The Benefits and Costs of the
Clean Air Act from 1990 to 2020
Final Report - Rev. A
U.S. Environmental Protection Agency
Office of Air and Radiation
April 2011
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
ABSTRACT
Section 812 of the 1990 Clean Air Act Amendments requires the U.S. Environmental
Protection Agency to develop periodic reports that estimate the benefits and costs of the
Clean Air Act. The main goal of these reports is to provide Congress and the public with
comprehensive, up-to-date, peer-reviewed information on the Clean Air Act's social
benefits and costs, including improvements in human health, welfare, and ecological
resources, as well as the impact of the Act's provisions on the US economy. This report
is the third in the Section 812 series, and is the result of EPA's Second Prospective
analysis of the 1990 Amendments.
The Clean Air Act Amendments (CAAA) of 1990 augmented the significant progress made
in improving the nation's air quality through the original Clean Air Act of 1970 and its
1977 amendments. The amendments built off the existing structure of the original Clean
Air Act, but went beyond those requirements to tighten and clarify implementation goals
and timing, increase the stringency of some federal requirements, revamp the hazardous
air pollutant regulatory program, refine and streamline permitting requirements, and
introduce new programs for the control of acid rain and stratospheric ozone depleters.
The main purpose of this report is to document the costs and benefits of the 1990 CAAA
provisions incremental to those costs and benefits achieved from implementing the
original 1970 Clean Air Act and the 1977 amendments.
The analysis estimates the costs and benefits of reducing emissions of air pollutants by
comparing a "with-CAAA" scenario that reflects expected or likely future measures
implemented under the CAAA with a ""without-CAAA" scenario that freezes the scope
and stringency of emissions controls at the levels that existed prior to implementing the
CAAA. There are six basic steps undertaken to complete this analysis: 1. air pollutant
emissions modeling; 2. compliance cost estimation; 3. ambient air quality modeling; 4.
health and environmental effects estimation; 5. economic valuation of these effects; and
6. results aggregation and uncertainty characterization.
The results of our analysis, summarized in the table below, make it abundantly clear that
the benefits of the CAAA exceed its costs by a wide margin, making the CAAA a very
good investment for the nation. We estimate that the annual dollar value of benefits of air
quality improvements will be very large, and will grow over time as emissions control
programs take their full effect, reaching a level of approximately $2.0 trillion in 2020.
These benefits will be achieved as a result of CAAA-related programs and regulatory
compliance actions estimated to cost approximately $65 billion in 2020. Most of these
benefits (about 85 percent) are attributable to reductions in premature mortality
associated with reductions in ambient particulate matter; as a result, we estimate that
cleaner air will, by 2020, prevent 230,000 cases of premature mortality in that year. The
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
remaining benefits are roughly equally divided among three categories of human health
and environmental improvement: preventing premature mortality associated with ozone
exposure; preventing morbidity, including acute myocardial infarctions and chronic
bronchitis; and improving the quality of ecological resources and other aspects of the
environment, the largest component of which is improved visibility.
The very wide margin between estimated benefits and costs, and the results of our
uncertainty analysis, suggest that it is extremely unlikely that the monetized benefits of
the CAAA over the 1990 to 2020 period reasonably could be less than its costs, under any
alternative set of assumptions we can conceive. Our central benefits estimate exceeds
costs by a factor of more than 30 to one, and the high benefits estimate exceeds costs by
90 times. Even the low benefits estimate exceeds costs by about three to one.
ESTIMATED MONETIZED BENEFITS AND COSTS OF THE 1990 CLEAN AIR ACT AMENDMENTS
ANNUAL ESTIMATES
2000
2010
2020
PRESENT VALUE
ESTIMATE
1990-2020
Monetized Direct Compliance Costs (millions 2006$):
Central a
$20,000
$53,000
$65,000
$380,000
Monetized Direct Benefits (millions 2006$):
Lowb
Central
Highb
$90,000
$770,000
$2,300,000
$160,000
$1,300,000
$3,800,000
$250,000
$2,000,000
$5,700,000
$1,400,000
$12,000,000
$35,000,000
Net Benefits - Benefits minus Costs (millions 2006$):
Low
Central
High
$70,000
$750,000
$2,300,000
$110,000
$1 ,200,000
$3,700,000
$190,000
$1 ,900,000
$5,600,000
$1,000,000
$12,000,000
$35,000,000
Benefit/Cost Ratio:
Lowc
Central
Highc
5/1
39/1
115/1
3/1
25/1
72/1
4/1
31/1
88/1
4/1
32/1
92/1
Compliance Costs per Premature Mortality Avoided (2006$):
Central
$180,000
$330,000
$280,000
Not estimated
a The cost estimates for this analysis are based on assumptions about future changes in factors
such as consumption patterns, input costs, and technological innovation, which introduce
significant uncertainty. The degree of uncertainty associated with many of the key factors,
however, cannot be reliably quantified. Thus, we are unable to present specific low and high
cost estimates.
b Low and high benefits estimates correspond to 5th and 95th percentile results from statistical
uncertainty analysis, incorporating uncertainties in physical effects and valuation steps of
benefits analysis.
c The low benefit/cost ratio reflects the ratio of the low benefits estimate to the central cost
estimate, while the high ratio reflects the ratio of the high benefits estimate to the central
costs estimate.
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE OF CONTENTS
ACKNOWLEDGEMENTS
CHAPTER 1 - INTRODUCTION
Background and Purpose 1-1
Relationship of this Report to Other Analyses 1-2
Analytical Design and Review 1-5
Review Process 1-14
Report Organization 1-14
CHAPTER 2 - EMISSIONS
Overview of Approach 2-3
Emissions Estimation Results 2-9
Comparison of Emissions Estimates with the First Prospective Analysis 2-14
Uncertainty in Emissions Estimates 2-16
CHAPTER 3 - DIRECT COSTS
Overview of Approach 3-2
Direct Compliance Cost Results 3-7
Comparison of Cost Estimates with the First Prospective Analysis 3-9
Uncertainty in Direct Cost Estimates 3-11
CHAPTER 4 - AIR QUALITY BENEFITS
Overview of Approach 4-1
Air Quality Modeling Tools Deployed 4-3
Air Quality Results 4-13
Uncertainty in Air Quality Estimates 4-22
CHAPTER 5 - ESTIMATION OF HUMAN HEALTH EFFECTS AND ECONOMIC
BENEFITS
Overview of Approach 5-2
Health Effects Modeling Results 5-24
Avoided Health Effects of Air Toxics 5-28
Comparison of Health Effects Modeling with First Prospective Analysis 5-34
Uncertainty in Health Benefits Estimates 5-36
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 6 - ECOLOGICAL AND OTHER WELFARE BENEFITS
Overview of Approach 6-1
Qualitative Characterization of Effects 6-3
Distribution of Air Pollutants in Sensitive Ecosystems of the United States 6-11
Quantified Results: National Estimates 6-17
Uncertainty in Ecological and Other Welfare Benefits 6-42
CHAPTER 7 - COMPARISON OF BENEFITS AND COSTS
Aggregating Benefit Estimates 7-1
Annual Benefits Estimates 7-3
Aggregate Monetized Benefits 7-6
Comparison of Benefits and Costs 7-7
Overview of Uncertainty Analyses 7-10
Quantifying Model, Parameter, and Scenario Uncertainty 7-13
Lessons Learned and New Research Directions 7-16
CHAPTER 8 - COMPUTABLE GENERAL EQUILIBRIUM ANALYSIS
EMPAX-CGE 8-2
Development of Model Inputs 8-9
EMPAX-CGE Model Results 8-17
Analytic Limitations 8-23
REFERENCES
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
LIST OF ACRONYMS
ACS American Cancer Society
AEO Annual Energy Outlook (from the US Department of Energy)
AERMOD American Meteorological Society/Regulatory Model
AIM Architectural and Industrial Maintenance
AMI Acute myocardial infarction
APEEP Air Pollution Emissions Experiments and Policy analysis model
AQMS Air Quality Modeling Subcommittee (of the Council)
AMET Atmospheric Model Evaluation Tool
ANC Acid Neutralizing Capacity
BenMAP Environmental Benefits Mapping and Analysis Program
CAA Clean Air Act of 1970
CAAA Clean Air Act Amendments of 1990
CAIR Clean Air Interstate Rule
CAMR Clean Air Mercury Rule
CARB California Air Resources Board
CAVR Clean Air Visibility Rule
CDC Centers for Disease Control
CGE Computable General Equilibrium
CMAQ Community Multi-scale Air Quality [System]
CO Carbon monoxide
COI Cost of illness
CONUS Continental United States (domain in CMAQ model)
Council Advisory Council on Clean Air Compliance Analysis
C-R Concentration-Response
CTG Control Techniques Guideline
CV Contingent valuation
DDT Dichlorodiphenyl-trichloroethane
DOE United States Department of Energy
EC Elemental carbon
EE Expert elicitation
EES Ecological Effects Subcommittee (of the Council)
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
EGU Electric Generating Unit
EMPAX-CGE Economic Model for Policy Analysis - Computable General Equilibrium
EPA United States Environmental Protection Agency
EUS Eastern United States (domain in CMAQ model)
EV [Hicksian] equivalent variation
eVNA Enhanced Voronoi Neighbor Averaging
FACA Federal Advisory Committee Act
FASOM Forest and Agriculture Sector Optimization Model
FRM Federal Reference Method
GDP Gross Domestic Product
GHG Greenhouse gas
F£AP Hazardous Air Pollutant
HAPEM6 Hazardous Air Pollution Exposure Model, Version 6
HDDV Heavy-Duty Diesel Vehicle
HES Health Effects Subcommittee (of the Council)
I&M Inspection and maintenance
IC/BC Initial and boundary conditions
IMPROVE Interagency Monitoring of Protected Visual Environments
IPM Integrated Planning Model
LEV Low-Emission Vehicle
LML Lowest measured level
MACT Maximum Available Control Technology
MAGIC Model of Acidification of Groundwater in Catchments
MATS Modeled Attainment Test Software
MCIP Meteorology-Chemistry Interface Processor
MM5 Fifth Generation Mesoscale Model
MSA Metropolitan statistical area
NAA Non-Attainment Area
NAAQS National Ambient Air Quality Standards
NAICS North American Industry Classification System
NAPAP National Acid Precipitation Assessment Program
IV
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
NEI
NEMS
NESHAP
NH3
NH4
NMMAPS
NO3
NOX
NPV
NSPS
O&M
OC
OTC
Pb
PCB
PM
PM25
PM10
PPB
PRB
PSU/NCAR
RACT
RADM/RPM
REMSAD
RfC
RFP
RIA
RSM
RUM
SAB
SANDWICH
National Emissions Inventory
National Energy Modeling System
National Emission Standard for Hazardous Air Pollutants
Ammonia
Ammonium
National Morbidity, Mortality, and Air Pollution Study
Nitrate
Nitrogen oxides
Net present value
New Source Performance Standard
Operation and maintenance
Organic carbon
Ozone Transport Commission
Lead
Polychlorinated biphenyl
Particulate matter
Particulate matter with an aerodynamic diameter less than 2.5 microns
Particulate matter with an aerodynamic diameter less than 10 microns
Parts per billion
Powder River Basin
Pennsylvania State University/National Center for Atmospheric Research
Reasonably Available Control Technology
Regional Acid Deposition Model/Regional Particulate Model
Regulatory Modeling System for Aerosols and Acid Deposition
Reference concentration
Rate of Further Progress
Regulatory Impact Analysis
Response Surface Model
Random Utility Model
Science Advisory Board
Sulfates, Adjusted Nitrates, Derived Water, Inferred Carbonaceous mass,
and estimated aerosol acidity (H+)) process
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
SCAQMD
SIP
SMAT
SMOKE
SO2
SOX
SOA
STN
SUV
TAG
TSP
UVb or UVB
VMT
VNA
voc
VSL
WTAC
WTP
WUS
Ogm"3 or ®g/m;
South Coast Air Quality Management District
State Implementation Plan
Speciated Modeled Attainment Test
Sparse-Matrix Operator Kernel Emissions
Sulfur dioxide
Sulfur oxides
Secondary organic aerosol
Speciation Trends Network
Sport-Utility Vehicle
Total Annualized Cost
Total Suspended Particulates
Ultraviolet B radiation
Vehicle miles traveled
Voronoi Neighbor Averaging
Volatile organic compound
Value of statistical life
Willingness-to-accept-compensation
Willingness-to-pay
Western United States (domain in CMAQ model)
Micrograms per cubic meter (unit for PM2 5 measurement)
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
ACKNOWLEDGEMENTS
The Project Team for the Second Prospective Study was comprised of EPA staff, and
staff from a number of organizations working under contract to EPA. The project
manager was Jim DeMocker, Senior Policy Analyst, EPA Office of Air and Radiation.
Under EPA direction, Project Team members designed and implemented the study, and
authored the study's full report, summary report, and supporting technical reports and
technical memoranda. In particular, the full report and summary report of the overall
Second Prospective Study were authored by Jim DeMocker of EPA and Jim Neumann of
Industrial Economics, Incorporated. Major contributions to the main reports and/or key
supporting reports were made by Rob Brenner and Jeneva Craig of EPA; Henry Roman,
Jason Price, Maura Flight, Tyra Walsh, Lindsay Ludwig, and Nadav Tanners of Industrial
Economics, Incorporated; Leland Deck of Stratus Consulting; Jim Wilson and Frank
Divita of E.H. Pechan and Associates; Sharon Douglas and Boddu Venkatesh of ICF
International; Neil Wheeler of Sonoma Technologies; and Brooks Depro and Robert
Beach of RTI International.
Many current and former EPA and contractor staff also made helpful contributions to the
development and/or review of the study. Those who made particularly significant
contributions included EPA staff Bryan Hubbell, Neal Fann, Amy Lamson, Lisa Conner,
Charles Fulcher, Rich Cook, Joe Touma, Chad Bailey, Ted Palma, Norm Possiel, Brian
Timin, Marc Houyoux, Larry Sorrels, Ken Davidson, and Jason Lynch; and contractor
staff Andrew Bollman, Maureen Mullen and Kirstin Thesing of E.H. Pechan and
Associates; Belle Hudischewskyj, Tom Myers, Yi Hua Wei, and Jay Haney of ICF
International; and Martin Ross and Lauren Davis of RTI International.
During all phases of the study, from initial design to final report drafting, the Project
Team and the Second Prospective Study benefitted immensely from the thoughtful,
rigorous, and expert advice of the Advisory Council on Clean Air Compliance Analysis
(Council) and its technical subcommittees. The Council was organized under the
auspices of EPA's Science Advisory Board, which provided staff support supervised by
Vanessa Vu, Director of the SAB Staff Office. The Designated Federal Official for the
final Council reviews was Stephanie Sanzone of the SAB Staff Office. Other SAB Staff
Office personnel who assisted in the coordination of Council reviews included Holly
Stall worth, Marc Rigas, Ellen Rubin, Angela Nugent, and Anthony Maciorowski.
The Council panel providing final review of the study was chaired by Professor James K.
Hammitt of Harvard University. Council members serving during the final review of this
report include John Bailar (Chair of the Health Effects Subcommittee), Michelle Bell,
Sylvia Brandt, Linda Bui, Dallas Burtraw, Ivan J. Fernandez (Chair of the Ecological
Effects Subcommittee), Shelby Gerking, Wayne Gray, D. Alan Hansen, Nathaniel
Keohane, Jonathan Levy, Richard L. Poirot, Arden Pope, Armistead (Ted) Russell (Chair
of the Air Quality Modeling Subcommittee), and Michael Walsh.
In addition to the Chairs listed above, members of the technical subcommittees serving
during the final review of this report included David T. Allen, David Chock, Paulette
Middleton, Ralph Morris, James Price, and Chris Walcek; Elizabeth Boyer, Charles T.
Vll
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Driscoll, Jr., Christine Goodale, Keith G. Harrison, Allan Legge, Stephen Polasky, and
Ralph Stahl, Jr.; John Fintan Hurley, Patrick Kinney, Michael T. Kleinman, Bart Ostro,
and Rebecca Parkin.
In addition, valuable advice and ideas in the early stages of project design and
implementation, as well as review of interim products of the study, were provided by
former Council members: Trudy Ann Cameron (former Council Chair), Maureen Cropper
(former Council Chair), Lauraine Chestnut, Lawrence Goulder, F. Reed Johnson,
Katherine Kiel, Charles Kolstad, Nino Kuenzli, Lester B. Lave, Virginia McConnell,
David Popp, and V. Kerry Smith. Former subcommittee members include: Mark Castro,
Harvey E. Jeffries, Morton Lippmann, and Scott Ollinger. The Council also consulted
with a number of invited experts and past panel members, including Aaron Cohen, John
Evans, Christopher Frey, Dale Hattis, D. Warner North, Thomas S. Wallsten, and Ronald
Wyzga.
vin
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 1 - INTRODUCTION
BACKGROUND AND PURPOSE
Section 812 of the 1990 Clean Air Act Amendments established a requirement that EPA
develop periodic reports that estimate the benefits and costs of the Clean Air Act (CAA).
The main goal of these reports is to provide Congress and the public with comprehensive,
up-to-date, peer-reviewed information on the Clean Air Act's social benefits and costs,
including improvements in human health, welfare, and ecological resources, as well as
the impact of CAA provisions on the US economy. This report is the third in the Section
812 series, and is the result of EPA's Second Prospective analysis of the 1990
Amendments.
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
Retrospective analysis comprehensively assessed benefits and costs of requirements of
the 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 gained in the form of increased health, environmental quality, and productivity. The
aggregate benefits of the CAA during the 1970 to 1990 period exceeded costs by a factor
of 10 to 100.
A second Section 812 report, The Benefits and Costs of the Clean Air Act: 1990 to 2010,
was completed in November of 1999 and addressed the incremental costs and benefits of
the Clean Air Act Amendments (CAAA) enacted by Congress and signed by the
President in November of 1990. This First Prospective analysis addressed
implementation of the CAAA over the period 1990 to 2010, and found that aggregate
benefits of the Amendments alone, excluding provisions in place prior to 1990, exceeded
the costs by a factor of four.
Similar to these prior analyses, this document has one primary and several secondary
objectives. The main goal is to provide Congress and the public with comprehensive, up-
to-date, peer-reviewed information on the CAAA's social costs and benefits, including
health, welfare, and ecological benefits. Data and methods derived from the
Retrospective and First Prospective analysis have already been used to assist policy-
makers in refining clean air regulations over the last several years, and we hope the
information continues to prove useful to Congress during future Clean Air Act
reauthorizations. 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 analysis will help better target
1-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
efforts to improve the accuracy and usefulness of future prospective analyses, generated
either as part of this series or as part of EPA's ongoing responsibility to estimate benefits
and costs of major rulemakings.
RELATIONSHIP OF THIS REPORT TO OTHER ANALYSES
The Clean Air Act Amendments of 1990 augmented the significant progress made in
improving the nation's air quality through the original Clean Air Act of 1970 and its 1977
amendments. The amendments built off the existing structure of the original Clean Air
Act, but went beyond those requirements to tighten and clarify implementation goals and
timing, increase the stringency of some federal requirements, revamp the hazardous air
pollutant regulatory program, refine and streamline permitting requirements, and
introduce new programs for the control of acid rain and stratospheric ozone depleters.
Because the 1990 Amendments represented an additional 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 benefits
assessed in the Retrospective analysis. In economic terminology, this report addresses
the marginal costs and benefits of the 1990 CAAA. Figure 1-1 below outlines this
relationship among the section 812 Retrospective, the First Prospective, and the Second
Prospective.
As illustrated in Figure 1-1, this report effectively updates and augments the First
Prospective. This report addresses essentially the same scenario and target variables as
the First Prospective, but incorporates a number of significant enhancements. First, this
report extends the time period of analysis an additional ten years relative to the First
Prospective, covering the period from the signing of the amendments in 1990 through
2020. Second, this report reflects updated cost and emissions estimation methods,
including use of a new model suited to nonroad engine regulation and incorporation of
the effects of learning-by-doing on projections of direct costs. Third, this report
incorporates new information on the benefits of air pollutant regulation, including use of
an integrated national-scale air quality model, more comprehensive characterization of
ecological benefits, and an air toxics case study. Fourth, the report reflects investments in
more comprehensive uncertainty analysis, including quantitative analyses where feasible.
Finally, this report incorporates a sophisticated economy-wide model to estimate effects
of the CAAA on such measures as GDP, prices, and consumer welfare. The
Retrospective analysis employed a similar model for assessing the direct costs of
compliance, but for the first time in this study the Agency has explored the economy-
wide implications of both the direct costs and the health benefits of the CAAA on
economic productivity, providing a much more complete picture of the full implications
of CAAA regulations.
The scope of this analysis is to estimate the costs and benefits of reducing emissions of
criteria pollutants under two scenarios, depicted in schematic form in Figure 1-1 below:
1-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 1-1. CLEAN AIR ACT SECTION 812 SCENARIOS: CONCEPTUAL SCHEMATIC
Second P
id Prospective
in
c
o
'w
v>
1
LLI
1970
1990
2000
2010
2020
Time
1. An historical, "with-CAAA" scenario control case that reflects expected or likely
future measures implemented since 1990 to comply with rules promulgated
through September 2005!; and
2. A counterfactual "without CAAA" scenario baseline case that freezes the scope
and stringency of emissions controls at their 1990 levels, while allowing for
changes in population and economic activity and, therefore, in emissions
attributable to economic and population growth.
The Second Prospective analysis required locking in a set of emissions reductions to be
used in subsequent analyses at a relatively early date (late 2005), and as a result we were
compelled to forecast the implementation outcome of several pending programs. The
most important of these was the then-promulgated Clean Air Interstate Rule (CAIR),
which took major steps to further reduce SOx and NOx emissions from electric
generating units. The rule has subsequently been vacated, and then remanded; EPA is
currently considering a proposed rule to modify areas identified by the court as
The lone exception is the Coke Ovens Residual Risk rulemaking, promulgated under Title III of the Act in March 2005. We
omitted this rule because it has a very small impact on criteria pollutant emissions (less than 10 tons per year VOCs)
relative to the overall impact of the CAM. The primary MACT rule for coke oven emissions, however, involves much larger
reductions and therefore is included in the with-CAAA scenario.
1-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
problematic. As a result, the emissions forecasts for electric generating units
incorporated in the with-CAAA scenario may not reflect the controls that are ultimately
implemented in a modified program. We acknowledge and discuss these types of
discrepancies and their impact on the outcome of our analysis in the document.
In addition, despite our efforts to comprehensively evaluate the costs and benefits of all
provisions of the Clean Air Act and its Amendments, there remain a few categories of
effects that are not addressed by the Retrospective or either prospective analysis. For
example, this Second Prospective analysis does not assess the effect of CAAA provisions
on lead exposures, primarily because the 1990 Amendments did not include major new
provisions for the control of lead emissions until the NAAQS for lead was recently
revisited and made significantly more stringent; the NAAQS revision was finalized after
our emissions inventory development had been completed, too late for inclusion in our
analysis. In addition, persistent data and model limitations preclude a full quantitative
treatment of some costs and many benefits of other clean air programs. Therefore, while
we considered all potentially relevant effects of the Clean Air Act and related programs,
the quantitative results we present are not fully comprehensive, even for programs
included in our assessment. Other, more modest omissions are acknowledged in the
supporting documentation for this effort.2
REQUIREMENTS OF THE 1990 CLEAN AIR ACT AMENDMENTS
This Second Prospective analysis, within the limitations discussed above, presents a
comprehensive estimate of costs and benefits of the key regulatory 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 Standards.
Title II. Regulates mobile sources and establishes requirements for reformulated gasoline
and clean fuel vehicles.
Title III. Expands and modifies regulations of hazardous air pollutant emissions; and
establishes a list of 189 hazardous air pollutants to be regulated.
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 enforcement.
Titles VIII through XI. Establish miscellaneous provisions for issues such as
disadvantaged business concerns, research, training, new regulation of outer continental
1 See www.epa.gov/oar/sect812 for a complete list and opportunity to download supporting documentation for this Second
Prospective analysis.
1-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
shelf sources, and assistance for people whose employment opportunities shift 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 established a requirement that EPA analyze the costs and benefits
to human health and the environment that are attributable to the Clean Air Act. In
addition, section 812 directed EPA to measure the effects of this statute on economic
growth, employment, productivity, cost of living, and the overall economy of the United
States.
This analysis does not provide updated information on the costs and benefits of CAAA
Title V regulations, which were thoroughly assessed in the First Prospective. Although
Title V is believed to have yielded benefits in the efficiency of air permitting, those
benefits are largely unquantified - as a result, the main effect of including Title V in the
First Prospective was to increase the cost estimate by about $300 million. Similarly, we
omit further consideration of Title VI regulation of the emissions of stratospheric ozone
depleting substances, which was also assessed in the First Prospective. Although
regulations under Title VI are continually updated and refined, the major components of
Title VI were in place prior to the First Prospective and were thoroughly analyzed as part
of that effort, resulting in the finding that the benefits of Title VI vastly exceeded its cost.
As a result, EPA chose to focus resources in the Second Prospective on other areas and
refinements. Because Titles V and VI have been previously assessed, and because Titles
VII through XI are largely procedural and have mostly modest effects on air pollutant
emissions and costs, this Second Prospective analysis is focused on the major emissions
regulatory programs of the CAAA, which make up Titles I through IV of the statutory
language.3
ANALYTICAL DESIGN AND REVIEW
TARGET VARIABLE
The Second 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 overall effects of the Clean Air Act, this analysis
complements the Regulatory Impact Analyses (RIAs) developed by EPA over the years
to evaluate individual regulations. We relied on information about the costs and benefits
of specific rules provided by these RIAs, as well as other EPA analyses, in order to use
resources efficiently. For this analysis, although costs can be reliably attributed to
individual programs, the broad-scale approach adopted in this prospective study largely
precludes reliable re-estimation of the benefits on a per-standard or per-program level.
Similar to the Retrospective and First Prospective benefits analysis, this study calculates
3 Note that some elements of Title VII enforcement efforts, such as settlements for historical violations of CM provisions,
particularly in the electric utility and petroleum refining sectors, are included in the emissions inventories of the with-CAAA
scenario. For more information, see EPA's detailed emissions report supporting this study at www.epa.gov/oar/sect812
1-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
the change in incidences of adverse effects implied by changes in ambient concentrations
of air pollutants. However, pollutant emissions reductions achieved contribute to changes
in ambient concentrations of those, or secondarily formed, pollutants in ways that are
highly complex, interactive, and often nonlinear. Although it would be possible to design
specific scenarios that focused analyses only on a subset of regulations (for example, all
of Title IV), those policy scenarios are not realistic. For example, exclusion of major
components of the Federal rules required under the CAAA would then trigger a much
greater need for reductions at the local level, in order to achieve NAAQS standards which
apply at the metropolitan area scale. Further, emissions reductions achieved by the
provisions of each Title, or more broadly by regulations across the CAAA provisions that
apply to a specific category of emitting sources, interact with other regulations to affect
the benefits implications of any emissions reduction. Therefore, benefits cannot be
reliably isolated or matched to provision-specific changes in emissions or costs.
Focusing on the broader target variables of overall costs and overall benefits of the Clean
Air Act, the EPA Project Team adopted an approach based on construction and
comparison of two distinct scenarios, briefly mentioned above: a ""without-CAAA" and a
"with-CAAA" scenario. The without-CAAA scenario essentially freezes federal, state, and
local air pollution controls at the levels of stringency and effectiveness which prevailed in
1990. The with-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 specific construction of the
scenarios and their relationship to historical trends, see Chapter 2 of this document.
KEY ASSUMPTIONS
Similar to the Retrospective and First Prospective analyses, we made two key
assumptions during the scenario design process to avoid miring the analytical process in
endless speculation. First, as stated above, we froze air pollution controls at 1990 levels
throughout the "without-CAAA" scenario. Second, we assumed that the geographic
distributions of population and economic activity remain the same between the two
scenarios, although these distributions could be expected to change overtime under both
scenarios in response to differences across scenarios in income and air quality.
The first assumption is an obvious simplification. 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, particularly since some states, most notably California, 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 meaningless
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 without-CAAA
scenario has been structured to reflect the assumption that states and localities would not
1-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
have invested further in air pollution control programs after 1990 in the absence of the
federal CAAA. Thus, this analysis accounts for all costs and benefits of air pollution
control from 1990 to 2020 and does not speculate about the fraction of costs and benefits
attributable exclusively to the federal CAAA. Nevertheless, it is important to note that
state and local governments and private initiatives are responsible 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 government and with the active participation
and cooperation of private entities and individuals.
The second assumption concerns changing demographic patterns in response to air
pollution. In the hypothetical without-CAAA scenario, air quality is worse than the actual
1990 conditions and the projected air quality in the with-CAAA scenario. It is possible
that under the without-CAAA scenario more people, relative to the with-CAAA case,
would move away from the most heavily polluted areas. Rather than speculate on the
scale of population movement, the analysis assumes no differences in demographic
patterns between the two scenarios. Similarly, the analysis assumes no differences
between the two scenarios with respect to the level or spatial pattern of overall economic
activity. Both scenarios do, however, reflect recent Census Bureau projections of
population growth and the distribution of population across the country.
ANALYTIC SEQUENCE
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
3. air quality modeling
4. health and environmental effects estimation
5. economic valuation
6. results aggregation and uncertainty characterization
Figure 1-2 summarizes the analytical sequence used to develop the prospective results;
we describe the analytic process in greater detail below.
The first step of the analysis is the 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 inventory for the base year (1990); (2) projection of
emissions for the without-CAAA case for three target years ~ 2000, 2010, and 2020 ~
assuming a freeze on emissions control regulation at 1990 levels and continued economic
progress, consistent with sector-specific Department of Energy Annual Energy Outlook
economic activity projections; and (3) construction of with-CAAA estimates for the same
three target years, using the same set of economic activity projections used in the without-
CAAA case but with regulatory stringency, scope, and timing consistent with EPA's
CAAA implementation plan (as of late 2005). The analysis reflects application of utility
1-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
and other sector-specific emissions models developed and used in various offices of
EPA's Office of Air and Radiation. These emissions models provide estimates of
emissions of five criteria air pollutants2 from each of several key emitting sectors. We
provide more details in Chapter 2.
FIGURE 1-2. ANALYTIC SEQUENCE FOR THE SECOND PROSPECTIVE ANALYSIS
Scenario Development
Sector Modeling
Emissions
Air Quality Modeling
i
Health
i
Welfare
Economic Valuation
Supplemental Analyses:
Benefit-Cost Comparison
Air Toxic case study
Ecological lit review
Ecological case study
Uncertainty Analyses
Macroeconomic
modeling
The emissions modeling step is a critical component of the 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 subsequent output from the emissions
estimation models, depending on the model used. For example, the Integrated Planning
Model used to analyze the utility sector reflects a financially optimal allocation of
reductions of sulfur and nitrogen oxides - taking into account the regulatory flexibility
2 The five pollutants are particulate matter (separate estimates for each of PM10 and PM2.5), sulfur dioxide (S02), nitrogen
oxides (NOX), carbon monoxide (CO), and volatile organic compounds (VOCs). One of the CM criteria pollutants, ozone
(03), is formed in the atmosphere through the interaction of sunlight and ozone precursor pollutants such as NOX and VOCs.
We also develop estimates for ammonia (NH3) emissions. Ammonia is not a criteria pollutant, but is an important input to
the air quality modeling step because it affects secondary particulate formation. A sixth criteria pollutant, lead (Pb), is not
included in this analysis since airborne emissions of lead were mostly eliminated by pre-1990 Clean Air Act programs - the
recent tightening of the Pb NAAQS, necessitated by an enhanced understanding of the effects of even small exposures to
airborne lead, was finalized too late to include in our scenarios. However, available estimates of the benefits and costs of
the updated Pb NAAQS could be viewed as approximately additive to the results presented here.
1-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
inherent in the Title IV trading programs - thereby estimating emissions reductions and
compliance costs simultaneously. Direct costs are addressed in Chapter 3.
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. For secondary particulate matter, ozone, and other air quality
conditions that involve substantial non-linear formation processes and/or long-range
atmospheric transport and transformation, the EPA Project Team employed EPA's
Community Multi-scale Air Quality (CMAQ) system. This modeling system, for the first
time in the series of Section 812 studies, provides a fully national, integrated analysis of
multiple emissions and their interactions. The result is a consistent estimate of air quality
for both primary and secondarily formed pollutants, as well as deposition and visibility
outcomes that represent the core of the subsequent benefit analyses. Air quality modeling
is covered in Chapter 4.
Up to this point of the analysis, modeled conditions and outcomes establish the without-
CAAA and with-CAAA scenarios. However, at the air quality modeling step, the analysis
returns to a foundation based on actual historical conditions and data, providing a form of
"ground-truthing" of the results. Specifically, actual 2000 historical air quality
monitoring data are used to define the baseline conditions from which the without-CAAA
and with-CAAA scenario air quality projections are constructed. We derive air quality
conditions under each of the projected years of the with-CAAA scenario by scaling the
historical data adopted for the base year (2000) by the ratio of the modeled with-CAAA
and base year air quality. We use the same approach to estimate future year air quality
for the without-CAAA scenario. This method takes advantage of the richness of the
monitoring data on air quality, 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
characterizing air quality conditions at each monitoring site in the lower 48 states. This
procedure also provided a means for calibrating model results in those grid cells where no
monitors exist, combining model results with nearby monitor data to yield a "surface" of
air quality that avoids the problems with direct extrapolation of results from monitors not
located within a grid cell boundary.
The without-CAAA and with-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 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 distribution of unit values derived from the economic
1-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
literature 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
nonmonetized effect. In many other cases, available data and techniques are insufficient
to support anything more than a qualitative characterization of the change in effects.
Health effects estimation and valuation are addressed in Chapter 5, and welfare effects,
including ecological impacts, visibility, and agriculture and forest productivity effects,
and their valuation, are addressed in Chapter 6.
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 estimate reflecting quantified uncertainties associated with the
physical effects and economic valuation steps. The monetized benefits used in the net
benefit calculations reflect only a portion of the total benefits due to limitations in
analytical resources, available data and models, and the state of the science. For example,
in many cases we 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 demonstrate 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. Details are provided in Chapter 7.
In addition to the sequence of analyses outlined in Figure 1-2, which are focused on
generating the key target variable of national net monetized benefits, a number of
supplemental analyses were also conducted to provide further insights on the impacts of
CAAA provisions for natural resources, health, and economic output. The first of these
supplemental analyses uses the Second Prospective's national direct cost, health
incidence, and health benefits valuation results to conduct further national-scale
economy-wide modeling using what is known as a Computable General Equilibrium
(CGE) model. The CGE model simulates, in a simplified way, shifts in markets and
transactions throughout the economy that might result from CAAA provisions. It is
therefore useful in assessing impacts on Gross Domestic Product (GDP), prices, and
sector shifts in production (e.g., from "dirty" to "clean" industries). Most past
applications of CGEs have focused on the economy-wide implications of the costs of
complying with regulations - as a result, many prior applications, including the use of
CGE in the Retrospective study, tell only half the story. Air pollution regulations not
only impose direct costs, but also yield benefits, and at least some of these benefits (e.g.,
reduced medical expenditures, improved labor productivity owing to better health) affect
market transactions in ways that can be assessed in the CGE framework. Not all benefits
are amenable to analysis in a CGE, however - for example, nonmarket effects such as
willingness-to-pay to avoid pain and suffering of air pollutant-linked disease cannot be
incorporated. Nonetheless, this study represents one of the first broad applications of a
CGE tool to regulatory costs and benefits. More details are provided in Chapter 8.
1-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Two other supplemental analyses represent local-scale case studies of difficult-to-
quantify benefits of air pollution regulation. One is a case study of health benefits
associated with air toxics control. In prior section 812 studies, benefits of air toxics
programs have been largely limited to their effects on criteria pollutant outcomes. For
example, many air toxics are also volatile organic compounds, and so contribute to ozone
formation, an effect which can be fairly readily quantified. The direct effects of air toxics
on health, however, have been more difficult to quantify, partly because of data
constraints, and partly because the highly localized effects of air toxics require a level of
emissions and air quality modeling resolution that is currently infeasible for a national
analysis. The air toxics case study, the results of which are presented in Chapter 5,
provides an example of the benefits of air toxics control for a pollutant (benzene) and
geographic scope (Houston area) that is both relatively data rich and computationally
manageable.
A second case study involves ecological effects, focused on the Adirondack region of
New York State. This region was carefully chosen, based on the recommendation of the
Advisory Council on Clean Air Compliance Analysis Ecological Effects Subcommittee
(Council EES), because of its relatively high sensitivity to the effects of deposited air
pollutants, because those same effects are relatively well-studied, and because methods
exist to quantify and, in many cases, monetize the benefits of air pollution controls.
Using the same emissions and air quality scenarios as in the overall national study, the
ecological case study assesses the impact of sulfur and nitrogen deposition in the
Adirondack region on aquatic resources, particularly lakes and ponds that support
recreational fishing, and on commercial timber resources.
Uncertainty analyses are also conducted at each phase of the analyses. Where applicable,
we present the results of a series of quantitative uncertainty analyses that test the effect of
alternative methods, models, or assumptions that differ from those we used to derive the
primary net benefit estimate. The primary estimate of net benefits and the range around
this estimate, however, reflect our current interpretation of the available literature; our
judgments regarding the best available data, models, and modeling methodologies; and
the assumptions we consider most appropriate to adopt in the face of important
uncertainties.
Finally, throughout the report, at the end of each chapter, we discuss the major sources of
uncertainty for each analytic step. Although the impact of many of these uncertainties
cannot be quantified, 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; andprobabfy minor factors likely have effects less than five
percent of total net benefits.
The Second Prospective involved a much greater effort in uncertainty analyses than prior
reports in this series. Figure 1-3 illustrates the Project Team's approach to uncertainty
analysis in the Second Prospective, superimposed on the overall analytic chain for the
study presented above. The grey box in Figure 1-3 represents the extent of uncertainty
1-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
analysis in the first section 812 prospective analysis, which was largely limited to
analysis of parameter uncertainty in the concentration-response and valuation steps of the
benefits analyses. Those parameter uncertainty analyses have become standard practice
in EPA analyses of air pollution program benefits, and are an integral part of the
BenMAP benefits assessment tool. The results of the probabilistic modeling of these
uncertainties constitute the "primary low" and "primary high" estimates presented in
Table 5-7 in Chapter 5 as well as in Chapter 7.
Enhancements employed in the current analysis include both "online" analyses (shown in
color), that feed information on uncertainty into the analytical chain at various points and
propagate it through the remaining steps in the chain, and separate "offline" analyses and
research that provide insights into the uncertainty, sensitivity, and robustness of results to
alternative assumptions that are currently most easily modeled outside the main analytical
process.
The online analyses consist of the selection of alternative inputs for mortality
concentration-response and valuation in BenMAP, as well as an analysis of the effect on
benefits of sector specific, marginal changes in PM-related emissions from the core
scenarios. This online analysis substitutes EPA's Response Surface Model (RSM) for
CMAQ. RSM is a less resource intensive meta-model of CMAQ used to rapidly
approximate PM concentrations from alternative emissions inputs. Those analyses are
described in much greater detail in the supporting uncertainty analysis report, referenced
at the end of this chapter.
The bottom box in Figure 1-3 lists additional offline research and analysis we
incorporated into the current study. As with the online analyses, these analyses were
chosen because they address uncertainty in key analytical elements or choices that may
significantly influence benefit or cost estimates. Most of these are described in this
integrated report, some only briefly, but full descriptions of the data, models, and
methods applied in these analyses are included in the underlying uncertainty analysis
report.
1-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 1-3. SCHEMATIC OF UNCERTAINTY ANALYSES
Analytic Design
Scenario
Development
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
REVIEW PROCESS
The 1990 CAA Amendments established a requirement that EPA consult with an outside
panel of experts during the development and interpretation of the 812 studies. This panel
of experts was originally organized in 1991 under the auspices of EPA's Science
Advisory Board (SAB) as the Advisory Council on Clean Air Compliance Analysis
(hereafter, the Council). Organizing the review committee under the SAB ensured that
highly qualified 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 2003
with a review of the analytical design plan. Since the initial meetings, the Council and its
subcommittees have met many times to review proposed data, proposed methodologies,
and interim results. While the full Council retains overall review responsibility for the
section 812 studies, some specific issues concerning physical effects and air quality
modeling were referred to subcommittees comprised of both Council members and
members of other SAB committees. The Council's Health Effects Subcommittee (HES),
Air Quality Modeling Subcommittee (AQMS), and Ecological Effects Subcommittee
(EES) held both in-person and teleconference meetings to review methodology proposals
and modeling results and conveyed their findings and recommendations to the parent
Council.
REPORT ORGANIZATION
The remainder of the main text of this report summarizes the key methodologies and
findings of our prospective study.
Chapter 2 summarizes emissions modeling and provides important additional detail
on design of the regulatory scenarios.
Chapter 3 discusses the direct cost estimation.
Chapter 4 presents the air quality modeling methodology and results.
Chapter 5 describes the approaches used and principal results obtained through the
human health effects estimation and valuation processes.
Chapter 6 summarizes the ecological and other welfare effects analyses, including
assessments of commercial timber, agriculture, visibility, and other categories of
effects.
Chapter 7 presents aggregated results of the cost and benefit estimates and describes
and evaluates important uncertainties in the results.
Chapter 8 presents estimates of the effect of the Clean Air Act Amendments on
economic growth, productivity, prices, household economic welfare, and the overall
economy of the United States, through the application of an economy-wide
economic simulation model.
1-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Note that additional details regarding the methodologies and results of this study can be
found in a series of supporting reports, available at EPA's Section 812 website
(www.epa.gov/oar/sect812). These reports include the following:
Emission Projections for the Clean Air Act Second Section 812 Prospective
Analysis.
Direct Cost Estimates for the Clean Air Act Second Section 812 Prospective
Analysis.
Memorandum to the Files Re Documentation of Second Prospective Study Air
Quality Modeling.
Health and Welfare Benefits Analyses to Support the Second Section 812 Benefit-
Cost Analysis of the Clean Air Act.
Effects of Air Pollutants on Ecological Resources: Literature Review and Case
Studies.
Section 812 Prospective Study of the Benefits and Costs of the Clean Air Act: Air
Toxics Case Study - Health Benefits of Benzene Reductions in Houston, 1990-2020.
Uncertainty Analyses to Support the Second Section 812 Benefit-Cost Analysis of the
Clean Air Act.
1-15
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 2 - EMISSIONS
Estimation of pollutant
emissions, a key component of
this prospective analysis, serves
as the starting point for
subsequent benefit and cost
estimates. We focused the
emissions analysis on six major
pollutants that are regulated by
the Clean Air Act Amendments:
volatile organic compounds
(VOCs), nitrogen oxides (NOX),
sulfur dioxide (SO2), carbon
monoxide (CO), particulate
matter with an aerodynamic
diameter of 10 microns or less
(PMio), and fine particulate
matter (PM2 5). Estimates of
current and future year ammonia
(NH3) emissions are also
included in this study because of
their importance in the
atmospheric formation of fine particles in the ambient air. For each of these pollutants
we projected emissions to the years 2010 and 2020 under two different scenarios:
i. An historical "with-CAAA" scenario control case that reflects expected or
likely future measures implemented since 1990 to comply with rules
promulgated through September 2005; and
2. A counterfactual ""without-CAAA" scenario baseline case that freezes the
scope and stringency of emissions controls at their 1990 levels, while
allowing for changes in emissions attributable to economic and population
growth.4
.
Emis
^
Air Quality
1
Health
i
Economic
E
Scenario Development
I
Sector Modeling
I
r i
r
sions Direct Cost
-
Modeling
r
Welfare
r
Valuation
I
I
r
tenefit-Cost Comparison
4 Implementing this approach has occasionally required some difficult decisions on what constitutes 1990 levels of emissions
controls. In general, we have interpreted any rules that were promulgated as final prior to 1990 to be part of the without-
CAAA scenario baseline. The residential wood stove New Source Performance Standard, however, was promulgated in 1988,
but is not part of the without-CAAA scenario, because EPA did not certify NSPS compliant wood stoves until 1992. In this
2-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 2-1.
We projected emissions for five major source categories: utilities, or electricity
generating units (EGUs); non-EGU industrial point sources; onroad motor vehicles;
nonroad engines/vehicles; and area sources, which are smaller, more diffuse sources of
pollutants that derive from many sources.5 Table 2-1 gives examples of emissions
sources for each of the five categories examined in this analysis and indicates which
major pollutants are targeted by CAAA requirements in each category. The primary
purpose of emissions analysis in this study is to estimate how emissions change over time
and across our scenarios, so we can estimate costs of reducing emissions and the benefits
of those emissions reductions for each of our target years.
MAJOR EMISSIONS SOURCE CATEGORIES
SOURCE CATEGORY
Electricity Generating Units
(EGUs)
Non-EGU Industrial Point
Sources
Onroad Motor Vehicles
Nonroad Engines/Vehicles
Area Sources
EXAMPLES
electricity producing utilities
boilers, cement kilns, process
heaters, turbines
buses, cars, trucks (sources
that usually operate on roads
and highways)
aircraft, construction
equipment, lawn and garden
equipment, locomotives,
marine engines
agricultural tilling, dry
cleaners, open burning,
wildfires
POLLUTANTS WITH
SUBSTANTIAL EMISSIONS
REDUCTIONS FROM CAAA
COMPLIANCE
NOX, S02
NOX, VOC, S02, PM10 PM2.5
NOX, VOC, CO
NOX, VOC, CO
NOX, VOC, PM10, PM2.5
This chapter consists of four sections. The first section provides an overview of our
approach for developing emissions estimates. The second section summarizes our
emissions projections for the years 2000, 2010, and 2020, and presents our estimates of
changes in future emissions resulting from the implementation of the 1990 Amendments.
The third section compares these results with estimates from the First Section 812
Prospective Analysis. Finally, we conclude this chapter with a summary of the key
uncertainties associated with estimating emissions.
case, perhaps incorrectly, we interpreted the effective date of 1992 as the determining factor in whether the level of
emissions stringency in 1990 should include the wood stove NSPS.
5 Area sources are also commonly referred to as nonpoint sources. We estimated utility and industrial point source emissions
at the plant/facility level. We estimated nonroad engine/vehicle, motor vehicle, and area source emissions at the county
level.
2-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
OVERVIEW OF APPROACH
For four out of the five major source categories described in this report—all except
electric generating units—we applied the following general method to estimate
emissions:
1. Select a "base" inventory for a specific year. This involves selection of an
historical year inventory from which projections will be based.
2. Select activity factors to project growth in the level of pollution-generating
activity in the target years. The activity factors should provide the best
possible means for representing future air pollutant emissions levels in the
absence of controls.
3. Develop a database of scenario-specific emissions control factors, to
represent emissions control efficiencies under the two scenarios of interest.
The control factors are "layered on" to the projected emissions levels absent
controls to estimate future emissions levels, taking into account those
controls required for CAAA compliance .
Air pollutant emissions for the fifth category, EGUs, were estimated by application of the
Integrated Planning Model (IPM), a model developed by ICF Consulting. IPM estimates
EGU emissions in the 48 contiguous states and the District of Columbia through an
optimization procedure that considers costs of electricity generation, costs of pollution
control, and external projections of electricity demand to forecast the fuel choice,
pollution control method, and generation for each unit considered in the model. We used
IPM to estimate EGU emissions in both the with-CAAA and without-CAAA scenarios for
2000, 2010, and 2020.
SELECTION OF BASE YEAR INVENTORY
The without-CAAA scenario emission projections are made from a 1990 base year, while
the with-CAAA scenario emission projections use a base year of 2000. The logic for these
base year inventory choices relates to the specific definitions of the scenarios themselves.
The with-CAAA scenario tracks compliance with CAAA requirements overtime; as a
result, the best basis for projecting the with-CAAA scenario is a current emissions
inventory that incorporates decisions made since 1990 to comply with the act. The
without-CAAA scenario, on the other hand, freezes the stringency of regulation at 1990
levels. The analysis therefore uses 1990 emission rates as a base and adjusts those
emissions to account for economic activity over time. We determined that this method
was less problematic than basing projections on a recent emissions inventory and trying
to simulate the effect of removing CAAA emission controls currently in place. Table 2-2
summarizes the key databases that were used in this study to estimate emissions for
historic years 1990 and 2000. Note that, in some cases, we determined that the best
representation for year 2000 emissions was actually a later year, either 2002 or 2001.
Those decisions are explained below.
2-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 2-2. BASE YEAR EMISSION DATA SOURCES FOR THE WITH- AND WITHOUT-CAAA
SCENARIOS
SOURCE CATEGORY
Electricity Generating
Units (EGUs)
Non-EGU Industrial Point
Sources
Onroad Motor Vehicles
Nonroad Engines/Vehicles
Area Sources
WITHOUT-CAAA SCENARIO -
1990
1990 EPA Point Source NEI1
1990 EPA Point Source NEI
MOBILE6.2 Emission Factors and
1 990 NEI VMT Database
NONROAD 2004 Model
Simulation for Calendar Year
1990
1990 EPA Nonpoint Source NEI3
WITH-CAAA SCENARIO - 2000
Estimated by the EPA Integrated
Planning Model for 2001
2002 EPA Point Source NEI
(Draft)
MOBILE6.2 Emission Factors and
2000 NEI VMT Database2
NONROAD 2004 Model Simulation
for Calendar Year 2000
2002 EPA Nonpoint Source NEI
(Final)
1 The NEI is EPA's National Emissions Inventory, conducted every three years.
2 The California Air Resources Board (ARE) supplied estimates for California.
3 Adjustments were made to the 1990 nonpoint source NEI file for priority source categories.
For EGUs and non-EGU industrial point sources, we estimated 1990 emissions using the
1990 EPA National Emission Inventory (NEI) point source file. This file is consistent
with the emission estimates used for the First Section 812 Prospective and is thought to
be the most comprehensive and complete representation of point source emissions and
associated activity in that year. Similarly, the 1990 EPA NEI nonpoint source file - with
a few exceptions - was used to estimate 1990 area source sector emissions.6
For base year emissions estimates in the with-CAAA scenario, we drew emissions from a
variety of sources. Due to resource constraints and the quality of available data, we relied
on emissions estimates for years other than 2000. In the case of with-CAAA emissions
from industrial point sources and area sources, we used the point source and nonpoint
source files from the 2002 EPA NEI.7 We chose the 2002 NEI to represent the year 2000
estimates primarily because the 2002 inventory incorporated a number of refinements in
emissions estimation methods that were not included in the previous inventory, which
covered 1999 emissions. We judged that the improved quality of the 2002 NEI data
justified the small expected difference between emissions for these source categories in
6 The exceptions are where 1990 emissions were re-computed using updated methods developed for the 2002 National
Emissions Inventory (NEI) for selected source categories with the largest criteria pollutant emissions and most significant
methods changes.
7 We used the draft NEI point source file because the final version of that file was not available at the time the analysis was
performed. For area sources, we used the final NEI nonpoint source file.
2-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
2000 and in 2002. To estimate with-CAAA EGU emissions, we used data from a
modified version of IPM that retrospectively modeled emissions for the year 2001.8
The project team estimated 1990 and 2000 emissions for the onroad and nonroad
vehicle/engine sectors independently using consistent modeling approaches and activity
estimates. For example, emission factors from EPA's MOBILE6.2 model were used
together with data from the 1990 and 2000 NEI vehicle miles traveled (VMT) databases
to estimate onroad vehicle emissions for 1990 and 2000. Similarly, EPA's NONROAD
2004 model was used to estimate 1990 and 2000 emissions for nonroad vehicles/engines.
SELECTION OF ACTIVITY FACTORS FOR PROJECTIONS
After specifying base year emissions, we projected emissions to 2000 (for the without-
CAAA scenario), 2010, and 2020. To model emissions in the absence of controls, our
general approach was to multiply an emission factor - derived from base year emissions
estimates - by the level of emission-generating activity upon which the emission factor is
based. These emission-generating activities vary by source category, but they are
generally related to economic activity, such as transportation, energy consumption, and
industrial output. Specifically, economic growth projections entered the emissions
analysis in three places:
• an electricity demand forecast (included in IPM);
• a fuel consumption forecast for non-utility sectors; and
• economic growth projections that serve as activity drivers for several other
sources of air pollutants.
For this analysis, we used fully integrated economic growth, energy demand, and fuel
price projections to model economic growth in both the with-CAAA and the without-
CAAA scenarios. The primary advantage of this approach is that it allowed us to conduct
an internally consistent analysis of economic growth across all emitting sectors. To
implement this integrated approach, we chose the Department of Energy's National
Energy Modeling System (NEMS), which is used to produce DOE's Annual Energy
Outlook (AEO) projections. Our emissions estimates primarily rely on AEO's 2005
"reference case" scenarios. We supplemented these projections with additional forecasts
from other data sources for emissions sources where we determined that AEO's energy
and socioeconomic forecasts would not adequately represent growth in emissions-
generating activities.9 Table 2-3 presents the values that we used for the AEO 2005
projections for population, GDP, energy consumption, and oil price values in 2010 and
2020. For reference, the table also presents the historical values for each variable in
8 Due to resource constraints and model limitations, we relied primarily on a validation analysis EPA conducted on 2001
emissions, rather than developing a new analysis for the year 2000.
9 These emissions sources include agricultural production-crops, fertilizer application, and nitrogen solutions; agricultural
tilling; animal husbandry; aircraft; forest wildfires; prescribed burning for forest management; residential wood fireplaces
and wood stoves; and unpaved roads.
2-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 2-3.
2002, as reported in AEO 2005. For each variable, the table shows the implied annual
growth rate that AEO 2005 used to project population, GDP, energy consumption, and oil
prices from 2002 to 2010 and from 2010 to 2020.10
SUMMARY OF KEY DRIVER DATA APPLIED IN EMISSIONS PROJECTIONS
VARIABLE
Population (millions)
GDP (billion 2000 chain-weighted
dollars)
Energy Consumption (quadrillion Btu
per year)
World Oil Price (1999$ per barrel)
HISTORICAL
DATA
2002
288.6
$10,075
97.99
$22.17
AEO 2005
PROJECTIONS
2010
310.1
$13,084
111.27
$23.00
2020
337.0
$17,634
125.60
$26.22
IMPLIED ANNUAL GROWTH
RATE
2002-2010
0.90%
3.32%
1.60%
0.46%
2010-2020
0.83%
3.03%
1 .22%
1.32%
One notable exception to the above involves the specification of PM2 5 emissions from
non-EGU point sources and area sources. After initially attempting to model PM2 5
emissions in the without-CAAA scenario in 2000, 2010, and 2020 using the process
described above, we determined that the resulting estimates over-attributed emissions
reductions to the amendments. We applied two separate approaches to correct these
emissions estimates: For emissions from area sources, we projected emissions from the
two sectors responsible for the majority of emissions - construction and wood stoves -
using source-specific data. For emissions from non-EGU point sources, the project team
determined that emissions reductions from CAAA-mandated controls would be negligible
in 2000, so we set without-CAAA PM2 5 emissions equal to with-CAAA emissions in that
year.
APPLYING CONTROLS TO THE WITH-CAAA SCENARIO
To estimate the impact of CAAA controls on projected emissions in the with-CAAA
scenario, we modeled the application of controls required by CAAA programs, including
(among others):
• Title IVOC and NOX reasonably available control technology (RACT)
requirements in ozone nonattainment areas (NAAs);
• Title II on-road vehicle and nonroad engine/vehicle provisions;
• Title III National Emission Standards for Hazardous Air Pollutants (NESHAPs);
• Title IV programs focused on emissions from EGUs.
10 The table presents 2002 data in order to be consistent with EPA's 2002 NEI, which we used to estimate emissions from
industrial point sources and area sources.
2-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
• Additional EGU regulations, such as the Clean Air Interstate Rule (CAIR), the
Clean Air Mercury Rule (CAMR), and the Clean Air Visibility Rule (CAVR).
As a general rule, we incorporated the effects of CAAA rules promulgated through
September 2005.n As such, we did not account for the impacts of rules promulgated
after that date, such as the revised NAAQS for lead. Additionally, we modeled
reductions from rules that have since been vacated, like the Clean Air Mercury Rule
(CAMR) and the Clean Air Interstate Rule (CAIR), though CAIR has since been
remanded. Rather than attempting to estimate the impacts of whatever rules might
replace CAMR and CAIR, we modeled the rules as promulgated because that was the
best information available when we made analytic commitments.
A full list of the CAAA programs modeled for each source category is presented in Table
2-4, together with the pollutants targeted by each program. For each source category, we
identified factors to use in modeling the effect of emission controls required by the
CAAA. For EGUs, onroad motor vehicles, and nonroad engines/vehicles, we used
control factors included in the three EPA models we used to estimate base year
emissions: IPM, MOBILE, and NONROAD, respectively. For non-EGU industrial point
sources and area sources, we relied on control factors developed by the five Regional
Planning Organizations funded by EPA to address regional air pollution issues, as well as
factors developed by the California Air Resources Board.
One exception is the Coke Ovens Residual Risk rulemaking, promulgated under Title III of the Act in March 2005. We
omitted this rule because it has a very small impact on criteria pollutant emissions (less than 10 tons per year VOCs)
relative to the with-CAAA scenario. The primary Maximum Achievable Control Technology (MACT) rule for coke oven
emissions, however, involves much larger reductions and therefore is included in the with-CAAA scenario. In addition, we
also modeled emissions reductions from local controls implemented to comply with the 8-hour Ozone NAAQS, the PM2.5
NAAQS, and the Clean Air Visibility Rule, using the proposed or promulgated forms of these rules as of January 2008.
2-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 2-4.
MAJOR CAAA PROGRAMS MODELED IN THE WITH-CAAA SCENARIO
SECTOR
POLLUTANT
CAAA PROGRAMS
Electricity Generating
Units (EGUs)
NOX/SO2
NOX
Title IV acid rain emission allowance program;
Clean Air Interstate Rule (CAIR); Clean Air Mercury
Rule (CAMR); Cases and Settlements; Additional
measures to meet PM and ozone NAAQS;
NOx SIP Call post-2000
Non-EGU Industrial Point
Sources
NOX/VOC/SO2
NOX
VOC
Measures required to meet PM and ozone National
Ambient Air Quality Standards (NAAQS)
Ozone Transport Commission (OTC) small NOx
source model rule (where adopted); NOx SIP Call
2-, 4-, 7-, and 10-year maximum achievable control
technology (MACT) standards;
NOX/VOC/SO2
NOX/VOC
Onroad Motor Vehicles
PM/SO2
Tier 1 tailpipe standards (Title II); Tier 2 tailpipe
standards;
National and California low-emission vehicle (LEV)
program (Title I); Federal and California
reformulated gasoline for ozone NAAQS NAAs (Title
I); I/M programs for ozone and CO NAAQS NAAs
(Title I); NOx and VOC measures included in ozone
NAAQS SIPs
Heavy-duty diesel vehicle (HDDV) standards; Diesel
fuel sulfur content limits (Title II) (1993); Gasoline
fuel sulfur limits; Additional measures to meet new
PM NAAQS
Nonroad Engines/
Vehicles
NOX/VOC/PM
NOX/PM
NOX/PM/SO2
Federal Phase I and II compression ignition (Cl) and
spark-ignition (S-l) engine standards; Federal
commercial and recreational marine vessel
standards
Federal locomotive standards
Nonroad Diesel Rule
Area Sources
NOX/VOC/PM
NOX/VOC
VOC
RACT requirements; NOx and VOC measures
included in ozone SIPs; Additional measures to
meet PM and ozone NAAQS
Ozone Transport Commission (OTC) model rules
(where adopted)
2-, 4-, 7-, and 10-year MACT Standards; Federal
VOC rules for architectural and industrial
maintenance (AIM) coatings, autobody refinishing,
and consumer products
Note: See Hubbell et al. (2010) for additional information regarding rules and regulations attributed to
the 1990 CAAA.
