REGULATORY IMPACT ANALYSIS
ON
THE NATIONAL AMBIENT AIR QUALITY STANDARDS
FOR
PARTICULATE MATTER
FEBRUARY 21, 1984
Prepared by
Strategies and Air Standards Division
Office of Air, Noise and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, N.C.
-------
TABLE OF CONTENTS
Page
Executive Summary i
I. Introduction 1-1
II. Statement of Need & Consequences II-l
III. Alternatives Examined III-l
IV. Cost & Environmental Impacts IV-1
V. Economic Impact Analysis V-l
VI. Benefits Analysis VI-1
VII. Cost/Benefit Analysis VII-1
VIII. Rationale for Choosing the Proposed Action VIII-1
IX. Statutory Authority IX-1
Addendum A-l
-------
i
Executive Summary
Regulatory Impact Analysis, on
The National Ambient Air Quality Standards
for Particulate Matter
I. Background
The particulate matter (PM) Regulatory Impact Analysis (RIA) was
prepared to fulfill the requirements of Executive Order 12291 (E.O. 12291).
The RIA attempts to quantify and inform the public of the costs, benefits
and economic impacts of various levels of current and revised PM National
Ambient Air Quality Standards (NAAQS). The Clean Air Act requires that
ambient standards be based on scientific criteria relating to the level of
air quality needed to protect public health and welfare. The Act and
subsequent court decisions preclude consideration of cost and technological
feasibility in determining the level of ambient standards. Therefore, EPA
has not considered the results of the RIA in selecting the ranges of standards
that are being proposed. It should be noted at the outset that in conducting
this analysis EPA had to overcome a number of data and analytical problems.
Some of the analytic assumptions can have a major impact on the final results.
Such assumptions are noted in this Executive Summary and discussed in more
detail in the text. The Agency invites public comment on this RIA and its
supporting analyses.
11. Statement of Need and Consequences
The Clean Air Act as amended sets out requirements and provides authority
for the listing of certain ambient air pollutants which may endanger public
health or welfare and for the setting and revising of NAAQS. Particulate
matter represents a broad class of diverse substances that exist as discrete
particles and are emitted to the atmosphere from a wide variety of sources.
It also can be formed in the atmosphere from gaseous pollutants. At elevated
concentrations PM can adversely affect human health and welfare.
-------
11
The need for regulatory action arises from the failure of the market
system to deal effectively with PM. Most emitters treat the atmosphere as
a free good and dispose of unwanted by-products by venting them to the
outdoor air. In the atmosphere PM can cause real costs to be
incurred by others. This is generally known as a negative externality.
III. Alternatives Examined
Pursuant to E.O. 12291 the following alternatives are examined:
a) No regulation
b) Regulations beyond the scope of present legislation
c) Market oriented alternatives
d) Alternative stringency levels and implementation schedules.
The first three alternatives were not found to be promising principally
due to statutory constraints. Therefore, in light of E.O. 12291 language
which requires that only the most promising alternatives be analyzed in
detail, the RIA focuses on stringency levels. The following alternatives
are examined:
Table I
PM Alternatives Analyzed*
Annual Arithmetic Mean 24-hour Expected 2nd Maximum Value
55
55 ng/m3 150
55 ng/m3 200 iag/m3
55 ng/m3 250
70 ng/m3 250
48 ng/m3 183
PM^Q refers to particles with an aerodynamic diameter less than or equal
to a nominal 10 \an.
-------
The current Total Suspended Particulate (TSP)2 standards are also examined.
The current primary standards are 75 ng/m^ annual geometric mean and
260 ng/m^ 24-hour observed 2nd maximum. The current secondary is 150 ng/m^
24-hour observed 2nd maximum.
IV. Cost and Environmental Impacts
The first steps in the RIA process are the estimation of future air
quality, the calculation of the air quality improvement needed to attain
the current or alternative PM standards, and the estimation of the cost of
improving air quality (over and above current control programs). This air
quality and cost analysis is county specific and includes some 1230 counties
where ambient PM data are available. These counties account for some 70
percent of nationwide point source emissions and some 82 percent of the
U.S. population. A summarized description of the process is as follows:
1) Particulate matter air quality data in the form of a single TSP
design value were obtained for each county. The design values nominally
represent the highest observed levels in a county and reflect 1978 conditions.
To represent air quality for PMig, a 0.55 conversion factor was applied to
the TSP data. While the use of a single design value simplifies the analysis,
it can be a poor measure of actual concentrations across an entire county.
Moreover, since these analyses were completed, it has been found that the
mean PM \Q/1SP ratio is closer to 0.46 but there is considerable variation in
the ratio both temporally and spatially. Limited sensitivity analyses have
been performed which suggest that the new ratio will reduce both the costs
and benefits of the alternative standards under consideration.
2TSP refers to particles measured in the ambient air by the method described
in 40 CFR Part 50, Appendix B which effectively collects particles up to
25-44 ^ aerodynamic diameter.
-------
IV
2) For each county where air quality data were available an inventory
of point and area source emissions was developed. The emissions inventory
is also nominally representative of 1978 conditions. PM^g emissions were
derived from the TSP data on a source category specific basis using a very
limited emissions testing data base and supplemented with engineering
estimates. As with the air quality data, the PMjg conversion of TSP emissions
is a source of uncertainty.
3) Future air quality was estimated by projecting emissions growth
and using a linear rollback model to calculate the impact on air quality.
The use of rollback presumes that each source in the county contributes to
the design value proportional to its emissions and inversely to its stack
height. The use of this simplified modeling technique was dictated by the
need to analyze a large number of areas.
4) For each county projected to be in non-attainment in step #3 a
source by source control strategy was developed. A list of control options
and associated costs was developed for each source and the air quality
improvement which could be expected from each option was estimated using
the rollback model. In developing a county wide strategy, source control
options were selected on a least cost basis (i.e., lowest dollar per
microgram of air quality improvement) until the standard under consideration
was met.
5) For roughly half of the areas initially in non-attainment (where
the inventory was incomplete or too few control options were available) the
control strategy did not result in full attainment. To estimate the additional
cost of full attainment, the cost of the strategy was multiplied by the
national ratio of the remaining air quality improvement needed to the air
quality improvement already achieved. This assumes that the cost of the
-------
remaining improvement (per unit of air quality) is comparable to the unit
cost of the improvement already achieved. However, if the residual
non-attainment areas include a number of areas with severe air quality
problems—areas, where the States have been striving to achieve improvements
in air quality for the past decade-cost estimation may substantially
understate the actual costs of attainment.
The preceding steps are described in more detail in Section IV.B of
the full RIA. Limitations imposed by the data bases and by various analytic
assumptions are discussed in Sections IV.C and 0 respectively. The most
important assumptions (e.g. the use of ratios to estimate PM^Q data, the
use of a simple roll-back model, and particularly the treatment of residual
non-attainment) are noted above and can have a very substantial impact on
the final results. The two major data bases used (i.e. SAROAD for air
quality and NEDS for emissions) presented problems in terms of data
completeness and quality. Although the estimates of non-attainment and
cost presented are thought to capture the relative differences among
various standard levels, they should be used and interpreted with care.
Section IV.E. of the RIA presents the detailed results of the cost
analysis for the nation and for specific industries and regions. Table II
below summarizes the national cost of full attainment for the alternatives
examined and Table III summarizes regional results.
-------
vi
Scenario?
Table II
Total Estimated Nationwide Costs1
Full Attainment
($ billions)
Capital
Annual Cost^
Discounted Present
Value4
PM10(70, 250)/89
PM10(55,-)/89
PM10(55,250)/89
PM10(55, 200)789
PM10{55, 150)789
PM1Qb(48, 183)789
2.2
5.8
5.9
5.8
8.2
7.8
0.4
1.0
1.0
1.0
1.4
1.3
1.4
3.8
3.9
3.8
5.3
5.0
TSP(75/260)/89
TSP(-,150)789
TSP(75,260)787
TSP(-,150)787
6.0
14.0
6.6
15.0
1.0
2.3
1.1
2.4
3.9
9.0
5.1
11.0
were calculated in 1980 dollars and do not include the cost of: 1) pre-
1979 controls and 2) New Source controls tied to meeting NSPS or PSD requirements.
2Key: TSP (x,y)/z - x=annual standard, y=24-hour standard, z=Attainment Year.
•^Annual costs include operation and maintenance costs and annual ized capital
charges. Annual capital charges were derived using an assumed 15-year equipment
life and a 10% real discount rate.
4Discounted Present Value represents the summation of the stream of annual
O&M and capital payments discounted back to 1982 using a 10% real discount rate.
5This PM10 alternative approximates the current primary TSP standards. The
values were derived from the TSP values by applying the regression equations
to estimate missing values in the air quality file and by applying the
conversion ratio of 0.55. The costs are lower than 55, 150
alternative because the 24-hour standards are controlling in most counties.
used
-------
Table III. Regional Discounted Present Value (UPV)Costs and Attainment Status1
1982 Discounted Present Value (106$) and
Scenario^
TSP(75,260)/87
TSP{75.150)/87
TSP( 75.2601/89
PM10(70, 250)/89
PM10(55. 250)/89
PM10(55, 200)789
DPV3
23
89
17
0
1
1
I II
NAA4 DPV3 NAA4
4/2 41 5/1
20/11 101 13/9
4/2 33 5/1
0/0 <1 1/0
2/0 3 2/0
3/0 6 3/0
1) Costs calculated In 1980 dollars and do
2) Key: TSP (x,y)/Z - x=annual standard.
3) Discounted Present Value represents the
discounted back to 1982 using a 10% real
4) Non-attainment areas (NAA) entries give
III IV V
DPV3 NAA4 DPV3 NAA4 DPV3 NAA4
294 24/9 265 39/16 827 70/28
384 45/21 477 88/42 1406 134/71
229 24/7 213 39/15 660 70/28
13 6/1 77 7/2 355 23/6
117 11/1 256 15/10 595 40/15
118 14/2 258 15/11 609 44/16
not Include the cost of: 1) pre-1979
y=24-hour standard, Z=atta1nment year.
summation of the stream of annual O&M
discount rate.
(Initial NAA)/(Res
-------
viii
V. Economic Impact Analysis
The economic impact of the control costs were assessed. It should
be noted that the methodology and procedures used in the cost and control
strategy analysis impose limitations on the economic analysis. The most
stringent PM standard analyzed resulted in the imposition of costs on
some 280 industries. Since it was infeasible to complete an economic
analysis of all affected industries, a screening analysis was performed
to identify those industries which might experience relatively large
adverse economic impacts. On the basis of the screen, ten industries
were selected for further analysis. Six others were added because of
the absolute magnitude of their costs, their vulnerability to competition
or because of significant control costs imposed by other environmental
regulations. The sixteen industries are listed below in Table III.
Specific control costs by industry and for each standard are reported in
Section IV.E. in the full RIA.
-------
1x
Table IV
Description SIC
Steel 3312
Primary smelting of copper 3331
Primary smelting of lead 3332
Primary smelting of aluminum 3334
Electric utilities, steam supply 4911, 4961
Cement, hydraulic 3241
Crushed and broken limestone 1422
Crushed and broken stone, NEC 1429
Construction sand and grvel 1442
Paving mixtures and blocks 2951
Cut stone and stone products 3281
Minerals and earths 3295
Grain and flour milling 2041
Gray iron foundries 3321
Lime 3274
Government sector
The analysis for each industry begins with a descriptive profile.
The profile is used to make an assessment of market structure, trends in
output and demand, employment, capacity, etc. Because of the uncertainty
regarding the characteristics (i.e., size, production, ownership, mix of
products, etc.) of the affected plants, "model plants" were used. Each
model plant has a balance sheet and income statement, developed from
industry data. Costs of the present PM NAAQSs are applied to the income
statement and balance sheet so that the incremental effect of the regulation
can be estimated. It is important to note that the analysis was conducted
-------
X
using the estimated costs for the TSP (-.150)87 partial attainment case.
All other NAAQSs considered in the RIA require less ambient air quality
improvement and result in lower costs, suggesting smaller economic impacts
than the base case.
Section V.F. presents the findings of the analysis. For 10 of the
16 industries, absorption of the control costs with no price changes or
net adjustment to production seems the most likely outcome. Firms in the
remaining industries are judged able to pass through a significant portion
of the control costs resulting in price increases ranging from 0.6* to
7.0* and net decreases in production no greater than 1%. Although the
outlook varies from industry to industry, the alternatives examined seem
unlikely to result in any major change in production or structure.
Under the Regulatory Flexibility Act, 5 U.S.C., 600 et seq., the
Agency must prepare a regulatory flexibility analysis assessing the impact
of any proposed or final rule on small entities.
EPA has made an effort to address the cost of control for small
entity groups. This effort is, however, limited and does not permit
definite findings with respect to all potentially affected small entities.
The extremely large number of potentially affected emitters of particulate
matter (over 280 different industries at the 4-digit SIC code level) led
to the development of a cost model based on the information from the
National Emission Data System (NEDS). For the particulate matter model,
completion of the economic analyses would have required a plant by plant
matching of NEDS with economic data bases. The large number of plants
and industries in question made this matching of data files, and therefore
a detailed small entity economic analysis, impractical at present.
In an attempt to develop some of the information necessary for a
regulatory flexibility analysis, the screening analysis was used to select
-------
xi
16 industries out of the 280 affected industries for more detailed
analysis. This preliminary assessment based on selected industries does
not suggest that the proposed revisions will significantly affect a
substantial number of small entities for the group of potentially affected
industries. This finding, however, is not conclusive.
VI. Benefit Analysis Estimates
The benefit analysis quantifies some of the economic benefits
attributable to health and welfare improvements of PM reductions. The
reader should be aware that, like the other portions of the RIA, neither
the methodology used nor the results obtained have been subjected to
formal peer review.
The methodology employed involved the use of existing health and
welfare studies to extract information needed to estimate benefits. The
first step was to identify categories of effects and for each category to
screen and classify existing research literature. In the screening of
literature the most important criteria were the analytic quality of the
study and its potential for extrapolation to benefits estimates. For most
categories the screen resulted in one or more studies being selected for
benefits estimation. However, for several important categories of effects
(e.g., most soiling and materials damage in the industrial/commercial sector)
adequate studies could not be identified and the category was dropped. It
should be noted that this screen results in the inclusion and subsequent
use of some studies which the Clean Air Scientific Advisory Committee
(CASAC) found unsuitable for quantitative purposes in standard setting.
For each of the studies selected, benefit estimates are derived based
on the ambient concentrations and the distribution of air quality changes
estimated from the cost and control strategy analysis described above.
The approaches used to estimate air quality and air quality changes may
-------
X11
have introduced an upward bias to national benefits estimates. For those
categories in which several studies could be used in the analysis, the
estimates produced cover a very wide range of values. In addition, in
those categories where only one study existed, several estimates could
usually be derived on the basis of different data sets, different functional
forms, and different threshold assumptions. Consequently, for each scenario
analyzed, the categories examined produce a range of estimates. In each
category, the studies and results are reviewed and a "best" estimate from
the study or studies which most adequately satisfy the evaluation criteria
is selected.
The procedure just discussed is applied on a county-by-county basis
to each area estimated to be in nonattainment during the period analyzed.
In order to go from the range of estimates in a specific benefit category
to a range across categories, it is necessary to aggregate results. Several
factors precluded simple aggregation by summing the data across categories. The
critical factors were whether or not the procedures used result in double counting
of benefits and whether they provide complete coverage of potential benefits.
Double counting exists in part because it is sometimes impossible to separate
benefits associated with unique effects where studies address a range of effects.
Potential lack of complete coverage of benefits results from the fact that the
necessary studies may not exist or may not be adequate to estimate all benefits.
A fundamental assumption underlying the benefits analysis is that
the particulate matter air quality changes which result from the cost analysis
involve the same types of particles which produced the observed effects in
the underlying health and welfare studies. For example, the cost analysis
which was conducted on contemporary U.S. sources would have to result in
-------
xi ii
the control of the same distribution of particles (both size and chemical
composition) as those which produced the effects in the health studies
some of which involved conditions in London in the 1960's. To the extent
that the make-up of the particles controlled are different, the estimated
benefits may be either over-or underestimated depending on how the make-up
varies.
Six different aggregation procedures (A-F) were developed and are
presented in Table V. A detailed discussion of each aggregation
scheme is presented in Section VI.F of the full RIA. Problems arising in
each aggregation scheme make it difficult to provide a precise estimate
of benefits for alternative standards. Nevertheless, the omission of
significant benefit categories in A and B and the potential double counting
in E and F suggest that these aggregation procedures are not likely to
provide best estimates of benefits.
Aggregation procedures A and B are based only on those studies which
have gone through the Criteria Document and CASAC review process and were
found to provide quantitative evidence of effects. However, procedures A
and B are likely to underestimate benefits since they omit a number of
significant benefit categories (such as certain health effects categories,
soiling and materials damage). Aggregation procedure C provides more
complete coverage of benefits categories, but includes a study of acute
morbidity that although it has been peer reviewed and published in a
journal,* it has not undergone formal Criteria Document and CASAC review.
Aggregation procedure D provides still more complete coverage, but in
addition to the acute morbidity study includes studies of mortality effects
*0stro, B.D. (1983) "The Effects of Air Pollution on Work Loss and
Morbidity" Journal of Environmental Economics and Management, December 1983.
-------
XIV
Table V
INCREMENTAL BENEFITS FOR ALTERNATIVE PM10 AND TSP STANDARDS
Full Attainment
($ Billions)
1
Alternative Standard2
PM10 (70,250)/89
PMlQ (55,-)/89
PM10 (55,250)789
PM10 (55,200)789
PM10 (55,150)789
PM104 (48,183)789
TSP (75,260)789
TSP (-,150)789
TSP (75,260)787
TSP (-,150)787
1. 1982 discounted present values 1n bill
the 7-year time horizon Is 1989-95 and
PMlo and TSP standards are 1n terms of
2. Key: PM 9x,y)/z - x=annual standard,
3. For each aggregation procedure/standar
Aggregation Procedure^
A B C D
.37/2.0 1.7/3.5 12/14 15/33
.51/3.1 2.9/5.3 20/24 25/61
.51/3.1 2.9/5.5 21/23 26/60
.51/3.1 2.9/5.7 21/25 26/62
.55/3.5 3.5/6.5 25/29 31/73
.57/3.6 3.7/6.7 27/31 34/78
.58/3.6 3.7/6.7 27/31 34/80
.65/4.2 4.6/8.2 34/38 42/100
.78/5.0 5.0/9.4 39/43 48/114
.87/5.7 6.4/11.4 48/52 59/141
E
30/48
56/92
57/91
58/94
69/111
76/120
77/123
93/147
107/173
131/209
F
42/62
79/117
79/117
81/119
98/142
106/154
107/153
137/183
151/229
191/269
Ions of 1980 dollars at a 10 percent discount rate.
the 9-year horizon Is 1987-95. Comparisons between
TSP stringency, not particle size.
y=24-hour standard, z=atta1nment year.
d combination two values are reported.
The first 1s
based
on a $0.36 for a unit reduction of 1.0 x 10~6 In annual mortality risk while the second 1s based
on $2.80 for the same unit reduction.
4. This PMio alternative approximates the current primary TSP standards. The PM^g values were derived
from the TSP values by applying the regression equations used to estimate missing values 1n the air
quality file and by applying the PM^o/TSP conversion ratio of 0.55. The costs are lower than 55, 150
alternative because the 24-hour standards are controlling in most counties.
-------
XV
(Lave and Seskin, Lipfert, etc.) some of which CASAC found were not refined
enough to provide quantitative concentration/response estimates. Procedures
E and F include all benefits categories for which underlying studies were
available. It should be stressed, however, that a number of the studies
overlap each other. Therefore, it is likely that procedures E and F
"double count" some benefits (e.g. mortality effects), resulting in an
overestimate of the total for the benefits categories covered. It is
important to note again that, due to the absence of adequate underlying
studies, some potential benefit categories were omitted altogether in all
aggregation schemes.
The full RIA and its technical appendices describe the benefits analysis
in more detail and lay out the strengths and limitations of the approach
used. In the full RIA, Section VI.C describes the air quality data used,
while Section VI.D outlines the key analytic assumptions. The results of
the analysis are reported in Section VI.6.
VII. Benefit-Cost Analysis
The benefit-cost analysis evaluates the alternative standards in terms
of economic efficiency. Comparisons are made on the basis of incremental
benefits and incremental costs (i.e., the benefits and costs of attaining
an alternative standard starting from "current" air quality).
The analysis of the incremental benefits and costs associated with the
alternative PM NAAQS is conducted in the following manner:
0 The estimated incremental benefits associated with each alternative
PM NAAQS are computed.
0 Similarly, the estimated incremental costs associated with each
alternative standard are computed.
0 The estimated net incremental benefits (i.e., incremental benefits
minus incremental costs) generated by each alternative standard
are computed.
0 The estimated net incremental benefits for each of the several
alternative standards are compared.
-------
XVI
Any alternative PM NAAQS that is associated with higher incremental costs,
but the same or smaller incremental benefits than some other standard or is
associated with the same costs and lower benefits, is cost-ineffective and
is not evaluated further in terms of efficiency. Any alternative PM NAAQS
that produces positive net incremental benefits will provide a more efficient
allocation of resources than would occur under the baseline air quality
scenario. The PM NAAQS that produces the largest positive net benefits
will produce the most efficient allocation of resources among the standards
considered. When net incremental benefits are negative for all considered
alternatives, no standard is identified as efficient. When the net incremental
benefits associated with a standard are negative, the baseline air quality
scenario yields a more efficient allocation of resources.
Table VI presents the range of incremental net benefits for full attainment
of the standards in question. It should be noted that before they were
subtracted from the benefits, the costs reported earlier were modified so
that they reflected the same accounting period as the benefits.
It should be noted that any conclusions drawn as to the efficiency
of an alternative standard must be caveated with the limitations of the
methods applied in both the cost and benefits analysis. Distributional
considerations (e.g., those who enjoy the benefits are not always the same
as those who bear the costs) may be important. In addition, it is especially
important to keep in mind that comparisons between PMjo and TSP are based on
TSP stringency. The analysis provides no information on preferred particle
size indicator. Thus, while Table V indicates that under certain circumstances
the TSP 150 standard is preferred on economic efficiency grounds, this
should be interpreted as a preference for the most stringent standard
analyzed. A more complete discussion of this important limitation is to be
found in Section VII.F. of the full RIA.
-------
vi
INCREMENTAL NET BENEFITS FOR ALTERNATIVE PM1Q AND TSP STANDARDS
Full Attainment
($ Billion'^)
1
Alternative Standard^
PM10 (70,250)/89
PM10 (55,-)/89
PM10 (55,250)789*
PM10 (55,200)/89
PM10 { 55,150)/89*
PM]n4 (48,183)789*
TSP (75,260)789
TSP (-.150)789**
TSP (75,260)787
TSP (-.150)787
Aggregation Procedure^
A B C D E
-.58/1.1 .75/2.6 11/13 14/32 29/47
-2.0/.58 .38/2.8 17/21 22/58 53/89
-2. I/. 53 .33/2.9 18/20 23/57 54/88
-2.0/.56 .36/3.2 18/22 23/59 55/91
-3.0/-.03 -.03/3.0 21/25 27/69 65/107
-2.S/.24 .34/3.3 24/28 31/75 73/117
-2.0/.99 1.1/4.1 24/28 31/77 74/120
-5.3/-1.8 -1.4/2.2 28/32 36/94 87/141
-3.2/.9S .98/5.4 35/39 44/110 103/169
-8.1/-3.3 -2.6/2.5 39/43 50/132 122/200
1. 1982 discounted present values In billions of 1980 dollars at a 10 percent discount rate
Time horizons for 7- and 9-year standards are, respectively, 1989-1995 and 1987-1995.
Comparisons between PM^o and TSP standards are in terms of TSP stringency, not particle
2. Key: PM 9x,y)/z - x=annual standard, y=24-hour standard, z=atta1nment year.
3. For each aggregation procedure/standard combination two values are reported. The first
F
41/61
76/114
76/114
78/116
94/138
103/151
104/150
131/177
147/225
182/260
•
size.
is based
on a $0.36 for a unit reduction of 1.0 x 10~b in annual mortality risk while the second 1s based
on $2.80 for the same unit reduction.
4. This PMjo alternative approximates the current primary TSP standards. The PM^g values were derived
from the TSP values by applying the regression equations used to estimate missing values in the air
quality file and by applying the PM^Q/TSP conversion ratio of 0.55. The costs are lower than 55, 150
alternative because the 24-hour standards are controlling in most counties.
*This alternative is dominated by another alternative with the same time horizon. An alternative
is considered to be dominated when another alternative provides the same dollar benefits for
smaller costs or greater dollar benefits for the same or smaller costs.
**This alternative 1s dominated by another alternative with a longer time horizon.
X
<
-------
xviii
VIII. Summary of Rationale for Choosing Proposed Action
In accordance with sections 108 and 109 of the Clean Air Act, EPA
has reviewed and revised the criteria upon which the existing primary and
secondary participate matter standards are based. The existing primary
standards for participate matter (measured as "total suspended particulate
matter" or "TSP") are 260 ng/m3, averaged over a period of 24 hours and
not to be exceeded more than once per year, and 75 ^g/m3 annual geometric
mean. The secondary standard (also measured as TSP) is 150 iig/m3,
averaged over a period of 24 hours, and not to be exceeded more than once
per year.
As a result of its review and revision of the health and welfare
criteria, EPA proposes the following revisions to the particulate matter
standards:
1) that TSP as an indicator for particulate matter be replaced for both
of the primary standards by a new indicator that includes only those particles
with an aerodynamic diameter smaller than or equal to a nominal 10 micrometers
2) that the level of the 24-hour primary standard be changed to a
value to be selected from a range of 150 to 250 jig/m3 and that the
current deterministic form of the standard be replaced with a statistical
form that permits one expected exceedance of the standard level per year;
3) that the level and form of the annual primary standard be changed to
a value to be selected from a range of 50 to 65 |ig/m3, expressed as an
expected annual arithmetic mean; and
4) that the current 24-hour secondary TSP standard be replaced by
an annual TSP standard selected from a range of 70 to 90 ug/m3, expected
annual arithmetic mean.
-------
XIX
Because no scientific consensus exists on the most appropriate levels
of the standards, and the analytical and policy bases for making these
decisions under the statute are limited and unclear, the Administrator is
not proposing specific standard levels within the above ranges. Rather, he
is soliciting additional comment and information from the public to be
considered in promulgating the final regulation, which will specify a
specific level for each of the standards. Given the precautionary nature
of the Act, the Administrator is inclined to select the levels of primary
standards from the lower portion of the above ranges. A more detailed
discussion of the rationale for this proposal is available in the Federal
Register preamble.
IX. Statutory Authority
The authority for the proposed revision of the PM NAAQS is contained
in Sections 108 and 109 of the Clean Air Act.
X. Addendum
Following the completion of the draft RIA and its supporting analyses,
a range of levels for the primary and secondary standards was selected for
proposal which had been only partially addressed in the basic cost and
benefit analyses. As a result additional analyses were made of the upper
and lower bounds of both the primary and secondary standard ranges. These
analyses were based on the same methodology used in the full RIA. However,
in this supplemental work the results are shown using the 0.46 PM^Q/TSP
ratio as well as the 0.55. The net benefits, as estimated in this analysis
are shown in Tables VII and VIII. The limitations and caveats discussed in
the preceding sections also apply here. The order of standards is rarely
changed by the use of the 0.46 ratio. However, when it changes, the result
is a preference toward the more restrictive standard. Like the analysis
-------
XX
contained 1n Section VII, as more comprehensive aggregation procedures
(i.e., C, D, E, F, and sometimes B) are adopted, the more restrictive
standard becomes preferred on economic efficiency grounds.
Other new supporting analyses are also presented. These were air
quality analyses which employed a more recent data base and which used a
statistical approach to account for the variability of the PMio/TSP ambient
ratio. These analyses tend to support the use of the 0.46 ratio and suggest
that the use of the 0.55 ratio may result in an overestimate of both costs
and benefits. However, preliminary analyses suggest this overestimate is
generally not sufficient enough to change the ranking of standards on economic
efficiency grounds. Where it is great enough to change the rankings more
stringent standards are preferred to the 0.46 ratio.
-------
Table VII '-'.-
Estimated Incremental Net Benefits for the Proposed Range of
Using 0.55 and 0.46 TSP to PMjo Ratios
Full Attainment
{$ Billion)
Primary Standards
Aggregation Procedures
Standards/Conditions
D
PM10 (50,150)
Full attainment with lower bound mortality
risk valuation
0.55 PM10 to TSP ratio
0.46 PM10 to TSP ratio
Full attainment with upper bound mortality
risk valuation
0.55 PMio to TSP ratio
0.46 PMio to TSP ratio
-2.87
-1.37
.21
.95
.25
.87
3.33
3.39
23.84
17.95
26.92
20.46
30.11
22.42
76.69
55.20
71.86
49.72
117.44
82.51
102.95
71.33
151.61
106.63
PM10 (65,250)
Full attainment with lower bound mortality
risk valuation
0.55 PMio to TSP ratio
0.46 PMio to TSP ratio
Full attainment with upper bound mortality
risk valuation
0.55 PMio to TSP ratio
0.46 PMio to TSP rat1°
- .60
- .32
1.4
.86
1.0
.53
3.0
1.71
13.25
6.95
15.25
8.13
16.47
8.70
40.33
21.76
35.89
16.29
49.74
29.35
50.92
24.04
76.79
35.30
*1982 discounted present values 1n billions of 1980 dollars at a 10 percent discount rate. The time horizon 1s
the 7 year period starting January 1, 1980 and ending December 31, 1995. The TSP annual arithmetic mean lower
bound of 110 ng/m3 when Imposed is applied to all health studies.
X
X
-------
Table VIII
Estimated Incremental Net Benefits for the Proposed Range of TSP Secondary Standards
Full Attainment
($ Billion)
Aggregation Procedures
Standards/Condi t1 ons
B
D
TSP (70)
Full attainment with lower bound mortality
risk valuation
Full attainment with upper bound mortality
risk valuation
-5.38** -1.03** 31.51** 40.40** 98.87** 144.62**
-1.68** 2.67** 35.20** 104.32** 162.79** 222.20**
TSP (90)
Full attainment with lower bound mortality
risk valuation
Full attainment with upper bound mortality
risk valuation
-1.55
1.41
1.27
4.23
22.89
25.85
28.66
70.72
66.32
94.69
108.38 139.71
X
X
*1982 discounted present values in billions of 1980 dollars at a 10 percent discount rate. The time horizon 1s
the 7 year period starting January 1, 1980 and ending December 31, 1995. The TSP annual arithmetic mean lower
bound of 110 ng/m^ when Imposed is applied to all health studies.
**The alternative standard 1s dominated by an alternative analyzed in Section VII of the RIA but which 1s
outside the proposed range. An alternative 1s dominated when another standard provides the same dollar
benefits for smaller costs; or, greater dollar benefits for the same or smaller costs.
-------
1-1
I. INTRODUCTION
This draft Regulatory Impact Analysis (RIA) has been prepared in
accordance with the requirements of Executive Order 12291. The Executive
Order requires preparation of an RIA for every "major rule." The
Environmental Protection Agency (EPA) has determined that the proposed
revisions to the particulate matter national ambient air quality standards
(NAAQS) are a major action because any changes to the existing standards
currently under consideration may result in an annual effect of $100
million or more on the economy.
As provided for in sections 108 and 109 of the Clean Air Act as amended,
EPA has reviewed and revised the criteria upon which the existing primary
(to protect public health) and secondary (to protect public welfare)
standards are based. As a result of this review and subsequent revisions
to the health and welfare effects criteria, EPA proposes to revise both the
primary and secondary NAAQS for particulate matter.
The Clean Air Act specifically requires that primary and secondary
NAAQS be based on scientific criteria relating to the levels of air quality
that should be attained to protect public health and welfare adequately.
The Act precludes consideration of the cost or technological feasibility
of achieving such standards in determining the level of the ambient
standards. Accordingly, EPA has not considered the results of this draft
RIA in selecting the standards that are being proposed.
This draft RIA examines the impacts of alternative levels of both
primary and secondary standards in terms of the benefits to be derived,
the cost and environmental impacts, and the economic impacts as well as
other potential impacts on both the public and private sectors. In
specifying the alternatives examined EPA took account of the legislative
-------
1-2
requirements affecting the development and revision of NAAQS and the
OAQPS staff paper Review of the National Ambient Air Quality Standards
for Particulate Matter: Assessment of_ Scientific and Technical Information
- OAQPS Staff Paper (EPA-450/5-82-001, January, 1982). The OAQPS staff
paper for particulate matter interprets the most relevant scientific and
technical information reviewed in the revised Air Quality Criteria for
Particulate Matter and Sulfur Oxides (EPA-600/8-82-029a,c; December,
1982). The OAQPS staff paper, which has undergone careful review by the
Clean Air Scientific Advisory Committee (an independent advisory group)
serves to identify those conclusions and uncertainties in the available
scientific literature that should be considered in selecting a particulate
pollutant indicator, form, and level for both the primary and secondary
standards.
-------
II-l
II. Statement of Need and Consequences
This section of the analysis summarizes the statutory requirements
.fi. affecting the development and revision of NAAQS and briefly describes
the nature of particulate pollution and the need for regulatory action
at this time.
A. Legislative Requirements Affecting the Development and
Revision of NAAQS
Sections 108 and 109 of the Clean Air Act as amended provide
authority and guidance for the listing of certain ambient air pollutants
which may endanger public health or welfare and the setting and revising
of NAAQS for those pollutants. Section 108 instructs EPA to document
the scientific bases (health and welfare criteria) for such standards
and section 109 provides requirements for reviewing criteria and
establishing standards.
Primary standards must be based on health effects criteria and be
set to protect public health with an adequate margin of safety. The
legislative history of the Act indicates that the standards should be set
at "the maximum permissible ambient air level. . . which will protect the
health of any (sensitive) group of the population." Also, margins of
safety are to be provided such that the standards will afford "a reasonable
degree of protection. . . against hazards which research has not yet
identified" (U.S. Senate, 1974). In setting the primary standards, the
Administrator of EPA must make a policy judgment regarding the implications
of all the health-effects evidence and decide upon a level that provides
an adequate margin of safety.
Secondary standards must be adequate to protect public welfare from
any known or anticipated adverse effects associated with the presence
-------
II-2
of a listed pollutant in the ambient air. Welfare effects, which are
defined in section 302(h) of the Act, include effects on vegetation,
visibility, water, crops, man-made materials, animals, economic values,
and personal comfort and well-being. In specifying a level for secondary
standards the Administrator must determine at which point these effects
become "adverse" and base his judgment on the welfare effects criteria.
Recent judicial decisions (Lead Industries Association, Inc.vs. EPA
, 1980; American Petroleum Institute vs.EPA , 1981) make clear that the
costs and technological feasibility of attainment are not to be considered
in setting primary or secondary NAAQS. Such factors can be considered to
a limited degree in the development of State plans to implement such
standards. Under section 110 of the Act, the States are to submit to EPA
for approval State Implementation Plans (SIPs) that provide for the
attainment and maintenance of NAAQS by certain deadlines.
Finally, section 109(d) of the Act directs the Administrator to
review existing criteria and standards at 5-year intervals. When warranted
by such review, the Administrator is to revise NAAQS.
B. Nature of the Particulate Matter Problem
Particulate matter represents a broad class of chemically and
physically diverse substances that exist as discrete particles (liquid
droplets or solids) over a wide range of sizes. Anthropogenic sources
of particles include a variety of stationary and mobile sources. Particles
may be emitted directly to the atmosphere or may be formed by transformations
of gaseous emissions such as S02. The major chemical and physical
properties of particulate matter vary greatly with time, region, meteorology
and source category, thus complicating the assessment of health and
welfare effects as related to various indicators of particulate pollution.
-------
II-3
At elevated concentrations, participate matter can adversely affect
human health, visibility, climate, materials, economic values, personal
comfort and well-being, and vegetation. Components of particulate matter
(e.g., sulfuric acid) also contribute to acid deposition.
More specifically, key findings concerning the health effects
associated with particulate pollution, as assessed in the criteria
document for particulate matter and sulfur oxides and the OAQPS staff
paper, can be summarized as follows:
1) Health risks posed by inhaled particles are affected both by
the penetration and deposition of particles in the various regions of
the respiratory tract, and by the biological responses to these deposited
materials.
2) The risks of adverse effects associated with deposition of
ambient fine and coarse particles in the thorax (tracheobronchial and
alveolar regions of the respiratory tract) are markedly greater than for
deposition in the extrathoracic (head) region. Maximum particle penetration
to the thoracic regions occurs during oronasal or mouth breathing.
3) The major effects categories that can be associated with high
exposures to particulate matter include: (a) changes in lung function
and respiratory symptoms, (b) aggravation of existing respiratory and
cardiovascular disease, (c) alterations in the body's defense systems
against foreign materials, (d) damage to lung tissues, (e) carcinogenesis,
and (f) mortality.
4) The major subgroups of the population that appear likely to be
most sensitive to the effects of particulate matter include: (a) individuals
with chronic obstructive pulmonary or cardiovascular disease, (b) individuals
-------
II-4
with influenza, (c) asthmatics, (d) the elderly, (e) children, (f)
smokers, and (g) mouth or oronasal breathers.
5) Review of the available epidemiological studies suggests the
need for both short term (24-hour) and long term (annual) health-based
primary standards in order to prevent adverse health effects from being
experienced by the most sensitive population subgroups.
This information on particle size (developed since EPA promulgated
the current NAAQS for total suspended particulate matter (TSP)) led to
the conclusion that TSP is not the most appropriate pollutant indicator
for the primary standard. TSP, as measured by the hi-volume air sampler,
includes particles up to 45 micrometers (urn) in diameter, the smaller
of which are of greater concern to health. As a result, the Clean Air
Scientific Advisory Committee (CASAC) as well as others within the scientific
community, recommended that the particle pollutant indicator be changed
to one that includes only those particles capable of penetrating the
thoracic region during mouth breathing, specifically those particles less
than or equal to a nominal size of 10 \im. While these particles are
called thoracic particles in the staff paper, they will be referred to as
PM-|Q throughout the remaining discussion of the proposed revisions to the
NAAQS for particulate matter.
