United States Office of Air Quality EPA-450/5-83-001a
Environmental Protection Planning and Standards August 1982
Agency Research Triangle Park NC 27711
Air
c/EPA Benefit Analysis
of Alternative
Secondary
National Ambient
Air Quality
Standards
for Sulfur Dioxide
and Total
Suspended
Particulates
Volume I
-------
FINAL ANALYSIS
BENEFITS ANALYSIS OF ALTERNATIVE SECONDARY
NATIONAL AMBIENT AIR QUALITY STANDARDS FOR
SULFUR DIOXIDE AND TOTAL SUSPENDED PARTICULARS
VOLUME I
BENEFITS ANALYSIS PROGRAM
ECONOMIC ANALYSIS BRANCH
STRATEGIES AND AIR STANDARDS DIVISION
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
U-S- ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK
NORTH CAROLINA 27711
AUGUST 1982
-------
-------
FINAL ANALYSIS
BENEFITS ANALYSIS OF ALTERNATIVE SECONDARY
NATIONAL AMBIENT AIR QUALITY STANDARDS FOR
SULFUR DIOXIDE AND TOTAL SUSPENDED PARTICULATES
By:
Ernest H. Manuel, Jr.
Robert L. Horst, Jr.
Kathleen M. Brennan
William N. Lanen
Marcus C. Duff
Judith K. Tapiero
With the Assistance of:
Richard M. Adams
David S. Brookshire
Thomas D. Crocker
Ralph C. d'Arge
A. Myrick Freeman, III
Shelby D. Gerking
Edwin S. Mills
William D. Schulze
MATHTECH, Inc.
P.O. Box 2392
Princeton, New Jersey 08540
EPA Contract Number 68-02-3392
Project Officer:
Allen C. Basala
Economic Analysis Branch
Strategies and Air Standards Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
August 1982
-------
PREFACE
This report was prepared for the U.S. Environmental Protection
Agency by MATHTECH, Inc. The report is organized into six volumes
containing a total of 14 sections as follows:
Volume I
Section 1:
Section 2:
Sect ion 3 :
Volume II
Section 4:
Section 5:
Section 6:
Volume III
Section 7:
Section 8:
Volume IV
Section 9:
Volume V
Section 10:
Section 11:
Volume VI
Section 12:
Section 13:
Section 14:
Executive Summary
Theory, Methods and Organization
Air Quality and Meteorological Data
Household Sector
Residential Property Market
Labor Services Market
Manufacturing Sector
Electric Utility Sector
Agricultural Sector
Extrapolations
Bibliography
Summary of the Public Meeting
Analysis of Pollutant Correlations
Summary of Manufacturing Sector Review
The analysis and conclusions presented in this report are those
of the authors and should not be interpreted as necessarily reflecting
the official policies of the U.S. Environmental Protection Agency.
11
-------
ACKNOWLEDGMENTS
This report and the underlying analyses profited considerably
from the efforts of Allen Basala, who served as EPA Project Officer,
and V. Kerry Smith, who served as a reviewer for EPA. Allen provided
the initiative and on-going support to conduct an applied benefits
analysis. Kerry's technical insights and suggestions are reflected in
nearly every section of the report.
James Bain and Tom Walton of EPA, and Jan Laarman and Ray
Palmquist, who served as reviewers for EPA, also contributed
substantially to individual report sections through their advice and
comments during the course of the project. Also providing helpful
comments and assistance were Don Gillette, Fred Haynie, Neil Frank and
Larry Zaragosa, all with EPA.
Several other members of the Mathtech staff contributed to the
project during various stages of the work. They included Robert J.
Anderson, Jr., Neil Swan, John Keith, Donald Wise, Yaw Ansu, Gary
Labovich, and Janet Stotsky.
The production of the report was ably managed by Carol Rossell,
whose patience remained intact through countless drafts and deadlines.
Carol was assisted by Sally Webb, Gail Gay, and Deborah Piantoni.
Finally, we extend our appreciation to the many dozens of
individuals, too numerous to list here, who provided advice,
suggestions, and data during the course of the project.
111
-------
CONTENTS
1. EXECUTIVE SUMMARY
Background 1-1
Study Scope and Objectives 1-3
Theoretical Base 1-4
Empirical Base 1-6
Efficient Use of Resources 1-6
Reliability Checks in the Analysis 1-7
Public Meeting 1-7
General Assumptions Underlying the Benefits
Estimates 1-9
Air Quality Scenarios 1-9
Economic Scenario 1-10
Summary of Estimated Benefits 1-12
Scope of the Estimates 1-12
Benefits of the Current S0« Secondary Standard . 1-14
Benefits for the Alternative SO- Secondary
Standard 7 1-15
Benefits for the Current TSP Secondary
Standard 1-16
Geographic Distribution of Benefits 1-17
Organization of the Report 1-18
References 1-19
2. THEORY, METHODS AND ORGANIZATION
Introduction 2-1
Physical Effects of S02 and TSP 2-1
Behavioral Responses to Air Pollution 2-3
Economic Benefits of Improved Air Quality 2-4
Overview of Later Subsections 2-7
-------
CONTENTS (continued)
•
2. THEORY, METHODS AND ORGANIZATION (continued)
The Definition of Economic Benefits 2-10
Willingness to Pay 2-11
Consumers' Surplus 2-14
Producers' Surplus 2-15
Benefits of a Particular Action or Event 2-17
Measuring Benefits by Cost Savings 2-18
Alternative Measures of Economic Benefits 2-20
Measuring Benefits in Intermediate Markets 2-22
Summary 2-24
Indirect-Market Approaches for Estimating Air
Quality Benefits 2-24
Air Pollution Effects on Firms 2-26
Air Pollution Effects on Households 2-28
Other Indirect-Market Approaches 2-35
Air Quality Benefits Not Observable in
Market Behavior 2-37
Aggregation and Coverage of Benefits
Categories 2-39
Organization of the Study and Report 2-41
Organizing Framework 2-41
Selection of Sectors 2-4 3
Coverage of Sectors 2-49
Validation of Results 2-52
References 2-56
3. AIR QUALITY AND METEOROLOGICAL DATA
Introduction 3-1
Air Quality Data 3-1
Total Suspended Particulates (TSP) 3-7
Sulfur Dioxide (SO-) 3-15
Other Pollutants .7 3-27
Meteorological Data 3-27
References 3-31
-------
FIGURES
Number Page
2-1. An illustration of the calculation of total
willingness-to-pay 2-14
2-2. Illustration of the calculation of consumers'
surplus 2-15
2-3. Producers' surplus 2-16
2-4. Consumers' and producers' surplus 2-17
2-5. Change in economic surplus 2-19
2-6. Change in consumers' and producers' surplus due to
an improvement in air quality from S to S1 2-28
2-7. Change in consumers' and producers' surplus due to
an improvement in air quality from S to S' 2-30
2-8. Benefits of an improvement in ambient air quality ... 2-33
2-9. Organizing framework for the study 2-42
VI
-------
TABLES
Number Page
1-1. Scope of the Study 1-8
1-2. Current National Ambient Air Quality Standards for
S02 and TSP 1-11
1-3. Coverage of Economic Activity in Each Sector 1-13
1-4. Estimated Benefits in Sectors Analyzed for
Current SO- Secondary Standard 1-14
1-5. Estimated Benefits in Sectors Analyzed for
Alternative SO^ Secondary Standard 1-16
1-6. Estimated Benefits in Sectors Analyzed for
Current TSP Secondary Standard 1-17
2-1. Coverage of Economic Activity in Each Sector 2-50
3-1. National Ambient Air Quality Standards for
Particulate Matter 3-9
3-2. National Ambient Air Quality standards for
Sulfur Dioxide 3-16
3-3. An Alternative Set of Ambient Air Quality
Standards for Sulfur Dioxide 3-17
vn
-------
SECTION 1
EXECUTIVE SUMMARY
-------
SECTION 1
EXECUTIVE SUMMARY
BACKGROUND
Traditionally, the U.S. Environmental Protection Agency has
emphasized the evaluation of the negative economic consequences of
alternative air pollution control regulations. Hence, while control
costs, price increases, and output declines have been examined for the
candidate alternatives, the beneficial economic aspects of the regula-
tions have been largely ignored. This is in spite of the acknowledged
fact that information about the economic benefits would provide
decision makers with a more balanced view of the consequences of
regulation. Ideally, the improved balance would result in regulations
which are more economically efficient.
The failure to include comprehensive and rigorous treatment of
the economic benefits of environmental policies has not been without a
supporting rationale. Economic benefits have been difficult to
measure due to a wide variety of methodological, data, and ethical
problems. Some of the more important problems that have prevailed
included: the nonmarket nature of the clean air commodity; the need
to separate health and welfare benefits; the need to consider how the
affected populations can reduce the effects of pollution through their
1-1
-------
own actions (i.e., "substitution possibilities"); and the distribu-
tional considerations of any accrued benefits.
This study was initiated by EPA's Office of Air Quality Planning
and Standards in order to allow broader consideration of potential
economic benefits. It was commissioned following a determination that
research and technical advances in the 1970's had reached the point
where measurement of selected economic benefits was feasible. These
advances included:
• The development of more reliable air pollution
measurement techniques and a broader network of air
pollution monitoring sites.
• Advances in economic theory that clarified the
requirements for measuring the economic effects of
nonmarketed commodities such as clean air.
• Developments in physical effects research that identi-
fied more of the mechanisms and determinants of air
pollution damage.
• Advances in economic theory and data that provided new
techniques for understanding consumer and producer
economic behavior.
• New types of computer software that made practical the
use of advanced statistical techniques.
This study has thus drawn heavily upon new data and research
advances from the past decade or so. We believe the study consider-
ably expands the information available about the economic benefits of
selected air pollution control regulations. The details of the study
are presented in the report sections which follow.
1-2
-------
STUDY SCOPE AND OBJECTIVES
Under authority provided by the Clean Air Act of 1970, as amended
(1), the U.S. Environmental Protection Agency (EPA) has promulgated
regulations that require the attainment of certain standards of
ambient air quality throughout the country. One set of standards has
been established at levels designed to protect human health; these are
referred to as the Primary National Ambient Air Quality Standards
(PNAAQS). A second set of standards has been established at levels to
protect the public welfare (e.g., other items of value such as vegeta-
tion and materials); the latter are referred to as the Secondary
National Ambient Air Quality Standards (SNAAQS). The Clean Air Act
also requires the EPA to review these standards on a periodic basis.
This study was initiated by EPA's Office of Air Quality Planning
and Standards as part of its review of the current secondary standards
for sulfur dioxide (S02) and total suspended particulates (TSP). The
primary goal of this study was to estimate selected non-health
benefits which would result from achieving alternate S02 and TSP
secondary standards. The rationale for excluding the human health
benefits was that the primary standards presumably protect human
health with an adequate margin of safety. Hence the health effects
associated with the more restrictive secondary standards were expected
to be small or nonexistent.
1-3
-------
Other benefits not covered in this study are those associated
with reduced acidic deposition and visibility improvement. Acidic
deposition effects were not considered because the role of ambient SCu
in acidic deposition has not been clearly identified. Visibility
effects were omitted because fine particulate concentrations rather
than TSP are believed to be more directly associated with visual
quality impacts.
The scope of the analysis was further refined to meet the
following four objectives:
• A sound theoretical and empirical base for estimating
benefits.
• Efficient use of available project resources.
• Incorporation of reliability checks into the analysis.
• Peer review of study results.
Theoretical Base
The correct starting point for a benefits analysis is a consid-
eration of the value which society places on improved air quality.
Economists generally agree that attempts to measure this value should
be based on individuals' own valuations, as evidenced by their
"willingness-to-pay" for improved air quality. For conventional types
of goods and services, evidence of willingness-to-pay can be found by
analyzing supply and demand conditions in the markets where those
goods and services are traded. Unfortunately, however, no such market
1-4
-------
for air quality exists. In the absence of a market for air quality,
we are unable to observe directly the price, and thus the willingness-
to-pay, for improvements in air quality.
An alternative approach is to observe how air quality influences
the behavior of the members of society in related markets. Although
members of society may not directly interact in a market for air
quality, they do interact in markets for many other commodities and
services potentially affected by air quality. As one example, we know
from field studies that sulfur oxides cause corrosion damage to
certain types of steel structures used by industrial firms. Unless
the various responses to corrosion (e.g., maintenance and repair) are
costless, the effect of air pollution will be to increase the costs
and* lower the productivity of these firms. Thus, by observing the
extent to which costs (or productivity) vary in areas with differing
air quality, while taking into account other factors that influence
cost, one can infer the economic effects of air pollution on these
firms.
The theoretical base for this benefits analysis is a series of
models structured to simulate optimizing behavior by members of
society in markets where air quality has an indirect influence. We
refer to groups of these members of society as "economic sectors".
Consideration of optimizing behavior allows for the possibility of
differing responses to air pollution. This reduces the potential for
1-5
-------
biased estimates and also allows inferences consistent with the
willingness-to-pay criterion.
Empirical Base
The development of a sound empirical base for the study involved
exploration, identification, evaluation, and, where necessary, augmen-
tation of economic and aerometric data bases. This entailed extensive
discussions with representatives of the Bureau of Economic Analysis,
Bureau of Labor Statistics, Bureau of the Census, the Monitoring and
Data Analysis Division of EPA, and other public and private organiza-
tions. As a result of these procedures and discussions, the quality
and limits of the economic and aerometric data were better understood.
Efficient Use of Resources
In April of 1980, a document was distributed describing the
various economic sectors which could benefit from reduced SC>2 and TSP
concentrations and the detailed plans for estimating benefits in each
of these sectors. In May of 1980, a meeting was held to discuss the
document and to assess the likelihood of successfully estimating
benefits in each economic sector. Over 60 persons representing
several agencies and disciplines attended the May meeting and provided
valuable feedback. As a result, selected economic sectors offering
the greatest prognosis for success were identified.
1-6
-------
The selected sectors included the household, manufacturing,
electric utilities, and agricultural sectors. Within each of these
sectors, statistical models were constructed to estimate some of the
benefits. These estimates were then extrapolated geographically and
categorically to broaden the scope of the study. The overall scope of
the study is summarized in Table 1-1.
Reliability Checks in the Analysis
The sector approach is significantly different from many of the
dose-response analysis techniques which have been traditionally used
to estimate benefits. Consequently, several validation checks were
built into the study. These included assessments of model form, data
inputs, and model performance. Not only environmental economists and
econometricians, but also plant pathologists, materials effects
engineers, and air quality data specialists provided suggestions
regarding these checks. This doesn't mean our approach or the
analysis results will satisfy all. However, we have recognized at the
outset the multidisciplinary nature of benefits analysis and accounted
for it in model design, input data selection, and interpretation of
results.
