UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                            WASHINGTON, D.C. 20460
                                   '   '                      OFFICE OF TW ADMINISTRATOR
                                                              SCIENCE ADVISORY BOARD
EPA-SAB-CAACAC-LTR-93-007

Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, DC 20460

      Subject:     Science Advisory Board's review of the Office of Policy, Planning,
                  and Evaluation 's (OPPE) and the Office of Air and Radiation's
                  (OAR) progress on the retrospective and prospective studies of  ,  .  .
                  the impacts of the Clean Air Act.

Dear Ms. Browner:

      On March 25, 1993,  the Clean Air Act Compliance Analysis Council (CAACAC)
met to address a variety of issues related to the retrospective and prospective Clean
Air Act (CAA) impact studies required by Section 812 of the CAA amendments of
1990.  The discussions at that meeting reflected both the Charge provided to the
Council and issues raised by the  background  documents that were also provided in
advance. The Council's reactions to the three major topics covered at the meeting
follow,

a) Estimation of Costs and  Macromodeling

      At our previous two meetings, the Council had expressed a number of concerns
regarding the methods and assumptions of macroeconornic modeling employed in the
retrospective study. At its March 25 meeting, the CAACAC was very impressed by
the progress that has been made in addressing the five major concerns raised in our
earlier reviews.

      1) The CAACAC had observed that the change in Gross Domestic Product
(GDP) is  not a conceptually valid  measure of economic cost.  In response, a new
equivalent variation (EV) measure was presented. This measure, based on the work

                                                                   Recyckd/Ftecydabta

-------
of Jorgenson and Slesnlck, relies on an intertemporal utility function for infinitely-lived
"dynasties" with different economic characteristics.  The Council believes that this is a
very sound basic approach (The "dynasty" formulation has a distinct air of unreality
about it.  However, the CAACAC  does not believe this approach is biased in any
obvious way and recommends against attempting to develop a superior approach
within the Section 812 study). We have not yet seen the details of the models and
calculation methods employed in  this analysis, however, and, accordingly, the
CAACAC cannot put its imprimatur on the numbers provided.

      Four additional points were raised regarding the maeroeconomic cost analysis.
First, the calculation of EV seems to depend on extrapolated future compliance costs,
while traditional income measures do not. It would be useful to have some  assurance
that results are not sensitive to alternative plausible extrapolations. Second, the
measures presented mix stocks and flows.  The reported EV measure is a wealth-like
stock, as opposed to the flow concepts (output, consumption, etc.) used elsewhere in
the maeroeconomic modeling. It  would be highly desirable to present a flow magni-
tude that would bear some family resemblance to GDP. Within the current EV
structure, perhaps the annuitized  value of the lifetime wealth change ("annuitized
augmented income," or AAt) would be preferable for the purpose of presentation.

      Third, the presentation made to the CAACAC employed numerous different
measures to deal with the range of issues considered (output, capital, consump-
tion/leisure, consumption, and so  on). As far as possible,  the final written presentation
should rely on a single preferred measure of impact (such as the AAI just mentioned)
to tie together the different issues. Fourth, while we believe that the estimated
relations between EV and such variables as income and family size are of interest, we
urge the Agency to de-emphasize the Jorgenson-Slesnick summary measures of
equity in the final presentation. These are both complex and controversial measures
and seem unlikely to provide much information to most readers beyond that contained
in the results for different family types.

      2) The treatment and presentation of direct and indirect costs was responsive to
our expressed concerns.  The presentation could be unified by presenting the results
for the AAI concept discussed above.  Some CAACAC members noted that  adaptation
and substitution might imply that indirect costs should be negative, whereas in the
Jorgenson-Slesnick analysis these costs are positive because of impacts on capital
accumulation.

-------
      3} This analysis now dearly distinguishes endogenous technical change (ETC)
from factor substitution {FS}, The CAACAC had argued that this distinction is crucial
because the ETC effect is not as well-established in the economic literature as is the
FS effect. As we now understand the analysis, estimates of the cost of the CAA
without the ETC effect are roughly double the estimates with ETC.  Because the ETC
effect is apparently of critical importance, the CAACAC strongly recommends that runs
with and without ETC be made for each of the major variations, sensitivity analyses,
and experiments in the macroeconomic analysis.

      4} In response to our earlier suggestions, the modelers made alternative
assumptions about net capital flows from abroad. These had no  significant effect on
the results of the analysis. Although none of these  assumptions are strictly consistent
with the model's strong emphasis on maximizing behavior, the  CAACAC does not
recommend that this topic be pursued further in the context of this study.

