Air Pollution: EPA's Progress in Determining the Costs and Benefits of
Clean Air Legislation (Letter Report, 02/11/94, GAO/RCED-94-20).

In the 1990 Clean Air Act amendments, the most recent effort to bolster
the nation's air quality, Congress added more requirements for attaining
national ambient air quality standards and reducing hazardous air
pollutants and precursors to acid rain.  The Business Roundtable has
estimated that it could cost as much as $104 billion annually to
implement three of the amendments' major provisions: those on acid
deposition, air toxics, and the attainment of ozone standards.  In light
of this and other high cost estimates, the Environmental Protection
Agency (EPA) was required to conduct cost-benefit analyses of the
legislation.  This report describes EPA's (1) methodology for conducting
the cost-benefit analyses, (2) progress in completing the analyses, and
(3) costs incurred for contractors and EPA staff resources used to
complete these analyses.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  RCED-94-20
     TITLE:  Air Pollution: EPA's Progress in Determining the Costs and 
             Benefits of Clean Air Legislation
      DATE:  02/11/94
   SUBJECT:  Air pollution
             Cost effectiveness analysis
             Environmental monitoring
             Environmental legislation
             Environmental policies
             Air pollution control
             Mission budgeting
             Standards evaluation
             Regulatory agencies
             Comparative analysis
IDENTIFIER:  EPA Regional Acid Deposition Model
             
**************************************************************************
* This file contains an ASCII representation of the text of a GAO        *
* report.  Delineations within the text indicating chapter titles,       *
* headings, and bullets are preserved.  Major divisions and subdivisions *
* of the text, such as Chapters, Sections, and Appendixes, are           *
* identified by double and single lines.  The numbers on the right end   *
* of these lines indicate the position of each of the subsections in the *
* document outline.  These numbers do NOT correspond with the page       *
* numbers of the printed product.                                        *
*                                                                        *
* No attempt has been made to display graphic images, although figure    *
* captions are reproduced. Tables are included, but may not resemble     *
* those in the printed version.                                          *
*                                                                        *
* A printed copy of this report may be obtained from the GAO Document    *
* Distribution Facility by calling (202) 512-6000, by faxing your        *
* request to (301) 258-4066, or by writing to P.O. Box 6015,             *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time.                                    *
**************************************************************************


Cover
================================================================ COVER


Report to Congressional Committees

February 1994

AIR POLLUTION - EPA'S PROGRESS IN
DETERMINING THE COSTS AND BENEFITS
OF CLEAN AIR LEGISLATION

GAO/RCED-94-20

Costs and Benefits of Clean Air Legislation


Abbreviations
=============================================================== ABBREV

  ACCACA - Advisory Council on Clean Air Compliance Analysis
  ARGUS - Argonne Utility Simulation Model
  CEUM - Coal and Electric Utilities Model
  CO - carbon monoxide
  CRESS - Commercial/Residential Energy and Emissions Simulation
     Systems
  EPA - Environmental Protection Agency
  GAO - General Accounting Office
  GDP - gross domestic product
  HC - hydrocarbons
  HPA - Hierarchically Partitioned Assessment model
  ICE - Industrial Combustion Emissions model
  IEc - Industrial Economics, Inc. 
  IMS - Integrated Model Set
  ISTUM - Industrial Sector Technology Use Model
  NAPAP - National Acid Precipitation Assessment Program
  NOx - nitrogen oxide
  OAQPS - Office of Air Quality Planning and Standards
  OAR - Office of Air and Radiation
  OPPE - Office of Policy, Planning, and Evaluation
  ORD - Office of Research and Development
  OZIP - Ozone Isopleth Model
  PROMPT - Process Model Projection Technique
  R&D - research and development
  RADM - Regional Acid Deposition Model
  SO2 - sulfur dioxide
  SOx - sulfur dioxide
  TAMM - Timber Assessment Market Model
  TEEMS - Transportation, Energy, and Emissions Modeling Systems
  TSP - total suspended particles
  VOC - volatile organic compound
  VOCM - Volatile Organic Compounds Model

Letter
=============================================================== LETTER


B-253479

February 11, 1994

The Honorable Max Baucus
Chairman
The Honorable John H.  Chafee
Ranking Minority Member
Committee on Environment and
 Public Works
United States Senate

The Honorable John D.  Dingell
Chairman
The Honorable Carlos J.  Moorhead
Ranking Minority Member
Committee on Energy and Commerce
House of Representatives

Each year, millions of Americans face adverse health risks as a
result of exposure to airborne pollutants.  The Congress has
attempted to improve the quality of the nation's air by enacting and
updating clean air legislation, but air quality problems have proved
difficult to alleviate.  In the 1990 Clean Air Act amendments--the
most recent attempt to improve air quality--the Congress enacted
additional requirements for, among other things, attaining the
national ambient air quality standards and reducing hazardous air
pollutants and precursors to acid rain. 

During the debate over the 1990 amendments, the Business Roundtable
estimated that it could cost as much as $104 billion annually to
implement three of the amendments' major provisions:  those on acid
deposition, air toxics, and the attainment of ozone standards.  In
light of this and other high cost estimates, the 1990 amendments
required that the Environmental Protection Agency (EPA) conduct
cost-benefit analyses of the 1990 amendments and any previous
amendments to the Clean Air Act.  By November 15, 1991, EPA was to
begin reporting its findings of the costs and benefits of the Clean
Air Act prior to the 1990 amendments.  By November 15, 1992, EPA was
to begin biennially updating its initial report and estimate future
costs and benefits of the Clean Air Act including the 1990
amendments. 

The Congress also mandated that, beginning in 1992, GAO annually
examine the costs and benefits of the 1990 amendments.  During
discussions with your offices, it became apparent that this
requirement was likely to duplicate EPA's analyses.  Consequently, it
was agreed that we would provide the Congress with a status report on
EPA's efforts by describing EPA's (1) methodology for conducting the
cost-benefit analyses, (2) progress in completing the analyses, and
(3) costs incurred for contractors and EPA staff resources used to
complete these analyses. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

EPA plans to conduct two major studies, with biennial updates of the
second, to fulfill its mandate.  To determine the costs and benefits
of legislation enacted before the 1990 amendments, EPA is undertaking
what it has termed a retrospective study, using a sequential,
six-step methodology.  In this study, EPA is comparing the economic,
health, and environmental conditions that have resulted from the
amendments enacted before 1990 with projections of what these
conditions would have been without the amendments.  In step one, EPA
is estimating the direct costs of compliance with clean air
regulations.  EPA will maintain a data base of these costs.  In steps
two, three, and four, EPA is using computer models to estimate
changes in economic activity, emissions levels, and ambient air
quality resulting from actions taken in response to the amendments. 
In the last two steps, EPA is estimating the adverse effects of air
pollution on human health and the environment and assessing the
economic value of these effects.  For the second study, known as the
prospective study, EPA will contrast the effects of air quality
standards imposed by legislation enacted before 1990 with the effects
of all potential post-1990 standards.  EPA also plans to update the
prospective study biennially.  As of December 1993, EPA had begun to
develop a methodology for the prospective study but had not yet begun
any analysis. 

Although the 1990 amendments specified that EPA complete the
retrospective study by November 15, 1991, the agency has estimated
that it will not complete this study until 1994.  EPA has spent 2
years developing and carrying out the methodology for the
retrospective study.  The agency's efforts have been delayed
primarily because of the size and complexity of the study.  Precisely
isolating the effects of federal clean air legislation--some of which
the Congress enacted over 20 years ago--is a difficult analytical
task.  According to agency officials, EPA will not concentrate its
efforts on the prospective study until the retrospective study is
finished.  These officials could not estimate when the prospective
study will be completed. 