2-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
EMISSIONS ESTIMATION RESULTS
Table 2-5 summarizes the national emission estimates by pollutant for each of the
scenario years evaluated in this study: 2000, 2010, and 2020. As a reference, the table
also presents total emissions for each pollutant in 1990. Figures 2-1 through 2-4 provide
a detailed breakdown of the emissions reductions in each target year by source category
for NOX, VOC, SO2, and primary PM2 5. We show the breakdown of emissions
reductions by source category for these pollutants because they constitute (or are
precursors of) the two main air quality impacts that drive the analysis of the benefits of
the CAAA: ozone and particulate matter pollution. The table and figures also incorporate
our estimates of emissions reductions from local controls required to meet attainment
requirements for 8-hour ozone and PM2 5 national ambient air quality standards
(NAAQS). Reductions needed for compliance, but for which we have not identified a
specific pollutant reducing measure or sector to achieve the reduction, are incorporated in
Table 2-5 and are presented as a separate category in Figures 2-1 through 2-4, labeled
"unidentified measures."
For five of the pollutants examined—NOX, SO2, PM10, PM2 5, and NH3—we estimate that
emissions in the absence of the amendments would increase steadily from 1990 through
2000, 2010, and 2020, suggesting that emissions controls in place by 1990 would not be
sufficient to prevent increases in pollutant emissions due to projected growth in economic
activity. For the remaining two pollutants—VOC and CO—emissions decrease between
1990 and 2000 as a result of automobile tailpipe controls enacted prior to 1990, but which
have delayed effects through the 1990s, before increasing from 2000 onward.
In the with-CAAA scenario, we estimate that emissions of SO2 and NOX will decrease
steadily from 1990 to 2020, while emissions of VOC, CO, PM10, and PM2 5 will decrease
from 1990 to 2010 before leveling off between 2010 and 2020. We also estimate that
emissions of NH3 will increase even in the presence of CAAA regulations, though at a
slightly slower pace than in the without-CAAA scenario. NH3 is not a specific target of
CAAA regulations, but some reductions result from efforts to control other pollutants.
The net result of these trends in the two scenarios is that we estimate that emissions
reductions, relative to the without-CAAA scenario, will increase for all pollutants
throughout the 2000 to 2020 period.
As Figure 2-1 shows, we estimate that reductions in NOX emissions will increase
substantially from 2000 to 2010 and from 2010 to 2020. All five major source categories
contribute to these reductions in 2010 and 2020, though the largest reductions come from
EGUs and on-road motor vehicles. Reductions in NOX emissions from EGUs are driven
largely by cap-and-trade programs, such as Phase II of the Ozone Transport Commission
memorandum of understanding and the Clean Air Interstate Rule.12 In the motor vehicle
sector, the large reductions in NOX emissions in 2010 and 2020 reflect both the delayed
12 Under Phase II of the OTC memorandum of understanding, eleven eastern states committed themselves to achieving
regional reductions in NOX emissions through a cap-and-trade system similar to the S02 trading program established under
Title IV of the amendments.
2-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
impact of Tier 1 NOX tailpipe standards as well as the impact of Tier 2 standards, which
went into effect in 2004.
Figure 2-2 shows increasing VOC emissions reductions from 2000 to 2020, with
contributions from all source categories, with the exception of EGUs. The figure also
shows a marked increase in on-road and nonroad emissions reductions between 2000 and
2010, reflecting both the delayed impact of Tier 1 VOC standards and the effect of low-
sulfur gasoline regulations. Additionally, about half of the rules affecting nonroad
sources came into effect between 2000 and 2010, explaining the increase in emissions
reductions during that time. Area sources also show large emissions reductions across all
three target years, driven primarily by regulations controlling evaporative emissions from
solvents, though residential fireplace and woodstove emissions are also projected to
decline as obsolete woodstoves are replaced with low-emitting models required by the
CAAA.13
In Figure 2-3, SO2 emissions reductions increase by more than 60 percent between 2000
and 2010, with a smaller increase between 2010 and 2020. Most reductions in SO2
emissions in all three target years come from EGUs, with smaller contributions from non-
EGU point sources and area sources as well. As with reductions in NOX emissions, the
CAIR and the Title IV cap and trade program are partly responsible for SO2 reductions
from EGUs, along with the revised PM2 5 NAAQS.
Figure 2-4 presents reductions in PM2 5 emissions for the three target years, with a steady
increase in reductions from 2000 through 2020, as PM2 5 NAAQS requirements ramp up.
Reductions in primary fine particulate emissions are expected to come from area sources,
nonroad and onroad vehicles, and EGUs. Reductions from area sources are driven
largely by the replacement of obsolete residential fireplaces and wood stoves, as well as
local controls on construction sites for PM NAAQS compliance. As noted above, we set
PM2 5 emissions at non-EGU industrial point sources in the without-CAAA scenario to be
equal to emissions in the with-CAAA scenario, so we do not estimate that there will be
any significant direct PM2 5 emissions reductions from that source category.
As noted earlier in this chapter, the woodstove NSPS was interpreted as part of the differential between the with- and
without-CAAA scenarios. NSPS compliance is required only for new units, which in practice are replaced very slowly. We
estimate that, almost 20 years after NSPS implementation, in 2010, about 70 percent of the wood stoves in use are pre-
NSPS uncertified models; by 2020, we estimate that turnover will reduce non-certified unit usage to just under 65 percent.
2-10
-------
TABLE 2-5.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
EMISSION TOTALS AND REDUCTIONS BY POLLUTANT - ALL SECTORS (THOUSAND TONS PER YEAR)
POLLUTANT
voc
NOX
CO
S02
PM10
PM2.5.1
NH3
1990
25,790
25,917
154,513
23,143
25,454
5,527
3,656
2000
WITHOUT-
CAAA
24,477
26,688
127,093
25,129
26,418
5,822
4,136
WITH-CAAA
17,798
20,837
107,691
15,319
21,143
5,489
3,983
REDUCTION
6,679
5,85?
19,403
9,810
5,275
333
153
2010
WITHOUT-
CAAA
26,742
28,517
134,151
26,831
26,405
5,924
4,405
WITH-CAAA
14,117
13,640
86,705
10,347
20,413
5,241
4,224
REDUCTION
12,626
14,877
47,447
16,484
5,992
682
181
2020
WITHOUT-
CAAA
31,288
31,740
155,970
27,912
28,280
6,368
4,787
WITH-CAAA
13,704
10,092
84,637
8,272
20,577
5,297
4,587
REDUCTION
17,584
21,647
71,332
19,640
7,702
1,072
200
1 PM2.s without-CAAA emissions were adjusted from previously reported values by reducing emissions from non-EGU industrial point sources and area
sources.
2-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 2-1. NOX REDUCTIONS ASSOCIATED WITH CAAA COMPLIANCE BY SOURCE CATEGORY
25,000,000
20,000,000
ra
01
15,000,000
Ol
Q.
1/1
C
o
10,000,000
5,000,000
2000
2010
2020
D Unidentified Measures
D Area
D Nonroad
• Onroad Vehicle
• Non-BGU Industrial Point
nBGU
FIGURE 2-2. VOC REDUCTIONS ASSOCIATED WITH CAAA COMPLIANCE BY SOURCE CATEGORY
. 12,000,000
Ol
Q.
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 2-3. SO2 REDUCTIONS ASSOCIATED WITH CAAA COMPLIANCE BY SOURCE CATEGORY
20 000 000 -
ra
01
*" 1 5 000 000 -
01
Q.
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
COMPARISON OF EMISSIONS ESTIMATES WITH THE FIRST PROSPECTIVE ANALYSIS
DIFFERENCES IN METHODOLOGY
In comparison with the First Prospective 812 Analysis, the Second Prospective includes a
number of refinements and improvements in emissions estimation methods, as well as a
different set of regulatory assumptions.
1. Updated Emissions and Economic Activity Data: Because the Second Prospective
analysis was developed ten years after the First Prospective, it incorporates
additional information that was not available when the First Prospective was
developed. This information includes with-CAAA emissions estimates for the
historical year 2000 as well as additional historical trend data used to project
economic activity from 1990 to 2000.
2. Additional Regulatory Requirements: The Second Prospective Analysis accounts
for several major CAA regulations that were not yet promulgated in 1996, when
decisions were made about which regulations to include in the First Prospective.
These regulations include, but are not limited to, the Clean Air Interstate Rule
(CAIR); the Clean Air Visibility Rule (CAVR); Tier II vehicle rules and heavy-
duty diesel vehicle rules, and the local controls required for the revised 8-hour
ozone and PM25 NAAQS. Because of this difference, the Second Prospective
Analysis models greater emissions reductions in 2000 and 2010 than were
predicted in the First Prospective, as we discuss in the following section.
3. Integrated Economic Modeling Approach: In the First Prospective Analysis, we
relied on a number of modeling tools to project future emissions, including
projections of economic activity and population growth from the Bureau of
Economic Analysis, and vehicle miles traveled from EPA's MOBILE fuel
consumption model. By using fully-integrated economic growth, energy
demand, and fuel price projections from DOE's AEO 2005, we were able to
achieve a greater degree of internal consistency in the Second Prospective
Analysis.
DIFFERENCES IN EMISSIONS RESULTS
Figures 2-5 and 2-6 show estimates from the First and Second Prospective Analyses of
cumulative criteria pollutant emissions and emissions reductions for 2000 and 2010, the
two years that were modeled in both analyses. The figures present emissions data for the
four pollutants presented in Figures 2-1 through 2-4: VOC, NOX, SO2, and primary PM2 5.
As Figure 2-5 shows, the Second Prospective Analysis estimates slightly higher 2000
emissions in the without-CAAA scenario, and slightly lower emissions in the with-CAAA
scenario. VOC and primary PM2 5 emissions estimates are approximately the same in
both analyses, but the Second Prospective estimates reductions in combined emissions of
NOX and SO2 of about three million tons more than in the First Prospective. As noted
above, most of the difference in SO2 emissions reductions is attributable to SO2 controls
from CAIR, but there are also substantial additional reductions attributable to reduced
2-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 2-5.
fuel sulfur content regulations. The difference in NOX emissions reductions is due
primarily to differences in the onroad and nonroad engine and EGU rules included in the
Second Prospective, but also to corrections made in the Second Prospective to more
accurately characterize the impact of the NOX SIP Call provisions for electric generating
units.
In Figure 2-6, the difference between emissions estimates in the First and Second
Prospective Analyses is much more noticeable. Although the without-CAAA scenario
emissions estimates for VOC, NOx, and SO2 are virtually identical for the two analyses,
estimates ofwith-CAAA emissions of these pollutants are all substantially lower in the
Second Prospective Analysis than in the First Prospective, yielding a difference in
cumulative emissions reductions of about 15 million tons. As discussed above, the
Second Prospective estimates much larger emissions reductions primarily because it
accounts for a number of major control programs that were not yet in place when the last
analysis was published.
FIRST AND SECOND PROSPECTIVE 2000 EMISSIONS AND EMISSIONS REDUCTIONS
(EXCLUDING CO AND PM10)
100
s
t
o
2-15
-------
FIGURE 2-6 .
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIRST AND SECOND PROSPECTIVE 2010 EMISSIONS AND EMISSIONS REDUCTIONS
(EXCLUDING CO AND PM10)
100
t
o
E
LU
UNCERTAINTY IN EMISSIONS ESTIMATES
Table 2-6 lists several sources of uncertainty associated with generating the emissions
estimates discussed in this chapter, as well as the expected direction of bias introduced by
each uncertainty (if known), and the relative significance of each uncertainty in the
overall 812 benefits analysis. These uncertainty sources are organized by the three
factors that drive our results: identifying base-year emissions, forecasting growth in
emissions-related activity, and modeling emissions controls in future years.
UNCERTAINTIES RELATED TO BASE-YEAR EMISSIONS
We estimated emissions from onroad motor vehicles, nonroad engines, and area sources
at the county level, since these source categories are generally not tied to a specific
location. Accordingly, our estimates of the spatial location of these emissions are less
precise than for EGUs and industrial point sources. This uncertainty affects our ability to
model changes in air quality associated with emissions reductions attributed to the
CAAA. However, we expect that this uncertainty has a minor impact on the overall net
benefit projections of the analysis.
A potentially major factor contributing to uncertainty in emissions estimates is our
specification of the without-CAAA scenario. The Project Team tested the influence of an
alternative scenario specification by first developing a with-CAAA scenario using
continuous CEM data available on EPA's Clean Air Markets website.14 Working from
this scenario as a base emissions estimate for each EGU, we estimated EGU data for the
M U.S. Environmental Protection Agency. Clean Air Markets - Data and Maps
Accessed March 2009.
2-16
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
without-CAAA scenario using an alternative counterfactual approach based on work done
by Dr. A. Denny Ellerman of Massachusetts Institute of Technology.15 The with-CAAA
results using the alternative EGU data appear very similar to the results using the IPM
EGU data, but air quality difference maps indicate that overall PM2 5 exposures are
slightly lower using the CEM data for the with-CAAA scenario in 2000, and PM2s
exposures are substantially higher using the data derived using the Ellerman
counterfactual method for the without-CAAA scenario compared to the corresponding
core scenarios.
These exposure differences carry over into benefits calculations. The health benefits of
the CAAA in 2000 arrived at using the alternative EGU emissions are approximately 50
percent greater than the benefits in the 2000 core scenario. For the alternative EGU
emissions scenarios, the substantial, 50 percent difference in air quality outcomes and
benefits results appears to be derived from our construction of a substantially different
without-CAAA scenario. The original motivation of the analysis was concern that the
spatial pattern of emissions for the with-CAAA scenario for 2000 predicted by an IPM run
for a historical year differed from the spatial pattern observed in the emissions monitor
data for the same year. The analysis illustrated that the difference in benefits results is
instead due primarily to differences in the without-CAAA scenario among the two
alternative scenario specifications. Not surprisingly, uncertainty in estimating a
counterfactual scenario is much larger than uncertainty in estimating the factual case, at
least for the EGU sector.
UNCERTAINTIES RELATED TO GROWTH FACTORS
When projecting future growth in economic activity, even the most thorough projection
model must tolerate a high amount of uncertainty. The factors we used to model growth
in this analysis reflect uncertainty both in the economic activity forecasted and in how
this activity translates into emissions of criteria pollutants. For example, because the
AEO 2005 economic growth projection predates the recent economic downturn, it is
possible that we overestimate emissions in both the with-CAAA and without-CAAA
scenarios. However, because we use the same growth factors to project emissions under
the with-CAAA and without-CAAA scenarios, this source of uncertainty probably has a
minor effect on our overall net benefits estimates. In addition, we considered projecting
emissions under high-growth and low-growth AEO projection scenarios, but we did not
find sufficient variation in our conclusions to justify such an analysis. For these reasons,
we do not believe this is a significant factor in our results.
15 Dr. A. Denny Ellerman's approach relies on multiplying a "baseline" pre-Title IV emissions rate by 2001 CEM heat input
observations for each electric generating unit.
2-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Similarly, our projected emissions from on-road motor vehicles are based on vehicle fleet
compositions included in the MOBILE6.2 model. Any change in fuel prices that might
cause a shift away from low-fuel-efficiency vehicles could cause us to overestimate
emissions from this sector. However, we expect that the impact of this uncertainty on our
estimate of net benefits is minor.
UNCERTAINTIES RELATED TO EMISSIONS CONTROL MODELING
When modeling the with-CAAA scenario, we incorporated the effects of rules
promulgated through September 2005. Accordingly, we did not fully account for rules
promulgated since that time, such as the revised NAAQS for lead, and we modeled
reductions from rules that have since been vacated, like the Clean Air Mercury Rule
(CAMR) and the Clean Air Interstate Rule (CAIR), though CAIR has since been
remanded. We estimated that CAMR would have only a modest impact on the pollutants
we examined in this analysis, since mercury controls do not have large co-control benefits
with other pollutants. However, our analysis projects that CAIR would have a large
impact on NOX and SO2 emissions at EGUs in 2010 and 2020. Ultimately, a new rule
will be promulgated to replace CAIR, and the emissions reductions, compliance costs,
and locations of emissions reductions could all be different from what we modeled in this
analysis. As a result, it is unclear whether our analysis overestimates or underestimates
the net benefits of CAAA provisions on EGU emissions.
Estimates of emissions of volatile organic compounds are also a source of uncertainty
because VOCs can be emitted through fuel combustion—like SO2 and NOX—as well as
evaporation of volatile materials. Because evaporation rates depend largely on
temperature, our estimates of future VOC emissions are influenced by the inherent
difficulty of predicting future temperatures. The analysis uses projections of average
daily minimum and maximum temperatures in order to predict average VOC emissions,
but the resulting estimates do not adequately capture the variability of such emissions.
The likely significance of this uncertainty, in terms of its impact on the overall net
benefits estimated in this analysis, is probably minor.
Our future-year control assumptions are also a source of uncertainty. The flexibility
allowed by the CAAA in achieving air quality standard target emission levels allows for
emissions control schemes that may differ significantly from the controls modeled in this
analysis. This is particularly true in the case of reductions needed for NAAQS
compliance for which we have not identified a specific sector target. This analysis treats
those reductions as if they come from area sources, but they could come from any of the
five source categories we consider. We are not able to determine the direction of any
possible bias caused by this uncertainty, but we do not expect it to have a major effect on
our net benefits estimate.
2-18
-------
TABLE 2-6.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
KEY UNCERTAINTIES ASSOCIATED WITH EMISSIONS ESTIMATION
POTENTIAL SOURCE OF ERROR
DIRECTION OF POTENTIAL BIAS FOR
NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
Uncertainties Related To Base-Year Emissions
Uncertainties in modeling a
counterfactual emissions scenario.
Estimating ECU emissions using an
alternate counterfactual
projection approach yielded
increases in air quality impacts and
health benefits of 50% relative to
the core scenario's IPM-generated
estimates.
Underestimate. The IPM-based
counterfactual generated
substantially lower benefits than
the alternative counterfactual
scenario specification we tested,
which was based on published and
readily replicated methodologies.
It is possible, however, that other
counterfactual specifications
would yield lower benefits. It is
also possible that the direction of
effect might be different for other
pollutant source categories where
this is no accepted basis to
generate an alternative
counterfactual scenario estimate.
Potentially major. Analysis
confirmed that IPM performs well
when estimating with-CAAA
emissions, but also highlighted a
high degree of uncertainty in
estimating counterfactual
emissions. Similar uncertainties
exist for emissions from other
emitting sectors. There is no clear
way, however, to determine what
approach to estimating
counterfactual emissions is
superior.
Uncertainties in biogenic emissions
inputs increase uncertainty in the
air quality modeling estimates.
Uncertainties in biogenic emissions
may be large (± 80%). The
biogenic inputs affect the
emissions-based VOC/NOx ratio
and, therefore, potentially affect
the response of the modeling
system to emissions changes.
Unable to determine based on
current information. The biogenic
emissions change overall
reactivity, leading to either an
underestimate or overestimate of
the model's response to emission
reductions.
Probably minor. Impacts for ozone
and PM2.s results. Both oxidation
potential and secondary organic
aerosol formation could influence
PM2.5 formation significantly.
However, biogenic emissions are
assumed to be unaffected by the
CAAA, so this uncertainty should
not significantly affect net
benefits. Furthermore, ozone
benefits contribute only minimally
to net benefit projections in this
study.
Emissions estimated at the county
level (e.g., low-level 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.
Unable to determine based on
current information.
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.
2-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF ERROR
DIRECTION OF POTENTIAL BIAS FOR
NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
Uncertainties Related To Growth Factors
Economic growth factors used to
project emissions are an indicator
of future economic activity. These
growth factors reflect uncertainty
in economic forecasting as well as
uncertainty in the link to
emissions. IPM projections may be
reasonable regionally but may
introduce significant biases locally.
Also, the Annual Energy Outlook
2005 growth factors do not reflect
the recent economic downturn or
the volatility in fuel prices since
the fall of 2005.
Unable to determine based on
current information.
Potentially major. The same set of
growth factors are used to project
emissions under both the Without-
CAAA and With-CAAA scenarios,
mitigating to some extent the
potential for significant errors in
estimating differences in
emissions. Some specific locations
may be more significantly
influenced. We estimated gross
benefits using AEO low-growth and
high-growth scenarios and found
differences of ±20%. However, due
to nonlinearities in the benefits
estimation model, we could not
reliably determine in what
direction over- or underestimating
growth might bias net benefits
estimates.
The on-road source emissions
projections reflect MOBILE6.2 data
on the composition of the vehicle
fleet. If recent volatility in fuel
prices persists or if fuel prices rise
significantly (like they did in 2007
and 2008), the motor vehicle fleet
may include more smaller, lower-
emitting automobiles and fewer
small trucks (e.g., SUVs).
Overestimate
Probably minor. Overall, fuel
prices affect fleet composition at
the margin, and we expect changes
in fleet composition to occur
gradually over long periods,
suggesting that any effect would
take several years to fully
manifest.
Uncertainties Related To Emissions Control Modeling
The With-CAAA scenario includes
implementation of the Clean Air
Mercury Rule (CAMR), which has
been vacated, and Clean Air
Interstate Rule (CAIR), which was
vacated but has since been
remanded.
Unable to determine based on
current information.
Potentially major. Significance in
2020 will depend on the speed and
effectiveness of implementing
potential alternatives to CAIR and
CAMR. In some areas, emissions
reductions are expected to be
overestimated, but in other areas,
NOX inhibition of ozone leads to
underestimates of ozone benefits
(e.g., some urban centers).
VOC emissions are dependent on
evaporation, and future patterns
of temperature are difficult to
predict.
Underestimate. Higher
temperatures in the future are
more likely than lower
temperatures because of climate
change, and higher temperature
would lead to more emissions in
the without-CAAA case but
controls would keep the with-CAAA
emissions roughly constant.
Probably minor. The analysis uses
meteorological data from 2002 to
characterize temperatures during
the 30-year period from 1990 to
2020. An acceleration of climate
change (warming) could increase
emissions but the increase relative
to 2002 levels would not likely be
significant.
2-20
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF ERROR
DIRECTION OF POTENTIAL BIAS FOR
NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
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
With-CAAA scenarios because of
more stringent evaporative
controls that are in place by 2000
and 2010.
Uncertainties in the stringency,
scope, timing, and effectiveness of
With-CAAA controls included in
projection scenarios.
Unable to determine based on
current information.
Probably minor. Future controls
could be more or less stringent,
widely applicable, or effective
than projected. Timing of
emissions reductions may also be
affected.
The location of the emissions
reductions achieved from
unidentified measures is uncertain.
We currently treat these
reductions as if they are achieved
from non-point sources, but this
may not be correct in all cases.
Unable to determine based on
current information.
Probably minor. Impacts from
these uncertainties would be
localized and would not
significantly change the overall net
benefit estimate.
2-21
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 3 - DIRECT COSTS
Scenario Development
Sector Modeling
Emissions
I
Air Quality Modeling
Health
Welfare
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 impact, initially experienced
through the direct costs imposed by
regulations promulgated under the
amendments, will also be seen in
patterns of industrial production,
research and development, capital
investment, productivity,
employment, and consumption. The
purpose of the analysis summarized
in this chapter is to estimate the
incremental change in direct annual
compliance costs from 1990 to 2020
that are attributable to the 1990
Clean Air Act Amendments.
As a measure of the direct
expenditures associated with CAAA
compliance, the estimates presented
here represent a key stand-alone output of the Second Prospective Analysis. In addition,
we use the direct cost estimates presented in this chapter to generated estimates of
CAAA-related private costs that will serve as inputs in the computable general
equilibrium (CGE) model used to estimate the net social costs of the CAAA on the
economy as a whole.16 Use of a CGE model allows us to estimate how compliance
costs—along with expected benefits of the CAAA, such as increased labor supply—
Economic Valuation
Benefit-Cost Comparison
16 Private costs differ from the direct cost estimates presented in this chapter in two important ways: (1) they reflect private
interest rates rather than the 5 percent social discount rate used throughout this report and (2) they reflect transfers (e.g.,
excise taxes on fuel) not included in our direct cost estimates.
3-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
have a net impact on social welfare through interactions with labor markets and other
areas of the economy. Further discussion of the CGE modeling conducted to estimate the
impacts of the CAAA on net social welfare is presented in Chapter 8.
This chapter consists of four sections. The first section summarizes our approach to
estimating direct compliance costs. In the second section we present the results of the cost
analysis. In the third section, we discuss how cost estimates in the Second Prospective
Analysis differ from those generated for the First Prospective Analysis. We conclude the
chapter with a discussion of the major analytic uncertainties, including a summary of the
results of quantitative sensitivity tests of key data and assumptions.
OVERVIEW OF APPROACH
The scope of this analysis is to estimate the incremental direct costs for all criteria and
hazardous air pollutant regulations issued under CAAA programs. Our approach to
estimating the direct costs of CAAA compliance is closely integrated with our estimates
of emissions reductions attributable to the amendments. In general, our analysis of
compliance costs is driven by the results of our analysis of CAAA-related emissions
reductions, and in some cases, costs and emissions reductions are measured concurrently.
As with the emissions analysis presented in the previous chapter, we modeled CAAA
compliance costs in 2000, 2010, and 2020 by comparing the costs of air pollution
abatement in two scenarios:
• An historical "with-CAAA" scenario control case that reflects expected or likely
future measures implemented since 1990 to comply with rules promulgated
through September 2005; and
• A counterfactual "without-CAAA" scenario baseline case that freezes the scope
and stringency of emissions controls at their 1990 levels, while allowing for
changes in emissions attributable to economic and population growth.17
In addition, we also estimated costs separately for five major source categories: utilities,
or electricity generating units (EGUs); non-EGU industrial point sources; onroad motor
vehicles; nonroad engines/vehicles; and area sources. Table 2-1 gives examples of
emissions sources for each of the six categories examined in this analysis. Additionally,
the cost analysis considers the costs of local controls required to achieve further progress
with the 8-hour Ozone NAAQS and the PM2 5 NAAQS as a separate category. Another
difference between the emissions analysis and the direct cost analysis discussed in this
chapter is that, whereas the emissions analysis considered emissions of six major criteria
pollutants (VOCs, NOX, SO2, CO, PM10, and PM2 5) and one other pollutant which is not
currently regulated under the CAAA in any form (NH3), the cost analysis addresses
CAAA provisions issued to control emissions of both criteria pollutants and hazardous air
pollutants (HAPs).18
17 A full list of the regulations incorporated in the with-CAAA scenario is presented in Table 2-3.
18 Except to the extent they are co-controlled by VOC limits or other measures focused on criteria pollutants, reductions in
emissions of hazardous air pollutants were omitted because our benefits analysis focuses on the effect of criteria
3-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
We estimated direct compliance costs in each source category using one of two
approaches:
1. Cost Estimates Based on Unit Costs - Costs were estimated by collecting
information on the costs associated with specific control measures required by
CAAA regulations, or costs were calculated using estimates of the average cost
per ton of pollutant emission reduced.
2. Cost Estimates Based on Optimization - Costs were estimated concurrently with
emissions estimation through a cost minimizing algorithm that modeled
attainment with specified emissions reduction targets. This approach was used
for electric generating units, for example, where costs and emissions outcomes
are outputs of the Integrated Planning Model.
COST ESTIMATES BASED ON UNIT COSTS
To estimate the cost of compliance CAAA regulations for most source categories, we
obtained unit costs of control devices and other measures from various sources. For costs
related to the 1-hour Ozone and PMi0 National Ambient Air Quality Standards
(NAAQS), we used cost data from EPA's AirControlNET database. AirControlNET
links detailed data on control technologies and pollution prevention measures with EPA's
National Emissions Inventory (NEI) to compute the costs associated with source- and
pollutant-specific emission reductions. To calculate the cost of emissions controls on
nonroad engines and vehicles, we multiplied unit cost estimates by estimates of
vehicle/equipment sales and fuel consumption from the 2004 edition of EPA's
NONROAD model. The NONROAD model was also used to estimate CAAA-related
emissions reductions in this sector, and direct cost estimates were developed consistent
with those results. For these nonroad engine and fuel rules, as well as for controls
required under other parts of the CAAA, we obtained unit cost estimates from EPA's
regulatory impact analyses (RIAs) as well as analyses commissioned by other
organizations, such as the Ozone Transport Commission and the California Air Resources
Board. Additional details on the specific data sources used to estimate unit costs for each
source category are provided in the Second Prospective Cost Report.19
pollutants. Benefits of HAP emissions reductions are discussed in the context of a limited case study, however, in Chapter 5
of this document. In addition, no CAAA emissions control measures are currently targeted to control NH3 emissions, so no
costs for NH3 control are included in our overall CAAA cost estimates.
19 See the report, Direct Cost Estimates for the Clean Air Act Second Section 812 Prospective Analysis. Available at
www.epa.gov/oar/sect812.
3-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
COST ESTIMATES BASED ON OPTIMIZATION
We estimated control costs for EGUs using EPA's Integrated Planning Model (IPM),
which determines the utility sector's least-cost strategy for meeting energy and peak
demand requirements over a specified period of time, accounting for CAAA-mandated
emissions caps. In the process of estimating the SO2 and NOX emissions that we
discussed in the previous chapter, IPM also produced cost estimates for NOX, SO2, and
mercury controls at EGUs.
We also used a least-cost optimization process to estimate the costs of local controls
required to achieve further progress toward and, ultimately, approximate attainment of
the 8-hour Ozone NAAQS. For each designated nonattainment area, we first modeled the
application of reasonably available control technology (RACT) and inspection and
maintenance (I/M) programs. Then, in areas where further emission reductions were
necessary, a least-cost algorithm was used to identify and apply the control measures to
meet progress and attainment requirements.20
Table 3-1 summarizes the cost estimation methods that we used for each source category,
organized by major rules within each category.
ADDITIONAL COST ESTIMATION CONSIDERATIONS
In addition to the general cost estimation methods described above, we also considered
additional factors when estimating CAAA compliance costs, such as how to account for
cost savings from "learning by doing," how to represent the annual costs of control
measures requiring initial capital investment, and how to estimate the costs of required
emissions reductions for which control measures have not yet been identified.
Learning - A significant body of literature suggests that the per unit cost of producing or
using a given technology declines as experience with that technology increases over
time.21 The mechanism through which these reductions occur is not well understood, as
decreases in costs may reflect several different effects, including returns to research and
development, productivity spillovers from outside an industry, economies of scale, or
efficiency improvements associated with increased experience with a given technology
(i.e., learning-curve impacts). Given the multitude of factors that may lead to cost
reductions over time, it is unclear whether such reductions should be modeled as
learning-curve effects or as some other form of technological change. Nordhaus (2008)
suggests that it is difficult to distinguish learning-curve effects from exogenous
20 For PM NAAQS compliance, an optimization approach was not possible, because target emissions reductions were not
available for each non-attainment area. Instead, we developed a model SIP for all PM nonattainment areas, and estimated
costs for those measures in the model SIP for each nonattainment area.
21 These studies include John M. Dutton and Annie Thomas, "Treating Progress Functions as a Managerial Opportunity,"
Academy of Management Review, 1984, Vol. 9, No. 2, 235-247; Dennis Epple, Linda Argote, and Rukmini Devadas,
"Organizational Learning Curves: A Method for Investigating Intra-plant Transfer of Knowledge Acquired Through Learning by
Doing," Organizational Science, Vol. 2, No. 1, February 1991; International Energy Agency, Experience Curves for Energy
Technology Policy, 2000; and Paul L. Joskow and Nancy L. Rose, "The Effects of Technological Change, Experience, and
Environmental Regulation on the Construction Cost of Coal-Burning Generating Units," RAND Journal of Economics, Vol. 16,
Issue 1, 1-27, 1985.
3-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 3-1.
technological change and that learning effects, as estimated separately from technological
change, will typically be overestimated. Nevertheless, the most detailed peer-reviewed
empirical studies examining these cost reductions quantify a "learning rate" for different
technologies and industries that represents the percentage reduction in costs associated
with each doubling in the cumulative production of a technology. Based on the strength
of the evidence in this literature, we incorporated the concept of the learning effect into
our assessment of CAAA costs.
COST ESTIMATION METHODS BY SOURCE CATEGORY AND RULE (WHERE
APPLICABLE)
SOURCE CATEGORY
COST ESTIMATION METHOD
EGUs
IPM Least-cost optimization
Non-EGU Industrial Point Sources
Ozone Transport Commission State Model Rules
(NOX/VOC):
NOX SIP Call:
MACT Rules:
Refinery Cases & Settlements:
1-hour Ozone NAAQS:
Federal Rules (RACT, Control Technique Guidelines,
National VOC Rules):
Additional Measures:
PM10 SIP Measures:
Ozone Transport Commission -sponsored 2001 analysis
AirControlNET
EPA cost estimates (from 1987-1998)
AirControlNET
Cost/ton from 1st Prospective
AirControlNET Least Cost Module
SIP control cost estimates;
AirControlNET
Onroad Engines and Fuels
Title I NAAQS Tailpipe & Evaporative Control
Standards:
California and National LEV:
Fuels:
I/M Programs:
EPA RIA unit costs
California Air Resources Board (CARB) unit cost
estimates
Unit costs from First Prospective Analysis, EPA RIAs,
CARB (for California standards)
Costs based on information from current I/M programs
Nonroad Engines and Fuels
EPA RIA Unit Costs applied to sales and fuel
consumption data provided by the NONROAD model,
consistent with growth projections used to estimate
emissions
Area Sources
Ozone Transport Commission State Model Rules
(NOX/VOC):
1-hour Ozone NAAQS:
RACT & Control Technique Guidelines:
Additional Measures:
Ozone Transport Commission-sponsored 2001 analysis
Cost/ton from 1st Prospective
AirControlNET
Local Controls
8-hour Ozone NAAQS:
RACT a I/M:
Additional (Identified) Measures:
Unidentified Measures:
PM2.5 NAAQS:
AirControlNET
AirControlNET using a least-cost algorithm
Assumed $15,000/ton
Model SIP approach with AirControlNET unit costs
Note:
Unit costs taken from earlier EPA analyses are inflated to 2006$ and adjusted to account for cost savings from
learning curve impacts.
Some cost estimates for onroad and nonroad engines and fuel also reflect costs and/or savings from changes in
fuel economy. These costs and savings are estimated using AEO 2005 fuel price projections.
3-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Where possible, we based our learning curve adjustments on learning rates presented in
the empirical literature. For some sectors, however, empirical estimates of learning rates
were not available. We identified learning rate estimates for SC>2 and NOX control
technologies in the EGU sector and in the onroad vehicle sector, where we used learning
rates for vehicle production to estimate the impact of learning on motor vehicle engine
controls. For other technologies and industries affected by the amendments, we applied a
default learning rate of 10 percent, consistent with the recommendation of the Council
that advised EPA on this study.22'23
Cost Accounting - The costs presented in this analysis are expressed as total annualized
costs (TAC) in 2000, 2010, and 2020. Annualized costs include both operation and
maintenance (O&M) costs and, for CAAA provisions that require investment in pollution
control equipment, capital investment costs. In order to make appropriate comparisons of
costs in 2000, 2010, and 2020, we annualized these investment costs over the expected
life of the control equipment, rather than assigning total capital investment costs to the
year in which the investment is expected to be made. We applied a discount rate of five
percent to annualize capital costs over an estimated equipment life.24 These annualized
capital costs, combined with the annual O&M costs for a given pollution control measure,
make up the total annualized cost estimates that we present for the three target years.
Because some control measures require more capital investment than others, the degree to
which our discount rate assumption affects our cost estimates varies by source category.
For CAAA-related rules that affect fuel economy, we also incorporate fuel savings or
losses into our cost estimates. Where possible, we estimate the value of these benefits or
costs based on fuel price projections presented in the Energy Information
Administration's Annual Energy Outlook 2005 (AEO 2005). In addition, for rules that
affect the fuel economy of an engine over a period of several years, we estimate these
benefits or costs as the present value of the fuel economy impacts realized over the entire
life of the engine.
Local Controls for NAAQS Compliance - When estimating the costs of compliance with
the 8-Hour Ozone and PM2 5 NAAQS, we first estimated the cost of applying known and
commercially available control technologies in nonattainment areas. We limited the
application of these known controls to those with an estimated cost not exceeding
$15,000 per ton for PM and ozone precursors (i.e., SO2, NOX, and VOCs). The rationale
for incorporating this threshold into the analysis is that controls more costly than $15,000
The Council recommended that we apply a default learning rate of 5 to 10 percent to sectors for which no empirical data
are available. We chose 10 percent as a default learning rate because this value is more consistent with the learning rates
presented in the empirical literature than the low end of the Council's recommended range.
23 The Project Team makes no learning curve adjustments for motor vehicle inspection and maintenance programs. Because
most states either run centralized inspection centers themselves or regulate the fees charged by decentralized inspection
centers, it is unclear whether the learning curve impacts for I&M programs would be significant.
24 Note that the discount rate we use to annualize capital investment costs is distinct from the discount rate used to
calculate the total net present value of costs and benefits incurred through the full 1990 to 2020 study period. The net
present value of costs and benefits is examined separately in Chapter 7 where we compare total costs to total benefits.
3-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
per ton may not be cost effective. Thus, local air quality agencies would seek reductions
from other (unidentified) control measures. This is roughly consistent with the practice
of the South Coast Air Quality Management District (SCAQMD 2006) in California,
which attempts to identify viable alternatives for any control requirements with an
estimated cost exceeding $16,500 per ton. When costs are above this threshold, the
SCAQMD also conducts more detailed cost-effectiveness and economic impact analyses
of the controls.
For areas projected to remain in nonattainment with the 8-Hour Ozone NAAQS with
identified controls, we estimated the costs associated with reducing emissions using
additional controls not yet identified. To estimate the cost of these unidentified controls,
we assumed that the cost of implementing these measures is $15,000 per ton of pollutant
reduced, consistent with the cost threshold for identified controls.
DIRECT COMPLIANCE COST RESULTS
In this section we summarize the compliance cost analysis results by source category. As
noted above, the control measures included in this analysis are consistent with our
assumptions in the emissions analysis and reflect any post-1990 regulations promulgated
(or reasonably anticipated, such as controls to meet RFP requirements) after passage of
the 1990 CAAA. In general, the emissions analysis and this cost analysis reflect all of the
regulations that were promulgated before September 2005. Similar to the emissions
projection analysis, regulations promulgated after September 2005 (e.g., the revised Lead
NAAQS) are not reflected in this report, in an effort to make the costs and benefits
analyses as consistent as possible.
Table 3-2 summarizes the estimated costs of the CAAA by sector for the three analysis
years: 2000, 2010 and 2020. The table shows that the direct compliance costs in 2000 are
estimated to be approximately $20 billion and that these costs are dominated by the costs
of motor vehicle-related provisions of the CAAA as well as MACT standards and electric
utility controls. The major components of motor vehicle-related control costs in 2000 are
for emission standards, fuel standards, and vehicle emission inspection programs in
nonattainment areas. Motor vehicle emissions standard costs in 2000 are primarily for
low emission vehicle programs, Tier 1 tailpipe standards, and on-board diagnostics.
Prominent motor vehicle fuel control programs in 2000 include Federal and California
reformulated gasoline. These two reformulated gasoline programs are focused primarily
in serious, severe and extreme 1-hour ozone NAAQS nonattainment areas.
Table 3-2 shows that the estimated costs of complying with 1990 CAAA provisions are
expected to more than double between 2000 and 2010 as areas develop and implement 8-
hour ozone and PM2 5 NAAQS State Implementation Plans (SIPs). One of the major
components of CAAA compliance costs in 2010 is the estimated cost to achieve
sufficient reductions of ozone precursor emissions to demonstrate 8-hour ozone NAAQS
attainment. As noted above, we estimated 8-hour ozone compliance costs in two phases:
first, we estimated the cost of applying known and commercially available control
technologies in nonattainment areas; second, we estimated the costs associated with
additional emissions reductions required to reach NAAQS attainment using controls not
3-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 3-2.
yet identified, at an assumed cost of $15,000 per ton. There is considerable uncertainty in
this element of the cost analysis because it is unclear how individual areas will approach
this issue. Because of the significant degree of uncertainty associated with estimating the
costs of unidentified controls, this component of the cost analysis is reported separately in
Table 3-2.
SUMMARY OF 1990 CAAA COMPLIANCE COSTS BY SECTOR
SOURCE CATEGORY
Electric Utilities
Non-EGU Industrial Point Sources
NOX SIP Call
MACT
National VOC Rules, RACT, and New CTGs
Refinery Settlements
1 -Hour Ozone SIP Measures
PM10 SIP Measures
Onroad Vehicles and Fuels
Motor Vehicle Emission Standards
California and National LEV
Fuels
Motor Vehicle I /M programs
Nonroad Vehicles and Fuels
Nonroad Engines/Vehicle Standards
Fuels
Area Sources
RACT and New CTGs
Ozone Transport Commission Model Rules
1 -Hour Ozone NAAQS
Local Controls
8-Hour Ozone NAAQS
PM2.5 NAAQS
Clean Air Visibility Rule
Sub-Total Excluding Unidentified Measures
ANNUAL COST (MILLION 2006$)
2000
$1,370
$3,130
$0
$1,500
$439
$0
$1,030
$163
$14,400
$4,400
$562
$4,820
$4,630
$298
$298
$0
$663
$446
$134
$82
$0
$0
$0
$0
$19,900
2010
$6,640
$5,190
$134
$3,010
$464
$295
$1,130
$152
$25,700
$7,650
$2,030
$9,830
$6,250
$359
$219
$140
$693
$442
$181
$70
$5,260
$4,270
$977
$0
$43,900
2020
$10,400
$5,140
$133
$2,920
$534
$324
$1 ,090
$146
$28,300
$7,760
$2,090
$11,200
$7,260
$1,150
$320
$831
$766
$490
$212
$64
$6,180
$4,390
$687
$1,100
$52,000
Additional Estimated Costs for Unidentified Controls for 8-Hour Ozone Compliance
Non-California areas
California areas
TOTAL
$19,900
$8,700
$318
$53,000
$8,500
$5,030
$65,500
Note: All values are rounded to no more than three significant digits.
3-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
The growth in costs between 2000 and 2020 partially reflects population growth during
this period and the corresponding increase in emissions-generating activity (e.g.,
increased vehicle miles traveled). Normalized for population growth, annual costs
increase from approximately $70 per capita in 2000 to $170 per capita in 2010 and $190
per capita in 2020. These results suggest that annual costs per capita grow by
approximately 170 percent between 2000 and 2020, whereas annual costs (not normalized
for population) grow by approximately 230 percent during this period.
COMPARISON OF COST ESTIMATES WITH THE FIRST PROSPECTIVE ANALYSIS
In many areas, cost estimation methods in the Second Prospective Analysis were identical
to those in the First Prospective, even to the point of using the same unit costs (adjusted
for inflation). In general, the Second Prospective improves on the First Prospective by
using more current cost estimates (where available) and more advanced least-cost
optimization tools. In addition, a major methodological innovation included in the
Second Prospective is the adjustment of compliance costs to account for the learning
curve effects of increased experience with pollution control measures.
Figure 3-1 shows the estimated compliance costs in 2000 and 2010 from the First and
Second Prospective Analyses, organized by source category. Overall, the year 2000 cost
estimate presented in Table 3-2 is considerably lower than the corresponding cost
estimate in the First Prospective ($27.6 billion), while the 2010 cost estimate presented in
Table 3-2 is higher than the corresponding First Prospective estimate ($37.8 billion).
Costs for electric utilities and area sources are significantly lower than were estimated in
the First Prospective. The significant difference for utilities likely reflects differences in
assumptions about the cost of obtaining low-sulfur coal from the Powder River Basin
(PRB) in Wyoming. Although the Project Team was aware of the downward trend in
PRB coal costs when the First Prospective was completed, this effect was not fully
addressed in the data and models available at the time of the First Prospective study.
It is useful to note that the Second Prospective's $1.37 billion estimate for EGU
compliance cost in 2000, which represent the pre-CAIR Title IV program requirements,
fits well within the range of costs estimated in a series of ex-post econometric studies of
compliance cost, which yield results of costs in 2000 of $1 to $1.4 billion.25 In addition,
the National Acid Precipitation Assessment Program's (NAPAP) 2005 assessment of the
Clean Air Act Title IV requirements provides another basis for evaluating the
reasonableness of the EGU cost estimates presented in this report (NSTC 2005). The
2005 NAPAP assessment summarizes the findings of several economic studies that
estimated the cost of fully implementing the Title IV SO2 provisions. According to
25
See, for example, A Denny Ellerman, 2003, "Ex Post Evaluation of Tradable Permits: The U.S. S02 Cap-and-Trade Program,"
MIT Center for Energy and Environmental Policy Research Working Paper number WP-2003-003, available at:
web.mit.edu/ceepr/www/publications/workingpapers_2000_2004.html#2003. Ellerman cites two papers for these
estimates: Curtis P. Carlson, Dallas Burtraw, Maureen Cropper, and Karen Palmer, (2000) "S02 Control by Electric Utilities:
What are the Gains from Trade?" Journal of Political Economy, 108 (6):1292-1326; and A. Denny Ellerman, Paul L. Joskow,
Richard Schmalensee, Juan-Pablo Montero, and Elizabeth Bailey (2000). Markets for Clean Air: The U.S. Acid Rain Program.
Cambridge University Press.
3-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
NAPAP, these studies estimate annual costs ranging from $1.2 billion to $2.3 billion for
full implementation in 2010, but these estimates exclude the cost of CAIR, CAMR, and
some other regulations that are part of the Second Prospective estimate for 2010.26
Overall, the Second Prospective cost estimates for 2010 are higher than those estimated
for the First Prospective mainly because many federal motor vehicle control programs not
included in the First Prospective study with-CAAA scenario have been promulgated since
the First Prospective was completed. For the same reason, the Second Prospective cost
estimates are also higher for motor vehicles in 2000, though to a lesser degree. In
addition, cost estimates in the current analysis are higher than in the First Prospective
because they include the costs of meeting the 8 hour ozone, PM25 NAAQS and Clean Air
Visibility Rule requirements in 2010. In both 2000 and 2010, estimated costs at area
sources are higher in the First Prospective than in the Second Prospective, by roughly a
factor of three, even though estimated emissions reductions are roughly a factor of three
greater in the Second Prospective. This difference is due primarily to a much lower
estimated cost per ton to reduce PM2 5 emissions in the Second Prospective - on average,
cost per ton of PM25 reduced is approximately $2,000 in the Second Prospective, and was
almost $20,000 in the First Prospective. One reason for the reduction is that the controls
in the Second Prospective are better targeted at fine particulate control - controls in the
First Prospective were actually focused on sources of PMi0, with PM25 emissions
reductions as a co-benefit. In addition, we have learned that pre-2002 NEI emissions
estimates for PM2 5 were very uncertain, suggesting that perhaps the estimated PM2 5
emissions reductions in the First Prospective were understated.
26 The NAPAP assessment cites a range of $1 billion to $2 billion, in year 2000 dollars. Adjusting for inflation using the GDP
deflator, this range increases to $1.2 billion to $2.3 billion in year 2006 dollars.
3-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 3-1. FIRST AND SECOND PROSPECTIVE ANNUAL CAAA COMPLIANCE COSTS: 2000 AND
2010
$60,000
• Local Controls (Unidentified)
D Local Controls (Identified)
• Area Sources
D Industrial Point Sources
D Nonroad Vehicles and Fuels
• Onroad Vehicles and Fuels
D Electric Utilities
First Prospective cost estimates from U.S. EPA, The Benefits and Costs of the Clean air Act 1990
to 2010, EPA-410-R-99-001, November 1999.
UNCERTAINTY IN DIRECT COST ESTIMATES
In a broad analysis of prospective regulatory impacts it is not possible to verify the ac-
curacy of the full range of assumptions regarding changes in consumption patterns, input
costs, and technological innovation used to estimate costs in future scenarios. Moreover,
for many of the factors contributing to uncertainty, 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 characterized. In this section of the chapter, we first discuss several
quantitative sensitivity analyses undertaken to characterize the impact of key assumptions
on the ultimate cost analysis. The quantitative analyses presented below were chosen
either because the parameter in question was a topic of discussion in the Council's review
of the direct cost analysis or because we identified the parameter as potentially influential
and/or uncertain. We then conclude the chapter with a qualitative discussion of the
impact of both quantified and unquantified sources of uncertainty.
QUANTITATIVE SENSITIVITY TESTS
We performed four quantitative sensitivity tests to estimate the impact of alternate
assumptions on our overall cost estimates. These tests covered our assumptions
regarding the cost of unidentified controls, the composition of motor vehicle sales and
fleet fuel efficiency, the failure rate of I/M tests, and the default learning rate applied to
3-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
sectors for which we could not identify a rate in the empirical literature. The results of
these sensitivity tests on our 2020 cost estimates are presented in Table 3-3,27
Local Controls Analysis - Unidentified Controls
As indicated above, when estimating the cost of local controls required for further
progress with the 8-hour Ozone and PM25 NAAQS, we used a cost cap of $15,000 per
ton to estimate the costs of identified local controls and also applied a cost of $15,000 per
ton to unidentified controls. To assess the sensitivity of the local controls analysis to
changes in these values, we estimated the costs of local controls based on a $10,000-per-
ton cost cap for identified controls and a $10,000-per-ton estimated cost for unidentified
controls. As indicated in Table 3-3, this alternative approach would yield lower cost
estimates for both identified local controls and unidentified measures. The estimated costs
of identified controls decline when the $10,000 cap is applied because controls that cost
between $10,000 and $15,000 per ton are not implemented. In addition, although the
application of the $10,000 cost cap increases the emissions reductions to be achieved
through unidentified controls (relative to when the $15,000 cost cap is used), reducing the
cost of unidentified controls to $10,000 per ton more than offsets the costs associated
with these additional emissions reductions. Based on preliminary analyses conducted
early in the development of the direct cost estimates, we found that in general higher
thresholds do not change the emissions reductions to be achieved by unidentified
controls, because few identified controls have a cost per ton higher than the $15,000
threshold used in the analysis. Accordingly, the major effect of increasing the cost cap
would be to increase the estimated cost of reductions achieved by unidentified controls,
whose cost is estimated based on the dollar per ton cap.
Composition of Motor Vehicle Sales and Fleet Fuel Efficiency
Our analysis of the costs associated with motor vehicle tailpipe and fuel rules is based on
sales and fuel efficiency projections from the 2005 version of DOE's Annual Energy
Outlook. Since the release of AEO 2005, however, fuel prices have been more volatile
than in previous years, leading many consumers to shift to more fuel efficient vehicles,
and the Department of Transportation revised the Federal Corporate Average Fuel
Economy (CAFE) standards. Given these developments, AEO 2008 projects that
passenger cars will make up a greater portion of light-duty vehicle sales in 2010 and 2020
than is projected by AEO 2005. AEO 2008 also assumes that the light-duty vehicle fleet
will be nearly 15 percent more fuel efficient relative to the projections in AEO 2005. To
assess the extent to which our cost estimates for the on-road sector would change under
the alternative AEO 2008 assumptions, we estimated the cost of motor vehicle tailpipe
and fuel rules for both the 2010 and 2020 target years based on the AEO 2008 data. As
indicated in Table 3-3, using AEO 2008 projections increases the estimated cost of motor
vehicle tailpipe standards and reduces the estimated cost of motor vehicle fuel rules in
2020. Although the alternative estimated cost of fuel rules is about 9 percent less than the
27 We present sensitivity test results for 2020 estimates because the differences between the primary cost estimates and the
alternative cost estimates discussed in this section are most pronounced in 2020.
3-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
primary estimate presented in Table 3-2, the reduction in estimated costs of both tailpipe
and fuel CAAA motor vehicle programs in aggregate is more modest, at 3.6 percent.28
Vehicle Inspection Failure Rate
Our estimates of the repair costs associated with motor vehicle I&M programs employed
program- and year-specific inspection failure rates derived from 2003 and 2004 data for
Wisconsin I&M programs. In its June 2007 review of the Draft Direct Cost Report, the
Council noted that a 2001 National Research Council report referenced a failure rate
about one-seventh the value derived from the Wisconsin data.29 To assess the sensitivity
of the I&M cost analysis to the assumed failure rate for annual dynamometer-based
programs, we developed alternative cost estimates for CAAA-mandated I&M programs
based on the failure rate reported by the NRC. We found that the estimated cost of these
programs declined by more than 40 percent when the alternative failure rates were used in
place of those supporting the Second Prospective Cost Report. In addition, as indicated
in Table 3-3, using these alternative values reduced total CAAA-related costs for the on-
road sector by about 12 percent in 2020. This suggests that the cost estimates for the on-
road sector are fairly sensitive to the assumed failure rate for I&M programs, in light of
the range of failure rates obtained from readily available data sources.
Default Learning Rate
As discussed above, we adjusted total program costs to account for "learning curve"
impacts (i.e., the extent to which the costs of a technology decline as experience with that
technology increases over time). Wherever possible, we employed technology- or
industry-specific learning rates obtained from the literature. Where industry-specific
learning rates were not readily available in the empirical literature, we applied a default
rate of 10 percent to the following technologies:
• Selective non-catalytic reduction at electric generating units (EGUs) (O&M costs
only);
• Activated carbon injection at EGUs;
• Motor vehicle fuel rules;
• Non-road engine and fuel rules;
• Non-EGU point source controls;
• Area source controls; and
• Local controls: EGU, non-EGU point source, and area source.
28 Note that in both our central case estimates and in our sensitivity analysis for fleet composition, the same fleet
composition is assumed in the with-CAAA and without-CAAA scenarios. It is likely that, as compliance costs increase, the
CAAA could have a significant effect on fleet composition, but our current analysis does not address that factor.
29 Committee on Vehicle Emission Inspection and Maintenance Programs, Board on Environmental Studies and Toxicology,
Transportation Research Board, National Research Council. Evaluating Vehicle Emissions Inspection and Maintenance
Programs. 2001.
3-13
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 3-3.
We tested the sensitivity of the cost analysis to the choice of a default learning rate by re-
estimating the total costs of the amendments using alternative default learning rates of 5
and 20 percent for the program areas listed above. The five percent default rate
represents the low end of the range recommended by the Council, while the 20 percent
value represents the central tendency presented in the peer-reviewed literature for several
technologies.30 For the sensitivity test, we did not adjust the cost estimates of program
areas where the empirical literature supplied specific and applicable learning rates. As
indicated in Table 3-3, the use of alternative default learning rates had only a small effect
on the estimated costs of the amendments in 2020. Using a five percent default learning
rate in 2020 increased the estimated cost of the amendments by 3.2 percent, while a 20
percent default learning rate reduced costs by six percent.
RESULTS OF QUANTITATIVE SENSITIVITY TESTS
PROVISION
Local Controls
(Identified and
Unidentified)
Motor Vehicle Costs
Motor Vehicle Costs
Total Costs (All
Source Categories)
Total Costs (All
Source Categories)
PRIMARY
ANNUAL COST
ESTIMATE FOR
2020 (BILLIONS
2006 $)
$20.39
$28.28
$28.28
$65.48
$65.48
STRATEGY FOR SENSITIVITY
ANALYSIS
$10,000/ton capon
identified controls and
$10,000/ton for unidentified
controls
Use AEO 2008 projections of
motor vehicle sales and fleet
fuel efficiency
Use Inspection Failure Rates
reported by the National
Research Council
Use alternate default
learning rate of 5 percent
Use alternate default
learning rate of 20 percent
ALTERNATIVE
2020 ESTIMATE
FROM SENSITIVITY
TEST (BILLIONS
2006 $)
$16.79
$27.25
$24.82
$67.60
$61.54
PERCENT CHANGE
FROM PRIMARY
COST ESTIMATE
-17.6%
-3.6%
-12.2%
3.2%
-6.0%
30 For an analysis of the learning rates estimated in the empirical literature, see John M. Dutton and Annie Thomas, "Treating
Progress Functions as a Managerial Opportunity," Academy of Management Review, Vol 9, No. 2, 1984.
3-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
QUALITATIVE ANALYSIS OF KEY FACTORS CONTRIBUTING TO UNCERTAINTY
In addition to the uncertainties outlined above, we identified several other areas of
uncertainty related to the direct compliance costs of the amendments that we did not
address quantitatively. These include the Project Team's projections of economic
activity, the impact of CAAA compliance on productivity, product quality degradation
resulting from the CAAA, the influence of technological innovation on CAAA
compliance costs, and the impact of input substitution on the costs of complying with the
amendments.
Economic Activity Projections. The cost of the amendments in 2010 and 2020 will
depend in large part on the future size and composition of the U.S. economy. If the AEO
2005 economic growth projections used to estimate emissions reductions in 2010 and
2020 underestimate or overestimate economic activity, we could likewise overestimate or
underestimate the costs of CAAA compliance. In addition, the particular composition of
economic output in 2010 and 2020 may deviate from the AEO 2005 projections, which
would also cause our cost projections to differ from the actual costs of the amendments.
Industrial Productivity. As stated in the introduction to this chapter, our cost estimates
represent the direct costs of the CAAA, i.e., the expected expenditures of regulated
facilities to comply with the amendments. Several peer-reviewed studies have suggested,
however, that the direct costs of pollution control measures do not adequately represent
the total costs of environmental protection, due to the effects of pollution abatement on
industrial productivity.31 Although our cost estimates do not capture these productivity
effects, the literature is not clear on the magnitude and direction of these effects. While
some studies have found that pollution control negatively affects productivity, others
have found that the productivity impact is positive or ambiguous.32
Effects of the CAAA on Product Quality: In addition to increasing the cost of producing
goods and services, CAAA requirements may also affect product quality. For example,
motor vehicle emission control requirements may reduce the performance of automobiles,
and changes in paint formulations (to reduce VOC emissions) may adversely affect how
well paint adheres to unfinished surfaces. On the other hand, changes in product quality
may also have unquantified benefits - while we capture the fuel saving benefits of many
motor vehicle engine changes, the benefits of low-VOC paint in improving indoor air
quality and human health are not captured in our estimates. As a result, product quality
31 Barbera, A.J. and McConnell, V.D. (1986) "Effects of Pollution Control on Industry Productivity: A Factor Demand
Approach." The Journal of Industrial Economics. Vol. XXXV, 161-172.
Barbera, A.J. and McConnell, V.D. (1990) "The Impact of Environmental Regulations on Industry Productivity: Direct and
Indirect Effects." Journal of Environmental Economics and Management. Vol. 18, 50-65.
Gray, W.B. and Shadbegian, R.J. (1994) "Pollution Abatement Costs, Regulation, and Plant-Level Productivity." Center for
Economic Studies.
Morgenstern, R.D., Pizer, W.A., and Shih, J-S. (2001) "The Cost of Environmental Protection." Review of Economics and
Statistics Vol. 83, No. 4, 732-738. (doi:10.1162/003465301753237812).
32 Barbera and McConnell (1986) found a negative impact of pollution control on productivity, while Barbera and McConnell
(1990) and Gray and Shadbegian (1994) found an ambiguous impact, and Morgenstern et al. (1998) found a positive impact.
3-15
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
effects may reduce the welfare of households that consume products affected by the
CAAA, or they may improve welfare. Households that substitute to other products due to
CAAA-related quality changes (e.g., households that substitute from automobiles to light-
duty trucks due to CAAA requirements that affect the performance of automobiles more
than light-duty trucks) may also experience welfare losses or gains, as they would have
otherwise preferred the product(s) that they would have consumed in the absence of the
CAAA but may, in the balance, experience previously unrecognized gains.
Technological Innovation: The CAAA could serve as in impetus for technological
innovation in the development of new, low-cost technologies or processes to reduce
emissions. As indicated above, our cost estimates reflect the impact of experience-driven
improvements in the productivity of existing control technologies—by accounting for
learning curve impacts—but not the impact of technological innovation. Because we did
not attempt to model technological innovation that might be spurred by incentives to
minimize compliance costs, the Second Prospective Analysis may overestimate costs.
Input Substitution: To minimize the cost of complying with the amendments, regulated
facilities may alter the mix of inputs used in the production of goods and services. With
the exception of fuel switching by EGUs (as part of compliance with the Title IV Acid
Rain Program and CAIR), we did not capture input substitution as a control strategy in
the Second Prospective Cost Report. Ignoring the possible impact of input substitution
could also cause our estimates to overstate CAAA compliance costs.