With respect to welfare effects, fine particles (<2.5um) have been
clearly associated with the impairment of visibility over urban areas
and large multi-state regions. Fine particles, or major constituents
thereof, also are implicated in climatic effects, materials damage,
soiling and acid deposition. While a relatively well defined quantitative
relationship exists between fine particle mass and visibility impairment,
-------
II-5
regulatory action on fine particles has been deferred until it is possible
to link the need and a standard with a coherent, scientifically based
strategy for these related regional air quality problems.
At high enough levels, both large and small particles may soil
household and other surfaces and become a nuisance. The available data
base on soiling and nuisance effects is largely qualitative and therefore
selecting a standard level is a matter of judgment.
Particulate pollution is a problem affecting localities, both urban
and non-urban, in all regions of the United States. Man-made emissions
that contribute to airborne particulate matter result principally from
stationary point sources (fuel combustion and industrial processes),
industrial process fugitive particulate emission sources, non-industrial
fugitive sources (roadway dust from paved and unpaved roads, wind erosion
from cropland etc.) and transportation sources. In addition to man-made
emissions, consideration must also be given to natural emissions including
dust, sea spray, volcanic emissions, biogenic emanation (e.g. from plants)
and emissions from wild fires when assessing particulate pollution and
devising control strategies.
Control strategies developed by the various States for attaining
the current NAAQS for particulate matter have focused primarily on
stationary point sources and to a lesser extent on the control of industrial
process fugitives and other fugitive sources such as paved and unpaved
road emissions and emissions from storage piles. Additional information
describing the available control techniques and their cost and effectiveness
are contained in the document Control Techniques for Particulate Emissions
from Stationary Sources (EPA-450/3-81-055a,b; September, 1982).
-------
II-6
C. NEED FOR REGULATORY ACTION
In the absence of government regulation, market systems have failed
to deal effectively with air pollution, because airsheds have been
treated as public goods and because most air polluters do not internalize
the full damage caused by their emissions. For an individual firm,
pollution is usually an unusable by-product which can be disposed of at
no cost by venting it to the atmosphere. However, in the atmosphere,
pollution causes real costs to be incurred by others. This is generally
referred to as a negative externality in economic theory.
The fact that the producer, or consumer, whose activity results in
air pollution, does not bear the full costs of his action leads to a
divergence between private costs and social costs. This is referred to
as "market failure" because it causes a misallocation of society's
resources, with more resources being devoted to the polluting activity
than would be if the polluter had to bear the full cost.
There are a variety of market and nonmarket mechanisms available to
correct this situation. Some of the principal market mechanisms are
briefly described in Section III of this RIA ("Alternatives Examined").
Other than regulation, nonmarket approaches would include negotiations
or litigation under tort law and general common law. In theory, these
latter approaches might result in payments to individuals to compensate
them for the damages they incur.
Such resolutions may not occur, however, in the absence of government
intervention. Two major impediments block the correction of pollution
inefficiencies and inequities by the private market. The first is high
transaction costs when millions of individuals are affected by thousands
of polluters. The transaction costs of compensating individuals adversely
-------
II-7
impacted by air pollution Include contacting the Individuals affected,
apportioning Injury to each from each pollution source, and executing the
appropriate damage suits or negotiations. If left to the private market,
each polluter and each affected individual would have to litigate or
negotiate on their own or organize into groups for these purposes. The
transaction costs involved would be so high as probably to exceed the
benefits of the pollution reduction.
The second factor discouraging private sector resolution of the
particulate matter pollution problem is that pollution abatement tends
to be a public good. That Is, after particulate matter pollution has
been abated, benefits of the abatement can be enjoyed by additional
people at no additional cost. This constitutes the classic "free rider"
problem. Any particular individual is reluctant to contribute time or
money to reduce particulate matter emissions knowing that his actions
will have little impact on how clean the air he breathes actually is.
In view of the clear legal requirements placed on EPA by the Clean
Air Act and the market failure discussed above, the Agency 1s proposing
to revise the primary and secondary NAAQS for particulate matter to
provide adequate protection of public health and welfare. As this
regulatory analysis shows, there are resource costs associated with this
governmental intervention (see Section IV, "Cost & Environmental Impacts
Analysis"). However, they are required by the Clean Air Act. In addition,
EPA believes that the cost of this abatement through government action is
less expensive than with any reasonably available private sector alternatives.
Finally, these standards will mitigate the negative externalities which
would otherwise occur due to the failure of the marketplace.
-------
III-l
III. ALTERNATIVES EXAMINED
This section briefly presents potential alternatives to the proposed
revisions of NAAQS for participate matter. The outline for the section
is adopted from Executive Order 12291 which requires that at a minimum
the following alternatives be examined:
a) No regulation
b) Regulations beyond the scope of present legislation
c) Market oriented alternatives
d) Alternative stringency levels and implementation schedules.
Although Executive Order 12291 requires that all alternatives be examined,
only the most promising ones need be analyzed in detail.
A. NO REGULATION
Abandoning current regulatory requirements for particulate matter
would result in a reliance on private efforts to reduce emissions and on
the absorptive capacity of the atmosphere. The most likely avenues for
private efforts would be either negotiation or litigation under tort and
general common law. Generally speaking there is no incentive for a
single company to enter into negotiations with individuals to reduce
particulate matter emissions. For an individual firm, the cost of reducing
emissions would leave that firm at a competitive disadvantage. Litigation
by those damaged could be pursued either to obtain a reduction in emissions
or to obtain payment for damages incurred (or both). The costs of such
litigation would likely be very high since the individual or classes of
individuals bringing suit would have to prove damages. Moreover, there
is little incentive for all those affected by air pollution to join
together in such a suit since everyone would enjoy the benefits of a
successful suit to reduce emissions regardless of the extent of their
participation.
-------
III-2
Because the Clean Air Act requires the Agency to prescribe standards
for pollutants, such as particulate matter, which have adverse effects on public
health and because of the impracticality of private efforts, the option
of no regulation has not been analyzed in any further detail.
B. OTHER REGULATORY APPROACHES
Other regulatory approaches include such options as performance and
technology standards and regional or State air quality standards.
Performance and technology standards are required by the present law in
a variety of forms (e.g. new source performance standards (NSPS) for new
and modified sources, lowest achievable emission rate (LAER) and reasonably
available control technology (RACT) in non-attainment areas, etc.). They
are not based solely on health and welfare criteria but are designed, in
part, to augment control strategies for attainment of the air quality
standards. These standards generally specify allowable emission rates
for specific source categories. The LAER and RACT requirements are
intended to allow growth in non-attainment areas while promoting progress
towards eventual attainment. NSPS help to reduce the likelihood of future
pollution problems by controlling new sources. EPA is required to consider
technology and cost in setting NSPS and RACT requirements.
Performance and technology based standards serve as useful adjuncts
to ambient standards. However, they cannot serve as substitutes for
ambient standards since even perfect compliance with them may not produce
acceptable air quality levels. Despite the application of such standards,
local meteorology, the interaction of multiple sources, and the level of
the standard itself (the standards are set on the basis of technology and
cost) could produce air quality levels that do not protect public health
and welfare.
-------
III-3
Regional or State differences in terrain and meteorology as well as
valuations of clean air have been cited as reasons for adopting regional
or State air quality standards. Variations in terrain and meteorology
are considered in setting SIP emission limits to achieve a NAAQS. Such
variations do not generally change the effect of particular levels of
pollution on public health and welfare. However, in the case of particulate
matter the composition of the pollutant and its effects can vary from city
to city. On the other hand, transport of pollutants across boundaries
would make a system of regional or State air quality standards difficult
to enforce. Moreover, the Clean Air Act requires national not regional
standards. Nonetheless to provide information to the public, EPA is
investigating several analyses to evaluate the potential merits of regional
standards in comparison to national standards. These will be incorporated
in the RIA at the time of promulgation.
In summary, the regulatory alternatives outlined above have not been
analyzed in detail in this draft because of the limitation of present
legislation. However, the performance and technology based standards are
helpful in augmenting control strategies for meeting ambient standards.
The role and cost of such standards are discussed in Section IV. E "New
Source Controls" below.
C. MARKET-ORIENTED ALTERNTIVES
There are several market-oriented approaches which can be considered
as alternatives to achieve the NAAQS for particulate matter, but none
of which are specifically authorized by the Clean Air Act. These approaches
include pollution charges, marketable permits, and subsidies and are
briefly discussed below.
Charges. This policy would involve a charge (or tax) being set on
each unit of a pollutant emitted. Firms would then choose the amount of
-------
III-4
abatement that minimizes their total cost, including the pollution charge.
Pollution is abated until the marginal cost of abatement is equal to the
pollution charge. The regulatory agency would have to set the level of
the charge or charges in a manner that would result in the desired air
quality. This could be quite difficult and might require continued
adjustments to account for inflation and growth.
Permits. A permit system would allow a pollution source to purchase
a permit in order to emit a specified amount of a pollutant over a specified
period of time at a specified location. A fixed number of permits could
be issued and auctioned off to the highest bidders. Alternatively, the
permits can be distributed among the sources, who could then trade these
permits as they see fit. If the number of permits and emission allowances are
correctly fixed in advance, then the desired ambient standard will be
achieved. Again, this may be difficult in areas with numerous and diverse
sources.
Subsidies. A subsidy system pays sources for each unit of pollution
that they do not emit. This can take the form of direct payments or tax
credits. Subsidies and charges are similar in that both increase the
opportunity cost of polluting, the former by causing each unit of pollution
to entail forgoing the subsidy which could be received if it were not
emitted. Thus, the subsidy is similar to a charge, except in two
respects:
(a) Administratively, there is the problem of determining the
actual abatement of each source. There must be a determination
of what its pollution would have been in the absence of the
subsidy, and this determination must be adjusted as conditions
change. These determinations are difficult and may depend
upon information from the polluters, who obviously have the
incentive to strategically misrepresent their intended emissions.
-------
III-5
(b) The long-run effects of subsidies may be quite different than
for permits or charges, since the former increase profit
levels or incomes for polluting industries and the latter
decrease them. Thus, in the long-run, subsidies will result
in firms entering the polluting industry, but permits or
charges will cause firms to leave it. Consequently, a fixed
subsidy rate will yield greater pollution levels than equivalent
permits or charges, although each individual firm will behave
identically under the two systems. Also subsidies would likely
increase dependance on capital intensive solutions.
Summary. Although the Clean Air Act does not permit these market-
oriented alternatives to be considered in setting the NAAQS, it does not
prevent States from using such approaches in attaining air quality standards.
Thus, to the extent they are permitted by EPA's regulations for development
of SIPs, the States may consider such market-oriented alternatives as
described above in the implementation of the NAAQS.
0. REGULATORY ALTERNATIVES WITHIN THE SCOPE OF PRESENT LEGISLATION
The Clean Air Act requires that primary NAAQS be set at levels
which protect the health of sensitive individuals with an adequate margin
of safety. The secondary NAAQS must be adequate to protect public welfare
from any known or anticipated adverse effects.
The assessment of the available quantitative and qualitative health
effects data presented in the criteria document and the OAQPS staff
paper, together with recommendations of CASAC and other public commenters,
suggest a range of alternatives for both short term (24-hour) and long
term (annual) particulate matter standards to protect sensitive individuals
from adverse effects. For a comprehensive discussion of the scientific
data that served as the basis for these alternatives as well as the
rationale for the Administrator's approach to this decision, the reader
-------
III-6
is referred to the OAQPS staff paper and the preamble to the proposed
revisions. Listed below in Table III.D.3 are the alternative primary
standards which are featured in this analysis. No alternative levels of
TSP were analyzed because of the clear recommendation for PM^Q primary
standards expressed in the Staff Paper.
Table III.D.3. PM]Q Alternatives Analyzed
Annual Arithmetic Mean 24-hour Expected 2nd Maximum Value
55 pg/m3 —
55 pg/m3 150 pg/m3
55 pg/m3 200 pg/m3
55 pg/m3 250 pg/m3
70 pg/m3 250 pg/m3
48 pg/m3 183 pg/m3
-------
IV-1
IV. COSTS AND ENVIRONMENTAL IMPACTS
A. INTRODUCTION
This section presents a summary of the direct costs and environmental
impacts'associated with controlling sources of particulate matter air
pollution in order to attain alternative NAAQS. The alternatives
examined correspond to those outlined in Section III D. The analysis of
costs and environmental impacts focused on some 1231 counties and sub-
county areas for which particulate matter ambient air quality data was
available. These counties and sub-county areas account for some 70
percent (prior to screening) of the nationwide point source emissions of
TSP reported in EPA's National Emission Data Systems, and some 82 percent
of the nation's population. Since monitors have generally been placed
in all polluted areas, the counties that lacked air quality data are
generally believed not to have significant particulate matter air quality
problems. Therefore, in the absence of the requisite air quality data,
no attempt has been made to extend the base analysis to include all of
the nation's 3118 counties. The following discussion is divided into
sections on the Methodology (Section IV.B), Data Bases employed (Section IV.C),
Key Analytic Assumptions (Section IV.D), the results of the analysis
(Section IV.E) and Resource Impacts Analysis (Section IV.F). A more
detailed treatment of these subjects is provided in the technical appendices
to the final RIA (Energy and Environmental Analysis, 1981; Smith and Brubaker,
1982). These appendices have not undergone formal peer review.
B. METHODOLOGY
The principal steps in the methodology employed to estimate future
air quality, to calculate the degree to which air quality must be improved
to attain the current (TSP) standards or alternative (PM]g) standards, to
-------
IV-2
develop source-specific control strategies, and to estimate the costs of
bringing all non-attainment counties or sub-county areas into full attainment
are depicted in Figure IV.B.I. It should be noted that cost savings that
might result from the removal of controls in clean areas are not estimated.
A brief description of the principal steps in the process is as follows
(more detailed information is provided in the subsequent discussion):
1) County and sub-county area level estimates of TSP and PM]g
emissions for both point and area sources were developed. TSP
emissions were based on screened and augmented data from EPA's
National Emissions Data Systems (NEDS). The PM]g emissions
were derived from the TSP data base by using a conversion
process which utilized a limited set of size-specific emissions
data and engineering estimates assembled by EPA's Industrial
Environmental Research Laboratory. The conversions were made by
source category. The emissions inventory is nominally representative
of 1978 conditions.
2) TSP Air Quality Data, in the form of a single design value
(24-hour and annual), were obtained for each county or sub-
county area from EPA's Storage and Retrieval of Aerometric
Data System (SAROAD). To insure conformity with the 1979
Regional Office designations of attainment or non-attainment,
design values were adjusted as appropriate. Similar to the
inventory, air quality data nominally represented 1978 conditions.
Estimates of PM]Q air quality data were derived from TSP data by
employing a 0.55 conversion factor. This conversion factor was
based on analysis of data collected by EPA's Inhalable Particulate
Network in 1981 (Pace, 1981). More recent data indicates that a
0.46 factor is more realistic (see section IV.D "Ambient
Ratio" for a fuller discussion).
-------
Figure VI.B.I
Analytical System
•AM Tt»« » AN8 TO
WMT SOUMCI WVIMTOMlll
•AM T|A« »
AMO T»» AMI*
MUMCI MVtNTOACS
7
•AU YIA* » AMD
T«» MHOM VAIUCI
AI0 •*CKO*OUNO
«AtUCI
7
7
•0 CO«T«OL COSTS
ISTO»»
7
J
MOJICT fVTURI COUNTY
•MISSIONS »0«
«•*!. IT. U. M
7
••OWTM AMO
WTMMINT HATfS
7
y
O» §CIN««iO
•TAMOAKOI AND
• *UTUNI AM OUALfTY
PHOJICTIO TO IICIIO TNI
•TANSAROS t
COw»uTt >»v|D AMO
UM»AV(0 MOAO IMISSIOMS
ROLLBACK TO
ISTIMATI IMISSIONS-
AM QUALITY «ILATIONSM»S
roii TMOSI souncts
/
£_
S /
/.
,
CONTROL OWONS /
T TO CONTROL /
CONTROL OTTX)N». COSTS.
OtVtlO* CONSTIIAMZO
UAST-COST STRATIOY
tO* ATTAMMINT
COUNTY. fTATl.
*NO NATIONAL SUMMARIES
COSTS. IMISSiON MDUCTIONS
AND MSDUAL NOMATTAMMtNT.
V
BTIMTI COSTS FOR RESIDUAL
MATTAJWCNT COUNTIES
-------
IV-4
3) Future emissions growth was projected and used to estimate
future air quality by means of a modified linear rollback or
proportional modeling on a county or sub-county basis. The
modeling procedures presumed that all sources of particulate
matter in a county or sub-county area, together with background
contributions, account for air quality in direct proportion to
their emissions and inversely to stack heights. The estimates
of future emissions take into account both entirely new sources
and the replacement of sources on-line in 1978.
4) For county or sub-county areas projected in Step 3 to be in
non-attainment with the current (TSP) standards or the alternatives
(PM]o) in a specified implementation year, a list of sources/control
options and associated emissions was developed for each
option combination. The associated costs were determined on the
basis of a "model plant" and then were applied to all other
facilities of that type with appropriate adjustments to account
for differing sizes and operating parameters. For many process
sources only one control option was determined to be available.
However, for both utility and industrial boilers as many as
three levels of control were estimated. During this process,
certain control options and/or sources were dropped from
further consideration. Control options were dropped if they
were not cost-effective, (i.e. if the cost per ton of emission
reduction was extraordinarily high based on current experience,
see Section IV.D for a more thorough discussion of this),
if they cost more but reduced emissions less than another option, or
if they were less efficient than the controls already in place.
-------
IV-5
Sources were not considered as candidates for control if
options were not available or if there were insufficient data
on file to permit the calculation of costs. As a result, in
some scenarios analyzed, more than one-half of the sources
were never considered as candidates for control.
5) Using modified rollback, (assuming all sources impact air quality
directly proportional to their emissions and inversely proportional
to their effective stack height) an emissions to air quality
relationship was established for all sources within each non-
attainment area. This relationship reflected the impact on air
quality that would result from a change in emissions at a
particular source.
6) For each non-attainment county or sub-county area, a source-
by-source control strategy was developed using the list of
available control options from Step 4. The air quality impact
of applying a control option was estimated by using the established
emissions-to-air quality relationship (Step 5) and the emission
reduction achieved by the option. Selection of the preferred control
strategy was determined by a "least cost" approximation, in terms
of dollars per microgram of air quality improvement (assuming roll-
back), that attempted to minimize total control costs within each
area while driving the non-attaining monitor towards attainment.
Additional control options were applied until the area was
brought into attainment or until the supply of controllable emissions
was exhausted and the area was determined to be in residual
non-attainment status.
-------
IV-6
7) The results of applying these individual control strategies
were used to develop national and regional summaries of cost,
emission reductions achieved, solid waste generated,
and residual non-attainment.
8) The cost of bringing residual non-attainment counties or
sub-county areas into full attainment was then estimated by
multiplying the cost of the strategy by the national average
ratio of the remaining air quality improvement needed to the
air quality improvement already achieved (see Section IV.D for a
fuller discussion). The resulting additional costs were then
added to those costs reported for the base analysis to produce
an estimate of the total cost to bring the full sample of 1231
county and sub-county areas into attainment.
C. DATA BASES
Before the results of this analysis are examined, it is important
to understand how the data bases and analytic assumptions may affect the
projected outcomes. This section briefly examines the air quality data
and the emission inventory used.
Air Quality Data
The TSP air qality data used in this analysis was obtained from
EPA's Storage and Retrieval of Aerometric Data System (SROAD) and is
nominally representative of 1978 conditions. Within each county or sub-
county area, the monitor with the highest annual average for the years
1977 or 1978 (or the two most recent years back to 1975) was selected as
the design value monitor and served as the basis for the annual and 24-
hour design values. Design values for expected ambient levels (24-hour
averaging period) were estimated by EPA's Monitoring and Data Analysis
Division by fitting the annual records from each design value monitor to
-------
IV-7
exponential curves and then reading the expected 2nd maximum value from
the curve. Background values (the air quality contribution from natural sources
and from man-made sources outside the county or sub-county area), were in
most cases defaulted using an average regional value taken from nonurban/rural
monitoring sites.
At the time this analysis was initiated, there were no measured PM]Q
data and only limited PMi5 data from EPA's inhalable particulate network.
PM^o air quality was estimated by applying a 0.55 conversion factor to the
TSP design value data. This factor was derived by analyzing data collected
by EPA's inhalable particulate (PM-js) network (Pace, 1981). Following
completion of the RIA analysis it was found that 0.46 may in fact be a better
point estimate (see Section IV.D). Although it is likely that this
factor may vary from one region of the U.S. to another and from one point
in time to another, analyses of the limited data now available have not
suggested any consistent method for handling the variation. The use of
this ratio approach to PM-|Q air quality introduces some uncertainty.
However, until sufficient ambient measurements are made, this appears to
be the only reasonable approach available for estimating PM-JQ air quality.
A sensitivity analysis on the use of this ratio is reported in Section
IV.0. "Key Analytic Assumptions."
The overall scope of the analysis was constrained by the availability
of valid air quality data. When valid design values could not be
obtained for counties designated in the Federal Register as being in
attainment or "unclassifiable", they were eliminated from the analysis.
In those instances where counties were missing valid data for one of the
averaging periods or statistical forms, defaults were applied and the
county was retained. The default values were obtained through the use of
regression equations relating one averaging period and statistical form
to another. Finally, the design values for each county were compared to
-------
IV-8
the 1979 Federal Register designations of attainment and non-attainment.
Where inconsistencies with the official designation were found, the
design values were adjusted to conform with the Federal Register (i.e.
either by obtaining an alternative value from SAROAD or by arbitarily
setting the design value equal to or one ug/m^ above the standard).
This overall process resulted in design values being obtained for some
1236 counties or sub-county areas.
While the use of design values obtained from the highest monitor
greatly simplifies the analysis, it can be a poor measure of actual
ambient concentrations across an entire county. In many instances,
design values reflect "hot spots" and other monitors in the county can
show much lower values. In general, particulate concentrations reveal
significant variations over short distances. Therefore, when a county is
projected to be in non-attainment it should not be construed that the
entire area or population will be exposed to levels in excess of the
alternative standard being considered. In the context of this analysis,
the use of the highest observed value in the county as the design value
potentially results in an overestimate of the degree of control needed,
since other areas of the county may be below the standard.
It must be noted that the air quality data used here are now at
least 5 years old. In making actual non-attainment decisions once a
standard is set, the States and Regional Offices will use more recent
data. In addition, they will use other types of data such as emission
trends, compliance data, and monitor siting information. Therefore,
actual attainment status may be expected to differ from the estimates
presented in IV.D below. Nonetheless, the estimates below do capture the
relative impacts of various standard levels.
-------
IV-9
Emission Inventory.
The emission inventory was derived from 1978 data in EPA's National
Emission Data System (NEDS). To prepare the inventory for use, the
following steps were taken to screen and augment the nationwide NEDS
inventory.
1) NEDS point source inventory was screened to remove those
emissions not located in "design value counties."* This step
reduced emissions from 7.2 million tons per year
(TRY) to 5.1 million TPY of TSP.
2) To simplify the analysis, the inventory was screened further
to remove emissions from sources that emit less than five TPY.
This step eliminated some 182,000 sources but only reduced
emissions coverage by some 66,000 TPY of TSP and is not expected
to bias final cost estimates.
3) Some 430,000 TPY of TSP non-traditional fugitive emissions
were added to the inventory to account for fugitive
emissions from storage and handling operations at coal
fired industrial and utility boilers, selected process
sources, as well as from inplant traffic (paved and unpaved
roads).
4) The NEDS area source file was used to identify area source
emissions within the sample counties. This resulted in
some 240,000 TPY of TSP area source emission being added
to the base inventory.
*1236 "counties" had available air quality data and served as the basis
for inventory preparation. Subsequently five county or sub-county areas
were dropped, thus, the analysis was based on 1231 county or sub-county
areas as noted earlier.
-------
IV-10
5) For the PM^o alternatives, the final step in inventory
preparation process was to compute emissions of participate
matter less than 10 micrometers for both point and fugitive
sources. This was accomplished by adjusting the TSP emission
inventory. The adjustment of point source emissions accounted
for the primary emissions of PM^g and the level of PM^g control
achieved by the existing particulate control technology. The
adjustments were source category specific and were based on a
limited set of size-specific emissions data assembled by EPA's
Industrial Environmental Research Laboratory and on engineering
judgment. PMig/TSP emission ratios were also used to adjust
fugitive and area source emissions.
-------
IV-11
A summary of the base inventory for both TSP and PM^g is presented
in Table IV.C.I.
Table IV.C.I. 1978 Base Inventory
Source Category
Point Sources
Nontraditional fugitives
Area source emissions
Total
Emissions
TSP
5.04
0.43
0.24*
5.72
(10° TPY)
PM10
3.67
0.23
0.06*
3.96
The base inventory shown in Table IV.C.I was used to project future
emissions. The projected emissions reflected the retirement of old
plants, and the addition of new plants based on replacement and growth
projections. The emission inventory, after growth, was then used to
project future air quality at the county or sub-county level. This, in
turn, determined the future attainment status and whether additional
controls would have to be applied. The projected PM]Q emissions, prior
to the application of control strategies, are shown in Table IV.C.2.
For point sources, the projected emissions show a slow decrease through
1989 followed by an upturn in 1995. The 1995 point source emissions,
however, are still below the base 1978 level. This phenomenon is a
function of the assumptions used in projecting growth. As new sources
at BACT control levels replace existing ones at current levels, emissions
decrease. Eventually, the cumulative effect of growth offsets the new
control technology and emissions begin to increase. Area and non-
*These figures represent the "effective fraction" of area emissions. The
full inventory for TSP area emissions is 24 x 106 TPY and 6 x 106 TPY
for PMio- See Section IV.D., Treatment of Area Sources for full discussion,
-------
IV-12
traditional sources were not assumed to be subject to BACT. Details
concerning growth and BACT assumptions are discussed in Section IV.D
below.
Table IV.C.2. Projected PMig Emissions after Growth and Retirement
(106)
Source Category
Point Sorces
Nontraditional
Area Source Emissions
Total
1982
3.53
0.26
0.06
3.85
1985
3.47
0.29
0.06
3.82
1987
3.44
0.31
0.06
3.82
1989
3.43
0.33
0.06
3.83
1995
3.50
0.42
0.07
3.98
The resulting emissions inventory, although the best currently
available, reflects several short-comings that are inherent in NEDS.
First, the basic NEDS 1978 data may not be representative of actual
conditions for that year since many records are not updated annually.
These sources would then show control levels and emissions for some
earlier year. As a result, the inventory may not accurately reflect the
degree of control actually being achieved in the base year. Second, the
NEDS inventory does not separately identify process fugitive emissions
and therefore they could not be accounted for in developing control
strategies. Directionally, this may result in an understatement of control
costs, particularly in the steel industry (See Sections IV.D and IV.E.3
for a fuller discussion). Third, incomplete and/or inaccurate point
source data resulted in only 3.55 million TPY of TSP emissions (out of
the 5.72 million TPY included in the analysis for purposes of projecting
future air quality) being considered in the control strategy. Again,
-------
IV-13
this introduced an unknown bias to costs for point source control and
tended to make attainment more difficult. Finally, the inventory does
not account for secondarily formed aerosols. As much as 60% of ambient
PMio can be in the fine fraction (<2.5 urn) and the fine fraction itself is
dominated by secondarily formed aerosols. The effect of not including
the secondary contribution is to estimate a greater degree of control
over ambient PM]Q than could be obtained in practice through traditional
strategies.
With respect to area sources, because of their low release height
and small zone of influence, only a fraction were considered capable of
influencing the design value. Among area sources only paved road emissions
were considered as candidates for control. Controls for other area
sources (e.g. home heating, construction, airports, etc.) were either
non-existent or beyond the scope of this analysis.
The exclusion of large fractions of both point and area source
emissions from the control strategies would tend to increase residual
non-attainment in the base analysis. At the major industrial category
level, except for the steel industry where the omission of process
fugitive emissions clearly biased cost downward in comparison to other
available studies, it is not clear in which direction the limitations of
the emission inventory may have influenced control costs.
D. KEY ANALYTICAL ASSUMPTIONS
As in the case of any analysis of this scope, a number of assumptions
had to be made that could potentially affect the results. In this
instance, some of the assumptions are inherent in the methodology and
the structure of the model used, while others relate to input variables.
-------
IV-14
Roll-Back.
To determine future air quality and the effect of emission reductions,
a modified roll-back model was used. The use of roll-back, modified to
account for effective stack height, assumes that all emission sources
within a county or subcounty area contribute proportionally to the air
quality design value. The approach does not take into account actual
source/receptor relationships. Thus, sources regardless of their distance
from the design value monitor are included when establishing the emissions
to air quality relationships used to project future air quality and in
the subsequent control strategies. Assuming equivalent stack heights
this means that distant sources are treated the same as nearby sources.
This can result in overprediction of future air quality concentrations,
in the application of controls on distant sources that may have little or
no impact, and in an indeterminant bias in the costs of complying with
alternative standards. Roll-back was used in spite of its inherent
limitations because it is the most practical means of assessing the
impacts associated with the alternative standards in some 1231 counties.
More sophisticated source/receptor modeling approaches require much more
accurate inventories, detailed meteorological data, and make more intensive
use of the computer. Such approaches are used by States in designing
SIPs for well defined areas, usually less than a county. However, given
the available data bases and geographic scope of this analysis, it was
determined that such models were impractical.
"Least-Cost" Control Strategy.
In implementing a NAAQS, States have relied on detailed modeling
studies and have attempted to strike a balance among a variety of competing
social, economic, and environmental goals. That balance often reflects
-------
IV-15
local conditions and there is considerable variation from State to
State. To replicate the decision criteria of each State was clearly an
infeasible task. In the absence of other criteria, it was assumed in this
analysis that each State would devise and implement a control strategy which
minimized control costs. Least cost analysis is also required for a correct
cost/benefit analysis that attempts to measure true resource costs. As a
result, the estimates of control costs would tend to be lower than those
based on other decision criteria.
The "least cost" approximation used in this model chooses control
options on a cost per ug/m-* of air quality improvement basis (i.e.
the lowest cost per wg/m^ improvement option will be chosen first).
This was determined to be the most rational approach that could be
handled within the scope of the study. When coupled with modified roll-
back (which assumes all sources impact the design value), the least-cost
control strategy tends to focus the application of controls on area
sources and nontraditional fugitive emissions, since they have lower
control costs per unit of air quality improvement. This has the effect
of lowering estimated costs further still. In recognition of this, area
sources were handled differently in the roll-back modeling as is discussed
below under "Treatment of Area Sources."
It should be noted that the strategy selected only approximates a
true "least-cost" solution. A true "least-cost" solution would consider
all possible combinations of options which would meet the NAAQS and
select the least-cost combination. The model, on the other hand, relies
on a ranking of discrete options and does not consider all possible
combinations of options. A true "least cost" solution would also take
into account individual source/receptor relationships.
-------
IV-16
Air Quality Maintenance.
Under the provisions of the Clean Air Act, once a county or subcounty
attains the level of the standards, additional controls may need to be
imposed in future years to maintain the required air quality level.
Such air quality maintenance controls and costs occur as the result of
emissions growth. To estimate these costs, an "accomodative SIP" approach
was employed in this study. That is, emissions were grown out to 1995
and the resulting air quality level was estimated. Then the controls
necessary to reduce the projected air quality level to the standard
level were put in place in the implementation year (e.g. 1989). This in
effect brings about more control than is necessary to attain the standard
in the implementation year, thereby accommodating future growth. The
additional costs incurred in the implementation year thus serve as a
crude surrogate for air quality maintenance costs A sensitivity analysis
showed a 14% drop in initial non-attainment and a 16% drop in base costs
(DPV) if growth were stopped in the implementation year (1989). Given the
assmptions made here this is equivalent to saying that approximately
16% of the costs reported below are due to air quality maintenance.
Ambient PMm/TSP Ratio
As discussed previously, PM^o a"ir quality was estimated in this
analysis through the application of an estimated PMiQ/TSP ratio of 0.55.
Since the analysis was initially run, EPA has collected additional ambient
PMiQ data and currently estimates the national average ratio to be 0.46.
Although resources were not available to re-run the entire analysis,
several standards were re-run employing the new ratio. In comparison to runs
using the old ratio, the estimated cost (DPV) dropped between 55 and 60%
and the initial non-attainment showed an approximate 40% drop.
-------
IV-17
It is important to note that both this analysis and the sensitivity
analysis discussed above treat the ambient PMio/TSP ratio as a fixed
quantity. All reviews of measured data indicate that the ratio varies
both temporally and spatially. More recent assessments have attempted
to account for this variability and have used probabalistic methods to
estimate non-attainment. When this analysis was initiated there were
insufficient data to conduct a probabalistic assessment.
Partial Attainment.
In this study, not all counties or sub-county areas are brought into
attainment after the application of the control strategy. This stems
from the limitations of the emission inventory and the control options
available (discussed in Section IV.C) as well as from actual air quality
problems. Therefore, the number of projected residual non-attainment
counties is believed to be in excess of the actual non-attainment problems
that might be encountered under the alternatives considered. In the
context of this analysis, the base costs (or Scenario A) do not fully
account for the cost of bringing all counties or subcounty areas into
full attainment. In order to do so, national control strategy costs were
multiplied by the national average ratio of the remaining air quality
improvement needed to the air quality improvement already achieved. The
use of this ratio assumes that cost per unit of air quality improvement
in residual nonattainment situations will be the same as the cost incurred
in the strategy. The ratio was computed for both the national median and
national mean cases and the results are presented in Section IV. E. In
this report the mean case is used for discussion purposes, because it
appears to be heavily weighted by marginal costs and therefore represents a
more reasonable upper estimate.
-------
IV-18
This provided a very rough estimate of total incremental attainment costs
for the some 1231 county or subcounty areas. Due to the method of deriving
this estimate, it was not possible to disaggregate it for major industrial
categories and source types. Therefore, all cost estimates for specific
industries and source types reflect partial attainment, while aggregrate
national costs are presented with both partial and full attainment.
New Growth.
State-level estimates of future growth through 1995 were obtained
from the Bureau of Economic Analysis, U.S. Department of Commerce for
major industrial categories. State level census data were used to
estimate population growth. The population growth estimates were used
to project growth in paved roadway emissions.
All new growth in point source stack emissions was assumed to be
subject to new source control programs. Therefore, for new sources, Best
Available Control Technology (BACT) emission levels were used in projecting
future air quality levels. BACT for TSP was assumed to be 98% for iron
and steel and 99% for all other source categories. Where current controls
were more efficient than the assumed BACT, the current control levels
were used.
New sources of fugitive emissions were not subject to BACT. Such
new emissions were incorporated into the emission inventory and were
candidates for control under the control strategy.
Retirement/Replacement.
Existing point source stack emissions were assumed to retire and be
replaced in accordance with rates presented in Table IV.D.I. These data
were derived from national retirement estimates by Energy and Environmental
Analysis, Inc. Unlike emission increases resulting from new growth, it
-------
IV-19
TABLE IV.D.I INDUSTRIAL RETIREMENT RATES BY
STANDARD INDUSTRIAL CLASSIFICATION (SIC)
Average Annual Average Annual
SIC Retirement Rates SIC Retirement Rates
01 .0426 31 .0409
07 .0426 32 .0493
08 .0426 33 .0497
09 .0426 34 .0323
10 .0426 35 .0410
11 .0426 36 .0461
12 .0426 37 .0409
13 .0426 38 .0483
14 .0426 39 .0441
15 .0426 40 .0426
16 .0426 41 .0426
17 .0426 42 .0426
20 .0456 43 .0426
21 .0335 44 .0426
22 .0320 45 .0426
23 .0318 46 .0426
24 .0637 47 .0426
25 .0368 48 .0426
26 .0413 49 .0426
27 .0492 91 .0426
28 .0507 92 .0426
29 .0448 93 .0426
30 .0297 100 .0426
Source: Energy and Environmental Analysis, Inc.
-------
IV-20
is uncertain whether all replacement emissions would be subject to the
new source control programs. In the base analysis it was assumed that
50% were controlled to BACT levels and 50% to current levels. The
fraction not subject to BACT was assumed to emit at their old rate and
were considered in the control strategy for additional control. The
fraction subject to BACT were not considered in the control strategy.
In the results section, all results are based on the assumption that
half of the replacement emissions are controlled at BACT level and half
at current levels. A series of sensitivity runs were made which varied
the fraction of replacement emissions subject to BACT. These runs
showed a 47% drop in control strategy costs if all replacement sources were
controlled at BACT level and a 52% increase if all replacement were
controlled at current levels. It should be noted that this analysis
does not directly assess the BACT costs. These costs were estimated
separately, however, and are reported below under "New Source Controls".
Treatment of Area Sources.
As noted in the inventory Section (IV.C) several assumptions were
made to treat area sources. For this analysis area sources were divided
into two categories: 1) paved roads, and 2) all others (primarily
unpaved roads but also home heating, construction activity, airports,
etc.). It was decided that controls, in the form of wet suppression,
were available for municipal paved roads and were estimated in the
analysis. The "other" category sources generally either had no controls
available or had controls which were infeasible to model and cost in this
analysis. As a result, no controls are placed on the "other" component.
As noted earlier, population growth rates were used to estimate growth
in paved road emissions. Since the "other" category is heavily dominated
-------
IV-21
by unpaved road emissions which are believed to either remain constant
or decrease as a county grows, no growth was assumed for the "other"
component.
A major assumption with respect to area sources was that only a
small fraction of their emissions would affect the design value. Studies
have clearly shown that roadway emissions, which heavily dominate the
area sorce inventory, affect air quality over very small areas. Therefore,
it was decided that only 1% of the area emissions would be used in
projecting future air quality and that only 1% of municipal paved roads
would be candidate for control. While this assumption deviates from the
underlying premise of roll-back (that all emissions affect air quality
equally), it was made to introduce a greater degree of realism and to
preclude an undue bias in the results. A sensitivity analysis showed
that increasing the effective fraction to 10% and allowing all sources
to grow with population would result in a 18% increase in initial (1989
pre-control) non-attainment and a 26% increase in costs (DRY). Increasing
the effective fraction to 100% and assuming growth for all area sources
resulted in a 28% increase in initial non-attainment and an approximate
300% increase in base costs. In these cases the large cost increases are
primarily the result of controlling many more miles of municipal road.
