EPA-450/5-31-001
Visibility Benefits
Assessment Guidebook
Benefits Analysis Program
Office of Air Quality
Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park,
North Carolina 27711
August 1981
-------
EPA-450/5-81-001
FINAL REPORT
VISIBILITY BENEFITS ASSESSMENT GUIDEBOOK
by
Robert D. Rowe, Ph.D.
Lauraine G. Chestnut
Abt/west
1410 Grant Street
Suite C-207
Denver, Colorado 80203
(303) 830-0181
August, 1981
EPA Contract Number 68-02-3528
Project Officer
Tom Walton
Economic Analysis Branch
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park., North Carolina 27711
Abt west
-------
PREFACE
This guidebook was prepared for the Office of Air Quality Planning
and Standards as part of its Economic Benefit Program. The U.S. EPA
recently promulgated visibility regulations persuant to the Clean Air Act.
These regulation do not, as of yet, explicitly mandate the use of benefit
cost analysis (BCA) in the implementation process. But, they are implicitly
required, have frequently proved appealing and may currently be required by
virtue of Executive Order 12291 for major new regulations. The guidebook
enhances the likelihood of high quality BCA in visibility related environmental
work by demonstrating how explicit measurement of aesthetic benefits can be
accurately estimated and by providing a comprehensive and critical assess-
ment of the current state-of-the-art.
This effort has greatly benefited from Myrick Freeman's book
The Benefits of Environmental Improvement; however, we have attempted to
assist a larger audience by giving a broader foundation in economics and
research methodology. Also, by focusing upon visibility benefit analysis,
we are able to give an in-depth treatment to technical non-economic aspects
of the aesthetic impacts of air pollution, a balanced examination of
alternative techniques used in air quality benefit analysis, and 10 de-
tailed case studies.
We would like to thank Mr. Tom Walton, who served as the EPA
Project Manager and provided many helpful comments. Drs. V. Kerry Smith
and Ray Palmquist provided considerable assistance from project design to
commenting upon the entire manuscript. Ron Henry of ERT, Inc. signifi-
cantly contributed through his development and writing of Chapter 3. We
would also like to thank Jan Laarman, Allen Basala, Alan Randall, Hugh
Devine, Anne Bresnock, Edna Loehman and William Schulze for technical
comments; Eric Walther and John Molenar for the sample slides; Winnie
Hinton for editorial assistance; and Claudia Collins for production support.
The analysis and evaluations expressed in this guidebook are those
of the authors and should not be interpreted as necessarily reflecting the
official policies of the U.S. Government.
Abt west
-------
TABLE OF CONTENTS
Preface
CHAPTER 1: INTRODUCTION
I 1 Objectives of the Guidebook 1-1
1. 2 The Need for Economic Benefit Analysis 1-1
1.3 Overview of Visibility Benefit Analysis 1-5
1 4 Guidebook Organization 1-7
CHAPTER 2: ECONOMIC BENEFIT MEASURES
2.1 Concepts and Measures of Benefits 2-1
2.1.1 Concepts of Benefits 2-1
2.1.2 Monetary Measures of Benefits 2-3
The Demand Curve Approach to Monetary Measures of
Benefits 2-4
2.2 Refined Consumer Surplus Monetary Measures of Benefits 2-9
2.2.1 The Utility Map Framework Used to Define Consumer
Surplus Measures 2-10
2.2.2 Why the Ordinary Consumer Surplus Measure is Incorrect 2-12
2.2.3 Four Refined Consumer Surplus Measures 2-14
Consumer Surplus Measures of Price Changes 2-16
Consumer Surplus Measures of Quantity Changes 2-17
2.2.4 Comparisons Among the Consumer Surplus Measures 2-19
Size 2-19
Reference Level of Welfare 2-20
Other Points of Comparison 2-21
Selection of the Appropriate Measure 2-22
2.3 Aggregation to Total Benefit Measures 2-22
2.3.1 Aggregation Across Individuals 2-23
2.3.2 Aggregation Across Time 2-24
-------
Page
2.4 Technical Consumer Behavior Models and Assumptions Used to
Derive Consumer Surplus Measures 2-26
A Simple Utility Map Approach 2-26
Household Production Function Approach 2-27
O 9 Q
Expenditure Function Approach i~£.?
Weak Complementarity 2-31
CHAPTER 3: INTRODUCTION TO VISION THROUGH THE ATMOSPHERE
3.1 The Importance of Atmospheric Science Fundamentals to
Visibility Benefits 3-1
Factors Affecting Visibility Impairment 3~3
3. 2 Human Perception of Visibility 3-5
Perceptual Thresholds 3-6
Content of the Scene 3-7
Psychological and Other Factors 3-7
Summary 39
3.3 Sources of Visibility Impairment 3-9
3.4 Definitions and Indices of Visibility Impairment 3-11
Definitions of Visibility Indices 3-12
Direct Measures of Human Perception 3-13
Measures of Light Intensity 3-14
Measures of Optical Properties of the Air and Airborne Particles..3-16
3.5 Visibility Monitoring and Natural Conditions Affecting Visibility... 3-17
Visibility Monitoring Technology 3-18
Visibility Monitoring Programs 3-21
Visibility and Natural Conditions 3-21
3. 6 Presentation of the Visual Environment 3-26
Review of Recent Applications 3-26
General Principles 3-27
Other Specific Problems 3-28
Other Alternative Approaches 3-28
3.7 Visibility Modeling 3-29
3.8 The Effect of Definition, Measurement, and Presentation of Air
Quality Conditions Upon Benefit Measures 3-30
-------
Page
CHAPTER 4: PRACTICAL APPROACHES TO MEASURING ECONOMIC BENEFITS
ASSOCIATED WITH CHANGES IN VISIBILITY AESTHETICS
4.1 Actual Market Versus Contingent Market Approaches 4-1
4. 2 Bidding Methods 4-3
4.2.1 Introduction to Bidding Methods 4-3
4.2.2. Theoretical Basis 4-4
Property Rights in Bidding Methods 4-4
4.2.3 Application 4-5
General Introduction and Statement of Purpose 4-6
Introductory Non-Valuation Questions 4-7
Scenario Development and Market Definition 4-7
Bidding or Valuation Questions 4-11
Evaluation of Problem Bids 4-14
Special Options, Closing Questions and Remarks 4-14
4.2.4 Strengths of the Bidding Questionnaire Approach 4-15
4.2.5 Weaknesses of the Bidding Approaches 4-16
Hypothetical Biases 4-16
Strategic Biases 4-18
Information Biases 4-20
Contingent Market Rejection and Problem Bids 4-21
Other Potential Problems and Biases 4-22
4.2.6 Survey Procedures and Post Survey Data Analysis 4-23
Survey Type 4-24
Sampling Procedures 4-25
Sample Size Determination and Verification 426
Post Survey Data Verification and Analysis 4-28
4.3 Residential Property Value Studies 4-29
4.3.1 Introduction to the Hedonic Price Technique 4-30
4.3.2 Application 4-37
The Hedonic Price Function: Dependent Variables 4-37
The Hedonic Price Function: Independent Variables 4-39
The Hedonic Price Function: Functional Form 4-42
The Hedonic Price Function: Market Segmentation 4-46
Estimation of Willingness to Pay 4-47
-------
Page
4.3.3 Variations in the Property Value Approach .................... 4-54
Matched Pairs Technique
Residential Location Model
4.3.4 Strengths and Weaknesses
When Property Value Studies Can Be Used .................... 4-56
What Property Value Studies Can Measure .................... 4-58
4.4 Introduction to Alternative Approaches For Measuring Economic
Benefits of Visibility Aesthetics ................................... 4-60
4.4.1 Ranked At tributes /Market Share ............................... ^-60
4.4.2 Travel Cost and Site Substitution Approaches ................. 4-61
4.4.3 Household Production Function Approach ....................... 4-68
4.4.4 Wage and Salary Differentials ................................ 4-68
4.4.5 Voting Approaches ............................................ 4-70
4.5 Can Benefit Measures Be Transferred Across Studies? ................. 4-72
CHAPTER 5: VISIBILITY BENEFIT ASSESSMENT GUIDELINES
5.1 Purpose of the Guidelines 5-1
5.2 Guidelines 5-1
5.2.1 Step 1 - Problem Formulation 5-1
5.2.2 Step 2 - Scenario Development 5-4
5.2.3 Step 3 - Selecting the Approach to Estimating
Economic Benefits 5-5
5.2.4 Step 4 - Application of Economic Methods 5-11
5.2.4.1 Application of Bidding Methods 5-11
5.2.4.2 Application of Hedonic Approaches 5-13
5.2.5 Step 5 - Aggregation of Benefits Across All
Affected Populations 5-15
5.2.6 Step 6 - Benefit-Cost Analysis 5-16
CHAPTER 6: CASE STUDIES OF VISIBILITY BENEFIT ANALYSIS
6.1 Case Studies Using Bidding Methods 6-1
6.1.1 The Four Corners Study 6-2
Problem Formulation and Scenario Development °~2
Application of the Bidding Method 6~4
-------
Page
Analysis of Results 6-4
Aggregate Benefits Measures 6-5
Evaluation Comments 6-6
6.1.2 The Lake Powell Study 6-6
Problem Formulation and Scenario Development 6-6
Application of the Bidding Method 6-7
Analysis of Results 6-7
Aggregate Benefits Measures 6-9
Evaluation Comments 6-9
6.1.3 The Farmington Study 6-10
Problem Formulation and Scenario Development 6-11
Application of Bidding Method 6-11
Analysis of Results 6-12
Aggregate Benefits Measures 6-15
Evaluation Comments 6-15
6.1.4 The South Coast Air Basin Study 6-17
Problem Formulation and Scenario Development 6-17
Application of the Bidding Method 6-18
Analysis of Results 6-19
Aggregate Benefit Measures 6-19
Evaluation Comments 6-21
6.1.5 The Grand Canyon/Southwest Parks Study 6-22
Problem Formulation and Scenario Development 6-22
Application of the Bidding Method 6-23
Analysis of Results 6-25
Aggregate Benefit Measures 6-27
Evaluation Comments 6-28
6.1.6 A Comparison of Visibility Related Bidding Method Studies 6-29
, 2 Hedonic Approach Case Studies 6-33
6.2.1 The Washington D.C. Study 6-35
Hedonic Price Function Estimation 6-36
WTP Function Estimation 6-36
Benefit Estimates 6-37
Evaluation Comments 6-37
-------
6.2.2 The Boston Study 6-38
Study Area Description and Data Selection 6-38
Hedonic Price Function Estimation 6-39
WTP Function Estimation 6-39
Benefit Estimates 6-40
Evaluation Comments 641
6.2.3 The Denver Study 6-41
Study Area Description and Data Selection 6-41
Hedonic Price Function Estimation 6-42
WTP Function Estimation 6-43
Benefit Estimates 6-43
Evaluation Comments 643
6.2.4 The Souch Coast Air Basin Study 6-45
Study Area Description and Data Selection 6-45
Hedonic Price Function Estimation 6-46
WTP Function Estimation 6-47
Benefit Estimates 6-47
Evaluation Comments 6-47
6.2.5 The San Francisco Bay Area Study 6-48
Study Area Description and Data Selection 6-48
Hedonic Price Function Estimation 6-49
WTP Function Estimation 6-50
Benefit Estimates 6-51
Evaluation Comments 651
6.2.6 Comparison and Review 6-52
6.3 Comparison of Property Value and Bidding Method Benefit Measures....658
6.4 Visibility Benefits in Benefit-Cost Analysis 6-60
6.5 Applicability of Benefit Estimation Techniques for Prescribed
Burning , 6-63
Glossary
Bibliography
Appendix 1 Sample Bidding Method Questionnaires and Picture Sets
Index
-------
LIST OF FIGURES
1.1 Types of Pollution Impacts for Protected Class I Areas 1-3
1.2 Inputs to the Economic Visibility Benefit Assessment Process 1-6
2.1 Demand Curve for a Typical Good X 2-5
2.2 Willingness to Pay for Visibility in the Recreation Experience 2-7
2.3 Demand for Activity Days as a function of Price and Visual Range....2-8
2.4 Indifference Map of XI, X2 2-11
2.5 Marshall's Consumer Surplus 2-13
2.6 Ordinary Consumer Surplus Using Ordinary Demand Curves 2-13
2.7 Derivation of Ordinary and Income Compensated Demand Curves 2-15
Panel a: Utility Map
Panel b: Demand Curves
2.8 EV and CV Consumer Surplus Measures 2-16
2.9 CS and ES Consumer Surplus Measures 2-18
2.10 Example of Weak Complementarity 2-32
3.1 Framework for Understanding Visibility Questions and the
Context for the Benefit Analysis 3-2
3.2 Project VIEW - EPA/NPS Southwest Visibility Monitoring Network 3-22
3.3 Cumulative Frequency of Occurrence for Standard Visual Range
derived from teleradiometer observations of Navajo Mountain
140 km distant from Bryce Canyon National Park 3-23
3.4 Standard visual range data for month of July 1979 as determined
from observations of Navajo Mountain from Bryce Canyon National
Park 3-24
3.5 Map of Existing Visual Range (from EPA Report to Congress,
page 4) 3-25
3.6 The Effect of Varying Definition and Presentations Upon Utility
Valuation 3-31
4. la Hedonic Price Function for Air Quality 4-34
4.1b Marginal Implicit Price Function and Willingness to Pay
Functions for Air Quality 4-34
4.2 An Example of the Relationship Between Property Values and
Air Pollution 4-43
4.3 A Semi-Log Functional Form 4-45
4.4a The Hedonic Price Function for Residential Property 4-48
4.4b The Marginal Implicit Price Function for Air Quality at
the Residence 4-48
4. 5 Identification of Willingness to Pay 4-52
4.6 Approximating Benefit Measures From the Marginal Implicit
Price Function 4-53
5.1 Visibility Benefit Assessment Guidelines 5-2
6.1 1978 existing and proposed coal-burning power plants in the Four
Corners region of southwestern USA 6-3
-------
LIST OF TABLES
2.1 Present Value of $1,000 Benefits Per Year 2-25
3.1 Manmade Sources of Visibility Reducing Gases and Particles and
Their Precursors 3-11
3.2 Commonly Used Indices of Visibility 3-12
3.3 Visibility Monitoring Methods 3-19
3.4 Some Natural Sources of Reduced Visibility 3-26
4.1 Correspondence Between Willingness to Pay Measures and
Consumer Surplus Measures 4-4
4.2 Appropriate Iterative Bidding Measure of Welfare Changes 4-5
4.3 Interpretation of Zero Bids 4-14
4.4 Steps in Survey Research 4-23
4.5 Property Attributes Found Significant in Air Quality Studies 4-40
5.1 Selection of Visibility Benefit Estimation Method
A Summary 5-6
5.2 Assumptions in Visibility Benefit Estimation 5-10
6.1 Aggregate Benefits for Abatement of Aesthetic Environmental
Damages Associated With the Four Corners Power Plant, 1972 65
6.2 Lake Powell Power Plant Siting Values 6-8
6.3 Farmington Study Results (Per Month, Per Family or Unrelated
Individual) 6-14
6.4 South Coast Air Basin Study Bidding Results 6-20
6.5 Bidding Method Estimates of Total Benefits for Air Quality
Improvement in the South Coast Air Basin (Approximate 30%
Improvement in Ambient Air Quality) 6-21
6.6 Grand Canyon/Southwest Parks Study Results 6-26
6.7 Aggregate Benefit Measures From the Grand Canyon/
Southwest Parks Study 6-27
6.8 A Summary Comparison of Bidding Method Visibility Studies 6-30
6. 9 National Ambient Air Quality Standards 6-35
6.10 Variables in the Hedonic Price Function 6-54
6.11 Marginal Implicit Prices and Willingness to Pay Measures
in Property Value Studies 6-56
6.12 WTP Elasticities 6-57
6.13 Costs of Controlling Visibility Impacts in Class I Areas
in the Southwest 6-61
-------
CHAPTER 1
INTRODUCTION
1.1 Objectives of the Guidebook
This guidebook presents concepts and techniques that can be used to
estimate monetary benefits for changes in visibility aesthetics resulting
from alternative levels of air pollution control. Although monetary values
are not commonly associated with visibility aesthetics, it is generally
recognized that visibility conditions affect people's moods and influence
many of their decisions. This indicates that people value visibility
aesthetics and that visibility should be considered when examining the
costs and benefits of changes in air quality.
There are several defensible methodologies that can be used to
place a monetary value on visibility aesthetics. This guidebook focuses
upon this one aspect of air quality analysis, which can be combined with
other aspects, such as health damages of air pollution and costs of emis-
sion controls to producers, to assist in policy decision making relating to
air quality management. Several best-practice techniques are discussed in
terms of their application for evaluating existing or potential visibility
impacts from single sources or for regional haze problems in recreational,
rural and/or urban settings.
