pay is unit
elastic (i.e., ten percent higher income level implies a
ten percent higher willingness to pay to reduce risk
changes), empirical evidence suggests that income
elasticity is substantially less than one.
The effects of income changes on WTP estimates
can influence benefit estimates in two different ways:
(i) as longitudinal changes that reflect estimates of
income change in the affected population over time,
and (ii) as cross-sectional changes based on differences
in income between study populations and the attracted
populations. Empirical evidence of the effect of
income on WTP gathered to date is based on studies
examining cross-sectional data. Income elasticity
adjustments to better account for changes over time,
therefore, will necessarily be based on potentially
inappropriate data.23
The need to adjust wage-risk-based WTP
estimates downward because of the likely upward bias
introduced by the age discrepancy has received
significant attention (see Chestnut, 1995; lEc, 1992).
If the age difference were the only difference between
the population affected by pollution changes and the
subjects in the wage-risk studies, there might be some
23For more information on the potential impact of income
elasticity on the valuation of health benefits, see the following
section, "Sensitivity Test for Impact of Income Changes Over
Time."
justification for trying to adjust the point estimate of
$4.8 million downward. Even in this case, however,
the degree of the adjustment would be unclear. There
is good reason to suspect, however, that there are
biases in both directions. Because in each case the
extent of the bias is unknown, the overall direction of
bias in the mortality values is similarly unknown.
Adjusting the estimate upward or downward to
compensate for any one source of bias could therefore
increase the degree of bias. Therefore, the range of
values from the 26 studies is used in the primary
analysis without adjustment.
VSLY
An alternative valuation of avoided premature
mortality is to use the VSLY. This approach uses life-
years lost as the unit of measure, rather than
estimating a single value of a statistical life lost
(applicable to all ages). With statistical life-years lost
as the unit of measure, the valuation depends on (1)
how many years of expected life are lost, (2) the
individual's discount rate, and (3) whether the value of
an undiscounted statistical life-year is the same no
matter which life-year it is (e.g., the undiscounted
value of the seventy-fifth year of life is the same as the
undiscounted value of the fortieth year of life).
We estimate the value of a statistical life-year
assuming that the value of a statistical life is directly
related to remaining life expectancy and a constant
value for each life-year. Such an approach results in
smaller values of a statistical life for older people, who
have shorter life expectancies, and larger values for
younger people. For example, if the $4.8 million
mean value of avoiding death for people with a 35
year life expectancy is assumed to be the discounted
present value of 35 equal-valued statistical life-years,
the implied value of each statistical life-year is
$293,000. This values assumes a five percent discount
rate and that the undiscounted value of a life-year is
the same no matter when it occurs in an individual's
life.
To obtain estimates of the number of air
pollution-related deaths in each age cohort, it is
preferable to have age-specific relative risks. Many of
H-34
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
the epidemiological studies, however, do not provide
any estimate of such age-specific risks. In this case,
the age-specific relative risks must be assumed to be
identical. Some epidemiology studies on PM do
provide some estimates of relative risks specific to
certain age categories. The limited information that is
available suggests that relative risks of mortality
associated with exposure to PM are greater for older
people. Most of the available information comes
from short-term exposure studies. There is
considerable uncertainty in applying the evidence from
short-term exposure studies to results from long-term
(chronic exposure) studies. However, using the
available information on the relative magnitudes of the
relative risks, it is possible to form a preliminary
assessment of the relative risks by different age
classes.
The analysis presented below uses two alternative
assumptions about age-specific risks: (1) there is a
constant relative risk (obtained directly from the
health literature) that is applicable to all age cohorts,
and (2) the relative risks differ by age, as estimated
from the available literature. Estimates of age-specific
PM-10 coefficients (and, from these, age-specific
relative risks) were derived from the few age-specific
PM-10 or TSP coefficients reported in the
epidemiological literature. These estimates in the
literature were used to estimate the ratio of each age-
specific coefficient to a coefficient for "all ages" in
such a way that consistency among the age-specific
coefficients is preserved — that is, that the sum of the
health effects incidences in the separate, non-
overlapping age categories equals the health effects
incidence for "all ages." These ratios were then
applied to the coefficient from Pope et al. (1995).
Details of this approach are provided in Post and
Deck (1996). Because Pope et al. considered only
individuals age 30 and older (instead of all ages), the
resulting age-specific PM coefficients may be slightly
different from what they would have been if the ratios
had been applied to an "all ages" coefficient. The
differences, however, are likely to be minimal and well
within the error bounds of this exercise. The age-
specific relative risks used in the example below
assume that the relative risks for people under 65 are
only 16 percent of the population-wide average
relative risk, the risks for people from 65 to 74 are 83
percent of the population-wide risk, and people 75
and older have a relative risk 55 percent greater than
the population average. Details of this approach are
provided in Post and Deck (1996).
The life-years lost approach also requires an
estimate of the number of life-years lost by a person
dying prematurely at each given age. The approach
developed for this analysis assumes that exposure to
elevated levels of PM increases the probability of
dying at a specific age. Increasing the probability of
dying at each age lowers the life expectancy for each
age cohort. The average number of life-years lost will
depend on the distribution of ages in the population
in a location. In addition, this analysis incorporates
the five-year PM mortality lag structure described in
Chapter 5 and Appendix D. It distributes the
mortality for each cohort across a five-year period (25
percent in each of the first two years, 16.7 percent in
each of the remaining years) and adjusts the loss of
life expectancy accordingly. That is, when applying
the lag assumption within a given cohort, individuals
who die later are expected to lose fewer life years than
those who die earlier. Further, this analysis applies a
five percent discount rate when calculating the present
discounted value of the avoided losses of life
expectancy in each cohort over the five-year lag
period.
The life-years lost approach used here assumes
that people who die from air pollution are typical of
people in their age group. The estimated value of the
quantity of life lost assumes that the people who die
from exposure to air pollution had an average life
expectancy. However, it is possible that the people
who die from air pollution are already in ill health, and
that their life expectancy is less than a typical person
of their age. If this is true, then the number of life
years lost per PM-related death would be lower than
calculated here, and the economic value would be
smaller.
The extent to which adverse effects of particulate
matter exposure are differentially imposed on people
of advanced age and/or poor health is one of the
most important current uncertainties in air pollution-
related health studies. There is limited information,
H-35
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
primarily from the short-term exposure studies, which
suggests that at least some of the estimated premature
mortality is imposed disproportionately on people
who are elderly and/or of poor health. Rowlatt, et al.
(1998) indicate that at risk individuals include those
who have suffered strokes or are suffering from
cardiovascular disease and angina. The Criteria
Document for Particulate Matter (U.S. EPA, 1996),
however, identifies only two studies which attempt to
evaluate the disproportionality in premature mortality
among people who are elderly and/or sickly. Spix et
al. (1994) suggests that a small portion of the PM-
associated mortality occurs in individuals who would
have died in a short time anyway. Cifuentes and Lave
(1996) found that 37 to 87 percent of the deaths from
short-term exposure could have been premature by
only a few days, although their evidence is
inconclusive.
Prematurity of death on the order of only a few
days is likely to occur largely among individuals with
pre-existing illnesses. Such individuals might be
particularly susceptible to a high PM day. To the
extent that the pre-existing illness is itself caused by or
exacerbated by chronic exposure to elevated levels of
PM, however, it would be misleading to define the
prematurity of death as only a few days. In the
absence of chronic exposure to elevated levels of PM,
the illness would either not exist (if it was caused by
the chronic exposure to elevated PM) or might be at
a less advanced stage of development (if it was not
caused by but was exacerbated by elevated PM levels).
The prematurity of death should be calculated as the
difference between when the individual died in the
"elevated PM" scenario and when he would have died
in the 'low PM" scenario. If the pre-existing illness
was entirely unconnected with chronic exposure to
PM in the "elevated PM" scenario, and if the
individual who dies prematurely because of a peak PM
day would have lived only a few more days, then the
prematurity of that PM-related death is only those few
days. If, however, in the absence of chronic exposure
to elevated levels of PM, the individual's illness would
have progressed more slowly, so that, in the absence
of a particular peak PM day the individual would have
lived several years longer, the prematurity of that PM-
related death would be those several years.
Long-term studies provide evidence that a portion
of the loss of life associated with long-term exposure
is independent of the death from short-term
exposures, and that the loss of life-years measured in
the long-term studies could be on the order of years.
If much of the premature mortality associated with
PM represents short term prematurity of death
imposed on people who are elderly and/or of ill
health, the estimates of the monetary benefits of
avoided mortality may overestimate society's total
willingness to pay to avoid particulate matter-related
premature mortality. On the other hand, if the
premature mortality measured in the chronic exposure
studies is detecting excess premature deaths which are
largely independent of the deaths predicted from the
short term studies, and the disproportionate effect on
the elderly and/or sick is modest, the benefits
measured in this report could be underestimates of
the total value. At this time there is insufficient
information from both the medical and economic
sciences to satisfactorily resolve these issues from a
theoretical/analytical standpoint. Until there is
evidence from the physical and social sciences which
is sufficiently compelling to encourage broad support
of age-specific values for reducing premature
mortality, EPA will continue to use for its primary
analyses a range of values for mortality risk reduction
which assumes society values reductions in pollution-
related premature mortality equally regardless of who
receives the benefit of such protection.
Sensitivity Test of Benefits Due to
Reduced Premature Mortality Valuation
Examining the sensitivity of the total benefits of
reduced premature mortality to alternative valuation
techniques does provide some illumination to the
potential impacts of alternative approaches. This
section presents alternative results to our primary
estimate of mortality valuation using the life-years lost
approach, and also examine the effects of alternative
discount rates.
The life-years lost approach also requires an
estimate of the number of life-years lost by a person
dying prematurely at each given age. The approach
developed for this analysis assumes that exposure to
H-36
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
elevated levels of PM increases the probability of
dying at a specific age. Increasing the probability of
dying at each age lowers the life expectancy for each
age cohort. The average number of life-years lost will
depend on the distribution of ages in the population
in a location. In addition, this analysis incorporates
the five-year PM mortality lag structure described in
Chapter 5 and Appendix D. It distributes the
mortality for each cohort across a five-year period (25
percent in each of the first two years, 16.7 percent in
each of the remaining years) and adjusts the loss of life
expectancy accordingly. That is, when applying the lag
assumption within a given cohort, individuals who die
later are expected to lose less live expectancy than
those who die earlier. Further, this analysis applies a
five percent discount rate when summing the value of
the avoided losses of life expectancy in each cohort
over the five-year lag period.