Cost-Effectiveness Cap.
In order to prevent the model from choosing unrealistically cost
ineffective controls, a cost effectiveness cap of $12,000/ton of TSP
reduced was added to the model. This means that controls which reduce
emissions at a cost of more than $12,000/ton of TSP (before tax annualized
cost) are not considered in the model. A sensitivity analysis showed
that removing the cap resulted in a 40 fold increase in costs with a
negligible improvement in residual non-attainment. In this case, the
model was estimating the cost of removing an efficient control device
-------
IV-22
(for which no salvage value was imputed) and replacing it with one only slightly
more efficient.
New Source Controls (BACT).
As noted in the section on growth, BACT was assumed to be 99$ control
in all industries except iron and steel where 98% was assumed. In cases where
current (1978 NEDS) controls were more effective, the current control levels were
used. These BACT assumptions were based on a review of decisions recorded in
EPA's New Source Review Clearinghouse. They are intended as a reasonable
approximation only, since the Clearinghouse refects considerable variation in
BACT decisions. A sensitivity analysis indicated that increasing the assumed
BACT efficiency to 99.7% cut the base cost of the control strategy for existing
sources by 37%.
The cost of the new source controls was not estimated directly in this
analysis. For the most part the requirements for these controls are independent
of the NAAQS (e.g NSPS, PSD). However, clearly these controls do improve
air quality and make the job of attaining the NAAQS easier. For the counties
used in this analysis, an indirect estimate of the annalized costs for BACT
through 1995 resulted in a range from $0.9 billion to $3.4 billion depending
on various assumptions regarding cost and control levels.*
The cost analysis assumed a single level of BACT control for new sources
within a given category. While this greatly simplified the analysis, the new
source review program actually examines each major new source to determine BACT
(or LAER) levels. In doing so, consideration is given to local conditions
including the air quality where the facility is to be located. Because the
potential exists for variation in control requirements from one area to another,
EPA plans between proposal and promulgation to examine whether it is feasible
and practical within the scope of the cost analysis to take into account
the specific air quality situation when applying controls to new sources.
*Estlmate was prepared by the Economic Analysis Branch, Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency.
-------
IV-23
1978 Baseline.
As noted previously, this analysis uses 1978 TSP air quality and
emissions as a baseline. However, the 1978 air quality and emissions
levels both reflect the controls in place at that time. Estimating the
cost of those particulate controls was beyond the scope of this analysis.
However, a review was made of EPA's Cost of Clean Air and Water (December,
1979) to derive an approximation of these costs. The figures below were
the result of this effort.
$millions (1978)
1978 Capital Investment $1 ,710
1970-77 Cumulative Capital Investment $11,313
1978 Annualized Cost $ 4,290
1970-77 Cumulative Annualized Cost $ 7,735
It should be pointed out that these historical TSP costs were derived by
a very different methodology than those for the alternate stanards and the
two are not directly additive. They do, however, help to place the costs
reported below in perspective.
Diesel Emissions Growth.
In 1980 diesel automobiles accounted for only 5% of all new light
duty vehicles. However, current projections show that by 1995 they may
account for as much as 25% of the new car fleet. The growth projections used
in this model would underestimate future particulate air quality if this
rapid growth in diesel automobiles takes place. EPA's Office of Policy
Analysis undertook a study to estimate the increments to particulate air
quality concentrations which would result from a diesel penetration of this
magnitude (Steckler, 1982). These estimates assumed that diesels would
meet Federal Motor Vehicle Standards and would account for 25% of new car
sales by 1995. To assess the potential impact of this increase, a sensitivity
analysis was performed. The results indicate that the inclusion of diesel
-------
IV-24
emissions would increase both initial nonattainment and the Discounted
Present Value (DPV) by approximately 2.5% under the current secondary
•5
standard. For the PM10 alternative of 70 i^g/m annual average and 250
24-hour average the increase was estimated to be 7.6%. These increases
would be in addition to the costs incurred to meet the Federal Motor
Vehicle Standard for diesels which would be $280-421 millions per year
assuming the diesel penetration rate noted above.
Process Fugitives.
It was noted above (Section IV.C) that the NEDS inventory does not
separately identify process fugitive emissions and that the cost of their
control was not directly estimated in this analysis. Process fugitives
are nonducted emissions which escape to the atmosphere from roof monitors,
windows, building ventilating systems, etc. They are distinguished from
non-traditional fugitive sources which include open dust sources such
as roads, material handling, and storage piles. Table IV.D.2 provides a
listing of industries with significant fugitive particulate emissions.
The table indicates that in eight of the industries non-traditional
sources dominate the fugitive fraction. The cost of controlling non-
traditional sources is accounted for directly in the model.
Of the remaining industrial categories, the three associated with
steel manufacturing (i.e. coke manufacturing, iron production, steel
manufacturing) have the greatest uncontrolled fugitive process emissions.
A number of studies are available on the emissions and control efforts
of the iron and steel industry. These studies, some of which addressed
process fugitive controls, were compared and the results are presented
in Section IV.E "Iron and Steel." Table IV.D.2 also indicates that
foundries have significant fugitive emissions. The fugitive emissions
and controls for foundries vary from source to source and there is
little data available on the hundreds of small foundries across the
-------
TABLE IV.D.2. FUGITIVE EMISSIONS FROM INDUSTRIAL CATEGORIES
Uncontrolled fugitive
participate emissions
Percent of
total plant
uncontrolled
Industrial Category
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Foundries
Portland cement
Minerals extraction and
beneficiatlon.
Iron production
Secondary lead
Primary aluminum
Asphaltlc concrete
Lime manufacturing.
Coke manufacturing
Secondary aluminum
Secondary brass/bronze
Secondary zinc
Lumber and furniture
Concrete batching
Primary copper
Grain elevators
Primary zinc
Primary lead .
Steel manufacturing
(tons/yr)
(117,872)
(768,961)
(714,096)
(110,070)
( 4,684)
( 57,890)
( 51,638)
( 49,410)
(145,400)
( 1,995)
( 842)
( 472)
( 9,549)
( 34,200)
( 22,024)
(1,364,803)
( 1,991)
( 12,945)
( 68,250)
(jar 1 1 t_u i a ic
emissions
50.3
5.7
100.0
0.9
6.2
24.4
0.7
1.3
100.0
5.6
10.9
6.9
52.9
100.0
22.0
100.0
2.1
6.1
2.8
rv
tn
Fugutive emissions from this category are primarily from non-traditional sources and are addressed
in the basic analysis.
323,720 TPY for integrated iron and steel.
-------
IV-26
country. Thus, within the scope of this study, it was not possible to
conduct a reasonable analysis of the cost of controlling these sources.
It was assumed that process fugitives from primary and secondary lead
smelters would be controlled under the.lead NAAQS. The control of these
smelters was studied by EPA as a part of the lead regulatory impact
analysis. The remaining categories have relatively insignificant fugitive
emissions and were not analyzed further in this study.
In summary, a review of industries with significant process fugitive
sources showed iron and steel to be the most important. A more detailed
analysis of this industry is presented below in Section IV.E "Iron and
Steel."
-------
IV-27
E. RESULTS
1. National Costs.
Tables IV.E.I and IV.E. 2 present the estimated national costs
of meeting the current primary TSP standards as well as various alternate
levels of a PM^g standard. As discussed in the section on "Methodology and
Data Bases", these estimates represent the cost of going from 1978 air
quality and emissions levels to attainment of a given standard. In the
case of the current TSP standards, two attainment years (1987 and 1989)
were assumed; in both cases maintenance was required through 1995.
For the PM]Q alternatives a 1989 attainment year was assumed with maintenance
through 1995. To keep the costs presented below in perspective, it
should be noted that pre-1978 capital investment in particulate controls
is estimated to be approximately $11 billion.
Because changes in air quality values are not estimated after the
year 1995, the time horizon for the benefit analysis (see Section VI)
ends with that year. However, the pollution control equipment installed
in either '87 or '89 has an expected life of 15 years; the cost analysis
presented here includes operating and maintenance costs for the expected
life of the equipment and annualizes the capital cost over the same 15
years. In order to provide a common time horizon for the comparison of
benefits and costs, only the portion of the annualized costs which will
be incurred through the end of 1995 are included for the benefit-cost
comparison in section VII. Because of this adjustment the present value
costs reported here and in section V are greater than the costs reported
in section VII. The same underlying annualized costs are used in all
sections, however.
As noted in Section I "Partial Attainment", not all counties are
brought into attainment through the application of the control strategy.
-------
IV-28
Table IV.E.I
Total Estimated Nationwide Costs1
(Smillions)
Scenario2
Capital
Annual Cost-*
Discounted Present
Value4
TSP5(75, 260)787
TSP (-,150)787
TSP5(75, 260)789
TSP (-,150)89
PM10(55,-)/89
PMio(55,150)/89
PM108(48, 183)789
PM10( 55, 200)789
PM10(55,250)/89
PM10(70,250)/89
6,590
14,510
6,020
13,760
5,810
8,200
7,750
5,840
5,930
2,230
1,080
2,410
1,000
2,290
970
1,360
1,290
980
990
370
5,100
11,370
3,920
8,950
3,780
5,300
5,040
3,810
3,860
1,430
*Costs were calculated in 1980 dollars and do not include the cost of: 1) pre-
1979 controls and 2) New Source controls.
: TSP (x,y)/z - x=annual standard, y=24-hour standard, z=Attainment Year.
^Annual costs include operation and maintenance costs and annualized capital
charges. Annual capital charges were derived using an assumed 15-year equipment
life and a 10% real discount rate.
^Discounted Present Value represents the summation of the stream of annual
O&M and capital payments discounted back to 1982 using a 10% real discount
rate.
Approximates a PM1Q standard of 48 yg/m3 AAM/180 yg/m3 24-hour.
"This PMiQ alternative approximates the current primary TSP standards. The
PMlQ values were derived from the TSP values by applying the regression
equations used to estimate missing values in the air quality file and by
applying the PM^g/TSP conversion ratio of 0.55.
-------
Table IV.E.2. Detailed Nationwide Costs and Attainment Status
Scenario^
Initial
Non-Attainment
Areas
Cost of Strategy
Discounted present
Value3
(106$)
Annual
Cost4
Residual
Non-attainment
Areas
Estimated DPV Cost for Reduction
Residual Non-attainment (106$)
Median Average
of
TSP5(75,260)787
TSP(-,150J/87
TSP5(75,260)/89
TSP(-,150)/89
PM10(55,-)/89
PM10j[55/150)/89
PM1(J6(48,183)/89
PM10(55,200J/89
PM10(55,250)789
PM10(70,250)789
300
525
299
522
182
329
298
205
185
105
2,046
3,343
1,624
2,675
1,356
1,900
1,977
1,430
1,390
757
433
708
416
685
347
487
507
365
356
194
150
304
147
301
88
165
160
102
92
50
356
651
0
506
441
161
291
408
431
83
3,052
8,027
2,296
6,277
2,422
3,404
3,068
2,379
2,467
676
10
1)
2)
3)
4)
5)
6)
Costs were calcualted in 1980 dollars and do not include the cost of: 1) pre-1979 and 2) New Source controls.
Key: TSP (x,y)/z - x=annual standard, y=24-hour standard, z=Attainment Year.
Discounted Present Value represents the summation of the stream of annual O&M and capital payments discounted back
to 1982 using a 10% real discount rate.
Annual costs include operation and maintenance costs and annualized capital charges. Annual capital charges were
derived using an assumed 15-year equipment life and a 10% real discount rate.
Approximates a PMig standard of 48 ug/m 24-hour.
This PM^o alternative approximates the current primary TSP standards. The PM^g values were derived from the
TSP values by applying the regression equations used to estimate missing values in the air quality file and
by applying the PM^o/TSP coversion ratio of 0.55.
-------
IV-30
Table IV.E.2 shows the initial number of non-attainment areas, the cost
of the modeled strategy, and the number of areas left in non-attainment.
The estimated incremental costs of bringing all areas into complete attainment
is shown in the two right hand columns. These costs were obtained by
multiplying the cost of the strategy by the national ratio of the remaining
air quality improvement needed to the air quality improvement already achieved.
The median value column used the median ratio for all counties and the mean
value column employed the mean ratio. The results presented in Table IV.E.2
show wide differences between the two estimates. This indicates that the
distribution of costs is skewed by a few very costly controls and that the
average is heavily weighted by the more costly controls. In this sense
the average is affected by the marginal cost of control; accordingly it was
used in all the national cost estimates discussed below.
It should be noted here that the estimated cost of reducing residual
non-attainment carries with it considerable uncertainty. As discussed
in Section IV.D., the residual non-attainment is felt to be due in large
part to gaps in the inventory and to a lack of control options in the
model. Therefore it is uncertain whether the cost of the remaining
reductions would approximate the average as was assumed here. However,
since the average is so heavily weighted by the marginal, it was decided
that it could be used as a reasonable upper estimate. The total estimated
cost of a standard was derived by adding the cost of the modeled strategy
(or Scenario A as it is often referred to) to the average cost for reducing
residal non-attainment and is shown in Table IV.E.I. It is the total
cost estimates which are used for proposes of comparison in the discussion
which follows and are referred to as Scenario B. The more detailed
information in Table IV.E.2 is intended to show the derivation of the
-------
IV-31
total estimated cost and to convey some of the uncertainties associated
with the analysis.* As an examination of Table I.E.2 shows, 50% or less of
total cost of any given standard comes from the control strategy (Scenario
A), the balance coming from estimated cost of reducing non-attainment.
Table IV. E.I indicates that the current primary standards implemented
in 1987 would result in higher control costs than most of the alternative
standards analyzed. Within the range of alternatives analyzed only the
low end of the range (55/150) shows costs which approach those of the
current standard. If the same implementation year (1989) is assumed for TSP,
then the PM^g 55/150 alternative produces appreciably higher costs and
initial non-attainment. The higher costs are a function of both the
number and type of non-attainment areas as well as the fact that, in this
modeling analysis, PM^g control strategies tended to focus slightly more
on stack controls, which tend to cost more. Although the level of this
alternative standard (PM]g55/150) appears to be less stringent than the
current primary, the statistical form of the 24-hour alternative makes it
somewhat harder to meet.
All of the remaining alternative PM-|g standards analyzed resulted
in cost decreases compared to the current TS? primary regardless which
implementation year is chosen. For example, the 55/200 combination
resulted in a 25% decrease in costs (DPV) and a 31% decrease in the
initial non-attainment when compared to the TSP standard attained in
1987. If a common attainment year is used, the 55/200 alternative shows
only a 3% decrease in costs (OPV) and a 30% decrease in initial non-attainment,
*This uncertainty can be illustrated by comparing the total costs of the
55/200 PMiQ alternative with those of the 55/250. Although 55/200 is
the more stringent of the two, its total costs are less. This result is
an artifact of the way the residual non-attainment costs were computed.
In fact, if one refers to the cost of the strategy, the 55/200 is seen
to be the more costly.
-------
IV-32
At the midpoint of the Staff Paper range of alternatives (70/250) estimated
costs drop by 72% and initial non-attainment by 65% as compared to the
current primary NAAQS. In terms of non-attainment, the range of alternatives
analyzed run from rough equivalency to the current primary to an approximate
65% decrease. Depending on the TSP year of implementation selected,
costs show either a 71% decrease (1987) or a 60% decrease (1989). These
cost reductions are due to the fact that there are fewer non-attainment
areas requiring control and that less stringent control is needed in the
few remaining non-attainment areas. In the case of different implementation
years, the time value of money is also a factor. It should be noted that
it was assumed here that controls already in place would be maintained
regardless of the standard level. No savings due to removal of controls
are estimated. The cost reductions represent controls not yet installed
that would no longer be needed if the alternative were adopted.
Clearly from Tables IV.E.I and IV.E.2 the most stringent standard
analyzed is the current secondary TSP standard of 150 yg/m^. Assuming
implementation in 1989 in both cases the PM^Q 55/150 alternative would
result in 40% drop in costs and a 37% decrease in initial non-attainment
when compared to the current secondary At higher levels of PM^o alternatives
the decreases become even more significant. The dominance of the current
secondary is reflected in all of the following tables.
In summary, this analysis indicates that within the range of primary
alternatives analyzed, a PM -\Q combination of 55/150 would approximate
the current primary standard in terms of non-attainment. However, the
composition of the counties does change. All of the remaining alternatives
showed significant reductions in cost and non-attainment. The most
stringent standard analyzed is the current secondary standard. This
-------
IV-33
standard produces higher costs and non-attainment than any of the
alternatives examined.
Table IV.E.3 presents the national costs and emission reductions
achieved as a function of source type (e.g. stack, fugitives, and roadways).
It should be remembered that these costs reflect the control strategy
only (Scenario A) and do not include full attainment (Scenario B). The
cost of full attainment was estimated on the basis of air quality and
unfortunately data limitations precluded disaggregation by source type.
These tables show that stack controls account for 85-90% of the control
strategy costs regardless of the standard being analyzed. The fugitive-type
sources (both plant fugitives and municipal roads) account for the remaining
10-15% of the costs, but represent some 20-30% of the emission reductions
achieved. However, there is a slight (>1%) decrease in reliance on
roadway controls and other non-traditional controls in the PM^g alternatives
compared to the current standards.
It should be emphasized here though that the magnitude of the costs
and reductions in emissions from municipal paved roads are both dependent
on the assumption that only a small fraction of roadway emissions affect
the design value. Sensitivity analyses have shown that as the effective
fraction of roadways is increased the initial non-attainment costs and
emission reductions will increase (see Section IV.D "Treatment of Area
Sources"). The model tends to control roadways and nontraditional fugitive
sources first, since they are relatively inexpensive to control. However,
since these sources impact air quality over only very small areas, their
treatment in this analysis is thought to be reasonable. Finally, it
should be noted that, since the NEDS inventory does not separately identify
-------
IV-34
Table IV.E.3
Nationwide Discounted Present Value (DPV) Costs
and Emission Reductions by Source Type
(SMillions and 103 TRY)
Scenario^
Source Type
Nontradi- Paved
tional Municipal National
Stack Fugitive Roads Total
EW~ DPV3 ERA4 DW3 ERA7*" DPV3 ERA4
TSP5(75,260)/87
TSP(-,150)787
TSP5(75,260)789
TSP (-.150)789
PM10(55,-)/89
1780 524 229 159 37 7 2046 690
2932 812 360 244 51 9 3343 1066
1407 495 187 158 30 7 1624 660
2338 781 295 243 42 9 2675 1033
1210 240 125 58 21 3 1356 302
PM10( 55, 150)789
PM1Q6(48, 183)789
PM10(55, 200)789
PM10( 55, 250)789
PM10(70, 250)789
1679
1603
1271
1242
676
331
344
247
242
116
192
185
133
127
69
84
82
61
59
40
28
30
22
21
13
6
5
4
3
2
1900
1977
1426
1390
758
419
431
312
304
158
*Costs do not reflect full attainment; see text for full explanation. All costs
calculated in 1980 dollars and do not include the cost of 1) pre 1979 controls
and 2) new source controls.
^Key: TSP (x,y)/Z x = annual standard, y = 24-hour standard, and z = attainment
year
JDiscounted present value represents the sumation of the stream of annual
O&M and capital payments discounted back to 1982 using a 10% real discount rate.
^Emission Reduction Achieved (ERA) given for pollutant corresponding to standard.
^Approximates a PMig standard of 48 ug/nr AAM/180 ug/rTr 24-hour.
"This PMin alternative approximates the current primary TSP standards. The PM
10.
values were derived from the TSP values by aplying the regression equations used
to estimate missing values in the air quality file and by applying the PM}Q/TSP
conversion ratio of 0.55.
-------
IV-35
fugitive process sources, the impact and cost of their control could not
be directly estimated. This issue is discussed in more detail in the
iron and steel section (Section IV.E.3).
As noted in Section IV.0. "New Source Controls", it was assumed
here that all new sources and half of the replacement sources would emit
at "NSPS" or "BACT" levels. Since the various new source control and
review programs are not directly related to attainment of a NAAQS, the
cost of these programs were not included in any of the tables in this
section. As previously noted, new source costs were estimated to be
between $0.9 and $3.4 billion. These costs would be in addition to the
cost estimates presented in Tables IV.E.I.
2. Industry Specific Costs.
Table IV.E.4 presents costs and emission reductions for selected
major industrial categories. These categories were selected on the basis
of a ranking of all industries by OPV cost for the current secondary
standard. Municipal paved roads and primary copper smelters were added
to the list because of their general interest. It should be understood
that this ranking was done on the basis of raw costs and not the economic
impact those costs might have. The costs reported by industry are taken
from the control strategy and, therefore, reflect partial attainment. As
noted earlier, since the cost of full attainment was computed on the
basis of air quality, it is not possible to break it out by industry.
In terms of cost, the two major industrial categories are the utilities
(SIC 4911) and iron and steel (SIC 3312). Together these two industries
account for 50% or more of the costs for most of the standards analyzed.
This is not unexpected since the emissions from these two industries
make-up some 44% of the base inventory used in this analysis. The costs
-------
Table IV.E.4. Discounted Present Value (OPV) Costs and Emission Reductions for Major Categories
Scenario
TSP(7b
,260)/
87
SIC
.
4911
3312
2951
3241
2621
1422
3295
4961
5153
3321
3331
All
Category
Municipal Paved Roads
Utility Power Plants
Iron and Steel
Paving Mixtures
Hydraulic Cement
Paper Mills. Except
Building Paper
Crushed and Broken
Limestone
Ground or Treated
Minerals
Steam Supply
Wholesale Grain
Gray Iron Foundries
Primary Copper
Smelters
SICs National Total
DPV*
37
706
366
57
59
29
28
49
51
32
43
5
2046
ERAJ
7
246
137
16
53
b
11
34
6
11
3
1
690
TSHl-150)/
87
DPV
51
1243
525
93
84
72
69
62
61
54
53
14
3343
ERA
9
428
197
21
6
10
16
42
6
20
3
4
1066
TSP(/5/260)/
U9
DPV
30
561
291
46
48
23
23
39
41
25
34
4
1624
ERA
7
235
127
15
bl
5
11
33
b
10
2
2
660
TSP(-.lbU)/
89
DPV
42
997
423
7b
66
57
55
49
46
42
42
11
2675
EKA
9
422
191
21
58
9
15
41
6
19
3
4
1033
PMiu(bb.-)/ PMnllU/183)/
89
DPV*
21
464
263
32
3b
18
6
35
43
11
19
6
1356
ERA-5
3
123
49
5
15
2
2
23
3
5
1
1
302
tiPV*
30
610
398
45
64
24
15
43
50
25
38
12
2977
ERAJ
5
161
70
6
33
3
5
23
4
9
2
2
431
PMm(55/20)/ PMml/0, 250)7
89
DPV
22
487
264
33
40
19
6
37
43
14
29
6
1426
ERA
5
125
50
5
17
2
2
23
3
6
1
1
312
89 ,
13
228
175
18
30
11
2
24
39
8
11
4
767
ERA"
2
54
26
2
10
1
1
21
2
3
<1
<1
158
Icosts do not reflect full attainment; see text for explanation. Costs calculated 1n 1980 dollars
and do not Include the cost of: 1) pre-1979 controls and 2) New Source controls.
^Discounted present value represents the summation of the stream of annual O&M and capital
,payments discounted back to 1982 using a 10% real discount rate.
•'Emission Reduction Achieved (ERA) given for pollutant corresponding the standard.
oo
cr>
-------
IV-37
associated with iron and steel are discussed in more detail in a separate
section below. For utilities, approximately 20% of the emissions reductions
and 6% of the costs (DPV) come from non-traditional fugitive controls
(storage piles and plant roads). To help place the remaining utility
costs in perspective, 93 power plants were assigned stack controls in the
analysis of the current secondary standard. On average this would mean
approximately $12.5 million per plant. The impacts of these costs will
be discussed in more detail in the economic analysis.
For most industries, the PM-jg alternative standards produce the
same relative decreases in costs as were shown in the national totals.
There are some notable exceptions, however. For example, the Crushed
and Broken Limestone industry (SIC 1422) shows much more dramatic decreases
in costs. Also under the PM]g 70/250 alternative national costs show a 53%
decline from the current primary, but the limestone industry shows a 91%
decline. In part, this is due to the fact over that half of the emission
reductions and control costs in this industry come from fugitive sources
which emit larger particles (under PM^g fugitive emissions tend to be less
important). However, other industries with large fugitive emissions
(e.g. Ground or Treated Minerals) do not show this same effect. It is
believed that the limestone operations are located primarily in counties
with moderate air quality that become attainment counties under most
alternatives. The reverse of this effect is noted in the iron and
steel section.
3. Iron and Steel Industry Costs.
For the standards analyzed, the discounted present value costs to
iron and steel ranged between $525 million for the current secondary
TSP standard down to approximately $175 million for a 75/250 PM^g alternative.
These figures are lower than other recent estimates of steel industry
-------
IV-38
compliance costs. However, there are several features in this analysis
which caused the differences. In the first place the estimates shown in
Table IV.E.4 do not include the cost of controls installed prior to 1979.
Also, since process fugitive emissions are not identified in the NEDS
inventory the cost of their control could not be estimated. In most
cases, controls for process fugitives are more costly than either the
stack control or the non-traditional fugitive controls examined in this
analysis. Finally, the costs in Table IV.E.4 do not represent
full attainment. As noted before, the cost of full attainment could not
be broken down by industry. A number of other recent studies, both
within EPA and by the industry, have examined the cost to the steel
industry of meeting the current TSP standard. The bases for these other
estimates of steel industry costs are discussed below and insofar as
possible are reconciled.
Arthur D. Little, Inc. prepared a study of the steel industry environmental
costs in 1981 for the American Iron and Steel Institute (Arthur D. Little,
1981). This study reported all costs for air pollution controls that had
been incurred by integrated steel plants up to 1980. The plants in the
A. D. Little analysis accounted for some 88% of national production capacity
in the year 1976. In addition, the study projected the cost of additional
controls that would be needed to meet the current primary and secondary
standards. In doing so, the study included the costs of controlling
process fugitive emissions. Its cost estimates were based on meeting the
current primary and seondary standards. A second study was prepared by
Temple, Barker, and Sloane, Inc. for EPA's Office of Policy Planning and
Evaluation (Temple, Barker, Sloane, Inc. 1982). This study, completed in
1982, estimated the cost to the steel industry of complying with current
-------
IV-39
SIPs. Similar to A. D. Little, this report included all costs that had
been incurred prior to 1979 and addressed process fugitive controls.
Temple, Barker and Sloane also estimated control needs for other pollutants
which were assumed to be associated principally with boilers and therefore
boiler control costs were excluded in the comparison below. For 1980
(the only common year for which aggregated costs were presented), sunk
capital costs were estimated to be $3.2 billion (excluding costs ascribed
to boiler controls) by Temple, Barker and Sloane, and $3.5 billion by A.
D Little. In a third study, PEDCo Environmental, Inc. estimate the
costs of implementing the consent decrees for EPA's Office of Air Quality
Planning and Standards (Pedco,198). This study did not address costs
previously incurred by the industry, but did include process fugitive
controls. Finally, EPA's Division of Stationary Source Enforcement
estimated the cost of both SIPs and the consent decrees.
Attempting to compare the costs reported in these studies raises a
number of problems. For example, while A. D. Little explicitly estimates
the cost of meeting the primary standard, the other studies estimate the
cost of meeting certain regulatory reqirements, usually SIP requirements.
For purposes of comparison, it was assumed here that the SIP and consent
decree requirements reflected an equivalent level of control to that
which would be required to meet current primary standards. Therefore,
A.D. Little estimates of cost of achieving the current secondary standard
were not considered in the reconciliation analysis. As a point of reference,
however, A.D. Little estimated the cost of the current secondary NAAQS to
be an additional $1.7 billion.
-------
IV-40
Table IV.E.5 presents an effort to reconcile these various estimates
by placing them on a common analytic basis.* As the footnotes indicate,
pre-1979 costs (as estimated by Temple, Barker and Sloane) were added
where appropriate and fugitive process costs were also added. The A. D.
Little estimates are clearly greater than the others. The other four
estimates, as adjusted, cluster around $4.0 billion. This table shows
the magnitude of error introduced by the inability to estimate process
fugitive costs in the PM^g RIA analysis. Approximately 30% of the total
cost or $1.3 billion are due to process fugitive controls. It also
indicates the assumptions made in analyses of this kind can have a
substantial impact on the final results. The estimates of process
fugitive control costs show the widest variation. Controls for such
sources are site specific and are difficult to estimate on a general
industry wide basis.
None of the studies cited above addressed the impact of a PM^g
standard on the steel industry. In the present study the potential
impact of a PM^o standard was addressed. Referring to the unadjusted
costs shown in Table IV.E.4, the PM-|g alternative of 70/250 shows that
steel industry costs could decline by about 40% when compared to the
* Following the completion of the RIA EPA's Office of Policy, Planning
and Evaluation contracted with GCA Technology Division to assess steel
industry compliance costs. As with the other studies discussed above,
the GCA study is based on a number of assumptions and is difficult to
compare directly. The study was thorough and estimated both pre-1970
costs and projected capital requirements out to 1990. It also accounted
for the Steel Stretchout Act (SICEA), the Bubble Policy and recent SIP
revisions. By accounting for these factors, which result in delayed
equipment purchases, GCA estimates higher future costs than other studies,
Their total capital cost estimate was $5.4 billion. This is towards the
upper end of the range reported in Table IV.E.5.
-------
IV-41
Table IV.E.5
RECONCILED STEEL COSTS
CURRENT PRIMARY STANDARDS
($ Billion)
Non-Traditional Process
Stack Fugitive Fugitive Total
A.D. Little 3.5
Temple, Barker^ 3.7
& Sloane
PEDCo2 2.4
Argonne (PM1Q RIA)3 — 2.8
DSSE4
0 2.4 5.9
0 0.9 4.6
.2 1.1 3.7
1.3 4.1
3.6
1-Does not include "miscellaneous" source costs, i.e., boilers
which included controls for other pollutants.
Adjusted to reflect pre-1979 costs.
^Adjusted to reflect pre-1979 costs and process fugitive controls.
Adjusted to reflect pre-1979 costs.
-------
IV-42
current primary. The same alternative would result in approximately 58%
cost decrease when compared to the current primary and secondary. In
general, the cost decreases in the steel industry tend to be proportionally
lower than other industries and the national average. This is primarily
due to the fact that the steel industry counties tend to have high particulate
matter levels and, thus, many are predicted to be in non-attainment even at
relatively high levels of a PM^g standard.
A more detailed analysis was performed to determine the impact of
alternative PMig standards on one example steel facility (PEDCo, 1981). The
plant in question is a medium sized facility with most of the particulate
sources found in the steel industry. The plant is relatively well controlled
when compared to the rest of the industry. The study showed that the
current primary standard and the 55/150 PMig alternative would be roughly
equivalent in terms of control requirements. The current primary TSP
standard resulted in controls with an annualized cost of $4.9 million;
adding approximately $2.13 per ton of finished steel. The 55/150 PM^g
alternative showed annualized costs of $5.2 million or approximately $2.25
per ton of finished steel. The study also indicated that, for this
particular plant, no new controls would be required by the PM^g alternatives
above 55/225. This finding is specific to the plant studied and is a
function of the controls already in place and the general air quality
situation around the plant. The study further showed that the bubble
concept would still be effective with a PMig standard. Controls
on fugitive emissions were selected for both the TSP base case and the
PMig alternative. The current secondary standard could not be met in
this study even with the addition of all controls considered in the
-------
IV-43
model. Addition of all controls resulted in an annualized cost of $17.7
million and a 24-hour average of 185 yg/m^(TSP).
In summary, this comparison illustrates the fact that cost estimates
for a specific industry are very sensitive to the assumptions made in
deriving the estimates. For the steel industry, the treatment of pre-
1979 costs and fugitive process costs can significantly alter the estimate
made of total industry costs. The costs to the steel industry as estimated
by the national model used in this analysis are significantly low in comparison
to other studies due to the inability to estimate process fugitive costs
and the exclusion of pre-1979 costs. However, when estimates of these
cost categories are aMeM in, the steel cost estimates presented here
fall in line with other recent studies.
4. REGIONAL
Tables IV.E.6 and IV.E.7 present two different geographic breakouts
of the costs. Table IV.E.6 is organized by the standard Federal regions;
while Table IV.E.7 is organized by groupings of Bureau of Census regions.
An examination of both tables reveals relatively high non-attainment in
the Western regions in comparison to the East for all standards. This
is easier to see in the maps which are organized using the Bureau of
Census groupings (Figures IV.E.1-5). The four sections west of the
Mississippi (Pacific, Mountain, Northern Midwest, and Southern Midwest)
contain roughly 45-65% of the initial non-attainment. However, these
sections incur only 30-40% of the cost.
-------
Table IV.E.6. Regional Discounted Present Value (UPV) Costs and Attainment Status1
1982 Discounted Present Value (106J) and Initial/Residual Nonattalnment Status
II
III
IV
VI
VII
VIII
IX
Scenario**
TSP(75,260)/87
TSP(7b,150)/B7
TSP(75,260)/89
PM10(70,250)/B9
PMl0(55,250)/89
PM10(55.200)/89
DPV3 NAA4 DPV3 NAA4 DPV3 NAA4 DPV3 NAA4 DPV3 NAA4
23 4/2 41 5/1 294 24/9 26b 39/16 827 70/28
89 20/11 101 13/9 384 4b/21 477 88/42 1406 134/71
17 4/2 33 b/1 229 24/7 213 39/lb 660 70/28
0 0/0 <1 1/0 13 6/1 77 7/2 355 23/6
1 2/0 3 2/0 117 11/1 2b6 lb/10 59b 40/15
1 3/0 6 3/0 118 14/2 2b8 lb/11 609 44/16
1) Costs calculated In 1980 dollars and do not Include the cost of: 1) pre-1979
2) Key: TSP (
3) Discounted
discounted
x.y)/Z - x°annual standard. y=24-hour standard, Z=atta1nment year.
Present Value represents the summation of the stream of annual O&M
back to 1982 using a 10% real discount rate.
DVP3 NAA4
94
Ib2
7b
7b
87
88
28/13
50/31
28/13
15/9
29/1 b
31/17
controls and
and
capital
DVP3 NAA4 DVP3 NAA4
102
257
76
15
24
33
2)
38/7
7/18
38/7
6/2
15/2
16/2
133
172
106
59
82
82
19/14
33/22
19/14
8/2
13/6
15/6
DVP3 NAA4
204
210
164
136
186
188
41/36
47/42
40/36
25/19
37/31
37/32
DVP3 NAA4
63
104
51
27
37
42
32/24
44/41
32/24
14/9
21/12
26/16
New Source Controls.
payments
4) Non-attainment areas (NAA) entries give (Initial NAA)/ (Res) dual NAA).
-------
Table IV.E./. Sectional Discounted Present Value (UPV) Costs and Attainment Status1
1982 Discounted Present Value (106$) and Initial/Residual
Pacific
Scenario2
TSP(75,260)/87
TSP(7b,lbO)/87
TSP(7b,260)/89
PM10(70,2bO)/89
PM10(bb,2bO)/89
PM10(bb,200)/89
DPV3
109
IbO
88
88
104
109
NAA4
39/29
b2/46
38/29
17/12
24/18
29/22
Mountain
DPV3
308
360
246
141
213
214
NAA4
60/51
82/68
60/bl
3b/21
bb/36
b7/37
Northern
Midwest
DPV3
104
302
78
Ib
2b
33
NAA4
47/8
83/23
47/8
7/2
16/3
18/3
Southern
Midwest
DPV3
78
126
62
67
76
77
NAA4
20/7
37/22
20/7
10/6
21/10
23/12
North
Central
DPV3
826
1363
6b8
3bb
b9b
608
NAA4
62/27
121/66
62/2/
22/6
39/14
b2/lb
Nonattainment Status
Northeast
DPV3
34b
b41
2/0
13
120
124
NAA4
21/7
b6/30
21//
6/0
10/0
13/0
Southeast
DPV3
277
blO
221
77
2b7
260
NAA4
bl/21
110/b3
bl/18
8/3
20/11
23/13
* Costs calculated in 1980 dollars and do not include the cost of: 1) pre-1979 controls and 2) New Source
Controls.
2 Key: TSP (x,y)/Z - x=annual standard, y=24-hour standard, Z=attainment year.
3 Discounted Present Value represents the summation of the stream of annual O&M and capital payments
discounted back to 1982 using a 10% real discount.
4 Non-attainment areas (NAA) entries give (Initial NAA)/(Residual NAA).
-------
FIGURE IV.E.l
NON-ATTAllftiff COUNTIES BY REGlOfl (75 ANWAL/260 24-HOUR, TSP)
NATOIAL TOTAL - 300
cri
-------
FIGURE IV,E,2
COUNTIES BY REGlOfl (75 JMJ/t/150 24-HOUR, TSP)
NORTHERN
MIDWEST
NORTHEAST
56
SOUTHEAST
**
1
SOUTHERN
MIDWEST
37
fWTIOilAL TOTAL
-p.
-------
FIGURE IV,E,3
Nffl-ATTAMtOT COUNTIES BY RHGlOf! (55 ANtm^SO^I-HOUR. PM1Q)
NORTHERN
MIDWEST
NORTHEAST
10
SOUTHEAST
SOUTHERN
MIDWEST
NATIONAL TOTAL - 135
00
-------
FIGURE 1V.E.1
N»l-AtTAMMT COUNTIES BY REGION (70AWW/250 21-HOUR, PM10)
NORTHERN
MIDWEST
NORTH
—CENTRAL
•— *xt
22
SOUTHERN
MIDWEST
10
-------
FIGURE IV.E.5
NoM-ATTAMIfT COUNTIES BY REGION (55 ANNML/200 2'1-HOUR,
NORTHERN
MIDWEST
NORTHEAST
NORTH
CENTRAL
SOUTHEAST
2
SOUTHERN
MIDWEST ^l '
23
IAL TOTAL- 205
en
o
-------
IV-51
A review of the emissions growth prior to imposition of any control
stragegy (Table IV.E.8 and IV.E.9) shows that the Western and Southern
areas of the nation are projected to experience more growth. This analysis
(based on the growth and retirement assumptions outlined in Section IV.D)
shows an overall decline in emissions in the Northeast and North Central sections.