As of December 1993, EPA had spent approximately $1.3 million for
contract work on the retrospective study, and an agency official
estimated that the total cost of contract work could reach $1.6
million.  In addition, EPA had used about 12 staff-years to oversee
the study.  This estimate of staff-years includes the use of staff in
the three EPA offices involved in the project--the Office of Air and
Radiation (OAR); the Office of Policy, Planning, and Evaluation
(OPPE); and the Office of Research and Development (ORD).  EPA
officials could not estimate how much staff time would be spent
during fiscal year 1994.  Agency officials believe that the cost of
the prospective study will be comparable to the cost of the
retrospective study.  Thus, the total cost for both studies could
reach $3.2 million for contract work and 24 staff-years. 


   BACKGROUND
------------------------------------------------------------ Letter :2

The Congress enacted the Clean Air Act in 1963 to protect the quality
of the nation's air and promote public health and welfare.  In the
1970 Clean Air Act amendments, the Congress created a comprehensive
program under which (1) EPA established national ambient air quality
standards and (2) the states developed plans describing how they
would control emissions from vehicles and stationary sources of
pollution to meet and maintain the national standards.  The 1977
Clean Air Act amendments added new requirements, including provisions
to help areas that failed to comply with deadlines for the national
ambient air quality standards (known as nonattainment areas) achieve
attainment.  For example, permits were required for the construction
of new or modified major stationary sources of pollution. 

The 1990 Clean Air Act amendments added stricter provisions
addressing, among other things, the attainment and maintenance of the
national ambient air quality standards, motor vehicle emissions,
hazardous air pollution, acid rain, chlorofluorocarbons, permits, and
enforcement.  As noted above, the 1990 amendments also required that
EPA assess the costs and benefits of these and previous
amendments--an expansion of a requirement in the 1970 Clean Air Act
amendments that EPA submit annual cost estimates to the Congress. 
Specifically, under the 1970 amendments, EPA was to estimate (1) the
cost of carrying out the act's requirements; (2) the cost to federal,
state, and local governments of implementing the programs; and (3)
the economic impact of air quality standards on industries and
communities, including the costs of controlling emissions.  EPA was
also required to reevaluate these estimates annually. 

Cost-benefit analysis has been an integral part of EPA's regulatory
process since Executive Order 12291 was issued in 1981.  This order
requires federal agencies to prepare a cost-benefit analysis for all
proposed major rules--rules requiring expenditures that could exceed
$100 million a year.  However, the requirement for the cost-benefit
analyses in the 1990 amendments is much more extensive than the
requirements in either the 1970 Clean Air Act amendments or Executive
Order 12291.  Under the 1990 amendments, EPA is required to conduct
comprehensive analyses of the impact of clean air regulations and
programs on the public health, economy, and environment of the United
States.  In doing so, EPA is to take into account the costs,
benefits, and other effects of complying with each standard issued
under the authority of the 1990 amendments and previous Clean Air Act
amendments. 

In early 1991, EPA decided to establish a joint management structure
involving OPPE, ORD, and OAR to perform the required analyses.  OPPE
was assigned lead responsibility.  As it became clear that no one
office had the skills and resources needed to complete the analyses,
OAR's role increased. 

In conducting the cost-benefit studies, EPA is to consult with the
Departments of Labor and Commerce and with an EPA-appointed advisory
council, specified in the 1990 amendments as the Advisory Council on
Clean Air Compliance Analysis (ACCACA).  The ACCACA consists of
recognized experts in the fields of health and the environmental
effects of air pollution, economic analysis, environmental sciences,
and other disciplines that EPA believes are appropriate.  The role of
the ACCACA is to review (1) EPA's methodology for conducting the
cost-benefit analyses, (2) the data used in the analyses, and (3) the
findings of the cost-benefit reports that result.  The ACCACA can
also recommend appropriate changes. 


   METHODOLOGY FOR THE
   COST-BENEFIT ANALYSES
------------------------------------------------------------ Letter :3

For the retrospective study--which will assess the costs and benefits
to public health, the economy, and the environment of legislation
enacted before the 1990 amendments--EPA is using a six-step
methodology.  As shown in figure 1, the methodology is generally
sequential because each step depends on the previous step for input. 
EPA is also assessing the uncertainty associated with each of the six
steps.  (App.  I contains a detailed description of the methodology
used for the retrospective study.) As of December 1993, EPA had begun
to develop a methodology for conducting the prospective study--which
will contrast the effects of the existing pre-1990 standards with the
effects of all potential post-1990 standards--but no analysis was
under way. 

   Figure 1:  EPA's Six-Step
   Methodology for Conducting the
   Retrospective Study

   (See figure in printed
   edition.)

Note:  In each step, EPA estimates the changes that have occurred as
a result of the Clean Air Act amendments. 

In step one of the retrospective study, EPA estimated the direct
costs of compliance with clean air regulations.  These data then
became input for the second step of the study.  For this second step,
EPA opted for a general equilibrium macroeconomic modeling approach
rather than the partial equilibrium approach the agency had used for
previous cost-benefit assessments.\1 EPA believes that the general
equilibrium approach allows it to assess more accurately the likely
effects of compliance with clean air regulations.  The partial
equilibrium approach does not address a number of these effects.  For
example, air pollution regulations that apply to one industry may
affect other industries that are linked to the regulated industry
through market transactions.  In addition, clean air regulations may
affect the distribution of labor, capital, and other factors of
production within the economy as well as the distribution of goods
and services.  Furthermore, the general equilibrium approach accounts
for possible actions that both businesses and households take to
avoid unnecessary costs.  The partial equilibrium approach does not
account for these effects. 

EPA's contractors used the Jorgenson-Wilcoxen general equilibrium
model of the U.S.  economy.\2 In the Jorgenson-Wilcoxen model,
expenditures that industries have made to comply with pre-1990
regulations are used to estimate the effects of those regulations. 
Using this model, EPA generated estimates of U.S.  economic activity,
including the costs of compliance, for two policy scenarios:  (1)
conditions with the pre-1990 Clean Air Act amendments in place, or
the base-case scenario, and (2) conditions without the pre-1990 Clean
Air Act amendments, or the counterfactual scenario.  EPA then
calculated the net changes in economic activity between the two
scenarios.  Although EPA has evaluated and will continue to evaluate
the uncertainties associated with using the Jorgenson-Wilcoxen model
(uncertainties are inherent in all models), the agency believes that
the current modeling results represent its best estimates and are
therefore valid input for the third step of the retrospective study. 

In step three, using the base-case and counterfactual estimates, EPA
estimated emissions and calculated the net changes in emissions
resulting from the legislation.  To estimate emissions, EPA used the
Argonne National Laboratory's Integrated Model Set; the Trends
methodology of EPA's Office of Air Quality, Planning, and Standards
(OAQPS); and ICF Resources' model of the electric utility industry,
the Coal and Electric Utilities Model (CEUM).  These approaches
assume that economic activity requires energy consumption and that
consuming energy generates emissions.  In general, higher levels of
economic activity result in higher levels of emissions.  In these
approaches, emissions estimates are also generated for noncombustion
sources, such as industrial processes, consumer solvents, and others. 