The alternative central estimates for avoided PM-
related premature mortality using a five percent dis-
count rate are $33 billion in 2000 and $53 billion in
2010. The VSLY approach results in estimates that
are almost 50 percent lower than our primary est-
imates of benefits due to avoided pre-mature mor-
tality. The sensitivity analysis, however, indicates that
the pattern of monetized mortality benefits with each
valuation procedure is essentially invariant to the dis-
count rate. We summarize these results in Table H-6.
We emphasize that the results of the VSLY
approach to valuing avoided mortality benefits
represent a crude estimate of the value of changes in
age-specific life expectancy. These results should be
interpreted cautiously, due to the several significant
assumptions required to generate a monetized
estimate of life years lost from the relative risks
reported in the Pope et al., 1995 study and the
available economic literature. These assumptions
include, but are not limited to: extrapolation of the
age distribution of the U.S. population in future years;
assumptions about the age-specificity of the relative
risk reported by Pope et al., 1995; assumptions about
the life expectancy of different age groups;
assumption of a particular lag structure; assumptions
about the age-specificity of the lag period (if any);
derivation of VSLY estimates from VSL estimates;
assumptions about the variation in VSLY with age;
and selection of an appropriate rate at which to
discount the lagged estimates of life years lost.
Changes in any of these assumptions could
significantly affect the VSLY benefit estimate. For
example, if we were to assume no lag period for PM-
related mortality effects instead of the five-year lag
structure, VSLY benefit estimates would increase
from $53 billion to $61 billion.
Table H-6
Sensitivity Analysis of Alternative Discount Rates on the Valuation of Reduced Premature
Mortality
Benefit Category &
Discount Rate
2000 (in millions, 1990$) 2010 (in millions, 1990$)
5th %ile Central 95th %ile 5th %ile Central 95th %ile
VSL Approach
3% Discount Rate
5% Discount Rate
7% Discount Rate
VSLY Approach
3% Discount Rate
5% Discount Rate
7% Discount Rate
$8,900
8,600
8,300
$4,600
5,000
5,400
$65,000
63,000
61,000
$30,000
33,000
35,000
$150,000
150,000
150,000
$68,000
74,000
80,000
$ 14,000
14,000
14,000
$ 7,400
8,100
8,800
$100,000
100,000
97,000
$48,000
53,000
57,000
$ 250,000
250,000
240,000
$ 110,000
120,000
130,000
Note: The discount rate affects the benefits estimates of VSL and VSLY approach differently. With the VSL approach, higher
discount rates lead to lower estimates because of the lag structure. With the VSLY approach, the higher discount rates lead to
higher estimates because of its affect on the annualized values.
H-37
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Sensitivity Test for Impact of Income
Changes Over Time
As an illustrative calculation, we adjust
willingness-to-pay (WTP) measures to reflect the
expected increase in real income over the full period
of the analysis, 1990 to 2010. Our procedure results
in an upward adjustment to more accurately reflect the
valuation of improved health as income increases over
time. In this section, we describe the procedure we
use and the results of our illustrative calculation.
Background and Methodology
Economists use income elasticity to evaluate how
private and public goods are valued based on the
interaction between income changes and demand. A
negative relationship between income and demand for
a good implies that the good is an inferior good. An
individual demands less of a good as income rises. A
positive relationship between income and the demand
for a good implies that the good is normal (i.e.,
income elasticity is greater than zero). As income rises
an individual demands more of a good. Depending on
the relative responsiveness of demand to income
changes, normal goods are characterized as a necessity
or a luxury. When income elasticity is between 0 and
+ 1, the good is considered a necessity (i.e., demand is
not significantly responsive to income). In contrast,
when income elasticity exceeds +1, the good is
considered a luxury (i.e., the relative increase in the
good's demand exceeds the increase in income).
The determination of a public good as inferior or
normal based on income elasticity is complicated by
its nonrival nature. In the case of a private good,
varying the level of consumption is measured as a
marginal change and implies that an individual will
adjust his or her consumption level of other good(s).
Consequently, income elasticity of demand estimates
a change in quantity consumed, and not necessarily a
change in utility (or the individual's well-being). With
public goods, the conceptual logic is different.
Income elasticity of WTP for public goods measures
changes in consumer surplus. For example, one
person enjoying the benefits of cleaner air does not
reduce the probability of another person enjoying the
same benefits. There are no apparent mechanisms for
regulating who specifically will enjoy the benefits. In
other words, there is no direct relationship between an
individual's WTP and level of consumption.24 The
consumption level of public goods is exogenous to
the individual's budget constraint. At the same time,
WTP for a public good is not exogenous. An
individual, therefore, must consider how his or her
WTP affects the allocation of income among private
and public goods.25
Flores and Carson (1997) provide examples of
how income elasticity can change depending on how
the good is defined (i.e., private or public). Given the
divergence between private and public goods, they
conclude that income elasticity of WTP and income
elasticity of demand are related. The relationship does
not imply that knowledge of income elasticity of
demand is sufficient to estimate income elasticity of
WTP given that the income elasticity of WTP depends
on factors that cannot be observed.
In addition to the theoretical issues associated
with WTP for public goods, there are important
empirical issues. We are interested in how WTP
changes with respect to increases in U.S. median
income. Measuring changes due to growth in median
income reflect shifts in overall preferences and utility
(or in the case of public goods, social welfare). This
type of analysis requires time series data.
Unfortunately, there are very few relevant studies that
use this approach to estimate income elasticity.26
Consequently, we must rely on income elasticities
estimated from cross-sectional data. The estimates
24The nonrival nature of public goods implies that the
marginal social cost of consuming an additional unit of benefit is
zero.
25CV studies solicit WTP estimates that are subject to the
respondent's current budget constraint. The budget share factor
requires that the income elasticities (for all consumed goods) sum
to one. This generally implies that income elasticity of any single
good is substantially less than one.
26Available studies using time series data estimate income
elasticity of public health care expenditures by analyzing changes
in government spending relative to gross domestic product
(GDP). These studies are not particularly applicable to the
valuation methodology used in the present study.
H-38
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
reflect differences in willingness to pay for improved
health among various income levels. They are
measures of an individual's preferences and expected
utility given the person's current state (i.e., in the
present).
There are several issues associated with the
application of cross-sectional results to estimate
longitudinal changes (i.e., changes over time). Most
important is the potential for misinterpretation of our
recommended application of income elasticity
adjustment. Although we outline an approach that
uses income elasticities derived from cross-sectional
data, the adjustment is solely a proxy for how
preferences and utility may change as projected overall
average income (i.e., real GDP per capita) increases
from 1990 to 2010. Application of these income
elasticity estimates does not imply a strategy for
adjusting benefits valuation by level of household
income in any given year.
Derivation of Elasticity Estimates
Based on our review of the available income
elasticity literature, we conducted sensitivity analyses
that characterize how the valuation of human health
benefits may increase with a rise in real U.S. income.
Given the range of different methodological
approaches and limited available research, we calculate
a range of illustrative values. Table H-7 summarizes
the income elasticities we used to conduct the
sensitivity analysis.
Table H-7
Elasticity Values for Conducting Sensitivity Analysis
Health Endpoint
Lower Estimate
Central Estimate
Upper Estimate
Minor Health Effect 0.04
Severe and Chronic n 9C-
Health Effects
Premature Mortality 0.08
Note: Sources for the derivation of these values can be found
0.14
0.45
0.40
in Industrial Economics 1999.
0.30
0.60
1.00
Reported income elasticities suggest that the
severity of the morbidity endpoint is a primary
determinant of the strength of the relationship
between changes in income and the willingness to pay.
Without accounting for severity, there is a fairly wide
range of values for income elasticity, 0.04 to 0.60.
Estimates are more closely clustered if we account for
the seriousness of the health effect. For the purposes
of a sensitivity analysis, we use two different ranges
based on whether morbidity endpoints are minor or
severe. With respect to minor health effects, we use
lower and upper values of 04 and 0.30, respectively.
The central estimate is 0.14. For conducting a
sensitivity test of the income elasticity effect on WTP
to avoid severe health effects, we use a lower and
upper estiamtes of 0.25 and 0.60, with 0.45 as the
central estimate. The lower and upper estimates
reflect the lowest and highest estimates derived from
our literature review. The central estimate is the
midpoint of the averages from each study.
With respect to VSL, estimates of income
elasticity range from 0.08 to 1.10. We use lower and
upper estimates that reflect the full range of values.
The central estimate, 0.40, represents the midpoint
between the average low value and the average high
value of the studies we reviewed.
H-39
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Illustrative Calculations —
Morbidity Benefits Estimates
Table H-8 provides a simplified example of how
application of the elasticity ranges we derive could
affect benefits estimates. For illustrative purposes, we
use the WTP to avoid an asthma attack to represent a
minor health effect and "WTP to avoid a case of
chronic bronchitis to represent a severe health effect.
By the year 2010, the effect of income growth on
WTP for a minor health effect can increase between
one and eight percent, with the central estimate
indicating three percent growth. The WTP to avoid a
severe health effect grows faster with 2010 estimates,
ranging between seven and sixteen percent and with
the central estimate increasing by thirteen percent.
Table H-8
Illustrative Adjustment to Estimates of WTP to Avoid Morbidity
WTP Estimate (1990 Dollars)1
Year
Minor Health
1990
2000
2010
US Population
(in millions)
Effect- Asthma
249,440
274,634
297,716
Severe Health Effect- Chronic
1990
2000
2010
249,440
274,634
297,716
Real GDP
(in millions)
5,744
7,123
8,959
Bronchitis
5,744
7,123
8,959
Income
23,026
25,936
30,092
23,026
25,936
30,092
Lower
Estimate
Ey=0.04
$32
$32.20
$32.30
Ey=0.25
$260,000
$267,850
$277,990
Central
Estimate
Ey=0.14
$32
$32.50
$33.20
Ey=0.45
$260,000
$274,300
$293,280
Upper
Estimate
Ey=0.30
$32
$33.20
$34.70
Ey=0.60
$260,000
$279,240
$305,290
Note:
1 WTP estimates are reported in undiscounted 1990 dollars and represent value per case.
H-40
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Illustrative Calculations —
VSL Estimate
We characterize the potential effect of income
elasticity on the VSL estimate in Table H-9. An
income elasticity of 0.08 demonstrates the effect of a
slight adjustment to the VSL estimates as median
income gradually rises. As shown in the figure,
between 1990 and 2010, the VSL estimates increase by
approximately two percent. The central estimate, 0.40,
demonstrates that by 2010, a thirty percent increase in
median income would result in VSL increasing by
approximately eleven percent. The upper bound value
demonstrates the effect of assuming one as the value
of income elasticity. In this twenty year period of the
prospective analysis, the VSL estimate would increase
from $4.8 to $6.3 million if income elasticity equals
one.