This guidebook introduces these benefit estimation techniques in
recognition of the EPA's need to provide technical support to those who
must evaluate impacts related to clean air regulations and in recognition
that improved benefit measurement will lead to more accurate benefit-cost
analysis. While visibility benefit analyses are not currently mandated
by clean air regulations, this guidebook serves as a vehicle for making
the best-practice techniques available to those who may undertake these
efforts and serves as a quality control mechanism for those who may review
these studies. Finally, documentation of the assumptions, strengths, and
limitations of each method provides inexperienced users with some technical
basis for interpreting the results of the various approaches.
1.2 The Need for Economic Benefit Analysis
The value of preserving scenic and natural settings has long been
recognized in maintaining national parks, monuments and other recreational
areas. Section 169A of the Clean Air Act, added in 1977, established a
1-1
-------
national goal of remedying existing and preventing future manmade visi-
bility impairment in mandatory Class I Federal areas. Recently, 156 of
these areas were identified in which visibility is an important value.
Figure 1.1 shows these areas and the types of pollution affecting them.
According to Federal land managers, roughly one-third of these areas have
been found to have undesirable visibility conditions.
The value of air quality in urban areas has also been recognized.
National Ambient Air Quality Standards that apply to urban areas are
primarily aimed at the health aspects of air quality. Although government
regulations do not now emphasize the aesthetic impacts of air pollution in
urban areas, there is growing awareness that these impacts do occur and are
closely related to residents' perceptions of air quality conditions.
Visibility benefit analysis is not specifically mandated in any
clean air legislation, although the regulations do require consideration of
costs and benefits for proposed changes in air quality. On December 2,
1980, the EPA published final regulations to implement selected visibility
protection provisions in Section 169A and 165(d) of the Clean Air Act.
These regulations provide for a phased approach to visibility protection.
The first phase, now being implemented, deals with controlling impairment
that can be traced to a single major source or a small group of sources.
It specifically requires states with mandatory Class I Federal areas to
incorporate certain measures into their State Implementation Plans (SIPs)
that will assure reasonable progress toward the national goal of preventing
future and remedying existing impairment of visibility in mandatory Class I
Federal areas. The regulations further state that integral vistas, defined
as important views from within mandatory Class I Federal areas of scenic
landmarks or panoramas that extend beyond the boundaries of the areas, are
to be considered in visibility impact determinations and that states and
Federal land managers are to coordinate with each other through a formal
process in impact determinations. The second phase regulations, which will
deal with multiple source regional hazes and urban plumes as they affect
mandatory Class I Federal areas, will be issued at a later date subject to
research progress.
These regulations indicate that while there is a national goal of
complete protection from visibility impairment in mandatory Class I Federal
areas, there is some flexibility in determining whether any impairment
should be allowed to exist or increase. This approach is reiterated in
sections of the regulations that mandate control of existing sources,
prevention of significant deterioration from new sources (PSD), new-source
review requirements, protection of integral vistas, and state long-term
strategies. For example:
1-2
-------
BURNING. FOREST PRODUCTS
EASTERN
HAZE
PLUMES
SMELTERS
URBAN
PLUMES
- GENERALLY DESIRABLE VISIBILITY
- SOME UNDESIRABLE CONDITIONS
- FREQUENT UNDESIRABLE CONDITIONS
L.I: Types of Pollution Impacts for Protected Class I Areas.
Soun-e: Protecting Visibility, An EPA Report to Congress, 1979; page 5.
-------
Today's action requires an analysis of visibility
impacts by all new sources which might impair visibility
in a mandatory Class I Federal area irrespective of
their proposed location. However, unlike review under
the PSD provisions, the state may, for these sources,
consider costs, energy, and other relevant factors in
determining whether to permit construction of the new
source. (45 FR 80088)
The EPA (1979) report to Congress summarizes this approach as
follows:
By requiring consideration of "significant" impairment
in BART decisions, "adverse" effects of proposed new
sources and "reasonable progress" in implementating
the national goal, Congress has, in effect, mandated
that judgements be made on the value of visibility in
the context of specific decisions on control and loca-
tion requirements for sources of visibility impairing
air pollution. (Page 1.)
The importance of these clauses is that some visibility impairment
in mandatory Class I Federal areas may be tolerated if the costs of com-
pliance and other factors are considered too high. However, the magnitude
of these measures of costs and other factors are only relevant when compared
to the primary benefits of improved visibility aesthetics to be gained from
control, or lost from the lack of control. This therefore requires some
judgment regarding the magnitude of benefits in order to make appro-
priate decisions about whether to tolerate existing or additional visi-
bility impacts.
Judgments regarding the appropriate levels of environmental goods
fall within the public sector's domain, because these resources are not
privately owned and there are no privately functioning markets to assure
that values are reflected in resource allocation decisions. The govern-
ment's role in such decisions is usually considered to be one of assuring
that resources are allocated to their highest valued uses and/or distri-
buted according to some system of equity. Benefit-cost analyses offer a
pragmatic method of developing the information necessary to make these
resource allocation decisions.
A complete benefit-cost analysis weighs all measurable impacts in
a resource balancing, multi-objective planning framework through the use of
a common monetary unit of measure. For a regulation or program to be
economically efficient in terms of resource allocation, the total benefits
must exceed the total costs with the optimal resource allocation level at
the point where the costs of diverting additional resources to air pollu-
tion control are offset by the additional benefits that would be obtained.
1-4
-------
The distribution among different population groups and different geographi-
cal regions of the benefits and costs of the program or regulation can also
be considered in benefit-cost analysis. Among the costs that must be
considered for this type of analysis are those of compliance, regulation
and other environmental and energy impacts. The benefits include improved
visual aesthetics and reduced damages to health and property.
Air quality analysis and legislation have often focused upon
obviously measurable variables, such as chemical and physical character-
istics and control costs, for decision-making purposes. These decisions
implicitly set values on other seemingly non-quantifiable effects, such as
aesthetic benefits, even though they may never have been explicitly con-
sidered. 1 When only the costs of air quality control are estimated and
never the benefits, the decisions on the level and location of emission
controls nonetheless reveal implicit benefit measures made by the decision
makers (under the assumption that each decision was made rationally).
However, there is no assurance that such implicit valuations are accurate
reflections of society's actual benefits or consistent from one decision to
the next. For example, in mandatory Class I Federal areas the national
goal is to have no visibility impairment, implying that the benefits, which
are primarily aesthetic, of eventually eliminating all impairment are
expected to outweigh costs of control; in other situations some impairment
is allowed, implying that the benefits of complete control do not outweigh
the costs. Because few, if any, explicit benefit measures were used in
these determinations, there is no assurance that the implied size of
benefits was either correct or consistent, and therefore that the decisions
were socially desirable.
The value of the research techniques discussed in this guidebook
is that they allow explicit measurement of the aesthetic benefits to be
achieved from air pollution control so that more accurate, economically
efficient, and equitable resource allocation decisions can be made within
the benefit-cost framework of analysis.
1.3 Overview of Visibility Benefit Analysis
A visibility benefit analysis is a complex research project that
requires inputs from many disciplines. Figure 1.2 illustrates the steps
that must be considered in a thorough economic analysis of visual aesthe-
tics .
Steps 1 and 2. A benefit analysis must fully define the visibility
problem under consideration. This includes describing the sources and
characteristics of the alternative pollution levels, and dispersion modeling
to determine when, where, and at what magnitude the impacts will occur.
This is required if the benefit measures for visibility improvements are to
be precisely related to the emission rates and costs of controls necessary
to achieve the improvements.
1-For a discussion of the problem of non-quantifiables in benefit-cost
analyses see Freeman 1970; and Sassone and Schaffer 1978.
1-5
-------
1. Source of pollutant, type, intensity, and frequency.
2. Spatial dispersion of pollutant and type of impact (plume, haze).
3. Scenic content of affected view.
4. Atmospheric conditions and transmission of visual images affected
by lighting, cloud cover, etc.
5. Human perception of impacts.
6. Psychological effect of perceived impacts.
7. Economic valuations of impacts in terms of willingness to alter
time or dollar expenditures.
Figure 1.2 Inputs to the Economic Visibility Benefit Assessment Process
1-6
-------
Steps 3 and 4. It must be determined how the pollutants affect
visibility as perceived by observers. This will depend upon the types of
pollutants many cannot be perceived by the human observer and upon
atmospheric conditions. For example, the light scattering effects of small
particles cause more whitening of the horizon when the sun is located in
front of the observer rather than directly overhead.
Steps 5 and 6. It must be established what aspects of the perceived
visibility degradation will impair the observer's enjoyment or use of the
view or location under study. The loss of the scenic content of an ob-
structed view may be important, as may be the loss of visual range or
discoloration that may occur with increased pollution levels.
Step 7. How much effort the observer is willing to expend to
avoid, reduce, or eliminate these impacts depends on the importance to him
of the various visibility impacts of the pollutants, including the psycho-
logical effect. Economic benefit analysis techniques are the procedures
used to quantify what this effort is worth in dollar terms, thus measuring
part of the benefits of pollution control.
1.4 Guidebook Organization
The topics of each chapter of the guidebook are as follows:
Chapter 2. The conceptual economic framework and definitions of
monetary benefit measures for visibility improvements derived from consumer
utility theory and demand analysis.
Chapter 3. An overview of required inputs from other disciplines
to aid the economic analyst in communicating with physical scientists,
psychologists and sociologists in performing a complete benefit analysis.
Chapter 4 The applications, theoretical foundations and strengths
and weaknesses of several economic techniques currently used to measure
benefits of visibility and overall air quality changes.
Chapter 5. A summary of the start to finish research tasks for a
complete visibility benefit analysis with suggested step-by-step decision
and application guidelines.
Chapter 6. Five case studies of actual applications for both of
the two predominant techniquesbidding methods and the hedonic property
value approachand a comparison of their procedures and results.
A glossary, annotated bibliography, and an appendix with sample
picture sets and questionnaires for the bidding method are also included.
The nontechnical reader who desires an introduction to the esti-
mation methods and their application should focus upon Sections 2.1, 2.2,
4.1 to 4.3, 5.1 and the relevant case studies in Chapter 6. The hedonic
approach case studies and the South Coast Air Basin bidding method case
1-7
-------
studies are the most relevant for urban area pollution problems, while the
bidding method case studies are more relevant to air quality issues in
mandatory Class I Federal and other recreational and rural areas. Tech-
nical material is presented in Section 2.4 and the last portion of 4.3.2,
and non-economics material is presented in Sections 3.1-3.7 Actual appli-
cation of the benefit estimation methods will require mastery of much of
the material in these sections.
Though visibility aesthetics are the focus of this guidebook, some
of the techniques discussed are equally applicable for estimating other
benefits from the control of air pollution, such as reduction of health
and materials damage. The guidebook also emphasizes visibility in recrea-
tional areas, particularly in the West, because the focus of the regula-
tions is mandatory Class I Federal areas, the majority of which are located
in the West, and because the most recent empirical work has been in the
western U.S.
Several other EPA reference documents will be especially useful
for visibility benefit analysis:
1. Protecting Visibility: An EPA Report to Congress (EPA-450/5-
79-008).
2. The Development of Mathematical Models for the Prediction of
Anthropogenic Visibility Impairment (EPA-450/3-78-110 a, b, c).
3. Guidelines for Determining Best Available Retrofit Technology
for Coal-Fired Power Plants and Other Existing Stationary
Facilities (EPA-450/3-80-009b).
4. Assessment of Economic Impacts of Visibility Regulations
(EPA-450/2-80-084).
5. User's Manual for the Plume Visibility Model (PLUVUE) (EPA
450/5-80-032).
6. Workbook for Estimating Visibility Impairment (EPA 450/4-80-031).
7. Interim Guidance for Visibility Monitoring (EPA 450/2-80-082).
8. Visibility Protection for Federal Class I Areas (45 FR 80084-
80095).
All the above documents are available from National Technical
Information Service, 5285 Port Royal Road, Springfield, Virginia 22161.
-------
CHAPTER 2
ECONOMIC BENEFIT MEASURES
Performing a benefit-cost analysis of changes in visibility aesthe-
tics, or in other air quality conditions, first requires a careful defini-
tion of what the benefits are and how monetary measures of these benefits
are defined. This is particularly important for visibility aesthetics,
because the concepts of visibility as an economic good and monetary bene-
fits of visibility improvements may not be immediately apparent. This
chapter draws upon economic theory to describe the concept of an indivi-
dual's welfare and to develop correct monetary measures of benefits for
visibility aesthetics. The aggregation of benefits across individuals and
through time to obtain total benefit measures for use in benefit-cost
analysis is presented in a subsequent section. The chapter concludes with
technical models of consumer behavior that can be used to derive benefit
measures. The application of these theories of value requires precise
definitions and measures of visibility. This topic is deferred to Chapter
3 where physical measures of visibility, what aspects of visibility are
demanded and how the objects viewed affect visibility are discussed.
Before proceeding into a substantial volume of material, matters
will be simplified by the use of several conventions. First, since damages
from environmental degradation may be viewed as negative benefits, these
termsbenefits and damagesmay be used interchangeably. The term "costs,"
meaning the monetary charges incurred to implement a change, is not inter-
changeable with damages or benefits. In benefit-cost analysis, it is the
difference between benefits and costs which is examined to determine
whether a project represents an economically efficient allocation of
resources to their highest valued uses. Second, the economic analyses
presented can be used for most any economic good or environmental quality
variable including air quality characteristics. Hence, the terms "air
quality," "visibility aesthetics," and "environmental quality," while
having different connotations in application, are used interchangeably in
this chapter.
2.1 Concepts and Measures of Benefits
2.1.1. Concepts of Benefits
Economists generally perceive all things as having value. A
measure of value is the well-being, or what economists call utility,
derived from the consumption of a good or service. Similarly, any change
in the level of consumption also has a value associated with it as long as
someone's utility is affected. This change in utility may be viewed as
either a benefit or damage, depending upon whether the individual's well-
being is enhanced or diminished. Goods and services that are bought and
2-1
-------
sold in the marketplace, such as automobiles, haircuts, and blue jeans, are
easily recognized as having economic value because individuals part with
their scarce income to purchase them at market prices and forego purchasing
other goods and services that could have also increased their well-being.
The market price they are willing to pay for these goods therefore repre-
sents a minimum monetary measure of value to the consumer from their
consumption in terms of alternative goods and services foregone.
Most environmental goods, such as air quality, are public goods.
Once public goods are provided to one individual, it is difficult to
exclude others from their consumption. Also, once provided to one indi-
vidual the additional cost of providing them to others is zero. For
example, if the air quality in a region is improved for the benefit of one
group of individuals, it would be hard to exclude others in the region from
also sharing the benefits or to charge a price for the consumption of the
good. Consequently, public goods such as air quality are not exchanged on
a market and do not have explicit prices. This does not mean they do not
have value. People change their recreation patterns, move their residences,
or simply alter their moods due to the existing level of air pollution.
These non-market goods affect our well-being, and consequently, they have
implicit values. By analyzing how individuals react to air quality changes,
the value they place on air quality may be revealed.
There are three types of air quality values commonly addressed:
activity value, option value, and existence value. Activity value (also
called user value) is the value in use, i.e., enjoyment of the visibility
at a site when one is actually there. Option value is the value assigned
to the option of preserving some air quality level in anticipation of
potential future activity at the site. It is the value of preserving the
option that the activity value will be available at a later time based upon
some non-zero yet uncertain probability that one will enjoy the visibility
of the site at a later date.l It is important to note that option value
is a value above and beyond the activity value. Existence value is the
value assigned to the existence of a certain level of visibility aesthetics
at a site even though one does not ever intend to participate in activity
at the site. This may be tied to the philanthropic goal of preservation so
that future generations may have the option of enjoying consumption of the
good. Some authors also call this "preservation value," or "bequest value."
Activity, option and existence values are not unique to visibility
aesthetics; they can also be applied to wildlife, Rolls Royces or any
economic good. An aspect which is not often considered for air quality is
its secondary consumption, such as in calendar pictures and post cards.
These are on a fine line between activity and existence values but are
usually assumed to have a significantly smaller value than on-site acti-
vity, option and existence values, and subsequently are less frequently
analyzed.
'A discussion of option values is found in Cicchetti and Freeman (1971),
and Schamlensee (1972). A related concept, "quasi-option value", has also
emerged. It is the expected value of remaining flexible as opposed to
being committed to a longer course of action, or of making step-by-step
decisions rather than once and for all decisions. See Arrow and Fisher
(1974) and Miller and Lad (1981).
2-2
-------
2.1.2 Monetary Measures of Benefits
The benefits to an individual from a change in visibility aesthe-
tics equal the change in his well-being, a value which is difficult to
quantify. Economic benefit measures address the problem by attempting to
put a monetary quantification on changes in well-being. The monetary
measure of benefits for a change in air quality is said to equal the change
in income that would yield the same change in an individual's well-being.