Table 3-4 lists the key sources of uncertainty noted in the quantitative and qualitative
discussions above and indicates—where possible—the expected impact of the uncertainty
on the net benefits estimate of the Second Prospective Analysis.
3-16
-------
TABLE 3-4.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
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 ESTIMATE1
Uncertainty in the
maximum per ton costs for
local controls to comply
with the 8-hour Ozone and
PM2.5 NAAQS.
Unable to
determine based
on current
information
Probably minor. Our analysis of local controls
assumes a maximum cost of $15,000 per ton
for local controls implemented to comply
with 8-hour Ozone and PM2.5 NAAQS
requirements.5 Local areas may implement
more costly controls to comply with the
NAAQS, but technological innovation may
lead to the development of less expensive
controls.
Uncertainty in the
projected composition of
motor vehicle sales and
the fuel efficiency of the
motor vehicle fleet.
Unable to
determine based
on current
information
Probably minor. We projected the
composition of motor vehicle sales and the
fuel efficiency of the motor vehicle fleet
based on AEO 2005 data. The sensitivity
analysis of alternative sales and fuel
efficiency projections presented in this
report suggests that this uncertainty has a
small impact on net benefits.
Uncertainty regarding
failure rates for motor
vehicle inspections.
Unable to
determine based
on current
information
Probably minor. The repair costs for vehicles
that fail emission inspections represent a
small fraction of the estimated net benefits
of the amendments. The failure rate
sensitivity analysis presented in this report
suggests that alternative failure rate
assumptions could have a large effect on the
costs for this component of the CAAA, but
only a minor effect on the estimated net
benefits of the amendments as a whole.
Costs for some
technologies and emissions
sectors reflect default
assumptions about the
rates at which learning
affects costs because
empirical information is
unavailable.
Underestimate
Probably minor. Based on the advice of the
Council on Clean Air Compliance Analysis, we
used a conservative learning rate of 10
percent for those sectors where no empirical
data were available.2 In contrast, the
learning curve literature suggests that the
average learning rate is approximately 20
percent, suggesting that learning will reduce
costs more than is reflected in the present
analysis.3
Uncertainties in the
economic growth
projections that form the
basis of the cost analysis.
Unable to
determine based
on current
information
Probably minor. The project team used AEO
2005 economic growth projections, which
suggest that the economy will grow at an
annual rate of 3.1 percent through 2025.4
This growth rate is in line with historical GDP
growth.
Incomplete
characterization of certain
indirect costs, such as
productivity impacts for
regulated industry.
Unable to
determine based
on current
information
Probably minor. The literature on the
productivity impacts of the CAAA is unclear
with respect to the direction and magnitude
of these effects.
3-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
DIRECTION OF
POTENTIAL BIAS
FOR NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO KEY
UNCERTAINTIES ON NET BENEFITS ESTIMATE
Product quality
degradation associated
with emission control
technology.
Unable to
determine based
on current
information
Exclusion of the impact of
technological innovation
and input substitution on
compliance costs.
Unable to determine based on current
information. Conceptually, the potential for
CAAA requirements to affect product quality
could result in an underestimate or
overestimate of the welfare effects of
compliance costs, and therefore an
indeterminate effect on net benefits.
Unfortunately, few studies exist that address
the potential product quality effects of CAAA
regulations.
Underestimate
Probably minor. Minimal information is
available on the potential effects of
technological innovation on costs. Though
input substitution is a potential source of cost
savings, the analysis primarily models mature
industries and compliance strategies which
have been established as least-cost
compliance paths. In addition, many
regulations, such as RACT, are technology-
based and may not allow for much input
substitution.
Partial estimation of costs
for compliance with the
PM2.5 NAAQS, due to the
unavailability of emission
reduction targets for non-
attainment areas.
Overestimate
Probably minor. The 2006 PM2.5 NAAQS RIA
estimates that the incremental costs of
residual non-attainment (i.e., costs of
additional reductions from unidentified
controls needed to reach attainment) are
approximately $4.3 billion in 2020, yielding
total cost estimates that exceed the
estimates presented here by a factor of five
or more.6 However, we estimate that the
costs of the PM2.s NAAQS represent less than 5
percent of the net benefits of the
amendments.7
Uncertainty in the
emission reduction
estimates used to estimate
the costs for select rules.
Unable to
determine based
on current
information
Probably minor. Costs for many rules are not
dependent on the corresponding emissions
reductions (e.g., fuel sulfur limits, tailpipe
standards, etc.)
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.
3-18
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
DIRECTION OF
POTENTIAL BIAS
FOR NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO KEY
UNCERTAINTIES ON NET BENEFITS ESTIMATE
Potential for
overestimation biases in
engineering cost
estimates.
Underestimate
Probably minor. A study by Harrington,
Morgenstern, and Nelson (1999) evaluated the
accuracy of EPA and OSHA estimates of 25 ex
ante regulatory cost estimates relative to ex
post studies of actual costs, and concluded
that initial cost estimates by EPA
tend to overstate costs. The source of these
biases include a built-in conservative bias,
inaccuracies in estimating the size of the
affected universe, the effect of learning on
reducing costs, the effect of innovation on
reducing costs, and cost-reducing features of
regulatory design. Some of these factors are
discussed elsewhere in this table. The
magnitude of these biases varies
substantially, but in no case would we expect
the overall impact to exceed five percent of
overall net benefits.
The classification of each potential source of error reflects the best judgment 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."
U.S. Environmental Protection Agency Science Advisory Board, EPA-SAB-COUNCIL-ADV-07-002, "Benefits
and Costs of Clean Air Act - Direct Costs and Uncertainty Analysis", Advisory Letter, June 8, 2007.
Available at http://www.epa.gov/sab/pdf/council-07-002.pdf.
For an analysis of the learning rates estimated in the empirical literature, see John M. Dutton and Annie
Thomas, "Treating Progress Functions as a Managerial Opportunity," Academy of Management Review,
Vol9, No. 2, 1984.
U.S. Department of Energy, Energy Information Administration, Annual Energy Outlook 2005, February
2005.
The Project Team uses this maximum unit cost value in two ways. First, the Project Team assumes that
local areas would not implement identified controls costing more than $15,000 per ton. Second, the
Project Team assumes a cost of $15,000 per ton for unidentified controls.
U.S. Environmental Protection Agency. Regulatory Impact Analysis for the Particulate Matter NAAQS.
October, 2006.
For detailed estimates of the costs of PM2.s NAAQS compliance, see E.H. Pechan and Associates, Inc. and
Industrial Economics, Inc., Direct Cost Estimates for the Clean Air Act Second Section 812 Prospective
Analysis, prepared for U.S. EPA, March 2009.
3-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 4 - AIR QUALITY BENEFITS
Air quality modeling links changes
in emissions to changes in the
atmospheric concentrations of
pollutants that may affect human
health and the environment. A
crucial analytical step, air quality
modeling is one of the more complex
and resource-intensive components
of the prospective analysis. This
chapter outlines how we estimated
future-year pollutant concentrations
under both the with-CAAA and
without-CAAA scenarios.
The first section of the chapter
begins with a discussion of some of
the challenges faced by air quality
modelers and a brief description of
the models we used in this analysis.
The following section provides more
details on the specific air quality
modeling tools we deployed to estimate future-year ambient concentrations. This
methodology section includes a description of how we use modeling results to adjust
monitor concentration data and estimate ambient concentrations for years and scenarios
where no monitoring yet exists - the projected and counterfactual (without-CAAA) target
years and scenarios. 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 brief discussion of the key uncertainties associated with air quality
modeling concludes the chapter.
OVERVIEW OF APPROACH
As we outlined in the First Prospective, air quality modelers face two key challenges in
attempting to translate emission inventories into pollutant concentrations. First, they
must model the dispersion and transport of pollutants through the atmosphere. Second,
they must model pertinent atmospheric chemistry and other pollutant transformation
processes. These challenges are particularly acute for those pollutants that are not
emitted directly, but instead form through secondary processes. Ozone is the best
Scenario Development
V
Sector Modeling
I
I
Emissions Direct
I
Air Quality Modeling
i '
Health Welfare
1 '
Economic Valuation
T
I
Benefit-Cost Compariso
•
Cost
i
4-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
example; it forms in the atmosphere through a series of complex, non-linear chemical
interactions of precursor pollutants, particularly certain classes of volatile organic
compounds (VOCs) and nitrogen oxides (NOX). We faced similar challenges when
estimating PM concentrations. Atmospheric transformation of gaseous sulfur dioxide and
nitrogen oxides to particulate sulfates and nitrates, respectively, contributes significantly
to ambient concentrations of fine particulate matter. In addition to recognizing the
complex atmospheric chemistry relevant for some pollutants, air quality modelers also
must deal with uncertainties associated with variable meteorology and the spatial and
temporal distribution of emissions.
Air quality modelers and researchers have responded to the need for scientifically valid
and reliable estimates of air quality changes by developing 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 studies,
State Implementation Plans (SIPs), and individual air toxic source risk assessments. In
this analysis, we focused our air quality modeling efforts on estimating the impact of
with- and without-CAAA emissions on ambient concentrations of ozone, PM10, and
PM2 5, as well as acid deposition and visibility for each of the target years: 2000, 2010,
and 2020. The focus on these pollutants is consistent with the result in the First
Prospective that most of the quantified benefits of the CAAA are attributable to PM and
ozone. The ideal model for this analysis is a single integrated air quality model capable of
estimating ambient concentrations for all of these key pollutants throughout the U.S. In
the prior First Prospective study, such a model had not yet been sufficiently developed
and tested. This analysis is the first Section 812 prospective analysis to use an integrated
modeling system, the Community Multiscale Air Quality (CMAQ) model, to simulate
national and regional-scale pollutant concentrations and deposition. The CMAQ model
(Byun and Ching, 1999) is a state-of-the-science, regional air quality modeling system
that is designed to simulate the physical and chemical processes that govern the
formation, transport, and deposition of gaseous and particulate species in the atmosphere.
The emissions data were processed for input to the CMAQ modeling using the Sparse-
Matrix Operator Kernel Emissions (SMOKE) emissions processing system (CEP, 2004).
The model-ready emission inventories for each scenario and year were then used to
obtain base year and target year estimates of the key criteria pollutants, as well as many
other species. The air quality modeling analysis was designed to make use of tools and
databases that have recently been developed and evaluated by EPA for other national-
and regional-scale air quality modeling studies. In particular, model-ready meteorological
input files for 2002 were provided by EPA for use in this study. For fine particulate
matter (PM2 5) and related species, the CMAQ model was applied for an annual
simulation period (January through December). A 36-km resolution modeling domain
that encompasses the contiguous 48 states was used for the annual modeling. For ozone
and related species, the CMAQ model was applied for a five-month simulation period
that captures the key ozone-season months of May through September. Two 12-km
resolution modeling domains (that when combined cover the key, ozone-significant areas
of the contiguous 48 U.S. states) were used for the ozone-season modeling. Altogether,
4-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
model-ready emission inventories were prepared and the CMAQ model was applied for a
total of 21 simulations (comprising seven core scenarios and three modeling domains).
The outputs from the CMAQ model provide the basis for the calculation of health and
ecological benefits of the CAA. The airborne criteria pollutants of interest include ozone
and fine particulate matter (PM2 5), where PM2 5 consists of particles less than 2.5 microns
in diameter. For health benefits analysis, it has become standard EPA practice to calibrate
the CMAQ results monitor data, rather than use the CMAQ results directly - the process
is sometimes called, "monitor and model relative adjustment." We follow that approach
in this analysis as well, applying a tool called the Modeled Attainment Test Software
(MATS) to develop and apply the calibration factors for particulate matter results relative
to nearby monitors. For ozone, the MATS procedure is not necessary; instead we use an
inverse distance squared weighting procedure called Enhanced Voronoi Neighbor
Averaging (eVNA), which calibrates the CMAQ model ozone results by weighing data
from monitors closer to the grid cell more heavily than monitors that are further away.
The eVNA interpolation and model to monitor calibration process is accomplished within
the BenMAP benefits analysis tool, which is described in Chapter 5. Visibility is also an
air quality parameter of interest and this was calculated using a variety of the CMAQ
output species. In addition, deposition of nitrogen and sulfur was also extracted from the
model outputs. An overview of the modeling approach is provided in Figure 4-1, which
summarizes the emissions processing and air quality components. The CMAQ modeling
components and application of the MATS tool are explained in further detail in the next
section.
AIR QUALITY MODELING TOOLS DEPLOYED
THE CMAQ MODELING SYSTEM
The Community Multiscale Air Quality (CMAQ) model is a state-of-the-science, regional
air quality modeling system that can be used to simulate the physical and chemical
processes that govern the formation, transport, and deposition of gaseous and particulate
species in the atmosphere (Byun and Ching, 1999). The CMAQ tool was designed to
improve the understanding of air quality issues (including the physical and chemical
processes that influence air quality) and to support the development of effective
emissions control strategies on both the regional and local scale. The CMAQ model was
designed as a "one-atmosphere" model and this concept refers to the ability of the model
to dynamically simulate ozone, particulate matter, and other species in a single simulation
which captures interaction effects among these pollutants. In addition to addressing a
variety of pollutants, CMAQ can be applied to a variety of regions with varying
geographical, land-use and emissions characteristics, and for a range of different space
and time scales.
4-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-1. SCHEMATIC DIAGRAM OF SECTION 812 AIR QUALITY MODELING ANALYSIS
812 Emissions Data
i
F
SMOKE Emissions
Processing Systems
T
CMAQ-Ready 36-km
Emission Inventories
(USDomain)
1990
2000 with out CAM
2000 with CAM
2010 with out CAM
2010 with CAM
2020 with out CAAA
2 020 with CAM
F ^
CMAQ-Ready12-km
Emission Inventories
(Eastern US Do main)
19
yj
2 000 with out CAM
2 000 with CAM
2010 with out CAM
2010 with CAM
2020 with out CAM
2 020 with CAM
r
CMAQ-Ready 12-km
Em i ss i on I nv entories
(Western US Domain)
19
90
2000 with out CAM
2000 with CAM
2010 with out CAAA
2010 with CAM
2020 with out CAM
2 020 with CAM
(b) CMAQ .Application for the 36-km Continental U. S. (CONUS) Domain
CMAQ-Ready 36-km
Emissionlnv entories
iUS Domain!
1990
2 000 with out CAM
2000 with C.AM
2010 with out CAM
2010 with CAAA
2 020 with out CAM
2020 with CMA
2002 Meteorological Inputs
Geophysical&IC/BC Inputs
Annual PM;;. Visibility
& Deposition
lUSDomaini
1990
23K5 with out CAM
2000 with CAAA
2010 without CAM
2010 with CMA
2020 with out CAM
2020 with CAM
Health and Ecological Assessments
(c) CMAQ Application for the 12-km Eastern and Western U.S. Domain (EUS and WUS)
CMAQ-Ready12-km
Emission Inventories
(Eastern and Western
USDomainj
1990
2:000 with out CAM
2 000 with CAM
2010withoutCAM
2010 with CAM
2020 with out CAM
2 020 with CAM
2002 Meteorological Inputs
Geophysical & IC/BC Inputs
Ozone metrics
(EUS and WUS)
1990
2003 with out CAM
2000 with CAM
2010 with out CAM
2010withCMA
2020 with out CAAA
2020 with CAM
Health and Ecological .Assessments
4-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
The CMAQ model numerically simulates the physical processes that determine the
magnitude, temporal variation and spatial distribution of the concentrations of ozone and
particulate species in the atmosphere and the amount, timing, and distribution of their
deposition to the earth's surface. The simulation processes include advection, dispersion
(or turbulent mixing), chemical transformation, cloud processes, and wet and dry
deposition. The CMAQ science algorithms are described in detail in Byun and Ching
(1999).
The CMAQ model requires several different types of input files. Gridded, hourly
emission inventories characterize the release of anthropogenic, biogenic and, in some
cases, geogenic emissions from sources within the modeling domain. The emissions
represent both low-level and elevated sources and a variety of source categories
(including, for example, point, onroad mobile, nonroad mobile, area, and biogenic
emissions). The amount, spatial distribution, and temporal distribution of each emitted
pollutant or precursor species are key determinants to the resultant simulated air quality
values.
The CMAQ model also requires hourly, gridded input fields of several meteorological
parameters including wind, temperature, mixing ratio, pressure, solar radiation, fractional
cloud cover, cloud depth, and precipitation. A full list of the meteorological input
parameters is given in Byun and Ching (1999). The meteorological input fields are
typically prepared using a data-assimilating prognostic meteorological model, the output
of which is processed for input to the CMAQ model using the Meteorology-Chemistry
Interface Processor (MCIP). The prescribed meteorological conditions influence the
transport, vertical mixing, and resulting distribution of the simulated pollutant
concentrations. Particular meteorological parameters, such as mixing ratio, can also
influence the simulated chemical reaction rates. Rainfall and near-surface meteorological
characteristics govern the wet and dry deposition, respectively, of the simulated
atmospheric constituents.
Initial and boundary conditions (IC/BC) files provide information on pollutant
concentrations throughout the domain for the first hour of the first day of the simulation,
and along the lateral and top boundaries of the domain for each hour of the simulation.
Photolysis rates and other chemistry related input files supply information needed by the
gas-phase and particulate chemistry algorithms.33
33 The latest available version of CMAQ, version 4.6, was used for this study. This version of the model supports several
different gas-phase chemical mechanism, particle treatment, aerosol deposition, and cloud treatment options. All
simulations conducted as part of this study used the CB05 chemical mechanism. For particles, the AER04 particle
treatment, which includes sea salt, was applied. Finally, the plume-in-grid feature of CMAQ was not used for this study.
More details are available in Second Prospective Analysis of Air Quality in the U.S.: Air Quality Modeling, available at
www.epa.gov/oar/sect812
4-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CMAQ APPLICATION PROCEDURES FOR THE SECOND PROSPECTIVE ANALYSIS
This specific application of CMAQ includes modeling domain specification and key
input files. The three modeling domains that were used for this analysis are shown in
Figure 4-2.
FIGURE 4-2. CMAQ MODELING DOMAINS FOR THE 812 MODELING STUDY
NOTE: CONUS IS THE CONTINENTAL US GRID USED FOR PM MODELING; WUS IS THE
WESTERN US GRID AND EUS IS THE EASTERN US GRID USED FOR OZONE MODELING.
The 36-km resolution continental U.S. (CONUS) domain is the large area that is covered
by the outer grid box in Figure 4-2. The CONUS domain includes 148 x 112 grid cells
(the total number of cells is 16,576). The tick marks denote the 36-km grid cells. For this
domain, the model was run for the entire 2002 calendar year, using 2002 meteorology but
varying the emissions inputs as outlined in each of the Second Prospective scenarios
listed in Figure 4-1. In running the model, the annual simulation period was divided into
two parts covering January through June and July through December, respectively. Each
part of the simulation also includes an additional five start-up simulation days, which are
intended to reduce the influence of uncertainties in the initial conditions on the simulation
results.
The Eastern U.S. (EUS) domain is comprised of 213 x 188 grid cells (total = 40,044
cells) and the Western U.S. (WUS) domain includes 213 by 192 grid cells (total = 40,896
cells). Together these two domains cover most of the continental U.S. with 12-km
horizontal resolution. There is some overlap in the central part of the country. For both
the EUS and WUS domains, the CMAQ model was run for the months of May through
September. This five-month period is intended to represent the ozone season - runs using
this domain provide the ozone inputs for subsequent steps of the analysis. The seasonal
simulation period was also divided into two parts covering May and June and July
4-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
through September, respectively. Each part of the simulation also includes an additional
ten start-up simulation days.
The 36- and 12-km resolution meteorological input files to support modeling in these
domains were prepared using the Pennsylvania State University/National Center for
Atmospheric Research (PSU/NCAR) Fifth Generation Mesoscale Model (MM5). The
MM5 outputs were postprocessed by EPA for input to CMAQ using the Meteorology-
Chemistry Interface Processor (MCIP) program. The meteorological input preparation
methodology and some information on MM5 model performance are provided by
Dolwick et al. (2007). Existing initial condition, boundary condition, land-use and
photolysis rate input files prepared by EPA for use in CMAQ modeling for the selected
modeling domains and simulation period were used.
After the initial CMAQ results were generated, the original primary PM emissions
estimates generated for area and non-EGU point sources were found to be inaccurate due
to two issues:
1) As described in Chapter 2, some of the fine particulate emissions estimates
derived from the 1990 NEI, on which the without-CAAA emissions estimates
were based, were discovered to be inconsistent with those from the 2002 NEI, on
which the with-CAAA emissions estimates were based.
2) The original emissions estimates did not include application of transport factors
for area source fine particulate emissions. These transport factors are county-
specific adjustment factors that are applied to specific types of emissions
estimates to account for the fact that only a fraction of total fugitive dust
emissions remain airborne and are available for transport away from the vicinity
of the source after localized removal (i.e., some of the particles are captured by
the local vegetation or other surface obstructions).
To correct these two errors, we first made the necessary adjustments to the primary
PM2.5 emissions estimates for the affected non-EGU point and area sources, focusing on
the PM2.5 species that contribute most significantly to primary PM emissions: elemental
carbon (EC), organic carbon (OC), and crustal material. We then calculated species-
specific adjustment factors for the CMAQ data, re-compiled the species-specific
estimates to generate an adjusted version of the original CMAQ results, and then
generated new MATS input files. All details of the procedure are described in a
memorandum prepared by the Project Team, which was reviewed in detail by the
Council's Air Quality Modeling Subcommittee.34
34
Memorandum of June 14, 2010 to Jim DeMocker, EPA, from Tyra Walsh, Henry Roman, and Jim Neumann, Industrial
Economics, Inc. (lEc), "Description of the Adjustment to the Primary Particulate Matter Emissions Estimates and the
Modeled Attainment Test Software Analysis (MATS) Procedure for the 812 Second Prospective Analysis." The memo is
available at www.epa.gov/oar/sect812.
4-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
MATS PROCEDURE
Rather than using the direct CMAQ results as the basis for the health and ecological
effects analyses, the Project Team conducted additional analyses using a speciated
monitor and model calibration technique to generate PM2 5 air quality estimates. The
PM2 5 estimates used in the Second Prospective health analysis were prepared using
EPA's Modeled Attainment Test Software (MATS, Version 2.1.1, Build 807). MATS
estimates quarterly mean PM2 5 chemical component concentrations at monitor locations
by conducting a Speciated Modeled Attainment Test (SMAT) analysis. MATS can also
estimate quarterly mean concentration estimates for each PM2 5 chemical component
concentrations at all grid cells in a grid model such as CMAQ.
Five of the six MATS PM25 concentration estimates for the Second Prospective scenarios
were prepared using the MATS' spatial and temporal relative adjustment method. The
MATS estimates for the 2000 with-CAAA scenario, which represents a historical year for
which monitor data are available, used a spatial only relative adjustment method, relying
on available monitor data and a single year of CMAQ modeling. The MATS procedure
was not applied for the 1990 base year scenario.
MATS estimates the PM2 5 concentrations in CMAQ grid cells by interpolating values
from nearby monitors using the inverse distance squared weighting option in the Voronoi
Neighbor Averaging (VNA) procedure in MATS. This is an algorithm that identifies a
set of monitors close to the grid cell (called "neighbors") and then estimates the PM
species concentration in that grid cell by calculating an inverse-distance weighted average
of the monitor values (i.e., the concentration values at monitors closer to the grid cell are
weighted more heavily than monitors that are further away). As noted above, for
calibrating ozone model results to nearby monitors, only the VNA component of the
procedure is used, because there is no need for the speciated interpolation approach
required for PM.
The spatial MATS analysis conducted for the PM2 5 estimates used the following input
information:
• observed quarterly PM2 5 data from 1,232 Federal Reference Method (FRM)
monitors with sufficient data in 2002 - sufficient data is defined as at least one
quarter of PM2 5 data. The year 2002 was used because it corresponds to the
vintage of the emissions estimates, which, as described in Chapter 2, were
derived from the 2002 National Emissions Inventory;
• observed daily chemically speciated fine particle mass data from both the PM2 5
Speciation Trends Network (STN) and the Interagency Monitoring of Protected
Visual Environments (IMPROVE) network, providing a total of 273 monitors
with sufficient data in 200235:
35
Most FRM monitors (about 75 percent) are not co-located with a speciation monitor. Therefore, we also used data
providing speciated PM mass from the STN and IMPROVE monitors. The MATS analysis used speciated data from 273 STN or
IMPROVE monitors with at least two valid quarters of speciated data in 2002.
4-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-3.
• speciated CMAQ estimates for 6 PM2 5 species (SO4, NO3, elemental carbon,
organic carbon, NFL,, and crustal material) at the 36 kilometer PM2 5 CMAQ grid
cell level for each of the Second Prospective scenarios (from CMAQ speciated
output data files).
The MATS procedure enables the use of monitor data to effectively calibrate the results
of air quality modeling for use in subsequent steps of the analysis. To illustrate the
effects of the MATS procedure, compare Figure 4-3, which is a scatter plot comparing
the direct CMAQ results for those 1,058 PM25 monitors with at least two quarters of data
for 2002, and Figure 4-4, which is a similar scatter plot, comparing the MATS results to
the same set of PM25 monitors. The agreement between monitor and model values in
Figure 4-4 is greatly improved by the MATS procedure.
SCATTER PLOT OF DIRECT CMAQ ESTIMATES AND 2002 PM2.5 FEDERAL REFERENCE
METHOD (FRM) MONITORS
o
CMAQ vrs 2002 FRM Monitors
5 10 15 20 25
2002 FRM Monitors (1,058 Monitors), PM2.5 annual mean (ug/m3)
Figure 4-5 provides a further illustration of the effect of the MATS procedure, and the
importance of individual PM species in achieving an effective calibration of the CMAQ
results to monitor data. The figure provides detailed species-specific CMAQ and MATS
results for a CMAQ grid cell in the three largest cities and metropolitan areas in the US -
New York, Los Angeles, and Chicago - and for Tucson, Arizona, a much smaller city but
4-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-4
one for which one component of PM, crustal (shown in brown), plays a critical role in our
analysis. For each city, the two leftmost bars provide the 2002 FRM and STN annual
average PM2 5 monitor data for a monitor of that type within the grid cell. FRM monitors
provide only total PM2 5 mass, while the STN monitors provide data for the seven PM
species (plus estimated water) indicated at the bottom of each graph.36 The remaining 12
bars in each panel show the CMAQ and MATS-adjusted results for the grid cell for the
with-CAAA and without-CAAA scenarios, for target years 2000, 2010, and 2020.
SCATTER PLOT OF MATS-ADJUSTED CMAQ ESTIMATES AND 2002 PM2.5 FEDERAL
REFERENCE METHOD (FRM) MONITORS
MATS with Adjustments vrs 2002 FRM Monitors
5 10 15 20 25
2002 FRM Monitors (1,058 Monitors) (annual mean PM2.5, ug/m3)
30
36 The STN bar charts include an estimated water component, which the MATS input monitor files include to make STN and
IMPROVE monitor data consistent with FRM monitor data. The water component is not an STN component, but was
estimated using the SANDWICH (Sulfates, Adjusted Citrates, Derived Water, inferred Carbonaceous mass, and estimated
aerosol acidity (H_+)) process.
4-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-5. COMPARISON OF CMAQ, MATS, AND MONITOR DATA FOR FOUR SELECTED CITIES
2002 FRM
2002 STN (8 components)
2000 With CAAA. CMAQ
2000 With OAAA, MATS
2000 Without CAAA, CMAQ
2000 Without CAAA, MATS
2010 With CAAA. CMAQ
2010 With CAAA, MATS
2010 Without CAAA, CMAQ
2010 Without CAAA, MATS
2020 With CAAA, CMAQ
2020 With CAAA, MATS
2020 Without CAAA, CMAQ
2020 Without CAAA, MATS
Manhattan, NY
10 20 30 40
Annual Mean PM2.S (ug/m3)
Los Angeles, CA
2002 FRM
2002 STN (8 components)
2000 With CAAA, CMAQ
2000 With CAAA, MATS
2000 Without CAAA, CMAQ
2000 Without CAAA, MATS
2010 With CAAA, CMAQ
2010 With CAAA, MATS
2010 Without CAAA, CMAQ
2010 Without CAAA, MATS
2020 With CAAA. CMAQ
2020 With CAAA, MATS
2020 Without CAAA, CMAQ ]
2020 Without CAAA, MATS
10 20 30 40
Annual Mean PM2.5 (ug/m3)
50
2002 FRM
2002 STN (8 components)
2000 With CAAA. CMAQ
2000 With CAAA, MATS
2000 Without CAAA, CMAQ
2000 Without CAAA, MATS
2010 With CAAA, CMAQ
2010 With CAAA. MATS
2010 Without CAAA, CMAQ
2010 Without CAAA, MATS
2020 With CAAA, CMAQ
2020 With CAAA, MATS
2020 Without CAAA, CMAQ
2020 Without CAAA, MATS
Chicago, IL
Tucson, AZ
2002 FRM
2002 STN (8 components)
2000 With CAAA, CMAQ
2000 With CAAA, MATS
2000 Without CAAA, CMAQ
2000 Without CAAA, MATS
2010 With CAAA, CMAQ
2010 With CAAA, MATS
2010 Without CAAA. CMAQ
2010 Without CAAA, MATS
2020 With CAAA. CMAQ
2020 With CAAA, MATS
2020 Without CAAA, CMAQ
2020 Without CAAA, MATS
10 20 30 40
Annual Mean PM2.5 (ug/m3)
50
10 20 30 40
Annual Mean PM2.5 (ug/m3)
50
HPM2.5 SO4 BN03 BNH4 B EC • OC BCrustal • Water Salt B Blank
The Manhattan panel in the upper left corner shows that both the FRM and STN monitors
indicate atotal PM concentration just greater than 15 (ig/m3. The next bar shows that the
CMAQ data for the 2000 with-CAAA simulation overestimates the PM concentration, by
about 4 (ig/m3. Comparing the 2002 STN bar with the 2000 with-CAAA CMAQ bar, we
see that the CMAQ simulation overestimates most constituents in this location, compared
to the monitor data, but underestimates organic matter (or OC, shown in green). The
MATS procedure, applied to the STN and CMAQ data, generates species-specific scaling
factors that result in a MATS-adjusted concentration for the 2000 with-CAAA scenario,
4-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
shown in the next bar. As a result, the species-specific constituents in the MATS
adjusted bar are in very nearly the same proportion as they appear for the STN monitor.
It would also appear from this figure that MATS "overcorrects" in Manhattan because the
2000 with-CAAA MATS bar is lower than the 2002 STN monitor bar. However, the
MATS procedure is estimating the concentration at the center of the grid cell, not at the
location of the STN monitor. In a 36 km grid cell, the monitor location can be many
kilometers away from the center of the grid cell. MATS considers not only monitors in
the same grid cell, but also the data at other nearby FRM and STN monitors, and makes a
spatial interpolation to estimate concentrations at the grid centroid. The Manhattan STN
monitor is near the intersection of four grid cells, which contain a total of 25 FRM and
STN monitors, all of which influence the MATS result.
The remaining MATS estimates for Manhattan, for the 2000 without-CAAA and the 2010
and 2020 projections, are based on scaling of the corresponding CMAQ simulation by the
species-specific factors developed from the 2000 with-CAAA to 2002 STN monitor
comparison. The effect of MATS in Manhattan is to adjust the CMAQ simulation
concentrations downward. Interestingly, the opposite is generally true in Los Angeles,
because in that city CMAQ tends to underestimate the monitor data for 2002. The mix of
species in both cities is similar in 2002, but strikingly different over time, particularly in
the without-CAAA scenario, where organic carbon (shown in green) in Los Angeles
derives from mobile sources, and sulfates (shown in yellow) in Manhattan derives from
long-range transport from coal-burning electric generating units.
In Chicago, the effect of MATS is more complex, and the importance of considering PM
species is highlighted. In the with-CAAA scenarios, MATS yields a downward
adjustment to the CMAQ simulations, because the 2000 with-CAAA CMAQ simulation is
higher than the 2002 STN monitor value. In the without-CAAA scenarios, however, there
are much higher emissions of organic carbon, because certain OC emissions controls are
not in place in the without-CAAA simulations that are in place in the with-CAAA scenario.
Because CMAQ underestimates the ambient OC component in the 2000 with-CAAA
(shown in green), the factor for OC that is applied to other scenarios yields an increase in
concentration in the MATS-adjusted values. That increase is large enough to dominate
the overall adjustment across all eight species, yielding an overall PM2 5 mass increase for
the without-CAAA scenarios relative to the CMAQ data.
The data for Tucson also illustrates the importance of the species-specific scaling factors.
If it were not for changes to one PM species, crustal (shown in brown), there would be
only a relatively modest difference between the with-CAAA and without-CAAA scenarios
in future years. In Tucson the crustal component derives largely from construction
activity, which in this relatively fast growing area of Arizona, and absent more stringent
dust control measures, could become a larger issue in the projection years. CAAA
controls on fugitive dust emissions in the construction sector, however, yield a substantial
difference in this component of PM concentrations, when comparing the with-CAAA and
without-CAAA scenario results. Other species differ much less across scenarios. In many
other places like Tucson, the species-specific MATS procedure likely yields a more
4-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
accurate projection of the impact of the CAAA than a calibration procedure that did not
take into account the impact of these species-specific control strategies.
AIR QUALITY RESULTS
PARTICULATE MATTER
As mentioned above, the CMAQ modeling results for the 36-km continental U.S.
(CONUS) modeling domain provide the basis for particulate matter air quality used in the
calculation of PM-related health effects and to calculate visibility, as well as sulfur and
nitrogen deposition. Summary results are presented in the maps in Figure 4-6 below,
representing annual average concentrations across the CONUS domain for each of the
seven scenario/target year combinations modeled. The rows of Figure 4-6 show modeled
PM2 5 concentrations for 2000, 2010, and 2020, contrasting the without-CAAA results on
the left and the with-CAAA results on the right.
As the figure indicates, over the thirty-year 1990-2020 simulation period air quality is
projected to worsen somewhat in the absence of CAAA regulations, particularly in the
Midwest and California, but with CAAA regulations in place air quality is estimated to
improve markedly as early as the year 2000 and to show continued improvements
through 2020. In general, the with-CAAA results reflect a calibration of the 2002 model
year results to monitor values, but as the accompanying Box 4-1 illustrates, such direct
comparisons are not possible for the counterfactual without-CAAA results. We conclude
for the analyses described in the text box that the without-CAAA results, with a few
exceptions, seem to imply a return of air quality conditions comparable to those that
prevailed in the 1980-1990 period prior to implementation of the CAAA. Such
comparisons are limited, however, by the sparse PM2 5 monitoring data for this period and
the uncertainty in adjusting available monitor data for other species. Although the
improvements attributed to the CAAA are nationwide, the most substantial gains are
made in those areas that had the worst PM air quality in 1990, suggesting the CAAA has
been and will continue to be effective in targeting improvements to the areas that would
have experienced the worst air quality in the absence of the amendments.
4-13
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
BOX 4-1: EVALUATING THE WITHOUT-CAM SCENARIO RESULTS
The two scenarios used in this study, the with-CAAA and without-CAAA scenarios, are designed to simulate and forecast air quality
conditions in the US as we expect them to unfold with full implementation of the CAAA (the with-CAAA), and alternatively as if regulations
authorized by the CAAA had not been implemented. In effect, the methods we use tie the with-CAAA scenario to monitored air quality in
the year 2000, providing some measure of credibility for the air quality conditions reflected in our with-CAAA simulation. It is more
difficult to evaluate the credibility of the without-CAAA scenario, because that scenario simulates hypothetical air quality conditions that
cannot be observed. The plausibility of the without-CAAA scenario and its differences from the with-CAAA scenario nevertheless can be
assessed through comparison to other similar air quality conditions.
One possible analog for conditions in the without-CAAA scenario is areas outside the US that have not implemented air quality regulations
that match the stringency of those in the US. The problem with comparing US to non-US areas is the difficulty of standardizing factors
which define air quality, such as meteorology, terrain, and the distribution of air pollutant emission sources. Another major challenge is that
monitoring networks for fine particle species are sparse or not available for the annual average measure.
A preferable, though still imperfect, comparison is between without-CAAA forecasts and historical concentrations in US cities. A key issue
arising for within-US comparisons is that prior to 1990 particulate matter monitors measured total suspended particulates (TSP), or PM10,
rather than PM2 5. The new PM standard is based on PM2 5, which is now recognized as better correlated with adverse health effects. PM2 5
is therefore the focus of our air quality simulations. Furthermore, the ratios of TSP and/or PM10 to PM2 5 vary considerably by location and
over time, so a simple transformation of the available monitor data may not be reliable. Nonetheless, it is possible to find times and
locations in the historical monitor data where at least two and sometimes all three of these measures were simultaneously collected,
providing a means to estimate a time and location-specific ratio that can be used to infer PM2 5 values. We use this type of information to
develop estimates of historical PM2 5 concentration in selected U.S. cities for comparison to our without-CAAA scenario projected values.
The table suggests that our estimates of without-CAAA PM2 5 concentrations in New York, Pittsburgh, and Los Angeles are reasonably
consistent with estimated historical concentrations in the 1980 to 1990 pre-CAAA period. In Chicago, however, the without-CAAA case
yields estimates that are much higher than historical estimates. One reason may be the potentially strong influences of projected
uncontrolled sulfur dioxide emissions from electric power plants near Chicago in the without-CAAA case. In the absence of Title IV these
plants are projected in our study to use relatively high sulfur, locally mined coal and would not have been required to install scrubber
technology.
(ANNUALAVG
MICROGRAMS PER
CUBIC METER)
CITIES
New York -
Manhattan
New York -
Queens/Brooklyn
Pittsburgh
Chicago
Los Angeles
ESTIMATED PM2.5 CONCENTRATIONS FOR THIS STUDY
2000
W-
CAAA
12.9
13.2
14.0
15.5
18.5
W/O-
CAAA
20.6
24.8
19.2
47.7
25.5
2010
W-
CAAA
10.9
11.0
11.0
13.7
17.1
W/O-
CAAA
21.0
25.2
19.7
47.6
29.7
2020
W-
CAAA
10.0
10.1
10.0
13.4
17.5
W/O-
CAAA
22.1
26.7
20.3
48.9
35.5
ESTIMATED HISTORICAL PM2.5
1980
(EST)
N/A
N/A
29.3
25.7
38.5
1990
(EST)
22.4
21.5
22.3
20.4
29.4
MAXIMUM
1980-90
N/A
N/A
29.8
25.7
41.9
4-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-6. CMAQ SIMULATED AND MATS ADJUSTED ANNUAL AVERAGE PM2.5 SPECIES
CONCENTRATION (MICROGRAMS PER CUBIC METERS) FOR THE CONUS DOMAIN
OUTPUTS FOR THE 1990 TO 2020 PERIOD
Avg Cone: 2000 Without CAAA Avg Cone: 2000 With CAAA
Avg Cone: 2010 Without CAAA
Avg Cone: 2010 With CAAA
Avg Cone: 2020 Without CAAA
Avg Cone: 2020 With CAAA
^^J.
o"
Figure 4-7 makes the gains in 2020 more clear, by illustrating the differences in PM2 5
concentrations between the with-CAAA and without-CAAA scenarios in 2020. The gains
in some areas, particularly in the eastern half of the US, in California, and in urban
centers nationwide, are dramatic, with reductions of more than 20 (ig/m3 in some areas.
These are consistent with the large decreases in PM precursor emissions for those areas,
described in Chapter 2. In some of these areas, the without-CAAA scenario
concentrations also reach high levels because of the absence of without-CAAA controls
4-15
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-7.
(see accompanying text box for a discussion of the without-CAAA scenario). There are
also some surprisingly large reductions in a few less populous areas, such as, west central
Idaho and central Virginia. The reductions in Idaho, as well as in a few other isolated
areas of the rural West, are associated with CAAA requirements to limit emissions from
agricultural burning operations. The reductions in central Virginia are attributable to
local controls on a large coal-burning industrial boiler.
DIFFERENCE IN CMAQ SIMULATED MATS ADJUSTED ANNUAL AVERAGE PM2.5 SPECIES
CONCENTRATION (MICROGRAMS PER CUBIC METER) FOR THE CONUS DOMAIN:
2020 WITH-CAAA MINUS 2020 WITHOUT CAAA SCENARIOS
-36. -20 -15 -10 -S
Hg/m3
Some areas also experience modest increases in PM concentrations with the CAAA -
these areas show up in light orange on the map. Some of the smallest estimated
increases, less than 1 (ig/m3, can be introduced by the MATS adjustment procedure,
particularly when the locations are far from monitors and/or have very low modeled or
monitored concentrations of a PM species. We interpret very small increases such as
these as effectively "no change" so adjusted the map legend to group these cells with
others where are small benefits.37 There remain five cells with disbenefits greater than 1
37 There is one area in northeastern Utah where the MATS procedure yields results for the without-CAAA scenario that are so
large as to be not plausible. The result was associated with increases in agricultural burning in the without-CAA scenario,
4-16
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
(ig/m3. The three cells of these five with the smallest disbenefit estimates did not have
disbenefits in the CMAQ modeling - we therefore conclude that the disbenefit result was
introduced by the MATS procedure.
In the remaining two cells, we conclude that implementation of the CAAA led to negative
benefits, associated with actual increases in emissions resulting in the with-CAAA case
relative to the without-CAAA case. The largest disbenefit, of 4.1 (ig/m3, is in the
northwestern corner of New Mexico, in the cell which includes the Four Corners Power
Plant, one of the largest coal-burning power plants in the West. The emissions data
indicate sulfur dioxide emissions for that plant that are 14,000 tons greater in the 2020
with-CAAA case, probably as a combined result of changes in dispatch and sulfur content
of coal for this plant, which as of December 2010 does not have a sulfur scrubber. The
other cell shows a disbenefit of 1.25 (ig/m3, and is located in Sweetwater County in south
central Wyoming, which includes the Pacificorp-Jim Bridger Power Plant. The air
quality result here is also attributable to a difference in sulfur dioxide emissions from a
power plant, in this case 2,000 tons greater in the 2020 with-CAAA scenario. The
dispatch of this unit appears to be identical in both scenarios, so the result is most likely
attributable to a marginal reallocation of higher sulfur coal. Note that, as indicated in the
with-CAAA maps in Figure 4-6, these are areas that nonetheless would continue to
experience PM2 5 concentrations below the 15 (ig/m3 PM2 5 annual standard. These
relatively modest and geographically limited exceptions notwithstanding, it is clear that
by 2020 the air quality benefits of the CAAA in reducing ambient concentrations of
particulate matter are large and widespread.
OZONE
Figures 4-8 through 4-11 present similar CMAQ output data for ozone, with two
important differences: (1) the ozone results are reported for the Eastern (BUS) and
Western (WUS) 12-km modeling domains; and (2) the results presented are the average
of daily maximum 8-hour ozone concentration, in ppb, over the course of a modeled
ozone season (May 1 through September 30). The average daily 8-hour maximum may
seem like an odd metric for evaluating ozone concentrations, but because this is the
metric used in epidemiological estimation of mortality risks of ozone this metric is
closely correlated with the major mortality incidence and economic benefits associated
with ozone precursor controls. Results for the Eastern US are in Figures 4-8 and 4-9, and
for the Western US in Figures 4-10 and 4-11.
For the Eastern US, Figure 4-8 shows a similar pattern for ozone as was illustrated for
particulate matter in Figure 4-6. That is, while there are relatively modest increases in
coupled with otherwise low organic carbon monitor values in nearby monitors - the application of MATS therefore led to
unusually high organic carbon and PM2.5 measures for that area. For those three cells, we performed a moving average
smoothing procedure to re-estimate the without-CAAA concentrations, using PM estimates from adjoining cells. The
adjustment is used only for the purposes of generating the maps in this chapter; for the purposes of health benefits
modeling and valuation of benefits, we excluded these three suspect cells. The cells represent very rural, sparsely
populated areas in the Wasatch Mountains, and so we believe that excluding them from the benefits calculations is both
prudent and has only a modest underestimation effect on the overall health benefits estimates.
4-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
ozone concentrations in the absence of the CAAA, the with-CAAA maps on the right side
of the graphic show significant and widespread gains in air quality throughout the region,
with air quality benefits increasing over time. By 2020, Figure 4-9 shows that the
difference in ozone concentrations is large in most areas of the east, with gains as large as
30 ppb for this simulated day.
Two other patterns in Figure 4-9 are also worth noting. First, although the region-wide
benefits of the CAAA are large, in many urban areas concentrations in the with-CAAA
case are higher than in the without-CAAA case, in some cases near the Gulf Coast and in
New York City by as much as 15 to 20 ppb. Second, some of the areas with the largest
improvements, such as those in the heart of the Midwest, include pockets of much smaller
gains, particularly in some urban centers. In both cases, these results are not unexpected.
The complex chemistry of ozone includes a phenomenon known as "NOx-scavenging",
whereby nitrogen oxides, while participating as an ozone precursor, can also serve to
scavenge or reduce ozone, particularly during the peak ozone season and in urban centers
where ozone levels might otherwise be quite high. The CAAA, in reducing the nitrogen
oxide precursors, may in some cases reduce ozone on a regional level while leading to
much smaller reductions or even increases in ozone in the center of certain urban areas.
This effect explains both these results. Nonetheless, as Figure 4-9 makes clear, the
overall area (and population exposed) of ozone reductions is far greater than the
corresponding areas with ozone increases.
Ozone results in the Western US, in Figures 4-10 and 4-11, indicate a similar pattern to
those for the Eastern US when examining concentrations in urban areas, although in the
West the largest ozone air quality gains are restricted to a smaller area, centered in the
areas in California that have historically struggled with ozone attainment. In addition, in
the Western US there are some more extensive areas in Figure 4-11 with ozone
disbenefits attributed to the CAAA, particularly in Los Angeles.38 Another interesting
result, not shown in Figure 4-10, is that we estimate that ozone concentrations will
actually increase from 1990 to 2000 in most parts of California, in both the without-
CAAA and with-CAAA scenarios, before reductions in 2010 and 2020 bring ambient
levels below those seen in 1990, at least in most areas. This result is largely attributable
to the longer attainment deadlines for the severe non-attainment areas in California - our
scenario assumes that emissions will increase for some period before aggressive regional
mobile source tailpipe standards and non-road fuel and engine standards, and local-scale
ozone attainment plans, have their full effect later in our simulation period.
38 We examined this result further and found that, in cells with the largest disbenefits, the 2020 without-CAAA scenario
yields concentrations of approximately 45 ppb, while concentrations in outlying areas are as high as 100 ppb or slightly
higher. One effect of CAAA controls is to suppress N0x-scavenging in the city center, where disbenefits are largest, yielding
with-CAAA concentrations in the 60 to 65 ppb range. The main effect of the CAAA, however, is large decreases in ozone in
the outlying areas, to concentrations of 60 to 75 ppb. The net effect on a population weighted basis remains a lowering of
overall exposures.
4-18
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-8. CMAQ SIMULATED AND VNA ADJUSTED DAILY MAXIMUM 8-HOUR OZONE (PPB) FOR
THE EUS DOMAIN
Max 8-hr Cone: 2000 Without CAAA
Max 8-hr Cone: 2000 With CAAA
Max 8-hr Cone- 2010 With CAAA
Max o-hr Cone: 2010 Without CAAA
Max 8-hr Cone: 2020 Without CAAA
Max 8-hr Cone: 2020 With CAAA
4-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-9. DIFFERENCE IN SIMULATED DAILY MAXIMUM 8-HOUR OZONE CONCENTRATION (PPB)
FOR THE EUS DOMAIN FOR 15 JULY: 2020 WITH-CAAA MIN US 2020 WITHOUT-
CAAA SCENARIOS
-37.5
-30.0
-22.S
-1S.O
•7.S
7.S 1S.O 22.S
parts per billion (ppb)
4-20
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-10. CMAQ SIMULATED AND VNA ADJUSTED DAILY MAXIMUM 8-HOUR OZONE (PPB) FOR
THE WUS DOMAIN
Max 8-hr Cone: 2000 Without CAAA
Max 8-hr Cone: 2000 With CAAA
Max 8-hr Cone: 20 0 With CAAA
Max K-hr Cone: 2010 Without CAAA
Max 8-hr Cone: 2020 Without CAAA
Max 8-hr Cone 2U20Wlth CAAA
4-21
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 4-11. DIFFERENCE IN SIMULATED DAILY MAXIMUM 8-HOUR OZONE CONCENTRATION (PPB)
FOR THE WUS DOMAIN FOR 15 AUGUST: 2020 WITH-CAAA MINUS 2020 WITHOUT-
CAAA SCENARIOS
-37.S -30.0 -22.S -1S.O -7.S
7.S 15.0 22.S
parts per billion (ppb)
UNCERTAINTY IN AIR QUALITY ESTIMATES
Unlike the air quality modeling conducted over a decade ago for the first Section 812
prospective analysis, which used two different models for ozone and particulate matter.
the modeling conducted for the Second Prospective analysis utilized EPA's Community
Multiscale Air Quality (CMAQ) model, a "one-atmosphere" model that simulates the
chemical formation, transport, and deposition of ozone and particulate matter together in
one comprehensive system.39 The use of this comprehensive air quality modeling system
provides a consistent platform for evaluating the expected responses to changes in
precursor emissions, reducing many of the uncertainties which pertained in the First
Prospective as a result of the limited ability of the models to capture important interaction
effects among the ozone and PM precursor pollutants.
39 Use of an integrated model such as CMAQ for the current study was one of the recommendations made by the Council in
their review of the First Prospective analysis.
4-22
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Nonetheless, air quality modeling is a complex process and, as such, involves many
uncertainties. We provide a summary of some of the more important classes of air
quality modeling uncertainties in Table 4-1 below. These include a known
meteorological bias in the 12-km eastern MM5 domain, which leads to a general
tendency to underestimate the monthly observed precipitation; uncertainties in secondary
organic aerosol (SOA) chemistry which lead to underestimation of SOA formation in the
CMAQ simulations; issues in the detailed CMAQ modeling of some PM precursors;
reliance for ozone modeling on a 12-km grid, suggesting NOX inhibition of ambient ozone
levels may be under-represented in some urban areas; and some emissions estimation
geographic scale/resolution issues. In all cases but the ozone grid resolution and
modeling of SOA formation, the effect of these uncertainties on our estimate of net
benefits is of uncertain direction. In addition, in all but one case, modeling of SOA
formation, we believe the impact of these uncertainties is probably minor, or of an
influence less than five percent of the total net benefits, based on current information.
Use of the CMAQ model platform, which has been evaluated in many contexts and used
extensively by EPA for broad regulatory analyses such as the Second Prospective, has
been a major factor enhancing our understanding of the impact of air quality modeling
exercises such as this.
Another factor contributing to our understanding of key uncertainties is that the air
quality modeling analysis conducted for the second Section 812 prospective study used
national-scale modeling databases originally prepared by EPA for use in other recent
modeling exercises conducted to support national rulemaking, including the latest
available meteorological and other input databases (for 2002). Given that the modeling
databases were originally prepared and utilized by EPA in other analyses, a
comprehensive performance evaluation was not undertaken as part of this Section 812
prospective analysis; though the overall projections were assessed using the Atmospheric
Model Evaluation Tool (AMET), which showed bias and error statistics for our results
were within the acceptable range for model performance.40 As noted in Table 4-1, biases
or uncertainties could be manifest in the simulated concentration fields due to the use of
the 36- and 12-km resolution grids, which might not be sufficiently detailed to resolve
certain sub-grid scale processes in portions of the modeling domain. All air quality
modeling exercises are affected by inherent uncertainties in model formulation,
meteorological inputs, and emission inventory estimates. Nevertheless, the modeling was
conducted following current EPA guidelines and in a manner consistent with EPA
approaches/practice for similar national-scale modeling exercises.
One factor identified in Table 4-1 involves uncertainties associated with corrections to the
air quality outputs completed coincident with the Council review of the study outputs.
These corrections, reflecting the need to adjust some categories of direct fine particulate
emissions for the without-CAAA scenario, and to incorporate adjustments to take account
of processes that remove fugitive dust from the ambient air at or close to the source of
emissions, owing to the effect of forests, vegetation, and urban structures on fugitive dust,
40ICF International, Evaluation of CMAQ Model Performance for the 812 Prospective II Study, November 24, 2009, page 31
4-23
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
were necessary because of issues identified through quality control assessments the
Project Team completed. As noted in the table, we believe these factors have been
addressed through carefully designed post-hoc adjustment of the CMAQ results, however
in both cases it would have been preferable to have made the adjustments prior to running
the CMAQ model. Resource and time limitations unfortunately prevented the Project
Team from re-estimating the CMAQ results to account for these adjustments.
Perhaps surprisingly, our assessment is that only one of these factors, uncertainty in
secondary organic aerosol formation, constitutes a major source of uncertainty. This
result could reflect our inability to apply alternative quantitative air quality modeling
tools in this already resource-intensive step in the analytic chain, although it is also clear
that the CMAQ model best reflects the state-of-the-art for the type of national scale air
quality modeling necessary to support this benefit-cost analysis. As we discuss in
Chapter 7, the overall contribution of this step in the analytic chain to uncertainty in net
benefits, compared to other steps, may be considerably less, because of the ability to
calibrate model results to monitor values for at least the year 2000 with-CAAA scenario.
It is worth noting, however, that as a whole the air quality modeling process very likely
contributes a greater than 10 percent uncertainty, of indeterminate direction, to the overall
uncertainty in benefits estimates. In addition, it is clear there are uncertainties introduced
by the ex post adjustment of some primary PM emissions estimates and the procedure
used to re-calibrate the CMAQ air quality to account for this emissions adjustment.
Although we argue that the overall effect of this source of uncertainty on the net benefits
is probably minor, in some locations ambient PM from primary PM emissions can be
more important than secondarily formed fine particles. Overall, we believe that our
application of the MATS monitor calibration procedure, which provides a speciated
calibration to ensure better agreement between air quality modeling results and
comparable monitor data, provides the best attainable consistency between our air quality
simulation results and monitored values - the ability to calibrate our results to detailed
monitor data in this step of the analytic chain provides considerably greater confidence
that our results are "ground-truthed" as much as possible to real world conditions.
TABLE 4-1. KEY UNCERTAINTIES ASSOCIATED WITH AIR QUALITY MODELING
POTENTIAL SOURCE OF ERROR
Unknown meteorological
biases in the 12-km western
and 36-km MM5 domains due
to the lack of model
performance evaluations.
DIRECTION OF
POTENTIAL BIAS FOR NET
BENEFITS
Unable to determine
based on current
information.
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
Probably minor. Other evaluations
using 2002 and similar meteorology
and CMAQ have shown reasonable
model performance, but significant
effects on nitrate results in western
areas with wintertime PM2.s
problems.
4-24
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF ERROR
DIRECTION OF
POTENTIAL BIAS FOR NET
BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
Known metrological biases in
the 12-km eastern MM5
domain. MM5 has a cold bias
during the winter and early
spring, and has a general
tendency to underestimate
the monthly observed
precipitation. MMS's under
prediction was greatest in the
fall and least in the spring
months.
Unable to determine
based on current
information.
Probably minor. These biases would
likely influence PM2.5 formation
processes, which was modeled on
the 36-km domain.
Secondary organic aerosol
(SOA) chemistry. CMAQ
version 4.6 has known biases
(underprediction) in SOA
formation.
Underestimate.
Possibly major. The modeling
system underpredicts SOA, which
has both biogenic and
anthropogenic components.
Reductions in NOx can reduce both
biogenic and anthropogenic SOA and
reductions in VOC will reduce
anthropogenic SOA. Since both of
these precursors are significantly
impacted by the CAAA, there may
be large benefits from SOA related
reductions that are not currently
captured by the modeling system.
The CMAQ modeling relies on
a modal approach to modeling
PM2.s instead of a sectional
approach. The modal
approach is effective in
modeling sulfate aerosol
formation but less effective in
modeling nitrate aerosol
formation than the sectional
approach.
Unable to determine
based on current
information.
Probably minor in the eastern U.S.
where annual PM2.5 is dominated by
sulfate. Potentially major in some
western U.S. areas where PM2.5 is
dominated by secondary nitrate
formation.
Limited model performance
evaluation of CMAQ for 2002.
Unable to determine
based on current
information.
Probably minor. While a
comprehensive model evaluation
was not completed, the overall
results of the CMAQ runs for the
Second Prospective were assessed
using AMET, and bias and error
statistics were within acceptable
ranges. Further, our application of
the MATS procedure provides
further assurance that air quality
results used in the subsequent
health assessments are consistent
with available monitor data.
4-25
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF ERROR
DIRECTION OF
POTENTIAL BIAS FOR NET
BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
Ozone modeling relies on a
12-km grid, suggesting NOX
inhibition of ambient ozone
levels may be under-
represented in some urban
areas. Grid resolution may
affect both model
performance and response to
emissions changes.
Unable to determine
based on current
information.
Probably minor. Though potentially
major 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.
Emissions estimated at the
county level (e.g., low-level
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.
Unable to determine
based on current
information.
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.
Use of MATS relative response
factors to calculate changes in
PM2.5
Indeterminate
Probably minor. Using MATS, air
quality modeling results were
projected in a "relative" sense. In
this approach, the ratio of future
year model predictions to base year
model predictions are used to
adjust ambient measured data up or
down depending on the relative
(percent) change in model
predictions for each location. The
use of ambient data as part of the
calculation helps to reduce
uncertainties in the future year
predictions, especially if the
absolute model concentrations are
over-predicted or under-predicted.
4-26
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF ERROR
DIRECTION OF
POTENTIAL BIAS FOR NET
BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO
KEY UNCERTAINTIES ON NET
BENEFITS ESTIMATE*
Modeling artifacts created by
changes in emissions inventory
estimation methods between
the 1990 inventories used for
the without-CAAA scenario
and the 2002 inventories used
for the with-CAAA scenarios
were mitigated through
application of adjustment
factors for primary PM from
non-EGU point sources, and
for the certain subsectors of
area sources, in the without-
CAAA case. Application of
these adjustments may result
in overestimated or
underestimated changes in
primary PM contributions to
ambient concentrations for
these particular sources.
Unable to determine
based on current
information.
Probably minor. While primary PM
can make a significant contribution
to ambient PM2.5 in some locations,
secondarily formed fine particles
dominate the estimates for ambient
concentration change in this
analysis. In addition, the effect of
the inventory adjustments was to
significantly reduce the differentials
between the control and
counterfactual scenarios, implying
any residual error is more likely to
reflect an underestimation bias than
an overestimation bias, particularly
since the non-EGU primary PM
reductions were adjusted to a
scenario differential of zero.
Adjustments to take account
of processes that remove
fugitive dust from the
ambient air at or close to the
source of emissions, owing to
the effect of forests,
vegetation, and urban
structures on fugitive dust.
Analysis of the chemical
species collected by ambient
air samplers suggests that the
modeling process may
overestimate PM-2.5 from
fugitive dust sources by as
much as an order of
magnitude, if not adjusted for
this effect. The Project Team
incorporated adjustments
post-CMAQ modeling but prior
to use of PM air quality
estimates in subsequent steps
of the analysis.
Unable to determine
based on current
information.
Probably minor. If adjustment
factors had been applied as part of
the CMAQ modeling, evidence
suggests the entrainment effect
would have been adequately
accounted for. The largely linear
processes of direct PM emissions to
air quality suggest that our post-hoc
adjustment should also be adequate
to account for this factor. Further
assurance that this factor has been
accounted for is our application of
the MATS monitor calibration
procedure, which provides a
speciated calibration to ensure
better agreement between air
quality modeling results and
comparable monitor data, and the
fact that the adjustment applies to
both scenarios, further mitigating
the impact of this source of
uncertainty.
* The classification of each potential source of error is based on those used in the First
Prospective Analysis. The classification of "potentially major" is used if a plausible alternative
assumption or approach could influence the overall monetary benefit estimate by approximately
5% or more; if an alternative assumption or approach is likely to change the total benefit
estimate by less than 5%, the classification of "probably minor" is used.