In step four, EPA is estimating changes in air quality on the basis
of the emissions estimates from step three.  According to the Chief
of the Air Quality Modeling Section, EPA is estimating the
differences in air quality between the base-case and the
counterfactual scenarios.  For the base-case scenario, EPA is using
historic concentrations of air pollutants to develop profiles for air
quality in each county or urban area.  For the counterfactual
scenario, EPA is using several air quality models and statistical
estimating techniques to assess the effects on air quality of the
absence of air pollution controls.\3

In step five, EPA is estimating the effects of projected air quality
on human health and welfare and on the environment.  These effects
are being estimated using concentration-response functions.\4
Starting with the air quality estimates from step four, EPA will
estimate concentration-response functions for each of the criteria
air pollutants using available scientific literature.\5 EPA will then
use these functions to estimate how air quality affects human
health--as measured by morbidity and mortality (sickness and
death)--and how it affects the environment--as measured by
visibility, surface water quality, agricultural productivity, forest
quality, and damage to materials.\6

In step six, EPA is estimating the economic value of these estimated
effects on health and welfare and on the environment.  For example,
EPA's contractors have recommended estimating people's willingness to
pay to avoid the symptoms commonly associated with poor air quality,
including eye irritation and throat congestion.  To estimate the
values of these effects, EPA is relying on evaluations published in
the economic literature. 

The estimates in steps one through six include considerable
uncertainty.  According to EPA, uncertainties in the retrospective
study are due to, among other things, poor data quality, modeling or
sampling error, incomplete knowledge of key processes and
relationships, and major uncertainties about the details of the
counterfactual scenario.  In addition, according to the ACCACA,
incomplete knowledge about many of the underlying relationships in
each of the steps in the analysis leads to uncertainties.  Moreover,
the uncertainty associated with each step's estimates carries through
to the next step's estimates.  As part of its analysis, EPA plans to
evaluate the uncertainty in the estimates from each of the six steps
and to identify key uncertainties and their implications for policy
decisions. 

EPA has consulted with the ACCACA and other experts in the fields of
public health, economics, and environmental science in deciding on
and implementing the methodology for the retrospective study.  In
April 1992, the ACCACA reviewed EPA's work plan for the study and
determined that it was "basically sound." In addition,
representatives from the Departments of Labor, Commerce, and Energy
as well as representatives from the Council of Economic Advisers, the
Council on Environmental Quality, and the Office of Management and
Budget reviewed EPA's work plan and agreed with the ACCACA that it
was sound. 


--------------------
\1 In this context, the term "general equilibrium" refers to an
economywide analysis, while the term "partial equilibrium" refers to
an industry-specific analysis. 

\2 Professor Dale W.  Jorgenson of Harvard University and Professor
Peter J.  Wilcoxen of the University of Texas at Austin developed the
Jorgenson-Wilcoxen model. 

\3 These models include a set of interrelated equations that simulate
and analyze the effects of wind speed, wind direction, and other
atmospheric conditions on the movement of airborne pollutants. 

\4 Concentration-response functions are mathematical functions used
to predict changes in health effects on the basis of specified air
pollution concentrations. 

\5 Criteria air pollutants are air pollutants for which EPA has
promulgated ambient air quality standards to protect public health. 
There are six criteria air pollutants:  ozone, carbon monoxide,
particulate matter, sulfur dioxide, nitrogen dioxide, and lead. 

\6 EPA can derive concentration-response functions (which are based
on scientific literature) at the same time air quality estimates are
made.  But the estimates of air quality must be complete before EPA
can generate estimates of the effects because concentration-response
functions and air quality estimates are used together to estimate the
effects of degradation in air quality on human health and the
environment. 


   PROGRESS IN COMPLETING THE
   STUDIES
------------------------------------------------------------ Letter :4

As of December 1993, EPA had completed steps one, two, and three of
its six-step retrospective study; preliminary work for steps four
through six was well under way.  Although the 1990 amendments
required EPA to complete the study by November 15, 1991, the agency
now estimates that it will complete the study in 1994.  The study's
size and complexity are the primary reasons for the delay.  EPA's
choice of a sequential, general equilibrium approach does not seem to
be a major contributing factor in the delay.  According to the
Chairman of the ACCACA, a different methodology could have been as
costly and time-consuming as the approach adopted by the agency. 

According to agency officials, EPA has initiated planning for the
prospective study and expects to begin work on it before the
retrospective study is completed.  These officials could not estimate
when the prospective study will be finished. 

For step one of the retrospective study, EPA completed development of
a data base on the costs of compliance in early 1992.  These data
were used as input for step two; that is, modeling of U.S.  economic
activity using the Jorgenson-Wilcoxen model. 

For step two, although additional analysis of the macroeconomic model
is being conducted to further assess its uncertainties, EPA and the
ACCACA were confident that modeling results presented in March 1993
were adequate to serve as input for step three.  Initial modeling was
completed in April 1992.  EPA reviewed the contractors'
results--which included estimates of economic activity for both the
base-case and the counterfactual scenarios--and the internal
consistency of the model before consulting with the ACCACA in
December 1992.  The ACCACA raised several concerns about the modeling
and requested additional analysis using the model.  This analysis was
intended to help EPA and the ACCACA assess the uncertainties inherent
in the model and increase the ACCACA's confidence in the model's
results.  Specifically, the ACCACA was concerned about the (1)
assessment of direct and indirect costs, (2) treatment of
productivity growth, (3) accuracy of cost estimates for stationary
and mobile sources of pollution, and (4) treatment of foreign savings
and investment behavior.  EPA's contractors presented additional
results from the model to the ACCACA in March 1993.\7 The ACCACA was
impressed with the progress on its concerns and endorsed the general
approach but still had concerns about certain assumptions inherent in
the model.  EPA said that it is confident that these concerns can be
addressed in its final report on the retrospective study. 

For step three, OPPE retained the Argonne National Laboratory, using
its Integrated Model Set, as the primary contractor for the emissions
estimates.  OPPE also employed OAQPS' Trends methodology to help with
emissions modeling.  Argonne presented EPA with preliminary emissions
estimates in June 1992 and revised estimates in September 1992. 

OAR disagreed with Argonne's original estimates of emissions from
electric utilities.  Because OAR considered these emissions
particularly important, it contracted with ICF Resources to provide
additional estimates of utility emissions to serve as a check on and
to help resolve any problems that might emerge with Argonne's
estimates.  The results of the Argonne and ICF models were similar
for particulates and nitrous oxides but very different for sulfur
oxides.  For example, the difference between the counterfactual and
base-case scenarios that ICF estimated for sulfur oxides for 1990 was
over twice the difference that Argonne estimated, and the difference
that ICF estimated for 1980 was almost 10 times the difference that
Argonne estimated for that year.  According to EPA, this
inconsistency arose because Argonne's estimates did not reflect fuel
switching--that is, utilities changing from one fuel to
another--among other factors.  In December 1992, the ACCACA expressed
its preference for ICF's approach and estimates.  Once ACCACA made
its preference known, OAR expanded ICF's work, requesting estimates
for more pollutants over more years to substitute for Argonne's
utility emissions estimates.  These emissions estimates were
completed in June 1993. 

For step four, in 1992 EPA's air quality modeling contractor began
developing a software program to make the emissions data compatible
with EPA's main air quality model, the Regional Acid Deposition Model
(RADM).  At the December 1992 ACCACA meeting, EPA presented its
initial ideas for using RADM.  However, EPA's assessment of the
effects on air quality was delayed because of the delay in completing
the utility emissions estimates noted above.  Thus air quality
modeling did not begin until July 1993.  According to the Chief of
the Air Quality Modeling Section, because of the delay in completing
the utility emission estimates and because ICF's estimates included
additional details, EPA's contractor was required to rewrite the
software program to make the final data on emissions compatible with
RADM.  As of December 1993, modeling using RADM was ongoing.  EPA
expects to employ other models and methodologies to assess air
quality in early 1994, and all air quality assessment activities are
to be completed by May 1994. 