This monetary quantification has advantages and disadvantages. The ad-
vantages include the use of a common unit of measurement which, in many
cases, can be readily observed in real market situations and is easily
understood, added, and compared to costs of proposed changes by policy
makers. A principal disadvantage of monetary quantification for visibility
is that units of quantification are not as easily defined or measured, as
is the case with market goods. The ability to use market data is a deter-
minant in the choice of methodology used to obtain benefit measures for
visibility valuation (see Chapter 4.) A second disadvantage relates to the
issues of equity and intensity of preference across individuals. In order
to measure the total welfare change society undergoes as the result of an
action, it is necessary to combine or compare the welfare changes for all
affected individuals. Even if the benefits or damages of a proposed change
can be measured in dollars for each affected individual, combining these
values requires some explicit consideration of the relative importance of
impacts to the individuals. This issue is discussed in Section 2.3.
In some instances the researcher will simply be unable to obtain
monetary measures of benefits. There are other methods which may be used
to arrive at decisions about the relative value of visibility that do not
use dollars, including preferability orderings, semantic differentials,
weighting of multiple objectives, and the like (see Sinden 1979). These
methods do not allow measurements of the change in utility, but they do,
for instance, indicate the concentration of haze in the Blue Ridge Moun-
tains at which the recreation experience ceases to be pleasant (or utility
ceases to be positive), or the relative importance of air quality levels
versus other concerns of society. Thus, these methods indirectly aid
decision making by accounting for reactions to proposed changes. This is
not picking up "value" in its true sense, but is indeed capturing "valua-
tion." In many cases these techniques are worthwhile undertakings before a
monetary benefit analysis is performed. By determining the characteristics
of an experience that are most important to affected individuals, they help
focus the subsequent monetary benefit analysis upon these important charac-
teristics. A recent example of one such non-monetary valuation for poten-
tial visual and visibility impacts of nearby strip mining upon Bryce Canyon
National Park can be found in U.S. NFS (1980). While these approaches are
useful, it is the focus of this guidebook to examine monetary measures of
benefits for incorporation into benefit-cost frameworks.
Three monetary measures of value used by economists in benefit
analysis are willingness to pay (WTP), expenditures, and consumer surplus.
"WTP" is the maximum amount an individual would be willing to pay for a
2-3
-------
good or service, representing its maximum monetary value to the indivi-
dual. In many cases price is an accurate reflection of the maximum WTP for
additional units of a good or service. "Expenditures" represents the
actual amount a consumer spends for a good or service (price times quanti-
ty), which in a situation of free choice would never be greater than his
maximum WTP for all of the units consumed. However, an individual may be
fortunate enough to spend less than his maximum WTP. This surplus value,
or the difference between the maximum WTP and expenditures, is commonly
referred to as "consumer surplus."
Consumer surplus is an especially important concept because changes
in consumer surplus equal the monetary measure of benefits from changes in
air quality. In many cases visibility aesthetics do not have a market
price and there are no expenditures, so that consumer surplus equals the
maximum WTP for all units consumed and is all that is affected by changes
in visibility conditions.
The Demand Curve Approach to Monetary Measures of Benefits
Demand curves can be used to measure maximum WTP, expenditures, and
what is called ordinary consumer surplus.2 A demand curve is the rela-
tionship between the amount of a good or service demanded and the maximum
WTP for it in terms of price. As shown in Figure 2.1, where PI, P2, Ql and
Q2 are specific price and quantity levels of some good, a demand curve
typically is downward sloping. While this need not be the case, empirical
research nearly always supports this law of demand. This is, in part, due
to the law of diminishing marginal utility, which is that each additional
unit consumed yields a smaller increase in total utility, and therefore the
individual's WTP for additional units also decreases. The quantity demanded
by a consumer depends on the price of the good, the prices of other goods,
the consumer's income, and his tastes and preferences. Because the demand
curve is the relationship between price and quantity demanded, a change in
the price of the good results in a movement along the demand curve, whereas
a change in the other things that determine demand will shift the demand
curve. For example, an increase in income may increase the quantity that a
consumer is willing and able to buy at each price.
A demand function can be mathematically illustrated as follows:
Qd = f (Px, P0, M, T)
Where:
Qd = quantity of good x demanded
Px = prices of good x
^Those unfamiliar with the fundamentals presented in this section are
referred to any standard intermediate microeconomic theory text such as
Hirshleifer (1976).
2-4
-------
P = Price of X
PI
P2
Demand Curve
Figure 2.1
Demand Curve For A Typical Good X
Q = Quantity
of X
2-5
-------
P = prices of other goods and services
M = individual's income
T = individual's tastes
From the demand curve the maximum WTP for an additional unit of a
good is represented by the corresponding price. For example, in Figure 2.1
the WTP for additional X at level Ql is PI. Expenditures are that area
under the demand curve which equals price times quantity. At consumption
level Ql, expenditures equal the shaded rectangle. Ordinary consumer
surplus (DCS), which is one approximation of the consumer surplus concept,
equals the difference between the area under the ordinary demand curve up
to the quantity consumed and expenditures. At Ql, OCS equals the cross-
hatched area under the demand curve above expenditures.
To demonstrate the use of demand curves for air quality benefit
analysis, consider the effect of visual quality upon recreation experience
in national parks and other similar areas. The individual's maximum WTP
for various visibility levels (perhaps represented by visual range) that
may be experienced on a recreation day in a particular park is represented
by demand curve DD in Figure 2.2. For example, the monetary value placed
upon additional miles of visual range at level VI is Wl. However, this
monetary value is not paid by the consumer and represents a surplus value
commonly referred to as OCS. In Figure 2.2, if VI is the prevailing
visibility, the dotted area represents the OCS measure, or the total unpaid
monetary value of visual range. If visual quality is improved to the level
V2, the recreationalist's well-being is improved, but no additional payments
are made. The monetary value of this improvement in well-being for each
recreation day equals the increase in consumer surplus, shown by the cross-
hatched area.
Figure 2.3 gives an equivalent and perhaps more operational pre-
sentation of willingness to pay for visibility. Here the demand for
recreation days is represented as a function of the price per day, in-
cluding entrance fees, food, variable expenses, value of time, and the
like. In this approach the demand for recreation days shifts as a function
of visual quality at the recreation site. If P represents the daily
entrance price and other daily costs, and visual quality is 50 kilometers,
then Rl recreation days are demanded. Total expenditures are represented
by the area OPAR1 and consumer surplus by the area PCA. If visibility
increases to 100 kilometers, demand to participate in recreation at all
prices can be expected to increase, perhaps to D2. At price P, R2 days are
now demanded and expenditures increase to the area OPBR2. Total consumer
surplus increases to the area PDB. It can be seen that not only is it
likely that the consumer surplus attached to each recreation day increases
but the number of days spent in the activity also increases.
Because the increased expenditures occur at the expense of de-
creased expenditures on other goods and services, they do not represent an
2-6
-------
Willingness to Pay
Wl
Demand Curve
Visual Quality
(perhaps visual
range in miles)
Figure 2.2
Willingness To Pay For Visibility In The Recreation Experience
-------
$ Willingness to Pay Per Activity Day = W
D
Rl
D2
(Visual range = 100K)
(Visual range = 50K)
Days spait in Activity R
(Recreation)
Figure 2.3
Demand for Activity Days as A Function Of Price And Visual Range
2-1
-------
increase in utility, but rather a shift in consumption from one good to
another. The change in OCS, represented by the area CDBA, is a monetary
measure of the change in an individual's utility for the change in visi-
bility because it represents the WTP for the change in excess of expen-
ditures. This change in OCS accurately reflects the monetary measure of
the change in an individual's utility under special conditions, such as
small income effects and weak complementarity, described further after
more theoretical foundations are established.
Factors affecting the location, steepness and shifts in the demand
curve and, consequently, in the consumer surplus measure, include avail-
ability and cost of alternatives, special characteristics of the site,
user population location, income, and other socioeconomic variables. The
travel cost approach described in Chapter 4 is an attempt to derive bene-
fits for visibility changes by estimating this type of demand curve for
recreation and by estimating changes in this demand as visibility changes.
In many markets, such as housing, the WTP for housing may similarly
be influenced by local air quality conditions. As described in Chapter 4,
by investigating market prices, house and buyer characteristics, and air
quality characteristics, price functions for property values as a function
of visibility can be estimated, from which implicit demand curves for
visibility, such as that shown in Figure 2.2, can be estimated and monetary
benefit measures obtained.
Unfortunately, demand curves for market goods as a function of air
quality characteristics are often very difficult to estimate because appro-
priate data are often unavailable. Researchers may attempt to estimate
changes in consumer surplus by direct questioning of individuals about
their WTP or their behavior as air quality changes (described in Chapter
4). The correct application of this and the above mentioned techniques to
obtain benefit measures requires very carefully defined measures of con-
sumer surplus other than the OCS measure thus far described.
2.2 Refined Consumer Surplus Monetary Measures of Benefits
Consumer surplus plays a critical role in valuing a change in air
quality. Unfortunately, the OCS measure illustrated in the previous
section is not a technically correct measure of the consumer surplus
concept. Because of this, several refined measures have been developed to
accurately define and estimate this monetary measure of benefits. Under-
standing the deficiencies in the OCS measure and using and understanding
the refined measures first requires knowledge of the basic economic frame-
work used to analyze consumer behavior.
2-9
-------
2-2.1 The Utility Map Framework Used to Define Consumer Surplus Measures
All theories of consumer behavior begin with the assumption that
consumers make rational choices concerning the allocation of their resources
with the ultimate goal of maximizing their own satisfaction or total
utility, which is the sum of the utility from all goods consumed. Because
the choice facing each consumer is not usually whether to buy all or none
of a particular commodity, marginal utility, or the additional satisfaction
derived the last additional units of a good consumed, is as important to
decision making as is total utility.
The standard paradigm used to illustrate the utility maximization
problem, which is the allocation of income to alternative uses so as to
maximize total utility, is the indifference map illustrated for two goods,
XI and X2, in Figure 2.4. (This type of analysis can be applied to any
number of goods or services.) An indifference map is a set of indifference
curves, such as II, 12, and 13, each representing various combinations of
the two goods which yield the same level of utility to the consumer.
Curves further away from the origin represent higher levels of utility.
The shape of the curves are based upon standard and generally well substan-
tiated assumptions about the theory of choice and consumption. One of the
most important assumptions is that of diminishing marginal utilitythat
the additional utility derived from consuming an additional unit of a good
decreases as the total consumption of the good increases. The shape of the
indifference curves reflects diminishing marginal utility, because as
consumption of one good increases, the increase in utility will be offset
by smaller and smaller decreases in consumption of the other good.
If the prices of goods XI and X2 are given as PX1 and PX2 and the
consumer's income is M, a budget line AC may be drawn. All points on or
to the left of AC represent bundles of goods XI and X2 which may be pur-
chased with income M. The highest utility level that may be reached with
this income constraint, representing the consumer utility maximum, is on 12
at point B. By varying the price of one of the goods and locating each new
consumer maximum point, it is possible to trace out the quantities of the
good that would be purchased at each price, and with these price and
quantity combinations to plot the previously illustrated classic demand
curve for that good.
Returning to consumer surplus, Alfred Marshall defined it in his
Principles of Economics (1930) as the excess of what the consumer would be
willing to pay rather than go without the good, over that which he actually
does pay. This clearly is the difference between the individual's maximum
willingness to pay for all units and his actual expenditures. Marshall's
"consumer's surplus" is illustrated in Figure 2.5, where two of the indi-
vidual's indifference curves showing the trade-off between all other goods,
represented by income, and some good X are presented as II and 12.
Let us assume an individual is at point A consuming zero units of
X. This might occur because the good does not exist or is priced at such a
2-10
-------
Quantity of X2
A =
X2
M
PX2
13
12
II
Quantity of XI
C =
PX1
Figure 2.4
Indifference Map of XI, X2
2-11
-------
high price, such as PA, that the consumer chooses to consumer zero units of
X. If X is introduced at some price PX, income M yields the budget line
AZ. Point E is then the utility maximizing point for the individual,
because it is on the highest utility level obtainable given income M and
price PX. Thus, at price PX the individual would choose to consume Q units
of X causing utility to increase from Ul to U2.
To consume Q units of X the individual would expend AB of his
income. However, the individual could have foregone AC in income while
still consuming Q units of X and remained on the original indifference
curve II at utility level Ul. The individual would not be willing to spend
more than AC for Q units of X because to do so would put him on a lower
indifference curve (i.e., reduce his utility from his original level at
point A). Therefore AC is the maximum willingness to pay for Q units of X
while the actual expenditure is AB. The difference, BC, is equivalent to
Marshall's definition of consumer's surplus and is a monetary measure of
the change in utility value derived from the move from point A to point E.
It was originally believed that Marshall's consumer's surplus was equal to
the area under the ordinary demand curve to the left of Q and above the
price line, as shown by the shaded area in Figure 2.6. Hicks (1941) and
Henderson (1941) showed that the ordinary consumer surplus measure of the
area under the ordinary demand curve would equal consumer's surplus as
defined by Marshall only under special circumstances. This argument is
based on the existence of an income effect upon utility levels that is not
accounted for in ordinary demand curves (ODC).
2.2.2 Why the Ordinary Consumer Surplus Measure is Incorrect
OCS does not accurately represent the consumer surplus concept
because utility is not held constant for points along the ODC. Recall that
consumer surplus is to be a dollar measure of the change in utility because
the individual's expenditures are less than his maximum WTP. To appropri-
ately measure consumer surplus a demand curve must depict price and quantity
combinations that represent the maximum WTP that keeps utility constant at
all points in the demand curve. However, as can be seen on Figure 2.5, the
consumer equilibrium price quantity combinations represented at point A and
point E and used to plot points A and E on the demand curve in Figure 2.6
are not at the same level of utility. In general, the level of utility
corresponding to points on the ODC increases moving down the curve. The
OCS measure is not true consumer surplus because it measures the differences
between expenditures and the maximum WTP to remain at the same level of
utility plus the change in WTP due to the change in utility. This latter
effect is what is known as the income effect.
Recognizing these problems, Hicks (1944) developed an alternative
demand curve called the income compensated demand curve (ICDC) and several
alternative consumer surplus measures. Whereas the ODC gives the quantity
that a utility maximizing consumer with a given income level will demand at
each price, the ICDC shows the quantity a consumer will demand at each
2-12
-------
Figure 2.5
Marshall's Consumer Surplus
Price of X
PA
PX
12 (U = U2)
II (U = Ul)
Quantity of X
Demand Curve for X
0 Q
Figure 2.6
Ordinary Consumer Surplus Using Ordinary Demand Curves
Quantity
of X
2-13
-------
price, assuming his income is adjusted so that he remains at his original
utility level. The difference is illustrated in Figure 2.7- In panel a,
assume an individual is at A, with utility Ul, consuming Ql of X at price
PI. This corresponds to point A on the ODC in panel b. Next, if price is
decreased to P2, the consumer maximum is reached at point B at higher
utility level U2, consuming Q3 of X. This corresponds to point B on the
ODC below.
This price induced change in consumptions from Ql to Q3 can be
broken down into two parts: a substitution effect and an income effect.
The substitution effect is the change in quantity due to a price change
holding utility constant. By making the new price line P2 tangent to the
old utility level, the consumer optimum point is at C where P2* is tangent
to Ul and Q2 of X is consumed. Again, this is the pure price substitution
effect with utility held constant. The income effect is the change in the
quantity consumed and subsequent change in utility because real purchasing
power changes as prices change. This is represented by the shift from P2*
to P2, quantity increase from Q2 to Q3, and utility increase from Ul to U2.
It is this income effect upon utility and consumption which invalidates
using the ODC to measure consumer surplus. The ICDC adjusts for the income
effect by plotting points A and C (see panel b) so that utility is held
constant along the demand curve and appropriate consumer surplus measures
can be derived. For the ICDC1 the consumer surplus measure is the area to
the left of the ICDC1 and above price for a price change from PI to P2 and
is less than OCS measured under the ODC.
For "normal goods," goods for which as income increases consumption
of the good also increases, the ICDC will be steeper than the ODC. Where
there are no income effects, i.e., changes in income do not affect consump-
tion levels, the ICDC equals the ODC. The question of how large the income
effect is and how large the difference is between consumer surplus measures
using the ODC rather than the ICDC is addressed below. Unfortunately ICDC's
are not unique as are ordinary demand curves, but rather an ICDC exists for
each level of utility implied by each point on the ODC. For example, an
ICDC2 could also be derived at point B for utility level U2. Another
important point is that ICDC's are not observable in real market situations
as are ODC's and must be derived from the estimation of ODC's, estimated
income effects and restrictive assumptions.
