REGULATORY IMPACT ANALYSIS
ON
THE NATIONAL AMBIENT AIR QUALITY STANDARDS
FOR
SULFUR OXIDES (SULFUR DIOXIDE)
Draft
March 1988
Prepared by
Air Quality Management Division
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, N.C.
-------
REGULATORY IMPACT ANALYSIS
ON
THE NATIONAL AMBIENT AIR QUALITY STANDARDS
FOR
SULFUR OXIDES (SULFUR DIOXIDE)
Draft
March 1988
Prepared by
Air Quality Management Division
Office of Air and Radiation
.S. Environmental Protection Agency
Research Triangle Park, N.C.
-------
PREFACE
A major component of the benefit analysis is visual range improvement.
The analysis procedures, results and qualifications appear in Chapter VI.
Visual range improvement results because of decreased fine particle (sulfate)
concentrations. The U.S. Environmental Protection Agency has received
comments critical of certain aspects of visibility benefit analysis.
The final 862 NAAQS RIA will incorporate information gleaned from public
comments on the draft RIA as well as the results of additional visibility
studies Wiich become available in the interim.
-------
TABLE OF CONTENTS
Section Page
Executive Summary .................. i
I. Introduction 1-1
II. Statement of Need and Consequences II-l
A. Legislative Requirements II-l
6. Nature of the Sulfur Oxides Problem II-2
C. Need for Regulatory Action 11-5
III. Alternatives Examined III-l
A. No Regulation III-l
B. Other Regulatory Approaches II1-2
C. Market Oriented Alternatives III-3
D. Regulatory Alternatives Within the Scope of
Present Legislation 111-5
IV. Cost and Environmental Impacts /IV-1
A. Introduction IV-1
B. Problem Characterization IV-1
C. Control Strategy and Cost Methodology IV-6
1. Utility Power Plants IV-6
2. Primary Copper Smelters , IV-18
3. Primary Lead Smelters IV-23
4. Industrial Boilers IV-25
5. Regional Scale Model ing IV-27
D. Results IV-32
1. National IV-32
2. Utilities IV-37
3. Industrial Boilers IV-42
4. Environmental Results IV-44
-------
Page
References
V. Economic Impact
A. Introduction V-l
B. Utilities V-l
References
VI. Benefit Analysis Estimates
A. Introduction Vl-1
B. Methodology VI-2
C. Air Quality Data Vl-8
1. 504 Air Quality Assessment VI-9
2. PM Air Quality Assessment Vi-H
3. SO? Air Quality: Assessment of Four Point Sources... VI-12
D. Study Selection, Application, Qualifications and
Plausibility Checks VI-14
1. SQ4 Benefits Vl-15
2. PM Benefits VI-23
3. SO? Benefits VI-32
E. Estimates VI-42
1. Benefits for 31 States VI-42
2. S02 Benefits for 4-Point Sources VI-44
3. Cost of Delay , VI-44
F. Findings VI-44
1. Estimated 504 and PM Related Benefits are Larger
Than S02 Related Benefits VI-44
2. Direct S02 Welfare Related Benefits for Alternatives
Considered Appear Greater Than S02 Health Related
Benefits VI-47
3. Air Quality Improvement Delays Mean Foregone
Benefits VI-47
List of References VI-48
-------
VII. Benefit-Cost Analysis
A. Introduction . VII-1
t B. Economic Efficiency VI1-2
C. .Methodology-Incremental Net Benefit Analysis VI1-3
D. Limitations, of the Methodology VI1-4
1. Feasibility Test VII-4
2. Cost-Effectiveness VII-5
3. Distributional Impacts ...-. VII-5
E. Scope of the Benefit-Cost Analysis VII-7
F. Measurement of Benefits and Costs: Conceptual Issues ... VI1-8
G. Estimates of Benefits and Costs VII-9
H. Limitations and Assumptions VII-13
I. Net Benefits VI1-19
J. Limitations to the Analysis VII-2b
K. Qualifications and Findings VI1-26
1. qualifications VII-26
'2. Findings VII-28
List of References VII-33
VIII. Summary of Rationale for Choosing the Proposed Action
IX. Statutory Authority
Appendix A
1. Introduction A-l
2. S04 Benefits A-l
3. PM Benefits A-8
4. S02 Benefits A-9
•v
List of References A-28
-------
Appendix 8
1. Introduction 8-1
2. Conceptual Framework for Valuing Risk Reductions 6-1
3. Inference of Risk From Health Studies 8-3
4. Calculation Procedures B-6
5. Results B-7
List of References 3-10
-------
LIST OF TABLES
Table Page
III.D.I S02 Alternatives Analyzed III-6
IV.B.I Emissions from Major Source Categories of
S02 IV-3
IV.C.I Primary Copper Smelters IV-21
IV.C.2 Summary of Potential Sources of Error and
Uncertainty , IV-28
IV.C.3 Comparison of Model/Matrices Used in Regional
Analysis of Air Quality IV-29
IV.D.I Total Estimated National Cost Summary IV-35
IV.D.2 Estimated National Cost Summary by Source
Category IV-36
IV.0.3 Total Estimated Reductions by Source Category , IV-38
IV.D.4 Total Estimated Utility Costs IV-39
IV.D.5 Estimated Annual Utility S02 Emissions IV-40
IV.D.6 Estimated Changes in Scrubber Capacity IV-41
IV.D.7 Industrial Boiler Annualized Costs and Emission
Reductions IV-43
IV.D.8 1980 Baseline Sulfate Estimates IV-45
IV.D.9 1995 Base Case Sulfate Estimates IV-46
IV.D.10 1995 Sulfate Estimates
Strict Interpretation of Current NAAQS IV-47
IV.0.11 199b Sulfate Estimates
0.5 ppm 1-hour Alternative IV-48
IV.D.12 1995 Sulfate Estimates
0.2b ppm Alternative NAAQS IV-49
IV.D.13 Estimated 1980 Baseline Visibility and
Estimated 1995 Baseline Visibility IV-51
f
IV.0.14 1995 Estimated Visibility-
Strict Interpretation of Current NAAQS IV-52
-------
Table Page _
IV.0.15 1995 Estimated Visibility
0.5 ppm 1-hour Alternative NAAQS IV-53
IV.0.16 1995 Estimated Visibility
0.25 ppm Alternative NAAQS IV-54
IV.0.17 Estimated Scrubber Sludge Production IV-56
V.B.I Absolute Change in Electricity Rates Based
on Annualized Costs in 1995 V-5
V.B.2 Percent Change in Electricity Rates Based on
Annualized Costs in 1995 V-6
V.B.3 Utility Industry Revenues and Capital
Expenditures V-8
V.B.4 Total Estimated Utility Cost Increases V-8
V.B.5 Coal Production and Transportation - 1995 V-10
V.B.6 Coal Mine Employment - 1995 V-ll
VI.B.I Alternative SOg NAAQS Potential Benefit Categories VI-4
VI.C.I Power Plant Characteristics VI-13
VI.E.I Thirty-One State Benefit Assessment for
Alternative Standards VI-45
VI.E.2 4-Point Source S02 Benefit Assessment VI-47
VI.E.3 Loss .of Benefits Oue to Air Quality Improvement Delays .. VI-48
VII.G.I 31 Eastern State Benefit Assessment for Alternative
Standards - 10% VII-10
VII.G.2 31 Eastern State Benefit Assessment for Alternative
Standards - 5% VII-11
VII.G.3 31 Eastern State Benefit Assessment for Alternative
Standards - 2% VII-12
VII.G.4 31 Eastern State Cost Assessment for Alternative
Standards - 10% VII-14
VII.G.5 31 Eastern State Cost Assessment for Alternative
Standards - 5% VII-15
VII.G.6 31 Eastern State Cost Assessment for Alternative
Standards - 2% VII-16
-------
Table Paye
VII. H.I Limitations and Assumptions of the 31 Eastern State
Benefit Analysis ........................................ VII-17
VII. H. 2 Limitations and Assumptions of the 31 Eastern State
Cost Analysis ........................................... VII-18
VII. I.I 31 Eastern State Net Benefit Assessment for Alternative
Standards - 10% ......................................... VI I -20
VII. 1. 2 31 Eastern State Net Benefit Assessment for Alternative
Standards - 5% .......................................... VI 1-21
VII. 1. 3 31 Eastern State Net Benefit Assessment for Alternative
Standards - 2% .......................................... VII-22
VII. I. 4 Implicit Valuation of the SQ$ Mortality Risk Reduction
Coef f i ci ent ............................................. VI I -24
VII. K.I Ordering of Preferred Standards ......................... VII-30
A.I Variables Used in Analysis of Results From Contingent
Valuation Studies ....................................... A-4
A. 2 Data Used in Analysis of Results From Contingent
Valuation Studies ....................................... A-b
A. 3 Regression Estimates of Equation A.I for all Contingent
Valuation Studies with Dummy Variables for Differences
in Studies .............................................. A-7
A. 4 Input Data that are Fixed in Time and Space ............. A-10
A. 5 Data Sources ............................................ A-13
A. 6 1982 Production Within Range of the Power Plant ......... A-6
A. 7 1982 Price Per Bushel by State .......................... A-15
A. 8 Demand Equations for Which S02 is a Significant
Explanatory Variable .................................... A-24
8.1 Mean Values of Distributions in the Urban/Rural
Analysis ................................................ B-5
B.2 31 Eastern State S04 Mortality Risk Reduction Benefit
Estimates - Valuation Coefficient of $420,000 ........... 8-8
B.3 31 Eastern State 504 Mortality Risk Reduction Benefit
Estimates - Valuation Coefficient of $7,300,000 ......... 8-9
-------
Executive Summary
•Regulatory Impact Analysis on
The National Ambient Air Quality Standards
for Sulfur Dioxide
I. Background
The sulfur dioxide (S02) Regulatory Impact Analysis (RIA) was
prepared to fulfill the requirements of Executive Order 12291 (E.G. 12291).
The RIA attempts to quantify and inform the public of the costs, benefits
and economic impacts of various levels of current and revised SO^ 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. 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.
II. 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.
Sulfur dioxide (SOg) is a reactive gas. that is quite soluble in water.
It is emitted principally from the combustion of sulfur bearing fuels and
the processing of sulfur bearing ores. In the U.S., utility power plants,
non-ferrous smelters, and industrial boilers are the major sources of S02.
At elevated concentrations, SQ2 can adversely effect human health, vegetation,
materials, economic values, and personal comfort and well-being. S02 and
its transformation products are also major contributors to acidic deposition
and regional visibility degradation.
-------
The need for regulatory action arises from the failure of the market
system to deal effectively with S0£. Most emittors treat the atmosphere as
a free good and dispose of unwanted by-products by venting them to the
outdoor air. In the atmosphere S02 can cause real costs to be
incurred by others. This is generally known as a negative externality.
Ill. Alternati ves 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
Alternatives Analyzed
Annual Arithmetic Mean 24-Hour Observed 3-Hour Observed 1-Hour Expected
(Primary) 2nd Maxima 2nd Maxima Exceedance Rate
(Primary) (Secondary) of 1
1) 0.03 ppm 0.14 ppm 0.5 ppm
2) - - - 0.5 ppm
3) - - - 0.25 ppm
Item 1 above is the current NAAQS, while items 2 & 3 correspond to the
range of alternative 1-hour standards initially recommended by the staff
and CASAC in 1982.
-------
ill
IV. Cost and Environmental Impacts
Since 502 "is primarily a point source pollutant the cost analyses have
focused on the largest point source categories. The categories analyzed
were utility power plants, primary non-ferrous smelters, and industrial
boilers. Together these source categories account for approximately 85%
of national SOg emissions. In designing the RIA principal consideration
was given to the most likely NAAQS options and in particular the 1-hour
options (See Table I above). A number of studies indicated that 1-hour
concentrations in the range being considered (0.5-0.25 ppm) occurred
primarily around point sources. These and other studies also indicated
that the peak 1-hour concentrations usually, but not always, resulted from
the impact of single sources. In other words, for the 1-hour averaging
period, source interaction did not appear to be a major problem. As a
result of this review it was decided to model point sources individually
and to omit explicit consideration of source interaction. However, it
should be noted that background factors were used where appropriate. The
following discussion briefly summarizes the approach taken in each of the
major source categories. In each case the analysis attempted to replicate
as closely as possible the techniques being used in the SIP program.
A. Utilities
Control Strategy: Utilities make up approximately 67% of all
emissions in the U.S. For this analysis a new emissions data base was
assembled and cross checked with industry data bases. Due to data problems
plants accounting for about 16% of utility emissions could not be analyzed.
Following the establishment of a data base, the next step in the analysis
was the determination of air quality impacts around each plant. It was
beyond the 'scope of the RIA to complete a detailed dispersion modeling of
-------
i v
each plant. Therefore a screening methodology which assumed worst case
meteorology and calculated impacts as a function of stack height, buoyancy
flux and emission rates was developed. The screen considers each plant
individually and from limited comparison to detailed dispersion modeling is
believed to slightly overpredict maximum impacts. Following determination
of the air quality impact of each plant, appropriate emission limits for
each of the alternatives listed in Table 1 were calculated. As discussed
in Section IV of the RIA, the emission limits were designed to account for
background, fuel sulfur variability, and load condition. As noted in
Section IV, the assumptions regarding sulfur variability may represent a
conservative bias and result in higher estimated costs when compared to actual
practice at some plants. Conversely, the inability to account for local scale
terrain effects in this national analysis results in an unknown, but downward
bias in the estimated costs. It should be noted, however, that emission
limits were calculated only for those plants exceeding the standard being
analyzed. Potential emission increases were not analyzed.
Cost Analysis: The cost analysis was accomplished using the ICF,
Inc. Coal and Electric Utilities Model (CEUM). This model attempts to
define coal supply and coal demand and generates an equilibrium solution
through standard linear programming techniques. Compliance with emission
limits can be achieved either through switches to lower sulfur fuels -or
through scrubbing.
B. Copper Smelters
Control Strategy: Primary copper smelters emit approximately 6% of
total U.S. emissions of S02. Although most major point sources now have
emission limits established through dispersion modeling, EPA has accepted
the use of multi-point rollback (MPR) for copper smelters. MPR uses a
di stribution" of. monitored air quality data to determine how much the
-------
emissions distribution needs to be "rolled back." The result is not a
single emissions limit, but rather an allowed emissions distribution.
Eleven of the thirteen copper smelters now operating were analyzed using
MPR. Of the remaining two, one (Phelps-Dodge, Hidalgo) is well controlled
already. For the other (ASARCO, El Paso) sufficient data were not available
to complete an MPR demonstration.
Cost Analysis: In the copper smelter cost .analysis, estimates of
control system costs were made based on smelter throughput. Where sufficient
control could be achieved scrubbing systems were used. Where additional
control was needed process modifications and acid plants were employed. It
should be noted that the process changes result in a cost savings leading
to lower annualized costs.
It should be noted that since this analysis was originally conducted in
the early 1980's, the primary copper smelting industry has undergone significant
change. A number of smelters, including those with the greatest emissions and
least controls (e.g., Douglas), have since ceased operations. Since the
smelters that are still operating tend to be the ones that have better controls
and because there are fewer smelters operating, the costs shown here for each
of the alternatives have been reduced.
C. Lead Smelters
Cost Strategy: Although lead smelters account for only 0.3% of
total U.S. emissions, their local air quality impacts can be large. The KIA
addressed the three smelters still in general operation. Standard EPA
dispersion models were used to estimate the local impacts of each smelter.
Cost Analysis: The lead smelter cost analysis was similar in
methodology to the copper smelter analysis. However, process changes were
not appropriate in the case of lead smelters. The control options evaluated
-------
VI
were limited to scrubbing of weak stream gases and acid plants for strong
stream gases.
D. Industrial Boilers
Control Strategy: The industrial boilers account for about 11% of
total U.S. emissions of SO^ and were analyzed in a manner similar to the
utilities. Specifically, a screening methodology based on standard Gaussian
dispersion algorithms was developed and used to estimate emission limits
for the standards under consideration.
Cost Analysis: Since the control options for industrial boilers
are essentially limited to fuel switches it was decided to analyze their
costs in the CEUM model simultaneously with the utilities.
E. Results
Table 2 displays the total estimated costs associated with the
three alternatives analyzed. As might be anticipated given their dominance
Table 2
Estimated National Cost Summary by Source Category
($ Millions)1
Current NAAQS 1-Hour 0.5 pprn 1-Hour 0.25 ppm
Source Catgegory
Utilities
Copper Smelters
Lead Smelters
Industrial Boilers
Total
Capital
$0-400
$200
$150
NA2
$350-750
Annual
0-700
$150
$45
0-$200
$200-1,100
Capital
$2,200
460
170
NA2
$2,800
Annual
$1,800
110
50
200
$2,200
Capital
15,900
750
170
NA2
17,000
Annual
b,OOU
50
50
300
5,400
1-All costs are calculated in 1984 dollars and do not include the cost of:
1) pre-1980 controls, or 2) new source controls tied to meeting NSPS,
NSR, or PSD requirements.
2Control options were limited to fuel switches; therefore no capital costs
were estimated.
-------
vii
in emissions, the utilities account for the majority of the costs. The
range of utility costs associated with the current NAAQS arises from
uncertainties regarding compliance determination and is discussed in Section
IV. The increase in utility capital costs with the more stringent alternatives
arises from the increased reliance on scrubbers. The copper and lead
smelters show substantial costs to meet the current NAAQS. Table 3 shows
the emission reductions associated with these costs.
Table 3
Total Estimated Emission Reductions by Source Category
(Millions TRY)
Utilities
Copper Smelters
Lead Smelters
Industrial Boilers
Total
F. Environmental
Current NAAQS
0-2.4
1.4
0.2
0-0.2
1.6-4.2
Impacts
0.5 ppm
4.4
1.6
0.2
0.2
6.4
0.25 ppm
9.0
1.7
0.2
0.3
11.2
In addition to analyzing local scale S02 air quality improvements
and emissions reductions, the RIA also addresses regional scale air quality
improvements (804). A variety of regional scale air quality models/transfer
matrices were evaluated and two selected for use in the analysis. The two
models/matrices used were ASTRAP and MONTE CARLO. It should be understood
that although these models provide some insight into the magnitude and
nature of regional-scale $04 air quality changes, they are uncertain and
the subject of some controversey. In particular there is considerable
debate in the scientific community as to whether the atmospheric processes
-------
VI 1 1
being modeled are linear. The models being used are linear. If non-linear
processes were found, the estimates provided below would be high. Despite
these limitations the models used are felt to provide the best estimates
given the current state of scientific information. Table 4 below reports
the 1995 base case 864 derived from each model and the percentage change
predicted for each of the alternatives examined.
-------
Table 4
1995 Sulfate Estimates
Alternative Standards
Base0Case
My/
I
'm° S04
A
1 Current
% Change from Base Case
NAAQS
0.5
ppm NAAQS
0.2'5
ppm NAAQS '
Strict Interpretation
STATE
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois--'
Indiana
Iowa
Kentucky
Louisiana
Maine ;
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsyl vania
Rhode Island
South Carolina
Tennessee
Vermont
Vi rginia
West Virginia
Wisconsin
ASTRAP
b.2
3.1
8.2
11.5
4.0
b.7
b.b
8.1
2.7
8.5
2.6
3.6
10.9
6.5
4.5
1.4
3.7
4.0
5.6
10.5
7.5
7.8
9.6
11.3
7.1
6.8
b.9
5.1
9.3
11.5
3.1
MONTE CARLO
10.9
7.3
8.6
9.9
4.5
9.4
7.7
10.0
4.1
12.5
6.1
5.8
9.9
8.6
4.5
2.9
7.9
7.2
7.4
10.3
10.3
1U.9
11.2
11.7
8.6
10.8
10.9
7.4
11.8
14.5
3.9
ASTRAP
-12
- 6
- 7
- 7
-10
-11
-16
-14
- 7
-15
- 8
- 6
- 8
- 6
- 9
- 0
-11
-13
- 5
- 7
- 8
-10
-12
- 9
- 7
- 9
-14
- 6
-11
-12
- 6
MONTE CARLO
-10
-10
- 6
- 6
- 7
-10
-10
-11
- 7
-12
- 7
- 5
- 6
- 6
- 7
- 3
- 9
-11
- 5
- 8
- 8
- 8
-11
- 9
- 6
- 8
-13
- 5
- 8
-10
- 3
ASTRAP
-17
-10
'-15
-17
-18
-18
-24
-25
-11
-26
- 8
-11
-19
-14
-16
- 0
-16
-20
-12
-16
-16
-19
-23
-19
-15
-16
-23
-10
-22
-23
- 6
MONTE CARLO
-17
-14
-12
-14
-13
-16
-16
-20
- 7
-22
- 8
-10
-14
-12
-11
- 3
-13
-17
-11
-15
-15
-17
-19
-17
-12
-16
-21
-11
-18
-19
- 5
ASTRAP
-35
-16
-32
-38
-33
-39
-36
-42
-15
-45
-15
-22
-40
-29
-31
- 7
-27
-28
-25
-34
-33
-41
-43
-40
-32
-38
-41
-24
-43
-45
-19
MONTE CARLO
-37
-30
-26
-32
-27
-36
-32
-38
-20
-43
-20
-21
-32
-26
-22
-10
-28
-36
-22
-30
-30
-37
-38
-36
-26
-36
-43
-22
-38
-42
-13
-------
X
V. Economic Impacts
The economic impacts analyzed in this RIA are limited to those arising
from the utility costs. This limitation arises from several factors. In
the first place an industry by industry review of the industrial boiler
costs indicated that no one industry was substantially impacted. Therefore
a detailed economic impact analysis was felt to be unwarranted. Secondly,
due to current industry economic conditions and uncertainties regarding the
degree and timing of compliance arising out of Section 119 of the Clean Air
Act, an economic impact analysis was not performed for the smelters.
There are several assumptions in the utility cost model (ICF's CEUM)
which constrain the economic analysis. Specifically, CEUM assumes a
perfectly inelastic demand for electricity which is exogenously specified.
This assumption is likely to result in an overstatement of price increases
and an understatement of changes in output. In addition, it is assumed in
CEUM that the industry capital supply is perfectly elastic. This assumption
may be doubtful under the 0.25 ppm alternative where the incremental capital
requirements are significant. With these limitations the economic impacts
analyzed are discussed below.
Utility Rate Impacts: The estimated rate changes were, on average, relatively
small for both the current NAAQS and the 0.5 ppm alternative. They were
0.5% and 1.0% respectively. On the other hand, the 0.25 ppm alternative
resulted in an average increase of 3.0%, with eight states showing increases
>. 5%.
Comparison to Total Industry Costs and Revenues: A comparison of the costs
estimated in this analysis was made to actual industry capital costs and
revenues (1980-82). Compared to actual industry capital costs for 1981,
-------
xi
the current NAAQS would result in a 0-1% increase while the 0.5 ppm NAAOS
results in an approximate 4.8% increase. The 0.25 ppm alternative which
required substantial scrubbing is estimated to result in a 34% increase.
Coal Production and Mine Employment: As might be expected all of the
alternatives result in shifts in coal production and mining employment from
areas with high sulfur coal to areas with lower sulfur coal. For the
current NAAQS the shift is relatively insignificant. However, for the
0.5 ppm and 0.25 ppm alternatives the shifts are more significant.
VI. Benefit Analysis
The benefits analyzed in this RIA represent the improvement in
society's well-being as a result of improved air quality. These benefits
do not represent the total improvement that results from going from zero
control to full compliance with the alternative standards. Rather, they
represent the incremental improvement in going from a baseline reflecting
current operating practice with respect to State Implementation Plans,
New Source Performance Standards, New Source Review, and similar control
requirements to full complaince with the alternative standards.
Economic benefits should be estimated using data, assumptions, and
modeling techniques developed specifically for the analytic objective.
In the case of this benefit analysis the ideal approach is precluded by
project structure, time, and resource constraints. Therefore, estimates
are based upon existing studies which address some aspects of the health
or welfare implications of ambient sulfur dioxide, sulfates, or particulate
matter. These existing studies were screened on the basis of analytic
quality and potential for extrapolation of estimates for benefit analysis.
-------
xi i
As can be seen in Table 5 only a limited subset of benefit categories
is estimated in this RIA. In addition, benefit coverage is often incomplete
for those categories that are covered. For example, the SOg agricultural
effects analysis only covered three crops (soybeans, wheat, and oats).
The coverage of benefits is also limited by the assumption of perfectly
price inelastic demand as mentioned in Section V. Finally, the benefit
analysis only covers the 31 eastern states and is not a nationwide analysis.
A. Air Quality Data
There are three types of air quality data used for the benefit analysis:
$04 data used for visual range improvement and mortality risk reduction
calculations, PM air quality data used for morbidity risk reduction and
soiling reduction calculations, and S02 data used to estimate mortality
and morbidity risk reductions, materials damage reduction, and increased
agricultural yield benefits. The geographic area for the sr>2 ambient air
quality data is the region modeled around four point sources. Specifically,
in the S02 benefit assessment the air quality changes result from simulated
compliance with the 0.5 ppm 1-hour alternative S02 NAAQS by four utility
power plants. The benefits are then extrapolated to the 31 eastern states
and the other two S02 NAAQS that are examined.
1. 504 Air Quality Assessment
The effect of S02 emission reductions on sulfates was modeled on a
regional scale. Modeling of regional scale transport, dispersion, chemical
transformation and removal of sulfur oxides is quite complex and involves
a number of uncertainties and assumptions. For instance, the two models
used in the transfer matrix--ASTRAP and MONTE CARLO--are linear. It is
unknown whether the processes being modeled are in fact linear. The
-------
XI 11
matrix being used also treats emission changes as if they occur uniformly
over an entire state. The lack of source to source emission variation
could be a serious limitation.
2. PM Air Quality Assessment
The 504 changes for each state are translated into TSP changes by
multiplying the sulfate change by 1.4 for the low estimate, 1.5 for the
middle estimate and 1.6 for the high estimate. No estimates of direct PM
emission reductions were calculated for this RIA.
3. S02 Air Quality Assessment
Four utility power plants were modeled to examine the direct benefits
of S02 emission reductions. The benefits based on the air quality changes
for these four point sources were extrapolated to the 31 eastern states.
There are several important qualifications to this air quality
assessment. Benefits may be biased downward because the outer most
receptor ring is only 20 km from the source. Small air quality changes
may occur beyond this point. For the four point source analysis, one year
is assumed to represent the air quality changes around those plants for
the period 1990-2000. No variations in meteorological factors are considered,
B. Study, Selection and Application
1. SC>4-Visual Range
Benefit estimation of visibility requires the establishment of a
hypothetical market or identification of complementarity between visual
range and an existing market. Five contingent valuation studies were
chosen for this analysis. They are: Brookshire et al. (1980), Rae
(1983), Rowe et al. (1980), Loehman et al. (1981), and Tolley et al.
(1986). These studies cover ten U.S. cities and only cover user values.
-------
xi v
Option and existence values are not quantified.
The application of these studies involves estimating visual range
changes for each of the 31 states. This is done using the 804 air quality
data discussed earlier. A visual range improvement valuation coefficient
is derived from the five selected studies and applied to the predicted
changes in visual range. The average annual household values, from each
of the studies, for each hypothesized change in visual range were compiled
to estimate a benefit function equation.
2. 504 - Mortality Risk Reduction
No estimates are given for $64 benefits associated with reduced
mortality risk. This is due to uncertainties associated with the available
studies and the lack of biological plausibility. Nevertheless, the data
clearly suggest a risk at current ambient levels, and it is probable that
reducing S02 emissions would reduce episodic peak acid aerosol exposures
and thus reduce the risk. To provide some idea of the nature of the risk
reduction three epidemiology studies (Chappie and Lave-1982, Evans et al .-
1984a, and Lipfert-1977) and responses from three experts were used to
produce hypothetical estimates of reduced mortality benefits.
3. PM - Chronic Morbidity
The longitudinal study by Ferris et al. (1962, 1973, and 1976) was
selected to estimate PM benefits of reduced chronic morbidity. The health
end point used by Ferris et al. is chronic respiratory disease. A
relationship is established between chronic respiratory disease and work
loss or restricted activity days. The valuation of damages included an
assessment of lost productivity and increased medical care charges.
Benefits are not calculated for reductions in pain and suffering.
-------
XV
4. PM - Acute Morbidity
Ostro (1986) was selected to estimate PM benefits of reduced acute
morbidity. Three measures of morbidity were used by Ostro: work loss
days, restricted activity days, and respiratory-related restricted activity
days. Ostro estimates a concentration response function that used a
lagged two-week average PM measure. The study also uses fine particles
as the measure of exposure. Fine particles are probably the most
appropriate indicator of PM exposure since small particles are most
damaging to health. Benefits were not calculated for reduction in pain
and suffering.
5. PM - Household Soiling
The Mathtech study (1982) on 24 Metropolitan Statistical Areas was
chosen to calculate household soiling benefits. The household soiling model
analyzes prices, socio-demographic characteristics, and environmental
variables at the county level. Benefits are calculated by changing the
value of the environmental variable to reflect alternative standards.
6. S02 - Mortality Risk
The Martin and Bradley study (1960) is used to calculate the
upper-bound mortality benefits. The lower-bound and mid-point estimates
of zero are based on Mazumder et al. (1981). Benefit calculations are
based on averages of predicted daily S02 concentrations averaged across
all receptors for the four power plants. Benefits are then extrapolated to
the 31 eastern states.
7. SOg - Morbidity
Clinical studies and the macroepidemiology study by Graves et al.
(1980) were selected to estimate S02 benefits of reduced morbidity. For
-------
XVI
the clinical studies an estimate of $50 per hour of symptom reduction is
used as the maximum. A minimum estimate of zero is applied with a mid-
point estimate of $25.
8. S02 - Agriculture
Benefits from increased yields of wheat, soybeans, and oats are
analyzed. Study selection is limited to yield effects. The major study
used for soybeans is Sprugel et al. (1980). For oats and wheat, the
principle source is a series of reports by Guderian and Stratnann (1962
and 1968). Average air quality for each receptor area for the growing
season is applied. The maximum benefit estimate applies the dose response
function of the study varietal to all of that crop. Zero is used as the
minimum estimate. The mid-point estimate is one-half the maximum estimate.
9. S02 - Materials Damage
The Mathtech (1982) analysis is used to calculate residential benefits
in this category. S02 air quality changes are evaluated using the
estimated coefficient contained in the household model. Economic demand
and supply curves are estimated to reflect the effects of SO;? on residential
materials damage.
C. Benefit Estimates for 31 States
Ranges of benefit estimates are presented in Table 6 to reflect
uncertainty concerning air quality and economic valuation.
D. Findings
Three major findings emerge from this benefit analysis:
1. 504 and PM benefits are significantly larger than S02 benefits
2. S02 welfare benefits are greater than S02 health benefits
3. Air quality improvement delays mean foregone benefits
-------
XV11
VII. BENEFIT-COST ANALYSIS
The benefit-cost analysis is presented to provide a consistent
framework for evaluating the economic effects of alternative regulatory
policies. An air quality regulation is efficient in an economic sense,
if as a result in its implementation, at least one individuals well-being
is improved without reducing the well-being of any other member of society.
In order to evaluate the relative efficiency of alternatives an analysis
of incremental benefits and costs is needed. Any alternative S02 NAAOS
that produces positive net incremental benefits will provide a more
economically efficient allocation of resources than would occur under the
baseline air quality scenario. The alternative that results in the
largest positive incremental net benefits will represent the most efficient
allocation of resources among those alternatives considered.
A. Limitations of the Methodology
j
The incremental net benefit analysis does not include those costs
and benefits associated with achieving and maintaining the baseline level
of air quality. Therefore, it is possible that the total costs could
exceed the total benefits of an alternative even if incremental net
benefits are positive. The distributional or equity effects have also
not been analyzed in this RIA. The focus of this benefit-cost analysis
is on the relative efficiency of a limited selection of alternatives and
not the identification of the most efficient of all feasible and cost-
effective S02 NAAQS.
B. Measurement of Benefits and Costs
Where possible, willingness-to-pay is the measure of benefits that
•' o • . •
is used in this analysis. Mortality risk valuations are obtained from
• ; ' s ,'',-'' -.. f • "'••> : ,S "' ,'>."'. , -' .
-------
xviii
occupational risk studies.
The cost analysis also requires the identification of society's
wil1ingness-to-pay for those directly foregone consumption opportunities
that would otherwise be available. Lack of consideration of direct price
elasticity effects and the overlap between control requirements for
different regulatory programs are two areas where estimated costs depart
from the conceptually correct willingness-to-pay measure.
C. Estimates of Benefits and Costs
Benefits and costs are calculated using three real interest rates--
10%, 5%, and 2%. The interest rate of 10% is used as directed by the
Office of Management and Budget in their guidance for implementing
Executive Order 12291. Alternative real rates of 5% and 2% are offered
for purposes of comparison. Tables 7, 8, and 9 report benefit estimates
for the 31-state area using 10%, 5%, and 2% real interest rates respectively
Tables 10, 11, and 12 present estimates of incremental costs. These costs
are presented as mid-point estimates although uncertainty is no less
critical or non-existent for the cost estimates as compared to the benefit
estimates.
D. Net Benefits
The estimated net benefits are presented in Tables 13, 14, and 15.
Of the 27 combinations of interest rates, benefit assumptions, and
alternative standards 25 have positive net incremental benefits. An
implicit valuation analysis was conducted to estimate how big the 504
mortality coefficient would have to be for incremenal benefits to equal
incremental costs for the low estimate 0.25 ppm standards using 10% and
5% discount rates. The results of th's analysis are presented in Table 16.
-------
xix
E. Findings
Four findings can be drawn from the information presented in the
benefit-cost analysis:
1. There is large uncertainty surrounding the analytic process
2. Within the limits of this analysis, more S02 control is supported
on grounds of economic efficiency -
3. Within the limitations of this analysis, the degree of additional
control warranted remains ambiguous.
4. There could be significant overlap between the PM and S0j> NAAQS
and other regulatory programs.
-------
XX
Table 5
Alternative S02 NAAQS Potential Benefit Categories
Health Effects
- Mortality Due to Chronic Exposure
- Mortality Due to Acute Exposure
- Morbidity Due to Chronic Exposure
- Morbidity Due to Acute Exposure
Soiling and Materials Damage
- Residential Facilities
- Commercial and Industrial Facilities
- Governmental and Institutional Facilities
Climate and Visibility Effects
- Local Visibility
Non-Local Visibility
- Climate
- Visibility at Parks
- Transportation Safety
Non-Human Biological Effects
- Agriculture
- Forestry
- Fishing
- Ecosystem
1. Estimated but coverage limited
2. Not estimated; benefits possible
3. Not estimated; benefits unlikely
* Benefits for this category are not estimated in the main body of this chapter.
However, ranges of estimates for this category are provided in Appendix B.
Also, an implicit valuation of mortality risk is presented in Chapter VIl!
:t SO?
2
1
2
1
1
2
2
3
3
2
3
3
1
2
2
2
SQ4
2*
2
2
2
2
2
2
1
2
2
2
2
2
2
2
2
Other
Participate
Matter
2
2
1
1
1
2
2
2
2
2
2
2
2
2
2
2
-------
Table 6. 31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS*
(DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
S02 Health
S02 Welfare
Material Damage
Agriculture
504 Visibility Benefits
PM Morbidity & Soiling
Total
Current Standards
(Strict Interpretation)
Low Middle
>0 .002
.029 .142
0 .004
2.1 2.5
2.1 2.4
4.2 5.1
High
.225
.262
.008
3.1
2.7
6.3
0.5 ppm
1-hour standard
Low Middle
0 .004
.052 .254
0 .007
3.7 4.6 5
3.9 4.3 4
7.7 9.2 11
0.25 ppm
1-hour standard
High
.402
.468
.014
.7
.8
.4
Low
0
Middle
.008
.104 .507
0 .014
8.3
7.6
16.0
9.8
8.8
19.1
High
,.805
.935
.028
11.8
10.2
23.8
1 The assessment only includes a subset of related benefits.
2 The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .2728.
-------
Table 7. 31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS1
(10% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
S02 Health
S02 Welfare
Material Damage
Agriculture
504 Visibility Benefits
PM Morbidity & Soiling
Total
Current Standards
(Strict Interpretation)
Low
0
.029
0
2.1
2.1
4.23
Middle
.002
.142
.004
2.5
2.4
5.05
High
.225
.262
.008
3.1
2.7
6.30
0.5 ppm
1-hour standard
Low Middle
0 .004
.052 .254
0 .007
3.7 4.6 5
3.9 4.3 4
7.65 9.17 11
High
.402
.468
.014
.7
.8
.38
0.25 ppm
1-hour standard
Low
0
.104
0
8.3
7.6
16.0
Middle
.008
.507
.014
9.8
8.8
19.13
High
.805
.935
.028
ll.fi
10.2
23.77
xxii
1 The assessment only includes a subset of related benefits.
2 The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .2728.
-------
Table 8. 31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS1
(5% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
S02 Health
S02 Welfare
Material Damage
Agriculture
SO'4 Visibility Benefits
PM Morbidity & Soiling
Total
Current Standards
(Strict Interpretation)
Low
0
.050
0
3.3
3.4
6.75
Middle
.004
.242
.007
4.0
3.8
8.05
High
.381
.447
.014
5.0
4.3
10.14
0.5 ppm
1-hour standard
Low Middle
0 .007
.090 .433
0, .012
6.0 7.4 9
6.3 7.0 7
12.39 14.85 18
0.25
1-hour
High
.681
.798
.024
.2
.8
.50
Low
0
.180
0
13.5
12.2
25.88
ppm
standard
Middle
•
•
15.
14.
31.
014
866
024
9
2
0
Hi
1.
gh
36
1.6
.049
19.
16.
38.
1
4
51
The assessment only includes a subset of related benefits.
The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 5 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .1613.
-------
Table 9. 31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS*
(2% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
S02 Health
S02 Welfare
Material Damage
Agriculture
S04 Visibility Benefits
PM Morbidity & Soiling
Total
Current Standards
(Strict Interpretation)
Low
0
.071
0
4.5 5
4.6 5
9.17 11
Middle
.006
.342
.01
.5
.2
.06
High
.535
.630
.019
6.8
5.9
13.88
0.5 ppm
1-hour standard
Low Middle
0 .01
.126 .611
0 .017
8.2 10.0 12
8.6 9.5 10
16.93 20.14 25
0.25 ppm
1-hour standard
High
.956
1.1
.034
.6
.6
.29
Low
0
.253
0
18.4
16.7
35.35
Middle
.02
1.2
.034
21.7
19.4
42.35
High
1.9
2.3
.069
26.0
22.3
52.57
AT XX
1 The assessment only includes a subset of related benefits.
2 The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 2 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .1297.
-------
Table 10. 31 EASTERN STATE COST ASSESSMENT FOR ALTERNATIVE STANDARDS
(10% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)*
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
0.5 ppm
1-hour standard
0.25 ppm
1-hour standard
Costs
3.3
7.0
18.0
The discounted present value of an eleven year stream of costs occurring from Janury 1, 1990 to December 31,
2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of costs for 1990 to
2000, multiply by .2728. These annualized costs are not directly comparable to those presented in Chapter IV
due to the use of different interest rates.
-------
Table 11. 31 EASTERN STATE COST ASSESSMENT FOR ALTERNATIVE STANDARDS
(5% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
0.5 ppm
1-hour standard
0.25 ppm
1-hour standard
Costs
5.6
11.2
27.3
The discounted present value of an eleven year stream of costs occurring from Janury 1, 1990 to December 31,
2000 using a real discount rate of 5 percent in 1984. To convert to an annualized stream of costs for 1990 to
2000, multiply by .1613. These annualized costs are not directly comparable to those presented in Chapter IV
due to the use of different interest rates.