Table H-9
Illustrative Adjustment to Estimates of The Value of Statistical Life
Value of Life Estimate (in thousands)1
Year
1990
2000
2010
US Population
(in millions)
249,440
274,634
297,716
Real GDP
(in millions)
5,744
7,123
8,959
Income
23,026
25,936
30,092
Lower
Estimate
Ey=0.08
$4,800
$4,848
$4,905
Central
Estimate
Ey=0.40
$4,800
$5,036
$5,345
Upper
Estimate
£,=1.0
$4,800
$5,410
$6,271
Note:
1 Value of life estimates reported in undiscounted 1990 dollars.
H-41
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
References
Abbey, D.E., F. Petersen, P. K. Mills, and W. L. Beeson. 1993. "Long-Term Ambient Concentrations of
Total Suspended Particulates, Ozone, and Sulfur Dioxide and Respiratory Symptoms in a
Nonsmoking Population." Archives of Environmental Health 48(1): 33-46.
Abt Associates, Inc. 1992. The Medical Costs of Five Illnesses Related to Exposure to Pollutants. Prepared
for U.S. EPA, Office of Pollution Prevention and Toxics, Washington, DC.
Abt Associates, Inc. 1996. Section 812 Retrospective Analysis: Quantifying Health and Welfare Benefits.
Draft. Prepared for U.S. EPA, Office of Policy Planning and Evaluation, Washington DC. May.
Alberini, A., A. Krupnick, M. Cropper, and W. Harrington. 1994. "Air Quality and the Value of Health in
Taiwan." Paper presented at the annual meeting of the Eastern Economics Association, Boston,
Massachusetts, March.
Beggs, Steven D. 1984. Diverse Risks and the Relative Worth of Government Health and Safety Programs:
An Experimental Survey. U.S. Environmental Protection Agency, EPA-223-04-85-005. June.
Blumenschein, Karen and Magnus Johannesson. 1998. "Relationship Between Quality of Life Instruments,
Health State Utilities, and Willingness to Pay in Patients with Asthma." Annals of Allergy, Asthma,
and Immunology 80:189-194.
Brookshire, David S., Ralph C. d'Arge, William D. Schulze and Mark A. Thayer. 1979. Methods
Development for Assessing Air Pollution Control Benefits, Vol. II: Experiments in Valuing Non-
Market Goods: A Case Study of Alternative Benefit Measures of Air Pollution Control in the South
Coast Air Basin of Southern California. Prepared for the U.S. Environmental Protection Agency,
Office of Research and Development.
Chestnut, Lauraine G. 1995. Dollars and Cents: The Economic and Health Benefits of Potential Particulate
Matter Reductions in the United States. Prepared for the American Lung Association.
Chestnut, Lauraine G. and Robert D. Rowe. 1989. "Economic Valuation of Changes in Visibility: A State of
the Science Assessment for NAPAP," as cited in National Acid Precipitation Assessment Program,
Methods for Valuing Acidic Deposition and Air Pollution Effects. NAPAP State of Science and
State of Technology Report No. 27, Part B. December.
Chestnut, Lauraine G. and Robin Dennis. 1997. "Economic Benefits of Improvements in Visibility: Acid
Rain Provisions of the 1990 Clean Air Act Ammendments." Journal of Air and Waste Management
Association 47:395-402.
Cifuentes, L. and L.B. Lave. 1996. "Association of Daily Mortality and Air Pollution in Philadelphia, 1983-
1988." J. Air Waste Manage. Assoc.: in press.
Crocker T. D. and R. L. Horst, Jr. 1981. "Hours of Work, Labor Productivity, and Environmental
Conditions: a Case Study." The Review of Economics and Statistics 63:361-368.
H-42
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Cropper, M.L. and AJ. Krupnick. 1990. "The Social Costs of Chronic Heart and Lung Disease," Resources
for the Future Discussion Paper QE 89-16-REV.
Cropper, Maureen L. and Frances G. Sussman. 1990. '"Valuing Future Risks to Life." Journal of
Environmental Economics and Management. 19:160-174.
Cutler, David M. and Elizabeth Richardson. 1998. "The Value of Health: 1970—1990." The American
Economic Review. 88(2):97-100. May.
Desvouges, William H., F. Reed Johnson, and H. Spencer Banzhaf. 1998. Environmental Policy Analysis with
Limited Information, Principles and Applications of the Transfer Method. Edward Elgar Publishing
Limited: Northampton.
Dickie, M. et al. 1991. Reconciling Averting Behavior and Contingent Valuation Benefit Estimates of
Reducing Symptoms of Ozone Exposure (draft), as cited in Neumann, J.E., Dickie, M.T., and R.E.
Unsworth. 1994. Industrial Economics, Incorporated. Memorandum to Jim DeMocker, U.S. EPA,
Office of Air and Radiation. March 31.
Elixhauser, A., R.M. Andrews, and S. Fox. 1993. Clinical Classifications for Health Policy Research:
Discharge Statistics by Principal Diagnosis and Procedure. Agency for Health Care Policy and
Research (AHCPR), Center for General Health Services Intramural Research, U.S. Department of
Health and Human Services.
Fisher, Ann, Lauraine G. Chestnut, and Daniel M. Violette. 1989. "The Value of Reducing Risks of Death:
A Note on New Evidence." Journal of Policy Analysis and Management. 8(1):88-100.
French, Michael T. and Josephine A. Mauskopf. 1992. "A Quality-of-Life Method for Estimating the Value
of Avoided Morbidity." American Journal of Public Health. 82(11):1553-1555. November.
Garber, Alan M. and Charles E. Phelps. 1997. "Economic Foundations of Cost-Effectiveness Analysis."
Journal of Health Economics. 16:1-31.
Gerking, S., M. DeHaan, and W. Schulze. 1988. "The Marginal Value of Job Safety: A Contingent Valuation
Study." Journal of Risk and Uncertainty 1: 185-199.
Industrial Economics, Incorporated (lEc). 1992. Approaches to Environmental Benefits Assessment to
Support the Clean Air Act Section 812 Analysis. Prepared by Robert E. Unsworth, James E.
Neumann, and W. Eric Browne, for Jim DeMocker, Office of Policy Analysis and Review, Office of
Air and Radiation, U.S. Environmental Protection Agency. 6 November.
Industrial Economics, Incorporated (lEc). 1993a. "Analysis of Visibility Valuation Issues for the Section
812 Study," Memorandum to Jim DeMocker, Office of Policy Analysis and Review, Office of Air
and Radiation, U.S. Environmental Protection Agency, prepared by Jim Neumann, Lisa Robinson,
and Bob Unsworth. September 30.
H-43
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Industrial Economics, Incorporated (lEc). 1997. "Visibility Valuation for the CAA Section 812
Retrospective Analysis," Memorandum to Jim DeMocker, Office of Policy Analysis and Review,
Office of Air and Radiation, U.S. Environmental Protection Agency, prepared by Michael H. Hester
and James E. Neumann. 18 February.
Industrial Economics, Incorporated (lEc). 1999. "Recommended Approach to Adjusting WTP Estimates to
Reflect Changes in Real Income," Memorandum to Jim DeMocker, Office of Policy Analysis and
Review, Office of Air and Radiation, U.S. Environmental Protection Agency, prepared by Naomi
Kleckner and James E. Neumann. 3 June.
Ireland, Thomas R. and Roy Gilbert. 1998. "Supramonetary Values, the Value of Life, and the Utility
Theory Meanings of Tort Recovery." Journal of Forensic Economics. 11:189-201.
Irwin, Julie, William Schulze, Gary McClelland, Donald Waldman, David Schenk, Thomas Stewart, Leland
Deck, Paul Slovic, Sarah Lictenstein, and Mark Thayer. 1990. Valuing Visibility: A Field Test of the
Contingent Valuation Method. Prepared for Office of Policy, Planning and Evaluation, U.S.
Environmental Protection Agency. March.
Johannesson, Magnus and Per-Olov Johansson. 1996. "To Be or Not to Be, That is the Question: An
Empirical Study of the WTP for an Increased Life Expectancy at an Advanced Age." Journal of
Risk and Uncertainty. 13:163-174.
Johannesson, Magnus and Per-Olov Johansson. 1997. "The Value of Life Extension and the Marginal Rate of
Time Preference: A Pilot Study." Applied Economic Letters. 4:53-55.
Johannesson, Magnus, Per-Olov Johansson, and Karl-Gustav Lofgren. 1997. "On the Value of Changes in Life
Expectancy: Blips Versus Parametric Changes." Journal of Risk and Uncertainty. 15:221-239.
Johansson, Per-Olov. 1995. Evaluating Health Risks: An Economic Approach. Cambridge University
Press: Great Britain.
Jones-Lee, M.W., et al. 1985. "The Value of Safety: Result of a National Sample Survey." Economic Journal
95(March): 49-72.
Jones-Lee, M.W. 1992. "Paternalistic Altruism and the Value of Statistical Life." The Economic Journal.
102:80-90. January.
Jones-Lee, M.W., M. Hammerton, and P.R. Philips. 1985. "The Value of Safety: Results of a National Sample
Survey. Economic Journal. 95: 49-72. March.
Krupnick, AJ. and M.L. Cropper. 1992. "The Effect of Information on Health Risk Valuations," Journal of
Risk and Uncertainty 5(2): 29-48.
Krupnick, AJ. and RJ. Kopp. 1988. Health and Agricultural Benefits of Reductions in Ambient Ozone in
the United States. Resources for the Future.
H-44
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Loehman, E.T., S.V. Berg, A.A. Arroyo, R.A. Hedinger, J.M. Schwartz, M.E. Shaw, R.W. Fahien, V.H. De,
R.P. Fishe, D.E. Rio, W.F. Rossley, and A.E.S. Green. 1979. "Distributional Analysis of Regional
Benefits and Cost of Air Quality Control." Journal of Environmental Economics and Management
6: 222-243.