At the same time emission levels in the Western sections show strong
positive growth. This pattern of emissions growth helps to explain the
relatively large non-attainment figures predicted in the West for both
TSP and PM10.
A regional or sectional comparison of the PM]Q alternatives to the
current standards reveals a potential shortcoming of the present air
quality analyses. Again for the four Western sections, the PM]Q alternative
of 70/250 shows roughly 40% of the cost and 65% of the initial non-
attainment compared to 30% of cost and 55% of non-attainment for the
current primary. In other words the model indicates that the PM^g
alternative may shift a higher portion of the costs to the West. This
same effect is shown in the other combinations presented. It is possible
that the model is accurate on this point. However, it is suspected that
this effect may well be an artifact of analytic assumptions. Specifically,
the application of the 0.55 ratio to derive PM^Q air quality in the West
as well as the East probably resulted in an upward bias to Western PM^Q
levels. It is expected that more intensive monitoring will show that,
outside of the major urban areas of the West, a lower ratio obtains.
This would have the effect of lowering the design value for such areas.
Unfortunately, at this time there is not sufficient data to develop
regionally specific ratios.
-------
Year
1978
1987
1989
1995
, IV-52
Table IV. E. 8. Regional PM-jQ Emission Projections
PM10 Emissions (103 TPY)
Region
I II III IV V VI VII VIII IX
82 176 577 1015 1035 334 274 163 197
74 156 537 1040 980 327 247 161 193
73 153 533 1062 977 330 244 163 194
72 146 533 1166 989 349 238 172 205
National
X Total
103 3956
102 3817
104 3833
111 3981
^Before application of control strategy.
-------
Table IV.E.9. Sectional PM^0 Emission Projections^
PM1Q Emissions (103 TPY)
Section
Year
1978
1987
1989
1995
Pacific
221
211
211
218
Mountain
298
299
303
326
Northern
Midwest
348
316
312
306
Southern
Midwest
255
250
253
268
North
Central
985
934
932
945
Northeast
620
558
549
534
Southeast
1229
1249
1272
1383
National
Total
3956
3817
3833
3981
^Before application of control strategy.
in
CO
-------
IV-54
F. Population Exposure Analysis
An important measure of a standard's impact is its effect on population
exposure. This section presents an analysis of population exposure to
PMio The analysis utilizes the air quality, control strategy, and population
data from the basic cost analysis. In considering the estimates provided
below, the limitations on the air quality and control strategy data discussed
in Sections IV.B-D should be kept in mind. The estimates presented here
are useful in considering the impacts on large populations of both the
current and alternative standards. It is felt that the data presented here
are particularly helpful in illustrating the impact of the form of the
standard.
Methodology
The population exposure estimates below were obtained from an
extrapolation model derived from a more detailed modeling effort in four
cities - Chicago, Los Angeles, Philadelphia, and St. Louis. The more
detailed model is known as the NAAQS Exposure Model (NEM) (PEDCo, 1981).
The NEM operates with two basic data bases: 1) air quality; 2) human
activity. The air quality data, in the NEM particulate matter exercise,
consisted of monitored data, interpolated using a time series analysis to
fill in missing values. The time series analysis attempted to account for
seasonal, week-day/weekend, and random variation in the data. The basic
data along with transformation factors derived from an extensive literature
review were used to estimate air quality levels in various microenvironments
(e.g., indoors, in the workplace, etc.). In some runs of the model, the
basic data were also "rolled back" to reflect compliance with various NAAQS
levels. The human activity data provides information on population movement
-------
IV-S5
about an area and through various microenvironments. The NEM then accounts
for the intersection of people as they go about their daily activities and
air quality microenvironments. These intersections are tracked over a year
and exposure distributions are developed.
Since the NEM is data intensive and expensive to run, the extrapolation
model was developed to make national estimates (Biller, 1983). When NEM
exposure estimates in each of the four cities were made assuming compliance
with a standard, it was found that the exposure distributions (normalized
to the same population) tended to be quite similar. It was possible,
therefore, to develop "universal" exposure curves. Given the universal
curves, a peak 24-hour or annual average concentration for an area, and the
area population it was possible to estimate the PM^g exposure that would be
calculated by NEM. Although both models produce various indices of exposure,
the measure reported here is the number of people who experience concentrations
at or above a given level at least once in the calendar year.
In order to obtain results comparable to those produced in the cost
and benefits analyses, the extrapolation model was run on the same set of
air quality and population data. The air quality data used reflected full
attainment (i.e., Scenario B). The population data were grown out to 1989
and beyond using the growth rates used in the benefits analysis.
Before presenting the results of this analysis several qualifications
need to be discussed. In the first place, the extrapolation model can be
no more accurate than NEM. In this regard it is significant that NEM has
only been applied to four cities. The use of more cities would have improved
the accuracy of the extrapolation model. However, even with the addition
of more cities, it should be stressed that the NEM is a relatively crude
-------
IV-56
model at present. Secondly, the use of a fixed ratio (0.55) to convert TSP
to PMiQ air quality limits the usefulness of the analysis. Specifically,
since the ratio may vary considerably (both spatially and seasonally), the
distributions below may not reflect the added range of exposures that would
result from this variance. The comparisons of the current TSP standard to
various PM^Q levels reported below are still useful in that they illustrate
the impact in terms of exposure of various forms of the standard. It
should be stressed, however, that the analysis does address the question of
particle size indicators. It should also be noted that although the NEM
model has been peer reviewed, its application to PM^g and the extrapolation have
not been peer reviewed.
Results
Table IV.F.I shows the effect on 24-hour PM^o exposures of a series of
PM}Q standards in which the annual component is held constant while the 24-
hour is made more stringent. With only an annual standard of 55 yg/m3,
24-hour exposures reach 357 yg/m3. The 24-hour peak exposure can be
controlled to any desired level by setting an appropriate expected 2nd
maximum standard. Table IV.F.2 shows the effect of the same series of
standards on annual average exposures. In this case, the annual standard
controls the annual average exposure distribution until the 24-hour standard
falls below 200 yg/m3. At the 150 yg/m3 level, the 24-hour component
significantly reduces the number of people exposed to annual levels above
30 ug/m3.
Table IV.F.3 and 4 compare the present TSP primary standards with
two sets of PMio alternative NAAQS (55/200 and 55/150) in terms of their
impact on PM^g exposures. The distributions show a large disparity in the
relative abilities of a deterministic (TSP) and a statistical (PMig) standard
-------
IV-57
to control peak 24-hour PMig exposures. On the basis of the 0.55 ratio, one
would expect that the current TSP standard would be roughly comparable to a
55/150 PMio standard and somewhat better than a 55/200. However, both PMio
standards produce markedly lower peak exposures. On the other hand, as one
moves into the distribution, the TSP standard does protect larger numbers
of people at lower concentrations. For example, the 55/150 PMio alternative
provides better protection down to 120 yg/m^. However, at 100 yg/m^
and below the TSP deterministic standard provides better protection to
more people. Turning to annual exposures (Table IV.F.4) the same effects
although less pronounced can be seen. This is attributable to the different
averaging techniques employed. Within the range analyzed, the PMiQ annual
arithmetic mean results in lower peak exposures when compared to the TSP
annual geometric mean.
These data suggest strongly that a deterministic standard (based on
the observed second high) provides poor control of peak 24-hour exposure.
For peak concentrations much better control is provided by statistical
forms. It is possible that the poor performance of the deterministic (TSP)
standard is overstated here because of the high variability in this air
quality data set between the observed second high and the estimated expected
second high. Such variability is to be expected but in this data set it is
relatively higher than expected. Nonetheless, the conclusion set out above
is unlikely to change. This conclusion has also been drawn from other data
sets and for other pollutants.
Since most of the health studies on particulate matter are epidemiological,
Tables IV.F.5 and 6 provide an alternative measure of exposure. In
these tables it is assumed that the entire county population is exposed to
-------
IV-58
the expected second maximum or annual mean at the design value monitor.
This comes closer to approximating the assumptions present in most community
studies than does the NEM type analysis. It is important to note that, for
both the deterministic (TSP) and statistical (PMioK the peak exposure
values are markedly higher in these tables than the proceeding ones. This
is expected since the NEM type analysis accounts for movement away from the
monitor and into indoor microenvironments where particulate levels are
typically lower. However, the higher concentrations in Tables IV.F.5
and 6 are more directly comparable to the ranges of concentrations discussed
in the Criteria Document and Staff Paper. The same points made above also
apply to these distributions. In fact the difference between the
"deterministic" and "statistical" forms of the standard appear even more
markedly here. As a point of reference Tables IV.F.5 and 6 also include
"base case" exposures (i.e., the exposures estimated to occur in 1989 if
the standard were not met). As can be seen, any of the standards considered
provide improved air quality to a substantial part of the population when
compared to the "base case."
-------
IV-59
TABLE IV.F.I
COMPARISON OF PEOPLE EXPOSED FOR ALTERNATIVE PM10 STANDARDS
24-Hour Exposure with No Smoking
Year: 1989
Concen- Number of
tration Std.*=55/ -
(ug/m3)
340
320
300
280
260
250
240
225
220
200
180
160
155
140
120
100
80
60
40
20
0
Highest Cone.
No. of Counties
Scenario B
No. Seen. B
Population:
Scenario B
No. Seen. B
501
8,120
28,100
38,400
54,500
80 ,900
113,000
165,000
190,000
365,000
912,000
2,050,000
2,550,000
5,200,000
16,600,000
37,500,000
75,000,000
135,000,000
186,000,000
196,000,000
196,000,000
357
•
167
1,055
64,858,200
131,394,380
People Exposed
55/250
0
0
0
0
0
0
0
4
667
99 ,400
644,000
1,740,000
2,220,000
4,810,000
16,200,000
37,300,000
74,900,000
135,000,000
186,000,000
196,000,000
196,000,000
225
163
1,053
65,056,201
131,196,379
At or Above Given Cone.
55/200 55/150
0
0
0
0
0
0
0
0
0
0
369
452 ,000
991 ,000
3,830,000
15,300,000
35,500,000
74,100,000
134,000,000
186,000,000
196,000,000
196,000,000
180
182
1,034
67,618,417
128,634,163
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3,350,000
26,600,000
59,800,000
128,000,000
185,000,000
196,000,000
196,000,000
136
297
919
84,791,912
111,460,668
•Standard designated by: Annual PMjo Cone./24-Hr PMjQ Cone.
-------
IV-60
TABLE IV.F.2
COMPARISON OF PEOPLE EXPOSED FOR ALTERNATIVE PM10 STANDARDS
Annual Exposures with No Smoking
Year: 1989
Concen-
tration
(ug/m3)
50
45
40
35
30
25
20
15
10
5
0
Number of
Std.*=55/ -
226,000
3,330,000
17,800,000
48,900,000
93,500,000
150,000,000
188,000,000
196,000,000
196,000,000
196,000,000
196,000,000
People Exposed At
55/250
224,000
3,310,000
17,700,000
48,600,000
93,200,000
149,000,000
188,000,000
196,000,000
196,000,000
196,000,000
196,000,000
or Above Gi ven
55/200
210,000
3,140,000
17,100,000
47,400,000
92,000,000
149,000,000
188,000,000
196,000,000
196,000,000
196,000,000
196,000,000
Cone.
55/150
58,700
1,670,000
11,400,000
36,800,000
80,500,000
143,000,000
186,000,000
196,000,000
196,000,000
196,000,000
196,000,000
High Cone. 52
No. of Counties:
Scenario B 161
No. Seen. B 1,055
Population:
Scenario B
No. Seen. B
64,858,200
131.394,380
52
163
1,053
65,056,201
131.196,379
52
182
1,034
67,618,417
128.634,163
52
297
919
84,791,912
111,460,668
Standard designated by: Annual PM^g Cone./24-Hr PMjg Cone.
-------
IV-61
TARLE IV. F.3
COMPARISON OF PEOPLE EXPOSED TO PM10 FOR TSP & ALTERNATIVE PM10 STANDARDS
24-Hour Exposures with No Smoking Year: 1989
Concen-
tration Std.*
(ug/rc3)
300
280
260
250
240
225
220
200
180
160
140
120
100
80
60
40
20
0
Highest Cone.
No. of Counties:
Scenario B
No Seen. B
Population:
Scenario B
No Seen. B
Number of People
= TSP 75/260
763
20,400
85,200
110,000
123,000
135,000
140,000
161,000
364,000
1,070,000
2,650,000
5,690,000
22,000,000
54,600,000
116,000,000
181,000,000
196,000,000
196,000,000
314
279
936
88,585,773
107,649,128
Exposed At or
PM10 55/200
0
0
0
0
0
0
0
0
369
452,000
3,830,000
15,300,000
35,500,000
74,100,000
134,000,000
186,000,000
196,000,000
196,000,000
180
182
1,034
67,618,417
128,634,163
Above Gi ven Cone.
PM10 55/150
0
0
0
0
0
0
0
0
0
0
0
3,350,000
26,600,000
59,800,000
128,000,000
185,000,000
196,000,000
196,000,000
136
297
919
84,791,912
111,460,668
*TSP Standard: Annual geometric mean TSP Cone./2nd Highest TSP 24-Hr Cone.
Standard: Annual arith. mean PMjg Cone./expected value 24-Hr PM^Q Cone.
-------
IV-62
TABLE IV.F.4
COMPARISON OF PEOPLE EXPOSED TO PM10 FOR TSP & ALTERNATIVE PN^g STANDARDS
Annual Exposures v/ith No Smoking Year: 1989
Concen-
trati on
60
55
50
45
40
35
30
25
20
15
10
5
0
Number of People
Std.* = TSP 75/260
9,690
61,600
249,000
770,000
2,800,000
11,500,000
36,100,000
79,700,000
133,000,000
177,000,000
195,000,000
196,000,000
196,000,000
Exposed At or Above
PM10 55/200
0
0
210,000
3*140,000
17,100,000
47,400,000
92,000,000
149,000,000
188,000,000
196,000,000
196,000,000
196,000,000
196,000,000
Gi ven Cone.
PM10 55/150
0
0
58,700
1,670,000
11 ,400,000
36,800,000
80,000,000
143,000,000
186,000,000
196,000,000
196,000,000
196,000,000
196,000,000
Highest Cone.
No. of Counties:
Scenario B
No Seen. B
Population:
Scenario B
No Seen. B
62
279
936
88,585,773
107,649,128
52
182
1,034
67,618,417
128,634,163
52
297
919
84,791,912
111,460,668
*TSP Standard:Annual geometric mean TSP Cone./2nd Highest TSP 24-Hr Cone.
Standard: Annual arith. mean PM^g Cone./expected value 24-Hr PM^g Cone.
-------
IV-63 ,
TABLE IV.F.5
COMPARISON OF POPULATION WEIGHTED EXPECTED SECOND HIGHEST 24-HOUR PM10
CONCENTRATIONS FOR TSP AND ALTERNATIVE PM10 STANDARDS
Year: 1989
Concen-
tration
(ug/m3)
690
600
500
400
340
320
300
280
260
240
220
200
180
160
140-
120
100
80
60
40
20
0
Poeple Exposed
Base Case
73,800
122,000
748,000
921,000
7,710,000
10,300,000
15,200,000
20,600,000
34,500,000
38,200,000
43,100,000
49,100,000
57,100,000
75,100,000
88,800,000
109,000,000
143,000,000
169,000,000
191,000,000
196,000,000
196,000,000
196,000,000
At or Above
TSP 75/260*
0
0
0
0
240,000
350,000
350,000
350,000
350,000
350,000
1,770,000
2,880,000
6,540,000
9,670,000
23,900,000
67,100,000
117,000,000
160,000,000
184,000,000
193,000,000
196,000,000
196,000,000
Given Cone, i f
PM10 55/200
0
0
0
0
0
0
0
0
0
0
0
3,380,000
10,100,000
34,500,000
58,000,000
96,100,000
141,000,000
168,000,000
190,000,000
196,000,000
196,000,000
196,000,000
all at Monitor
PM10 55/150
0
0
0
0
0
0
0
0
0
0
0
0
0
0
37,800,000
87,900,000
135,000,000
166,000,000
188,000,000
196,000,000
196,000,000
196,000,000
Highest Cone.
No. of Counties:
693
348
200
151
Scenario B
No Seen. B
Population:
Scenario B
No Seen. B
0
1,216
0
196,252,580
279
936
88,585,773
107,649,128
182
1,034
67,618,417
128,634,163
297
919
84,791,912
111,460,668
*TSP Standard: Annual geometric mean TSP Cone./2nd Highest TSP 24-Hr Cone.
Standard: Annual arith. mean PM^g Cone./expected value 24-Hr PM^g Cone.
-------
IV-64
TABLE IV.F.6
COMPARISON OF POPULATION WEIGHTED EXPECTED SECOND HIGHEST 24-HOUR PM10
CONCENTRATIONS FOR TSP AND ALTERNATIVE PM10 STANDARDS
Year: 1989
Concen-
tration
(ug/m3)
180
150
100
80
65
60
55
50
45
40
35
30
25
20
15
10
5
0
Poeple Exposed
Base Case
48,400
175,000
7,140,000
34,500,000
47,100,000
51,700,000
61,700,000
74,300,000
92,200,000
114,000,000
140,000,000
173,000,000
187,000,000
194,000,000
196,000,000
196,000,000
196,000,000
196,000,000
At or Above Gi
TSP 75/260*
0
0
0
0
2,010,000
2,050,000
2,940,000
7,170,000
45,000,000
98,200,000
132,000,000
169,000,000
187,000,000
194,000,000
196,000,000
196,000,000
196,000,000
196,000,000
ven Cone, i f all
PM10 55/200
0
0
0
0
0
0
14,300,000
51,700,000
77,300,000
109,000,000
138,000,000
171,000,000
187,000,000
194,000,000
196,000,000
196,000,000
196,000,000
196,000,000
at Monitor
PM10 55/150
0
0
0
0
0
0
1,600,000
37,800,000
61,700,000
95,400,000
132,000,000
170,000,000
185,000,000
193,000,000
196,000,000
196,900,000
196,000,000
196,000,000
Highest Cone.
No. of Counties:
185
66
56
56
Scenario B
No Seen. B
Population:
Scenario B
No Seen. B
0
1,216
0
196,252,580
279
936
88,585,773
107,649,128
182
1,034
67,618,417
128,634,163
297
919
84,791,912
111,460,668
*TSP Standard:Annual geometric mean TSP Cone./2nd Highest TSP 24-Hr Cone.
Standard: Annual arith. mean PM^g Cone./expected value 24-Hr PM^g Cone.
-------
IV-65
G. Resource Impact Analysis
This section addresses the impact on public agency resources which a
PM^o standard might have. The analysis is limited to State and local air
pollution control agencies and to EPA Regional Offices. The estimates
presented serve to bound the relative impacts by examining the most and
least stringent PMig alternatives considered in the standard review process,
For comparative purposes, the impacts associated with the current TSP
standards are also provided. The level of resources needed to implement
revised PM standards will be determined not only by their level but by the
way they are implemented.
State and local agency resources were estimated using EPA's Air
Pollution Control Strategy Resource Estimator (APCSRE) model (Donaldson,
1981). This model estimates resource needs as a function of the number of
sources and the number of monitors. In addition an estimate is made of the
cost of developing new control strategies. The estimate for control
strategies is especially uncertain, because no PM^g strategies have yet
been developed. The level of resources needed for monitoring activities
was modified to reflect more detailed estimates made by EPA's Monitoring
and Data Analysis Division (MDAD) (Sleva, 1982 and Sleva, 1983). Table
IV.F.I below shows the modified results of the model runs. It shows that
over the long run any of the alternatives considered would require fewer
resources to implement than the current TSP standard.
-------
IV-66
Table IV.G.I
State and Local Agency Resource Needs
PMig NAAQS
($million)
1st year 2nd year 3rd year
PM10 (55/150) $49.9 $47.4 $40.4
PMlO (90/300)* $35.6 $33.4 $28.2
TSP (75/150) $45.6 $45.6 $45.6
However, the low end of the range initially will require more State and
local resources than the current standard but by the third year shows a
drop in need. The high initial costs reflect the fact that many States
will need to design new SIPs, purchase monitoring equipment, and develop
new inventories. The high end of the range is considerably less from the
outset and by the third year would result in an approximate 38% drop in
State and local need.
Several factors underlie these lower costs. In the first place,
although these agencies will have first year expenses in buying new
monitoring equipment for PMig, over the long run monitoring and laboratory
expenses are expected to be considerably less than with TSP. This is due
to the fact that a smaller network is currently being planned for PM^o-
The estimates presented here are based on the assumption that there will
be 200 NAMS sites and 500 SLAMS sites. In addition, at higher levels of
a PM^Q standard less source compliance and enforcement activity will be
needed. First year costs will also be high because of the need to develop
control strategies and SIP regulations. After the first several years
these costs will also fall.
The initial cost of samplers is somewhat uncertain since the precise
Federal Reference Method and sampling frequency has not been established.
*NOTE:This alternative was not featured in the previous cost discussion
because benefit and benefit/cost analyses were not performed for this option.
-------
IV-67
In these estimates it was assumed that a 1 in 6 day sampling frequency
would be maintained and that existing hi-volume samplers would be fitted
with size selective inlets. It is quite possible that everyday sampling
may be required for areas close to or above the standard. The number of
monitoring sites which might be involved in more frequent sampling has
not been determined yet. However, on a per site basis the annual operating
cost would likely rise from $2,000 to between $6 and $8,000. Depending
on the number of instruments per site capital costs could go from $1,000
to $2,000.
In addition to State and local agency resources, an assessment was
made of EPA Regional Office resource requirements. In the case of the
Regional Office the estimates are based on the Agency's workload model
used in planning and allocating Agency resources (Renner, 1982). The
Regional Office, workload estimates show that the PM}Q alternative of
55/150 would be roughly equivalent to the current standard over the long
run. However, in the second and third years as SIPs are processed and
promulgated, PM^g costs would be significantly greater than current TSP
costs. The upper end of the range (90/300) would over the long run result
in an approximate 85% drop in Regional needs.
Table IV.G.2
EPA Regional Office Resource Needs
PM10 NAAQS
($thousands)
PM10 (55/150)
PM10 (90/300)
TSP 75/150)
1st year
$ 700
$ 350
$2,450
2nd year
$3,850
$ 575
$2,450
3rd year
$4,550
$ 700
$2,450
4th year
$2,450
$ 350
$2,450
-------
IV-68
It should be noted, however, that the models and analytic techniques used
to prepare the estimates above do not permit them to be compared directly
to the other cost estimates. Specifically, these estimates do not reflect
a stream of payments from 1989 (87) to 1995. Therefore in Section VII
below, implementation costs were assumed to be 4% of the control cost.
This assumption was based on an analysis of administrative and management
costs (Vatavuk, 1981). A comparison was made of the impact on net
benefit/cost of using either estimate. It was found that neither estimate
changed final conclusions, since these costs are minor compared to total
costs and total benefits.
-------
IV-69
1. Arthur D. Little, Inc., (1981). Environmental Policy for the 1980's:
Impace on the American Steel Industry, report to The American Iron
and Steel Institute, 1981.
2. Biller, W.F. , (1983). Estimated Population Exposures to Particulate
Matter for Alternative Standards, contractor report, May 1983.
3. Donaldson, T. , (1981). Estimated Resource Needs of State and Local
Agencies for Implementing Ambient Inhalable Particulate Matter
Standards Within the Range of 55 ug/m-* and 110 yg/m3 for Particles
_< 10 urn, Control Programs Development Division, September, 1981.
4. Energy and Environmental Analysis (1981), Documentation of the TSP
Nationwide Model Used in Particulate Ambient Standards Analysis,
final draft report June 22, 1981.
5. Energy and Environmental Analysis (1981). A Nationwide Study of
Control Strategies to Attain Current Ambient TSP Standards, final
draft report, September 22, 1981.
6. Energy and Environmental Analysis, (1981). S02/TSP Regulatory Analysis;
Emission Factors and Control Cost Methodology final draft report,
April 17, 1981.
7. Johnson, Ted, and Roy Paul (1981). The NAAQS Exposure Model (NEM)
and Its Application to Particulate Matter, draft report PEDCo
Environmental, Inc., August 31, 1981.
8. Pace, Thompson G. , Air Management Technology Branch, 1981. Memorandum
to Henry Thomas. Subject: Review of the Relationships of IP 10, IP
15 and TSP. July 6, 1981.
9. PEDCo Environmental, Inc., (1981). Alternative NAAQS for Particulate
Emissions from Steel Industry Sources: Impacts on an Example Plant,
final report, October 1981.
10. PEDCo Environmental, Inc., (1982). Estimate of Steel Industry Control
Costs: 1978-1985, draft report February 1982.
11. Renner, Fred, Office of Regional Programs, (1982). Memorandum to
Henry C. Thomas. Subject: Regional Office Resource Estimates for
Three Candidate PMio Ambient Standards. February 10, 1982.
12. Sleva, Stanley, Chief, Monitoring Section, (1982). Memorandum ;to
Henry C. Thomas. Subject: PM^g Monitoring Costs. November 18,
1982.
13. Sleva, Stanley, Chief, Monitoring Section, (1983). Memorandum to
Henry C. Thomas. Subject: TSP Annual i zed Operating Costs.
February 18, 1983.
-------
IV-70
14. Smith, A.E., and K.L. Brubaker, 1982. Costs and Air Quality Impacts
of Alternative National Ambient Air Quality Standards for Particulate
Matter, Technical Support Document, Argonne National Laboratory,
October 1982.
15. Steckler, Steve, Air Economics Branch, 1982. Memorandum to Henry
Thomas. Subject: Analysis of Ambient Diesel Particulate Projections.
March 26, 1982.
16. Temple, Barker, & Sloane, Inc., (1982). An Economic Analysis of
Final and Pretreatment Standards for the Iron and Steel Manufacturing
Point Source Category, unnumbered report, January 28, 1982.
17. Vatavuk, William, Economic Analysis Branch, 1981. Memorandum to Tom
Walton. Subject: Development of AME Estimates. April 13, 1981.
-------
V. Economic Impact Analysis
A. Introduction
The Economic Impact Analysis (EIA) addresses the direct resource
allocation consequences of the costs discussed in Section IV of the RIA.
The following sub-sections address: Purpose (B), Background (C),
Methodology (D), Other Qualifications (E), Findings (F), and Conclusions
(G) of the EIA.
B. Purpose
The objectives of the economic impact analysis (EIA) for the re-
consideration of the present national ambient air quality standards (NAAQS)
for particulate matter (PM) are two-fold. The first objective is to
fulfill requirements of Executive Order 12291 by describing the adverse
economic effects of the direct pollution control costs. The second is
to assess the relationship between direct pollution control costs and
cost requirements for the benefit-cost analysis.
In order to accomplish the first objective, economic impacts of
pollution control costs are calculated for selected 4-digit SIC (Standard
Industrial Classification) code industries. Among all industries requiring
control, these may experience relatively greater impacts as a result of
the PM NAAQS. The following categories of impacts are evaluated for the
selected industries: product prices; product output; industry employment;
industry structure (e.g., plant closures, small business impacts);
industry investment, productivity, and innovation; and, foreign trade.
The EIA assesses the relationship between the direct pollution
control cost and the cost required in the benefit-cost analysis by answering
-------
V-2
two questions. First, how do the post-control production volumes of
emitters compare with the pre-control production volumes assumed in the
direct pollution control cost analysis? Second, as a result of additional
control requirements, how do the costs in the transition period (from
pre-control to post-control market equilibrium) correspond to the zero
cost transition assumed using comparative static analysis?3
C. Background
1. Breadth of EIA
Industries affected by the most stringent NAAQS alternative
for particulate matter (i.e., the present primary and secondary standards)
are found in almost every sector of the U. S. economy. Direct control
costs are anticipated to be faced by an estimated 280 industries, when
industries are defined at the 4-digit level of the SIC. Affected
industries include minerals, manufacturing, construction, services, and
commercial and institutional establishments. Also affected are agencies
of municipal, county, state, and federal governments.
2. Control Cost Model
The cost model in Section IV estimated aggregate costs of
attainment by state, for each 4-digit SIC industry. These data were
used in the screening analysis (described later in Section V D) to identify
industries requiring a detailed economic impact assessment.
The EIA utilizes 4 outputs of the cost model: capital costs;
operating and maintenance costs; before-tax annualized costs; and, after-
a Comparative static analysis focuses on equilibrium values before and
after a change. It does not deal with intermediate values during the
change.
-------
V-3
tax annualized costs. The cost data, in constant 1980 dollars, are
aggregated to state and national levels. Capital costs are assumed to
be incurred at the beginning of the attainment year (1987 or 1989,
depending on whether a TSP or PM^g standard is being considered).
Operating and maintenance costs are assumed to be incurred in the
attainment year, and then to continue at the same level for the 15-year
life of the control equipment.
Development of Costs for "Model Plants". The cost model (See
Section IV for more details) used data from the National Emissions Data
System (NEDS) to estimate the direct costs of pollution control. The
NEDS data include source-specific engineering parameters (e.g., fuel use
and stack height for a boiler). Affected plants (i.e., those incurring
additional cost for pollution control) are identified by only a 4-digit
SIC code and a geographic location (i.e., county or city). The NEDS
data used in the cost model do not specify production capacity and utili-
zation, ownership type, and other data that would be desirable for the
EIA. Consequently, the EIA had to depend on the construction of "model
plants" to represent the "average" scale and product mix of all firms in
an industry.
Distribution of Additional Costs for Reaching Complete Attainment.
The initial control strategies (Scenario A) defined in the cost model
result in residual non-attainment. A side calculation was made to estimate
the additional costs of reaching complete attainment (Scenario B) in
those geographical areas where the pollution control is applied. However,
these additional costs are specified only in terms of dollars per unit of
-------
V-4
air quality improvement. And, hence, they cannot meaningfully be ascribed
to specific industries, let alone specific plants. Although the cost of
removing residual non-attainment cannot be incorporated in the El A, it is
included in the benefit-cost analysis (see Section VII).
Adjustment of Costs for Retirement/Replacement of Plants.
In Scenario A the cost model adjusts aggregate control costs downward
for retirement of old plants and their replacement by new ones subject
to new source controls. However, lack of data precluded meaningful
adjustments to costs at the individual plant level. Consequently,
aggregate industry control costs in the El A exceed aggregate control
costs from the cost model for an otherwise identical scenario. That is,
some plants characterized as affected in the EIA will never actually
have to incur control costs because of retirement or modification which
makes them subject to new source controls (which are independent of the
NAAQS). Moreover, some of the affected plants identified in the study
may have already undertaken pollution control investments since 1978.a
The effect of not adjusting for retirement and replacement in the EIA is
to provide an upward bias to the projections of adverse impacts resulting
from the pollution control costs associated with the current PM NAAQSs.
Assumptions of the Least-Cost Control Strategy. The cost
model attempts to simulate decision-making behavior that minimizes
direct control costs. Limitations of the cost model and input data
preclude estimation of a true least-cost strategy. On the other hand,
the states may design their implementation plans with some objective(s)
aControl requirements as of 1978 form the baseline control level for
existing plants from which additional controls requirements are estimated.
-------
V-5
other than direct control cost minimization. For example, the states
may wish to minimize certain types of economic impacts, or to reduce
administrative and enforcement costs. Consequently, for a particular
plant or industry, actual control costs may be less or greater than the
control costs used in the EIA.
D. Methodology
1. Screening Analysis
As noted previously, the most stringent alternative standards
(the present PM NAAQS) are expected to impose control costs on 280 indus-
tries (at the 4-digit SIC code level). Given the number and diversity
of affected industries, a full-scale economic analysis for each industry
is not feasible. A screening analysis was conducted to identify those
industries that might experience relatively large adverse economic
impacts. In the screening procedure, alternative measures of economic
performance are compared in order to isolate industries that are likely
to experience relatively large economic impacts. The screening ratios
used in the analysis are described below.
o The first is the ratio of capital costs of additional pollution
control equipment to capital expenditures: a rough indicator
of the magnitude of required pollution control capital expendi-
tures compared to the magnitude of general capital expenditures
for the industry. This measure is limited by the fact that
the capital expenditures in the denominator are for one specific
year which therefore may or may not be typical.
o The second is the ratio of before-tax annualized costs of
additional pollution control to the value of shipments: an
indicator of the maximum price increase assuming the cost of
control is completely passed on to buyers of the industry's
products.
o The third is the ratio of the after-tax annualized cost of
additional pollution control to after-tax profits that results
if costs are completely absorbed: an indication of potential
cost absorbtion burden. This measure is limited by the use
-------
V-6
of profits for one specific year which therefore may or may
not be typical.
o The fourth is the ratio of after-tax annualized cost of pollution
control to value added: this measure provides another indicator
of the impact on the firm if the additional control cost is
absorbed.
Each screening ratio is calculated twice for an industry.
First, the ratios are calculated using national aggregate cost data.
Then they are also estimated at the individual state level.
All eight ratio calculations are used to help select the indus-
tries to be analyzed in more depth. By assigning weighted scores to the
industries ranking in the top 15 for each ratio (e.g., 15 for 1st, 1 for
15th), a total score for each industry is compiled. The total score is
then used as the main criterion to determine those industries for which
economic impacts are likely to be more adverse.
Sixteen 4-digit SIC industries, and parts of the public
sector, were identified by the screening study for more detailed analysis
in the EIA. Included in these 16 industries are the 8 industries
with the highest scores from the screening study. Also included are
several industries that are important because of the absolute magnitude
of the estimated control costs they could incur, their vulnerability to
competition from other products and/or imports, or because they may face
significant control costs due to other environmental regulations.
2. Industry Impacts
The analysis for each industry begins with a descriptive
profile. The profile is used to make an assessment of market structure,
trends in output and demand, employment, capacity, plant size, prices,
firm profitability, and the importance of foreign trade. Because of the
-------
V-7
uncertainty regarding the characteristics (i.e., size, production,
ownership, mix of products, etc.) of the affected plants, the EIA constructs
"model plants" (as discussed previously). Each model plant has a balance
sheet and income statement, developed from industry data. Costs of the
present PM NAAQSs are applied to the income statement and balance sheet
so that the incremental effect of the regulation can be estimated.
Detailed economic impacts are calculated for all 16 industries under
Scenario A for the current PM NAAQS--the "base case."
All other NAAQSs considered in the Regulatory Impact Analysis
require less ambient air quality improvement, suggesting smaller economic
impacts than the base case.
The current PM NAAQSs regulatory alternative is likely to have
different impacts across states and counties because of variations in
the degree to which emission controls are required in each geographic
region to comply with the NAAQS. Thus national average control costs
are substantially lower than the control costs faced by plants in some
non-attainment regions. In order to take account of these geographic
variations, economic impacts are computed for each industry for each of
the following two cases. In the first case, the "model plants" are
assumed to incur national average control costs for the plants subject
to control in the industry. In the second case, the "model plants" are
assumed to incur control costs faced by plants located in the state with
the highest average control costs for the industry.
The economic impacts for the above cases are estimated under
the following assumptions about market conduct. In the first, firms are
-------
V-8
assumed to pass through all costs. In the cost absorption case, the EIA
examines a "worst-case" decline in firm profitability, measured by change
in net income. In the cost pass-through case, the EIA examines a "worst-
case" price increase. These 2 cost pass-through cases encompass the
range of possibilities. A qualified judgement is provided to indicate
the most likely cost pass-through situation. The most likely situation
is based on the particular industry's structure and market circumstance.
With the most likely situation identified, the EIA then assesses
additional economic impacts. They include possible plant closures and
impacts on productivity, innovation, foreign trade, and small businesses.
3. Use of Direct Pollution Control Costs for the Benefit-Cost
Analysi s: Methodology
The benefit-cost analysis uses the direct control cost developed
in the cost model. However, the appropriate measure of cost in a
benefit-cost analysis should measure the loss in economic surplus rather
than the direct control cost.3 Under certain conditions, direct control
cost serves as a good proxy for the change in economic surplus. For
example, if prices are marked up to cover control costs and passed forward
with no decrease in market equilibrium output quantity, the direct pollu-
tion control costs and the change in economic surplus are identical.b In
aEconomic surplus is the sum of producer and consumer surplus. Consumer
surplus is the difference between the maximum amount consumers are willing
to pay for a given quantity of a good and the amount the consumers actually
pay. Similiarly, producer surplus is the difference between the amount
producers actually receive for the given quantity of a good and the
minimum amount for which the producers would be willing to produce that
quantity.
bThe key assumption here is that direct pollution control costs reflect
economic costs not engineering or historical costs.
-------
V-9
competitive situations with quantity adjustments, direct pollution control
costs over-estimate the loss in economic surplus. In these cases, market
equilibrium output quantities decline and/or affected plants close. In
noncompetitive situations, the direct pollution control cost may either
underestimate or overestimate surplus changes when there is a quantity
adjustment as a result of the control costs. For regulated monopolies
however, the direct pollution control cost will most likely result in an
overestimate of surplus changes when there is a quantity adjustment.
The market structure of each individual industry and the distribution of
direct control costs form the basis of judgements regarding the validity
of using direct control cost in the benefit/cost analysis.
Transition cost refers to the valuation of resources required to
get from one state (e.g., pre-additional control) to another (e.g.,
post-additional control). In particular it may include such short run
phenomena as employee and capital displacement. The comparative static
framework of benefit-cost does not provide for direct inclusion of
transition cost in the analysis. But, transition cost may be indicative
of some of the distributional consequences of alternative NAAQSs. The
detailed analyses for 16 industries (e.g., industry production volume
adjustments and employee displacements) are used to reach a judgement
regarding the relative size of the transition cost.
E. Other Qualifications
The results of the economic impact analysis (EIA) are subject
to many qualifications. First, the EIA examines each industry separately.
As such, it is a partial equilibrium analysis. Hence, it does not capture
beneficial or adverse impacts due to interactions among industries producing
-------
V-10
complements and substitutes. Additionally, the EIA does not address
impacts on the suppliers of materials and other factor inputs purchased
by the affected plants. This partial equilibrium analysis results in an
incomplete categorization of impacts, and can imply biases in either
direction for those impacts that are addressed.