2.2.3 Four Refined Consumer Surplus Measures
The consumer surplus concept of a measure of the change in utility
that occurs when one pays less than the amount that would maintain constant
utility is unambiguous; whereas there are several monetary measures of
consumer surplus. This section discusses four more measures which are
technically correct in that they hold utility constant, but which differ in
terms of which utility level is used for comparison and if the individual
can affect the quantity consumed.
2-14
-------
Income
Quantity of X
Panel a: Utility Map
Price of X
PI
P2
\
\
^ *-
\
\j , D
N^ ft lp-
V . OCD
X
ICDC1 x ICDC2
L_J i
01 02 03
Quantity of X
Panel b: Demand Curves
Figure 2.7
Derivation Of Ordinary And Income Compensated Demand Curves
2-15
-------
Consumer Surplus Measures of Price Changes
Figure 2.8 illustrates two alternative measures of the impact of
real or implicit price changes. Following Figure 2.7a, point A is the
consumer optimum for income M and price PI. For a lower price, P2, point B
represents the new consumer optimum point.-'
Income
EV
CV
Quantity of
Visibility
EV and CV Consumer Surplus Measures
definitions and presentations can be generalized to multiple price
changes. See Layard and Walters (1978).
2-16
-------
Two monetary measures of this change in utility from Ul at point A
to U2 at point B are "compenating variation" and "equivalent variation."
1. Compensating variation (CV) is the change in income, given the
new price, which offsets the change in utility induced by the
price change if the individual is able to vary the quantity
consumed. Given the new P2, income equalling CV could be
withdrawn and the individual would substitute consumption to
achieve maximum utility at the original level Ul at point C.
2. Equivalent variation (EV) is the change in income, given the old
price, which yields the same change in utility as the price
change if the individual is able to vary the quantity consumed.
Given the old price PI, income equalling EV could have been
given to the individual and by substituting consumption, utility
level U2 would be reached at point D.
As long as the income effect is non-zero, EV and CV measures will not equal
each other or OCS. It has been demonstrated by many authors (Hicks 1944;
Patinkin 1963; and Freeman 1979a) that the CV and EV measures are equiva-
lent to the area under income compensated demand curves. In general, EV is
measured by change in the area under the ICDC that corresponds to the new
level of utility and CV is measured by changes in the area under the ICDC
that corresponds to the original utility level. In Figure 2.7b for a price
change from PI to P2, CV is measured using ICDC1 and EV using ICDC2. It is
readily observable that for a price decrease the equivalent variation
exceeds the compensating variation measures if Q is a normal good. This is
due to the income effect.
A positive income effect implies that as real income increases the
marginal WTP for additional units of the good increases. For example, as
income increases, the marginal WTP for increased air quality may also
increase. For the higher level of utility, U2. represented in ICDC2, the
real income is larger than for utility level Ul represented in ICDC2.
Therefore the maximum WTP for changes in air quality based upon utility
level U2 is larger than the maximum WTP based upon Ul. Consumer surplus,
as the difference between this maximum WTP and expenditure is therefore
larger when based upon U2 (EV measure for a price decrease) than when based
upon Ul (CV measure for a price decrease). For the OCS measure, utility
varies from Ul to U2. Consequently, the OCS measure is between the EV and
CV measures.
Consumer Surplus Measures of Quantity Changes
The individual is not always able to choose the quantity consumed.
This is particularly true for air quality at a particular site. The air
quality at a national park is seldom under the influence of any one indi-
vidual the day he recreates there. Two measures"compensating suplus" and
''equivalent surplus''are applicable in cases where quantity of the good
cannot be adjusted regardless of whether there are any market prices
2-17
-------
! Compensating surplus (CS) is the change in income that offsets
the change in utility induced by the change in the quantity
of the good. The individual is at the new quantity level and
old utility level.
2. Equivalent surplus (ES) is the change in income that yields the
same change in utility as did the change in the quantity of the
good. The individual is at the old quantity level and the new
utility level.
CS and ES are illustrated in Figure 2.9 for an increase in visi-
bility from VI to V2, where income represents an aggregate of all market
goods and Ul and U2 are indifference curves. ES equals the addition to
income that at the original visibility level would yield a utility change
equivalent to the visibility quality improvement. The CS measure equals
the reduction in income necessary to return the individual to the original
utility level at the new visibility level. It is again the amount of
income that offsets the utility change caused by the change in visibility
while the individual remains at the new visibility level.
Income
M
VI V2
Figure 2.9
CS and ES Consumer Surplus Measure
Visibility
CS and ES could also be depicted in Figure 2.8 as the distances BF
and AE, respectively. These measures will also be of different magnitudes,
which again depends upon the size of the income effect.
Freeman (1979a) has suggested that when the ability to substitute is not
present, which distinguishes the CV and EV measures from the CS and ES
measures, CV and EV measures logically reduce to the CS and ES measures
shown on Figure 2.9.
2-18
-------
2-2.4 Comparisons Among the Consumer Surplus Measures
Ordinary consumer surplus and the four refined measures can also be
compared in terms of their relative size, their implied reference level of
welfare, and other more technical aspects.
Size
If there are no income effects, or if changes in real income do
not affect consumption levels of the good in question, then all measures
are equal. If, however, the commodity is a normal good (positive income
effect), then for a price decrease we have:
ES > EV > CV > CS
For a quantity increase we have:
ES > CS.
By examination of the definitions it can be shown that the compensating and
equivalent measures are symetrical. In other words, CV for a price de-
crease is equal to EV for a comparable price increase; and CS for a quantity
decrease is equal to ES for a comparable quantity increase. Therefore, for
a price increase of a normal good, we have:
CS > CV > EV > ES
and for a quantity decrease:
CS > ES.
The issue, then, is how large the differences are between each of
the refined measures and OCS. Generally the OCS lies somewhere between
the EV and CV measures for any particular price or quantity change (the OCS
measure does not apply as directly for quantity changes). In some cases
the OCS measure is a close approximation of the other measures, as shown by
Willig (1976) and by Randall and Stoll (1980).
Willig (1976) first addressed this issue for changes in prices and
found that OCS yields a satisfactory estimate of the welfare change unless
the income effect is very large, expenditures on the good are a large
component of income, and the price or quality change is large. In fact,
the error in approximation is probably much smaller than the error in
estimation of most demand curves for visibility.
2-19
-------
To illustrate the size differences between OCS and either EV or CV
measures Willig provides the following formula, where the income elasticity
of demand equals the percent change in quantity demanded for each one
percent increase in income:
OCS - Wp QCS*N
OCS ~ 2M
Wp = EV or CV as appropriate
OCS = ordinary consumers surplus
N = income elasticity of demand
M = initial level of income
OCS = $500 = 25
As an example, if income elasticity =1.1, and 2M $20,000
then the percent error in using ordinary consumer surplus is 2.75 percent.
This holds quite closely if ° '^MN < .04. Exact formulas are also
provided for other, more general cases.
Blocksteal and McConnell (1980) point out that Willig's formula is
very limited for cases where the choice is between unalterable quantity
levels (perhaps large "all or nothing" choices), as is often the case for
changes in the environment. In these cases the large changes will further
increase the size differences of the measures due to substantial nonlineari
ties not accounted for in Willig's calculations. The calculations are also
limited when demand cannot be obtained from observable data. In these
instances the authors argue that the appropriate measurement of the impacts
would be made with the CS and ES measures.
Randall and Stoll (1980) undertook an analysis similar to Willig's
for changes in quantities using ES and CS measures. The results are very
similar for small changes in quantities; but where changes are large, the
good is highly valued, and where income elasticity of WTP is large, the
differences are likely to be substantial. Willig's formulation can be used
for this analysis by substitution of Wp for ES or CS; and N = income
elasticity of WTP (see Randall and Stoll 1980, page 454); however, the
authors support Blocksteal and McConnell by suggesting that these formulas
underestimate the difference in size of the measures for large quantity
changes.
Reference Level of Welfare
Let us concentrate upon the ES and CS measures for changes in
visibility. The CS measure tells what compensating change in income is
necessary to make the individual indifferent between the original environ-
mental quality and the new situation. Regardless of the direction of
2-20
-------
change, the CS measure takes the original utility point as the basis of
comparison. In this sense it also implies that the individual has what has
become known as property rights to the initial level of visibility.
The ES measure tells what income payment would lead to the same
utility change as the change in visibility. In other words, ES is the
amount of additional income necessary to induce the individual to forego an
increase in visibility or the amount of income taken from the individual to
yield an equivalent utility change as a visibility degradation. Therefore
the ES measure takes the new or proposed position as the reference level of
utility, and implicitly vests the individual with the property rights
level associated with the new or proposed visibility levels.
From a theoretical standpoint the choice of ES or CS measures
primarily depends upon the choice of an appropriate reference level of
utility, however; the implied property right aspect of the comparison among
the measures greatly affects the application of the welfare measures to
actual changes in visibility and must be considered. Section 4.2.2 below
examines this issue and the appropriate choice of measures in more detail.
In the marketplace the individual is a price taker and has no
designated right to any particular set of prices. For example, if a price
change were to occur, the correct measure of value would be based upon the
existing situation and therefore would be the compensating variation
measure (see NUREG 1980). However, in situations where other rights have
been established, such as legally mandated lower rents for low-income
housing, the equivalent variation measure would be appropriate.
For visibility the establishment of property rights is not clear
cut. Individuals may believe that they have the property rights to dif-
ferent levels of visibility. Further, if no rights are assigned, the state
that the individual desires will often determine the appropriate surplus
measure. For example, advocates of the proposed new level (of visibility)
may believe that level to be the appropriate property rights allocation.
For this group the equivalent variation measures are appropriate. On the
other hand, dissenters to the change may believe the current level is
appropriate; consequently, the compensating measures are appropriate for
this group.
Other Points of Comparison
Other points of comparison have also been thoroughly examined.
These include the concepts that the EV measure relates to the new utility
level for proposed changes and consequently is more in keeping with con-
ceptual welfare measures of a proposed change than the CV concept of
preventing a change (Silberberg 1972); that if multiple sequential changes
are proposed the EV measure is not independent of the order in which
changes are considered, except under special assumptions, while the CV
measure is (Silberberg 1972); and that the EV measure does yield rankings
2-21
-------
of multiple alternatives that are consistent with the underlying utility
preferences while the CV measure may not (Hause 1975 and Freeman 1979).
Selection of the Appropriate Measure
The points of comparison discussed support the refined measures
over the OCS measure as monetary measures of benefits associated with a
change in the quantity or implicit price of visibility. However, the
fact that the measures are, in certain circumstances, approximately equiva-
lent theoretically often justifies the use of whatever measure is most
readily and accurately available. Where there are large price or quantity
changes, or where WTP is very responsive to income changes, the refined
measures should be used. CS and ES measures should be used where the
individual cannot affect the quantity consumed; otherwise the CV and EV
measures are appropriate. The selection of CV and CS versus EV and ES
measures is not clear cut and must be determined based upon how the benefit
measures are to be used. Because several techniques used to estimate
monetary benefit measures have specific implications about property rights
this criteria may dominate other considerations when choosing an appropriate
measure. This point is discussed in more detail in Section 4.2.
2.3 Aggregation to Total Benefit Measures
Benefit-cost analysis is based upon the proposition that if the
total benefits of a proposed change exceed total costs, the project is an
economically efficient reallocation of society's resources. Thus far, this
chapter has focused only on monetary measures of benefits of air quality
changes for an individual. Because air quality is a public good where
changes affect many individuals, the appropriate total benefit measure is
some combination of benefits for all individuals affected by the proposed
change. The problem is to identify all potential benefits and affected
individuals and how their benefits are to be aggregated to a total or
weighted for comparisons. The number and type of individuals affected will
depend on the location, frequency and timing of the air quality impact.
For example, a permanent urban plume would affect all the residents and
visitors in the urban area whereas a prescribed burn that occurs one week
out of the year affects only those who are present at that time.
There are many types of benefits from air quality control, in-
cluding prevention of agricultural and materials damage, health impacts and
aesthetics impacts. This guidebook concentrates on the latter, within
which activity values, option values, and existence values should all be
considered. The types of value being examined determine the affected
population. "Activity value" applies to residents of and tourists to the
location where the change will occur, "option value" applies to all indivi-
duals who could reside at or visit the location, and "existence value"
2-22
-------
applies to all individuals. Usually the largest set of individuals for
whom benefits are measured are those for whom benefits could potentially be
captured. For example, all activity value benefits could potentially be
captured through taxes, entrance fees, and the like. Option and existence
value benefits could potentially be captured for all individuals of the
same country through taxation, which could then be used to pay the costs of
a program. Option and existence values for citizens of foreign countries
are unlikely to be captured and are usually not considered for projects
which do not cross international boundaries.
The selection of the appropriate methodologies to aggregate and
compare benefits across individuals and through time is a difficult prob-
lem. Different methodologies can result in variations in the total
benefit measures far outweighing variations from selecting alternative
benefit measures. Because these methodologies are not the focus of this
guidebook and are extensively addressed elsewhere, the next sections
highlight problems in aggregating and comparing benefits, suggest direc-
tions that can be taken, and provide references for additional reading.5
2.3.1 Aggregation Across Individuals
The first problem that must be addressed is how to compare and
aggregate all the individual benefits which occur in the same time period;
this is the aggregation to obtain total yearly benefits. A frequent
practice in benefit-cost analysis is to simply sum the benefits across all
affected individuals. The advantage of this approach is that if total
benefits exceed total costs there is a potential net improvement in the
monetary measure of society's well-being, which will yield what is called
an efficient reallocation of society's resources. This is because one
version of what is called a pareto optimal change could occur: that is,
some benefits could be taxed away and redistributed to cover damages and
costs so that some individuals are better off and no individuals are worse
off. A problem with this efficiency criterion is that the system of taxing
benefits to pay costs may be impossible to implement on a project-by-project
basis, resulting in an unequal distribution of impacts. On the other hand,
because the federal government undertakes a large number of projects, each
of which may have positive or adverse impacts to any individual, varying
impacts from project to project may balance the distributional inequities
of any one project (see Freeman 1979a; Polinsky 1972; and Freeman and
Haveman 1977 for additional discussions).
There are several shortcomings to this simple aggregation ap-
proach. Any aggregation is an implicit weighting of benefits. Simple
aggregation implies that one dollar of benefits or damages is of equal
importance to all individuals regardless of income or location.
^Useful discussions on the concepts and methods of aggregating benefits
may be found in Sassone and Schaffer 1970; Mishan 1972; Freeman 1979a;
Mikesell 1977; and Lind, et al. 1982.
2-23
-------
Another problem with simple aggregation is that measured WTP by
lower-income individuals is likely to be less than that of higher-income
individuals. This is because the greater financial flexibility of the
latter group may yield a lower marginal utility of money and increased
willingness (and ability) to pay. In this situation a simple aggregation
of benefits tends to support more projects which provide disproportionately
greater benefits to higher-income groups. Because the use and enjoyment of
environmental quality, such as visibility, in urban and recreation settings
is often not income constrained, the simple aggregation of WTP measures may
lead to inequities in the distribution of benefits.
An example of the equity issue is a case in which a high-income
individual (say $100,000) receives an economic benefit of $1,100 which
exceeds damages of $100 each to 10 lower-income individuals. While a tax
of $1,000 of the benefits could cover the damages, this may not be a
politically and socially acceptable action. Hence, the distribution of
impacts of a proposed project cannot be ignored.
The same concern holds for locational distribution of impacts. For
example, the Rocky Mountain West is faced with potential environmental
degradation because the benefits of energy independence and potentially
lower energy costs to the rest of the country may exceed the local damages.
Another example is the issue of whether an individual should be restrained
from developing a tract of private land because it may spoil the view from
a national park used by many citizens across the country. In both cases
the incidence of benefits and damages fall upon different groups of people
at different locations.
Technically, what is needed to appropriately compare the distribu-
tional impacts of benefits is an acceptable weighting scheme, which is
often a political rather than an economic issue. One approach which has
been accepted in some instances is the system of multiple objectives and
accounts established by the Water Resource Council's Principles and Stand-
ards (1979), which displays benefits and damages by categories of "region"
and "rest of nation" and specifically separates out impacts to low-income,
unemployed or minority individuals. With this approach the simple aggre-
gate benefit measure may be more meaningfully evaluated.
2.3.2 Aggregation Across Time
Once benefits are aggregated to yearly totals, these totals must be
aggregated through time. The standard procedure is to discount future
benefits to a present value following the logic that a dollar held now is
worth more than a dollar to be held sometime in the future.
The choice of the appropriate discount rate is a complex and often
unresolved issue.6 In practice the choice of a discount rate is usually
from among a set of prevailing market rates for financial instruments of
6Useful in-depth discussions can be found in Mikesell 1977; Lind, et al.