X
H-
-------
Table 12. 31 EASTERN STATE COST ASSESSMENT FOR ALTERNATIVE STANDARDS
(2% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
0.5 ppm
1-hour standard
0.25 ppm
1-hour standard
Costs
6.2
13.9
31.6
H
<
The discounted present value of an eleven year stream of costs occurring from Janury 1, 1990 to December 31,
2000 using a real discount rate of 2 percent in 1984. To convert to an annualized stream of costs for 1990 to
2000, multiply by .1297. These annualized costs are not directly comparable to those presented in Chapter IV
due to the use of different interest rates.
-------
Table 13. 31 Eastern State Net Benefit Assessment for Alternative Standards
(10% Discounted Present Value in Billions of January 1984 Dollars)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
Low Middle High
0.5 PPM
1-Hour Standard
Low Middle High
0.25 PPM
1-Hour Standard
Low Middle High
Net Benefits .9 1.8 3.0 .7 2.2 4.4 -2.0 1.1 5.8
The discounted present value of net benefits using a 10% real discount rate. These are derived from x
incremental benefits (Table VII.G.I) and incremental costs (Table VII.G.4). H
-------
Table 14. 31 Eastern State Net Benefit Assessment for Alternative Standards
(5% Discounted Present Value in Billions of January 1984 Dollars)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
Low Middle High
0.5 PPM
1-Hour Standard
Low Middle High
0.25 PPM
1-Hour Standard
Low Middle High
Net Benefits 1.2 2.5 4.5 1.2 3.7 7.3 -1.4 3.7 11.3
1
1 The discounted'present value of net benefits using a 5% real discount rate. These are derived from
incremental benefits (Table VII.G.2) and incremental costs (Table VII.G.5). .
-------
Table 15. 31 Eastern State Net Benefit Assessment for Alternative Standards
(2% Discounted Present Value in Billions of January 1984 Dollars)1
Alternative SO? NAAQS
Current Standards 0.5 PPM 0.25 PPM
(Strict Interpretation) 1-Hour Standard 1-Hour Standard
Low Middle HhTgFLow MTddTeHigh Low" MTddTeHigh
Net Benefits 3.0 4.9 7.7 3.1 6.2 11.4 3.8 10.8 21.0
The discounted present value of net benefits using a 2% real discount rate. These are derived from
incremental benefits (Table VII.G.3) and incremental costs (Table VII.G.6).
-------
XXXI
Table 16
Implicit Valuation of the SO^ Mortality Risk Reduction Coefficient*
(Lives saved/100,000 population/ugnr reduction in SO^ annually)
Alternative SO? NAAQS
Implicit Valuation Coefficient?
$420,000 $7,300,000
Low estimate 0.25 ppm - 10%
Low estimate 0.25 ppm - 5%
.26042
.11272
.01496
.00648
1 The value of the mortality risk reduction coefficient required for
incremental benefits to equal incremental costs. This coefficient
is derived assuming the value of a statistical life saved is $7.3
mil lion.
2 Valuation coefficients of $.420 and $7.30 for decreased mortality
risks of 1.0 X lO"6. See Tables B.I and B.2 in Section VI,
Appendix B for more detail.
-------
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
decision not to revise the sulfur oxides national ambient air quality
standards (NAAQS) is a major action. Full compliance with the current
standards could result in an annual effect of $1UO million or more on the
economy. Moreover, the principal alternative examined, the addition of a
one^-hour primary standard, had larger impacts. 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.
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 primary standards in setting the ambient standards. The Act
requires that secondary standards be set at level(s) that protect against
both known and anticipated adverse effects on public welfare. The Agency,
in its particulate matter rulemaking and in its ANPR for fine particles,
solicited comment on the role of economic analyses in secondary standard
reviews. The question is still under consideration by the Agency. However,
in the case of this RIA the Agency has decided not to consider the results
in standard setting. This decision is based primarily on the fact that
the RIA methodology has not been subjected to public and CASAC review.
*
This draft RIA examines the impacts of alternative levels standards
in terms of the benefits to be derived, the cost and environmental impacts,
-------
1-2
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 requirements affecting the development
and revision of NAAQS and the OAQPS staff papers: Review of the National
Ambient Air Quality Standards for Sulfur Oxides: Assessment erf Scientific
and Technical Information - QAQPS Staff Paper (EPA-4bO/5-82-U07, November,
1982) and The Addendum to the QAQPS Staff Paper (1986). The OAQPS staff
papers for sulfur oxides interpret the most relevant scientific and technical
information reviewed in the revised Ai^r Quality Criteria for Particulate
Matter and Sulfur Oxides (EPA-600/8-82-029a,b,c; December, 1982) and in
The Second Addendum to_ the Air Quality Criteria for Participate Matter and
Sulfur Oxides (EPA 450/5-86-012, 1986). The OAQPS staff papers, which have
undergone careful review by the Clean Air Scientific Advisory Committee (an
independent advisory group), serve to identify those conclusions and
uncertainties in the available scientific literature that should be considered
in determining the 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
affecting the development and revision of NAAQS and briefly describes
the nature of sulfur oxides 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.
Finally, section !U9(d) of the Act directs tne 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 Sulfur Oxides Problem
The principal focus of this standard review was on the health and
welfare effects of S02, alone and in combination with other pollutants.
Other sulfur vapors (e.g., 803) are not commonly found in the atmosphere.
Information on the effects of the principal atmospheric transformation
products of SOg (i.e., sulfuric acid and sulfates) was considered in the
review of the particulate matter standard and addressed in the March 20,
1984 proposed revisions to those standards.
Sulfur dioxide (S02) is a reactive gas that is quite soluble in water.
It is emitted principally from the combustion of sulfur bearing fuels and
the processing of sulfur bearing ores. In the U.S., utility power plants,
non-ferrous smelters, and industrial boilers are the major sources of 502.
At elevated concentrations, SO2 can adversely effect human health, vegetation,
materials, economic values, and personal comfort and well-being. SO2 and
its transformation products are also major contributors to acidic deposition
and regional visibility degradation.
The existing standards for S02 were set in 1971 based on health and
welfare data assessed in the 1969 criteria document. Current primary
-------
II-3
(health-based) standards and secondary (welfare-based) standard for this
pollutant are:
Primary:
365 ug/m (U.14 ppm) 24-hour average (not to be exceeded
more than once per year)
80 pg/in^ (O.U3 ppm) Annual Arithmetic Mean
Secondary:
1300 ug/m^ (0.5 ppm) 3-hour average not to be exceeded
more than once per year)
An annual secondary standard of 60 yg/m^ (0.02 ppm) was also set in 1971
but was later revoked in 1973 following a court challenge. In its decision,
the court did not rule on the substantive basis for such a standard, but
rather remanded it because the Agency failed to supply an "implementing
statement" providing the basis for the standard.
Largely as a result of EPA and State efforts since 1971, current air
quality is reasonably good with respect to these standards. Only 63 areas
are now designated as non-attainment for S02- In most of these cases the
designation applies only to limited areas in the immediate vicinity of
major point sources. Since 1975 national emissions of SOg have declined by
approximately 16%. The decline in emissions has been accompanied by an
improvement in measured air quality levels in most urban areas, thus reducing
public health risks in these areas. In recent years EPA and the States
have tended to implement the standards through dispersion modeling around
point sources rather than by monitoring actual air quality. Although some
problems still persist, most major point sources now comply with SIP regulations,
The present review of the S02 criteria and standards was initiated in
1978. The Clean Air Scientific Advisory Committee (CASAC) closed on the
Criteria Document (which also addressed parti cul ate matter) in January
-------
II-4
1982. The EPA staff paper, which Identifies critical issues and summarizes
the staff's interpretation of key studies, received verbal closure at a
CASAC meeting in August 1982 and formal written closure in August 1983.
Subsequently, addenda to both the criteria document and staff paper were
prepared in 1986 and reviewed by CASAC in October 1986.
Based on the comprehensive examination of all available scientific
information on the health and welfare effects of sulfur oxides in the criteria
document and analyses of current and alternative standards, the EPA staff
and CASAC recommended that the Administrator focus consideration on a discrete
range of scientifically supportable policy options for retaining or revising
the S02 standards. The Administrator has relied heavily on these recommenda-
tions, and on the detailed rationale contained in the staff paper and CASAC
closure letter in reaching his decision to propose not to revise the
current NAAQS's and to solicit broad public comment on the alternative of
setting a 1-hour NAAQS and making revisions to the current NAAQS.
Taken one at a time, the staff and CASAC recommendations could have
led to a large number of potential combinations of retention, modifications,
and additions to the current standards. The major choice that was considered
by the Administrator was between the following two principal alternatives:
1. Reaffirmation of the current standards in their present form, and
2. Addition of a new 1-hour standard and consideration of revisions
to all of the existing standards.
-------
II-5
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 might 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-6
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 sulfur
dioxide pollution problem is that pollution abatement tends to be a- public
good. That is, after sulfur dioxide 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 sulfur dioxide emissions
knowing that his actions will have little impact on how clean the air he
breathes actually is.
As this regulatory analysis shows, there are resource costs associated
with this governmental- intervention (see Section IV, "Cost & Environmental
Impacts Analysis"). However, in view of the scientific data on SO2 health
and welfare effects the proposed action is 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 S02- 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 S02 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 S02 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 Administrator to prescribe
standards for pollutants, such as S02, which have adverse effects on public
health or welfare 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. Moreover, transport of pollutants
across boundaries would make a system of regional or State air quality
standards difficult to enforce. The Clean Air Act requires national not
regional standards.
In summary, the regulatory alternatives outlined above have not been
analyzed in detail in this draft because of the Clean Air Act requirements
for setting and revising NAAQS. 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 means, for achieving the NAAQS for S02, but not as a substitute for NAAQS
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
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.
-------
III-4
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 would 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 could have the incentive
to strategically misrepresent their intended emissions.
(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
-------
III-5
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 contemplate consideration
of these market-oriented alternatives 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 approaches as
described above in the implementation of the NAAQS.
D. REGULATORY ALTERNATIVES WITHIN THE SCOPE OF PRESENT LEGISLATION
The Clean Air Act requires that primary NAAQS be set at levels which
protect the public health, including that 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., suggested
two major alternatives. The two major alternatives are briefly discussed
in Section II of this RIA. 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
is referred to the Criteria Document, the OAQPS staff paper, and the
preamble to the proposed decision not to revise. Listed below in Table
III.D.3 are the alternative standards which are featured in this analysis.
-------
III-6
Table III.D.I. S02 Alternatives Analyzed
Annual Arithmetic Mean 24-Hour Observed 3-Hour Observed 1-Hour Expected
(Primary) 2nd Maxima 2nd Maxima Exceedance Rate
(Primary) (Secondary) of 1
1) - - 0.25 ppm
2) - - - 0.5 ppm
3) 0.03 ppm 0.14 ppm 0.5 ppm
Items 1 and 2 correspond to the upper and lower bounds of the original
range of interest in the 1982 Staff Paper. Item 3 is the current NAAQS
which is being reaffirmed.
-------
IV. Cost and Environmental Impacts
A. Introduction
This section presents a summary of the estimated costs and environmental
impacts associated with meeting various levels of S02 NAAQS. The
alternatives examined correspond to those outlined in Section III.D.
Since S02 is primarily a point source pollutant the cost and environmental
impact analyses have focused on the largest point sources. The major
source categories analyzed were utility power plants, non-ferrous smelters,
and industrial boilers. Together these source categories account for
approximately 85% of the National Emission Data Systems (REDS) inventory
of S02« A more thorough discussion of the rationale and justification
for this approach is discussed in sub-section B. "Problem Characterization."
The following sub-section, C. "Control Strategy and Cost Methodology,"
outlines the analytic approaches used for each source category. Sub-
section D. reports the results in terms of costs and environmental impacts.
B. Problem Characterization
Sulfur dioxide emissions result primarily from the combustion of
sulfur bearing fuels and the processing of sulfur bearing ores. Over the
past ten to fifteen years there has been a marked reduction in S02
emissions in the U.S. Since 1975, emissions of this pollutant have
declined by almost 17%. Over an even longer time period, stretching back
to the 1940's, there has also been a change in the relative importance of
certain source categories. For example, in 1950 residential fuel combustion
contributed nearly 10% of total emissions whereas today it contributes
slightly less than 1%. On the other hand, electric utility emissions
increased from 20% of the total in 1950, to 67% today. In general, large
-------
IV-2
point sources of S02 have become more important, while emissions from smaller
dispersed sources have declined. This change is in part the result of
the control efforts of State and local air pollution agencies as well
as EPA. Most localities now require the use of "clean" fuels (e.g.
natural gas, low sulfur oil, etc.).
As a consequence of this process EPA's control efforts for S02 now
focus largely on major point sources of SO-^. Of the sixty three areas
currently designated as non-attainment for $03, nearly half involve only
a single point source. In the remaining areas multi-point sources
(sometimes interacting with each other) are implicated. In determining
emission limits for existing or new sources of SOg the Agency generally
uses dispersion models. The use of dispersion models arises from the fact
that short-term (ie. <_ 24-hour) SO2 levels show considerable temporal
and spatial variation. In this situation, fixed monitoring networks tend
not to be as reliable in determining peaks as are dispersion models. Fixed
monitors are, however, useful in characterizing trends and longer term
averages.
In designing the studies which underlie this Regulatory Impact
Analysis, principal consideration was given to the most likely NAAQS options.
Specifically, in addition to the current NAAQS, this meant considering a
1-hour NAAQS in the range of 0.25-0.75 ppm. Analyses of monitored air
quality have shown that short-term peaks of such levels are rare at
population oriented monitoring sites. In a study of 1377 site years of
data, only 123 site years contained 2nd highest 1-hour values > 0.5 ppri
(Frank and Thrall, 1982). Of these only 10 were population oriented
sites; the balance being source oriented sites. On the other hand, both
modeled and monitored data indicate that large point sources do produce
-------
IV-3
1-hour levels _>VQ.5 p.pm (Frank and Thrall, 1982, SAI,V 1982 and SAI1983).
Accordingly, it was, decided to; focus the analysis on major point source
categories shown in Table IV;.l. In the case of electric utilities, monitoring
data and the modeling data cited above indicated that power plants do produce
1-hour concentrations in the range of concern. For industrial boilers, the
data indicated that large boilers and those with poor dispersion characteristics
(e.g., short stack or relatively low buoyancy flux) might also produce such
Table IV.B.I
Emissions From
Major Source Categories of Sulfur Dioxide-'-
(106 Metric Tons Per Year)
1982
Electric Utilities 14.3 (67%)
Industrial Boilers , 2.3 (11%)
Primary Copper & Lead Smelting 1.2 ( 5%)
Petroleum Refineries .7 ( 3%)
Other • .. 2.9 (13:%)
Total 21.4 (100%)
1) Source: "National Air Pollution Emissions Estimates, 1940-1982"
OAQPS (EPA-450/4-83-024, February, 1984). N.B. This reference .
is used in this analysis to estimate trends and the relative
., importance, of specific source categories. The inventories actually
used to estimate air quality and emissions limits are described
below in Section IV.C.
levels. However, even a preliminary and conservative screening analysis
indicated that the number of such facilities would likely be small (_< 3% of
all industrial boilers). Monitoring data collected around primary copper
smelters indicated high 1-hour levels* A combination of both monitored and
modeled data were ,used to determine that primary lead smelters do produce
elevated 1-hour concentrations. With respect to petroleum refineries, a
detailed modeling study of four refineries, chosen because they processed
relatively high sulfur crude, showed maximum,1-hour impacts well below 0.5
ppm given-their current emissions (TRW, 1981). As a result, the decision
-------
IV-4
was made to exclude refineries from further analysis. Subsequent analyses,
sponsored by the American Petroleum I nstitute (API), have shown the same
order of estimated 1-hour impacts (Hayes et al . 1984). In summary, the
cost analyses discussed below are based on studies of three major source
categories: electric utilities, industrial boilers, and primary copper
and lead smelters.
Since most State implementation plan (SIP) regulations for S02 are
now established through dispersion models, it was decided that the RIA
analyses should approximate dispersion modeling results as closely as
possible. A brief discussion of the modeling techniques used for each
source category follows in Section IV.C "Control Strategy and Cost Methodology."
However, in the context of this discussion of problem characterization,
it is important to note that within each source category sources were
modeled individually. In other words, sources were assumed not to
interact. Several factors lie behind this decision. In the first place,
a review of non-attainment areas under the current NAAQS showed that
single source problems predominate. Source interaction was flagged as a
problem for a number of areas, however, and especially for the longer 24-
hour averaging period. Since a primary concern in designing the RIA was
the potential for a 1-hour NAAQS, the question of source interaction for
1-hour became important. In theory, it is unlikely that neighboring
sources would produce a combined impact _>. 0-5 ppm. The meteorological
conditions which usually produce maximum 1-hour impacts (i.e. A-stability
and 3.5 m/s winds) make it very unlikely although not impossible that
neighboring (not co-located) sources will produce their maximum impact (>_
0.5 ppm) at the same location. The monitoring and modeling analyses
cited above tend to support this conclusion.
However, if the 1-hour concentration of concern drops to U.25 ppm,
-------
IV-5
source interaction could become more of a factor. In light of these
concerns a screening analysis was undertaken to try to determine the
extent of source interaction (ANL, 1984). This analysis covered 129
counties chosen because they were either currently non-attainment due to
source interaction or had large and diverse S02 inventories. Standard
dispersion equations were used to estimate impacts at a finite number of
receptors (20-80 per urban area). Of the 930 receptors examined only 38
exceeded 0.5 ppm 1-hour and in half of those cases a single source caused
the violation. A total of 38 receptors showed violations of the current
3-hour NAAQS and 11 violated the current 24-hour. In summary, this screening
analysis, designed to find as many multi-source problems as possible,
found no more than 5% of the receptors to have such problems. Given the
differences in analytic techniques and in data bases used, it is extremely
difficult to compare these screening results to the single source studies.
However, in the utility analysis, depending on the assumptions used,
between 25% and 50% of the plants modeled exceeded 0.5 ppm and for the
smelters this becomes 100%. A further analysis showed that populations
likely to be affected by single source problems were considerably larger
than those affected by multi-source problems.
In conclusion, given the technical evidence pointing toward single
sources and the considerable expense which would be incurred to model
source interaction in detail, it was decided to proceed with a single
point source analysis. Given that decision, it is legitimate to ask
whether it introduced a significant bias to results. It is not possible
to provide a precise quantitative answer to that question. However, the
multi-source screen does provide enough information to give a qualitative
answer. For the current NAAQS and for a potential 1-hour NAAQS of 0.5
ppm the single source assumption may have resulted in a quite small downward
bias in the final cost estimates. In light of other analytic assumptions
-------
IV-6
and data base problems this bias is negligible. For a 1-hour NAAQS of
0.25 the bias is probably larger. However, it was beyond the scope of
this effort to quantify it precisely.
C. Control Strategy and Cost Methodology
This section provides a summary of the control strategy and cost
methodologies used for each of the source categories addressed in this
RIA. The methodologies used varied somewhat from source category to source
category and were selected with characteristics of the category in mind.
In general, however, there are several features that are common to all
of the methodologies. In the first place all of the control strategies
were developed by modeling each source individually. This means that
emissions and dispersion data specific to each individual source were
used in estimating the air quality impact of that source and the amount
of emission reduction needed. Secondly, as noted in IV. B above, sources
were analyzed in isolation with no explicit consideration of source
interaction. However, where appropriate, background factors were added
to account, in general, for the contribution of other sources. The
handling of background is discussed in each of the source category sections
below.
1. Utility Power Plants
a) Control Strategy/Emission Limit Analysis
EPA's National Emission Data System (NEDS) indicates that, with
67% of total S02 emissions, the utilities are the most important source
of S02 today. The first step in analyzing the local impacts of utili-
ties was the development of a utility emissions data base. A stack-
by-stack inventory was developed. The file, referred to as the STACK
-------
IV-7
file, was developed from available data sources; primarily FPC Form 67
data and the Department of Energy's Generating Unit Reference File ("GURF")
(Pechan, 1981). Following initial development of the STACK file, an
effort was made to cross-check it with an inventory maintained by the
utility industry for both accuracy and completeness. This cross-check
resulted in the inclusion of additional stacks and missing data points
(UDI, 1982). The resulting STACK file covers generating units >_ 25 MWe
capacity with fuel quality/emissions data from 1980. Despite the efforts
that were made to assure a complete file on each plant/stack, missing
data points still proved to be a problem. In particular, for the analysis
to proceed, detailed information on each stack (e.g., height, diameter,
temperature, exit velocity) was required. Of the 777 utility power
plants represented on the file, final calculations were possible for 505.
Thus, some 272 plants were excluded because of missing or erroneous data.
Subsequent review showed that these were primarily smaller units. Specifically,
the review showed that the stacks and units analyzed accounted for 80% of
total steam capacity and 85% of total coal fired capacity (Braine, 1984).
With respect to total utility S02 emissions, the units analyzed re-
presented 84%.
The second step in the utility analysis involved a determination of
the air quality impacts around each unit. In a regulatory setting, such
a determination would be made using a dispersion model which calcuates im-
pacts for an array of receptors for each hour of the year. A complete
modeling of each stack in the inventory would have required resources well
beyond the scope of this project. Therefore, a screening methodology was
developed to produce credible estimates of the maximum 1-, 3-, and 24-hour
impacts associated with each stack and plant. The screening methodology
-------
IV-8
was designed to be conservative. For the purposes of the RIA it was impor-
tant that the screen be as accurate as possible and that it not underestimate
the impact of a standard. The methodology required the definition of a
"worst case" meteorological scenario (i.e., the meteorological condition which
produces the maximum impact) for each stack and averaging period and a calcula-
tion of ma-ximum impact, given that scenario, as a function of: 1) Stack
height (H), 2) Buoyancy flux (F), and 3) Emission rate (Q) (Anderson, 1984).
The development of this procedure for maximum 1-hour impacts was relatively
straightforward, since meteorological data are defined in 1-hour blocks and
therefore the procedure had only to address single events. For 3- and
24-hour impacts the procedure had to define adverse sequences of meteorological
events. The development of the screening methodology is described in
detail in the supporting documentation to the RIA (Anderson, 1984). However,
it is important that a number of issues be discussed here:
Comparison to Dispersion Model Results: The screening model used in
the RIA was designed to yield results similar to those which would
have been obtained with CRSTER (the Agency's principal model used in
SIP work). There are several important differences between the screen
and CRSTER. An important difference is that CRSTER generates a range
of predicted ambient concentrations depending on meteorological conditions,
whereas the screen generates a point estimate assuming worst-case
conditions. In the first analysis of the 24-hour standard, fairly
conservative assumptions regarding worst case meteorology were used in
order to produce predictions of ambient concentrations in the high end
of the range that would have been produced by CRSTER. The screening
model results were then compared with CRSTER results for 5 plants
and 10 different meteorological data sets. This comparison showed that
the screening model estimated 24-hour second maxima some 10-35% higher
-------
IV-9
than CRSTER. After reviewing the results, all of the predictions
were adjusted downward by 33% in order to generate a best-guess,
rather than a conservative, estimate of the impacts. This adjustment,
however, obviously resulted in lower estimated cost and emission
changes. For example, the estimated emission change dropped by some
38% and annual utility cost changes dropped by 50%.
The issue of whether or not the screening model is conservative
for the analysis of alternative one-hour standards is a bit more
complicated. For a one-hour standard, CRSTER and the screen operate
similarly. The distribution that CRSTER produces for a 24-hour standard
results from varying individual hours; for a one-hour standard, there
is no smaller time increment over which the model can vary the
meteorological data. High one-hour events generally result from a
limited number of meteorological conditions. Unstable conditions
with high turbulence (A-stabjlity) can cause high 1-hour concentrations.
For A-stability, the screening model uses an algorithm based on CRSTER's
treatment of this condition. Thus for A-stability, the screen and
CRSTER will yield similar results. Several other low persistance
events are also thought to produce high 1-hour concentrations. These
include fumigation and limited-mixing. Meteorological field studies
of these events have produced somewhat contradictory findings and as
a result these conditions are not modeled in CRSTER or any other EPA
model. However, if a 1-hour standard were set, consideration would
likely be given to modifying the models to include these conditions.
In light of this, a decision was made to include fumigation and limited
mixing in the screen. A review of the screen results shows that the
inclusion of the other events (i.e. those not modeled by CRSTER) did
-------
IV-10
not affect the final results. For example, of the plants controlled
under the U.5 ppm NAAQS, all were limited by A-stability. In the 0.2b
ppm case, only 2 of the 19b stacks controlled were limited by conditions
other than A-stability. Since as noted above the screen and CKSTEK
treat A-stability identically, the 1-hour results from the screening
model results should be largely identical to those which would be
obtained with CRSTER.
Fuel Sulfur Variability: A critical question in determining compliance
with S02 NAAQS concerns the treatment of fuel (specifically coal) sulfur
variability. The sulfur content of coal varies considerably. The
quality of a coal (with respect to sulfur content) is often described
as the long term average sulfur content. The STACK file used in this
analysis records the long-term average. However, a quantity of coal
with a given average sulfur content is made up of coals with sulfur
contents both higher and lower than the average. Therefore, it is
probable that if average sulfur content is used in modeling a plant
the actual short term peaks will be higher than those predicted. To
account for this phenomenon in the RIA, an estimated sulfur content
greater than the average was used for each plant. The estimated
coal quality was calculated using data on actual coal sulfur distributions
taken from 12 different power plants. The GSDs used were:
24 hours: 1.15
12 hours: 1 .175
3 hours: 1.19
1 hour: 1.20
The ratio of the maximum ground level concentration with peak
sulfur content to the maximum ground level concentration with mean
sulfur content was calculated for a range of time periods. It was
determined that a ratio of 1.25 results in a peak SOg impact with only
-------
IV-11
about a 10% probability of exceedance (Anderson, 1984). This factor
was then used in calculating maximum ground level impacts and therefore
the amount of emission reduction required.
It is EPA's policy that emissions limits are to be complied with
on a short term basis. However, in actual practice there is a wide
range of compliance procedures. Many plants appear to be complying
with what ought to be a short-term limit on a long-term average
basis. If this misinterpretation of emission limits were to continue,
then the use of the 1.25 factor described above would introduce a
conservative bias into the analysis and implies that the costs and
emission changes are over-estimated.
Background: To account for the possibility that smaller, dispersed
sources of S02 might add to the estimated concentrations around a
power plant, a background factor of 0.007 ppm was used. This factor
was estimated from monitored data. EPA trends data for 1982 were
used and annual average concentrations were found to range from 0.002
ppm to 0.045 with a mean of 0.007 ppm. The 0.007 ppm background factor
was used in all calculations of 3- and 24-hour impacts. Since the
meteorological conditions used to estimate 1-hour maximum impacts (A
stability and 3.5 m/s winds) provide for rapid break-up of background,
no background factor was used in the 1-hour case. However, had the
general background factor been used, it would have resulted in the
addition of only one more plant to the control strategy for the
l^hour standards. Use of the maximum annual average (0.045 ppm) as the
background factor instead of the mean (0.007 ppm) would also have
increased the number of sources controlled. For example, in the current
NAAQS case, some 32 plants would have been added to the strategy.
Terrain Impacts: The maximum impact from a given plant depends
not only on its dispersion characteristics and emissions but also
-------
IV-12
on the surrounding terrain features. Impingement of a plume on
nearby terrain features can produce high short term impacts. As a
check on the RIA analysis a screening procedure to estimate imping-
ment impacts was developed and a data file on terrain features was
assembled (gridded to 1 krn by 1 km cells). It must be noted, how-
ever, that only the single most significant terrain feature in each
grid cell was listed for that cell. Moreover, its exact location
within the cell was not available. This meant that the location of
the terrain feature with respect to a plant located in the same grid
cell could not be precisely established. This relatively simple
terrain analysis procedure indicated that the number of plants
estimated to violate a standard would increase if terrain were
considered. However, in view of the inherent difficulties in pre-
dicting terrain impacts with the best of data, and the very imprecise
data available for this analysis, it was determined that the cost
analysis should proceed under a flat terrain assumption. It is
likely that a detailed plant by plant terrain assessment would
result in additional plants being controlled and some of those
already controlled needing additional control. This, of course,
would imply higher costs as well.
Multi-Stack Plants: For short averaging periods, a power plant
plume impacts at ground level in a relatively small "footprint".
Where stacks are co-located variations in the dispersion charac-
teristics from stack to stack will lead to displacement of the
footprints. If the displacement is sufficiently large, the impacts
will not be additive and only the most "adverse" stack's emissions
will contribute to the maximum short term impact. If, on the other
hand,, trie stacks are nearly identical, their impacts will be additive
-------
IV-13
and the entire plant's emissions will contribute to the estimated
maximum. A simple algorithm was developed to determine whether
stacks at a given plant were additive or not (Anderson, 1984). At
plants where the stacks were additive the entire plant's emissions
were assumed to contribute to the maximum impact. Where the stacks
were not additive only the emissions from the most "adverse" stack
were assumed to contribute.
Load Condition: Plume impacts were estimated for 100%, 75%, and
50% load conditions. Although the lower load conditions result in
lower emissions, they also can result in more adverse dispersion
characteristics for the plume (eg. temperature). For each plant
and averaging period the most adverse load condition was assumed.
For the majority of plants 100% load was most adverse, but there were
some plants where other load conditions were most adverse.
The third step in the analytic process, following the development of
the utility STACK file and the determination of air quality impacts, was
the calculation of appropriate emission limits. Using the methods described
above, maximum concentrations for each averaging time at each plant were
calculated accounting for background, coal sulfur variability, additive vs.
non-additive stacks, and load condition. Then, for each of the standards
analyzed (see Table III. 0. 3), the most critical averaging period and
load condition combination at each plant was determined. A new allowable
emission limit was then calculated for all plants which exceeded the stan-
dard^) under consideration by multiplying the current emissions times the
ratio of the standard to the estimated impact (Anderson, 1984). It is
important to note that plants which could potentially increase their emission
limits were not analyzed further. This was done because of the many
-------
IV-14
uncertainties involved in determining whether a plant would take advantage
of an opportunity to increase its emissions. For example, many plants
which could increase emissions under the current NAAQS have not, for
reasons ranging from favorable long term coal contracts to an unwillingness
to consume PSD increment.
b) Cost
In general, utilities can meet SC>2 emission limits with either a shift
to a lower sulfur fuel or scrubbing (or a combination of both). The possi-
bility that a significant number of utilities might shift to lower sulfur
fuel implies that there may be a shift in demand for various fuels large
enough to affect prices. This in turn implies that the cost of compli-
ance cannot be analyzed on a plant-by-plant basis. The analysis of cost
has to consider both shifts at individual plants and possible changes in
the market price of various fuels. To accomplish this, the Coal and Electric
Utilities Model (CEUM) developed by ICF Inc. was used. This model is do-
cumented and described in detail elsewhere (ICF, 1981). In its simplest
terms, the model attempts to define coal supply, including transportation,
and both utility and non-utility coal demand. It generates an equilibruim
solution through standard linear programming techniques balancing supply
and demand at least cost.
The coal supply component of CEUM offers a variety of coal types
based on Btu/volatility levels and sulfur levels. Some 38 coal supply
regions are defined and supply curves for each coal type available in a
region are also developed. The supply curves are a function of coal
reserves data and mine engineering costing algorithms. The coal
transportation network represented in CEUM has some 1,000 links. Modes
of transportation include unit trains, barges, Great Lakes steamers,
-------
IV-15
intercoastal barges, ocean vessels and trucks. Cost functions (dollar/ton
or dollar/ton/mile) are developed for each transportation mode.
The demand component of the model is broken into 47 demand regions.
Demand is further structured into a utility sector and five non-utility
sectors. The utility sector is driven by exogenously specified assumptions
regarding electricity demand and growth rates. The model minimizes the
cost of generating and distributing electricity and determines the extent
to which 1) powerplants are operated at various load categories, 2) new
power plants of various types are built and operated at various load
levels, and 3) electrical power is transmitted between regions.
Environmental standards are specified as constraints.
Although the CEUM model is described in detail elsewhere, a number
of the assumptions and uncertainties deserve to be discussed here:
Short-Run Coal Production and Scrubber Constraints: The analysis
assumed that the coal producers and scrubber manufacturers could
respond to new NAAQS requirements by 1990. For current NAAQS and
possibly the 0.5 ppm 1-hour alternative this assumption may not be
unrealistic. For example, a 0.5 ppm alternative would increase scrubbed
capacity by some 9 gigawatts (Gw) and it has been estimated by ICF
that the scrubber industry can accommodate up to 20 Gw by 1990. However,
the scrubber industry itself feels that it can expand to readily meet
,, increases in demand. The assumption of no short-run constraints in
the case of a 0.25 ppm standard is more likely to be unrealistic. A
0.25 ppm 1-hour NAAQS results in an increase of 60 Gw scrubbed capacity,
compared to the estimated potential industry capability for 20 Gw.
Moreover the 0.25 ppm NAAQS resulted in a forecast of significant
shifts in.coal production and shipments. It is estimated that all of
these short-run constraints could be handled if implementation were
-------
IV-16
deferred until 1995. Implementation in 1990 would likely result in one
or a combination of the following situations arising 1) to meet the surge
in scrubber demand, scrubber prices might rise, 2) hurriedly designed
and installed scrubber systems might not achieve present reliability
levels, and 3) compliance deadlines might be missed.
Changes in Demand for Electricity: As noted above demand and changes
in demand for electricity are exogenously specified in CEUM. In
addition, the model assumes inelastic demand which means that price
changes will not result in any change in demand. Although demand
for electricity is relatively inelastic studies have shown it is not
perfectly inelastic. Use of elastic demand curves would result in
lower consumption of electricity. The impact of this on emissions
would depend on how the reduced load were dispatched. However, it was
beyond the scope of this analysis to derive industry specific demand
curves for electricity.
New Plant Construction: One of the features of CEUM is that it
brings on new plants over time. The model assumes that new utility
power plants will meet the appropriate NSPS. Available information
was used to determine which plants would be subject to Subpart D
requirements (i.e. 1972 NSPS) and which would be subject to Subpart
D(a) requirements (i.e. 1978 NSPS). The inclusion of new NSPS plants
in the cost analysis raises the issue of whether such plants could
meet the current and alternative standards being considered. Detailed
modeling of four actual or proposed plants subject to Subpart Da
indicate that such plants would not exceed any of the standards
under consideration. Therefore, for all of the alternatives examined
-------
IV-17
the Subpart Da plants were assumed to come on line with their NSPS
controls. In the analysis for 0.5 ppm, no further reductions were
needed at plants made up of exclusively Subpart D units. However, an
analysis of plants with a mixture of units (some subject to Subpart D
and some only subject to SIP requirements) indicated that some of those
plants might exceed 0.5 ppm, and therefore would be required to obtain
additional control under that alternative. In the analysis of the
costs, the reductions were assumed to be obtained from the SIP unit(s)
first. In the case of the 0.25 ppm alternative, approximately 20 NSPS
units would be required to obtain additional control. Once again,
reductions were first obtained from any SIP unit(s) in the same plant,
if possible, in the cost analysis.
Cost of the Current NAAQS: For the utilities the analysis of the
current standards presented a special difficulty. As noted above the
treatment of sulfur variability as well as other modeling issues can
have a large influence on the degree of emission reduction needed to
meet a given standard. EPA's Compliance Data System (CDS) and ICF's
CEUM data both indicate that the overwhelming majority (>90%) of
utilities comply with their SIP limit. The SIP limit theoretically
guarantees attainment of the current NAAQS. Reflecting this, CEUM's
"normal" base case forecast shows attainment of the current NAAQS
with no cost. However, it is known that for many plants sulfur
variability was not explicitly considered in establishing emission
limits. Some of these plants comply with what should be a short-
term limit only on a long-term average basis. Also many plants
comply with emission limits set in the early 1970's without dispersion
modeling. Some of these would require tighter limits if they were
-------
IV-18
modeled today. Finally, a small number of plants have limits that
reflect compliance with the primary but not the secondary NAAQS.
For all these reasons, as well as the inherent conservatism of the
screening analysis outlined above, it is not surprising that the screen
shows additional control needed at some 20 to 3D plants to meet the current
NAAQS. The cost analysis examined these screen-predicted reductions
associated with the current IMAAQS as well as the standard CEUM baseline.
It should be noted that the cost analysis focuses on the results obtained
with the "adjusted" screening model (see page IV-8 & 9).
The data presented in Section IV.D show a range of costs and emissions
reductions in the current NAAQS case for the utilities. In each case
the "zero" estimate could be considered to reflect current practice,
while the positive cost and emission reduction figures represent a
more rigorous and uniform practice. No range is presented for the 1-hour
alternatives, since it was assumed that such a major change in NAAQS would
also entail changes in emission limit definition and compliance practices.
Least Cost Approach: The KIA approach to the utilities assumes emission
limits will be set on a plant-by-plant basis and CEUM assumes that
utilities will meet these limits at least cost. At present most new
utility emission limits are set on a plant specific basis. However in
the past, SIP's sometimes imposed county-wide or state-wide emission
limits to all sources within a given category. To the extent that
a SIP takes such an approach this analysis could underestimate costs.
2. Copper Smelters
a) Control Strategy/Emission Limit Analysis
Primary copper smelters emit approximately 6% of total U.S.
emissions of S0£. Although EPA has generally moved toward the use of
-------
IV-19
dispersion models in establishing emission limits for point sources,
primary copper smelters represent an important exception to this trend.
In recent copper smelter actions EPA has accepted the use of multi-point
rollback (MPR). Several different factors underlie this acceptance. .In
the first place, many of the copper smelters are located in complex terrain.
Significant terrain features make dispersion modeling considerably more
difficult and controversial. In addition, copper smelters have highly
variable emissions over relatively short (24-hours) time periods. Although
utility and industrial boilers also have variable emissions, the temporal
scale is longer. The variability in smelter emissions results primarily
from the batch process nature of their .production cycle, with higher emission
rates occurring at different points in the production cycle. Depending on
the derivation of the emission rates used in the models, standard dispersion
modeling could result in overly strict or overly lax emission limits being
predicted. Since MPR is now a part of the institutional framework, it was
used in this RIA.
Multi-point rollback (MPR) assumes that if an emissions distribution
is "rolled back" by a given percent, then the distribution of observed air
•quality impacts will also be "rolled back" by the same percent. To use MPR
several things are required. Continuous ambient S02 data from as many
monitors as possible around a smelter are required to determine maximum
impact and thereby the margin by which the standard being analyzed is
exceeded (or met). Continuous emissions data for all the sources of S0£
within the smelter are also required. The emissions and ambient data must
represent the same time period. Moreover, the data must come from a period
when intermittent controls were not used (i.e., emissions and dispersion
must be independent). The maximum expected concentration is calculated
from the ambient data and used to determine the degree of SO2 reduction
-------
IV-2U
needed to protect the alternative standard under consideration. A cumulative
frequency distribution of emission rates is constructed from the emissions
data. This distribution (or the function describing it) is then reduced
at every point by the calculated reduction factor. The result is an
allowable emissions distribution rather than a single emissions limit.
The thirteen primary copper smelters in the U.S. are listed in Table
IV.C.I. The detailed emissions and air quality data needed to complete
an MPR demonstration were obtained, in most cases, from the appropriate
SIP. In those cases where MPR was not used in the SIP, the data were
obtained directly from the smelter. It should be noted, however, that
the Agency was unable to obtain the detailed emissions and air quality
data needed to complete a MPR demonstration for either the ASARCO, El Paso
facility or the Phelps-Dodge Hidalgo facility for this RIA. Therefore no
estimates could be made of the needed emission reductions for those smelters.