The EIA does not address the positive impacts of control costs on
firms not incurring these costs. For example, while those plants which
incur control costs may suffer reduced profitability, other plants in
that same industry may increase their profitability as a direct con-
sequence of product price increases by controlling plants. The EIA does
not address this increased profitability. A second example of a positive
consequence of control costs is the impact on firms which provide equip-
ment and services for pollution control.
Another category of economic impacts not addressed in this EIA
stems from the benefit side -- the improvement in air quality itself.
These impacts may be positive for some (e.g., increased property values
for owners of property where air quality improves) and negative for
others (decreased revenues for medical service suppliers due to a re-
duction in pollution-related illnesses).
F. Findings
This portion of Section V summarizes the findings of the analysis
for the 16 industries selected for more detailed investigation. It also
discusses the implications of the EIA for the benefit-cost analyses
presented in Section VII.
1. Industry Impacts Under "Base Case" Regulation
Table V-l shows the number of affected plants (i.e., those
-------
V-ll
TABLE V-l. INDUSTRIES COVERED IN THE ECONOMIC ANALYSIS
Description
Steel
Primary smelting of copper
Primary smelting of lead
Primary smelting of aluminum
Electric utilities, steam supply
Cement, hydraulic
Crushed and broken limestone
Crushed and broken stone, NEC
Construction sand and gravel
Paving mixtures and blocks
Cut stone and stone products
Minerals and earths
Grain and flour milling
Gray iron foundries
Lime
Government sector
SIC
3312
3331
3332
3334
4911, 4961
3241
1422
1429
1442
2951
3281
3295
2041
3321
3274
_.—
Number of
affected
plants
75
8
4
4
243
67
62
34
61
384
23
161
65
46
22
__
Affected plants
as a percent
of total plants
15
30
57
13
39
41
4
8
2
38
2
35
16
14
14
__
Technical
appendix
chapter
number
2
3
3
3
4
5
6
6
7
8
9
10
11
12
13
14
-------
V-12
requiring additional pollution control) in each of the 16 industries,
assuming "base case" regulation (defined previously). The ratios of
affected plants to total plants is highly variable by industry, ranging
from 2 percent to 57 percent.
Table V-2 estimates increases in production costs attributable
to pollution control. The figures are for "model plants" (defined
previously) in each industry, under each of the following (previously
noted) assumptions:
o The plant incurs control costs equal to the industry-wide
(i.e., national) average for all affected plants in that
industry.
o The plant incurs control costs equal to those in the
highest-cost (average per plant) state.
The use of "model plants" provides a measure of the average
production cost increases incurred by affected plants. Of course,
production cost increases for actual plants will vary according to their
scale and the magnitude of control costs they face.
To improve characterization of cost absorption versus cost
pass-through (to consumers in the form of higher prices), judgements
were made on the basis of the industry profiles and cost distribution.
The 16 industries were classified into 3 groups:
o One, industries with firms that are likely to absorb
control costs, creating the possibility of plant closures;
o two, industries with firms that are likely to absorb
control costs with little potential for plant closure;
and,
-------
V-13
TABLE V-2. POTENTIAL INCREASE IN PRODUCTION COSTS DUE TO NAAQS CONTROL COSTS
Description
Steel
Primary smelting of copper
Primary smelting of lead
Primary smelting of aluminum
Electric utilities
Steam supply
Cement, hydraulic
Crushed and broken limestone
Crushed and broken stone, NEC
Construction sand and gravel
Paving mixtures and blocks
Cut stone and stone products
Minerals and earths
Grain and flour milling
Gray iron foundries
Lime
SIC
3312
3331
3332
3334
4911
4961
3241
1422
1429
1442
2951
3281
3295
2041
3321
3274
National
average3
(Percent)
O.?b
0.3
0.1
0.1
0.6C
0.3C
2.3
31.1
18.3
13.0
3.1
26. 6d
4.3
2.2^
3.0
10.4
Highest -Cost
state3
(Percent)
2.8b
1.9
0.1
0.2
2.7C
1.9C
41.5
80.4
89.9
37.5
13.1
41. 9d
66.4
4.2d
28.4
28.4
aAt an average-size plant (except as noted below). Increases are at affected
plants only.
t>At an average-size plant among integrated and partially integrated steel companies.
cAt an average-size private utility company, assuming one affected plant per utility.
dAt an average-size plant; the technical chapter also presents figures for larger .
plants.
-------
V-14
o three, industries with firms that are likely to at least
partially pass through control costs.
Impacts on Group I. Table V-3 summarizes the "base case"
NAAQSs for industries in Group I. Affected plants in Group I are unlikely
to be able to pass through costs of control to the consumer, and are
likely to experience relatively large declines in net income. An upper
bound by industry of potential closures ranges between 3 percent and 100
percent of all affected plants. However, these potential closures would
be 2.5 percent or less of all the plants in each industry. Furthermore,
if closures occur, they will result in a transfer of production and
employment within the impacted industry. Consequently, there is not
likely to be any net change in either output or employment on an industry-
wide basis.
Impacts on Group II. Table V-4 summarizes the "base case"
NAAQS for industries in Group II. Affected plants in this group are
unlikely to be able to pass through control costs in the form of higher
prices to consumers. However, the decreases in net income because of
cost absorption are not large enough to suggest plant closure. Decreases
in net income range from or 3 percent to 68 percent for plants in the
highest-cost state. No price changes, output changes, or plant closures
are anticipated because projected revenues for the firms cover the estimated
variable production costs.
Impacts on Group III. Table V-5 summarizes the "base case"
NAAQS for industries in Group III. Industries in this group have vary-
ing capabilities for passing control costs through to the consumer.
-------
TABLE V-3. IMPACT SUMMARY FOR GROUP I
Impacts
1422
Crushed and
broken
limestone
1429
Crushed and
broken
limestone Nee
Industry3
3281
Cut stone
and stone
products
3295
Minerals
and
earth
2041
Grain and
flour
milling
3321
Gray
iron
foundries
Change in price (%) at affected 0
plants when control costs equal
the national average
Change in price (%) at affected 0
plants in the highest-cost state
Potential change in industry 0
production (%)
Potential plant closures (number) 18
Potential plant closures (% of 29
affected plants)
Potential plant closures (X of 1
all plants in industry)
Potential number of employees 270
displaced*'
Decrease in net Income (%) at 100
affected plants in the
highest-cost state
3
9
45
100
0
0 to 23
0 to 100
0 to 2
300
100C
5
3
130
100
0
0
10
16
2.5
400
90C
0
0
0 to 8
0 to 17
0 to 1
0 to 1,800
100
affecterl plants in these industries are assumed not to be able to pass through control costs. Closures are
possible at some plants.
^Displaced employees are employees at plants that might close. The figure does not include employment increases
at other plants necessary to maintain industry output. The employee displaced figure is obtained by multiplying
the number of anticipated closures by the average number of employees per plant.
cFigures shown are for average plants. Technical chapters also show figures for large plants.
-------
TABLE V-4. IMPACT SUMMARY FOR GROUP II
Impacts
3331
Primacy
3312 smelting
Steel of copper
Industry8
3332
Primary
smelting
of lead
3334
Primary
smelting
of aluminum
Change In price (Z) at affected
plants when control costs equal
the national average
Change In price (%) at affected
plants In the highest-cost state
Potential change In Industry
production (7,)
Potential plant closures (number
Potential plants closures
(% of affected plants)
Potential plant closures
(% of all plants In Industry)
Potential number of employees
dlsplaced
Decrease In net Income (%) at
affected plants In the highest-
cost state
0
0
0
0
0
0
53
0
0
0
0
0
68
0
0
0
0
0
0
0
i
i—*
en
aAffected plants In these Industries are assumed to not be able to pass through control costs. No plant
closures are likely because revenues are expected to cover costs (net Income remains positive).
-------
TABLE V-5. IMPACT SUMMARY FOR GROUP III
Industry3
Impacts
1442
Construction
sand and gravel
4911, 4961
Electric utilities
including steam supply**
2951
Paving
mixtures
and blocks
3274
Lime
3241
Cement
hydraulic
Change in price (%) at affected
plants when control costs equal
the national average
Change In price (%) at affected
plants in the highest-cost state
Potential change in industry
production (%)
Potential plant closures (number)
Potential plant closures (Z of
affected plants)
Potential plant closures (% of
all plants in Industry)
Potential number of employees
dlsnlacedc
Decrease In net income (%) at affected
plants In the highest—cost state
0.6
15
0.1
9
15
0.13
114
100
11
0.5
0
0
2,110
1
0 to 39
0 to 10
0 to 4
620
80
0
0
60
38
1
1
2
330
100
"Affected plants in these Industries are assumed to have various pass-through ability. Utilities are assumed to
pass through 100 percent of costs for SIC 4911 plants and 50 percent of costs for SIC 4961 plants. Lime plants
are assumed to pass through 100 percent of costs at all plants, subject to a cap of no more than an 11 percent
price increase at anv plant. The national average hydraulic cement plant Is assumed to pass through 100 per-
cent of control costs, but the plant in the highest-cost state is assumed to pass through only 10 percent of
control costs. Construction sand and gravel plants are assumed to pass through 50 percent of control costs. The
national average paving mixtures and blocks plant Is assumed to pass through 100 percent of control costs, but
the plant in the highest-cost state Is assumed to pass through only 50 percent of control costs.
^Figures reflect one affected plant in each of SICs 4911 and 4961 at an average private utility.
cRecatise of predicted production volume decreases at the industry level, plant closures or operating rate
declines may.result in employment displacement. To avoid understating potential displacement, the figures
are calculated bv summing the predicted employment losses at the plants that are closure candidates and
the predicted losses associated with the forecast decline in industry production. The predicted dis-
placement figures should be viewed as upper bound estimates of gross or net employment changes.
-------
V-18
This uncertainty about the cost pass-through assumption makes the esti-
mated impacts more tentative than for the other two groups. Adverse
impacts for Group III include both price and closure effects as a result
of partial cost pass-through.
Impacts on all groups. Two sets of findings are common to each of
Groups I, II, and III. First, no significant impacts on foreign trade
are anticipated. Second, impacts on investment, productivity, and
innovation are probably adverse but small in magnitude.
Impact on Small Entities. Under the Regulatory Flexibility Act, 5
U.S.C., 600 et seq., the Agency must prepare a regulatory flexibility
analysis assessing the impact of any proposed or final rule on small
entities. Under 5 U.S.C. 605(b) this requirement may be waived if the
Agency certifies that the rule will not have a significant economic
effect on a substantial number of small entities. Small entities include
small businesses, small not-for-profit enterprises, and governmental
entities with jurisdiction over populations of less than 50,000.
EPA has made an effort to address the cost of control for small
entity groups. This effort is, however, limited and does not permit
definite findings with respect to all potentially affected small entities.
The extremely large number of potentially affected emitters of particulate
matter (over 280 different industries at the 4-digit SIC code level) led
to the development of a cost model based on the information from the
National Emission Data System (NEDS). NEDS does not provide, however,
the detailed information on plant size or ownership which is necessary
for a detailed assessment of the distribution of cost for each small
entity group. For the particulate matter model, completion of the small
entity analyses would have required a plant by plant matching of NEDS
with various economic data bases. The large number of plants and industries
-------
V-19
in question made this matching of data files, and therefore a detailed
small entity analysis, impractical at present.
The degree of control simulated by the cost model is dependent on
several factors. These include the required emission reduction for the
particular geographic area of the plant, the level of emissions of the
plant, and the costs of control of these emissions. These factors combine
to make it likely that, everything else being equal, a higher percentage
of larger plants will incur control costs than will smaller plants. If
smaller plants are a proxy for smaller business entities, then one would
also expect smaller affected businesses to be less likely to incur control
costs than their larger business counterparts.
In an attempt to determine if a substantial number of small entities
are potentially affected, the screening analysis (described in section
V.D.I) was used. This analysis selected 16 industries out of the 280
affected industries for more detailed analysis. The design of the screening
analysis is such that the industries selected for detailed analysis are
more likely to have a higher percentage of affected plants than the
nonselected industries. For the 16 industries as a group the percentage
of affected plants is less than 20 percent. Therefore, it is likely that
the percentage of small entities affected for the 280 industries as a
group is also less than 20 percent. This suggests i'.'Mt. t.h- percentage of
small entities affected within the 280 industries would not be substantial
(i.e., not greater than 20 percent).
Additionally, after promulgation of national ambient air quality
standards, the control measures necessary to attain and maintain them are
developed by the respective states as part of their state implementation
plans. In selecting such measures, the states have considerable discretion
so long as the mix of controls selected is adequate to attain and maintain
-------
V-20
significant affect on a substantial number of small entities then depends
to some extent on how the states would choose to implement it. For these
reasons, any assessment performed by EPA at this time is uncertain.
This preliminary assessment based on selected industries does not
suggest that the proposed revisions will significantly affect a substantial
number of small entities for the group of potentially affected industries.
This finding, however, is not conclusive.
2. Industry Impacts Under Regulation Other Than "Base Case"
As noted previously, the "base case" represents the most
stringent control alternative considered in the EIA. With two exceptions,3
the base case produces the highest cost of all regulatory alternatives
considered. Table V-6 presents costs for the base case, the next most
stringent (costly) alternative, and the least stringent (costly) alternative.
Compared with the base case, the other regulatory alternatives should
result in smaller increases in product prices, smaller declines in
profitability, reduced numbers of plant closures, and a reduced amount
of employee displacement.
3. Impacts on the Government Sector
The current PM NAAQSs also require compliance by the government
sector. Chapter 14 of the technical appendices provides a detailed
description of the government sector analysis. The methodology is
different than for an industry due to differences in financing and pricing
aNote the control costs for primary smelting of copper (3331) and minerals
and earth (3295 in Table 15-6. For these two industries, the PMio
standard of tg/m^ annual mean and 150 tg/nr> 24 hour expected value has a
greater capital cost than the base case capital cost.
-------
TABLE V-6. CAPITAL AND AFTER-TAX ANNUALIZED COSTS (ATAC).
OF THREE DIFFERENT NAAQS ALTERNATIVES
(1980 $106)
1422
1429
1442
2041
2951
3241
3274
3281
3295
3312
3321
3331
3332
3334
4911
4961
aNAAOS
bNAAOS
CNAAOS
Base case
Capital
58,220.6
20,900.3
7,951.8
71,153.6
63,774.3
103,782.3
38,874.8
10,152.5
60,156.2
644,549.9
85,667.7
16,013.9
0.0
663.7
1,818,903.3
70 519.4
cost3
ATAC
9,100.4
2,677.0
1,867.2
7,269.9
12,221.7
12,351.5
4,792.6
1,292.1
8,486.4
76,593.2
8,633.4
1,980.3
118.6
401.3
191,710.6
8,848.5
are: 75 |jg/mj annual mean; 150 pg/mj
are 55 pg/m^ annual mean
Second most stringent
costs**
Capital
14,068.8
6,945.8
9,161.1
21,534.1
39,658.3
76,105.6
22,028.0
4,890.0
61,189.8
481,388.8
75,050.3
17,869.8
0.0
492.0
1,102,918.5 1
68,934.6
24-hour observed 2nd high;
ATAC
2,392.2
1,159.5
1,700.3
2,231.6
7,412.4
9,120.4
2,713.2
553.0
7,839.4
57,914.4
7,533.3
2,086.1
73.3
184.4
16,954.5
8,043.8
total suspended
Least
stringent
costs0
Capital ATAC
135.0
1,323.8
104.0
533.1
5,987.0
1,161.2
1,614.0
2,070.5
5,595.6
6,650.7
4,021.0
1,542.9
0.0
0.0
179,529.3
339.6
partlculates.
15.0
181.0
179.0
60.9
1,045.3
426.7
191.8
246.6
712.4
1,129.3
408.9
348.2
67.1
0.0
18,489.6
108.3
; 150 pg/m^ 24-hour expected value; PMIQ-
are: 90 pg/m3 annual mean; 300 Pg/m3
24-hour expected value; PM
in'
I
rv>
-------
V-22
between the private and the public sectors. The principal object of
control at the level of state governments is road dust.
Two financial ratios were calculated to estimate the effects of the
present NAAQS standard on government finances: (1) capital costs of
control as a ratio of government outlays by the states, and (2) annualized
costs of control as a ratio of the sum of state and local tax revenues.
The first measure helps assess the effect of the NAAQS on governments'
capital budgeting decisions. The second helps determine what changes,
if any, are needed in tax rates to fund the cost of the NAAQS.
In all states, the ratio of capital costs to state outlay is
less than 0.2 percent. In all states, annualized costs are less than
0.03 percent of tax revenues. These figures imply that state governments
should not find it difficult to finance and pay for controls of road
dust. However, the estimated cost of these controls is directly dependent
on the assumptions made regarding the effective emission fraction (i.e.,
0.01). This assumption is discussed in Section IV.D. "Treatment of Area
Sources."
4. Use of Direct Pollution Control Costs for the Benefit-Cost
Analysis: Implications
The results of the EIA's for the 16 industries provide
qualitative information for the interpretation of control costs. The
control costs are assessed for potential upward bias and potential
limitations of not including transition costs.
The EIA's for Groups I and III show potential plant closures
and/or quantity adjustments under "base case" regulation. Thus, the
direct control costs may overstate the actual social costs of regulation
-------
V-23
in these two groups. For Group II, however, control costs appear to be
a good proxy for the decrease in economic surplus. Thus, on net, it
appears direct pollution control costs are greater than the social costs
of the regulation.
Transition costs are, by definition, omitted from a comparative
static benefit-cost analysis. They refer to short-run phenomena such as
the costs of employee displacement, and the costs of disequilibrium
price and quantity adjustments. For the sixteen industries analyzed
under "base case" regulation, transition costs do not appear large
enough to represent a serious omission from the benefit-cost analysis.
G. Conclusions
In the economic impact analysis, adverse impacts of the PM NAAQS
are estimated for 16 specific industries selected by a screening procedure.
Quantitative impacts are estimated for the industries under the most
restrictive or base case standard. Adverse impacts on the other industries
among the 280 industries affected by the base case standard are likely
to be less severe, since the screening identified the industries with
the most significant control costs. Virtually all industries under the
less stringent alternative standards should also have less adverse
impacts, if any.
For 10 of the 16 industries analyzed absorption of pollution control
with no price changes or net adjustments in production quantities is the
most likely outcome. Because affected plants absorbing control costs
would have lower profits, a significant portion of the affected plants
may become possible closure candidates. However, because most plants
in the industries with potential closures do not require control,
-------
V-24
the upper bound estimate for potential closures among all the firms from
any one industry is 4 percent. Also, any potential closures for the 10
industries are not expected to affect total industry output because the
loss production of any closing plant is expected to be offset by increased
output by remaining plants.
Firms in 6 of the 16 industries are judged to be able to pass
through a significant amount of pollution control costs associated with
the standards, resulting in price increases ranging from 0.6 percent to
7.0 percent and net decreases in industry production of no more than 1
percent. Hence, while the outlook for different industries varies, the
impacts resulting from PM NAAQS are unlikely to involve any major change
in industry production or structure.
Finally, use of direct pollution control cost for the benefit-cost
analysis may overestimate costs due to closures and quantity adjustments.
Furthermore, neglect of transition costs does not appear to significantly
impair the findings of the benefit-cost analysis.
-------
VI. BENEFIT ANALYSIS ESTIMATES
A. INTRODUCTION
This section presents a summary of the benefits which may be associated
with the alternative NAAQS for participate matter (PM). Benefits represent the
improvement in society's well-being as a result of lower PM concentrations.
The analysis is based on a study prepared by Mathtech, Inc. under contract
to EPA.* The purposes of this section are to present the analytic methodology
and the resulting benefit estimates for selected alternative courses of
action. The benefit analysis examined several alternative ambient PM
standards which are shown in Table VI.E.I.
While cost and environmental impacts have been traditionally accom-
plished by the Agency for other regulatory actions, the estimation of
economic benefits for ambient air standards was not a formal requirement
until the issuance of E.O. 12291 in February of 1981. Two of the reasons
the Agency has not accomplished them in the past are the analytic limita-
tions imposed by available data and methodologies and the difficulties
related to quantifying the economic value of health and welfare benefits.
In an effort to overcome these technical problems the EPA Office of Air Quality
Planning and Standards initiated a program in 1979 to quantify the economic
benefits attributable to welfare effects of air regulations. This program
produced reports on soiling, materials damage, and visibility benefits of
particulate matter reductions.
It is important to stress that the Mathtech study has not had a role
to date in the development of proposed revisions to the NAAQS for particulate
matter. Staff recommendations for revised NAAQS currently under consideration
*See Reference 1
-------
VI-2
are based on the scientific and technical information contained in two EPA
documents. They are the "Air Quality Criteria for Particulate Matter and
Sulfur Oxides"* and the "Review of the National Ambient Air Quality Standards
for Particulate Matter: Assessment of Scientific and Technical Information,
OAQPS Staff Paper."** These documents have undergone extensive review by
the public and the Clean Air Scientific Advisory Committee in accordance
with the Agency's established scientific review policy. Although the
Mathtech study reflects the "state-of-the-art" in particulate matter benefit
analysis, the approach and results have not been subjected to a comparable
extensive peer review process.
The Agency is currently considering the generic issue of the role,
if any, of benefit analysis, or parts thereof, in setting ambient standards.
If the Agency concludes that information from benefit analyses may be
legally and technically relevant to standard setting, Agency staff will
review the benefits information in the particulate matter RIA and contractor
reports to determine whether or not any portions of it appear to be relevant
in this rulemaking. Because the approaches used and the results have not
been subject to extensive peer review, EPA would then submit the portions
thought to be relevant to CASAC and the public for comment on their scientific
and technical adequacy and whether they should be considered in the final
decision on the particulate matter standards.
* Hereafter referred to as the Criteria Document, Reference 2.
** Reference 3.
-------
VI-3
The discussion which follows is divided into sections on Methodology
(Section VLB), Air Quality Data (Section VI.C), Key Analytic Assumptions
(Section VI .D), Scenarios Analyzed (Section VI.E), Alternative Aggregation
Procedures (Section VI.F), Selected Results (Section VI.G), and Findings
and Conclusions (Section VI.H) from the analysis. A detailed treatment
of the analysis is presented in the Mathtech Technical Report entitled
"Benefit and Net Benefit Analysis of Alternative National Ambient Air Quality
Standards for Particulate Matter."
B. METHODOLOGY
Ideally, the estimation of potential economic benefits would be
accomplished using data, assumptions, and modeling techniques developed
specifically for the analytic objective. In the case of the particulate
matter NAAQS, the ideal approach was precluded by time and resource con-
straints. Alternatively, estimations could be based upon existing research
and studies which address some aspect of the health or welfare implications
of ambient particulate matter. This latter approach, which involves trans-
formation and extrapolation, cannot be accomplished in a thorough and
comprehensive manner without first recognizing the many technical problems
associated with drawing inferences from studies not necessarily designed
for that purpose. The problems stem from a variety of sources, which
include limited and sometimes conflicting scientific information, paucity
of data, and analytic techniques which have not always been thoroughly
tested. The technical problems which exist result in a relatively high
degree of uncertainty regarding the magnitude and precision of the empirical
economic benefit estimates. In order to deal explicitly with this uncertainty,
the extrapolation approach to benefit estimation requires:
-------
VI-4
o Identification and use of the best data currently available;
o Accomplishment of sensitivity analysis when alternative data or
assumptions exist;
o Incorporation of plausibility and validation checks whenever
possible; and
o Development of ranges of estimates to demonstrate the level of
uncertainty associated with different assumptions.
The above four elements served as the basis for the analytic strategy
which was employed to develop estimates of the benefits for a PM NAAQS. The
approach which was designed first began with a thorough literature search
for existing studies which could possibly be used in the extrapolation
process. The quantitative relationships which were contained in or derived
from the best of the available studies were used to develop the PM benefit
estimates presented in this analysis. A summary description of the approach
used is presented in subsequent paragraphs.
1) All categories of potential effects which might be attri-
buted to ambient concentrations of particulate matter were
identified. A review of the Criteria Document and other
reports provided a comprehensive listing of possible adverse
effects of PM. A listing of significant effects categories
is presented in Table VI.B.I below.
2) The existing research literature on the potential effects was
identified, classified and reviewed. This portion of the ana-
lysis involved about one hundred different studies drawn from
both the Criteria Document and the economic literature.
All identified studies were screened on the basis of several
criteria, most notable of which were analytic quality and
their potential for extrapolation to benefits estimates. As
a result of this screening analysis, it was determined that
only some of the categories shown in Table VI.B.I could
possibly be estimated.
-------
VI-5
Table VI.B.I
EFFECTS CATEGORIES POTENTIALLY RELEVANT TO ALTERNATIVE PM STANDARDS
o Health Effects
- Mortality
- Acute Morbidity
- Chronic Morbidity
o Soiling & Materials Damage
- Residential Facilities
- Commercial & Industrial Facilities
- Governmental & Institutional Facilities
o Visibility Effects
- Regional Haze
- Plume Blight
o Acidic Deposition
- Aquatic Life
- Crops & Forests
- Materials
o Climatic Effects
- Temperature
- Precipitation
o Non-Users Benefits
- Bequest Value
- Existence Value
- Option Value
Those categories were: mortality, acute and chronic morbidity,
residential facility soiling, and soiling and materials damage
to some manufacturing facilities. Currently available
studies preclude consideration of soiling and materials
damage to commercial, governmental, institutional, and most
industrial facilities. Visibility studies were available and
examined, but they may not be applicable to the PM standards
under consideration. Consequently, although visibility
benefit estimates are addressed in the Mathtech Technical
Report, they are not included in any of the alternative
benefit aggregations. The reason for this is that visibility
impairment as well as acidic deposition are associated with
fine particles such as sulfates which are not expected to be
appreciably affected by the control strategies examined.
However, incidents of plume blight may be reduced but are not
considered in the analysis. Visibility is not the only
omitted benefit category. The available information in the
Criteria Document does not permit quantitative assessment of
the relationship between the presence of PM and effects on climate.
In addition, the non-user benefit categories of bequest,
existence, and option value are also omitted from the analysis.
The coverage of potentially relevant effects categories is
thus only partially complete.
-------
VI-6
3) Benefit estimates were developed using the quantitative
relationships from each individual study and the air
quality improvements postulated in the environmental impact
analysis. These estimates were accomplished according to a
four- or five-step procedure as shown in Figure VI.B.I. The
first step was to identify the magnitude of the ambient air
quality improvement that is estimated to occur in each county
and year. This was the improvement achieved after imple-
mentation of a particular ambient standard, relative to a
baseline situation reflecting controls already in place.
The second step involved estimating the health and welfare
improvements that are expected to occur as a result of the
improvement in ambient air quality. This step made use of
the research findings extracted from the literature review
discussed previously. These findings included either
linear or nonlinear relationships between health or welfare
status and ambient concentrations of particulate matter.
These relationships are discussed in detail in the Mathtech
Technical Report. Note that estimates are required for
each county and year in which there is an air quality
improvement.
The third step was to impute an economic value to the estimated
changes in health and welfare status. This is shown in the
figure as Step 3a. For the health studies, this was perhaps the
most difficult step conceptually. There is limited evidence on
how to estimate the economic value of some changes in health
status (e.g., improved lung function). For this reason, a variety
of methods were employed in order to evaluate how they influence
the results. Note also that in some cases it was possible to
estimate economic values directly from the air quality improvement.
This is shown in the figure as Step 3b. Property value studies
are an example of this approach. These studies analyze variations
in residential property values within a metropolitan area to
identify the variation due to perceived differences in environ-
mental amenities such as air quality, while statistically control-
ling for variation due to differences in housing quality and other
locational attributes. Studies of this type allow direct estimates
of the perceived economic value of environmental improvements.
»
Benefit estimates by covered effect categories were developed
from selected underlying studies. For each scenario analyzed,
the categories examined produced a range of estimates. In each
category, the studies and results were reviewed and a "best"
estimate selected from the study(ies) which most adequately
satisfied the analytic quality criteria. Estimates drawn from
the other studies were used to gauge the potential range of
benefits for the effects category under consideration.
The fourth step was to aggregate results over a specific period
of years to obtain discounted present values. The number of
-------
VI-7
STEP
1
Identify air quality improvement
in county i at year t
STEP
2
Estimate health or welfare
improvement in county i at year t
3a
3b
STEP
3
i
Estimate economic value of the health
or welfare improvement for i, t
STEP
4
Aggregate over t to obtain
discounted present values
STEP
5
Aggregate over i to obtain
regional and national totals
Figure VI.B.I. Basic.steps in estimating benefits for an individual study
-------
VI-8
years depends on the particular standard under consideration. In
Step 5, benefits for each county were summed to obtain regional
and national totals.
4) Total incremental benefit estimates were developed by combining
or aggregating estimates from the appropriate effects categories.
Total incremental benefit estimates for each of the alternative
standards under consideration were required in order to complete
a benefit-cost analysis of the various policy alternatives. The
estimates are incremental in the sense they are derived from the
air quality change associated with going from present existing
and new source control requirements to controls required for the
alternative National Ambient Air Quality Standards analyzed.
The aggregation of benefits cannot be accomplished by indiscriminate
summing of results from the different effects categories. Any
procedure of aggregation must carefully consider factors such as
completeness of coverage across effects categories, the possibility
of double counting, and the relative strength of evidence between
studies and across different categories. The various alternative
aggregation procedures examined and the underlying rationale for
each is presented in Section VI.F.
C. AIR QUALITY DATA
The estimation of economic benefits for alternative standards requires
an understanding of changes in exposure for the affected population.
This, in turn, requires knowlege about projected air quality changes. The
air quality data used in this analysis were developed in the cost and
environmental impact analysis (See Section IV). Limitations on that data
(single design value monitor, county basis, and use of modified rollback)
may have a significant impact on the benefit estimates. This section
focuses on the air quality data, the analytic adjustments made to the data
to reduce the limitations, and the major biases which still remain.
The cost and environmental impact analysis provided estimates of ambient
air quality conditions at the "design value" pollution monitors for each
county. The design value monitor in a county is defined as the monitor
which generally recorded the highest ambient concentration (TSP annual mean
-------
VI-9
and/or 24-hour second high) in 1977 or 1978 (or the two most recent years
back to 1975). Although nonmonitored areas in a county may experience
higher or lower concentrations, lower concentrations are more likely since
monitors are generally sited to try to reflect high concentrations. Air
quality conditions at other monitors and locations were not addressed in
the cost and environmental impact analysis. The cost and air quality
analysis also omits the possibility of air quality improvement due to
controls applied in an adjacent county.
Since projections of air quality and economic data are available
primarily for county areas, the county is the basic unit of analysis in
this study. This approach provides a good match with the original studies
that used either county or SMSA data. The air quality change in the county
is then taken to be either the change at the design value monitor, or the
average change across all monitors in the county, depending on the measure
used in the original study. The proportionality factor for the all monitor
to design monitor relationship on average was 0.74 for the annual mean and
0.70 for the 24-hour second-highest concentration.
Note that the above method for estimating the average air quality
improvement across monitors in a county is only approximate. This is
because the improvement at a particular monitor may or may not be proportional
to the improvement at the design value monitor. That is, the dispersion
properties of PM attenuate the air quality improvement occurring at various
distances from the emissions sources being incrementally controlled. The
proportionality assumption is an approximation to the actual attenuation.
The degree of attenuation cannot be assessed without detailed dispersion
modeling of the PM emission controls applied in each county. Such modeling
-------
VI-10
is very time and resource intensive, and was not practical in this nationwide
study. As a result, the cost analysis relies on the more approximate
county-wide modified rollback method for predicting air quality improvements
at the design monitor in each county. County-wide modified rollback methods
generally assume that the change in air quality (net of background concen-
trations at any location in a county is proportional to the change in total
county emissions. Thus, the use of proportionality in the benefit analysis to
estimate the average air quality improvement across other monitors in a
county is essentially a further application of county-wide modified
rollback; and is required for consistency with the cost analysis.
Lack of dispersion modeling also hinders use of studies which measured
air quality effects at the city or census tract level. For most appropriate
use of the results of these studies, it would be desirable to have
dispersion modeling results for subcounty areas. For example, for the
same reasons noted previously, use of the air quality change projected at
the county design value monitor may overstate the change which occurs
elsewhere in the county. To provide a better air quality measure for
studies measuring air quality of the census tract or city level, the
average air quality change across all monitors in a county is used.
The average was thought more typical of the population based monitored
readings than the design monitor change. The average change is estimated
as described previously.
Plausibility Check
As noted above, the ideal situation would be to identify the change
in PM concentration experienced by the population in each part of the
county. In the absence of dispersion modeling results, the approximations
-------
VI-11
described above attempt to improve upon the available concentration
measure, the design monitor. Two plausibility checks were performed on
these approximations. The first plausibility check concerned the accuracy
of estimating the change in county average air quality based on a proportion-
ality relationship to the change at the county design monitor. An analysis
was undertaken to evaluate the quality of this approximation. The analysis,
which is discussed further in Section 10 of the Mathtech Technical Report
(Reference 1), centers on the Chicago, Illinois area (Cook County). This
area was chosen because it accounts for the largest fraction of total national
benefits among all counties in the analysis and also had been the subject
of detailed dispersion modeling previously. The analysis compares benefits
estimated using two methods. First, benefits are calculated based on
detailed dispersion modeling results for 31 subareas within Cook County.
In the second method, only the predicted change at the design monitor is
used and proportional changes are assumed through the rest of the county.
The second method, which is comparable to the approach used in the national
benefit analysis, leads to estimates which are a factor of 2 to 4 higher
than the first method.
The Chicago analysis suggests that the assumption of proportional
air quality changes may introduce an upward bias in the national benefit
analysis. This suggestion is consistent with other studies that have
shown that particulate concentrations decrease rapidly with distance from
the source. However, there are a number of differences between the
Chicago analysis and the national analysis which preclude an accurate
estimate of the magnitude and direction of the bias.
-------
VI-12
First, the results of the Chicago analysis reflect the actual
geographic distribution of the population residences within Cook County
relative to the location of the emission sources. Within Cook County,
human activity patterns may be such that residences are poor proxies for
pollution exposure. Even if the residences were good proxies, in other
counties the distribution of population relative to emission sources could
be considerably different, leading to either higher or lower estimates.
Second, the two analyses differ in the method used to select emission
sources that will be controlled (i.e., dispersion model vs. modified linear
rollback, which omits consideration of sources to monitor distance.) The
different methods could lead to different sources being selected for control,
different levels of control for each source, and thus different changes in
population exposures. Consequently, failure to account for source control
differences in comparing dispersion modeling and modified rollback derived
benefit estimates could bias the estimates and subsequent comparisons. Third,
the PM problem in the Chicago area is associated with concentrated major point
sources compared to areas where the PM problem is more geographically
dispersed (e.g., Los Angeles). In areas of the latter type, proportion-
ally more sources may require control, leading to a more equally distributed
concentration reduction over the entire population. Fourth, the Chicago
control strategy analysis accounts for transport of pollution across county
borders and this possibility is not considered in the national analysis.
Fifth, certain selected underlying studies are based on wider area air
quality indexes. Application of coefficients from those studies to census
tract level air quality changes may bias the results. The national
analysis provided for consistent use of air quality indexes in the
-------
VI-13
underlying studies. These various differences make it difficult to extrapo-
late the results of the Chicago analysis to the national analysis.*
The second plausibility check concerned the use of the county
average as an additional measure of the PM concentration. This check
involved comparing the county averages in the 1978 base year with corre-
sponding readings at monitors designated by EPA as "population-oriented"
monitors. This analysis found that readings at population monitors are
typically much closer to the county average than to the design value
monitor readings.** This is not too surprising since 73 percent of all
monitors in 1978 are designated as population-oriented. This suggests
that the county average air quality is likely to be more representative
than the design monitor as an estimate of the actual population exposure
(given human activity patterns are limited to areas typical of the popu-
lation based monitoring site).
D. KEY ANALYTIC ASSUMPTIONS
The economic benefit estimates which were developed are based upon
extrapolations from several, independent health and welfare studies, each
of which is based upon its own set of assumptions. Consequently, when all
of the studies are considered, the total set of assumptions becomes quite
large. The Mathtech Technical Report (Reference 1) presents a comprehensive
listing and discussion of the most important assumptions as well as their
*A study on the impact of new source performance standards in Jefferson County,
Texas showed benefits estimates were greater using dispersion modelling.
However, like the Chicago analysis there are enough differences to make
extrapolation of those results to the present analysis suspect.
**The correlation between the average of all population monitors in a
county and the average of all monitors in a county ranged between 0.95
and 0.96, depending on the averaging time. For population monitors and
design values, the correlation ranged between 0.48 and 0.50.
-------
VI-14
potential impact on results. The subset of assumptions that most critically
affects the results includes population exposure, range of applicability
for the individual concentration-response functions,* valuation of health
benefits, and alternative aggregation procedures. Population exposure
assumptions and implications were the subject of the preceding section.
This section focuses on the ranges of applicability for the concentration-
response functions and the valuation of health effects. Section VI.F
focuses on the alternative aggregation procedures.
Applicable Concentration Ranges for Concentration-Response Functions
The results of community health effects studies can often be expressed
in the form of concentration-response functions. Such functions represent
a quantitative statement of the relationship between health status and
ambient concentrations of an environmental pollutant such as PM.
Concentration-response functions are useful in a benefit analysis.
As used in this analysis, concentration-response functions simulate the
individual's perception of the air pollution-damage relationship and thus
enable generation of consistent willingness to pay estimates for various
improvements in air quality.** This subsection is concerned with the
* Following the practice of the EPA/OAQPS Staff Paper, the RIA uses
the term "concentration-response function" when referring to community
studies of ambient air pollution effects. The more traditional term
"dose-response function" implies greater knowledge of actual exposure,
as would be the case in a laboratory study.
**The benefit analysis only simulates the willingness to pay for individuals
experiencing the air quality improvement. The non-user benefits of bequest,
existence, and option value are omitted from the benefit analysis. Further
if the property right for cleaner air is vested with the beneficiaries,
the appropriate question is willingness to accept compensation for air quality
degradation. Empirical studies have shown that because of income and wealth
considerations willingness to pay is less than willingness to accept
compensation.