1982; Mishan 1971; and Sassone and Schaffer 1978. Other readings of
interest include Baumol and Oates 1979; Sandier and Smith 1976, 1977;
Arrow and Fisher 1974; and Krutilla, et al. 1972.
2-24
-------
duration equal to the project length, for 50 years, or for some other pre-
determined rate. This rate represents an intertemporal weighting scheme
and can greatly affect the aggregate benefit measure. Table 2.1 shows the
total present value of a constant stream of $1,000 in benefits per year for
alternative combinations of discount rates and time horizons. These data
show that the choice of discount rates and time horizons can radically
affect the total benefit measure and overwhelm the measurement problems in
the individual value measures.
Table 2.1: Present Value of $1,000 Benefits Per Year
Discount Rate
Years
25
35
50
75
0
$25,000
$35,000
$50,000
$75,000
5
$14,094
$16,374
$18,256
$19,485
10
$9,077
$9,644
$9,915
$9,992
15
$6,460
$6,617
$6,661
$6,664
The choice of discount rates is further complicated by population
growth, which must be accounted for in either the yearly benefit figure or
the discount rate, by inflation rates, and by changes in prevailing market
interest rates over time. Particular care must be taken to account for
the effects of inflation. Either all benefits should be expressed in real
(inflation adjusted) dollars for each year and a real discount rate used,
or all benefits should be expressed in nominal dollars with a nominal,
usually market, discount rate used. Exact formulation can be found in
Mikesell (1977).
Discounting procedures are explicit intertemporal weighting schemes
subject to a great deal of uncertainty about future generations. Should
impacts to future generations be weighed (discounted) to lower values than
current impacts? Will future generations have the same, or similar valu-
ations of impacts? Future tastes and preferences, income distributions and
the degree of aversion to risk are all unknown, but discounting procedures
implicitly project current values as expected future values. If individuals
are risk averse, and possibly in other circumstances, it has been shown
that the discounting procedure, with the use of expected future values,
biases decision making toward accepting development rather than conserva-
tion. In essence, the problem is that the future wishes are unknown while
the current generation has the power in the decision-making process.
The problem of intertemporal choice and allocation of resources is
the most critical and difficult where a change is irreversible or a resource
is irreplaceable. Few things are totally irreversible with unlimited eco-
nomic and political resources, but many things are effectively irreversible.
For example, mineral extraction from pristine environments, flooding of
2-25
-------
canyons, and the demolition of historic buildings are all examples of
actions which cannot normally be reversed. Due to natural chemical pro-
cesses, wind, and gravity, most air pollution problems can be rectified if
the source is controlled; however, the effects of the pollutant may be ir-
reversible. For example, exposure to asbestos is apparently linked to
higher rates of cancer and mortality long after the source is removed.
In summary, there is no discount rate which is appropriate for all
instances, although in many cases there are required discount rates for
federal projects. Where possible, the discount rate selection should be a
policy decision. If an appropriate rate is not available, it is advisable
to present total benefits using several alternative rates so the effect of
alternative rates can be ascertained.
2.4 Technical Consumer Behavior Models and Assumptions Used to
Derive Consumer Surplus Measures
Demand curves and consumer surplus measures may be theoretically
derived through the use of several alternative mathematical constructs and
assumptions about consumer behavior. This section introduces several
alternative constructs that are used by researchers to model the economic
benefits of changes in air quality. While the technical researcher will
need to be familiar with these models of consumer behavior, others need not
be familiar with this material to obtain an overview of how visibility
benefit analyses are conducted.
A Simple Utility Map Approach
In general, an individual tries to maximize his utility subject to
constraints. In the case of visual quality, one constraint may be the
inability to alter the level of quality that is incurred. By using
income as the composite index for all other goods, we may specify utility
as a simple function of visual quality Q and income M, where IT is the
initial income level for the individual:
U = U(Q, M) (1)
If one considers an improvement in visual quality from state
QA to state QB, the surplus measures are defined as follows:
U(QA,M) = U(QB,M - CSAB) = UO (2)
U(QB,M) = U(QA,M + ESAB) = Ul (3)
2-26
-------
where UO is the original utility level and Ul is a higher level of utility
obtained as visual quality is improved as depicted in Figure 2.9 on page 2-18.
CSAB and ESAB are the CS and ES measures for a change from QA to QB. For a
decrease in visual quality from state QA to state QC one may define the
measures as follows:
U(QC,M) = U(QA,M - ESAC) = U2 (4)
U(QA,ff) = U(QC,M + CSAC) = UO (5)
where U2 is less than UO. If the exact utility map specification is
known or assumed one can derive the functional form of the ES and CS
functions. For example, as presented in Rowe et al. (1980a), if:
u = aVi + b*Vi*M + C*M * Edi*Xi*Vi -I- fi*Xi
where Vi represents the ith visibility level, Xi is the ith socioeconomic
variable, M is income and a, b, c, di and fi are parameters, the ES and CS
functions can be derived for a proposed visibility decrease from VI to V2.
For example:
(Vj ~ V2) (a + b*M + £di+Xi^ (7)
V]_*b + c
Drawbacks to this approach are that the specification of the sur-
plus functions is contingent upon specifications of the utility function,
which is unknown, and that the individual is seen as not being able
to take mitigating actions to incur visibility levels other than the
current level. If, however, the individual does take mitigating actions
that allow him to choose the visibility level he will incur, EV and CV
measures would be correct and this simplified approach would not be appro-
priate. The benefit of this approach is its simplicity and guidance as to
the likely determinants of the ES and CS values.
Household Production Function Approach'
This approach draws upon the work of Lancaster (1966), Becker
(1965), Muellbauer (1974), Muth (1966), and Pollack and Wachter (1975). It
is based on the assumption that the individual does not derive utility
directly from the consumption of commodities and environmental goods, as in
the standard utility approach, but rather from outputs and experiences that
are produced through a household production process that combines the
market and non-market commodities as inputs. In this sense, visibility is
A good explanation of th<=> current work in this area is found in Eubaaks
and Brookshire (1980".
2-27
-------
not valued in and of itself but rather for how changes in the level of Q
affect the desired experience. For example, air quality at alternative
recreation sites will affect the value to the individual of recreation at
each site and will therefore affect recreation demand at each site.
In the simple case where time constraints are not considered, the
model may be written as:
maximize U(Z1,Z2...,Zn)
subject to F(Zl,Z2...,Zn;XlX2,...,Xm,Q) = 0
m
M = £ PiXi
i = 1
where:
Zj = the jth output or experience, j = l,2,...,n
Xi = ith market input good, 1, = l,2,...,m
Pi = the price of ith market input, i = l,2,...,m
Q = an environmental good
M = the individual's fixed income
U(') = the individual's utility function
F(') = the individual's transformation function, which summarizes
the household technology to produce the Zi's from Xi's
and where the utility and transformation functions are assumed to have
classic properties. The individual's problem is to choose the levels of
final outputs and experiences he desires and the subsequent mix of input
goods and visibility to produce the outputs given market prices and
the household production technololgy. The individual must therefore
maximize utility from the produced goods and experiences and simultaneously
minimize input costs to produce these final goods and experiences from the
market inputs and the environmental good. This maximization/minimization
problem can be solved to yield demand function for the output goods and
experiences.
Zi = Zi (PI, P2,...,Pn,Q,M)
2-28
-------
By analyzing the effects of Q on the derived demand function for
Zi, EV and CV measures for each change in Q can be derived. The introduc-
tion of time constraints can only be made at the cost of more complicated
mathematics and implementation requirements.
In their analysis of the impact of congestion upon recreation
demand, Cicchetti and Smith (1976) note that without specific information
on the forms of the utility function and consumer production technology,
it is impossible to distinguish operationally between a standard utility
framework and the household production framework in terms of testable
hypotheses. Each of the models generates the result that there is a
positive willingness to pay for air quality increases due to associated
utility increases. Without assuming specific functional forms or without
obtaining empirical information about consumer production technology,
nothing more can be said about this relationship.
Expenditure Function Approach^
The utility maximization problem for the individual may be conven-
tionally expressed as:
maximize U = U(X)
N
subject to I PiXi = M
i = 1
where X is the vector of quantities of goals and services (X = XI, . . .,
Xi, . . -,Xn), P is a vector of prices (P = P1.P2,...,Pn), and M is money
income. The solution to this problem leads to a set of ordinary or
Marshallian demand functions
Xi = Xi(P,M) for i = l,i...,N
Maler (1974) for the development of the expenditure function as it
relates to the welfare effects for changes in environmental quality and
Diamond and McFadden (1974) for applications in public finance. Also
see Freeman (1979a) for a slightly expanded version of the material
presented in this section.
2-29
-------
and some maximum utility level Urn. From the ordinary demand function one
may derive ordinary consumer surplus measures. The dual of this problem is
to:
N
minimize E PiXi
i = 1
subject to U(X) >^ UO
Solving this minimization leads to an expenditure function relating
the minimum dollar expenditure necessary, given market prices, to obtain
a specified utility level UO:
E = E(P,UO)
where E is the dollar expenditure, P is the price vector, and UO is the
specified utility level. The solution of this minimization problem yields
a set of income compensated demand curves:
Xi = Xi (P,UO)
The utility level chosen for the basis of comparison, which is one of
the main considerations in the choice of surplus measures, is specified in
the expenditure function. The expenditure function readily explicates the
effect of changes in prices on the EV and CV measures. For example, con-
sider a price decrease for visibility, with price PI, from P10 to Pll
while all other prices are held constant at P = (P2.P3,...,Pn):
CV = E (P10, P, UO) - E (Pll, P, UO)
EV = E (P10, P, Ul) - E (Pll, P, Ul)
where Ul is greater than UO and represents the higher utility level
that results from the price decrease.
The advantage of this approach is that if visibility, Q, enters the
utility function either directly or indirectly it will also enter the
expenditure function. The derivative of the expenditure function, E(P,Q,UO)
= M, with respect to Q, with the appropriate change of sign, yields the
income compensated inverse demand function or the marginal WTP function,
Eq = 3E/90 for changes in Q. Therefore, for changes in the level of
visibility from Ql to Q2, the Riemann integral of the Eq function yields
the appropriate benefit measure:
2-30
-------
Q2
Eq (P1.Q1, UO) dq
Ql
where b = EV or CV depending upon the choice of UO or Ul.
The disadvantages of this approach are again that the exact form of
the expenditure function is contingent upon the form of the unknown
utility function. Special restrictions on the interaction of the elements
are also necessary to separate and estimate an implicit price for the
visibility.
Weak Complementarity
In earlier examples of OCS (Section 2.1.2) and in the previous
models of consumer behavior, if a demand curve for some measurable good X,
such as recreation days or housing, could be estimated as a function of
prices, P, income, M, and environmental quality, Q, as:
X = X(P,Q,M)
the change in the maximum WTP for X as Q changes could often be used to
accurately estimate the maximum willingness to pay function for changes in
Q. This is the case when the conditions of weak complementarity hold.
Weak complementarity, as defined by Maler (1974) and explicated in Freeman
(1979a) involves two conditions: (1) there is some price P2 for which
the quantity demanded of X equals zero, and (2) at price P2 expenditures
do not change with small changes in environmental quality, or for E =
E(P2,P,Q,UO)
3E/3Q = 0
Equivalently, changes in Q do not affect utility at price P2 because Q only
has value in combination with the consumption of X.
Figure 2.11 demonstrates why weak complementarity is necessary in
order to use changes in demand to measure the value of environmental
quality. Assume Dl is the estimated demand curve for recreation days at a
site with air quality at level Ql and assume the price remains fixed at PO.
The OCS measure is the area ABC. Next, assume air quality increases to Q2,
causing the demand for recreation to shift to D2. The benefits of this
change can be calculated in three steps:
2-31
-------
1. Given the original demand curve, Dl, if the price were to
increase from PO to P2, the individual would have to be com-
pensated by income equal to the area ABC to be no worse off.
2. With the condition of weak complementarity, the increase in air
quality to Q2 would not affect utility when the price of X is P2
because the consumption of X is zero.
3. Next, let price return to the original level, PO. The indivi-
dual gains a consumer surplus equal ADE. In order to return to
the utility level obtained when the consumption of X was zero,
this amount of income would need to be withdrawn. The net
effect, or benefit, of the change in Q is therefore the amount
of income equalling the area BDEC.
Price of X
P2
PI
PO
N.A
D2 (Q= Q2)
_X = Recreation
Days
Figure 2.10
Example of Weak Complementarity
As pointed out by Freeman (1979a), with symbols changed for
continuity:
If weak complementarity did not hold, there would be a positive
benefit associated with the increase in Q even though the quantity
demanded of X were zero in step 2 above. In this case, the area
BCED would be an underestimate of the benefits of increasing Q.
2-32
-------
If Dl is the Hicks-compensated demand curve, this area (BCDE) is an
exact measure of the CV or maximum willingness to pay for the
increment to Q. If an expression can be formulated for this area
as a function of Q, the derivative of this function with respect to
Q is the marginal demand price for Q. Typically, Dl is an ordinary
demand curve, not a Hicks-compensated demand curve (page 74).
2-33
-------
CHAPTER 3
INTRODUCTION TO VISION THROUGH THE ATMOSPHERE1
3-1 The Importance of Atmospheric Science Fundamentals to Visibility
Benefits
Everyone would agree that good visibility is desirable, but often
the more commonplace the subject, the more difficult it is to think scien-
tifically and quantitatively about it. Some kind of quantitative measure
or measures of visibility are absolutely necessary in a benefit analysis.
There must also be a known relationship between measures of visibility and
human values, on the one hand, and physical environmental factors on the
other. Only with this type of knowledge can one determine those parameters
to which visibility benefits are sensitive and whether or not they are con-
trollable parts of the environment. For example, would there be any
benefit in improving visual range from 100 km to 125 km when the most
important view in the area is 50 km distant or when visual range is reduced
to less than 10 km by natural fog or rain 70% of the time? Though everyone
understands what visibility is, it becomes a difficult and complex topic
when put in quantitative terms.
In performing a benefit analysis, the analyst may need to use or to
take into account technical information from many diverse areas, such as
photographs, visibility monitoring data or model simulations of visibility
impacts. It is the purpose of this section to provide a sound, fundamental
framework which will allow the analyst to place this information in proper
perspective for use in benefit analysis. This framework must span human
perception, atmospheric physics and chemistry, optics and air quality
meteorology. It is not possible to outline here all the relevant technical
information contained in, for example, Protecting Visibility: An EPA
Report to Congress (EPA 1979). However, this section will introduce the
non-expert to the field.
Visibility impairment analysis is best thought of as a sequential
process as outlined in Figure 3-1. Visibility impairment is caused pri-
marily by fine suspended particles and gases which scatter and absorb
light. The possible anthropogenic (manmade) sources of these pollutants
can be determined by performing an emission inventory- The composition and
distribution of these gases and particles is determined by air quality
monitoring or can be roughly predicted by air quality modeling. The effect
on light viewed by the observer can be monitored by instruments and cal-
culated by established theories. How that light is perceived by the eye is
the study of psychophysics. Interpretation of the perceptions are matters
Sections 3.1-3.7 of this chapter were prepared by Dr. Ronald C. Henry,
ERT, Inc., 2625 Townsgate Road, Westlake Village, California 91361.
3-1
-------
Source of Information
External World
Internal World
Discipline
Emission Inventory
Usual Air Quality
Modeling and/or
Monitoring
Monitoring and/or
Modeling of Optical
Properties
Source
Dispersion and
Atmospheric Chemical
and Physical Changes
Light scattering and
Absorption Processes
Willingness to pay to
improve visibility,
costs of control
Psychological
Interpretation
Human Perception
Processes
Economics
Benefit
Analysis
Psychology
Visual
Psychophysics
Figure 3-1. FRAMEWORK FOR UNDERSTANDING VISIBILITY QUESTIONS
AND THE CONTEXT FOR THE BENEFIT ANALYSIS
-------
of psychology., sociology and anthropology. The value of Che visibility
conditions to humans as measured by willingness to pay to obtain improved
visibility; is in the domain of economics and visibility benefit analysis.
The circle is then closed by decisions of what, if any, sources to control
and how to do it.
The remainder of this section gives the analyst an overview of the
types of information to be gained and inputs required for each step of the
visibility analysis process. In keeping with the current focus of visi-
bility regulations the discussion is, at times, more oriented toward
visibility in the context of point source impacts on Class I recreation
areas.