Loehman, E.T. and Vo Hu De. 1982. "Application of Stochastic Choice Modeling to Policy Analysis of
Public Goods: A Case Study of Air Quality Improvements." The Review of Economics and
Statistics 64(3): 474-480.
Manuel, E.H., R.L. Horst, K.M. Brennan, W.N. Lanen, M.C. Duff, and J.K Tapiero. 1982. Benefits Analysis
of Alternative Secondary National Ambient Air Quality Standards for Sulfur Dioxide and Total
Suspended Particulates, Volumes I-IV. Prepared for U.S. Environmental Protection Agency, Office
of Air Quality Planning and Standards, Research Triangle Park, NC.
McClelland, Gary, William Schulze, Donald Waldman, Julie Irwin, David Schenk, Thomas Stewart, Leland
Deck and Mark Thayer. 1991. Valuing Eastern Visibility: A Field Test of the Contingent Valuation
Method. Prepared for Office of Policy, Planning and Evaluation, U.S. Environmental Protection
Agency. June.
Miller, T.R., C. Calhoun, and W.B. Arthur. 1990. Utility-adjusted Impairment Years: A Low-Cost Approach
to Morbidity Valuation. Federal Highway Administration. March.
Mitchell, R.C. and R.T. Carson. 1986. "Valuing Drinking Water Risk Reductions Using the Contingent
Valuation Methods: A Methodological Study of Risks from THM and Giardia." Paper prepared for
Resources for the Future, Washington, DC.
Moore, MJ. and W.K. Viscusi. 1988. "The Quantity-Adjusted Value of Life". Economic Inquiry 26(3): 369-
388.
Murray, Christopher J.L. and Alan D. Lopez, eds. 1996 The Global Burden of Disease. Harvard School of
Public Health on behalf of the World Health Organization and the World Bank.
National Acid Precipitation Assessment Program (NAPAP). 1991. Acidic Deposition: State of Science and
Technology (Summary Report). (Washington, DC: NAPAP). September.
Neumann, J.E., M. T. Dickie, and R.E. Unsworth. 1994. Industrial Economics, Incorporated.
Memorandum to Jim DeMocker, U.S. EPA, Office of Air and Radiation. Linkage Between Health
Effects Estimation and Morbidity Valuation in the Section 812 Analysis — Draft Valuation
Document. March 31.
O'Conor, Richard M. and Glenn C. Blomquist, 1997. "Measurement of Consumer-Patient Preferences
Using a Hybrid Contingent Valuation Method." Journal of Health Economics 16:667-683.
Ostro, B.D., MJ. Lipsett, J.K. Mann, H. Braxton-Owens, and M.C. White. 1995. "Air Pollution and Asthma
Exacerbations Among African American Children in Los Angeles." Inhalation Toxicology.
H-45
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
Pope, C.A., III, MJ. Thun, M.M. Namboodin, D.W. Dockery, J.S. Evans, F.E. Speizer, and C.W. Heath, Jr.
1995. "Particulate Air Pollution as a Predictor of Mortality in a Prospective Study of U.S. Adults."
Am. J. Respir. Grit. Care Med. 151: 669-674.
Post, Ellen and L. Deck. 1996. Abt Associates Inc. Memorandum to Tom Gillis, U.S. EPA, Office of
Office of Policy Planning and Evaluation. September 20.
Rowe, R.D. and L.G. Chestnut. 1986. "Oxidants and Asthmatics in Los Angeles: A Benefits Analysis--
Executive Summary." Prepared by Energy and Resource Consultants, Inc. Report to the U.S. EPA,
Office of Policy Analysis. EPA-230-09-86-018. Washington, B.C. March.
Rowlatt, Penelope, Michael Spackman, Sion Jones, Michael Jones-Lee, and Graham Loomes. 1998. Valuation
of Deaths from Air Pollution. For the Department of Environment, Transport and the Regions and
the Department of Trade and Industry. February.
Salkever, D.S. 1995. "Updated Estimates of Earnings Benefits from Reduced Exposure of Children to
Environmental Lead." Environmental Research 70: 1-6.
Schwartz, J. 1994. "Societal Benefits of Reducing Lead Exposure." Environmental Research 66: 105-124.
Selvin, Steve. 1996. Statistical Analysis of Epidemiologic Data. Second Edition. Monographs in
Epidemiology and Biostatistics, Volume 25. Oxford University Press.
Smith, David H., Daniel C. Malone, Kenneth A. Lawson, Lynn J. Okamoto, Carmelina Battista, and William
B. Saunders, 1997. "A National Estimate of the Economic Costs of Asthma," Am. J. Respir. Crit.
Care Med. 156:787-793.
Smith, V. Kerry, George Van Houtven, Subhrendu Pattanayak, 1999. "Benefit Transfer as a Preference
Calibration." Resources for the Future, Discussion Paper 99-36. May.
Spix, C., J. Heinrich, D. Dockery, J. Schwartz, G. Volksch, K Schwinkowski, C. Collen, and H.E. Wichmann.
1994. Summary of the Analysis and Reanalysis Corresponding to the Publication Air Pollution and
Daily Mortality in Erfurt, East Germany 1980-1989. Summary report for: Critical Evaluation
Workshop on Particulate Matter—Mortality Epidemiology Studies; November; Raleigh, NC.
Wuppertal, Germany: Bergische Universitat-Gesamthochschule Wuppertal.
Taylor, T.N., P.H. Davis, J.C. Torner, J. Holmes, J.W. Meyer, and M. F. Jacobson. 1996. "Lifetime Cost of
Stroke in the United States." Stroke 27(9): 1459-1466.
Tolley, G.S. et al. 1986. Valuation of Reductions in Human Health Symptoms and Risks. University of
Chicago. Final Report for the U.S. Environmental Protection Agency. January.
Tolley, George, Donald Kenkel, and Robert Fabian, eds. 1994. Valuing Health for Policy: An Economic
Approach. The University of Chicago Press.
U.S. Department of Commerce, Economics and Statistics Administration. 1992. Statistical Abstract of the
United States, 1992: The National Data Book. 112th Edition, Washington, D.C.
H-46
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
U.S. Environmental Protection Agency (U.S. EPA). 1994. Documentation for Oz-One Computer Model
(Version 2.0). Office of Air Quality Planning and Standards. Prepared by: Mathtech, Inc., under
EPA Contract No. 68D30030, WA 1-29. August.
U.S. Environmental Protection Agency (U.S. EPA). 1995. Human Health Benefits From Sulfate Reductions
Under Title IV of the 1990 Clean Air Act Amendments. Prepared by Hagler Bailly Consulting, Inc.
for U.S. EPA, Office of Air and Radiation, Office of Atmospheric Programs. November 10.
U.S. Environmental Protection Agency (U.S. EPA). 1996. Air Quality Criteria for Particulate Matter,
Volume III of III. Office of Research and Development, Washington DC. EPA/600/P-95/001cF
U.S. Environmental Protection Agency (U.S. EPA). 1997 The Benefits and Costs of the Clean Air Act, 1970
to 1990. Prepared for U.S. Congress. October.
Violette, D.M. and L.G. Chestnut. 1983. Valuing Reduction in Risks: A Review of the Empirical Estimates.
Report prepared for the U.S. Environmental Protection Agency, Washington, D.C. EPA-230-05-83-
002.
Viscusi, W.K. 1992. Fatal Tradeoffs: Public and Private Responsibilities for Risk. (New York: Oxford
University Press).
Viscusi, W.K., W. A. Magat, and J. Huber. 1991. "Pricing Environmental Health Risks: Survey Assessments
of Risk-Risk and Risk-dollar Tradeoffs." Journal of Environmental Economics and Management
201: 32-57.
Viscusi, W. Kip. 1992. Fatal Tradeoffs: Public & Private Responsibilities for Risk. Oxford University Press:
New York.
Viscusi, W. Kip. 1993. "The Value of Risks to Life and Health." Journal of Economic Literature. 31(4): 1912-
46. October.
Wittels, E.H., J.W. Hay, and A.M. Gotto, Jr. 1990. "Medical Costs of Coronary Artery Disease in the United
States." The American Journal of Cardiology 65: 432-440.
World Health Organization (WHO). 1996. Final Consultation on Updating and Revision of the Air Quality
Guidelines for Europe. Bilthoven, The Netherlands 28-31 October, 1996 ICP EHH 018 VD96
2.11.
H-47
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
[This page left blank intentionally.]
H-48
-------
Implications for
Future Research
C
0)
Q.
Q.
Throughout this report we have attempted to
accurately characterize and, where possible, quantify
the major sources of uncertainties that affect our
primary estimate of the costs and benefits of the
CAAA. In many cases, these uncertainties are the
result of gaps in data or methods that might be
addressed through additional research. In this
Appendix, we provide a summary of important areas
for new research which, if carried out, have the
potential to increase accuracy and reduce uncertainty
in future assessments.
Overview
The uncertainties in the primary estimates and the
controversies which persist regarding model choices
and valuation paradigms highlight the need for a
variety of new and continued research efforts. Based
on the findings of this study, the highest priority
research needs are:
• Improved emissions inventories and
inventory management systems
• Improved tools for assessing the full range of
social costs associated with regulation,
including the tax-interaction effect
• A more geographically comprehensive air
quality monitoring network, particularly for
fine particles and hazardous air pollutants
• Development of integrated air quality
modeling tools based on an open, consistent
model architecture
• Increased basic and targeted research on the
health effects of air pollution, especially
particulate matter
• Continued efforts to assess the cancer and
noncancer health effects of air toxics
exposure
• Development of tools and data to assess the
significance of wetland, aquatic, and
terrestrial ecosystem changes associated with
air pollution
• Continued development of economic
valuation methods and data, particularly
valuation of changes in risks of premature
mortality associated with air pollution
We discuss each of these research needs in greater
detail in the sections that follow.
Emissions Modeling
Our analysis of emissions suggests several areas of
research that could improve emissions data and
modeling tools. The overall importance of ambient
particulate matter estimates to the results of this
analysis makes improved modeling of particulate
matter and precursor emissions a high priority.
Ambient monitoring of particulate composition, for
example, indicates that particulate matter of crustal
origin (e.g., from agricultural tilling, construction, and
wind erosion) may be over-represented in our
emissions inventories. As we discuss in the report, one
possible explanation for this apparent inconsistency
may be the extent to which these emissions are
transported beyond their point of emission. Some
preliminary evidence suggests that the mobility of a
large fraction of these particles may be relatively
limited, but further research is needed to confirm this
hypothesis.