Public Meeting
In July of 1981, a public meeting was held to review a draft
version of this report. An announcement concerning the meeting was
made in the Federal Register to encourage public attendance and
1-7
-------
p
B
CO
a
^
o
Ot
rH
1
rH
a
EH
II
II
II
11
II
11
II
II
II
II
II
II
II
II
II
II
11
II
II
II
II
II
11
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
11
II
II
II
II
II
II
II
II
i!
H
II
II
n
II
tl
H
n
n
n
11
||
n
n
H
ii
ij
|i
n
n
n
n
n
}|
n
j|
n
n
||
n
n
n
1]
||
u
11
"8
4-*
03 g
•rH -3
ni ii
CD cn
C u
cn
r-H
O
1
1
rJ
0)
5
i-H
rH
>4
3
4J
O
03
UH
3
C
§
^J
CD
c cn
0 0)
•iH
cn u
4J 4J
u cn
CD 3
UH ^n
UH C
U -H
rj,
C
•tH
3
4J
O
UH
i
TJ cn
CD CD
4J -iH
rH 4J
CD cn
!-i 3
CN -r-i
2*1
_£j
C 03
0 C
•
O
c cn
O CD
•r-l
cn in
4-> 4J
o cn
CD 3
UH ^O
UH C
H-rH
C
_0
1 4-1
C— jQ
(0 -rH
4J 4J
, cn
r-l -rH
O^ C -rH
C C
C CD
10 § 1
i— 1 C7> 03
E C rH
•rH O 03
C -fH
i
i
i
i
i
i
I
03 cn u cn if
LJ
ii o
II til
II CD
II G
P
•H C
iJ.S
u cn
CD cn
i-H -r-l
CJ £
i i
1
1
CD
5 .
4J
UH U
i! ° a
ii JJ
•-* "O Ii U O
4-> O II 0) CD
CD
C
£
V4
U
a
u
u
4J
o
0)
rH
63
t) 1
QJ I
cn cn
II C C
C 1
0 1
4-1 1
4-J 1 1
•r-t -rH
13 T3
3j (Tl
81 || 4-1 Jj
II !u u cn cn
O i
cn i
i
1
i
CD CD
OJ CS
•K «
II -K
1-8
-------
comments. A panel of experts in environmental benefits analysis was
also assembled to critically discuss the report at the meeting.
Comments from the panel and audience were generally favorable.
Following the meeting, various revisions were made to the analysis and
this final version of the report was prepared. Two additional plausi-
bility checks were also undertaken at the suggestion of the panel.
The findings of the plausibility checks did not appreciably change the
original conclusions of the study. A summary of the public meeting
and the results of the two additional plausibility checks are included
in a new volume of the study (Volume VI).
GENERAL ASSUMPTIONS UNDERLYING THE BENEFITS ESTIMATES
Air Quality Scenarios
The estimates reported in this study represent the benefits of
enforcing compliance with secondary standards (SNAAQS) for S02 and
TSP, as compared to a situation in which only the primary standards
(PNAAQS) are enforced. The basic idea, then, is one of calculating
the benefits of improving air quality from the level of the primary
standard to the level of the secondary standard. There are certain
definitional problems in describing this scenario, however, because of
the different averaging times used in the primary and secondary
standards.*
* Averaging time refers to the period of time over which the pollution
measurements are averaged.
1-9
-------
As shown in Table 1-2, the current standards for SCU and TSP are
stated in terms of several different averaging times, and in the case
of S02f there is no primary standard at the averaging time used in the
secondary standard. As discussed in Section 3, there are computa-
tional procedures which allow approximate conversions to be made
between pollution measurements at different averaging times. These
techniques were used in the case of SC>2 to establish an equivalent 24-
hour secondary standard. Benefits for SC>2 were then calculated for
two different scenarios — one based on the 24-hour equivalent of the
current 3-hour secondary standard, and one based on an alternative 24-
hour secondary standard. Both sets of benefits estimates are reported
in the next section. In the case of TSP, the estimate of benefits is
based on the 24-hour averaging time since there is a primary standard
corresponding to the secondary standard at that averaging time. For
both S02 and TSP, compliance with the primary standard is assumed to
occur by 1985 and with the secondary standard by 1987.
Economic Scenario
The benefits estimates reported for each sector are based on
detailed economic scenarios specific to each sector. The details of
these scenarios can be found in later sections of the report. All of
the sector estimates, however, incorporate the same three-part calcu-
lation. First, the benefits in each year after 1985 are calculated.
Time horizons of 50 years or more are used in each sector. Second,
all annual values are then converted to 1980 dollars. Third, the
1-10
-------
a
en
EH
Q
Z
H
EH
M
J
<
^
O
OS
r- 1
<
PH
Z
Cd
M
03
^
<
J
<
Z
O
1— 1
EH
<
Z
EH
Z
H
OS
OS
£3
C.)
•
CN
1
rH
U
J
03
<
EH
II
II
II
II
II
II
II
II
II
W
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
||
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
, — .
f")
e.
en
a.
-~*
T3
^4
03
T3
C
03
4->
CO
>1
4J
• H
rH
03
3
D1
jj
•rH
1
J-|
03
'O
^
O
CJ
en
>,
i_i
03
e
.rH
S_|
PU
03
T3
C
03
4-1
C
03
4->
3
rH
i— 1
O
Cu
o
O
1 ro
1 -
0 1
00 1
•K
C
0
.r-|
4J
03
J_l
C 4J
03 C
cy a;
e u
c
0 O
•rH O
4J
(D iJ
e 3
oj x: o
T3 4-> -C
•H -rH |
X rJ 00
O 03
•- > e
T3 r-H 3
03 g
SJ 3 --H
3 C X
>W C 03
-H < 2
3
en
i
i
LO
VO
m
•K
c
O
• H
4J
03
i_i
•U
C
0)
u
c
o
o
Jj
3
O
.c
1
CN
e
3
e
•i-i
X
03
g
O
in
1 r-l
1
in o
r*- {&
CN
•K
C
0
• r-l
cn 4-1
0) 03
4-1 lj
03 4-1
r-l C
3 CO)
O 03 O
•r4 0) C
•u SO
!J 0
03 O
Qj •'H rJ
U 3
T3 4J O
d) CJJ £
T3 E 1
C O rr
0) < S
O
EH
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
•
jj
03
0)
\j
8,
4-)
o
Z
•K
>
-— ,
CN
— '
0)
u
c
0)
Vl
(U
4H
OJ
OS
• •
-------
annual benefits in 1980 dollars are then converted to a discounted
present value in 1980, using a (real) discount rate of 10 percent.*
The benefits numbers reported are thus discounted present values in
1980, stated in 1980 dollars.
SUMMARY OF ESTIMATED BENEFITS
Scope of the Estimates
The economic sectors considered in this study, as described
previously, do not include the full range of sectors in the economy.
The specific coverage of sectors is identified in Table 1-3. The top
half of the table identifies the final demand sectors in the economy;
the bottom half shows the producing sectors. The first and second
columns in the table identify the specific sectors and the percent of
economic activity accounted for by each sector. The third column
indicates the percent of each sector covered by the basic statistical
analysis in each sector. For example, the basic analysis in the
household sector covered 24 major metropolitan areas and a subset of
total consumption expenditures. This represented approximately 17
percent of total activity in the household sector. Through extrapola-
tion of the results for the basic analysis, coverage of this sector
was expanded to about 45 to 55 percent of the household sector, as
shown in the fourth column.
* Estimates using discount rates of 2 and 4 percent were also
developed and are reported in the individual sector reports.
1-12
-------
TABLE 1-3. COVERAGE OF ECONOMIC ACTIVITY IN EACH SECTOR
Percent coverage
Final demand sector
Percent of
final demand Basic Basic plus
analysis extrapolation
Households*
Government
Other
Totals
Producing sector
Agriculture, forestry
and fisheries
Mining and
construction
Manufacturing
Transportation,
63.5
20.5
16.0
100.0
Percent of
GNP
3.1
7.1
23.9
9.0
17
0
• o
11**
Percent
Basic
analysis
2-15
0
4-8
8-11
45-55
0
0
29-35**
coverage
Basic plus
extrapolation
2-15
0
25-30
15-20
communication and
utilities
Commercial and
services
Government and other
Totals
43.6
2-3**
* Goods and services consumed by individuals and certain nonprofit
institutions. Includes rental of dwellings but not purchases of
dwellings. The latter are included with "other".
**
Weighted average coverage.
Source: Estimates of final demand and GNP shares are from U.S.
Department of Commerce, Bureau of Economic Analysis. Survey
of Current Business. July 1979. Tables 1.1 and 6.1.
1-13
-------
It is important to note that this study does not provide complete
coverage of all possible sectors. Nor does it include consideration
of effects such as impacts on climate or the ecosystem. In this
respect, the benefits reported in the study are conservative estimates
of the benefits of the secondary ambient air quality standards.
Benefits of the Current SO,, Secondary Standard
The estimated benefits for the current S02 secondary standard are
shown in Table 1-4. As indicated previously, these are discounted
TABLE 1-4. ESTIMATED BENEFITS IN SECTORS ANALYZED, FOR CURRENT S02
SECONDARY STANDARD* (discounted present values for 1980
in millions of 1980 dollars)**
Basic Basic analysis
Sector analysis* with extrapolation
Households — 2.0
Agricultural 0.2 0.2
Manufacturing — 6.4
Electric Utilities — 0.1
Current secondary standard for SO- is 1,300 Mg/m3, based on a 3-
hour averaging time. Standard not to be exceeded more than once
per year.
**
Discount rate of 10 percent is assumed.
+ Estimates shown are for effects which were statistically signifi-
cant at the 10 percent level or less. Estimates would be larger if
higher significance levels are used.
— Equals zero.
1-14
-------
present values in 1980, assuming a 10 percent discount rate and an
infinite time horizon. The benefits are predicted to be quite small.
This is in part because not all of the adverse effects of SCu have
been measured. It also occurs because so few counties are out of
compliance with the standard. In particular, only 18 counties were
out of compliance in 1978 for the 3-hour averaging time standard.
This compares with over 3,000 counties in the U.S. When the
conversion is made to an equivalent standard based on a 24-hour
averaging time, only five counties were out of compliance.
Benefits for the Alternative SO,, Secondary Standard
The current SC>2 secondary standard, based on a 3-hour averaging
time, was apparently established to prevent vegetation damage. For
materials damage, a longer averaging time is believed to be more
appropriate. Estimates were thus developed for an alternative
secondary standard (260 /xg/m3) based on a 24-hour averaging time. The
estimated benefits for this standard are shown in Table 1-5. These
estimates are also discounted present values. Compared to the
previous table, the benefits are considerably larger because many more
counties would be out of compliance with this alternative standard.
1-15
-------
TABLE 1-5. ESTIMATED BENEFITS IN SECTORS ANALYZED FOR ALTERNATIVE SO2
SECONDARY STANDARD* (discounted present values for 1980 in
millions of 1980 dollars)**
Sector
Household
Agricultural
Manufacturing
Electric utilities
Basic
analysis+
920
22
345
56
EJasic analysis
with extrapolations
1,101
22
1,912
124
* Alternate secondary standard is 260 /xg/m^, based on a 24-hour
averaging time. Standard not to be exceeded more than once per
year.
** Discount rate of 10 percent is assumed.
+ Estimates shown are for effects which were statistically signifi-
cant at the 10 percent level or less. Estimates would be larger if
higher significance levels are used.
Benefits for the Current TSP Secondary Standard
The estimated benefits for the current TSP secondary standard are
shown in Table 1-6. As can be seen, the benefits of this standard are
considerably larger than for either of the S02 standards. In large
part, this is due to the fact that more counties are out of compliance
with the standard, as well as to the difference in SO- and TSP
economic effects. As in the previous tables, the entries are
discounted present values.
1-16
-------
TABLE 1-6. ESTIMATED BENEFITS IN SECTORS ANALYZED FOR CURRENT TSP
SECONDARY STANDARD* (discounted present values for
1980 in millions of 1980 dollars)**
Basic Basic analysis
Sector analysis+ with extrapolations
Household 2,299 3,630
Agricultural — —
Manufacturing 920 8,029
Electric utilities —
* Current TSP secondary standard is 150 Mg/m , based on a 24-hour
averaging time. Standard not to be exceeded more than once per
year.
** Discount rate of 10 percent is assumed.
+ Estimates shown are for effects which were statistically signifi-
cant at the 10 percent level or less. Estimates would be larger if
higher significance levels are used.
— Equals zero.
Geographic Distribution of Benefits
As described more fully in later sections of the report, the
estimated benefits for the various standards exhibit significant
geographic variation. The variation reflects both the distribution of
households, industries and farm activity, as well as the geographic
variations in air quality. For the current 3-hour S02 standard, most
of the estimated benefits arise in the East North Central region
1-17
-------
(Ohio, Indiana, Illinois, Michigan, Wisconsin).* For the alternative
24-hour SC>2 standard, benefits are heavily concentrated in the East
North Central region and the Mid-Atlantic region (New York, New
Jersey, Pennsylvania), with additional benefits arising in several
other regions. For the current 24-hour TSP standard, benefits are
predicted in all nine regions of the country. The largest benefits
arise in the East North Central region, the Pacific region
(Washington, Oregon, California, Alaska and Hawaii), and the Mid-
Atlantic region.
ORGANIZATION OF THE REPORT
The remainder of the report is comprised of 13 sections, bound in
six volumes. Section 2 describes the general theory of benefits
analysis and presents an overview of the approaches taken in this
study. Section 3 describes the air quality and meteorological data.
Section 4 describes the structure, estimation, and results of the
household sector model. Sections 5 and 6 focus on other approaches to
household sector benefits analysis. These include the analysis of
residential property values and wage differentials to assess the
willingness to pay for air quality. In Section 7, the details of the
manufacturing sector model are explained. The electric utilities
sector model is the subject of Section 8, while the agricultural
sector model is covered in Section 9. The extrapolation procedures
* The regional definitions are those used by the U.S. Bureau of
Census.
1-18
-------
and results for national benefits estimates are described in Section
10. Section 11 presents a comprehensive bibliography of reference
material to this study. A summary of the public meeting is included
in Section 12, and two additional plausibility checks undertaken
following the public meeting are summarized in Sections 13 and 14.
REFERENCES
1. Clean Air Act (42 USC 7401 et seq) as amended by the Clean Air
Amendments of 1970 (Public Law 91-604), December 31, 1970; and
the Clean Air Act Amendments of 1977 (Public Law 95-95), August
7, 1977.
2. U.S. Code of Federal Regulations, Title 40, Part 50, as revised,
July 1, 1979.