Factors Affecting Visibility Impairment
The components of the visibility problem can be classified broadly
into physical, perceptual and psychological factors. The physical factors
have received by far the most study. Those which must be determined
for a visibility analysis include:
characteristics of source emissions
- quantities of each type of emitted particles and gases for
various sources, i.e., emissions inventory
- size distribution and chemical composition of emitted par-
ticles
- optical properties of emitted gases and particles
transport and transformations of source emissions
- dilution by dispersion and diffusion in the atmoshphere
- losses by wet and dry deposition
- formation of particles by chemical reactions, e.g., sulfur
dioxide to sulfates, nitrogen oxides to nitrates
- changes in particle size distributions as pollutants dilute and
react
- effects of relative humidity on particle growth
optical properties of polluted and pristine atmospheres
- horizon-sky intensities
- contrast of plumes and natural objects to the horizon
- characterization of existing scattering and absorption proper-
ties of natural and polluted atmospheres
- time of day (solar angle)
3-3
-------
Perceptual and psychological questions have only recently been the
subject of systematic study. Important factors to be considered here
are:
content of the scenic vista
- size and distance of important objects
- overall scenic beauty
- cloud coverovercast or clear
- ground coversnow, rock or vegetation
- time of day (solar angle)
- visible sources of pollution
perceptual thresholds of the human eye-brain system
- lower limits of detection of visible haze and distant objects
- relative limits of detection of changes in brightness and color
of the horizon sky and objects seen through the air
- the degree to which texture (fine detail) is visible inside
viewed objects
- the distance to which natural objects are visible (visual
range)
psychological and other factors
- previous experience and knowledge of the scenic view
- personal background, sex, age, work experience, education,
income level
- reason for visit to Class I areabiking, boating, hang-
gliding, etc.
- political views
Changes in any or all of these affect the good itself, the context of
enjoyment of the good and, thus, valuation of visibility by the observer.
In addition to the EPA Report to Congress cited above, several
guidance documents have also been made available and are listed at the end
of Chapter 1. These documents should be consulted for detailed information
requirements for their respective areas of the visibility analysis. The
Monitoring Guidelines and the Workbook are particularly useful sources of
general information.
Finally, it should be stressed that the term "visibility impair-
ment" is used as "any humanly perceptible change in visibility (visual
range, contrast, coloration) from that which would have existed under
3-4
-------
natural conditions." (45 FR 80091) Thus, reductions in visibility are not
considered visibility impairment unless they are substantial enough
to be humanly perceived. Visibility reductions caused by natural sources
are not considered impairment whatever their magnitude.
3.2 Human Perception of Visibility
"There is light in shadow and shadow in light, and black in the
blue of the sky" (Lucy Larcom, "Black in Blue Sky"). This accurately des-
cribes the process of visibility degradation as experienced by an observer.
When looking at a dark object (or shadow) light is scattered into the eye
by gases and particles in the air and the dark object appears brighter-
Conversely, when observing a bright object (or light) its brightness is
dimmed by loss of light scattered or absorbed by the atmosphere. The net
effect is for all objects near the horizon to eventually blend into the
horizon.
In purely descriptive terms, when looking at an object near the
horizon through haze, the observer notices that the object or its texture
(fine detail) may be obscured or its colors changed; the haze itself may be
visible as a dark or light layer which may be colored. Since the effect of
haze is for both the object and its background to approach the horizon
brightness, the contrast between them decreases, making objects more
difficult to see. Thus, haze affects a human observer by changing the
brightness of objects, their texture and colors. If conditions are right,
one may be able to perceive the haze itself. It may appear as a coherent
plume or several visible layers (so called "plume veil") or the haze may be
visible as a general or regional haze. The visibility regulations currently
emphasize only those cases of impairment which can be unambiguously assigned
to a single source. Therefore, the case of a visible plume or plume veil
will be of immediate interest.
It is important in this section to reserve the terms "brightness"
and "color" for the humanly perceived sensations of lightness and color.
This is standard psychophysical terminology (Wyszecki and Stiles 1967).
Brightness and color are contents of consciousness and cannot be measured
by an instrument. The physical antecedents of brightness and color, which
can be objectively measured, are the amount of light energy entering the
eye and its distribution among visible wavelengths. A later section will
focus on the relationship between perception and their physical antecedents.
This section will be devoted to discussion of perceptual clues of visibility
impairment in the environment and how they are processed by the eye-brain
system. The appropriate results of three recent studies will be heavily
relied upon (Malm et al. 1980 a, b; Latimer et al. 1979).
3-5
-------
In valuing visibility improvements it is important to understand
the effects of perceptual thresholds, content of the scenic vista, and
psychological and other factors so that they do not unintentially affect
the valuation (as discussed in Section 2.4). As this topic is less
thoroughly discussed in the Report to Congress and other guideline docu-
ments, it will receive extended discussion below.
Perceptual Thresholds
A threshold is defined as "a point at which a physiological or
psychological effect begins to be produced." (the Merriam-Webster Dic-
tionary.) Without a visibly perceivable effect there can be no benefit
measure derived for enhanced visibility aesthetics from changes in air
quality (although other benefits, such as health and materials impacts, may
occur). Therefore, perception thresholds are basic to the quantification
of the good. A later section looks at this question of perceptual thresh-
olds using various visibility indices. At this point, some of the more
general aspects of thresholds are noted. Most important is to realize that
for perception of visible pollution there are at least three thresholds:
absolute thresholds, incremental thresholds and identification thresholds.
Absolute thresholds are the limits of perception below which there is no
sensation. An example of an absolute threshold is the often used assump-
tion that if the contrast between the sky and a large black object is below
0.02 then the object disappears. A contrast of 0.02 means that the dif-
ference between the light intensity coming from the object and the sky is
2% of the intensity coming from the sky. Contrast is especially important
because it is contrast in brightness or color that is actually perceived.
An absolute threshold for contrast is also called the liminal contrast.
An incremental threshold is linked to the concept of a just-
noticeable difference. In this case, we are concerned with the minimum
change that can just be noticed. For example, if the contrast of an object
on the horizon is 0.20 and a decrease to 0.10 is just noticeable, then its
incremental threshold is 10%. This is not an absolute threshold, since
the object remains visible. These two types of thresholds are commonly
discussed in the psychophysical literature.
A third threshold is unique to air quality studies. This is the
level of change necessary to assign a source to the impairment. If a plume
is a visible layer, at this point the observer can assign the visual
impairment to this observed layer. If conditions are right, he may be able
to distinguish the emission source responsible for the plume or haze, e.g.,
campfires or an industrial plant.
In general, each of these thresholds depends in a complex way on
many factors, including the age and experience of the observer, ambient
lighting conditions and content of the scene, as discussed below.
3-6
-------
Content of the Scene
Some contextual factors that are known to affect perception of
visibility are obvious. Size and distance of objects of interest are
important, since small objects are harder to see, or have larger thresholds
(Blackwell 1946), and the further away they are the greater the intervening
mass of the atmosphere. Effects of cloud cover, sky color, ground cover
and time of day are less intuitive, but are known to exist. Even less
understood is the impact of scenic beauty on the perception of visibility
impacts.
Cloud cover can vary from clear skies to dense fog. Obviously,
clouds associated with rain or fog will seriously reduce visibility. Less
obvious is the potential importance of cloud shadows. These may shade an
otherwise bright mountain or make a dark background for the same mountain.
Cloud shadows can reduce the light scattered into the eye and increase
visibility. Finally, clouds add interest and scenic beauty to most vistas.
Two recent studies have found clouds to be an important aspect of the
visual environment, Malm et al. (1980a) and Latimer et al. (1980). Both of
these studies ask observers to rate the visual air quality of a series of
slides of scenic vistas. The Latimer study also ranked scenic beauty.
Although they do not report quantitative estimates of cloud cover effects,
both reports identify it as important. The Malm et al. report uses cloud
cover as a basis for stratifying these data. Similarly, both reports
indicate the importance of ground cover, i.e., rock, vegetation, snow, or
some combination of these. Whether or not snow is present on a mountain is
an important means of classification for Malm et al. (1980a).
The time of day and direction of the sun with respect to the view
(solar angle) have been identified as important to the appearance of a
scenic vista. Solar angle influences the depth and direction of shadows.
Views looking at a small angle from the sun are much influenced by bright
forward scattered sunlight.
Psychological and Other Factors
Previous experience or knowledge of a scene can drastically alter
ones perception. If a visitor does not know there are any mountains to be
seen, then he does not miss them. Thresholds are greatly affected by
experience. A mountain may be just visible to an experienced observer who
knows just where to look and what to look for, but it may be totally
invisible to the casual observer.
Other factors, such as work experience, education, income and
political views, potentially influence an observer's view. The Latimer et
al. (1980) study compared subjective ratings of Visual Air Quality Index
(VAQI) (scale of 1 to 10) of the same 20 slides of western vistas for 13
3-7
-------
special interest groups ranging from air pollution professionals and oil
company staff to an art class and EPA personnel. An intergroup correlation
of 0.48 was the lowest foundbetween a group of 26 Spanish speaking
observers and 12 persons in an art class. Most groups showed correlation
of VAQI ratings of 0.8 and above; but there were a number of lower corre-
lations. There does appear to be some indication of important differences,
although the Latimer report concludes that there was not any clear-cut case
to be made for such differences. This report does claim, however, that the
overall scenic beauty of a vista can influence the visual air quality
estimates.
The Malm et al. (1980a) report examined the effect of age, educa-
tion, sex, frequency of visits to National Parks and location of residence
on estimates of visual air quality. Again, this study relied on photo-
graphic slides. The data consisted of repeated subjective estimates of
visual air quality (scale of 1-10) of the same slide projected among
other slides. This control slide had a visual air quality index (VAQI) of
about 5. There was a mean difference of 0.3 in VAQI rating between men and
women. For the number of observations (426 and 305), this is significant
at above the 99% level by the usual t-test. Some other differences in
visual air quality perception are evidenced between age groups. The 18-24
group rated the control slide significantly higher than the 45-54 age
group. On the other hand correlations between demographic groups were 0.98
or higher while within group correlations always exceeded 0.81. Most
importantly, all correlations are very high and positive, indicating very
good agreement both within and between demographic groups. The researchers
conclude that there is not a difference between groups of observers.
To complete this review of current work in the area of human
perception of the visual environment, mention must be made of the percep-
tual studies carried out in Denver, Colorado, by personnel of the National
Center for Atmospheric Research (NCAR) and the Institute for Behavioral
Science at the University of Colorado. The NCAR study is reported in
Mumpower et al. (1980). It is the only recent study that has systema-
tically examined the relative importance of various perceptual clues in the
natural environment. This study does not rely on photographs as do the
other studies. Ratings of the visual air quality were made along with
subjective estimates of several perceptual clues: distance at which
objects were visible, clarity of the air, color of the air and existance of
a distinct border to the haze. Observations were made by the same indivi-
duals several times a day in each of the four directions. These observers
elicited perceptual judgments from passers-by, referred to as respondents.
The study was carried out in summer and repeated in slightly modified form
in winter. In all cases, the perceptual clues which correlated most highly
with judgments of subjective visual air quality were subjective clarity of
the air and distance to which objects could be seen. Visible haze color
and border had more effect in winter than in summer. The correlation
between perceptual clues as judged simultaneously by regular observers and
passer-by respondents was often very low. This indicates that perceptions
of the man-in-the-street are rather different than a trained, experienced
observer.
3-8
-------
Finally, visibility of texture, i.e., fine detail inside the border
of objects, is an important factor in judgment of clarity and distance.
Although this factor has not been quantitatively studied as yet, Malm et
al. (1980a) and Henry (1979a) hypothesize its potential importance. A
group of landscape arhitects for the U.S. Forest Service have also identi-
fied texture as important to perception of visibility (Paulson 1979).
Summary
In assessing human perception and its relation to visibility
benefits, several important points stand out.
The limits of perceptions, thresholds, must be defined quanti-
tatively. Thresholds are different for disappearance, noticeable
changes, and source assignment.
Perception of visibility is influenced by the content of the
scene, clouds, snow cover, scenic beauty, vegetation, etc.
« There is evidence that estimates of visual air quality from
slides does not depend strongly on age, sex, and special in-
terests of the observers.
Major perceptual clues to judgments of visual air quality in an
urban area (Denver) are clarity of the air and the distance to
which objects can be seen. The color and noticeable border of
the haze are of less importance.
Two major areas of concern to human perception will be treated in
other sections. The thresholds of various indices of visibility is
discussed in Section 3.4. The importance of adaptation of the visual
system to its environment is central to Section 3.6, which discusses
various presentations of the visual environment.
3.3 Sources of Visibility Impairment
This section covers sources of visibility reducing components of
the air. Natural variations in visibility and their causes will be ad-
dressed in Section 3.7 on visibility monitoring. The importance of meteor-
ology and atmospheric chemistry in producing particles from gases is
stressed.
3-9
-------
Any human activity that produces airborne particles, or gases that
can react to form particles, can contribute to visibility impairment.
Particles the size of bacteria, 0.1 to 1 micrometers (urn) in diameter, are
by far the most efficient on a per unit mass basis in producing visibility
impairment. Sources of the colored gas nitrogen dioxide or its chemical
percursors are also potential sources of discoloration. In addition,
interaction of light and nitrogen dioxide is a key link in the chain of
reactions that form photochemical smog and its associated particulates.
Atmospheric chemistry plays a key role in understanding the sources
of visibility reduction. The fact is that in most urban and rural areas,
the majority of the mass of particles in the efficient light-scattering
size range are not directly emitted by a source, but are the product of
reactions occurring in the atmosphere involving invisible gases. Secondary
particulates are those formed by atmospheric reactions, while primary
particulates are those directly emitted by sources. Gaseous emissions
which contribute to the production of secondary particles are sulfur
oxides, in nitrogen oxides and other organic vapors. Atmospheric reac-
tions can also produce new gaseous species, such as ozone.
The atmospheric chemistry of pollutant gases is extremely complex.
For example, concentrations of the red-brown gas nitrogen dioxide are
determined in part by the associated concentrations of hydrocarbons, ozone,
and nitric oxide, as well as ambient temperature, intensity of sunlight
and moisture content of the air. Because the effects of atmospheric
dilution add to the complexities of atmospheric chemistry, predicting or
modeling of source-receptor relationships for secondary particles and gases
is extremely difficult. Unfortunately, the majority of visibility-reducing
particles in the air, those in the size range of 0.1 to 1.0 pm diameter,
are the product of atmospheric chemical reactions and thus not subject to
direct control methods.
Both natural and man-made sources produce primary visibility-reducing
particulates and gases which convert to secondary particulates. The
natural causes will be discussed later. Table 3.1 summarized sources of
pollutants affecting visibility. The major anthropogenic sources of
visibility-reducing particulates are combustion sources. Combustion
produces primary fine particles, such as coal fly ash or diesel exhaust
particultes. Combustion sources also emit reactive gases that convert to
particulates in the atmosphere. Also, any high temperature combustion
using air will produce some nitrogen oxides, with a certain portion being
or becoming nitrogen dioxide. In addition, some industrial process sources
produce visibility-reducing particles. For example, primary smelting of
ores involves roasting of sulfide ores with resultant large emissions of
particles and sulfur oxides.
3-10
-------
Agricultural and forestry activity is another major source type.
Emissions of soil dust from ground preparation are large but of short
range, usually. Application fertilizers and pesticides can affect air
quality by adding nitrogen oxides and hydrocarbons to the atmoshphere. An
important land management tool is prescribed burning, whereby intentionally
ignited fires are allowed to burn under controlled conditions for manage-
ment objectives, including hazard reduction, forestry and wildlife manage-
ment, and grazing. Such burning produces emissions of gases and particu-
lates that may cause temporary visibility degradation.
Table 3.1: Manmade Sources of Visibility Reducing Gases and Particles and
Their Precursors
Source Type
Examples
Composition of Emissions (in
rough order of importance)
Industrial process
and stationary
combustion
Mobile sources
coal-fired power
plants, smelters,
refiners, synfuel
plants
autos, diesels,
trains, aircraft,
boats
sulfur oxides, nitrogen ox-
ides, primary particulates,
hydrocarbons
nitrogen oxides, hydrocarbon
gases, elemental carbon,
primary particulate, sulfur
oxides
Residential sources
Agricultural and
forestry sources
gas water heaters,
oil burners
soil preparation,
field burning,
prescribed
burning
nitrogen oxides, elemental
carbon (soot), hydrocarbons,
primary particulates
soil dust, nitrogen oxides,
elemental carbon, hydro-
carbons, sulfur oxides,
primary particulates
3.4
Definitions and Indices of Visibility Impairment
To quantitatively estimate benefits associated with increments in
visibility, a measure or measures of visibility are needed. Also, moni-
toring technology must rely on objective measures of visibility. Since
only humanly perceptible increments are defined as impairment, the relation-
ship of visibility measures to human perception must be known. If mathe-
matical simulations are used to relate emissions to physical measures, the
dependence of visibility indices on mathematically predictable parameters
must be established.