Comparison of emissions inventories with
monitoring data also suggests that our inventory may
1-1
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
underestimate the organic and elemental carbon
fraction of directly emitted particulates, particularly in
urban areas. In this case, it is more difficult to assess
the potential sources of underestimation. One
hypothesis is that current emissions estimation tools
may underestimate organic particulate emissions from
mobile sources. Continued research into emissions
rates for mobile sources could yield increased accuracy
for particulate emissions. Additional tailpipe
emissions studies may be needed, and emissions
estimation techniques need to be developed to better
reflect the results of those studies. In general,
continued research to better reconcile monitoring data
on the composition of ambient particulate matter on
the one hand, with emissions estimates for primary
and secondary sources of particulate matter on the
other hand, should help in improving our ability to
predict changes in fine PM concentrations.
One other emissions uncertainty that could be
reduced by additional research involves volatile
organic compound (VOC) emissions. Estimates of
VOC emissions tend to be highly variable — in the
summer months especially, they can be closely linked
to variations in temperature. As ambient ozone
modeling becomes more sophisticated, however,
better temporal and spatial resolution of VOC
emissions inventories may be needed to take
advantage of the increased capabilities of air quality
models to process more highly resolved data.
In a broader sense, our current inability to
quantitatively characterize and carry through the
analysis the impact of key uncertainties in emissions
estimation may give the misleading impression that
these uncertainties are less important than other
quantifiable sources of uncertainty. For example, the
statistical simulation modeling analysis we present here
reflects only quantifiable sources of uncertainty in the
concentration-response and economic valuation steps
of the analysis. Uncertainties in emissions estimates,
however, may be among the most important in the
entire analysis. Emissions estimates are a critical first
step in our approach, so errors in this step can
magnify as we work through the subsequent steps of
the analysis. One way to enhance the quantification of
emissions estimation uncertainties in future
assessments, and to reduce any potential errors of
inconsistency with the subsequent air quality modeling
steps, is to develop a tool that both integrates
emissions and air quality analyses and provides a
means to more cost-effectively perform multiple
scenario analyses. The Models-3 development effort,
described below, may provide a modeling platform
that is more amenable to sensitivity testing of
alternative emissions results.
Cost Estimation
The first prospective analysis relies on direct
expenditure estimates to characterize the costs of the
CAAA. As we state in the report, this approach
probably does not represent a large source of error in
our estimate of social costs, though there is some
evidence it may provide conservative estimates. The
direct cost approach does not provide information on
other potential categories of impact that may be of
interest, however, including total employment,
employment by sector, capital accumulation patterns,
and the pace of technological change. Additional
cost-effective tools are needed to better estimate the
secondary impacts of direct cost estimates for broad,
programmatic assessments such as the section 812
series.
One potentially important area where research
may enhance our ability to conduct broader
assessments is development of computable general
equilibrium (CGE) models that can be implemented
in a resource-effective manner. The potential for
introducing additional error when using such a
forecasting tool, however, demands the model be
capable of processing many scenarios of important
economic inputs (e.g., alternative interest rate
scenarios) to better bracket the range of future
outcomes relevant to CAAA implementation.
A well-designed CGE model may also enhance
our capability to estimate the effects of the tax-
interaction effect, both on the cost and the benefit
side. Additional empirical work will also be needed to
confirm that the magnitude of the effect estimated in
1-2
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
the current literature, which is largely confined to the
electric utility sector, is applicable for other economic
sectors where the competitive dynamics and capital-
intensity of production may differ from those in the
electric utility industry.
Air Quality Modeling
Our current limited ability to disaggregate the
overall benefits of the CAAA is largely attributable to
the complexity of the relationships between changes
in precursor emissions and the ambient concentration
outcomes. For example, nitrogen oxides are pre-
cursors of both fine participate matter and ozone, and
their presence in the atmosphere also affects the
conversion of sulfur dioxide to fine particles. In
addition, while low levels of nitrogen oxides can
contribute to elevated ozone concentrations, very high
levels of nitrogen oxides, in the right combination
with VOCs and certain meteorological conditions, can
suppress ozone concentrations. These complex inter-
relationships among pollutants affected by the CAAA,
coupled with the national scope of the analysis
conducted here, demand the use of an air quality
modeling technique that accurately reflects the
complexities of atmospheric chemistry. Estimating
the incremental impact of various combinations of
emissions changes would require the repeated exercise
of the model for each alternative set of emissions
scenarios of interest.
The models we chose for this analysis, while they
represent the current state of the art in modeling
atmospheric chemistry, are difficult and expensive to
run for a wide range of scenarios. To improve our
ability to disaggregate the benefits of the CAAA, we
need a fully integrated air quality modeling and
emissions input system that accounts for the full range
of pollutant interactions and relevant atmospheric
chemistry. The current Models-3 effort holds promise
in this area, but must be adequately funded to achieve
these goals. Pursuit of a fully integrated modeling
system also holds promise for generating more
accurate ambient particulate matter estimates. The
current best modeling systems for this purpose
provide estimates based solely on changes in the
concentrations of sulfate- and nitrate-derived particles,
with limited abilities to assess changes associated with
organic precursors of fine particles. Gaining a good
understanding of organic particle formation may also
be an important goal in better characterizing the full
range of impacts of efforts to control air toxics under
Title III. In addition, a more cost-effective air quality
modeling tool may also enhance our ability to conduct
comparative analyses and explore the sensitivity of air
quality modeling and emissions estimation outcomes
to alternative assumptions and modeling paradigms.
Improvements in exposure analysis might also be
made with additional research into techniques for
extrapolating the results of monitor-based analyses to
unmonitored areas. In particular, we suggest further
exploration and development of methods that base
extrapolation on the causes of ambient air quality (e.g.,
local land use, emissions characterizations,
meteorology, and terrain), rather than the outcomes of
air quality modeling (e.g., simple extrapolation of air
quality concentrations). In the course of developing
this analysis, we began development of such an
approach, termed the "homology mapping"
technique. Continued development of this tool could
improve the accuracy of our estimates in future
analyses for those areas that are distant from
monitors.
Human Health Effects Estimation
The results of our analysis clearly highlight the
importance of the link between premature mortality
and air pollution. The wide range of current research
on this link, including the several short-term and long-
term cohort studies, provides a strong basis for
establishing that particulate matter contributes to
premature mortality among the exposed population.
The existing studies, however, are limited by the
availability and resolution of air quality monitoring
data, data on the characteristics of exposed
populations, and, in the case of the long-term studies,
extensive time-series of these data. The continued
enhancement of our air quality monitoring network,
I-3
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
particularly for fine particles, is critical to the better
understanding of the relationships between fine
particles and human health effects. Developing long-
time series of fine particle data will take time,
however. In the meantime, it is important to continue
efforts to better isolate the separate and joint impacts
of ambient pollutants on premature mortality,
including better resolution of the incremental impacts
of ozone, carbon monoxide, nitrogen oxides, sulfates,
and particles in the ultrafine and fine fraction, as well
as the coarse particles within PM10.
In addition, the sensitivity analyses presented in
Appendix D show that a resolution of competing
alternative hypotheses about the presence and
potential time period of a lag in the incidence of
premature mortality following exposure may be
important. Although in our judgment assuming a
distributed five-year lag period may be warranted,
there is no scientific basis for either the assumption of
a lag or for determining the appropriate time period.
We believe it will continue to be important to evaluate
the existing evidence and develop new studies to
clarify the extent to which the premature mortality
outcomes reflected in the existing epidemiological
literature ought to reflect a lag period between
exposure and the mortality effect.
For premature mortality and for other health
effects, our analysis is based on the premise that the
available literature provides broadly applicable
characterizations of the relationships between
exposure to air pollutants and the incidence of health
effects. We use the results of available studies on a
national basis, although in many cases the underlying
literature may be based on analysis of the
concentration-response relationship in a particular
region. It is possible, however, that region-specific
factors may play a role in the results of these studies.
For example, the composition of air pollutants such as
particulate matter varies by region, and it is possible
that other, perhaps unobservable factors may have a
synergistic or mitigating effect on the incremental
incidence of health effects. The literature on air
pollution's influence on health is not yet broad enough
for us to implement a regional approach to health
effects estimation. As the literature base develops,
however, a regional approach may be an option for
future assessments. In the meantime, it is important
to continue to develop a broader base of regional
estimates of the effects of multiple pollutants on key
health outcomes, including mortality, chronic
bronchitis, and hospital admissions, to better reflect
the impact of potentially important regional
differences in the composition of particulate matter
and other human health stressors. Expanding the
current literature base may also provide a better means
for evaluating the effects of air pollution on sub-
populations of individuals, such as children and the
infirm, that may be of increasing importance in the
Federal government regulatory effort.
Evaluation of the Effects of Air
Toxics
In order to develop a meaningful estimate of the
benefits of air toxics controls for future 812
Prospective analyses, we must address existing
knowledge gaps and other methodological barriers
that prevent more realistic analyses of the benefits of
air toxics control. We have already begun developing
a detailed research plan for improving the assessment
of air toxics in future prospective analyses. For
example, EPA has agreed to sponsor workshops that
bring together experts in toxicology, exposure and risk
assessment, and economics with the goal of
establishing a framework for air toxics benefit
estimation. In establishing such a framework, we will
need to address issues and research needs related to
estimating both air toxics exposure and the hazard
posed by individual air toxics.
Exposure-related research needs include both the
development of a database of air toxics measurements
and the extent to which individuals, on average, would
be exposed to the measured concentrations. To
address the first issue, we plan to explore the potential
for compiling a database of air toxics data from
established state air toxics monitoring networks. We
also plan to explore design options for the "super-
site" monitoring programs that will permit them to be
exploited to better understand exposure to air toxics
1-4
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
linked to key health effects categories, and to improve
the performance of ambient concentration modeling
efforts.
More generally, there is a need to continue to
pursue research aimed at the following goals: (a)
improving methods to estimate current levels and
future changes in acute and chronic ambient exposure
conditions nationwide; (b) evaluating the full
distribution of concentration-response relationships
linking exposure and health outcomes, with the goal
of providing a better estimate of the central tendency
of the relationships to support primary benefits
estimation; and (c) tailoring economic valuation
methods for the broad array of potential serious
health effects such as renal damage, reproductive
effects and fatal and non-fatal cancers, including
accurate characterization of the impact on valuation of
latency periods for these effects, where applicable.