1-19
-------
SECTION 2
THEORY, METHODS AND ORGANIZATION
-------
SECTION 2
THEORY, METHODS AND ORGANIZATION
INTRODUCTION
The primary objective of this study is to estimate selected non-
health benefits of attaining the secondary national ambient air
quality standards for SO^ and TSP. The methods employed in this
effort are consistent with the mainstream of current work in benefit-
cost analysis. Thus, expositions of both the general theory and the
application to environmental problems are widely available.* The
purpose of this section is to provide a brief review of the underlying
theory and then to show how application of the theory provides the
organizing structure for the analysis and for the overall report.
More specialized discussions of the theory can be found in the main
sections of the report.
Physical Effects of SO^ and TSP
To set the context for the discussion, it will be useful to
consider first the various potential effects of S02 and TSP on the
* A general reference on benefit-cost analysis is Mishan (1). Baumol
and Gates (2), Crocker e_t al. (3), Freeman (4), Maler (5), and
Smith (6) are general references on environmental benefits analysis.
2-1
-------
physiqal environment. Generally speaking, the following categories of
effects have been suggested in the literature:
• Soiling
— Interior and exterior surfaces
— Fabrics
— Equipment
• Damage to materials
— Metals
— Coatings
— Building materials
— Electronic components
— Fabrics
— Paper
— Leather
• Damage to vegetation
— Crops
— Forests
— Ornamental plants
—• Native vegetation
• Damage to animals
— Livestock
Pets
Fish and wildlife
— Other organisms
• Effects on climate
— Temperature
Precipitation
• Reduced aesthetics
— Visibility
It should be noted that the evidence for these effects is not uniform.
For example, the increased corrosion of certain metals exposed to £©
2-2
-------
is well established.* However, the effects on climate are less well
known. Similarly, it should be noted that some effects may be
beneficial. For example, certain plant species growing in sulfur-
deficient soils have been found to benefit from exposure to ambient
SC>2.** However, adverse effects are believed to be the more usual
case.
Behavioral Responses to Air Pollution
The economic consequences of the physical effects listed above
will depend significantly on how individuals and organizations respond
to air pollution. Hence, these responses must be taken into account
when estimating the benefits of improved air quality. Possible
responses to air pollution may include one or more of the following:
• Ameliorative actions — These include actions taken in
response to pollution damage. In the case of
particulate soiling, the response might be increased
frequency of cleaning. For metal corrosion, it may be
sandblasting and repainting. For plant injury, it
could be increased fertilization.
• Preventative actions — These include actions taken to
prevent air pollution damage. In the case of soiling,
this may include air conditioning and filtering.
Metal corrosion can be prevented or reduced through
the use of coatings. Crop damage may be reducible
through use of pollution-resistant species.
* See Section 7 of this report for references.
** See Section 9 of this report for references.
2-3
-------
Relocation — Because pollution varies from place to
place, one defensive measure is to relocate to a
cleaner area. This may include changing residences,
job locations, relocating from one building or factory
to another, and traveling to more distant recreation
sites. In the case of climate or aesthetic effects,
this may be the only possible response other than no
response at all.
No action — One alternative always available is no
action. This may be the case if the effects of
pollution are not perceived. Or it could be the case
if the cost of the other possible actions exceeds the
perceived benefits of the action.
The important point is that households, businesses, and
agricultural enterprises may take positive steps to prevent or reduce
air pollution damage. In cases where such actions are taken, the
appropriate measure of economic damage is not the economic damage that
would have occurred in the absence of preventative or ameliorative
actions; rather it is the cost of these actions plus any residual
damage. Presumably this cost will be less than the damage in the no
action case, or the action would not have been undertaken. It is
therefore important to take these alternative behavioral responses
into account — otherwise, estimates of damages and benefits will be
biased. This requires a technical approach which can incorporate
optimizing behavior on the part of individuals and organizations. The
various models used in this study have this capability.
Economic Benefits of Improved Air Quality
As discussed above, air pollution can produce physical effects
which may have economic consequences. As also noted above, the
2-4
-------
magnitude of the economic effect will depend on how individuals and
organizations respond to air pollution. In addition, however, the
measured effect will also depend on the method used in placing a value
on the physical damages.
In the case of soiling and materials or vegetation damage, two
rather common approaches to valuation have been used. One approach
measures the value of the damage by the cost of ameliorative or
preventative actions, and in some cases an adjustment is made for
residual damage. Another approach assigns a value equal to the
intrinsic value of the material lost (e.g., tons of corroded metal
times price per ton, or tons of crop lost times price per ton); and in
some cases an adjustment is made for the labor cost for installation
of the material. With either of the two approaches, the resulting
estimate of economic damage is viewed as the economic benefit that
would result if air pollution were reduced.*
An alternative method of valuation, and the one generally
subscribed to by economists, is based on the concept of willingness-
to-pay. The basic idea in this case is that the value of a program to
reduce air pollution should be measured by the aggregate amount that
members of society would be willing to pay to be in a preferred state
* An example study which used the first approach is Fink _et_ al_. (7).
An example of the latter approach is Salmon (8). In a moreTecent
study, SRI (9) used one or the other approach depending on the
availability of data. None of these studies demonstrated that the
valuation method used was in fact consistent with how individuals
and organizations have actually responded to air pollution effects.
2-5
-------
(e.g., clean air) as opposed to a less preferred state (e.g., dirty
air). As will be shown in a later subsection, the willingness-to-pay
measure of benefits includes two components. The first of these is
the conventional notion of a reduction in damages (i.e., savings in
cost). The second allows for the possibility that the cost savings
may reduce prices and thus stimulate an increase in the quantity of
goods and services demanded. Since the increased quantity demanded is
directly attributable to improved air quality, the willingness-to-pay
measure also includes this as a benefit.
The willingness-to-pay concept is generally the more valid
measure of benefits compared to the two other approaches described
earlier. This is the case for at least two reasons. First, benefits
measured by cost savings, if the latter are measured accurately, will
understate true benefits by the value of the increase in demand
mentioned above. Similarly, benefits measured by value of lost
product may overstate benefits by failing to take into account the
price reduction mentioned above. In cases where demand is relatively
insensitive to price, the degree of under- or overestimate may be
small. If demand is sensitive to price, the inaccuracy can be larger.
This will be illustrated further in the next subsection.
The other reason for preferring the willingness-to-pay concept is
that, if measured correctly, it will reflect the optimal behavioral
responses. In contrast, using, for example, maintenance and repair
cost to estimate potential cost savings will be valid only if
2-6
-------
maintenance and repair is the optimal behavioral response. If it is
not optimal (e.g., if preventative actions would lead to lower cost),
then benefits will be overestimated.
Overview of Later Subsections
The remainder of Section 2 is divided into three subsections.
The next subsection provides a more specific definition of the
willingness-to-pay concept. As will be seen, willingness-to-pay turns
out to have a very natural interpretation in terms of the somewhat
more widely known concepts of market supply and demand. In
particular, the willingness-to-pay for a good or service is
identically equal to the integral of the demand function, over the
quantity of the good or service demanded. This means that the
benefits of improved air quality could be readily estimated if one
could determine the demand function for air quality. Unfortunately,
there is no established "market" for air quality in which one can
observe such a demand function directly. Thus, much of the recent
work in air quality benefits analysis has focused on developing other
methods of estimating willingness-to-pay.
Among the various methods that have been developed for estimating
the willingness-to-pay for air quality, two classes of approaches can
be defined. One approach involves looking at behavior in other
markets which may be affected by air pollution. Examples include the
residential property market, the market for maintenance services, and
2-7
-------
the markets for goods and services in which production efficiency
(e.g., agricultural crop yields) may be influenced by air pollution.
We will refer to these various approaches which rely on behavior
observed in related markets as the indirect-market approach. The
second broad class of techniques is often referred to as the
non-market approach. The majority of these techniques involve survey
methods in which individuals are asked about their willingness-to-pay
for air quality. Non-market approaches have-been increasingly used to
evaluate aesthetic benefits of improved air quality.
There is a third broad class of techniques that have been widely
used in air quality benefits analysis. These will be referred to as
the damage function approach. Previous efforts using this approach
have not generally resulted in willingness-to-pay estimates. However,
it is possible to extend these techniques in that direction (see
Section 9 of this report). Damage function approaches typically
couple mathematical dose-response relationships* with estimates of
receptor** inventories and, for example, unit maintenance and repair
costs to estimate economic damages.
Most of the analyses in this study are examples of the use of
market data and indirect-market approaches to estimate benefits. None
of the non-market approaches were attempted because of the time and
* A mathematical function relating physical damage to air pollution
exposure.
**
Items which "receive" (are affected by) air pollution.
2-8
-------
resources that would be required to undertake a survey. The damage
•
function approach was used in the agricultural sector analysis (see
Section 9) and coupled with a market model to evaluate benefits.
Damage functions were not used more widely, however, for several
reasons. These included:
The problems discussed earlier in valuing the
identified physical damages.
The absence of data on the distribution of receptor
inventories.
The unavailability of damage functions for many types
of materials.
The absence of damage functions for more subtle types
of effects such as efficiency losses in machinery and
equipment containing materials affected by air
pollution.
The second subsection following is thus concerned with indirect-market
approaches for estimating air quality benefits.
Given the indirect-market approaches used in this study, a
natural organizing framework for the study was along the lines of
economic sectors; that is, the various groups that participate in the
markets potentially affected by air pollution. Division of society
into various sectors is analytically very useful. This is because it
allows separate characterization of the optimizing behavior, and thus
the response to pollution, in each sector.
2-9
-------
The final subsection discusses the economic sectors so defined
and identifies the particular sectors analyzed in detail. On the
final demand side of the market, the household sector is analyzed in
detail. On the supply side of the market, subsets of the
manufacturing, electric utility and agricultural industries are
analyzed in detail. Potentially important sectors not considered in
this study are the commercial and government sectors. One reason for
omitting the latter sector was that the appropriate form of optimizing
behavior was not clear.
THE DEFINITION OF ECONOMIC BENEFITS
In broad terms, the economic process involves the conversion of
society's stock of resources into goods and services, and the sale or
exchange of these goods and services in the marketplace. This
activity generates economic benefits by allowing people to consume and
produce desired combinations of goods and services. For example,
manufacturing firms consume capital, labor and raw materials in order
to produce saleable goods and services. Households can also be viewed
as producing services by consuming various inputs. One example would
be the consumption of gasoline, tires and automobiles to produce
transportation services. Or perhaps more to the point, they also
consume cleaning supplies and services in order to "produce"
attractive housing services.
2-10
-------
Changes in ambient air quality can alter these market
relationships, including the efficiency of production activities, in a
variety of ways. These alterations can thus increase or decrease the
benefits enjoyed by society. However, before we address the linkage
between air quality and economic benefits, it will be useful first to
consider the concept of economic benefits in a broader sense.
Willingness to Pay
In the most general case, one can say that economic benefits are
generated whenever a transaction, such as the sale of a good or
service, takes place. Economists generally agree that any attempt to
measure these benefits should be based on individuals' own valuations
of the benefits, as evidenced by their "willingness to pay" for the
opportunity to engage in the transaction. Evidence of willingness-to-
pay can thus often be observed by analyzing market transactions.
To consider a more specific case, suppose that we observe sales
of a particular good taking place at a price of $1 per unit. We can
infer that each purchaser of this good is willing to pay at least $1
to have it rather than go without it. Total expenditures for this
good thus represent a lower bound on the sum of all purchasers'
willingness to pay for the good. In fact, of course, they may be
willing to pay much more.
To be able to infer actual willingness-to-pay from observed
behavior clearly requires more information. Exactly the needed
2-11
-------
information is provided by the fact that the marginal buyer (i.e., the
buyer who would not buy if the prince were any higher) is willing to
pay exactly the price he pays and not one penny more. By using this
fact, we can find a differential equation representing market behavior
which, in principle, can be used to measure willingness-to-pay
exactly.
Let q denote the total quantity of the good purchased per unit of
time, and let p(q) represent the market price corresponding to q. In
this case, the marginal willingness-to-pay when (say) Qg units are
purchased is simply the prevailing market price:
£ • p
-------
have denoted by p(q)] , we will have exactly the relationship that is
needed to measure willingness-to-pay.
This relationship between p and q is nothing other than what
economists call the "market demand function" which relates price and
quantity demanded for goods and services. Of course, variables other
than price and quantity are important in this relationship. For
example, we know that income, population size, income distribution,
and many other factors influence demand.* For ease of exposition and
notation here, however, we shall retain our convention of expressing
explicitly only price and quantity.
This procedure for estimating individuals' willingness-to-pay can
be illustrated using simple diagrams from elementary economics. In
Figure 2-1, we have drawn a demand curve (a "demand curve" is a demand
function with all variables other than price and quantity held
constant). At quantity QQ, total willingness-to-pay is given by the
area under the demand curve up to QQ, which is the shaded area in
Figure 2-1.
* Note that the calculation of total willingness-to-pay in Equation
(2.2) involves aggregating individuals' willingness-to-pay. Thus,
population size and income are important influences not only on
demand, but also on total willingness-to-pay.
2-13
-------
Price per
unit
Market demand curve
(D)
0 '
Quantity per
unit time
Figure 2-1. An illustration of the calculation of total willingness-
to-pay.
Consumers' Surplus
Frequently in cost-benefit analysis, willingness-to-pay is
measured net of any charges levied upon customers for the good or
service in question. When this is done, the result is called "net
willingness-to-pay", or more frequently, consumers' surplus. The
"consumers' surplus" measure represents what customers would be
willing to pay over and above what they do pay. This concept is
illustrated in Figure 2-2, where it is assumed that a price of PQ is
charged for each of the QQ units purchased. The shaded area in this
figure represents consumers' surplus (or net willingness-to-pay). The
total user expenditures that have been netted out of willingness-to-
2-14
-------
Price per
unit
D
Quantity per
unit time
Figure 2-2. Illustration of the calculation of consumers' surplus.
pay are given by the rectangle O?QAQQ. (Compare Figure 2-2 with
Figure 2-1.)
Producers' Surplus
Corresponding to consumers' surplus is the concept of
producers' surplus which can be simply illustrated by a figure such as
Figure 2-3. In that figure, D is the demand curve. However, in
Figure 2-3, we have added the marginal cost curve, MC, which
represents the cost of producing one additional unit. As drawn, some
firms are able to produce at lower cost than others. Thus, with price
2-15
-------
Price/cost
per unit
MC
Quantity per
unit time
Figure 2-3. Producers' surplus.
set at PQ — where the demand curve intersects the marginal cost curve
— these firms earn a surplus represented by the difference between
price and marginal cost, or the shaded area in Figure 2-3. Since this
surplus ultimately flows back to members of society (as income to
fixed factors), the producers' surplus is considered to be as much a
benefit as the consumers' surplus.