However, the State approved emission limit was available for the ASARCO
unit and for the sake of completeness the cost of meeting this limit was
determined. The Hidalgo smelter is already well controlled with a double
contact acid plant. It could obtain further reductions by improved operation
and maintenance of its control equipment, additional control of fugitives,
and by curtailing production. No costs were estimated for Hidalgo.
The MPR has been described in general elsewhere (Peterson and Moyers,
1980) and its specific application in analyzing alternative ambient S02
standards has also been documented (Peterson, 1983). In addition, discussions
of MPR have appeared in the literature (Mage, 1982). Several points are
worth noting here:
Fugitive Emissions: Fugitive emissions can make up 2-10% of smelter
emissions of SOg. These emissions are generally released at or near
ground level and do not have the same "effective" stack height as stack
-------
IV-21
Table IV.C.I
Primary Copper Smelters
Current estimated maximum
Current SO? emissions, Ib/h
Phelps-Dodge 13,450
Ajo, Ariz.*
Phelps-Dodge 124,351
Douglas, Ariz.*
Phelps-Dodge 53,455
Morenci, Ariz.*
Phelps-Dodge 6,800
Hidalgo, N.M.
Magma, Ariz. 55,583
Inspiration, Ariz. 3,116
Kennecott-Garfield, Utah* 8,250
Kennecott-Hayden, Ariz.* 12,098
Kennecott-Hurley, N.M. 23,033
Kennecott-McGill, Nev.* 72,7U2
ASARCO-Hayden, Ariz. 48,600
ASARCO-Tacoma, Wash.* 25,560
ASARCO-E1 Paso, Tex. 8,737
*Not currently (1987) operating
emissions. MPR does not distinguish between stack and fugitive
emissions in calculating needed reductions. Fugitive emissions were
not specifically controlled in this analysis and the entire emission
reduction was obtained with stack controls. In fact, stack controls
tend to be more cost-effective for smelting. However, if the monitor(s)
used in MPR were strongly influenced by fugitive emissions, the
stack controls may not result in as large a reduction in ambient
impact as predicted.
-------
IV-22
Intermittent Controls: Every effort was made to assure that the
data used in this MPR analysis was not collected during a period when an
intermittent control strategy (ICS) was being employed. The use of
ICS is not permitted by the Clean Air Act. However, in the past, before
the ICS issue was decided, smelters used such techniques to avoid
violations. If there is not independence between emissions and dispersion
the result could be to understate the degree of control needed.
Changes in Stack Parameters: The use of MPR assumes that the dispersion
characteristics of a smelter will remain unchanged by the controls
placed on it. In fact some of the controls used (e.g., scrubbers)
resulted in cooler, less buoyant plumes. In such cases PTMAX was used
to estimate the effect of the temperature decrease. Allowable
emissions were then reduced further to compensate for the lowered
stack gas temperature. In a few cases where no further reductions
could be achieved, additional stack gas reheat (beyond the bU°F
routinely applied) was provided.
Background: Due to their location in isolated rural areas of the
Southwest, background is considered, for all practical purposes, to be
non-existent for most smelters. It should be noted that the ASARCO
facility in Tacoma and the Kennecott facility in Garfield are both
close enough to urban areas that they might have measurable background.
However because the background levels would be small with respect to
srnelter impacts and because of the step function nature of smelter
controls, background values were not used.
b) Cost Analysis
Determination of true least cost controls for each smelter would
require a complex analysis of many factors such as age and current configuration
-------
IV-23
of the smelter, concentrate supply, copper and sulfuric acid markets, fuel
costs, etc. The approach adopted in this study involved a simpler estimate
of control system costs based on smelter throughput (PEDCo, 1984). Where
sufficient control could be achieved scrubbing systems were used. Lime
based scrubbing systems with a removal efficiency of 90% were used. Where
greater control was required smelting process changes to yield more concentrated
SOg streams for treatment in acid plants were used. The process modifications
considered included a fluidized bed dryer, a flash furnace, a converter
arrangement, and an oxygen-sprinkle reverberatory furnace/converter
arrangement. Wherever possible the choice of which system to use was
guided by compliance plans submitted to EPA by the company. A comparison
of the annualized cost for scrubbing with the annualized cost for process
modification requires a comparison of the total annualized cost of smelting.
As the detailed data in the supporting documentation indicate, in a number
of cases the process modification is estimated to result in a cost savings.
Process modifications do in fact result in lower costs as compared to some
current systems with scrubbing added. Nonetheless this result must be
viewed with some caution since the initial capital investment is quite
large and the length of the payback period will depend on factors such as
the cost of capital, copper and sulfuric acid markets, ore supplies, foreign
competition, etc. The depressed economic condition of the primary copper
industry raises the question of whether or not industry could afford to make
the initial capital investment at all. Thus even for those smelters where
process modifications might result in a theoretical cost savings, the cost
estimates presented below assume the smelters will scrub to meet an alterna-
tive standard if possible. Only in cases where scrubbing does not achieve
compliance is process Codification considered.
-------
IV-24
It should be noted that since this analysis was originally conducted
in the early 1980's, the primary copper smeltiny industry has undergone
significant change. A number of smelters, including those with the greatest
emissions and least controls (e.g., Douglas), have since ceased operations.
Since the smelters that are still operating tend to be the ones that have
better controls and because there are fewer smelters operating, the costs
shown here for each of the alternatives have been reduced.
3. Primary Lead Smelters
a) Control Strategy/Emission Limit Analysis
Although primary lead smelters account for only 0.3% of total
U.S. emissions of S02, their local air quality impacts can be significant.
The RIA analysis addressed the three primary lead smelters still in
operation in this country. The three, all located in Missouri, are the
ASARCO smelter at Glover, the AMAX smelter near Bixby, and the St. Joe smelter
at Herculaneum. It should be noted that ASARCO's facility in El Paso smelts
lead as well as copper; however, this facility was addressed along with the
other primary copper smelters. In contrast to the copper smelters, MPR has
not been used to determine emission limits for lead smelters. Evaluations
of the lead smelters have used dispersion models. Accordingly the RIA
analysis used a screening model based on the dispersion algorithms in
VALLEY and PTPLU. The ASARCO smelter was evaluated in a complex terrain
mode (it is located within 5 km of significant terrain features). The AMAX
and St. Joe facilities were evaluated in a flat terrain mode. All smelters
were modeled with their estimated current emissions and stack characteristics
and with their projected emissions and stack characteristics under various
control options. The details of the modeling analysis are provided in
the supporting documentation (Radian, 1984). Similar to the copper
smelters, background values were not used in the analysis of lead smelters.
-------
IV-25
b) Cost Analysis
Similar to the copper smelters only a limited number of control
options are available to the lead smelters. Essentially three levels of
control were assumed (Radian, 1984). Level 1 control is a sulfuric acid
plant and is already employed by AMAX and St. Joe. Level 2 control
involves scrubbing (Wellman-Lord process) of the lower S02 concentration
gases and assumes an acid plant already in place for strong stream gases.
The Wellman-Lord process was selected because it is a regenerable system
and the concentrated S02 gas it produces can be treated in the acid
plant. Level 3 control actually involves no further emission reduction
but requires plume reheat. Reheat increases the exhaust temperature and
volume flow and therefore increases plume rise.
4. Industrial Boilers
a) Control Strategy/Emission Limit Analysis
The control strategy and emission limit analysis for industrial
boilers is similar to that conducted for the utilities. Specifically a
screening methodology incorporating standard Gaussian dispersion algorithms
was developed to determine the sources which might violate the standards
under consideration and to then estimate emission limits (ANL, 1984).
The emissions data base used in this analysis was the MAP3S inventory,
which was largely derived from EPA's NEDS inventory. The MAP3S inventory
was checked for consistency and completeness. Where data were found to
be inconsistent or missing, values were calculated from other data or
default values were applied. As a result the initial analysis proceeded
with slightly over 19,QUO stacks to be modeled.
Due to the large number of stacks the modeling proceeded in two phases.
The first phase involved an intentionally very conservative screen. Specifically
worst case_wind speeds of each stability class were assumed and worst case
-------
IV-26
meteorology incorporating maximum possible wind direction persistence was
also assumed. In addition, maximum emission rates were used and plume
centerline concentrations were calculated. This screening resulted in a
list of stacks that had some potential to cause a violation of the standards
being considered.
In the second phase of the analysis a conservative, but more realistic
modeling was conducted for those stacks caught in the first phase. In
place of worst case meteorology a statistical sample of meteorological
conditions was used (sample size:l-hour:35,OOU; 3-hour :700U; 24-hour:1000).
The samples were derived from frequency distributions derived from 1964
Evansville, Ind. meteorological data. The samples defined wind direction,
persistence, and stability class. Worst case wind speeds were assumed for
each stability class. In place of maximum emissions, an emission rate
equivalent to an average for each of its annual hours of operation was
calculated. In most cases this value was between the maximum rate and the
annual average reported in NEDS/MAP3S. Finally, in place of plume centerline
concentration a sector average was used. The sector average is based on
the assumption that wind direction varies randomly within the reported 10°
sector. A statistically expected value for the concentration at a receptor
is then derived from the distribution of wind direction. An array of 216
receptors was used for each source. For those sources which violated a
given alternative standard, appropriate emission limits were derived from
the computed maximum concentration. Background values identical to those
used for the utilities were employed with industrial boilers.
b) Cost Analysis
Similar to the utilities, industrial boilers can meet S02 emission
limits with either a shift to a lower sulfur fuel or scrubbing (or a
combination of both). Therefore, it was decided to analyze industrial
-------
IV-27
boiler control costs in the CEUM model simultaneously with the utilities.
The CEUM model has been described above and need not be described further
here. The advantage that this simultaneous analysis affords is that all
the fuel demand shifts are considered together. This means that changes
in utility demand are reflected in industrial boiler costs. It should be
pointed out, however, that the industrial boilers were constrained to
fuel choice options and scrubbing was not considered. Because of their
size and the retrofit penalites associated with scrubbing, it was felt
that most existing industrial boilers would buy lower sulfur fuels to reach
compliance. Therefore, the refinement of considering scrubber options was
not believed to be worth the added resources needed for that analysis.
Table IV.C.2 below summarizes the various sources of error and un-
certainty in the cost and control strategy analysis. It summarizes both
the source of the error as well as the possible direction and magnitude
of the error on calculated emission reductions and costs.
5. Regional Scale Modeling
The standards analyzed in this RIA all relate to the local scale
impacts of sulfur oxides sources. However, it is generally recognized
that S02 emissions also have regional scale effects. In the atmosphere
S02 can be transformed into sulfates (804) and both can be transported for
long distances (>300 km). As a part of the overall RIA the effect of
S02 emissions reductions on: 1) sulfates and 2) visual range were modeled
on a regional scale. The modeling of source-receptor relationships on any
geographic scale and for any pollutant involves some uncertainty. Modeling
of regional scale transport, dispersion, chemical transformation and removal
of sulfur oxides involves a number of uncertainties. Nevertheless over the
past several years a number of models have been developed. These models
are now at-a stage of development that they can be used to provide some
-------
IV-28
Table IV.C.
Summary of Potential Sources of
Source
Error and Uncertainty
Direction and Magnitude of
Error on Calculated Emission
Reductions and Costs
General
Source Categories Not Analyzed
No Source Interaction
Air Quality/Control Strategy Analysis
Utilities and Industrial Boilers
Use of Screening Model
Treatment of Sulfur Variability
Background Factors
Flat Terrain Assumption
Inventory Completeness
Smelters
Use of MPR (copper)
Use of dispersion model (lead)
Background treatment
Treatment of Fugitive Impacts
Contamination of AQ data by
Intermittant Control Strategies
Cost Analyses
Utilities and Industrial Boilers
Elasticity of Demand
Scrubber Constraints
Least-Cost Control Strategy
Smelters
Use of Process Modification
-, likely to be small
-, small for current £ 0.5
ppm NAAQSs, larger for 0.25
ppm case
-, likely to be small
-, unknown potentially moderate to laryt
-, likely to be smal1
0, replicates current regulatory practii
0, replicates current regulatory practii
-, likely to be negligible
-, unknown
-, unknown
, unknown
-, more significant for 0.25 ppm case
-, unknown
-, unknown
-------
IV-29
Table IV.C.3
Comparison of Models/Matrices Used in Regional Analysis of Air Quality
Model
Criteria
1) Model vs. Matrix
2) Model Domain
3) Transformation Rates*
4) Return Flows
5) Treatment of
Elevated Emissions
6) Background Factors
7) Meteorology
8) Western State Emissions
(outside 31 -State Region)
9) Canadian Emissions
4-= , = , :
RTM-LT
Full Model
30 Eastern States,
plus portions of
Eastern Canada
Winter 0-0.25 2
Spring 0-0.9
Summer 0-1.6
Fall
Addresses return
flows from Atlanti c
Considers elevated
release
Large background
(2-5 ug/m3 factor
to account for
Western emissions
October 1977 to
September 1978
Not addressed (see
#2 & #6 above)
1995 projected;
unchanged by strat-
egy; truncated to
Southern Ontario
and Quebec
ASTRAP
Matrix
48 States, plus Canada
Winter 0.5 2
Spring 1.0
Summer 2.0
Fall 0.7
Addresses return flows
from Atlantic
Considers elevated
release
None
1976-1981
1995 projected;
unchanged by strategy
1995 projected;
unchanged by strategy
ENAMAP
Matrix
31 Eastern states.plus
Western states and
Canada east of 105°W
longitude
Winter 1.0 3
Spring *
Summer 2.0
Fall *
Addresses return flow
from Atlantic
Elevated release
not considered
None
January & July
1978
1995 projected;
unchanged by strategy
1995 projected;
unchanged by strategy
MONTE CARLO
Matrix
48 states, plus
Canada
Winter
Spring-
Summer
Fall
Addresses return
flows from Atlantic
Considers elevated
release
l.b ug/m3 S04
background used to
account for bio-
genie contribution
and return flows
from the Gulf
4 Months 1978
1995 projected;
unchanged by
1995 projected;
unchanged by
strategy
Expressed as % of SOg converted to sulfate per hour.
^Approximate daily averages, actual rates exhibit strong diurnal variation.
^Matrix addressed 2 months only.
-------
IV-30
Insight into the magnitude and nature of air quality changes that might
result from emission reduction scenarios.
Given the uncertainties inherent in regional scale modelling, several
different approaches were utilized in this RIA. The four different approaches/
models used are shown in Table IV.C.3. At the outset it should be noted
that only one, RTM-LT, was used in its full form. The others, ASTRAP,
ENAMAP, and MONTE CARLO, were not directly used; rather transfer matrices
derived from them were. It should also be pointed out that only RTM-LT
includes the calculation of visual range as a subroutine of the model. In
the other cases, visual range was estimated outside the model /rnatri x using
the matrix generated 804 concentrations. Finally the report on this specific
application of RTM-LT underwent an extensive peer review, while the
others did not.
Transfer matrices assume that concentrations at a given receptor
equal the sum of all partial contributions. Thus the concentration C at
receptor A could be represented as:
CA = TA1 * El + TA2 " E2 + ••• + TAJ ' EJ
where EJ is the emission rate from the Jth source region and T/\j is a pro-
portionality coefficient relating region J emisisons to region A concentra-
tions. An array of these T/\j coefficients can be generated from running the
full model and constitutes a transfer matrix. Transfer matrices are extremely
inexpensive and easy to apply compared to full regional scale model runs.
However, they also carry with them a number of limitations including:
* The matrices used here employ state level source and receptor
regions. This means that there is some "averaging out" of differences
which might be apparent in the finer grids most models use.
* When adjusted emissions are considered in a matrix, the matrix
will treat them as if the change occurs uniformly over an entire
state. This limitation could be important in this RIA because the
emission changes being modeled vary significantly from source to
source within a given region.
-------
IV-31
* A matrix is inherently linear and will not reflect any non-1'inear
responses. There is presently disagreement in the scientific
community as to whether or not linearity is a good approximation.
It should also be noted that this limitation applies to RTM-LT as
well. If non-linear processes are found to be significant, then
present models and matrices are not producing accurate estimates.
* Matrices are limited to the meteorological record used in their
derivation. Clearly a longer record is more apt to be more
representative of long term expectations. This is also true of
ful 1 models as we!1.
A more complete discussion of regional modeling and transfer matrices can
be found in the Memorandum of Intent on Transboundary Air Pollution (MOI 1982).
Table IV.C.3 summarizes some of the basic features of models/matrices
used and also describes the input data employed. A review of the table reveals
a number of key differences. The most critical differences concern RTM-LT's
model domain (item 2) as well as the treatment of Western State emissions
(item 8) and background (item 6). In this particular application RTM-LT
addressed only 30 Eastern states plus portions of eastern Canada. Since
Western state emissions were not modeled, background factors (as high as
5 ug/nP) were used to obtain better agreement with observed concentrations
in the base year. However, the introduction of a large background term
will reduce the estimated effect of control strategy on future air quality.
In fact, as will be seen in Section IV.D. below the relative improvement
predicted by RTM-LT is much lower than that predicted by the other models/
matrices. As noted above, this particular application of RTM-LT underwent
an extensive peer review. A number of reviewers had fundamental reservations
regarding the study because of the reduced size of the domain and the
introduction of large background factors.
A third difference worth noting concerns the length of the
meteorological record (item 7). Here the six year record underlying the
ASTRAP matrix is a strong point in favor of that matrix. At the other
end of the.scale, the present version of the ENAMAP matrix is based on
-------
IV-32
only two months of data. This raises serious concerns regarding the
representativeness of ENAMAP results in this particular application.
A fourth comparison worth noting concerns biogenic contributions. Natural
sources of sulfur add to regional levels of SOg and $04. MONTE CARLO
adds a small background term to account for these biogenic contributions.
The other models in their current forms do not. Finally, all the models
except ENAMAP consider elevated release of emissions. This means they
model at least two transport layers and handle elevated emissions differently
As noted above RTM-LT includes a routine to calculate visual range.
These calculations are described in detail elsewhere (SAI, 1984). The
procedures used are complex and used regressions of local airport data
against modeled sulfate in an attempt to account for local variations.
However, a number of peer reviewers found significant problems with the
approach and noted that it would seriously underestimate changes in
visibility (in addition to the tendency noted above to underestimate
sulfate changes). On the other hand, ASTRAP, MONTE CARLO, and ENAMAP do
not predict visual range. With these models/matrices a relatively simple
approach was taken to estimating a range of possible visual ranges from
504 concentrations (Bachmann, 1985). The approach requires making
assumptions regarding the sulfate/non-sulfate ratio in extinction over
time and space. However, it does take advantage of available empirical
data and allows specification of uncertainties. Office of Research and
Development/Atmospheric Sciences Research Laboratory reviewers agreed
with the stated limitations of the simpler approach, but nonetheless
found it appropriate for this application.
Results from all models are presented and discussed in Section IV.D.4
below. However, in view of the limitations noted above regarding RTM-LT
and ENAMAP and in view of time and budget constraints, only ASTRAP and
MONTE CARLO were used in the benefits calculations in Section VI.
-------
IV-33
D. Results
1. National
Table IV.0.1 presents the total estimated national costs for the current
NAAQS and for the two alternative one hour standards. These costs represent
the additional controls needed to move from a 1980 baseline to compliance
with the various standard levels considered. They include neither:
1) the costs of control incurred prior to 1980, nor 2) the costs of new
source controls associated with meeting NSPS, NSR, and PSD requirements.
It is assumed that all standards must be met by 1990. As noted above this
assumption is probably unrealistic for the 0.25 ppm case and also
for the smelters when process modifications are required. However,
the assumption of a common implementation year for all source categories
and standards does provide a consistent base for comparison. It should
also be kept i'n mind that the costs reported are for four major source
categories which account for 84% of all S02 emissions in the U.S. The
air quality impacts of each source were assessed individually and
appropriate background factors were employed. Due to the nature of the
inventory, the type of air quality problem being evaluated, and resource
constraints, source interaction was not analyzed. Finally the costs
reported here represent the cost of full attainment for all of the sources
analyzed. In no case was a source left in non-attainment.
As discussed above in Section IV.C (see page IV-17) a range of cost
estimates are provided for the current NAAQS case. This range derives from
uncertainties in defining compliance by the utilities. The lower estimates
(i.e. $0.4 billion in capital costs and $0.2 billion in annualized costs) are
based on the assumption that current emission limits for utilities and industrial
boilers do produce attainment of the current NAAQS and will not be revised.
On the other hand the upper estimates (i.e. $0.7 billion in capital costs
-------
IV-34
and $1.1 billion in annualized costs) are based on the assumption that
not all current emission limits produce attainment. [N.B. This upper
estimate is, however, based on the "adjusted" air quality screening
model. The unadjusted model had a distinct conservative bias and would
have resulted in utility capital costs of $1.5 billion and utility annual
costs of $1.4 billion. (See paye IV-17 for a fuller discussion.]
Factors such as failure to consider sulfur variability or inadequate
modeling may have resulted in some utility power plants or industrial
boilers being assigned limits that will not produce attainment. The
upper estimate can be thought of as representing a more rigorous and
strict implementation of the NAAQS.
Table IV.D.I presents the incremental costs of attaining each of
the alternatives reviewed. The control costs for meeting the current
implementation of the current SO;? NAAQS - which are significant - are not
included. Therefore, these incremental costs should not be used for direct
comparisons of the various alternatives. The high annual incremental costs
relative to the incremental capital costs for the U.b ppm alternative are
the result of the use of fuel switching - rather than control equipment to
obtain the required emissions reductions. For the 0.25 ppm alternative, a
significant number of utilities (175 units) would not be able to obtain
sufficient reductions by fuel switching alone, and would therefore need to
obtain further reductions by other means. In determining the incremental
cost of the 0.25 ppm alternative, the cost for scrubbers was used to estimate
the costs of control for those facilities.
Table IV.D.2 displays the national cost estimates broken out by
major source category. The largest share of the costs are borne by the
utility industry. As noted above the range of costs shown for utilities
under the current NAAQS results from ambiguities in compliance determinations.
-------
IV-35
Table IV.D.I
Total Estimated National Co$t Summary!
'($ Millions)
Standard
Current NAAQS
1-Hour 0.5 ppm
1-Hour 0.25 ppm
Capital
$350-$750
$2,800
$17,000
Annual! zed Cost
$200-$!, 100
.$2,200
$b,400
IAH costs are calculated in 1984 dollars and do not include the .cost of;
1) pre-1980 controls, or 2) new source controls tied to meeting NSPS,
NSR, or PSD requirements.
^Based on analyses of four source categories: Utilities, copper and
lead ,smelters, and industrial boilers o-r 84% of total SO;? emissions.
-------
IV-36
Table IV.D.2
Estimated National Cost Summary by Source Category
($ Millions)1
Current NAAQS
1-Hour 0.5 ppm
1-Hour 0.25 ppm
Source Catgegory
Utilities
Copper Smelters
Lead Smelters
Industrial Boilers
Total
Capital
$0-400
$200
$150
NA2
$350-750
Annual
0-700
$150
$4b
0-$200
$200-1,100
Capital
$2,200
460
170
NA2
$2,800
Annual
$1,800
110
50
200
$2,200
Capital
15,90U
7bO
170
NA2
17,000
Annual
5,000
60
50
300
5,400
•'•All costs are calculated in 1984 dollars and do not include the cost of:
1) pre-1980 controls, or 2) new source controls tied to meeting NSPS,
NSR, or PSD requirements.
2Control options were limited to fuel switches; therefore no capital costs
were estimated.
-------
IV-37
With a 1-hour U.b ppm alternative the utilities account for some 80% of
the estimated annualized costs; while for a 1-hour 0.2b ppm alternative
the utilities account for over 90%. This result was not unanticipated in
that the utilities constitute approximately 67% of the national SO^
emissions inventory. Moreover, unlike a source category such as industrial
boilers, the sources in the utility sector are individually large and can
produce significant ground level impacts. The ratio of capital to annual
costs in the utility sector reflects the reliance on fuel switching as a
control measure. However, in the 0.25 ppm case a very significant number
of utilities would have to scrub their emissions and the capital costs
become quite large. Another point to be made here concerns the annualized
costs estimated for copper smelters. As can be seen, while the capital
costs rise with the more restrictive one-hour standards the annualized
costs fall. This is due to the increasing use of process modifications as
a control measures. These modifications allow for a lower cost of production
and this savings is reflected in the annual cost. Again this result should
be viewed with caution; the industry in light of its current financial
straits might well focus on the initial capital- costs rather than the
annualized savings. Such a focus could result in decisions to close plants
rather than incur the fairly substantial capital cost. Finally, it should
be noted that because industrial boiler controls were limited to fuel
switches, no capital costs were incurred.
Table IV.D.3 reports the emission reductions that were estimated for
each of the three standards analyzed. Not unexpectedly, the emission
reductions achieved in the utility sector are significantly larger than
those in other source categories. Another point worth noting is the range
of emission reductions reported for the utilities under the current NAAQS
reflecting-the uncertainty in compliance definition. The one-hour standards
-------
IV-38
Table IV.D.3
Total Estimated Emission Reductions by Source Category
(Millions TRY)
Utilities
Copper Smelters
Lead Smelters
Industrial Boilers
Current NAAQS
0-2.4
1.4
0.2
0-0.2
0.5 ppm
4.4
1.6
0.2
0.2
0.25 ppm
9.0
1.7
0.2
0.3
Total 1.6-4.2 6.4 11.2
-------
IV-39
Table IV.0,4
Total Estimated Utility Costs1
($ Millions)
Utility Annual Costs
Capital
O&M
Fuel
Total
Utility Cumulative Capital Costs
31-Eastern States
17-Western States
Total
Current
NAAQS
0-100
0-100
0-500
0-700
0.5PPM-lhr
300
200
1,300
1,800
0-400
0^400
2,100
100
2,200
0.25PPM-1hr
1,800
1,000
2,200
5,000
15,100
800
15,900
Average Cost Per Ton SO? Removed
($/ton)
294
403
557
costs are calculated in 1984 dollars and do not include the cost of:
1) pre-1980 controls, or 2) new source controls tied to meeting NSPS, NSR,
or PSD requirements.
-------
IV-40
Table IV.D.5
Estimated Annual Utility Sulfur Dioxide Emissions
(Millions TRY)
Change from Base Case 1995
Base Case Current 0.5 PPM 0.25 PPM
198U 1995 NAAQS 1 Hour 1 Hour
31 Eastern States
Existing
Coal
Oil/Gas
Total Existing
New
Total 31-Eastern States
Total 17-Western States
Total U.S. 17.4 10.1 -2.4 -4.4 -9.0
14.9
1.3
16.2
16.2
1.2
15.9
1.1
17.0
0.9
17.9
2.2
-2.3
-0.1
-2.4
_
-2.4
_
-4.2
-0.1
-4.4
_
-4.4
_
-8.6
-0.3
-8.9
+0.1
-8.8
-0.2
-------
IV-41
Table IV.D.6
Estimated Changes in Scrubber Capacity
Current
Scrubber Capacity (Gw) NAAQS 0.5PPM-1hr 0.25PPM-lhr
31-Eastern States 0- 1.8 8.4 53.8
17-Western States - 0.3 4.0
Total 1.8 8.7 57.8
No. of Units Scrubbed
31-Eastern States 0- 2 36 164
17-Western States - __2_ 11
Total 0- 2 38 175
-------
IV-42
both require successively larger reduction from the utilities. This is not
the case for the copper and lead smelters. For these sources attainment of
the current NAAQS results in most of the emission reductions needed to
attain the one-hour standards.
2) Utilities
The data reported below provide more detail on the nature and distri-
bution of the costs and emission reductions estimated in the utility
industry. Table IV.0.4 provides a more detailed breakout of utility
costs, while IV.D.5 provides a similar accounting of emission reductions.
Table IV.0.6 provides information on the estimated increases in the use
of scrubbers. As measured both by scrubber capacity and number of units
scrubbed, scrubbers become an important part of the national control
strategy with the U.25 ppm alternative. It was noted above that short
run constraints in the scrubber industry could result in either higher
scrubber costs or delayed compliance. The distinction between the 31
Eastern states and 17 Western states is made to facilitate comparison
of these results to results from studies of regional sulfur strategies
(e.g., acid deposition and visibility strategies). The bulk of the costs
and emission reductions are realized in the Eastern United States.
It must be realized, however, that there are uncertainties in the utility
analysis. Although these are discussed in some detail in Section IV.C.I above,
it is worthwhile to note them again here. The treatment of sulfur variability
results in higher estimated impacts than would be the case if this factor
were ignored. The national scope of this analysis precluded treatment of
terrain effects. If terrain were considered, the estimated impacts would be
higher. Use of higher background values at some or all of the plants could
also have resulted in higher estimated impacts. However, as discussed in
Section VI.C.I, choices were made on the ensemble of these factors to try to
produce as unbiased an estimate of impact as possible.
-------
IV-43
Table IV.D.7
Industrial Boiler Annualized Costs and Emission Reduction
by Major Industry Groupl»2
($ Millions and Thousands TRY)
Industry Group (SIC)
Food (20)
Paper (26)
Chemicals (28)
Petroleum Refining (29)
Primary Metals (33)
Other
Total
Current
NAAQS
Annual Cost TPY
0-4
0-54
0-38
0-7
0-34
0-47
0-184
0-2
0-40
0-56
0-2
0-36
0-27
0-164
1-Hour
0.5 JDjDITl
Annual Cost
4
64
39
7
43
47
203
TPY
3
49
62
3
55
29
199
1-Hour
0.25 j3j>m
Annual Cost
9
82
72
7
44
114
328
i
TPY
5
67
102
4
75
53
306
costs are calculated in 1984 dollars and do not include the cost of: 1) pre-1980
controls, or 2) new source controls tied to meeting NSPS, NSR, or PSD requirements.
^Control options were limited to fuel switches; therefore no capital costs were
estimated.
-------
IV-44
3) Industrial Boilers
The total costs associated with industrial boilers are small compared
to other source categories. However, unlike the utility and smelters
source categories, the industrial boiler source category does not coincide
with a Standard Industrial Classification (SIC) code . Industrial boilers
are found in wide range of industries. Table IV.D.7 presents a break out
of the industrial boiler costs and emission reductions for the top 5 two
digit SIC codes. The industries with the largest impacts are paper,
chemicals, and primary metals. This break out by SIC code further indicates
that the impacts in any one industry are likely to be small.
4) Environmental Results
In addition to emission reduction estimates, other environmental results
were assessed. The ambient standards analyzed in the RIA are meant to result
in a certain level of local scale S02 air quality. The emission reductions
associated with each alternative result in the attainment of the NAAQS being
considered. However, it is generally recognized that reductions in SOg
emissions can also result in improvements to regional scale air quality. As
discussed in Section IV.C.5 above, four different models were used to assess
the impact of these emissions changes on regional $64 concentrations and
visibility. The models were used to predict changes in the 31-State Eastern
U.S. region (N.B. the RTM-LT model currently makes predictions for a 3U-State
region and does not address Florida.) The air quality estimates derived from
these models are key inputs to the economic benefits calculations presented
in Chapter VI below.
The 1980 Baseline sulfate estimates are displayed in Table IV.0.8. In
general the MONTE CARLO model produces the highest S04 estimates and ENAMAP
the lowest. However, as can be seen there is considerable variation from
state to state and the ordering of the models can change. Table IV.0.9
-------
IV-45
STATE
Table IV.D.8
1980 Baseline Sulfate Estimates
n3
ASTRAP
tug/fir* S04)
MONTE CARLO
ENAMAP
RTM-LT
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Mai ne
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshi re
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
West Virginia
Wisconsin
4.4
2.3
8.4
11.3
3.3
5.1
5.5
8.1
2.3
8.5
1.8
4.0
10.6
6.9
4.3
1.1
2.8
3.4
6.1
10.5
7.8
7.5
9.5
11.3
7.4
6.2
6.6
5.7
9.0
11.2
2.6
9.9
6.2
8.9
9.7
4.2
8.6
7.3
9.7
3.6
12.1
5.0
6.1
9.7
8.9
4.2
2.6
6.6
6.3
7.9
10.7
10.7
10.4
10.9
11.5
8.9
10.1
10.4
7.9
11.3
13.9
3.4
2.8
1.6
4.7
7.7
2.8
4.8
5.1
6.7
1.6
6.3
1.6
3.5
7.7
4.7
3.9
0.5
1.6
3.3
5.8
7.7
9.5
6.3
10.4
10.5
4.7
4.8
5.5
5.8
6.3
8.6
1.6
6.6
6.1
6.6
8.6
-
7.3
6.4
8.3
4.6
8.5
5.7
4.3
8.9
6.0
6.4
3.1
5.7
6.2
5.4
7.9
7.1
8.1
8.5
8.6
6.3
7.5
7.8
5.2
8.5
8.8
4.6
-------
Table IV.D.9
1995 Base Case Sulfate Estimates
(pg/m3 S04 and % change from 1980 Baseline Estimate)
STATE
ASTRAP
MONTE CARLO
ENAMAP
RTM-LT
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
West Vi rginia
Wisconsin
pg/nr
5.2
3.1
8.2
11.5
4.0
5.7
5.5
8.1
2.7
8.5
2.6
3.6
10.9
6.5
4.5
1.4
3.7
4.0
5.6
10.5
7.5
7.8
9.6
11.3
7.1
6.8
6.9
5.1
9.3
11.5
3.1
% change
18
35
-2
2
21
12
0
0
17
0
44
-10
3
-6
5
27
32
18
-8
0
-4
4
1
0
-4
10
5
-11
3
3
19
pg/rn^
10.9
7.3
8.6
9.9
4.5
9.4
7.7
10.0
4.1
12.5
6.1
5.8
9.9
8.6
4.5
2.9
7.9
7.2
7.4
10.3
10.3
10.9
11.2
11.7
8.6
10.8
10.9
7.4
11.8
14.5
3.9
% change
10
18
-3
2
7
9
5
3
14
3
22
-5
2
-3
7
12
20
14
-6
-4
-4
5
3
2
-3
7
5
-6
4
4
15
pg/m
3.1
1.8
4.6
8.0
3.3
5.3
5.0
6.5
1.8
6.2
1.8
3.4
8.0
4.6
3.8
0.6
1.8
3.3
5.7
8.0
9.1
6.8
10.2
10.5
4.6
5.3
5.4
5.7
6.8
8.8
1.8
% change
11.0
12.5
-2.1
3.9
17.8
10.4
-1.9
-2.9
12.5
-1.6
12.5
-2.9
3.9
-2.1
-2.6
20.0
12.5
-0
-1.7
3.9
-4.2
7.9
-1.9
-0
-2.1
10.4
-1.8
-1.7
7.9
2.3
12.5
pg/m
6.7
6.1
6.7
8.7
7.4
6.3
8.1
4.6
8.2
5.7
4.2
8.9
5.9
6.5
3.1
5.8
6.1
5.2
8.1
7.2
8.2
8.5
8.8
6.3
7.6
7.5
5.0
8.6
8.9
4.8
% change
1.5
0.3
1.1
1.0
1.5
-1.3
-2.2
0.2
-3.1
0.5
-3.0
-0.1
-2.0
1.2
0.6
2.0
-2.2
-3.5
1.9
0.8
1.4
0.4
2.2
0.2
1.2
-3.5
-3.8
1.5
1.4
3.5
I
-p-
-------
Table IV.0.10
1995 Sulfate Estimates
Strict Interpretation of Current NAAQS
(ug/m3 and % change from 1995 Base Case)
STATE
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Caroli na
Tennessee
Vermont
Virginia
West Virginia
f Wisconsin
ASTRAP
ug/rn^
4.6
2.9
7.6
10.7
3.6
5.1
4.6
7.0
2.5
7.2
2.4
3.4
10.0
6.1
4.1
1.4
3.3
3.5
5.3
9.8
6.9
7.0
8.4
10.3
6.6
6.2
5.9
4.8
8.3
10.1
2.9
% change
-12
- 6
- 7
- 7
-10
-11
-16
-14
- 7
-15
- 8
- 6
- 8
- 6
- 9
- 0
-11
-13
- 5
- 7
- 8
-10
-12
- 9
- 7
- 9
-14
- 6
-11
-12
- 6
MONTE CARLO
ug/nP
9.8
6.6
8.1
9.3
4.2
8.5
6.9
8.9
3.8
11.0
5.7
5.5
9.3
8.1
4.2
2.8
7.2
6.4
7.0
9.5
9.5
10.0
10.0
10.7
8.1
9.9
9.5
7.0
10.8
13.1
3.8
% change
-10
-10
- 6
- 6
- 7
-10
-10
-11
- 7
-12
- 7
- 5
- 6
- 6
- 7
- 3
- 9
-11
- 5
- 8
- 8
- 8
-11
- 9
- 6
- 8
-13
_ 5
- 8
-10
- 3 .
ENAMAP
ug/m3
2.7
1.5
4.2
7.4
3.0
4.7
4.2
5.6
1.6
5.3
1.5
3.2
7.4
4.2
3.4
0.5
1.5
2.6
5.2
7.4
8.2
6.1
8.9
9.4
4.2
4.7
4.5
5.2
6.1
7.8
1.6
% change
-13.8
-14.8
- 6.8
- 7.8
-10.0
-11.5
-16.5
-14.0
-10.3
-15.3
-14.8
- 7.5
- 7.8
- 6.8
-10.0
- 4.2
-14.8
-21.2
- 8.4
- 7.8 ' "
- 9.3
- 9.3
-12.8
- 9.9
- 6.8
-11.5
-17.1
- 8.4
- 9.3
-11.3
-10.3
RTM-LT
O ' • '
ug/nr
6.5
6.0
6.4
8.3
-
7.0
6.1
7.6
4.6
7.6
5.8
4.1
8.4
5.7
6.3
3.1
5.7
5.8
5.1
7.8
6.9
7.8
8.0
8.2
6.1
7.3
7.0
4.9
8.1
8.2
4.8
% change
-6.8
-2.1
-5.2
-6.2
-
-7.2
-4.6
-6.4
-0.2
-9.5
-0.3
-3.7
-7.0
-4.2
-3.4
0.0
-2.0
-5.0
-4.2
-5.4
-5.1
-7.2
-7.3
-7.3
-4.8
-6.8
-9.1
-3.9
-7.2
-8.5
-1.9
-------
Table IV.D.ll
1995 Sulfate Estimates
0.5 ppm 1-hour Alternative
(pg/m3 and % change from 1995 Base Case)
STATE ASTRAP MONTE CARLO ENAMAP
ug/m3 % change pg/rn^ % change pg/m^ % change pg/m^
Alabama 4.3 -17 9.1
-17 2.4 -21.5 6.2
Arkansas 2.8 -10 6.3 -14 1.4 -22.6 5.9
Connecticut 7.0 -15 7.6 -12 3.8 -15.9 6.0
Delaware 9.5 -17 8.5 -14 6.6 -17.9 7.6
RTM-LT
% change
-11.2
- 3.8
-10.8
-13.6
Florida 3.3 -18 3.9 -13 2.8 -16.9
Georgia 4.7 -18 7.9 -16 4.4 -18.3 6.6
Illinois 4.2 -24 6.5 -16 3.7 -25.6 5.9
Indiana 6.1 -25 8.0 -20 4.9 -25.4 7.2
Iowa 2.4 -11 3.8 - 7 1.5 -15.9 4.6
Kentucky 6.3 -26 9.8 -22 4.5 -27.1 6.9
Louisiana 2.4 -8 5.6 -8 1.4 -22.6 5.7
Maine 3.2 -11 5.2 -10 2.8 -17.5 4.0
Maryland 8.8 -19 8.5 -14 6.6 -17.9 7.8
Massachusetts 5.6 -14 7.6 -12 3.8 -15.9 5.4
Michigan 3.8 -16 4.0 -11 3.1 -18.1 6.0
Minnesota 1.4 - 0 2.8 - 3 0.5 - 6.5 3.1
Mississippi 3.1 -16 6.9 -13 1.4 -22.6 5.6
Missouri 3.2 -20 6.0 -17 2.3 -30.0 5.7
New Hampshire 4.9 -12 6.6 -11 4.7 -18.2 4.9
New Jersey 8.8 -16 8.8 -15 6.6 -17.9 7.3
New York 6.3 -16 8.8 -15 7.4 -18.2 6.5
North Carolina 6.3 -19 9.1
Ohio 7.4 -23 9.1
Pennsylvania 9.1 -19 9."