Ecological Effects Estimation
The research needs for future analysis of the
CAAA's ecological benefits can be viewed from two
perspectives. The first is the valuation of additional
first order, acute ecological effects that change the
level of service flows society receives from
ecosystems, and the second is assessing and valuing
the broader changes to the structure and function of
ecosystems. Our analysis reflects the state of the
current data and methods in this area by characterizing
and quantifying ecosystem service flows affected by air
pollutants, but many gaps remain. Pursuing a strategy
of enlarging the array of quantified service flows
would entail further development of economic models
and collection of data. However, notably absent in
this report is quantitative treatment of the changes of
ecosystem structure and function that do not
measurably affect the provision of service flows to
society, such as nutrient cycling, species composition,
and the resistance and resilience of ecosystems to
disturbance. Because many ecological benefits of the
CAAA fall into this category, future research could
adopt a strategy of developing analytic tools to assist
in the valuation of these impacts.
Our current analysis suggests several ways we can
enhance the comprehensiveness of coverage for
potentially important service flows. For example,
while we can develop estimates of the changes in
mercury emissions attributable to CAAA provisions,
and there is an extensive literature on the effects of
mercury in ecological systems, there are great
uncertainties in estimating the fate and transport of
incremental increases in airborne mercury emissions.
The persistence of this element, the potentially long
recovery times for ecological systems contaminated
with mercury, and the potentially global scale of
mercury transport suggest that overcoming this barrier
will be challenging; however, existing tools may
provide a good starting point for bounding analyses.
Similar issues are present in assessing the ecological
effects of air toxics. Some toxics are persistent in
natural systems, can be attributed to multiple airborne
and other sources, and have been accumulating in the
environment for many years. Analyses aimed at
characterizing effects at the watershed level, however,
may be more successful in capturing many of the
complexities of source-receptor relationships and
receptor sensitivities than the national analyses we
have traditionally pursued.
Analyses of nitrogen deposition in the current
report are based on a displaced cost approach. The
uncertainties and potential circularity of this approach
limit its applicability to a subset of the aquatic systems
susceptible to eutrophication. A more widely
applicable and therefore more promising approach for
future analyses will be an avoided damages analysis.
To complete such an approach, however, further
research is needed to better explain the dose-response
relationships between increased nitrogen loads,
thresholds of nitrogen loading that lead to
eutrophication, and the ecological mechanisms that
lead to the loss of service flows such as recreational
and commercial fishing. Some recent analyses attempt
to bridge this gap through the development of
reduced form relationships between nitrogen loads
and service flow disruption, but these types of
approaches also have only limited applicability unless
they can be shown to hold for long periods of time
and across a wider range of marine environments,
climate conditions, and species types.
I-5
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
In some cases acute impacts to ecosystems, such
as the disappearance of game species from particular
ecosystems or foliar damage to trees, attract the
attention of the public and policy makers. Such acute
impacts generally have effects that are observable and
alter the provision of ecosystem services in a
measurable way. Less often noticed are the
ecosystem-level ecological impacts that change
ecosystem structure and function but do not
immediately affect service flows received from that
ecosystem. By focusing on acute impacts it is possible
to lose sight of ecosystem-level changes to structure
and function that could eventually lead to large-scale
impacts far greater in degree and geographic extent
than the contemporaneous, acute effects.
Ecosystems generally maintain multiple
interchangeable elements that may drive a particular
process, as hypothesized by Odum (1985) for forest
ecosystems and Howarth (1991) for aquatic systems.
This allows for natural variation in these elements as
well as long-term cycles in which some elements
dominate over others. Explicit in the definition of
ecological structure and function is the ability of an
ecosystem to adapt to natural changes in its
environment. When pollution affects ecosystem
functions such as nutrient cycling, water filtration,
biological diversity, and provision of habitat, it may
also be precluding the system's ability to adapt and
respond to change and perform these functions in the
future. The ultimate effects of such changes in
ecosystems are sometimes unpredictable in scale and
nature. Ecosystems impacted by mankind may
respond in a discontinuous manner around critical
thresholds that are boundaries between locally stable
equilibria (Common and Perrings 1992; Constanza et
al 1993). Complexity in ecosystems prevents analysts
from using linear methods to "add up" the discrete
ecological effects of pollution.
Additional research is also needed to develop
economic valuation methods that can adequately
characterize the monetized benefits of maintaining
ecosystem structure and function in their current
states. Contingent valuation approaches may prove
valuable, but the scientific basis for evaluating changes
in ecosystems needs to be sufficiently advanced that
analysts can construct plausible scenarios of
alternative ecosystem outcomes for respondents to
react to. To lay the groundwork for these efforts,
there is an immediate need to identify the key
attributes of ecosystems that are most valued by
individuals. The results of those types of scoping
analyses might be useful in targeting subsequent
scientific and ecological research, with the goal of
developing pilot analyses that integrate robust and
realistic characterizations of the changes in ecological
resources attributed to air pollution with careful
economic valuation approaches to assess the value of
avoiding those changes.
The isolation of service flows, while a useful
interim tool for quantifying and monetizing the effects
of air pollutants on ecosystems, may imply an
oversimplified cause and effect relationship between
pollution and the provision of the service flow. As
our analysis suggests, often the service flow is affected
by complex non-linear relationships that govern
ecosystem structure and function. Pursuit of the
short-term goal of enhancing our understanding of
ecological effects on service flows may ultimately
provide new insights into our understanding of these
complex relationships. At the same time, we suggest
that it will continue to be important to pursue
methods to estimate the effects of air pollution on
other ecological indicators of concern, including those
that may not be directly linked to service flows,
recognizing that accurate assessment of changes in
nutrient cycling, water filtration, biological diversity,
provision of habitat, and other valuable aspects of
ecosystems may ultimately demand a broader view of
these effects.
Economic Valuation
The importance of the economic valuation step in
our benefits assessment is highlighted by our analysis
of the influence of key variables on the overall range
of uncertainty from our statistical simluation modeling
analysis. Our analysis clearly shows that uncertainty in
the measurement of the value of statistical life
dominates the quantifiable uncertainty in our overall
1-6
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
estimates. In addition, there remain several important
non-quantified uncertainties in the use of labor market
studies to value avoidance of environmental risks
from air pollution. These uncertainties were also
highlighted in the section 812 retrospective analysis
report, and as a result many are currently being
pursued as part of federally sponsored and
independent research studies.
First, to the extent we must continue to rely on
the value of statistical life (VSL) approach to value
avoided mortality risks, we need more advanced
methods for discerning an appropriate VSL from the
extensive literature on this topic. The application of
meta-analysis techniques may help us better
understand the impact of important methodological
and measurement choices that are made in conducting
these studies, and provide a basis for narrowing the
range of appropriate VSL estimates for environmental
risk estimates. Second, we need to develop a better
means for adjusting VSL estimates to address the key
benefits transfer issues from the risk scenarios
presented in existing studies to the specific mortality
risk presented by particulate matter air pollution.
Pursuit of this goal would include developing better
adjustment techniques for differences in age, health
status, and the risk context, including such attributes
of risk perception as dread and the involuntary nature
of environmental exposures.
Third, and perhaps most importantly, we need to
develop a better literature basis for directly valuing the
commodity provided by air pollution reductions, that
is, reductions in the risk of dying prematurely or, put
another way, changes in individual's survival
probabilities. Several research efforts are currently
underway that are attempting to directly value life
extensions similar in magnitude to those provided by
air quality improvements. These research efforts
necessarily rely on stated preference methods, which
in most cases are considered less reliable than the
revealed preference estimates used as the basis for
VSL approaches. Substantial analytic challenges
remain in making the risk reduction scenarios
presented to respondents clear and understandable to
the lay person, but the results of this new work will
need to be carefully considered for their implications
for a new paradigm for valuation of mortality risk
reduction.
Beyond the valuation of avoided premature
mortality, there are several other areas of research
that, if pursued, can enhance our ability to value the
health outcomes of reductions in air pollutants. For
example, we must develop a broader research base for
valuation of avoided effects to children, including
construction of an overall framework for considering
the welfare and utility of children within the broader
family context, to better characterize the effects of air
pollutants to this important sub-population. It may
also be fruitful to pursue the potential cost-effective
advantages of developing a more flexible means of
valuing health effects through health-state utility
approaches. More research is needed to assess the
trade-offs in accuracy and precision of these results
with the advantages of a broader set of WTP
estimates to apply to relevant endpoints (for example,
to value the avoidance of hospital admissions).
Additional research is also needed to enhance our
ability to value important welfare effects. Visibility
continues to be one of the most important welfare
endpoints for analyses of particulate air pollutants, but
it would be useful if the literature base were
periodically updated to better reflect the current state-
of-the-art in stated preference technique. For example,
an important research direction would be to pursue
development of additional estimates for residential
visibility valuation to corroborate those currently
available, and develop insights into the potential for
double-counting in application of the location-specific
residential and recreational visibility valuation
estimates. The literature on materials damage
valuation is also in need of updating. We chose not to
include an estimate for household soiling effects in
our primary benefits estimates because of the age of
the original research, its reliance on an older measure
of particulate air pollution (total suspended
participates), and its reliance on outdated household
expenditure data. Updating existing estimates of the
effects of air pollutants on household soiling
expenditures would be a relatively straightforward
research project. Agricultural analyses could also
benefit from a broader assessment of the crops
1-7
-------
The Benefits and Costs of the Clean Air Act, 1990 to 2010
potentially affected by ozone and other pollutants, and
by the joint analysis of not only the damaging effects
of some pollutants but also the yield-enhancing effects
of others (e.g., nitrogen deposition).
The results of this first prospective analysis
continue to suggest that our nation's investment in
clean air has been a wise one. At the same time, we
recognize that we should continue to assess the
progress of the clean air program, as implemented
under the Clean Air Act, to ensure that benefits are
achieved in the most cost-effective means possible.
Pursuit of the research goals outlined above will
continue to enhance our ability to provide accurate
and timely assessments of the costs and benefits of all
provisions of the Clean Air Act.