The sum of consumers' and producers' surplus if often referred to
as economic surplus. Economic surplus can be illustrated as the
shaded triangle in Figure 2-4.
2-16
-------
Price per
unit
Quantity per
unit time
Figure 2-4. Consumers' and producers' surplus.
Benefits of a Particular Action or Event
The economic surplus illustrated previously in Figure 2-4
represents the economic benefit to society that results from
transactions for a particular good or service. Note also that any
action or event which leads to a change in market supply or demand
will also change the magnitude of the economic surplus. In
particular, as can be seen in Figure 2-4, the economic surplus will be
increased by any action or event which causes an upward shift in the
market demand curve or a downward shift in the market supply curve.
The economic surplus will be decreased by any action or event which
causes the opposite shift in either curve. Since an increase
2-17
-------
(decrease) in economic surplus represents an increase (decrease) in
economic benefits to society, this observation provides the basis for
evaluating the benefits (costs) of a particular action or event.
To be more specific, suppose the particular event is an innova-
tion in manufacturing which lowers the cost of producing some consumer
product. Thus, the effect of the innovation is to produce a downward
shift in the marginal cost (supply) curve for the product from MC to
i
MC ; i.e., the cost of producing each additional unit is less. This
effect is shown in Figure 2-5 which represents the market for the
consumer product. As can be seen, the downward shift increases the
economic surplus in the market by an amount given by the shaded area.
The magnitude of this increase represents the economic benefits of the
innovation per unit of time (e.g., per year). Of direct benefit to
consumers of the product is the fact that the price of the product
*
declines from PQ to PQ.
Measuring Benefits by Cost Savings
The change in economic surplus shown previously in Figure 2-5
represents the benefits of the innovation that led to a lowering of
i
production costs from MC to MC . Referring back to Figure 2-5, note
that the change in surplus consists of two components. Area OAC
represents the savings in the cost of producing the original level of
output QQ. Area ABC represents the additional surplus generated when
2-18
-------
Price
per unit
Quantity per
unit time
Figure 2-5. Change in economic surplus.
i
the decline in price from PQ to PQ led to an expansion of sales from
Q to
The discussion above thus indicates that the use of cost savings
to evaluate the economic benefits of an action or event is consistent
with the concept of economic surplus. Benefits measured by cost
savings will always provide a lower bound estimate of the benefits
measured by the change in economic surplus. The difference between
the two measures is given by the area ABC.
2-19
-------
For given supply conditions, the size of the area ABC will depend
on the slope of the demand curve (sensitivity of demand to price).
When demand is insensitive to price, the demand curve will be steeply
sloped, area ABC will be small, and cost savings will closely
approximate the change in surplus. When demand is sensitive to price,
the demand curve will be flatter, area ABC will be larger, and the use
of cost savings to approximate the change in surplus will be less
accurate.
Alternative Measures of Economic Benefits*
The definition of economic benefits developed in the previous
sections is based on the concepts of consumers' and producers'
surplus. These concepts are widely used in empirical benefit-cost
analysis because they can be readily calculated from knowledge of
"ordinary" demand and supply curves. From a more theoretical
viewpoint, the concept of consumers' surplus (C'S) has a certain
imprecision, which has led to the development of four alternative
definitions of benefits. These include the concepts of compensating
variation (CV) and equivalent variation (EV), which are appropriate
for measuring the benefits of a price change; and compensating surplus
(CS) and equivalent surplus (ES), which are appropriate for quantity
* This section discusses certain theoretical refinements which the
general reader may wish to skip over.
2-20
-------
change. These four measures are generally viewed as theoretically
more correct than ordinary consumers' surplus (C'S).*
In a recent paper, Willig (10) has provided an analytical basis
for comparing the C'S, CV, and EV measures of economic benefits. The
general conclusion is that for price decreases, the measures are
related as follows:
CV <_ C'S <_ EV .
Furthermore, the differences between the three measures are small if:
(1) the magnitude of the measured surplus is small relative to the
consumer's income; or (2) the income elasticity of demand for the good
or service under consideration is small. Since air quality benefits
are believed to be small relative to income, the Willig results
support the use of the ordinary consumers' surplus measure as an
approximation to the CV and EV measures. In one part of this study
(see Section 4), it is actually possible to use the more exact CV
measure.
More recently, Randall and Stoll (11) have established
corresponding results for the C'S, CS, and ES measures. Since a
reduction in air pollution is a quantity change, it might seem that
their results are more appropriate here. However, as will be shown
* Precise definitions of the CV and EV measures can be found in any of
the references cited at the beginning of Section 2.
2-21
-------
later in this section, it is possible to transform the problem of
analyzing quantity changes in air pollution to one of analyzing price
changes. Thus, it is the Willig results that are of interest here.
Measuring Benefits in Intermediate Markets*
In the discussion to this point, it has been assumed that the
demand side of the market is represented by consumers, and thus that
the goods and services involved are final goods. It is possible, of
course, for actions or events to affect directly the prices of
intermediate goods (i.e., goods used in the production of other goods)
and through them the prices of final goods. The demand for the
intermediate goods is a derived demand, based solely on their value in
producing other final goods for consumption. Conceptually, consumers'
surplus can only accrue to consumers of these final goods. As a
result, measuring the benefits of an action or event which lowers the
cost of an intermediate good can be more complex than is indicated by
the example presented earlier. For example, in cases where many final
good markets are affected, the measurement of these benefits would
seem to be an extremely involved task requiring the calculation of
benefits accruing in all of these final good markets.
Fortunately, however, the benefits that accrue in the form of
consumers' surplus in the various final good markets can be
* This section discusses certain theoretical refinements which the
general reader may wish to skip over.
2-22
-------
approximated using data on demand and cost relationships in the
intermediate market alone. This is the case because of the
relationship between derived demand for an intermediate product and
demand in the final good market. The critical issue then is how
accurately the benefits measured in the intermediate market
approximate the benefits in the final good markets.
A partial answer to the above question was developed by
Schmalensee (12). He showed that two conditions were sufficient to
guarantee that the approximation would be exact if there were one
final good: competitive markets and infinitely elastic input
supplies. This result has been independently generalized to the case
of an intermediate good used in the production of more than one final
good (13). In the case where markets are monopolistic, Schmalensee
also showed that benefits measured in the input market would
underestimate actual benefits, and thus represent a conservative
estimate.
A more recent generalization of the Schmalensee results was
provided by Just and Hueth (14). Their generalization was in two
directions: (1) a partial relaxation of the assumption of infinitely
elastic input supplies, and (2) an extension to the case of a sequence
of intermediate markets. Their conclusion was that the overall impact
of an event in some intermediate market can be measured by the change
in economic surplus in that market alone if the general equilibrium
supply and demand curves for the market are used. If ordinary supply
2-23
-------
and demand curves are used, then benefits in that market alone will be
measured and total benefits will be underestimated; i.e., estimates
will be conservative.
Summary
This subsection has shown how the concept of willingness-to-pay
can be used to derive a measure of economic benefits, and thus provide
a basis for evaluating the benefits of a particular action or event.
The following subsection describes how these concepts can be applied
to the problem of measuring the benefits of improved air quality.
INDIRECT-MARKET APPROACHES FOR ESTIMATING AIR QUALITY BENEFITS
As indicated in the preceding subsection, one can evaluate the
benefits of a particular action or event by calculating the change in
economic surplus caused by the action or event. Thus, for example, if
one could observe the demand curve for air quality, it would be
possible to calculate the willingness-to-pay for an improvement in air
quality. Unfortunately, however, no market for air quality exists and
thus the demand curve for air quality cannot be observed directly.
An alternative approach is to observe how air quality influences
the behavior of the members of society in other markets. Although
members of society may not directly interact in a market for air
quality, they do interact in markets for many other commodities and
2-24
-------
services potentially affected by air quality. As one example, we know
from field studies that sulfur oxides cause corrosion damage to steel
structures used by industrial firms. By observing the extent to which
increased air pollution causes firms to undertake more frequent
maintenance, or to undertake other preventive measures, one can infer
the extent to which an improvement in air quality would reduce the
costs incurred by these firms. The savings in cost, or more
specifically, the downward shift in the marginal cost curve for these
firms, could then be used to calculate the benefits of improved air
quality.
The purpose of this subsection is to show how the effects of air
quality on the supply or demand for other marketed goods and services
can be used to measure the benefits of improved air quality. Of
particular interest in this section will be two issues:
How can data observed in related markets be used to
estimate air quality benefits?
Are there benefits which cannot be observed in market
behavior?
The key to answering the first question has already been
suggested. Air pollution may influence costs of production for firms
and the demand for products and services by individuals. We also know
from the previous section that any action or event which alters supply
or demand relationships can alter the economic surplus in individual
2-25
-------
markets and thus generate benefits or costs. This leads us to
consider two broad categories of effects:
• The effect of air pollution on the costs of production
by firms.
• The effect of air pollution on the demand for goods
and services by households.
The key to answering the second question requires determining
whether there are air pollution effects which do not result in changes
in market supply and demand conditions. It turns out that such
effects do exist. These are also addressed in this section.
Air Pollution Effects on Firms*
Businesses, industrial firms, and agricultural enterprises
combine labor, capital and materials inputs to produce goods and
services. If these firms are adversely affected by air pollution,
then their costs of production may also be affected. Several examples
of this have already been mentioned: increased maintenance and repair
activity as a result of corrosion damage, increased cleaning activity
in response to soiling, reduced crop yields because of vegetation
damage, etc. Each of these effects, if present, would result in
higher costs of production, for a given level of output, compared to a
situation with less air pollution.
* More detailed discussions of these effects can be found in Sections
7,8, and 9.
2-26
-------
Recall from the previous section that one way of representing
production relationships is by means of the marginal cost or supply
curve for an industry. The marginal cost curve depicts the cost of
producing each additional unit of output. If the industry is
characterized by constant unit costs at every level of output, the
marginal cost curve will be horizontal. Or, as indicated previously
in Figures 2-3 to 2-5, marginal cost may increase at higher levels of
output.
If firms in the industry are affected by air pollution, then, in
these cases, one can view the marginal cost curve for the industry as
functionally dependent on the level of air pollution. This is shown
in Figure 2-6 where the marginal cost curve MC is taken to be a
function of both the level of output Q and the level of air pollution
S. Also shown in the figure is the marginal cost curve when air
i
pollution has declined from S to S . As indicated in the preceding
t
section, the benefits of the improvement from S to S are given by the
change in economic surplus which is shown in the figure as the shaded
area. Note that although the pollution effect is on the supply side
of the market, consumers share in the benefits as a result of the
i
price decrease from PQ to PQ.
It is very possible, of course, that air pollution may affect
production costs in more than one industry. In this case, an
improvement in air quality would shift the marginal cost curve in each
of these industries. Hence, calculating the benefits of a change in
2-27
-------
Price
MC(Q,S)
MC(Q,S )
Quantity
Figure 2-6. Change in consumers' and producers' surplus due to an
improvement in air quality from S to S'.
air quality requires calculating the change in economic surplus in
each directly affected industry. The basic method in each case,
however, is the same.
Air Pollution Effects on Households
The pollution effects described in the previous paragraphs were
assumed to be incident directly on the supply side of the market —
businesses, industries, agricultural enterprises, etc. It is also
possible, of course, for pollution to affect the demand side of the
market, and in particular, to affect households (consumers) directly.
2-28
-------
Effects in this case may include those already mentioned such as
soiling and materials damage, reduced aesthetic values, etc.
Demand Shifts—
Consider, for example, the possibility that demand for use of a
recreation area may depend on the environmental attractiveness of the
area. In this case, if the environmental attractiveness improves,
e.g., through improved air quality, then one would expect the demand
for use of the area to increase.
In terms of the earlier diagrams, the relationship between air
quality and demand for recreation can be depicted as in Figure 2-7.
Note that in the figure, the demand for the recreation activity is
represented by a demand curve which depends parametrically on air
quality S, just as production costs did in the previous section. This
implies that with an improvement in air quality from S to S , the
demand curve shifts upward, indicating that consumers demand more
recreation services at every price. For this type of an effect, it
can be shown (15) that under certain reasonable circumstances, the
shaded area between the two demand curves in the figure is the benefit
of the improved air quality. Note that the shaded area is simply the
change in economic surplus as defined previously.
Implicit Price Shifts—
In some respects, the effect of air quality on recreation demand
is atypical of the effect on the demand for other household goods and
2-29
-------
Price of
recreation
Supply of
recreation
Quantity of
recreation activity
Figure 2-7.- Change in consumers' and producers' surplus due to an
improvement in air quality from S to S1.
services. For example, one might expect soiling and materials damages
due to air pollution to influence household demand for goods and
services such as soaps and detergents, laundry and dry cleaning
services, interior and exterior paints, and so on. In these cases,
however, an improvement in air quality may lead to reduced demand for
these items, rather than increased demand as was the case with
recreation.
The difference arises because of the differing interpretation as
to the exact way in which various goods and services, and therefore
2-30
-------
"air quality", enter consumers' utility functions.* Since recreation
is a final good, one would expect it to enter directly in the utility
functions. Similarly, one might also assume that soaps and detergents
are final goods and thus appear in the utility functions. If we do
this, however, we must then allow that air quality can also enter the
utility function, since we suspect that dirtier air may increase
demand for cleaning items, or at a minimum, affect the satisfaction
associated with the existing level of consumption of these items.
In another interpretation, however, soaps and detergents and the
like are "final goods" only in the way the term is used by economists
in constructing the national income accounts. Another way to look at
the problem is that cleaning items are really derived demands based on
a more fundamental consumer demand for "cleanliness". That is,
consumers derive utility from cleanliness, not from detergent.
Detergent, and clean air, contribute to satisfying the final demand
for cleanliness.
The implication of the second interpretation is twofold. First,
in this interpretation, goods and services like cleanliness enter the
utility function rather than individual goods and services like
detergent. To make this distinction clear, we will use the phrase
"final goods and services" to refer to things like cleanliness, and
the phrase "intermediate goods and services" to refer to things like
* A utility function is a relationship expressing a consumer's
preference for different combinations of goods and services.
2-31
-------
soap and detergent (and air quality). A second implication of this
approach is that we therefore impose certain restrictions on the form
of the consumer's utility function. In particular, it means that we
have assumed that air quality enters the utility function only
indirectly as opposed to appearing directly. We discuss the implica-
tions of this further below.
To illustrate the situation more clearly, suppose we define a
final good called "household operations", which is presumed to enter
directly into consumers' utility functions. Household operations
might include all laundry and cleaning activities, and other forms of
household maintenance. We also assume that some natural index or
measure of the quantity of household operations can be defined.*
Under this assumption, it is then possible to think in terms of a
demand function for household operations, which relates the quantity
of household operations demanded by consumers to the price of house-
hold operations. A graph of such a function is shown in Figure 2-8.