-17 5.5 -18.3 7.2
-19 7.9 -23.3 7.5
' -17 8.3 -20.4 7.7
Rhode Island 6.0 -15 7.6 -12 3.8 -15.9 5.8
South Carolina 5.7 -16 9.1
-16 4.4 -18.3 6.8
Tennessee 5.3 -23 8.6 -21 3.9 -27.9 6.5
Vermont 4.6 -10 6.e
Virginia 7.3 -22 9.1
) -11 4.7 -18.2 4.7
' -18 5.5 -18.3 7.5
West Virginia 8.9 -23 11.7 -19 6.8 -22.5 7.6
Wisconsin 2.9 - 6 3./
' - 5 1.5 -15.9 4.7
-12.1
- 7.4
-11.9
- 0.9
-16.8
- 1.0
- 7.5
-14.0
- 9.1
- 8.1
- 0.3
- 4.9
- 7.3
- 8.3
-12.2
-11.0
-13.8
-13.1
-13.8
-10.2
-12.7
-15.6
- 8.1
-14.2
-15.5
- 3.5
I
-p-
oo
-------
Table IV.D.12
1995 Sulfate Estimates
0.25 ppm Alternative NAAQS
(ug/m3 and % change from 1995 Base Case)
STATE
Al abama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Mi chigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Vi rginia
West Vi rginia
Wisconsin
ug/m3
3.4
2.6
5.6
7.1
2.7
3.5
3.5
4.7
2.3
4.7
2.2
2.8
6.5
4.6
3.1
1.3
2.7
2.9
4.2
6.9
5.0
4.6
5.5
6.8
4.8
4.2
4.1
3.9
5.3
6.3
2.5
ASTRAP
% change
-35
-16
-32
-38
-33
-39
-36
-42
-15
-45
-15
-22
-40
-29
-31
- 7
-27
-28
-25
-34
-33
-41
-43
-40
-32
-38
-41
-24
-43
-45
-19
MONTE
ug/m3
6.9
5.1
6.4
6.7
3.3
6.0
5.2
6.2
3.3
7.1
4.9
4.6
6.7
6.4
3.5
2.6
5.7
4.6
5.8
7.2
7.2
6.9
6.9
7.5
6.4
6.9
6.2
5.8
7.3
8.4
3.4
CARLO
% change
-37
-30
-26
-32
-27
-36
-32
-38
-20
-43
-20
-21
-32
-26
-22
-10
-28
-36
-22
-30
-30
-37
-38
-36
-26
-36
-43
-22
-38
-42
-13
ENAMAP
ug/m3 %
2.0
1.1
3.0
4.9
2.3
3.3
3.0
3.7
1.2
3.3
1.1
2.1
4.9
3.0
2.5
0.5
1.1
1.9
3.5
4.9
5.7
4.1
5.8
6.1
3.0
3.3
3.0
3.5
4.1
4.8
1.2
change
-36.8
-36.6
-34.8
-39.0
-32.4
-38.1
-39.8
-43.3
-30.3
-46.1
-36.6
-37.9
-39.0
-34.8
-33.2
-13.9
-36.6
-42.3
-39.0
-39.0
-37.7
-39.7
-43.6
-42.1
-34.8
-38.1
-44.8
-39.0
-39.7
-45.6
-30.3
ug/m3
5.4
5.7
5.2
6.3
-
5.6
5.4
6.3
4.5
5.9
5.6
3.6
6.4
4.7
5.4
3.1
5.2
5.4
4.3
6.1
5.6
5.9
6.4
6.4
5.0
5.6
5.6
4.2
6.1
6.1
4.4
RTM-LT
% change
-22.8
- 6.4
-22.9
-28.8
-
-25.8
-14.5
-22.1
- 3.3
-29.7
- 2.9
-15.9
-29.4
-20.2
-17.3
- 1.0
-11.1
-11.2
-18.3
-26.1
-22.5
-30.2
-26.0
-28.0
-21.7
-28.5
-27.6
-17.3
-30.4
-31.8
- 9.7
<
-------
IV-5U
presents the 1995 Base Case sulfate estimates (i.e. the estimated sulfate
levels associated with normal growth and retirement of sources but no change
in the NAAQS). As can be seen all of the models predict an increase in $04
levels for most states, with only a few states experiencing minor decreases.
It should also be noted that, in general, RTM-LT predicts a smaller percent
change than the other models. This pattern is also seen in Tables IV.Q.10
through 12. It is believed that the relative insensitivity of RTM-LT to
emission changes results from the large background term used to account for
Western U.S. emissions. In the other models Western U.S. emissions were
modeled directly. As discussed in Section IV.C.5 above, this and other
problems with RTM-LT led to its being dropped in this analysis as the basis
for benefits calculations.
Tables IV.D.10 through 12 report the model results for the three NAAQS
cases analyzed. All of the alternatives lead to substantial estimated $04
changes. However, the "strict interpretation" of the current NAAQS does not
result in a large enough estimated decrease to regain 1980 levels for some
model/state combinations. The 1-hour NAAQS alternatives result in larger
changes and, in most instances, an improvement over the 1980 baseline.
The estimated sulfate levels were in turn used to calculate visual range.
The RTM-LT results were used to compile a 1980 baseline. Although various
reviewers have pointed out flaws in the RTM-LT visual range calculation, the
procedure did produce results which very closely agreed with actual 1980
readings. Since these data were in the needed format and since other problems
precluded obtaining actual readings, the 1980 RTM-LT baseline was used as an
input to the calculations associated with the other models as well. Although
the use of the actual readings would have been preferrable it is felt that
the usd of RTM-LT data did not unduly bias the final results. Table IV.D.13
presents the 1980 Baseline and the 1995 Base Case for RTM-LT, ASTRAP, and
-------
Table IV.D.13
Estimated 1980 Baseline Visibility and
Estimated 1995 Baseline Visibility
(Median Annual Visual Range in km)
STATE
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
West Virginia
Wisconsin
1980*
Baseline
14.4
14.6
27.2
25.1
18.9
17.8
17.7
16.8
20.7
18.7
14.6
33.5
25.0
28.5
18.8
24.5
14.8
18.0
30.2
25.3
26.3
17.1
16.8
19.1
27.6
17.5
15.7
30.4
24.7
19.8
20.0
RTM-LT
14.3
14.6
27.0
24.8
-
17.5
17.7
16.9
20.7
18.8
14.6
33.8
24.8
28.6
18.6
24.5
14.6
18.2
30.5
24.9
26.0
16.9
16.7
18.8
27.5
17.3
15.7
30.7
24.3
19.6
19.7
1995
ASTRAP2
13.3
12.4
27.5
24.9
17.1
16.8
17.7
16.8
19.0
18.7
11.9
35.3
24.7
29.4
18.4
21.6
12.8
16.5
31.5
25.3
26.8
16.8
16.7
19.1
28.2
16.7
15.4
32.1
24.3
19.5
18.2
MONTE CARLO
13.9
13.6
27.6
24.9
18.4
17.2
17.3
16.6
19.6
18.5
13.4
34.2
24.8
28.9
18.3
23.4
13.7
17.0
31.0
25.7
26.7
16.8
16.6
19.0
28.0
17.0
15.4
31.2
24.3
19.5
18.9
-------
Strict
(Median Annual
Table IV.D.14
1995 Estimated Visibility
Interpretation of Current NAAQS
Visibility km and % from 1995 Base
Case)
STATE
ASTRAP1
MONTE CARLO2
AVERAGE3
,
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Vi rginia
West Virginia
Wisconsin
^-Assumes a non-sui
Visibility
14.1
12.9
28.6
25.8
18.0
17.7
19.3
18.0
19.8
20.2
12.4
36.3
25.7
30.3
19.2
21.6
13.5
17.6
32.4
26.2
27.9
17.7
17.8
20.0
29.2
17.5
16.6
33.1
25.7
20.8
18.9
Ifate/sulfate
% change
6
3
4
4
5
6
9
7
4
8
4
3
4
3
5
0
6
7
3
3
4
5
7
5
4
5
8
3
6
6
3
ratio of 1 .0.
Visibi lity
14.5
14.2
28.2
25.5
18.9
17.8
18.1
17.4
20.2
19.4
13.8
34.9
25.4
29.6
18.8
23.7
14.2
17.8
31.7
26.5
27.6
17.4
17.4
19.6
28.6
17.6
16.2
31.9
25.1
20.2
19.1
% change
4
4
2
2
3
4
4
5
3
5
3
2
2
2
3
1
4
5
2
3
3
3
4
4
2
3
5
2
4
4
1
Visibility
14.4
13.8
28.4
25.6
18.5
17.8
18.5
17.6
20.0
19.7
13.4
35.4
25.5
29.9
19.0
23.1
14.0
17.8
32.0
26.4
27.7
17.5
17.6
19.8
28.9
17.6
16.4
32.3
25.3
20.5
19.0
% change
5
4
3
3
4
5
6
6
3
6
3
2
3
3
4
1
4
5
2
3
4
4
5
4
3
4
6
3
4
5
2
^Assumes a non-sulfate/sulfate ratio of 1.5.
3Employs the average of the state level ASTUAP
non-sulfate/sulfate ratio of 1.25.
and MONTE CARLO sulfate estimates and assumes a
i
Ul
-------
Table IV.D.15
1995 Estimated Visibility
0.5 ppm 1-Hour Alternative NAAQS
(Median Annual Visibility km and % from 1995 Base Case)
STATE
ASTRAP1
MONTE CARLO2
AVERAGE3
Alabama
Arkansas
Connecticut
Delaware
Florida
Georgia
Illinois
Indiana
Iowa
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
South Carolina
Tennessee
Vermont
Virginia
West Virginia
Wisconsin
Visibility
14.6
13.1
29.7
27.2
18.7
18.4
20.1
19.2
20.2
21.5
12.4
37.3
27.3
31.5
19.9
21.6
13.9
18.4
33.6
27.5
29.1
18.5
18.9
21.2
30.5
18.2
17.4
33.7
27.2
22.0
18.9
% change
9
5
8
10
10
10
13
14
6
15
4
6
11
7
8
0
9
11
7
9
9
11
13
11
8
9
13
5
12
13
3
Visibility
14.9
14.4
28.9
26.4
19.4
18.3
18.5
18.0
20.2
20.2
13.9
35.6
26.3
30.3
19.1
23.7
14.5
18.2
32.4
27.3
28.4
18.0
18.0
20.4
29.3
18.2
16.8
32.6
26.1
21.1
19.3
% change
7
6
5
6
6
7
7
9
3
9
3
4
6
5
5
1
5
7
5
6
6
7
8
7
5
7
9
5
8
8
2
Visibility
14.8
14.1
29.3
26.8
19.1
18.4
19.1
18.5
20.2
20.7
13.5
36.3
26.7
30.8
19.5
23.1
14.3
18.3
32.9
27.4
28.7
18.2
18.3
20.7
29.8
18.2
17.0
33.1
26.6
21 .5
19.1
% change
8
6
6
8
7
8
9
11
4
12
4
5
8
6
6
1
7
9
5
7
7
9
10
9
6
8
11
5
9
10
3
lAssumes a non-sulfate/sulfate ratio of 1.0.
2Assumes a non-sulfate/sulfate ratio of 1.5.
3Employs the average of the state level ASTRAP
non-sulfate/sulfate ratio of 1.25.
and MONTE CARLO sulfate estimates and assumes a
<
CO
-------
Table IV.D.16
1995 Estimated Visibility
0.25 ppm 1-Hour Alternative NAAQS
(Median Annual Visual Range - km and % change from 1995 Base Case)
STATE
ASTRAP1
MONTE CARLO2
AVERAGE3
V i s i b i 1 i ty
Alabama 16.1
Arkansas 13.5
Connecticut 32.7
Delaware 30.8
Florida 20.4
Georgia 20.8
Illinois 21.6
Indiana 21.3
Iowa 20.6
Kentucky 24.1
Louisiana 12.9
Maine 39.7
Maryland 30.9
Massachusetts 34.4
Michigan 21.8
Minnesota 22.4
Mississippi 14.7
Missouri 19.2
New Hampshire 36.0
New Jersey 30.5
New York 32.2
North Carolina 21 .1
Ohio 21.2
Pennsylvania 23.8
Rhode Island 33.6
South Carolina 20.6
Tennessee 19.3
Vermont 36.4
Virginia 31.0
West Virginia 25.2
Wisconsin 20.2
^-Assumes 4 non-sulf ate/sulf ate
^Assumes a nor\-sulf ate/sulfate
^Employs the average of the sta
i non-sulfate/sulfate ratio of 1
% change
21
9
19
24
19
24
22
27
8
29
8
13
25
17
18
4
16
16
14
21
20
26
27
25
19
24
25
13
27
29
11
ratio of 1 .0.
ratio of 1.5.
te level ASTRAP
.25.
Visibility
16.3
15.5
30.7
28.6
20.6
20.1
19.9
19.6
21.3
22.3
14.6
37.3
28.5
32.2
20.1
24.4
15.4
19.9
33.9
29.2
30.4
19.7
19.6
22.1
31 .2
19.9
18.6
34.1
28.6
23.4
19.9
and MONTE CARLO
% change
17
14
11
15
12
17
15
18
8
21
9
9
15
11
10
4
13
17
9
14
14
17
18
17
11
17
21
9
18
20
5
sulfate estimates
Visibility %
16.3
15.0
31.6
29.6
20.5
20.4
20.6
20.3
21.0
23.0
14.1
38.2
29.5
33.1
20.8
23.8
15.3
19.7
34.8
29.8
31.1
20.2
20.3
22.9
32.2
20.2
18.9
35.0
29.6
24.2
20.0
and assumes a
change
19
13
15
19
15
20
18
21
9
24
9
10
19
14
13
4
14
17
11
17
16
21
22
20
15
20
23
11
22
24
8
Ul
-p-
-------
IV-55
MONTE CARLO. A review of the results shows that RTM-LT predicts a smaller
change than the other models. For reasons noted previously RTM-LT results
were not used in benefits calculations and are therefore not included in
subsequent tables (N.B. Because the ENAMAP results are based on only two
months of meteorology, no visual range changes were calculated from them).
Tables IV.0.14 through 16 present the estimated visibility associated
with the three alternative NAAQS cases analyzed. Each table reports results
for ASTRAP, MONTE CARLO, and an "AVERAGE." The AVERAGE case employs the
average of the state level ASTRAP and MONTE CARLO 804 estimates and a non-
sulfate/sulfate ratio of 1.25. This case was constructed to help present
a balanced benefits analysis.
The analysis also produced estimates of the amount of scrubber sludge
produced. These estimates are displayed in Table IV.0.17. As can be seen,
the current NAAQS and the 0.5 ppm alternative produce relatively small
increases over the 1995 base case. However, the 0.25 ppm alternative
nearly doubles the amount of scrubber sludge compared to 1995 Base Case
projections. This results from increased reliance on scrubbers as a control
strategy (see Table IV.0.6).
-------
IV-56
Table IV.D.I7
Estimated Scrubber Sludge Production
1995
(millions tons per year)
New England
Middle Atlantic
Upper South Atlantic
Lower South Atlantic
East North Central
East South Central
West Bank of Mississippi
West North Central
West South Central
Mountain
Pacific
Total U.S.
Base Case
1995
0.6
3.5
0.6
0.7
3.3
4.6
0.9
1.1
5.3
2.5
0.3
23.4
Changes from Base
Current
NAAQS
0.5ppm-lhr
NAAQS
0.25ppm-lhr
NAAQS
+0.1
+0.2
-
+0.6
+1.1
-0.2
+0.3
-
-
+0.1
-
+2.2
+0.1
-0.1
+0.5
+1.0
+ 1.2
+1.0
+0.4
-
-
+0.2
-
+4.3
+2.3
+2.5
+4.7
+8.2
+1.3
+ 1.6
+0.1
+0.2
+0.4
+0.3
+21.6
-------
REFERENCES
Anderson, Gerald E., et al. (1984). Estimation of Power Plant Impacts and
Emission Limits at Various Averaging Times, final report, Systems
Applications Inc. August 1984.
Bachrnann, John (1985). Memorandum to Tom Walton. Subject: Calculation of
Visual Range Changes from Predicted Sulfate Changes. June 10, 1985.
Braine, bruce (1984). Memorandum to Henry C. Thomas. Subject: Coverage
of Powerplant Capacity from the STACK file. January 30, 1984.
Brubaker, K.L. and A.E. Smith (1984). Air Quality Analyses in Support of
Regulatory Impact Analysis for S02 NAAQS: Technical Notes, draft
report, Argonne National Laboratory, August 3, 1984.
E.H. Pech'an & Assoc. (1981). Development of an Electric Utility Stack
Inventory for Sulfur Oxides Analyses, final report, September 1981.
Frank, N.H. and A.D. Thrall (1982). Relationships Among S02 Averaging
Times and Ambient Standards, MDAD Report, Office of Air Quality
Planning and Standards, August 1982.
Hayes, S.R., et al. (1984). An Evaluation of Alternative Formulations of
the National Ambient Air Quality Standards for S02, paper presented
at the 77th Annual Meeting of APCA, June 1984.
ICF Inc. (1984). Analysis of Alternative Sulfur Dioxide Ambient Standards,
final draft report, August 1984.
Mage, David T. (1982). Emissions Limits for Variable Sources by Use of
Multipoint Rollback, Discussion, Atmospheric Environment Vol. 16
pp. 1273-1274, 1982.
Memorandum of'Intent on Transboundary Air Pollution (1982). Atmospheric
Sciences and Analysis Work Group 2 Final Report, November 1982.
-------
PEDCo Environmental, Inc. (1984). Costs Incurred by Primary Copper Smelters
to Meet Selected Ambient Air S02 Standards, final draft report,
hay 1984.
Peterson, T.W. and J. L. Moyers (1980). Emission Limits for Variable
Sources by Use of Multi Point Rollback, Atmospheric Environment
Vol. 14, pp. 143-1444, 1980.
Peterson, T.W., and Jeffery Hagen (1983). The Effect of A Proposed One-Hour
Standard for SOg on the Ultimate Emission Limits for Western Smelters,
final report, Dept. of Chemical Engineering, University of Arizona,
September 1983.
Radian Corp. (1984). Use of S02 Controls at Lead Smelters to Comply with
Short-term SO^ Standards, draft final report, July 1984.
Systems Applications, Inc. (1982). On a New Short-term Standard for Sulfur
Dioxide: The Protection From Calculated Peak One-Hour Concentrations
Provided by the Existing Three- and 24-hour NAAQS in the Vicinity of
a Hypothetical Power Plant, final report, August 1982.
Systems Applications, Inc. (1983). Protection Provided by Existing Short-
term NAAQS from Calculated Peak 1-hour Concentrations in the Vicinity
of Five Power Plants, draft report, September 1983.
TRU Inc. (1981). Analysis of Ambient Short-term SQ2 Impacts from Petroleum
Refineries, draft report, 1981.
UDI (1982). Letter from Chris Bergesen to E.H. Pechan November 9, 1982.
-------
V. ECONOMIC IMPACTS
A. Introduction
This section of the RIA reports on the economic effects of the
direct costs of control reported in Section IV. The economic impact
assessment below is limited in several respects. In the first place, it
focuses on the utility sector with little treatment of other industrial
categories. Specifically, due to current industry economic conditions,
and substantial uncertainties over both the degree and timing of industry
compliance, no economic impact assessment was performed for the smelters.
The uncertainty over the timing and degree of compliance arises from the
fact that the copper smelters were allowed additional time to comply with
the Clean Air Act (§119). As has been noted above, reliance on process
modification as a control strategy introduces additional uncertainty. Speci-
fically, although such controls would result in both attainment of the
standards and a reduction in production costs, the initial capital costs
may make them economically unfeasible in many cases. Following an examination
of the industry by industry costs associated with industrial boiler costs,
it was determined that an economic impact assessment was not warranted.
Secondly, the analysis is limited with respect to the types of economic
effects that are examined. This second limitation results in part from
the way in which the utility industry costs were assessed. In particular,
assumptions in the CEUM model preclude the analysis of certain effects.
B. Utilities
As discussed in Section IV.C.I, ICF's CEUM model was used to
estimate direct costs for the utility sector. CEUM attempts to project
coal supply and coal demand (both utility and non-utility) and then
balance the two at least cost through standard linear programming techniques.
-------
V-2
Although CEUM is the only model presently available capable of providing
a systematic national-scale cost analysis, it has a number of features
which constrain the scope of economic impact analyses on costs it estimates.
Several of the more important features are discussed below:
Demand for Electricity: As noted in Section IV.C.I, demand for
electricity is estimated outside of CEUM and is specified as an input
to the model. All cost increases will be passed on by the utilities
to the consumers. Furthermore, the model assumes an inelastic demand
for energy. The assumption of perfectly inelastic demand for energy
implies that there will be no change in energy output as a result of
the rate increases discussed below in this Section. A number of
studies have indicated that energy demand is not perfectly inelastic.
However, development of an energy demand curve would have implied an
industry by industry examination of demand. Such an effort was beyond
the scope of this RIA. It should be recognized that the assumption
of inelastic demand results in a bias in the results. It is likely
to result in an overstatement of price increases. It also implies
that changes in direct employment in the utility industry cannot be
correctly estimated.
Although the development of an actual energy demand curve was
beyond the scope of this analysis, a sensitivity analysis was
performed. A number of studies have given a range of estimated
elasticities (-0.04 to -0.49 for short run and -0.45 to -1.89 for
long-run) (Bohi, 1981). For the purposes of this sensitivity analysis
a long-run elasticity at the mid-point of the range or -1.2 was
assumed. Using the percent changes in electricity rates shown in
Table V.B.2 below state by state percent changes in demand were then
-------
V-3
calculated. It is, of course, difficult if not impossible to. calculate
the impact such changes in demand will have on emissions. The CEUM
model incorporates a great deal of data and assumptions regarding
dispatch of plants, (i.e., the order in which plants are brought on-
line to satisfy demand, which are base load and which are peaking).
A change in demand would undoubtedly change the dispatch order.
These changes in dispatch would, therefore, also indicate whether
the percent change in emissions would be greater or lesser than the
percent change in demand. Since a detailed analysis of dispatch was
not possible it was simply assumed that the change in emissions
would equal the change in demand. The estimated reductions in
emissions (OOO's TRY) given these assumptions were:
Current NAAQS 0-160
0.5 ppm - 1-hour NAAQS 290
0.25 ppm - 1-hour NAAQS 530
These reductions are in addition to those shown in Table IV.D.5 and
were calculated for the 31-Eastern States only. As can be seen, these
reductions, although significant on an absolute basis, are still an
order of magnitude less than those in Table IV.D.5. It should be
kept in mind that these results are shown only as a sensitivity analysis,
The considerable uncertainties regarding the estimated elasticity of
demand and the relationship of demand to emissions imply that the
results do not give an accurate estimate, but do give an indication
of the sensitivity of the analysis to. the assumption of inelasticity.
Industry Supply: Interest rates and capital availability are
specified exogenously in the ICF model and do not change from one
alternative standard to another. The implication of this is that
the utilities can acquire additional capital at rates equivalent to
-------
V-4
the projected weighted average cost of capital. The large capital require-
ments associated with the 0.2b ppm alternative may make this assumption
doubtful. The total capital expenditures of the industry are compared
to the increases associated with these standards in a following section.
The cost of new supplies of cleaner coals is equated to the cost
of production and accounts for depletion effects. This implies that
increased demand for low sulfur coal will not by itself raise the price.
This could possibly have resulted in an overestimate of the degree
to which fuel switches will be used in place of control equipment.
Despite these limitations the data presented below do help to capture
the expected economic impacts of the alternatives examined.
Utility Rate Impact: Tables V.B.I presents the estimated absolute
change in utility rates based on the estimated annualized costs in
1995. These changes were calculated by subtracting the base case
annualized costs from the annualized costs for the standard being
analyzed and dividing by projected electricity sales. Table V.B.2
displays the same rate changes expressed as a percent of 1980 average
residential electricity. In general the rate changes are less than
1% in the current standards case. The 0.5 ppm alternative resulted
in rate changes in a few states which approached 5.0%. However, in
most states the rate changes were less than 3%. In the 0.25 ppm
alternative rate changes approached 10% in one state and were above
5% in 8 states. It must be kept in mind that these estimates were
derived from a model which assumes an inelastic demand for electricity.
The estimated rate changes for the current standards case would be
unlikely to result in any change in output. However, in the two one
hour alternative cases the rate changes are significant enough in
some states that they might result in lowered demand.
-------
V-5
Table V.B.I
Absolute Change in Electricity Rates Based on
Annualized Costs in 1995
(1984 mills/kwh)
Changes from Base
State or CE'UM Region
Maine/Vermont/New Hampshire
Massachusetts/Connecticut/Rhode Isl and
New York
Pennsylvania
New Jersey
Maryland/Delaware/District of Columbia
Virginia
West Virginia
North and South Carolina
Georgia
Florida
Ohio
Michigan
Indiana
Illinois
Wisconsin
Kentucky
Tennessee
Al abama
Mississippi
Minnesota
Iowa
Mi ssouri
Arkansas
Louisiana
Current
NAAQS
_
0.5
0.3
-
-
0.1
0.5
0.4
0.1
0.3
0.2
0.9
0.1
0.4
0.3
0.1
0.8
0.6
0.2
0.4
-
0.1
2.5
_
0.5ppm-lhr.
NAAQS
_
0.8
0.6
0.9
0.6
0.7
1.9
2.4
0.3
0.5
0.5
1.2
0.6
1.2
0.6
-
1.7
1.8
0.1
0.8
0.3
1.5
2.3
0.2
0.1
0.25ppm-lhr-
NAAQS
0.3
2.1
1.9
3.3
1.7
2.9
3.8
5.1
2.0
2.9
1.6
4.5
3.3
1.1
2.9
1.5
3.8
3.2
0.8
2.4
-
2.1
5.1
0.1
0.1
1/Calculated as follows:
1995 Ambient Standard Case Annualized Cost-
1995 Base Case Annualized Cost
1995 Electricity Sales
-------
V-6
Table V.B.2
Percent Change in Electricity Rates Based on
Annualized Costs in 1995
(percent)
Changes from Base
State or CEUM Region
Maine/Vermont/New Hampshire
Massachusetts /Connect! cut /Rhode Island
New York
Pennsylvania
New Jersey
Maryland/Delaware/District of Columbia
Virginia
West Virginia
North and South Carolina
Georgi a
Florida
Ohio
Michi gan
Indiana
Illinois
Wisconsi n
Kentucky
Tennessee
Al abama
Mississippi
Mi nnesota
Iowa
Missouri
Arkansas
Loui siana
Averge Change
I/Calculated as follows:
1995 Ambient Standard Case Annualized
1995 Base Case Annualized Cost
1995 Electricity Sales
Current
NAAQS
—
0.6
0.3
-
-
0.2
0.8
0.8
0.2
0.6
0.4
1.3
0.1
0.8
0.4
0.1
1.7
1.5
0.3
0.8
-
0.2
4.3
_
-
0.4
Cost-
.
.
0.5ppm-lhr. 0
NAAQS
—
1.0
0.6
1.4
0.6
1.1
2.8
4.7
0.6
0.9
0.8
1.8
0.9
2.2
0.8
-
3.8
4.4
0.2
1.4
0.6
2.4
3.9
0.4
0.2
1.3
1980 Average
Residential
Electricity
.25ppm-lhr
NAAQS
0.4
2.4
2.0
4.8
1.7
4.2
5.6
9.9
3.9
5.3
2.4
7.0
4.9
5.5
1.6
2.7
8.2
7.7
1.4
4.4
-0.1
3.5
8.8
0.3
0.1
3.8
Rates
-------
V-7
Comparison to Total I-ndustry Capital Costs and Revenues: To place the
costs associated with these standards in perspective, Table V.B.3
presents total industry revenues and capital costs for the years
1980-82. These data were obtained from industry publications and are
stated in 1984 dollars. For comparison, summary cost information on
the standards analyzed is shown in Table V.B.4. Several observations
can be made regarding these Tables. Focusing on the capital costs,
both the current and 0.5 ppm NAAQS result in comparatively moderate
increases over present capital expenditures. -Compared to 1981 capital
expenditures the current NAAQS might require a 0-1% increase; while
the 0.5 ppm alternative shows an approximate 4.8% increase. On the
other hand, the 0.25 ppm is estimated to result in a 34% increase.
Although for analytic purposes the ICF model assumes that capital
costs are incurred in the year of attainment (1990), in reality it is
quite likely that these costs would be spread over two to three years.
This spreading of capital costs over several years would soften the
impact of the standard on utilities. Nonetheless, a 0.25 ppm NAAQS
could be expected to have a significant impact on the industry. Due
to the assumptions of the cost model (perfectly elastic capital supply
curves) and the fact that neither of the 1-hour alternatives is being
proposed, the result of these impacts is not assessed.
Tables V.B.3 and V.B.4 also display total revenues (1980-82) and
incremental annualized cost. The annual costs divided by total revenues
closely approximates the forecasted average percent change in electricity
rates in the U.S. Of course these are average changes and the change
can be more significant in some states and regions as was shown in
Table V.B.2.
-------
V-8
Table V.B.3
Utility Industry Revenues and
Capital Expenditures
($ billions, 1984)
1980 1981 1982
Total Revenues1 $118.6 $125.6 $129.7
Total Capital Expenditures2 $ 44.1 $ 46.3 $ 41.1
^Source: Edison Electric Institute
^Source: Electrical World, September 1983
Table V.B.4
Total Estimated Utility Cost Increases
1995
($ billions)
0.5 ppm 0.25 ppm
Current NAAQS 1-hour NAAQS 1-hour NAAQS
Increase in Annual Costs $0-0.7 $1.8 $ 5.0
Increase in Capital Costs $0-0.4 $2.2 $15.9
Average Percent Change 0.4% 1.0% 3.0%
in Electricity Rates
-------
V-9
Coal Production:
Table V.B.5 summarizes the national impacts of these alternatives on
coal production. The table indicates, as expected, that production
in higher sulfur regions (i.e., Midwest and Northern Appalachia) will
likely be curtailed, while production in lower sulfur regions (i.e.,
Central Appalachia and the West) will likely increase. In some
regions (e.g., the Midwest) coal production does not fall as significantly
as might be expected between the 0.5 ppm and 0.25 ppm alternatives.
This is due to the fact that so many plants shift to scrubbing and a
number of them find it more economic to scrub higher sulfur coals.
V,
Shifts in production of this magnitude will clearly have an effect
on employment in the coal mining industry. These effects are discussed
below.
Coal Mine Employment: A secondary impact which these standards would
have is on coal mining employment. Table V.B.6 summarizes the
changes in mining employment in 1995 by coal supply region. Several
points should be made regarding these data. In the first place, the
changes are measured against a 1995 baseline and not against current
employment levels. This is important because mining employment is
projected to grow in most regions between now and 1995. In most
instances the losses shown in Table V.B.6 reflect a decline in new
jobs and not a loss of existing jobs. A second point concerns the
congruence between shifts in mining employment and in coal production.
Although the changes in employment by region are generally similar
to the changes in production, this is not always the case.
Disproportionate changes occur when production shifts to mines and
mine-types having significantly different levels of productivity.
-------
V-10
Table V.6.5
Coal Production and Transportation - 1995
(Millions of Tons)
Coal Production
Northern Appalachia
Central Appalachia
Southern Appalachia
Midwest
West
Total U.S. a/
Coal Transportation
Western Coal Shipped
East
Eastern Coal Shipped
West
1980
185
233
26
134
251
83U
37
25
Base
Case
1995
223
287
29
160
539
1237
73
31
Current
NAAQS
0/-3
0/+21
0/+1
0/-29
0/+10
0/-2
0/+4
0/-6
Change from Base
0.5ppm-lhr .
NAAQS
-22
+39
+1
-40
+20
_2
+12
-8
0.25pp:ii-lh"r
NAAQS
-39
+52
+1
-51
+34
-4
+28
-6
a/Totals may not add due to independent rounding.
-------
V-ll
Table V.B.6
Coal Mine Employment - 1995
(Thousand Workers)
Change from Base
Supply Region
N. Appalachia
C. Appalachia
S. Appalachia
Midwest
Northern Great Plains
Central West
Gulf
Rocky Mountains
Southwest
Northwest
Alaska
Total U.S.
Base
1995
73.3
105.7
12.3
40.9
9.8
3.4
8.5
19.6
2.1
0.5
0.3
277.0
Current
NAAQS
-0.9
+6.6
+0.3
-8.4
-
+0.1
-
+0.1
+0.1
-
-
-0.6
0.5ppm-lhr.
NAAQS
-5.9
+12.0
+0.4
-11.3
-
-
-0.2
+0.1
+0.1
-
-
-0.9
0.25ppm-lhr.
NAAQS
-11.9
+16.5
+0.3
-14.4
-
-0.1
-
+0.1
+ 0.1
-
-
-2.3
Note: Totals may not' add due to independent rounding.
-------
V-12
Impact on Smal 1 Enti tles
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.
The decision not to revise the current NAAQS for SOg will impose
additional control costs and economic effects only on those areas and
sources which are currently designated as non-attainment for S02. A
preliminary assessment of remaining non-attainment areas indicates that
major sources such as utilities, primary smelters, and refineries owned
by large businesses are generally implicated. In addition, the total
number of sources is very limited. These assessments suggest that the
proposed reaffirmation will not significantly affect a substantial number
of small entities.
Furthermore, 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
the ambient standards. Whether a particular standard would have a significant
effect 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 would necessarily be somewhat
speculative.
-------
REFERENCES
Bohi, Douglas R. (1981). Analyzing Demand Behavior: A Study of Energy
Elasticities, published for Resources for the Future, the Johns
Hopkins University Press, Baltimore, MD, 1981.
-------
VI. BENEFIT ANALYSIS" ESTIMATES
A. Introduction
This section presents a summary of estimates for some of the benefits
associated with attaining and maintaining alternative S02 NAAQS. Benefits
represent the improvement in society's well-being as a result of improved
air quality. The benefits estimated in this analysis do not represent
the total improvement that results from going from zero control to
full compli ance with the alternative standards. Rather, they represent
the incremental improvement in going from a baseline reflecting current
operating practice with respect to State Implementation Plans, New Source
Performance Standards, New Source Review, and similar control requirements
to full compliance with the alternative standards.
The purposes of this section are to present the analytic methodology
and the resulting benefits estimates for the alternative S02 NAAQS which
are analyzed. The alternative S02 NAAQS examined include (1) strict
interpretation of the current 3-hour, 24-hour, and annual standards
(2) the addition to the current standards of a single 1-hour, 0.5 ppm
standard, and (3) the addition to the current standards of a single
1-hour, 0.25 ppm standard. The estimated benefits assume full compliance
with each alternative standard on January 1, 1990 and maintenance of that
compliance state through December 31, 2000. However, the loss of benefits
resulting from delaying the start of air quality improvements from
one to ten years is also examined.
The discussion which follows is divided into sections on Methodology
(Section VLB), Air Quality Data (Section VI.C), Study Selection,
Application, Qualifications and Plausibility Checks (Section VI.D),
Estimates (Section VI.E). and Findings (Section VI.F).
-------
VI-2
B. Methodology
Ideally, the estimation of potential economic benefits would be
accomplished using data, assumptions, and mod-el ing techniques developed
specifically for the analytic objective. In the case of the SC>2 NAAQS,
the ideal approach is precluded by project structure, time, and resource
constraints. Therefore, estimates are based upon existing studies which
address some aspects of the health or welfare implications of ambient
sulfur dioxide, sulfates and particul ate matter. This approach, which
involves transformation and extrapolation of existing research and studies,
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 the purposes of this
analysis. 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 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:
o Identification and use of the best data currently available;
o Accomplishment of sensitivity analysis when alternative data
or assumptions exist;
o Development of ranges of estimates to demonstrate the level of
uncertainty associated with different assumptions.
The above elements serve as the basis for the analytic strategy
that is used to develop estimates of the benefits for the alternative
-------
VI-3
S02 NAAQS. The approach begins with a thorough literature search for
existing studies that could possibly be used in the extrapolation process.
The quantitative relationships which are contained in or derived from
the best of the available studies are then used to develop the benefit estimates
presented in this analysis. A summary description of the approach used
is presented in subsequent paragraphs.
1) All categories of potential benefits that might result from
control strategies needed to attain and maintain the alternative
SO? NAAQS are identified. A review of the Criteria Document
and other reports provided a comprehensive listing of possible
adverse effects. A listing of effects categories is presented
in Table VI.B.I below.
2) The existing research literature on the potential effects is
identified, classified and reviewed. All identified studies
are screened on the basis of several criteria, the most notable
of which are analytic quality and potential.for extrapolation of
estimates for benefits analysis (e.g., requisite air quality
data available). As a result of this screening analysis, and
because of time and resource contraints, it is determi ned that
only some of the categories shown in Table VI.B.I can be
estimated. Table VI.3.1 organizes the potential benefits by
pollutant and effects categories.
To comply with alternative S02 NAAQS, sulfur dioxide
emission controls are applied. This in turn reduces concentra-
tions of S02 directly and sulfates and particulate matter
-------
VI-4
Table VI.B.I
Alternative SC>2 NAAQS Potential Benefit Categories
Health Effects
- Mortality Due to Chronic Exposure
- Mortality Due to Acute Exposure
Morbidity Due to Chronic Exposure
- Morbidity Due to Acute Exposure
Soiling and Materials Damage
- Residential Facilities
- Commercial and Industrial Facilities
- Governmental and Institutional Facilities
Climate and Visibility Effects
Local Visibility
- Non-Local Visibility
- Climate
- Visibility at Parks
- Transportation Safety
Non-Human Biological Effects
- Agriculture
- Forestry
- Fisheries
Ecosystem
1. Estimated but coverage limited
2. Not estimated; benefits possible
3. Not estimated; benefits unlikely
* Benefits for this category are not estimated in the main body of this chapter.
However, ran-jes of estimates for this category are provided in Appendix B.
Also, an implicit valuation of mortality risk is presented in Chapter VII.
:t SO?
2
1
2
1
1
2
2
3
3
2
3
3
1
2
2
2
SO/i
2*
2
2
2
2
2
2
1
2
2
2
2
2
2
2
2
Other
Particulate
Matter
2
2
1
1
1
2
2
2
2
2
2
2
2
2
2
2
-------
VI-5
indirectly (by reducing the S02 precursor). Furthermore, when
scrubbers are used to reduce S02 emissions, particulate matter
emissions may also be reduced below baseline levels. Consequently,
compliance with the alternative S02 NAAQS may result in indirect
and direct reductions in PM concentrations. However, PM benefits
are only estimated for the indirect reductions.*
The coverage of benefits -is also limited in other ways.