I-8
-------
of the 19, 1999
by on
on
of of
of
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
November 19, 1999
EPA-SAB-COUNCIL-ADV-00-003
OFFICE OF THE ADMINISTRATOR
SCIENCE ADVISORY BOARD
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
RE: Final Advisory by the Advisory Council on Clean Air Compliance Analysis on
the 1999 Prospective Study of Costs and Benefits (1999) of Implementation of the
Clean Air Act Amendments (CAAA)
Dear Ms. Browner:
On October 1, 1999, the Advisory Council on Clean Air Compliance Analysis (Council)
held a public teleconference to review a draft Agency document, The Benefits and Costs of the
Clean Air Act, 1990 to 2010; EPA Report to Congress (U.S. EPA, Office of Air and Radiation
and Office of Policy, September 1999) and held a follow-up teleconference on October 15, 1999
to review an October draft of that same document. These two closure meetings represented the
culmination of a multi-year series of review meetings during which the Council provided advice
to the Agency on the study design, methodologies, and intermediate results. The Council
submits this Advisory to complete its review responsibilities as defined in Section 812 of the
CAAA.1
The Council believes that The Benefits and Costs of the Clean Air Act, 1990 to 2010 is a
serious, careful study that, in general, employs sound methods and data. While we do not
endorse all details of the study, we believe that the study's conclusions are generally consistent
with the weight of available evidence. The Council also appreciates the Agency's
responsiveness over the many years of this study's development to advice conveyed by the
Council and its technical subcommittees. While the Project Team has not followed our advice in
every instance, we believe that they have done a remarkable job on an extremely difficult
project.
1 Specifically, subsection (g) of CAA §312 (as amended by §812 of the amendments) states:
"(g) The Council shall -- (1) review the data to be used for any analysis required under this section
and make recommendations to the Administrator on the use of such data, (2) review the methodology
used to analyze such data and make recommendations to the Administrator on the use of such
methodology; and (3) prior to issuance of a report required under subsection (d) or (e), review the
findings of such report, and make recommendations to the Administrator concerning the validity and
utility of such findings."
Recycled/Recyclable
Printed with Soy/Canola Ink on paper trial
contains at toast 50% recycled fiber
-------
We would, however, like to bring to your attention two major issues that arose in our
review of the study. These pertain to the study's measurement of costs and representation of
uncertainty regarding costs. Following our discussion of these points, we present suggestions to
improve future Prospective Studies. Because of their importance, we would like to highlight
these suggestions here:
a) We believe that benefits and costs must be disaggregated by individual provision
of the Clean Air Act if benefit-cost analysis is to be useful in informing
regulation.
b) Future studies must attempt to quantify uncertainties about regulatory costs, as
well as uncertainties about the benefits of regulations. Failure to quantify cost
uncertainties may give the impression that costs cannot exceed point estimates.
c) Cost estimates should include tax-interaction effects; i.e., they should reflect the
fact that environmental regulations may exacerbate the disincentive effects of the
personal and corporate income taxes. This may raise cost estimates considerably.
d) The Agency should revise its estimates of the Value of a Statistical Life.
e) The impact of air quality regulations should be stated in terms of a Net Cost per
Life Saved and a Net Cost per Life-year Saved to facilitate comparisons with
other health and safety regulations.
f) Attempts should be made to increase the set of ecosystem benefits valued and to
improve estimates of the exposure and effects of air toxics.
1. Comments on the Drafts Provided for Council Review
a) The Relationship between Direct and Social Costs of Compliance. Social cost is
the type of cost that is most relevant to the evaluation of the 1990 Clean Air Act.
However, the draft Prospective Study relies primarily on estimates of the "direct"
compliance costs for affected industries or pollution sources. The reliance on
direct costs is understandable, since it is more difficult to assess the social costs.
At the same time, it is important to articulate clearly and without bias the
relationship between direct and social costs.
The Council believes that the study's discussion of this issue lacks balance and is
prone to misinterpretation. The study describes in detail two factors that might
cause direct costs to overstate true social costs (absence of attention to producer
and consumer responses, and the assumption of a static technology). The October
review draft contained a discussion of the tax-interaction effect, which can cause
direct costs to understate social costs, possibly by a very large amount. In the
Council's view, this effect merited discussion in the text and should not be
relegated to a footnote. There is now a substantial body of published theoretical
-------
and empirical research that indicates that, under typical conditions,
tax-interactions can cause social costs to exceed direct costs by at least 25
percent, and in some cases by 100 percent or more. Table 3-3 further contributes
to potential misinterpretation. It explicitly mentions a factor that would cause
direct cost to overstate social cost (lack of attention to producer and consumer
responses) but fails to mention explicitly the factor (tax interactions) that works in
the opposite direction. By minimizing attention to the tax-interaction effect, the
study gives readers the erroneous impression that the EPA's use of direct costs is
likely to overstate social costs.
Tax interactions occur when environmental regulations exacerbate the distortions
in labor and capital markets caused by prior income, profit, or sales taxes. These
interactions may result from any regulations that raise production costs and
thereby lower the real purchasing power stemming from given real wages. Even
"small" regulations can produce significant tax-interaction effects. The
Prospective Study fails to indicate the general relevance of these effects. The
study states that general equilibrium effects are important where the regulatory
action is known to have an impact on many sectors of the US economy. Although
this statement is technically correct, it allows for the impression that such general
equilibrium effects are unusual. It fails to point out the key finding from the
tax-interaction literature: namely, that all regulatory actions have impacts on other
sectors (particularly labor and capital markets) and that these general equilibrium
impacts, under typical conditions, raise social costs substantially relative to direct
costs.
In sum, the Council would urge the EPA to give more attention to the tax-
interaction effect in order to achieve a more balanced and straightforward
presentation of the relationship between direct costs and overall social cost. This
is necessary to avoid giving the false impression that direct costs are likely to
overstate social cost.
b) Characterization of Uncertainty with Respect to Cost Estimates. The main results
of the first Prospective Study are summarized in a table of costs and benefits that
appears both in the Executive Summary and in Chapter 8. Uncertainties about the
benefits of the CAAA are nicely illustrated by a lower bound and an upper bound
(90% confidence interval). In contrast, the cost of this environmental protection
is represented only by a central estimate. (Cost uncertainties are discussed via
sensitivity analyses in other tables, but these uncertainties are not combined into
an overall set of bounds on the central cost estimate.)
Thus the benefit-cost ratios in that main summary table vary only with
uncertainties about benefits. These ratios would vary even more if they
incorporated some uncertainty about costs. Since costs are indeed uncertain, the
table implicitly understates the true degree of uncertainty about the net benefits of
the CAAA.
3
-------
Even rough representations of uncertainty about these costs would be better than
the current implication that costs are certain. One possibility would be to assume
a uniform distribution about each element of cost, ranging from 50% to 150% of
the central estimate. A second possibility is to show an additional row of
benefit-cost ratios where the costs have been multiplied by 1.3 to account for the
tax-interaction effect. A third possibility is suggested by reference to the fact that
the Retrospective Study produced a central estimate of direct cost equal to $523
billion, while the modeling approach provided welfare effects between $493
billion and $621 billion. Since these bounds are 6% below and 19% above the
central estimate, the same percentage bounds could be applied to the central
estimate of costs in the Prospective Study. True bounds on costs in the
Prospective Study would be preferred, but one of these rough estimates of bounds
is better than using no bounds at all.
2. Suggestions to Improve Future Prospective Studies
a) Disaggregate Benefits and Costs by Title or Provision. The Council reiterates its
strong recommendation for presenting the benefits as well as the costs of the
CAAA by title and, preferably, by provision, in future studies. Without this level
of disaggregation, the study cannot be used directly to identify how the CAAA
might be improved in the future. The Council recognizes that a thorough
disaggregation analysis was not feasible for the current study since resources
were not available for exercising several air quality models to create the needed
data base for the analysis. Future studies should not be limited in this regard
since more universal and versatile platforms for air quality modeling, such as
Models-3, are expected to be available. With careful design, using such a system,
a small number of additional comprehensive modeling simulations can provide
the information needed for a thorough bottom-up assessment of the CAAA
benefits by individual title and even by some provisions. If, in the design phase
of the next prospective study, it becomes apparent that resources cannot be
allocated for these analyses, then an alternative design strategy combining use of
top-down or screening model approaches combined with carefully selected
essential comprehensive model simulations should be pursued.
b) Characterize Uncertainty about Costs. The costs imposed by air pollution
regulations are highly uncertain. For example, the costs of sulfur dioxide
abatement under the 1990 Clean Air Act have turned out to be a fraction of what
was estimated in 1990. Unfortunately, uncertainty can lead to higher as well as
lower costs.
EPA has relied on engineering estimates of abatement costs. Even if these
estimates were accurate estimates of the cost of equipment and operating costs,
they would understate social costs because of tax-interaction and other effects.
EPA needs to discuss and to quantify the following sources of uncertainty:
-------
(1) Uncertainty in the engineering cost estimates.
(2) Costs in addition to the engineering estimates, such as tax-interactions.
(3) Technical change due to the technology forcing that lowers costs.
(4) Changes in the wage rate or prices of materials due to the changes in
demand.
c) Include Tax-Interaction Effects in Future Cost Estimates. One of the most
important insights to emerge in Environmental Economics in the past 25 years is
that regulations, by exacerbating existing distortions in the economy, can have
social costs considerably in excess of direct compliance costs. An environmental
regulation that raises the price of purchased goods and lowers the real wage will
tend to, other things equal, cause a substitution of leisure for labor. This
compounds the deadweight loss of the tax system, which, by driving a wedge
between the gross and net of tax wages, causes individuals to substitute leisure for
labor. This tax-interaction effect can, in some cases, double the costs of a
regulation (Goulder et al. 1999, Parry et al. 1999).2
It is important for tax-interaction effects to be included in future Prospective
Studies for two reasons. First, these costs are real. They represent real losses in
output, and they occur even for small regulations. Second, the tax-interaction
effect can at least to some degree be offset if the environmental program raises
revenues, which are used to reduce the rates of other, distorting taxes. This
implies that the costs of a program will depend on how a standard is achieved,
which has implications for the choice of regulatory approach. For example, a
permit market will have lower social costs if permits are auctioned and revenues
recycled than if permits are given away (Goulder et al. 1997; Parry 1997).3
d) Revise Mortality Risk Estimates. The Council is uncomfortable with the
Agency's use of $4.8 million (1990 U.S. dollars) for the Value of a Statistical Life
(VSL) and $293,000 for the Value of a Statistical Life-year (VSLY) to value
mortality risk reductions from reduced air pollution. We question the
2 Goulder, Lawrence H., Ian W. H. Parry, Roberton C. Williams III, and Dallas Burtraw, 1999, "The Cost-
Effectiveness of Alternative Instruments for Environmental Protection in a Second-Best Setting," Journal of Public
Economics 72(3):329-360; and Parry, Ian, W. H., R. C. Williams III, and L. H. Goulder, 1999. "When Can Carbon
Abatement Policies Increase Welfare? The Fundamental Role of Distorted Factor Markets," Journal of
Environmental Economics and Management 37:52-84.
