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.

-------