In this formulation, the remaining question is how to define the
unit cost (or price) of "supplying" household operations. As a start,
we know that this cost will depend on the prices of the intermediate
goods used in household operations, namely, the prices for detergent,
dry cleaning, etc. In addition, if air pollution increases the amount
* For a detailed discussion of how such indices can be formed, see
Section 4 of this report.
2-32
-------
Price of
household
operations
•Unit cost at S
Unit cost at S1
Benefits
Demand for
household operations
Quantity of
household operations
Figure 2-8. Benefits of an improvement in ambient air quality.
of detergent, etc., required to,maintain a given level of household
operations (e.g., a given level of cleanliness), then the price will
also depend on the level of air quality. Thus, for a given set of
fixed prices for detergent, etc., the unit cost or price of supplying
household operations will depend parametrically on air quality S.
This relationship is shown in Figure 2-8 for two different levels of
i
air quality, S and S . Note that when air quality improves from S to
i
S , the unit cost curve for household operations shifts downward. The
shaded area in the figure is a measure of the economic benefits
generated.
2-33
-------
A comparison of Figures 2-7 and 2-8 highlights the differences in
the two approaches. In Figure 2-7, recreation and air quality both
enter the utility function. Hence, as air quality improves, demand
for recreation shifts upward and benefits are generated. In Figure 2-
8, the final good "household operations" enters the utility function
and the demand for household operations is independent of air quality.
However, in the latter case, the cost of household operations depends
on air quality. Thus, as air quality improves, the cost of household
operations shifts downward and benefits are generated. The latter
situation is thus analogous to the earlier discussion for firms. That
is, one might view households as "producing" household operations by
using inputs such as detergent and labor. Air pollution increases the
cost of this production by increasing the quantity of inputs required
to produce a given level of household operation.
In general, the consumers' utility functions will contain as
arguments a variety of final goods and services in addition to recrea-
tion and household operations. The use of some of these other goods
and services may also be influenced by air quality. In these cases,
improvements in air quality will lead to shifts in demand or shifts in
the prices for the final goods. Benefits can thus be calculated by
using one or the other of the techniques described above.
In a more advanced analysis, one will also want to take into
account the interrelationship among the expenditures for different
categories of goods and services. This interrelationship is imposed
2-34
-------
by the constraint that total expenditures be less than or equal to
total income. The household sector analysis in Section 4 incorporates
this feature.
Other Indirect-Market Approaches
The two previous sections described the indirect-market
approaches applicable to firms and households. While the techniques
described have broad applicability to markets in general, considerable
attention in the literature has also been given to examination of two
more specialized markets — the residential property market and the
labor market. For completeness, these are reviewed briefly below.
More detailed discussions can be found in Sections 5 and 6 of the
report.
The Residential Property Market—
Many previous studies have developed estimates of air quality
improvement benefits to households by analyzing differences in
residential property values.* The underlying hypothesis in these
studies is that residential property values will reflect not only
housing quality, but also site-specific attributes such as location,
neighborhood characteristics, availability of services, and
environmental quality including air quality. Various studies have
thus attempted to estimate the willingness to pay for air quality by
* See Section 5 for specific references.
2-35
-------
examining differences in property values (or rents) while
statistically controlling for the other determinants of property
value. Estimates of willingness to pay can then be used to estimate
the benefits of an improvement in air quality.
An analysis of residential property values is not easily
adaptable to the problem of estimating the non-health benefits of the
secondary standards. This is because property values are likely to
reflect all perceived effects of air pollution — both health and
non-health effects. Hence, benefits estimates based on property value
differentials are likely to be larger than estimates based on analysis
of expenditures for household operations and other items, using the
techniques described earlier.
The fact that property value-based benefits estimates include a
broader range of effects is still useful information, however. In
particular, by comparing estimates using this approach with estimates
based on the approach described earlier, the plausibility of the
latter can be assessed. This is the use made of property value
techniques in this study (see Section 5).
The Labor Services Market—
A number of studies have also developed air quality benefits
estimates by analyzing geographic differences in labor wage rates.*
See Section 6 for specific references.
2-36
-------
The underlying hypothesis in this case is that wage rate variations
will reflect differences in individual attributes (e.g., educational
attainment), differences in job characteristics (e.g., health and
safety hazards), and locational amenities such as climate, access to
services, and environmental quality. Various studies have thus
attempted to estimate air quality benefits by examining wage rate
variations across geographic areas while controlling statistically for
the other wage rate determinants.
As in the case of the property value technique, air quality
benefits based on wage rate variations are likely to reflect both
health and non-health benefits. Estimates based on this technique are
thus used in this study in the same way as the property value
estimates. That is, they are used to provide an upper-bound estimate
of benefits as a cross-check on estimates derived by examining air
quality effects on the prices for final goods and services.
Air Quality Benefits Not Observable in Market Behavior
Recall that in the absence of a market for air quality, the
technical approach used in this study relies on analyses of other
affected markets. As one might expect, some of the benefits of air
quality improvement cannot be identified in this way. For example,
some members of society may attach a value to environmental quality,
per se, independent of any directly received benefit or enjoyment.
This may arise, for instance, as a desire to preserve the environment
2-37
-------
for later generations, or for later enjoyment by existing generations.
*
These benefits are thus sometimes referred to as "bequest values" or
"option values", respectively. These benefits cannot be observed by
looking at behavior in other markets.
Some of the aesthetic values associated with a clean environment
may also be difficult to detect in other markets. One example of this
is the value of improved visibility. This value is difficult to
observe in other markets because improved visibility has less of a
direct association with the consumption expenditures of households or
production decisions by firms. The residential property value and
labor market techniques may capture some visibility benefits. The
recreation example described previously may also be useful. However,
the other indirect-market approach based on household consumption
expenditures is not likely to capture visibility benefits.
Pollution effects not offset by preventative or ameliorative
actions (e.g., increased cleaning frequency) may also not be captured
by analysis of household expenditures. For example, suppose pollution
causes increased soiling but no action is taken in response. In this
case, a loss of utility may occur, but it would not be observable in
market behavior since no behavioral change has occurred.
In contrast, pollution effects on firms may be observable even if
no behavioral adjustment takes place. For example, if air pollution
reduces crop yields, the loss of output can be observed even if no
2-38
-------
attempts are made to increase yield by, say, increased fertilization.
Or, if air pollution causes an unperceived loss of efficiency in
equipment or machinery, the loss of productivity can be observed in
the firms' economic performance even if no ameliorative action is
taken. Thus, the possibility of underestimating air quality benefits
using indirect-market approaches is more likely to be a problem in the
case of households than in the case of firms.
Lastly, it seems likely that certain effects of air pollution on
the ecosystem, or on climate, would be difficult to observe in market
behavior. For example, if air pollution led to the extinction of
certain plant or animal species, the loss would be measurable only to
the extent that it eventually had a later effect on, say, the food
chain, and thus ultimately on agricultural markets. In other
circumstances, the effects may not be observable in market data,
Aggregation and Coverage of Benefits Categories
The previous sections described a number of techniques for
estimating air quality benefits by observing market behavior. In the
case of air pollution effects on businesses, industries, and
agricultural enterprises, the idea was to look for changes in
production and cost relationships. In the case of households, it was
suggested that air pollution effects may change demand relationships,
or the implicit price of composite household activities, depending on
how one views air pollution as affecting consumers' utility.
2-39
-------
The methods described for both firms and households are basically
partial equilibrium techniques. That is, the analysis for firms is
focused on production relationships, while assuming that demand
schedules are fixed;* the analysis for households focuses on
consumption activities while assuming that market prices are fixed.
Under these assumptions, total benefits are approximately given by the
sum of the benefits calculated in the two sets of analyses. All
previous analyses of air quality benefits appear to have been based on
partial equilibrium approaches. Ideally, one would like to consider
the problem using a general equilibrium approach which would take into
account the interrelationship of the effects in the different sectors.
However, such an approach has not been feasible in empirical applica-
tions to date.
In the case of the household sector, it was noted that two
additional indirect-market approaches also exist, based on the
residential property and labor services markets. It was also noted,
however, that benefits estimated with these techniques include some
overlap with each other and with estimates based on household
consumption decisions. Hence, it is not appropriate to add these
benefits estimates together in developing a total for the household
sector.
* That is, there may be movements along the demand curve, but the
curve itself remains fixed.
2-40
-------
Lastly, it was suggested that indirect-market approaches may not
be able to identify certain types of air quality benefits. These are
believed to include benefits generally referred to as "option" values
and "bequest" values; certain aesthetic benefits; and climate or
ecosystem effects which do not ultimately impact economic markets.
ORGANIZATION OF THE STUDY AND REPORT
The previous sections have reviewed the general theory underlying
benefits analysis in general and the indirect-market approach in
particular. In applying these broad concepts to a practical problem,
many specific decisions are required concerning scope, methods and
validation. The purpose of this section is to summarize the decisions
that were made, and in so doing, to provide an overview of the
remainder of the report.
Organizing Framework
The overall logic of the study can be summarized as shown in
Figure 2-9. At the top of the figure, the overall scope of the study,
the non-health effects of TSP/S02/ is identified. At the second level
in the figure, the alternative methods for estimating non-health
benefits are listed. As indicated previously, the basic approach in
this study can generally be viewed as a combination of indirect market
2-41
-------
^^^fmm
MHMM
•^•^M
^^^••B
^ l*>
0) -1
B ci
°i
-- tn
2 R_
0 -*
U ? y;
y '11 m
g 'J) -«r
O .a ~
II ;H
t|3S
1 § J
r-
51
u -»
(0 r*l
I
S
o< *.> —
s " 2 -.
0! -a
3 >. OJ —
JJ U — «a
3 en oj •
u O -^
fl
h.
T3
3
4J
03
O
M-4
1-1
0
(1)
(0
M-l
C
•H
N
•H
C
I
CN
0)
1-1
3
Oi
•H
2-42
-------
approaches.* Given this approach, the next step is to identify the
various markets, or economic sectors, that might be affected. Recall
that within a partial equilibrium framework, we consider final demand
effects and supply side effects separately. The third level in the
figure indicates the separation, and the fourth level lists the
various economic sectors associated with each side. The percent
figures included in the boxes indicate the percent of final demand
accounted for by each demand sector and the percent of gross national
product (GNP) accounted for by each producing sector (16).
Selection of Sectors
Given the sectors identified in the figure, the next step in the
study involved assessing the availability of data and models for
analyzing effects in each sector. As will be discussed, we did not
conduct analyses for all of the sectors. Some sectors were excluded
because available data were limited (e.g., air quality data in timber-
growing regions). Others were excluded because the nature of the
industry made it difficult to define the location of the industry with
respect to air quality (e.g., the transportation industries). The
specific conclusions concerning each sector are described below.
* The analysis for the agricultural sector (Section 9) is a hybrid of
a damage function and a market model.
2-43
-------
The Household Sector (63.5%)—
In the household sector, it was determined that available data
would allow development of a model incorporating the approach
described previously. Recall that in this approach, the basic idea is
that air pollution affects the implicit price of household activities
such as "household operations". Higher levels of air pollution induce
greater use of certain market goods, such as detergents, and thus lead
to higher costs in carrying out household operations. Benefits are
generated when reductions in air pollution reduce these implicit
prices. Data were considered to be less adequate for consideration of
demand shifting effects involving activities such as recreation.
In addition to the basic household sector analysis described
above, two supplementary analyses were undertaken for this sector.
One of the analyses examines property value differentials and the
other examines wage rate differentials. The purpose of the two
supplementary analyses is to provide a cross-check on the results from
the basic household sector analysis. The basic analysis is reported
in Section 4. The property value and wage analyses are reported in
Sections 5 and 6 of the report, respectively.
The Government Sector (20.5%)—
The government sector on the final demand side includes the
purchase of goods and services by all levels of government. To a
lesser extent, the government sector also shows up on the producing
2-44
-------
side and accounts for part of the "other" category in the previous
figure.
During the course of the study, it was judged that an analysis
for county and municipal governments might be feasible using the same
basic approach as in the household sector analysis. It was decided,
however, to postpone this analysis until the household sector model
had been completed and the properties of that model were more well
defined. As of this writing, the government sector analysis has not
been initiated. Analysis of the Federal sector was considered to be
somewhat more problematical. The additional difficulty in this case
comes in identifying the location of Federal activities with respect
to air quality conditions. No analysis of the Federal sector was
therefore undertaken.
Agriculture, Forestry and Fisheries (3.1%)—
The constraining factor in conducting analyses of these sectors
is the lack of air quality data in many rural areas. In particular,
S02, which is the more important pollutant (compared to TSP) in terms
of vegetation effects, is monitored in only about 10 percent of the
counties in the U.S. This basically ruled out analysis of forestry
and fisheries. For similar reasons, it was decided to limit the
analysis of agriculture to a selected number of crops which met two
criteria: (1) an economically significant amount of production occurs
within or near metropolitan areas (and is thus more likely to be in
areas where air quality is monitored); and (2) S02 is believed to have
2-45
-------
damaging effects on that crop. Two crops which met these criteria
were cotton and soybeans. The analysis for this sector thus focuses
on these two crops.* The results are reported in Section 9 of this
report.
Mining and Construction (7.1%)—
No analysis was undertaken for these sectors. In the case of
mining, one serious constraint is that a considerable amount of mining
occurs in remote areas where air quality data are limited. With
construction, the problem is one of location variability. Except in
the very largest construction projects, construction equipment is
likely to be moved from one job site to another on a short-term basis
so that matching of construction activity with air quality data would
be problematical.
Manufacturing (23.9%)—
As in the case of the household sector, fairly extensive analyses
of the economic characteristics of manufacturing industries have taken
place and considerable manufacturing activity occurs in metropolitan
areas where air quality monitoring is done. In view of these points
and in view of the economic importance of the sector, it was decided
that an analysis of this sector should be undertaken.
* Cotton and soybeans account for about 15 percent of the "value
added" in the agriculture, forestry and fisheries industries. Thus,
even though only two crops are considered in the study, they
represent a significant fraction of the economic activity in these
sectors. Value added is defined to be the value of production less
the cost of raw materials.
2-46
-------
Two difficult problems that were viewed as confronting the
analysis were as follows: (1) The manufacturing sector is comprised
of many hundreds of industries with widely differing processes.
Available data sources generally provide either industry detail or
geographic detail but are more limited when both kinds of detail are
required. (2) Economic detail available is in relatively aggregate
categories such as "labor" and "materials", with no breakdown into
maintenance, operation, etc.