4-27
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 5 - ESTIMATION OF HUMAN HEALTH EFFECTS AND
ECONOMIC BENEFITS
A large portion of the overall
benefits of the Clean Air Act
Amendments (CAAA) of 1990 are
due to human health benefits from
improved air quality. As part of
the Second Prospective analysis of
these amendments, we identified
and, where possible, estimated the
magnitude of health benefits
Americans are likely to realize in
future years as a result of the
CAAA. We express these health
benefits as avoided cases of air
pollution-related health effects,
such as premature mortality, heart
disease and respiratory illness.
Human health benefits of the 1990
CAAA can be attributed to reduced
emissions of criteria pollutants
(Titles I through IV), and reduced
emission of ozone depleting
substances (Title VI), however as highlighted in Chapter 1 the Second Prospective
focuses primarily on human health effects attributed to the reduction of criteria pollutants,
and within that category, health benefits associated with reduced exposure to fine
particulate matter (PM2 5) and ozone, as these are the largest contributors to the overall
health benefits estimates.
The goal in a benefit-cost analysis such as the Second Prospective is to develop estimates
of the monetary value of benefits wherever possible - doing so facilitates comparison and
aggregation of monetized health benefits across endpoints. Therefore, we assigned a
dollar value to avoided incidences of each health effect. We obtained valuation estimates
from the economic literature and report them in "dollars per case avoided." We report
each of the monetary values of benefits applied in this analysis in terms of a central
estimate and a probability distribution around that value. The statistical form of the
probability distribution varies by endpoint.
AirC
He;
Ecor
Scenario Development
I
Sector Modeling
I
J
.
Emissions Direct Cost
1
uality Modeling
1 '
alth Welfare
1 '
lomic Valuation
1
1
r
Benefit-Cost Comparison
5-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
This chapter presents an overview of our approach to modeling changes in adverse health
effects and applying monetary value to these benefits, summarizes the results for major
health effect categories and discusses key uncertainties related to the analysis. As noted
above, the chapter focuses primarily on the human health effects associated with
exposure to criteria pollutants, however we also present the methodology and results of a
case study of health benefits from a single air toxic pollutant (benzene) for a particular
area of the United States (the Houston metropolitan area).
OVERVIEW OF APPROACH
We estimate the impact of the CAAA on human health by analyzing the difference in the
expected incidence of adverse health effects between a "with-" and a "without-CAAA"
regulatory scenario. As described in Chapter 1, the without-CAAA scenario assumes no
further controls on criteria pollutant emissions aside from those already in place in 1990,
while the with-CAAA scenario assumes full implementation of the 1990 CAAA. The
analysis uses a sequence of linked analytical models to estimate health benefits, also
described in Chapter 1, which includes forecasts of implementation activities undertaken
in response to the CAAA, estimates of pollutant emissions associated with each scenario
(see Chapter 2) and air quality modeling of criteria pollutant emissions under each
scenario (see Chapter 4).
Estimating health effects benefits from air quality modeling results involves three key
steps, described in greater detail below. The first step involves estimating the exposure of
individuals to air pollutants. Although exposure to air pollutants can occur in both
outdoor and indoor environments, for our purposes it is appropriate to focus on outdoor
air pollution concentrations as a measure of human exposure. The main reason is that, in
the second step of our approach, estimating the human response to exposure, the exposure
measures used in the epidemiological studies used to derive human response are typically
based on outdoor concentrations. These "concentration-response functions" were
developed to relate outdoor concentrations to changes in the incidence of health effects
and mortality in response to pollutant exposure. The third step, valuation of avoided
human health risk, is accomplished by application of estimates from the literature to
characterize unit values per case avoided.
A critical tool in EPA's analyses of health benefits is the Environmental Benefits
Mapping and Analysis Program (BenMAP), developed and continuously maintained by
EPA's Office of Air and Radiation.41 BenMAP is capable of accepting a wide range of
air quality inputs, and then performing exposure analysis that includes calibration of
model results to monitor data for historical years, assessing the changes in health effects
incidence resulting from those exposures, and estimating the monetized value of those
avoided health effects. Health effects in BenMAP are based on differences in two
scenarios of exposure, and health effects and valuation estimates reflect the implications
of the difference in exposure across scenarios, rather than absolute estimates of incidence
41 For more information, see the BenMAP User's Manual and Appendices, September 2008, Prepared for the Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC, by Abt Associates Inc.
5-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
associated with in any given scenario. BenMAP required three types of inputs for this
analysis: 1) forecasted changes in air quality from the without-CAAA to the with-CAAA
scenarios in 2000, 2010 and 2020; 2) health impact functions that quantify the
relationship between the forecasted changes in exposure and expected changes in adverse
health effects; and 3) health valuation functions that assign a monetary value to changes
in specific health effects. We describe each of these inputs in greater detail below. The
outputs of BenMAP for this analysis include central estimates and distributions of health
effects incidence and valuation, at the national and county level, for each of the three
target years of analysis.
The Project Team also estimates two other outputs related to avoided premature mortality
attributed to the CAAA: life-years lost, and changes in life expectancy. EPA developed a
separate model, the Population Simulation model, to generate these outputs. As
described below, the population simulation approach provides some advantages over the
BenMAP model in terms of simulation of the dynamic effects of mortality across a
population through time, but also has several significant disadvantages relative to
BenMAP in terms of the spatial resolution of pollutant exposure estimates. As a result,
the population simulation approach operates as a supplement to the BenMAP-based
primary estimates for selected measures of the impact of reducing risks of premature
mortality.
EXPOSURE ASSESSMENT
As described in Chapter 4, the Project Team used the Community Multi-scale Air Quality
(CMAQ) integrated modeling system to simulate the physical and chemical processes
that govern the formation, transport, and deposition of gaseous and particulate species in
the atmosphere. The CMAQ results serve as the basis of the air quality inputs required
for BenMAP. For particulate matter, the CMAQ model was applied for an annual
simulation period (January through December) and utilized a 36-km resolution modeling
domain that encompasses the contiguous 48 states. For ozone and related species, the
CMAQ model was applied for a five-month simulation period that captures the key
ozone-season months of May through September, and used two 12-km resolution
modeling domains (that when combined cover the contiguous 48 U.S. states).
We also described in Chapter 4 the adjustment of the CMAQ results generated by
combining those results with observed monitoring data, using a method known as the
monitor and model relative adjustment procedure. This technique was applied for the PM
estimates using a program called the Modeled Attainment Test Software (MATS) (see
Chapter 4 for a detailed description of this process). The resulting 36 km grid cell
concentrations for PM were then used as inputs for BenMAP. For ozone, a similar
adjustment process was completed, but the analysis was done directly within BenMAP,
using the enhanced Voronoi Neighbor Averaging (eVNA) procedure.42 The eVNA and
As noted in Chapter 4, eVNA and VNA are procedures for interpolating values from nearby monitors using inverse distance
squared weighting using Voronoi Neighbor Averaging. This is an algorithm that identifies a set of monitors close to the grid
cell (called "neighbors") and then estimates the PM species concentration in that grid cell by calculating an inverse-
distance weighted average of the monitor values (i.e., the concentration values at monitors closer to the grid cell are
weighted more heavily than monitors that are further away). See the BenMAP manual for further information on the eVNA
5-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
MATS procedures provide gridded estimates of outdoor air quality at the same grid
resolution as the CMAQ results. These procedures also provide a means for calibrating
model results in those grid cells where no monitors exist, combining both model results
with nearby monitor results to yield a "surface" of air quality that avoids the problems
with direct extrapolation of results from monitors not located within a grid cell boundary.
HEALTH IMPACT FUNCTIONS
Health impact functions estimate the change in a health endpoint of interest, such as
hospital admissions, for a given change in ambient pollutant concentration. A standard
health impact function has four components: 1) the size of the potentially affected
population; 2) a baseline incidence rate for the health effect (obtained from a source of
public health statistics, such as the Centers for Disease Control, or sometimes from an
epidemiological study itself); 3) a concentration-response (C-R) function (derived from
epidemiological studies), which relates the change in the number of individuals in a
population exhibiting a "response" to a change in pollutant concentration experience to
the size of the exposed population; and 4) the estimated change in the relevant pollutant
concentration. The first three of these components are discussed in further detail below.
The fourth is generated through the air quality modeling and exposure estimation
procedure discussed above.
Potentially Affected Populations
Health benefits resulting from the CAAA are related to the change in air pollutant
exposure experienced by individuals. Because the expected changes in pollutant
concentrations vary from location to location, individuals in different parts of the country
may not experience the same level of health benefits. This analysis apportions benefits
among individuals by matching the change in air pollutant concentration in a grid cell
with the size of the population that experiences that change.
BenMAP incorporates 2000 U.S. Census Bureau block-group population data to
determine the specific populations potentially affected by ozone and PM2 5. For future
years (2010 and 2020), BenMAP scales the 2000 Census-based population estimates
using the ratio of forecasted and 2000 county-level population estimates provided by
Woods and Poole (2007).43
procedure. Abt Associations (2008). BenMAP: Environmental Benefits Mapping and Analysis Program User's Manual.
Prepared for the U.S. Environmental Protection Agency's Office of Air Quality Planning and Standards, Research Triangle
Park, NC, September.
43 Woods 6t Poole Economics Inc., 2007. Complete Demographic Database. Washington, DC.
http://woodsandpoole.com/index.php.
5-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Baseline Incidence Rates
Baseline incidence rates are needed to convert the relative changes of a health effect in
relation to a specific change in air pollution, which are reported in epidemiological
studies, into the number of avoided cases. For instance, an epidemiological study might
report that for a 10 ppb decrease in daily ozone levels, hospital admissions decrease by
three percent. This estimate must then be multiplied by a baseline incidence rate (i.e., an
estimate of the number of cases of the health effect per year) and the total population to
determine how this three percent decrease translates into the number of fewer cases.
For this analysis, we used nationally-representative age-specific incidence and prevalence
rates, where available, for each health endpoint. We obtained these data from a variety of
sources, such as the CDC, the National Center for Health Statistics and the American
Lung Association. Information from individual epidemiological studies was used if data
from other sources were not available, as these data are often specific to the study
population and location and therefore may not be as nationally representative.44 For
future years, mortality rates are projected based on available Bureau of the Census
projections - other projected baseline incidence rates are generated to be consistent with
the projections of population growth incorporated into BenMAP.
Con cent ration-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 functions are equations that relate the change in the number of individuals in
a population exhibiting a "response" (in this case an adverse health effect such as
respiratory disease) to a change in pollutant concentration experienced by that population.
PM2 5 and ozone have been associated with a number of adverse health effects in the
epidemiological literature, such as premature mortality, hospital admissions, emergency
room visits, and respiratory and cardiovascular disease. The published scientific
literature contains information that supports the estimate of some, but not all, of these
effects. Thus, it is not possible currently to estimate all of the human health benefits
attributable to the CAAA. In addition, for some of the health effects we do quantify, the
current economic literature does not support the estimation of the economic value of
these effects. Table 5-1 lists the human health effects of these pollutants that have been
identified, indicating which have been included in our benefits estimates and those that
we did not quantify. See Chapter 2 of Health and Welfare Benefits Analyses to Support
the Second Section 812 Benefit-Cost Analysis of the Clean Air Act, for a specific list of
the C-R functions used for each health endpoint.
H See Health and Welfare Benefits Analyses to Support the Second Section 812 Benefit-Cost Analysis of the Clean Air Act,
February 2011, for a list of data sources and average baseline incidence rates for each health effect.
5-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-1. HUMAN HEALTH EFFECTS OF OZONE AND PM2.5
POLLUTANT/EFFECT
QUANTIFIED AND MONETIZED IN BASE
ESTIMATES*
UNQUANTIFIED EFFECTS°'H-CHANGES IN:
PM/Healthb
Premature mortality based on both cohort
study estimates and on expert elicitationc>d
Bronchitis: chronic and acute
Hospital admissions: respiratory and
cardiovascular
Emergency room visits for asthma
Nonfatal heart attacks (myocardial
infarction)
Lower respiratory symptoms
Minor restricted-activity days
Work loss days
Asthma exacerbations (asthmatic population)
Upper Respiratory symptoms (asthmatic
population)
Infant mortality
Subchronic bronchitis cases
Low birth weight
Pulmonary function
Chronic respiratory diseases other than
chronic bronchitis
Morphological changes
Altered host defense mechanisms
Cancer
Non-asthma respiratory emergency room
Visits
UVb exposure (+/-)e
Ozone/Health
Premature mortality: short-term exposures
Hospital admissions: respiratory
Emergency room visits for asthma
Minor restricted-activity days
School loss days
Outdoor worker productivity
Cardiovascular emergency room visits
Asthma attacks
Respiratory symptoms
Chronic respiratory damage
Increased responsiveness to stimuli
Inflammation in the lung
Premature aging of the lungs
Acute inflammation and respiratory cell
damage
Increased susceptibility to respiratory
infection
Non-asthma respiratory emergency room
Visits
UVb exposure (+/-)e
a Primary quantified and monetized effects are those included when determining the primary estimate of total
monetized benefits of the alternative standards.
b In addition to primary economic endpoints, there are a number of biological responses that have been
associated with PM health effects including morphological changes and altered host defense mechanisms. The
public health impact of these biological responses may be partly represented by our quantified endpoints.
c Cohort estimates are designed to examine the effects of long-term exposures to ambient pollution, but relative
risk estimates may also incorporate some effects due to shorter term exposures (see Kunzli et al., 2001 for a
discussion of this issue).
d While some of the effects of short-term exposure are likely to be captured by the cohort estimates, there may
be additional premature mortality from short-term PM exposure not captured in the cohort estimates included in
the primary analysis.
e May result in benefits or disbenefits.
f In addition to primary economic endpoints, there are a number of biological responses that have been
associated with ozone health including increased airway responsiveness to stimuli, inflammation in the lung,
acute inflammation and respiratory cell damage, and increased susceptibility to respiratory infection. The public
health impact of these biological responses may be partly represented by our quantified endpoints.
g The categorization of unqualified health effects is not exhaustive.
h Health endpoints in the unqualified benefits column include both a) those for which there is not consensus on
causality and b) those for which causality has been established but empirical data are not available to allow
calculation of benefits.
5-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
We rely on the most recently available, published scientific literature to ascertain the
relationship between air pollution and adverse human health effects. We use a set of
criteria outlined in Table 5-2 to evaluate potential studies to use as the basis for the C-R
function. These criteria include consideration of whether the study was peer-reviewed,
the study design and location, and characteristics of the study population, among others.
In addition, we consider the input of the Council advising EPA for this study, as well the
specific advice of the Health Effects Subcommittee (HES) of the Council, which
explicitly focused on the health effects estimation component of the study. Overall, the
selection of C-R functions for benefits analysis is guided by the goal of achieving a
balance between comprehensiveness and scientific defensibility.
Epidemiological studies provide the basis for the C-R functions used in the health impact
functions for assessing benefits of the CAAA. These studies also provide an indication of
a portion of the uncertainty associated with the C-R function, by reporting a confidence
interval around the mean value, which we use to derive a low, central and high estimate
of avoided cases. However, this range only represents the statistical error in the
estimates, which is related to the study population size and frequency of outcome.
Several other sources of uncertainty exist in the relationship between ambient pollution
and the health outcomes, including model uncertainty, potential confounding by factors
that are both correlated with the health outcome and each other, and potential
misclassification of the study population exposures. For a full list of uncertainties related
to application of a C-R function to estimate benefits, see the Uncertainty section of this
chapter and the Second Prospective Uncertainty Report, Uncertainty Analyses to Support
the Second Section 812 Benefit-Cost Analysis of the Clean Air Act.
EPA recently conducted an expert elicitation (EE) study, which is the formal elicitation of
subjective judgments, in order to more fully characterize the uncertainty surrounding the
PM2 5/mortality C-R function. This study allowed experts to consider and integrate
several sources of uncertainty in the form of a probability distribution of the C-R
function. As discussed further below, the EE study results helped to inform our selection
of a primary C-R function to estimate avoided premature mortality due to CAAA-related
PM25 exposure reductions.
Avoided premature mortality is the largest contributor to the monetized health benefits of
PM2 5 and ozone. Therefore, we describe below in further detail the specific C-R
functions selected to quantify CAAA-related avoided deaths.
5-7
-------
TABLE 5-2.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
SUMMARY OF CONSIDERATIONS USED IN SELECTING C-R FUNCTIONS
CONSIDERATION
Peer-Reviewed Research
Study Type
Study Period
Population Attributes
Study Size
Study Location
Pollutants Included in
Model
Measure of PM
Economically Valuable
Health Effects
Non-overlapping
Endpoints
COMMENTS
Peer-reviewed research is preferred to research that has not undergone the peer-
review process.
Among studies that consider chronic exposure (e.g., over a year or longer),
prospective cohort studies are preferred over ecological studies because they
control for important individual-level confounding variables that cannot be
controlled for in ecological studies.
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 lifestyle over time. However, when there are
only a few studies available, studies from all years will be included.
The most technically appropriate measures of benefits would be based on impact
functions that cover the entire sensitive population but allow for heterogeneity
across age or other relevant demographic factors. In the absence of effect
estimates specific to age, sex, preexisting condition status, or other relevant
factors, it may be appropriate to select effect estimates that cover the broadest
population to match with the desired outcome of the analysis, which is total
national-level health impacts. When available, multi-city studies are preferred to
single city studies because they provide a more generalizable representation of
the C-R function.
Studies examining a relatively large sample are preferred because they generally
have more power to detect small magnitude effects. A large sample can be
obtained in several ways, either through a large population or through repeated
observations on a smaller population (e.g., through a symptom diary recorded for
a panel of asthmatic children).
U.S. studies are more desirable than non-U.S. studies because of potential
differences in pollution characteristics, exposure patterns, medical care system,
population behavior, and lifestyle.
When modeling the effects of ozone and PM (or other pollutant combinations)
jointly, it is important to use properly specified impact functions that include
both pollutants. Using single-pollutant models in cases where both pollutants are
expected to affect a health outcome can lead to double-counting when pollutants
are correlated.
For this analysis, impact functions based on PM2.5 are preferred to PM10 because
of the focus on reducing emissions of PM2.5 precursors, and because air quality
modeling was conducted for this size fraction of PM. Where PM2.5 functions are
not available, PM10 functions are used as surrogates, recognizing that there will
be potential downward (upward) biases if the fine fraction of PM10 is more (less)
toxic than the coarse fraction.
Some health effects, such as forced expiratory volume and other technical
measurements of lung function, are difficult to value in monetary terms. These
health effects are not quantified in this analysis.
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.
5-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
PM Mortality C-R Function
The estimated relationship between participate matter exposure and premature mortality
is one of the most important parameters in the overall quantified and monetized benefit
estimate for this study. An extensive base of literature exists to support development of
the C-R function linking fine particulate matter exposure with premature mortality. Our
knowledge of both the potential biological mechanisms linking PM2 5 exposure with
mortality and the potential magnitude of this effect has grown since the First Prospective
was completed as the result of continued research and follow-up of existing study
populations. Both short-term and long-term epidemiological studies have been conducted
to examine the PM/mortality 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 concentrations. Long-
term exposure studies examine the potential relationship between longer-term (e.g.,
annual) changes in exposure and annual mortality rates. Although positive, significant
results have been reported using both of these study types, we rely exclusively on long-
term studies to quantify PM mortality effects. This is because cohort studies are able to
discern changes in mortality rates due to long-term exposure to elevated air pollution
concentrations. This provides a better match to the benefits of air pollution control
programs under the CAAA, which are also focused on reducing long-term exposure.
These effect estimates may also include some of the mortality changes due to short-term
peak exposures.45 Therefore, the use of C-R functions from long-term studies is likely to
yield a more complete assessment of the effect of PM on mortality risk.
Among long-term PM studies, we prefer those using a prospective cohort design to those
using an ecologic or population-level design. Prospective 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 of daily
mortality) and ambient levels of air pollution. Prospective cohort studies are preferred
because they are better at controlling a source of uncertainty known as "confounding."
Confounding is the mis-estimation of an association that results if a study does not
control for factors that are correlated with both the outcome of interest (e.g., mortality)
and the exposure of interest (e.g., PM exposure). For example, smoking is associated
with mortality. 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 cohort studies are better at
controlling for confounding 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.
45 See Kunzli et al. (2001) for a discussion of this issue.
5-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Two major prospective cohort studies have been conducted in the U.S.: the American
Cancer Society (ACS) study and the Six Cities study. These two cohorts are large,
produce consistent results, provide broad geographic coverage and have been
independently reexamined and reanalyzed. Strengths of the ACS study over the Six
Cities study include greater geographic coverage (50 U.S. cities) and larger sample size.
However, a key limitation of this study is a recruitment method that led to a study
population with higher income, more education, and greater proportion of whites than the
general U.S. population. In addition, available monitoring data was often assigned to all
of the individuals within a large metropolitan area, potentially allowing for exposure
misclassification.46 Both of these limitations could imply that the ACS results are
potentially biased low. The Six Cities study included a more representative sample of
subjects within each community and set up monitors purposefully for the study. It was
therefore able to assign exposures at a finer geographic scale. However, this study only
included six cities and therefore may not be representative of the entire U.S. population,
mix of air pollutants, and other potentially important factors.
The extensive epidemiological literature is complemented by EPA's 2006 expert
elicitation (EE) study that asked 12 leading experts in PM health effects to integrate this
pool of knowledge with the various sources of uncertainty that hinder our ability to
precisely identify the true mortality impact of a unit change in annual PM2 5 concentration
(lEc, 2006). The results of the expert elicitation study showed three important findings:
first, that advances in the scientific literature led many of the interviewed scientists to
espouse greater confidence in the linkage between PM2 5 exposure and mortality; second,
that many of the experts believed that the central estimate of the mortality effect was
considerably higher than the Pope et al. (2002) result used in the First Prospective; and
third, that most of the experts' uncertainty distributions of the mortality effect reflected a
much wider range of possible values, both high and low, than were used in the First
Prospective study. The expert elicitation study does not, however, provide an integrated
distribution across all 12 experts of possible values for the PM-mortality C-R function.
Based on consultations with the Council's Health Effects Subcommittee (HES), the 812
Project Team developed a distribution of C-R function coefficients (i.e., the percent
change in annual all-cause mortality per one ug/m3 change in annual average PM2 5) for
use in the PM-mortality C-R function for the Second Prospective study. This distribution
is rooted in the epidemiological studies that most inform our understanding of the PM-
mortality C-R function, but reflects the broader findings of the EE study. We based the
primary C-R coefficient estimate of the Second Prospective study on a Weibull
distribution with a mean of 1.06 percent decrease in annual all-cause mortality per one
ug/m3. This mean is roughly equidistant between the results of the two most well-studied
PM cohorts, the ACS cohort (0.58, as derived from Pope et al., 2002) and the Six Cities
cohort (1.5, as derived from Laden et al., 2006), both of whose results have been robust to
continued follow-up and extensive re-analysis. Half of the coefficient values in this
* Studies have shown that greater spatial resolution of exposures can result in increased effect estimates (Jerrett et al.,
2005).
5-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
distribution fall between these two studies, one-quarter are higher than the Laden mean
estimate, and one-quarter are lower than the Pope mean estimate; however all coefficient
values are greater than zero. This distribution is consistent with the EE results described
above, showing considerable support for higher values based on results from more recent
studies (e.g., the Laden et al. (2006) Six Cities follow-up) and concerns cited by the
Council HES that the ACS cohort results may underestimate the true effect. The use of
all positive values is consistent with both the increased confidence in a causal link
between PM2 5 exposure and mortality shown in the EE study and the lack of evidence in
general to support a threshold for mortality effects of PM25 in the U.S. population.47
The results of two recently published cohort studies provide additional support for the
selection of the Weibull distribution as the primary estimate for the PM Mortality C-R
function. The first is a large retrospective cohort study of over 13 million Medicare
participants (i.e., those aged 65 and above) throughout the US (Eftim et al. 2008; Zeger et
al. 2008). When the entire Medicare cohort was analyzed, authors found a 6.8 percent
change in annual all-cause mortality in the eastern US (95% CI: 4.9-8.7) and a 13.2
percent change in the central US (95% CI: 9.5-16.9) per 10 ug/m3 change in the long-term
(six-year) average annual PM2 5. There was no association found in the western US
(Zeger et al., 2008). These results are similar to the interquartile range of the Weibull
distribution selected for the primary estimate for the Second Prospective. An analysis
restricted to those living in the locations corresponding to the ACS and Six Cities cohort
study analyses yielded percent changes in annual all-cause mortality per 10 ug/m3 of
PM25 of 10.9 (95% CI: 9.0-12.8) and 20.8 (95%CI: 14.8-27.1) respectively, which are
somewhat higher than the estimates reported in the original studies (Eftim et al., 2008).48
One possible explanation for this difference is the lack of control for lifestyle factors in
the analyses by Eftim et al., such as smoking, potentially leading to confounded results.
The second study is a prospective cohort of female nurses in the Northeastern and
Midwestern regions of the US (Puett et al. 2008 and 2009). An increase of 10 ug/m3 of
PM2 5 in the previous year was associated with a 26 percent increase in annual all-cause
mortality (a hazard ratio of 1.26 with a 95% CI ranging from 1.02 to 1.54).49 This
estimate is at the upper end of our primary estimate Weibull distribution (roughly
equivalent to the 95th percentile). However, this study covered only two regions of the
country and included only females and therefore may not be generalizable to the general
population of the US.
A final topic concerns EPA's choice to estimate avoided mortality and morbidity
associated with reductions in fine particles using estimates of changes in exposure to fine
47 See "Health Effects Subcommittee of the Council. Review of EPA's Draft Health Benefits of the Second Section 812
Prospective Study of the Clean Air Act." (EPA-COUNCIL-10-001), available at http://www.epa.gov/advisorycouncilcaa
48 Note that these results are based on a slightly different air quality dataset than the analysis of the full cohort. The
nationwide estimate is based on a six-year average (2000-2005) and the ACS and Six Cities location-specific results are
based on two years of data (2000-2002).
49 Biennial questionnaires on lifestyle factors were administered to participants, allowing for control of a number of
individual-level confounders.
5-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
particle mass as the exposure input in the damage function. The implication of this
approach is that we assume that all fine particles, regardless of their chemical
composition, are equally potent per unit concentration in producing premature mortality
and other health outcomes. If it could be shown that fine particle species exhibit
significantly differentiated toxicity, then from a benefits analysis perspective, treatment
of all fine particle species as equally toxic would lead to biased benefits estimates,
because the composition of fine particle mass varies over space and time, as do the fine
particle reductions resulting from different air pollutant control strategies. We believe
that these biases would likely be minor in an analysis such as the 812 study, which
evaluates a blended particle reduction strategy targeting multiple particle types across the
entire spectrum of control programs authorized under the Clean Air Act Amendments.
Nonetheless, we conducted a careful evaluation of the potential for characterizing
uncertainty in the differential toxicity of the components of fine particle pollution.
There exists a limited but growing literature addressing the health effects of various fine
particle components, including sulfate, nitrate, elemental carbon (EC), organic carbon
(OC), and metals.50 A number of epidemic logical studies, mostly time-series studies,
have associated one or more of the components of fine particle pollution individually
with mortality; however, so far no clear picture has emerged to implicate specific
components as being consistently more toxic than fine particles in general or to classify
any individual components of fine particle pollution as non-toxic. However, the
epidemiological evidence base is limited by the high correlations among many fine
particle components (and between those components and fine particles as a whole). It is
difficult to corroborate this evidence lexicologically, given the fact that human exposure
to single particle components is not a realistic scenario. The literature base continues to
expand, but significant investments in both epidemiological and toxicological research
are needed to understand the potentially complex systems of particle interactions that may
be responsible for the observed health effects of fine particle pollution.
Thus, while treatment of all fine particle components as equally toxic may lead to biases
in benefits estimates, we also acknowledge that any arbitrary assumption about the
differential toxicities of specific fine particle types may also lead to biases in benefits
estimates. Any of these biases may mask important spatial variation in the distribution of
benefits of Clean Air Act programs across the U.S. due to regional variation in fine
particle species mixes, which could affect selection of the most health beneficial
measures to meet Clean Air Act requirements such as the National Ambient Air Quality
Standards. However, the "equal toxicity" fine particle approach is rooted in both
biological considerations (i.e., the importance of particle size to toxicity) and in largely
consistent findings across an extensive set of epidemiological studies conducted across
countries, states, and cities that show PM2 5 concentrations are associated with increased
mortality and morbidity rates. This consistency of results across a variety of fine particle
50 For specific examples of research addressing differential toxicity of PM components, see Chapter 5 of Uncertainty Analyses
to Support the Second Section 812 Benefit-Cost Analysis of the Clean Air Act.
http://www.epa.gov/oar/sect812/mav10/IEc Uncertaintv.pdf
5-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-3.
mixes in different locations implies an equivalence of risk resulting from exposure to fine
particle masses with different concentrations of component species. We conclude that the
current evidentiary base from the epidemiological and toxicological literatures supports
the use of an equal toxicity assumption for the present study, especially since the fine
particle pollution reductions estimated herein reflect a variety of fine particle mixtures
across different locations and time frames. Furthermore, we conclude that current
information does not support specification of alternative concentration-response functions
that would be both scientifically sound and useful for development of policy-relevant
insights.
To provide further confidence that the results presented in this chapter are not likely to be
substantially affected by the possibility that PM2 5 species exhibit differential toxicity, the
Project Team developed and evaluated estimates of the overall population-weighted
exposure to PM species. The results are presented in Table 5-3 below, and graphically in
the two panels of the accompanying Figure 5-1. The results in Figure 5-lindicate that the
population-weighted composition of fine particulate matter is affected by the control
strategies applied in the CAAA, but the changes are relatively modest.51 We therefore
conclude that, even if species-specific toxicity estimates could be derived from the
existing literature, applying them in this study would not have a large effect on the
mortality results presented later in this chapter.
ESTIMATED POPULATION WEIGHTED EXPOSURE FOR PM2.5 SPECIES (MICROGRAMS
PER CUBIC METER)
Crustal
NO3
NH4
EC
OC
S04
2000
NO
CAAA
1.18
1.06
1.87
0.74
5.18
4.84
2000
WITH
CAAA
0.82
0.89
1.26
0.62
3.94
3.11
2010
NO
CAAA
1.27
1.17
1.96
0.77
5.36
5.02
2010
WITH
CAAA
0.86
0.81
1.03
0.48
3.86
2.48
2020
NO
CAAA
1.51
1.32
2.05
0.9
6.02
5.17
2020
WITH
CAAA
0.96
0.69
0.92
0.41
3.99
2.22
51 Note that data presented in Table 5-3 are for the most important PM2.5 components; some less important species, with
lower concentrations, are omitted.
5-13
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 5-1. DISTRIBUTION OF POPULATION WEIGHTED EXPOSURE TO PM2.5 SPECIES AS
PERCENTAGE OF TOTAL (TOP PANEL) AND IN MICROGRAMS PER CUBIC METER
(BOTTOM PANEL)
100%
80%
60%
40%
20%
ISO4
IOC
I EC
INH4
IN03
I Crustal
2000 2000 2010 2010 2020 2020
No With No With No With
CAAA CAAA CAAA CAAA CAAA CAAA
ISO4
IOC
I EC
INH4
IN03
I Crustal
2000 2000 2010 2010 2020 2020
No With No With No With
CAAA CAAA CAAA CAAA CAAA CAAA
Ozone Mortality C-R Function
Several recent epidemiological studies suggest that ozone exposure likely contributes to
premature mortality.52 Epidemiological data are also supported by recent human and
52 See, for example, National Research Council, 2008, Estimating Mortality Risk Reduction and Economic Benefits from
Controlling Ozone Air Pollution. A key recommendation of this NAS panel was that ozone mortality estimates from
available epidemiological studies represent a separate and additive effect to those from PM/mortality epidemiological
studies.
5-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
animal experimental data, which suggestive evidence for plausible pathways by which the
risk of respiratory or cardiovascular mortality could be increased by ambient ozone.
Multiple time-series epidemiological studies explore the relationship between short-term
ozone exposure and premature mortality. Most notably, a large multi-city study known
as the National Morbidity, Mortality, and Air Pollution Study (NMMAPS) was designed
to explore the association between several pollutants, including ozone, and daily
mortality that focused on large cities across the US where levels of pollutants were varied
(Samet et al., 2000). Two recently published studies based on the NMMAPS database
that focus on the ozone/premature mortality relationship are Bell et al. (2004) (95 U.S.
cities) and Huang et al. (2005) (19 U.S. cities). Another multi-city study by Schwartz
(2005) examined the relationship between short-term ozone exposure and mortality in 14
U.S. cities.
In addition to these multi-city estimates, C-R functions for short-term ozone mortality can
be derived from meta-analyses, which combine the results of several studies. Three
meta-analyses were performed to obtain a summary estimate of ozone-related mortality
risks and to attempt to describe heterogeneity in risk estimates (Ito et al., 2005; Levy et
al., 2005; Bell et al., 2005). Each of these studies used different statistical techniques and
datasets and examined statistical concerns, such as confounding, collinearity and possible
interaction effects.53
In general, effect estimates from the meta-analyses are higher than the multi-city results.
This could potentially be due to publication bias, as the meta-analyses relied solely on
published studies, which could be more likely to contain statistically significant results.
NMMAPS generally produces lower estimates than other epidemiological time-series
studies, however, which could reflect specific methodological choices made by these
investigators. Since these studies are associated with different strengths and limitations
and no single study emerges as the most suitable to use as the basis for our primary
estimate, we opted to use a pooled estimate, equally weighting the C-R functions from all
six of these studies.
In addition to time-series epidemiological studies, a limited number of studies examine
the cumulative effect of long-term exposure to ozone on mortality. One such recent study
(Jerrett et al., 2009) used study population data from the ACS cohort study along with
ozone monitoring data and reported a significant association between deaths from
respiratory causes and long-term ozone exposure. In a recent review of the 812 Second
Prospective Analysis methodology, the Council HES found the use of the Jerrett et al.
estimate as the primary estimate premature at this time, due to a lack of corroboration
from other cohort studies ,54
National Research Council (NRC) (2008). Estimating Mortality Risk Reduction and Economic Benefits from Controlling Ozone
Air Pollution. The National Academies Press, Washington, DC.
M See "Health Effects Subcommittee of the Council. Review of EPA's Draft Health Benefits of the Second Section 812
Prospective Study of the Clean Air Act." (EPA-COUNCIL-10-001), available at http://www.epa.gov/advisorycouncilcaa
5-15
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
HEALTH VALUATION FUNCTIONS
In environmental benefit-cost analyses, the dollar value of an environmental benefit, such
as improved health or avoidance of a case of illness, is the dollar amount necessary such
that the person would be indifferent between experiencing the benefit and possessing the
money. In most cases, the dollar amount required to compensate 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. Therefore, in economic terms, the "willingness-to-pay" (WTP) is the
appropriate measure of the value of avoiding an adverse effect. For example, the value of
an avoided respiratory symptom would be a person's WTP to avoid that symptom.
For most goods, WTP can be observed by examining actual market transactions. For
example, if a gallon of bottled drinking water sells for one dollar, it can be observed that
at least those persons who choose to purchase that good are willing to pay at least one
dollar for the water. For goods that are not exchanged in the market, such as most
environmental goods, valuation is not so straightforward. 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 in similar effects
or risk reductions, (e.g., non-toxic cleaners or bike helmets). Alternatively, surveys may
be used in an attempt to directly elicit WTP for an environmental improvement.
Wherever possible in this analysis, we use estimates of mean WTP. In cases where WTP
estimates are not available, we use the cost of treating or mitigating the effect as an
alternative estimate.
For example, 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 admission. These costs
of illness (COI) estimates generally 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. We use a range
of values for most environmental effects, to support the primary central estimate of net
benefits. Table 5-4 summarizes the mean unit value estimates that we use in this
analysis.
Valuation of Premature Mortality
Some forms of air pollution increase the probability that individuals will die prematurely.
We use C-R functions for mortality that express the increase in mortality risk as cases of
"excess premature mortality" per year. The benefit provided by air pollution reductions,
however, is the avoidance of small increases in the risk of mortality. By summing
individuals WTP to avoid small increases in risk over enough individuals, we can infer
the value of a statistical premature death avoided.55 For expository purposes, we express
this valuation as "dollars per mortality avoided," or "value of a statistical life" (VSL),
55 Because people are valuing small decreases in 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 the average
individual's WTP for that risk reduction is $50, then the implied value of a statistical premature death avoided in that
population is $50 x 100,000 = $5 million.
5-16
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
even though the actual valuation is of small changes in mortality risk experienced by a
large number of people. The economic benefits associated with avoiding premature
mortality were the largest category of monetized benefits in the First Prospective
Analysis and continue to be the largest source of monetized benefits for this Second
Prospective Analysis. Mortality benefits, however, are also the largest contributor to the
range of uncertainty in monetized benefits.
Because avoided premature mortality benefits are such an important part of this study's
results and findings, the remainder of this section provides an expanded discussion of
some of the issues in valuing the avoidance of mortality risks from air pollution. We first
discuss some characteristics of an "ideal" measure of the value of mortality risk
reductions from air pollution, and then review several dimensions in which the current
estimates fall short of the ideal measure for this study. For a more detailed discussion of
the factors affecting the valuation of premature mortality see the Uncertainty section of
this chapter and the Uncertainty Analyses to Support the Second Section 812 Benefit-Cost
Analysis of the Clean Air Act.
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 appear to be 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). An ideal economic benefits estimate of mortality risk
reduction would reflect these human characteristics, in addition to an individual's WTP to
improve one's own chances of survival plus WTP to improve other individuals' survival
rates.56 The ideal measure would also take into account the specific nature of the risk
reduction 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. 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 characteristics as age, health
state, and the current age to which the individual is likely to survive.
' For a more detailed discussion of altruistic values related to the value of life, see Jones-Lee (1992).
5-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-4. UNIT VALUES FOR ECONOMIC VALUATION OF HEALTH ENDPOINTS (2006$)
HEALTH ENDPOINT
CENTRAL ESTIMATE OF VALUE
PER STATISTICAL INCIDENCE
1990 INCOME
LEVEL
2020 INCOME
LEVEL
DERIVATION OF DISTRIBUTIONS OF ESTIMATES
Premature Mortality
(Value of a Statistical
Life)
Chronic Bronchitis
(CB)
$7,400,000
$8,900,000
Mean Value of Statistical Life (VSL) based 26 wage-risk and contingent valuation studies. A Weibull
distribution, with a mean of $7.4 million (in 2006$), provided the best fit to the 26 estimates. Note that
VSL represents the value of a small change in mortality risk aggregated over the affected population.
$399,000
$490,000
WTP =WTP -e~
The WTP to avoid a case of pollution-related CB is calculated as x 13 , where x is
the severity of an average CB case, WTP13 is the WTP for a severe case of CB, and B is the parameter
distribution of WTP for an air pollution-relevant, average severity-level case of 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, with the most likely value at severity level
6.5 and endpoints at 1.0 and 12.0; 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)). This process and the rationale for choosing it is described in detail in the Costs and Benefits of
the Clean Air Act, 1990 to 2010 (EPA, 1999).
Nonfatal Myocardial
Infarction (heart
attack)
7% discount rate
Age 0-24
Age 25-44
Age 45-54
Age 55-65
Age 66 and over
$84,171
$93,802
$98,366
$166,222
$84,171
No distributional information available. Age-specific cost-of-illness values reflect lost earnings and direct
medical costs over a 5-year period following a nonfatal Ml. Lost earnings estimates are based on Cropper
and Krupnick (1990). Direct medical costs are based on simple average of estimates from Russell et al.
(1998) and Wittels et al. (1990).
Lost earnings:
Cropper and Krupnick (1990). Present discounted value of 5 years of lost earnings (2006$):
age of onset: at 7%a
25-44 $9,631
45-54 $14,195
55-65 $82,051
Direct medical expenses: An average of (2006$):
1. Wittels et al. (1990) ($141,124-no discounting)
2. Russell et al. (1998), 5-year period ($28,787 at 3% discount rate; $27,217 at 7% discount rate)
5-18
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
HEALTH ENDPOINT
CENTRAL ESTIMATE OF VALUE
PER STATISTICAL INCIDENCE
1990 INCOME
LEVEL
2020 INCOME
LEVEL
DERIVATION OF DISTRIBUTIONS OF ESTIMATES
Hospital Admissions
All respiratory (ages
65+)
All respiratory (ages
0-2)
Chronic Obstructive
Pulmonary Disease
(COPD) (ages 65+)
Asthma Admissions
(ages <65)
Pneumonia
Admissions (ages 65+)
COPD, less asthma
(ages 20-64)
All Cardiovascular
(ages 65+)
All Cardiovascular
(ages 20-64)
Ischemic Heart
Disease (ages 65+)
Dysrhythmia (ages
65+)
Congestive Heart
Failure (ages 65+)
Emergency Room
Visits for Asthma
$23,711
$10,002
$17,308
$10,040
$23,004
$15,903
$27,319
$29,364
$33,357
$19,643
$19,619
$369
$23,711
$10,002
$17,308
$10,040
$23,004
$15,903
$27,319
$29,364
$33,357
$19,643
$19,619
$369
No distributions available. The COI point estimates (lost earnings plus direct medical costs) are based on
ICD-9 code level information (e.g., average hospital care costs and average length of hospital stay)
reported in Agency for Healthcare Research and Quality, 2000 (www.ahrq.qov). As noted in the text, no
adjustments are made to cost of illness values for income growth.
No distributional information available. Simple average of two unit COI values (2006$):
(1) $401.62, from Smith et al. (1997) and
(2) $336.03, from Stanford et al. (1999).
As noted in the text, no adjustments are made to cost of illness values for income growth.
5-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
HEALTH ENDPOINT
CENTRAL ESTIMATE OF VALUE
PER STATISTICAL INCIDENCE
1990 INCOME
LEVEL
2020 INCOME
LEVEL
DERIVATION OF DISTRIBUTIONS OF ESTIMATES
Respiratory Ailments Not Requiring Hospitalization
Upper Respiratory
Symptoms (URS)
Lower Respiratory
Symptoms (LRS)
Asthma
Exacerbations
Acute Bronchitis
Work Loss Days
(WLDs)
Minor Restricted
Activity Days (MRADs)
$28.8
$18
$50
$416
Variable (U.S.
median =
$149)
$59
$30.7
$19
$54
$512
$64
Combinations of the three symptoms for which WTP estimates are available that closely match those
listed by Pope et al. result in seven different "symptom clusters," each describing a "type" of URS. A
dollar value was derived for each type of URS, using mid-range estimates of WTP (lEc, 1994) to avoid
each symptom in the cluster and assuming additivity of WTPs. In the absence of information surrounding
the frequency with which each of the seven types of URS occurs within the URS symptom complex, we
assumed a uniform distribution between $10.8 and $50.5 (2006$).
Combinations of the four symptoms for which WTP estimates are available that closely match those listed
by Schwartz et al. result in 1 1 different "symptom clusters," each describing a "type" of LRS. A dollar
value was derived for each type of LRS, using mid-range estimates of WTP (lEc, 1994) to avoid each
symptom in the cluster and assuming additivity of WTPs. The dollar value for LRS is the average of the
dollar values for the 11 different types of LRS. In the absence of information surrounding the frequency
with which each of the 1 1 types of LRS occurs within the LRS symptom complex, we assumed a uniform
distribution between $8.1 and $28.6 (2006$).
Asthma exacerbations are valued at $50 per incidence, based on the mean of average WTP estimates for
the four severity definitions of a "bad asthma day," described in Rowe and Chestnut (1986). This study
surveyed asthmatics to estimate WTP for avoidance of a "bad asthma day," as defined by the subjects.
For purposes of valuation, an asthma exacerbation is assumed to be equivalent to a day in which asthma
is moderate or worse as reported in the Rowe and Chestnut (1986) study. The value is assumed have a
uniform distribution between $18.3 and $82.9 (2006$).
Assumes a 6-day episode, with the distribution of the daily value specified as uniform with the low and
high values based on those recommended for related respiratory symptoms in Neumann et al. (1994).
The low daily estimate of $20.5 (2006$) is the sum of the mid-range values recommended by lEc (1994)
for two symptoms believed to be associated with acute bronchitis: coughing and chest tightness. The
high daily estimate was taken to be twice the value of a minor respiratory restricted activity day, or $1 1 8
(2006$). The low and high daily values are multiplied by six to get the 6-day episode values.
No distribution available. Point estimate is based on county-specific median annual wages divided by 50
(assuming 2 weeks of vacation) and then by 5— to get median daily wage. U.S. Year 2000 Census,
compiled byGeolytics, Inc.
Median WTP estimate to avoid one MRAD from Tolley et al. (1986). Distribution is assumed to be
triangular with a minimum of $24 and a maximum of $94, with a most likely value of $59 (2006$). Range
is based on assumption that value should exceed WTP for a single mild symptom (the highest estimate for
5-20
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
HEALTH ENDPOINT
School Loss Days
CENTRAL ESTIMATE OF VALUE
PER STATISTICAL INCIDENCE
1990 INCOME
LEVEL
$89
2020 INCOME
LEVEL
$89
DERIVATION OF DISTRIBUTIONS OF ESTIMATES
a single symptom— for eye irritation— is $24) 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.
No distribution available. Point estimate is based on (1 ) the probability that, if a school child stays home
from school, a parent will have to stay home from work to care for the child, and (2) the value of the
parent's lost productivity. Calculated using U.S. Bureau of Census data. School loss days, similar to cost
of illness estimates for emergency room visits and hospital admissions, are not adjusted for changes in
longitudinal income.
a These values are presented using a seven percent discount rate for this draft report, however these results will be presented using a five percent discount rate in
the final report.
5-21
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
A survival curve approach provides a theoretically preferred method for valuing the
economic benefits of reduced risk of premature mortality associated with reducing air
pollution, but the approach does not align well with current estimates of individual
willingness to pay to avoid mortal risks. We have adopted the survival curve approach in
the population simulation model that we use to generate estimates of life years lost and
reduced life expectancy associated with air pollution, but implementing that approach
requires that we use a national measure of the change in air pollution exposure, and also
does not include a valuation component. As a result, the population simulation model
results are not used for the primary results.
The Project Team also considered whether other evidence might support an adjustment to
the VSL used in this study, particularly to account for the age of individuals affected. In
general, studies of WTP to reduce mortality risk do not provide information on how VSL
varies with life expectancy, but there are a few studies that attempt to assess the impact of
age on VSL.57 Some economic models in the theoretical literature suggest that VSL
follows an inverted U, rising through middle age and falling at older ages, though this
model is only partially supported by the relevant empirical evidence (Johansson 2002,
Hammitt 2007). For example, revealed preference studies of the wage-risk literature
support the inverted-U hypothesis (Aldy and Viscusi, 2007). These studies are limited,
however, in that they necessarily include only employed workers and thereby exclude the
elderly and those in poor health. Stated-preference studies, which can include a broader
population, yield mixed results. Some suggest little or no effect of age on VSL and others
suggest a modest decrease at older ages (Krupnick, 2007). Some studies, such as those
by DeShazo (with Cameron, 2004), Chestnut (et al., 2004), and Alberini (et al., 2004)
have found the effect of age on VSL to be statistically weak, suggesting a flatter
relationship of VSL and age with a decline in VSL at much older ages. Consistent with
Hammitt (2007), we conclude that there is insufficient evidence in the empirical VSL
literature at this time to support an adjustment to the base VSL for the age of the affected
population.
In sum, the economic valuation literature does not yet include good estimates of the value
of this particular risk reduction commodity. As a result, in this study we value avoided
premature mortality risk using the value of statistical life approach. As in the First
Prospective Analysis, we use a mortality risk valuation estimate which is based on an
analysis of 26 policy-relevant value-of-life studies (see Table 5-5). Five of the 26 studies
are contingent valuation (CV) studies, which directly solicit WTP information from
subjects; the remaining studies are wage-risk studies, which base WTP estimates on
estimates of the additional compensation demanded in the labor market for riskier jobs.
57
For a review of these studies, and this issue in particular see, for example, Hammitt (2007), Aldy and Viscusi (2007), and
Krupnick (2007).
5-22
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
We used the best estimate from each of the 26 studies to construct a distribution of
mortality risk valuation estimates for the section 812 study. A Weibull distribution, with
a mean of $7.4 million (in 2006$), provided the best fit to the 26 estimates.
An additional uncertainty that is pertinent for this study's results is the potential bias in
using estimates of VSL that correspond to small changes in risk for the relatively larger
changes in mortality risk estimated in this study. As the results section below indicates,
the large changes in PM25ihat represent the difference between the with-CAAA and
without-CAAA scenarios by 2020 lead to a change in annual mortality risk of
approximately 1 in one thousand for adults aged 25 and older, or 7 in ten thousand for all
ages, which corresponds to a roughly ten percent change from the national baseline
mortality risk of approximately 1 in one hundred.58 This risk change is large compared to
the mean mortality risk faced by subjects in the wage-risk studies that underlie our
estimate of VSL - the mean risk for individual studies in our group of 26 varies from 4 in
10,000 to 5 in 100,000, although clearly some individuals in those samples face higher
individual risks.59 Economic theory suggests that individuals' incremental willingness to
pay to reduce mortality risk declines with an increasing size of the risk increment, but the
rate at which it declines is uncertain.60 Estimates of differences in VSL across individuals
in wage-risk study samples are also not informative, because they reflect variability in
individuals' risk tolerance rather than differences in WTP across a population for varying
increments of risk reduction. Further, it is not clear whether, in this context, the external
risk imposed by air polluters on the exposed population implies that willingness-to-
accept-compensation (WTAC) to forgo air quality improvement may be the more
relevant measure. There is some theoretical work which suggests that, while valuation of
a large risk increment may lead WTP estimates to be overestimated, it may lead WTAC
estimates to be underestimated.61 Although the Project Team remains concerned that
there may be a potentially important disparity between the large increment of risk valued
in this study and relatively smaller increments of risk valued in the underlying VSL
literature, we conclude that the current literature does not provide a sufficient basis to
make a quantitative adjustment to our base VSL values to account for this factor.
When valuing premature mortality for PM, we assume a lag between reduced PM
exposure and the resulting reductions in incidences of premature mortality.62 This lag
Note that we are here reporting the total risk change that results from changes in 2020 exposures. As outlined below, this
risk is not immediate - instead we model this risk as occurring with latency over the course of the ensuing 20 years.
59 See W. Kip Viscusi, 1992, Fatal Tradeoffs, (Oxford University Press: New York), Table 4-1.
60 This issue is discussed to some extent in Thomas J. Kniesner, W. Kip Viscusi, and James P. Ziliak (2010), "Policy relevant
heterogeneity in the value of statistical life: New evidence from panel data quantile regressions," Journal of Risk and
Uncertainty 40:15-31.
61 See discussion papers provided in support of a recent EPA risk valuation workshop at
http://www.epa.gov/air/toxicair/2009workshop.html (accessed November 24, 2010) in particular the papers and
presentations by W. Kip Viscusi.
62 Note that we do not employ a cessation lag for ozone mortality due to our reliance on short-term studies to estimate these
benefits.
5-22
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-5.
does not affect the number of estimated incidences, but 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. The
primary estimate reflects a 20-year distributed lag structure, which was recommended by
the Council HES (2004). Under this scenario, 30 percent of the mortality reductions
occur in the first year, 50 percent occur equally in years two through five, and the
remaining 20 percent occur equally in years six through 20. Our valuation of avoided
premature mortality applies a five percent discount rate to the lagged estimates over the
periods 2000 to 2020, 2010 to 2030 and 2020 to 2040. We discount over the period
between the initial PM exposure change (2000, 2010, or 2020) and the timing of the
resulting change in incidence.
SUMMARY OF MORTALITY VALUATION ESTIMATES PER STATISTICAL INCIDENCE OF
PREMATURE MORTALITY (MILLIONS OF 2006$)
STUDY
Kneisner and Leeth (1991) (US)
Smith and Gilbert (1984)
Dillingham (1985)
Butler (1983)
Miller and Guria (1991)
Moore and Viscusi (1988a)
Viscusi, Magat, and Huber (1991b)
Gegaxetal. (1985)
Mann and Psacharopoulos (1982)
Kneisner and Leeth (1991) (Australia)
Gerking, de Haan, and Schulze (1988)
Cousineau, Lacroix, and Girard (1988)
Jones-Lee (1989)
Dillingham (1985)
Viscusi (1978, 1979)
R.S. Smith (1976)
V.K. Smith (1976)
Olson (1981)
Viscusi (1981)
R.S. Smith (1974)
Moore and Viscusi (1988a)
Kneisner and Leeth (1991) (Japan)
Herzog and Schlottman (1987)
Leigh and Folson (1984)
Leigh (1987)
Garen (1988)
Source: Viscusi, 1992 and EPA analysis.
TYPE OF ESTIMATE
Labor Market
Labor Market
Labor Market
Labor Market
Contingent Valuation
Labor Market
Contingent Valuation
Contingent Valuation
Labor Market
Labor Market
Contingent Valuation
Labor Market
Contingent Valuation
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
Labor Market
VALUATION
(MILLIONS 2006$)
$ 0.9
$ 1.1
$ 1.4
$ 1.7
$ 1.9
$ 3.9
$ 4.2
$ 5.1
$ 4.3
$ 5.1
$ 5.2
$ 5.6
$ 5.9
$ 6.0
$ 6.3
$ 7.1
$ 7.2
$ 8.0
$ 10.0
$ 11.1
$ 11.3
$ 11.7
$ 14.0
$ 15.0
$ 16.0
$ 20.8
5-23
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
HEALTH EFFECTS MODELING RESULTS
This section presents a summary of the differences in health effects resulting from
improvements in air quality between the with-CAAA and the without-CAAA scenarios.
Table 5-6 summarizes the CAAA-related avoided health effects in 2020 for each health
endpoint included in the analysis and the associated monetary benefits. The mean
estimate is presented as the primary central estimate, the 5th percentile observation is
presented as the primary low estimate and the 95th percentile is presented as the primary
high estimate.63 In general, because the differences in air quality between the with- and
without-CAAA scenarios are expected to increase from 1990 to 2020 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 2020. More detailed
results can be found in Health and Welfare Benefits Analyses to Support the Second
Section 812 Benefit-Cost Analysis of the Clean Air Act, February 2011.
AVOIDED PREMATURE MORTALITY ESTIMATES
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. As described in detail in this chapter, there are also significant
reductions in other short-term and chronic health effects and a substantial number of
health benefits that we could not quantify or monetize. Mean results for all three target
years are provided in Table 5-6, and the mean, primary low, and primary high estimates
for 2020 are presented in Table 5-7.
As shown in Table 5-7, our primary central estimate implies that PM and ozone
reductions due to the CAAA in 2020 will result in 230,000 avoided deaths, with a
primary low and primary high bound on this estimate of 45,000 and 490,000 avoided
deaths, respectively. These avoided deaths are valued at $1.8 trillion (2006$), with
primary low and primary high bounds on this estimate of $170 billion to $5.5 trillion. To
provide some context for these large values, we estimated the per capita risk change and
monetized benefits. The estimated 230,000 avoided deaths in 2020 are equivalent to a
total annual mortality risk reduction of 6.8 x 10"4 for the full estimated US population in
2020. With approximately 2.4 million estimated deaths in 2002, the avoided deaths in
2020 would increase total deaths by about 9.5 percent. The 230,000 avoided deaths are
about 16 percent of the total mortality from the top four causes of death in the US in
2002: heart disease (over 600,000 deaths); cancer (over 550,000 deaths); stroke (over
130,000 deaths); and chronic lower respiratory disease (just less than 130,000 deaths).
The monetized benefit per capita in 2020 is about $6,000, increasing from $2,700 in 2000
and $4,200 in 2010. Monetized benefits per household would be approximately $16,000
in 2020, increasing from $7,300 in 2000 and $11,000 in 2010.
63 The distribution of incidence results represent the uncertainty associated with the coefficient of the C-R function for each
health endpoint. The distribution around the monetized benefits estimate reflects both uncertainty in the incidence as
well as uncertainty associated with the valuation estimate.
5-24
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-6. MEAN CAAA-RELATED AVOIDED ANNUAL INCIDENCE OF HEALTH EFFECTS AND
ASSOCIATED MONETARY VALUATION IN 2000, 2010, AND 2020
ENDPOINT
POLLUTANT
INCIDENCE
2000
2010
2020
VALUATION (MILLIONS 2006$)
2000
2010
2020
Mortality
Mortality - adults 30
and older
Mortality - infant
Mortality - all ages
PM
PM
Ozone
110,000
160
1,400
160,000
230
4,300
230,000
280
7,100
$710,000
$1,300
$10,000
$1 ,200,000
$1 ,900
$33,000
$1,700,000
$2,500
$55,000
Morbidity
Chronic Bronchitis
Non-fatal Myocardial
Infarction
Hospital Admissions,
Respiratory
Hospital Admissions,
Cardiovascular
Emergency Room
Visits, Respiratory
Acute Bronchitis
Lower Respiratory
Symptoms
Upper Respiratory
Symptoms
Asthma Exacerbation
Minor Restricted
Activity Days
Work Loss Days
School Loss Days
Outdoor Worker
Productivity
PM
PM
PM, Ozone
PM
PM, Ozone
PM
PM
PM
PM
PM, Ozone
PM
Ozone
Ozone
34,000
79,000
20,000
26,000
58,000
96,000
1,200,000
980,000
1,200,000
49,000,000
8,000,000
1 ,200,000
N/A
54,000
130,000
41 ,000
45,000
86,000
130,000
1,700,000
1 ,400,000
1,700,000
84,000,000
13,000,000
3,200,000
N/A
75,000
200,000
66,000
69,000
120,000
180,000
2,300,000
2,000,000
2,400,000
110,000,000
17,000,000
5,400,000
N/A
$14,000
$8,100
$290
$760
$21
$42
$22
$30
$61
$2,900
$1,300
$110
$30
$24,000
$14,000
$640
$1,300
$32
$61
$30
$42
$90
$4,900
$2,000
$290
$100
$36,000
$21 ,000
$1,100
$2,000
$44
$94
$42
$60
$130
$6,700
$2,700
$480
$170
Note: All incidence and valuation results are rounded to two significant figures. All estimates are annual estimates for
individual target years of the analysis. Mortality valuation estimates reflect a delay in mortality incidence from the time
of the exposure change in the target year, reflecting application of a 20-year distributed cessation lag as described in the
text and a 5 percent discount rate.
It may also be worth noting that most of the changes in mortality risk we estimate occur
in locations where both the with-CAAA and without-CAAA concentrations are above the
lowest measured level (LML) in the underlying epidemiological studies. As noted above,
standard EPA practice is to estimate PM-related mortality without applying an assumed
concentration threshold, and the LML is itself not a threshold either. The LML approach
summarizes the distribution of avoided PM mortality impacts according to the baseline
PM2 5 levels experienced by the population receiving the PM2 5 mortality benefit. Unlike
an assumed threshold, the LML is a characterization of the fraction of benefits that are
more uncertain. In general, our confidence in the estimated PM mortality decreases as we
5-25
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-7.
consider air quality levels further below the LML in the two underlying PM-mortality
epidemiological studies, Pope et al. (2002) and Laden et al. (2006).