-------
VI-15
ranges over which the concentration-response functions should be applied.
In other words, what information is relevant for the individual members
of society to consider in revealing their preferences for cleaner air.
The EPA/OAQPS staff Paper explicitly addressed the question of
concentration ranges over which health effects may occur. The Staff
Paper identified one set of ranges at or above which health effects are
"likely" to occur among sensitive groups in the population; it identified a
second, lower set of ranges over which effects may be "possible." The
numerical values of these ranges are identified in Table VI.D.I. The
ranges were derived from the EPA Criteria Document and reflect consideration
of the epidemiology studies judged to provide the most reliable quantitative
evidence for purposes of standard setting. The staff paper concluded
that the data do not, however, show evidence of a clear threshold in
exposed populations. Instead, they suggest a continuum or response with
both the likelihood (risk) of effects occurring and the magnitude of any
potential effect decreasing with concentration. Thus, according to the
Staff Paper effects may be "possible" at levels below those listed in the
"effects likely" row; but, because the evidence is less clear, the nature
and extent of risks at lower levels are much less certain. An analysis
(Ostro, Reference 4) completed after CASAC closure on the Criteria Document
and Staff Paper further evaluates the short-term (London mortality) data
and examines whether clear effects thresholds can be identified. The
analysis adds to evidence that would suggest that the possibility of
effects extends to levels even below 75 yg/m^ BS (75 to 175 yg/m^ as
PM}Q). Ostro (Reference 5) also evaluated long term data regarding U.S.
morbidity. He found no evidence of a threshold in his study using annual
average TSP concentrations ranging from 42 to 130
-------
VI-16
The Clean Air Scientific Advisory Committee (CASAC) in reviewing the
ranges of interest reached the following conclusions as summarized in the
CASAC Chairman's closure letter. "At the upper bound of the proposed
ranges of 150-350 ug/m^ for the 24-hour and 55-110 ug/m-* for the annual
averages, detectable health effects occur in the populations evaluated in
the epidemiological studies. Since the upper end of these ranges contain
little or no margin of safety, it would be appropriate to consider lower
values for revising the 24-hour and annual standards. In addition, the
stated ranges are based solely on quantitative evidence reported in
epidemiological studies. A final decision on a revised standard should
also incorporate information generated through controlled human, animal
toxicology, and from other less quantitative epidemiological studies
discussed in the criteria document." In addition, the CASAC noted
absence of a clearly definable exposure-response relationship for particles.
Table VI.D.I
SUMMARY OF CONCENTRATION RANGES IN EPA/OAQPS STAFF PAPER**
"Effects Likely"
"Effects Possible"
Short-Term Exposure
BS
250-500
150-250
PM10
350-600
150-350
Long-Term Exposure
TSP
>. 180
110-180
PM10
90-110
55-110
**A11 entries are in ug/n. Concentrations are given in both original
study units and approximately equivalent PM^Q units.
Source: EPA/OAQPS Staff Paper, op_. cit., Tables 1 and 2.
As a result of the preceding, in the absence of data that shows a
clear threshold, the concentration-response functions were applied to
-------
VI-17
potential exposures below the effects "possible" level. However, in no
selected study (medical or economic) is the concentration response
relationship extended beyond the observed data of the underlying study.
Also, for the selected medical epidemiological studies a weighted point
estimate for benefits is derived from the nonlinear and linear functions
which are fitted to the data sets of the four underlying studies. The
linear function, which yields higher benefit estimates, predicts constant.
health changes for a given air quality change for all concentrations.
The quadratic function predicts that health changes for a given air
quality change will be lower at lower concentrations than at higher
concentrations. In deriving point estimates a geometric mean is used
which gives greater weight to the lower estimates provided by the quadratic
form. For the selected economic epidemiology studies, the functional
forms selected by the original authors were used. In two cases these
were linear functions while the third was a logistic function.
MEASUREMENT AND ECONOMIC VALUATION OF HEALTH IMPROVEMENTS
Comparison of costs and benefits in commensurate terms requires
placing an economic value on reductions in the risks of mortality and
mordidity. The alternative selected in this Regulatory Impact Analysis (RIA)
is to explicitly provide benefit estimates including valuation of mortality
risk.
The benefits of a reduction in health risk are estimated using a
four-part calculation.
o Reduction in ambient PM concentration;
o Reduction in health risk per capita corresponding to the reduction
in ambient PM;
-------
VI-18
o Number of individuals experiencing the health risk reduction
(i.e., experiencing the PM concentration reduction); and
o Dollar value(s) placed by an individual on a unit reduction in
health risk.
Reduced Health Risk
The reductions in health risk are estimated using results from selected
health and economic epidemiological studies. For example, assume the
yearly mortality rate is 15 deaths per 100,000 people or 0.00015. This
rate (r) includes deaths from all causes including air pollution. An
individual chosen at random would have a 15 in 100,000 chance of dying that
year or a mortality risk of 0.00015.
Now suppose epidemiological studies allow an individual to infer
that reducing air pollution by 100 yg/m^ annual average would cause r to
decline by 0.00001 or Ar. That is the change in mortality risk for the
i ndividual.
By multiplying the population at risk by Ar, one estimates the
aggregate reduction in mortality risk associated with incremental
improvement in air quality. If 100,000 individuals experience the 100
pg/m^ air quality improvement, then 100,000 x 0.00001 is the aggregate
reduction. In this example the result is one statistical live saved. In
the case of morbidity the units vary depending on the nature, severity,
and population at risk. Units may be work days lost, incidents of acute
respiratory disease, etc.
Valuation of Reduced Health Risk
The economic value of a small reduction in health risk (Ar) is estimated
by adding together the values which individuals assign to the reduction.
The valuation of morbidity risk should account for the potential effects of lost
-------
VI-19
work days, lost non-work days (reduced activity days), medical care
services, and residual pain, suffering, and inconvenience. However, residual
pain, suffering, and inconvenience estimates were not available. Conse-
quently, by omitting these the valuation process probably underestimates
the willingness to pay for morbidity risk reduction. The method used to
value each of the other effects is listed below. The valuation factors are
county specific.
Effect Method of Valuation
Lost workday Average daily wage
Reduced activity day One-half of the
average daily wage
Change in direct Proportional to change in
medical expenditures expected morbidity incidence
The valuation of reduction in mortality risk is affected by risk preferences
(i.e., neutral, averse, taking), information quality, and information processing
abilities of individuals. Data constraints currently prohibit reflecting
the distribution of these relevant factors across the population at risk.
However, the valuation of mortality risk reductions is made more manageable
by virtue of the very small changes in the probability of death across
the range of standards being analyzed. This means an individual's marginal
willingness to pay for a unit reduction in mortality risk can be approximated
by a constant valuation factor.
Wage compensation studies examine the relationship between wages and
the probability of job fatality. Other things remaining the same, as
probability of a fatality increases the real wage, it is argued, should
be higher. Such studies support that hypothesis and allow derivation of
-------
VI-20
willingness to pay functions for a reduction in risk. However, wage
compensation studies are based on workers who have voluntarily chosen
high risk occupations. Consequently, they may understate the willingness
of the general population to involuntarily bear additional mortality
risk. Also, applying these studies to the general population
when large changes in mortality rates are limited to a few individuals
may bias the benefit estimates.*
A survey and assessment of 6 wage compensation studies by Mathtech
(Reference 1) yielded a valuation factor ranging from $0.36 for a unit
reduction of 1.0 x 10"^ in annual mortality risk to $2.80 (all figures in
1980 $'s). To reflect data constraints regarding risk preference, information
quality, and information processing abilities of individuals, the two
factors are used to derive a range of benefits estimates for reduced
mortality risk. While not representing Agency position regarding the
range, it is thought to be a generally representative example.
Measurement and Economic Valuation of "Welfare" Improvement
Existing research on PM soiling and materials effects has mainly
focused on household soiling, and to a lesser extent on specific materials
(such as paint and building stone) and certain manufacturing industries.
The benefit estimates address the household sector with limited coverage
of the manufacturing sector. Because the commercial, institutional, and
governmental sector are omitted from the analysis and the manufacturing
sector is only partially covered, the benefits of reduced soiling and
materials damage may be underestimated.
With large changes in mortality, risk income effects are likely. By not
taking that into acccount (i.e., by not deriving an income compensated
demand curve for reduced mortality risk) the estimates could be biased.
The direction of bias is a function of the starting and ending air quality
values.
-------
VI-21
For the household sector three benefit estimation techniques (studies)
are examined. One is multi-regional and seeks to measure benefits as the
change in household expenditures required to maintain cleanliness in
varying air quality conditions. The other two focus on one area
(Philadelphia). One of these two studies estimates the change in well-
being given no change in out-of-pocket cleaning expenditures with varying
air quality levels. The remaining study imputes a time value to the
frequency and duration of cleaning tasks as well as out-of-pocket costs
for varying air quality levels. The greatest weight is given to the
multi-regional study in part because of its broader geographic base. The
multi-region study yields lower benefit estimates than the other two studies.
For the manufacturing sector one study was amenable to benefit
analysis. This study estimated benefits by statistically explaining the
variation in production costs for particular industries by variations in
air quality levels. However, because of disclosure and air quality data
limits the study only covered about 6% of the economic activity from manu-
facturing. Hence, the manufacturing sector benefits may be underestimated.
E. SCENARIOS ANALYZED
Baseline Ai r Quality
The baseline for all scenarios is a projection of ambient air quality
in each county for the period 1989 through 1995 (or 1987 to 1995). The
baseline projection reflects the following major components: background
ambient concentrations, area source emissions (e.g., roadway dust),
emissions from current stationary sources and emissions from new stationary
sources coming on-line during the period. Thus, ambient air quality in a
county may improve or deteriorate in the baseline scenario, depending on
the relative growth rates for area sources and new stationary sources and
retirement/replacement rates for existing stationary sources.
-------
VI-22
In the baseline scenario, some pollution controls are assumed to be
in place. In particular, unretired sources and one-half of any replacement
sources are assumed to be controlled at 1978 control levels. The other
replacement sources and all net new sources are assumed to be controlled
at new source control levels. New source controls include BACT (Best
Available Control Technology) which represents NSPS (New Source Performance
Standards) and other new source control requirements.
The baseline scenario reflects some air quality improvement relative
to the no-control situation. Specifically, present existing and new source
regulations are factored into the baseline. The alternative PM standards
under consideration represent an incremental improvement in air quality
compared to the baseline scenario. For this reason, the benefits associated
with the alternative PM standards are incremental benefits. The total
benefits of both baseline controls and the alternative PM standards are
not estimated.
Alternative Standards
The various ambient standards considered in the benefit analysis are
listed in Table VI.E.I.
In both the cost and benefit analyses, pollution controls are assumed
to be implemented at the beginning of the year (1987 or 1989). Two scenarios
are considered. In the Scenario B cases, all counties are assumed to
achieve attainment of the standard at the beginning of the implementation
year and to maintain the standard through the end of 1995. In the Scenario
A cases, some counties may not be in attainment during some or all of the
period. In this case, sufficient control options are not available to attain
and maintain the standard throughout the applicable time period. As a
-------
Table VI.E.I
ALTERNATIVE AIR GOALITT STANDARDS
•
(la pi/"3)
PolUtant
PN10
PN10
PH10
PN10
PN10
PN10
TSP
TSP
TSP
TSP
Annual Standard
Concentration
70
55
55
33
55
48
75
—
75
—
Parameter
AAN
AAM
AAM
AAN
AAM
AAN
AON
. —
AON
--
24-Hour Standard
Concentration
250
—
250
200
150
183
260
150
260
150
Parameter
ENV
ENV
ENV
ENV
' ENV
ENV
2nd High
2nd Bigh
2nd Bigh
2nd nigh
Implementation
Date
1989
1989
1989
1989
1989
1989
1989 .
1989
1987
1987
Attainment
Statna
A. B
A. B
A. B
A. B
A. B
A. B
A. B
A. B
A. B
A. B
AAN - Annual arithmetic aiean.
AfiM - Annual geonetrio aiaan of all 24-hour average valnea.
EMV - Eipeoted 2nd maximum value of all 24-hour average valnea.
2nd High • Second highest 24-hour average value obterved per year.
A • Some conntiea eiperienoe non-attainment.
B - All conntiea attain and maintain the standard.
-------
VI-24
result, both costs and benefits are lower than if complete attainment occurred.
As indicated in the table, all standards were evaluated under both partial
attainment (Scenario A) and complete attainment (Scenario B) conditions. A more
complete discussion of the derivation of these two conditions is presented in
Section IV.
Several of the key concepts defined above can be illustrated
graphically as shown in Figure VI.E.I. The upper curve in the figure
represents the projection of baseline air quality. As noted previously,
the baseline reflects sources and controls in place -in 1978 plus growth
and retirement/replacement of sources after that date. The dashed line
represents the standard. The bottom curve identifies the improved air
quality after implementation of additional controls (in this case in 1989).
Benefits are generated by the improvement in air quality indicated by the
entire hatched area. Note that some improvement below the standard can
occur. This results from approximations required in the cost analysis.
The impact of the approximation is to overstate the discounted present
value of benefits (and costs). Constraints are imposed to insure that
the approximations do not result in predicted air quality improvements
below background air quality levels (broader dashed line, crossed-hatched
area).
Figure VI.E.I illustrates the situation with complete attainment of
the standard (type "B" scenarios). Figure VI.E.2 illustrates the case of
partial attainment (type "A" scenarios).
As the previous figures indicated, benefits occur over a period of several
years. It is thus convenient to express benefits in discounted present
value terms. In calculating the discounted present value of benefits,
the following conventions are employed:
-------
Ambient
Concentration
(Mg/m3)
Baseline
Projected Air Quality-
Improvement Below Background
Precluded
Source of
Benefits
With Incremental Controls
I
1978
I
1989
I
1995
Standard
Background
i
no
en
Year
Figure VI.E.I. Typical Air Quality Scenario
-------
Ambient
Concentration
Baseline
Source of
Benefits
ual Non-Attainment
Standard
i
rv>
cr>
Background
Project Air Quality
Improvement Below
Background Precluded
With Incremental Controls
1978
1989
1995
Year
Figure VI.E.2. Residual Non-Attainment (Scenario "A")
-------
VI-27
o Time horizons corresponding to the standard (7 or 9 years);
o Present value as of January 1, 1982;
o Real discount rate of 10 percent;
o Estimates stated in 1980 dollars; and
o Exclusion of benefits occurring after 1995.
F. ALTERNATIVE AGGREGATION PROCEDURES
The final task is to combine the estimates for the individual
categories in order to develop an estimate of aggregate incremental
benefits. Aggregate numbers are required in order for any benefit-cost
analysis to proceed. However, aggregation cannot be accomplished by
simply summing all of the categories. Inherent in the summation process are
assumptions regarding the scientific strength of underlying studies, the
separability of benefit estimates within and among categories, and the
perceptions of individuals regarding revealed preferences for cleaner air.
One way of dealing with these issues is to establish explicit criteria
for developing an aggregation procedure. Specifically, consider:
o Does the procedure give due consideration to the relative strength
of the evidence across different benefit categories?;
o Does the procedure avoid double counting of benefits?; and
o Does the procedure provide complete coverage of benefits?
Since no one procedure best satisfies all three criteria, it is useful to
consider various alternative approaches. Six possible alternatives are
identified in Table VI.F.I. They give differing weights to the three
criteria. Each alternative is discussed further below and in Section 10
of the Mathtech Technical Report. The names in the table are of studies
selected to form the basis for benefit estimates for specific effects
categories. Recognize that each procedure requires a judgement regarding
-------
Table VI.F.I
ALTBtNATITB A00IB0ATTON PtOCBIIOm
••••fit Catofor?
Mortal it?
A««t« •ortMlty
Ckrofllo Rortlilty
•o«i«kol4 •••to*
SoilUf • NatorUlt
/
• ••••f ••toriaf S*«tor
Soiliaf ft MfttorMl*
Otk*r •••tor*
Soiliaf 1 MtUrUU
Fro«»J«r«
A
Mamdcr
•t •!.
—
P«rri«
•t •!.
—
—
—
B
••s«»d*r
•t •!.
••••t
•t ml.
P«rri«
•t ml.
— •
—
—
C
lUimdcr
• t •!.
Octro
F.rrU
•t •!.
••tkt«ok
—
—
D
Ur« 1 •••&!•!
Lipfcct
Oitro
P«r*l»
•t •!.
••tkttok
—
—
1 '
LOT* i •••kin
Llpf.rt
Oftro
Crook* r
•t ml.
Omem. mmmm of
Col«. D ft P
•t Qommtj
Imvml
••tkt«ok
'. .—
P
Col. C •
Col. D
Col. • •
Col. C
Crook* r
•t ml.
fittoa §
Jakcok 4-
Cmmlmn
• t •!.
••tkt«ek
—
I
ro
oo
-------
VI-29
what information is appropriate to simulate revealed preferences of
individuals with respect to lower PM concentrations.
Procedure A
Aggregation Procedure A is designed to be consistent with the CASAC's
review of the scientific literature for the purposes of standard setting.
In the Criteria Document for PM and SOX, EPA distinguished between studies
providing quantitative evidence of pollution-related effects and those
providing less quantitative evidence. Among the health studies considered
adaptable for use in this benefit analysis, the studies by Mazumdar et al.
and Ferris et_ aj_. were judged by EPA to be in the quantitative category
(Reference 2). Procedure A includes only these two health studies.
Other studies were judged by EPA as providing quantitative evidence, but
they could not be adapted for benefit calculations.
Procedure A also excludes all non-health studies. While the Criteria
Document discusses a number of non-health effects studies and uses a
number of them to provide crude estimates of magnitude and direction of
the benefits associated with reduced soiling, they were not included in
Procedure A because the estimates are not comparable (i.e., calculated on
the same basis) to those appearing in later aggregation procedures.
It is likely that Procedure A involves an underestimate of benefits.
This conclusion follows because of its incomplete coverage of benefit
categories.
Procedure B
This procedure is similar to the previous one with the exception
that the health study by Samet et £]_. has been included. This modification
is consistent with the conclusions of the EPA Staff Paper concerning
-------
VI-30
studies providing reasonable evidence of concentration-response relationships
for health effects. The Staff Paper includes the Samet et_ aj_. study in
this category, as well as the studies in Procedure A.
As with Procedure A, non-health studies are excluded. The Staff
Paper, like the Criteria Document, is noncommittal regarding the availability
of quantitative welfare effects evidence.
Procedures C Through F
The studies included in Procedures A and B are among those judged by
the EPA Criteria Document and Staff Paper as providing the most reliable
quantitative evidence of air pollution effects. Procedures C, D, E, and
F make increasing degrees of use of other studies. This includes studies
such as the one by Ostro (Reference 4) which had not yet been formally pub-
lished prior to closure of the Criteria Document. It also includes
studies that the EPA Criteria Document and Staff Paper judged to be less
reliable but that still had some merit. With these latter studies, it is
of interest to trace out their implications. Even though these studies
were judged by the Criteria Document and Staff Paper to be less reliable
for purposes of standard setting, they provide the best (according to
study selection process used in the benefit analysis) evidence of the
potential magnitude of benefits in some categories (e.g., chronic exposure
mortality). Including calculations based on these studies provides
greater assurance that potentially important effects of PM have not been
omitted from the benefit analysis. At the same time, the distinction is
maintained between these studies and studies deemed quantitative by the
staff paper by including them in different aggregation procedures.
-------
VI-31
Procedures C through F differ in the degree to which the additional
studies are used (C is the more selective). The procedures also differ in
the importance assigned to avoidance of double counting versus completeness
of coverage (C stresses avoidance of double counting).
Procedure C
Procedure C brings in the study of acute morbidity by Ostro. This
recent study is not cited in the Criteria Document. In fact, the benefit.
analysis later uses supplemental results not available prior to the Criteria
Document final draft. Among the available health studies not already
included in Procedures A and B, the Ostro study was rated very high by
Mathtech for purposes of an economic benefit analysis. Specifically, the
study is broad-based (90 cities), micro-epidemological (individual data
not city averages) with controls for smoking, occupation, residence, air
quality indexes as well as other economic and demographic variables.
With the addition of the Ostro study in Procedure C, the study by
Samet et^ aj_. is omitted to avoid possible double counting of benefits.
The Samet et^ aj_. study is concerned with acute respiratory effects, which
may be a subset of the effects captured in the Ostro study. The Ostro study
is also given preference over the acute morbidity study by Saric et al.
The latter study provides less complete coverage of diseases (respiratory
disease only) and limited concentration-response information (e.g., no
separation of PM/S02 effects, no information on PM effects below 200 pg/nr
annual mean, etc.).
Procedure C also incorporates the Mathtech study of household soiling
effects. The Criteria Document and Staff Paper reference this recent study
but are noncommittal as to its use for quantitative purposes in standard
-------
VI-32
setting. The study is included here in order to provide some coverage of
soiling effects, in recognition of its strong analytical features, and
because of the favorable peer review which the study has received. With
respect to the latter, the reviewers found the theoretical structure to be
at the frontier of applied welfare economics. However, because of data
limitations inherent in all applied studies, they were less confident
regarding the benefit estimates. Subsequent investigations regarding the
study results have not altered this view.
Note that Procedure C excludes possible chronic exposure mortality
effects, provides limited coverage of chronic morbidity effects (respiratory
illness only), and offers limited coverage of soiling effects.
Procedure D
Procedure D addresses some of the possible incompleteness of coverage
present with Procedure C. In particular, Procedure C omits the possibility
of benefits from reduced mortality risk due to long-term exposure. There
are many existing, cross-sectional studies of long-term exposure mortality
effects. The majority conclude that mortality effects exist. However, these
studies are not highly regarded by the Criteria Document and Staff Paper,
so care is taken in how the results of these studies are used in the benefit
analysis. This includes carefully assessing the limitations of each study
and ultimately using results which are at the low end of the studies finding
effects.*
* The Lave and Seskin mortality risk studies were criticized in part for
degree of aggregation and omission of smoking control. Findings of subsequent
analyses by Lipfert having less aggregation and approximate control for
smoking were consistent with those of Lave and Seskin. Other studies by
Gregor and Koshal and Koshal found similar relationships but found pollution
to be an even more important factor. The pollution mortality risk coefficient
used in Procedure D is at the lower end of the range of those estimated by
Lave and Seskin.
-------
VI-33
Summarizing briefly, nine chronic exposure mortality studies are
reviewed in detail in the Mathtech technical report. As noted previously,
the majority (six) of those studies found statistically significant associa-
tions between mortality rate and PM (generally measured as TSP). Two
others found that it was the sulfate component of PM, rather than TSP, that
i
had an association. The ninth found little evidence of association between
mortality and air pollution.
The three studies finding no PM effects have been faulted for inclusion
of an excessive number of pollution variables and/or control variables.
This leads to multicol1inearity and reduces the possibility of finding
statistically significant PM effects. At least one of the three no-effects
studies also used relatively more recent data when mean and variance levels
of TSP were reduced compared to earlier years, thus reducing the possiblity
of finding significant effects. These negative results (no effects = no
benefits) are retained in the Mathtech Report (Reference 1) for use as a
lower-bound estimate of zero for this category.
The six positive-effect studies (and the no-effect studies) have been
faulted on other grounds. Some have used relatively aggregate data. Some
have been criticized for the pollution monitoring used and the biological
plausibility of their results. All of the studies have incomplete statistica1
controls. However, that is also true of the studies included in Procedures
A through C.
With the addition of chronic exposure studies, the acute exposure
study by Mazumdar et aJL is dropped from this procedure. This is because
of the possibility for overlap between these estimates. In particular,
the chronic exposure studies are based on annual mortality rates. Annual
-------
VI-34
mortality rates will include all deaths during the year, including those
deaths that may be due to acute exposures. Thus, it is possible for the
chronic exposure studies to be capturing the mortality effects of both
acute and chronic exposure. The extent to which this may happen is unknown,
however. It depends on a variety of factors such as the functional forms for
the acute and chronic concentration-response relationships and the statistical
correlation between the measures of acute and chronic exposure. Assuming the
acute exposure mortality is a subset of chronic exposure estimates, the
acute exposure study is eliminated from this procedure. Consequently, the
difference in monetized benefits for procedures C and D is accounted for by
the difference between the acute and chronic exposure mortality studies
(regarding reduced mortality risk).
Procedures E and F
Procedures E and F are most easily considered together. Procedure E
addresses the incomplete coverage of the chronic morbidity category and the
underestimation of soiling effects. In Procedures A through D, chronic
morbidity estimates are based on the study by Ferris e_t _a_K which includes
only respiratory diseases. In Procedure E, the Ferris ejt^ aj_. study is
replaced by the Crocker ^t a]_. study which includes more chronic illnesses.
In Procedure E, the Mathtech study of soiling and materials damage in
parts of the manufacturing sector is included. The coverage of household
sector soiling is also expanded. The latter is done by taking into
consideration the results of innovative studies by Watson and Jaksch and by
-------
VI-35
Cummings et. al.* The studies measuring household soiling use techniques
which, although they probably overlap benefits covered, do not measure the
same component of benefits.** Both of these studies are used in Procedure
F, which may result in an overestimate for residential soiling user benefits.
In contrast, the Mathtech household study probably underestimates benefits.
Procedure E thus uses a compromise estimate for household soiling: the
geometric mean of the Mathtech estimate and the sum of the estimates based
on Cummings et_ aj_. and Watson and Jaksch. The geometric mean is used as a
conservative measure of the average of the two estimates giving greater
weight to the lower estimate.
The remainder of the estimates used in Procedure F also seek to provide
more complete coverage of benefits, with the possible risk of some double
counting.*** In particular, the benefits for the acute and chronic exposure
* The Watson and Jaksch and Cummings et al. estimates are developed using a
fairly detailed survey for one city-Philadelphia. Like the use of Samet,
(Steubenville, Ohio) Ferris, (Berlin, New Hampshire) and Mazumdar, et. al.
(London, England) one must infer that the relationship which holds for the
city in the underlying study also holds for the nonattainment areas in the
benefit analysis. Questions have been raised regarding the legitimacy of
the soiling function and cleaning task definitions employed in these studies.
Hence, one should interpret the results with appropriate caution.
** Watson-Jaksch measure the benefits associated with a unitarily elastic
demand for cleanliness—no change in expenditures accompany an air quality
change. Cummings et_ £l_ focus on the benefits derived from changes in the
frequency and duration of cleaning tasks—benefits include out-of-pocket
expenditures as well as the value of time expended for the various tasks.
***The possible risk of double counting refers only to those benefit
categories covered in this analysis and not to all of the potential benefits
of PM NAAQS. Consequently, even Aggregation Procedure F omits some potential
benefits.
-------
VI-36
studies for mortality are added together; the same is also done with the
estimates from the acute and chronic exposure studies of acute morbidity.
As a result, Procedure F provides the most complete estimate of monetized
benefits possible within the selected studies and benefit categories examined.
However, it may involve some double counting within those selected studies
and benefit categories. The possibility of double counting is less likely
to arise with any of the Procedures A through E.
G. RESULTS
Incremental benefits for a variety of alternative standards considered
in this analysis are shown in Tables VI.G.I and VI.G.2 The alternatives include
all of the standards defined previously in Table IV.E.I. They include six
PMiQ standards with 1989 implementation dates, two TSP standards with 1989
or 1987 implementation dates. All standards extend through 1995.
Estimates are displayed to two or three significant digits.
As noted previously, health studies available for the benefit analysis
do not incorporate particle size information. Benefits shown in the table
for the PM^o standards are based on the TSP change that results. Comparisons
across PM^Q and TSP standards thus reflect only differences in relative
stringency in terms of the TSP reductions; they do not reflect differences
in particle size. If PM^Q standards lead to proportionately larger reductions
in PM^g relative to TSP, benefits for the PM^g standards may be underestimated.
Data from the cost and air quality analysis suggest that proportionately
larger reductions do not generally occur. However, approximations in that
analysis are such that the comparisons should still be interpreted with caution.
This is signified by the line in the table separating the two groups of standards.
-------
Rev. 12/20/83
Table VI.G.I
INCREMENTAL BENEFITS FOR ALTERNATIVE PM1Q AND TSP STANDARDS1
(B Scenarios)
Alt"prnati\/p Stands rc\*~
PM10 (70,250)/89
PM10 (55,-)/89
PM10 (55,250)/89
PM10 (55,200)/89
PM10 (55,150)789
PM10 (48,183)789
TSP (75,260)/89
TSP (-,150)/89
TSP (75,260)787
TSP (-.150)787
Aggregation Procedure
A
.37/2.0
.51/3.1
.51/3.1
.51/3.1
.55/3.5
.57/3.6
.58/3.6
.65/4.2
.78/5.0
.87/5.7
B
1.7/3.5
2.9/5.3
2.9/5.5
2.9/5.7
3.5/6.5
3.7/6.7
3.7/6.7
4.6/8.2
5.0/9.4
6.4/11.4
C
12/14
20/24
21/23
21/25
25/29
27/31
27/31
34/38
39/43
48/52
D
15/33
25/61
26/60
26/62
31/73
34/78
34/80
42/100
48/114
59/141
E
30/48
56/92
57/91
58/94
69/111
76/120
77/123
93/147
107/173
131/209
F
42/62
79/117
79/117
81/119
98/142
106/154
107/153
137/183
151/229
191/269
1. 1982 discounted present values in billions of 1980 dollars at a 10 percent discount
rate. The 7-year time horizon is 1989-95 and the 9-year horizon is 1987-95. Comparisons
between PMjg an^ TSP standards are in terms of TSP stringency, not particle size.
2. Key: PM (x,y)/z - x=annual standard, y=24-hour standard, z=attainment year.
3. Key: V/W - V=benefits estimated assuming a mortality risk reduction factor of $0.36 per
1.0 x 10~6 annual reduction. W=benefits estimated assuming a mortality risk reduction factor of
$2.80 per 1.0 x 10"^ annual reduction.
GO
-------
Rev. 12/20/83
Table VI .G.2
INCREMENTAL BENEFITS FOR ALTERNATIVE PM1Q AND TSP STANDARDS, WITH LOWER BOUND APPLIED1
(A Scenarios)
PM10 (70,250)/89
PM10 (55,-)/89
PM10 (55,250)/89
PM10 (55,200)/89
PM10 (55,150)/89
PM10 (48,183)/89
TSP (75,260)/89
TSP (-,150)/89
TSP (75,260)/87
TSP (-,150)/87
Aggregation Procedure
A
.29/1.5
.33/1.9
.34/1.9
.34/1.9
.36/2.0
.35/2.0
.36/2.0
.40/2.4
.49/2.9
.53/3.3
B
1.2/2.2
1.5/3.1
1.5/3.1
1.6/3.2
1.9/3.5
1.9/3.7
2.0/3.6
2.4/4.4
2.7/5.1
3.3/6.1
C
7.6/8.8
10/12
10/12
11/13
13/15
13/15
13/15
17/19
18/22
23/27
D
9.4/23
13/31
13/31
14/32
16/36
16/38
16/38
21/47
23/53
29/67
E
18/30
26/44
26/44
27/45
32/54
33/53
33/55
42/70
47/77
59/97
F
25/39
38/56
38/56
39/59
46/70
47/71
48/72
62/90
68/100
88/132
1. With TSP AAM lower bound of 110 pg/m3 applied to all health studies. 1982 discounted
present values in billions of 1980 dollars. Comparisons between PM^g and TSP standards
are in terms of TSP stringency, not particle size.
2. Key: PM (x,y)/z - x=annual standard, y=24-hour standard, z=attainment year.
3. Key: V/W - V=benefits estimated assuming a mortality risk reduction factor of $0.36 per
1.0 x 10~6 annual reduction. W=benefits estimated assuming a mortality risk reduction factor of
$2.80 per 1.0 x 10"6 annual reduction.
CO
CD
-------
VI-39
The standards are listed in approximate order of increasing stringency
with the PM10 (70> 25°) as tne 1east stringent and the TSP(-, 150) as the
most stringent. The ordering is only approximate for two reasons. First,
the ordering is different depending on whether the annual or 24-hour
concentration is used. Second, the PM],0 and TSP standards use different
statistical measures. The PM^Q standards are stated in terms of arithmetic
means, expected second maximum values, and particles less than or equal to
lOpm in diameter; the TSP standards use geometric means, observed second
highest values and total suspended particles. This means that relative
stringency will vary from county to county, depending on the particle size
distribution and the temporal distribution of air quality in each county.*
B Scenario
An examination of Table VI.G.I reveals several general patterns. First,
incremental benefits generally increase as the stringency of the standard
increases. This results because more stringent standards lead to larger
improvements in air quality, and also increase the number of counties where
air quality improvements occur. Second, the ordering of the most lenient
and most restrictive standards in terms of incremental benefits is insensitive
to the choice of a particular aggregation procedure. That is, the ordering
is the same for all six procedures. Third, standards with earlier implementation
dates produce larger incremental benefits. This is also to be expected
* As an example, it is possible for air quality in one county to have a
larger annual geometric mean and a smaller annual arithmetic mean than in
another county. In this case, it would be possible for a geometric mean
standard to be binding in the first county but not binding in the second
and vice versa for an arithmetic mean standard. The same is true for com-
parisons between different monitors.
-------
VI-40
since earlier implementation leads to improvements in air quality which are-
earlier and of longer duration analytically.
A Scenario
Table VI.G.2 contains estimates of the same ten alternative standards
for the "A" scenario case. The A scenarios include some counties where simulated
available control options are insufficient to attain the standards. In the
B scenarios, these counties are forced into attainment. The degree of air
quality improvement is thus lower in the A scenarios relative to the corre-
sponding standards in the B scenarios. Hence, benefits (and costs) are
also lower with the A scenarios.
The A scenario results exhibit the same general patterns as the B
scenarios. Benefits are larger with more stringent standards and with
earlier implementation dates, and these patterns exist for all six aggregation
procedures. As with the B scenarios, costs of control associated with each
standard must also be considered before drawing conclusions about the
economic attractiveness of a particular standard.
Sensitivity to Staff Paper Lower Bounds
It may be of interest to calculate benefits under the assumption that
no health benefits are produced by reducing PM below the Staff Paper lower
bounds of 55 vg/m3 annual PM1Q (110 yg/m3 annual TSP) and 150 ug/m3 for 24-hour
PM^g. However, there are several practical problems with doing the calculation,
ranging from probable statistical bias to lack of daily observed or forecast
data on PM concentrations for the 7 and 9 year time horizons of the analysis.
Keeping those limitations in mind, Tables 11-41 through 11-80 of Volume V of
Mathtech Report (Referenced) contain the benefit estimates that result when
all health benefits associated with reducing PM concentrations below 110 pg/m3
annual average TSP are excluded.
-------
VI-41
The effect of imposing the Staff Paper lower bound is to reduce the
benefits associated with all of the standards. The process is equivalent
to not counting the hatched area below the standard displayed in Figures
VI.E.I and VI.E.2. Furthermore, the lower bound is fixed thus reducing the
benefits of more stringent standards proportionately more than those of
the less stringent standards. This results because the more stringent
standards produce air quality improvements which are increasingly below the
lower bound concentration. Hence, benefits show little increase since no
credit is taken for below-lower bound air quality improvements.
Generally, incremental benefits increase for all aggregation procedures
between the least and most restrictive standards. (The few exceptions
arise when lower bounds are applied.) The increase results from two changes:
1) the larger improvement in air quality in each geographic area, and 2)
the addition of more areas that will experience an improvement. However,
this does not necessarily imply that the most restrictive standard is the
economically preferred standard. The offsetting costs of pollution control
must be considered in making such a decision.
H. GEOGRAPHIC DISTRIBUTION OF BENEFITS
As indicated in the Mathtech Report (Reference 1), the benefits of reduced
PM concentrations vary considerably among the different regions of the
country. Differences in baseline air quality, population, and growth help
explain this variation. Under the least restrictive standard (PM^o 70
AAM/24-hour), Federal administrative Region I (New England) receives no
benefits. Regions V (East North Central), VI (South Central), and IX
(South Pacific) receive most of the benefits. The relative share of benefits
among regions is also sensitive to the benefit aggregation procedures
-------
VI-42
employed. For the same standard Region IX's shares for aggregation procedures
A and E were 54 percent and 32 percent respectively.
Analysis of other standards indicates regional shares also change with
the stringency level of the standard. In general, benefits are more widely
shared as the stringency level increases. For the least restrictive standard
benefits accrue to about 95 counties. For the most restrictive standard
benefits accrue to 500 counties.
I. Other Benefit Measures
The valuation of health risk reduction was explained earlier. However,
because of ethical considerations and uncertainties regarding valuation,
other indexes of health benefits may prove useful. Table VI.I.I provides
estimates of the expected reduction in adverse health effects using
Aggregation Procedures C and D for the PM10 (70,250), PM10 (55,200), PM10
(48,183), and TSP (-.150) alternatives under conditions of full attainment.
Volume V of the Mathtech Technical Report provides a more exhaustive
treatment of these other health benefit indexes.
Not all valuation methods permit a readily accessible means of "backing-
out" physical effects indexes. This is especially true for the approach to
estimating welfare related benefits.
-------
VI-43
Table VI.I.I
Expected Incremental Reduction in Adverse Health Effects
Aggregation Procedure C* between 1989 and 1995
PM10(70,250) PM10(55,200) PM10(48,183) TSP(-,150)
Statistical
Lives
103 Work
Loss Days
103 Reduced
Activity Days
103 Incidences
of Chronic
Respi ratory
Disease
2000
82,000
390,000
1100
3100
150,000
710,000
1100
3600
190,000
910,000
1100
4100
240,000
1,200,000
1100
Procedure D differs from C only in regard to the underlying evidence for
mortality risk. Adopting the Lave and Seskin, Lipfert, etc. evidence as
the basis for calculating mortality risk reductions, the estimates of
reduced statistical lives lost are as follows: 22,589 for the PMio(?0,250),
41,146 for the PM10(55,200), 52,528 for the PM10(48,183), and 64,771 for the
TSP (-.150).
Again, these are comparisons on based standard stringency, not particle size.