      5) The CAACAC had asked that the Agency analyze the implications of alterna-
tive cost estimates produced by other U.S. government agencies -- in particular, the
estimates of mobile source compliance costs produced by the Bureau of the Census.
As we had expected, total cost estimates are very sensitive to which mobile source
cost series is employed.   Because no study of this sort can be  stronger than the data
on which it rests, we urge the Agency to assign a high priority to careful examination
and analysis of at least all alternative cost estimates published by U.S. government
agencies.  The final report should contain a careful analysis of the reasons for
important differences between these estimates, to the extent these reasons can be
ascertained, if it is not possible to reduce the range of credible estimates to a single
point by strong theoretical and empirical arguments, the final report should show the
dependence of all main results on choice of cost estimates,

b) Health Effects of Lead and Other Air Toxics

      A wide range of issues were discussed under this general heading. The
CAACAC's reactions and recommendations are as follows.

      1) In the analysis presented to the CAACAC,  health effects estimates associat-
ed with exposures to outdoor air toxics have been restricted to  cancer incidence for 27
pollutant categories.  The  estimated total annual cancer cases number between 1,720
and 2,706, of which one pollutant category, products of incomplete combustion (PIC)

-------
accounts for between 438 and 1,120 cases.  We urge the EPA to develop methods to
deal with a range of important toxics that are not carcinogens. Such methods will be
important in the prospective study and could be significant in the retrospective study
as well. In addition, in order to evaluate both the retrospective and prospective
studies, the CAACAC would like a poHutant-by-poIiutant indication as to whether the
assumed exposure-response relationships were based on human, other primate, or
rodent studies.

      2.) The numbers presented to the Council were based on sums of 95% upper
confidence limits for individual incidence rates.  While it may be appropriate to work
with such upper bounds in a regulatory setting, measures of central tendency are
required for cost/benefit analysis. The only valid way to determine an unbiased
estimate for the total impact is to sum a central tendency estimate, such as the mean
or median number of cases in each category. (This problem may also arise for
population distributions as well as exposure-response relationships.)  The Agency
must develop and apply the methods necessary for an analysis based on measures of
central tendency, not 95% upper bound limits.  When this is done, we expect the totat
impact of the 27 carcinogens considered would most likely be well below 1,000 cases
per year.

      3) It should also be noted that the incidence estimates presented to the
CAACAC were based on some assumptions we find to be untenable, such as nearly
continuous exposure to the concentrations found out-of-doors.  In fact, most people
spend about 90% of their time indoors, and the concentrations of pollutants  having
outdoor sources are much lower in indoor air, especially for chemically reactive vapors
and all forms of particulate matter.  If exposures are really less than  those currently
assumed, then the risk will be less in proportion to the reduced exposure. In order to
evaluate responsibly the Agency's estimates of GAA costs and benefits, the  CAACAC
would like  a pollutant-by-pollutant indication of the extent to which the modeling uses
actual exposures rather than some hypothetical exposure. Ideally; of course, all risk
estimates for cost/benefit analysis should be based on sound estimates of actual
exposures, rather than the sort of worst case scenarios often employed in regulatory
contexts,

      4) With reasonable adjustments for exposure and means, rather than 95%
upper bound risks, the total annual cancer impact of ambient air toxics is probably
closer to a few hundred than a few thousand cases. In contrast, EPA has estimated

-------
that there are 5,000 to 20,000 radon-associated lung cancers per year in the U.S.,
approximately 3,000 long cancers associated with environmental tobacco smoke
(ETS), and about 2,000 cancers associated with volatile organics (VOCs). VOCs
include many chemicals also included among ambient air toxics, but, for these agents,
the indoor concentrations are generally much higher than outdoor concentrations.  In
addition, the adverse health effects of ETS and VOCs are not limited to cancer.
Finally, all of these health impacts are, however, small in  comparison to those
associated with exposures to three of the criteria pollutants: lead (Pb), particulate
matter (PM),  and ozone (03).  This reinforces the importance of developing methods
to deal with adverse health effects other than cancer.

       5) The data bases on lead exposure, human  health effects, and benefits are
more comprehensive and more thoroughly validated than those for any of the other
pollutants under consideration, We urge the Agency to take full advantage of the
large amount of research that has been done on this important pollutant. Lead fs also
interesting as a paradigm for the CAACAC evaluation because it is a persistent
pollutant that can reach humans by direct inhalation, by inhalation after re-suspension
from soil, and via incorporation into the food supply.