3-11
-------
In this section commonly used visibility indices will be defined
and discussed. What is known of their connections to human perception and
physically measureable quantities will be outlined. No one visibility
measure is appropriate for all times and locations. Hard and fast criteria
for levels of perceptible changes are not at hand and may never be. The
best that can be done at present is to present an array of measures to
select from on the basis of sound technical advice.
Definitions of Visibility Indices
In the past there has been much confusion because of imprecise use
of visibility-related terms. In this section we will define the most
commonly used indices. It is first of all very important to distin-
guish three types of visibility-related indicesthose directly related
to human perception, those which are calculated from measures of light
intensities, and those which are measures of optical properties of the
atmosphere and airborne particles. Table 3.2 gives the commonly used,
measured or calculated indices separated by this criterion.
Table 3.2: Commonly Used Indices of Visibility
Category
Examples
Reference
Measures of human
perception
Measures of light
intensities
Measures of optical
properties of air
and airborne
particles
airport visual range
Visual Air Quality
Index
contrast, blue-red
luminance ratios,
color contrasts,
C.I.E. chromaticities,
C.I.E. combined color
and brightness scales, AE
particle light-scattering
coefficient, particle
light-absorption co-
efficient, total light-
extinction coefficient,
standard visual
range
Meteorological Handbook
No. 1, (U.S. Dept. of
Commerce, 1975)
Craik (1979), Latimer
et al. (1980), Malm et al.
(1980a)
EPA (1978), EPA (1979),
Malm (1979)
Middleton (1954), U.S.
EPA (1979)
3-12
-------
Direct Measures of Human Perception
Currently only two indices unambiguously related to the human
observer have been routinely measured. At some airports observers hourly
estimate the maximum distance to which they can see in eight directional
sectors. The prevailing visibility is determined from these observations
and is the greatest distance to which large objects can be seen, which is
equalled or exceeded over half of the horizon, not necessarily uninterrupted.
This is also called airport visual range. It is important to distinguish
this human observer determined visual range from standard meteorological
range and standard visual range, defined below.
The usual definition of meteorological range is the distance at
which the contrast of a black object to the ideal horizon is 2%. Since
standard visual range is a somewhat artificial construction, it need
not indicate the greatest distance at which real objects may actually be
seen. Common experience has shown that this definition of visual range
overstates the distance to which natural objects can be seen. This is
because natural objects at a great distance are usually small, and small
objects need a contrast greater than 0.02 to be seen. Also natural objects
are never black. As a rule-of-thumb, one may expect to be able to observe
natural objects out to about 75% of the meteorological range. (Snow-
covered mountains are visible for longer distances because of their
larger contrast.) If, in the study area, the geometric mean standard
visual range is about 200 km, then targets up to 150 km should be typically
observed. Meteorological range is theoretically related to the extinction
coefficient of the atmosphere, a subject covered later.
The U.S. EPA and NPS have defined a further variant of meteoro-
logical range, the standard visual range, which takes into account the
dependence of visual range on altitude. Visual range increases with the
altitude of the observer because the air through which light passes becomes
less dense. As some Class I areas are at high altitudes and some at sea
level, the meteorological ranges cannot be directly compared. The standard
visual range overcomes this problem by normalizing the meteorological range
to an altitude of about 2,000 m. A precise definition is found in U.S. EPA
(1980a).
A number of investigators have used airport prevailing visibility
to determine long-term trends and geographical distribution of visibility
(Trijonis 1979, Latimer et al. 1978). Trijonis has used airport visibili-
ties in the Southwest to assign sources to visibility impairment.
The second direct measure of human perception of visibility, a
Visual Air Quality Index (VAQI), is a simple subjective rating of the
visual air quality on a scale, usually of 1 to 10. The Denver study
(Mumpower 1980) elicited such subjective responses directly. The recent
NPS/EPA studies of Malm et al. (1980a, b) had observers determine VAQI's
from slides of scenic vistas. Similarly, an American Petroleum Institute
study of Latimer et al. (1980) used VAQI's from slides.
3-13
-------
The major problem with both humanly observed visual range and
VAQI's are the subjective, unreproducible nature of the measurement. Malm
et al. (1980a) found that for a midrange slide of VAQI about 5, 98% of the
respondents assigned a value to it in the range of 2-8a rather large
spread. This means that for the same scene subjective ratings of 2 to 8 on
a scale of 10 can be expected from different observers. Airport visual
range is similarly hampered by subjective determinations as well as dif-
ficulty in finding suitable targets.
More serious, since visibility is a subjective phenomenon, is the
difficulty in relating VAQI's and airport" visual range to objective physi-
cal parameters. The studies of Malm et al. (1980a, b) and Latimer et al.
(1980) attempt to answer these questions, but with equivocal success. For
this reason it is difficult to relate results of mathematical models to
either of these measures. However, Malm et al. (1980a) concluded that
VAQI's of isolated scenic elements decreased exponentially with the product
of extinction coefficient and distance to the scenic element. Furthermore,
it was argued that the perception of a scene containing many scenic elements
is equal to the combined perceptions of isolated scenic elements weighted
in accordance with the fraction of the total area subtended by each specific
scenic component. This study further concluded that the sensitivity of a
scene to increases in air pollution was not affected by variation in the
kind or amount of foreground features, nor was the sensitivity affected by
cloud cover.
Finally, it is very difficult to establish thresholds for these
measures since controlled experiments are almost impossible to conduct.
Using slides Malm et al. (1980a) have made an attempt at determining a
human perception threshold. However, this work is of a research nature and
needs further study and review.
While the idea of VAQI does not lend itself to the determination of
a threshold, a guess as to a reasonable threshold for visual range would be
-h. 05 km as a perceptible change for an object near its limit of detection.
This is not a percentage change, i.e., for an object of contrast -0.15 a
noticeable degradation might be to a contrast of -0.10. This is based on
experience with a large number of slides of scenic vistas in the Southwest
with associated objective measurement of target contrast.
Measures of Light Intensity
In an attempt to bridge the gap between human perception and the
world of physically reproducible measures, much attention has been focused
on measurements of light and various measures calculated from these obser-
vations. Also, several recent visibility models (U.S. EPA 1980c, Drivas et
al. 1980, Williams et al. 1979) predict light intensities for several wave-
lengths (colors) of light. Thus, these visibility indices based on light
measurements offer the hope of connecting models with human perception of
visibility degradation. The vital link between light and perception is the
realm of visual psychophysics and its application to the natural environ-
ment is not simple.
3-14
-------
In this work, "brightness" will refer to the humanly perceived
sensation of lightness and darkness and ''color" to the humanly perceived
sensation of color. However, these human perceptions cannot be measured by
instruments. The physical antecedents of brightness and color, which can
be objectively measured, are the amount of light energy entering the eye
and its distribution among visible wavelengths. The multiwave-length
telephotometer or, more accurately, teleradiometer, is the commonly used
instrument which measures the light energy, or radiance, coming from
a distant area in several narrow wavelength bands. Physical contrast (or
simply contrast) is determined by the ratio:
where;
C(x) = contrast of the object to background at wavelength \;
Lo(X) = light intensity of the object at wavelength A; and
LB(X) = light intensity of the background (usually sky) at
wavelength X.
The relationship between perceived contrast (relative lightness and dark-
ness) and the measured physical contrast has never been directly studied in
the natural environment. Extensive laboratory studies (Cornsweet 1970)
have shown that the relationship is not simple but is a complex function of
ambient lighting and spatial interactions between the object (or its
texture) and its background.
The dependence of contrast on wavelength is important, and usually
green light of wavelength 550 nanometers (nra) is specified in the definition.
Contrasts of grey objects are between 0 and -1 and are often quoted as a
percentage, e.g., -10%, -20% contrast. Bright objects, such as snow-covered
mountains or bright rocks can be brighter than the sky, they have contrasts
that are positive and can range as high as +4 or +5.
Blue-red ratios are simply defined as the ratio of blue light
intensity to red light intensity. The wavelengths most often used are 450
run for blue and 650 nm for red. In most applications the blue-red ratio is
normalized to the blue-red ratio of the horizon, i.e., the blue-red ratio
of the horizon is defined to be 1.0.
Chromaticities and combined brightness and color difference formulas,
J.E, are functions of light intensities over several wavelengths which have
been related to color perception under controlled laboratory-like conditions.
The definitions are highly technical and can be found in texts such as
Wysyecki and Stiles (1970), or for a brief account, see Latimer et al.
(1979) or the EPA Report to Congress.
3-15
-------
The previously quoted studies of Malm et al. (1980b) and Latimer et
al- (1980) have made considerable effort to relate these measures to human
perception of visibility as judged by VAQI estimates from slides. No clear-
cut thresholds emerge from these efforts, however. In general, absolute
thresholds of 0.02 for contrast, -0.1 for blue-red ratio and +4 for AE are
acceptable but perhaps conservative estimates. Since AE is calculated
from chromaticities, no separate thresholds for chromaticities are
needed.
Measures of Optical Properties of the Air and Airborne Particles
Other measures of visibility that are often mentioned are light-
scattering coefficient, light-extinction coefficient, and light-absorption
coefficient (Middleton 1954). These concepts are twice removed from the
human perception of visibility, since they do not describe either human
sensations or light itself. They are properties of the atmosphere and
describe how light interacts with the gases and particles in the atmosphere.
Light passing through the atmosphere is either absorbed, scattered, or
transmitted unchanged. The extinction coefficient is the fraction of
incident light which is not transmitted by a small unit length of the
atmosphere, i.e., an extinction coefficient of 0.04 km-1 means that over
a 1 km path length, roughly 4% of the incident light would be absorbed or
scattered by gases and particles and 96% would be transmitted.
It is interesting to note that for an important special case,
contrast is reduced by the same law as light intensity; i.e., in the
example above, the contrast of an object would be reduced about 4% in each
kilometer of distance. This is the case of horizontal views of objects
against the horizon. The atmosphere must be homogeneous and the curvature
of the earth is ignored. This last constraint is not well satisfied for
views over 75 km. Also, many objects are not seen against the true hori-
zon, but against a higher portion of the sky or other objects. Still, this
law of contrast change, known as Middleton's Law (Middleton 1952), is a
very useful tool. Middleton's law can be used to derive a relationship
between standard visual range and the extinction coefficient of the at-
mosphere; this is called Koshmieder's Law, which states that the visual
range is 3.9 divided by the extinction coeffienct.
The total extinction coefficient of the atmosphere has four com-
ponents: scattering and absorption by both gases and particles. Each
wavelength of light is scatteredd and absorbed in differing amounts. In
the following paragraphs, each of these components is discussed.
Pure scattering of light by colorless gases (such as oxygen and
nitrogen, 99% of the atmosphere) is known as Rayleigh scattering; it is
about four times larger for blue light than red light. For this reason the
sky overhead is perceived as blue; pure white sunlight is composed of all
colors of light, and the blue portion of the spectrum is efficiently
scattered by the atmospheric gases, giving a diffuse blue color to the
3-16
-------
visible sky, while the red and yellow light is, for the most part, trans-
mitted completely through the atmosphere, giving sunlight a warm golden
color. At sunset, when the sunlight must travel through a greater atmo-
spheric depth because the sun is on the horizon, the sky acquires a reddish
tint due to decreased transmittal of yellow and green and increased scat-
tering of red light. Since Rayleigh scattering depends on the number of
molecules in the air, it also varies with ambient temperature and pressure.
For green light (wavelength of 550 nm) about 2,000 m above sea level, it is
0.01 km-1.
The only ambient gas able to absorb visible light is nitrogen
dioxide. It affects blue light to a much greater extent than other wave-
lengths. In high concentrations, it will make white light look red-brown.
In plumes of modern power plants, nitrogen dioxide can be the dominant
cause of visibility impairment.
Particle light-scattering is usually the major physical cause of
visibility degradation by pollution. Small particles, in the size range of
bacteria (0.1 to 1.0 pm diameter) are by far the most efficient in scattering
light and reducing visibility on a per unit mass basis. Unfortunately,
these particles are in the size range that keeps them suspended in the
atmosphere for most of the time. Pollution aerosols can also cause dis-
coloration by preferential backscattering of red light (Husar and White
1976).
Particle light-absorption is typically only 5 to 10% of total
extinction unless significant amounts of elemental carbon are present
in the suspended particles. In some urban areas, e.g., Denver, the par-
ticle absorption coefficient can, at times, equal or exceed the scattering
coefficient (Heisler et al. 1979). Because absorption reduces the total
amount of light energy transmitted through the atmosphere, it produces a
grey haze which is often particularly offensive to observers. As pointed
out in Henry (1979a), a low, grey haze layer seen next to a deep blue sky
(a situation which often occurs in the West) may take on the complementary
color, a yellow-brown. This color is produced by the overcompensation of
the eye to the blue sky.
3.5 Visibility Monitoring and Natural Conditions Affecting Visibility
To those assessing visibility impairment and the benefits of
improved visibility; monitoring existing visibility and understanding the
natural conditions which can affect visibility are critical. This is
emphasized in the visibility regulations. The determination of significant
impairment must take "into account the geographic extent, intensity,
duration, frequency and time of the visibility impairment, and how these
factors correlate with (1) times of visitor use of the mandatory Class I
Federal area, and (2) the frequency and timing of natural conditions that
reduce visibility." (Federal Register, Vol. 45, No. 233, p. 80091).
Currently, monitoring is the only reliable source of information concerning
frequency of occurrence, duration and geographical extent of visibility
impairment and natural variability of visibility.
3-17
-------
There are many examples of the importance of these concepts.
Improved visibility at night or during off-season periods is of limited
benefit. In some areas, natural fog or haze may limit the noticeable
impairment of visibility. Also, long-term monitoring in pristine areas is
the only obvious way to directly determine baseline natural conditions and
the range of natural variation in visibility.
In this section, a brief overview of visibility monitoring tech-
nology and programs are presented followed by a brief discussion of natural
impediments to visibility.
Visibility Monitoring Technology
The output of a monitoring program of greatest utility to those
engaged in benefit analysis are the tables and graphs showing the frequency
of occurrence and behavior of visibility-related parameters. The methods
and associated parameters, advantages, limitations and uses are summarized
in Table 3.3, which is taken from Protecting Visibility; An EPA Report to
Congress (EPA, 1979). Detailed recommendations for a visibility monitoring
program can be found in the EPA's Interim Guidance For Visibility Monitoring
(EPA 1980a).
Most of the methods and parameters have been discussed in Section
3.4. However, some discussion of the importance and use of the airborne
particulate measurements is appropriate. The Guidelines for Visibility
Monitoring recommends separate collection of coarse, greater than 2 pm, and
fine, less than 2 ym, diameter particles and some analysis of their chemi-
cal composition. This is because airborne particles typically exist in at
least two mass modes, which tend to have fairly distinct sources and
effects on visibility.
The fine particle mode (<1 or 2 pm) consists chiefly of primary
combustion particulates or particles formed from combustion generated gases
and other man-made sources. Sulfates, organic matter and nitrates are
important contributors to fine particle mass. Elemental carbon (soot) and
industrial process emissions are often in this size range. Since these
particles efficiently scatter or absorb light, they are a very important
and are often the dominant source of visibility impairment.
The coarse particles, greater than 1 or 2 urn diameter, are prin-
cipally derived from soil dust. In rural areas this can be natural wind-
blown dust or dust raised by vehicles or farm equipment. Unless a dust
storm situation exists, the particles in this mode are seldom responsible
for more than 10% of the particle light-extinction coefficient.
3-18
-------
TABLE 3.3: VISIBILITY MONITORING METHODS
Method
Hi cm,in observer
Parameters Measured
Advantages
Perceived visual quality
Atmospheric Color
Plume bl ight
Visual range
flexibility, judgment;
large existing data base
(airport visuul range).
Limitations
Labor intensive; variability
in observer perception;
suitable targets for visual
range not generally
avallable.
Preferred Use
Complement to instru-
nental observations;
areas with frei|uunt
plume blight, discolor-
ation; visual ranges
uviii luble target
distances.
Integrating ncpliclometer
I
I'
^o
Scattering Coefficient
(b ) at site
scat
relatablti to fine aerosol
Continuous readings;
unaffected by clouds.
night; b directly
neglectssiStinctlon from
night; b directly
relatabliC?o fine aerosol
concentration at a point;
seni-portable; used in a
number of previous studies;
sensitive models available;
automated.
Point measurement, requires
assumption of homogeneous
distribution of particles;
distribution of particles;
neglects extinction from
absorption, coarse particles
>J to 10 urn; must consider
humidity effects at high III).
Areas experiencing
periodic well mixed
general huze; medium
general naze; medium
to short viewing
distances; small
absorption coefficient
(babs); relating to
point composition
measurements.