-------
VI-44
REFERENCES
1. Manuel, E. H., Jr., R. L. Horst, Jr., K. M. Brennan, J. H. Hobart,
C. D. Harvey, J. J. Bentley, M. C. Duff, D. E. Klingler, J. K. Tapiero.
Benefit and Net Benefit Analysis of Alternative National Ambient Air
Quality Standards for Particulate Matter. Report to U.S. Environmental
Protection Agency on EPA Contract Number 68-02-3826 by Mathtech Inc.,
Princeton, New Jersey, March 1983.
2. U. S. Environmental Protection Agency, Environmental Criteria and
Assessment Office. Air Quality Criteria for Particulate Matter and
Sulfur Oxides: EPA-600/8-82-029c. Research Triangle Park, North . .
Carolina, December 1982.
3. U. S. Environmental Protection Agency, Office of Air Quality Planning
and Standards. Review of the National Ambient Air Quality Standards
for Particulate Matter: Assessment of Scientific and Technical
Information. OAQPS Staff Paper (EPA-450/5-82-001). Research Triangle
Park, North Carolina, January 1982.
4. Ostro, B.D. A Search for a Threshold in the Relationship of Air
Pollution to Mortality in London: A Reanalysis of Martin and Bradley.
Working Paper Dated September 1982.
5. Ostro, B.D. Morbidity, Air Pollution and Health Statistics. Paper
presented at the Joint Statistical Meetings of the American Statistical
Association and Biometric Society. Detroit, Michigan, August 1981.
The benefit analysis also uses later results supplied by the author.
-------
VII. BENEFIT-COST ANALYSIS
A. INTRODUCTION
This section of the report presents comparisons of the estimated
benefits and costs of several alternative National Ambient Air Quality
Standards for particulate matter (PM NAAQS). These comparisons of benefits
and costs are referred to as benefit-cost analyses. They provide a frame-
work for evaluating the economic effects of alternative regulatory policies
and are presented as a response to Executive Order 12291 which requires the
identification of the regulatory alternative which will produce the maximum
net benefits (benefits minus costs) to society. As indicated previously,
two recent judicial decisions state that technical and economic feasibility
of attaining NAAQS are not to be considered in setting them. The inference
is made that the courts were referring to a criterion of societal economic
feasibility (i.e., net benefits positive; benefits minus costs greater than
zero); not private economic feasibility concerns (i.e., the emitter stays
in business after controls are imposed). Thus the Agency has not considered
net benefits in the proposed rulemaking.
Some necessary background concepts are developed immediately below.
In particular, the economic criteria employed to evaluate the alternative
PM NAAQS are described (Section VII.B). Proper methods of conducting
benefit-cost analyses are reviewed next (Section VII.C). . Following this,
the estimates of both benefits and costs are described; comments on the
appropriateness of these estimates for use in the benefit-cost analysis are
provided (Section VII.D). Next, the limitations of benefit-cost analysis
are discussed (Section VII.E). The results of the benefit-cost analyses
-------
VII-2
are then presented (Section VII.F). Finally, summary remarks, conclusions,
and qualifications are offered (Section VII.G).
B. BENEFIT-COST ANALYSIS: EVALUATION CRITERIA
Air quality regulations affect society's economic well-being by causing a
reallocation of productive resources within the economy. Specifically,
resources are allocated towards the production of cleaner air and away from
other goods and services that would otherwise be produced. Benefit-cost
analysis provides a method for assessing the desirability of the alternative
PM NAAQS in terms of their impacts on the allocation of economic resources.
These air quality standards are evaluated in terms of the established
criterion of economic efficiency.
Incremental Benefits and Costs
An analysis of the incremental benefits and costs associated with each
of the alternative PM NAAQS is required for an evaluation of both the
cost-effectiveness and the relative efficiency of the alternative ambient
air quality standards. These are defined as follows:
o The incremental benefits associated with a given PM NAAQS are
defined as the additional benefits resulting from improvements in
air quality over baseline air quality levels.
o The incremental costs associated with a given PM NAAQS are defined
as the additional costs that are incurred to achieve and maintain
improvements in air quality over the baseline levels.
Both the incremental benefits and incremental costs are computed relative
to baseline air quality levels.* The term "incremental net benefits"
refers to the difference between incremental benefits and incremental costs.
*Recall the "baseline" air quality levels are intended to reflect those air
quality levels that would prevail in the absence of any further attempts to
comply with any of the alternative PM NAAQS. However, compliance with BACT
and NSPS requirements are factored into the "baseline". (See Section VI
for a more detailed description).
-------
Vll-3
Cost-Effectiveness
An analysis of the cost-effectiveness of alternatives is generally
employed to reduce the set of alternatives that are evaluated in terms of
economic efficiency. Inferior standards are those which require higher
costs to achieve the same or smaller benefits than dominant alternatives;
or require the same cost to achieve a smaller benefit than a dominant
alternative. An inferior standard is not "cost-effective" since the same
or higher incremental benefits can be achieved by adopting a less costly
standard. However, the cost-effectiveness of the various PM NAAQS alternatives
is not analyzed in this RIA.
Efficiency Criterion
The efficiency criterion is used to evaluate the economic desirability
of the real location of resources that occurs as the result of adopting a
particular PM NAAQS. That air quality standard is efficient in an economic
sense if, as a result of its implementation, at least one individual's
well-being is improved without reducing the well-being of any other member
of society. Consequently, the allocation of resources associated with an
efficient standard is economically preferred to the allocation that exists
prior to its implementation.
It should be recognized, however, that those individuals enjoying the
benefits of improved air quality may not generally be the same as those who
bear the cost of controlling pollution emissions. As a result, the personal
comfort and well-being of some members of society may, in a welfare economic
theory context, be reduced.* This need not be the case, however, if those
* In welfare economics, the performance of an economic system is measured
by its ability to satisfy the perceived needs and wants of individuals
(Reference 1). With greater cost, there is less income available to satisfy
these desires. Consequently, personal comfort and well-being decline.
-------
VII-4
individuals benefiting from improved air quality are required to compensate
those who bear the costs of pollution control. Under such a compensation
scheme, a given pollution standard would be judged efficient if those
individuals receiving benefits could potentially compensate exactly those
individuals bearing the costs of the standard, and still realize a net gain
in economic well-being. As is typically the case with applied benefit-cost
analysis, this is the efficiency criterion that is adopted in this analysis.
In addition to determining whether a given air quality standard is
efficient, it is also possible to rank the alternative PM NAAQS in terms of
relative efficiency. The PM NAAQS that is most efficient, relative to the
alternatives considered, is the one which provides the largest incremental
net benefits. An analysis of the relative efficiency of the alternative PM
NAAQS considered is described later in this section.
Scope of Analysis
Efficiency is only a necessary but not a sufficient condition for
establishing the economic desirability of an air quality standard. Since
there are generally both benefits and costs associated with achieving and
maintaining baseline air quality levels, it is possible that the total
costs could exceed the total benefits associated with a PM NAAQS, even if
incremental net benefits are positive.* This could occur if the cost of
baseline controls exceeds the benefit associated with baseline air quality
levels. If this were the case, society might be better off if no air
quality standards are adopted, including those already in place to achieve
baseline levels of air quality.
*Total benefits include the benefits associated with baseline controls as
well as the incremental benefits realized as a result of the adoption of a
PM NAAQS. Similarly, total costs include both baseline control costs and
incremental costs.
-------
VII-5
A standard for which total benefits exceed total costs is termed
"feasible". As this definition indicates, estimates of both baseline and
incremental benefits and costs are required to assess the feasibility of
the alternative PM NAAQS. However, estimates of baseline benefits consistent
with the methodology employed in this RIA have not yet been developed. Con-
sequently, no feasibility tests have been conducted.
Ideally, all feasible and cost-effective PM NAAQS should be evaluated
in order to identify the one which maximizes society's well-being as a
result of improvements in ambient air quality. Within the context of
applied benefit-cost analysis, however, it is usually possible to consider
only a limited set of discrete alternatives. Such is the case with this
analysis. Consequently, the focus is on the relative efficiency of a
limited selection of alternatives and not the identification of the most
efficient of all possible feasible and cost-effective PM NAAQS.
C. BENEFIT-COST ANALYSIS: METHODOLOGY
Appropriate methods of testing for both the cost-effectiveness and
efficiency conditions are described immediately below. As was stated
previously, the tests for cost-effectiveness and efficiency require analyses
of the incremental benefits and costs associated with each of the alternative
PM NAAQS. These benefit-cost analyses are limited in that they are not
employed to evaluate the distributional impacts of these air quality
standards.
Incremental Benefits And Costs
An analysis of the incremental benefits and costs associated with the
alternative PM NAAQS can be conducted in the following manner:
o Compute the estimated incremental benefits associated with each
alternative PM NAAQS.
-------
VII-6
o Similarly, compute the estimated incremental costs associated with
each alternative standard.
o Compute the estimated net incremental benefits (i.e., incremental
benefits minus incremental costs) generated by each alternative standard.
o Compare the estimated net incremental benefits for each of the
several alternative standards.
Any alternative PM NAAQS that is associated with higher incremental costs,
but the same or smaller incremental benefits than some other standard (or
the same incremental cost but smaller incremental benefits than some other
standard), is cost-ineffective and need not be further evaluated in terms
of efficiency. However, all specified alternative PM NAAQS are evaluated
in terms of economic efficiency in this analysis. Any alternative PM NAAQS
that produces positive net incremental benefits will provide a more efficient
allocation of resources than what would occur under the baseline air quality
scenario. The PM NAAQS that produces the largest positive net benefits
will produce the most efficient allocation of resources among the standards
considered. When net incremental benefits are negative for all considered
alternatives, no standard is identified as efficient. When the net incremental
benefits associated with a standard are negative, the baseline air quality
scenario yields a more efficient allocation of resources.
Distributional Effects
There are two reasons why the distributional impacts associated with
the alternative PM NAAQS are an important issue. These reasons are:
o The benefits and costs associated with the alternative standards
are not likely to be distributed evenly across various sectors of the
economy, thus raising equity issues.
o The distribution of adverse impacts may affect the measurements of
costs that are appropriate for use in the benefit-cost analyses.
Both of these points are discussed below in greater detail.
-------
VII-7
The distributional or equity effects are not typically evaluated
within the framework of applied benefit-cost analysis. In order to do so,
it would be necessary to obtain estimates of the values that society places
on the distributions of economic well-being associated with each of the
alternative PM NAAQS. Estimates of these values are unavailable.*
The potential distribution of adverse economic impacts associated with
the alternative PM NAAQS should also be considered in measuring appropriate
costs for use in the benefit-cost analysis. The cost estimates provided
by the emissions control phase of the study, for example, are based on the
assumption that no plants close or reduce production under the burden of
additional emission control costs. The potential for plant closures could
likely produce an upward bias in these costs estimates. If a significant
number of plants would, in fact, cease operation or reduce output when
faced with additional control costs, downward adjustments to these earlier
cost estimates may be necessary before they can be used appropriately in
the benefit-cost analysis.**
D. MEASUREMENT OF BENEFITS AND COSTS
A clear understanding of several conceptual issues is necessary for a
proper interpretation of the estimated benefits and costs that are compared
in the benefit-cost analysis. These conceptual issues are discussed
immediately below. Following this, the estimates of both baseline and the
* However, one can, provided adequate data and other resources, compare the
predicted distributions with various assumed distributional standards of
wel1-bei ng.
** Descriptions of several potentially adverse impacts associated with the
alternative PM NAAQS are described in a separate report by Energy and
Environmental Analysis, Inc. (Reference 2). See also Section V of the
Regulatory Impact Analysis.
-------
VII-8
incremental benefits and incremental costs associated with each of the
alternative PM NAAQS are discussed. The scope of benefits and costs included
in these estimates are also described and the estimation techniques employed
are reviewed briefly. Finally, a discussion of the consistency between
these benefit and cost estimates is provided.
Measurement of Benefits and Costs: Conceptual Issues
The beneficial effects of improved ambient air quality can be measured
as the value that individuals place on the opportunity to consume cleaner
air. The conceptually correct valuation of this opportunity requires the
identification of individuals' willingness to pay for cleaner air (or to be
compensated for deterioration of air quality).* Where possible, willingness
to pay is the measure of benefits that is adopted in this analysis.
Estimates of society's willingness to pay for cleaner air do not exist for
all benefit categories addressed in this analysis (e.g., morbidity).
Alternative measures are used for those benefit categories included in the
analyses for which estimates of willingness to pay are unavailable (e.g.,
partial compensation).
Similarly, an appropriate measure of the cost of pollution emissions
control can be measured as the value that society places on those goods and
services not produced as a result of resources being diverted to the production
of improved air quality. Again, the conceptually correct valuation of these
costs requires the identification of society's willingness to pay for these
directly foregone consumption opportunities that would otherwise be available.
The appropriateness of willingness to pay versus willingness to be
compensated depends on the property right endowments of receptors. Because
of income constraint considerations, willingness to be compensated may be
greater than willingness to pay.
-------
VII-9
It is also possible that air quality standards may generate indirect
economic impacts. One example of an indirect effect might be the transfer of
income to owners of recreational facilities. For example, individuals whose
health is improved as a direct result of better air quality may increase
their use of the recreational facilities. Indirect costs and benefits
which represent real (as opposed to pecuniary*) effects should be included
in benefit and cost estimates. As a practical matter, however, it is often
very difficult to measure all indirect effects and to determine whether
they are real or pecuniary in nature. Consequently, the benefits and costs
considered in this analysis include only the direct effects of the alter-
native PM NAAQS. The exclusion of possible indirect effects may mean that
these estimated benefits and costs misstate the total effects of the
alternative air quality standards.
The indirect consequences of the costs are, however, addressed in
Section IV on economic impact. Furthermore, the results of the economic
impact analysis are used in part to qualify the cost estimates used in the
benefit-cost analysis.
Estimates of Benefits
The benefit-cost analyses described later in this section represent
a partial range of benefits and point estimates of costs associated with
alternative PM NAAQS. As a result, the uncertainty that surrounds both
estimated benefits and costs may be obscured. Nonetheless, the validity of
the benefit-cost analyses depends critically on the accuracy of estimated
benefits and costs. Any uncertainty embedded in these estimates will carry
*"Income transfers" refer to redistributions of income that do not result
directly from production of goods or services. These transfers are referred
to as "pecuniary".
-------
VII-10
over to the benefit-cost analyses. The sources of uncertainty in the
benefits estimates have already been described in detail earlier in
Section VI.
Estimates of Costs
The costs of achieving and maintaining air quality levels associated
with the alternative PM NAAQS were discussed in Section IV. As is the case
with the benefit estimates, emissions control costs are not estimated with
certainty. Consequently, a proper interpretation of the benefit-cost
analyses requires an understanding of how emissions control costs are
estimated so that uncertainties and potential biases introduced by method-
ological compromises and data quality can be assessed.
Because changes in air quality values are not estimated after year 1995,
the time horizon for the benefit analysis ends with that year. Because the
pollution control equipment installed in either '87 or '89 has an expected
life of 15 years, the cost analysis includes operation and maintenance
costs for the expected life of the equipment and annualizes the capital
cost over the same 15 years. In order to provide a common time horizon for
the comparison of benefits and costs, only the portion of the annualized
costs which will be incurred through the end of '95 are included for the
benefit-cost comparison. Because of this adjustment the present value
costs reported in Sections IV and V are greater than the costs reported in
Section VII. The same underlying annualized costs are found in all sections,
however.
As noted in Section IV.F "Resource Impacts Analysis", the cost estimates
used here assumed that implementation (i.e., administration, monitoring,
and enforcement-AME) costs were 4% of annualized control costs. Other
estimates were also made and presented in Section IV.F.
-------
VII-11
Estimated costs for each alternative PM NAAQS for both the A and B
scenarios are summarized in Table VII.D.l.* As in the case with the benefit
estimates, these costs are "incremental" in the sense that they are calculated
relative to the baseline controls. Although direct pollution control costs
do exhibit some sensitivity to various assumptions and conditions which
underlie their derivation, no attempt is made to develop a range of
incremental costs estimates which reflects this uncertainty.**
The incremental costs associated with the most stringent of the PMjg
standards are greater than those associated with the TSP 75 AGM/260 standard,
even though the latter is listed as being more stringent. Three possible
explanations of this phenomenon are:
o The counties projected to be in nonattainment with the PMjg standards
sometimes differ from those expected to violate the TSP standard.
o PM}Q and TSP control options differ because of different emission
sources and different particle sizes.
o The PM^o standards are in the form of annual arithmetic means and
expected 2nd high 24-hour averages while the TSP standards are in
the form of annual geometric means and observed 2nd high 24-hour
averages. Other things remaining the same, annual geometric mean
and expected 2nd high 24-hour average are more restrictive NAAQS
forms.
The figures displayed in Table VII.D.l represents estimates of the
incremental or additional costs of achieving and maintaining the air quality
levels associated with the alternative PM NAAQS. These estimates displayed
in the table do not include the costs of achieving and maintaining baseline
air quality levels. Baseline costs are discussed in Section IV (p.38).
* Recall that some areas remain in residual nonattainment under the A
scenario while all areas are in full attainment with the alternative
standards under the B scenario.
** The results of several sensitivity analyses are documented in the Argonne
National Laboratories Report (Reference 3). Some are discussed in Section IV.
-------
VII-12
Table VII.D.I
SUMMARY OF COST ESTIMATES1
Alternative Standard^
PM10 (70,250)/89
PM10 (55,-)/89
PM10 (55,250)/89
PM10 (55,200)789
PM10 (55,150)789
PM 0 (48,183)789
TSP (75,260)789
TSP (-,150)789
TSP (75,260)787
TSP (-,150)787
Point Estimate of Incremental Costs
A Scenario
("Partial
Attainment")
0.50
0.90
0.93
0.95
1.26
1.32
1.08
1.78
1.61
2.63
B Scenario
("Full
Attainment")
0.95
2.52
2.57
2.54
3.53
3.36
2.61
5.96
4.02
8.95
1982 discounted values in billions of 1980 dollars at a 10 percent discount
rate. The 7-year time horizon is 1989-95 and the 9-year horizon is 1987-95.
Comparisons between PM^g and TSP standards are in terms of TSP stringency,
not particle size.
Key: PM (x,y)/z - x=annual standard, y=24-hour standard, z=attainment year.
-------
VII-13
Consistency Of Benefits And Cost Estimates
Although the incremental benefit and cost estimates used in this
benefit-cost analysis are estimated using different methodologies, they are
consistent and comparable in the following respects:
o Each is based on the same air quality information.
o Each is valued as 1980 dollars in 1982.
o Each reflects the same time horizons.*
o Each uses a real discount rate of 10 percent to obtain a present
discounted value and an estimate of annualized benefits and costs.
The air quality data used in the benefit analysis were generated as
part of the procedure employed to estimate the costs associated with each
standard. Thus, estimates of both benefits and costs are based on identical
assumptions regarding air quality levels.
Both benefits and costs accrue over time. When appropriate, similar
growth factors are used to project both benefits and costs in order to ensure
consistency between the two time streams.** For convenience, they are ex-
pressed in terms of their present discounted values in 1980 dollars in 1982.
For each alternative PM standard under consideration, benefits and
costs are measured from the implementation date through 1995. It was judged
impractical to extend the benefit-cost analysis past this date, given the
modeling uncertainties which arise when growth factors are incorporated
into the analyses. Fortunately, the process of discounting causes both
* As noted earlier, an adjustment factor is applied to the incremental
cost estimates in order to make them consistent with the time-horizon
used in the benefit analysis.
**
Both analyses used projections of future industry and population growth.
Although the projections were obtained from different sources, they were
very similar in magnitude. For a comparative analysis, see Reference 4.
-------
VII-14
benefits and costs which may accrue in the distant future to be small
relative to the present. Thus, the truncation of benefit and cost estimates
in 1995 may not greatly affect the outcome of the benefit-cost analyses.
A constant real discount rate of 10 percent is used to discount the
time streams of benefits and costs back to 1982. This real discount rate
is prescribed by the Office of Management and Budget. Much debate surrounds
the proper choice of the discount rate, and there is no clear concensus on
this issue.* However, a real discount rate of 10 percent is probably too high.
The benefits and costs compared in the benefit-cost analyses are also
inconsistent in some respects. These inconsistencies include the following:
o The costs of controlling emissions in nonattainment areas are likely
to improve air quality in adjacent areas; however, these air quality
improvements are not estimated, and thus the resulting benefits are
omitted from the analysis.
o Because of the limited scope of the benefit analysis, all possible
direct benefits (e.g., visibility benefits such as plume blight,
reduced soiling and materials damages to the commercial, institutional,
and governmental sectors as well as some manufacturing industries)
associated with emissions controls have not been estimated, while,
in principle, all potentially significant direct costs associated
with the alternative standards are included in the estimates.
o Distinctions between particle sizes are not made in the benefit
analysis. The emissions control options selected in the cost
analysis, however, do depend on particle size differences.
The first two of the inconsistencies listed above could cause an over-
statement of the incremental costs relative to the incremental benefits
associated with the alternative PM NAAQS. The implications of the third
inconsistency, however, are less straightforward and deserve additional
comment.
* For a discussion of this issue, see R. C. Lind et^ £l_. (Reference 5).
-------
VII-15
Most of the studies used to estimate health benefits are based on TSP
measures of air quality. As a result, the benefit analysis cannot distinguish
between effects caused by different particle sizes. The benefits associated
with the alternative PM^g standards are estimated based on transforming,
where necessary, the estimated concentration levels associated with PM^Q
controls to the pollutant indicator of the underlying studies. The indicators
used in the underlying studies included TSP and British Smoke. Evidence
cited in the EPA/OAQPS Staff Paper, however, suggests that the PM^g fraction
of TSP is primarily responsible for adverse respiratory health effects.
If the respirable particulate air quality index is PM^g not TSP, one
might anticipate that control agencies and emitters would alter their
control strategies. With a PMjg Primary NAAQS (PNAAQS), the agencies and
emitters would focus more on the effectiveness of controls with respect to
particle size. Carrying this line of reasoning a step further, other things
remaining the same, one might expect a PM^g PNAAQS to reduce more PM^g than
a TSP PNAAQS. The control strategies forming the basis for the cost analysis
simulated such behavior by adjusting the source emissions inventory and control
device effectiveness files to distinguish between PM^g and TSP PNAAQS.
However, an analysis of PM}g to TSP ratios after controls are selected and
implemented suggests the average ratios are unchanged between pre and post
control and between PMjg PNAAQS and TSP PNAAQS conditions. The counter
intuitive results of this analysis should be interpreted in view of the
methodological and data limitations employed to select emission control
options and to project future air quality levels. It may be that modified
roll-back modelling is not sensitive enough to pick up a shift in ambient
ratios. Alternatively, non-attainment problems may be of such a form and
-------
VII-16
and magnitude that lower PM^g to TSP emission ratio sources also have to be
markedly controlled to ensure attainment, thus leaving the pre and post
control ratios virtually unchanged.
Later in this section, comparisons among alternative PM NAAQS standards
including PM^g and TSP are offered. These comparisons for the benefit
analysis comparisons are based on approximate differences in the stringency
of standards (in terms of TSP) and not differences in particle sizes.
Since the cost analysis did adjust the source emissions inventory and
control device effectiveness files to distinguish between PM^g and TSP
PNAAQS, there is a potential difference for the cost analysis between the
TSP and PM^g standards. However, the analysis of the ratios of PM^g to TSP
both before and after control suggests that this difference is not important.
If these differences are important net benefits or cost effectiveness
comparisons between TSP alternative NAAQS standards and PM^g alternative
NAAQS would be inappropriate. In such a case, a comparison between TSP
alternative standards and PM^g alternative standards would require an
assessment of the impact on health benefits of the particle size distribution
differential between the two types of standards. To further qualify the
comparison process one should also recognize that the alternative standards
not only reflect different levels (i.e., numbers) but also different forms.
The alternative TSP NAAQS are in terms of annual geometric means and observed
24 hour values while the alternative PM^g NAAQS are in terms of annual
arithmetic means and expected 24 hour values. Both the level and form of
the standard influence the stringency and hence the benefits and costs.
-------
VII-17
E. LIMITATIONS AND QUALIFICATIONS OF BENEFIT-COST ANALYSIS
The preceding discussion in this section provides the necessary
background for an interpretation of the benefit-cost analyses that follow.
Before the results of these analyses are reported, however, one should be
cognizant of some limitations of this approach in evaluating the relative
desirability of the alternative PM NAAQS.
Many of the shortcomings of benefit-cost analysis have already been
described in the preceding text. A more complete list is as follows:
o The distributive or equity impacts of the alternative PM NAAQS
are not evaluated within the framework of the benefit-cost analyses.
o The benefit-cost analyses rely on the validity and scope of
estimates of benefits and costs.
o The benefit-cost analyses are limited in scope to an evaluation
of a limited set of ambient air quality standards.
o The feasibility of the alternative PM NAAQS has not been determined.
Each of these points is discussed briefly below.
As noted previously the air quality standards will have different
impacts on members of society; those who enjoy the benefits of the standards
will not always be the same as those who bear the costs. The benefit-cost
analyses conducted in this study do not evaluate these distributive or
equity impacts.
The results of the benefit-cost analyses depend, of course, on the
validity and scope of the estimates of both benefits and costs associated
with the alternative ambient air quality standards. The wide range of
variability in the estimated benefits associated with each air quality
standard has already been described in Section VI. Similarly, all costs
associated with the alternative PM NAAQS cannot be estimated with certainty.
Moreover, in some cases, the definitions of both benefits and costs employed
-------
VII-18
in the estimates do not correspond exactly to the conceptually correct
definitions (e.g., willingness to pay). Finally, the benefit and cost
estimates are limited in scope in that they do not include all possible
benefits (i.e., some direct and all indirect omitted) and costs (i.e.,
some indirect omitted) that may result from the alternative PM NAAQS.
The benefit-cost analyses described in this section are limited in
scope in that they consider only a limited selection of possible PM NAAQS.
As a result, the benefit-cost analysis will identify, with uncertainty,
only the most efficient PM NAAQS among those considered. The identifica-
tion of the most efficient possible standard requires an evaluation of
all cost-effective and feasible PM NAAQS.
The results of the benefit-cost analyses that follow should be
interpreted in light of these limitations. Specifically, these analyses
provide a qualified assessment of the relative economic efficiency of the
alternative PM NAAQS considered.
F. BENEFIT-COST ANALYSIS OF ALTERNATIVE PM NAAQS
The results of the benefit-cost analyses of the alternative PM NAAQS
are reported below in this subsection. The relative efficiency of each
standard is determined through an analysis of the incremental benefits
and incremental costs associated with the alternative PM NAAQS.
Benefit-Cost Analysis: Incremental Net Benefits
The results of the analyses of the cost-effectiveness and relative
economic efficiency of the alternative PM NAAQS are described below.
The analyses are conducted under each of the six benefit aggregation
procedures. The six aggregation procedures and the studies included in
-------
VII-19
them are described in Section VI.* They reflect judgements regarding what
information is appropriate for an individual to consider in revealing
preferences for environmental quality improvement. Briefly, these
aggregation procedures are:
o Procedure A. Only the Criteria Document (Reference 7) classified
quantitative studies amenable to benefit analysis (Mazumdar et al.
and Ferris ^t_aj_.) are used to estimate benefits under this
aggregation procedure. Acute exposure mortality and chronic
morbidity effects are addressed in those studies. Because non-health
benefits are not covered, an overestimate of net benefits is
unlikely.
o Procedure B. This aggregation procedure augments the quantitative
studies with an acute morbidity study (Samet ^£l_.) judged by the
Staff Paper (Reference 8) as appropriate for revealing exposure
response relationships. However, because non-health benefits
such as soiling and materails damage are omitted, an overestimate
of net benefits is unlikely.
o Procedure C. This procedure expands B's coverage of the net
benefit analysis to include household soiling (Mathtech) and broader
acute morbidity effects (Ostro). To avoid double counting of
acute morbidity effects, the Samet et_ ^1_. study is replaced by
Ostro. The Mathtech analysis was listed in the Criteria Document
but was not classified as quantitative or qualitative. The Ostro
analysis was not completed in time for inclusion in the Criteria
Document. Consequently, some of the underlying studies of Procedure C
do not have as strong an endorsement as those in A and B. However,
coverage of benefit categories is still incomplete. For example,
non-user benefits of existence, bequest, and option value;
commercial, institutional, governmental, and industrial soil and
materials damage; visibility improvement; acidic deposition; and
climatic effects are omitted.
o Procedure D. This procedure is distinguished from C by the replace-
ment of the acute exposure mortality study (Mazumdar et £l_.) in C
with a set of chronic exposure mortality studies (e.g. Lave and
Seskin, Lipfert).
o Procedure E. In this procedure the chronic morbidity study by
Ferris et^ aJL is replaced by a more comprehensive, but not Criteria
Document referenced, study by Crocker et_ ^1_. Expanded coverage
is given to the household soiling and materials damage effects to
reflect the results of studies by Watson and Jaksch and Cummings
* See especially Table VI.F.I, page 28; also Reference 6.
-------
VII-20
e_t a\_. The Watson-Jaksch study is Criteria Document referenced
but is not classified as quantitative or qualitative. The
Cummings et_ aj[., like the Crocker et^ a\_., study was not completed
in time for consideration in the Criteria Document. A Mathtech
study of soiling and materials damage in selected manufacturing
industries is also included. Like the Watson-Jaksch study, it is
referenced in the Criteria Document but is not classified as
quantitative or qualitative. Aside from these differences, this
procedure is the same as D. Consequently, Procedure E provides
broader coverage of benefits than Procedure A through D but does
so using studies which have not been identified as quantitative
in the Criteria Document review process. However, coverage of
benefit categories is still incomplete.
o Procedure F. This aggregation procedure includes the acute
(Mazumdar et_ aj_.) and chronic (Lave and Seskin, Lipfert, etc.)
exposure mortality studies. Both the Ostro and Samet e t a 1.
studies are used in the acute morbidity category. This may lead
to some double counting in this category. Results from the
Watson and Jaksch and Cummings et_ ^\_. studies are added together
for the household soiling and materials damage effect category.
This may also lead to some double counting. The Mathtech study
of soiling and materials damage for selected manufacturing industries
and the Crocker et_ aj_. study on chronic morbidity used in Procedure
E are also used in this procedure. Consequently, compared to
Procedure E, Procedure F provides more extensive coverage of
specific benefit categories at the risk of double counting benefits
for some of those categories. However, coverage of benefit
categories is still incomplete.
Results are reported for analyses conducted for full (B Scenario) and
partial (A Scenario) attainment states.
Comparisons of the alternative PM^o and TSP standards are provided
below. Recall that these comparisons are based on TSP stringency and not
particle sizes.
Incremental Net Benefits
The results of the incremental net benefit analysis are reported in
Tables VII.F.I and VII.F.2 and interpreted below. Any standard that
yields positive incremental net benefits will produce an improvement in
the efficiency of resource allocation relative to baseline controls.
Alternatively, a standard associated with negative incremental net
-------
INCREMENTAL NET BENEFITS
Table VII.F.I
FOR ALTERNATIVE PM1Q AND TSP STANDARDS1
(B Scenario)
PM10 (70,250)/89
PM10 (55,-)/89
PM10 (55,250)/89
PM10 (55,200)/89
PM10 (55,150)/89
PM10 (48,183)/89
TSP (75,260)/89
TSP (-,150)/89
TSP (75,260)787
TSP (-.150)787
Aqqreqation Procedure-^
A
-.58/1.1
-2. 0/. 58
-2. 1/. 53
-2. O/. 56
-3.0/-.03
-2.8/.24
-2. O/. 99
-5.3/-1.8
-3.2/.9S
-8.1/-3.3
B
.75/2.6
.38/2.8
.33/2.9
.36/3.2
-.03/3.0
.34/3.3
1.1/4.1
-1.4/2.2
.98/5.4
-2.6/2.5
C
11/13
17/21
18/20
18/22
21/25
24/28
24/28
28/32
35/39
39/43
D
14/32
22/58
23/57
23/59
27/69
31/75
31/77
36/94
44/110
50/132
E
29/47
53/89
54/88
55/91
65/107
73/117
74/120
87/141
103/169
122/200
F
41/61
76/114
76/114
78/116
94/138
103/151
104/150
131/177
147/225
182/260
I
rv>
1. 1982 discounted present values in billions of 1980 dollars at a 10 percent discount rate. Time
horizons for 7- and 9-year standards are, respectively, 1989-1995 and 1987-1995. Comparisons
between PMjg and TSP standards are in terms of TSP stringency, not particle sizes.
2. Key: PM (x,y)/z - x=annual standard, y=24-hour standard, z=attainment year.
3. Key: V/W - V=benefits estimated assuming a mortality risk reduction factor of $0.36 per
1.0 x 10~6 annual reduction. W=benefits estimated assuming a mortality risk reduction factor
of $2.80 per 1.0 x 10"6 annual reduction.
-------
Table VII.F.2
INCREMENTAL NET BENEFITS FOR ALTERNATIVE PM1Q AND TSP STANDARDS1
(A Scenario)
PMin (70,250)/89
PM10 (55,-)/89
PM10 (55,250)/89
PM10 (55,200)/89
PM10 (55,150)/89
PM10 (48,183)/89
TSP (75,260)/89
TSP (-,150)/89
TSP (75,260)/87
TSP (-,150)/87
Aggregation Procedures^
A
-.21/1.0
-.57/1.0
-.59/1.0
-.61/1.0
-.90/.74
-.97/.6S
-.72/.92
-1.4/.62
-1.1/1.3
-2. 1/. 67
B
.7/1.7
.6/2.2
.57/2.2
.65/2.3
.64/2.2
.58/2.4
.92/2.5
.62/2.6
1.1/3.5
.67/3.5
C
7.1/8.3
9/11
9/11
10/12
12/14
12/14
12/14
15/17
16/20
20/24
D
8.9/23
12/30
12/30
13/31
15/35
15/37
15/37
19/45
21/51
26/64
E
18/30
25/43
25/43
26/44
31/53
32/52
32/54
40/68
45/75
56/94
F
25/39
37/55
37/55
38/58
45/69
46/70
47/71
60/ 88
66/98
85/129
I
F\J
rv>
1.
2.
3.
1982 discounted present values in billions of 1980 dollars at a 10 percent discount rate. Time
horizons for 7- and 9-year standards are, respectively, 1989-1995 and 1987-1995. Comparisons
between PMig and TSP standards are in terms of TSP stringency, not particle sizes.
Key: PM (x,y)/z - x=annual standard, y=24-hour standard, z=attainment year.
Key: V/W - V=benefits estimated assuming a mortality risk reduction factor of $0.36 per
1.0 x 10~6 annual reduction. W=benefits estimated assuming a mortality risk reduction factor
of $2.80 per 1.0 x 10~6 annual reduction.
-------
VII-23
benefits will produce an inefficient allocation of resources relative to
baseline controls. The standard generating the largest positive incremental
net benefits is the most efficient of the alternative standards evaluated.
The figures presented in Tables VII.F.I and VII.F.2 represent estimates
of incremental net benefits corresponding to the full and partial attain-
ment scenarios (scenarios B and A, respectively).* The range of figures
results from the two valuation factors applied to reductions in mortality
risk. See Section VI .D. The results of the incremental net benefit
analysis shed light on three important issues regarding the alternative
PM NAAQS. These issues are:
o The relative economic efficiency of the alternative standards;
o The economic efficiency of complete versus partial attainment; and
o The economic efficiency of the implementation periods for the TSP
options.
Some conclusions regarding each of these issues are provided immediately
below. These findings, however, should be interpreted in view of the
previously mentioned caveats regarding benefit-cost analysis.
Economic Efficiency of Alternative Standards
The estimated incremental net benefits associated with the alternative
PM NAAQS may be affected by the aggregation procedures employed, the mortality
risk reduction valuation factor applied, and the attainment status considered.
A general understanding of how these three conditions affect estimated
* The figures reported in Table VII.'F.l are computed as the difference between
the incremental benefits reported earlier in Table VI.G.I and the incre-
mental costs for the B scenario reported in Table VII.D.I. Comparable
figures in Table VII.F.2 are computed as the difference between estimated
incremental benefits in Table VI .G.2 and the incremental costs for the A
scenario in Table VII.D.I.
-------
VII-24
incremental net benefits is helpful in evaluating the economic efficiency
of specific alternative standards.
Other things being the same, the estimated incremental net benefits
associated with a given standard increase as more comprehensive aggregation
procedures and the higher mortality risk reduction valuation factor are
applied. The estimated costs associated with a given standard,
however, remain constant across aggregation procedures and valuation
factors. Consequently, the estimated incremental net benefits associated
with all standards increase as more comprehensive aggregation procedures
and higher valuation factors are employed. It is also noteworthy
that, ceteris paribus, the estimated incremental net benefits associated
with more stringent standards typically increase relative to those associated
with less stringent standards as more comprehensive aggregation procedures
and higher mortality risk reduction valuation factors are applied.
The estimated incremental net benefits under the full attainment
scenario (B) are usually higher than those estimated under the partial
attainment scenario (A), other things being the same. This observation,
however, consistently does not hold for aggregation procedure A.
Given the general effects of aggregation procedures, mortality risk
reduction evaluation factors, and attainment states, on estimated incremental
net benefits, the evaluations of the relative economic efficiency of
specific standards are perhaps more clear.
The domain of economically preferred air quality standards is dis-
played in Table VII.F.3. Among the alternatives considered, the most
efficient standard (i.e., the standard associated with the largest posi-
tive estimated incremental net benefits) is indicated for each of the
several conditions under which incremental net benefits are estimated.