For step five, agency officials anticipate having preliminary
concentration-response functions available for review in February
1994.  EPA has used "criteria documents" as the basis of its
assessment of the available scientific literature to establish links
between pollution and its adverse effects.  These documents are used
primarily to support the national ambient air quality standards.  EPA
has also begun to review and summarize peer-reviewed scientific
literature and staff papers that became available after the criteria
documents were published.  In addition, EPA has reviewed the
scientific literature that assesses health risks resulting from
exposure to toxic air emissions.  According to EPA, it will quantify
some health benefits that result from reductions in air toxics and
attempt to place a dollar value on such benefits. 

Finally, so that EPA can place a value on the benefits in step six, a
contractor has recommended benefits categories and approaches to
setting dollar values on them.  These recommendations were presented
to the ACCACA in December 1992.  In a March 1993 letter, the ACCACA
suggested changes to the approaches developed by the contractor.  As
of December 1993, the contractor had incorporated these suggestions
and was continuing to refine the valuation methodologies. 


--------------------
\7 Professor Jorgenson presented the additional model results to the
ACCACA. 


   CONTRACT COSTS INCURRED AND EPA
   STAFF USED FOR THE
   RETROSPECTIVE STUDY
------------------------------------------------------------ Letter :5

EPA had spent approximately $1.3 million for contract work on the
retrospective study as of December 1993, and an agency official
estimated that the final cost for contract work would be about $1.6
million.  In addition, as of December 1993, EPA had used about 12
staff-years for EPA staff dedicated to the study.  EPA could not
estimate the staff-years needed for fiscal year 1994.  Although EPA
will not concentrate its efforts on the prospective study until the
retrospective study is finished, OPPE and OAR officials expect that
the costs of that study will be comparable.  Thus, expenditures for
the two studies could total $3.2 million for contract work and 24
staff-years.  Table 1 shows how much the three EPA offices had spent
on contract work for the retrospective study as of December 1993. 



                                     Table 1
                     
                     EPA's Estimates of the Cost of Contract
                     Work for the Retrospective Study, as of
                                  December 1993


Step in the retrospective study         OPPE         OAR         ORD       Total
--------------------------------  ----------  ----------  ----------  ==========
Direct compliance cost               $81,000     $20,000          \a    $101,000
 estimation (step one)
Macroeconomic modeling (step        $105,000          \a          \a    $105,000
 two)
Emissions modeling (step three)     $380,900     $67,000          \a    $447,900
Air quality modeling (step four)          \a  $191,000\b     $70,000    $261,000
Concentration-response functions     $70,000     $13,000          \a     $83,000
 (step five)
Benefits valuation (step six)        $75,000    $145,000          \a    $220,000
Uncertainty valuation\c                   \a     $64,000          \a     $64,000
================================================================================
Total by office                     $711,900    $500,000     $70,000  $1,281,900
--------------------------------------------------------------------------------
\a Not applicable. 

\b This figure includes $80,000 in contract money that OAR
transferred to ORD. 

\c Each step in the retrospective study includes some uncertainty. 
As reflected in this cost figure, EPA has already started to evaluate
these uncertainties, and it will continue to do so. 

As the table shows, EPA has spent $101,000 on step one of the study
(estimating direct compliance costs).  This includes the cost of
maintaining a data base of compliance costs for both mobile and
stationary sources of pollution.  For step two (macroeconomic
modeling), EPA has spent $105,000, including the cost of initial
modeling and additional analysis.  For step three (emissions
modeling), EPA has spent $447,900, including the costs of analysis
using Argonne's Integrated Model Set, ICF's CEUM, and OAQPS' Trends
methodology.  For step four (air quality assessment), EPA has spent
$261,000, including the costs of developing air quality profiles and
of initial air toxics modeling.  For steps five and six, EPA has
spent $83,000 on developing concentration-response functions and
$220,000 on establishing the value of benefits, including reviews of
the scientific and economic literature. 

The three EPA offices involved in the retrospective study funded
specific parts of the study.  OPPE has spent $711,900 on work
completed by contractors, including (1) researching the cost data on
mobile sources used in the macroeconomic model, (2) modeling the
base-case and counterfactual scenarios, (3) performing additional
analyses using the macroeconomic model, (4) estimating emissions, and
(5) developing concentration-response functions.  As of December
1993, OPPE had used 4.5 staff-years for the study.  OPPE officials
expect the prospective study to require a level of cost and effort
similar to that of the retrospective study. 

OAR has spent $500,000 for contract work on the study, including (1)
reviewing data on compliance costs; (2) estimating emissions from
electric utilities; (3) assessing air quality; (4) estimating
concentration-response functions for the effects of pollutants on
mortality, morbidity, and welfare; and (5) configuring a model to
analyze the uncertainties in the study's results.  OAR has used
approximately 6.5 staff-years for the study, according to an estimate
made by OAR in December 1993.  OAR officials also expect to use
comparable resources for the prospective study. 

To date, ORD's expenditures on the retrospective study have been
minimal.  As of December 1993, ORD had spent $70,000 and used
approximately 1 staff-year on the retrospective study. 


   CONCLUSIONS
------------------------------------------------------------ Letter :6

The 1990 Clean Air Act amendments require EPA to assess the costs and
benefits of federal clean air regulations, including compliance with
each standard associated with earlier Clean Air Act amendments as
well as with the 1990 amendments.  This task is difficult because of
the size and complexity of the analyses required.  As a result, EPA
is taking longer to complete the analyses than the Congress mandated. 
Isolating the effects of clean air legislation on human health and
the environment is an analytical task that requires a complex and
sophisticated methodology, and EPA has expended substantial effort in
developing and implementing an appropriate methodology. 

EPA consulted experts in the fields of economics, public health, and
environmental science before developing and implementing the
methodology it is using for the retrospective study--the first part
of its cost-benefit analyses.  Although EPA's choice of methodology
may be costly, time-consuming, and inherently uncertain, other
methodologies could have been as costly and time-consuming, according
to the Chairman of the ACCACA. 

Given the size, scope, and complexity of the required cost-benefit
analyses, it is not surprising that EPA has been unable to meet the
specified deadlines.  According to EPA, the retrospective study will
not be completed until sometime in 1994.  EPA will not concentrate
its efforts on the prospective study until the retrospective study is
completed; thus, the agency has not yet estimated when the
prospective study will be completed. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :7

In commenting on a draft of this report, EPA said the report
accurately describes several of the difficulties that the agency has
had in implementing the program, including the size and complexity of
the cost-benefit studies.  In addition, EPA offered a number of
technical corrections and clarifications, which have been
incorporated into the report where appropriate.  EPA's comments are
reproduced in appendix II. 


   SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :8

We conducted our review between December 1992 and December 1993 in
accordance with generally accepted government auditing standards.  To
describe EPA's methodology for the cost-benefit analyses mandated in
the 1990 amendments, we reviewed the agency's work plan and assessed
the extent to which this plan conformed with requirements established
in the 1990 amendments.  We also reviewed work completed by EPA and
its contractors as of December 1993.  We discussed EPA's methods with
agency officials, contractors, and experts in the fields of
economics, modeling, and environmental science.  To determine EPA's
progress toward fulfilling the mandate in the 1990 amendments, we
reviewed preliminary reports prepared by EPA's contractors and
internal agency documents describing the progress of the analyses. 
In addition, we discussed EPA's progress with agency officials,
contractors, and the Chairman of the ACCACA.  To describe the costs
of the cost-benefit analyses, we obtained documentation from EPA on
the costs of work by contractors and the expenditure of agency
staff-years and funds.  We obtained estimates of contract costs and
staff resources from EPA offices conducting the retrospective study,
including OPPE, OAR, and ORD.  We also discussed the costs of the
analyses with agency officials. 


---------------------------------------------------------- Letter :8.1

We are sending copies of this report to appropriate congressional
committees; the Administrator, EPA; the Director, Office of
Management and Budget; and other interested parties.  We will also
make copies available to others on request. 

This report was prepared under the direction of Peter F.  Guerrero,
Director, Environmental Protection Issues, who may be contacted at
(202) 512-6111 if you or your staffs have any questions.  Major
contributors to this report are listed in appendix III. 

Sincerely yours,

Keith O.  Fultz
Assistant Comptroller General


EPA'S METHODOLOGY FOR THE
RETROSPECTIVE STUDY
=========================================================== Appendix I

This appendix provides detailed information on the Environmental
Protection Agency's (EPA) methodology for the retrospective study. 
Information is presented on each step in the study's methodology. 

EPA has stated that the quantitative estimates resulting from each
step will include considerable uncertainty and that the agency is
committed to including quantitative measures of uncertainty
associated with the major parts of an analysis, identifying those
uncertainties that are important for policy decisions and assessing
the impact of uncertainty on final estimates of net benefits.  For
the retrospective study, EPA plans to use a variety of methods to
characterize uncertainty, including measures of central tendency,
standard deviations, probability distributions, and the judgment of
experts.  Furthermore, the Advisory Council on Clean Air Compliance
Analysis (ACCACA) recommended in March 1993 that EPA pay attention to
uncertainty associated with (1) model specifications and (2) input
data.  The ACCACA recommended that EPA not rely totally on the
Hierarchically Partitioned Assessment model (HPA).\1 The HPA cannot
substitute for another method of analyzing the Jorgenson-Wilcoxen
model--the principal model used in step two-- and other complex
models.  Nor can it substitute for the use of judgment in the
analysis of uncertainty.  The ACCACA also recommended that EPA use
probability distributions and expert judgment in the uncertainty
assessment. 


--------------------
\1 The Hierarchically Partitioned Assessment (HPA) model is a
spreadsheet accounting system that handles the straightforward,
mechanical propagation of confidence intervals. 


   STEP ONE:  COSTS OF COMPLIANCE
--------------------------------------------------------- Appendix I:1

For the first step in the retrospective study, EPA is estimating the
direct costs of complying with clean air legislation.  The data on
compliance costs will be used as a stand-alone analysis as well as
for input for step two.  Compliance costs average $18.1 billion (in
constant 1982 dollars) over the period 1973 to 1990.  Of this total,
capital and net operating expenditures for stationary sources of
pollution average $5.4 billion and $5.2 billion (in 1982 dollars),
respectively.  Government outlays average just over $0.5 billion, and
costs for mobile sources average about $7.0 billion (in 1982
dollars).  Private research and development (R&D) expenditures, which
were omitted from consideration in this analysis, average $1.2
billion (in 1982 dollars). 

These costs, while significant in absolute terms, are small in
comparison with the overall economy.  On average, they represent only
one-third of one percent of domestic output from 1973 to 1990. 
However, the cost burden falls somewhat more heavily early in the
period, as these costs account for almost one-half of one percent of
total output in 1973, but only one-quarter of one percent in 1990. 
In terms of real household income, the costs are only slightly more
significant, averaging just over two-thirds of one percent from 1973
to 1990. 

EPA's costs of compliance include direct and indirect costs.  The
agency's approach entails determining (1) direct costs through
estimates of expenditures for capital and operation and maintenance
of pollution control programs and (2) indirect costs from the
Jorgenson-Wilcoxen model.  The main source of data on direct
pollution abatement expenditures is EPA's Environmental Investments: 
The Cost of a Clean Environment (Nov.  1990).  Changes in direct
expenditures for compliance were processed through the
Jorgenson-Wilcoxen model to estimate indirect costs, such as changes
in employment, productivity, cost of living, and economic growth. 


   STEP TWO:  MACROECONOMIC
   MODELING
--------------------------------------------------------- Appendix I:2

The Jorgenson-Wilcoxen model, which EPA chose for step two of the
retrospective study,\2 divides the economy into the following
sectors:  business, household, government, and foreign.  The key
features of the Jorgenson-Wilcoxen model are summarized below. 

The business sector is divided into 35 industries.  These industries
produce commodities to fill the orders of the household, government,
and foreign sectors.  The industries also provide each other with
commodities or materials needed in their production processes.  Each
of the 35 industries is represented in the model by a mathematical
equation that explains how the costs of production are determined. 
Each of these cost equations is estimated from data spanning the
period 1947 through 1985.  These equations represent costs as
dependent on capital, labor, energy, materials, and technological
change or productivity growth.  In the model, technological change is
influenced by the prices of capital, labor, energy, and materials. 
The model assumes that producers in these industries use capital,
labor, and other inputs to produce 35 separate commodities at minimum
costs.\3

The household sector is divided into demographic groups that differ
by family size, the age of the head of the household, race, the
region in which the household resides, and urban versus rural
location.  The model assumes that each household decides how much of
its available time to pursue earning income and enjoying leisure and,
in turn, how much of its income to consume and to save so as to
maximize its satisfaction.  Mathematical equations describing this
behavior are also estimated from data spanning the period 1947
through 1985.\4 In the model, households buy energy, food, nondurable
consumer goods like clothing, and services.  To determine total
expenditures on all 35 commodities represented in the model, each
household goes through a two-step process.  First, it decides how
much of its total wealth\5 to use each year.\6 Then, for each year,
it decides how much of its wealth to spend on commodities and
services and in the form of leisure time.\7 The difference between
the household's total time available and leisure time determines how
much time it spends earning income in that year.  Finally, household
saving is equal to the difference between current income from the
supply of capital and labor services provided by the household and
personal consumption expenditures. 

In the Jorgenson-Wilcoxen model, investment is derived from the
behavior of households and businesses described above.  The model
includes a system of demand equations for investment goods by these
two sectors.  For example, the business sector purchases goods for
investments in producer durables, residential and nonresidential
structures, and inventories.  The household sector invests in
consumer durables like personal computers and housing.  In the model,
each new unit of capital is an aggregate of these commodities
purchased for investment.  Thus, the price of new capital depends on
commodity prices.\8 As in the case of households, intertemporal
behavior is also assumed on the part of investors.\9 For each year,
the supply of capital depends on past investment.  Investment during
the year is determined by household savings.  Would-be investors also
compare the price of these new investment goods with the present
value of future capital services.  In the model, commodity prices and
interest rates adjust to bring about an equilibrium in which the
returns on additional investment equal the cost of new capital goods,
and savings equal investment. 