As noted in Sections IV and V, the assumption of perfectly
price inelastic demand causes us to omit potential SO^, $04,
and PM concentration reductions resulting from decreases in
production volumes in the utility and industrial boiler sectors.
In addition, the benefits are often incomplete for those effects
categories where benefits are measured. For example, the S02
agricultural effects analysis only covered three crops, and the
morbidity analysis for PM ignored compensation for residual
pain and suffering; Finally, as noted previously, there are
many other categories of effects where no benefit estimates are
made. Hence, the coverage of benefits is incomplete. See
Table VI.B.I.
*,, The direct reduction of PM will result in lower concentrations closer to
the plant. The indirect reductions will result in lower concentrations
over a broader geographic area than the direct reduction because the
indirect reduction has a greater impact on fine particulate concentrations.
-------
VI-6
3) Benefits estimates are developed using the quantitative
relationships from each Individual study and the air quality
Improvements postulated for each alternative standard. These
estimates are accomplished according to a four- or five-step
procedure as shown in Figure VI.B.I. The first step is to
identify the magnitude of the ambient air quality improvement
that is estimated to occur in each area and year. This is the
improvement achieved due to implementation of a particular
ambient standard, relative to a baseline situation reflecting
controls already in place.
The second step involves estimating the health and welfare
improvements that might occur as a result of the improvement in
ambient air quality. This step makes use of the research findings
extracted from the literature review discussed previously. These
findings include either linear or nonlinear relationships between
health or welfare status and ambient concentrations of S02 and
SO^. Note that estimates are generally required for each area and
year in which there is an air quality improvement.
The third step is to impute an economic value to the estimated
changes in health and welfare status (Step 3a). For some classes
of benefits it is possible to estimate economic values directly
from the air quality improvement (Step 3b). The household sector
materials benefit estimate is an example of this approach. Studies
of this type allow direct estimates of the perceived economic value
of environmental improvements.
-------
VI-7
STEP
1
Identify air quality improvement
in area i at year t
STEP
2
Estimate health or welfare
improvement in area i at year t
3a
3b
STEP
3
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 totals
Figure VI.B.I Basic steps in estimating benefits for an individual study.
-------
VI-8
Benefit estimates for the covered effect categories are
developed from selected underlying studies. For each scenario
analyzed, the categories examined produce a range of estimates. In
all categories, the studies and results are reviewed and a "best"
estimate selected from the study(ies) which best satisfied the
analytic quality criteria. Estimates drawn from the other studies
are sometimes used to gauge the potential range of benefits for the
effects category under consideration.
The fourth step is to aggregate results over a specific period
of years to obtain discounted present values. In Step 5, benefits
for each area are summed to obtain totals commensurate with the
control strategy and air quality data.
4) Total incremental benefit estimates are developed by combining or
aggregating estimates from the appropriate effects categories.
Total increnental benefit estimates for each of the alternative
standards under consideration are required in order to complete a
benefit-cost analysis of the various policy alternatives. The
estimates are incremental in the sense that 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.
C. Ai r Quality Data
The estimation of economic benefits for alternative standards requires
an understanding of changes in potential exposure for the affected population.
This, in turn, requires knowledge about projected air quality changes.
-------
VI-9
This section provides a description of the three types of air quality
data used for the benefit assessment. The first is 504 data used for
mortality risk reduction estimates and visual range improvement benefits. The
geographic coverage is 31 eastern states. The air quality changes result from
utility and industrial boiler sector compliance with the alternative SOg NAAQS
in 31 eastern states. (The associated emission reduction impacts and control
strategy design are described in Section IV C.5 of the RIA). The second is
PM air quality information used for estimates of PM related morbidity risk
reduction and residential soiling reduction benefits. This second type of
air quality data uses the first type of air quality data as an input but trans-
forms the changes in $04 to changes in TSP. The third type of air quality is
SOg data used to estimate mortality risk reduction, morbidity risk reduction,
materials damage reduction, and increased agricultural yield benefits. The
geographic area for the S02 ambient air quality data is the region modeled
around 4 point sources. Specifically, in the S02 benefit assessment the air
quality changes result from simulated compliance with the 0.5 ppm 1-hour
alternative S02 NAAQS by four utility power plants. The benefits are then
extrapolated to the 31 eastern states and the other two S02 NAAQS that are
examined.
1. S04 Air Quality Assessment
The standards analyzed in this RIA all relate to the local scale
impacts of sulfur oxide sources. However, it is generally recognized that
S02 emissions also have regional scale effects. In the atmosphere S02 can
be transformed into sulfates (504) and both can be transported for long
distances (> 300 km). As a part of the overall RIA the effect of S02 emissions
reductions on sul fates and visual range were modeled on a regional scale. The
-------
VI-10
modeling of source-receptor relationships on any geographic scale and for
any pollutant is difficult and involves some uncertainty. Modeling of
regional seale transport, dispersion, chemical transformation and removal
of sulfur oxides is quite difficult and involves a number of uncertainties.
Nevertheless over the past several years a number of models have been
developed. These models are now at a stage of development that they can be
used to provide some insight into the magnitude and nature of air quality
changes that might result from emission reduction scenarios. The models
are compared and discussed in Section IV.C.5 and detailed results are
presented in Section IV.D.4. Nevertheless it is important to reiterate
some of the limitations discussed in those sections:
o Due to analytic problems associated with the specific application
of RTM-LT and ENAMAP these models/matrices were not used in the benefits
calculations discussed below.
o ASTRAP and MONTE CARLO were used in a transfer matrix and provide the
basis for all benefits calculations. (See Section IV.C.5).
o Since state level matrices were used there is some "averaging out"
of differences which might be apparent with the finer grids used
in the models thenselves.
o The models/matrices used are linear. There is substantial disagreement
as to whether the processes being modeled are in fact linear. If
non-linear processes were found then the estimates used here would not
be accurate.
o The matrices used treat changed emissions as if the change occurs
uniformly over an entire state. This limitation may be significant
in this RIA because the emission changes being modeled do vary from
source to source.
-------
VI-11
o In modeling $64 (and visual range) levels resulting from alternative
NAAQS only emission changes in the eastern 31 state region were
modeled. Western U.S. and Canadian emissions were modeled at their
1995 base level. This was done to allow easier comparison of costs
and benefits in the 31 state region.
In considering the benefits calculations and results reported below, the
reader should keep these limitations in mind.
2. PM Air Quality Assessment
The air quality information developed for the 864 benefit assessment is
also used for the PM benefit assessment. Specifically, the S04 changes for
each state described in the last section are used. The 8-04 changes are trans-
lated into TSP changes by multiplying the sulfate changes by 1.4 for the low
estimate, 1.5 for the middle estimate and 1.6 for the high estimate. The basis
for these estimates is that measured (or predicted) sulfate mass as S04 is
always accompanied by some mass of cations and water, neither of which would
be in particulate form without the hydroscopic sulfate aerosol. In urban areas
the predominant form of sulfate may be ammonium sulfate ((^4)2804) and the
associated ammonium ion alone would make the ratio of fine mass to sulfate =
(NHd)?S04 = 132_ = 1.4.
S04 96
Water would make the ratio still higher. In non-urban areas more charac-
teristic of regional conditions, the major form is thought to be ammonium
bisulfate (NH4HS04). The fine mass to sulfate ratio for this compound is
1.2. Based on limited measurements made in eastern non-urban areas in the
summer, associated water collected on the filter and weighed, ranges from 0
to 50% of sulfate levels. This varies with humidity, temperature, and sample
handling. The total fine mass/sulfate ratio can, therefore, be as high as 1.7
-------
VI-12
(Bachmann, 1985). The factor of 1.6 is used in this RIA for the high estimate
rather than 1.7, to reflect a national average and not an outlier. No estimates
of direct PM emission reductions for alternative standards due to SOg controls
such as scrubbing are available. Therefore no estimates of PM ambient air
reductions due to reduction of PM emissions are included in the analysis.
3. SC>2 Air Quality: Assessment of 4 Point Sources
To examine the direct benefits of S02 anissions reductions, the S02
air quality changes around four large point sources are examined. The
point sources chosen are all utility power plants. The plants are
selected to give a distribution of plant size and dispersion characteristics.
Table VI.C.I provides details concerning each of the four plants. Standard
EPA dispersion models, CRSTER and MPTER, are used to determine ground level
air quality. In all cases, MPTER is run in the CRSTER-equi val ent mode.
Air quality predictions are made for receptors laid-out in a polar
coordinate grid. In general, the receptors are placed where the maximum
1-hour impacts are calculated. For the Portage des Sioux and Potomac River
plants, the outer most ring in the grid is placed at 20 km. For Wansley, the
outer most ring is at 16 km and for Eddystone at 8 km. The models produced
1-hour air quality estimates which are then used to generate the air quality
indices needed for benefits calculations (e.g., annual average).
Various levels of spatial aggregation are required to use the underlying
benefit models. This is because of the need to maintain consistency with
the air quality indices used in the benefit models. For some studies, one
index is used for the whole area around the power plant (e.g., the highest
value of all the receptors or an average of all the receptors). For other
studies, the area around the power plant is divided into 180 smaller areas
-------
Table VI.C.I.
Power Plant Characteristics
Geometric Mean Emissions
Stack Parameters
Meterological Background1
Data (ug/m3)
Portage des Sioux
Eddystone
Potomac River
Wans! ey
MW
904
625
496
1740
Height
(m)
182.9
75.9
49.1
304.8
Diameter
(m)
5.72
7.16
2.59
10.70
Current
4.66
2.912
1.14
4.64
Compliant (0.5 ppm)
2.50
2.54
0.636
3.85
(Year)
1976
1977
1977
1973
234
286
150
80
•^Background concentrations used in determining compliant emissions.
-The actual anission rate for this plant is much lower due to the operation of scrubbers. The anissions shown here
represent unscrubbed operations.
-------
VI-14
around each receptor and the benefits calculated for each of these small
areas using the particular receptor value for the small area.
Several qualifications to the air quality analyses are potentially impor-
tant for the 4 point source based benefit analysis. One year is assumed to
represent the air quality changes around these plants for the 1990-2000 period.
However, variations in meteorological factors could generate different air
quality changes from year to year. In addition, small air quality changes may
occur beyond the outer most receptor ring that would result in a downward bias
in estimated benefits. For example, recent work done by EPA on S02 benefit
estimates from industrial boiler control, shows that increasing the distance
from the outer most receptor ring to the source will lead to larger benefit
estimates. The absence of source interaction combined with a constant back-
ground emission levels results in bias of unknown direction. The assumption
of a 100% capacity utilization rate results in larger predicted air quality
changes.
As mentioned above, the benefits based on the air quality changes for
these 4 point sources are used in an extrapolation of benefits to the 31
eastern states. The western states are not analyzed due to the lack of air
quality data.
D. Study Selection, Application, Qualifications and Plausibility Checks
The procedures used to select studies as a basis for the benefit calculations
are identified in this section. The procedures used for applying the selected
studies to generate benefit estimates are also described. In addition, the
qualifications associated with each application and the plausibility checks for
resulting benefit estimates are discussed. The section is organized by the
pollutant and effect category for which benefits are estimated.
-------
VI-15
1. $04 Benefits
a) Visual Range
1) Study Selection
Because visibility is not directly traded in existing markets,
benefit estimation methods require the establishment of a hypothetical
market or identification of complementarity between visual range and
existing markets. The former method manifests itself in contingent ranking
and contingent valuation surveys eliciting willingness-to-pay estimates
from survey respondents. The latter method generally infers a relationship
between visual range and residential property markets or visual range and
the cost of travel to alternative sites (e.g., national parks).
Several visual range studies have been accomplished in the western
U.S. The contingent valuation and ranking studies generally focused on
user and existence values at national parks (Rowe et al. 1980, Schulze
et. al. 1981, Rae 1983). User value is the value associated with the present
or future enjoyment of the visibility at a site while existence value is the
value an individual places on the preservation or existence of a resource,
such as air quality, even though they do not intend to use the resource
themselves. The western visual range studies also included hedonic property
value studies for two urban areas: the South Coastal Air Basin-Los Angeles
(Brook-shire, et al. 1979, 1980) and the San Francisco Bay Area (Loehmann, et
al. 1981). The hedonic technique is a method for estimating the implicit
prices of the characteristics which differentiate closely related products in
a product class such as property.
To date, three eastern visibility studies have been conducted. One
study (Randall, et al. 1981) assessed the wi Hi ngness-to-pay for visual range
improvement in the Chicago area. Another evaluated the wi Hi ngness-to-pay
-------
VI-16
for visual range improvement in the Cincinnati area (Rae, 1983). A third
(Tolley et al. 1986) assessed the wi 1 li ngness-to-pay for visual range improve-
ment in each of six cities (i.e., Atlanta, Boston, Cincinnati, Miami, Mobile,
and Washington, D.C.) as well as the rest of the east. All of the eastern
visibility studies used contingent valuation or contingent ranking methods.
The studies by Brookshire et al ., Rae, Rowe et al., Loehmann et al.,
and Tolley et al. are chosen for this analysis. They cover nine cities and
all use contingent valuation surveys. The nine cities are: Atlanta; Boston;
Cincinnati (analyzed by both Tolley and Rae); Miami; Mobile; Washington, O.C.;
Los Angeles; San Francisco; and Farmington, New Mexico (analyzed by Rowe, 1980).
2) Application
The visual range benefit assessment involves two sets of procedures.
More detail on these procedures can be found in Appendix A. The first involves
using the visual range estimates for each of the 31 states (Section IV.D.4) to
develop estimates of "local" visual range change. The predicted absolute changes
in annual average visual range vary from 0.20 to 6.23 kilometers depending on
the air quality model, alternative standard analyzed and geographic area.
The second set of procedures entails developing a visual range
improvement valuation coefficient from the five studies discussed above and
applying it to the predicted visual range changes. All of the studies used
personal interviews in which subjects were shown photographs of different
levels of visibility and were asked to estimate how much they would be
willing to pay each month to have one level rather than another. In some
cases, subjects were asked to give separate values for concerns related to
health effects of air pollution versus the visual aesthetic effects. In
other cases, subjects were asked to consider only the visual aesthetic
effects of air pollution. Most of the studies asked about two or more
-------
VI-17
different changes in visual range. In most cases, the subjects were told
that the payment would be in the form of different monthly utility bills.
The bids vary widely within and across the ten cities - ranging
f rom $1.66/year/household/kilometer in Farmington (Rowe et. al.) to $80.66/
year/household/ kilometer in Cincinnati (Rae). The average annual household
values from each of the studies for each hypothesized change in visual range
were compiled to estimate a benefits equation. This equation suggests that
the bids from the surveys can be expected to be a function- of the change in
visual range considered and of the base level and new level of visual range
hypothesized. The variables and data set used in this analysis can be found
in Appendix i\.
3) Qualifications and Plausibility Checks
A number of caveats need to be stressed in presenting the visibility
benefit estimates. One major concern is that the various methods used to
estimate the benefits of environmental improvements are evolving, and any
benefit estimates may be inaccurate due to a number of reasons.
The contingent valuation method may not be capable of measuring
visibility benefits. Although respondents are asked to provide estimates of
visibility, as opposed to health values, individuals may be unable to separate
health values from visibility values. Evidence from cognitive psychology suggests
that individuals may not be able to maintain an a priori distinction between
concerns over health and visibility since they have no need to make such a distinction
in prior thinking or decisionmaking. Attempts to obtain visibility values may,
in fact, include all of the value associated with the 'mental account1 maintained
for air quality. It has been documented by a number of psychologists, including
Kahneman and Tversky, that peopl e tend to^'keep associated concerns in 'mental
-------
VI-18
accounts' and behave as though items in the same account are, in economic
terms, perfect complements.
The contingent valuation method may also pose another concern. Some
researchers suspect that respondents typically overestimate their true willing-
ness to pay when asked to value only one environmental benefit category
(i.e., visibility). When asked for their total willingness to pay for a
number of simultaneous environmental improvements (e.g., better visibility,
cleaner water quality, endangered species preservation, etc.) and then asked
to allocate the total estimate among these categories the resulting willingness
to pay for visibility may be lower than the figures used in this study.
These concerns raise fundamental questions regarding the application
of the contingent valuation method to valuing visual quality improvement, a
good not traded in the market place. The EPA is currently evaluating these
concerns.
Aside from generic methodological concerns, certain factors specific
to this analysis might cause inaccurate benefit estimates. These factors can
be classified into three groups: factors that have a potential upward bias
(overestimating the benefits); factors that have a potential downward bias,
(underestimating the benefits); and factors that potentially cause errors of
unknown direction of bias.
Perception is a factor specific to the analysis used here which also
may cause upward bias. The legitimacy of the benefit estimates is dependent
in part on the ability of people to perceive the projected visual range
improvements. The perception threshold for a change in visual range is two
to five percent for a single visibility event. However, assuming a zero
benefit for a 10 year program that improves visual range one percent per year
-------
VI-19
may not be appropriate either. .Hence, depending on the variance of visual
range improvements throughout the year, and the number of years the program
is in force, the perception factor could influence the benefit estimate.
Therefore, to the extent the generic methodological and perception
factor criticisms are correct, the estimates do represent the total perceived
direct'consumption-benefits (health, materials damage, visibility, etc.) of
improved air quality, not just the benefits of improved visibility.
Option and existence value are factors that suggest that the estimates
of visual range improvement benefits are understated. No attempt has been
made in this RIA to quantify option or existence values. Option value is the
value of an option to preserve some level of air quality at a site in anticipation
of future use. Existence value is defined on page VI-15. Residents of the
Western U.S. may be willing to pay for visual range improvements in the
east, especially at the national parks. U.S. residents West of the Mississippi
have not been included in the benefit assessment. Likewise, the control
strategies implemented in the 31 eastern states may result in visual range
improvements in Canada and certain western states bordering the 31 eastern
states. These too are omitted from this analysis.
Finally, there are other factors- specific to this analysis that may
cause inaccurate estimates where the direction of bias is uncertain. The
extrapolation of the contingent valuation results from nine cities to other
areas of the east is a matter of concern. This extrapolation includes urban
to rural extrapolation which may be questionabl e. These estimates reflect
the average estimates obtained to date in contingent valuation studies concerning
values to residents of visibility in urban areas. As such, the range of the
data should be kept immind. For example, the smallest change in average
visual range considered in any of the studies was about 1 mile, while the
-------
VI-20
average change was about 5 miles and most of the changes were 20 miles or
less. The equation would therefore be less reliable for predicting values
for changes in average visual range of less than 1 mile or greater than 20
miles. The base level of visual range is also important. The average in
these studies was about 15 miles, with most of them falling between 7 and
19 miles. The equation would be less reliable for areas where the current
average visual range is less than 7 or more than 19 miles.
In addition, data collection and analysis methods have been questioned
for some of the underlying studies. For example, in the Tolley et al .
study there is an inconsistency between the visual range improvements
identified in the willingness to pay question and the photographs illustrating
that visual range improvement.
The EPA is currently investigating many of the sources of potential
bias and error.
b) Health Risk Reduction
As discussed in previous sections, a major air quality related
change expected to result from implementing some of the S02 NAAQS standard
alternatives is the reduction in the transformation products of SOg, chiefly
atmospheric sulfate species. These substances appear as both solid and
liquid aerosols with most of the mass being concentrated in the fine
(<2.5 um) fraction. The dominant chemical forms of sulfate observed in
-------
VI-21
the eastern U.S. include acidic species like sulfuric acid (H2S04)., ammonium
bisulfate (NH^HSO^) and the more neutral ammonium sulfate ((Nh^^SO^).
A substantial body of evidence derived from epidemiologic studies,
controlled human exposures, and animal studies suggests that at high enough
concentrations, various atmospheric sulfate species might produce a number
of adverse health effects in sensitive populations. Nevertheless, the
available information base on population exposures to important species,
effects of pollutant mixtures, and concentration-response relationships is
quite limited when applied to the problem of assessing the benefits associated
with reducing current U.S. loadings of sulfates.
A comprehensive assessment of health risk associated with alternative
sulfate levels would include examination of spatial and temporal distribution
of key sulfate species, the relative distribution of co-pollutants,
(particularly oxidants), estimates of population exposures and judgments on
the risk associated with each of the sulfate species Such an undertaking is
far beyond the scope of this analysis, and could be limited by the lack of
information in important areas. Because the data suggest some possibility of
health effects of some sulfates and related particulate matter at or near
current ambient loadings, however, it is important to consider some assessment
of the nature of plausible reductions in health risk associated with sulfate
reduction. Based on the available air quality model outputs, and health
effects information, several simplified approaches for assessing those risk
reductions have been examined. These can be divided into two categories: 1)
use of changes in total annual sulfate levels as a surrogate for potential
exposures of concern and 2) treatment of total sulfate as a component of
otherwise undi fferentiated particul ate matter. The basis for and application
of each of these approaches is outlined below.
-------
VI-22
1. Total Sulfate As Surrogate
Both animal and human studies indicate that the major atmospheric
sulfate species vary greatly in toxicity, with most studies indicating that
toxicity increases with aerosol acidity, i.e., h^SO^ > NH4HS04 > (NH4)2S04.
Major effects of acid sulfate aerosols in such studies include changes in
pulmonary mechanics after single brief exposures (of human asthmatics and
animals) (Utell et al. 1985; Amdur et al. 1983; Koenigetal, 1983), altered
clearance of particles from the lung after single exposures in humans and
animals, persistent shifts in clearance after repeated peak exposures in
animals (Schlesinger et al. 1983; Lippmann et al. 1982), and lung morphological
damages after long-term continued exposures in animals (Gillespie, 1980).
Some of these effects have been noted at single or repeated exposures
as low as 100 to 300 ug/nr of sulfuric acid, or well within the ranges that
occurred in the ambient air in London pollution episodes. A recent review
of the limited data on current U.S. sulfuric acid levels indicate that 12 hour
o
values as high as 40 ug/nr have been recorded in summertime hazy air masses
in the eastern U.S. (Lioy and Lippmann, 1985). Such peaks are probably
limited, but may occur periodically over large geographic areas, with
higher 1-hour values possible. Thus, the laboratory data show potential
acid aerosol effects at levels less than an order of magnitude higher than
short-term peaks that can occur in the eastern U.S. Although such laboratory
data cannot be readily extrapolated for the benefits analyses, they do provide
some qualitative support for associations between total ambient sulfates and
health effects observed in some epidemiological studies.
One of the first epidemiological indications of sulfuric acid effects
was derived from the above mentioned sulfuric acid measurements in London
-------
VI-23
during various episodes where increased mortality and morbidity were observed
(Lawther, 1972). Specific associations were confounded by limited acid aerometry
and historically high levels of other particles and S02- Recent work by
Bates (1985), Scheuher et al. (1983), and Lippmann et al. (1984) suggest
associations between current short-term acid sulfate aerosol exposures and
respiratory problems including symptoms and lung function decrements. None
of these studies by themselves demonstrate causal concentration response
relationships. Few other field studies have attempted to examine the results
of single or repeated exposures to sulfates and other components of hazy air
masses. Thus, no basis exists for providing quantitative estimates of the
effects of specific sulfate species on morbidity. Instead, assessement of
potential morbidity benefits is handled by treating sulfates as an ordinary
component of particulate matter and applying particulate matter morbidity
relationships. This is described in the subsequent section.
A separate set of epidemiological investigations has examined
associations between high sulfate levels and annual mortality statistics.
Substantial disagreement exists within the scientific and analytical communities
regarding the proper interpretation of the pollutant-mortality associations
reported in these so called "macro" -epid emi ological studies and their use in
making quantitative estimates of effects. Three such studies examined for this
analysis (Chappie and Lave 1981; Mendelsohn and Orcutt 1979; and Lave and
Seskiii 1977) find statistically significant relationships between sul fates and
annual mortality, but three other studies (Lipfert 1977, 1979, and 1980) do not
find a statistically significant relationship. A recent study (Evans et al.,
1984) falls somewhere between the aforementioned studies, finding that the
association is significant at the .05 level in 4 of 21 regressions. The
-------
VI-24
remaining 17 all had a positive association with 7 of the 17 significant at
the .10 level. The regression with the lowest significance level for the $64
coefficient was significant at the .32 level. In its review of the epidemio-
logical evidence in the PM-SOX Criteria Document, the Clean Air Scientific
Advisory Committee stated that such studies "suggest an association between
chronic exposures to high concentrations of sulfates and increases in the
level of mortality, but they do not indicate any threshold or safe level from
such exposures and they are not refined enough to provide estimates of the
quantitative effect of sulfate concentrations on mortality."
A major problem with using such studies to quantify health risk is related
to the exposure surrogate. Those studies use measured annual sulfate as
the pollutant indicator. Currently, annual total sulfate levels are less
o
than 10 ug/m in most of the eastern U.S. and acid equivalent levels are much
lower. None of the available laboratory data support the notion that
steady, longterm exposures to acid sulfates at levels under 3 to 5
O n
ug/nr or neutral sulfates at 5 to 7 ug/m produce any measurable effects.
The available data suggest effects of single or repeated peak acid aerosol
exposures at much higher levels. The association could be the result of
actual effects from periodic episodic exposures to acid aerosols.
Unqualified use of the studies would be misleading. In the first instance,
a predicted reduction in annual regional sulfate levels would result in a
small estimated risk reduction to a large number of people. If the emission
changes reduced annual levels but not peak acid levels, however, no health
improvements would be expected and the estimate would be biased high. If the
emissions change was more effective in eliminating peak acid levels than
annual levels, the health improvements could be understated. Moreover, the
-------
VI-25
use of the available studies cannot account for the real potential for individual
or population thresholds that might occur at or even above the
range of current peak exposures.
Some attenpts at risk assessments based on sulfate mortality relationship
have been conducted by Morgan et al. (1982). The examination was inconclusive
concerning the causality question because of the wide variability of risk
estimates given by three health experts and rejection of the process by some
others. In part, rejection of the process may have been due to the uncertainty
of the data, and in part may have been due to the attempt to force judgments
on risk to fit an annual sulf ate/mortality effect model.
Because of the uncertainties associated with the available studies and
the lack of biological plausibility that small changes in annual sulfates
would of themselves produce calculable reductions in health risks, no
estimates are given for benefits associated with reduced mortality risk
for 504. Nevertheless, the data clearly suggest a risk at current ambient
levels, and it is probable that reducing S02 emissions would reduce episodic
peak acid aerosol exposures and thus reduce the risk. Although it has not
been quantified, even a small risk reduction to the millions of people exposed
may have substantial economic benefit. The lack of valuation in this
analysis may significantly understate total benefits. To provide some idea
of the nature of the risk reduction, Appendix 8 uses the annual sulfate
studies to produce hypothetical estimates of reduced mortality benefits.
2. PM Benefits <•:
The PM benefits developed in this analysis are derived from willing-
O
ness-to-pay estimates in the form of dollars/ug/m reduction in TSP. These
estimates in turn are derived from earlier Agency directed studies, including
-------
VI-26
the PM NAAQS RIA. In this section the study selection, application,
qualification and plausibility check procedures used in the PM NAAQS work
are briefly summarized. In addition the procedures, qualifications, and
plausibility of using those studies to develop estimates of PM benefits
associated with attaining and maintaining alternative S02 NAAQS are also
described.
a) Chronic Morbidity
1) Study Selection
The epidemiological studies amenable for benefit analysis in this
category include the Ferris et. al. (1962, 1973, and 1976) and Crocker
et. al (1979) research. The longitudinal studies by Ferris et. al (1962,
1973, and 1976) in Berlin, New Hampshire, were classified by the criteria
document and staff paper as useful for quantitative purposes in PM standard
setting. The cross-sectional study by Crocker et. al, although analytically
appealing from a benefit analysis perspective, was viewed as not quantitative
for purposes of standard setting during the development of the Criteria
Document. The Crocker study yields higher benefit estimates. However,
the Ferris et. al study is adopted for purposes of estimating PM benefits
associated with attaining and maintaining alternative S02 NAAQS,
2) Application
In the PM NAAQS study, air quality concentration thresholds
are imposed to avoid going outside the TSP concentration range used by
Ferris et. al. in estimating the concentration response relationship.
The health end point used by Ferris et. al. is incidence of chronic
respiratory disease. After applying the Ferris et. al. concentration-
-------
VI-27
response relationship, the Health Interview Survey is used to establish
a relationship between the change in chronic respiratory disease incidence
and work loss or restricted activity days. The geographic unit of analysis
is the county. The valuation of chronic respiratory disease damages included
an assessment of lost productivity (i.e., foregone average daily wage), and
increased medical care service charges (i.e., direct medical expenditures).
Of course, the benefit of decreased PM concentrations are evaluated as the
reduction in the aforementioned damages. Benefits are not calculated for
reductions in pain and suffering prior and subsequent to the receipt of
medical care services. The average per capita benefit per ug/nr reduction in
annual TSP for the counties used in the PM NAAQS analysis ranged from $0.12
bo $0.18 for direct medical expenditures, $0.15 to $0.21 for reduced work
loss days per employed worker, and $0.33 to $0.49 per capita reduction in
reduced activity days. The percentage change in chronic respiratory disease
for a 1 ug/m3 change in annual TSP over 130 ug/m3 0.73 to 0.93 percent. More
detail on these procedures is provided in Appendix A.
3) Qualifications
Applying a study based on a small area in-New Hampshire to
estimate benefits in 31 eastern states is not without uncertainty.
However, the fact that the study is judged to be quantitative for purposes
of standard setting does suggest some underlying credibility in the
benefit estimates. Even so, the benefit estimates for this PM category
are limited. For example, no valuation factor is applied to the residual
pain and suffering category. Recent work by Rowe et. al (1986) indicate
that this may understate PM chronic morbidity benefits by a factor of three.
-------
VI-28
b) Acute Morbidity
1) Study Selection
The acute morbidity studies amenable for benefit analysis include
the Samet et. al (1981) longitudinal epidemiological study in Steubenvi11e,
Ohio, and the cross sectional mi croepidemi ological study by Ostro (1986)
which covers six years. The Samet et.al study focused on the relationship
of emergency room visits to acute respiratory disease. The study was
classified by the Particulate Matter Staff Paper (1982) as useful for deter-
mining concentration-response relationships in the standard setting. The
Ostro study focused on broader health end points which related work loss and
restricted activity days due to respiratory disease to fine particle concen-
trations. The PM Staff Paper states that this study provides strong qualitative
support for a relationship between current PM levels and restricted activities
in adults.
The Ostro study provides a more complete geographic and health end point
assessment than the Samet et. al. study and estimates work loss and restricted
activity days directly. Hence, it is selected for use in estimating PM bene-
fits associated with attaining and maintaining alternative SO? NAAQS.
2) Application
Ostro (1986) analyzed six years of individual data from
the National Center for Health Statistics Health Interview Survey (HIS)
to examine the relationship between air pollution and morbidity. Ostro's
sample included all adults age 18 to 65 from 49 metropolitan areas for
which pollution data and HIS sample data exist. The sample contained
approximately 12,000 adults for each of the six years from 1976 through
1981.
-------
VI-29
Three measures of morbidity were used in the analysis: days of
lost work (WLD), days of restricted activity (RAD) and days of respiratory-
related restricted activity (RRAD). Information on these morbidity measures
was obtained in response to a survey question asking the individual how many
days did illness in the previous two weeks prevent him (her) from working
or participating in his (her) usual activities.
The concentration-response functions estimated by Ostro regressed
a measure of the individual 's acute WLD, RAD, or RRAD against a measure of
PM and his/her personal, economic, and other characteristics. The measure of
PM was a two-week average lagged to represent the two-week exposure period
before the period under consideration by the HIS. In addition to the PM
measure, the variables included the individual's age, sex, race, education,
family income, marital status, existence of a chronic condition, quarter of
the survey, and average two-week minimum temperature. The concentration-
response functions that included WLD as the dependent variable also con-
trolled for paid sick leave and whether the individual worked in a blue or
white collar job. In addition to the basic set of variables, the RAD and
RRAD concentration-response functions included a variable reflecting whether
or not the individual was working. Benefits are not calculated for reduction
in pain and suffering prior and subsequent to the receipt of medical care
services.
3) Qualifications
Like all epidemi ological studies, there is the question of inferring
causality from the statistically significant relationship. Unlike many epi-
demi ological studies, Ostro has relatively better control for potentially
confounding factors. Ostro uses a lagged two-week average fine particle
-------
VI-30
concentration as the measure of exposure for the two-week period for
which acute illness is examined. This two-week lag resulted in the
strongest regression. Alternative lag structures (i.e., 0 to 3 periods)
were examined but they were either not significant of less significant
than the two-week lag. Results from Steubenvi11e, Spengler (1986)
and the Netherlands, Dassen (1986) which show an extended depression in
lung function following a single episode provide additional support for
a two-week lag.
The third significant feature of the Ostro study is the use of
fine particle data. Current evidence suggests that small particles (i.e.,
less than 10.0 microns), rather than total suspended particulates (TSP),
have an adverse impact on health. Ostro used data on particles less than
2.5 microns [i.e., fine particles (FP)J as an index of exposure to smaller
particles. Because data on FP were not routinely collected during the
1976 through 1981 time period, Ostro estimated FP concentrations using
equations developed by Trijonis (1982, 1983) and airport visibility data.
Two comments can be raised with respect to the fine particle air
quality data used by Ostro. First, estimated rather than actual FP concen^
trations, were used by Ostro. Although several quality control features
were undertaken to assure that the estimates of FP were reliable, the
reliability of the Trijonis equation itself was never examined by Ostro.
Second, the two-week average of FP used by Ostro corresponds to the
two-week period before the recall period considered in the HIS.
-------
VI-31
G) Household Soili ng
1) Study Selection :
The studies avail able for benefit assessment include the two Phila-
delphia studies by Watson-Gaksch and Cummings et al. (1981) and the 24
Standard Metropolitan Statistical Areas longitudinal study by Mathtech (1982).
The first two studies are developed from a soiling survey (Booz, Allen, Hamilton
1970) designed to determine out-of-pocket cost for air pollution induced soiling.
Watson-Jaksch (1978) analyzed the survey results and found that although cleaning
expenditures did not change a welfare loss could also be incurred. Cummings et
al (1981) analyzed the survey and found that by imputing a value for the time
of the do-it-yourself cleaning activities, which varied with pollution levels,
welfare losses could also be incurred.
The Mathtech study (1982) is referenced later in the SOg section
on materials damage. The structure of this approach is such that physical,
damage functions do not enter .directly into the analysis. Instead, dele-
terious effects of pollutants are reflected in changes in market demand
and supply relationships. Within that framework, household behavior in
terms of soiling perception, cleaning activity, and expenditures to main-
tain a, given degree of cleanliness is estimated econometrical ly. Prices,
environmental variables, and sociodemographic characteristics are .inputs to
the model.
The Mathtech model is chosen because it is based on more recent,
data, has broader geographic coverage, and uses the theoretically correct
measure of benefits, wi1lingness-to-pay.
-------
VI-32
2) Application
To apply the model in the PM NAAQS (Mathtech, 1983) analysis, values
of prices, environmental variables, and socio-demographic characteristics
are put into the estimated model at the county level. The benefits are
determined by changing the value of the environmental variable (i.e., TSP)
to reflect alternative ambient air quality standards or conditions with and
without TSP air quality improvements. Refer to Appendix A for more detail.
3) Qualification and Plausibility
The Mathtech model was peer reviewed in a public forum and a panel of
experts found the model and empirical methods to be sound (Mathtech, 1982).
The predicted level of benefits was judged to be reasonable as a percent
of household income and relative to previously developed estimates. The
model is also referenced in the Criteria Document. However, the framework
of the Mathtech household model does not capture soiling related benefits
such as the time cost incurred for the do-it-yourselfers or the residential
location decision. The Mathtech model is limited in other ways. For
example, the level of detail for the data inputs of the econometric esti-
mation could have been more refined in the absence of the disclosure restric-
tions on the Consumer Expenditure Survey. Application of a model based on
24 SMSAs to 491 counties is not without analytic uncertainty although the
same data base has been analyzed by Gilbert (1985) and Gilbert and Smith
(1985) with similar results.
d) Extrapolation
1) Method
The studies selected for estimating PM benefits associated with
reductions in chronic morbidity, acute morbidity, and household soiling
-------
VI-33
were used In a related particulate matter New Source Performance Standard
generic NSPS study (Mathtech 1985). Average benefits estimated for the PM
generic NSPS study are extrapolated to the PM changes associated with the three
NAAQS control scenarios. Of the counties covered in the related PM generic
NSPS study 491 are in the 31 eastern states addressed in this analysis. The
baseline for the PM generic NSPS analysis is current PM SIP levels. Projected
air quality is with and without 16 currently promulgated PM NSPSs. The chronic
and acute morbidity and household soiling benefits accruing in the year 1995,
to the population projected to reside in those 491 counties, are calculated for
purposes of this analysis. For each county, the estimated benefits are divided
by population to yield a benefit/person figure. This, in turn, is divided by
o
the annual change in air quality to yield a benefit/person/ug/m reduction in
ISP for each county. These figures are summed and divided by the number of
counties (i.e. 491} to give the average $/person/ug/m^ TSP reduction estimate
used in this analysis.
2) Qualifications and Plausibility Checks
To the extent the air quality scenario in the PM generic NSPS
analysis is different from that projected in the S02 NAAQS, analysis biases
of an unknown direction will result. Furthermore, population differences
between the 31 states and the 491 counties forming the basis of this analysis
could also impart biases of an unknown direction.
For example, controlling S02 emissions results in lower fine particle (i.e.,
<_2.5 urn) levels, while NSPS controls for PM emissions result in lower
concentrations of other particle sizes in addition to fine.
-------
VI-34
In the PM generic NSPS analysis and the PM NAAQS analysis,
plausibility checks are conducted by comparing benefits estimated using
independent methods (i.e., hedonic wage and property value studies) with
those developed using the methods described in Sections VI D.2.a, b and c.
In those cases, the independent methods resulted in higher benefit estimates.
The benefit/person/ug/m^ reduction inTSP estimates used in this analysis is
$3.21 in 1984 dollars for acute morbidity, chronic morbidity, and soiling
categories. This number is, of course, an average and because of non-
linearities in the underlying studies is only an approximation of the
individual benef it/person/ug/nr equivalent for a more di saggregated analysis.
3. SOg Benefits
a) Mo rt al i ty Risk
1) Study Selection
S02 mortality studies generally fall into two categories. Micro-
epidemiology studies analyze data on the deaths of specific individuals and
the characteristics of these individuals. Macroepidemiology studies analyze
data on the death rates and characteristics of aggregate populations. A large
number of studies are reviewed to identify the effects of SOg exposure on
mortality. This review is limited to macroepidemiological studies since no
usable concentration response functions could be developed from existing micro-
epidemiological studies.
The macroepidemiological studies are further classified into two
groups depending on the type of exposure (short-term exposure studies
and long-term exposure studies.)
Short-term exposure studies are generally divided into the two
categories of episodic and non-episodic.
-------
VI-35
Short-term Exposure Studies
Episodic Studies
o
The episodic studies indicate that SC^ in excess of 1,000 ug/in and
24-hour concentrations of PM are associated with increases in mortality. The
evidence from these studies is less clear in determining whether 24-hour concen-
3
trations exceeding 1,000 ug/m of S02 result in increased mortality. In addition,
the episodic studies do not provide any information on the mortality effects of
o
exposure to 24-hour concentrations of S02 that are lower than 1,000 ug/m . Since
current and projected 24-hour concentrations of S02 are much less than 1,000
o
ug/m , the episodic studies are not used in the analysis of the effects of
alternative S02 NAAQS.