CAAA-RELATED AVOIDED ANNUAL INCIDENCE OF HEALTH EFFECTS AND ASSOCIATED
MONETARY VALUATION IN 2020
ENDPOINT
POLLUTANT
INCIDENCE
5™ %ILE
MEAN
95™ %ILE
VALUATION (MILLIONS 2006$)
5™ %ILE
MEAN
95™ %ILE
Mortality
Mortality1
PM, Ozone
45,000
230,000
490,000
$170,000
$1 ,800,000
$5,500,000
Morbidity
Chronic Bronchitis
Non-fatal
Myocardial
Infarction
Hospital
Admissions,
Respiratory
Hospital
Admissions,
Cardiovascular
Emergency Room
Visits, Respiratory
Acute Bronchitis
Lower Respiratory
Symptoms
Upper Respiratory
Symptoms
Asthma
Exacerbation
Minor Restricted
Activity Days
Work Loss Days
School Loss Days
Outdoor Worker
Productivity
PM
PM
PM, Ozone
PM
PM, Ozone
PM
PM
PM
PM
PM, Ozone
PM
Ozone
Ozone
12,000
80,000
24,000
52,000
64,000
-7,000
1,200,000
620,000
270,000
91 ,000,000
15,000,000
2,200,000
N/A
75,000
200,000
66,000
69,000
120,000
180,000
2,300,000
2,000,000
2,400,000
110,000,000
17,000,000
5,400,000
N/A
130,000
300,000
110,000
84,000
180,000
340,000
3,300,000
3,300,000
6,700,000
140,000,000
19,000,000
8,600,000
N/A
$3,100
$6,200
$320
$1 ,400
$22
-$4
$18
$17
$15
$3,800
$2,300
$190
$170
$36,000
$21,000
$1,100
$2,000
$44
$94
$42
$60
$130
$6,700
$2,700
$480
$170
$130,000
$48,000
$1,800
$2,600
$69
$220
$76
$130
$390
$10,000
$3,000
$770
$170
Notes:
1 Includes adult and infant mortality for PM and all ages for ozone.
All incidence and valuation results are rounded to two significant figures. Mortality valuation estimates reflect a delay in
mortality incidence from the time of the exposure change in the target year, reflecting application of a 20-year
distributed cessation lag as described in the text and a 5 percent discount rate.
Using the Pope et al. (2002) study, approximately 98 percent of the mortality impacts
occur among populations with exposure to annual mean PM2 5 levels at or above the LML
of 7.5 ug/m3. Using the Laden et al. (2006) study, approximately 91 percent of the
mortality impacts occur at or above the LML of 10 ug/m3. These analyses confirm that
the great majority of the mortality benefits occur at or above the cohort study LMLs.
5-26
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Avoided premature mortality is one of the more commonly cited results of benefits
analyses for air pollution control. However, as noted in the valuation section of this
chapter, a more accurate description of the benefit of clean air is a reduction in the risk of
mortality for the exposed population over many years, which results in the extension of
lives (sometimes referred to as "lives saved"). Other useful metrics of the benefit of
cleaner air are the number of life years that are gained through the reduction of mortal
risks, and the number of years of life expectancy gained on average throughout the
population. We estimated these metrics through the application of a population
simulation tool - effectively, we simulated the process of gradually reducing mortality
risk from air pollution across all individuals in the US 30 years old and older, starting in
1990 and continuing through 2020. In addition, we tracked the impact of these effects,
held constant at the 2020 levels, for an additional 30 years, through 2050. Running the
simulation beyond 2020 allows us to estimate the full effect of changes that begin in
2020, which because of the cessation lag are not fully realized until many years after the
end of the study period. Comparing the estimated population in each age cohort across
the two scenarios allows us to estimate gains in life-years (i.e., one additional person in a
cohort for one year yields a life year gained), and summing across cohorts and years
yields cumulative estimates. In addition, analysis of the changes in mortality risk among
cohorts older than a specific age yields estimates of life expectancy gains at specific
ages.64
The results of these calculations are presented in Table 5-8 below, and provide further
evidence of the substantial benefits of CAAA during and after the 1990-2020 period. The
first panel of the table provides estimates of life-years gained for 2020 and 2040 - these
are estimates of the life-years gained only in that year of the simulation, but reflect the
cumulative effect of mortality risk reductions in prior years. The next panel provides
estimates of cumulative life years gained overall all years since 1990, first for the 1990-
2020 period, and then for the 1990-2040 period, inclusive.
As expected, life-years gained are largest in the older cohorts, particularly cohorts 60
years and older, and they increase over time as the effect of mortality risk reduction in
successive years increases survival rates among all individuals age 30 and over. By 2020,
the cumulative effects indicate 22 million life-years are gained from the air pollution
mortality risk reduction.
The last panel provides the life expectancy results. As early as 2010, the CAAA
increased life expectancy at 30 years by 0.65 years, with somewhat smaller gains among
older cohorts. By 2040, the full effect of the CAAA on life expectancy is realized, with a
total gain in life expectancy of almost one year at age 30 across the entire US population.
M For a detailed description of the model, see the related report, Uncertainty Analyses to Support the Second Section 812
Benefit-Cost Analysis of the Clean Air Act, March 2010, and Industrial Economics, Inc. (2006).
5-27
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-8. LIFE YEARS GAINED AND LIFE EXPECTANCY GAIN ESTIMATES FROM THE
POPULATION SIMULATION MODEL
AGE COHORT
START AGE
30
40
50
60
70
80
90
END AGE
39
49
59
69
79
89
99
100+
Total
LIFE-YEARS GAINED IN
SPECIFIC YEARS
(ANNUAL)
2020
17,000
60,000
150,000
330,000
470,000
470,000
320,000
60,000
1 ,900,000
2040
18,000
71 ,000
180,000
380,000
840,000
1 ,200,000
800,000
200,000
3,800,000
CUMULATIVE LIFE YEARS
GAINED THROUGH TARGET
YEAR
2020
260,000
910,000
2,000,000
3,500,000
5,000,000
6,000,000
3,600,000
490,000
22,000,000
2040
620,000
2,300,000
5,400,000
11,000,000
20,000,000
23,000,000
14,000,000
3,100,000
80,000,000
LIFE EXPECTANCY GAINS
(YEARS)
2010
0.65
0.63
0.59
0.53
0.44
0.32
0.19
0
2020
0.87
0.84
0.79
0.71
0.59
0.43
0.25
0
2040
0.91
0.88
0.84
0.76
0.64
0.48
0.27
0
Note: Column entries to not add to totals due to rounding. Life expectancy results are incremental period
conditional life expectancy gains at the start age of the cohort.
NON-FATAL HEALTH IMPACTS
We report non-fatal health effects estimates in a similar manner to estimates of premature
mortality - as a range of estimates for each quantified health endpoint, with the range
dependent on the quantified uncertainties in the underlying C-R functions. The range of
results for 2020 is characterized in Table 5-6 with 5th percentile, mean, and 95th percentile
estimates which correspond to the primary low, central, and high estimates. All estimates
are expressed as new cases avoided in 2020, with the following exceptions. Hospital
admissions reflect admissions for a range of respiratory and cardiovascular diseases and
these results, along with emergency room visits for respiratory disease, do not necessarily
represent the avoidance of new cases of disease (i.e., air pollution may simply exacerbate
an existing condition, resulting in an emergency room visit or hospital admission).
Further, each admission is only counted once, regardless of the length of stay in the
hospital. Minor restricted activity days, school loss days, and work loss days are
expressed in terms of person-days. For instance, one "case" of a school loss day
represents one person out of school for one day.
AVOIDED HEALTH EFFECTS OF AIR TOXICS
The prior discussion focuses on the effects of the 1990 CAAA on particulate matter and
ozone health effects, but the Amendments also address the control of air toxics or
hazardous air pollutants (FiAPs). F£APs are pollutants regulated under Title III of the
CAAA that can cause adverse effects to human health and ecological resources. The
Amendments establish a list of F£APs to be regulated, require EPA to establish air toxic
5-28
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
emissions standards based on Maximum Achievable Control Technology (MACT)
standards, and include a provision that requires EPA to establish more stringent air toxics
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, such as controls on
volatile organic compounds (VOCs) necessary to achieve the NAAQS for ambient
tropospheric ozone.
Both the Retrospective analysis and the First Prospective analysis omitted a quantitative
estimation of the benefits of reduced concentrations of air toxics, citing gaps in the
toxicological database, difficulty in designing population-based epidemiological studies
with sufficient power to detect health effects, limited ambient and personal exposure
monitoring data, limited data to estimate exposures in some critical microenvironments,
and insufficient economic research to support valuation of the types of health impacts
often associated with exposure to individual air toxics. Based on a recommendation by
the Council, EPA developed a case study of the benefits of CAAA controls on benzene
emissions in the Houston area (USEPA, 2001).65 The purpose of the case study was to
demonstrate a methodology that could be used to generate human health benefits from
CAAA controls on a single HAP in an urban setting, while highlighting key limitations
and uncertainties in the process. In addition, EPA hoped to gain insight into the use of
the case study methodology for characterizing benefits nationwide. The case study was
not intended, however, to provide a comprehensive assessment of the benefits of benzene
reductions due to the CAAA.
The case study involved calculating the reduction in the annual number of cases of
leukemia due to reductions in benzene levels resulting from the 1990 CAAA through the
year 2020 in the Houston metropolitan area. Benzene was selected for the case study due
to the availability of human epidemiological studies linking its exposure with adverse
health effects. The case study focused on Houston because of the presence of significant
large benzene emitting sources, such as petroleum refineries, as well as sources more
typical of other urban areas, such as gasoline refueling stations.
We conducted the case study using the same five steps used in the main 812 criteria
pollutant analysis:
1. Scenario Development: We assessed benefits from the reduction in benzene
concentrations between a without-CAAA scenario, which essentially freezes
federal, state, and local air pollution controls at the levels of stringency and
effectiveness that existed in 1990, and awith-CAAA scenario, which assumes that
all federal, state, and local rules promulgated pursuant to, or in support of, the
1990 CAAA were implemented.
65 A detailed report of the case study methodology and results was completed by Industrial Economics, Inc (lEc, 2009). This
report can be downloaded from the following website: www.epa.gov/oar/sect812
5-29
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
2. Emissions Estimation: We estimated benzene emissions in the Houston area
under both the with-CAAA and without-CAAA scenarios by extrapolating data
based on expected growth in emissions-generating activities over time, adjusted
for the impact of future year control assumptions under each scenario.
3. Air Quality and Exposure Modeling: We then applied EPA's American
Meteorological Society/Regulatory Model (AERMOD) dispersion modeling
system (USEPA, 2004) to convert emissions estimates to ambient benzene
concentrations at the Census block group level. The AERMOD output was then
run through EPA's Hazardous Air Pollutant Exposure Model, Version 6
(HAPEM6; ICF International, 2007) to generate benzene exposure concentrations
for the study population at the Census tract level, which reflect average benzene
concentrations likely experienced by the study population as they carry out their
daily activities.
4. Health Effects Modeling: We next estimated avoided cases of leukemia using a
life-table based risk assessment model. The life-table model assessed age-
specific risks at the Census tract level, based on county-level background rates of
leukemia, age-specific benzene exposure data from HAPEM6 and an
epidemiological dose-response function derived from a study of occupational
benzene exposures (Crump, 1994).66 The model yielded annual age-specific
Census tract-level avoided cases of leukemia (fatal and non-fatal) for each target
year. We also estimated the number of cases expected to occur after the end of
the study period resulting from CAAA-related benzene changes within the study
period, due to lagging effects of these changes on leukemia risks.
5. Valuation: We then applied valuation methods from the current economic
literature to assign monetary value to the avoided leukemia cases. This included
valuing fatal cancers using the VSL estimate used in the primary 812 analysis
(i.e., the Weibull distribution based on 26 studies) with an adjustment for medical
costs associated with the period of cancer illness leading up to death (i.e., "pre-
mortality morbidity").67 We valued non-fatal cancers using two bounding
estimates, a WTP value for chronic bronchitis and one from a health risk tradeoff
study that provided a value for avoiding a case of non-fatal lymphoma.68
Table 5-9 presents our primary estimate for avoided fatal and non-fatal cases of leukemia
due to CAAA-related changes in ambient benzene levels in the Houston area. It includes
the number of expected annual cases avoided in each study year as well as the total
cumulative avoided cases throughout the study period and the total cumulative avoided
cases expected to occur after 2020, due to changes in benzene occurring within the study
56 This study is also the basis for the Inhalation Unit Risk (IUR) published on EPA's Integrated Risk Information System (IRIS)
(USEPA, 1998).
57 This estimate was based on a value presented in EPA's Cost of Illness Handbook (USEPA, 1999) for a "typical" cancer case.
58 The chronic bronchitis value is the same as that used in EPA's Regulatory Impact Analysis (RIA) for the PM National Ambient
Air Quality Standards (NAAQS) (USEPA, 2006). The non-fatal lymphoma value was derived by using the risk-risk ratio from
Magat et al. (1996) along with our primary VSL estimate.
5-30
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
period. It also shows the monetary value (the 1990 net present value (NPV), using a five
percent discount rate) of these avoided leukemia cases.
Our results indicate that by the year 2020, the change in benzene-related population risk
due to the 1990 CAAA programs would be equivalent to a total of four cases of leukemia
in the Houston area, with three of those occurring in Harris County, the most densely
populated county included in the analysis. We estimated two of the four cases to be fatal
and two to be non-fatal. Our primary central estimate of total benefits due to CAAA-
related reductions in benzene is $8.9 to 13 million (in 2006$), $8.5 million of which is
due to fatal cases of leukemia, and $0.4 to 4.1 million of which is due to non-fatal cases.
In addition to the leukemia analysis, we evaluated the numbers of individuals likely to be
exposed to benzene at levels exceeding EPA's chronic reference concentration (RfC) for
benzene, which is based on changes in white blood cell counts, under the with-CAAA and
without-CAAA scenarios. We found no individuals exposed to benzene at concentrations
exceeding the RfC in either the with-CAAA or without-CAAA scenario. We also
conducted illustrative analyses of exposure and risk reductions to highly exposed
subpopulations in the study area, and found potentially significant individual risk
reductions due to the CAAA for individuals in these groups. For instance, a back-of-the
envelope calculation of residents living in homes with attached garages, who are expected
to have higher benzene exposures, suggests that adding attached garage-related benefits
to our primary estimate could result in an approximate doubling of our primary estimate.
The effect of the CAAA on lifetime risks of benzene-induced leukemia for Houston
residents at the Census tract level is explored in Figure 5-2. The map on the left displays
the distribution of leukemia risks based on benzene exposures levels expected in 2020
under the without-CAAA scenario. The highest risk levels (i.e., greater than one-in-one
hundred thousand) occur in Harris County in the downtown Houston area (within the
rings of the interstate), in the Texas City area of Galveston County where a number of
refineries and chemical facilities are located and in southeastern Brazoria County, which
also features major chemical manufacturing and petroleum refining facilities. The map
on the right shows the distribution in the magnitude of CAAA-related risk reductions
throughout the Houston area. The highest risk reductions (i.e., greater than a factor of
three) coincide with the areas identified as those with the highest risks in the first map.
For instance, the CAAA is expected to reduce risks significantly in the highly populated
downtown Houston area, where residents are expected to have risks on the order of one-
in-one hundred thousand or greater.
5-31
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-9. TOTAL ANNUAL BENEFITS FOR EACH STUDY YEAR FROM CAAA-RELATED CHANGES IN BENZENE EXPOSURE IN THE
HOUSTON AREA
ANNUAL AVOIDED CASES OF LEUKEMIA
AVOIDED
FATAL CASES
AVOIDED NON-
FATAL CASES
TOTAL AVOIDED
CASES
TOTAL MONETARY BENEFITS, 1990 TO 2010
(1990 NPV, MILLIONS OF 2006$, 5% DISCOUNT RATE)
BENEFITS FROM
FATAL CASES OF
LEUKEMIA
BENEFITS FROM
NON-FATAL
CASES OF
LEUKEMIA
TOTAL BENEFITS
Results by Study Year
2000
2010
2020
0.03
0.09
0.2
0.02
0.07
0.1
0.05
0.2
0.3
$0.12
$0.27
$0.31
$0.01 - 0.06
$0.01 -0.13
$0.01 -0.15
$0.13-0.18
$0.28-0.40
$0.32-0.46
Cumulative Results
Cumulative Cases Occurring
Within the Study Period
Additional Cumulative Cases
Occurring After 2020*
Total Cumulative Cases
2
1
3
2
1
3
4
2
6
$6.7
$1.8
$8.5
$0.32-3.3
$0.08-0.8
$0.40-4.1
$7.0- 10
$1.9-2.6
$8.9- 13
* Note: These avoided cases are due to changes in benzene exposure that took place within the study period. However, the cases occurred after
2020 due to lagging effects of these changes on leukemia risks, as described in the text.
5-32
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
In summary, this case study demonstrates that the 1990 CAAA controls on benzene
emissions are expected to result in reductions in the incidence of leukemia in the greater
Houston area over the period 1990 to 2020. The case study does have some limitations,
including possible underestimation of benzene emissions from large point sources (e.g.,
refineries), possible exclusion of unquantifiable adverse health effects of benzene (e.g.,
Hodgkin's and non-Hodgkin's Lymphoma), and exclusion of new programs established
after the case study (e.g., Mobile Source Air Toxics Rule). However, it successfully
demonstrates a methodology that can serve as a useful tool in EPA's evolving HAP
benefits assessment strategy. It can provide a comprehensive assessment of the impact of
benzene controls from multiple CAAA Titles on cancer incidence in an urban population,
using a combination of national and local data to conduct urban-scale modeling of air
quality and health impacts. Further, the life-table model allows for more careful
assessment of risk changes over time at the Census tract level, incorporating local, age-
specific baseline incidence data with age-specific exposure data and information on the
lag between exposure changes and risk reductions.
FIGURE 5-2. EFFECT OF THE CAAA ON LIFETIME RISKS OF BENZENE-RELATED LEUKEMIA IN THE
HOUSTON AREA
Legend
2020 Without CAAA Risk Uv«li
_ j Less iFian i in a
] B
] Bthvttn 1 in ww hundtd Ihouwnd wd 1 w» tw
2020 Risk Comparison
rnV from Wrthout to Wrth CAAA
[_ 1 BvtwMn i hctor cf MO md • Ida c* thrw
^| BetwBW • Factor cfftra Bod • *• dcr c< four
m Efetwem « fcdar & (Our ard a t>:'o of left
^H &*»1»r in*n | fatv 9(1vn
Determining where this approach might fit within EPA's HAP benefits assessment
strategy will require additional analysis and evaluation to determine the added value of
the detailed, urban-scale approach, as well as the potential pool of HAPs suitable for
assessment via the damage-function approach for cancer and/or non-cancer effects.
5-33
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
COMPARISON OF HEALTH EFFECTS MODELING WITH FIRST PROSPECTIVE ANALYSIS
DIFFERENCES IN METHODOLOGY
In comparison with the First Prospective 812 Analysis, the Second Prospective includes a
number of refinements and improvements in health benefits estimation methods.
• Targeted Criteria Pollutant Analysis: The Second Prospective excludes benefits
of CAAA-related reductions in carbon monoxide, nitrogen oxides, and sulfur
dioxide, which were included in the First Prospective, in an effort to streamline
the quantitative analysis to focus on the two criteria pollutants that yield the
greatest benefits - PM2 5 and ozone.
• New Cessation Lag Structure for PMMortality: The Second Prospective relies
on the use of a 20-year distributed lag structure assumption for the cessation lag
between changes in PM exposure and resulting changes in premature mortality.
This estimate represents a shift from the First Prospective, which applied a 5-year
distributed lag based on smoking cessation literature. The 20-year distributed lag
is based on recommendations from the Council HES, is derived from air
pollution literature and attempts to more closely reflect the disease processes that
occur from PM exposure.69
• New C-R Function for PMMortality: The First Prospective relied upon a C-R
function derived from the most recently published ACS cohort study at the time
(Pope et al., 1995). Since this time, additional follow-up has occurred for both
the ACS and Six Cities cohort studies. In addition, new evidence has emerged on
the ACS study results that suggest that this estimate is potentially underestimated.
Our new primary C-R function mean is based on the follow-up literature,
specifically the Pope et al. (2002) update of the ACS cohort and the Laden et al.
(2006) update of the Six Cities cohort. Our new C-R function also reflects the
results of an expert elicitation study, which allowed experts to incorporate
multiple sources of uncertainty in the C-R function and to adjust the C-R function
estimates to account for known biases.
• Ozone Mortality Benefits Estimates: The Second Prospective includes ozone-
related premature mortality. This additional endpoint, which was not included in
the First Prospective, was added because of advances that have occurred in the
epidemiological literature that provide consistent evidence for this health
endpoint.70
69 Science Advisory Board (2004). Advisory on Plans for Health Effects Analysis in the Analytical Plan for EPA's Second
Prospective Analysis—Benefits and Costs of the Clean Air Act, 1990-2020: Advisory by the Health Effects Subcommittee of
the Advisory Council on Clean Air Compliance Analysis. EPA-SAB-COUNCIL-ADV-04-002.
70 As noted earlier, a key recommendation of NRC (2008) was that ozone mortality estimates from available epidemiological
studies represent a separate and additive effect to those from PM/mortality epidemiological studies.
5-34
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
• New Health Benefits Modeling Program: The Second Prospective relies on
EPA's BenMAP health benefits modeling program. Key advantages of the
updated model are ease of use, allowing us to more readily perform multiple
sensitivity tests; updated population and baseline incidence estimates; new C-R
function options; and the ability to perform integrated exposure analysis using the
eVNA method described earlier.
• Air Toxics Case Study: The Second Prospective includes the results of a case
study demonstrating a methodology for assessing health benefits from a single
hazardous air pollutant.
DIFFERENCES IN HEALTH EFFECTS MODELING RESULTS
The health effects estimates for the Second Prospective are much larger than the
estimates EPA developed for the First Prospective. The 2020 estimates are new to the
Second Prospective, but the comparable mean estimate of health benefits in 2000 and
2010 for the First Prospective were $71 billion in 2000 and $110 billion in 2010, in
1990S71 - if updated to 2006$, these estimates would be $110 billion in 2000 and $170
billion in 2010. The Second Prospective results are larger by roughly a factor of 10.
There are four key reasons we have identified for the increase in benefits:
1. Scenario differences: The with-CAAA scenario, especially for the 2010 target year,
includes new rules with substantial additional pollutant reductions that were not
included in the comparable First Prospective scenario, such as the Clean Air
Interstate Rule (CAIR).
2. Improved air quality models: The First Prospective relied on the Regional Acid
Deposition Model/Regional Particulate Model (RADM/RPM) for PM and deposition
estimates in the eastern U.S., the Regulatory Modeling System for Aerosols and Acid
Deposition (REMSAD) for PM estimates in the western U.S., and the Urban Airshed
Model (versions V and IV) at various regional and urban scales to generate ozone
estimates. The Second Prospective relies on the integrated CMAQ modeling tool,
which reflects substantial improvements in air quality modeling, provides more
comprehensive spatial coverage, and achieves improved model performance.
3. Better, more comprehensive exposure estimates: The First Prospective relied on
first generation exposure extrapolation tools to generate monitor-adjusted exposure
estimates away from monitors. Since then, the monitor network, availability of
speciated data, and the performance of speciated exposure estimation tools have
improved substantially.
4. Updated dose-response estimates. Since 1999, some concentration response
functions have been updated, most notably the PM-premature mortality C/R function,
whose central estimate of the mortality impact of fine PM has nearly doubled. In
71 See The Benefits and Costs of the Clean Air Act 1990 to 2010, USEPA Office of Air and Radiation and Office of Policy, EPA-
410-R-99-001, November 1999.
5-35
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
addition, health effects research has addressed endpoints that were not covered in the
First Prospective, including premature mortality associated with ozone exposure.
Although the Agency has not yet conducted a rigorous quantitative analysis to assess the
impact of these methodology and data improvements, and the differences in study design
between the first and Second Prospective made such an analysis difficult to perform, the
impact of most of these factors is to increase the estimates of benefits, in some cases very
substantially.
UNCERTAINTY IN HEALTH BENEFITS ESTIMATES
A number of important assumptions and uncertainties in the health benefits analysis may
influence the estimate of monetary benefits presented in this study. In this section of the
chapter, we first discuss several quantitative sensitivity analyses undertaken to
characterize the impact of key assumptions on the ultimate health benefits estimates. We
then conclude with a qualitative discussion of the impact of both quantified and
unqualified sources of uncertainty.
QUANTITATIVE SENSITIVITY TESTS
We performed three quantitative sensitivity tests to estimate the impact of alternate
assumptions on our overall benefits estimates due to avoided premature mortality, the
largest contributor to our overall health benefits estimates. The three focal areas for
sensitivity analysis were: (1) the C-R function estimate; (2) the PM/mortality cessation
lag structure; and (3) the mortality valuation estimate (including both the VSL and the
discount rate). These are influential assumptions in our analysis and those for which
plausible alternative quantitative estimates are available. Table 5-10 below provides the
results of these sensitivity analyses.
Con cent ration-Response Function
Our monetized estimate of the benefits of reducing premature mortality from CAAA-
related pollution reductions is based on a single primary estimate C-R function for each
of the criteria pollutants included in our analysis, PM2 5 and ozone. This selection is
associated with uncertainty related to potential across-study variation. That is, different
published studies of the same pollutant/health effect relationship often do not report
identical findings; in some instances, the differences are substantial. These differences
can arise from differences in factors such as study design, random sampling for subject
populations, or modeling choices, such as inclusion of potential confounders.
In order to estimate the effect of across-study variation on our CAAA-related mortality
benefits from reductions in PM2 5 and ozone, we performed a sensitivity analysis on the
C-R functions selected. For PM2 5, our primary estimate is based on a Weibull
distribution of C-R coefficients with a mean of 1.06 percent decrease in annual all-cause
mortality per 1 |o,g/m3 and an interquartile range bracketed by the Pope et al. (2002) ACS
estimate (0.55 percent) on the low end and the Six Cities Laden et al. (2006) extended
follow-up estimate (1.5 percent) at the high end. We conducted a sensitivity analysis by
5-36
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
first substituting the primary C-R distribution with alternative C-R functions, one based
on the Pope et al. (2002) ACS study, one based on the Laden et al. (2006) Six Cities
cohort study as well as the C-R distributions provided by each of the 12 experts included
in the PM/mortality expert elicitation study.
For ozone, our primary estimate consists of a pooled estimate of six studies, three multi-
city studies (Schwartz, 2005; Bell et al., 2004; Huang et al., 2005) and three meta-
analyses (Ito et al., 2005; Levy et al., 2005; Bell et al., 2005). We conducted a sensitivity
analysis by substitute this primary C-R function with the C-R functions reported in each
of these six individual studies, and separately for the Jerrett et al. (2009) cohort study.
As shown in Table 5-10, substituting alternate PM C-R functions results in total mortality
benefits estimates that range from between 81 percent lower up to 78 percent higher than
the primary estimate. Substituting alternative ozone C-R function does not affect the total
mortality benefits estimate, since ozone does not contribute significantly to this estimate.
However, the C-R function selection does affect the ozone mortality estimates, ranging
from 63 percent lower up to 66 percent higher than the primary estimate for ozone
mortality incidence. As expected, the Jerrett et al. study yields estimates higher than the
primary pooled estimate. Cohort studies measure the effects of cumulative exposure and
so should reasonably yield higher estimates than the comparably parameterized time-
series study - but within the range of underlying six studies, albeit at the high end of that
range.
PM/Mortality Cessation Lag
The timing of the cessation lag between PM exposure and mortality remains uncertain.
Our primary monetized estimate of PM/mortality benefits assumes a 20-year distributed
lag (30 percent of the mortality reductions occur in the first year, 50 percent occur equally
in years two through five, and the remaining 20 percent occur equally in years six through
20). We tested the sensitivity of this assumption by calculating monetized mortality
benefits based on alternative cessation lag structures. We selected two alternative lag
structures - a 5-year distributed lag (which was employed in the First Prospective) and a
smooth function (which assumes an exponential decay model and is based on an analysis
by Roosli et al., 2005; see Chapter 6 of Uncertainty Analyses to Support the Second
Section 812 Benefit-Cost Analysis of the Clean Air Act for further details). We also
calculated benefits assuming no cessation lag. Application of alternative cessation lag
structures had a smaller impact on the benefits estimates than the C-R function, resulting
in benefits estimates that range from 22 percent lower up to 16 percent higher than the
primary estimate.
Mortality Valuation
We apply a VSL value to reductions in premature mortality based on a Weibull
distribution of 26 study estimates. The literature on VSL is extensive, and studies have
measured VSL using different methodological approaches (e.g., revealed versus stated
preference) on a variety of study populations (e.g., workers versus a general population
sample) in a variety of different risk contexts (e.g., fatal workplace accidents versus
5-37
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
mortality risk from disease). In addition, several meta-analyses of the literature have
been conducted in an attempt to synthesize the literature. As a result, there are many
options for alternative VSL estimates. We selected several alternative VSL estimates
derived from the literature for sensitivity testing, including two estimates from a meta-
analysis by Viscusi and Aldy (2003), an estimate used in past EPA regulatory analyses in
the form of a normal distribution, and an estimate from a wage-risk study by Viscusi
(2004). VSL did not affect the benefits results to the same degree as the C-R function,
with alternative monetized benefits ranging from 21 percent lower to approximately
equivalent to our primary estimate.
TABLE 5-10. RESULTS OF QUANTITATIVE SENSITIVITY TESTS
FACTOR
PM C-R Function
Ozone C-R Function
PM/Mortality
Cessation Lag
VSL
Discount Rate
STRATEGY FOR SENSITIVITY ANALYSIS
Alternative C-R functions - two from
empirical literature (Pope et al., 2002
and Laden et al., 2006) and 12
subjective estimates from the expert
elicitation study
Alternative C-R functions - three from
multi-city studies, three meta-
analyses, and the Jerrett et al. (2009)
cohort long-term exposure study
Alternative lag structures - one step
function and one smooth function
(based on an exponential decay
function)
Alternative VSL estimates
Alternative discount rates
RANGE OF PERCENT CHANGES
FROM MEAN PRIMARY MORTALITY
BENEFITS ESTIMATE1
-81% to 78%,
Based on most extreme
estimates from PM expert
elicitation study. Rest of
alternatives range from
-41% to 40%.
0% for total mortality benefits.
-63% to 66%
For ozone- related mortality.
-22% to 16%
-21%toO%
-6% to 6%
1 All values in the table represent the percent change from the mean primary estimate. Percent change
estimates to not vary by target year.
Our primary monetized benefits estimate of avoided premature mortality also assumes a
discount rate of five percent. We tested the sensitivity of our primary results by
substituting alternative discount rates of three and seven percent.72 This assumption has a
small effect on the benefits estimates; applying a discount rate of seven percent results in
benefits that are 6 percent lower than the default and applying a three percent discount
rate results in a benefits estimate 6 percent higher than the default.
72 Alternative discount rates of three and seven percent are recommended in U.S. Environmental Protection Agency (2000).
Guidelines for Preparing Economic Analyses, EPA 240-R-00-003, September.
5-38
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
QUALITATIVE ANALYSIS OF KEY FACTORS CONTRIBUTING TO UNCERTAINTY
In addition to the uncertainties outlined above, we identified several other areas of
uncertainty related to our health benefits analysis that we did not address quantitatively.
This includes sources of uncertainty in our estimation of avoided mortality, not related to
across-study variation; application of C-R functions for national benefits estimation;
projection of population and baseline incidence rates; and health valuation.
Table 5-11 provides a summary of the key uncertainties related to the Second Prospective
health effects modeling analysis. The first column provides a brief description of each
key assumption made in the analysis. The second column indicates the direction of the
potential bias with respect to the overall net benefits estimate. The third indicates the
magnitude of the impact of the potential bias on the net benefits. 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. "73 This assessment is intended to provide readers with a sense for the
quantitative impact on the net benefits estimate if an alternate assumption to that selected
by the Project Team were to be implemented. Finally, the fourth column provides our
level of confidence in the selected assumption, based on our assessment of the available
body of evidence. That is, based on the given available evidence, how certain we are that
the selected assumption is the most plausible of the alternatives. The Project Team uses
the following four qualitative categories to express the degree of confidence in the chosen
assumption:
• "High" - the current evidence is plentiful and strongly supports the selected
assumption;
• "Medium" - some evidence exists to support the assumption, but data gaps are
present; and
• "Low" - there are limited data to support the selected assumption.
• The Project Team uses "N/A" to indicate that the data was so limited that it was
excluded from the analysis entirely.
73 If the quantitative magnitude of the assumption's effect on the net benefits cannot be assessed, the Project Team
indicates that this is "Unknown."
5-39
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 5-11. KEY UNCERTAINTIES ASSOCIATED WITH HUMAN HEALTH EFFECTS MODELING
POTENTIAL SOURCE OF
ERROR
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
MAGNITUDE OF
IMPACT ON NET
BENFITS ESTIMATE
DEGREE OF
CONFIDENCE
UNCERTAINTIES RELATED TO PREMATURE MORTALITY BENEFITS ESTIMATES
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.
Analysis assumes a
causal relationship
between ozone
exposure and
premature mortality
based on strong
epidemiological and
experimental evidence
of an ozone/mortality
association.
Overestimate
Overestimate
Potentially major.
PM/mortality
effects are the
largest contributor
to the net benefits
estimate. If the
PM/mortality
relationship is not
causal, it would
lead to a significant
overestimation of
net benefits.
Probably minor.
Ozone mortality
effects are a large
contributor to the
net benefits
estimate, but total
monetized ozone
mortality benefits
remain less than
five percent of total
net benefits. If the
ozone mortality
relationship is not
causal, it would
lead to an
overestimation of
net benefits.
High.
The assumption of
causality is suggested
by the epidemiologic
and lexicological
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.
Medium.
Several
epidemiological
studies provide strong
evidence for
associations between
ozone and mortality.
This data is supported
by human and animal
experimental studies
that provide
suggestive evidence
for plausible
mechanisms. Overall,
the evidence is highly
suggestive, but
additional research is
needed to more fully
establish underlying
mechanisms.
5-40
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
It is possible that the
PM/mortality
relationship is modified
by socioeconomic
status (SES).
Exposure
misclassification due to
reliance on ambient
monitoring data to
estimate PM2.5
exposures rather than
measuring personal
exposures.
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Unable to determine
based on current
information.
Consideration of both
the Pope and Laden
studies avoids the
possible
underestimation
effect from the ACS
cohort, owing to the
demographics of that
study population, and
the possible
overestimation bias
associated with the
more limited
geographic scope of
the Six Cities cohort.
Underestimate.
Concentrations
measured at central
site monitors may not
accurately reflect
exposure experienced
by the population due
to variation in
ambient
concentrations over
space within a
geographic area,
incomplete
penetration of
ambient pollution into
homes and
workplaces, patterns
of population activity
and indoor sources
that can contribute
significantly to
individual PM2.5
exposures. Reducing
exposure error can
result in stronger
associations between
pollutants and health
effects than generally
observed in studies
having less exposure
detail.
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Potentially major.
Sensitivity analyses
reported in this
chapter indicate
the high sensitivity
of benefits results
to the choice of the
PM/mortality C/R
function.
Potentially major.
Recent analyses
reported in Krewski
et al. (2009)
demonstrate the
relatively
significant effect
that this source of
uncertainty can
have on effect
estimates.
DEGREE OF
CONFIDENCE
Medium.
Studies have found
effect modification of
the PM/mortality
effect by SES, as
assessed through
education attainment
(Krewski et al., 2000).
However, this effect is
likely to affect only
the Pope et al.
estimate. Our
inclusion of both the
Pope et al. and Laden
et al. (which does
includes a more
diverse population)
helps account for the
possible significance
of this uncertainty.
High.
The results from
Krewski et al. (2009)
and Jerrett et al.
(2005) suggest that
exposure error may
underestimate effect
estimates (PM ISA).
5-41
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
Exclusion of C-R
functions from short-
term exposure studies
in PM mortality
calculations.
Assumption that PM-
related mortality
occurs over a period of
20 years following the
critical PM exposure.
Analysis assumes that
30% of mortality
reductions in the first
year, 50% over years 2
to 5, and 20% over the
years 6 to 20 after the
reduction in PM2.5
Assumption of a linear,
no-threshold model for
PM and ozone mortality
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Underestimate
Unable to determine
based on current
information
Overestimate
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Potentially major.
PM/mortality is the
top contributor to
the net benefits
estimate. If short-
term functions
contribute
substantially to the
overall PM-related
mortality estimate,
then the net
benefits could be
underestimated.
Potentially major.
PM/mortality is the
largest contributor
to monetary
benefits. Our
quantitative
sensitivity analysis
indicated that
alternative
plausible cessation
lag structures could
alter the benefits
estimate between
23% lower to 1 6%
higher than the
primary estimate.
Probably minor.
Although
consideration for
alternative model
forms (Krewski et
al., 2009) does
suggest that
different models
can impact risk
estimates to a
certain extent,
generally this
appears to be a
moderate source of
overall uncertainty.
DEGREE OF
CONFIDENCE
Medium.
Long-term PM
exposure studies likely
capture a large part of
the impact of short-
term peak exposure on
mortality; however,
the extent of overlap
between the two
study types is unclear.
Medium.
Recent
epidemiological
studies (e.g.,
Schwartz, 2008) have
shown that the
majority of the risk
occurs within 2 years
of reduced exposure.
However, our default
lag assumes 43% of
mortality reductions
would occur within
the first 2 years. The
evidence directly
informing the
cessation lag structure
is somewhat limited,
but the current lag is
supported by the
Council HES.
High.
The current scientific
literature does not
support a population-
based threshold,
which consistently
shows effects down to
the lowest
measureable levels. If
a threshold does exist,
it is likely below the
range of
concentrations of
regulatory interest.
5-42
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
Mortality health impact
did not include
pollutants other than
PM or ozone.
Pooling with equal
weights of ozone
mortality incidence
estimates to present a
primary estimate.
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Unable to determine
based on current
information
Unable to determine
based on current
information
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Probably minor.
If other criteria
pollutants
correlated with PM
contribute to
mortality, that
effect may be
captured in the PM
estimate. This
uncertainty does
make it difficult to
disaggregate
avoided mortality
benefits by
pollutant.
Probably minor.
Pooling with equal
weights provides a
central estimate of
ozone mortality
benefits, but it is
not clear that the
six ozone mortality
incidence studies
should be combined
in this manner.
Relying on a
particular single
study or another
combination of
studies may result
in significantly
different estimated
benefits from ozone
reductions.
However, ozone-
related avoided
mortality benefits
are a minor
contributor to total
monetized benefits.
DEGREE OF
CONFIDENCE
High.
PM and ozone are the
two pollutants most
strongly linked to
mortality in the
epidemiological
literature. It is likely
that we've captured
the majority of
mortality benefits due
to criteria pollutants
in our analysis.
Medium.
All six studies are
associated with
different strengths
and limitation. No
single study has
emerged as solely
suitable to support a
primary estimate.
Therefore, a pooled
estimate provides a
central estimate of
the available
literature.
5-43
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
DEGREE OF
CONFIDENCE
No cessation lag was
used for ozone
mortality.
Overestimate
Probably minor.
If there is a time
lag between
changes in ozone
exposure and the
total realization of
changes in health
effects then
benefits occurring
in the future should
be discounted. The
use of no lag
assumes that all
mortality benefits
are realized in the
year of the
exposure change
and therefore no
discounting occurs.
This may lead to an
overestimate of
benefits.
High.
Due to the use of
short-term studies of
ozone mortality, use
of a no lag structure is
appropriate and
supported by the
Council HES.
UNCERTAINTIES RELATED TO APPLICATION OF C-R FUNCTIONS
Application of C-R
relationships only to
those subpopulations
matching the original
study population.
Underestimate
Probably minor.
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 30+)
and thus may
underestimate the
whole population
benefits of
reductions in
pollutant
exposures.
However, the
background
incidence rates for
these age groups
are likely low and
therefore would not
contribute many
additional cases.
High.
The baseline mortality
and morbidity rates
for PM-related health
effects are
significantly lower in
those under the age of
30 (other than
neonates).
5-44
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
Application of
regionally derived C-R
estimates to entire U.S.
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Unable to determine
based on current
information
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Probably minor.
This is likely to
affect morbidity
estimates rather
than mortality, as
mortality estimates
are based on
studies that include
multiple cities.
Since morbidity is
not as large of a
contributor to
overall benefits,
this is not likely to
have a large impact
on net benefits.
DEGREE OF
CONFIDENCE
Medium.
The differences in the
expected changes in
health effects
calculated using
different underlying
studies can be large.
If differences reflect
real regional
variation, applying
individual C-R
functions throughout
the U.S. could result
in considerable
uncertainty in health
effect estimates.
UNCERTAINTIES RELATED TO HEALTH VALUATION
Use of a Value-of-a-
Statistical-Life (VSL)
estimate based on a
Weibull distribution of
26 studies
Use of cost of illness
(COI) estimates to
value some morbidity
end points
Unable to determine
based on current
information
Underestimate
Potentially major.
Mortality valuation
generally dominates
monetized benefits.
Probably minor.
Mortality valuation
generally dominates
monetized benefits;
therefore specific
estimates used to
generate morbidity
benefits likely
would not have a
large impact on net
benefits.
Medium.
The VSL used in this
analysis is based on 26
labor market and
stated preference
studies published
between 1 974 and
1991. Although there
are many more recent
studies, including
meta-analyses,
sensitivity analyses
reported above
suggest that these
alternative sources
generate results that
are close to the
estimates used in the
analysis.
Low.
Morbidity benefits
such as hospital
admissions and heart
attacks are calculated
using COI estimates,
which some studies
have shown are
generally half as much
as WTP to avoid the
illness. However, WTP
estimate are currently
not available for all
health endpoints.
5-45
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
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/ozone
and the risks evaluated
in the available
economic studies.
Inability to value some
quantifiable morbidity
endpoints, such as
impaired lung function.
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Unable to determine
based on currently
available information
Underestimate
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Potentially major.
The mortality
valuation step is
clearly a critical
element in the net
benefits estimate,
so any uncertainties
can have a large
effect.
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.
However, the lack
of a clear
symptomatic
presentation of the
effect, however,
could limit
individual WTP to
avoid lung function
decrements.
DEGREE OF
CONFIDENCE
Medium.
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.
However, our VSL
estimate is based on a
distribution of the
results of 26 individual
studies, which cover a
range of
characteristics.
Low.
There currently is no
evidence to determine
the monetary value of
the benefits of
avoided lung function
reductions.
5-46
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
DEGREE OF
CONFIDENCE
UNCERTAINTIES IN FORECASTED DATA SUPPORTING HEALTH EFFECTS ESTIMATES
Uncertainty in
projecting baseline
incidence rates
Income growth
adjustments
Both
Both
Probably minor.
The magnitude
varies with the
health endpoint.
Mortality baseline
incidence is at the
county level and
projected for 5-year
increments.
Morbidity baseline
incidence has
varying spatial
resolution for year
2000 only.
Potentially major.
Income growth
increases
willingness-to-pay
valuation
estimates, including
mortality, over
time.
Medium.
The county- level
baseline incidence and
population estimates
were obtained from
databases where the
relative degree of
uncertainty is low.
The baseline data for
other endpoints are
not location specific
(e.g., those taken
from studies) and
therefore may not
accurately represent
the actual location-
specific rates.
Medium
It is difficult to
forecast future income
growth, owing to
unpredictability of
future business and
employment cycles.
These can have a
substantial effect on
short term growth rate
projections, although
over longer periods
economic growth rates
have tended to
converge. The use of
data from AEO 2005,
however, omits the
effect of the most
recent economic
downturn.
5-47
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
Population projections
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Both
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Probably minor.
The demographics
of population
forecasting are
relatively well-
established,
however migration
estimates are quite
uncertain,
particularly for
specific locations.
Overall, we believe
that population
projections are not
likely to vary more
than 5 percent at
the national level.
DEGREE OF
CONFIDENCE
Medium.
Population projections
cannot adequately
account for future
population migration
due to catastrophic
events. Projected
population and
demographics may not
well represent future-
year population and
demographics.
OTHER UNCERTAINTIES
Variation in effect
estimates reflecting
differences in PM2.5
composition
Very limited
quantification of health
effects associated with
exposure to air toxics.
Unable to determine
based on current
information
Underestimate
Unable to
determine based on
current information
Probably minor.
Studies have found
air toxics cancer
risks to be orders of
magnitude lower
than those of
criteria pollutants.
Medium.
Epidemiology studies
examining regional
differences in PM2.5-
related health effects
have found
differences in the
magnitude of those
effects. While these
may be the result of
factors other than
composition (e.g.,
different degrees of
exposure
misclassification),
composition remains
one potential
explanatory factor.
N/A
Current data and
methods are
insufficient to develop
(and value) national
quantitative estimates
of the health effects
of these pollutants.
5-48
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF
ERROR
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.
DIRECTION OF
POTENTIAL BIAS FOR
NET BENEFITS
ESTIMATE
Underestimate
MAGNITUDE OF
IMPACT ON NET
BEN FITS ESTIMATE
Probably minor.
The health and
economic benefits
of reducing lead
exposure can be
substantial (e.g.,
see section 81 2
Retrospective Study
Report to
Congress).
However, most
additional fugitive
dust controls
implemented under
the with-CAM
scenario (e.g.,
unpaved road dust
suppression,
agricultural tilling
controls, etc.) tend
to be applied in
relatively low
population areas.
DEGREE OF
CONFIDENCE
N/A
5-49
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 6 - ECOLOGICAL AND OTHER WELFARE BENEFITS
OVERVIEW OF APPROACH
Air pollution has important impacts not
only on human health, but on a wide
range of ecological and environmental
resources. Clean Air Act provisions
are designed to be protective of human
health and the environment, but as a
practical matter, because human health
impacts are more readily quantified,
many of EPA's air pollution analyses
have focused much more on human
health than on ecological health,
aesthetic effects, or natural resource
productivity. In general, as science and
economics have provided greater
insights into the effects of
anthropogenic stressors on ecological
systems, pursuit of environmental
programs targeted on reductions of
damage to the environment have
become more common. For example,
as we noted in the First Prospective, the original motivation for Title IV of the CAAA
was addressing the effects of acid rain on ecological resources - it was only after passage
that it became clear that these provisions also provide very large human health benefits.
In this chapter, we provide quantitative results for the effects of air pollution on
ecological health and natural resources where the science and economic base is strongest,
including the lake acidification effects that motivated Title IV, as well as a broad
qualitative characterization of effects that are more difficult to quantify. The first portion
of this chapter 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 multiple-stressor effects on ecosystems 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 ecosystem, community, population, organism, etc.). We used a similar strategy in
the First Prospective, which has been updated here to reflect new scientific literature
published since 1999, but we also supplement the literature review with a new mapping
Scenario Development
1
Sector Modeling
I
„
Emissions Direc
I
Air Quality Modeling
1 r
Health Welfare
1 r
Economic Valuation
i
i
Benefit-Cost Compariso
•
Cost
n
6-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
of air pollutant stressors relative to ecological systems that are most sensitive to those
stressors - for example, we relate atmospheric deposition of nitrogen to estuarine systems
that have been classified as sensitive to marginal nitrogen inputs.
The second portion of the chapter presents the results of a wide range of analyses that
quantitatively characterize specific effects of air pollution on ecological systems, as well
as other effects on natural and human systems that contribute to economic welfare. We
provide quantitative estimates of the benefits of the 1990 CAAA for the following
effects:
• Enhanced forest and agricultural plant growth associated with reduced exposure
to tropospheric ozone, on a national scale;
• Enhanced visibility in recreational and residential settings associated with
reduced particulate matter concentrations, also on a national scale;
• Reduced damage to certain building and structural materials associated with
reduced exposure to corrosive air pollutants, such as acid deposition, on a
national scale;
• Acidification of freshwater bodies and impairment of timber growth associated
with atmospheric nitrogen and sulfur deposition, for a case study area in New
York's Adirondack region.
The categories of effects ultimately chosen for quantitative assessment here are
necessarily limited by available methods and data. The scope is largely consistent with
the recommendations of the Ecological Effects Subcommittee (EES) of the Council,
which supported EPA's plans for qualitative characterization of the ecological effects of
CAA-related air pollutants, an expanded literature review, national analyses where
possible, and a quantitative, ecosystem-level case study of ecological service benefits. As
scientific understanding and impact assessment methods grow more comprehensive,
however, we expect that the focus of subsequent analyses will continue to broaden, and
also yield greater insight on which effects that can be avoided by air pollution controls
have the greatest potential ecological and/or economic value.
Because the breadth and complexity of air pollutant-ecosystem interactions do 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 apparent that the economic benefits of the
CAAA's reduction of air pollution impacts on ecosystems are substantial.
QUALITATIVE CHARACTERIZATION OF EFFECTS
The First Prospective summarized available information on the ecological effects of
criteria pollutants and hazardous air pollutants regulated under the 1990 Clean Air Act
Amendments. In this Second Prospective analysis we expand that effort, updating the
6-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
literature review to reflect published and peer-reviewed research that has become
available since the development of the 1999 analysis, through 2008. As data limitations
prevent the quantitative assessment of all potential ecological benefits, the goal of this
effort is to provide a broad characterization of the range of effects of major air pollutants
on ecological endpoints.
Ecosystem impacts can be organized by the pollutants of concern and by the level of
biological organization at which impacts are directly measured. We address both
dimensions of categorization in this overview. Table 6-1 summarizes the major
pollutants of concern, and the documented acute and long-term ecological impacts
associated with them.
The following discussion provides more specific information on ecological effects of
each pollutant class, including information on sources, sensitive ecosystems, and
summary tables of effects organized by level of biological organization.
ACIDIC DEPOSITION
The predominant chemicals associated with acidic precipitation are sulfuric and nitric
acid (H2SO4 and HNO3). These strong mineral acids are formed from sulfur dioxide
(SO2) and nitrogen oxides (NOX) in the atmosphere. Sulfur compounds are emitted from
anthropogenic sources in the form of SO2 and, to a lesser extent, primary sulfates,
principally from coal and residual-oil combustion and a few industrial processes. The
principal anthropogenic source of NOX emissions is fuel combustion. In the atmosphere,
SO2 and NOX are converted to sulfates and nitrates, transported over long distances, and
deposited over large areas downwind of urban areas or point sources.
6-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-1.
CLASSES OF POLLUTANTS AND ECOLOGICAL EFFECTS
POLLUTANT
CLASS
Acidic deposition
Nitrogen
Deposition
Ozone
Hazardous Air
Pollutants (HAPs)
MAJOR
POLLUTANTS AND
PRECURSORS
Sulfuric acid, nitric
acid
Precursors: Sulfur
dioxide,
nitrogen oxides
Nitrogen
compounds (e.g.,
nitrogen oxides)
Tropospheric ozone
Precursors:
Nitrogen oxides
and volatile
organic compounds
(VOCs)
Mercury, dioxins
ACUTE EFFECTS
Direct toxic effects
to plant leaves and
aquatic organisms.
Direct toxic effects
to plants.
Direct toxic effects
to animals.
LONG-TERM EFFECTS
Progressive deterioration of soil quality due to
nutrient leaching. Forest health decline.
Acidification of surface waters. Reduction in acid
neutralizing capacity in lakes and streams.
Enhancement of bioavailability of toxic metals
(aluminum) to aquatic biota.
Nitrogen saturation of terrestrial ecosystems,
causing nutrient imbalances and reduced forest
health. Soil and water acidification. Reduction in
acid neutralizing capacity in lakes and streams.
Progressive nitrogen enrichment of coastal
estuaries causing eutrophication. Changes in the
global nitrogen cycle.
Alterations of ecosystem wide patterns of energy
flow and nutrient cycling; community changes.
Conservation of mercury and dioxins in
biogeochemical cycles and accumulation in the
food chain. Sublethal impacts.
Acidification of ecosystems has been shown to cause direct toxic effects on sensitive
organisms as well as long-term changes in ecosystem structure and function. The effects
of acidification can be seen at all levels of biological organization in both terrestrial and
aquatic ecosystems. Adverse effects in terrestrial ecosystems include acutely toxic
impacts of acids on terrestrial plants and, more commonly, chronic acidification of
terrestrial ecosystems leading to nutrient deficiencies in soils, aluminum mobilization,
and decreased health and biological productivity of forests. These effects can lead to
changes in individual plant survival, as well as changes in forest populations and
communities.
In aquatic ecosystems, acidification-induced effects are mediated by changes in water
chemistry including reductions in Acid Neutralizing Capacity74 (ANC) and increased
availability of aluminum (A13+), which in turn can cause increased mortality in sensitive
species, changes in community composition, and changes in nutrient cycling and energy
flows. Acidic deposition has resulted in increased acidity in surface waters, especially in
areas where acid buffering capacity of soils is reduced and nitrate and sulfate have
M Acid Neutralizing Capacity (ANC) is a measure of overall buffering capacity of a solution or surface waterbody. A well-
buffered system will resist rapid changes in pH, while a poorly buffered system responds quickly to changes in pH.
Reductions in ANC put waterbodies at risk of acidification due to this inability to buffer excess H* ions.
6-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
leached from upland areas. While many fish species are acid-sensitive, the main lethal
agent is the increase in dissolved aluminum that occurs with falling pH levels.
Acid-sensitive ecosystems include those with high acidic deposition and low acid
neutralizing capacity. Many of these ecosystems occur downwind of emission sources,
often in mountainous areas where soils are thin and poorly buffered. High elevation sites
are also more vulnerable because mountain fog is often more acidic than rain.
Table 6-2 provides a summary of the potential ecological effects of acidification.
NITROGEN DEPOSITION
Along with its role in acidification of ecosystems, nitrogen deposition also affects
nitrogen biogeochemistry, which in turn affects the health of forest and coastal
ecosystems. Nitrogen is a naturally occurring element, and is essential to both plant and
animal life, but combustion processes cause this nitrogen to be "fixed" - that is,
converted from the unreactive N2 form to a reactive form such as nitrate (NO3) or
ammonia (NH3). The availability of reactive nitrogen limits plant growth in many
terrestrial ecosystems and is generally the limiting nutrient in marine and coastal waters
as well.
By 1990, human activities had more than doubled the amount of reactive nitrogen
available annually to living organisms. At present, more than 50 percent of the annual
global reactive nitrogen emissions are generated directly or indirectly by human
activities. Ammonia emissions to the atmosphere occur largely via volatilization from
animal wastes. Anthropogenic nitrogen oxide (NOX) emissions to the atmosphere are
generally a result of fossil fuel combustion, with electric power generation and
automobiles as the largest two sources.
Because most terrestrial and coastal ecosystems are nitrogen limited, increased supply of
nitrogen in terrestrial systems can stimulate uptake by plants and microorganisms, and
increase biological productivity. Moderate levels of nitrogen input can have a
"fertilizing" effect, similar to the application of nitrogen fertilizer frequently used in
timber production or agriculture. In the long run, however, chronic nitrogen deposition
adversely affects organisms, communities, and biogeochemical cycles of watersheds and
coastal waters. Biogeochemical cycles change when the nutrient balance is disrupted by
excess nitrogen because nitrogen is an important nutrient in biological systems.
6-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-2. EFFECTS OF ACIDIFICATION ON NATURAL SYSTEMS AT VARIOUS LEVELS OF
ORGANIZATION
SPATIAL SCALE
TYPE OF INTERACTION
EXAMPLES OF EFFECTS
FOREST ECOSYSTEMS
STREAMS AND LAKES
Molecular and
cellular
Chemical and biochemical
processes
Damages to epidermal layers
and cells of plants through
deposition of acids; alteration of
stomatal activity.
Organism
Direct physiological
response
In trees, increased loss of
nutrients via foliar leaching.
Indirect effects:
Acidification can
indirectly affect response
to altered environmental
factors or alterations of
the organism's ability to
cope with other kinds of
stress.
Cation depletion in the soil
causes nutrient deficiencies in
plants. Concentrations of
aluminum ions in soils can reach
phy to toxic levels. Increased
sensitivity to other stress factors
including pathogens and frost.
In birds, possible calcium
limitation and growth reduction.
Population
Change of population
characteristics like
productivity or mortality
rates.
Decrease of biological
productivity of sensitive
organisms. Selection for less
sensitive organisms.
Microevolution of resistance.
Community
Changes of community
structure and competitive
patterns.
Local Ecosystem
(e.g., landscape
element)
Changes in nutrient cycle,
hydrological cycle, and
energy flow of lakes,
wetlands, forests,
grasslands, etc.
Alteration of competitive
patterns. Selective advantage
for acid-resistant species. Loss
of acid sensitive species and
organisms. Decrease in
productivity. Decrease of
species richness and diversity.
Decline in Sugar Maple and red
spruce in Eastern U.S. and
Canadian forests.
Progressive depletion of nutrient
cations in the soil. Increase in
the concentration of mobile
aluminum ions in the soil.
Regional
Ecosystem (e.g.,
watershed)
Biogeochemical cycles
within a watershed.
Region-wide alterations of
biodiversity.
Leaching of sulfate, nitrate,
aluminum, and calcium to
streams and lakes. Change in
sulfur and nitrogen
biogeochemistry in northeastern
forests.
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.
Decrease of biological
productivity and increased
mortality of sensitive organisms.
Selection for less sensitive
organisms. Microevolution of
resistance.
Alteration of competitive
patterns. Selective advantage
for acid-resistant species. Loss
of acid sensitive species and
organisms. Decrease in
productivity. Decrease in
species richness and diversity.
Acidification of lakes and
streams. Decrease in acid
neutralizing capacity. Persistent
acidic conditions in lakes and
streams in some regions, despite
reduction in sulfate deposition.
Regional acidification of aquatic
systems due to high deposition
rates and nitrogen saturation of
terrestrial ecosystems and
increased nitrate leaching to
surface waters. Persistent acidic
conditions in lakes and streams
in some regions, despite
reduction in sulfate deposition.
6-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Because fresh waters are generally not nitrogen limited, the addition of nitrogen does not
lead to excessive eutrophication as it does in coastal waters. Coastal waters are an
extraordinarily important natural resource, providing spawning grounds/nurseries for fish
and shellfish, foraging and breeding habitat for birds, and generally contributing greatly
to the productivity of the marine environment. Critical to the health of coastal waters is
an appropriate balance of nutrients. If present in mild or moderate quantities, nitrogen
enrichment of coastal waters can cause moderate increases in productivity, leading to
neutral or positive changes in the ecosystem. However, because coastal waters are
generally nitrogen limited, too much nitrogen leads to excess production of algae,
decreasing water clarity and reducing concentrations of dissolved oxygen, a situation
referred to as eutrophication.
Table 6-3 summarizes the potential effects of nitrogen deposition on ecosystem structure
and function.
TROPOSPHERIC OZONE
Ozone is a secondary pollutant formed through the oxidation of volatile organic
compounds (VOCs) in the presence of oxides of nitrogen. Ozone is one of the most
powerful oxidants known but its impacts have been little studied in faunal species. The
limited available research has shown a variety of pulmonary impacts to specific
mammalian and avian species. In contrast, ozone's impacts on plants are much better
understood. Documented effects on forest trees include visible foliar damage, decreased
chlorophyll content, accelerated leaf senescence, decreased photosynthesis, increased
respiration, altered carbon allocation, water balance changes, and damage to epicuticular
wax. These can lead to changes in canopy structure, carbon allocation, productivity, and
fitness of trees.
Ozone sensitivity of plants varies between species, with evergreen species tending to be
less sensitive to ozone than deciduous species, and with most individual deciduous trees
being less sensitive than most annual plants. However, there are exceptions to this broad
ranking scheme, and there can be variability not only between species but even between
clones of some trees and within cultivars. Life stage also matters: in general, mature
deciduous trees tend to be more sensitive than seedlings, while the reverse is more typical
for evergreen trees.
6-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-3. EFFECTS OF NITROGEN DEPOSITION ON NATURAL SYSTEMS AT VARIOUS LEVELS OF
ORGANIZATION
SPATIAL SCALE
Molecular and cellular
Organism
Population
Community
Local Ecosystem
(e.g., landscape
element)
TYPE OF INTERACTION
Chemical and biochemical
processes.
Direct physiological
response.
Indirect effects: Response
to altered environmental
factors or alterations of
the organism'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.
EXAMPLES OF EFFECTS
FOREST ECOSYSTEAAS
Increased uptake of
nitrogen by plants and
microorganisms. With
chronic exposure, reduced
stomatal activity and
photosynthesis in some
species.
Increases in leaf size of
terrestrial plants.
Increase in foliar nitrogen
concentration in major
canopy trees. Change in
carbon allocation to
various plant tissues.
Decreased resistance to
biotic and abiotic stress
factors including
pathogens, insects, and
frost. Disruption of plant-
symbiont relationships
with mycorrhizal fungi.
Increase in biological
productivity and growth
rates of some species.
Increase in pathogens.
