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 0.99

0.50 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.

-------