Non-Episodic Studies
The non-episodic studies considered in the S02 NAAQS analysis examined
the air pollution-mortality relationship using data from London and New York
City. In general, the New York studies are considered to be less reliable than
the London studies because a single centrally located monitor was used as the
index of air quality. Consequently, the New York studies were given limited
weight in the S02 NAAQS analysis.
The London non-episodic studies are used to develop the range of
mortality effects associated with exposure to daily levels of S02- The lower-
bound and mid-point estimates of zero are based on the results of the Mazumdar
et al. (1981). The upper-bound estimate is based on a reanalysis of the London
mortality data for the winter of 1958-1959 by Mathtech.
Long-term Exposure Studies
Six studies on the effects of long-term exposure (i.e., annual) to
S02 were reviewed. These studies did not find a significant relationship
between long-term exposure and mortality. Consequently, no benefits for
reduced long-term exposure to S02 are estimated.
-------
VI-36
2) Application
Benefit calculations are based on averages of predicted daily S02
concentrations averaged across all receptors for each power plant. See
Chapter IV for more detail on the air quality modeling. A threshold of
.07 ppm (24-hour average) is used (i.e., no mortality changes for reduction
in ambient concentration below .07 ppm for a 24-hour average). For more
detail on these procedures see Appendix A.
3) Qualifications and Plausibility Checks
The upper bound estimate is derived from London data that is not
considered typical. Both S02 and particle (measured as British smoke) levels
were higher than those generally observed in the U.S. Also, an influenza
epidemic and higher fog conditions may have influenced mortality relationship.
Consequently, pollution coefficients may be picking up some of the covariation
of variables not controlled for in the specifications.
The use of 0 for lower-bound and mid-point estimates is plausible in
light of the U.S. EPA Criteria Document (19825) and Staff Paper (1982) review
of SOg mortality.
b) Morbidity
1) Study Selection
A large number of studies were reviewed to identify sources for
benefit functions relating S02 to morbidity. Based on the type of exposure,
the studies can be divided into three categories:
Short-term (1-hour),
Daily (24-hour), and
Long-term (annual).
The limitations of the available studies vary across categories.
-------
VI-37
Short-term Exposure
For the short-term exposure category, clinical studies are utilized.
In the relevant range of SOg, the major effects occur in asthmatics. Studies
of this group examine effects on both Tung function indicators and symptoms.
Unfortunately, lung function changes are difficult to value. There are no
estimates of willingness to pay for these changes or of the accompanying medical
costs. Therefore, the analysis considers only effects on symptoms. These
symptoms include wheezi ng, shortness of breath, nose and throat irritation,
and coughing.
The full range of clinical studies of effects of SOg on symptom
prevalence is utilized. The lower-bound estimate is based on the pooled
results of eight studies. The upper-bound estimate is based on work of
Kirkpatrick et al. (1982). The mid-point estimate averages the Kirtpatrick
(1982) estimate with the lower-bound estimate.
Daily Exposure
No microepidemi ology studies were found to be directly usable for
estimating effects of daily exposure. Among the limitations of the studies
were failure to derive quantitative concentration-response functions and
lack of control of confounding factors.
Several mac roepidemi ology studies examine the relationship between
S02 and the number of emergency room visits in an area using time-series
regression analysis. Graves et al. (1980) was selected to develop a range of
benefit estimates.
Long-Term Exposure
There are a number of microepidemi ology studies of the health effects
of annual exposure. All the stud.ies identified, however, either are not
-------
VI-38
suitable or found no effects. Most of the studies are rejected because of
failure to differentiate between the effects of S02 and PM. The observed
variations in morbidity could be due to differences in the levels of SOg, PM,
or a combination of pollutants. Due to the small number of observations in the
studies, statistical techniques could not be applied to the study results to
attempt to determine the effects of an individual pollutant. Therefore, no
benefits are estimated for this category.
A survey by Loehman et. al. (1979) identified a quantitative estimate
to aid in valuation of symptoms. This study indicated a willingness to pay for
an hour of minor symptoms reduction of under $1 for the general population.
Another study by Gerking and Dickie (1986) examined 26 symptoms and gathered
information on willingness to pay to avoid each of these symptoms for one day-
Results from this study indicated median bids ranging from $0 for avoidance of
nose bleeds to $262 for avoidance of nausea.
2) Application
For the short-term exposure category a minimum estimate of zero is
used. The maximum estimate is based on Kirkpatrick and is arbitrarily estimated
to be $50 per hour of symptom reduction. The midpoint estimate averages the
Kirkpatrick (1982) estimate with the zero lower estimate and uses a $25 per
hour valuation of symptoms. These valuation factors may be adjusted in the
final report as a result of ongoing Agency work on VOC benefits. While the
benefits from the daily exposure analysis are estimated using the Graves et al.
(1980) study, the results are three orders of magnitude smaller than those
based on Kirkpatrick and hence are not reflected in the S02 analysis due to
rounding. For more detail on these procedures see Appendix A.
-------
VI-39
3) Qualifications and Plausibility Checks
To estimate the benefits using clinical studies, assumptions about
activity levels are derived from the original PM NAAQS exposure analysis.
Doing this implicitly assumed the same activity patterns for asthmatics as the
rest of the population. Inhaled dose is estimated using an assumed ventilation
rate of 40 liters per minute and a peak concentration of twice the hourly
concentration. The symptom valuation is an arbitrary escalation of a study
valuing similar symptoms for the general population rather than for asthmatics.
The magnitude of the benefit estimates seems plausible in view of the
S02 exposure analysis.
c) Agriculture
1) Study Selection
The selection of crops for analysis is governed by the
following criteria:
o Crop sensitivity to exposure to S02;
o Information availability for the formulation of a
dose-response function; and
o Significant economic value of,the crop.
Based on application of these criteria, soybeans, wheat, and oats are
selected.
For calculation of benefits, the parameters which define the
response of a crop to S02 should reflect its economic value. Yield can
be given a value in economic terms. While foliar and yield effects
may occur together and,may\ in some instances, be highly correlated,
yield response cannot generally be accurately extrapolated from observed
foliar effects of variations in S02- Therefore, study selection is
limited to studies of yield effects. The additional criteria for study
-------
VI-40
selection are: study focus is on ambient S02 exposure and other ambient
environmental factors; control for the impact of other confounding
environmental conditions on the concentration relationship; and plausible
and quantifiable results which can be used to develop concentration response
functions.
Application of these cri ten a yi elded a very limited number of
studies for each crop. The major study used for soybeans is from Sprugel
et. al (1980). For oats and wheat, the principal source is a series of
reports by Guderian and Stratmann (1962 and 1968).
2) Application
Average air quality for each receptor area for the same part of
the growing season as the underlying study uses is applied. The estimate
is restricted to air quality changes within the underlying study range.
Zero is used as the minimum estimate. The maximum estimate applies the
dose response function of the study varietal to all of that crop. One-half
the maximum estimate is used for the middle estimate.
3) Qualifications and Plausibility Checks
The analysis values yield changes at recent state market prices.
No attempt is made to adjust for price changes that might result from
nationwide yield changes. No changes in cropping patterns are factored
into the analysis. No attempt is made to value contributions to yield
for sulfur deficient soils. No attempt is made to adjust agricultural
benefits for subsidy effects.
The yield functions only consider average sulfur concentrations for
the growing seasons. This ignores impacts of the time pattern of exposure
(e.g. size and timing of peaks), the impact of other pollutants, and
-------
VI-41
interactions with other determinants of yield such as temperature, light,
soil, and agricultural practice.
The choice of functional form for the yield function was necessarily
made with a very, limited set of data points. The yield changes are sensitive
to the functional form.
The four power plants may not be in any way representative of other
power plants with respect to the distribution of crops within their area of
air quality impact. The estimated yield changes for crops within the areas
of influence around the four power plants are all less than 1.1 percent
which does not seem implausibly large.
d) Materials Damage Models
1) Study Selection
There are two elements that are important in selecting models
that consider materials damage. First, one must evaluate the validity
of statistical relationships between measures of physical damage (e.g.,
corrosion rates) for a given material and plausible explanatory factors
such as SC>2 concentrations and relative humidity. Second, one must
consider the soundness of the procedures used to impart economic value to
the physical damage that is caused.
The first topic requires a review and critique of the literature
on physical damage functions. These functions have been reviewed in the
Criteria Document and the general consensus is that fairly reliable damage
functions have been developed for ferrous metals and zinc. However, cur-
rently available damage functions for exterior paints, stone, masonary,
concrete, textile^, leather, and paper are considered not to be as well
specified.
-------
VI-42
There are three general approaches that have been used to value
physical damage. Two of the approaches use the physical damage functions
directly. These approaches have been classified as the value of lost
material method and the cost of ameliorative/preventive action method.
Examples of the first approach include Salmon (1970) and SRI (1981).
Examples of the second valuation approach include Fink et al. (1971) and
TRC (1931). In the third approach physical damage functions do not enter
directly into the analysis. Instead, economic denand and supply curves
are estimated to reflect the deleterious effects of S02 (Mathtech, 1982).
Note however, that the physical damage functions can provide supporting
information on the types of behavioral responses implied by the economic
analysi s.
A prime concern in selecting benefit models for valuing material
damage is the data requirements. Both valuation methods that rely ex-
plicitly on physical damage functions require information on materials
inventory, exposure and distribution. These data are most likely
available only with significant approximation and are most appropriate
only for selected cities where materials surveys have been conducted. It
is principally for this reason that the calculation of benefits from reduced
materials damage is limited to the demand/supply model as estimated by
Mathtech. Although the data used in the Mathtech analysis are limited in
some respects and interpretations of some of the implied behavioral responses
have been questioned, the theoretical and empirical methods have been
judged by several groups to be sound. In addition to reduced data require-
ments, the Mathtech model of materials damage in the household sector has
-------
VI-43
other positive attributes that are not shared by the other valuation
approaches. The method of benefit calculation is based on willingness to
pay which is the theoretically correct procedure for measuring welfare
change. Also, the Mathtech model permits consideration of substitution
possibilities.
2) Application
The midpoint and range of estimated benefits comes from evaluating
the S02 air quality changes using the estimated expected coefficient contained
in the household model and plus and minus two standard deviations from the
expected coefficient. Because of the structure of the underlying model,
the receptor with the maximum 24-hour second high is used for all households
within the area of the air quality analysis for each plant.
3) Qualifications and Plausibility Checks
Using modeled air quality receptors rather than monitored data as
was used in the original study may bias the benefit estimate in an unknown
direction. Only residential material damage is estimated. Because of the
pollution index used in the underlying study (24-hour second high for an
SMSA) and the geographic unit of the underlying study (SMSA's), the impact
of applying these coefficients to a different area (a circle around a power
plant) is unclear. In addition, recent work by EPA (Industrial Boiler S02
NSPS RIA, 1986), indicates that the distances used in this analysis of
8-20 km from the outer most receptor ring to the source may result in a
downward bias.
The benefits calculated on a per-household basis range from $.77
per year to $6.89 per year depending on the particular power plant and
** •"
alternative standard. These numbers do not seem implausibly large.
-------
VI-44
e) Extrapolation of Results
1) Method
A crude procedure is developed to extrapolate the benefits for the
four power plants to the 31 eastern states. It is assumed that benefits
per ton are invariant with source type and location. Case study benefits
from the four power plants are divided by the associated reduction in tons
of S02. The resulting benefit per ton estimate is applied to the number of
tons of S02 controlled for the 31 eastern states. Further detail on this
procedure is provided in Appendix A.
2) Qualifications
This extrapolation procedure is subject to all the qualifications
of any analysis based on a constant benefits per ton assumption. This is a
questionable assumption for a number of reasons. First, many of the under-
lying benefit functions are non-linear with respect to ambient concentrations.
Second, ambient S02 concentration changes are not linearly related to gross
emissions changes. Finally, the distribution of people, crops, and materials
at risk cannot be expected to lead to a constant benefit per unit concen-
tration change. Each of the problems with the constant emissions ratio
assumption applies to both its use for extrapolation to increase geographic
coverage and its use to increase coverage of standards.
E. Estimates
1. Benefits for 31 States
The benefits estimated for the 31-state analysis are exhibited in Table
VI.E.I. Ranges are presented to reflect the uncertainty concerning air quality
and economic valuation. The benefit categories estimated include increased
visibility, reduced chronic and acute morbidity and reduced household (resi-
dential) soiling due to PM reductions, reduced mortality and morbidity due
-------
Table VI.E.I
31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS
1
(DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
S02 Health
S02 Welfare
Material Damage
Agriculture
504 Visibility Benefits
PM Morbidity & Soiling
Total
Current Standards
(Strict Interpretation)
Low Middle
0 .002
.029 .142
0 .004
2.1 -' 2.5
2.1 2.4
4.2 5.1
High
.225
.262
.008
3.1
2.7
6.3
0.5 ppm
1-hour stand
Low Middle
0 .004
.052 .254
0 .007
3.7 4.6
3.9 4.3
7.7 9.2
ard
High
.402
.468
.014
5.7
4.8
11.4
Low
0
0.25 ppm
1-hour stand
Middle
.008
.104 .507
0 .014
8.3
7.6
16.0
9.8
8.8
19.1
ard
High
.805
.935
.028
11.8
10.2
23.8
1 The assessment only includes a subset of related benefits.
2 The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .2728.
-------
VI-46
to S02 reductions, increased agricultural yields due to SOg reductions, and
decreased household sector materials damage due to SOg reductions.
2. SO? Benefits for 4-Point Sources
Table VI.E.2 is presented to better see the relative magnitudes of the
related benefits. The range and midpoint estimates are presented for reduced
mortality and morbidity risk, increased agricultural yields and decreased
household sector materials damage. The geographic coverage is 4-point sources
The standard analyzed is the 0.5 ppm 1-hour S02 NAAQS.
3. Cost of Delay
In Chapter IV (Costs and Environmental Impacts, Section C .l.b.) there is
a discussion of scrubber constraints. The 0.25 ppm 1-hour NAAQS results in
an increase of 60 GW scrubbed capacity compared to the estimated potential
industry capability for 20 GW. It is estimated that this short run constraint
could be handled if impl anentation of the NAAQS were delayed from 1990
until 1995. This section examines the potential costs of a 10 year delay
(1990 to 2000) in implementation of the NAAQS.
Delaying the air quality improvements associated with the alternative
SOg NAAQS results in a loss of benefits. Estimates of this cost of delay
are exhibited in Table VI.E.3. The range of benefits from Table VI.E.I for
each standard was used to generate these estimates.
F. Findings
Three major findings anerge from the abbreviated and qualified benefit
analysis.
1. Estimated $04 and PM Related Benefits Are Larger Than S02 Related Benefits
Control strategies designed to attain and maintain alternative S02
NAAQS also result in reduced concentrations of SO^ and PM. As can be seen
from Table VI.E.I, estimates of SOg health and welfare benefits are always
-------
Table VI.E.2 4 Point Source S02 Benefit Assessment
for a 0.5 ppm 1-hour S02 NAAQS
(Millions of January 1984 Dollars)
1
Reduced Mortality Risk
LOW MIDPOINT HIGH
p O
Present Value Annualized Present Value Annualized Present Value Annualized
0
0
0
0
11.8
3.2
Reduced Morbidity
.13
.04
.25
.07
Increased Agricultural Yields
.22
.06
.43
.12
Decreased Household Sector
Material Damage
1.6
0.44
7.6
2.1
14.1
3.8
2
3
The assessment only includes a subset of SOj benefits and is not extrapolated to the 31
eastern states.
The discounted present value of an eleven year stream of benefits occuring from January 1, 1990 to
Decanber 31, 2000 using a real discount rate of 10 percent in 1984.
The equivalent annual flo// of benefits occuring from January 1, 1990 to December 31, 2000. Benefits
actually increase over time because of projected changes in population and other socio-demographic
variables. . _ .
-------
Table VI.E.3
Loss of Benefits Due to Air Quality Improvement Delays
(Billions of January 1984 Dollars)
Air Quality Improvement Delay
(Date of Attainment)1
1 year (1991)
2 years (1992)
3 years (1993)
4 years (1994)
5 years ,(1995)
6 years (1996)
7 years (1997)
8 years (1998)
9 years (1999)
10 years (2000)
Current Standards
(Strict Interpretation),.
Discounted Present Value*1
.59 to .88
1.1 to 1.7
1.4 to 2.1
2.1 to 3.1
2.5 to 3.7
2.8 to 4.2
3.2 to 4.7
3.5 to 5.2
3.7 to 5.6
4.0 to 6.0
Alternative SOg NAAQS
0.5 ppm
1-Hour Standard
Discounted Present Value^
1.1 to 1.6
2.1 to 3.1
2.5 to 3.8
3.8 to 5.6
4.5 to 6.7
5.2 to 7.6
5.8 to 8.5
6.3 to 9.4
6.8 to 10.1
7.3 to 10.8
0.25 ppm
1-Hour Standard
Discounted Present Value^
2.2 to 3.3
4.3 to 6.4
5.3 to 7.8
7.8 to 11.6
9.4 to 13.9
10.7 to 16.0
12.0 to 17.8
13.1 to 19.6
14.2 to 21.1
15.1 to 22.5
1 Date of attainment is January 1 for each year.
^ Lost benefits are calculated in comparison with a 1990 date using the ranges for total benefits from Table VI.E.I.
-------
VI-49
less than 10% of the estimates of benefits for the total of PM morbidity
and soiling and 504 visibility.
2. Direct S02 Welfare Related Benefits for Alternatives Considered Appear
Greater Than SC>2 Health Related Benefits
The Clean Air Act enphasizes public health protection relative to welfare
(e.g., materials, aesthetic, agriculture, etc.) protection. However, the
4-point source assessment of S02 related benefits for the 0.5 ppm 1-hour
alternative suggests that society, in terms of its collective willingness-to-
pay, may not always put health first. The assessment, albeit uncertain,
indicates that foregoing selected quantities of materials and agricultural
damages may be valued more highly than foregoing lesser risks of mortality
and morbidity. Consequently, in this instance, welfare benefits could be
greater than health benefits.
3. Air Quality Improvement Delays Mean Foregone Benefits
If implementation of the S02 NAAQS is delayed from 1990 to later years,
associated benefits will be foregone or lost. These foregone benefits
will increase with the delay and will be greatest for the 0.25 ppm S02
1-hour alternative. However, this pattern is not sufficient evidence to
recommend, on economic efficiency grounds, expeditious attainment. When
attainment is delayed, control costs are also foregone.
-------
VI-50
List of References
Bachmann, J. Memorandum to Tom Walton, Estimate of Sulfate/Fine Mass
Relationship, 1985.
Booz, Allen and Hamilton, Inc. (1970) Study to Determi ne Residenti al Soiling
Costs of Particulate Air Pollution. APTD-0715, U.S. Department of
Health, Education and Welfare, National Air Pol lution Control Administration,
Raleigh, N.C.
Brookshire, D., d 'Arge, R., Schulze, W., and Thayer, M., (1979). Methods
Development for Assessing Air Pollution Control Benefits. Vol. 2:
Experiments in Valuing Non-Market Goods: A Case Study of Alternative
Benefit Measures of Air Pollution in the South Coast Air Basin of Southern
California, EPA-600/6-79-0016.
Brookshire, D., Thayer, M., Schulze, W., and d'Arge, R., (1980). Valuing
Public Goods: A Comparison of Survey and Hedonic Approaches, Resource
and Environmental Economic Laboratory, University of Wyoming, later
published in American Economic Review, 1982, 72:165-177.
Chappie, M. and Lave, L., (1982). The Health Effects of Air Pollution: A
Re-Analysis, Journal of Urban Economics, 12:346-376.
Crocker, T. D., Schulze, W., Ben-David, S. and Kneese, A., (1979). Methods
Development for Assessing Air Pollution Control Benefits, Vol. I,
Economics of Air Pollution Epidemiology, EPA-600/5-79-001a, U.S.
Environmental Protection Agency, Washington, D.C.
Cummings, R., Burness, H. and Norton, R. Methods Development for Environmental
Control Benefits Assessment, Volume V: Measuring Household Soiling
Damages from Suspended Air Particul ates, A Methodological Inquiry. Draft
Report, January 1981.
Dassen, W., Brunekreef, 8., Hoek, G., Hofschreuder, P., Staatsen, B., de Grout,
H., Schouten, E., Biersteker, K. (1986). Decline in Children's Pulmonary
Function During an Air Pollution Episode. Journal of the Air Pollution
Control Association (in press).
Dickie, M. Gerking, S. Schulze, W., Coulson, A., Tashkin, D. (1986). Improving
Accuracy and Reducing Costs of Environmental Benefit Assessments. Value
of Symptoms of Ozone Exposure: An Application of the Averting Behavior
Method. U.S. Environmental Protection Agency, Washington, D.C.
Evans, J.S., Tosteson, T. and Kinney, P.L., (1984a). Cross-Sectional
Mortality Studies and Air Pollution Risk Assessment, Environment Inter-
national , 55-83.
Ferris, B. G., Jr., Higgins, I., Higgins, M. W. and Peters, J. M., (1973).
Chronic Nonspecific Respiratory Disease in Berlin, New Hampshire,
1961-1967: A Follow-up Study. American Review of Respiratory Disease,
107:110-122.
-------
VI-51
Ferris, B. G., Jr., Chen H., Puleo, S., and Murphy, H., Jr., (1976). Chronic
Nonspecific Respiratory Disease in Berlin, New Hampshi re, 1967-1973: A
Further Follow-Up Study. American Review of Respiratory Disease, 113:
475-485.
Fink, F.W., Buttner, F. H., and Boyd W. K., (1971). Technical-Economic
Evaluation of Air Pollution Corrosion Costs on Metals in the U.S.
APTD-0654. U.S. Environmental Protection Agency, Research Triangle
Park, N.C.
Gilbert and Smith, Role of Economic Adjustment for Environmental Benefits
Analysis. Econometric Society Meetings, New York. December 1985.
Graves, P.E., Krumm R. J., and Violette, D. M., (1980). Estimating the Benefits
of Improved Air Quality. Report for Meeting of Benefit Methodology
Panel, National Commission on Air Quality, Decanber.
Guderian, R. and Stratman, H., (1962). Field Experiments to Determine the
Effects of S02 on Vegetation - Part I: Survey of Method and Evaluation
of Results. Research Reports by the State of North Rhine-Westphalia,
No. 1920. West German Press, Cologne and Opladen.
Guderian, R. and Stratman, H., (1968). Field Experiments to Determine the
Effects of S02 on Vegetation - Part III: Threshold Values of Harmful
S02 Emissions for Fruit and Forest Trees and for Agricultural and
Garden Plant Species. Research Report by the State.of North Rhine-
Westphalia, No. 1920. West German Press, Cologne and Opladen.
Kirkpatrick, M.B., Sheppard, D., Nadel, J.A., and Boushey, H.A., (1982). Effect
of Oronasal Breathing Route on Sulfur Dioxide-Induced Bronchoconstriction
in Exercising Asthmatic Subjects. American Review of Respiratory Diseases,
125:627-631.
Lave, L. B. and Seskin, E.P., (1977). Air Pollution and Human Health, Johns
Hopkins University Press, Baltimore, MD.
Lave, L.B. and Seskin, E.P;, (1973). An .Analysis of the Association Between
U.S. Mortality and Air Pollution, "Journal of the American Statistical
Assocociation, 68:284-290.
Lipfert, F. W. .. The Association of Air Pollution with Human Mortality:
Multiple Regression Results for 136 Cities, 1969. Paper presented at
the 70th Annual Meeting of the Air Pollution Control Association, June
20-24, 1977.
Lipfert, F.W. On the Evaluation of Air Pollution Control Benefits. Prepared
for the National Commission on Air Quality, November 1979.
Lipfert, F.W. (1980) Sulfur Oxides, Particulates, and Human Mortality:
Synopsis of Statistical Correlations. Journal of the Air Pollution
Control Association, 30:366-371.
-------
VI-52
Loehman, E., Boldt, D., and Chaikln, K., (1981). Measuring the Benefits of Air
Quality Improvements in the San Francisco Bay Area, SRI 8962.
Loehman, E.T., Berg, S.V., Arroyo, A.A., Hedinger, R.A., Schwartz, J.M.,
Shaw, M.E., Fahien, R.W., De, V.H., Fishe, R.P., Rio, D.E., Rossley, W.F.,
and Green, A.E.S., (1979). Distributional Analysis of Regional Benefits and
Cost of Air Quality Control. Journal of Environmental Economics and
Management, 6:222-243.
Mazumdar, S., Schimmel, H. and Higgins, I., (1981). Daily mortality, smoke
and S02 in London, England 1959-1972. Proceedings of the Proposed
SOx and Particul ate Standard Specialty Conference. Air Pollution Control
Association, Atlanta, Georgia.
Mazumdar, S. and Sussman, N., (1981). Relationships of air pollution to .
health: results from the Pittsburgh study. Proceedings of the 74th
Annual Meeting, Air Pollution Control Association, Philadelphia, Pa.
June 21-16, 1981.
Martin, A.E. and Bradley, W. H., (1960). Mortality, Fog, and Atmospheric
Pollution - An Investigation During the Winter of 1958-59. Monthly
Bulletin of the Ministry of Health. Public Health Laboratory Service,
19:56-72.
Mathtech, Inc., (1982). Benefits Analysis of Alternative Secondary National
Ambient Air Quality Standards for Sulfur Dioxide and Total Suspended
Particul ates. Final report prepared for U.S. Environmental Protection
Agency under Contract No. 68-02-3392. Research Triangle Park, N.C.
Mathtech, Inc., (1983). Benefit and Net Benefit Analysis of Alternative
National Ambient Air Quality Standards for Particulate Matter. Prepared
for the U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards under Contract No. 68-02-3826.
Mathtech, Inc., (1984). Benefit Cost Analysis of Selected New Source
Performance Standards for Particulate Matter. Draft Final Report
Proposed for U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards under Contract No. 68-02-3553, Research Triangle
Park, N.C.
Mendelson, R., and Orcutt, G., (1979). An anpirical analysis of air pollution
dose-response curves. Journal of Environmental Management. 666:85-106.
Morgan, Granger M. et.al., (1982). Technologial Uncertainty in Policy Analysis.
Final Report prepared for the Division of Policy Research and Analysis,
National Science Foundation under Grant No. PRA-79-13070.
Ostro, B. D., (1986) Air Pollution and Morbidity Revisited: A Specification
Test. Journal of Environmental Economics and Management, March 1987.
-------
VI-53
Rae, D., (1982). The Value to Visitors of Improving Visibility at Mesa
Verde and Great Smokey National Parks, a paper presented at a National
Park Service Conference, Keystone Colorado, May, later published in
Robert Rowe and Lauraine Chestnut, eds., Managing Air Quality and Scenic
Resources at National Parks and Wilderness Areas, Westview Press, Boulder,
Colorado, 1983.
Rae, D., (1983). Benefits of Visual Air Quality In Cincinnati: Results of
a Contingent Ranking Survey. Presented at the American Economic
Association Meetings, Washington, DC.
Randall, A., Hoehn, J., and Tolley, G. "The Structure of Contingent Markets:
Results of a Recent Experiment," a paper presented at the American
Economic Association Meetings, December 1981.
Roth et_al_., (1981). Summary of unpublished Roth et al. (1981): Year-by-Year
Analysis of London Mortality Data for Winters of 1958-1959 to 1971-1972.
Appendix 14-F of Air Quaity Criteria for Particulate Matter and Sulfur
Oxides, Volume III, EPA Office of Research and Development, Research
Triangle Park, North Carolina, December.
Rowe, R.D., d'Arge, R., and Brookshire, D., (1980). An experiment on the economic
value of visibility. Journal of Environmental Economics and Management.
March: 1-19.
Rowe, R. D., Chestnut, L. G., (1985). Oxidants and Asthmatics in Los Angeles:
A Benefits Analysis. U.S. Environmental Protection Agency, Washington, D. C.
Rowe, R. D., Chestnut, L. G. (1986). Addendum to Oxidants and Asthmatics in
Los Angeles: A Benefits Analysis. U.S. Environmental Protection Agency,
Washington, D.C.
Salman, A. Systems Analysis of the Effect of Air Pollution on Materials,
Final Report for the National Air Pollution Control Association under
Contract No. CPA-22-69-113. January 1970.
Samet, J.M., Speizer, F.E., Bishop, Y., Spengler, J.D., and Ferris, Jr., E.G.,
(1981). The relationship between air pollution and emergency room
visits in an industrial community. Journal of the Air Pollution Control
Association 31:236-240.
Schulze, W.D., Brookshire, D.S., Walther, E., and Kelley, K., (1981). Methods
development for environmental control benefits assessment. Vol. X:
The benefits of preserving visibility in the national parklands of the
Southwest. Washington, D.C.: U.S. Environmental Protection Agency,
Office of Research and Development.
Spengler, J.D., Briggs, S.C.K., Ozkaynak, H. (1986). Relationship Between TSP
Measuremnts and Size-Fractionated Particle Mass Measurements in Six Cities
Participating in the Harvard Air Pollution Health Study. U. S.
Environmental Protection Agency, Office of Air Quality Planning and
Standards. Draft.
-------
VI-54
Sprugel, D.6., Miller, J.E., Muller, R.N., Smith, H.J., and Xen'kos P.B.,
(1980). Sulfur Dioxide Effects on Yield and Seed Quality in Field-
Grown Soybeans. Phytopathology, 70:1129-1133.
SRI, Inc. An Estimate of the Non-Health Benefits of Meeting the Secondary
National Ambient Air Quality Standards. Prepared for the National
Commission on Air Quality, January 1981.
Tolley, G., Randall, A., Bloomquist, G., Fabian, R., Fishelson, G., Frankel ,
A., Hoehn, J., Krumm, R., Mensah, E., and Smith, T. , (1986). Establishing
and Valuing the Effects of Improved Visibility in Eastern United States,
Office of Research and Development, Environmental Protection Agency.
TRC (1981). Benefit Model for Pollution Effects on Material. Final report
prepared for U.S. Environmental Protection Agency under Contract No.
68-02-3447. Research Triangle Park, N.C. 27711.
Trijonis, J., (1982). Analysis of Particul ate Matter Concentrations and
Visibility in the Eastern United States. U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards.
Trijonis, J., (1983). Development and Application of Methods for Estimating
Inhalable and Fine Particul ate Concentrations from Routi ne Hi-Vol Data.
Atmospheric Environment, Vol. 17, 998-1008.
U.S. Environmental Protection Agency, Office of Air Quality Planning and
Standards. Review of the National Ambient Air Quality Standards for
Sulfur Oxides: Assessment of Scientific and Technical Information,
OAQPS Staff Paper. November 1982.
U.S. Environmental Protection Agency, Office of Air Quality Planning and
Standards. Review of the National Ambient Air Quality Standards for
Particul ate Matter: Assessment of Scientific and Technical Information.
OAQPS Staff Paper (EPA-450/5-82-001), Research Triangle Park, NC ,
January 1982.
U.S. Environmental Protection Agency, Environmental Criteria and Assessment
Office. Air Quality Criteria for Particulate Matter and Sulfur Oxides:
Volume III. Report No. EPA-600/8-82-029cf, Research Triangle Park, NC,
1982b.
U.S. Environmental Protection Agency, Office of Air Quality Planning and
Standards. Benefit-Cost Analysis of Selected New Source Performance
Standards for Particulate Matter-Draft Report. Research Triangle Park,
NC, July 1985.
Watson, W.D., Jr., and Jaksch, J.A., (1978). Household cleaning costs and
air pollution. Presented at the 71st Annual Meeting, Air Pollution
Control Association, Houston, Texas, June 25-30. Paper No. 78-52.3.
-------
VII. BENEFIT-COST ANALYSIS
A. Introduction
This chapter of the Regulatory Impact Analysis (RIA) presents
comparisons of the estimated incremental benefits and incremental costs
of alternative National Ambient Air Quality Standards for sulfur dioxide
(S02 NAAQS). Air quality regulations such as the S02 NAAQS 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. The benefit-cost analysis provides a consistent
framework for evaluating the economic effects of alternative regulatory
policies. The analysis is presented in 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. The
EPA Guidelines for Performing Regulatory Impact Analysis suggest that the
determination of which regulatory alternative is preferred in terms of
maximum net benefits is made difficult by uncertainties in data, by
inadequacies in analytical techniques, and by the presence of benefits and
costs that can be quantified but not monetized or that can only be
qualitatively assessed. Court decisions brought by the Lead Industries
Association (Lead Industries Association, Inc. V. EPA, 1980) and the American
Petroleum Institute (American Petroleum Institute V. EPA, 1981) clearly
limit the use of technical and economic feasibility in setting primary
national ambient air quality standards. In response to these judicial
decisions, the Administrator has not considered this benefit-cost analysis
in the proposed rule-making.
-------
VII-2
The criteria of economic efficiency which is employed to evaluate the
alternative S02 NAAQS is described in Section VII.B followed by the benefit-
cost analysis methodology in Section VII.C. Limitations of the methodology
are presented in Section VII.D and the scope of the analysis is detailed in
Section VII.E. Several conceptual issues concerning the measurement of
benefits and costs are raised in Section VII.F- Section VII.G includes
estimates of benefits and costs and Section VII.H presents limitations
to assumptions behind the estimates. Net benefits are provided in Section
VII.I along with additional limitations in Section VII.J. The analysis is
concluded with qualifications and findings in Section VII.K.
B. Economic Efficiency
When an air quality regulation is adopted, society's resources are
reallocated towards the production of cleaner air and away from other goods
and services. As a result society's economic well-being is affected. The
efficiency criterion is used to evaluate the economic desirability of this
reallocation. An 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 existed 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
-------
VII-3
theory context, be reduced.* This need not be the case, however, if those
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. This analysis adopts the conventional compensation
efficiency criterion.
In addition to determining whether a given air quality standard is
efficient, it is also possible to rank the alternative S02 NAAQS in terms of
relative efficiency. The S02 NAAQS that is most efficient, relative to the
alternatives considered, is the one which provides the largest positive
incremental net benefits. An analysis of the relative efficiency of the
alternative S02 NAAQS considered is described later in Section VII.I and K.
C. Methodology - Incremental Net Benefit Analysis
In order to evaluate the relative efficiency of the alternative S(>2
NAAQS, an analysis of the incremental benefits and costs associated with
each alternative S02 NAAQS i.s needed. The incremental benefits associated
with a given SC>2 NAAQS are the additional benefits resulting from improvements
in air quality over baseline levels of air quality. The incremental costs
associated with a given S02 NAAQS are the additional costs that are incurred
to achieve and maintain improvements in air quality over the baseline levels
of air quality. See Section VI for more detail on baseline air quality
* In welfare economics, the performance of an economic system is measured
•by its ability to satisfy the perceived needs and wants of individuals
(M-jlls, 1978). With greater cost, there is less income available to satisfy
these desires. Consequently, personal comfort and well-being decline.
-------
VII-4
levels. The term "incremental net benefits" refers to the difference
between incremental benefits and incremental costs.
Each alternative S0£ NAAQS is evaluated in terms of incremental benefits,
incremental costs, and relative economic efficiency. Any alternative S02
NAAQS that produces positive net incremental benefits will provide a more
economically efficient allocation of resources than would occur under the
baseline air quality scenario. The alternative S02 NAAQS that results in
the largest positive incremental net benefits will represent the most
efficient allocation of resources among those alternatives considered in
this analysis. If net incremental benefits were negative for all alternatives
considered, then no alternative considered would be economically efficient
and the baseline air quality scenario would yield a more efficient allocation
of resources.
D. Limitations of the Methodology
1. Feasibility Test
A standard for which total benefits exceed total costs is termed "feasible."
Incremental net benefit analysis does not include those costs and benefits
associated with achieving and maintaining the baseline level of air quality.
Therefore, it is possible that the total costs could exceed the total benefits
associated with an alternative SC^ NAAQS, even if incremental net benefits are
positive. This could occur if the cost of baseline controls exceeds the benefits
associated with baseline levels of air quality. If this were the case, society
would be better off, in terms of economic efficiency, if no air quality standards
were adopted, including those already in place to achieve baseline levels of
air quality. Since the criteria of economic efficiency only considers incremental
and not total benefits and costs, it is determined that economic efficiency is
-------
VII-5
a necessary but not sufficient condition for establishing the economic desirability
of an air quality standard. For purposes of this analysis, estimates of baseline
benefits and costs, consistent with methodologies employed in this RIA, have
not been developed. Consequently, total benefits and costs have not been
analyzed, and no feasibility tests have been conducted.
2. Cost-Effectiveness
An analysis of the cost-effectiveness of alternatives is generally
conducted 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 SC>2 NAAQS alternatives is
not analyzed in this RIA.
3. Distributional Impacts
The distributional impacts associated with alternative S02 NAAQS have
also not been analyzed in this RIA. The distribution of the benefits and
costs from a specific regulatory alternative are important because:
0 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.
0 The distribution of adverse impacts may affect the measurement of
the benefits and costs that are appropriate for use in the benefit-
cost analyses.
The potential distribution of adverse economic impacts associated with
the alternative SOg NAAQS should be considered when measuring benefits and
costs appropriate for use in a benefit-cost analysis. The cost estimates
-------
VII-6
developed in the emission control phase of the study were developed assuming
perfectly inelastic price demand elasticity. In other words, as prices
were marked up to cover control costs, the amount of energy demanded by
utility and industrial boiler customers did not change. However, recent
econometric estimates of direct price demand elasticity for the utility
sector show some responsiveness of the quantity of electricity demanded to
price changes. In the short-run the figure for the commercial/residential
and industrial sectors is estimated at about -O.3., while the long-run
estimates are about -0.7 for the commercial/residential and -1.0 for the
industrial sector (Bohi, 1981). For example, the long-run industrial
sector responsiveness to a 10% price increase would be a 10% decrease in
quantity demanded. There would be a 100% decrease in utility sector emissions
(i.e., S02, PM, NOx) associated with the production volume decrease required
to accomodate the decrease in quantity demand. However, in the near term,
this may be mitigated somewhat if utilities shift loads to dirtier plants
and away from cleaner ones or if industrial sources produce pollution in
the process of producing their own steam. But, given full capacity conditions,
operating permit constraints, NSPS, and a full compliance S02 NAAQS scenario,
it seems unlikely that emissions would increase in the face of some demand
responsiveness to price changes. Consequently, in the long-run, failure to
consider the economic impact of marking up prices to cover control costs
will cause an underestimate of benefits because those benefits that would
occur as a result of decreased demand for electricity are not being counted.
The production volume decrease mentioned in the previous paragraph
means that certain control costs will not be incurred. Consequently,
including these costs, as is done with the assumption of perfectly inelastic
-------
VII-7
demand, causes an overestimate of control costs.
The distributional or equity effects have not been analyzed in this
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 S02 NAAQS. Estimates of
these values are unavailable.
Ideally, all feasible and cost-effective S02 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 feasible and cost-effective S02 NAAQS.
E. Scope of the Benefit-Cost Analysis
The scope of the net benefit analysis is consistent with that of both
the benefit and cost analyses. The geographic area is the 31 states east
of and bordering the Mississippi River. The pollutants analyzed are S02,
504, and PM. A subset of health and welfare effects—including S02 mortality
and morbidity, S02 agriculture and materials damage, $04 visibility, and PM
soiling and morbidity—are considered for the purposes of analyzing net
benefits. Utility and industrial emission sources are the focus of the
cost analysis. The alternative S02 standards analyzed include the strict
interpretation of the current S02 NAAQS, the addition to the current standard
of a 0.5 ppm 1-hour S02 NAAQS, and the addition to the current standard of
a 0.25 ppm 1-hour S02 NAAQS. The assumption is made that attainment will
-------
VII-8
commence January 1, 1990 and be maintained through December 31, 2000.