In view of the data constraints present in this sector, it was
decided to limit the analysis to a few selected industries for which
the available data were most complete. The basic analytical approach
employed is the one described previously where an effort is made to
identify whether there is a relationship between production costs and
air quality not explainable by other factors. The results of this
analysis are reported in Section 7 of the report.
Transportation, Communication and Utilities (9.0%) —
Analysis of air pollution effects on the transportation sector is
difficult. This is because so much of the potentially affected
material is in the form of rolling stock (e.g., airplanes and trucks)
so that a matching of economic data to air quality data is generally
not feasible. No analysis of this sector was therefore attempted.
Within the communications sector, the most economically
significant industry is the telephone industry. As a regulated
2-47
-------
industry, considerable data are available concerning telephone
industry financial performance, physical plant, and output. A
significant effort was therefore made to determine whether such data
could be obtained for substate areas (e.g., exchanges) so that a
matching with local air quality data would be possible. Contact with
several individuals in the industry and the applicable regulatory
agency indicated that data were not available at the geographic level
required. No analysis of the communications sector was therefore
undertaken.
Within the utilities sector, the largest industry is the electric
utility industry. As a regulated industry, it is also well documented
in terms of financial and physical characteristics. In particular,
detailed cost and output data are available for individual generating
plants so that a matching with air quality data is feasible. An
analysis was therefore undertaken concerning the generation phase of
the industry (transmission and distribution equipment are
geographically dispersed so that an analysis of these phases was
judged less feasible). The analytical approach was similar to that
used for the manufacturing sector. In this case, however, available
data made it possible to analyze the effect of air quality on
maintenance cost as well as on total production costs. The results of
this analysis are reported in Section 8. Analysis of the gas and
sanitary utilities was not undertaken.
2-48
-------
Commercial and Services (43.6%)—
The commercial and services sector is a mixture of industries
including wholesale and retail trade; finance, insurance and real
estate; and a wide variety of other service industries. Two
approaches were initially considered for this sector, based on the
assumption that soiling and aesthetic effects were dominant. One
approach was to utilize data sources on maintenance activities by
commercial cleaning companies to estimate maintenance costs for
commercial buildings. The other was to look at commercial building
rent or property value differentials in areas with differing
concentrations of pollution. The latter approach is based on the same
hypothesis as the residential property value technique in the
household sector. Data sets along these lines were identified for
certain areas of the country but were not fully adequate for
developing national benefits estimates. No analyses were therefore
undertaken in this sector.
Coverage of Sectors
The decisions concerning the basic scope of the study, as
described in the previous section, are summarized in Table 2-1. The
first and second columns in the table identify the sectors and the
percent of economic activity accounted for by each sector. The third
column indicates the percent of each sector covered by the basic
analysis in each sector. For example, the basic analysis in the
household sector covered 24 major metropolitan areas and a subset of
2-49
-------
TABLE 2-1. COVERAGE OF ECONOMIC ACTIVITY IN EACH SECTOR
t inaj. ueiuana secuor
Households*
Government
Other
Totals
Producing sector
Agriculture, forestry
' and fisheries
Mining and
construction
Manufacturing
Transportation ,
ircrCSUL OI
final demand
63.5
20.5
JLS^O
100.0
Percent of
GNP
3.1
7.1
23.9
9.0
Percent
Basic
analysis
17
0
_0
11**
Percent
Basic
analysis
2-15
0
4-8
8-11
coverage
Basic plus
extrapolation
45-55
0
0
29-35**
coverage
Basic plus
extrapolation
2-15
0
25-30
15-20
communication and
utilities
Commercial and
services
Government and other
Totals
43.6
13.3
100.0
* Goods and services consumed by individuals and certain nonprofit
institutions. Includes rental of dwellings but not purchases of
dwellings. The latter are included with "other".
** Weighted average coverage.
Source: Estimates of final demand and GNP shares are from U.S.
Department of Commerce, Bureau of Economic Analysis. Survey
of Current Business. July 1979. Tables 1.1 and 6.1.
2-50
-------
total consumption expenditures. This represented 17 percent of total
activity in the sector. The agricultural sector analysis included two
major crops representing 15 percent of activity in that sector. The
manufacturing sector analysis covered six industries, accounting for 8
percent of that sector. The analysis of electric utilities focused on
the generation phase which accounts for 11 percent of the
transportation, communication and utilities sector. The table also
provides subtotals of the coverage for the final demand sectors and
the producing sectors.
In order to broaden the scope of the analysis, we made limited
extrapolations of the results of the basic analysis. For example, the
household sector analysis was extrapolated from the original 24 SMSAs
to other areas of the country. No extrapolation was attempted in the
agriculture sector. Results for the manufacturing sector were
extrapolated to closely related industries. And data in the open
literature were used to extend the electric utility sector results to
include the transmission and distribution phases. The details of the
extrapolation procedures are discussed in Section 10 of the report.
The other pertinent report sections are also identified in the table.
It is important to note that this study does not provide complete
coverage of all possible sectors. Nor does it include consideration
of effects such as impacts on the ecosystem. In this respect, the
benefits reported in the study are conservative estimates of the
benefits of the secondary ambient air quality standards.
2-51
-------
Validation of Results
A study of this type can have potentially important policy
implications. Consequently, considerable effort was made to identify
and incorporate methods for assessing the validity of the analytical
results. The basic philosophy adopted was to build an information
gathering process into the model development effort. This was done by
identifying numerous checkpoints in the analyses where important
information would be available. This information could then be used
both to assess the plausibility of the results at that stage and to
guide model development at the next stage. Although there were some
variations in the validation procedures used in each sector, the
validation procedures generally included in each sector analysis were
of the types summarized below.
Statistical Tests—
All of the sector models incorporate mathematical equations whose
coefficients have been statistically estimated. A key advantage of
this approach is that it allows formal tests of the structure and
contents of the models. In. particular, standard tests can be used to
assess: the importance of particular variables, the importance of
interactions between variables, the overall explanatory power of the
models, and the error properties of the models. These tests can thus
be used both to guide model development as well as to assess the
analytical results.
2-52
-------
Economic Tests—
All of the sector analyses are based on economic models. The
sector models are generically similar to ones developed by other
researchers for a variety of other purposes. Consequently, the basic
economic properties of the sector models can be examined to determine
whether they are consistent with economic theory and with other
results in the open literature. Clearly, if the sector models do not
measure up in terms of their basic economic properties, then one would
have little confidence in their ability to measure air pollution
effects. Depending on the sector, the economic properties which
provide a basis for assessment and comparison include such properties
as the magnitudes and signs of the price elasticities of demand, and
the magnitudes and signs of the elasticities of substitution.*
Sensitivity Tests—
In any model development effort, an important consideration is
the possible sensitivity of the results to the methods, data and
assumptions used. In an effort to judge the robustness of the
estimated sector models, a variety of sensitivity tests were
undertaken. Depending on the particular sector, these tests included:
varying the functional form of the equations, varying the ways in
which pollution variables enter the equations, using alternative
measures of pollution, and re-estimating the equations over different
subsamples of the data.
* Definitions of these parameters are provided in the individual
sector reports.
2-53
-------
Plausibility Checks—
Each of the sector models is designed to predict the economic
effect of a change in air pollution. It is thus possible to compare
the economic effect of air pollution, implied by the models, with
other economic characteristics of the sectors. Clearly, for example,
one would expect the economic effect on a household to be much smaller
than household income. Similarly, the effect, on an industry or farm
should be much smaller than their other economic costs.
One can also ask whether there is existing physical evidence to
support or reject the finding of an economic effect from air
pollution. This can range from asking firms whether they have a
problem with atmospheric corrosion, to identifying whether the implied
alteration in household expenditure patterns is consistent with
physical effects identified in the open literature.
Comparison of Benefits Estimates—
There have been a variety of previous studies which have
developed estimates of the benefits of air quality improvements.
Unfortunately, there is wide variation among these studies in the
methods used, the time periods covered, the measures of pollution, and
the assumed changes in air quality. As a result, we found it
difficult to compare the aggregate benefits we estimated in each of
the sectors with estimates from the earlier studies. Comparisons were
particularly difficult with the manufacturing and electric utility
sectors, and to a lesser extent in the other sectors.
2-54
-------
In the household sector analysis, it was possible to develop
gross comparisons of estimated benefits in a different way. As
indicated previously, the basic model for the household sector
analyzes household expenditure variations to develop benefits
estimates. Independent estimates for the household sector were also
developed as part of this study, based on the residential property
value and labor wage rate techniques described previously. These
three analyses are reported in Sections 4, 5, and 6 of this report.
2-55
-------
REFERENCES
1. Mishan, E. J. Cost-Benefit Analysis (2nd Edition). Praeger, New
York, 1976.
2. Baumol, William J. and Wallace S. Oat.es. The Theory of
Environmental Policy. Prentice-Hall, Englewood Cliffs, New
Jersey, 1975.
3. Crocker, Thomas D., et al. Methods Development for Assessing
Tradeoffs in Environmental Management (4 Vols.). Final report
under EPA Grant R805059010. Laramie, Wyoming, 1978.
4. Freeman, A. Myrick, III. The Benefits of Environmental
Improvement: Theory and Practice. Johns Hopkins University
Press, Baltimore, Maryland, 1979.
5. Maler, Karl-Go'ran. Environmental Economics: A Theoretical
Inquiry. Johns Hopkins University Press, Baltimore, Maryland,
1974.
6. Smith, V. Kerry. The Economic Consequences of Air Pollution.
Ballinger Publishing Company, Cambridge, Massachusetts, 1976.
7. Fink, F. W., F. H. Buttner and W. K. Boyd. Technical-Economic
Evaluation of Air Pollution Corrosion Costs on Metals in the U.S.
(NTIS: PB 198 453). Battelle Memorial Institute, Columbus,
Ohio, 1971.
8. Salmon, R. L. Systems Analysis of the Effects of Air Pollution
on Materials. Midwest Research Institute, Kansas City, Missouri,
1970.
9. Ryan, John W., _et. al. An Estimate of the Nonhealth Benefits of
Meeting the Secondary National Ambient Air Quality Standards.
Final report prepared for the National Commission on Air Quality,
SRI International, Menlo Park, California, 1981.
10. Willig, Robert D. Consumers' Surplus Without Apology. American
Economic Review, 66:589-597, 1976.
11. Randall, Alan and John R. Stoll. Consumer's Surplus in Commodity
Space. American Economic Review, 70:449-455, 1980.
12. Schmalensee, Richard. Another Look at the Social Valuation of
Input Price Changes. American Economic Review, 66:239-243, 1976.
2-56
-------
13. Anderson, Robert J., Jr., et al. Quantifying the Benefits to the
National Economy from Secondary Applications of NASA Technology.
NASA Contract NASW-2734, Mathematica, Inc., Princeton, New
Jersey, 1975. The extension involved the assumption of linear
demand functions in each final good market.
14. Just, Richard E. and Darrell L. Hueth. Welfare Measures in a
Multimarket Framework. American Economic Review, 69:947-954,
December 1979.
15. See, for example, Freeman, op. cit. pp. 72-75.
16. U.S. Department of Commerce, Bureau of Economic Analysis. Survey
of Current Business. Vol. 59, July 1979. Tables 1.1 and 6.1.
2-57
-------
SECTION 3
AIR QUALITY AND METEOROLOGICAL DATA
-------
SECTION 3
AIR QUALITY AND METEOROLOGICAL DATA
INTRODUCTION
This section describes the procedures that were used in
constructing the air quality/meteorological data base. Air quality
data were supplied by the U.S. Environmental Protection Agency from
the SAROAD data file maintained by the Agency. Meteorological data
were collected from various issues of the U.S. National Oceanic and
Atmospheric Administration's Local Climatological Data series.
The section is divided into two major parts. Initially, we
discuss the air quality data used in the analysis. The presentation
describes specific attributes of the air quality data, the temporal
and spatial scope of the data, and various transformations which were
required in order to bring the data into a useable form. Following
our discussion of the air quality data, we outline the meteorological
data available for the study.
AIR QUALITY DATA
Air quality data were obtained on ambient concentrations of
sulfur dioxide (S02) and total suspended particulates (TSP) on an
annual basis for the years 1972-78, inclusive, while quarterly data
3-1
-------
were supplied for SC>2 / TSP, total oxidants, and ozone over the same
time span. A variety of pollutant measurement techniques, averaging
times, and pollutant measurement indices (e.g., arithmetic mean,
geometric mean) were obtained. Site identification codes as well as
county, SMSA, and state codes were included with each data record.*
This diversity of location identifiers provided flexibility with
respect to defining the spatial unit of observation.
The spatial coverage of the air quality data was, of course,
limited by the placement of monitoring stations. Furthermore, only
those monitoring sites meeting summary criteria established by EPA
were included in the data file. The summary criteria represent
minimum data collection requirements for individual sites, and
statistics such as the annual arithmetic mean are reported for only
those sites satisfying these criteria. The summary criteria are
defined in Reference (1) as follows:
Criteria for continuous observations with sampling
intervals less than 24 hours are:
— Data representing quarterly (annual) periods must
reflect a minimum of 75 percent of the total
number of possible observations for the
applicable quarter (year).
* It should be noted that SMSA definitions are revised periodically,
and thus the SMSA codes reported in the SAROAD data base may vary
depending on when the codes were assigned. If the geographic unit
of observation is to be an SMSA, we would advise others using the
SAROAD data to aggregate to the SMSA level from the counties that
make up the SMSA.
3-2
-------
Criteria for noncontinuous observations with sampling
intervals of 24 hours or greater are:
— Data representing quarterly periods must reflect
a minimum of five observations for the applicable
quarter. Should there be no measurements in one
of the three months of the quarter, each
remaining month must have no less than two
observations reported for the applicable period.
— Data representing annual periods must reflect
four quarters of observation that have satisfied
the quarterly criteria.
With these constraints, the geographic range of sites was limited.
For example, in 1978 there were 3,042 counties in the United
States (2), yet valid TSP data were available for only 1,000 counties
and SO2 data were limited to 182 counties. The coverage is even less
when other pollutants such as oxidants are considered.*
These restrictions on air quality data availability have
implications for both the estimation phase and the benefits
calculation phase of the analysis. In the estimation phase, economic
data may be available for a location for which there is no air quality
data. In this case, potential observations would be lost unless the
"missing data" problem can be overcome. Further discussion of ways in
which the "missing data" problem was handled is presented within each
of the sector analyses.
* There does exist an ancillary air quality data set of "design
values" which provides more extensive geographical coverage of
counties. However, certain assumptions used in the construction of
the design values were felt not to be appropriate for the
statistical estimation phase of this study.