The behavior modeled for the two remaining sectors--the government
and foreign sectors--is less detailed.  The key assumptions about
these two sectors are that the government deficit and current account
surplus are exogenous; that is, predetermined outside the model.  In
the government sector, tax revenues are determined given a set of tax
rates.  These revenues, together with the given deficit, determine
government spending.  Government purchases of the various commodities
are based on historical spending patterns.  For the foreign trade
sector, imports are assumed to be imperfect substitutes for similar
domestic commodities.  The responsiveness of imports to prices is
estimated from historical data.  The prices of imports are determined
outside the model.  Exports are determined by the level of foreign
income and the foreign prices of U.S.  exports.  Foreign income is
determined outside the model, while foreign prices are calculated
from domestic prices and exchange rates.  The responsiveness of
exports to prices is also estimated from historical data.\10 The
foreign exchange rate is determined in the model. 

Given this outline of the Jorgenson-Wilcoxen model, the effects of
environmental regulation can be explained.  Using historical data,
the shares of the operating costs of pollution abatement to the total
costs for each industry are computed.  Then the share of total costs
excluding those pollution abatement outlays can be calculated.  To
simulate the effect of eliminating these operating costs, these
shares are inserted into the cost equations of each industry.\11 The
effect of this operation is to lower commodity prices.  Thus, the
price of investment goods is lower.  Lower-priced investment goods
lead to a higher rate of return on capital, stimulate savings, and
generate more rapid capital accumulation.  Additional capital leads
to a lower price of capital.  Cheaper capital services lead to
further declines in prices of other goods and services, and an
increase in consumption\12 and a greater gross domestic product
(GDP).  In addition, the exchange rate falls, increasing the
international competitiveness of the U.S.  economy. 

The next step is to simulate the effect of eliminating investments in
pollution control equipment.  This elimination is simulated as a
decrease in the price of investment goods.\13

This decrease leads to higher rates of return on capital, increased
capital accumulation, a lower price for capital, lower overall
prices, increased consumption, and higher GDP. 


--------------------
\2 Professors Dale W.  Jorgenson of Harvard University and Peter J. 
Wilcoxen of the University of Texas at Austin developed the model on
the basis of economic research they conducted over many years. 

\3 In the model, there is a price equation for each industry's
output.  This equation represents the price of the commodity produced
as depending on the prices of energy, materials, capital, labor
services, and a time variable.  In turn, the model represents the
price of energy as depending on the prices of coal, crude oil,
refined petroleum, electricity, and natural gas.  The price of
materials depends on the prices of all other commodities used by the
industries to produce their final products for sale.  In the model,
the amounts of capital, energy, labor, and materials used to produce
commodities are derived mathematically from these price equations. 

\4 In the model, each household's behavior is represented by an
equation in which the household's satisfaction depends on commodity
prices, total household expenditures, and differences in tastes
related to the demographic characteristics of the household. 
Assuming that the household maximizes its satisfaction, how much it
spends on individual commodities depends on the prices paid for the
commodities and the household's demographic characteristics and total
expenditures. 

\5 Wealth includes future earnings from the supply of capital and
labor services by the household, transfer payments from the
government, and an imputed value of leisure time. 

\6 In making these decisions, it turns out that the amount of wealth
that is consumed and saved in any given year depends on what the
household expects future prices and interest rates to be.  The model
represents this consumption of goods, services, and leisure time as
depending on the interest rate and the price of this consumption. 

\7 For each year, the model represents the household's satisfaction
as depending on the prices of leisure and an aggregate consumption
good.  The price of leisure is assumed to be equal to the after-tax
wage rate, and the price of the aggregate consumption good is equal
to a price index based on the commodities consumed.  From this
relationship, it follows that the household's demand for consumer
goods and leisure in each year depends on the prices of these goods
and leisure and the amount of wealth that the household has decided
to spend in that year. 

\8 New capital goods are produced out of individual commodities
according to a production equation estimated from investment data for
the period 1947 through 1985.  The technology for producing these new
capital goods is represented by a price equation for investment goods
that is estimated using the data described above. 

\9 In the model, households are also investors. 

\10 A set of foreign demand equations is included, in which foreign
consumption of U.S.  goods and services depends on foreign income and
prices for these goods and services. 

\11 The price equation described in footnote 1 is modified. 

\12 Recall that consumption includes leisure time as well as personal
consumption expenditures. 

\13 The price of investment goods is reduced by the proportion of
total investment attributable to pollution control. 


   STEP THREE:  EMISSIONS
   ESTIMATES
--------------------------------------------------------- Appendix I:3

To estimate emissions, EPA used several models.  EPA entered into a
cooperative agreement with Argonne National Laboratory to conduct the
emissions analysis for the retrospective study.  EPA used Argonne's
Integrated Model Set (IMS); the Office of Air Quality Planning and
Standards' (OAQPS) Trends methodology; and ICF Resources' model of
the electric utility industry, the Coal and Electric Utilities Model
(CEUM).  Output from the macroeconomic model became input for EPA's
efforts to determine net changes in emissions between the two
scenarios generated in step one; that is, the base-case and
counterfactual scenarios. 


      THE INTEGRATED MODEL SET
      (IMS)
------------------------------------------------------- Appendix I:3.1

The IMS, designed for the National Acid Precipitation Assessment
Program (NAPAP), estimates emissions for criteria air pollutants,
broken down by source for each state.\14 The IMS is driven by data
from a macroeconomic model, such as the Jorgenson-Wilcoxen model, and
predicts energy use and subsequent emissions for each state and
energy source on the basis of estimates of economic activity and
cost-effective fuel choices.\15 The IMS contains separate models for
emissions from mobile sources, utilities, industrial sources, and
commercial and residential sources.  (See fig.  I.1.)

   Figure I.1:  IMS Modeling
   Framework

   (See figure in printed
   edition.)

Note:  The J/W model is the Jorgenson-Wilcoxen Model. 

For the retrospective study, the IMS supplied emissions data for the
base-case and counterfactual scenarios for SO2, NOx, VOC, total
suspended particles (TSP), CO, hydrocarbons (HC), and lead for the
transportation sector, by state, by sector, by 5-year intervals over
the period 1975-90, and by fuel type.  Even though the base-case
scenario represents what actually happened--that is, implementation
of the Clean Air Act--EPA chose to estimate emissions for this
scenario rather than to use data on actual emissions.  EPA made this
choice because comparing actual and modeled emissions data may bias
any outcome, since actual emissions data may result from variables
not accounted for in the emissions models.  Argonne modeled
historical emissions (base-case) using assumptions similar to those
used for the counterfactual scenario so that the scenarios could be
compared. 


--------------------
\14 The 1970 Clean Air Act amendments provided authority to establish
ambient air quality standards.  Currently, there are six national
ambient air quality standards, or standards for criteria air
pollutants:  ozone, including precursor compounds, i.e., volatile
organic compounds (VOC); carbon monoxide (CO); particulate matter
(PM-10), formerly called total suspended particles (TSP); sulfur
dioxide (SO2); nitrogen dioxide (NO2); and lead. 

\15 The IMS accounts for the effect of the fuel choices of one source
on other sources.  For example, if utilities chose coal for fuel,
their decision may affect the fuel choices of industrial users, which
would in turn change emissions from these sources. 


      THE TRENDS METHODOLOGY
------------------------------------------------------- Appendix I:3.2

EPA supplemented estimates from the IMS with estimates from the OAQPS
Trends methodology.  The Trends methodology uses point-source
estimates and area-source estimates to develop emissions figures. 
Point-source estimates include estimates of emissions from specific,
identifiable sources.  Areas-source estimates include estimates of
emissions from many small sources, such as residential fuel
combustion, solid waste disposal, and fugitive dust emissions such as
those resulting from wind erosion of land.  The Trends methodology
uses estimates of economic activity as indicators of emissions.  Fuel
consumption and deliveries, tons of refuse burned, and raw material
processed are examples of some of the activities used as indicators. 
Emissions factors--that is, estimates of the average rate of
emissions from many sources combined--are then used to translate
these activity levels into the estimates of emissions.  The Trends
methodology is depicted in figure I.2.  Also, table I.1 includes
information on how IMS and Trends cover emissions estimates. 