F. Measurement of Benefits And Costs: Conceptual Issues
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 incremental benefits
and incremental costs associated with each of the alternative S02 NAAQS is
presented.
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' wi11ingness-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 (e.g.,
visual range improvement, decreased soiling and materials damage for the
household sector). Direct estimates of society's wi1lingness-to-pay for
cleaner air do not exist for the other benefit categories addressed in this
analysis (e.g., reduced mortality risk). In that instance, alternative
measures are used to estimate willingness to pay such as reduced mortality
risk valuation coefficients obtained from occupational risk studies.
Similarly, an appropriate measure of the cost of pollution emissions
control can be measured as the value that society places on those goods and
* The appropriateness of willingness to pay versus willingness to be
compensated depends on the property right endowments of receptors.
Empirical evidence consistently indicates that estimates of willingness
to be compensated are greater than estimates of willingness to pay.
For additional information on ex ante and ex post assessments of willingness
to pay see Smith, 1986.
-------
VII-9
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. As with the benefits analysis, the conceptually
correct measure of costs was not always employed. Lack of consideration of
the direct price elasticity effects and the overlap between control
requirements for different regulatory programs are two areas where estimated
costs depart from the conceptually correct measure. Some limitations to
the benefit and cost analyses have been discussed in more detail in Section
VII.D and will be further discussed in Section VII.J.
G. Estimates of Benefits and Costs
The estimates of incremental benefits for attaining and maintaining
the three alternative S02 NAAQS from January 1, 1990 through December 31,
2000 are presented in Tables VII.6.1-3. Benefits (and costs) are calculated
using three real interest rates—10%, 5%, and 2%. The interest rate of 10%
is used as directed by the Office of Management and Budget in their guidance
for implementing Executive Order 12291. There is some evidence that the
10% rate is too high. The appropriate rate depends on the value of foregone
consumption and investment opportunities. For purposes of this RIA an
exhaustive analysis has not been done on these foregone opportunities.
Rather, alternative real rates of 5% and 2% are offered for purposes of
comparison. Ranges of benefits are presented to reflect some of the
uncertainties concerning air quality and economic valuation. The benefit
categories estimated include increased visibility due to reduced $64,
reduced chronic and acute morbidity and reduced household soiling due to PM
-------
Table VII.G.I
31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS1
(10% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
>02 Health
502 Welfare
Material Damage
Agriculture
S04 Visibility Benefits
PM Morbidity & Soiling
Current Standards
(Strict Interpretation)
Low Middle
0 .002
.029 .142
0 .004
2.1 2.5
2.1 2.4
High
.225
.262
.008
3.1
2.7
0.5 ppm
1-hour standard
Low Middle
0 .004
.052 .254
0 .007
3.7 4.6
3.9 4.3
Hig_h
.402
.468
.014
5.7
4.8
0.25 ppm
1-hour standard
Low
0
.104
0
8.3
7.6
Middle
.008
.507
.014
9.8
8.8
High
.805
.935
.028
11.8
10.2
Total
4.23 5.05
6.30
7.65
9.17 11.38
16.0 19.13 23.77
1 The assessment only includes a subset of related benefits.
2 The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .2728.
I
I—1
o
-------
Table VII.G.2
31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS1
(5% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)2
Alternative SO? NAAQS
Benefit Categories
S02 Health
S02 Welfare
Material Damage
Agriculture
$04 Visibility Benefits
PM Morbidity & Soiling
Total
Current Standards
(Strict Interpretation)
Low r Middle
0 .004
.050 .242
0 .007
3.3 4.0
3.4 3.8
6.75 8.05
High
.381
.447
.014
5.0
4.3
10.14
0.5 ppm
1-hour standard
Low Middle
0 .007
.090 .433
0 .012
6.0 7.4
6.3 7.0
12.39 14.85
High
.681
.798
.024
9.2
7.8
18.50
0.25 ppm
1-hour standard
Low
0
.180
0
13.5
12.2
25.88
Middle
.014
.866
.024
15.9
14.2
31.0
High
1.36
1.6
.049
19.1
16.4
38.51
The assessment only include^ a subset of related benefits.
The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 5 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .1613.
-------
Table VII.G.3
31 EASTERN STATE BENEFIT ASSESSMENT FOR ALTERNATIVE STANDARDS1
(2% DISCOUNTED PRESENT VALUE IN BILLI ,
-------
VII-13
reductions, reduced mortality, morbidity, and household soiling due to
reduced S02, and increased agricultural yields due to S02 reductions. As is
noted in the table, benefit coverage is incomplete. Benefit categories
not covered are presented in Table VI.B.I. More discussion of incomplete
coverage is presented in Section VII.J.
The estimates of incremental costs for attaining and maintaining the
three alternative standards from January 1, 1990 through December 31, 2000
are presented in Tables VII.G.4-6. These costs are developed using information
in Section IV and in ICF (1984). Costs are also calculated using real
interest rates of 10%, 5%, and 2%. These costs are presented as mid-point
estimates without a range. This should not suggest to the reader that
uncertainty is less critical or non-existent for the cost estimates as
compared to the benefit estimates. As discussed in Section VII.F, lack of
consideration of the price demand elasticity effects and the cost overlap
between regulatory programs may cause costs to be overstated. Other issues,
such as potential scrubber capacity constraint, may cause costs to be
understated.
H. Limitations and Assumptions
The limitations and assumptions of the benefit and cost analyses for
the most part have been discussed in Section VI and IV respectively. They
are summarized here in Tables VII.H.I and VII.H.2 to remind the reader of
the uncertainties inherent in the inputs to the benefit-cost analysis.
Where possible the potential direction of bias imparted by the analytic
limitation or assumption is noted.
-------
Table VII.G.4
31 EASTERN STATE COST ASSESSMENT FOR ALTERNATIVE STANDARDS
(10% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
0.5 ppm
1-hour standard
0.25 ppm
1-hour standard
Costs
3.3
7.0
18.0
The discounted present value of an eleven year stream of costs occurring from Janury 1, 1990 to December 31,
2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of costs for 1990 to
2000, multiply by .2728. These annualized costs are not directly comparable to those presented in Chapter IV
due to the use of different interest rates.
-------
Table VII.G.5
31 EASTERN STATE COST ASSESSMENT FOR ALTERNATIVE STANDARDS
(5% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
0.5 ppm
1-hour standard
0.25 ppm
1-hour standard
Costs
5.6
11.2
27.3
The discounted present value of an eleven year stream of costs occurring from Janury 1, 1990 to December 31,
2000 using a real discount rate of 5 percent in 1984. To convert to an annualized stream of costs for 1990 to
2000, multiply by .1613. These annualized costs are not directly comparable to those presented in Chapter IV
due to the use of different interest rates.
en
-------
Table VI 1.6.6 31 EASTERN STATE COST ASSESSMENT FOR ALTERNATIVE STANDARDS
(2% DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)1
Alternative SO? NAAQS
Current Standards 0.5 ppm 0.25 ppm
(Strict Interpretation) 1-hour standard 1-hour standard
Costs 6.2 13.9 31.6
The discounted present value of an eleven year stream of costs occurring from Janury 1, 1990 to December 31,
2000 using a real discount rate of 2 percent in 1984. To convert to an annualized stream of costs for 1990 to
2.000, multiply by .1297. These annualized costs are not directly comparable to those presented in Chapter IV
due to the use of different interest rates.
-------
VII-17
Table VII.H.I
Limitations and Assumptions of the 31 Eastern State Benefit Analysis
Potential Direction of Bias
Limitations/Assumptions Understates Benefits Uncertain^ Overstates Benefits
Omitted Benefit Effects Categories X
Incomplete Geographic Coverage X
Isolated Point Source Problem X
Characterization
Perfectly Price Inelastic Demand X
for Electricity
Omission of Benefits Associated with X
Baseline Control Levels
No Gas Penetration Into Utility and X
Industrial Boiler Markets
Using Valuation Coefficients From Other X
Studies
Derivation of Midpoint Estimates , X
Use of 10% Real Discount Rate Versus 5% X
or 2%
Use of 1995 Predicted Air Quality Changes X
to Represent All Years of the Analysis
Flat Terrain X
Full Compliance and Rigorous Enforcement X
Worst Case Screening and Modeling X
Assumptions
-------
VII-18
Table VII.H.2
Limitations and Assumptions of the 31 Eastern State Cost Analysis
Potential Direction of Bias
Limitations/Assumptions Understates Costs Uncertain Overstates Co_sjts_
Omission of Costs Associated with X
Baseline Control Levels
Isolated Point Source Problem X
Characterization
Flat Terrain X
Fixed Scrubber Prices for the 0.25 X
ppm 1-hr S02 NAAQS (1990 - 2000)
Transition Cost Not Estimated X
Omission of Administrative, Monitoring, X
and Enforcement Cost
Incomplete Source Coverage X
Subpart D units not controlled for X
0.5 ppm alternative
0.1 Probability that the Fuel has a X
Higher Sulfur Content
0.007 ppm Background for the Current X
Standards
Allocation of Control Possibilities to X
Industrial Boilers, Subpart D units, and
Wet Bottom Boilers First
Multiple Stack Compliance by Assuring X
No Violations When the Most Adverse
Stack is Controlled
Use of 10% Real Discount Rate Versus 5% X
or 2%
Zer Direct Price Demand Elasticity for Electricity X
Full Compliance and Rigorous Enforcement X
Lack of Overlap Considerations Between X
PM and S02 NAAQS Requirements and other
Regulatory Programs (e.g., tall stacks)
No Gas Penetration Into Utility and X
Industrial Boiler Markets
Worst Case Screen for 1-hr Standards X
Most Adverse Load Conditions X
-------
VII-19
The reader is referred to the above mentioned sections and the
corresponding technical reports for a detailed description of the limitations
and assumptions.
I. Net Benefits
The estimates of the discounted present value of incremental benefit
minus incremental costs, or net benefits, are presented in Tables VII.I.
1-3. Net benefits are calculated for three interest rates - 10% , 5% and
2%. The ranges of net benefits presented reflect only some of the uncertainties
associated with the benefit analysis and none of the uncertainties associated
with the cost analysis. See Tables VII.H.I and VII.H.2 for more detail.
As can be seen from Table VII.I.I (10% case) the full range of the current
standards and the 0.5 ppm standard has positive net benefits. For the
0.25 ppm standard the middle and high end of the range has positive net
benefits while the low end of the range has negative net benefits of two
billion dollars.
The same pattern holds true for Table VII.1.2 using a 5% discount
rate for costs and benefits. All scenarios yield positive net benefits
with the exception of the low estimate 0.25 ppm case where net benefits
are a negative 1.4 billion. The net benefit calculations using a 2% discount
rate yields positive net benefits for all scenarios without exception.
An implicit valuation analysis was conducted to estimate how big the
504 mortality coefficient would have to be for incremental benefits to
equal incremental costs for the low estimate 0.25 ppm standards using 10%
and 5% discount rates. For the implicit valuation analysis to be completely
valid, 504 mortality risk must be the only unquantified benefit category.
As discussed earlier and shown in Table VI.B.I many benefit categories
-------
Table VII.I.I 31 Eastern State Net Benefit Assessment for Alternative Standards
H0% Discounted Present Value in Billions of January 1984 Dollars)1
Alternative SO? NAAQS
Current Standards
(Strict Interpretation)
0.5 PPM
1-Hour Standard
0.25 PPM
1-Hour Standard
Low Middle High
Low Middle High
Low Middle
High
Net Benefits
.9
1.8
3.0
.7
2.2
4.4 -2.0
1.1
5.8
1X3
O
The discounted present value of net benefits using a 10% real discount rate. These are derived from
incremental benefits (Table VII.G.I) and incremental costs (Table VII.G.4).
-------
Table VII.1.2 31 Eastern State Net Benefit Assessment for Alternative Standards
(5% Discounted Present Value in Billions of January 1984 Dollars)1
Alternative SO? NAAQS
Current Standards 0.5 PPM 0.25 PPM
(Strict Interpretation) 1-Hour Standard 1-Hour Standard
Low Middle High Low Middle High_ Low Middle High
,•.. F ,;•-
Net Benefits ';J 1.2 2.5 4.5 1.2 3.7 7.3 -1.4 3.7 11.3
The discounted present value of net benefits using a 5% real discount rate. These are derived from
incremental benefits (Table VII.G.2) and incremental costs (Table VII.G.5).
-------
Table VII.1.3 31 Eastern State Net Benefit Assessment for Alternative Standards
(2% Discounted Present Value in Billions of January 1984 Dollars)^
Alternative SO? NAAQS
Current Standards 0.5 PPM 0.25 PPM
(Strict Interpretation) 1-Hour Standard 1-Hour Standard
j-ow Middle MLi!l Low Middle High Low Middle High
Net Benefits 3.0 4.9 7.7 3.1 6.2 11.4 3.8 10.8 21.0
The discounted present value of net benefits using a 2% real discount rate. These are derived from
incremental benefits (Table VII.G.3) and incremental costs (Table VII.G.6).
-------
VII-23
have not been quantified in this RIA. In addition, the implicit valuation
analysis will be valid only as long as all other benefit and cost estimates
are accurate. Considering these constraints—the uncertainties surrounding
both the benefit and cost estimates and the obvious limitations to benefits
coverage—the $04 mortality risk coefficient from the implicit valuation
analysis could be overestimated.
Table VII.1.4 presents the results of the implicit analysis of
mortality risk coefficients. In this analysis, as in Section VI, Appendix
B, mortality valuations of $7.3 and $.420 million per statistical life
saved are used. The mortality risk coefficient that will make incremental
benefits and costs equal for the low estimate 0.25 ppm standard using a
10% discount rate are: .26042 for $.420 million valuation and 0.1496 for
$7.3 million valuation. The risk coefficient that makes incremental
benefits and costs equal using the 5% discount rate are: .11272 for $.420
million valuation and .00648 for $7.3 million valuation. These coefficients
are in terms of the number of deaths avoided per 100,000 population for a
n
pg/nr reduction in annual SO^ levels.
In relative terms the coefficients from the 10% real discount rate
case for the $.420 million and $7':3 million valuations are .06% and .004%
respectively of the mortality rate for cardiovascular disease, .6% and .03%
respectively of the mortality rate for respiratory disease, and .03% and
.002% respectively of the mortality rates for both types of diseases combined.
The risk coefficients from the 5% real discount rate case for both the
$.420 million and $7.3 million valuations are .03% and .002% respectively
of the mortality rate for cardiovascular disease, .24% and .01% respectively
of the mortality rate for respiratory disease, and .01% and .0008% respectively
of the mortality rates for both types of diseases combined.
-------
VII-24
Table VII.1.4
Implicit Valuation of the SO^ Mortality Risk Reduction Coefficient^
(Lives saved/100,000 population/ug/nr reduction in SO^ annually)
Alternative SO? NAAQS
Implicit Valuation Coefficient^
$420,000 $7.300,000
Low estimate 0.25 ppm - 10%
Low estimate 0.25 ppm - 5%
.26042
.11272
.01496
.00648
The value of the mortality risk reduction coefficient required for
incremental benefits to equal incremental costs. This coefficient
is derived assuming the value of a statistical life saved is $7.3
million.
Valuation coefficients of $.420 and $7.30 for decreased mortality
risks of 1.0 X 10~6. See Tables B.I and B.2 in Appendix B for more
detai1.
-------
VII-25
J. Limitations to the Analysis
There are some important limitations to the interpretation of this
analysis that need to be considered prior to discussing the findings. On
the cost side, not considering direct price demand elasticity effects is
one area where estimated costs depart from the conceptually correct
measure of costs. Another area that deserves more attention is the
overlap which exists between control requirements for the PM NAAQS and
those associated with the alternative S02 NAAQS. The utility sector
accounted for more control cost than any other emitter in both the PM and
S02 NAAQS analyses. Comparison of the utility power plants requiring
stack controls to meet the present TSP secondary NAAQS with those requiring
S02 emission controls to meet the 0.25 ppm S0£ 1-hour NAAQS alternative
revealed an overlap of at least 22 percent. The overlap between utility
power plants requiring fugitive emission controls, but not stack controls,
for the current TSP secondary NAAQS and S02 emission control for the 0.25
ppm S02 1-hour NAAQS alternative is at least 29 percent. Scrubbers will
reduce S02 as well as PM emissions. Moreover, they are predicted to be
used to some extent for all the considered S02 NAAQS alternatives.
Consequently, not all the scrubber costs estimated in this analysis may
be attributable to the S02 NAAQS; some may be justly apportioned to the
PM NAAQS. Considering the additional regulatory requirements of the
alternative PM and S02 NAAQS simultaneously should cause the estimated
cost of complying with both standards to fall.
A related argument can be made regarding jointness or overlap on the
benefit side where the preferred solution is simultaneous consideration
-------
VII-26
of standards. Ideally not only the measures of air quality (e.g., S02,
$04, PM]_Q, etc.) but also the levels would be indicative of economic
efficiency considerations. A less efficient alternative may be to order
the establishment of related NAAQS on the basis of their net benefit
rankings. For example, if the net benefit for the most efficient PM^g
alternative is greater than the corresponding alternative for $04 and
that is greater than the corresponding alternative of SC>2, the preferred
orderings might be PM^g first, $04 second, and the SC>2 third. Another
option (not necessarily mutually exclusive from the others) might be to
reflect the averaging time of the dominant benefit category in establishing
the air quality index. For example, if most of the benefits come from
reduced 24-hour or annual concentrations and not 1-hour peaks, perhaps
the longer term averaging times should be reflected in the standard.
Still, another option (not necessarily mutually exclusive from the others)
might be to allow each state to simultaneously develop State Implementation
Plans for PM]_Q, $04, and the S02 NAAQS as opposed to the sequential
process now practiced. An overriding limitation to be kept in mind while
reviewing the results of the net benefit analysis is the lack of complete
benefit category coverage. The reader is reminded of Table VI.B.I which
shows many benefit categories not included in this RIA and many categories
that are not fully covered.
K. Qualifications and Findings
1. Qualifications
As noted earlier in Section VII, the air quality standards will have
different impacts on different members of society; those who enjoy the
benefits of the standards will not always be the same as those who bear
-------
VII-27
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 SC>2 NAAQS cannot be estimated with
certainty. Moreover, in some cases, the definitions of both benefits and
costs employed 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 cases analyzed.
The benefit-cost analyses described in this section are restricted in
scope in that they consider only a limited selection of possible S02 NAAQS. As
a result, the benefit-cost analysis will identify, with uncertainty, only the
most efficient S02 NAAQS from among those considered. The identification of
the most efficient possible standard requires an evaluation of all cost-effective
and feasible S02 NAAQS.
The results of the benefit-cost analyses that follow should be interpreted
in light of these factors. Specifically, these analyses provide but a qualified
assessment of the relative economic efficiency of the alternative S02 NAAQS.
-------
VII-28
2. Findings
Four findings can be drawn from the information presented in Section
VII.
a) Large Uncertainty Surrounds the Analytic Process
The validity of the analyses conducted and summarized in Section VII
depends on the estimates of air quality, costs, and benefits. However,
the collection of limitations and assumptions presented in Tables
VII.H.I and 2 as a whole show no clear directional bias. Consequently,
whether the differences between incremental benefits and costs as
presented here are significant or represent over or underestimates of
the true differences is unknown. Furthermore, there is a possibility
that a statistical assessment of the uncertainty underlying these
estimates would suggest that the estimated differences between
incremental benefits and incremental costs are not significantly
different from zero.
b) Within the Limitations of This Analysis More S02 Control is Supported
on Grounds of Economic Efficiency
As can be seen from examining Tables VII.1.1-3 in all cases more S02
emission reduction is preferred to the status quo. In 25 of the 27
combinations of alternative standards, discount rates, and benefit
assumptions analyzed net benefits are positive. In these cases,
adoption of the alternative standards will improve society's well-
being relative to the baseline conditions. In these cases with
positive net benefits, the alternative S02 NAAQS provides a more
economically efficient allocation of resources than would occur under
the baseline air quality scenario. In the cases of the low 0.25 ppm
-------
VII-29
alternative with 10% and 5% real discount rates the regulatory
alternative is not economically efficient and is not preferred
over the baseline. This does not mean that no alternative
regulation using the low benefit assumption and either 10% or
5% real discount rates is preferred over the baseline. As can
be seen from the tables, both the current standard and the 0.5
ppm standard (low estimate, 10% and 5% real discount rates) are
preferred on economic efficiency grounds over the baseline.
For those two cases--.!ow 0.25 ppm 10% and 5%--where net
benefits are negative the implicit valuation of the mortality
risk coefficient indicates that if benefit coverage was expanded
to include mortality benefits the true net benefits could be
positive. Table VII.1.4 presents evidence for this observation.
c) Within the Limitations of this Analysis the Degree of Additional
Control Warranted Remains Ambiguous
The regulatory alternative that yields the highest net benefits
shifts between or among the three benefit assumptions and
between or among the three discount rates used. The following
table illustrates this point.
-------
VII-30
Table VII.K.I
Ordering of Preferred Standards
Benefit Assumptions
Low
Middle
High
Discount Rate
10%
Pref. 2nd 3rd
CS .5 BL
.5 CS .25
.25 .5 CS
5%
Pref. 2nd 3rd
CS/.5 CS/.5 BL
.B/.25 .57.25 CS
.25 .5 CS
2%
Pref. 2nd 3rd
.25 .5 CS
.25 .5 CS
.25 .5 CS
1 The preferred standard is the one yielding the greatest net benefits.
^ CS = current standards with strict interpretation
.5 = 0.5 ppm S02 NAAQS alternative
.25 = 0.25 ppm S02 NAAQS alternative
BL = Baseline
The above table provides nine situations in which preferred (hiqhest
positive incremental net benefits) standards can be identified. In five
of these nine situations the 0.25 ppm standard yields the highest positive
incremental net benefits. The 0.5 ppm standard and the current standard
are each preferred in one of the nine situations. In two situations it
cannot unequivocally be stated which standard is preferred. Hence, within
the bounds of this analysis no one alternative standard is preferred across
the board to the baseline conditions.
It is important to remember in analyzing the ordering of preferred
standards that while the low net benefit scenario reflects a low estimate
of benefits, it reflects a midpoint estimate of costs. The high net
benefit scenario reflects a high estimate of benefits, but a midpoint
estimate of costs. Therefore, the most equal situation to compare for
identification of the preferred standard is the middle estimate of incremental
-------
VII-31
net benefits for the 10%, 5% and 2% real discount rates. Upon examination
of Table VII.K.I one can see that no one regulatory alternative is
dominate. The 0.5 ppm alternative is preferred with a 10% rate, the 0.5
ppm and 0.25 ppm alternatives are equally preferred with the 5% rate, and
the 0.25 ppm alternative is preferred using the 2% real discount rate.
The sensitivity of net benefits to increases in the benefit and cost
estimates was tested. The reader should remember that the low benefit
estimate is comprised of low benefit assumptions but midpoint cost
assumptions. Likewise, the high benefit estimates are comprised of high
benefit assumptions but midpoint cost estimates. If the 0.25 ppm low
benefit estimate using a 10% real discount rate was increased by 19%,
the 0.25 ppm alternative would be preferred over the current standard or
the 0.5 ppm standard. For the 0.25 ppm low estimate 5% real discount
rate case, if benefits were increased by 11% the 0.25 ppm alternative
would be preferred over the other two alternatives. The low estimate
0.25 ppm alternative using a 2% real discount rate is already the preferred
standard. A similar sensitivity analysis was done to determine by how
much the costs for the high net benefit estimate would have to increase
for the ordering of preferred standards to shift from the 0.25 ppm to the
0.5 ppm standard. If the cost estimate in the 0.25 ppm high net benefit
case using a 10% real discount rate increased by 32% the 0.5 ppm alternative
would be preferred over the 0.25 ppm alternative. This percentage increases
to 41% and 67% for the 5% and 2% real discount rates.
d) Overlap of PM and S02 NAAQS Benefits and Costs
Given the overlap and relative magnitudes of estimated S04, PM,
-------
VII-32
and S02 benefits, economic efficiency gains may be realized by
simultaneous establishment of PM (FP, PM]_Q, TSP, etc.) and SC>2
NAAQS. Likewise, integrated (i.e., multiple pollutant) control
strategy development in the State Implementation Plan process may also
foster efficiency gains.
-------
VII-33
List of References
Bohi, D.R., (1981). A Study of Energy Elasticities. Johns Hopkins University
Press, Baltimore.
ICF, Inc., (1984). Analysis of Alternative Sulfur Dioxide Ambient Standards.
Prepared for the U.S. Environmental Protection Agency.
Mills, E.S., (1978). The Economics of Environmental Quality. W.W. Norton
and Company, New York.
Smith, V.K., (1986). Advances in Applied Micro-Economics, Volume 4. JAI
Press, Inc.
U.S. Environmental Protection Agency, Office of Policy Analysis. Guidelines
for Performing Regulatory Impact Analysis. Washington, D.C., December
1983.
-------
VIII. Summary of Rationale for Choosing Proposed Action
The full basis for the proposed decision not to revise the current
S02 NAAQS is contained in the Federal Register preamble. The following is
a brief summary of the preamble; for additional detail the reader is referred
to the preamble itself.
A comprehensive review of the scientific data in the criteria document
and its addenda, as well as a variety of analyses suggests that continued
implementation of the current suite of standards would provide substantial
protection against the direct health and welfare effects associated with
S02.
The basis for the current 24-hour standard rests largely on epidemiologies!
studies conducted in London in the 1950's and 1960's. The principal
effects identified were increased daily mortality and aggravation of bron-
chitis. More recent studies have suggested the possibility of decreased
lung function. Based on available scientific and technical information
as well as the recommendations of the Clean Air Scientific Advisory
Committee (CASAC) the Administrator concluded that the current 24-hour
primary NAAQS at 0.14 ppm should be retained.
With respect to the current annual standard, the scientific data
provide some qualitative evidence that long-term exposure to elevated SOg
concentrations may lead to potential health effects. However, there is no
single study or studies which would provide a clear quantitative basis for
the annual NAAQS. EPA staff analyses have shown that the annual standard
does limit emissions in a number of urban areas and thus moderates both
acute and chronic exposures. The CASAC agreed that there is a need to
protect against increases in ch'ronic exposures, but found little quantitative
-------
VIII-2
support in the literature for maintaining the present annual primary NAADS.
They recommended that the decision on the annual standard he taken "in
light of the total protection that is to he offered by the suite of
standards..." In addition, they pointed out that the most persuasive
scientific basis for the annual standard was in the potential welfare
effects. The Adrni nistrator proposes to maintain the current annual NAAOS.
The notice presents the case that despite the uncertainties in the information
base, the retention of the current annual standard is a prudent public
health policy choice that will limit any increase in acute or chronic
exposures in urban areas now meeting the standards. Since no revision to
the current annual primary standard is being proposed and since it is
attained virtually everywhere, there appear to he no practical benefits to
be gained from proposing an annual secondary standard.
The current secondary standard is a 3-hour standard of 0.5 ppm based
on acute effects on sensitive plants (e.g., reductions in growth and yield,
as well as foliar injury). Since the original standard was set there have
been significant additions to the scientific literature in this area. The
staff in their assessent found even stronger support for the current secondary
standard and with the CASAC recommended its retention. Considering both the
staff and CASAC recommendations, the Administrator proposes no revisions to
the current secondary standard.
Although both the EPA staff and CASAC concluded that the current suite
of SOg standards were adequate to protect against the health and welfare
effects associated with the 24-hour, annual, and 3-hour averaging periods,
the central concern of the MAAQS review was the potential effect of
short-term peak exposures on asthmatics. A considerable body of scientific
-------
VIII-3
studies has accumulated in the past few years showing Pleasurable changes in
respiratory function and symptoms in asthmatics exposed to 0.4 - 0.75 ppm
SOg or more for periods of 5 minutes to an hour. Based on these studies,
both the EPA staff and CASAC recommended that consideration should be given
to the option of setting a new short-term (1-hour) primary standard.
Several points should be made regarding the effects themselves. The
functional changes and symptoms reported in the controlled human exposure
studies appear to be transient and reversible. They are, in general,
not equivalent to the more severe responses that accompany an asthma
"attack." Due to their susceptibility to responses from other stimuli, many
asthmatics already routinely use medications which can prevent or ameliorate
response to S02 exposure. Thus, there is a divided opinion in the scientific
community as to whether and to what extent these effects should be considered
adverse health effects.
Since the effect was observed in a controlled clinical setting, a key
question to be answered in the NAAQS review was how likely were asthmatics
to be exposed given current conditions. A related question, of course,
was the likelihood of exposure to peak concentrations given attainment of
the current standards, as well as various alternative standards. The EPA staff
completed a number of analyses which examined current air quality levels
and simulated population exposures in the immediate vicinity of utility
power plants. In completing these analyses, the staff focused on 5-minute
peaks in excess of 0.5 ppm S02 as a benchmark of concern. This .benchmark
was chosen based on other analyses showing that 25% or less of asthmatics might
experience at least a doubling of airway resistance (with a smaller
percent experiencing noticeable symptoms) if exposed to such levels while
-------
VIII-4
at exercise. The staff analyses found that the current standards might
permit 1 to 14% of the asthmatics residing around major point sources
to be exposed to such peaks Wiile they were at exercise. Given that only
10 to 40% of the asthmatic population resides in the vicinity of such major
sources, Agency cal culated that fewer than 1% of all U.S. asthmatics would
experience such exposures. Other work has shown that 25% or less of asthmatics
would experience even moderate pulmonary function changes and symptoms if
exposed under such conditions.
Given these considerations, the Administrator is presently inclined
inclination to conclude that the current standards provide adequate protection
against potential short-term effects of S0£ and that a 1-hour primary
standard is not needed. Given the protection of the current NAAOS, the
occasional remaining short-term exposures that occur are not judged by the
Administrator to constitute a significant public health problem that requires
a new NAAOS. For the same reasons, the 1982 recommendation made by some
CASAC members, that in reaffirming the standards, the current 3-hour secondary
standard be made a primary standard, is not followed. Since the present
standards are widely implemented, no practical environmental benefit would
result from such a change.
In reaching these conclusions, the Administrator is mindful of the un-
certainties involved and the diversity of opinion on the subject. To promote
a full public discussion of the issues, the basis for selecting a 1-hour standard
and revising the current NAAQS is presented. Given typical 5-minute peak to
1-hour mean ratios of 2 to 1, a 1-hour NAAQS of 0.3 to 0.4 ppm could result in
5-minute peaks on the order of 0.6 to 0.8 ppm. Several CASAC members supported
1-hour standards in this portion of the overall range. Rased on a consideration
of the views of CASAC as well as the uncertain significance of the health
-------
VIII-5
effects and the infrequence of inducement of such effects by S02, a 1-hour
standard level of 0.4 ppm is suggested for public comment. If such a
standard were set, other revisions would also be considered. Specifically,
EPA would consider replacing the current 3-hour secondary with a 1-hour
secondary equal to the 1-hour primary. In addition, the Agency would give
serious consideration to adopting a statistical form (single expected
exceedance) for all of the SOg NAAQS.
A complete discussion of the rationale is to be found in the
Federal Register preamble.
-------
IX-1
IX. STATUTORY AUTHORITY
The statutory authority for the proposed sulfur oxides NAAO.S is
contained in the Clean Air Act. Two sections of the Act govern the
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.
-------
APPENDIX A
1. Introduction
The purpose of this appendix is to provide more information on
the procedures for calculating the benefits estimated quantitatively in
Section VI. Detail is provided on input data and the the benefit calcu-
lations. The appendix is organized by benefits related to 804, PM, and
S02 concentration reductions. In each section some sample calculations
are provided.
2. $04 Benefits
The $04 benefits estimated in Chapter VI were associated with
visual range improvement. These improvements are due to decreases in
$04 concentrations from compliance with alternative S02 standards for the
thirty-one eastern states. The visibility numbers used for each state are
found in Section IV.D.4 The numbers represent annual average visual range
estimates for each state. The high estimate of visual range change uses
the results of the Astrap model with a lower ratio of extinction coefficient
due to nonsulfate contribution than to sulfate contribution (N=l). The
middle estimate uses the arithmetic average of the results of the Astrap
and Monte Carlo models with a middle ratio (N=1.25). The low estimate
uses the results of the Monte Carlo model with a high ratio (N=1.5).
-------
A-2
The estimation of visibility benefits uses information on willingnes
to pay on a household basis. The visibility valuation data is used
together with the visual range improvements to generate the benefits of
visual range improvements for the 31 eastern states. The visibility
valuation data represents the annual willingness to pay of the mean
household for a one kilometer improvement in visual range. The average
annual household values from each of the studies (Brookshire, et al. 1979,
Rowe et al. 1980, Rae 1983, Loehmann et al. 1981, Tolley et al. 1986)
for each hypothesized change in visual range were compiled to estimate
Equation A.I.
Benefits = [(b/a) (VR2-VR1) + (c/a) (VR22-VR12)J f(x) A.I
This equation suggests that the bids from the surveys can be expected to
be a function of the change in visual range considered and of the base
level (VR1) and new level (VR2) of visual range hypothesized. The function
f(x) was presumed to be held constant across the different studies. The
variables used in this analysis are defined in Table A.I and the entire
data set used can be found in Table A.2.
Four dummy variables were defined for study characteristics that might
influence the value estimate. These are: RANK, DIST, PRETEST, and WEST
and are defined in Table A.I.
Table A.3 shows the OLS estimates of Equation A.I. Both DVR and DIFSQ
are significant. The negative coefficient for DIFSQ combined with the
positive coefficient for DVR indicates that over the range of visibility
considered in these studies, the value increases with bigger changes in
visual range, but at a decreasing rate.
-------
A-3
TABLE A.I
VARIABLES USED IN ANALYSIS OF RESULTS FROM CONTINGENT VALUATION STUDIES
Variable
Description
Mean
CITY
BID
VR1
VR2
DVR
PERHILE
DIFSQ
PERCENT
LOGRAT
RANK
DIST
PRETEST
WEST
Cities where the means were estimated—entered
separately if more than one city covered in one
study. Value codes as follows (date of survey):
1 Chicago - Tolley et al. (1981)
2 Atlanta - Tolley et al. (1982)
3 Boston - Tolley et al. (1982)
4 Cincinnati - Tolley et al. (1982)
5 Miami - Tolley et al. (1982)
6 Mobile - Tolley et al. (1982)
7 Washington DC - Tolley et al. (1982)
8 Los Angeles - Brookshire et al. (1978)
9 San Francisco - Loehman et al. (1980)
10 Cincinnati - Rae et al. (1982)
11 Farmington, New Mexico -Rove et al. (1977)
Annual willingness to pay per household in 1984 100.47
dollars. Mid points of ranges were used for Rae
study. Format C2 (payment card with private and
public goods)'was used for Tolley et al. pretest.
Visual range in miles that was the presumed 14.98
starting point for the hypothesized change
Visual range in miles that was the hypothesized new 19.48
level
Change in visual range hypothesized (VR2 - VR1) 4.50
BID/DVR 29.64
[(VR2)2 - (VR1)2] -4.74
DVR/VR -82
LOG(VR2/VR1) .34
1 - contingent ranking method was used. .17
1 - visual range was presented as a distribution of .17
several levels, not a single average value.
1 - study was a pretest for a larger effort. .19
1 - study conducted in the Western U.S. .19
-------
CITY
BID
TABLE A.2
DATA USED IN ANALYSIS OF RESULTS FROM CONTINGENT VALUATION STUDIES
VRl
VR2
DV/R
DIFSQ
PERCENT
LOGRftT
MONK
OIST
1
1
1
£
2
^
3
3
3
4
4
4
5
5
S
6
6
6
7
7
7
8
8
9
9
10
10
10
10
10
10
10
10
11
11
1 1
-323
369
520
-212
203
309
-137
130
185
-62
62
69
-107
93
112
-168
181
213
-251
237
327
94
132
91
-155
498
£3 a,.
133
94
464
239
117
162
-134
-97
-72
9.0
9.0
9.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10. 0
10.0
10.0
10.0
10.0
10.0
10.0
2.0
12.0
16.3
18.6
7.8
11.4
11.4
7.8
11.6
10.9
10.9
11.4
73.0
75.0
50.0
4.0
18.0
30.0
3.0
£0.0
30.0
3.0
20.0
30.0
3.0
20.0
30.0
3.0
20.0
30.0
3.0
20.0
30.0
3.0
20.0
30.0
12.0
28.0
18.6
16.3
23.2
21.0
13.8
23.2
16.4
14.4
11.8
16. 4
23.0
30.0
23.0
-5.0
9.0
21.0
-3.0
10.0
20.0
-5.0
10.0
20.0
-3.0
10.0
20.0
-5.0
10.0
20.0
-5.0
10.0
20.0
-5.0
10.0
20.0
10.0
16.0
2.3
-2.3
17.4
9.6
2.4
17.4
4.8
3.5
0.9
5.0
-50.0
-25.0
-23.0
64. 600
4 1 . 000
24.762
42. 400
20. 300
13. 450
31.400
13.000
9.230
12.400
6.200
3.450
21.400
9.500
3.600
33. 600
18. 100
10.650
50. 200
23. 700
16.350
9.400
8.250
39. 565
67.391 '
28.621
24.896
63. 750
5.402
96.667
74.000
130.000
32. 400
2.680
3.8B0
2.880
-65.0
243.0
819.0
-75.0
300.0
800.0
-75.0
300.0
800.0
-73.0
300.' 0
800.0
-75.0
300.0
800.0
-75.0
300.0
800.0
-73.0
300.0
800.0
140.0
640.0
80.3
-80.3
374.2
311.0
60.5
574.2
134.4
88.6
20.4
139.0
-5000. 0
-3125.0
-1875.0
-*. 53S3b
1. 0000*
£. 33333
-0. 30*00
1 . 00000
2. 00000
-0. 50000
1 . 00000
2. 00000
-0. 50000
1 . 00000
2.00000
-0. 50000
1.00000
2. 00000
-0. 50000
1.00000
2. 00000
-0. 50000
1.00000
2. 00000
3. 00000
1.33333
0. 14110
-0. 12366
2.23077
0.8421 1
0.21053
2. 23077
0.41379
0.3,2110
0.08257
0.43860
-0. 66667
-0.33333
-0. 5(3(300
-0. 8109
0.6931
1. 2040
-0.6931
0.6931
1.0986
-0. 693 1
0.6931
1.0986
-0.6931
0.6931
1.09B6
-0.6931
0.6931
1.0986
-0.6931
0.6931
1.0986
-0.6931
0.6931
1.0986
1.7918
0.8473
0. 1320
-0. 1320
1. 1727
0.6109
0. 1911
1. 1727
0.3463
0.2785
0.0793
0.3637
-1.0986
-0.4055
-0.6931
(2>
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
1
1
0
0
1
1
0
0
0
0
PHETEST
WEST
1
1
1
Id
0
Id
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
Id
0
0
a
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
1
1
1
-------
A-5
TABLE A.3 '
REGRESSION ESTIMATES OF EQUATION A.I FOR ALL CONTINGENT VALUATION STUDIES
WITH DUMMY VARIABLES FOR DIFFERENCES IN STUDIES
Variable
DVR
DIFSQ
RANK
DIST
PRETEST
WEST
N2 = 36
R = .77
F = 16.33
Coefficient
19.74
-.17
258.15
-91.41
-5.23
-6.39
t-statistic
5.62
-3.11
3.28
-1.14
-.10
-.09
-------
A-6
An example of the procedure calculating the middle estimate benefits
for the .25 ppm 1-hour $62 NAAQS alternative is described below. Given
the following data for Alabama: 1980 population - 3,890,000; 1980 number
of households - 1,342,000; 1990 projected population - 4,214,000; 2000
projected population - 4,415,000, we estimate the 1997 Alabama population
to be 4,353,714 people and the number of households in 1997 to be
1,501,988. The middle estimate for benefits is Equation A.I.