3-3
-------
With respect to the benefits calculation phase, the absence of a
complete set of air quality data implies that only a partial estimate
of national benefits can be derived. However, this would not be true
if data were available for all sites which exceed the secondary
standards. Although we cannot confirm that this is the case, it seems
reasonable to assume that monitoring sites would be located in those
areas that were believed to have high concentrations of various
pollutants. In this case, reasonable coverage of the geographic
dispersion of national benefits would be expected.
Air quality concentrations vary not only across locations, but
also across time. For example, analysts of air pollution data have
noted that there are diurnal cycles, weekend/weekday cycles, seasonal
patterns, and year-to-year trends. In an assessment of air pollution
effects, these time-dependent variations must be taken into account.
Thus, in the analysis of agricultural benefits, quarterly data were
used in order to better characterize the exposures faced by crops
during their growing period. In the other sectors, where the primary
concern was with soiling and materials damage estimates, annual data
were employed. This choice appeared reasonable given the exposure
durations typically associated with these effects, but it was also
conditioned by the fact that the available economic data in these
other sectors were reported on an annual basis.
In addition to the cyclic variations in observed pollution
patterns, one must also be concerned with the averaging times used to
3-4
-------
present summary information of air quality levels. The averaging time
represents the period over which air quality concentrations are
observed before an index is developed. The shorter the averaging
time, the more variation there will be in the data for a fixed period
of observation. For example, in the course of one year, a graph of
one-hour averaging time concentrations could have as many as 8,760
observations, while a graph of 24-hour averaging time concentrations
would record at most 365 data points (which are averages of the 24
one-hour observations for each day).
Averaging times become important in characterizing the possible
damaging effects of pollution. For example, it is believed that
damage to vegetation can occur with a single exposure to high
concentrations of, say, SCU. This implies that one would want a
measure of maximum concentrations occurring for fairly short averaging
times. Conversely, metal corrosion occurs over a long time period of
continued exposure so that an average based on longer averaging times
would be more appropriate. These considerations have been taken into
account by the developers of measurement instrumentation, so that the
form of the data when it is collected is typically recorded in an
averaging time that is reasonable for analyses involving the
particular pollutant. Because of this, we have made no adjustments to
obtain averaging times that do not appear in the SAROAD data file.
In the analysis phase of the study, air pollution data were
merged with economic data of the same year. In particular, no lag
3-5
-------
structure was introduced into the analysis. This implies that current
air quality levels serve as a proxy for the historical levels of air
quality in an area. Specific details on how the air quality data were
used in the analysis and benefits calculations appear in each of the
sector discussions.
There are several other general attributes of the air quality
data which are important for defining the scope of the analysis.
First, the data obtained are for recorded ambient concentrations of
air pollutants. In particular, the data are generated from discrete
receptors and not derived from an analysis of emission dispersion
characteristics. Thus, the implied exposure patterns imperfectly
reflect exposure experiences of individual economic units. However,
since our economic data typically represent only the behavior of the
average (or representative) economic unit for a geographical area
(e.g., a county), aggregation of site-specific air quality data across
an equivalently defined region would permit the definition of the
corresponding representative exposure levels for that area. More
detailed discussion of the spatial aggregation procedures used in this
study is provided in the reviews of the specific pollutants.
Finally, no distinction is made between ambient concentrations
and indoor pollution levels (self-generated or otherwise). The
ambient concentrations are viewed as proxies for the actual exposure
concentrations and any damaging effects which may accompany a
particular level of exposure.
3-6
-------
In the remaining parts of this discussion of the air quality
data, the specific attributes of the various pollutants included in
this study are examined more closely. Particular emphasis is placed
on reviewing various data transformations which were required in order
to bring the data into a useable form.
Total Suspended Particulates (TSP)
The reference measurement technique for TSP is the Hi-Volume
Gravimetric, 24-hour averaging time method, with concentration levels
reported in /ig/m . This methodology is the only method observed on
the SAROAD data tape. Because of this, no additional effort was
required to ensure that data compatibility was maintained.
Spatial Aggregation—
In the estimation phase of the analysis, the major effort of data
manipulation involved spatial aggregation procedures. Since the unit
of analysis in our economic data is a geographic area defined by
political boundaries (e.g., a county), exposure patterns can vary
substantially within a given area. This implies that a single-number
pollutant index for the area must be viewed as only an approximation
to the actual exposures experienced by individual economic units.
Freeman (3) provides a concise description of the biases that may
arise when such spatial aggregation procedures are employed. While we
recognize the possibility of bias in a single-number index, the
3-7
-------
aggregate nature of available economic data makes more sophisticated,
multiple index procedures such as isopleth mapping impractical.
The aggregation procedures used were as follows. For each site
within a county, data for the annual geometric mean and the second
highest annual reading were obtained. For both of these measures, the
arithmetic mean was calculated across all sites in the county. This
procedure yielded the county average of the annual geometric mean and
second-high readings. The second method used in defining a county
index for TSP involved finding the maximum reading from among all
sites in the county for the geometric mean and the second highest
concentrations. Thus, for each county, there were four separate
indices that could be used. For example, assume that four sites exist
in a particular county with annual geometric means of 40, 70, 50, and
40 and 24-hour second-high observations of 200, 380, 400, and 240.
Then the four indices would have the values:
Average annual geometric mean equals 50
• Maximum annual geometric mean equals 70
• Average second high equals 305
Maximum second high equals 400
Note that when an SMSA was the unit of observation, similar aggrega-
tion procedures were used. In particular, the county average and
maximum indices were aggregated to obtain the SMSA average and maximum
geometric mean and second-high indices.
3-8
-------
Air Quality Standards—
The use of the geometric mean and the second highest observed
concentration level was conditioned by the units of the air quality
standards. Table 3-1 lists the Federal primary and secondary national
ambient air quality standards that are currently in force for TSP.
Note that the annual secondary standard of 60 pg/m is only a guide
for assessing State Implementation Plan achievement of the legislated
24-hour secondary standard of 150 jug/m . It is important for our
analysis that both a primary and a secondary standard are defined,
since the benefits estimated are those that would accrue with a change
from the primary to a secondary standard. While this causes few
problems with TSP, complications do arise when SC>2 is considered.
TABLE 3-1. NATIONAL AMBIENT AIR QUALITY STANDARDS FOR PARTICULATE
MATTER
Averaging Primary Secondary
time standards standards
Annual 75 Mg/m3 60
(geometric mean)
24-hour** 260 ^g/m3 150
* To be used as a guide for assessing State Implementation Plan
achievement of the 24-hour secondary standard.
** Not to be exceeded more than one per year.
Source: Air Quality Data — 1977 Annual Statistics.
3-9
-------
Temporal Aggregation—
As noted earlier, air quality varies not only spatially but also
temporally. Exposures can vary dramatically on a fairly short-term
basis, so that a summary measure of the time path profile of exposure
patterns can be sensitive to the averaging time used to characterize
the temporal aspects of pollutant concentrations. The standards for
TSP have been set on the basis of a 24-hour filter sample, which is
appropriate for identifying longer-term effects of TSP (e.g., weekly,
monthly, or annual cycles). Since TSP is primarily associated with
soiling and materials damage effects, the longer averaging times
appear to be reasonable characterizations of the relevant time
profile. We have made no changes in the reported TSP data to account
for other averaging times.
Benefits Calculations—
TSP data are also used in the calculation of benefits. As was
described in Section 2, benefits can be identified by examining
changes in cost or demand functions before arid after a specified
change in air quality. For the most part, this process is
straightforward. One simply evaluates and compares the economic
functions at the different levels of air quality, where the units
(e.g., averaging time) of the air quality measures are consistent with
those used in the estimation phase of the analysis. Thus, if a demand
function is estimated which depends on the 24-hour second-high
concentration of TSP, benefits would be calculated for the achievement
of the 24-hour secondary standard of 150 jug/m . Ostensibly, the
3-10
-------
annual primary and secondary TSP standards have no place in the
calculation of benefits. However, this is not necessarily the case.
Consider the following example.
Assume that TSP concentrations in a particular county are at
levels that violate either of the two primary standards. Since the
benefits estimates derived in this analysis are conditional on the
attainment of the primary standard, TSP levels in the county must be
reduced to the primary standard prior to the calculation of benefits.
But which of the primary standards is to be chosen? If the analysis
leading to the evaluation of benefits is done in terms of the second-
high concentrations, then the 260 ^g/m standard would appear to be
appropriate. However, this choice neglects the fact that the annual
primary standard may be more stringent so that additional improvements
in TSP levels must be made if both standards are to be met. If the
annual standard is more stringent, then it becomes the "controlling
standard" since it imposes the binding constraint on acceptable
ambient concentrations of TSP. This means that when the annual
primary standard is met, the expected 24-hour second-high ambient
concentration levels would be less than the primary standard levels of
260 /ug/m . Consequently, even though the analysis is in terms of
second-high concentrations, evaluation of benefits from the primary
standard of 260 ^g/m to the secondary standard of 150 Mg/m would
lead to an overestimate of benefits.
3-11
-------
In going from the example posed above to real data, several
questions arise. Is the annual standard ever the controlling
standard? If it is, how can one identify the expected second-high
concentrations that would be consistent with attainment of the annual
primary standard? For both questions, the answer requires the ability
to express concentration levels in different averaging times.
Formulas for the desired transformations have been developed by
Larsen (4) under the assumption that pollutant concentrations are
distributed log normally. The use of the formulas requires
information on concentration levels and the standard geometric
deviation (SGD). The SGD is defined by Larsen as:
SGD = exp
I(In C - In M )2
n
1/2
(3.1)
where C is the level of concentrations
M is the geometric mean
n is the number of observations
In is the natural log (base e) function
exp is the exponential function.
Table 12 in Larsen (4) indicates that when the SGD is less than
1.53, the annual primary standard for TSP will be the controlling
standard. A review of our data reveals that this is the case in
approximately 20 percent of site observations. However, it should be
noted that the reported SGD is for current levels of TSP, while the
3-12
-------
SGD that should be used is that which would occur after the less
•
stringent standard has been achieved. Unfortunately, this measure is
not observed, and it is difficult to judge whether the observed SGD is
likely to be greater, less than, or equal to the SGD that would occur
if the second-high standard is attained. Because of this, any
computation of the expected second-high concentrations must assume an
explicit value for the SGD.
The actual computation of the expected second-high concentration
level, given that the "controlling" annual primary standard is
achieved, can be derived from Equations (13) and (72) in Larsen (4),
and presented here as Equations (3.2) and (3.3).
In M = In M+ 0.5 In2 SGD (3.2)
M = C(SGD)<°-5 ln SGD> ~ Z (3.3)
where M is the arithmetic mean
Z is the number of standard deviations between a particular
frequency and the median, and the other terms are as
previously defined.
With the variables in Equations (3.2) and (3.3) being site-
specific, each of the several sites within a county or SMSA may have a
different starting point for benefits calculations. Since our
benefits estimates are derived at the county or SMSA level, there is
some question as to the concentration level at which to begin the
3-13
-------
benefits evaluation. Several options are available. An average of
•
the "effective" 24-hour second-high standards for each site could be
evaluated, minimums or maximums from among the sites could be used, or
one could simply adopt the alternative secondary standard value of 260
/xg/m . Ultimately, it was decided to use the alternative second-high
standard in this study. As noted earlier, this leads to a slight
overestimate of benefits. The principle reason for adopting this
assumption was the realization that our estimates of exposure patterns
were approximations, and that the refinements suggested above would
likely give a false sense of precision to the numbers being used.
However, it was felt to be important to point out this particular
problem since we were not aware that other benefits analysts had
explicitly considered, it.
Summary of TSP Data—
This completes our review of the TSP data available for the
study. Overall, it is our opinion that the data are acceptable.
Spatial availability is widespread and the fact that one pollutant
methodology is (and has been) dominant makes working with the data
straightforward. This is in contrast to our experience with S02/
where several challenging problems had to be overcome before the data
were acceptable for the benefits analysis.
3-14
-------
Sulfur Dioxide (S02)
The construction of a data file for SC>2 was complicated by
several factors. These included:
The spatial coverage of S02 monitoring stations was
limited.
Different methodologies were used to measure S02
levels in the 1970's.
Current air quality standards do not include both
primary and secondary standards of the same averaging
time.
Before we discuss these specific problem areas, we will briefly review
the general characteristics of the available data and the S02 air
quality standards used in the calculation of benefits.
Air Quality Standards—
The regulatory standards for SO2 have changed in the past decade.
The current Federal primary and secondary standards are listed in
Table 3-2. One feature of the table that is important for our
analysis is that the averaging times differ for the primary standards
and the secondary standards. Since our benefit evaluations involve
movement between the two standards, some method must be devised to
find "equivalent" standards in different averaging times. For
example, we would want to find out what the expected second-high 24-
hour averaging time concentration would be given that the second-high
3-hour averaging time standard of 1,300 ^g/m was just met. The
3-15
-------
TABLE 3-2. NATIONAL AMBIENT AIR QUALITY STANDARD'S FOR SULFUR DIOXIDE
A'veraging time Primary standards Secondary standards
Annual 80
(arithmetic mean)
24-hour* 365 /*g/m3
3-hour* — 1,300
* Not to be exceeded more than once per year.
Source: Air Quality Data — 1977 Annual Statistics.
expected 24-hour second-high value could then serve as a pseudo-
standard. This topic will be discussed more fully below.
The fact that the only SO2 secondary standard that is currently
part of Federal regulation is a 3-hour averaging time standard
concerned us. Discussions with EPA personnel confirmed what we
suspected. The 3-hour standard was set on the basis of observed
vegetation damage. Because most of our sector analyses deal primarily
with soiling and materials damage effects, we believed that it was
more appropriate to look at alternative standards that are expressed
in longer averaging times. In fact, in the early 1970's, such
standards did exist. This expanded list of S02 standards is presented
in Table 3-3. In our sector analyses, benefits estimates are provided
for both the attainment of the "equivalent" pseudo-standard and the
alternative secondary standards shown in Table 3-3.
3-16
-------
TABLE 3-3. AN ALTERNATIVE SET OF AMBIENT AIR QUALITY STANDARDS FOR
SULFUR DIOXIDE
Averaging time Primary standards Secondary standards
Annual
(arithmetic mean)
24-hour*
3-hour*
80 Mg/m3
365 Mg/m3
—
60 MC
260 /u
1,300 MC
3/m3
3/m3
3/m3
* Not to be exceeded more than once per year.
Spatial and Temporal Considerations—
The various procedures described for the spatial aggregation of
TSP data were followed with SO-. In particular, for each county or
SMSA, four indices of S02 concentrations were developed. These
included the average and maximum arithmetic means, and the average and
maximum second-high concentration levels. Note that the units of
these indices are consistent with the units of the S02 primary and
secondary standards. As with TSP, the county averages and maximum
values were obtained by looking across sites, while SMSA indices were
derived from the estimated county statistics.