   Figure I.2:  Trends Methodology

   (See figure in printed
   edition.)

Note:  J/W is the Jorgenson-Wilcoxen Model. 



                          Table I.1
           
            Emissions Estimates Covered Using IMS,
              Trends Methodology (TR), and CEUM


Source                    SOx   NOx   TSP   VOC   CO    Lead
------------------------  ----  ----  ----  ----  ----  ----
Mobile sources

On-highway light-and      IMS   IMS   IMS   IMS   IMS   TR
heavy-duty vehicles

Off-highway T vehicles    R T   R T   R T   R T   R     TR

Utilities                 CEUM  CEUM  CEUM  IMS   IMS/  IMS/
                                                  TR    TR

Industrial sources

Fuel combustion boilers   IMS   IMS   IMS   TR    TR    IMS

Nonboilers                IMS   IMS   TR    IMS   TR    TR

Commercial/residential    IMS   IMS   IMS   IMS   IMS   TR

Waste disposal/           IMS   IMS   IMS   IMS   IMS   TR
miscellaneous
------------------------------------------------------------

   STEP FOUR:  AIR QUALITY
   ASSESSMENT
--------------------------------------------------------- Appendix I:4

For its overall approach to assessing air quality, EPA is estimating
the differences in air quality, by state, between the base-case and
the counterfactual scenarios.  For the counterfactual scenario, EPA
is using several air quality models and statistical estimating
techniques to assess the effects on air quality of not having air
pollution controls.  EPA is modeling the base-case scenario using
historical air quality concentrations to develop profiles for each
state's air quality. 

To assess air quality for the counterfactual scenario for the eastern
states, EPA is using the Regional Acid Deposition Model (RADM) as its
primary means for estimating SOx, NOx, ozone, and the sulfate and
nitrate components of total particulates.  RADM is a highly
sophisticated computer model whose output consists of estimates of
air quality in 20- to 80-kilometer zones, or grids, in the 31 eastern
states.  EPA will aggregate these substate grid measurements into
statewide estimates of air quality for the eastern states.  Ozone
measures will be supplemented by the Ozone Isopleth Model (OZIP). 
For the western states, EPA will use heuristic models to estimate
regional air quality and roll up or roll back historical
concentrations of air pollutants on the basis of these changes.\16
Heuristic models are simple models that assume that air quality
varies in direct proportion to changes in emissions.  According to
the Chief of EPA's Air Quality Modeling Section, the agency must
depend on heuristic models for the western states because other, more
elaborate models like RADM have not been developed for use in the
west. 


--------------------
\16 "Roll up" methodologies involve using local and county level
pollution concentrations and extrapolating these to the state level
on the basis of emission sources, prevailing weather patterns, and
pollutant type. 


   STEP FIVE: 
   CONCENTRATION-RESPONSE
   FUNCTIONS
--------------------------------------------------------- Appendix I:5

Once estimates of changes in air quality have been completed, EPA
plans to estimate the effects of those changes on human health and
the environment.  The relationship between air quality and its
adverse effects is being assessed using concentration-response
functions; that is, specific concentrations of air pollutants may
result in negative health or environmental responses.  EPA will
estimate the effects of air pollution concentrations on human health,
including morbidity and mortality (sickness and death) and on the
environment, including changes in visibility, surface water quality,
agricultural productivity, and forest quality, and damage to
materials.  EPA will develop concentration-response functions for
each of the criteria air pollutants on the basis of existing and
published scientific literature and the air quality estimates from
step three. 

According to EPA officials, the agency's criteria documents will
serve as the basis for developing concentration-response functions. 
Criteria documents are EPA's assessments of the available scientific
literature and are used to support setting national ambient air
quality standards.  The agency prepares separate criteria documents
that include toxicological, epidemiological, and clinical studies for
each standard.  Agency officials noted, however, that criteria
documents are sometimes outdated and may be less comprehensive than
what is needed for this step in the retrospective study.  As a
result, the agency intends to update its scientific understanding of
concentration-response functions by reviewing and summarizing the
scientific literature published since the criteria documents were
issued.\17

Agency officials stated that this approach will allow the three types
of research (toxicological, epidemiological, and clinical studies) to
complement one another and will result in a more complete
understanding of the concentration-response functions for each
criteria air pollutant. 

According to agency officials, toxicological, epidemiological, and
clinical studies contribute to an understanding of how exposure to
air pollution affects health and the environment.  In toxicology
research, animals are exposed to contaminants under laboratory
conditions.  In epidemiological studies, the effects of air pollution
on humans are measured in natural, uncontrolled settings, and
clinical analyses examine human responses to pollution under these
controlled conditions.  EPA plans to use all three types of research
in its assessment of concentration-response functions. 

Each research approach offers a different perspective on how
pollution affects people and the environment, and each has its
strengths and weaknesses relative to the other approaches.  The major
advantage of the toxicological approach is its ability to examine the
physiological effects of pollution over a long period of time, making
it the preferred methodology to study chronic effects.  In addition,
the toxicology approach enables researchers to test animals at higher
doses and to use more invasive techniques than can be used in human
studies.  The major disadvantage is that, in general, applying the
results of animal studies to humans is highly uncertain.  The
clinical approach is better for studying acute health effects in
humans because, under controlled conditions, these effects can be
isolated.  The drawback of this approach is that for cost reasons, it
usually involves only a few subjects.  As a result, clinical studies
are unlikely to adequately describe the average population.  In
addition, the role of actions taken to reduce exposure (e.g.,
installing air purifiers) complicates the interpretation of clinical
research because such mitigating behavior cannot be accounted for
with laboratory controls.  Typically, epidemiological studies use
ambient air concentrations as a proxy for exposure, taking into
account the population's activity patterns and defensive actions. 
Thus, the epidemiological approach implicitly addresses the effects
of actions people take to avoid pollution.  However, epidemiological
studies have weaknesses because observed statistical associations
between health effects and air pollution may be influenced by
unmeasured variables.  Epidemiological research is sometimes
questionable because associations between effects and pollution
concentrations may be influenced by unmeasured factors, or by not
accounting for all plausible factors because of measurement problems. 


--------------------
\17 Because criteria documents are used mainly to support the
national ambient air quality standards, they emphasize establishing
the lowest level of exposure that will result in adverse effects, not
on a more inclusive, general understanding (best guess) of the
relationship between concentrations of air pollutants and
physiological or environmental responses. 


   STEP SIX:  BENEFITS VALUATION
--------------------------------------------------------- Appendix I:6

To assess benefits, EPA hired Industrial Economics, Inc.  (IEc), to
recommend categories of the benefits of changes in air quality and to
recommend methodologies for setting values for these categories.  IEc
has identified seven categories of benefits, including reduced
morbidity, reduced mortality, increased visibility, improved surface
water quality, increased agricultural productivity, improved forest
quality, and decreased materials damage.  For each category of
benefit, IEc located a source document in which a literature review
(i.e., a broad sweep of the economic literature) was conducted.  IEc
then integrated empirical estimates from the literature with
estimates from EPA offices.  IEc stated that according to existing
models, changes in surface water quality, agricultural productivity,
and forest quality are the primary benefits being quantitatively
measured. 