Alteration of competitive
patterns. Selective
advantage for fast growing
species and organisms that
efficiently use additional
nitrogen. Loss of species
adapted to nitrogen-poor
or acidic environments.
Increase in weedy species
or parasites.
Changes in the nitrogen
cycle. Progressive
nitrogen saturation.
Mobilization of nitrate and
aluminum in soils. Loss of
calcium and magnesium
from soil. Change in
organic matter
decomposition rate.
ESTUARINE ECOSYSTEAAS
Increased assimilation of
nitrogen by marine
plants, macroalgae, and
microorganisms.
Increase in algal growth.
Injuries to marine fauna
through depletion of
oxygen in the water
column. Loss of physical
habitat due to increased
macroalgal biomass and
loss of seagrass beds.
Injury and habitat loss
through increased shading
by macroalgae.
Increase in algal and
macroalgal biomass.
Excessive algal growth.
Changes in species
composition with increase
in algal and macroalgal
species and decrease or
loss of seagrass beds.
Loss of species sensitive
to low oxygen conditions.
Changes in the nitrogen
cycle. Increased algal
growth leading to
depletion of oxygen,
increased shading of
seagrasses. Reduced
water clarity and
dissolved oxygen levels.
6-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
SPATIAL SCALE
Regional Ecosystem
(e.g., watershed)
Global Ecological System
TYPE OF INTERACTION
Changes in biogeochemical
cycles within a watershed.
Region-wide alterations of
biodiversity.
Changes in global
biogeochemical cycles;
increased availability of
reactive nitrogen to
plants.
EXAMPLES OF EFFECTS
FOREST ECOSYSTEAAS
Leaching of nitrate and
aluminum from terrestrial
sites to streams and lakes.
Acidification of soils and
waterbodies. Increased
emission of greenhouse
gases from soils to
atmosphere. Change in
nutrient turnover and soil
formation rates.
Increased input of reactive
nitrogen; loss of soil
nutrients. Nitrogen
saturation and leaching
throughout forests in
northeastern United States
and Western Europe.
Acidification of surface
waters.
ESTUARINE ECOSYSTEAAS
Additional input of
nitrogen from nitrogen-
saturated terrestrial sites
within the watershed.
Regional decline in water
quality in waterbodies
draining large watersheds
(e.g. Chesapeake Bay).
Changes in the regional-
scale nitrogen cycle.
Greatly increased transfer
of nitrogen to coastal
ecosystems; change in
structure and function of
estuarine and nearshore
systems.
Impacts to plant communities may occur as a result of ozone exposure, although such
effects have not been studied as extensively due to ecosystem complexity and the long
timeframes involved. Experiments with an early successional plant community found
that ozone reduced vegetative cover, vertical density, species richness, and evenness
relative to the control, although differences were less pronounced in a drought year.
Other observed community level effects include reduced competitive ability of sensitive
species, changed soil microbial communities, and altered species composition and
relative abundance.
Table 6-4 summarizes the potential effects of ozone exposure on ecosystems.
HAZARDOUS AIR POLLUTANTS
Hazardous air pollutants (HAPs) are a general category of toxic substances covered under
Title III of the Clean Air Act, which lists 189 HAPs. Of these 189 substances, the best
understood in terms of the potential for adverse ecological impacts include mercury,
polychlorinated biphenyls (PCBs), dioxins, and dichlorodiphenyl-trichloroethane (DDT).
The use of PCBs 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). With respect to mercury and
dioxins, regulatory actions have reduced, but have not eliminated, anthropogenic
emissions. This section discusses environmental effects associated with these two HAPs.
6-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-4.
EFFECTS OF OZONE ON NATURAL SYSTEMS AT VARIOUS LEVELS OF ORGANIZATION
SPATIAL SCALE
Molecular and
cellular
Organism
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
organism'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.
EXAMPLES OF EFFECTS
Oxidation of enzymes of plants, generation
of toxic reactive oxygen species (hydroxyl
radicals). Disruption of the membrane
potential.
Reduced photosynthesis and nitrogen
fixation. Increased apoptosis.
Visible foliar damage, premature needle
senescence, altered carbon allocation, and
reduced growth rates.
Increased sensitivity to bio tic and abiotic
stress factors such as pathogens and frost.
Disruption of plant-symbiont relationship
(mychorrhizae), and symbionts.
Reduced biological productivity and
reproductive success. Selection for less
sensitive organisms. Potential for
microevolution for ozone resistance.
Alteration of competitive patterns. Loss of
ozone sensitive species and organisms
leading to reduced species richness and
evenness. Reduction in productivity.
Changes in microbial species composition
in soils.
Alteration of ecosystem- wide patterns of
energy flow and nutrient cycling (e.g., via
alterations in litter quantity, litter
nutrient content, and degradation rates;
also via changing carbon fluxes to soils and
carbon sequestration in soils).
Potential for region-wide
phytotoxicological impacts and reductions
in net primary production.
Mercury
Mercury (Hg) is a toxic element found ubiquitously throughout the environment. About
50-80 percent of total emissions originate from anthropogenic sources, including fossil
fuel combustion, leaks from industrial activities, and the disposal or incineration of
wastes.
Mercury is generally released in its elemental and inorganic forms. However, it can
undergo various transformations in the environment, and its chemical form determines
not only its environmental fate but also its potency as a toxicant. From a biological
perspective, the most hazardous form of mercury is methylmercury both because of its
6-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
bioaccumulation and biomagnification potential, and also because organic forms of
mercury (including methylmercury) are the most toxic. Adverse effects on wildlife
include neurotoxicity as well as reproductive, behavioral, and developmental effects.
These types of effects have been observed in laboratory studies of mammals, birds, fish,
and aquatic invertebrates. While species sensitivity varies, within a species the early life
stages are generally the most sensitive.
Dioxins
Polychlorinated dibenzo-p-dioxins (PCDDs) are a group of 75 organochlorine
compounds, often referred to as dioxins. Although dioxins can be produced through
natural events such as forest fires and volcanic eruptions, most environmental inputs are
anthropogenic in origin. EPA categorizes dioxin sources into five broad groups:
combustion; metals smelting, refining, and processing sources; chemical manufacturing;
biological and photochemical processes; and reservoir sources (for example urban
runoff).
Dioxins and related compounds are thought to exert most of their toxic effects through
interaction with the aryl hydrocarbon receptor (AhR). In laboratory studies, particularly
of rodents, some dioxins have been shown to cause reproductive toxicity, neurotoxicity,
immune suppression, increased inflammatory responses, and cancer. Fish are among the
most sensitive species to the effects of dioxin, and early life stages are the most
vulnerable. The risk that dioxins pose to other wildlife is difficult to assess because both
laboratory and field studies are few.
Dioxins are extremely stable chemicals with a persistence that is measured in decades.
Dioxins are subject to photochemical degradation, but since the penetration of light into
soils and many natural water bodies is limited, this degradation is slow. Because of
dioxins' toxicity and persistence, their presence is likely to be an issue of concern for
decades.
DISTRIBUTION OF AIR POLLUTANTS IN SENSITIVE ECOSYSTEMS OF THE UNITED
STATES
This section describes the spatial and temporal trends of air pollutants regulated by the
CAAA, highlighting their distribution against sensitive ecosystems across the United
States. This information provides useful context regarding the geographic distribution of
potential ecological benefits of the CAAA, particularly for the ecological endpoints
described above for which data are not available to quantify impacts.
The maps presented illustrate changes in forecast pollutant levels under the current,
baseline scenario (with the CAAA) as compared to the counterfactual scenario (without
the CAAA). The three pollutant classes considered are: acidic deposition, nitrogen
deposition, and tropospheric ozone. Data are not available to map the distribution of
HAPs. The pollutant exposure maps presented in this discussion were created using data
from the Community Multiscale Air Quality Modeling System (CMAQ) Version 4.6,
6-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
which estimates tropospheric ozone concentrations as well as deposition in kilograms per
hectare for acidic deposition and total nitrogen.75
ACIDIC DEPOSITION
As described in the previous section, ecosystem sensitivity to acid deposition occurs in
areas with low ANC. High elevation sites tend to be more vulnerable because of thin,
poorly buffered soils coinciding with acidic deposition from rain, snow, and fog. Acid-
sensitive areas in the U.S. include the southern Blue Ridge Mountains of eastern
Tennessee, western North Carolina and northern Georgia; the mid Appalachian Region of
eastern West Virginia, western Virginia and central Pennsylvania; New York's Catskill
and Adirondack Mountains; the Green Mountains of Vermont; the White Mountains of
New Hampshire, and areas of the Upper Midwest (Wisconsin and Michigan).76 Montane
areas in the Adirondacks, Northern New England, and the Appalachian region have
experienced acidification of surface waters and soils, as well as forest decline.
Figure 6-1 presents acidic deposition from 1990 through 2020 for both with- and without-
CAAA scenarios. Acid deposition estimates are expressed as equivalents per hectare
(eq/ha).77 Under both regulatory scenarios, acidic deposition is highest in western
Pennsylvania, southern Ohio and Indiana, western West Virginia, and northern Kentucky.
Without the CAAA, acidic deposition in these areas increases overtime. Further, acidic
deposition increases overtime in the areas surrounding these hotpots. By 2020,
significant portions of the Northeast, Midwest, and South are projected to have elevated
levels of acidic deposition. Hotspots also exist in eastern Texas and southern Louisiana.
As shown in the right column of Figure 6-1, with the CAAA acidic deposition levels
lessen in and around the areas with the highest acidic deposition. By 2020, elevated
acidic deposition levels are primarily limited to much smaller areas in the Midwest,
Northeast, and Gulf Coast.
75 The CMAQ tool is described in more detail in Chapter 4 of this document.
76 U.S. Environmental Protection Agency (EPA). October 2003. Response of surface water chemistry to the Clean Air Act
Amendments of 1990. EPA 620/R-03/001.
77 Acid deposition is calculated using the hydrogen deposition derived from both sulfur and nitrogen deposition as described
in: U.S. Department of Agriculture, Forest Service, Rocky Mountain Region. January 2000. Screening Methodology for
Calculating ANC Change to High Elevation Lakes: User's Guide. The deposition estimates in Figures 6-2 and 6-3 include
combined wet and dry deposition for the stated years as estimated by the CMAQ modeling system version 4.6. These
modeled estimates are not calibrated with monitored deposition data such as the National Atmospheric Deposition Program
(NADP)data
6-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-1. COMBINED NOX AND SOX DEPOSITION ESTIMATES FOR 1990, 2000, 2010, AND 2020
WITH AND WITHOUT THE CAAA
[Scenario: No CAM; Year: 1990
Acid Deposition:
With & Without CAAA (eq/hectare):
0-2,581
2,582-2,793
2,794-3p43
3,044-3338
3,339-3,735
3,736-4200
4,201 -4,757
4,758-5541
| 5,542- 7049
I 7,050-20,724
The upper-bounds of the deposition classes range from the
77.5th percentile to the 100th percentile, increasing at
increments of 2.5 percent (i.e., 77.5,80.0,82.5,...,100th)
\Scenario: No CAAA; Year: 2000
[Scenario: With CMA; Year: 2000
IScenario: No CAAA; Year: 2010
[Scenario: With CAAA; Year: 2010
[Scenario: Mo CAAA; Year: 2020
''V'.. •
[Scenario: With CAAA; Year: 2020
610
1,220
1,830
2,440
Sources:
1.) CMAQ Version 4.6 (Provided by ICF International, October 2, 2008)
2.) Environmental Systems Research Institute, Inc.
Jr I EC
Map Projection: Lsmbert Conforms! Conic
Geodetic Reference System: N£D83
6-13
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
NITROGEN DEPOSITION
Atmospheric nitrogen deposition is highest in the northeastern and eastern central regions
of the U.S. Elevated nitrogen deposition in the western and southern United States is
limited to areas in the vicinity of large nitrogen sources (e.g., livestock production areas),
high-elevation areas on which cloud droplet deposition may contribute substantial
nitrogen inputs, and urban areas with relatively high levels of NOx emissions.
Figure 6-2 presents total nitrogen deposition from years 1990 through 2020 for both the
with-CAAA and without-CAAA scenarios. In general, total nitrogen deposition is less than
24 kg/hectare in the conterminous U.S. for each year and regulatory scenario presented.
However, "hot spots" exist across the U.S. where meteorological conditions and/or high
nitrogen emissions contribute to relatively high deposition rates. Two particularly
significant hot spots for nitrogen deposition are located in southern Louisiana and eastern
North Carolina. Total nitrogen deposition is estimated to increase in both hot spots over
time regardless of the regulatory scenario. Outside of the two hot spots, total nitrogen
deposition is highest without the CAAA in the Ohio River Valley (i.e., western
Pennsylvania, southern Ohio and Indiana, western West Virginia, and northern
Kentucky). Over time, the total nitrogen deposition increases around the Ohio River
Valley without the CAAA and decreases slightly with the CAAA. Outside of the Ohio
River Valley, nitrogen deposition with the CAAA decreases slightly over time in the
eastern U.S. In the western U.S., total nitrogen deposition with the CAAA remains
relatively constant over time.
Estuarine areas in the Northeast are less susceptible to injury from nitrogen loading than
estuaries in other parts of the country due to the rapid flushing characteristics of estuaries
in this region. Estuaries along the Southeastern Coast, Gulf Coast, and Southern
California Coast experience the greatest reduction in total nitrogen deposition. Total
nitrogen deposition along the West Coast, with the exception of southern California, is
relatively low in the absence of the CAAA.
TROPOSPHERIC OZONE
Areas within the U.S. with elevated tropospheric ozone levels include the Northeast, mid-
Atlantic, Midwest, and California. Combined ozone concentrations are reported for the
May through September period as ozone levels tend to increase during the spring and
summer. Figure 6-3 presents combined cumulative ozone season (W126) values for the
May through September period for both the with-CAAA and without-CAAA scenarios.
The W126 metric is a weighted sum of hourly concentrations observed between 8 a.m.
and 8 p.m. where hourly weights are a function of the hourly ozone concentration
observed.
6-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-2. TOTAL NITROGEN DEPOSITION ESTIMATES FOR 1990, 2000, 2010, AND 2020 WITH
AND WITHOUT THE CAAA78'79
:
Total Nitrogen Deposition:
With & Without CAAA
Deposition Classes (kg/hectare)
0.00-2.99 (31.3%) 10.00- 11.99 (83.9%)
3.00 - 3.99 (38.8%) | 12.00 - 1 3.99 (90.6%)
4.00-5.99 (51.1%) ^H 14.00-23.99 (99.5%)
6.00 - 7.99 (E2.5%) ^H Greaterthan 24.00 (100%)
8.00 - 9.99 (74.2%)
*Percentiles for the upper bound within each value bin
are presented within parentheses.
^Scenario: No CAAA; Year: 2000
Scenario: With CAAA; Year: 2000 \
—
Vi
^
[Scenario: No CAAA: Year: 2020
[Scenario: With CAAA: Year: 2020 |
600 1,200 1,800 2,400
^ ^— iMIIes
Sources:
1.) CMAQ Version 4.6 (Provided by ICF International, October 2, 200S)
2.) Environmental Systems Research Institute, Inc.
it
~~~i t CMa|ipro'ec!ion:Li
.smbert Conformal Ccnic
Geodetic Reference System: N£D 83
1HU5-Ri.H ECOMOMl^^, IKCD^^QRtTED
78 Value bins for nitrogen deposition taken from: Rea, A., J. Lynch, R. White, G. Tennant, J. Phelan and N. Possiel. 2009.
Critical Loads as a Policy Tool: Highlights of the NOx/SOx Secondary National Ambient Air Quality Standard Review. Slide 6:
Nationwide Total Reactive Nitrogen Deposition (2002). Available online at:
http://nadp.sws.uiuc.edu/meetings/fall2009/post/session4.html.
79 Percentiles are calculated using the combined nitrogen deposition data for all years and scenarios presented in the map.
6-15
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-3. W126 CUMULATIVE TROPOSPHERIC OZONE SEASON MEASURES FOR 2000, 2010,
AND 2020 WITH AND WITHOUT THE CAAA
Sources:
1.) W126 Estimates Provided by Stratus
Consulting on July 21, 2009.
2.) Environmental Systems Research
Institute, Inc.
Tropospheric Ozone Concentrations
W126 Values (ppm-hours)
0.00-5.00 25.01-50.00
5.01-10.00 50.01-75.00
10.01 - 15.00 75.01 - 100.00
15.01 - 20.00 • 100.01 - 200.00
20.01 -25.00 • 200.01 -315.00
lEc
hybp Projection: Lambert Confcnnal C
Geodetic Reference System: NAD 19:
6-16
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
In general, tropospheric ozone concentrations increase over time without the CAAA and
decrease over time with the CAAA. Elevated ozone concentrations are present in
California, mid-Atlantic states, and Corn Belt states in 2000 both with and without the
CAAA; ozone concentrations are, however, slightly less with the CAAA in 2000. In
2000, ozone hot spots are present in southern California, central Ohio, portions of
Virginia, North Carolina, and South Carolina, and western Tennessee. Without the
CAAA, these hot spots grow in size and magnitude. Under the with-CAAA scenario, the
hot spots decrease in size and magnitude. By 2020, the combined W126 values for nearly
the entire conterminous U.S. (outside of California) are less than 15 ppm-hours.
Tropospheric ozone concentrations within the California hot spot are reduced to 25 to 75
ppm-hours.80
As noted in the previous section, elevated tropospheric ozone levels may negatively
affect plants in a number of ways, including reducing plant photosynthesis and increasing
leaf senescence leading to reduced plant growth and productivity. Given the potential
effects of elevated tropospheric ozone concentrations on plant growth, forested and
cropland areas across the U.S. are considered particularly sensitive to the effects of
elevated tropospheric ozone. It follows that these same areas also stand to benefit the
most from reduced tropospheric ozone concentrations due to the implementation of the
CAAA. In particular, forested ecosystems in the San Bernardino and Sierra Nevada
Mountains of California have suffered ecological damages attributed to elevated ozone
levels. Forests in the southern portions of the Midwest and Northeast regions and the
Southeast region (except the southernmost areas where ozone concentrations are
relatively low without the CAAA) are also expected to benefit from reductions in
tropospheric ozone due to the implementation of the CAAA. In addition, crops in
California are expected to benefit the most from the implementation of the CAAA. The
cropland areas in California are located almost entirely within the tropospheric ozone hot
spot. Other cropland areas expected to benefit from reduced tropospheric ozone
concentrations associated with the implementation of the CAAA include the Corn Belt
region, the southern portion of the Midwest region, the Mississippi Valley, Texas, and
Oklahoma.
QUANTIFIED RESULTS: NATIONAL ESTIMATES
AGRICULTURE AND FOREST PRODUCTIVITY EFFECTS
A significant body of literature exists addressing the effects of tropospheric ozone on
plants, including commercial tree species and agricultural crops, as noted in the previous
section. In general, elevated levels of tropospheric ozone have been shown to reduce
Within the California hot spot, the modeled CMAQ ozone concentration estimates were low compared to the ozone
monitoring data. This may have resulted in the eVNA analysis overestimating future ozone concentrations. This
overestimate is expected to have occurred in this region for both the with-CAAA and without-CAAA scenarios, however, and
therefore the effect on the difference in ozone concentrations between the two scenarios is uncertain.
6-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
overall plant health and growth by reducing photosynthesis and altering carbon
allocation. Methods and data also exist to estimate the magnitude of plant growth
reductions due to elevated tropospheric ozone levels, based on laboratory studies that
developed exposure-response functions describing the functional relationship between
plant yield and ozone exposure for a variety of plant species.81 Applying exposure-
response functions, this analysis estimates yield losses in agricultural crops and
commercial tree species under the counterfactual, without-CAAA scenario relative to the
baseline, with-CAAA scenario. Relative yield losses (i.e., reductions in crop and tree
yield under the counterfactual scenario relative to the baseline scenario) measure the
amount crop and tree yields would be reduced in the absence of CAAA regulations, and
therefore, indicate a benefit of the CAAA.82
Table 6-5 provides a summary of estimated relative yield losses by crop/forest type and
year. Relative yield losses indicate a benefit of the CAAA; the larger the relative yield
loss without the CAAA, the greater the crop or tree yield with the CAAA. In addition,
Figures 6-4 and 6-5 provides maps of the crop-subregion-specific and tree-region-specific
relative yield losses for two representative species: potatoes and softwood trees. The
results presented generally follow the temporal and spatial pattern of ozone concentration
reductions attributable to the CAAA, as outlined in Chapter 4, with reductions in
tropospheric ozone concentrations being greatest along the East Coast, particularly the
Southeast, in the Midwest (within the Ohio River Valley), and in California. Several
other factors also affect yield changes in crops and trees, including sensitivity to ozone,
geographic distribution, growing period length, and the specific time of year the growing
period occurs. Potatoes and softwoods, as indicated in Table 6-5, suffer relatively larger
changes in growth than some other species in our analysis, and yield losses tend to
increase over time as differences in ozone concentrations increase between the with-
CAAA and without-CAAA scenarios. Across all crops, the largest relative yield losses for
both crops and trees occur in the Southeast, frequently in Virginia, North Carolina, South
Carolina, and Tennessee.
See, for example, E.H. Lee and W.E. Hogsett. 1996. Methodology for Calculating Inputs for Ozone Secondary Standard
Benefits Analysis: Part II. Prepared for the U.S. EPA, Office of Air Quality Planning and Standards, Air Quality Strategies and
Standards Division. The application of laboratory-derived functions is less preferable than functions developed from field
studies. However, the laboratory-derived functions frequently provide the best available information regarding the
relationship between ozone exposure and crop or tree growth. The exposure-response functions applied in this report have
been used in other EPA studies, such as: USEPA. July 2007. Review of the National Ambient Air Quality Standards for
Ozone: Policy Assessment of Scientific and Technical information. EPA-452/R-07-007.
82 Relative yield losses are estimated instead of relative yield gains because the baseline (with CAAA) scenario in this analysis
defines current conditions, whereas the counterfactual (no CAAA) scenario defines a change in current conditions. The
models applied in this analysis forecast changes in yield relative to current conditions (i.e., relative to the baseline
scenario).
6-18
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-5. MINIMUM, MAXIMUM, AND AVERAGE ANNUAL RELATIVE YIELD LOSSES ACROSS ALL FASOM SUBREGIONS FOR CROPS AND
ALL FASOM REGIONS FOR TREES BY YEAR (2000, 2010, 2020)
CROP/FOREST TYPE
Barley
Corn
Cotton
Oranges
Potato
Rice
Sorghum
Soybean
Processing Tomatoes
Spring Wheat
Winter Wheat
Hardwood Forests
Softwood Forests
2000
MINIMUM
0.00%
0.00%
0.00%
0.00%
0.00%
-0.08%
0.00%
0.00%
0.00%
0.00%
0.00%
1.60%
0.06%
MAXIMUM
0.02%
1.12%
6.60%
1 .95%
6.17%
0.14%
0.87%
3.60%
1.82%
1.50%
6.53%
7.16%
3.85%
AVERAGE
0.01%
0.18%
1.15%
0.09%
1 .76%
0.00%
0.14%
1 .24%
0.31%
0.06%
1 .00%
5.06%
1 .77%
2010
MINIMUM
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-0.55%
0.00%
0.00%
0.00%
4.20%
0.25%
MAXIMUM
0.06%
3.07%
16.67%
4.68%
17.54%
1 .03%
2.17%
11.73%
5.54%
3.67%
18.23%
19.12%
10.49%
AVERAGE
0.02%
0.44%
3.00%
0.25%
4.99%
0.11%
0.35%
3.07%
0.96%
0.15%
2.49%
13.86%
4.88%
2020
MINIMUM
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
6.61%
0.42%
MAXIMUM
0.07%
3.45%
20.31%
7.87%
20.80%
1.66%
2.65%
12.74%
8.21%
6.98%
19.23%
23.04%
12.27%
AVERAGE
0.02%
0.56%
3.81%
0.43%
6.50%
0.18%
0.47%
4.26%
1.47%
0.28%
3.29%
16.68%
6.11%
Note: Negative relative yield losses indicate yield reductions with the CAAA. For example, the minimum estimate for soybeans in
2010 reflects an estimated relative yield loss of -0.55 percent. The negative relative yield loss is due to reductions in W126 ozone
metric values under the counterfactual, no CAAA scenario in Florida in September of 2010 (the growing period for soybeans in
Florida is roughly mid-July through September). In other words, ozone exposure is greater under the with-CAAA scenario for that
month and region and, therefore, a net increase in soybean yield occurs assuming a rollback of the CAAA. Ozone concentrations are
lower under the baseline, with CAAA scenario in Florida for all other months in 2010.
6-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-4. RELATIVE ANNUAL YIELD LOSSES IN POTATOES UNDER THE COUNTERFACTUAL (NO
CAAA) SCENARIO BY FASOM SUBREGION AND YEAR BASED ON SUBREGIONAL-
SPECIFIC OZONE CONCENTRATIONS AND GROWING PERIODS
I 'I
A
Relative Yield Loss Under the
Counterfactual (No CAAA) Scenario
(Percent Reduction in Yield)
| 0.0% -1.0%
| 2.6% -5.0%
| 5.1% -7.5%
7.6% -10.0%
10.1%- 15.0%
15.1%- 20.0%
20.1%- 25.0%
I Crop Not Present in Subregion
FASOM Sub regions
Sources:
1.) W126 Estimates Provided by Stratus
Consulting on July 21, 2009.
2.) Yield Functions Found in EPA2007.
3.) FASOM Subregions Provided by RTI
International on February 19, 2009.
4.) Environmental Systems Research
Institute, Inc.
1,760
Mies
lEc
Map Projection: Lambert Conformal Conic
Geodetic Reference System: N,V11983
6-20
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-5. RELATIVE ANNUAL YIELD LOSSES IN SOFTWOOD FOREST TYPES UNDER THE
COUNTERFACTUAL (NO CAAA) SCENARIO BY FASOM REGION AND YEAR BASED ON
REGIONAL-SPECIFIC OZONE CONCENTRATIONS AND GROWING PERIODS
I 1
A
|Year= 20001
ffear= 20201
Relative Yield Loss Under the
Counterfactual (No CAAA) Scenario
(Percent Reduction in Yield)
] 0.0%-1.0%
| 1.1%-2.5%
] 2.6%-5.0%
] 5.1%-7.5%
j 7.6%-10.0%
| 10.1%- 15.0%
| 15.1%- 20.0%
| 20.1%- 25.0%
Forest Type Not Present in Region
FASOM Regions
Padlc.Ncllh sea Wert :i 'He
Sources:
1.) W126 Estimates Provided by Stratus
Consulting on July 21, 2009.
2.) Yield Functions Found in EPA2007.
3.) FASOM Subregions Provided by RTI
International on February 1 9, 2009.
4.) Environmental Systems Research
Institute, Inc.
lEc
Map Projection: Lambert Conform al Conic
Geodetic Reference System: N£D 1983
IMDUlTWAt. ICOMQHICt. INCOMOMTIt
6-21
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Commercial timber and agriculture operations generally manage their land to maximize
profits. As such, changes in crop yields between the baseline and counterfactual
scenarios may affect the distribution of commercial species planted; for example,
landowners may shift production towards plants that are less sensitive to elevated ozone
concentrations under the counterfactual scenario. This may occur at the individual plant
level, replacing one crop or tree species for another with a higher growth rate; or, it may
occur at the community level, converting agricultural lands to timberlands, or vice versa,
to adjust for combined yield losses to agricultural crops and commercial tree species.
Changes in the distribution and yield of crop and tree species may in turn affect the
supply of and demand for agricultural crops and commercial tree species, resulting in
changes in the welfare of consumers and within agricultural and timber sectors of the
economy. To quantify this economic benefit of cleaner air, we used the Forest and
Agriculture Sector Optimization Model (FASOM). FASOM development was funded by
EPA's Climate Economics Branch (CEB) and other EPA, U.S. government, and non-
governmental flinders over several decades as a partial equilibrium tool to evaluate the
welfare and market impacts of public policies affecting agriculture and forestry. The
model simulates biophysical and economic processes affecting land management and
land allocation decisions over time to potentially competing agriculture and forest
activities. Although the latest version of FASOM was developed to evaluate climate and
biofuels policies, the model is capable of assessing a broad range of factors that might
affect plant growth; for this project, we worked with the model's developers to develop
input files to characterize the impact of ozone on plant and tree growth at a regional and
crop-specific level, using the exposure-response results described above.83
Although FASOM has been widely applied to agricultural sector analysis and has been
peer reviewed in many contexts, it has not to date been subject to a validation exercise
comparing the model results for an historical period to historical data for that period.84
As a result, the performance of the model in forecasting future agricultural sector effects,
such as those estimated for this study, has not yet been assessed. Two other potential
limitations may pertain in EPA's application of FASOM for this study. First, FASOM
adopts a model simulation approach which assumes perfect foresight by economic actors
in the agricultural sector. A perfect foresight assumption may be of concern for some
Note that we performed two runs of the FASOM model, one where the response to ozone for those crop/region
combinations without specific individual concentration-response functions are assumed to be zero, and a second where
impacts on crop/region combinations without specific concentration-response functions were set to the values used in
adjacent regions and/or proxy crops where possible (for example, soft white wheat was used for barley and sugarbeets;
tomatoes for processing were used for potatoes; soybeans for fresh tomatoes; corn for fresh tomatoes if there is not a value
for soybeans; etc.). We found that the difference in the overall national results between these two runs was negligible,
however. As a result, in this chapter we report the results from the run that applies proxy crop/region concentration-
response functions. Note further that the version of FASOM used for this analysis is the version current as of July 21, 2010.
M See, for example, a review commissioned by USEPA for its application of FASOM to support regulatory analysis of
renewable fuels standards, concluded in July of 2010 and available at the following web site (accessed November 26, 2010):
http://www.epa.gov/otaq/fuels/renewablefuels/regulations.htm
6-22
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
long-term analyses, but is likely to be less problematic for this study because our time
horizon extends only to 2020. Furthermore, USDA projections of commodity prices and
outputs also extend nearly to 2020, and FASOM's projections for their base case agree
well with the USDA projections. As a result, the effect of perfect foresight on model
outcomes in the present study is reduced.85 A second potential limitation of FASOM is
its approach to estimating the sensitivity of imports to changes in domestic prices.
Although FASOM is not a full international model, it does incorporate an import
elasticity estimate for the largest and most important commodity crops. This allows the
model to capture, for example, increases in agricultural imports to the US under a
scenario in which domestic crop prices are projected to rise. For a number of minor
crops, traded in very small quantities, however, FASOM holds imports fixed. The effect
of this factor on our results is not clear, but we estimate that a more flexible import sector
for these much less important crops would have only a minor effect on our estimates of
the net benefits of reducing ozone exposure for US crops. We expect the directional bias
of holding minor crop imports fixed, while small, would be to slightly reduce our
estimates of the net welfare benefit of reducing ozone exposure, and thereby improving
productivity, of domestic agricultural crops.
The economic welfare results of the FASOM modeling are presented in Table 6-6.
FASOM generates total welfare estimates for the agricultural and forest sectors for each
of our scenarios, for each target year, reflecting the sum of total consumer and producer
surplus derived from agriculture and forest production. In general, higher ozone
concentrations in the without-CAAA scenario lead to reduced agricultural and forest
productivity, raising prices for these products, which in turn increases producer surplus
but reduces consumer surplus by a larger amount. As a result, FASOM estimates the net
welfare benefits of the CAAA to be approximately $1 billion in 2000, $5.5 billion in
2010, and $10.7 billion in 2020, increasing overtime as the differences in ozone
concentrations grows.86
85 Perfect foresight is a basic assumption of the modeling approach on which FASOM is based. Structuring the model based on
perfect foresight rather than a myopic (recursive) approach allows an expanded array of policy simulations and potential
insights, which is the main purpose of this type of model.
86 Note that the year 2000 in FASOM represents average annual activity over the 5-year period from 2000 to 2004; 2010
represents 2010 through 2014; and 2020 represents 2020 through 2024. Values provided for ozone impacts in 2000, 2010, and
2020 were applied to the 2000, 2010, and 2020 model periods in FASOM, respectively. The results presented here do not
includes losses Canada and the rest of the world; for example, in 2020, higher US prices in the without-CAAA scenario result
in additional consumer surplus losses to non-US consumers of $1.7 billion in the forest sector and $3.3 billion in the
agricultural sector.
6-23
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-6. SUMMARY OF FASOM RESULTS: TOTAL CONSUMER AND PRODUCER SURPLUS VALUES
FOR THE AGRICULTURAL AND FOREST SECTORS
VARIABLE
Annual Welfare,
US Forest Sector
Annual Welfare,
US Agriculture
Sector
Annual Welfare,
Forest and
Agriculture Sector
Combined
MODEL RUN
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Damage Estimate ($ billion)
Percent change
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Damage Estimate ($ billion)
2000 2010 2020
$637
$636
$1.5
$877
$875
$1.7
$1426
$1426
$0
0.24% | 0.20% | 0%
$1706
$1706
-$0.5
$1831
$1828
$3.8
Percent change -0.03% 0.21%
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Damage Estimate ($ billion)
Percent change
$2343
$2242
$1.0
$2708
$2703
$5.5
0.05% 0.20%
$1916
$1905
$10.6
0.55%
$3341
$3331
$10.7
0.32%
Notes:
1 . Results are expressed in year 2006 dollars.
In general, FASOM forecasts a relative shift towards forestry and away from agriculture
under the without-CAAA scenario, indicating that the net impacts of the ozone effects on
forests and agriculture would make forestry relatively more profitable than in the baseline
compared with agriculture, resulting in a shift in land use. The model forecasts a sizable
increase in cropland in the without-CAAA scenario, however there is an even greater
decline in pasture as the returns to crop production rise relative to livestock production
with higher crop prices.
As noted above, the model suggests that the damages attributed to higher ozone
concentrations indicate that producers gain in many cases, while consumers are always
substantially worse off with the ozone impacts reducing productivity. The reason that
producers often are better off is that most forest and agricultural products have relatively
inelastic demands, which means that a general decline in productivity will tend to
increase prices by more than the reduction in quantity, increasing revenue and often
profits as well. In general, FASOM attributes large price increases in response to the
reductions in productivity for these inelastic products, and production declines in the
without-CAAA scenario for most agricultural commodities, with larger declines in general
for those products experiencing larger ozone impacts, and also sizable reductions in
exports.
6-24
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FASOM also is capable of modeling land-use changes in response to the higher ozone
concentrations in the without-CAAA scenario. The model indicates changes in major land
use categories at the national level over time under the ozone impacts scenario, which is
leading to a net increase in forest of about 6.1 million acres by the 2020 model period and
an increase in cropland of 7.6 million acres by 2020 in response to the productivity
declines. At the same time, the model indicates that cropland pasture (high-quality land
that is suitable for cropland but is being used as pasture) and pasture (lower-quality land
that is not suitable for growing crops without improvement) decline by a total of 12.7
million acres and Conservation Reserve Program (CRP) land decreases by about 1
million acres. The crop experiencing the largest reduction in acreage is soybeans, while
there is an increase in wheat acreage and a number of smaller shifts between alternative
crops.
VISIBILITY
Air pollution impairs visibility in both residential and recreational settings, and an
individual's willingness to pay (WTP) to avoid reductions in visibility differs in these two
settings. Benefits of residential visibility 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
recreational visitors. For the purposes of this analysis, recreational visibility
improvements are defined as those that occur specifically in federal Class I areas, and
residential visibility improvements are those that occur within the boundaries of Census
Metropolitan Statistical Areas (MSAs).
We calculate household WTP for improvements in both residential and recreational
visibility. We base our calculations on simulations of future visibility conditions at the
36-km grid-cell level, as estimated by EPA's Community Multiscale Air Quality
(CMAQ) model. The relationship between a household's WTP and changes in visibility
is derived from a number of contingent valuation (CV) studies published in the peer-
reviewed economics literature. The approach we apply to estimate the benefit of
improvements in recreational visibility is consistent with methods EPA has used in
analyses conducted since EPA's First Prospective analysis was completed. In particular,
this chapter relies heavily on research completed for the PM NAAQS RIA (U.S. EPA,
2006) for the recreational visibility analysis. Our estimate of the benefit of residential
visibility is consistent with methods applied in past analyses as well, but in previous
reviews the Council had expressed concerns about residential visibility estimates based
on WTP estimates from the McClelland et al. (1991) study. As a result, our estimates in
this chapter rely on a new "benefits transfer" estimate of WTP derived from other
published sources of residential visibility WTP.
6-25
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
According to the CMAQ simulations, the CAAA has had and will continue to have a
substantial effect on visibility in both residential and recreational settings. The visibility
data used in this analysis is annual mean visibility data, by county, measured in
deciviews.87 Figure 6-6 depicts the change in visibility (measured in deciviews) over the
30-year time frame, from 1990 to 2020, along the with-CAAA scenario. This map shows
that, overall, changes in visibility due to the CAAA are greater in the eastern U.S. than
the western U.S. Additionally, the largest changes in visibility occur in the Midwestern
states. The county level data presented here are the basis for the residential visibility
improvements we present below.
Figure 6-7 summarizes trends in visibility at the 13 most-visited U.S. National Parks.
Visibility estimates (measured in deciviews) are provided for each of the seven core
CAAA scenarios. Note that deciviews are inversely related to visual range, such that a
decrease in deciviews implies an increase in visual range (i.e., improved visibility).
Conversely, an increase in deciviews implies a decrease in visual range (i.e., decreased
visibility). The figure illustrates that the CAAA greatly affects visibility at National
Parks - over the 1990 to 2020 period, visibility markedly improves with the CAAA, and
markedly declines without the CAAA. Particularly large differences in visibility between
the with-CAAA and without-CAAA scenarios are seen at Great Smoky Mountains National
Park, which is the most visited park in the U.S. Note that six of the 13 parks listed in
Figure 6-7 are not included in the primary monetized recreational visibility estimates
presented later in this chapter, because they were not included in the park regions studied
in the underlying economic valuation study. The six parks not included are in the
northern part of the country, and include Mount Rainier, Olympic, Glacier, Yellowstone,
Grand Teton, and Acadia.
87 The data was aggregated from the 36-km grid-cell level to the county level using the BenMAP version 3.0.15 "Air Quality
Grid Aggregation" algorithm. The fourth quarter data is corrected for a missing day (the CMAQ runs modeled 364 days,
omitting December 31) by reweighting the mean to account for the missing day.
6-26
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-6. ESTIMATED CHANGE IN VISIBILITY FOR WITH-CAAA SCENARIO, 1990 TO 2020
Without CAAA
WithCAAA
Visibility in Deciviews, 2020
0 3 6 9 15 18 24 27 30
6-27
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-7. VISIBILITY TRENDS FOR THE 13 MOST-VISITED U.S. NATIONAL PARKS
mo m =ID
Only one existing study provides defensible monetary estimates of the value of
recreational visibility (Chestnut and Rowe, 1990b; 1990c). Although the Chestnut and
Rowe study is unpublished, it was originally developed as part of the National Acid
Precipitation Assessment Program (NAPAP) and, therefore, has been subject to peer-
review as part of that program. The Chestnut and Rowe study measures the demand for
visibility in Class I areas managed by the National Park Service (NPS) in three broad
regions of the country: California, the Southwest, and the Southeast. Respondents in five
states were asked about their WTP to protect national parks or NPS-managed wilderness
areas within a particular region. The survey used photographs reflecting different
visibility levels in the specified recreational areas. The visibility levels in these
photographs were later converted to deciviews for the current analysis. The three regions
assessed in the study cover 86 of the 156 Class I areas in the United States. Given that
national parks and wilderness areas exhibit unique characteristics, it is not clear whether
the WTP estimate obtained from the Chestnut and Rowe study can be transferred to other
national parks and wilderness areas, without introducing additional uncertainty. As a
6-28
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
result, for the primary estimate, we value only those recreational benefits in the areas that
were directly analyzed in the original Chestnut and Rowe study.
In the First Prospective analysis, we omitted the results of the benefits estimate for
residential visibility from the primary benefits estimate due to technical concerns about
the methodology of the study upon which our original calculations were based
(McClelland et al., 1991).88 There exists a wide range of published, peer-reviewed
literature, however, that supports a non-zero value for residential visibility. As a result,
we have revised our methodology for valuing residential visibility, and now include these
benefits in our overall primary visibility benefits estimate.
For valuing residential visibility improvements, we rely upon a benefits transfer approach
that draws upon information from the published Brookshire (1979), Loehman (1984) and
Tolley (1986) studies. Each of the studies used provides estimates of household WTP to
improve visibility conditions from a status quo visual range to an improved visual range.
While uncertainty exists regarding the precision of these older, stated-preference
residential valuation studies, we believe their results support the argument that
individuals have a non-zero value for residential visibility improvements. The implied
annual per-household WTP estimates from these study, for a hypothetical 10-percent
improvement, ranges from $14 to $145, with a mean of $69 and median of $53. It is not
surprising that such a range of values exists, as the areas of the country covered feature
different landscapes and vistas, populations and prevailing visibility conditions.
Fortunately, the three recommended studies provide primary visibility values for a variety
of cities throughout the United States: Atlanta, Boston, Chicago, Denver, Los Angeles,
Mobile, San Francisco, and Washington B.C. We assign each of the 359 MSAs in the
contiguous U.S. a value based on geographic proximity to one of the eight study cities,
with two exceptions: 1) We apply the Loehman et al. (1984) value only to the six San
Francisco Bay area MSAs, because the study is unique among the three in the manner in
which visibility changes were described to respondents (i.e., a distribution of days versus
average conditions), and 2) Values associated with Denver are not assigned on the basis
of proximity but are instead assigned only to MSAs which meet an elevation range
threshold of 1500 meters within the MSA, because one would expect that residents of
Denver, with a dramatic view of the Rocky Mountains that is rarely obstructed by trees,
would have a greater interest in protecting visibility than a city without a dramatic skyline
or nearby mountains.89
88 Council review of early drafts of the First Prospective analysis noted that the McClelland et al. (1991) study 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. As a result of these concerns, a prior Council recommended that residential visibility be omitted from the overall
primary benefits estimate in the First Prospective.
89 The geographic proximity assignment is preserved for the Los Angeles and Riverside MSAs although these MSAs meet the
elevation range threshold of 1500 meters. The assignment is preserved because Los Angeles is one of the study cities and
6-29
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
The primary estimate of benefits of recreational and residential visibility improvements is
provided in Table 6-7. The primary estimate for recreational visibility only includes
benefits in the original study regions (i.e., California, the Southwest, and the Southeast).
The primary estimate for residential visibility includes benefits in all MSAs. In general,
benefits to visibility increase over time as visibility improves due to the CAAA. Benefits
to residential visibility are approximately three times as large as benefits to recreational
visibility.
TABLE 6-7. PRIMARY ESTIMATE OF BENEFITS TO VISIBILITY (BILLION 2006$)
Recreational Benefits
Residential Benefits
Total Benefits
2000 BENEFITS
$3.3
$11
$14
2010 BENEFITS
$8.6
$25
$34
2020 BENEFITS
$19
$48
$67
In Figures 6-8a and 6-8b below, we map the primary 2020 estimate of benefits of
recreational and residential visibility improvement by state. Overall, the spatial pattern of
benefits is similar for recreational and residential visibility. Recreational visibility
benefits are driven by population and park location, within the original study regions of
Chestnut and Rowe (1990a). These regions are California, the Southwest (Arizona,
Nevada, Utah, Colorado, and New Mexico), and the Southeast (Delaware, Maryland,
West Virginia, Virginia, Kentucky, Tennessee, North Carolina, South Carolina, Georgia,
Alabama, Florida, and Mississippi). Households express WTP for visibility
improvements in Class I areas located in-region as well as out-of-region. For this reason,
there may be high recreational benefits in a state that has no Class I areas. Although
household WTP is higher for in-region parks, this effect seems to be dominated by the
effect of population. For example, less populated states such as New Mexico and Utah
with Class I areas have low benefits to recreational visibility, while more populated states
such as New York without Class I areas have high recreational visibility benefits.
also because Los Angeles has a particular set of location-specific characteristics that set it apart from Denver. As a
conservative measure, Riverside MSA is also assigned to the Los Angeles study area because a significant portion of Riverside
County itself is located in the South Coast Air Quality Management District, and therefore is considered by at least some
measures to be part of the same regulated airshed as Los Angeles.
6-30
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-8A. PRIMARY ESTIMATE OF RECREATIONAL VISIBILITY BENEFITS IN 2020 (BILLION
2006$)
;
;'
HlfbMy BWMfltt 4»M1
™»— »* WifHii MNU-1-iSi
•Hcae-Vltt W31.B3B •IIIH-UIS
• 10«32-M» _ BfllC-M-tl- HI Etc? -CM
•;•*>;••
IGURE 6-8B. PRIMARY ESTIMATE OF RESIDENTIAL VISIBILITY BENEFITS IN 2020 (BILLION
2006$)
6-31
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Residential visibility benefits are driven by population and visibility improvements.
Overall, benefits are greater in the East. This is due in part to greater population levels as
well as greater visibility improvements. Benefits are also very high in California due to
the state's large population and visibility improvements, especially in and around Los
Angeles and San Francisco. Residential visibility is also dependent upon the WTP value
applied. Much of the West uses the WTP value for Denver, which is highest WTP value
being widely applied. Yet, the West still has lower overall benefits to residential
visibility.90 This impact shows that the effect of population and visibility improvement
dominates the effect of the WTP value applied.
MATERIALS DAMAGE
Since the mid-19th century air pollution has been suspected of accelerating the
degradation of natural and man-made materials that are exposed to the outdoor
environment. Concern over the effect of pollutants on materials has mainly been directed
towards the economic consequences of damage to materials used in construction, but
aesthetic damage to historic buildings and monuments is also a concern. Wet and dry
acidic deposition, alone or combined with other air pollutants, contribute to the increased
rate of materials damage. Acidic deposition has been shown to have an effect on
materials including zinc/galvanized steel and other metal, carbonate stone (as monuments
and building facings), and surface coatings (paints) (NAPAP, 1991).
Metal structures are usually coated by alkaline corrosion product layers and thus are
subject to increased corrosion by acidic deposition. In addition, research has
demonstrated that iron, copper, and aluminum based products are subject to increased
corrosion due to pollution, in particular SO2 (NAPAP, 1991), that acidic deposition
accelerates the rate of erosion of carbonate stone (marble and limestone), and that acidic
deposition has numerous negative effects on painted wood and, in general, increases the
weathering rate. This analysis focuses on quantifying the impact of sulfur dioxide
deposition on exterior building and infrastructural materials including carbonate stone,
galvanized steel, carbon steel, and painted wood, as outlined Table 6-8 below.
0 The WTP value for San Francisco is higher than Denver, but the San Francisco value is not applied to other MSA's.
6-32
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-8.
MATERIALS DAMAGE EFFECTS
POLLUTANT
Sulfur oxides
Hydrogen ion and
nitrogen oxides
Carbon dioxide
Formaldehyde
Particulate matter
Ozone
QUANTIFIED EFFECTS-DAMAGE TO:
Infrastructural materials -
galvanized and painted carbon steel
Commercial buildings - carbonate
stone, metal, and painted wood
surfaces
Residential buildings - carbonate
stone, metal, and painted wood
surfaces
UNQUANTIFIED EFFECTS"-
DAMAGE TO:
Monuments - carbonate stone
and metal
Structural aesthetics
Automotive finishes - painted
metal
Infrastructural materials -
galvanized and painted carbon
steel
Zinc-based metal products, such
as galvanized steel
Commercial and residential
buildings - carbonate stone,
metal, and wood surfaces
Monuments - carbonate stone
and metal
Structural aesthetics
Automotive finishes - painted
metal
Zinc-based metal products, such
as galvanized steel
Zinc-based metal products, such
as galvanized steel
Household cleanliness (i.e.,
household soiling)
Rubber products (e.g., tires)
a The categorization of unqualified effects is not exhaustive.
This analysis applies the Air Pollution Emissions Experiments and Policy (APEEP)
analysis model, described in Muller and Mendelsohn (2007, 2009), to link SO2 emissions
to ambient SO2 levels. Using emission inputs, the air quality model in APEEP forecasts
seasonal and annual average county concentrations for SO2, amongst other pollutants.91
As reported in Muller and Mendelsohn (2007) and detailed in the supporting online
material for that publication, APEEP's air quality modeling has been statistically tested
and calibrated against the predictions generated by the Community Multi-scale Air
Quality Model (CMAQ), using 1996 emissions data and a CMAQ run for 1996
The Project Team considered using the CMAQ S02 air quality results directly, but the decision to implement the materials
damage approach described here came too late to cost-effectively recover the relevant ambient S02 estimates from the
original CMAQ runs. The overall magnitude of the monetizable materials damage benefits is such a small part of the overall
benefits of the CAAA that the impact of using APEEP's air quality tool rather than CMAQ on the overall benefits estimates is
likely to be very small.
6-33
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
conditions. Muller and Mendelsohn (2007) also report comparisons of APEEP's results
with available monitor data for this period. The results for the SO2 air quality component
used in these materials damage calculations appear to suggest good agreement for APEEP
for concentrations near the mean, but APEEP appears to overpredict SO2 concentrations
for high-end concentrations. Overall, however, it is important to note that APEEP is
designed to be a fast-running alternative to CMAQ for use in an integrated assessment
model - the air quality component of APEEP is a statistical representation of relations
that are accomplished in a far more sophisticated manner in CMAQ.
The remaining general steps in the process of estimating materials damage effects are as
follows:
• Develop a national inventory of sensitive materials. A key piece of
information needed to apply the appropriate materials damage concentration-
response functions is the existing materials inventories. This analysis estimates
the inventory of four exterior building and infrastructural materials in each
county in the lower 48 states, including carbonate stone, galvanized steel, carbon
steel, and painted wood surfaces.
• Derive concentration-response functions that relate material mass loss to
ambient SO2. Dose-response functions for man-made materials damages are
obtained from two sources; the NAPAP studies (Atteraas, Haagenrud, 1982;
Haynie, 1986) and from the International Cooperative Programme on Effects on
Materials (ICP, 1998).
• Estimate the value of lost materials. Materials damage is valued as the cost of
future materials maintenance activities. The accelerated rate of materials decay
due to pollution exposure increases the frequency of regularly scheduled future
maintenance activities. The change in the present value of the maintenance
schedules extending into the future constitutes the monetary impact of an
emission change on materials damage.
Table 6-9 summarizes the benefits of reduced materials damage attributed to CAAA
programs in 2000, 2010, and 2020. Benefits are given by EPA region. Although the total
benefits are relatively small compared to other categories of effect, the benefits of CAAA
programs to materials damage increase over time as we would expect. The spatial
distribution of the benefits is primarily owing to the distribution of the materials
inventory and SO2 exposure. The effect of SO2 exposure is a more important driver of
results than the inventory. For example, the benefits in Region 5 are approximately twice
as large as those in any other EPA region. This is due to the significant decrease in SO2
exposure associated with the CAAA in this region.
6-34
-------
TABLE 6-9.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
BENEFITS OF REDUCED MATERIALS DAMAGE DUE TO CAAA PROGRAMS
EPA REGION
1:CT, ME, MA, NH, Rl, VT
2: NY, NY
3: DE, DC, MD, PA, VA, WV
4: AL, FL, GA, KY, MS, NC, SC, TN
5: IL, IN, Ml, MN, OH, Wl
6: AR, LA, NM, OK, TX
7: IA, KS, MO, NE
8: CO, MT, ND, SD, UT, WY
9: AZ, CA, NV
10: ID, OR, WA
Total
VALUATION (THOUSAND 2006$)
2000
$720
$9,000
$9,400
$8,400
$26,000
$2,200
$2,000
$400
-$100
$340
$58,000
2010
$2,100
$10,000
$19,000
$16,000
$38,000
$4,000
$1,600
$570
$490
$510
$93,000
2020
$2,100
$12,000
$23,000
$21,000
$38,000
$7,300
$1,600
$730
$640
$560
$110,000
Notes: Results are rounded to two significant figures. Totals may not sum due to rounding.
ADIRONDACK CASE STUDY RESULTS
The Project Team was encouraged to consider case study analysis of a set of ecological
effects for which national analyses might not be feasible, owing to lack of available data
or methods. EPA chose to conduct a case study in the Adirondack region of New York
State, focusing on two ecological service flows that provide benefits in terms of both
ecosystem health and economic terms: (1) acidification of surface waters and (2) reduced
yields of commercial timber. The Adirondack region of New York may exhibit the most
severe ecological impacts from acidic deposition of any region in North America - acid
deposition is the main cause of both of the effects we studied.92 Adirondack Park is a
State Park comprising 5,821,183 acres of State and privately owned land in upstate New
York and is nearly a 100 by 100 mile box of land, intersecting fourteen counties. The
Park was created in 1892 through an amendment to the State constitution, with the
purpose of forest and natural resource conservation. Federal programs addressing air
pollution have been particularly beneficial to the region as, due to its location downwind
of the highly industrialized Ohio River Valley, most of the acid deposition in the region
originates from out of state. In addition to its status as a region of particular sensitivity to
lake acidification and with some existing research on the effects of air pollutants on forest
32 Driscoll, Charles T. et al. May 2003. Chemical Response of Lakes in the Adirondack Region of New York to Declines in
Acidic Deposition. Environmental Science and Technology 37(10): 2036-2042.
6-35
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
growth, the Adirondack Region was selected as a setting for this case study due to the
existence of a regional economic random utility model describing recreational fishing
behavior.
Lake Acidification in the Adirondacks
Surface waters, such as lakes and streams, may be the most susceptible systems to acidic
deposition as they collect acidic precipitation not only from direct deposition on their
surfaces but also in the form of runoff from their entire watershed. Acid accumulates in
surface waters via three main pathways:
• precipitation, or wet deposition, in which pollutants are dissolved in rain or snow;
• dry deposition, or direct deposition of gases and particles on surfaces; and
• cloud-water deposition, involving material dissolved in cloud droplets and
deposited on vegetation.93
As acids accumulate, ecosystems gradually lose the ability to buffer them, resulting in
changes to ecosystem structure and function. Acidification of the surface water affects
the trophic structure of water contributing to declines in the abundance of zooplankton,
macroinvertebrates, and fish.94
The ecological service flow affected by lake acidification that is most amenable to
economic analysis is recreational fishing. Extensive research exists focused on both the
effects of lake acidification on fisheries and on individuals' willingness to pay to avoid
reductions in the quality or quantity of recreational fishing opportunities. This analysis
employs the following general steps to quantify the benefits of reduced lake acidification
on recreational fishing in the Adirondacks. A conceptual model depicting the analytic
steps in terms of inputs, outputs, and ecological and economic models is provided in
Figure 6-9.
• Forecast lake acidification levels consistent with the with-CAAA and without-
CAAA scenarios. EPA generated estimates of acidic deposition at a 36-
kilometer grid cell level across the Adirondack region using the CMAQ model.
We then implemented an ecological model, the Model of Acidification of
Groundwater in Catchments (MAGIC), to simulate the transport of the acidic
deposition through the hydrological and terrestrial ecosystems and forecast
acidification levels in a subset of Adirondack lakes.
• Extrapolate results of the ecological model within the Adirondacks region.
We developed a random effects model to explain the relationship between
acidification of lakes and their specific site characteristics.
93 The U.S. National Acid Precipitation Assessment Program. 1991. Integrated Assessment Report. The NAPAP Office of the
Director, Washington, DC.
94 Driscoll, Charles T. et. al. March 2001. Acidic Deposition in the Northeastern United States: Sources and Inputs, Ecosystem
Effects, and Management Strategies. BioScience 51(3): 180-198.
6-36
-------
FIGURE 6-9.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
• Apply ANC thresholds to classify lakes as either "fishable" or "impaired".
Fishable lakes are those for which water quality is not deteriorated to an extent
which limits recreational fishing. Impaired lakes' water quality is deteriorated so
as to reduce fish populations and preclude recreational fishing. Lakes are defined
as either fishable or impaired based on identified ANC thresholds. As
uncertainty exists regarding the ANC threshold at which effects are experienced,
this analysis considers three separate thresholds below which lakes are
considered impaired.
• Apply an economic random utility model (RUM) to quantify economic
benefits of the CAAA in terms of recreational fishing in the Adirondack
region. We employ a RUM that was developed to account for fishing site
choices made by recreational fishers based on attributes of sites specifically in the
Adirondack region. The difference in economic welfare values between the
value of fishable (i.e., not impaired) lakes in the with-CAAA scenario and the
without-CAAA scenario represents the benefits to recreational fishing in the
Adirondack region associated with the CAAA.
CONCEPTUAL APPROACH TO ESTIMATING THE ECONOMIC BENEFITS OF REDUCED
ACIDIFICATION ON ADIRONDACK LAKES
Deposition Scenario 1:
Full implementation
of 1990 CAAA
De position Scena rio 2:
No implementation
of 1990 CAAA
L
Random
Utility
Model
Listof Adirondack
Lakes that are
Impaired according
to Scenarbs 1 and2.
Scenario 1:
Economic
Impact
of lake
acidification
Scenarb 2:
Eco no mic
Impact
of lake
acdifbatbn
I
ANC Thr
20, 50, a
Jieg
Scenario 2 -Scenario 1 =
Economic Benefit
of CAAA on
Recreational
Fishing
Table 6-10 summarizes the results of this analysis. Present value cumulative benefits are
provided for 2000, 2010, and 2020, assuming a five percent discount rate. Single year
undiscounted benefits are also given for each year. Undiscounted single year benefits
6-37
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Table 6-10
increase over time but the benefits do not follow any particular trend across alternative
threshold assumptions. It should be noted that benefits in each year and under each
threshold assumption reflect a different subset of lakes. Therefore, benefits are not
expected to follow any particular trend across years or threshold assumptions.
SUMMARY OF ANNUAL AND CUMULATIVE ESTIMATED BENEFITS TO RECREATIONAL
FISHING IN THE ADIRONDACK REGION (MILLION 2006$)
YEAR
2000
2010
2020
ANC THRESHOLD
ASSUMPTION FOR DEFINING
"FISHABLE" LAKES
20
50
100
20
50
100
20
50
100
ADIRONDACK REGION
SINGLE YEAR UNDISCOUNTED
$7
$7
$5
$8
$8
$6
$9
$8
$6
CUMULATIVE FIVE PERCENT
DISCOUNT RATE
$62
$57
$44
$143
$132
$101
$197
$182
$136
Note:
1 ) Cumulative benefits in year 2000 are the cumulative benefits to recreational fishing of
implementing the CAAA from 1990 to 2000. Similarly, cumulative benefits in 2010 are
cumulative from 1990 to 2010 and cumulative benefits in 2020 are cumulative from 1990 to
2020. The single year undiscounted benefits are the benefits to recreation fishing of
implementing CAAA in that year (2000, 2010, or 2020).
2) Benefits in this case study are evaluated from 1990 (the year of the passage of the
CAAA) to 2050 (the forecast horizon for the lake ANC levels with and without the CAAA).
The benefits in this table are presented for years 2000, 2010, and 2020, however, to be
consistent with the benefits as calculated in the broader cost-benefit analysis of the CAAA.
Commercial Timber in the Adirondacks
Reductions in NOX and SOX emissions due to the implementation of the CAAA are also
believed to reduce forest soil acidity. Reductions in soil acidity have been shown by
scientists to increase tree growth and improve overall forest health. Such changes in
forest growth and health would have a positive effect on the timber industry within
Adirondack Park, potentially increasing the frequency and/or the volume of timber
harvests.
Quantifying the magnitude of these benefits requires a function to translate varying levels
of soil acidity into corresponding tree growth productivity. Unfortunately, species-
specific dose-response functions relating soil acidity levels with changes in tree growth in
Adirondack Park are not available. Our analysis instead characterized the existing timber
industry in Adirondack Park in terms of the types of tree species present, wood products
harvested, extent of timber harvest activities, and the overall value of timber harvests
6-38
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
within the Park. We then estimated changes in percent base saturation (a measure of soil
acidity) due to the implementation of the CAAA across the Park from 1990 to 2050,
focusing on soil acidity differences in areas subject to commercial timber activity.