-------
Table VII.F.3
DOMAIN OF ECONOMICALLY PREFERRED STANDARDS*
Conditions
LOWER VALUATION FACTOR**
Partial Attainment
(Scenario A)
LOWER VALUATION FACTOR**
Complete Attainment
(Scenario B)
HIGHER VALUATION FACTOR*
Partial Attainment
(Scenario A)
HIGHER VALUATION FACTOR*
Complete Attainment
(Scenario B)
Aggregation Procedure
A
No Standard
Efficient
No Standard
Efficient
r
TSP 75/260
[PM10 41/143]
r
Least
Stringent
Alternative
(7-year)
B
TSP 75/260
[PMin 41/43]
TSP 75/260
(7-year)
[PM10 41/143]
TSP 75/260-
[PM10 41/143]
TSP 75/260
(7 year)
[PMin 41/143]
C
Most
Stringent
Alternative
Most
Stri ngent
Alternative
Most
Stringent
Alternative
(7-year)
Most
Stringent
Alternative
D
Most
Stringent
Alternative
Most
Stringent
Alternative
Most
Stringent
Alternative
Most
Stringent
Alternative
E
Most
Stringent
Alternative
Most
Stringent
Alternative
Most
Stri ngent
Alternative
Most
Stringent
Alternative
F
Most
Stringent
Alternative
Most
Stringent
Alternative
Most
Stringent
Alternative
Most
Stri ngent
Alternative
I
rv>
tn
* All preferred standards, except as noted, are 9-year standards with the 1987-1995 time horizon. Comparisons between
PM^Q and TSP standards are in terms of TSP stringency, not particle sizes. Values shown in brackets [ ] are TSP
values multiplied by a PM^ to TSP ratio of 0.55. The incremental net benefit of the bracketed standards is not
equivalent to that of the TSP standard. Furthermore, the ordering of the bracketed standards vis a vis the other
standards analyzed could change. This is because control strategy design and costing may be sensitive to particle
size of the pollutant (i.e., TSP vs. PM10).
**These refer to the valuation factors for reduced mortality risk. The lower value is $0.36 for a unit reduction
of 1.0 x 10"6 in annual mortality risk. The higher factor is $2.80 for the same unit reduction.
-------
VII-26
As shown in Table VII.F.3, the findings are robust once one moves
beyond aggregation procedure A. For the other aggregation procedures, the
economically preferred standards are at the lower end of the range of
standards considered. Again, the inferences drawn relate to stringency
levels of the alternative standards and not to the desirability of PMjg
versus TSP as a respirable particulate indicator.
Aggregation procedures A and B which use the CASAC and staff paper
selected studies exclusively provide the least complete coverage of
potential benefits. Using the lower valuation factor for mortality risk
reduction, aggregation procedure A suggests baseline control is preferred
to any of the alternative standards considered; aggregation procedure B
indicates the TSP 75/260 standard is preferred. This swing from
no alternative standard to one of the most stringent being preferred
results from an expanded coverage of benefits. Procedure B augments
benefit estimates in the acute exposure mortality and chronic morbidity
category to also include estimates in the acute morbidity category.
Using the higher valuation factor for mortality risk reduction, the
findings for aggregation procedure B remain unchanged. However, those
for procedure A do change. Under partial attainment (scenario A) the TSP
75/260 standard is preferred; under complete attainment the PM^Q 70/250
standard is preferred. As noted previously, the incremental net benefits
under full attainment may be less than those associated with partial
attainment. This results from one of the studies underlying procedure A.
The chronic morbidity category benefits for procedure A were derived from
an epidemiological study whose observed lowest air quality concentration
was generally equalled or surpassed by employing the PMjg 70/250 scenario A
-------
VII-27
control strategy. Consequently, even though air quality improved in
moving to scenario B, strict application of the underlying study (i.e., not
extrapolating benefits beyond the range of observed concentrations) did
not permit a concommittent increase in estimated benefits. This relationship
does not influence the findings for the other aggregation procedures
using this study (i.e., B, C, and D). Such procedures provide a more
comprehensive estimate of benefits; and, estimated benefits in other
effects categories dampen the relative impact of the chronic morbidity
study.
Aggregation procedures C and D provide more complete coverage of
potential benefit categories. In particular, benefits are estimated for
the household soiling and materials, mortality, acute morbidity, and
chronic morbidity categories. However, coverage is still incomplete by
virtue of omitting benefits in categories such as soiling and materials
(manufacturing, commercial, institutional, and governmental sectors),
visibility, and climatic change. However, procedures C and D employ
studies which although acceptable for purposes of benefit analysis
were not deemed quantitative by CASAC or the Staff Paper for purposes of
standard setting.
The findings for procedures C and D are remarkably stable with the
most stringent standard TSP (-/150) being preferred on economic efficiency
grounds regardless of attainment state or mortality risk reduction
valuation factor conditions. Again, the alternative standards are compared
on the basis of stringency; not the desirability of PMjg versus TSP as an
indicator of respirable particulate.
-------
VII-28
Procedures E and F provide even broader coverage of potential benefits
by including soiling and materials effects for a small portion of the
manufacturing sector. Of course, coverage of potential benefits is still
incomplete. However, by trying to more completely account for benefits
in the mortality, acute morbidity, and household soiling and materials
categories the possibility of double counting in those categories is
enhanced.
Like the findings for procedures C and D, those for E and F are also
stable. The most stringent standard TSP (-/150) is preferred regardless
of attainment state or mortality risk reduction valuation factor conditions.
Throughout the analysis the standards are compared on the basis of
stringency. The desirability of PMjg versus TSP as an indicator of
respirable particulate is beyond the scope of this analysis.
If only of the alternative PM^g standards are evaluated for relative
economic efficiency, the most stringent PMjg standard (48,183) is generally
preferred. In particular, for aggregation procedures C, D, E, and F the
(48,183) standard yields the highest positive incremental net benefits
regardless of attainment state or mortality risk reduction valuation
factor conditions.
For procedure B, with the higher valuation factor, the PM^g (48,183)
standard is also preferred. However, with the lower valuation factor, the
least stringent PM^g (70,250) standard is preferred. Recall, when more
alternatives were considered under the same conditions, the relatively
stringent TSP 75/260 standard was preferred. This illustrates the potential
non-intuitive consequences of indiscriminately limiting the alternative
set of standards. In particular, it shows the implications of not performing
-------
VII-29
cost-effectiveness analysis to identify the dominant set of alternative
standards.
Using procedure A and limiting the set of considered standards to
the PM^Q alternatives, baseline conditions are preferable given lower
valuation factors and either complete or partial attainment. Using the
higher mortality risk reduction valuation factors and partial attainment
several of the less stringent PM^Q alternatives are equally preferable
(i.e., 70/250, 55/-, 55/250, 55/200). Recall again that under the same
conditions when more alternatives were considered, the relatively stringent
TSP 75/260 standard was preferred. This also illustrates the potential
non-intuitive consequences of limiting the alternative set of standards.
However, using the higher valuation factor and complete attainment conditions
the PM}Q (70/250) standard which was preferred when a more expanded set of
standards were considered is also preferred here.
When the standards which yield the highest net incremental benefit
are the least or most restrictive considered, it is desirable to evaluate
a wider range of alternative standards. Specifically, in the case where PMjo
(70/250) is preferred, more lenient alternatives should be examined to
determine if incremental net benefits peak at 70/250, or whether they
would be larger with a less stringent standard. In cases where the TSP 150
(or the PMio 48 AAM/183) standard is preferred, more stringent alternative
standards should be examined to see if incremental net benefits peak, or
if some more stringent standard is economically preferred.
As noted earlier in this section, applied benefit-cost analysis
typically can provide only a qualified economic assessment of regulatory
alternatives. The previously mentioned caveats regarding benefit, cost,
-------
VII-30
and air quality estimates as inputs to the analysis should be considered
in using the incremental net benefits to assess the relative efficiency
of the alternative PM NAAQS. First, in order to determine if a given
standard is efficient (i.e., relative to the baseline scenario), the
positive difference between estimated incremental benefits and costs
should be statistically significant. Second, in order to identify the
most efficient standard (among those evaluated), the difference between.
estimated incremental net benefits for the preferred and the next best
standard should also be statistically significant. However, because some
of the uncertainty in the estimated incremental benefits and costs cannot
be measured, no statistical tests have been conducted. The differences
in estimated incremental net benefits do become greater, both in
absolute and relative terms, as more comprehensive aggregation schemes
are employed. These differences, however, do not imply the presence or
the absence of statistical significance.
Economic Efficiency: Attainment Status and TSP Implementation Periods
Based on the results of the benefit-cost analysis, complete attainment
is generally preferred to partial attainment. Other things being equal,
estimated incremental net benefits are generally larger under the B scenario.
However, partial attainment is sometimes preferred for particular standards
evaluated, if aggregation procedure A or B is adopted. These conclusions
depend on the accuracy of the estimates of the differential cost associated
with complete attainment. As noted previously, these cost estimates are
highly uncertain.
The findings of the analysis regarding TSP implementation periods
for aggregation procedures D, E, and F indicate that the 9-year implementation
-------
VII-31 .
period (beginning in 1987 and extending to 1995) is preferred to the 7-
year implementation period (beginning in 1989 and extending to 1995).
However, those findings do not consistently hold for procedures A, B, and C.
For example, in aggregation procedure C where the TSP 150 standard is
preferred under conditions of partial attainment the 7-year period yields
higher positive incremental net benefits while under conditions of complete
attainment the 9-year period yields higher benefits. In procedure B using
the low mortality risk reduction valuation factor, the preferred TSP
75/260 standard yields higher benefits for the 9-year period given partial
attainment; however, with full attainment and the low valuation factor
under procedure B the 9-year period is preferred.
As stated earlier, these conclusions imply the ability to measure
statistically significant differences in benefits and costs. As was the
case for the analysis of the relative efficiency of specific standards,
differences in estimated incremental benefits and costs become greater as
more comprehensive aggregation procedures are applied. However, the
earlier caveat applies here as well: relatively large differences do not
imply statistically significant differences; smaller differences do not
necessarily imply the absence of statistical significance.
Distribution of Incremental Net Benefits
For alternative PM NAAQS yielding positive incremental net benefits,
those who benefit are assumed able to potentially compensate those who
incur cost and still retain some welfare gain. Because the compensation
does not take place, some people lose when air quality is improved. An
examination of some of the distributional consequences will allow one to
better judge the alternative standards. In view of the information
-------
VII-32
developed in References (2), (3), and (6), qualified statements can be
made about the regional, sectoral, and demographic aspects of the
incremental net benefit totals.
Mathtech (Reference 6) performed a regional incremental net benefit
analysis. However, because they used a different mortality risk reduction
valuation factor (i.e., the midpoint of our range), the results are not
exactly comparable. But, they found in some instances even though total
incremental net benefits were positive some regions (Federal administrative
regions) experienced negative incremental net benefits. Mathtech also
found the regional distribution of estimated incremental net benefits
varies across aggregation procedures and standards. However, these
differences become less pronounced as aggregation procedures become
more comprehensive and as standards become more restrictive. But, the
largest estimated incremental net benefits generally accrued to the East
North Central and South Pacific regions despite variations in aggregation
procedures or standard stringency levels.
Incremental net benefits will also accrue across various economic
sectors. For example, benefits and/or costs may be incident upon the
household, agricultural, manufacturing, mining, commercial, institutional,
and governmental sectors. In this analysis however, benefit estimates
were limited to the household and a small subset of the manufacturing
industry. Cost estimates included control of private sources in the
agricultural, mining, construction, manufacturing, service utilities,
wholesale goods and service industries as well as government-owned
utilities and roads. Consequently, a complete sectoral analysis of
incremental net benefits is not possible. However, the results of the
-------
VII-33
economic impact analysis for the most restrictive standard for Scenario A
does provide some information on the sectoral distribution of control
costs.* Two sectors, manufacturing and utility, are estimated to account
for the vast majority of initial control cost incidence. However, passing
some of these costs onto the purchasers of manufacturing goods and utility
services is estimated to reduce some of the initial cost incidence.
But, for more restrictive standards this may not be enough to prevent
closing some establishments in certain other industries. However, no net
adverse impact on industry output or employment is anticipated. Furthermore,
foreign trade, investment, productivity, and innovation effects are not
expected to be significant.
G. CONCLUSIONS AND QUALIFICATIONS
The benefit-cost analyses reported earlier in this section were
conducted to assess the economic efficiency of the alternative standards.
The economic efficiency considerations of attainment status and the
implementation period for the TSP standards were also evaluated. The
conclusions offered from the interpretation of these analyses, however,
are subject to qualifications.
Conclusions
The major economic efficiency conclusions that emerge from the
benefit-cost analyses are described briefly below for each of the several
aspects of the alternative PM NAAQS that were evaluated.
*See Section V and Reference 2.
-------
VII-34
The ranking of the standards in terms of economic efficiency is
fairly stable across aggregation procedures B, C, D, E and F.
Adopting any of these regardless of attainment state or mortality
risk reduction valuation factor conditions leads to the relatively
more restrictive standards being preferred on economic efficiency
grounds.
The ranking of standards for aggregation procedure A is sensitive
to the valuation factor (for mortality risk reduction) and
attainment state conditions. Preferred alternatives range from
baseline control levels to one of the more stringent (TSP)
standards. However, despite the credibility ascribed to the
studies underlying procedure A by the Criteria Document, the
coverage of potential benefit categories is relatively incomplete.
Consequently, the rankings are expected to be relatively less
stable.
o When only PM^g standards are evaluated on economic efficiency
grounds, the most stringent PM^g standard 48/183 is generally
preferred (i.e., procedures C, D, E, and F). However, because
the PM^g 48/183 is dominated by the TSP standards on cost
effectiveness grounds, the PMjg 48/183 does not always replace
the TSP standards in the rankings of preferred standards for
aggregation procedures A and B.
Qual i f ications
The general limitations of applied benefit-cost analysis and some
specific qualifications to this analysis have already been described in
this section. The conclusions summarized immediately above should be
assessed in full view of these limitations and qualifications. Some of
those as well as other specific qualifications to this analysis are
summarized briefly below.
o The validity of the benefit-cost analyses depends on the estimates
of benefits, costs, and air quality. Sources of potential biases
and uncertainty in each of these estimates have already been
described. It is noted that these biases and uncertainties can
bear directly on the conclusions summarized above. These
conclusions depend on the ability to detect meaningful differences
between benefits and costs and differences in net benefits across
the alternative standards. As was stated previously, tests for
statistically significant differences have not been conducted
because not all of the uncertainty associated with the estimated
benefits and cost is known.
-------
VII-35
A fundamental premise of benefit-cost analysis is consumer
sovereignty. Benefits are associated with alternative levels of
air quality if consumers perceive the benefits to exist. Such
benefits are measured appropriately by society's willingness to
pay to prevent the perceived adverse effects of air quality
deterioration. However, the revealed preferences of consumers
have not been measured directly in this analysis.* Instead, the
consequences of alternative revealed preference states have been
simulated via alternative aggregation procedures and mortality
risk reduction valuation factors.
As indicated in the findings, the results of the incremental
benefit-cost analyses are fairly stable. This is important because
proper accounting of consumer sovereignty becomes less of an
issue. The appropriate set of assumptions under which the benefit-
cost analysis is conducted depends on the perceptions of consumers.
Risk taking consumers may adopt relatively incomplete aggregation
procedures A and B because of the "certainty" and "credibility"
associated with the Criteria Document and Staff Paper endorsement
of the underlying studies. Risk-neutral consumers may adopt the
broader but still incomplete procedures C and D. Whereas, risk-
averting consumers may argue that procedures E and F understate the
intensity of their revealed preferences because of the omitted
and incompletely covered benefit categories. Likewise, the
risk-taking consumer may adopt the lower mortality risk reduction
valuation factor while the risk-averting consumer may adopt the
higher valuation factor.
The results for aggregation procedures B, C, 0, E, and F are
stable pointing to the more stringent standards as being
economically preferred regardless of the mortality risk reduction
valuation factor employed. Consequently, it seems unlikely that
the sum of the dollar weighted votes of sovereign risk-taking,
risk-neutral, and risk-averting consumers regarding appropriate
aggregation schemes and mortality risk reduction valuation factors
would alter the findings. In particular, even if consumer
sovereignty were explicitly considered, the more stringent standards
appear to be the probable social choice on economic efficiency grounds.
The findings of the incremental benefit-cost analysis were
developed on the assumption that more complete coverage of the
benefits and costs would not change the rankings. However, the
limited coverage of direct benefits and indirect costs as well
as the omission of indirect benefits in their entirety should be
kept in mind in interpreting the findings.
*However, specific components of the various aggregation procedures have
attempted to do so in an indirect manner (i.e., Mathtech soiling and
material damage studies; Crocker et_ a_l_. chronic morbidity study).
-------
VII-36
In most cases, the alternative PM NAAQS ranked as most efficient are
either the most (i.e., aggregation procedures C, D, E, and F for all
conditions) or least (i.e., aggregation procedure A for 1 set of conditions)
stringent of the standards evaluated. Because of this, additional information
on efficiency could have been obtained if a wider range of standards had
been evaluated. However, the analysis was limited to options that were
no less restrictive than the middle of the EPA Staff Paper range (i.e.,
PM 70 AAM/250 24-hour expected value) and no more restrictive than the
current TSP secondary standard (150 24-hour second high).
-------
VII-37
REFERENCES
1. Mills, Edwin S. The Economics of Environmental Quality. New York:
W. W. Norton & Company, 1978.
2. Viola, J., R. Silver-man, S. Schechter, J. Wagner, E. Hurley, J.
McGovern and G. Scarborough. Technical Appendices, Economic Impact
Analyses of Alternative National Ambient Air Quality Standards for
Particulate Matter. Final Report to the U.S. Environmental Protection
Agency by Energy and Environmental Analysis, Inc. Arlington, Virginia,
December 1982.
3. Smith, A.E., and K. L. Brubaker, 1982. Costs and Air Quality Impacts
of Alternative National Air Quality Standards for Particulate Matter,
Technical Support Document. Argonne National Laboratory, October 1982.
4. Laarman, J. Comparison of Mathtech with BEA Data on SIC Growth Rates.
Memorandum to T. Walton of U.S. Environmental Protection Agency,
Durham, North Carolina, February 17, 1982.
5. Lind, R. C., K. J. Arrow, G. R. Corey, P. Dasgupta, A. K. Sen, T.
Stauffer, J. E. Stiglitz, J. A. Stockfisch, and R. Wilson. Discounting
for Time and Risk in Energy Policy. Baltimore: Johns Hopkins University
Press, 1982.
6. Manuel, E. H. Jr., R. L. Horst Jr., K. M. Brennan, J. H. Hobart, C.
D. Harvey, J. T. Bentley, M. C. Duff, D. E. Klinger, J. K. Tapiero.
Benefit and Net Benefit Analysis of Alternative National Ambient Air
Quality Standards for Particulate Matter. Report to U.S. Environmental
Protection Agency under EPA Contract Number 68-02-3826 by Mathtech
Inc., Princeton, New Jersey, March 1983.
7. U.S. Environmental Protection Agency, Environmental Criteria and
Assessment Office. Air Quality Criteria for Particulate Matter and
Sulfur Oxides: EPA-600/8-82-029c. Research Triangle Park, North
Carolina, December 1982.
8. U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards. Review of the National Ambient Air Quality Standards
for Particulate Matter: Assessment of Scientific and Technical
Information. OAQPS Staff Paper (EPA-450/5-82-001), Research Triangle
Park, North Carolina, January 1982.
-------
VIII-1
VIII. Summary of Rationale for Choosing the Proposed Action
In accordance with sections 108 and 109 of the Clean Air Act, EPA
has reviewed and revised the criteria upon which the existing primary and
secondary particulate matter standards are based. The existing primary
standards for particulate matter (measured as "total suspended particulate
matter" or "TSP") are 260 ug/m3, averaged over a period of 24 hours and
not to be exceeded more than once per year, and 75 pg/nr annual geometric
mean. The secondary standard (also measured as TSP) is 150 pg/m3,
averaged over a period of 24 hours, and not to be exceeded more than once
per year.
As a result of its review and revision of the health and welfare
criteria, EPA proposes the following revisions to the particulate matter
standards:
1) that TSP as an indicator for particulate matter be replaced for both
of the primary standards by a new indicator that includes only those particles
with an aerodynamic diameter smaller than or equal to a nominal 10 micrometers
2) that the level of the 24-hour primary standard be changed to a
value to be selected from a range of 150 to 250 pg/m3 and that the
current deterministic form of the standard be replaced with a statistical
form that permits one expected exceedance of the standard level per year;
3) that the level and form of the annual primary standard be changed to
a value to be selected from a range of 50 to 65 pg/m3, expressed as an
expected annual arithmetic mean; and
4) that the current 24-hour secondary TSP standard be replaced by
an annual TSP standard selected from a range of 70 to 90 pg/m3, expected
annual arithmetic mean.
-------
VIII-2
Because no scientific consensus exists on the proper levels of the
standards, and the analytical and policy bases for making these decisions
under the statute are limited and unclear, the Administrator is not proposing
specific standard levels within the above ranges. Rather, he is soliciting
additional comment and information from the public to be considered in
promulgating the final regulation, which will specify a specific level for
each of the standards. Given the precautionary nature of the Act, the
Administrator is inclined to select the levels of primary standards from the
lower portion of the above ranges. A more detailed discussion of the
rationale for this proposal is available in the Federal Register preamble.
-------
IX-1
IX. STATUTORY AUTHORITY
The statutory authority for the proposed revision of the particulate
matter NAAQS is contained in the Clean Air Act. Two sections of the Act
govern development of NAAQS. Section 108 (42 U.S.C. 7408) requires EPA
to document the most recent scientific basis (criteria) for setting an
ambient standard. Section 109 provides authority for reviewing the
criteria and establishing primary (health based) and secondary (welfare
based) NAAQS. A more complete discussion of the legal authority for this
proposed regulation is contained in the proposal preamble.
-------
ADDENDUM
Regulatory Impact Analysis
on
The National Ambient Air Quality Standards
for
Particulate Matter
I. Introduction
The purpose of this Addendum is to draw together and briefly summarize
all information in the RIA relevant to the range of particulate matter
standards proposed in the Federal Register and to report the results of
more recent analyses. The Regulatory Impact Analysis (RIA) for Particulate
Matter was prepared to fulfill the requirements of Executive Order 12291
(E.O. 12291). The draft RIA addresses the cost, benefit, and economic
impacts of various levels of a particulate matter (PM) standard. The Clean
Air Act specifically requires that ambient standards be based on scientific
criteria relating to the level of air quality that should be attained to
protect public health and welfare adequately. The Act also precludes
consideration of costs or technological feasibility in determining the
level of ambient standards. Accordingly, EPA has not considered the results
of the draft RIA in selecting the range of standards that are being proposed.
Following completion of the draft RIA and its supporting analyses, a
range of levels for the primary and secondary standards was selected for
proposal which had been only partially addressed in the basic cost and
benefit analyses. Specifically, it was decided, based on available scientific
and technical information, that the Total Suspended Particulate (TSP)
indicator for PM for the primary standards should be replaced by a new
indicator that includes only those particles with an aerodynamic diameter
less than or equal to a nominal 10 urn (PM^), while TSP should be retained
-------
as the indicator for the secondary standard. Because combined standards
(PM^o and TSP) had been considered unlikely when the control cost model on
which the RIA is based was developed, the model was not structured to
consider the two pollutant indicators simultaneously. As a result, benefits/costs
could not be directly estimated for a joint PM^Q/TSP standard. However,
rough benefit/cost estimates were made by comparing the results for the two
indicators run separately.
In addition to the combined PM^Q/TSP standard, changes were also proposed
for the form of the proposed annual standards. Specifically, the annual
arithmetic mean primary and secondary standards were changed from a
deterministic form to an expected value form. The air quality data used as
input to the control cost model assume a deterministic form. Because of
the manner in which the data were reported and recorded, it has not been
possible to directly calculate the difference this change might have on the
cost. In the RIA the annual component was analyzed as the highest mean in
a multi-year record, whereas the proposed range involves the mean of several
years (usually 3 years).
A more recent air quality data base has been used to assess potential
non-attainment under a combined standard. In addition to employing more
recent air quality data, this assessment used more sophisticated statistical
methods to relate observed TSP levels to estimated PMjg levels. Finally,
this assessment can estimate the impact of changing the statistical form of
the annual standards.
This Addendum addresses the benefit/cost of the proposed range of
primary PM^g standards (Section II) and also of the proposed range of
secondary TSP standards (Section III). Section IV presents other
-------
considerations relative to benefits and costs, while Section V presents
related air quality and non-attainment analyses.
II. Primary National Ambient Air Quality Standards
The range of primary standards proposed for public comment is PM^g
(50, 150) to PMio (65, 250). However, as noted above, this range was not
analyzed in the full RIA. The range was analyzed later and the results
are reported below.
This portion of the addended analysis compares the incremental
benefits and costs associated with the range of proposed PM^g standards.
The aggregation procedures, attainment state, and mortality risk
reduction valuation conditions are the same as described in Section VI of
the RIA.
The estimated incremental net benefits are presented in Table I I.I.
The findings are consistent with those derived from an analysis of the
standards/conditions analyzed previously. Specifically, as one adopts
more comprehensive benefit aggregation procedures (i.e. C, D, E, F and
in certain circumstances, B) the more stringent standards are preferred
on economic efficiency grounds.
The original analysis of PMjg standards and this one assumed a PM^g
to TSP ratio of 0.55. Since more recent and complete ambient data show
the mean national ratio to be closer to 0.46, an analysis of the proposed
range of PM^g standards was conducted using that ratio. The results are
presented on Table II.2. For 5 of the 24 cases (i.e. 6 aggregation
procedures x 2 standards x 2 mortality risk reduction valuation
factors) adopting the .46 ratio results in greater incremental net
benefits. For 4 of those cases (i.e. aggregation procedure A with the
higher valuation factor for both standards; aggregation procedure B with
-------
TABLE I I.I
Estimated Incremental Net Benefits for the Proposed Range of PMjQ Primary Standards'*
Full Attainment
($ Billion)
Aggregation Procedures
Standards/ Conditions
PM10 (50,150)
Full attainment with lower mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
Full attainment with higher mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
PM10 (65,250)
Full attainment with lower mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
Full attainment with higher mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
A
-2.9
.21
-.60
1.4
B
.30
3.3
1.0
3.0
C
23
27
13
15
D
30
76
16
40
E
72
120
36
60
F
103
152
51
77
*1982 discounted present values in billions of 1980 dollars at a 10 percent discount rate. The time horizon is
the 7 year period starting January 1, 1989 and ending December 31, 1995. The TSP annual arithmetic mean lower
bound of 110 pg/m^ when imposed is applied to all health studies.
-------
TABLE 11.2
Estimated Incremental Net Benefits for the Proposed Range of PM^Q Primary Standards
Using 0.55 and 0.46 TSP to PM10 Ratios*
Full Attainment
($ Billion)
Aggregation Procedures
Standards/ Conditions
PM10 (50,150)
Full attainment with lower mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
0.46 PM10 to TSP ratio
Full attainment with higher mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
0.46 PM10 to TSP ratio
PM10 (65,250)
Full attainment with lower mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
0.46 PM10 to TSP ratio
Full attainment with higher mortality risk
reduction valuation factor
0.55 PM10 to TSP ratio
0.46 PM10 to TSP ratio
A
-2.9
-1.4
.21
.95
-.60
-.32
1.4
.86
B
.25
.87
3.3
3.4
1.0
.53
3.0
1.7
C
24
18
27
20
13
7.0
15
8.0
D
30
22
76
55
16
8.7
40
22
E
72
50
120
83
36
16
60
29
F
100
71
150
110
51
24
77
38
en
*1982 discounted present values in billions of 1980 dollars at a 10 percent discount rate. The time horizon is
the 7 year period starting January 1, 1989 and ending December 31, 1995. The TSP annual arithmetic mean lower
bound of 110 pg/m3 when imposed is applied to all health studies.
-------
the lower valuation factor for both standards) the difference is great
enough to change the ordering of incremental net benefits between standards.
Where the ordering is changed by adopting the 0.46 ratio the more restrictive
PM^o (50,150) standard yields the highest incremental net benefits. In the
remaining 20 cases the ordering of standards is unaffected by the choice
of PM]_Q to TSP ratios.
III. Secondary National Ambient Air Quality Standards
The range of secondary standards proposed for public comment is TSP
(70 annual expected arithmetic mean) to TSP (90 annual expected airthmetic
mean). The current TSP (-, 150 24-hr, second high) standard was the only
secondary standard analyzed earlier.
This portion of the addended analysis compares the incremental
benefits and costs associated with the range of proposed TSP standards.
Again, the benefit aggregation procedures, attainment state, and mortality
risk reduction valuation conditions are the same as described in Section
VI of the RIA.
The estimated incremental net benefits are presented in Table III.l.
The findings are generally consistent with those appearing in Section VII
of the RIA. Specifically, more comprehensive benefit aggregation procedures
(i.e., C, D, E, and F) lead to selection of the more stringent standard
on economic efficiency grounds.
IV. Other Considerations
The analysis of the proposed range of PM^g and TSP standards described
earlier does not permit definitive statements regarding which portion of
the range, if any, yields larger or maximizes incremental net benefits.
To address this issue the incremental net benefits for the proposed range
-------
TABLE 11 I.I
Estimated Incremental Net Benefits for the Propose Range of TSP Secondary Standards*
Full Attainment
($ Billion)
Aggregation Procedures
Standards/ Condi ti ons
TSP (70)
Full attainment with lower mortality risk
reduction valuation factor
Full attainment with higher mortality risk
reduction valuation factor
TSP (90)
Full attainment with lower mortality risk
reduction valuation factor
Full attainment with higher mortality risk
reduction valuation factor
A
'
-5.4
-1.7
-1.6
1.4
B
-1.0
2.7
1.3
4.2
C
32
35
23
26
D
40
104
29
71
E
99
163
66
110
F
145
212
95
140
*1982 discounted present values in billions of 1980 dollars at a 10 percent discount rate.
is the 7 year period starting January 1, 1989 and ending December 31, 1995. The TSP annual
lower bound of 110 pg/nv* when imposed is applied to all health studies.
The time horizon
arithmetic mean
-------
of PM^g and TSP standards were compared to the other PM^ and TSP standards
analyzed in Section VII of the RIA. These other standards include some
within and beyond the proposed range for the PMjg and TSP standards. In
addition, another TSP standard (i.e., 80 annual arithmetic mean) was also
included. The standards were compared under conditions of full attainment
assuming a 7 year (i.e., 1989 to 1995) project time horizon and a PM^g to
TSP ratio of 0.55.
The results are presented in Table IV.1. The limitations and
qualifications to the incremental net benefit analysis discussed in
Section VII of the RIA also apply here and in Sections II and III of this
addendum. For example, uncertainties in the calculations could affect
the ability to show meaningful differences across standards and aggregation
schemes. Furthermore, the findings of the incremental benefit-cost
analysis are developed assuming more complete coverage would not change
the results. Hence, the omission of indirect benefits in their entirety
and the limited coverage of direct benefits and indirect costs should be
kept in mind. In addition, the standards are compared on levels of TSP
"equivalent" stringency. Consequently, no inferences can be drawn on
economic efficiency grounds regarding the merits of TSP versus PM^g
indicators. Finally, because the full set of cost-effective PM^g and TSP
NAAQS alternatives was not analyzed, our statements regarding the portion
of the proposed range which yields relatively larger net benefits must be
interpreted accordingly.
With the more comprehensive aggregation procedures C, D, E, and F
the findings remain unchanged regardless of the mortality risk reduction
valuation factor employed. The lower portion of the proposed ranges of
-------
TABLE IV.1
Economically Preferred Standards and Portions of the Proposed Range
Aggregation Procedures
Standards/ Conditions
With Lower Mortality Risk Reduction
Valuation Factor
- Preferred Standard
- Preferred Portion of the Proposed
Range
-Primary Standard
-Secondary Standard
With Higher Mortality Risk Reduction
Valuation Factor
- Preferred Standard(s)
- Preferred Portion of the Proposed
Range
-Primary Standard
-Secondary Standard
A
No
Standard
Feasible
No Portion
No Portion
TSP (90)
Upper
Upper
B
TSP (90)
Upper
Upper
TSP (80)
Lower
Middle
C
TSP (70)
Lower
Lower
TSP (70)
Lower
Lower
D
TSP (70)
Lower
Lower
TSP (70)
Lower
Lower
E
TSP (70)
Lower
Lower
TSP (70)
Lower
Lower
F
TSP (70)
Lower
Lower
TSP (70)
Lower
Lower
-------
10
primary and secondary standards is always preferred. The preferred
standard, comparing all alternatives on the basis of stringency, is the
TSP(70).
With the less comprehensive procedures A and B, the findings are
sensitive to both the aggregation procedure and mortality risk reduction
valuation factor used. With procedure A and the lower factor, no portion
of the proposed range of primary and secondary standards yields positive
incremental net benefits. Furthermore, none of the alternative standards
outside the upper end of the proposed range (e.g., PM^o (70/250) had
positive incremental net benefits. With the same procedure and the higher
valuation factor the upper end of the proposed ranges for the primary and
secondary standards is preferred. In that situation the TSP (90) is the
alternative standard yielding the greatest positive incremental net
benefits. Under aggregation procedure B and lower valuation factor
conditions, the upper portion of the proposed ranges for the primary and
secondary standards is preferred while the TSP (90) is the alternative
yielding the highest positive incremental net benefits. With the same
procedure and the higher valuation factor, the lower portion of the
proposed primary standard range and the middle portion of the proposed
secondary standard range are preferred. The TSP (80) standard yields the
highest positive incremental net benefits.
-------
11
V. Supporting Analyses
In addition to the RIA, EPA has conducted other analyses to determine
the impact of alternative PM standards. Most notably EPA has assembled an
air quality data base for use in analyzing potential non-attainment. There
are several factors which distinguish this more recent analysis of non-attainment
from that in the RIA. In the first place, the new data base reflects more
recent air quality than does the RIA (1979-81 versus 1975-78). More importantly,
however, the new analysis accounts for the variability of the PM^o/TSP
ratio (see RIA Section IV.D). The RIA analysis used a fixed ratio of 0.55
with limited sensitivity analysis using 0.46 as reported above. The new
analysis uses a full distribution of observed PMio/TSP ratios (with a mean
value of 0.46) to calculate a probability of PM^o non-attainment from TSP
data. The calculated probability represents the likelihood that a site
would have been classified as non-attainment if it had measured PM^Q data.
However, the new analysis was not designed to consider the effect of growth.
Moreover, unlike the RIA, the new analysis was not designed to produce cost
estimates and therefore only produces non-attainment estimates. On balance,
it is felt that these new non-attainment estimates are more reliable.
Of interest in considering the proposed range of standards is the fact
that the new analysis can be used to consider combined TSP/PM^g standards and
can also consider varying statistical forms. Table V.I summarizes the results
of analysis of the proposed ranges. As can easily be seen in the table the
probability level chosen can significantly affect the estimate of non-attainment.
For example in the case of PMjg 50/150, the p>0.20 level results in more
than a 250% increase over the p>0.95 level. When the secondary TSP standard
is factored in the shifts become much less dramatic. This results from the
fact that TSP non-attainment is determined directly from the data and
-------
12
therefore is unaffected by PM^g probability level. In the following discussion
the p>0.5 data are used to present as reasonable an estimate as possible. A
related analysis examined the difference between the current deterministic
form of the annual standard (highest mean in a two year record) and the
proposed expected value form. It was found that on average the deterministic
form yielded values 9% higher, although variations were found.
The PMiQ 50/150 standard produces significantly higher non-attainment
estimates than the PMio 65/250 (a 246% increase at p>0.50). Of the standards
considered separately, it is the TSP 70 secondary standard that produces
the highest non-attainment estimate. This standard coupled with, a PM^Q
50/150 is the most stringent of all standards considered and across all
probability levels (N.B. In combined standards, the non-attainment total
is not the sum of the two standards considered separately. Double counting
has been eleminated.) In this case the non-attainment totals are heavily
influenced by the TSP component of the standards. If on the other hand the
TSP 90 standard is coupled with a PMio 50/150 the non-attainment totals are
more heavily influenced by PMiQ.
Since the new analysis is based on more recent air quality data and
uses a different methodology than the RIA it is to be expected that the
results would also be different. Table V.2 displays the results of a
limited comparison. As noted in the full RIA, the use of the 0.55 ambient
ratio was based on the limited data available in 1981; it now appears that
a better national point estimate of the ratio is 0.46. In Table V.2, the
RIA estimates using the 0.46 ratio and before growth fall into the 0.990.50
range of the newer estimates. The use of the 0.55 ratio produces estimates
that fall in the 0.50 0.20 range. Once again it should be noted that the
two analysis differ not only in the way the PMiQ/TSP ratio was handled, but
also the age of the data bases.
-------
13
Since the 0.46 ratio is supported by more observed data and produces
results more consistent with the new analysis, it is reasonable to consider
carefully its impact. As can be seen its use produces non-attainment
estimates some 40% lower than the base RIA. In summary, the analysis of
the PMio/TSP ratio and the comparison to new data, suggest that costs and
benefits may be overstated in the RIA and that heavier weight should be given
to the results from the use of the 0.46 ratio.
-------
14
PM10 50,150
PM10 65,250 ?
TSP 70
TSP 90
PM10 50,150 and TSP
PM10 50,150 and TSP
PM10 65,250 and TSP
TSP 75.2603
TSP -,1503
Table V.I
Estimated Non-Attainment Counties
PM^/TSP1
p>0.952
107
18
448
168
70 453
90 188
90 170
301
525
p>0.502
246
71
448
168
484
275
184
301
525
p>0.202
381
106
448
168
523
389
191
301
525
estimates based on a sample of 880 counties and 1979-81 data. Annual
arithemetic means (PM^g and TSP) are calculated as the mean of all three
years. The PM^Q 24-hour value is computed as the expected value.
^Probabi lity of PM^Q non-attainment greater than the stated value. Note
that TSP non-attainment does not change by probability level.
3Current TSP standard: The annual average is computed as the maximum Annual
Geometric Mean (AGM) out of the 2 most recent years. The 24-hour value is
the highest 2nd maximum out of the 2 most recent years.
-------
15
Table V.2
Comparison of Non-Attainment Estimates
Non-Attai nment
Analysis^ 50/150 65/250
(p>0.95) 107 18
(p>0.50) 246 71
(p>0.20) 381 106
RIA Analysis2
Without Growth: 0.55 ratio 315 109
0.46 ratio 191 53
With Growth: 0.55 ratio 346 127
0.46 ratio 215 71
ifiased on 880 county sample, 1979-81 data, probabilistic assessment of
PMlO/TSP ratio, and expected AAM.
2Based on 1230 county sample, 1975-78 data, fixed PM10/TSP ratios (as
indicated), and highest AAM in two years.
------- |