      Among the benefits to the analytic program from further refinement of the lead
exposure and effects evaluations are the lessons from these analyses that can be
applied to evaluations of the exposures and effects of other pollutants.  For example,
the atmospheric dispersion and lead uptake-biokinetic models can be improved
through the determination and validation of transfer coefficients beyond those needed
in earlier applications of the models. Thus, the results of  the forthcoming modeling of
lead dispersion from coal burning power plants could be extended to the modelling of
the dispersion of other toxic metals (such as mercury, arsenic, and cadmium) from the
same power plants, and similar extrapolations could be made on biological fate and
metabolism.

      6) The CAACAC believes it was appropriate to review contingent valuation (CV)
studies eliciting wiliingness-to-pay for avoiding exposures to carcinogens; such studies
might serve either to corroborate or to correct estimates of the value of a statistical life
(VSL) derived from hedonic wage studies.   In view of the state of the CV literature,
however, the  CAACAC recommends that statistical lives saved as a result of air toxics
control be valued using hedonic techniques.  Any bias that might be introduced as a

-------
result is likely to be very small compared to the uncertainties attached to the corre-
sponding dose-response functions.

      The major complication is that the number of life-years saved must be approxi-
mately the same.  Since the median age of workers in the typical labor market study is
about 43 years, preventing an immediate fatality saves approximately 30 life-years. If
the cancers that may be prevented by controlling air toxics would claim an equivalent
number of life-years, no problem exists.  If,  as seems possible, these cancers claim
fewer life-years, an appropriate adjustment would have to be made in the VSL.

c) Design of the Prospective Assessment

      The CAACAC was pleased to have an opportunity to discuss a range of issues
affecting the design of the prospective study early in the design process.  Our main
substantive recommendation is that the Agency should learn from the retrospective
study that as a rule, resources are better spent on developing sound, comprehensive
data relevant to key issues than on elaborate modeling or literature review efforts.  On
process, the CAACAC feels strongly that the Agency should actively seek to involve
CAACAC, its individual members, and other outside experts throughout the design of
the prospective study.  It will be a major challenge to do this highly visible study well,
and the Agency will be well-served by frank, early discussions of important problems
and issues at a range  of informal expert gatherings.

      We appreciate the opportunity to review the progress to date on the CAA
impact analysis and look forward to receiving your response to the major points raised
in this letter.
                                          Sincerely,
                              Dr. Richard Schmalensee
                              Chair
                              Clean Air Act Compliance Analysis Council

-------
                               Distribution List
Administrator
Deputy Administrator
Assistant Administrators
Deputy Assistant Administrator for Research and Development
Deputy Assistant Administrator for Water
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries

-------
 U.S. ENVIRONMENTAL PROTECTION AGENCY
 SCIENCE ADVISORY BOARD
 CLEAN AIR ACT COMPLIANCE ANALYSIS COUNCIL
CHAIR

Dr. Richard Schmatensee, Department of Economics, Massachusetts Institute of Technology,
Cambridge, MA

MEMBERS

Dr. Ronald Cummings, Department of Economics, University of New Mexico,
Albuquerque, NM

Dr. Daniel Dudek, Environmental Defense Fund, New York City, NY

Dr. A. Myrick Freeman, Department of Economics, Bowdoin College,
Brunswick, ME

Dr. Robert Mendelsohn, Yale School of Forestry, New Haven, CT

Dr. William Nordhaus, Department Of Economics, Yale University, New Haven, CT

Dr. Wallace E. Gates, Department of Economics, University of Maryland
College Park, MD

Dr. Paul R. Portney, Resources for the Future, Washington, DC

Dr. Thomas H. Tjetenberg, Department, of Economics, Colby College, Waterville, ME

Dr. W. Kip Viscusi, Department of Economics, Duke University, Durham, NC

SAB COMMITTEE LIAISONS

Dr. William Cooper (EPEC)
Mr. Richard Conway (EEC)
Dr. Morton Lippmann (LAQC)
Dr. Roger McClellan (CASAC)

DESIGNATED FEDERAL OFFICIAL
Mr. Samuel Rondberg, Executive Secretary, Environmental Health Committee, Science
Advisory Board (A101F), U.S. Environmental Protection Agency, Washington, DC 20460
(202) 260-6552

-------
STAFF SECRETARY
Ms. Mary L Winston, Environmental Protection Agency, Science Advisory Board, A101-F),
Washington, DC  20460  {202} 260-6552

ASSISTANT STAFF DIRECTOR
Mr. Robert Flaak, Science Advisory Board, U.S. Environmental Protection Agency

STAFF DIRECTOR
Dr. Donald G. Barnes, Science Advisory Board, U.S. Environmental Protection Agency
                                       9

-------

-------