[(19.74) (16.3-13.3)-(.17)(16.32-13.32)J times 1,501,988 households times
1.047 (1982 to 1984 Consumer Price Index inflator) or $69,388,583.
This number is the estimate of visibility benefits, in Alabama, in 1997,
expressed in constant January 1984 dollars.
Visibility benefits are aggregated across time by discounting
back to January 1984 with a 10% real discount rate. The discount formula
for year n is 1 . Therefore, the Alabama 1997 local
(1 + .I)"'1984
1 13
visibility benefits become $69,388,583* ( 1.1 ) = $20,099,401. These
discounted values for the years 1990 through 2000 are added to reach an
estimate for each state and the totals for each of the thirty-one states
are added to give the thirty-one state total.
3. PM Benefits
The particulate matter benefits estimated in Chapter VI included
reduced acute morbidity, reduced chronic morbidity, and reduced house-
hold soiling. The analysis of particulate matter (PM) benefits due to
S02 changes associated with alternative standards for the thirty-one
eastern states is computed using the sulfate numbers found in Section
IV.D.4. The low estimate uses the Astrap model results, the middle
-------
A-7
estimate uses the average of Astrap and Monte Carlo results, and
the high estimate uses Monte Carlo results. The 504 changes at a state
level are computed by subtracting the S02 concentration estimated for each
standard from the 864 ambient concentration of the baseline. The $04
changes are translated into TSP changes by multiplying by 1.4 for the low
estimate, 1.5 for the middle estimate, and 1.6 for the high estimate.
The population estimates are obtained in the same way as described
above in the explanation of estimating number of households. Since
population, not number of households is needed, no conversion of population
to number of households is needed.
The benefit calculation employs an average benefit per person per
microgram number of $3.21. Thus the benefit calculation for a given
state and year consists of multiplying the population estimate times the
change in 804 concentration times 1.4, 1.5, or 1.6 times $3.21.
For example, the low estimate for the .25 ppm 1-hour alternative for
Alabama in 1997 is 4,353,714 people * (6.940 * .22723 micrograms per meter
cubed) * 1.4 * $3.21 per microgram per meter cubed per person =
$30,854,471.00. As in the visibility example this number is discounted
at a 10% real rate to reflect a present value in 1984 of $8,937,367.
4. S02 Benefits
The SOg benefit estimates in Table VI.E.3 are based on a case study
analysis of four power plants. Below, the method of estimating benefits in
the case, study is described. Then, the extrapolation of the case study
results to the analysis of 31 eastern states is discussed.
There are two basic types of inputs required to perform the benefit
analysis. First, baseline and post-control levels of air quality must be
-------
A-8
available. Second, quantitative concentration-response functions must be
identified for each benefit category of interest.
Ai r Quality Data
Data were provided by SAI for four power plants. The power plants
analyzed include: Eddystone (near Philadelphia); Wansley (Georgia);
Potomac River (Washington, D.C.); and Portage des Sioux (near St. Louis).
The information available for each plant included:
o Background concentrations (ug/m3), fixed over time and
space
o Baseline (pre-control) average emission rates, fixed in
time (Ib/MMBtu)
o Post-control average emission rates (Ib/MMBtu), fixed in
time
o Normalized concentrations (pg/m3), indexed to a unit
emission rate (1 Ib/MMBtu).
Table A.4 summarizes the background and emissions rate data by plant. How-
ever, since the concentration data vary in time and space due to variations
in meteorology, they are more difficult to portray. Normalized concentra-
tion data were provided for 180 receptors around each power plant. The
receptors were arranged in a polar coordinate grid system consisting of
five concentric rings with the power plant at the center of the grid. The
radius of the outermost ring ranged from 12 kilometers at the Eddystone
plant to 30 kilometers at the Potomac River power plant.
With 180 receptors and 8,760 hours in a year, over 1.5 million
normalized concentrations would have been predicted for each plant. Given
that the emissions rates (pre- and post-control) are to be multipled by the
normalized concentrations to obtain both pre- and post-control concentra-
tions, over 3 million pieces of data would be generated.
-------
A-9
To reduce the scope of the analysis, a cut-off was imposed on the
normalized concentrations associated with unit emission rates at each power
Table A.4
INPUT DATA THAT ARE FIXED IN TIME AND SPACE
Plant
Eddystone
Wansley
Potomac River
Portage des Sioux
Background
(Mg/m3)
286
80
150
234
Pre-Control
Emissions
(Ib/MMBtu)
2.91
4.64
1.14
4.66
Post-Control
Emissions
(Ib/MMBtu)
2.54
3.85
0.64
2.50
plant. If the modeled normalized concentration fell below the threshold, a
value of zero was assumed at that receptor and for that hour. This implies
that concentrations would default to background for that receptor and hour
in both pre- and post-control scenarios. The cut-offs were 30 ug/m3 at the
Eddystone and Portage des Sioux plants and 10 ug/m3 at the other two power
plants. Use of the cut-offs reduced the volume of data by 90 percent.
Concentration-response Functions
Benefits are calculated for four benefit categories:
o increased agricultural yields
o reduced mortality risk
o reduced morbidity
o decreased household sector materials damage
-------
A-10
For each category, a concentration-response function is applied to the
air quality data. A concentration-response function relates a physical
effect to a measure of SC^:
y = f(x;S02)
where y is the effect response; S02 is a measure of S02 concentration, and
x is a vector of other relevant explanatory variables. Subsequent discus-
sion identifies more clearly examples of the y and x variables. In
general, benefits are generated by valuing the change in y associated with
the change in S02- A single calculation may be done for each case study
area using an area-wide index of SC^. Alternatively, calculations may be
done for each grid separately. The selection of the geographic area
covered by a calculation is dictated by the size of the geographic area
used in the original study.
In the case study, benefits are estimated for each year in the period
1990-2000. Below, the method of estimation for a single year is described
in detail for each benefit category. A description of the data sources
for variables in the concentration-response functions is given in Table
A.5. The derivation of valuations attached to the physical effects
end-points also is summarized in the table.
a. Increased Agricultural Yield
The case study considers three crops: wheat, oats, and soybeans.
The effect of reduced S02 on the yield of each crop is estimated at the
receptor grid level.
-------
A-ll
Table A.5
DATA SOURCES
Variable
Source
Population
Households
Income
Asthmatics
Population Share
Share of Asthmatics
at High Exercise
Level
Valuation of
Mortality Risk
Reductions
Valuation of
Symptom Reduction
Cost of Doctor Visit
in Emergency Room
Crop Prices
SAI provided base grid-level population data. County-
level growth rates from the Bureau of Economic Analysis
(1981) were used to increase population over time.
Base county household data were taken from Bureau of the
Census (1980),, Projections of household growth were
based on population projections and projections of
average national household size.
State-level income projections reported by the U.S.
Department of Commerce News (1980) were employed.
Regional shares were taken from the National Health
Interview Survey (1970). The 1970 estimates were
updated using the national growth rate implied by the
NHIS data and NIAID (1979) data.
Activity pattern data from the PM exposure analysis of
Pedco (1981) were used to identify the share of popula
tion at a high exercise level for each hour. It is
assumed that results for the general population can be
applied to asthmatics.
Several hedpnic wage studies were used to develop an
estimate of willingness to pay for risk reduction. Use
of the studies is detailed in Manuel et _al_. (1983).
There is limited information on willingness to pay for
symptom reduction. A survey by Loehman j|t Jj_. (1979)
indicates a willingness to pay for an hour of symptom
reduction of under $1. The survey, however, has many
shortcomings for our use. In the absence of other
information, a $50 value is used as an upper bound to
reflect the possibility that symptoms are accompanied by
high discomfort, medical costs, or productivity losses.
Rossiter and Walden (1979).
Recent state-level crop prices were obtained from county
agricultural statisticians.
Crop Yield
County-level crop yields were obtained from state agrv
cultural reports. Conversations with agricultural
extension workers were used to allocate county output
within the case study areas.
-------
A-12
To perform benefit calculations for a selected crop and plant, two
piece of data are required. First, the baseline level of production must
be identified for each grid. The following procedure was employed:
i) The counties that fell within the power plant range were determined.
Since the grid boundaries are unrelated to county lines, often
counties are only partially contained within the grid area.
ii) Using state agricultural reports, county-level data on 1982 pro-
duction was obtained.
iii) County agricultural agents were consulted to find the share of
each county's crop production contained within the power plant
range.
iv) Production within the power plant range was estimated for each
county by:
where gp-j = production in county i within the power plant
range.
sh-j = share of production of county i contained within
power plant range.
cp-j = production of county i.
Table A. 6 presents the results of this estimation.
v) Within each county3 production was assumed to be uniformly .di s-
tributed across receptors.
Thus, if there are n receptors in county i, the production associated with
each one is given by gp-j/n.
Second, the price of the crop must be identified. 1982 state level
prices, shown in Table A. 7, are used.
Based on these data items, benefits can be calculated. The following
functions are applied to each grid:
-------
A-13
Table A.6
1982 PRODUCTION WITHIN RANGE OF THE POWER PLANT
(Bushels)
County
Wheat Oats Soybean
Production Production Production
Eddystone
Wansley
Potomac River
Portage des Sioux
Carroll, GA
Coweta, GA
Heard, GA
Prince Georges, MD
Charles, MD
St. Charles, MO
St. Louis, MO
Jersey, III
Madison, IL
78,800
4,350
25,517
27,360
16,800
936,150
272,272
309,985
1,051,578
528
6,992
11,400
3,660
4,699
37,600
8,540
38,570
145,730
51,040
1,837,080
314,853
631,300
1,140,300
-------
A-14
Table A.7
1982 PRICE PER BUSHEL BY STATE
(1982 price inflated by the CPI to 1984 dollars)
State
Wheat
Oats
Soybean
Maryland
Georgi a
Missouri
111 inois
3.10 (3.34)
3.05 (3.28)
3.00 (3.23)
3.10 (3.34)
1.85 (1.99)
1.55 (1.67)
2.10 (2.26)
1.85 (1.99)
5.60 (6.03)
5.45 (5.86)
5.50 (5.92)
5.75 (6.19)
-------
Wheat
Oats
Soybeans
A-15
B = [-8.002(S02 - SOI) + 16.492(S022 - SOI2)] .
. Pw
1.404 - (8.002)(SOI) + (16.492)(S01)2
B = C-0.338(log S02 - log SOI) .
[ ____ _ 1 . Po
0.038 - (0.338)(log SOI)
= [ e-0.034 - 1.89 ; S02 - e -0.034 - 1.89 ; SOI ]
ys
r
PS
e-0.034 - 1.89 ; SOI
where SOI = mean pre-control S0£ in ppm.*
S02 = mean post-control S02 in ppm.*
7W = baseline wheat production in bushels.
y~o = baseline oats production in bushels.
y~s = baseline soybean production in bushels.
Pw = price per bushel of wheat for relevant state.
PO = price per bushel of oats for relevant state.
Ps = price per bushel of soybeans for relevant state.
The crop varieties on which these concentration-response functions are
based may differ from those grown around the four power plants. For a
minimum estimate of benefits, it is assumed that the functions are not
* Benefits only are calculated for S02 ranges in the range of the original
study: 0.058-0.238 ppm, 0.056-0.178 ppm, and 0.022-0.284 ppm concentra-
tions for wheat, oats, and soybeans, respectively-
-------
A-16
applicable to the varieties in the case study areas. Benefits are zero.
For a maximum estimate, it is assumed that the functions are applicable to
all growth of the crop. The point estimate is the arithmetic mean of the
minimum and maximum.
As an example of the calculation procedure, we will estimate benefits
for a single crop, wheat, and a single grid in a hypothetical Missouri
county -
The 1982 county production within the power plant range is 100,000
bushels. There are 100 receptors in the county within plant range. Thus,
baseline production per receptor is 1,000 bushels. The price per bushel is
$3.23 per bushel (1984 dollars).
The maximum estimate of benefits of a change in S02 from 0.10 to 0.07
ppm is:
B = [-8.002(0.07 - 0.10) + 16.492(0.072 - 0.102)]
1,000
r 1 [3.23] = $656
1.404 - (8.002)(0.10) + (16.492)(0.10)2
The minimum estimate of benefits for the grid is zero. The point
estimate is $328 (0 + 656/2).
b. Reduced Mortality Risk
Benefits of reduced mortality risk are estimated at the area level.
That is, a single-valued area index is developed using the individual
receptor data. The following procedure is used to estimate annual benefits,
-------
A-17
First, the change in mortality risk is estimated for each day of the year
using a concentration-response function. Second, a willingness-to-pay
valuation derived from hedonic wage studies is attached to the daily
mortality risk reductions. Third, benefits across days are summed to get
the annual benefit.
The minimum and point estimates of the impact on mortality risk are
zero. The maximum estimate is found by applying the following concentra-
tion-response function, derived by Mathtech from the 1958-59 data of Martin
and Bradley (1960), to daily S02 levels:
AMd = [-0.568 (S01d - S02d) + 0.041 (SOI2 - SO2)] .
[pop/3,204,000]
where AMd = change in mortality risk on day d
SOl^ = mean pre-control S02 concentration on day d in pphm*
S02,-| = mean post-control S02 concentration on day d in pphm*
pop = population for the grid area around the power plant; the
factor 3,024,000 represents London population in 1958-59.
Thus, the ratio in the last term is a normalizing factor.
The S02 concentrations are based on an average across receptors. Changes
in mortality risk are valued at $7.3 million (1984 dollars) per unit.
As an example, consider an area with population of 100,000 in a given
year. On the first day of the year, the SOg measure is reduced from 10 to
7 pphm. The maximum estimate of benefits, MB1, for the day is:
ml = [-0.568 (10 - 7) + 0.041 (102- 72)] .
[100,000/3,204,000][7,300,000] = $88,174
A similar calculation must be performed for each day of the year to get the
annual estimate.
* No benefits are estimated for S02 changes below 7 pphm.
-------
A-18
c. Reduced Morbidity
Two types of reduced morbidity effects are considered: reductions
in emergency room admissions and reductions in prevalence of asthmatic
symptoms.
Reductions in Emergency Room Admissions
Effects on emergency room admissions are also estimated at the area
level. Benefit estimation involves the following steps. First, the
daily changes in admissions are calculated using a concentration-response
function. Second, each admission is valued at the average charge.
Third, benefits across days are summed to get annual benefits.
The minimum estimate of the effect on admissions is zero. The maximum
estimate is based on a concentration-function developed by Graves et al.
(1980):
AAd = [ e 4.726 + (0.18)(ln S01d) - (0.027)(ln S01d)2 -
4.726 + (0.18)(ln S02d) - (0.027)(ln S02d)2
pop
6.5 x 105
where AAd = change in number of emergency room admissions on day d
S01(j = mean pre-control concentration on day d in pphm
S02d = mean post-control concentration on day d in pphm
pop = population; the value 6.5 x 10^ represents the Chicago
population served by the hospital studied by Graves et_
al. Thus, the last term is a normalizing factor.
-------
A-19
The S02 concentrations are based on an average across receptors. Each
admission is valued at $85 (1984 dollars). The point estimate is the
arithmetic mean of the minimum and maximum.
The original Graves _et_ al_. function is dependent on data for a number
of weather and pollution variables. It would have been very time-consuming
to develop a daily series for these data for the study areas. Therefore,
some approximations had to be made. Data for the four areas and the
original study were examined to develop a range of possible values. Then,
to yield a maximum estimate, the ends of the range that yielded highest
benefits were selected for each variable. The function presented above
incorporates these values in the constants.
As an example of the estimation procedure, assume a population of
100,000 in a given year. On the first day of the year, the 8^2 level is
decreased from 0.10 to 0.07 ppm. The maximum estimate benefit of MB^ for
the day is:
MB1 = [ e 4.726 + (O.l8)(ln 10) - (0.027)(ln 10)2 -
4.726 + (O..18)(ln 7) - (0.027)(Tn 7)2 ]
_ ][ 85 ] = $45
;. . . 6,
100,000
6.5 x 105
The point estimate for the day is $22 (0 '+ 45/2). A similar calculation
must be performed for each day of the year to get the annual benefit.
Reduction of Asthmatic Symptom Prevalence
Benefits of reduced asthma symptom prevalence are estimated at the grid
level. The calculations require several steps. First, the asthmatic population
residing in each grid is identified. Second, the asthmatic population at a high
exercise level >in each grid is determined for each hour of the year. Third, a
-------
A-20
concentration-response function is used to estimate the reduction in
symptom prevalence for each hour. Fourth, the symptom reductions are
valued. Fifth, results across hours and grids are summed to get the annua'
benefit.
The minimum estimate of the impact on symptoms is zero. For the
maximum estimate, the following concentration-response function is applied
to the relevant population:
apoph
ASh = (g)(31)(S01h - S02h)
100
where ASh = change in number of asthmatics with symptoms during hour
h
3 = coefficient with a value of 7.1, 11.3, and 8.5 for
changes in S02 within the range of 0-0.23, 0.23-0.28, and
0.28-0.58 ppm, respectively. No benefits are calculated
for higher levels of S02-
SOlf, = pre-c ,ntrol S02 concentration for hour h in ppm
S02n = post-control S02 concentration for hour h in ppm
apopn = asthmatic population at high exercise level during hour h
The function is based on Kirkpatrick et_ a_l_. (1982). Each change in symptom
prevalence is valued at $50 (1984 dollars). The point estimate is derived
by replacing 3 with g/2 in the concentration-response function and using a
valuation of $25. These valuations may be adjusted in the final report
as a result of ongoing Agency work.
As an example, assume a grid.with an asthmatic population of 5 at a
high exercise level. The S02 level for the hour is reduced from 0.10 to
0.07 ppm. The maximum estimate of benefits, MBh, for the hour is:
MBh = (7.1)(31)(0.10 - 0.07)(5/100)(50) = $17
-------
A-21
A similar calculation must be performed for each grid and hour to get the
annual benefit.
d. Materials Damage
Estimates of materials damage due to SC>2 are developed from the
household sector model presented in Manuel .et. _§_]_. (1982). This model is
an economic demand model which describes the relationships between S02
and various goods and services that may be purchased by a consumer.
Thus, the concentration-response function relates S02 directly to economic
parameters; there is no intermediate calculation of physical damage.
Demand equations are estimated for seven commodity groups: food,
shelter, household operations, home furnishings, clothing, transportation,
and personal care. Two additional plausible demand -.ategories, property
expenditures and recreation are not included because of data problems. For
each of the seven commodity groups, from 2 to 5 goods are separately
identified and demand equations are also estimated for each of these goods.
For example, separate equations are estimated for household textiles,
furniture, appliances, and housewares in the home furnishings commodity
group. In all, 21 demand equations for goods are estimated.
Data for the original estimation of the household sector model were
obtained from several sources. The expenditure data for individual goods
and specific SMSA's are taken from the U.S. Bureau of Labor Statistics,
Consumer Expenditure Survey (1978). Retail price data for the same SMSA's
and goods are also from a Bureau of Labor Statistics publication (1973).
Air quality data are developed from data maintained by EPA on the SAROAD
• . *. •- -* / • _• '' - - ' ,';
data base. The data base that is constructed covers 24 SMSA's with budget
and environmental data representative of conditions in 1972 and 1973. It
-------
A-22
is assumed that the preferences revealed in the 1972-73 data can be applied
to the time period and areas considered in the case study analysis. That
is, budget allocations and relative prices remain unchanged.
The procedures used to develop estimates of benefits are quite com-
plex, involving simultaneous evaluation of systems of demand equations.
For this reason, no attempt is made here to present a sample calculation
involving all the steps. Instead, the major steps of the calculation
routine are highlighted.
The first step involves aggregation of market data to form price and
quantity indices for the 21 goods for which separate demand equations are
to be estimated.
Second, the 21 goods are grouped into seven commodity groups. The
seven commodity groups are consistent with the major divisions defined by
the Bureau of Labor Statistics for household budget allocations.
As noted earlier, each of the seven commodity groups contains from 2
to 5 goods. The third step thus involves the simultaneous estimation of
demand equations for each good in a given commodity group. It is at this
stage that the S02 variables are entered to determine whether they are
relevant explanatory variables. In addition, several socioeconomic vari-
ables are also considered to determine whether they help explain the varia-
tion in demand.
The econometric modeling of the goods demand equations indicated that
S02 is a significant explanatory variable in three of the equations: home
repair activities; textiles excluding clothing; and transportation ser-
vices. These equations are reproduced in Table A.8. The measure of S02
found to be most robust is the annual arithmetic mean.
-------
A-23
Table A.8
DEMAND EQUATIONS FOR WHICH S02 IS A SIGNIFICANT EXPLANATORY VARIABLE
1. Home Repai r
Xi = 0.23288 + 0.767 * (-33.314 + 0.03286 * S02) * Zi
- 0.23288 * (-153.27 + 0.11979 * TSP) * Z2
where X^ = the share of home repair expenditures relative to all
expenditures for shelter.
I\ = the ratio of the home repair price index to shelter
expenditures.
Z2 ~ the ratio of the utility price index to shelter
expenditures.
S02 = the 24 hour second high measure of S02 for the monitor
recording the maximum value of this variable within the
case study area.
TSP = the 24 hour second high measure of Total Suspended Par-
ticulates for the monitor recording the maximum value of
this variable within the case study area. This value is
defaulted to the current primary NAAQS.
2. Textiles excluding clothing
X2 = 0.106456 + 0.89354 * (3.23302 + 0.020286 * S02) * Z3
- 0.106456 * ((-21.5796 - 37.8934 * FAMSZ + 55.375 *
REGION) * Z4 + (-50.0434' + 10.3395 * FAMSZ + 2.4711 *
REGION) * ZP5 - 7.33839 * Z6)
where X2 = the share of textile expenditures relative to all
expenditures for home furnishings.
Z3 = the ratio of the textile price index to home furnishing
expenditures.
Z4 = the ratio of the furniture price index to home
furnishing expenditures.
Zs = the ratio of the appliance price index to home
furnishing expenditures.
-------
A-24
Table A.8 (continued)
DEMAND EQUATIONS FOR WHICH S02 IS A SIGNIFICANT EXPLANATORY VARIABLE
75 = the ratio of the housewares price index to home
furnishing expenditures.
S02 = the 24 hour second high measure of S02 for the monitor
recording the maximum value of this variable within the
case study area.
FAMSZ = family size.
REGION = dummy variable for location of case study area
(l=Northeast or North Central, 0=otherwise).
3. Transporation Services
Z8
S02
= 0.53466 + 0.46534 * (-675.609 + 37.1853 * S02) * Z7
- 0.53466 * (-377.413 + 28.3191 * TEMP) * Z8
where X3 =
the share of gasoline expenditures relative to all
expenditures for transportation.
the ratio of the gasoline price index to all
tra; sportation expenditures.
the ratio of the price index for the "other"
transportation demand category to all transportation
expenditures.
the 24 hour second high measure of S02 for the monitor
recording the maximum value of this variable within the
case study area.
TEMP = average annual temperature in degrees Celsius.
-------
A-25
It is important to note that these equations are estimated as part of
a system of demand equations, and by themselves do not lead directly to a
measure of benefits for reductions in S02. In fact, a series of steps is
required before benefits can be ascertained.
Given the demand equations for goods, the next step involves using
these equations to define price and quantity indices for the commodity
category in which they have been placed. Note, in particular, that the
price index for a commodity will be sensitive to the level of S02 that is
used to evaluate the goods' demand equations. Since demand for only three
goods is found to be associated with SC>2, only three of the commodity price
indices will be dependent on S02- However, since the seven commodity
groups also form a demand system, the S02 variable indirectly plays a role
in the overall budget allocation across commodities.
With aggregate price and quantity indices defined for the commodities
the next step involves the simultaneous estimation of the commodity demand
functions. With the parameters of the commodity demand functions identi-
fied, it is now possible to calculate benefits. The goal is to use the
commodity demand functions to define an expenditure function that is a
function of S02« This is accomplished in the following way.
First, the parameters of the system of commodity demand functions are
used to define a compensated commodity demand system. The compensated
curves result from a consumer optimization problem in which the consumer
minimizes expenditures subject to a constant level of utility. The expend-
iture function is then formed as the inner product of the commodity
compensated demand function and the associated price index for all seven
commodities. This procedure is performed with SC>2 first at baseline (pre-
-------
A-26
control) levels and then at post-control levels. The difference between
the two expenditure functions is a measure of the compensating variation, a
welfare measure that is consistent with the correct concept of a benefits
measure: willingness to pay.
5. Method of Extrapolation from Case Studies
To extrapolate the benefit estimates for the four power plants to the
31 eastern states, it is assumed that benefits per ton are invariant with
source type and location. Case study benefits are divided by the asso-
ciated reduction in tons of S02- The resulting benefit per ton estimate is
applied :o the number of tons of S02 controlled for the 31 eastern states.
For example, the case study midpoint estimate of the annualized bene-
fits of a 0.5 ppm 1-hour standard is $3.6 million. These benefits are
associated with an annual reduction of 137,330 tons of S02. Thus, benefits
per ton are $26 (3.6M/137,330).
Imposition of a 1-hour standard of 0.5 ppm in the 31 eastern states
will result in an annual reduction in tons of S02 emitted of 4,530,000.
Thus, the annualized benefits is given by $26 x 4,530,000 = $118 million.
-------
A-27
List of References
Graves, P.E., Krumm, R.J., and Violette, D.M., (1980). Estimating the Benefits
of Improved Air Quality. Report for Meeting of Benefit Methodology
Panel, National Commission on Air Quality, December.
Hausmann, J. (1981). Exact Consumer's Surplus and Deadweight Loss. Ameri-
can Economic Review, 71:662-676.
Hobart, J.M., Horst, R.L., Jr., Brennan, K.M., Manuel, E.H., Jr., et_ _a_[.,
(1984). Benefit Analysis of National Air Quality Standards for Sulfur
Dioxide. Report by Mathtech, Inc. for EPA, Office of Air Quality
Planning and Standards, Research Triangle Park, North Carolina,
December.
Kirkpatrick, M.B., Sheppard, D., Nadel, J.A., and Boushey, H.A., (1982).
Effect of Oronasal Breathing Route on Sulfur Dioxide-Induced Broncho-
constriction in Exercising Asthmatic Subjects. American Review of
Respiratory Diseases, 125:627-631.
Loehman, E.T., Berg, S.V., Arroyo, A.A., Hedinger, R.A., Schwartz, J.M.,
Shaw, M.E., Fahien, R.W., De, V.H., Fishe, R.P., Rio, D.E., Rossley, F.W.
and Green, A.E.S., (1979). Distributional Analysis of Regional Benefits and
Cost of Air Quality Control. Journal of Environmental Economics and
Management, 6:222-243.
Manuel, E.H., Jr., Horst, R.L., Jr., Brennan, K.M., et^aU, (1982). Benefits
Analysis of Alternative Secondary National Ambient Air Quality Stan
dards for Sulfur Dioxides and Total Suspended Particulates, Volume II.
Final report by Mathtech, Inc. for EPA, Office of Air Quality Planning
and Standards, Research Triangle Park, North Carolina, May.
Manuel, E.H., Jr., Horst, R.L., Jr., Brennan, K.M., Hobart, J.M., et al.
(1983). Benefit and Net Benefit Analysis of Alternative National
Ambient Air Quality Standards for Particulate Matter, Volume II.
Final report by Mathtech, Inc. for EPA, Office of Air Quality Planning
and Standards, Research Triangle Park, North Carolina, March.
Martin, A.E. and Bradley, W.H., (I960). Mortality, Fog, and Atmospheric
Pollution - An Investigation During the Winter of 1958-&9. Monthly
Bulletin of the Ministry of Health. Public Health Laboratory Service,
19:56-72.
National Health Interview Survey (1970). Household Interview of Civilian,
Non-Institutional, Population, Chart - Asthma With and Without Hay
Fever.
NIAID (1979). Asthma and Other Allergic Diseases. NIH Publication 79-387.
Washington, D.C.
-------
A-28
Pedco, Inc. (1981). The NAAQS Exposure Model (NEM) and its Application to
Participate Matter. Draft report prepared for U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina, August.
Rossiter, L.F. and Walden, D.C., (1979). Pediatric Care: Charges, Payments
and the Medical Setting. Paper presented at APHA Annual Meeting,
Health Administration Section, New York, New York, November.
U.S. Bureau of the Census (1980). Population and Households by States and
Counties: 1980 (PC80-51-2).
U.S. Bureau of Labor Statistics (1973). Average Retail Prices of Selected
Commodities and Services. U.S. Government Printing Office, Wash
ington, D.C.
U.S. Bureau of Labor Statistics (1978). Consumer Expenditure Survey:
Integrated Diary and Interview Survey Data, 1972-73. Bulletin 1992,
U.S. Government Printing Office, Washington, D.C.
U.S. Bureau of Economic Analysis (1981). Projections of the Population
1976-2000. Memorandum, March.
U.S. Department of Commerce News (1980). Projections of Personal Income to
the Year 2000. December.
-------
Appendix B
1. Introduction
The purpose of this appendix is to address the issue of the potential
size of benefits resulting from decreased mortality risk associated with
ambient 804 reductions. Section D.2.b of Chapter Vldiscusses the evidence
for a link between mortality risk and transformation products of S02,
chiefly atmospheric sulfate species. Because of the uncertainties
associated with the available studies and the lack of biological plausibility
that small changes in annual sulfates would of themselves, produce
calculable reductions in health risks, no estimates are given in Chapter
VI for benefits associated with reduced mortality risk for $04. Since
the data clearly suggest a risk at current levels, and it is probable
that reducing S02 emissions would reduce episodic peak acid aerosol
exposures and thus mortality risk, this appendix provides some illustrative
estimates of reduced mortality benefits.
The appendix is organized into the following sections: Conceptual
framework for valuing risk reduction, inferences of risk from health
studies, calculation procedures, and results.
2. Conceptual Framework for Valuing Risk Reductions
Benefit analysis focuses on willingness to pay for small reductions
in risk of death; no attempt is made to value avoiding the death of a
specific person* There are several different types of uncertainty
associated with risk or mortality reductions due to 564 reductions.
First, even if we had perfect information regarding the distribution of
mortality risk under all $04 scenarios (i .e., population risk is known),
we would not know the outcome for any specific individual. This presents
-------
B-2
no problem however because the willingness to pay estimates are derived
from studies of the relationship of wages and occupational risk. These
wage studies associate pay with known population risk (not individual risk).
A second type of uncertainty involves extrapolation of willingness
to pay for one type of mortality risk for one group of people to another
type of mortality risk to another group of people. Differences in pain
and suffering before death, medical costs, baseline expected risk, age,
size of risk reduction, etc. are all expected to influence willingness to
pay for mortality risk reduction. While this kind of uncertainty is
important to remember in interpreting results, it is present in all
benefit analyses of mortality risk reduction.
Third, there is the uncertainty surrounding the mortality risk for a
population. Essentially, the statistical relationship between $04
concentrations and mortality rates potentially includes annual 864 levels
acting as a surrogate for things that will change with S02 control and
things that will not change. As mentioned in Chapter VI, annual $04 may
be a surrogate for shorter exposures to specific $04 species. If S02
control is more efficient at controlling such exposure, then use of statis-
tically derived annual SOg coefficient results in a downward bias on benefit
estimates. If SOg control is less efficient for the things actually
causally related to mortality than for annual $04 concentrations, then an
upward bias results. The extreme case would be annual 864 concentration
acting as a surrogate for something independent of SC>2 emission control.
As long as there is some chance that the reductions in 504 due to S02
control will result in lower mortality risk then there is an ex ante willing-
ness to pay for such a potential reduction. If further research causes a
-------
B-3
change in the state of information such that the probability of such
mortality reductions being linked to 864 reduction approaches
zero, then willingness to pay for such mortality reduction should also approach
zero. Therefore, given an imperfect state of information there are ex ante
benefits for reducing a potential source of risk (even if perfect information
would indicate a zero risk for the potential source of risk). This third
source of uncertainty results in uncertainty concerning the magnitude, not the
sign of the benefits of reducing mortality risks.
3. Inference of Risk from Health Studies
Attempts to quantify and evaluate changes in mortality risk associated
with S02 control must rely on a weak data base for which the interpretation
is inherently controversial. The major evidence suggesting an association
between regional sulfur oxide transformation products (sulfates) and
mortality is derived from a series of large scale retrospective cross
sectional epidemiology studies. Substantial disagreement exists within
the scientific and analytical community regarding the proper interpretation
of the pollutant-mortality associations reported in these studies and
their use in making quantitative estimates of effects. The approach
adopted here selects alternative concentration response coefficients from
epidemiological studies and from expert judgments. Hypothetical effects
are then calculated assuming that coefficients are correct and that no effects
threshold exists. Because there is no information that can tell us what part
of this range for the mortality coefficient is most probable, no coefficient
is identified as a .best estimate,.
Seven epidemiological: studies are considered in choosing alternative
504 concentration response coefficients. Three studies (.Chappie and Lave
-------
B-4
1981; Mendelsohn and Orcutt 1979; and Lave and Seskin 1977) find statistically
significant relationships between $04 and mortality risk. But, three
other studies (Lipfert 1977, 1979, and 1980) do not find a statistically
significant relationship. In fact, although statistically insignificant,
the $64 coefficient in many of Lipfert's regressions is implausibly signed
(i.e. lower $04 concentrations increase the risk of mortality). A recent
study (Evans et.al. 1984) falls somewhere between the aforementioned
studies. The study finds that in all 21 regressions the $04 is "plausibly"
signed (i.e., higher $04 concentrations increase the risk of mortality).
However, the coefficient is statistically significantly different from
zero (at the .05 level) in only four of the regressions. The Lipfert
study s selected to represent the lower range of benefits in this category.
The Chappie and Lave study is selected as the basis for the higher benefit
estimate. This latter study is selected over the Mendelsohn and Orcutt
(1979) and Lave and Seskin (1977) studies because of fewer criticisms
regarding both the underlying air quality data and omission of important
independent variables. The Evans study (1984) is also included.
Morgan et. al. (1982) prepared a report for the National Science
Foundation involving extended interviews with leading health effects
experts. The experts were asked several questions some of which dealt
with the relationship between mortality risk and sulfate levels. This
analysis uses the information found on page 82 of the Morgan, et. al.
report which is reproduced here as Table B.I. The geographic weighted
average annual $04 concentration for baseline conditions in the 31 eastern
states used in this analysis ranges from 5.59 to 8.07 ug/m^ depending on
which model is used. That range most closely corresponds to the 7 ug/m^
504 rural scenario which appears on Table B.I. As noted in the table,
-------
B-5
Table B.I
Mean Values of Distributions in the Urban/Rural Analysis
pg/m3
20
15
13
10
7
5
TYPE
City
Urban
Urban
Urban
Rural
Rural
EXP1
8.32
8.04
7.74
7.29
6.49
5.21
SLOPE1
EXP2
.398
.167
.155
.085
.032
.021
EXP3
.883
—
.371
—
0.00
___
Percent Excess Deaths'
EXP1
13.68
9.97
8.04
5.29
2.86
1.64
EXP2
.234
.082
.047
.017
.007
.006
EXP3
1.90
.368
0.00
The information in this table was taken from page 82 of "Technological Uncertainty in
Policy Analysis," M. Granger Morgan et.al., 1982. Prepared for the Division of Policy
Research and Analysis, National Science Foundation under grant number PRA-7913070.
Deaths per 100,000 people per ug/m3.
-------
B-6
responses in terms of predicted deaths/100,000 population for a 1 pg/m3
increase in annual sulfate concentrations were elicited from experts one
two, and three.
4. Calculation Procedures
Developing a benefit estimate for reduced mortality risk involves
two sets of procedures. The first establishes the relationship between
mortality, $04 concentration changes, air quality data and selected
studies. The average of the Astrap and Monte Carlo results are used for
the air quality concentration change. As mentioned in the previous
section, an estimate of 0 deaths/100,000 population for a 1 pg/m3 increase
in annual $04 concentration is taken from the Lipfert findings. The only
equation from the Chappie and Lave study that could be used with the air
quality data available for this analysis (i.e., predicted annual average
$04 concentrations) has a coefficient of 10.547 deaths/100,000 population
for a 1 tg/m3 annual increase. The Evans coefficients ranged from 1.41
to 3.72. The coefficient of 2.77 is chosen because it comes from the only
model that included mean $04 and TSP as the pollution variables. Co-
efficients are also taken from the subjective probabilities of three experts
(Morgan et. al., 1982). Those coefficients are applied to the predicted
air quality improvements (804 concentration decreases) to yield estimates
of statistical lives saved.
The second set of procedures in developing estimates of $04 mortality
risk reduction benefits is to apply valuation coefficients to the estimates
of statistical lives saved. The valuation coefficients used in this
study are $.420 and $7.30 for decreased mortality risks of 1.0 X lO'6.
This range is based on recommendations of the EPA RIA Guidelines
-------
B-7
(1983). More recent work by the Office of Policy Planning and Evaluation
suggests that the $420,000 estimate may be too low.
5. Results
As can be seen in Tables B.I and B.2 the estimates of benefits due
to 804 mortality risk reductions are very sensitive to the choice of
coefficients. Benefit estimates, using a valuation coefficient of $420,000
range from zero to 81 billion dollars for the 0.25 ppm 1-hour standard.
Benefit estimates, using a valuation coefficient of $7,300,000 range from
zero to 1,410 billion dollars for the 0.25 ppm 1-hour standard.
For the non-zero benefit estimates, using valuation coefficients of
$420,000 or 7,300,000 the ratio of benefits for the 0.5 ppm 1-hour standard
to the current standards (strictly applied) is 1.82. The ratio of the
0.25 ppm 1-hour standard to the current standards (strictly applied) for
both valuation coefficients is 3.71.
-------
Table B.2
31 EASTERN STATE S04 MORTALITY RISK REDUCTION BENEFIT ESTIMATES - VALUATION COEFFICIENT OF $420,000
(DISCOUNTED PRESENT VALUE IN BILLIONS OF JANUARY 1984 DOLLARS)1
Alternative SO? NAAQS
Current Standards 0.5 ppm 0.25 ppm
Benefit Categories (Strict Interpretation) 1-hour standard 1-hour standard
Low Middle Hl_gh__ Low Middle High Low Middle High
Valuation Coefficient
$420,000
Chappie & Lave 22 40 81
(10.55)2
Evans 6 11 21
(2.77)2
-------
Table B.3
31 EASTERN STATE S04 MORTALITY RISK REDUCTION BENEFIT ESTIMATES - VALUATION COEFFICIENT OF $7,300,000
Current Standards
(Strict Interpretation)
Alternative SO? NAAQS
Benefit Categories
0.5 ppm
1-hour standard
0.25 ppm
1-hour standard
Low
Middle
Low
Middle
Low
Middle
Valuation Coefficient
$7,300,000
Chappie & Lave
(10.55)2
Evans
(2.77)2
Lipfert + Expert 3
(0)
Expert 1
(6.49)
Expert 2
(.032)
379
100
0
233
1
693
182
0
426
2
1410
370
0
867
4
CO
I The discounted present value of an eleven year stream of benefits occurring from January 1, 1990 to
December 31, 2000 using a real discount rate of 10 percent in 1984. To convert to an annualized stream of
benefits for 1990 to 2000, multiply by .2728.
2 Deaths per 100,000 people per ug/nr SO^ annually. For purposes of comparison deaths per 100,000 people due
to respiratory disease are 46.8 and deaths per 100,000 people due to cardiovascular disease are 417.6.
------- |