To assess the effects on morbidity, IEc began its work with a
literature survey conducted by David Weitzel of National Economic
Research Associates, Inc.\18 The survey includes information on
symptoms and effects, as well as dollar estimates of the cost of
avoiding symptoms.\19 Also, estimates of benefits in the survey are
presented as ranges to reflect variability in the underlying
estimates.  IEc recommended that EPA consider the estimates of
economic benefits presented in table I.2; all the reported estimates
reflect a willingness to pay to avoid one day of each symptom.  To
assess the effects on mortality, IEc reviewed three existing surveys
of literature on the value of life, including Viscusi (1992), Miller
(1990), and Fisher et al.  (1989).\20 Each survey evaluates three
types of value-of-life estimates:  wage-risk studies, contingent
valuation studies, and hedonic studies.\21 IEc recommended a
lognormal distribution for the value of a statistical life.\22



                          Table I.2
           
              IEc's Recommended Morbidity Values


Symptom                                  Low    Best    High
------------------------------------  ------  ------  ------
Throat congestion                      $3.77  $16.35  $36.44
Head congestion/sinus                   4.40    8.20   65.41
Coughing                                1.26    4.98   52.83
Asthma attack                          11.81   32.48   53.80
Eye irritation                         15.72   15.72   34.88
Headache                                1.26   25.16   50.44
Shortness of breath                        0   10.57   98.12
Chest tightness                         6.29    6.29   22.71
Nausea                                 22.01   22.01   63.25
Drowsiness                             18.87   18.87   39.51
Allergy (chronic)                       5.66   15.72   25.79
Bronchitis and emphysema (chronic)     55.35   84.28  111.96
MRRAD\a                                38.37   38.37   82.52
Angina pectoris                        83.12  106.71  124.60
------------------------------------------------------------
\a Minor respiratory restricted activity day. 

To assess the value of environmental effects (reduced visibility,
changes in surface water quality, changes in agricultural
productivity, changes in forest quality, materials damage), IEc
considered several approaches and recommended that EPA use parts of
the National Acid Precipitation Assessment Program (NAPAP) analysis. 
To assess the value of visibility, IEc used contingent valuation
studies from the economic literature that calculate individuals'
willingness to pay for changes in visual range.  It then generated
benefit curves on the basis of the economic literature.  To assess
the value of surface water quality, IEc considered existing models of
the effect of acid rain on recreational fishing, as well as
contingent valuation studies of other values.  To place a value on
recreational fishing, IEc recommended generating new estimates using
the NAPAP model set for the New England/Adirondack and Mid-Atlantic
regions or using the existing NAPAP results to generate estimates for
these regions.\23 IEc recommended considering qualitative approaches
to assess the value of other environmental effects. 

To assess the value of effects on agricultural productivity, IEc
recommended considering the effects on economic welfare of changes in
agricultural productivity and using agricultural sector models and
estimates of economic impact.  To place a value on agricultural
productivity, IEc recommended generating benefits using agricultural
sector models.  IEc also recommended applying estimates from the
literature for agricultural benefit categories not captured by the
model. 

To assess the value of effects on forest quality, IEc recommended
using existing forest sector models to assess the implications for
economic welfare of changes in forest productivity.  IEc also
conducted a literature review and related analysis to determine the
potential magnitude of other categories of benefits for the forest. 
Furthermore, to place a value of the effects on timber production,
IEc recommended using the most recent version of the Timber
Assessment Market Model (TAMM) to generate estimates of benefits.\24
Finally, to assess the value of the effects on materials, IEc
recommended using existing models designed to capture the effects on
economic welfare of materials damage induced by air pollution.  IEc
conducted a literature review to determine the potential magnitude of
economic benefits not addressed by these models. 

To include the benefits of reduced air toxics, IEc conducted a
literature review to determine if these benefits had been estimated;
it determined that no studies had been conducted.  IEc is working
with Professor Mark Dickey of the University of Georgia to research
ways to set a value on the benefits of reduced air toxics.  In the
meantime, the ACCACA recommended that EPA use the same mortality
valuation measure for effects related to the criteria pollutants. 



(See figure in printed edition.)Appendix II

--------------------
\18 David L.  Weitzel, Economic Valuation of Environmental Health
Benefits:  A Review of the Literature, National Economic Research
Associates, Inc., report to the Washington State Department of
Ecology, Dec.  31, 1990. 

\19 The estimates of benefits related to morbidity were drawn from
existing contingent valuation studies cited in the Weitzel survey. 
In contingent valuation studies, individuals are asked how much they
value specific changes in environmental quality.  For example,
individuals are asked how much they would be willing to pay for
increased visibility in the Grand Canyon. 

\20 W.K.  Viscusi, Fatal Tradeoffs:  Public and Private
Responsibilities for Risk (New York:  Oxford University Press, 1992);
T.  Miller, "The Plausible Range for the Value of Life:  Red Herrings
Among the Mackerel," Journal of Forensic Economics, vol.  3, no.  3
(1990), pp.  17-39; A.  Fisher, L.G.  Chestnut, and D.M.  Violette,
"The Value of Reducing Risks of Death:  A Note on New Evidence,"
Journal of Policy Analysis and Management, vol.  8, no.  1, (1989),
pp.  88-100. 

\21 Wage-risk studies assume that people will demand higher wages for
taking riskier jobs.  If higher wages can be correlated with riskier
jobs and vice versa, a value for the risk can be derived.  Hedonic
studies rely on the prices people pay for commodities being affected
by changing environmental quality. 

\22 IEc identified three sources of bias that should be considered in
assigning a distribution to the value of a statistical life:  risk
perception, age, and income.  It has been found that such values
depend on differences in the perception of the risk faced, the age of
the person affected, and personal income. 

\23 To use the existing NAPAP results to generate estimates for the
New England/Adirondack region, EPA would have to include wide-error
bounds to account for the limited scope and uncertainties introduced
by transferring one set of benefits (e.g., the benefits of acid rain
reductions) to this problem. 

\24 TAMM does not address the effects on federal land or on other
nations, e.g., acid rain on Canadian forests. 


COMMENTS FROM THE ENVIRONMENTAL
PROTECTION AGENCY
=========================================================== Appendix I



(See figure in printed edition.)

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 



(See figure in printed edition.)

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 2. 

See comment 1. 



(See figure in printed edition.)

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 



(See figure in printed edition.)

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 



(See figure in printed edition.)

See comment 1. 

See comment 1. 

See comment 3. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 

See comment 1. 



   GAO'S COMMENTS
--------------------------------------------------------- Appendix I:7

1.The report has been amended to include updated information as well
as clarifications in response to EPA's technical and editorial
concerns. 

2.We clarified the language in the report to reflect that ACCACA
expressed its preference for the ICF Resources utility emissions
estimates.  However, because ACCACA is an advisory committee, EPA is
not bound to accept its recommendations. 

3.We revised the language in the report to correct the possible
impression that IEc played a role other than a strictly advisory one. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix III

RESOURCES, COMMUNITY, AND ECONOMIC
DEVELOPMENT DIVISION, WASHINGTON,
D.C. 

William F.  McGee, Assistant Director
Charles W.  Bausell, Jr., Assistant Director
Chester F.  Janik, Assignment Manager
Eugene W.  Wisnoski, Evaluator-in-Charge
Robin S.  Langdon, Staff Evaluator

OFFICE OF THE GENERAL COUNSEL

Doreen Stolzenberg Feldman, Senior Attorney