3 Goulder, Lawrence H., Parry, Ian W. H., and Dallas Burtraw, "Revenue-Raising vs. Other Approaches to
Environmental Protection: The Critical Significance of Pre-Existing Tax Distortions," RAND Journal of
Economics, Winter 1997; and Parry, Ian W. H., "Environmental Taxes and Quotas in the Presence of
Distorting Taxes in Factor Markets," Resource and Energy Economics, Winter 1997,19:203-20.
-------
appropriateness of the $4.8 million VSL even as a measure of prime-aged
individuals' willingness to pay (WTP) for risk reductions, and we question the
application of a WTP estimate for prime-aged individuals to a population of older
individuals and people who are in poor health. Time limitations did not permit a
thorough treatment of this issue prior to completing the first Prospective Study;
hence we recommended that the Project Team use the $4.8 million figure. For
future studies, however, we recommend that the Agency revisit the literature on
the value of mortality risk reductions. The following points should be kept in
mind when examining this literature:
(1) It is WTP for risk reductions that is the appropriate concept when valuing
the mortality benefits of environmental regulations. The costs of
environmental regulations are spread broadly over many individuals who,
indeed, are paying for the resulting risk reductions.
(2) Labor market studies measure willingness to accept (WTA) compensation
for increased risk of death. This is likely to exceed what people will pay
(WTP) for the same risk reductions.
(3) Averting behavior and consumer product safety studies, which are omitted
from the current list of 26 studies, do measure WTP. These studies should
be considered in the review.
(4) In reviewing studies the population whose preferences are measured
should be noted, as should the magnitude of the risk reduction valued.
Studies should be identified that measure WTP for risk reductions among
the populations that benefit from air quality regulation, especially older
people, and that value risk reductions of the same magnitude as those in
future Prospective Studies.
(5) There should be well-defined criteria for selecting studies, which are
clearly stated and consistently applied. For example, compensating wage
studies should adequately control for inter-industry wage differentials;
contingent valuation studies should test for sensitivity to scope.
e) Present Cost-Effectiveness Results. Improvements in human health remain a
major motivation behind air quality regulation and account for over 90% of the
quantified benefits from Titles I-IV of the 1990 CAAA. Reductions in premature
mortality, in turn, account for over 90% of these health benefits. Because
mortality risk reductions are such a large component of the benefits from air
quality regulation, the Council urges the Agency to express the outcomes of the
CAAA in terms of: (1) Net Cost per Life Saved, and (2) Net Cost per Life-Year
Saved. These are calculated by subtracting the value of non-mortality benefits
from costs and dividing the result by: (1) the number of statistical lives saved, and
(2) the number of statistical life-years saved.
-------
By taking this approach the Agency would: (1) provide a measure of program
effectiveness that avoids the use of flawed measures of VSL and VSLY and, more
generally, avoids the controversies surrounding the valuation of mortality risk
change; (2) be in line with standard practice in the public health community,
where different programs are routinely compared using cost-effectiveness
analysis; and (3) facilitate comparisons of the cost effectiveness of various health
and safety programs with health-based environmental regulations. The Council
feels such comparisons are necessary for improving public decisions about the
allocation of society's scarce resources among competing ends.
f) Increase Set of Ecosystem Benefits Valued. The current Prospective Study has
made important advances in identifying ecosystem services that can be linked to
air pollution, and in trying to value these endpoints. For the purposes of
valuation, it is convenient to categorize the impact of pollution on ecosystems as
follows: (1) impacts that occur through markets (e.g., impacts of pollution on
commercial timber stands or fish populations); (2) impacts that affect recreation
(e.g., damage to National Parks from air pollution or to recreational fishing from
acid rain); (3) impacts on ecosystems for which people have well-defined non-use
values (e.g., damage to forest canopy, the value of reduced fish populations to
non-anglers); and (4) other impacts on ecosystem functions and services, not
otherwise classified, for example water and nutrient recycling, maintenance of
biodiversity, climate stabilization. These indirect and more subtle effects may not
be well understood or even perceived by people; yet they may have important
impacts on human well being.
Techniques for valuing the first 3 categories of benefits are well-established, but
the set of applied studies is sparse. The Agency might consider funding new
studies, after determining which categories of benefits are likely to have the
largest impact on regulatory decisions. When commissioning studies to measure
non-use values, care should be taken in defining: (1) the geographical scope of
what is to be valued (for example, are people asked only for non-use values in .
their state?); (2) the nature of substitutes (i.e., conditions at other locations); and
(3) how many endpoints to value at the same time. For example, in regard to this
last point, if a regulation to reduce nitrogen oxides (NOx) affects forests through
ozone and fish population through acid rain, people should be asked to value the
entire package of ecosystem benefits brought about by NOx reduction. Adding
up WTP values from individual studies might overstate the value of the NOx
reduction program if there are important substitution effects across ecosystem
services.
A problem for policy analysis is that the endpoints that affect markets or for
which people have well-defined use (recreation) and non-use values (e.g.,
damages to forests, fish and wildlife populations) do not capture the totality of
ecosystem damages associated with pollution control decisions. In particular,
they do not capture ecosystem functions and services such as nutrient recycling
7
-------
and habitat provision. Nor do they capture the more subtle changes in ecosystem
functioning that may to lead to non-marginal changes in ecosystem performance.
Before these changes can be valued, however, it is essential that ecologists
characterize the ecosystem outcomes (or indicators) that are important to
ecosystem functioning and then relate these outcomes (or indicators) to particular
activities or pollutants. This information is an essential foundation for economic
valuation.
g) Estimate Exposure and Effects of Air Toxics. The Retrospective Study and the
first Prospective Study do not contain any quantitative benefit-cost analyses of
Toxic Air Pollutants (TAP). As the Council's Health and Ecological Effects
Subcommittee (HEES) has stated,4 the Agency does not currently have analytical
methodologies available to establish population exposure estimates, or to define
realistic risk estimates for the general population. The HEES, with approval by
the full Council, suggested an approach to identifying the research and
methodological developments needed to overcome these deficiencies. The effort
requires coordination with the SAB Executive Committee, various SAB
Committees, the Office of Research and Development and Office of Air Quality
Planning and Standards. Implementation of the plan of action outlined for the
Agency will begin a process that can lead to quantitative estimates of health, and
possibly ecological benefits for the next Prospective Study.
3. Conclusion
The purpose of conducting benefit-cost analyses is to improve the efficiency of
regulation. The suggestions we have made in this Advisory are designed to help achieve this
goal. Increasing the accuracy of benefit-cost analyses will entail measuring certain categories of
benefits (e.g., certain ecological benefits, benefits of reduced exposure to hazardous air
pollutants) and costs (tax-interaction effects) not included in the current Prospective Study. It
will also entail refining estimates of the value of mortality benefits, which continue to dominate
the monetized benefits of improved air quality. Of all the suggestions made above, however, we
believe that disaggregating the benefits and costs of individual provisions of the CAA is,
perhaps, the most important. If our recommendation to provide more disaggregated benefit-cost
estimates can be implemented, the specific programs which have the highest potential payoff to
society can be more readily identified. We strongly encourage the Agency to make the research
investment and analytical commitments required to ensure this objective is met in the next
prospective study.
4
See HEES Letter Advisories, "The Clean Air Act Amendments (CAAA) Section 812 Prospective Study of Costs and Benefits
(1999): Advisory by the Health and Ecological Effects Subcommittee on Initial Assessments of Health and Ecological Effects, Part 1", EPA-
SAB-COUNCIL-ADV-99-012 and "Part 2", EPA-SAB-COUNCIL-ADV-00-001.
-------
We thank the Agency for the opportunity to review the first Prospective Study and to
make recommendations to improve the methods and data to be used in future prospective
studies. We look forward to your response to this Advisory.
Sincerely,
?7 ~X /""
/M^^^^r^n. ^L^^^-^
Dr. Maureen L. Cropper, Cnpr^
Advisory Council on Clean Air Compliance Analysis
Science Advisory Board
-------
NOTICE
This report has been written as part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the
Administrator and other officials of the Environmental Protection Agency. The Board is
structured to provide balanced, expert assessment of scientific matters related to problems facing
the Agency. This report has not been reviewed for approval by the Agency and, hence, the
contents of this report do not necessarily represent the views and policies of the Environmental
Protection Agency, nor of other agencies in the Executive Branch of the Federal government, nor
does mention of trade names or commercial products constitute a recommendation for use.
Distribution and Availability: This Science Advisory Board report is provided to the EPA
Administrator, senior Agency management, appropriate program staff, interested members of the
public, and is posted on the SAB website (www.epa.gov/sab). Information on its availability is
also provided in the SAB's monthly newsletter (Happenings at the Science Advisory Board).
Additional copies and further information are available from the SAB Staff.
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD (SAB)
ADVISORY COUNCIL ON CLEAN AIR COMPLIANCE ANALYSIS
CHAIR
Dr. Maureen L. Cropper, The World Bank, Washington, DC
MEMBERS
Dr. Gardner M. Brown, University of Washington, Seattle, WA
Dr. Trudy Ann Cameron, University of California, Los Angeles, CA,
Dr, Don Fullerton, University of Texas, Austin, TX
Dr. Lawrence H. Goulder, Stanford University, Stanford, CA
Dr. Jane V. Hall, California State University, Fullerton, CA
Dr. Charles Kolstad, University of California, Santa Barbara, CA
Dr. Paul Lioy, Robert Wood Johnson School of Medicine, Piscataway, NJ
Dr. Paulette Middleton, RAND Center for Science & Policy, Boulder, CO
CONSULTANTS
Dr. A. Myrick Freeman, Bowdoin College, ME
Dr. Alan J. Krupnick, Resources for the Future, Washington, DC
SCIENCE ADVISORY BOARD STAFF
Dr. Angela Nugent, Designated Federal Officer, Science Advisory Board, U.S. Environmental
Protection Agency, Washington, DC
Mrs. Diana L. Pozun, Management Assistant, Science Advisory Board, U.S. Environmental
Protection Agency, Washington, DC
------- |