Specifically, changes in percent base saturation levels in timber harvest areas were
mapped in relation to potential changes in the growth and health of tree species present in
these areas and the likely effects of altered tree growth and health on timber harvest rates
and volumes. In addition, we provide some perspective on the potential order of
magnitude of benefits of the CAAA on the timber industry in the Adirondacks,
summarizing existing, relevant research.
We used estimates of soil percent base saturation levels for 1990, 2000, 2010, 2020, and
2050 with and without the CAAA to characterize the effect on Adirondack forests.95
Percent base saturation is the proportion of cation exchange sites (exchange sites are
areas on soil particles where ions may be adsorbed) occupied by basic cations (Ca2+,
Mg2+, K+, and Na+). These basic cations buffer the soil by inhibiting the adsorption of H4"
ions. Thus, percent base saturation is a measure of the soil's buffering capacity. High
percent base saturation levels indicate large buffering capacity and low soil acidity levels,
while low percent base saturation levels indicate the converse. Percent base saturation
point estimates were generated using the same Model of Acidification of Groundwater in
Catchments (MAGIC) as used in the lake acidification analysis described above.
Figure 6-10 presents differences in percent base saturation levels with and without the
CAAA specifically within the timber harvest areas of the Park by year. There is a clear
temporal trend in the difference in percent base saturation levels with and without the
CAAA. Specifically, differences between percent base saturation levels with the CAAA
as compared to without the CAAA increase in each year in the analysis. However, there
is little spatial variability in percent base saturation differences within individual years.
The lack of spatial variability becomes more pronounced as time goes on, so that by 2050
the difference in percent base saturation is between 2.07 and 6.26 percent in almost all
forested resource management areas in the Park. The lack of spatial variability makes
sense given the relatively small geographic scope considered in this analysis. The minor
spatial variation in percent base saturation differences exhibited in 2000 and 2010 is most
likely related to microhabitat factors (i.e., different soil types and differing precipitation
levels).
95 While the timeframe for this Second Prospective analysis of the CAAA is through 2020, this case study reports benefits
through 2050 as we expect that reductions in emissions that occur in 2020 will continue to provide benefits to recreational
fishing through this time frame.
6-39
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 6-10. DIFFERENCES IN PERCENT BASE SATURATION VALUES WITH AND WITHOUT THE
CAAA IN FORESTED RESOURCE MANAGEMENT AREAS IN ADIRONDACK PARK 96'97
0 9 18
Map Pr.:(e.Mfcl:UTMZoi*
A
'•" Difference in Percent Base Saturation
D.OOOO - 0.0137 • 0.5712 - 0.8220
0.0138-0.0167 •0.8221 -0.9780
00168-0.3977 MO9781 -1.8321
0.3978-0.4567 •18322-2.0670
• 0.4568-0.5711 •2.0671 -6.2625
"Sources: 1.) U.S. Forest Setvice; 2.) N.Y. Dept. of Environmental Conservation; 3.) Environmental Protection Agency;
4.) Environmental System'; h-• .-'.j!•: r !:'s1itute. Inc.
96 The differences between percent base saturation levels with the CAAA and without the CAAA are presented rather than
absolute percent base saturation levels for each scenario to highlight the changes in percent base saturation attributable to
the implementation of the CAAA.
97 The ten ranges of difference in percent base saturation values presented in Exhibit 5-8 are equal to the 10th, 20th,..., and
100th percentiles for the combined distribution of difference in percent base saturation values across all years in the
analysis (2000, 2010, 2020, and 2050).
6-40
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-11.
Also of importance to this analysis is the magnitude of the increase in percent base
saturation levels in relation to specific forest types within resource management areas.
We focused on six forest types (sugar maple/beech/yellow birch, red maple/upland,
spruce/fir, eastern hemlock, eastern white pine, and paper birch) that are prevalent in the
Park relative to other forest types and contain tree species of commercial value. Table 6-
11 presents the area-weighted mean increase in percent base saturation levels in these
forest types per year. Of the forest types of interest, the paper birch forest type
experiences the greatest increase in percent base saturation due to the CAAA, followed
by the eastern hemlock and the sugar maple/beech/yellow birch forest types.
AREA-WEIGHTED MEAN DIFFERENCES IN PERCENT BASE SATURATION VALUES WITH
AND WITHOUT THE CAAA IN FOREST TYPES OF INTEREST
FOREST TYPE
Sugar Maple/Beech/Yellow Birch
Red Maple /Upland
Spruce/Fir
Eastern Hemlock
Eastern White Pine
Paper Birch
Other Forest Types
AREA-WEIGHTED DIFFERENCE IN PERCENT BASE
SATURATION
2000
0.023
0.025
0.028
0.028
0.018
0.018
0.015
2010
0.414
0.377
0.361
0.413
0.419
0.457
0.429
2020
0.820
0.758
0.736
0.827
0.814
0.891
0.829
2050
1.899
1.755
1.702
1.908
1.882
2.069
1.918
The area-weighted increase in percent base saturation levels in sugar maple/beech/yellow
birch forests is in line with increases in percent base saturation levels in other forest types
in Adirondack Park. This is an important point given the prevalence of sugar maple in
this forest type, and the fact that sugar maple is an economically important tree species in
the Park. Although dose-response functions, which would allow for estimates of growth
increases in sugar maples due to increased base saturation levels, do not exist, several
studies have estimated changes in sugar maple growth due to increases in soil acidity
stemming from elevated nitrogen and/or sulfur deposition.98 Changes in harvest volumes
comparable to those seen in those existing studies might lead to annual wood harvest
For example, Duchesne et al. (2002) found that sugar maple basal area growth rates were reduced by 17 percent, on
average, in forest stands exhibiting decreasing basal area growth rates over time (declining stands) compared to sugar
maple basal area growth rates in stands exhibiting increasing basal area growth rates over time (healthy stands). In
addition Mclaughlin (1998) found that the health of hardwood stands on shallow, poorly buffered soils similar to those
found in Adirondack Park declined during the 1990s due to decreasing pH and base saturation levels and increased
aluminum ion concentrations.
6-41
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
benefits of roughly $1 million to $1.5 million annually, based on the total stumpage
values for sugar maple pulpwood/chip wood we estimate for the region." Whether sugar
maple growth rate changes would mirror those reported in either of these studies,
however, is uncertain due to the lack of an established functional relationship.
Nonetheless, we expect that all tree species in the Park would benefit, in terms of
increased stand growth and vigor, from increased percent base saturation levels. In some
cases, increases in growth may allow for both more frequent and larger timber harvests
(i.e., more frequent timber harvests removing larger volumes of wood). Improved forest
health may also provide the added benefit of increasing the resiliency of forest stands and
limiting damage caused by disturbance events.
UNCERTAINTY IN ECOLOGICAL AND OTHER WELFARE BENEFITS
As noted above, limitations in the available methods and data mean that 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 suggests that great care must be
taken to consider those ecosystem impacts that are not quantified here. Significant future
analytical work and basic ecological and economic research is needed 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 6-12 below.
In general, our analysis focuses on more acute and readily observable effects. Chronic
ecological effects of air pollutants, on the other hand, may be poorly understood, difficult
to observe, or difficult to discern from other influences on dynamic ecosystems.
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.
99
We estimated stumpage values of commonly harvested species in the Adirondack Region by applying average stumpage
values to the pulpwood and wood chip and roundwood log harvest volume estimates. The average stumpage value for
pulpwood and wood chips is estimated to be $3 per ton; while, the average stumpage value for roundwood logs is estimated
to be $150 per thousand board-feet (MBF). Using these estimates, the annual harvest value of pulpwood and wood chips is
estimated to be approximately $5.4 million, and the annual harvest value of roundwood logs is estimated to be $15 million.
6-42
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 6-12. KEY UNCERTAINTIES ASSOCIATED WITH ECOLOGICAL EFFECTS ESTIMATION
POTENTIAL SOURCE OF ERROR
DIRECTION OF POTENTIAL BIAS
FOR NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO KEY
UNCERTAINTIES ON NET BENEFITS
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. Examples of categories of
potential ecological effects for which
benefits are not quantified include:
reduced eutrophication of estuaries,
reduced acidification of soils, reduced
bioaccumulation of mercury and
dioxins in the food chain.
Underestimate
Potentially major. The extent of
unqualified and unmonetized
benefits is largely unknown, but the
available evidence suggests the
impact of air pollutants on ecological
systems may be widespread and
significant.
Incomplete geographic scope of
recreational fishing benefits associated
with reduced lake acidification analysis
due to case study approach.
Underestimate
Probably minor. As a case study
limited to the Adirondack region of
New York State, the estimated
benefits to recreational fishing reflect
only a portion of the overall benefits
of reduced acidification on this
service flow, but based on the
magnitude of effects in the
Adirondacks the national estimate is
nonetheless likely be less than five
percent of total benefits.
Incomplete assessment of long-term
bioaccumulative and persistent effects
of air pollutants.
Underestimate
Potentially major. Little is currently
known about the longer-term effects
associated with the accumulation of
toxins in ecosystems. 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.
Omission of the effects of nitrogen
deposition as a nutrient with beneficial
effects.
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.
6-43
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
POTENTIAL SOURCE OF ERROR
DIRECTION OF POTENTIAL BIAS
FOR NET BENEFITS
LIKELY SIGNIFICANCE RELATIVE TO KEY
UNCERTAINTIES ON NET BENEFITS
ESTIMATE*
Use of CMAQ model to estimate air
pollutant deposition levels.
Unable to determine. As part
of a performance evaluation
of CMAQ, EPA compared
model predictions for some
forms of deposition relevant
to this analysis (wet SO2, NOx
and ammonium) to observed
deposition data).** The
evaluation indicated that
CMAQ overpredicted some
forms of deposition and
underpredicted others. The
relative accuracy of the
model's predictions varied
seasonally and geographically.
Probably minor.
The Adirondack lake acidification
analysis uses deposition estimates as
inputs, but they are calibrated to
lake-level monitoring data, and the
monetized benefits estimates for that
component are a small part of the
overall net benefits. We also use the
CMAQ deposition estimates to
generate maps that highlight the
relative distribution of deposition for
various air pollutants across the U.S.
With respect to net impacts, the
extent to which the forms of
deposition and geographic areas that
are overpredicted in the model are
offset by those that are
underpredicted is unknown.
* The classification of each potential source of error reflects the best judgment 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."
** See U.S. EPA, Office of Air Quality Planning and Standards, Emissions Analysis and Monitoring Division, Air
Quality Modeling Group. CMAQ Model Performance Evaluation Report for 2001: Updated March 2005. CAIR Docket
OAR-2005-0053-2149.
6-44
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 7 - COMPARISON OF BENEFITS AND COSTS
In this chapter we present our
summary of the primary
estimates of monetized benefits
of the CAAA from 1990 to
2020, compare the benefits
estimates with the corresponding
costs, and explore some of the
major sources of uncertainty in
the benefits estimates, including
a summary of outcomes using
alternative assumptions from
those employed in the primary
analysis.
The overall conclusion of our
analysis is that the benefits of the
CAAA substantially 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 2020
period could be less than its costs. The central benefits estimate exceeds costs by a factor
of more than 30 to one, whether we are looking at annual or present value measures.
By our measures, the programs associated with the 1990 Clean Air Act Amendments
have been, and will likely continue to be, a very good investment.
AirQ
Hea
Econ
Scenario Development
|
Sector Modeling
I
1
Emissions
1 r
uality Modeling
i '
Ith Welfare
i *
omic Valuation
r
Direct Cost
|
J
Benefit-Cost Comparison
AGGREGATING BENEFIT ESTIMATES
Our primary estimates of the monetized economic benefits for the 1990 to 2020 period
derive from two types of analyses: (1) the analysis of changes in human health effects
associated with reduced exposures to criteria pollutants and the valuation of these
changes, summarized and described in Chapter 5; and (2) the analysis of monetized
ecological and other welfare benefits (e.g., visibility), described in Chapter 6.100 We
measure the benefits and present the results from these analyses in slightly different ways,
1 Note that the direct costs were aggregated in Chapter 3.
7-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
in part because they derive from different tools. The main differences have to do with the
manner in which we conduct uncertainty analyses, as outlined below.
Although there are some differences in these two types of benefits analysis, in both cases
we generate annual estimates of benefits that result from a single set of emissions and air
quality modeling scenarios for the three target years of the study: 2000, 2010, and 2020.
The consistent use of scenarios across all the benefit and cost analyses allows us to
aggregate and directly compare monetized benefits estimates to the estimates of costs
incurred in the target years. In some cases, we need to apply a discount rate to compare
benefits to costs; for example, we model the effect of particulate matter on premature
mortality to occur over a period of twenty years from the time of exposure, even though
the costs to achieve that benefit are incurred at the time of the initial exposure change. 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. Some
ecological effects, such as the effects of acid deposition on Adirondack lakes, also occur
with a lag - again, we use a discounting procedure to standardize the benefits results for
these estimates.
The annual estimates for the three target years also provide an indication of the trend in
benefits we project will accrue over the 30-year study period. To generate a cumulative
measure of benefits over the full 30-year period, however, we must make an assumption
about the level of benefits that would be realized in the years between the target years.
We interpolate these values, assuming a trend in benefits accrual that roughly matches the
trend in emission reductions for PM precursors. Basing our estimate of the benefits
trajectory on PM precursor reductions acknowledges that the majority of monetized
benefits, including health and visibility, are attributable to reductions in ambient
particulate matter.
The distribution of estimates we generate for the monetized benefits of human health
effects incorporates both the quantified uncertainty associated with each of the health
effect estimates and the quantified uncertainty associated 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 benefits presented in this chapter, from the primary low estimate to the
primary high estimate, is narrower than would be expected with a complete accounting of
the uncertainties in all analytical components.101
In the health benefits analyses we estimate, for each endpoint-pollutant combination,
distributions of values for both the key parameter of the concentration-response function
and the valuation coefficients. We combine these distributions by using a computerized,
101 The characterization of the uncertainty surrounding economic valuation is discussed in detail in Industrial Economics, Inc.,
Uncertainty Analyses to Support the Second Section 812 Prospective Benefit-Cost Analysis of the Clean Air Act: Draft
Report, prepared for Office of Air and Radiation, US Environmental Protection Agency, April 2010.
7-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
statistical aggregation technique to estimate the mean of the monetized benefit estimate
for each endpoint-pollutant combination and to characterize the uncertainty surrounding
each estimate.102
The ecological and welfare results are not currently amenable to the same type of
uncertainty analysis. The modeling procedures for estimating the effects of sulfur and
nitrogen deposition in acidifying lakes, the effects of ozone in reducing timber and
agricultural production, and the effects of particulate matter on visibility are all subject to
uncertainty, but they require substantial resources simply to develop single point
estimates. We describe key uncertainties in these estimation procedures qualitatively in
Chapter 6, with some limited sensitivity analyses also presented to characterize the effect
of key assumptions. The sources of uncertainty in these estimates, however, cannot as
easily be disaggregated among physical effects modeling and valuation components, and
they have not been assessed with the BenMAP model used for health benefits uncertainty
analysis. As a result, we cannot reliably develop an aggregate estimate of the uncertainty
in the sum of health and welfare benefits estimates.
ANNUAL BENEFITS ESTIMATES
We present the results of our aggregation of primary annual health benefits estimates for
the CAAA in Figure 7-1 below. The figure provides a characterization of both the
primary central estimate and the range of values generated by the aggregation procedure
described above, for each of the three target years of the analysis (2000, 2010, and 2020).
The Primary High estimate corresponds to the 95th percentile value from the health
benefits aggregation, and the Primary Low estimate corresponds to the 5th percentile
value. The total benefits estimates are substantial; for example, the Primary Central
estimate in 2020 is $2.0 trillion.
Table 7-1 shows the detailed breakdown of benefits estimates for 2000, 2010, and 2020.
As shown in the table, $1.7 trillion of the $2.0 trillion total benefit estimate in 2020, or 85
percent, is attributable to reductions in premature mortality associated with reductions in
ambient particulate matter. The remaining benefits are roughly equally divided among
three broad categories of benefits: avoided premature mortality associated with ozone
exposure; avoided morbidity, the largest component of which is avoided acute myocardial
infarctions and avoided chronic bronchitis; and avoided ecological and other welfare
benefits, the largest component of which is improved visibility. Because of the
aggregation procedure used, and because we round all intermediate results to two
significant digits for presentation purposes, the columns of Table 7-1 may not sum to the
total estimate presented in the last row.
102 The statistical aggregation technique applied is commonly referred to as Monte Carlo analysis. The technique involves
many re-calculations of results, using different combinations of input parameters each time. For each calculation, values
from each input parameter's statistical distribution are selected at random to ensure that the calculation does not always
result in extreme values, or rely solely on low end or solely on high end input parameters. The aggregate distribution more
accurately reflects a reasonable likelihood of the joint occurrence of multiple input parameters.
7-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 7-1. ANNUAL MONETIZED BENEFITS IN 2000, 2010 AND 2020
Monetized Direct Benefits (million 2006$)
ice nnn nnn
2000 2010 2020
Examination of the emissions and aggregate exposure estimates suggests that most of
these benefits can be attributed to air quality improvements that result from CAAA
implementation, relative to conditions as they were in 1990, before the CAAA, rather
than from avoiding degradation of air quality that might have occurred without the
CAAA. For example, we estimate that emissions of NOX, SO2, and VOCs, three of the
most important PM and ozone precursors, would have grown just over 20 percent from
1990 to 2020 in the without-CAAA scenario, which corresponds to an annual growth rate
of about 0.65 percent. We also estimate that PM25 emissions would have grown
somewhat slower, at 0.5 percent annually, without the CAAA. Reductions along the
with-CAAA scenario over the same period, however, were more than 60 percent for SO2
and NOX, a reduction of roughly 3 percent per year, and were roughly 45 percent for
VOCs, a reduction of about 2 percent per year. As a result, about 75 percent of the
difference in emissions that we estimate would occur by 2020 between the with-CAAA
and without-CAAA scenario can be attributed to reductions in emissions relative to those
in 1990.
7-4
-------
TABLE 7-1.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
SUMMARY OF MEAN PRIMARY ANNUAL BENEFITS RESULTS
BENEFIT CATEGORY
ANNUAL MONETIZED BENEFITS (MILLION
2006$) BY TARGET YEAR
2000
2010
2020
NOTES
Health Effects
PM Mortality
PM Morbidity
Ozone Mortality
Ozone Morbidity
Subtotal Health
Effects
$710,000
$27,000
$10,000
$420
$750,000
$1 ,200,000
$46,000
$33,000
$1,300
$1,300,000
$1,700,000
$68,000
$55,000
$2,100
$1 ,900,000
- PM mortality estimates
based on Weibull distribution
derived from Pope et. al
(2002) and Laden et al., 2006.
- Ozone mortality estimates
based on pooled function
Visibility
Recreational
Residential
Subtotal Visibility
Agricultural and
Forest Productivity
Materials Damage
Ecological
Total: all categories
$3,300
$11,000
$14,000
$1,000
$58
$6.9
$770,000
$8,600
$25,000
$34,000
$5,500
$93
$7.5
$1,300,000
$19,000
$48,000
$67,000
$11,000
$110
$8.2
$2,000,000
Recreational visibility only
includes benefits in the
regions analyzed in Chestnut
and Rowe, 1990 (i.e.,
California, the Southwest, and
the Southeast).
Reduced lake acidification
benefits to recreational
fishing.
Note: See Chapters 5 and 6 of this report for detailed results summaries. Values presented are
means from results reported as distributions. Estimates presented with two significant figures.
PM2 5 exposure estimates also support the conclusion that more of the benefit in 2020 can
be attributed to air quality improvements from implementing CAAA programs than to
preventing degradation in air quality that might have resulted in the without-CAAA case.
Although we did not estimate 1990 air quality using the CMAQ/MATS system described
in Chapter 4, and the PM2 5 monitor network was very sparse in 1990, there was an
extensive PMi0 monitor network at that time. Using PMi0 monitor data and regional
PM2 s/PMio ratio estimates from the 1996 Particulate Matter Criteria Document, we
estimated population weighted average exposure to PM2 5 in 1990 of 19.0 ug/m3. In
addition, using the CMAQ/MATS system, we estimate population-weighted average
exposure to PM2 5 along the without-CAAA scenario is about 17 ug/m3 in 2000, and
increases to 17.7 (ig/m3 and 19.2 ug/m3 in 2010 and 2020. Along the with-CAAA
scenario, population weighted average exposure to PM2 5 is 12.2 (ig/m3 in 2000, and
declines to 10.9 ug/m3 in 2010, and 10.5 ug/m3 in 2020. In the without-CAAA scenario
some improvements in air quality occurred from 1990 to 2000 as a result of the
7-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 7-2.
continuing effect of the pre-1990 Clean Air Act requirements already on the books, but
after 2000 the without-CAAA scenario shows deterioration of air quality through 2020.
As shown in Table 7-2, there is considerable uncertainty in the estimates of health
benefits. As described above, the health benefit uncertainty analysis is based on
underlying statistical uncertainties in the concentration-response and valuation
coefficients. The low estimates are approximately an order of magnitude less than the
central estimate; the high estimate is three times the central estimate. Uncertainty
analyses for non-health benefits were not developed, but as they constitute only about
five percent of the central estimate, their contribution to the overall uncertainty in benefits
estimates is likely to be proportionately small.
DISTRIBUTION OF PRIMARY ANNUAL BENEFITS RESULTS FOR 2020
BENEFIT CATEGORY
PRIMARY ANNUAL MONETIZED BENEFITS
FOR 2020
(MILLION 2006$)
LOW CENTRAL HIGH
NOTES
Health Effects
PM Mortality
PM Morbidity
Ozone Mortality
Ozone Morbidity
Subtotal Health
Effects
$170,000 $1,700,000 $5,300,000
$17,000 $68,000 $190,000
$3,200 $55,000 $170,000
$780 $2,100 $3,600
$190,000 $1,900,000 $5,700,000
Low and high are 5th and 95th
percentile estimates from
health benefits uncertainty
analysis
Visibility
Recreational
Residential
Subtotal Visibility
Agricultural and
Forest Productivity
Materials Damage
Ecological
Total: all categories
$19,000
$48,000
$67,000
$1 1 ,000
$110
$8.2
$2,000,000
Only central estimates were
developed
Only central estimates were
developed
Reduced lake acidification
benefits to recreational fishing
Note: See Chapters 5 and 6 of this report for detailed results summaries. Estimates presented
with two significant figures; as a result, columns may not add to totals or subtotals.
AGGREGATE MONETIZED BENEFITS
As discussed earlier in this chapter, we interpolate benefit estimates between target years
and then aggregate the resulting annual estimates across the entire 1990 to 2020 period of
the study to yield a present discounted value of total aggregate benefits for the period. In
this section we present the results of the aggregation.
7-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 7-3.
In Table 7-3 we present the mean estimate from the aggregation procedure, along with
the Primary Low (i.e., 5th percentile of the distribution) and Primary High (i.e., 95th
percentile of the distribution) estimates, for all provisions we assessed. Aggregating the
stream of monetized benefits across years involved discounting the stream of monetized
benefits estimated for each year to the 1990 present value using a five percent discount
rate.
PRESENT VALUE OF MONETIZED BENEFITS OF THE CAAA
All Provisions, 1990 to 2020
PRESENT VALUE (MILLIONS 2006$)
PRIMARY LOW
$1,400,000
PRIMARY CENTRAL
$12,000,000
PRIMARY HIGH
$35,000,000
Note: Values presented in this table are in millions of 2006$, discounted to 1990 using a 5
percent discount rate.
COMPARISON OF BENEFITS AND COSTS
Table 7-4 presents summary quantitative results for the prospective assessment, with
costs disaggregated by emissions source category and benefits disaggregated by type.
We present annual, Primary Central estimate results for each of the three target years of
the analysis, with all dollar figures expressed as inflation-adjusted 2006 dollars. The final
columns provide net present value estimates for costs and benefits from 1990 to 2020,
discounted to 1990 at five percent. The results indicate that the Primary Central estimate
of benefits clearly exceeds the costs of the CAAA, for each of the target years and for the
cumulative estimates of present value over the 1990 to 2020 period.
As Table 7-4 indicates, a very high percentage of the benefits is attributable to reduced
premature mortality associated with reductions in ambient particulate matter and ozone.
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 dioxide and nitrogen oxides as well as, to a lesser extent, organic
constituents) and directly emitted PM (i.e., dust particles). Because the effects of these
constituents on ambient PM are nonlinear, and because some precursor pollutants interact
with each other in ways which influence the total concentration of particulates in the
atmosphere, separating the effects of individual pollutants on the change in ambient PM
would require many iterations of our air quality modeling system. Even with such a tool,
the interactive effects of pollutants are complex - as a result the marginal impact of any
particular pollutant is dependent on the levels of other pollutants as well. These factors
make it difficult to reliably link specific costs to specific aggregate benefits for the
pollutant source-specific components of the CAAA (e.g., electric utilities or additional
local controls).
7-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 7-4. SUMMARY OF QUANTIFIED PRIMARY CENTRAL ESTIMATE BENEFIT AND COSTS
(ESTIMATES IN MILLION 2006$)
COST OR BENEFIT CATEGORY
Costs:
Electric Utilities
Industrial Point Sources
Onroad Vehicles and Fuels
Nonroad Engines and Fuels
Area Sources
Local Controls to Meet NAAQS
Total Costs
Monetized Benefits:
Avoided Mortality
Avoided Morbidity
Ecological and Welfare Effects
Total Benefits
ANNUAL ESTIMATES
2000
$1,400
$3,100
$14,000
$300
$660
$0
$20,000
$720,000
$27,000
$15,000
$770,000
2010
$6,600
$5,200
$26,000
$360
$690
$14,000
$53,000
$1,200,000
$47,000
$39,000
$1,300,000
2020
$10,000
$5,100
$28,000
$1,200
$770
$20,000
$65,000
$1,800,000
$70,000
$78,000
$2,000,000
PRESENT VALUE
$49,000
$43,000
$220,000
$4,500
$7,600
$53,000
$380,000
$11,000,000
$410,000
$310,000
$12,000,000
Table 7-5 provides the results of our more detailed comparison of primary benefits
estimates to primary cost estimates. In the top half of the table we show both annual and
present value estimates. The cost estimates presented in the table reflect estimates
presented in Chapter 3. The monetized benefits indicate both the Primary Central
estimate (the mean) from our statistical aggregation procedure and the Primary Low and
Primary High estimates (5th and 95th percentile values, respectively). In the bottom half
of the table we present three alternative methods for comparing benefits to costs. "Net
benefits" reflect estimates of monetized benefits less costs. The table also notes the
benefit/cost ratios implied by the benefit ranges, and our estimates of the costs per
premature mortality avoided.
The results in Table 7-5 make it abundantly clear that the benefits of the CAAA exceed
its costs by a wide margin, making the CAAA a very good investment. Our estimates
rely on a particular set of data, models and assumptions we believe are most appropriate
at this time. It is possible that another set of data, models, or assumptions might yield
different estimates of benefits, costs, and benefit-cost comparisons. Nonetheless, the very
wide margin between estimated benefits and costs, and the results of the uncertainty
analysis, suggest that it is extremely unlikely that the monetized benefits of the CAAA
over the 1990 to 2020 period reasonably could be less than its costs, under any alternative
set of assumptions we can conceive. The central benefits estimate exceeds costs by a
factor of more than 30 to one, whether we are looking at annual or present value
measures, and the high estimate exceeds costs by roughly 90 to one.
7-8
-------
TABLE 7-5.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
SUMMARY COMPARISON OF BENEFITS AND COSTS
ANNUAL ESTIMATES
2000
2010
2020
PRESENT VALUE
ESTIMATE
1990-2020
Monetized Direct Costs (millions 2006$):
Low3
Central
High a
$20,000
$53,000
$65,000
$380,000
Monetized Direct Benefits (millions 2006$):
Lowb
Central
Highb
$90,000
$770,000
$2,300,000
$160,000
$1,300,000
$3,800,000
$250,000
$2,000,000
$5,700,000
$1,400,000
$12,000,000
$35,000,000
Net Benefits (millions 2006$):
Low
Central
High
$70,000
$750,000
$2,300,000
$110,000
$1 ,200,000
$3,700,000
$190,000
$1 ,900,000
$5,600,000
$1,000,000
$12,000,000
$35,000,000
Benefit/Cost Ratio:
Lowc
Central
Highc
5/1
39/1
115/1
3/1
25/1
72/1
4/1
31/1
88/1
4/1
32/1
92/1
Costs per Premature Mortality Avoided (2006$):
Central
$180,000
$330,000
$280,000
Not estimated
a 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 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.
c 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.
It is also clear from Table 7-5 that costs for criteria pollutant programs grow more
quickly than benefits at the beginning of the CAAA compliance period, from 2000 to
2010, and that benefits grow more quickly at the end of the period, from 2010 to 2020.
This is consistent with the general statement that investments in clean air tend to involve
upfront costs and benefits that accrue over time. The present value estimates in Table 7-5
show, however, that the total aggregated value of benefits far exceeds the costs - by our
measures, therefore, the programs associated with the 1990 Clean Air Act Amendments
have been, and will likely continue to be, a very good investment.
7-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
As indicated in the table, the low estimate of net benefits for the year 2020 is positive
(i.e., benefits exceed costs) and of significant magnitude - $190 billion. Our uncertainty
modeling therefore indicates that the likelihood that the cost estimates of $65 billion in
2020 could exceed the benefits estimates is much less than five percent.
OVERVIEW OF UNCERTAINTY ANALYSES
Completion of a study of this breadth and complexity has required EPA to directly
confront the role of uncertainty in the key analytic outcomes of the study. While the
previous section establishes that the primary estimates of benefits of air pollution control
greatly exceed the primary estimates of costs of CAAA compliance, it is nonetheless
important to evaluate the extent to which alternative models, assumptions about
scenarios, and key parameter choices might affect both benefits and costs. Cognizant of
advice to the Agency from the National Research Council,103 the Project Team
developed a three step approach to uncertainty analysis:
1. Identify important sources of uncertainty in each analytical element, starting with
emissions profile development. At the end of each of the preceding chapters, we
provide a table of key uncertainties and our assessment of the direction and
potential magnitude of the impact of this uncertainty on the key analytic output of
the study, the monetized net benefits of the CAAA.
2. Quantify parameter and model uncertainty quantitatively where possible by using
alternative assumptions or models to estimate intermediate and/or overall net
benefit results. In addition, explore options for assessing scenario uncertainty
that propagate through the complete analytic chain.
3. Compare the results from these quantitative analyses to the primary results, to
inform the degree of confidence in the primary analytic results and to help
identify new research directions to address or reduce uncertain and influential
components of the analysis.
In the remainder of this section we review each of these three components of our
uncertainty analysis.104
IDENTIFYING IMPORTANT SOURCES OF UNCERTAINTY
Within each of the summary uncertainty tables in the prior chapters the Project Team has
distinguished sources of uncertainty that could have a potentially major impact on the
overall net benefits estimate presented in this chapter, based either on quantitative
analyses or, where quantitative assessments are unavailable or infeasible, the judgment of
Project Team analysts. Potentially major factors are those for which a plausible
alternative assumption or approach could influence the overall benefit or cost estimate by
See National Research Council (2002), Estimating the Public Health Benefits of Proposed Air Pollution Regulations, The
National Academies Press, Washington, DC, in particular Chapter 5, titled: "Uncertainty."
104 For a more thorough description of the methods and results of these uncertainty analyses see the accompanying report
Uncertainty Analyses to Support the Second Section 812 Benefit-Cost Analysis of the Clean Air Act, March 2009.
7-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 7-6.
five percent or more. We identify a total of 13 potentially major sources of uncertainty in
Chapters 2 through 6; these are listed in Table 7-6 below.
POTENTIALLY MAJOR SOURCES OF UNCERTAINTY FOR ESTIMATING THE COSTS
AND BENEFITS OF THE CAAA
POTENTIAL SOURCE OF ERROR
Estimated emissions rates under the
counterf actual Without-CAAA scenario
Estimated economic growth - a key driver of
total emissions - under both scenarios
Forecast of the final form and compliance with
EPA's revisions of the vacated Clean Air
Mercury Rule and the remanded Clean Air
Interstate Rule
Secondary organic aerosol (SOA) chemistry
Inability to conclusively state that PM
mortality outcome is causal based on
epidemiology
Effect of socioeconomic status on mortality
from PM exposure
Attribution of exposure to PM in epidemiology
studies based on monitor data
Omission of short-term effects of PM exposure
on mortality
Timing of reduction in mortality risk after
exposure is reduced (cessation lag)
Source of mortality risk valuation includes
many older studies
Scenario of mortal risk in available valuation
studies is generally different from that
presented by air pollution
Valuation of risk avoidance can change over
time and as income increases
Incomplete coverage of ecological effects of
air pollutants, including omission of several
short-term and virtually all long-term
bioaccumulative and persistent effects.
ANALYTIC STEP
Emissions
Emissions
Emissions
Air Quality Modeling
Health Effects
Health Effects
Health Effects
Health Effects
Health Effects
Valuation
Valuation
Valuation
Ecological
DIRECTION OF POTENTIAL BIAS FOR
NET BENEFITS
Under-estimate
Unable to determine
Unable to determine
Under-estimate
Over-estimate
Unable to determine
Under-estimate
Under-estimate
Unable to determine
Unable to determine
Unable to determine
Unable to determine
Under-estimate
* The classification of each potential source of error reflects the best judgment 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.
See tables at the end of Chapters 2 through 6 above for more information.
Perhaps not surprisingly, the key emissions estimation uncertainties involve forecasting
errors, particularly related to estimating future economic and regulatory activity as well
as estimating behavior under the counterfactual without-CAAA scenario. A key cost
estimation uncertainty involves estimating NAAQS compliance, particularly when
currently known emissions reductions measures are not sufficient to achieve full
compliance with the standard in the future. However, in order for any uncertainty to be
considered "major" the impact would need to be of a magnitude of approximately $100
7-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
billion to affect net benefits estimates by as much as five percent. In our judgment, while
there are several factors that could affect direct cost estimates by a significant percentage,
no cost estimation uncertainty has the potential to either more than double our current
total cost estimate of $65 billion, or to reduce the cost estimate to $0 or less, which is the
magnitude that would be required to constitute five percent of the net benefit estimate.
Several uncertainties that affect benefits estimates, however, could have an impact of
$100 billion or greater on the net benefits estimates. Both health effects and valuation
uncertainties center on estimation of the impact of air pollutants on mortal risk and the
valuation of that health endpoint. The key ecological uncertainty involves identifying
what is missing from our necessarily limited quantified ecological benefits. Only one
potentially major factor was identified for the air quality modeling step - this may be the
result of our inability to apply alternative quantitative air quality modeling tools in this
already resource-intensive step in the analytic chain. It is worth noting, however, that as
a whole the air quality modeling process very likely contributes a greater than 5 percent
uncertainty, of indeterminate direction, to the overall uncertainty in benefits estimates. In
addition, the AQMS highlighted uncertainties introduced by the ex post adjustment of
some primary PM emissions estimates and the procedure used to re-calibrate the CMAQ
air quality to account for this emissions adjustment. Although we argue that the overall
effect of this source of uncertainty on the net benefits is probably minor (see Table 4-10
in Chapter 4), in some locations ambient PM from primary PM emissions can be more
important than secondarily formed fine particles. Overall, we believe that our application
of the MATS monitor calibration procedure, which provides a speciated calibration to
ensure better agreement between air quality modeling results and comparable monitor
data, provides the best agreement possible between our air quality simulation results and
monitored values. In the end, however, there is no way to validate the counterfactual,
without-CAAA scenario estimates.
Examination of the last column of Table 7-6 suggests a limited ability to estimate the
joint effect of these factors on the direction of potential bias for net benefits. Seven of the
factors listed have an indeterminate direction of effect; five yield a potential
underestimate of net benefits; and one results in a potential overestimate of net benefits.
The large number of factors with an indeterminate direction imply that the direction of
the net effect of all factors taken together remains unclear, but the relative confidence that
the PM exposure-mortality concentration-response function is causal, based on weight-of-
evidence, that being the only uncertainty that yields a potential overestimate, suggests
that our primary results may be more likely to understate net benefits than overstate them.
A comparison of the qualitative uncertainty tables from the First and Second Prospective
studies indicates that significant advancements over the First Prospective include the use
of improved monitoring data for PM2 5, an improved understanding and treatment of
atmospheric chemistry and the composition of PM2 5 emissions, and the use of longer-
term simulations with integrated modeling of criteria pollutants using CMAQ rather than
a collection of separate air quality models.
7-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
QUANTIFYING MODEL, PARAMETER, AND SCENARIO UNCERTAINTY
The benefits values presented in this report are subject to a number of uncertainties
related to data limitations, analytical choices related to models and input parameters,
difficulties predicting future scenarios, and other factors. As noted above, among the
most significant model uncertainties is the extensive list of benefits categories, mostly in
the ecological area, for which we currently lack the data and/or tools to quantify and
monetize benefits. These categories are implicitly treated as having zero value though in
reality they may include physical benefits that have a positive economic value. Examples
of potentially important, but unquantified ecological effects include nitrogen deposition,
non-ozone effects on forest and agriculture vegetation, effects of HAPs on ecological
structure and function, and synergistic effects associated with exposures to mixtures of
pollutants and interactions of the effects of conventional pollutants such as ozone with
climate change. The unquantified and unmonetized benefits thus represent an important
underestimation bias in the summary benefit results.
The uncertainties in our quantified and monetized primary benefits estimates that are
most likely to significantly influence the primary benefit results are those affecting the
largest benefit category: the estimation and valuation of reductions in premature mortality
due to decreases in PM2 5. Three key uncertainties affecting economic estimates of
avoided PM mortality include: (1) the C-R function estimate; (2) the PM/mortality
cessation lag structure; and (3) the mortality valuation estimate. These are influential
assumptions in our analysis and those for which plausible alternative quantitative
estimates are available. The companion Second Prospective Section 812 report,
Uncertainty Analyses to Support the Second Section 812 Benefit-Cost Analysis of the
Clean Air Act, presents detailed quantitative analyses of the sensitivity of benefits results
to these and other factors.
Table 7-7 presents a tabular summary of the results of the full range of uncertainty
analyses for both costs and benefits, and Figure 7-2 presents a graphical illustration of the
impacts of effect of alternative assumptions and models on the central estimate and
distribution of monetized avoided mortality benefits, the primary contributor to
monetized benefits.
COST UNCERTAINTIES
Table 7-7 shows that the impact of our alternative assumptions about mobile source cost
parameters, learning curves, and unidentified local control costs each have relatively
modest impacts on total costs, while the I&M failure rate and learning curve assumptions
have a slightly larger impact on total costs.105 In addition, the assumptions underlying our
primary cost estimates tend to be conservative; most of the alternatives decrease total
compliance costs and none increase costs more than about three percent.
105 The estimate of the impact on total costs is derived from the relative contribution of the affected cost sector to the
overall costs of compliance, assuming all other sectors are unaffected.
7-13
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 7-7. QUANTITATIVE ANALYSES OF UNCERTAINTY IN THE 812 SECOND PROSPECTIVE
ANALYSIS
FACTOR AND LOCATION OF
UNCERTAINTY ANALYSIS
DISCUSSION IN THIS REPORT
(WHERE APPLICABLE)
TYPE OF
UNCERTAINTY
EVALUATED
ALTERNATIVE ASSUMPTIONS
IMPACT OF
ALTERNATIVE
ASSUMPTIONS ON 2020
PRIMARY ESTIMATE
UNCERTAINTIES RELATED TO COST ESTIMATES
Unidentified controls
(Chapter 3)
I &M program vehicle failure
rates(Chapter 3)
Learning curve assumptions
(Chapter 3)
Fleet composition and fuel
efficiency (Chapter 3)
Parameter
Parameter
Parameter
Scenario
Alternate assumption about the
threshold for, and cost of, applying
unidentified local controls to achieve
NAAQS compliance ($10,000/ton).
Alternative assumption about failure
rates for I&M program testing based
onNRC(2001).
Alternate assumptions about the
learning rate (5 and 20%)
Alternate assumption about future
fleet composition and fuel efficiency
using AEO 2008.
-18% of local control
costs; -2.1% of total
costs
-12% for mobile source
costs;
-6. 5% of total costs
-6.0% to 3. 2% of total
costs
-3.6% for mobile
source costs;
-2.0% of total costs
UNCERTAINTIES RELATED TO BENEFITS ESTIMATES
Alternate C-R function for PM
(Chapter 5)
Emissions from ECU sources
(Chapter 2)
PM/Mortality Cessation lag
(Chapter 5)
Value of Statistical Life
(Chapter 5)
Discount rates
Alternate C-R function for
ozone (Chapter 5)
Parameter
Scenario
Model and
parameter
Parameter
Parameter
Parameter
Alternative C-R functions - two from
empirical literature (Pope et al.,
2002 and Laden et al., 2006) and 12
subjective estimates from the expert
elicitation study
Use continuous emissions monitoring
(CEM) data in place of Integrated
Planning Model (IPM) results,
coupled with alternative
counterfactual consistent with CEM
approach.
Alternative lag structures - one step
function and a series of smooth
functions (based on an exponential
decay). Smooth functions in some
cases also require change in C-R
coefficient.
Alternative VSL estimates
Alternate discount rates (5% and 7%)
Alternative C-R functions - three
from multi-city studies, three from
meta-analyses, and one from Jerrett
et al. cohort study.
-83% to 76%,
Based on most
extreme estimates
from PM EE study.
Rest of alternatives
range from -43% to 41%
+50% in 2000
Due almost entirely to
the impact of the
alternative without-
CAAA scenario.
-23% to 16% when using
primary C-R function.
-52 to 50% when also
changing C-R function.
-20% to 0%
-4% to 4%
0% for total mortality
benefits.
-63% to 66%
For ozone-related
mortality.
7-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FACTOR AND LOCATION OF
UNCERTAINTY ANALYSIS
DISCUSSION IN THIS REPORT
(WHERE APPLICABLE)
Emissions changes by
emitting sector
Differential toxicity of PM
components
Dynamic population modeling
(Chapter 5)
TYPE OF
UNCERTAINTY
EVALUATED
Scenario
Parameter
Model
ALTERNATIVE ASSUMPTIONS
Altering each sector-specific
emissions by 10 percent
Potential alternative estimates of
toxicity for specific PM components
Incorporation of dynamic population
estimates to calculate life years
gained and changes in life
expectancy
IMPACT OF
ALTERNATIVE
ASSUMPTIONS ON 2020
PRIMARY ESTIMATE
$/ton marginal benefit
for proportional ECU
sector reductions is
about 3 times that for
nonroad and on-road
sectors, and 50%
higher than that for
area and non-EGU
point source sectors.
N/A. No quantitative
sensitivity analysis
performed due to
significant data gaps.
N/A. Life years gained
and changes in life
expectancy are
supplemental
estimates of
PM/mortality effects
and cannot be directly
compared to the
primary estimate.
A further overarching issue with our direct cost methodology is that, for EGU modeling
and for some components of the ozone NAAQS compliance cost assessment, the method
employed assumes specific optimizing behavior by polluters. In particular, the IPM
model used for EGU compliance cost assessment assumes a forward looking approach
and may incorporate only limited available information on real-world constraints on
optimizing behavior such as long-term fuel supply contracts. If polluters do not optimize
in the manner assumed in these models, the direct costs may under-estimate the true costs
of compliance. For other emitting sectors, where optimization approaches were not
feasible, the potential for under-estimation from this factor does not apply.
A potential issue in considering the uncertainty in cost estimates is our inability to
adequately consider the effects of the CAAA on the quality of goods overall. Our method
emphasizes that the CAAA does increase the cost of products, and we attempt to hold the
quality of products constant in the process. In reality it is likely that the CAAA affects
both price and quality of products. One of the more straightforward examples is that
motor-vehicle emission controls may reduce performance, though at the same time those
controls can increase fuel economy. Other examples include substitution of other devices
for charcoal lighting fluid, reformulation of VOC emitting paints, and other product
changes that may have altered the quality of those products to consumers. As discussed
briefly in Chapter 3, however, the CAAA could also change quality in ways that benefit
consumers, but which we do not capture in our estimates - for example, low VOC paint
7-15
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
contributes not only to lower ambient ozone levels, but also reduces consumer exposure
to VOCs in enclosed indoor environments. Unfortunately, it is very difficult to quantify
the effect of this factor on our overall cost and net benefit estimates.
BENEFIT UNCERTAINTIES
On the benefits side, Table 7-7 and Figure 7-2 show that the most influential assumptions
affecting benefits are the choice of the C-R function, the cessation lag model for the
accrual of benefits, and the VSL distribution. While the two most extreme results from
EPA's Expert Elicitation (EE) study imply substantial effects of C-R choice (about 80
percent in either direction) most of the alternatives from the EE study and the published
epidemiological studies suggest effects on benefits of about 40 percent or less in either
direction. By themselves, longer cessation lag alternatives can reduce monetized benefits
by as much as a 23 percent and if coupled with a change in the C-R function, by close to
half; however, the Council Health Effects Subcommittee advised that much of the risk
reduction benefits from PM2 5 controls are more likely to accrue sooner rather than later.
Accelerating benefits increases benefits by about 16 percent when maintaining the same
C-R function, but could increase them by as much as half when using a smooth function
based on the Laden Six Cities follow-up effect estimate. VSL distribution choices in one
case produce the same central estimate; in others they reduce VSL between 7 and 20
percent.
A review of the box plots in Figure 7-2 for the factors that have the greatest potential to
change the central estimate shows that most of the alternatives do not have a dramatic
effect on the spread of uncertainty. Some alternatives suggest the high end of the
distribution could be lower, including all of the alternative VSL distributions, which give
less weight to higher VSL values than the 26-study Weibull. On the other hand, only a
few alternatives (from EPA's particulate matter expert elicitation study) significantly
extend the upper end and hardly any extend the lower end, suggesting our primary
estimate is unlikely to understate greatly the uncertainty in avoided mortality benefits. In
all these cases, however, we are unable to develop a probabilistic representation of
uncertainty in the emissions and air quality modeling steps; incorporating uncertainty in
these factors would certainly increase the spread between the Primary Low and Primary
High estimates.
LESSONS LEARNED AND NEW RESEARCH DIRECTIONS
Many of the factors contributing to uncertainty in these estimates are the result of
scientific unknowns that might be addressed through additional research. Identification
of research directions to address current unknowns can serve an important function - in
the First Prospective, for example, we identified eight high priority research directions,
six of which were addressed in the Second Prospective.106
The six were: improved emissions inventories and inventory management tools (see Chapter 2 for a description of the
improvements in the 2002 NEI, and the AirControlNET tool used to estimate emissions reductions necessary for NAAQS
compliance); improved tools for assessing the full range of social costs associated with regulation, including the tax-
interaction effect (see Chapter 8 of this document for a description of the economic modeling tool EMPAX-CGE); a more
7-16
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 7-2. SUMMARY OF QUANTITATIVE ANALYSIS OF UNCERTAINTY IN MONETIZED MORTALITY
BENEFITS ESTIMATES
£*
CD rs
tl
C UJ
11
ii
Hi
c u
t\iu~fa -
350% -
300% -
250% -
200% -
150% -
100% -
50% -
-50% -
-100% -
1 ^A%
•
! •-
Altem alive C-F
I Functions
1
95th
I Mean
I 5th
Alternative Alternative Alternative
Lags VSLs Discount
Rates
r
I
j
i
iC'rntnrnmmmmrnriirrimrn zu-WW« < < z < !£T
Ililfffllfffi C I I 1 1 I I 1 1 S?
3 5
I §
11
° ?
I *
I I
i l
£ £
o o
to.
»
Note: These estimates renresentthe percent change for each alternative assumption on the total mean mortality benefits estimates. Note that the Project
Teamassessedthe impact Of alternative ozone C-R functions on the total mortality benefits and found that these did not alter the primary estimate. Therefore.
alternative ozone C-R functions are not included in this graph
The key lessons learned in this analysis lead directly to new research directions to inform
future assessments, assessments which include both Regulatory Impact Analyses of
specific rules and broader, policy-oriented documents such as this Second Prospective.
The key insight from this analysis is that rules that target precursors of fine particulate
matter are likely to be very cost-effective. Costs per ton of PM control are similar or less
than previously estimated, and benefits per ton emitted of these precursors are much
larger than previously thought or estimated in the First Prospective. There are three key
reasons for the large increase in benefits per ton of PM precursor emitted, involving
advances in our understanding of: 1. PM species emissions, 2. the fate of these emissions
as estimated by integrated national-scale air quality modeling systems, and 3. the
implications of fine particulate air quality for premature mortality.
geographically comprehensive air quality monitoring network, particularly for fine particulate matter (see discussion of the
MATS procedure in Chapter 4 of this document); development of integrated air quality modeling tools based on an open,
consistent model architecture (see Chapter 4 for a description of the CMAQ system); increased basic and targeted research
on the health effects of air pollution, especially particulate matter (see Chapter 5 for multiple examples of recent work
that was applied in this analysis); continued efforts to assess the cancer and noncancer effects of air toxics exposure (see
discussion of the air toxics case study in Chapter 5).
7-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
In addition, the results of the study also provide evidence of the significant benefits of
avoiding mortality associated with ozone exposure, avoiding degradation of visibility in
residential and recreational settings, and avoiding significant chronic and acute morbidity,
including chronic bronchitis and acute myocardial infarction. The last two of these
monetized benefits categories were shown, by themselves, nearly to equal the full costs of
all provisions of the CAAA. There also remain large categories of health and ecological
benefits for which we have no quantified or monetized benefit estimates. For example,
although there is an established literature linking air pollutant exposure with increased
risk of cerebrovascular accidents (stroke), as well as a literature on the medical costs of
this condition, that category of effect is not yet included in our estimates of the health
benefits of reducing air pollutant exposure.
Insights gleaned from completing this study suggest the following eight areas to be the
highest priority research needs:
• Improving cost analyses for rules that are technology-forcing. The overall cost
analysis in Chapter 3 is characterized by complete coverage of the costs of many
rules, but the Project Team acknowledges that in some cases, particularly
involving compliance with tighter future NAAQS standards, application of the
suite of known, cost-effective current pollutant control measures are not
sufficient to achieve compliance in all locations. This shortcoming remains one
of the important focal points for compliance cost research within the Agency.
One possible direction that the Agency is considering is analysis of historical data
on the cost and penetration rates of new emissions control technologies,
particularly those for NAAQS compliance, which could provide insights on the
process, cost, timing, and potential limits of induced innovation.
• Continuing efforts to incorporate a broader range of market benefits in
economy-wide modeling of the impacts of regulation. The results of Chapter 8
indicate that there are significant benefits to economic growth when we consider
the labor force and health expenditure implications of cleaner air. Our
demonstration of the importance of incorporating benefits-side effects in macro-
economic modeling efforts, however, does not incorporate all possible market
effects of cleaner air. For example, increased agricultural and forest productivity
might feasibly be incorporated in the model we employed. Ultimately, it will
also be important to develop new methods to characterize the large nonmarket
benefits of cleaner air in these models, including most importantly the welfare
enhancements (as opposed to simply the market implications) associated with
reductions in premature mortality.
• Understanding synergies and antagonistic effects of climate change in
realizing benefits, as well as for understanding co-benefits of greenhouse gas
(GHG) control policies. Consideration of climate change was outside the scope
of this Second Prospective effort, but designing effective and efficient regulatory
mechanisms for GHG emissions control has rapidly become an important priority
for the Agency. The methods, data, and results of this study are important for
7-18
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
modeling co-benefits of GHG control policies, as many policies targeted at GHG
reductions also reduce other, conventional pollutants, and those benefits are
realized sooner than the generally long-term benefits of GHG policies. In
addition, climate change likely alters the benefits achieved by conventional
pollutant policies, as for example increases in mean temperature as well as
increases the frequency of extreme temperature events creates conditions
conducive to ozone formation. Both areas are important for further research.
• Developing probabilistic representation of emissions and air quality to support
uncertainty analysis. As noted earlier in this chapter, a major shortcoming of
existing quantitative characterizations of uncertainty in benefits and costs of the
CAAA is the inability to integrate uncertainties in emissions and air quality
modeling steps. Two areas of research deserve further attention: 1. Developing
more nimble tools for assessing the air quality implications of emissions control
policies, or updating those that exist; 2. Developing probabilistic
characterizations of key parameters that contribute to overall uncertainty in
emissions and air quality analyses. Pursuit of the latter initiative will likely
require application of expert elicitation, either formal or informal, to make
progress.
• Understanding the potential for differential toxicity to play a role in benefits of
control programs and, by extension, policy priorities. The issue of species-
specific particulate matter toxicity remains very complex, involving the effects of
mixtures and synergies of species that are not currently well understood. It is
nonetheless important to understand the extent to which rules targeted at specific
PM species might yield similar benefits as rules targeting total PM mass.
• Continuing to pursue evidence of the real-world public health impact of
specific air quality actions. Sometimes referred to as accountability analyses,
tracking the real-world instances of rapid air quality changes, either
improvements or reductions in air quality, can yield important corroborating
evidence of the effects found in epidemiology studies. As we found in our
uncertainty analyses supporting the Second Prospective, these natural
experiments also provide insights for the nature of cessation lags, and might be
useful in better understanding species-specific toxicity.
• Expanding coverage of ecological benefits. There are potentially large
ecological benefits of air pollution control that are not currently quantified.
Some of the most important categories of unqualified effects include nitrogen
deposition effects on estuarine areas, sulfur deposition effects on vegetation and
other aspects of terrestrial systems, and long-term effects of air toxics. Perhaps
equally important, but much more subtle, are the long-term effects of a wide
range of air pollutants on ecosystem structure and function. Even potentially
beneficial effects of pollutants, such as deposition of the nutrient nitrogen in
terrestrial and even actively managed farms and forests, might have longer-term
7-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
detrimental effects on nutrient cycling and species selection that are currently
poorly understood.
• Expanding coverage of health benefits. Great effort has been expended to better
characterize the full range of health implications of air pollution. Despite this
effort, it is still difficult to quantify the link between air pollution and stroke, and
it is also difficult to assess the incremental effects of gaseous pollutant exposures,
in part because there are only a limited set of studies that characterize the
individual contributions of multiple pollutant exposures on health outcomes.
While the Agency has developed robust benefits analyses for programs that
control individual gaseous pollutants, such as carbon monoxide, it remains
difficult to incorporate these effects in multi-pollutant models that include PM,
ozone, and other gaseous pollutants typically present in many settings in the U.S.
The results of this Second Prospective study clearly provide strong evidence that the
nation's investment in clean air has been a wise and cost-effective policy. Continued
effort is needed to ensure that air pollution policies are pursued in the most cost-effective
manner possible. Pursuit of these research goals should continue to enhance our ability to
provide accurate and timely assessments of the costs and benefits of all provision
authorized under the Clean Air Act and its Amendments.
7-20
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
CHAPTER 8 - COMPUTABLE GENERAL EQUILIBRIUM ANALYSIS
The 1990 Clean Air Act Amendments (CAAA) represent a significant change in Federal
air pollution policy affecting virtually every sector of the U.S. economy, including
industry as well as individual households. The cost and benefit estimates presented in the
previous chapters reflect the direct impacts of the CAAA in terms of industry's and
households' direct compliance expenditures and the value of the direct human health,
visibility, ecological, and other benefits associated with CAAA-related improvements in
air quality. The cost-benefit information is central to EPA's analysis of the Amendments,
but policymakers and the public are also interested in the impact of CAAA programs on
overall economic performance. Therefore, to supplement the direct cost and benefit
estimates presented in the previous chapter, the Project Team applied an economy-wide
computable general equilibrium (CGE) analysis of the Amendments and estimated the
effect of the CAAA on U.S. gross domestic product and other macroeconomic measures.
The Project Team performed this analysis with the Economic Model for Policy Analysis
(EMPAX-CGE), a CGE model employed by EPA for several previous analyses of CAAA
regulations, including the National Ambient Air Quality Standards (NAAQS) for PM2 5,
the 8-hour Ozone NAAQS, and the Clean Air Interstate Rule.
The Project Team's CGE analysis for the Second Prospective represents a major step
forward in EPA's application of CGE models in the context of air pollution policy.
Unlike previous CGE analyses that focused exclusively on the macroeconomic impacts of
compliance expenditures, the Second Prospective incorporates impacts related to both
CAAA costs and some categories of benefits into EMPAX-CGE, to the extent feasible.
Because both the costs and benefits of CAAA regulations may affect the size and
composition of the U.S. economy, the Project Team's approach provides a more
comprehensive and balanced view of the macroeconomic impacts of air pollution policy
than previous assessments. To illustrate the extent to which including labor force and
medical expenditure impacts in EMPAX-CGE affects model results, we applied the
model in two ways: one model run that reflects only the costs of the CAAA (the cost-only
case) and a second model run that reflects both the costs and a subset of the total benefits
of the Amendments (the labor force-adjusted case).
This chapter presents the CGE analysis in four sections. In the first section, we provide
an overview of EMPAX-CGE, describing the model's overall structure and highlighting
the sectoral and geographic resolution of the model. The second section describes the
development of the cost- and labor force and health expenditure benefit-side inputs for
the analysis and documents how these inputs were incorporated into EMPAX-CGE. The
third section presents the results of our analysis, both in aggregate and by industry. To
8-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
conclude the chapter, we discuss the major uncertainties of the analysis and their
implications for results.
EMPAX-CGE107
EMPAX-CGE is a multi-industry, multi-region computable general equilibrium model of
the U.S. economy. Below we describe the main features of the typical CGE model,
followed by a more detailed overview of the structure and functionality of EMPAX-CGE.
OVERVIEW OF CGE MODELING
CGE models simulate the flow of commodities and factors of production (i.e., labor,
capital, and natural resources) among producers and households to assess how a change
in policy or an economic shock affects the size and composition of the economy. As
shown in Figure 8-1, households in CGE models own factors of production (capital,
labor, and natural resources) that they supply to firms in exchange for wages and other
forms of income. Firms use these factors in conjunction with intermediate inputs
purchased from other industries to produce goods and services, which are sold to other
industries as well as consumers. Goods and services can also be exported, and imported
goods can be purchased from other countries.
In modeling the circular flow of the economy depicted in Figure 8-1, CGE models
capture behavioral changes among households and firms in response to changes in prices.
At the producer level, CGE models simulate the substitution of inputs as the price of one
input, such as steel or labor, rises relative to the price of other inputs. This allows the
simulation of producer behavior in CGE models using minimization of the cost of
production as an objective, consistent with the behavior of firms in the real economy.
Similarly, as the price of one good rises relative to the prices of other products and
services, CGEs model the process whereby households consume less of the more
expensive good and more of other goods. Related to households' substitution between
different goods, CGE models also simulate household substitution between labor and
leisure as real wages change. Because the productive capacity of the economy is
dependent, in part, on labor supply, the labor-leisure tradeoff is critical in determining the
size of the economy.
' The description of CGE models, in general, and EMPAX-CGE included in this section is based on RTI (2008).
8-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 8-1. CGE MODEL SCHEMATIC
Imports
Households
purchase goods
& services
Exports
Households
Firms buy 1
goods & services |
as inputs w
Firms supply
goods &
services
Firms
Households
receive income
from facto?
sales
Firms
purchase
factors
Households
supply factors
Source: RTI International, EMPAX-CGE Model Documentation, prepared for U.S. EPA
Office of Air Quality Planning and Standards, March 2008.
The general equilibrium component of CGE modeling requires a comprehensive market
coverage in which all sectors in the economy are in balance and all economic flows are
accounted for. Establishing equilibrium conditions requires that every commodity that is
produced must be purchased by firms or consumers within the United States or exported
to foreign consumers. The requirement for all markets to be in equilibrium during the
time period of the model simulation is a simplifying assumption of the model, but is
nonetheless a condition which, over time, is consistent with production in the actual
economy. Prices of these goods reflect all costs of production. Households receive
payments for their productive factors and transfers from the government (not shown in
Figure 8-1), and this income must equal consumer expenditures and savings. In
aggregate, all markets must clear, meaning that supplies of commodities and factors must
equal demand, and the income of each household must equal its factor endowments plus
any net transfers received. An important implication of this market clearing assumption
is that CGE models assume that the economy is at full employment (i.e., there is no
involuntary unemployment). Therefore, CGE models do not typically provide insights
into the unemployment impacts of policy changes.