Although the existence of the 3-hour averaging time secondary
standard implies that average and maximum second-high values should be
calculated for this averaging time, this was not done. Each of our
3-17
-------
sector analyses relies on indices generated for the annual arithmetic
mean or the 24-hour second-high readings.*
While the general approach to developing an SO 2 data file was the
same as the used for TSP, the problems mentioned at the beginning of
this subsection required us to consider making several transformations
in the raw data.
Limited Scope of Data—
No matter which pollutant measurement methodology is considered,
there does not appear to be a single year in which there are more than
a couple of hundred counties for which valid (i.e., meet the summary
criteria) SCu statistics are available. This compares with the over
3,000 counties in the United States. Naturally,, if the counties for
which data are available represent all counties that have anything
more than background concentrations of S02, then no problem exists for
the benefits calculation part of the analysis. However, in the
analysis of the economic relationships, the absence of a large number
of counties with S02 data severely constrains the number of observa-
tions available for inclusion in the study.
One possible solution to this problem was to use a set of air
quality "design values" that have been developed. These design values
use the SAROAD data as a base, but augment the SAROAD values through
* Note that the arithmetic mean is the same for all averaging times.
3-1!
-------
county-by-county adjustments after a review of records and
consultations with local and regional air quality officials. With
this method, over 700 counties had data available in 1978, versus 182
with the unaltered SAROAD data file.
There were several problems, however, with using the design
values. In the economic analysis phase, we felt that it was important
for the air quality data to be drawn from the same year as the
behavioral data. The design values are defined as maximum values that
occur over a several-year period. This feature made them unattractive
for the economic analysis. On the other hand, characterizing county
pollution levels from data taken across several years made some sense
for the calculation of benefits. Consideration of several years might
give a better picture of what the long-term pattern of air pollution
is like in a given county. Unfortunately, because the design values
represent aggregate indices, it is not possible to recover site-
specific measures of the standard geometric deviation. Consequently,
it would not be possible to transform data of one averaging time to
another. Given these problems, we felt that it was more appropriate
to use only the SAROAD data base. A discussion of how the limited
observations affected the economic analysis is presented in each of
the sector discussions.
Different Pollutant Measurement Methodologies—
In the early 1970's, the dominant method for measuring S02
pollutant concentrations was the noncontinuous "gas bubbler"
3-19
-------
technique. As technical advances were made in instrumentation
throughout the decade, continuous methods became the principal
measurement methodology. Normally, if both methodologies provide an
accurate measure of ambient concentrations, then no compatibility
problems would exist and observations from the different methodologies
could be used together.
However, it has been shown that data derived from the gas bubbler
methodology is biased. In particular, the integrity of a sample
collected by the gas bubbler method is dependent on temperature. Up
to 50 percent of the sample can be "lost" if various steps in the
sampling process are not temperature controlled. This potential bias
does not appear to exist with the continuous monitoring methods.
One obvious solution to this problem would be to discard
concentrations generated from the gas bubbler measurement technique.
This was not practical since much of our economic data was available
only for the early 1970's. During this time period, if only
continuous data were used, the number of observations available for
analysis would fall to unacceptable levels.
Because the bias was identified with a temperature dependency,
another approach to overcoming the problem would be to develop a
relationship between concentrations recorded by the continuous methods
and concentrations recorded by the gas bubbler technique at the same
3-20
-------
site, for the same time period, and with temperature included in the
expression.
A data set was put together for 28 sites across the United States
that monitored SO with both methodologies in 1977. All sites
included in the data set met the summary criteria for reporting annual
'means.
A functional relationship between the concentrations of the two
pollutant measurement methodologies was developed by regression
analysis. Several specifications were tried, and the "best" (by a
criteria of highest R^) are reported as Equations (3.4) and (3,5).
The regression for the annual mean uses annual data, while the second-
high regression uses quarterly data.
AMCONT = 19.0034 + 1.61875 AMGAS
(5.222) (5.584)
- 0.05756 • (AMGAS
(-1.828)
TEMP)
(3.4)
SECHICONT =
42.621 + 2.0612 • SECHIGAS
(2.621) (8.875)
+ 0.033642 • (SECHIGAS
(1.994)
TEMP)
(3.5)
where AMCONT is the annual arithmetic mean measured by the
continuous method in ug/m .
3-21
-------
AMGAS is the annual arithmetic mean measured by the gas
bubbler method in ug/m .
TEMP is the average annual temperature in degrees
Celsius.
SECHICONT is the second highest reading recorded by the
continuous monitors.
SECHIGAS is the second highest reading recorded by the gas
bubbler method.
In Equation (3.4), the R2 is 0.802, while the R2 is 0.497 in
Equation (3.5). The numbers in parentheses are t-statistics.
The interpretation of this relationship is that expected
continuous readings can be predicted by observing gas bubbler
concentrations and temperature. The interaction term of TEMP with
AMGAS and SECHIGAS allows the level of TEMP to influence marginal
changes in measured concentration levels. Note that the sign on the
interaction term is negative in Equation (3.4) and positive in
Equation (3.5). A priori, a positive sign is expected since the gas
bubbler data are biased downward with increasing temperature. The
negative sign in Equation (3.4) may be due to the small sample used in
the analysis. In any case, in the subsequent sections of this report,
statistically significant air pollution impacts were found only in
those instances where second high data were used. While it is
plausible that second high measures represent the proper index for air
quality data in the economic relationships, the fact that annual data
were not statistically significant may be a reflection of the poorly
specified nature of Equation (3.4).
3-22
-------
Since all of the data used in estimating Equations (3.4) and
(3.5) are from 1977, the assumption is that this same relationship
would hold in earlier years of the 1970's, and that gas bubbler data
can be "corrected" to an unbiased measure of concentrations. Data for
1977 were used because earlier years had few sites monitoring with
both methodologies.
In the economic analysis phase of the study, these equations were
used to convert gas bubbler data prior to estimation.* It should be
noted, however, that the 1978 data which served as our base year in
the benefits calculations consisted of only the unbiased instrumental
data. Thus, no conversions were required.
Averaging Times and Air Quality Standards—
The final problem to consider involves the fact that current
Federal primary and secondary standards for SO- are defined for
different averaging times. In particular, the primary standards are
stated in terms of an annual arithmetic average and a 24-hour second
high, while the secondary standard is a 3-hour second-high standard.
If we are to determine the economic benefits associated with an
improvement in air quality from a primary to a secondary standard,
then it must be possible to express both standards in an equivalent
averaging time.
* In the agricultural sector, the data used in the analysis were
quarterly and separate conversion equations were estimated in this
sector. See Section 9.
3-23
-------
There are two ways of looking at this problem. First, one could
attempt to define the 3-hour averaging time concentration level that
would be expected to occur when the 24-hour averaging time primary
standard was just met. This expected concentration level could then
serve as a pseudo-primary standard. Alternatively, one could attempt
to define the 24-hour averaging time concentration level that would be
expected to occur when the 3-hour averaging time standard was
attained.* In either case, a relationship must be developed between
concentrations of one averaging time and those of a different
averag ing time.
The same type of problem was encountered in our discussion of the
TSP data file. There, we noted that Larsen (4) has developed sets of
equations that relate parameters of different averaging times. These
relationships are established by assuming that concentration levels
are log-normally distributed for all averaging times. Use of the
equations developed by Larsen requires information on concentration
levels and the standard geometric deviation (SGD), which was defined
in our discussion of the TSP data.
The transformation we have elected to use involves finding the
24-hour averaging time concentration that would be expected to occur
given the SGD of the site and assuming that the 3-hour standard of
1,300 ug/m is just met. This expected concentration level could then
* A similar relationship could be considered between the annual
arithmetic mean and the 3-hour secondary standard.
3-24
-------
serve as a pseudo-secondary standard which could be compared directly
to the 24-hour second-high primary standard.
The relationship we use to obtain this value is a version of
Larsen's equation (71). However, instead of solving for SGD, we want
to solve for the expected 24-hour concentration level. After a
rearrangement of terms, Larsen's equation can be written as:
In C24 = 0.175 In2 SGD - 1.1615 In SGD + ln(l,300) (3.6)
where In C^, is the natural log of 24-hour averaging time
concentration levels.
SGD is the standard geometric deviation for a 24-hour
period.
For typical values of SGD, Equation (3.6) yields values of C_4
that are mostly in excess of the current 24-hour primary standard.
For example, with SGD equal to 3.0, C-4 is found to be 448 ug/m .
Since the primary standard is 365 ug/m , and we assume that the
primary standard is met, there would be no benefits associated with
attaining a pseudo-secondary standard that reflected air quality
levels consistent with attainment of the 3-hour standard. This
observation is in accord with conclusions reached by Larsen. He notes
that except in areas with strong, high sources of sulfur dioxide,
standards based solely on 24 averaging time concentrations would be
expected to control source reduction plans. This is borne out in our
data since only five counties in the United States have an expected
3-25
-------
pseudo-standard less than the current primary standard. Although the
number of counties is small, benefits estimates are reported for these
counties in the sector analyses. The benefit numbers presented in
these analyses represent our estimates of the benefits that would be
achieved through attainment of the current 3-hour standard.
There was concern on our part as to the appropriateness of the 3-
hour secondary standard for SO-. As we noted earlier, the relatively
short averaging time reflects concern for items such as vegetation
that may be damaged by high short-term concentrations. Given that
many of our sector analyses focus on soiling and materials damage
effects, a standard defined for a longer averaging time may provide a
more appropriate basis for evaluating benefits of air quality
improvements. Consequently, we have also estimated the economic
benefits that would be realized by attainment of an alternate
secondary standard of 260 ug/m . This standard is stated in terms of
a 24-hour averaging time second high and can be compared directly to
the current primary standard of 365 ug/m^. Although the concentration
level of 260 ug/m is somewhat arbitrary, this number has been used in
the past as a guide in assessing implementation plan achievement of
the secondary standard.
Summary of S02 Data—
In its raw form, the SO- data available from the SAROAD data base
have several shortcomings. In this subsection, we have outlined the
procedures that were used to overcome these obstacles. While the S02
3-26
-------
data do not approach the completeness of the TSP data, we do believe
that what is available is adequate for deriving reasonable benefits
estimates.
Other Pollutants
Although only TSP and SCU concentrations were used in the
analyses reported in this study, air quality data for total oxidants
and ozone were also obtained. Originally, our intent was to include
these pollutants in the model developed for the agricultural sector,
since evidence exists that the principle damaging agent for many crops
is ozone. Unfortunately, the geographical scope of ozone data is so
limited that available observations were reduced to unacceptable
levels. Consequently, in the final specifications of the various
sectors, no pollutants other than SCU and TSP have been included.
METEOROLOGICAL DATA
The Criteria Document summarizes several studies which have
reported an interrelationship between meteorological variables and the
damaging effects of air pollutants. For example, Schwarz (5) reports
that the corrosion rate of a metal can be expected to increase by 20
percent for each increase of one percent in relative humidity above a
critical level, while Setterstrom and Zimmerman (6) find that plant
sensitivity increases with higher levels of relative humidity. Other
climatological variables may also be important. Temperature,
3-27
-------
rainfall, wind speed, and exposure to sunlight all have been shown to
have varying degrees of influence on the extent to which ambient
concentrations of various pollutants can be expected to promote
damaging effects.
The implication of these observations is that it may not be
correct to draw inferences about damages or responses of economic
units to alternative levels of concentrations without giving
consideration to the possible interactive influence of exogenous
meteorological variables. In effect, these other factors should be
controlled for by including measures of them in the model
specifications. Even more importantly, these other factors may also
have direct effects on economic units, independent of pollution. For
example, temperature and rainfall may influence agricultural yields
directly, as well as influencing the effect of pollution variables.
To this end, data were collected from various issues of the
National Oceanic and Atmospheric Administration's Local Climatological
Data Series. Two separate data files were created. First, data were
obtained from the 1972-1974 annual summaries of Climatological data.
These summaries provide statistics for approximately 250 urban areas
in the continental United States. The data coded for the present
study included the following variables:
• Average annual temperature in degrees Celsius.
• Total annual precipitation in millimeters.
3-28
-------
• Average percent relative humidity at 7 a.m. and 1 p.m.
• Average annual wind speed in meters per second.
• An index of average sky cover.
These data are typically reported for one weather station within a
given area so that no spatial aggregation is required. In those
instances where more than one weather station in an area was included
in the annual summaries, data from only the urban reporting station
were coded. No additional transformations to these data were
undertaken.
Despite the variety of weather parameters available in the annual
summaries, there were several drawbacks to the data which necessitated
an additional data collection effort. Specifically, the annual nature
of the data made it inappropriate for the agricultural sector analysis
since crop and air quality data from the second quarter were used in
that part of the study. In addition, the geographic scope of the
annual data was limited. Typically, data were available only for
counties which included fairly large urban areas. This placed a
restriction on the number of observations available for the analysis.
In order to overcome these difficulties, additional climatological
data were collected from various yearly State volumes of weather data.
These publications include daily accounts of weather patterns for all
the weather stations across the United States. Since this is an
extensive network, spatial coverage was very good. In fact, most
counties have multiple reporting stations. However, because these
3-29
-------
data were hand-coded, only a single station was chosen for each
county, with the choice based on the centrality of the site with
respect to county boundaries.
The advantage of detailed spatial coverage was offset by the fact
that most weather stations report only temperature and precipitation.
In fact, it appears that only those areas included in the annual
summaries record parameters other than these two standard indicators
of climate. Consequently, much of our analyses were limited to
consideration of temperature and precipitation.
In the construction of this second meteorological data file, the
decision of which counties to code was guided by the availability of
other data. In particular, since the air quality data tended to be
the limiting factor, in general, temperature and precipitation data
files were put together only for those counties with valid air quality
data.
3-30
-------
REFERENCES
1. U.S. Environmental Protection Agency. Air Quality Data - 1977
Annual Statistics.
2. U.S. Department of Commerce. Statistical Abstract of the United
States, 1978.
3. Freeman, A. M. The Benefits of Environmental Improvement. Johns
Hopkins University Press, Baltimore, 1979.
4. Larsen, R. I. A Mathematical Model for Relating Air Quality
Measurements to Air Quality Standards. U.S. Environmental
Protection Agency, Office of Air Programs Publication AP-89,
November 1971.
5. Schwarz, H. Uber die Wirkung des Magetits beim Atmospherischen
Rosten und beim Unterrosten von Austrichen. Werkst, Korros,
23:648-663, 1972.
6. Setterstrom, C. and P. Zimmerman. Factors Influencing
Susceptability of Plants to Sulphur Dioxide Injury. Contrib.
Boyce Thompson Inst., 10:155-186, 1939.
3-31
-------
'•- a -
I'
------- |