8-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
OVERVIEW OF EMPAX-CGE
Similar to other CGE models, EMPAX-CGE is structured to represent the complex
interactions between consumers and producers in the real economy. To model these
interactions, EMPAX-CGE performs thousands of calculations with the objective of
maximizing household utility (well-being) while simultaneously maximizing firm profits.
While complex, these calculations are a simplified representation of the real economy.
The behavior of households and firms is inherently multi-faceted and dependent on a
range of factors, many of which are not well understood. To model this behavior,
EMPAX-CGE uses a simplified, hierarchical representation of household and firm
decision-making that reduces the behavior of households and firms to a limited number of
structured decisions. For example, as shown in Figure 8-2, the first decision for the
household sector in EMPAX-CGE is the optimization of consumption and leisure. To
model this decision, EMPAX-CGE assumes that households are free to allocate their time
between labor and leisure to maximize their welfare. Time that households do not devote
to leisure represents household labor supplied to producers. Therefore, in effect, the
leisure-consumption decision also represents a tradeoff between leisure and labor force
participation. After the consumption-leisure decision, EMPAX-CGE simulates
household consumption as a series of hierarchical decisions involving consumption goods
and transportation.
EMPAX-CGE also models firm behavior as a series of hierarchical decisions. Similar to
EMPAX-CGE's treatment of households, this hierarchical structure represents a
simplification of how firms decide which inputs to use in the production of goods and
services. As illustrated in Figure 8-3, the first tier of this decision hierarchy is a choice
between: (1) an indeterminate mix of capital, labor, and energy and (2) goods and
services produced by other industries, such as steel or computer equipment. Producers
then optimize among capital, labor, and energy.
Consistent with simplifying household and firm decision-making into the structured
frameworks depicted in Figures 8-2 and 8-3, EMPAX also uses a simplified
representation of the overall structure of the economy. Firms in the U.S. are scattered
across thousands of industries and produce countless goods and services. Modeling each
of these sectors individually within an economy-wide model, however, is not feasible due
to data and computational processing constraints. To address this issue, EMPAX-CGE
aggregates the economy into 35 distinct industries, as listed in Table 8-1. The industry
classifications included in EMPAX-CGE were defined so as to maximize the level of
sectoral detail among energy-intensive and manufacturing industries. EMPAX-CGE also
separates the electricity industry into fossil fuel generation and non-fossil generation,
which is important for assessing the impacts of policies that affect only fossil fuel-fired
electricity, such as air pollutant regulations.
8-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 8-2. EMPAX-CGE DECISION HIERARCHY FOR HOUSEHOLDS
Household utility is a function of
consumption and leisure.
Consumption Leisure
Personal Purchased
Transport Transport
Petroleum
Services Manufactured
Goods
Consumption Goods
Energy
Goods and
Services
Domestic
Foreign
FIGURE 8-3. EMPAX-CGE NESTED STRUCTURE FOR PRODUCERS
Output
Capital/Labor/Energy
Goods and services produced by
other industries
Energy
Capital & Labor
Energy
(5 types)
Capital
Labor
8-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
EMPAX-CGE is also designed to reflect regional differences in the overall structure of
the economy. Because the availability and cost of different production inputs, such as
labor and energy, vary across different regions of the U.S., the response of a given
industry to changes in policy may vary by region. To account for this effect, EMPAX-
CGE models each industry separately in five different regions, as shown in Figure 8-4.
The specification of the five economic regions included in the model is based, as closely
as possible, on the structure of the electricity market regions defined by the North
American Electric Reliability Council (NERC).108
TABLE 8-1. INDUSTRIES IN EMPAX-CGE
EMPAX Industry
North American Industry Classification
System (NAICS)
Energy
Coal
Crude oila
Electricity (fossil and nonfossil)
Natural gas
Petroleum refining b
General
Agriculture
Mining (w/o coal, crude, gas)
Construction
Manufacturing
Food products
Textiles and apparel
Lumber
Paper and allied
Printing
Chemicals
Plastic and rubber
Glass
Cement
Other minerals
Iron and steel
Aluminum
Other primary metals
Fabricated metal products
2121
211111,4861
2211
211112,2212,4862
324,48691
11
21
23
311
313,314,315,316
321
322
323
325
326
3272
3273
3271,3274,3279
3311,3312
3313
3314,3316
332
108 Economic data and information on non-electricity energy markets are generally available only at the state level, which
necessitates an approximation of the NERC regions that follows state boundaries.
8-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Manufacturing equipment
Computers & communication equipment
Electronic equipment
Transportation equipment
Miscellaneous remaining
Services
Wholesale & retail trade
Transportation °
Information
Finance and real estate
Business/professional
Education (w/public)
Health care (w/public)
Other services
333
334
335
336
312,337,339
42, 44, 45
481-488
51
52,54
53,55,56
61
62
71,72,81,92
a Although NAICS 211111 covers both crude oil and gas extraction, the gas component of this sector is
addressed in the natural gas energy sector.
b EMPAX-CGE reports output for the petroleum refining industry based on the delivered price of petroleum
products. This reflects the value of pipeline transport.
0 Transportation does not include NAICS 4862 (natural gas distribution), which is part of the natural gas
industry.
FIGURE 8-4. EMPAX-CGE REGIONS
""West" also includes
Alaska and Hawaii
8-7
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
EMPAX-CGE assumes that households have perfect foresight of future changes in policy
and maximize utility over the full time horizon of the model. To adjust to future policy
changes, households may alter their decisions about labor force participation and modify
their consumption patterns in terms of their overall level of consumption and the mix of
goods and services they choose to consume. This is in contrast to static CGEs, which
model the economy without regard for time (i.e., they effectively model the economy for
a single time period).
EMPAX-CGE contains four representative households in each model region, classified
by income. These household income groups are:
• $0 to $14,999,
• $15,000 to $29,999,
• $30,000 to $49,999, and
• $50,000 and above.
These representative households are assumed to possess certain factors of production
including labor, capital, natural resources, and land inputs to agricultural production.
Factor prices are equal to the marginal revenue received by firms from employing an
additional unit of labor or capital, and households allocate income from sales of these
productive factors to purchases of consumption goods to maximize welfare.
The outputs generated by EMPAX-CGE include GDP, consumption, and an economic
welfare measure known as Hicksian equivalent variation (EV). EV is based on the
concept of willingness-to-pay, which is the maximum amount a household would pay for
a particular good or service (including leisure), given its budget constraint. Willingness
to pay reflects the value or welfare that a household derives from the consumption of a
good or service. For a given policy scenario, the change in EV represents the additional
money that a household would require (at original prices and income) to make it as well
off with the new policy as it was under baseline conditions; this amount is "equivalent" to
the change in utility the household derives from consumption and leisure time. It is
important to note, however, that EMPAX-CGE's estimation of EV captures welfare
associated with market goods and services but does not capture non-market effects. As a
result, the measure would not reflect some categories of household welfare that are
important to our cost-benefit analysis, such as avoided pain and suffering associated with
health effects incidence, improvements in visibility, and changes in service flows that
derive from well functioning ecological resources.
The baseline values for the outputs generated by EMPAX-CGE are adapted from the
economic forecast in the Department of Energy's Annual Energy Outlook 2007. These
baseline values represent the U.S. economy under the with-CAAA scenario for the Second
Prospective.109
109 As noted in Chapter 2, the emissions projections for the Second Prospective are based on the economic forecast from
Annual Energy Outlook (AEO) 2005, not AEO 2007. The AEO 2007 forecast, however, is similar to that in AEO 2005. For the
year 2020, the AEO 2007 GDP forecast is approximately 3 percent lower than the projection from AEO 2005.
8-8
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
DEVELOPMENT OF MODEL INPUTS
The Project Team estimated the macroeconomic impacts of the CAAA as the difference
between (1) the EMPAX-CGE reference case projections, which represent the with-
CAAA scenario, and (2) EMPAX-CGE projections for the without-CAAA scenario. To
conduct the model runs for the without-CAAA scenario, the Project Team developed
model inputs related to both the costs and benefits of the Amendments. To assess the
difference in costs associated with CAAA compliance, we estimated CAAA-related
compliance expenditures by industry and EMPAX region. Based on these estimates, the
Project Team reduced the cost of production for affected industries from the baseline
costs of production to develop industry-wide cost structures for the without-CAAA
scenario. The "cost-only" runs therefore estimate the loss in economic productivity
associated with CAAA compliance costs.
As noted above, however, the CAAA also yields benefits that result in potentially
substantial changes in economic production as well. The benefit-side inputs developed
by the Project Team include (1) medical expenditures associated with pollution-related
illness, (2) the change in workers' time endowment due to pollution-related mortality, and
(3) the change in workers' time endowment due to pollution-related morbidity. The
Project Team incorporated changes in medical expenditures into EMPAX-CGE as
changes in household expenditure patterns. To incorporate changes in the amount of time
workers can devote to labor or to leisure in the model, we first estimated how health
effects and mortality estimated in Chapter 5 would affect the exposed population's ability
to supply labor to firms. Estimates of lost work time associated with morbidity have been
estimated in prior work or are available from BenMAP.110 Next, we assumed that
pollution-related illness and mortality among the labor force reduce workers' overall time
endowment (labor and leisure) in proportion to the effect on labor supply. That is, if air
pollution would reduce labor supply by x percent in 2020, the Project Team assumed that
the overall time endowment of workers would also decline by x percent in 2020.
We did not attempt to incorporate time endowment effects for people outside the formal
economy (e.g., retirees, students, homemakers) into EMPAX-CGE. While the "non-
working" population is clearly affected by air pollution, and those effects are likely to
influence the level and composition of economic activity, the structure of EMPAX-CGE
is not conducive to assessing how these populations affect the economy. The results
presented in this chapter therefore likely underestimate the macroeconomic impacts
resulting from CAAA-related improvements in public health.
Below we describe the Project Team's approach for generating the EMPAX-CGE inputs
related to the costs and benefits of the CAAA. As noted above, the Project Team used
these inputs to conduct two analyses of CAAA-related macroeconomic impacts; the first
reflects only the costs of the CAAA (the cost-only case), while the second reflects both
the costs and selected human health benefits of the Amendments (the labor force-adjusted
case).
110 For example, Cropper and Krupnick (1999) estimate income losses resulting from chronic bronchitis and acute myocardial
infarction. Based on these estimates, we calculated the lost work time per case associated with each of these endpoints.
8-9
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
COST INPUTS
To assess the macroeconomic impacts of CAAA-related costs, the Project Team
incorporated CAAA compliance expenditures by industry and region into EMPAX-CGE.
Similar to other CGE models, EMPAX-CGE is an expenditure-based model and therefore
requires expenditure-based inputs to represent the costs of the Amendments. CAAA
compliance expenditures, however, are not always the equivalent of the direct costs of the
Amendments presented in Chapter 3. While the direct costs of the CAAA reflect the
value of the capital, labor, and other resources necessary for CAAA compliance,
compliance expenditures simply represent the financial resources exchanged for CAAA
compliance. For example, the direct costs of the Amendments do not include taxes,
because such payments represent transfers rather than resources expended to control air
pollutant emissions. In contrast, CAAA compliance expenditures include transfers
because they represent an exchange of financial resources from one party (e.g., a firm) to
another (e.g., the government) that can affect the choices made by firms.
To estimate the compliance expenditures associated with the Amendments, the Project
Team made three adjustments to the direct cost estimates presented in Chapter 3:
1. Inclusion of fuel excise taxes: The Project Team included fuel excise taxes in
the compliance expenditure estimates developed for the EMPAX-CGE analysis.
Excise taxes were excluded from the direct cost estimates presented in Chapter 3
because such taxes are transfers.
2. Industry-specific discount rates: Unlike the direct cost estimates presented in
Chapter 3, which reflect a 5 percent social discount rate, the compliance
expenditures presented in this chapter reflect the private discount rates of affected
industries. For each industry, we estimated the private discount rate based on the
industry-specific weighted average cost of capital as reported in Ibbotson
Associates' Cost of Capital Yearbook.111
3. Exclusion of motorist waiting time from cost estimates for inspection and
maintenance programs: The direct cost estimates for motor vehicle inspection
and maintenance (I&M) programs in Chapter 3 reflect the value of motorist
waiting time. Although waiting time represents a welfare loss to society, this
cost is not incurred as an expenditure. Because CGEs are expenditure-based
models, we exclude motorist waiting time from the cost-side inputs incorporated
into EMPAX-CGE. The exclusion of motorist waiting time is unlikely to
significantly affect the results of the CGE analysis, as these costs represent only
18 percent of direct CAAA costs associated with I&M programs and less than 5
percent of direct costs for the entire on-road sector.
Based on these adjustments, we developed the compliance expenditure estimates
presented in Table 8-2. For comparison, the exhibit also includes the direct cost estimates
summarized in Chapter 3. As indicated in the exhibit, the estimated CAAA compliance
Ibbotson Associates, Cost of Capital Yearbook, 1997 through 2006 editions.
8-10
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 8-2.
expenditures in 2010 are approximately $2.0 billion greater than the Project Team's
direct cost estimates for 2010. In 2020, the difference between the two increases to $3.0
billion. The estimates in Exhibit 8-6 also show that the distribution of compliance
expenditures across source categories is similar to the distribution of direct costs.112
SUMMARY OF ANNUAL CAAA COMPLIANCE EXPENDITURES AND DIRECT COSTS
(MILLIONS OF 2006$)
SOURCE CATEGORY
Electric Generating Units
On -road Sources
Non-road Sources
Industrial Point Sources
Area Sources
Local Controls (Identified)
Unidentified Local Controls
TOTAL
2010
COMPLIANCE
EXPENDITURES
(USED FOR EMPAX
ANALYSIS)
$8,470
$24,800
$750
$5,580
$693
$5,590
$9,020
$54,900
DIRECT
COSTS
$6,640
$25,800
$359
$5,180
$693
$5,250
$9,020
$52,900
2020
COMPLIANCE
EXPENDITURES
(USED FOR EMPAX
ANALYSIS)
$13,000
$27,200
$1,620
$5,600
$768
$6,790
$13,500
$68,500
DIRECT
COSTS
$10,400
$28,300
$1,150
$5,140
$767
$6,180
$13,500
$65,500
In most of EPA's recent EMPAX applications to air pollution rules, only a small portion of total costs have been accounted
for by expenditures in the household sector. In this application, however, a large portion of total compliance costs,
particularly for mobile source fuels rules, involve increased expenditures by the household sector. For this reason, the
Project Team gave special consideration to the treatment of these costs. Estimated household compliance expenditures
associated with petroleum products are implemented as price adjustments to reflect higher motor vehicle fuel prices. The
petroleum price adjustment is calculated to match compliance expenditures related to household transportation fuel use.
For other transportation compliance expenditures, the household utility function is adjusted to require additional
expenditures to achieve a given utility level. These adjustments reflect the additional automotive inspections, maintenance,
and technologies purchased by households to comply with the Clean Air Act. Other unidentified household compliance costs
not related to transportation (e.g. non-road related local controls) are treated as lump-sum reductions to household income.
8-11
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
BENEFIT INPUTS
As noted above, the Project Team's analysis of the macroeconomic impacts of CAAA-
related health improvements focuses on three specific effects: (1) the change in the
household time endowment from pollution-related mortality impacts, (2) the change in
the household time endowment from pollution-related morbidity, and (3) the change in
medical expenditures associated with pollution-related morbidity. The Project Team
incorporated these effects into the without-CAAA EMPAX-CGE model runs to estimate
the size and composition of the economy in the absence of the Amendments. The
methods employed to quantify these effects and convert them into useable inputs for
EMPAX-CGE are described below.
Mortality-related Labor Force Impacts
The Project Team incorporated pollution-related mortality impacts into EMPAX-CGE as
a percentage change in the time available to workers for labor and leisure activities (i.e.,
their time endowment). In estimating this percentage change, the Project Team focused
on the dynamic population effects of premature mortality from particulate matter (PM)
exposure. While ozone also leads to premature mortality, the benefits results in Chapter 5
show that reductions in ambient PM concentrations are responsible for approximately 98
percent of the avoided cases of premature mortality associated with the Amendments in
both 2010 and 2020. Because of the dominant effect of PM on mortality (relative to
ozone) and the lack of tools available to examine the dynamic population effects of PM
and ozone in an integrated fashion, the Project Team focused the mortality component of
the EMPAX-CGE analysis on changes in PM-related mortality.
The mortality-related inputs developed by the Project Team reflect the dynamic effects of
PM mortality on the population overtime. When PM concentrations change, the
resulting population impact grows over time, as the change in population for any given
year reflects changes in the incidence of PM-related mortality from prior years. For
example, if PM concentrations are reduced permanently in 2015, the population (and the
size of the labor force) in 2017 will reflect avoided cases of premature mortality in 2015,
2016, and 2017. Over time, this dynamic effect leads to a significant number of life years
saved as the reduction in pollution-related risk is applied to successively larger
populations each year (due to previous years' improvements in air quality).
To capture these dynamic effects, the Project Team used a spreadsheet-based dynamic
population simulation model described in Chapter 5.113 The model was designed to track
the effect of alternative assumptions about the mortality effects of PM25 on the U.S.
population, but may also be used to assess how changes in PM2 5 concentrations lead to
changes in the population over time. The tool incorporates detailed life table data for
historical years, by age, gender, and cause of death, obtained from the Census Bureau and
the Centers for Disease Control. It also incorporates Census mortality and population
projections for future years, again by age and gender, using the projected death and birth
113 For a detailed description of the model, see the related report, Uncertainty Analyses to Support the Second Section 812
Benefit-Cost Analysis of the Clean Air Act, March 2010, and Industrial Economics, Inc. (2006).
8-12
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
rates that underlie the Census Bureau's published population projections. For a given
model scenario, the model simulates the U.S. population by single year age group and
gender for each year through 2050.
To estimate changes in the labor force with the population simulation model, the Project
Team employed the following three-step approach:
1. CAAA-related change in population: First, the Project Team entered changes in
PM2 5 concentrations into the population simulation model based on the air
quality modeling analysis described in Chapter 4. Netting the model results from
baseline (with-CAAA) population projections, the Project Team estimated PM-
related changes in population by gender and single-year age group for both the
2010 and 2020 target years (and for every other year in the model time horizon).
These changes represent the estimated difference in population between the with-
CAAA and without-CAAA scenarios.
2. CAAA-related change in the labor force: To estimate the change in the labor
force associated with the CAAA, the Project Team applied age- and gender-
specific labor force participation rates from the Bureau of Labor Statistics to the
changes in population estimated in Step 1.
3. Percent Change in Labor Force: The Project Team estimated the percent change
in the labor force associated with pollution-related mortality by dividing the total
labor force changes estimated in Step 2 by baseline (with-CAAA) projections of
the total labor force. As indicated above, the Project Team assumes that this
percent change applies to the full time endowment (labor and leisure time) for the
labor force.
Morbidity-related Labor Force Impacts
Similar to pollution-related mortality, pollution-related morbidity was incorporated into
EMPAX-CGE as a percent change in the labor and leisure time available to workers.
Unlike the Project Team's PM-based approach for mortality, the approach for morbidity
accounts for both PM- and ozone-related impacts. The literature for the various PM and
ozone endpoints examined use several different metrics for quantifying labor force
impacts. To standardize these estimates, we converted the values obtained from the
literature to the number of work days lost per case, by endpoint. We then applied these
values to the yearly changes in the number of cases for each endpoint to estimate the total
work days lost for any given year. These values reflect the labor force participation rate
among those individuals afflicted by each health effect. Because the time endowment in
EMPAX-CGE measures time on an annual basis, we converted the estimated number of
work days lost to lost work years, based on an assumed work year of 235 work days.114
To express work years lost as a percent change in the labor force, we divided the
estimated work years lost for each target year by the projected size of the labor force.
The resulting value represents the percent change in workers' labor time.
IM This estimate is consistent with that used in Jorgenson et al. (2004).
8-13
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
As suggested above, estimating the number of work days lost per case for each endpoint
is a key step in the Project Team's methodology. Table 8-3 summarizes these endpoint
values for both PM and ozone. With the exception of chronic bronchitis and acute
myocardial infarction (AMI), the estimates presented in Table 8-3 were applied to the
annual change in incidence for each endpoint (i.e., the change in the number of new cases
per year), as the duration of disease for most endpoints is no more than several weeks.
Chronic bronchitis and AMI, however, affect individuals over multi-year time horizons.
We therefore apply the work loss day estimates for these endpoints to changes in the
prevalence of each disease (i.e., the change in the number of people with the disease,
relative to the baseline).
Medical Expenditures
To estimate the medical expenditures associated with changes in PM and ozone
concentrations, the Project Team relied upon cost-of-illness estimates from the published
literature. Table 8-4 presents the annual medical expenditures per case for those
endpoints for which medical expenditure data were available. We applied the estimates
presented in the table to the respective annual changes in incidence for each endpoint,
except for chronic bronchitis and AMI. For these two endpoints, we applied the values
from Table 8-4 to estimated changes in prevalence.
Summary of Benefit-Related Inputs
Table 8-5 summarizes the estimated changes in the labor force (i.e., the worker time
endowment) associated with the Amendments for the 2010 and 2020 target years. Using
the estimates in the table, the Project Team modified the time endowment for each model
household included in EMPAX-CGE. The estimates in the table suggest that the U.S.
OO
labor force would be 0.34 percent smaller in 2010 and 0.57 percent smaller in 2020 if the
Amendments had not been enacted. PM mortality effects would make up more than half
of this reduction. Among morbidity endpoints, AMI and chronic bronchitis would have
the most significant effect. The labor force impact of ozone pollution would represent
less than five percent of the reduction in the labor force for each target year.
Table 8-6 presents the estimated change in pollution-related medical expenditures
associated with the Amendments. As indicated in the table, the Project Team estimates
that medical expenditures related to air pollution would be approximately $12.9 billion
higher in 2010 and $21 billion higher in 2020 in the absence of the Amendments. Similar
to the labor force effects summarized in Table 8-5, PM-related morbidity, AMI in
particular, represents most of the estimated change in pollution-related medical
expenditures.
8-14
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 8-3. WORK DAYS LOST PER CASE, BY MORBIDITY ENDPOINT1
PM2
Acute Myocardial Infarction
Chronic Bronchitis
Hospital Admissions, Cardiovascular4
Hospital Admissions, Respiratory4
Emergency Room Visits, Respiratory
Work Loss Days
Age <25: N/A
Age 25-34: 17.7 days
Age 35-44: 14.5 days
Age 45-54: 23.7 days
Age 55-65: 137.0 days
Ago 65: 0 days
Age <25: N/A
Age 25-34: 50.3 days
Age 35-44: 42.2 days
Age 0-14: N/A
Age 15-44: 18.3 days
Age 45-54: 55.5 days
Age 55-65: 73.5 days
Age >65: 0 days
Age 45-64: 17.9 days
Age >64: 7.0 days
Age 0-14: N/A
Age 15-44: 30.7 days
Age 45-64: 30.1 days
Age >64: 7.5 days
Average across all age groups: 0.2 days
Average among working age population: 1 day
Ozone
School Loss Days7
Worker Productivity
Hospital Admissions, Respiratory9'10
Emergency Room Visits, Respiratory
Average across all age groups: 0.7 days
Not applicable
Age <2: 0 days
Age >64: 7.5 days
Average across all age groups: 0.2 days
Notes:
N/A indicates that the underlying C-R function does not provide incidence estimates for that age group.
1. Except for chronic bronchitis and acute myocardial infarction, the number of work days lost is applied to the
change in annual incidence. For chronic bronchitis and acute myocardial infarction, the work days lost presented
in this table are applied to annual changes in the prevalence of each disease.
2. We did not generate separate work loss day estimates for the following PM health endpoints discussed in Chapter 5:
acute bronchitis, acute respiratory symptoms, asthma exacerbation, lower respiratory symptoms, and upper
respiratory symptoms. The lost work days associated with these endpoints are already reflected in the work loss
day endpoint included in this table.
3. Derived from Cropper and Krupnick (1999).
4. Agency for Healthcare Research and Quality (2000), as cited in BenAAAP user's guide, Abt Associates (2008).
5. We assume that each E.R. visit equals one day of lost work time per worker affected. The estimate of 0.2 days per
case reflects the percentage of cases realized by the working-age population, the ratio of workdays to total days in
a year (235/365), and the percent of the working-age population in the labor force.
6. We did not estimate the number of work days lost per case of acute respiratory symptoms associated with ozone
exposure.
7. Derived from Abt Associates (2008). Note that 0.7 is the estimated average work loss days per school loss day,
incorporating work-force participation rates for caregivers.
8. The benefits analysis presented in Chapter 5 does not estimate the number of cases for the worker productivity
endpoint. Instead, worker productivity is estimated as the change in income associated with changes in ozone
concentrations. We estimated the work days lost per dollar of income lost based on the average daily wages of
outdoor workers.
9. Derived from Abt Associates (2008).
10. The dose-response function for ozone-related respiratory hospital admissions does not cover populations older than
two years old and younger than 65. For this endpoint we do not address potential caregiver time lost for incidence
in either cohort.
8-15
-------
TABLE 8-4.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
ANNUAL MEDICAL EXPENDITURES PER CASE, BY MORBIDITY ENDPOINT (2006$)1
2010
PM2
Acute Myocardial Infarction3
Chronic Bronchitis4
Emergency Room Visits, Respiratory5
Hospital Admissions, Cardiovascular6
Hospital Admissions, Respiratory6
$17,600
$715
2020
$17,300
$810
$369
$27,
$21,
Ozone7
Emergency Room Visits, Respiratory5
Hospital Admissions, Respiratory6
400
000
$369
$16,400
Notes:
$17,100
1 . Except for chronic bronchitis and acute myocardial infarction, medical expenditures per
case are applied to the change in annual incidence. For chronic bronchitis and acute
myocardial infarction, medical expenditures per case are applied to the annual changes in
the prevalence of each disease, to generate an annual rather than lifetime estimate of
costs for these chronic diseases.
2. Medical expenditure estimates for the following PM morbidity endpoints were not readily
available: acute bronchitis, acute respiratory symptoms, asthma exacerbation, lower
respiratory symptoms, upper respiratory symptoms, and work loss days.
3. Derived from Wittels et al. (1990) and Russell et al. (1998), both as cited
(2008).
4. Cropper and Krupnick (1999).
in Abt Associates
5. We assume that each E.R. visit equals one day of lost work time per worker affected. The
estimate of 0.2 days per case reflects the percentage of cases realized by the working-age
population, the ratio of workdays to total days in a year (235/365), and the percent of the
working-age population in the labor force.
6. Agency for Healthcare Research and Quality (2000), as cited in Abt Associates (2008).
7. Medical expenditure estimates for the following ozone morbidity endpoints were not
readily available: minor restricted activity days, school loss days, and outdoor worker
productivity.
8-16
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
TABLE 8-5. ANNUAL CHANGE IN LABOR FORCE DUE TO CAAA-RELATED CHANGES IN AIR
QUALITY (PERCENT CHANGE IN WORKER TIME ENDOWMENT)
Pollution-related Change in Worker Time Endowment
PM Mortality Subtotal
PM Morbidity Subtotal
Acute Myocardial Infarction
Chronic Bronchitis
Emergency Room Visits, Respiratory
Hospital Admissions, Cardiovascular
Hospital Admissions, Respiratory
Work Loss Days
Ozone Morbidity Subtotal
Emergency Room Visits, Respiratory
Hospital Admissions, Respiratory
Acute Respiratory Symptoms
School Loss Days
Worker Productivity
2010
0.34%
0.18%
0.15%
0.06%
0.05%
<0.01%
<0.01%
<0.01%
0.04%
0.01%
<0.01%
<0.01%
<0.01%
0.01%
<0.01%
2020
0.57%
0.31%
0.25%
0.09%
0.11%
<0.01%
<0.01%
<0.01%
0.05%
0.02%
<0.01%
<0.01%
<0.01%
0.01%
0.01%
TABLE 8-6.
CAAA-RELATED CHANGES IN ANNUAL MEDICAL EXPENDITURES (MILLION 2006$)
Pollution-related Change in Medical Expenditures
PM Morbidity Subtotal
Acute Myocardial Infarction
Chronic Bronchitis
Emergency Room Visits, Respiratory
Hospital Admissions, Cardiovascular
Hospital Admissions, Respiratory
Ozone Morbidity Subtotal
Emergency Room Visits, Respiratory
Hospital Admissions, Respiratory
2010
$11,900
$11,600
$9,500
$375
$29
$1,228
$467
$310
$2
$311
2020
$19,600
$19,000
$15,500
$919
$39
$1 ,900
$683
$580
$4
$575
EMPAX-CGE MODEL RESULTS
Using the inputs summarized in the previous section, the Project Team estimated the
macroeconomic impacts of the Amendments under both the cost-only case and the labor
force-adjusted case. As described above, the former captures the general equilibrium
effects of CAAA compliance expenditures, whereas the latter accounts for the impacts of
8-17
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
these expenditures as well as the labor force and medical expenditure impacts associated
with the Amendments. We present the results of both analyses below.
MACROECONOMIC IMPACTS OF CAAA COMPLIANCE EXPENDITURES
Table 8-7 summarizes the results of the EMPAX-CGE cost-only model run. As the
results in the table indicate, the Project Team estimates that the compliance expenditures
associated with the Amendments will reduce GDP and consumption by approximately 0.5
percent in 2010 and 2020, relative to the without-CAAA scenario. The total estimated
GDP reduction of $79 billion in 2010 and $110 billion in 2020 are 50 to 70 percent larger
than the total primary cost estimates of $53 billion in 2010 and $65 billion in 2020. The
difference is attributable to secondary effects of compliance costs on the overall
economy, a large portion of which are likely the result of increases in energy prices,
which has broad effects on overall production. Another factor is that investment in
pollution control capital can divert capital from the purpose of enhancing long-term
productivity within the industrial sector.
The percent reduction in equivalent variation is smaller than the corresponding reductions
in GDP and consumption, at approximately 0.4 percent for both target years. This
disconnect between the percent reduction in EV and the reductions in GDP and
consumption suggests that, under the with-CAAA scenario, households allocate a greater
share of their time endowment to leisure (and less to labor) than under the without-CAAA
scenario. This increase in leisure partially offsets the welfare loss associated with
reduced consumption.
TABLE 8-7. SUMMARY OF ANNUAL MACROECONOMIC IMPACTS: COST-ONLY CASE1
VARIABLE
GDP
Consumption
Hicksian EV
(annual)
MODEL RUN
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Change ($ billion)
% change
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Change ($ billion)
% change
Change ($ billion)
% change
2010
$15,027
$15,106
-$79
-0.52%
$10,969
$11,023
-$54
-0.49%
-$54
-0.38%
2015
$17,338
$17,430
-$93
-0.53%
$12,699
$12,761
-$62
-0.49%
-$62
-0.38%
2020
$20,202
$20,312
-$110
-0.54%
$14,881
$14,956
-$75
-0.50%
-$75
-0.39%
Notes:
1 . Results are expressed in year 2006 dollars.
Figure 8-5 presents the percent change in output by industry as estimated by EMPAX-
CGE for the year 2020. The values in the table are typically highest among those
8-18
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
industries with the most significant CAAA compliance expenditures relative to baseline
industry revenue. For example, the electricity industry accounts for approximately 20
percent of CAAA compliance expenditures (approximately $14 billion, or 3.3 percent of
benchmark electricity revenue); as a result, EMPAX-CGE estimates that output from the
electricity industry declines by just less than 4 percent under the with-CAAA scenario
relative to a U.S. economy without Clean Air Act programs. Because the power industry
is the largest consumer of coal in the U.S., the reduction in output from the electricity
industry also results in the secondary effect of reducing coal output by approximately 1.5
percent. The electricity industry's CAAA compliance expenditures also leads to higher
electricity prices that prompt energy-intensive industries to switch to other energy sources
(e.g., natural gas and oil) and/or seek energy efficiency improvements in their production
process. In addition, because of CAAA requirements for cleaner (more expensive) fuels,
petroleum sector output is projected to decline approximately 1.5 percent. The results in
Figure 8-5 also suggest that the other minerals sector experiences the largest reduction in
output, in proportional terms, among all industries (over 5 percent). This reflects the
industry's high compliance expenditures relative to its size and the industry's energy-
intensive production processes.
The industry-level results presented in Figure 8-5 also reflect the extent to which
economic activity associated with CAAA compliance, such as new purchases of
environmental protection goods and services, may partially offset the output losses
associated with CAAA compliance expenditures. As a result of the CAAA, the demand
for environmental protection goods and services will be higher relative to a U.S. economy
without the Amendments.
MACROECONOMIC IMPACTS OF CAAA COMPLIANCE EXPENDITURES AND HUMAN
HEALTH BENEFITS
Building upon the results presented above, Table 8-8 summarizes the results of the
EMPAX-CGE analysis for the labor force-adjusted case, which captures the full CAAA
compliance expenditures as well as the labor force and medical expenditure benefits of
the Amendments. The results presented in the table suggest that over time, the positive
macroeconomic impacts of CAAA-related labor force and medical expenditure impacts
slightly outweigh the negative macroeconomic effects of CAAA compliance costs.115
For 2010, the results for the labor force-adjusted case show a reduction in GDP and
consumption relative to the without-CAAA scenario, but the corresponding changes
become positive in 2020. This largely reflects the rapid growth in the CAAA labor force
effect between 2010 and 2020 (67 percent) relative to the growth in CAAA compliance
expenditures (25 percent). We expect the CAAA-related labor force effect to grow more
quickly than CAAA compliance expenditures during this period because, unlike
compliance expenditures, the labor force effect is cumulative for the health endpoints
with the most significant effect on the size of the labor force (i.e., premature mortality,
115 The EMPAX model results do not isolate the impact of the labor force effect on GDP or the impact of changes in medical
expenditures, as the two were modeled simultaneously.
8-19
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
chronic bronchitis, and AMI). In addition, the mortality effect is delayed relative to the
time costs are incurred to reduce exposures because of the impact of the cessation lag.116
FIGURE 8-5. PERCENT CHANGE IN INDUSTRY OUTPUT IN 2020: COST-ONLY CASE
-10.0%
-8.0%
Percentage Change with Clean Air Act
-6.0% -4.0% -2.0% 0.0%
2.0%
Coal
Crude Oil
Electricity
Natural Gas
Petroleum
Agriculture
Mining (other)
Construction
Food
Textiles & Apparel
Lumber
Pulp & Paper
Printing
Chemicals
Plastics & Rubber
Glass
Cement
Other Minerals
Iron & Steel
Aluminum
Other Primary Metals
Fabricated Metal Products
Machinery & Equipment
Computer Equipment
Electronic Equipment
Transportation Equipment
Miscellaneous Manufacturing
Wholesale & Retail Trade
Transportation Services
Information Services
Finance & Real Estate
Business Services
Education
Health Services
Other Services
116 Note that results for the labor force-adjusted case for years after 2020 indicate that the beneficial effects on the
economy grow over time, through 2030, from $5 billion in 2020 to $14 billion in 2025 to $24 billion in 2030. EMPAX results
for 2030, however, are considered less reliable because of the greater uncertainty in forecasting GDP and industry-level
productivity 20 years into the future.
8-20
-------
TABLE 8-8.
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
SUMMARY OF ANNUAL MACROECONOMIC IMPACTS: LABOR FORCE-ADJUSTED CASE1
VARIABLE
GDP
Consumption
Hicksian EV
(annual)
MODEL RUN
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Change ($ billion)
% change
With Clean Air Act ($ billion)
Without Clean Air Act ($ billion)
Change ($ billion)
% change
Change ($ billion)
% change
2010
$15,027
$15,059
-$32
-0.21%
$10,969
$10,972
-$3
-0.03%
$11
0.08%
2015
$17,338
$17,350
-$12
-0.07%
$12,699
$12,696
$3
0.02%
$22
0.13%
2020
$20,202
$20,197
$5
0.02%
$14,881
$14,876
$5
0.03%
$29
0.15%
Notes:
1 . Results are expressed in year 2006 dollars.
The results in Table 8-8 also suggest that the Amendments lead to an increase in
household welfare, measured as the change in EV, under the labor force-adjusted case for
both the 2010 and 2020 target years. The projected 0.8 percent increase in welfare for
2010 stands in contrast to the projected 0.21 percent reduction in GDP for that year and
the 0.03 percent reduction in consumption. The fact that welfare rises while economic
output declines indicates that, under the with-CAAA scenario, households allocate a
greater share of their time endowment to leisure (and less to labor) than under the
without-CAAA scenario. This reallocation of household time also occurs under the cost-
only case, but it only partially offsets the negative welfare impact of reduced
consumption. Under the labor force-adjusted case, the increase in leisure more than
offsets the welfare loss associated with reduced consumption.
Figure 8-6 presents, by industry, the estimated percent change in output in 2020 for the
labor force-adjusted case. The results in the figure indicate that, when labor force and
medical expenditure impacts are accounted for, the CAAA leads to increased output for
many industries and a decline in output for others. Consistent with the cost-only results,
output in the computer equipment industry increases. The other sectors projected to
experience an increase in output include many industries that tend to be labor-intensive
and would benefit from a larger labor pool, such as most service industries. Output for
health services declines, however, due to the reduction in health services demand
associated with CAAA-related health improvements. Most of the other industries
experiencing a reduction in output are either energy producers (e.g., electricity) or
industries with energy-intensive production processes (e.g., iron and steel).
8-21
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
FIGURE 8-6. PERCENT CHANGE IN INDUSTRY OUTPUT IN 2020: LABOR FORCE-ADJUSTED CASE
-10.0%
Percentage Change with Clean Air Act
i.0% -6.0% -4.0% -2.0% 0.0%
2.0%
Coal
Crude Oil
Electricity
Natural Gas
Petroleum
Agriculture
Mining (other)
Construction
Food
Textiles & Apparel
Lumber
Pulp & Paper
Printing
Chemicals
Plastics & Rubber
Glass
Cement
Other Minerals
Iron & Steel
Aluminum
Other Primary Metals
Fabricated Metal Products
Machinery & Equipment
Computer Equipment
Electronic Equipment
Transportation Equipment
Miscellaneous Manufacturing
Wholesale & Retail Trade
Transportation Services
Information Services
Finance & Real Estate
Business Services
Education
Health Services
Other Services
The results presented in Table 8-8 and Figure 8-6 show that the conclusions drawn from
macroeconomic analyses of air pollution policy depend significantly on whether such
analyses capture both the costs and at least some of the benefits of air policy. While the
results of the EMPAX-CGE cost-only case suggest that the CAAA reduces the output of
the U.S. economy by approximately 0.5 percent each year, the labor force-adjusted case
shows that analyzing benefits in conjunction with costs can either reduce the magnitude
or change the sign of model results. Therefore, general equilibrium analyses that
8-22
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
examine cost-side macroeconomic impacts but ignore or overlook the impacts of policy-
related labor force and health improvements may yield incomplete results that misinform
policymakers and the public. The analysis presented in this chapter illustrates the
feasibility of avoiding this outcome by examining both the costs and (a portion of the)
benefits of air policy in a general equilibrium framework. It is important to note,
however, that assessing expenditure-based output impacts should not replace the current
practice of estimating the welfare (i.e., willingness-to-pay) benefits of avoided health
effects. Unlike willingness-to-pay estimates, the results of CGE models do not reflect the
non-market value that people place on avoided adverse health impacts. The outputs of
such models represent a supplement to willingness-to-pay estimates rather than a
substitute for such estimates.
Further work is needed, however, to reflect a much broader set of benefits in CGE
models. As noted earlier, the results in this chapter are designed to supplement, but not
replace, the more complete primary estimates of benefits and costs. The CGE model
represents flows of products, labor, and capital between and among producers and
consumers, but it excludes improvements in well-being due to enhanced longevity and
health, except to the extent that these increase time available for labor and leisure among
the workforce and reduce some medical costs. As a result, the vast majority of monetized
benefits, many but not all of which represent benefits that are not traded in markets,
cannot currently be reflected in a CGE model. This is the main reason that the beneficial
results to the economy estimated in this chapter are substantially smaller than the primary
estimate of benefits based on willingness to pay estimates. It is nonetheless important to
realize that even the partial set of benefits-related impacts that are reflected in this chapter
(i.e., labor force and medical expenditure impacts) more than compensate for the market
costs we estimate to achieve CAAA compliance.
ANALYTIC LIMITATIONS
While the analysis presented in this chapter provides a reasonable approximation of the
macroeconomic impacts of the CAAA, we note the following limitations:
• Exclusion of labor force and leisure effects for individuals outside the formal
economy: Given the uncertainty surrounding the macroeconomic impacts of retirees,
children, and other populations who do not participate in formal labor markets and
the fact that CGE models are ill-suited to address these uncertainties, the inputs
developed by the Project Team for this analysis did not reflect changes in the time
endowment for these individuals. To the extent that people outside formal labor
markets contribute to the economy, we may underestimate the positive
macroeconomic impacts of the Amendments.
• Exclusion of ozone mortality. As described in the methods section, our analysis
captures PM-related changes in mortality but does not account for mortality impacts
from ozone exposure. Therefore, we likely underestimate the positive
macroeconomic impacts of the Amendments.
8-23
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
• Exclusion ofnonmarket and some market benefits: Our assessment of the
macroeconomic impacts of the Amendments also excludes several other CAAA-
related benefits that may improve economic performance or consumer welfare, such
as visibility improvements, productivity enhancements in the agriculture and forestry
sectors, reduced materials damage, and reduced pain and suffering from pollution-
related illness. Because we do not capture these effects, we very likely underestimate
the positive macroeconomic impacts of the Amendments.
• Assumption of separable benefits categories: Our modeling assumes labor supply
and environmental quality are separable components of the utility function for
households. This separability does not always hold, however; for example, cleaner
air may encourage leisure activities such as birding and fishing, making air quality a
complement to leisure. Prior work suggests that assuming separability may affect
benefits by up to 30 percent in some cases.117
• Perfect foresight: EMPAX-CGE assumes that households have perfect foresight of
future changes in policy and modify their current economic behavior accordingly. In
reality, households often have imperfect information of future policy changes.
Whether the assumption of perfect foresight leads to overestimation or
underestimation of impacts is uncertain.
• EMPAX-CGE parameter uncertainty. Similar to other CGE models, EMPAX-CGE
requires the specification of several model parameters (e.g., elasticity values).
Although the model relies upon credible values from the literature, the range of
published estimates for many parameters varies widely across studies. It is uncertain
whether the parameters included in EMPAX lead to overestimation or
underestimation of impacts.
117
We are grateful to the SAB Council for sharing this observation. For further information, see, for example, J.C. Carbone
and V.K. Smith. 2008. Evaluating policy interventions with general equilibrium externalities. J. Public Econ. 92:1254-1274.
8-24
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
REFERENCES
Abt Associates. (2009). Modeled Attainment Test Software User's Manual. Report
prepared for Brian Timmon, US EPA Office of Air Quality Planning and Standards,
Research Triangle Park, NC. March.
Abt Associates (2008). Environmental Benefits Mapping and Analysis Program
(BenMAP) User's Manual. Report Prepared for Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC,
Neal Fann, Project Manager.
Alberini, A., Cropper, M., Krupnick, A., Simon, N.B. (2004). "Does the Value of a
Statistical Life Vary with Age and Health Status? Evidence from the U.S. and
Canada". Journal of Environmental Economics and Management 48(1): 769-792.
Aldy, Joseph E. and W. Kip Viscusi. (2007). "Age Differences in the Value of Statistical
Life: Revealed Preference Evidence." Rev Environ Econ Policy 1: 241-260.
Barbera, A.J. and McConnell, V.D. (1986) "Effects of Pollution Control on Industry
Productivity: A Factor Demand Approach." The Journal of Industrial Economics.
Vol. XXXV, 161-172.
Barbera, A.J. and McConnell, V.D. (1990) "The Impact of Environmental Regulations on
Industry Productivity: Direct and Indirect Effects." Journal of Environmental
Economics and Management. Vol. 18, 50-65.
Bell, M. L., F. Dominici and J. M. Samet, 2005. A meta-analysis of time-series studies of
ozone and mortality with comparison to the national morbidity, mortality, and air
pollution study. Epidemiology. Vol. 16 (4): 436-45.
Bell, M.L., et al., 2004. Ozone and short-term mortality in 95 US urban communities,
1987-2000. JAMA. 292(19): 2372-8.
Byun, D. W. and J. K. S. Ching. (1999). "Science Algorithms of the EPA Models-3
Community Multiscale Air Quality (CMAQ) Modeling System." U.S. EPA Office of
Research and Development, Washington, D.C. (EPA/600/R-99/030).
Carbone, J.C. and V.K. Smith. (2008). "Evaluating policy interventions with general
equilibrium externalities." Journal of Public Economics. 92:1254-1274.
Carlson, Curtis P., Dallas Burtraw, Maureen Cropper, and Karen Palmer, (2000) "SO2
Control by Electric Utilities: What are the Gains from Trade?" Journal of Political
Economy, 108 (6): 1292-1326.
Chestnut, L.G., RD. Rowe, and W.S. Breffle. (2004). Economic Valuation of Mortality
Risk Reduction: Stated Preference Approach in Canada. Report prepared for Paul
De Civita, Health Canada by Stratus Consulting Inc., Boulder, CO, December.
Committee on Vehicle Emission Inspection and Maintenance Programs, Board on
Environmental Studies and Toxicology, Transportation Research Board, National
R-1
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Research Council. (2001). Evaluating Vehicle Emissions Inspection and
Maintenance Programs.
Crawford, S. 2009. Personal communication on September 4, 2009 and September 15,
2009. New York State Department of Environmental Conservation, Division of
Lands and Forests, Forest Utilization Program.
Crump, K.S. (1994). Risk of benzene-induced leukemia: A sensitivity analysis of the
Pliofilm Cohort with additional follow-up and new exposure estimates. Journal of
Toxicology and Environmental Health 42:219-242.
DeShazo, J.R., and T.A. Cameron. (2004). "Mortality and Morbidity Risk Reduction: An
Empirical Life-Cycle Model of Demand with Two Types of Age Effects."
Unpublished paper, Department of Policy Studies, University of California at Los
Angeles.
Driscoll, Charles T. et al. (2003). Chemical Response of Lakes in the Adirondack
Region of New York to Declines in Acidic Deposition. Environmental Science and
Technology 37(10): 2036-2042.
Driscoll, Charles T. et. al. (2001). Acidic Deposition in the Northeastern United States:
Sources and Inputs, Ecosystem Effects, and Management Strategies. BioScience
51(3): 180-198.
Dutton, John M. and Annie Thomas, (1984). "Treating Progress Functions as a
Managerial Opportunity," Academy of Management Review, 9(2): 235-247.
Eftim, S.E., J.M. Samet, H. Janes, A. McDermott, and F. Dominici. 2008. Fine particulate
matter and mortality: a comparison of the six cities and American Cancer Society
cohorts with a medicare cohort. Epidemiology 19(2):209-216.
Ellerman, A. Denny. (2003). "Ex Post Evaluation of Tradable Permits: The U.S. SO2
Cap-and-Trade Program," MIT CEEPR Working Paper WP-2003-003, available at:
web.mit.edu/ceepr/www/publications/workingpapers_2000_2004.html#2003.
Ellerman, A. Denny, Paul L. Joskow, Richard Schmalensee, Juan-Pablo Montero, and
Elizabeth Bailey (2000). Markets for Clean Air: The U.S. Acid Rain Program.
Cambridge University Press.
Epple, Dennis, Linda Argote, and Rukmini Devadas, (1991). "Organizational Learning
Curves: A Method for Investigating Intra-plant Transfer of Knowledge Acquired
Through Learning by Doing," Organizational Science, 2(1); February 1991.
Gray, W.B. and Shadbegian, R.J. (1994) "Pollution Abatement Costs, Regulation, and
Plant-Level Productivity." Center for Economic Studies.
Hammitt, James K. (2007). "Valuing Changes in Mortality Risk: Lives Saved Versus
Life Years Saved." Rev Environ Econ Policy 1: 228-240.
R-2
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Harrington, Winston, Robert D. Morgenstern, and Peter Nelson (1999), "On the Accuracy
of Regulatory Cost Estimates," Resources for the Future Discussion Paper 99-18,
January 1999.
Huang, Y., F. Dominici and M. L. Bell, 2005. Bayesian hierarchical distributed lag
models for summer ozone exposure and cardio-respiratory mortality.
Environmetrics. Vol. 16: 547-562.
Hubbell, Brian J., Richard V. Crume, Dale M. Evarts and Jeff M. Cohen. (2010).
"Regulation and Progress under the 1990 Clean Air Act Amendments" Review of
Environmental Economics and Policy 4(1): 122-138.
ICF International. (2007). The HAPEM6 User's Guide: Hazardous Air Pollutant
Exposure Model, Version 6. Prepared for the Office of Air Quality Planning and
Standards, US EPA, Research Triangle Park, NC.
Industrial Economics, Inc. (2006). Expanded Expert Judgment Study of the
Concentration-Response Relationship Between PM2.5 Exposure and Mortality.
Prepared for the Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, September.
Industrial Economics, Inc. (2009). Section 812 Prospective Study of the Benefits and
Costs of the Clean Air Act: Air Toxics Case Study - Health Benefits of Benzene
Reductions in Houston, 1990-2020. Prepared for the Office of Policy Analysis and
Review, U.S. Environmental Protection Agency, Washington, DC. July.
International Energy Agency, (2000). "Experience Curves for Energy Technology
Policy"
Ito, K., S. F. De Leon and M. Lippmann, 2005. Associations between ozone and daily
mortality: analysis and meta-analysis. Epidemiology. Vol. 16 (4): 446-57.
Jerrett, M., R.T. Burnett, et al. (2005). Spatial analysis of air pollution and mortality in
Los Angeles. Epidemiology 16(6): 1-10.
Jerrett, M.J., RT. Burnett, C. Arden Pope III, K. Ito, G. Thurston, D. Krewski, Y. Shi, E.
Calle, M. Thun. (2009). Long-Term Ozone Exposure and Mortality. New England
Journal of'Medicine 360: 1085-95.
Johansson, P.-O. (2002). "On the definition and age dependency of the value of a
statistical life." Journal of Risk and Uncertainty. 25: 251-63.
Jones-Lee, M.W. (1992). "Paternalistic Altruism and the Value of Statistical Life." The
Economic Journal. 102:80-90. January.
Jorgenson, Dale W., Richard J. Goettle, Brian H. Kurd, and Joel B. Smith, et al. (2004).
U.S. Market Consequences of Global Climate Change, prepared forthe Pew Center
on Global Climate Change, April 2004.
R-3
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Joskow, Paul L. and Nancy L. Rose, (1985) "The Effects of Technological Change,
Experience, and Environmental Regulation on the Construction Cost of Coal-
Burning Generating Units," RAND Journal of Economics, 16(1): 1-27.
Kniesner, Thomas J., W. Kip Viscusi, and James P. Ziliak, (2010) "Policy relevant
heterogeneity in the value of statistical life: New evidence from panel data quantile
regressions," Journal of Risk and Uncertainty 40:15-31.
Krewski, D., M. Jerrett, R.T. Burnett, et al. (2009). Extended follow-up and spatial
analysis of the American Cancer Society study linking particulate air pollution and
mortality. Res Rep Health Efflnst. 140: 5-114.
Krewski, D., R.T. Burnett, et al. (2000). Reanalysis of the Harvard Six Cities Study and
the American Cancer Society Study of Particulate Air Pollution and Mortality.
Special Report to the Health Effects Institute, Boston MA, July.
Krupnick, Alan. (2007). "Mortality-risk Valuation and Age: Stated Preference
Evidence." Rev Environ Econ Policy 1: 261-282.
Krupnick, A.J. and M.L. Cropper (1992). The effect of information on health risk
valuations. Journal of Risk and Uncertainty 5(1): 29-48.
Kunzli, N., S. Medina, et al. (2001). Assessment of deaths attributable to air pollution:
should we use risk estimates based on time series or cohort studies? American
Journal of Epidemiology 153(11): 1050-1055.
Laden, F., J. Schwartz, et al. (2006). Reduction in Fine Particulate Air Pollution and
Mortality: Extended Follow-up of the Harvard Six Cities Study. American Journal
of Respiratory and Critical Care Medicine 173: 667-672.
Lee, E.H. and W.E. Hogsett. (1996). Methodology for Calculating Inputs for Ozone
Secondary Standard Benefits Analysis: Part II. Prepared for the U.S. EPA, Office of
Air Quality Planning and Standards, Air Quality Strategies and Standards Division.
Levy, J. I., S. M. Chemerynski and J. A. Sarnat, 2005. Ozone exposure and mortality: an
empiric bayes metaregression analysis. Epidemiology. Vol. 16 (4): 458-68.
McLaughlin, D. (1998). A decade of forest tree monitoring in Canada: evidence of air
pollution effects. Environ. Rev. 6(3-4): 151-171
Magat, W.A., Viscusi, W.K., et al. (1996). A reference lottery metric for valuing health.
Management Science 42:1118-1130.
Morgenstern, R.D., Pizer, W.A., and Shih, J-S. (1998) "The Cost of Environmental
Protection." Discussion Paper 98-36. Resources for the Future.
National Research Council (NRC) (2008). Estimating Mortality Risk Reduction and
Economic Benefits from Controlling Ozone Air Pollution. The National Academies
Press, Washington, DC.
R-4
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
New York State Department of Environmental Conservation. (2009). Stumpage Price
Report (Winter 2009/#74). New York State Department of Environmental
Conservation, Division of Lands and Forests, Forest Utilization Program. Albany,
New York.
Pope, C. A., R. T. Burnett, et al. (2002). Lung cancer, cardiopulmonary mortality, and
long-term exposure to fine particulate air pollution. Journal of the American Medical
Association 287(9): 1132-1141.
Puett, R.C., J. Schwartz, J.E. Hart, J.D. Yanosky, F.E. Speizer, H. Suh, C.J. Paciorek,
L.M. Neas, and F. Laden. 2008. Chronic particulate exposure, mortality, and
coronary heart disease in the nurses' health study. Am JEpidemiol. 168(10): 1161-
1168.
Puett, R.C., J. Hart, J.D. Yanosky, C. Paciorek, J. Schwartz, H. Suh, F.E. Speizer, and F.
Laden. 2009. Chronic Fine and Coarse Particulate Exposure, Mortality, and
Coronary Heart Disease in the Nurses' Health Study. Environmental Health
Perspectives 117:1697-1701.
Rea, A., J. Lynch, R. White, G. Tennant, J. Phelan and N. Possiel. 2009. Critical Loads as
a Policy Tool: Highlights of the NOx/SOx Secondary National Ambient Air Quality
Standard Review. Slide 6: Nationwide Total Reactive Nitrogen Deposition (2002).
Available online at: http://nadp.sws.uiuc.edu/meetings/fall2009/post/session4.html.
Rowlatt, Penelope, Michael Spackman, Sion Jones, Michael Jones-Lee, and Graham
Loonies. (1998). Valuation of Deaths from Air Pollution. For the Department of
Environment, Transport and the Regions and the Department of Trade and Industry.
February.
RTI International. (2008). EMPAX-CGE Model Documentation, prepared for U.S. EPA
Office of Air Quality Planning and Standards, March 2008.
Samet, J. M., F. Dominici, et al. (2000). The National Morbidity, Mortality, and Air
Pollution Study Part I: Methods and Methodologic Issues. Research Report 94,
Health Effects Institute, Boston, MA.
Schwartz, J., (2005). How sensitive is the association between ozone and daily deaths to
control for temperaturelAmJRespir Crit Care Med. Vol. 171 (6): 627-31.
Schwartz, J., B. Coull, F. Laden, L. Ryan. (2008). The effect of dose and timing of dose
on the association between airborne particles and survival. Environmental Health
Perspectives 116(1): 64-9.
U.S. Department of Agriculture, Forest Service, Rocky Mountain Region. (2000)
Screening Methodology for Calculating ANC Change to High Elevation Lakes:
User's Guide. January 2000.
U.S. Department of Energy, Energy Information Administration, (2005). Annual Energy
Outlook 2005.
R-5
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
U.S. Environmental Protection Agency (2006). Regulatory Impact Analysis for the 2006
National Ambient Air Quality Standards for Particle Pollution.
http://www.epa.gov/ttn/ecas/ria.html
U.S. Environmental Protection Agency (2004). User's Guide for the AMS/EPA
Regulatory Model - AERMOD. Office of Air Quality Planning and Standards,
Research Triangle Park, NC, Report No. EPA-454/B-03-001.
http://www.epa.gov/scram001/7thconf/aermod/aermodugb.pdf
U.S. Environmental Protection Agency (2003). Response of surface water chemistry to
the Clean Air Act Amendments of 1990. EPA 620/R-03/001.
U.S. Environmental Protection Agency (2000). Guidelines for Preparing Economic
Analyses, EPA 240-R-00-003, September.
U.S. Environmental Protection Agency (1999). Cost of Illness Handbook.
http://www.epa.gOv/oppt/coi/pubs/I _l.pdf.
U.S. Environmental Protection Agency (1998). Carcinogenic Effects of Benzene: An
Update. Office of Research and Development, Washington, DC. EPA/600/P-
97/00 IF.
U.S. Environmental Protection Agency, Science Advisory Board (2007). "Benefits and
Costs of Clean Air Act - Direct Costs and Uncertainty Analysis", Advisory Letter,
June 8, 2007. Washington, DC. EPA-SAB-COUNCIL-ADV-07-002, Available at:
http://www.epa.gov/sab/pdf/council-07-002.pdf
U.S. Environmental Protection Agency, Science Advisory Board (2004). "Advisory on
Plans for Health Effects Analysis in the Analytical Plan for EPA's Second
Prospective Analysis—Benefits and Costs of the Clean Air Act, 1990-2020:
Advisory by the Health Effects Subcommittee of the Advisory Council on Clean Air
Compliance Analysis." Washington, DC. EPA-SAB-COUNCIL-ADV-04-002.
U.S. Environmental Protection Agency, Science Advisory Board (2001). "Review Of The
Draft Analytical Plan For EPA's Second Prospective Analysis - Benefits And Costs
Of The Clean Air Act 1990-2020". Washington, D.C. EPA-SAB-COUNCIL-ADV-
01-004.
The U.S. National Acid Precipitation Assessment Program. 1991. Integrated Assessment
Report. The NAPAP Office of the Director, Washington, DC.
U.S. Office of Management and Budget (2003). Regulatory Analysis. OMB Circular A-
4, September 17. http://www.whitehouse.gov/OMB/circulars/a004/a-4.pdf.
Viscusi, W.K. (1992). Fatal Tradeoffs, (Oxford University Press: New York), Table 4-1.
Viscusi, W. K. and J. E. Aldy. 2003. The Value of a Statistical Life: A Critical Review of
Market Estimates throughout the World. AEI-Brookings Joint Center for Regulatory
Studies. Washington, DC. January.
R-6
-------
The Benefits and Costs of the Clean Air Actfron 1990 to 2020
Viscusi, W.K. (2004). The value of life: Estimates with risks by occupation and industry.
Economic Inquiry 42(1): 29-48.
Wilson, J. H., M. A. Mullen, A. D. Bollman, K. B. Thesing, M. Salhotra, F. Divita, J.
Neumann, J. C. Price, and J. DeMocker. (2008). Emissions projections for the U.S.
Environmental Protection Agency Section 812 second prospective Clean Air Act
cost/benefit analysis. J. of Air & Waste Manage. Assoc. 58:657-672.
Woods & Poole Economics Inc., 2007. Complete Demographic Database. Washington,
DC. http://woodsandpoole.com/index.php.
Zeger, S.L., F. Dominici, A. McDermott, and J.M. Samet. 2008. Mortality in the
Medicare population and chronic exposure to fine particulate air pollution in urban
centers (2000-2005). Environ Health Perspect 116(12): 1614-1619.
R-7
------- |