Mnlt iwavelength
t e I erad i cuue t er
Sky and/or target
radiance, contrast at
various wavelengths
Measurement over long view
path (up to 100 km) with
suitable illumination and
target, contrast truns-
mittance, total extinction,
and chromatic! ty over sight
path can be determined; in-
cludes scattering and
absorption fion all sources;
can detect plume blight;
automated.
Sensitive to illumination
conditions; useful only
in daylight; relationship
to extinction, aerosol re-
lationship possible only
under cloudless skys; re-
quires large, uniform
targets.
Areas experiencing
mixed or inhonogcneous
haze, significant
fugitive dust; medium
to long viewing dis-
tances (1/4 of visual
range); ureas with
frequent discoloration;
horizontal sight path.
'I ransnii ssoinct c r
Long path extinction
coefficient (b )
Measurement over medium view
path (10-25 km); measures
total extinction, scattering
and absorption; unaffected
by c louds , night .
Calibration problems; single
wavelength; equivalent to
point measurement in areas
with long view paths (50-
100 km); limited appli-
cations to date sti11
under development.
Areas experiencing
periodic mixed general
haze, medium to short
viewing distance areas
with significant
absorption (babs).
-------
TABLE 3.3: VISIBILITY MONITORING METHODS (Continued)
Method
Parameters Measured
Advantages
Limitations
Preferred Use
Photography
OJ
I
Particle samplers
Ili-Vol (llif-.li Volume
Air Sampler)
Cascade impactor
Dichotomous and
fine particle
samplers (several
fundamentally
different types)
Visual quality
Plume blight
Color
Contrast (limited)
Particles
TSP (Total Suspended
Particulates)
Size segregated
particles
(>2 stages).
Fine particles
(<2.5 urn) coarse
particles (2.5 to
15 Mm) inhalable
particles (0 to 15 urn)
Related to perception
of visual quality;
documentation of vista
conditions.
Permit evaluation of
causes of impairment.
Large data base,
amenable to chemical
analysis; coarse
particle analysis.
Detailed chemical, size
evaluation.
Size cut enhances reso-
lution, optically im-
portant aerosol analysis,
low artifact potential,
particle bounce; amenable
to automated compositional
analysis; automated
versions available; large
networks under development.
Sensitive to lighting
conditions; degradation
in storage; contrast
measurement from film
subject to significant
errors.
Not always relatable
to visual air quality;
point measurement.
Does not separate sizes;
sampling artifacts for
nitrate, sulfate; not
automated.
Particle bounce, wall
losses; labor intensive.
Some large-particle pene-
tration; 24-hour or
longer sample required
in clean areas for mass
measurement; automated
version relatively un-
tested in remote locations.
Complement to human
observation, instru-
mental methods; areas
with frequent plume
blight, discoloration.
Complement to visi-
bility measurements.
Not useful for visi-
bility sites.
Detailed studies of
scattering by particles
<2 urn.
Complement to visi-
bility measurement,
source assessment
for general haze,
ground level plumes.
l:rom Protecting Visibility, An EPA tteport to Congress, p. J-5 (U.S. EPA 1979).
-------
Visibility Monitoring Programs
Historically, visibility monitoring has centered on airport visual
range. For the past ten years visibility monitoring as part of air quality
studies has been a case of running a nephelometer in an urban area. Only
recently has monitoring directed at air quality and visibility in pristine
areas begun. In the next section some of the results of the pioneering
Cedar Mountain study in central Utah are discussed. This study was begun
in 1976 by the National Oceanic and Atmospheric Adminstration (NOAA) and
the EPA. An extensive network of visibility monitoring sites has grown as
part of the VIEW (Visibility Investigative Experiement in the West) which
is a joint endeavor of the National Park Service (NPS) and EPA. Some
visibility data from this network begins in September of 1978. The periodic
reports, e.g., Walther and Newburn (1980), summarizing the teleradiometer
measurements of contrast and calculated visual range are a unique source of
information on frequency of occurrence, time, extent and intensity of
variations in visibility in the VIEW study area, shown in Figure 3.2.
Figures 3.3 and 3.4 are examples of the available information from the VIEW
program. In the next section natural variability in visibility and its
causes are discussed.
Visibility and Natural Conditions
Field studies have indicated the importance of large scale meteor-
ology patterns in determining visibility and natural variation in visi-
bility. Besides determining the direction of transport of pollutants and
their dilution, meteorology determines large-scale temperature, pressure
and humidity patterns. Data from the Cedar Mountain Study, in central
Utah, (DeNevers et al. 1979) have shown that variations in visibility are
to a large extent determined by these large-scale variables. Similarly,
widespread low visibilities in the eastern United States ar? often ac-
companied by warm moist air which originates in tropical seas (Husar et al.
1979).
Variations in synoptic meteorology represent the large driving
force behind natural variation in visibility. Latimer et al. (1978) uses
airport visibility data to estimate long-term trends in visibility and
geographical distribution of average airport visual range. The map of
existing visual range, Figure 3.5, shows that, in general, the West has the
best visual range. One must remember that this use of airport visibility
data is quite different from that for which it was designed and obtained.
There are many ambiquities and pitfalls to beware of interpreting this
data. But it does offer the only long-term record available of a visi-
bility-related parameter.
Some natural sources produce primary visibility-reducing particu-
lates and gases which convert to secondary particulates. Table 3.4 lists
some natural sources of reduced visibility. Soil dust consists mostly of
particles greater than 1 urn and larger which are outside the most efficient
light-scattering size range. In sufficient quantity, as in a dust storm,
these particles can cause significant visibility degradation. Quasi-natural
dust from roads or agricultural activities can be noticeable, but since
these particles settle out rapidly, their effect is usually local.
3-21
-------
UTAH
I
I
I
I
f
OINOSAUR COLORADO "I
I
CANYON LANDS I
BRYCE "« I
CANYON "^^^
I MESA VERDE ,
I
I
/ ^s_^< I
/-AQ I MESA V
- ' . \NAVAJO I C&L
I ARIZONA -X*1 . __ _LJL^
1 / '
^
GRAND C,
(
^CHACO CANYON
I GRAND CANYON^
BANDELIER
WHITE SANDS |
I I
TEXAS
I \
PH-» TELEPHOTOMETER VIEWS
LEGEND
^+~
-------
u
I
"\
o
I!
R
H
H
G
E
K
M
60C"
400
700
166
SO
e
50
46
30
20
10
o MOUNT ran
NUMBER OF STHHDnRO
3 2 I
lOH." FF'OM MEDIAN
1
J I
J I_J_L_L_1_L
I
LOCATION 4
SPRING 1379
TIME TOTAL
t _ I _ !
.01 .1
5 10 2030 50 rOSO ?0 '35
CUMULA T I UE FREQUENCY
9. 99
Figure 3-3. Cumulative Frequency of Occurrence for Standard Visual
Range derived from teleradiomcter observations of Mavajo
Mountain 140 Ion distant fron Bryce Canyon National Park.
(Walther and Newburn 1930)
3-23
-------
LOCATION' BRYCE CANYON AT BRYCE POINT MONTH JULY
XHAUAJO NT
1979
I
ro
(I
I
IJ
(t
L
R
A
II
G
E
11
400.
360.
320.
240.
200.
160.
120.
40.
n
METEOROLOGICAL CODES UP THROUGH 5
<
^r
^
<>l
X
>
X
X
>
X
*
X
_
<
X
X
$<
AJ*
^K
X
>
X
X
s
X
'
<
5
<-
X
?
4
_y
^
X
^
-
>$.
^
*
*
12345678 9 101112131415161?18i92G2122232425262728293fl31
DAYS 80XQ3/03,
Figure 3-4. Standard visual range data for month of July 1979 as determined from
observations of Navajo Mountain from Bryce Canyon National Park.
(Walther and Newburn 1980)
-------
I
10
i'i IMI h". V - L-'b mi les
Figure J.5. Map of Existing Visual Range (from EPA Report to Congress, page 4)
-------
Table 3.4: Some Natural Sources of Reduced Visibility
Products which contribute directly
Source Type Examples or indirectly to visibility reduction
Airborne soil dust dust storms coarse and fine particulates
Wood fires wildfire fine particulates, hydrocarbons
nitrogen oxides, sulfur oxides,
soot (elemental carbon)
Water hazes fog, rain, etc. water droplets and high humidity
Volcanic activity Mt. St. Helens, fine particulates (ash) and
Kilewa sulfur oxides
.Geothermal areas Yellowstone, water vapor, sulfurous gases
Geysers, CA
Biogenic Blue Ridge terpines, ethylene, isoprenes,
Mountains, hydrocarbons
Smokey Mountains
Wood fires can produce very significant visibility reductions even
at large distances. These fires produce many particles in the 1 to 0.1 urn
range. Some of these particles contain soot from incomplete combustion.
Gases given off include nitrogen oxides, hydrocarbons and, to a lesser
extent, sulfur oxides.
Water hazes and fog are important natural causes of visibility
reduction. Water droplets themselves interfere with visibility. Also,
high humidity conditions will encourage the growth of very small particles
below the light scattering region into a size range which is the most
efficient in scattering light.
3.6 Presentation of the Visual Environment
Review of Recent Applications
After reading the previous Section 3.4 visibility indices one may
wonder if a non-technical person can ever make decisions about visibility
based on numerical measures alone. In terms of benefit analysis, an
increase of AE of 4.6, for example, may be perceptible but what does it
really mean in terms of the appearance of the vista? To help decision
makers and designers of visibility benefit analyses, several attempts have
been made to use pictorial graphic displays of scenic vistas in visibility
studies.
3-26
-------
The first and most advanced attempt was that at Los Alamos Scien-
tific Laboratory (LASL). Williams et al. (1979) report how they used a
sophisticated computer generated display to simulate the appearances of
power plant plumes. Malm et al. (1980b) has used a similar technique to
study the thresholds of perception of layered haze. Of course, the pre-
viously much cited studies of Malm et al. (1980a) and Latimer et al. (1980)
are based on photographic slides with coincident physical quantities.
Similar "calibrated" photographic slides have been used to demon-
strate potential visual air quality degradation in the Kaiparowits Plateau
of southern Utah for various coal production scenarios on Bureau of Land
Management lands (BLM 1980).
Given all this activity and potentially very useful applications to
benefit analysis, a closer look at display techniques for visibility studies
is reasonable. Because of the great influence a photograph can have on a
non-technical person, it is only fair that in the review below a very
critical approach is taken.
General Principles
The basic goal of a graphic display for visibility work is to
reproduce as accurately as possible the original perceived view. Unfor-
tunately, this is an extremely complex problem. The major source of the
trouble lies in the extreme ability of the eye to adapt to its surrounding
conditions. Usually the reproduction is to be viewed under different
lighting conditions at different ambient luminance levels from the original,
and often the reproduction will be smaller or larger than the original
scene. These factors have to be taken into account in producing the
reproduction. Chapter 19 of The Theory of the Photographic Process (James
1977) describes in detail the psychophysical engineering that goes into
designing films which will produce pleasing pictures.
This brings up another issue. Henry (1979b) reviews several
excellent studies which show that the most preferred colors of skin, grass
and sky in photographs are quite different from the actual, natural colors.
The memory colors, i.e., colors of sky, etc., picked from memory are
different from the "real" colors and the preferred colors. This phenomenon
is widespread and well documented.
The result is that a reproduction with exactly the same relative
intensities of all colors (wavelengths) as the original scene will not look
like the original scene and will probably be judged less pleasing than the
real scene. The work of the LASL group and Malm et al. (1980b) do not
appear to account fully for these facts.
3-27
-------
On the other hand, if ordinary color film is used to make slides,
as in Malm et al. (1980a) and Latimer et al. (1980), then, since these
films have been designed to produce pleasing pictures, the reproductions
will look better than the original scene. In fact, Malm et al. (1980a)
does report this to be so. Further study of the use of photographs in this
manner is obviously called for.
Other Specific Problems
Loss of fine contrast detail by most reproduction is inevitable.
Since perception of fine detail is probably an important perceptual clue a
loss of detail will distort the reproduction. This is especially true of
video based techniques, such as with the LASL simulated photo technique.
Use of telephoto lenses tends to distort the size and perspective
of the view. Yet, most studies using slides have used telephoto lenses on
the camera. Processing and exposure are very important in producing
accurate photographs. Control photos of standard grey scales and other
color patches, such as those used in darkrooms, are recommended.
Other Alternative Approaches
Regular Photography. Experience from Malm et al. (1979) has shown
that regular photography twice a day of long-distance vistas using a
single-lens reflex camera with a 135 mm lens and Kodachrome 25 film will
produce during one year a good selection of slides showing various levels
of air quality, cloudiness and hazes under two different illumination
conditions.
Kodachrome 25 seems to produce an acceptable color and its stand-
ardized processing by Kodak, Inc. is very repeatable. Although, as
recommended above, pictures of standard scales should be taken for quality
control checks. Taking the pictures around 9 am and 3 pm local time
provides illumination from almost opposite directions. The 135 mm lens,
combined with a 135 mm lens in the slide projector and placing a person
near the projector while viewing a screen at a normal distance (20 feet),
makes a good image size and field of view.
Regular photography can capture hazes of varying intensity at
almost any location because of the large extent of regional hazes (many
hundreds of kilometers). Distinct plumes from power plants, smelters,
controlled forest burns and other sources may be photographed at relatively
close locations (up to few tens of kilometers). Sample regular photo-
graphy slides for one visibility benefits study may be found at the
back of the guidebook.
In addition to the more proven technique of regular photography,
there are a number of other approaches which could be useful to a benefit
analysis. It should be noted that these are only suggestions and that most
of the techniques are untried.
3-28
-------
Picture Modifications for Hypothetical Plumes. One approach to
representing visibility levels is through modification of existing pictures
or construction of simple artists renderings. These may range from very
simplistic attempts which merely retouch existing pictures to rather
elaborate systems such as the simulated photograph technique (see Malm et
al. 1980b).
Overlay Technique. One simple method involves overlaying layers of
translucent material over portions of the original picture. In this case a
key difficulty is the proper choice of the material to be overlayed and
the difficulty with proper gradations in material in order to represent
gradations in plume material or distances to terrain features. The color
effects and gradation caused by sun angle effects are also difficult to
represent in this fashion.
Retouch Technique. The method of retouching pictures is similarly
difficult. Furthermore, it is probably even more difficult to assure that
the retouching is a technically defensible representation of appropriate
visibility impairment levels.
Real Smoke Plumes. Another possibility is the use of infrequent
real smoke plumes to represent possible future smoke plumes. The diffi-
culty of this approach is that it may be very difficult to find existing
plumes with characteristics appropriate to a hypothetical source.
Artist's Rendering Technique. This approach involves computing the
sky color in terms of Munsell color chips and employing a commercial artist
to paint the scene with calculated sky colors.
Plume Superposition Technique. Another technique involves using
imaging technique to place an actual plume, photographed for a similar
source in another region during similar lighting and atmospheric conditions
on to an actual scene. This technique can not be used if there is a mixed
background of sky and terrain features. Furthermore, in the superposition
approach care must be taken to control the large number of parameters which
may influence the perception of the scene.
3.7 Visibility Modeling
A valuable means of relating source emissions to visibility im-
pacts are by mathematical simulation, modeling, and evaluation of moni-
toring data. Each of the components of the visibility problem, as shown in
Figure 3.1, can be reduced to mathematical relationships and programmed
into a computer. Such models must pass two criteria to be really useful:
1) they must be scientifically sound, i.e., they must be based on valid
assumptions and simplifications, and 2) they must be able to accurately
predict observed visibility degradation in the real world. The distinction
here is important, since a sound, valid model may give inaccurate results
because of poor input information. On the other hand, a grossly over-
simplified, unsound, and invalid model may produce reasonable predictions
by the well known process of cancellation of errors.
3-29
-------
Visibility models currently available concentrate on single source,
near source (less than 50 km) impacts on visibility. This is in line with
the phased approach of the EPA, of which Phase I focuses on control of
existing single sources strongly implicated in causing visibility degrada-
tion in protected Class I areas. The EPA has issued a User's Manual for
the single source model PLUVUE (U.S. EPA^SOc). The associated guideline
Workbook on Estimating Visibility Impairment (U.S. EPA 1980b) gives the
derivation and procedures for two less sophisticated levels of modeling, to
be used as screening tools to quickly sort out those sources which can
potentially, under worst case conditions, impair visibility in a protected
Class I area. Other models described so far in the literature are for the
most part similar to the PLUVUE model in basic assumptions and approach to
the optical calculations and in choice of calculated visibility indices.
The PLUVUE model has received limited validation as part of the
EPA's VISTTA (Visibility Impairment due to Sulfur Transformation and
Transport in the Atmosphere) program. Blumenthal et al. (1980) presents an
overall summary of the project, and Bergstrom et al. (1980) reports the
initial model validation studies. Further work is needed in this area.
Several organizations other than the EPA are expected to attempt field
validation of visibility models.
In the benefit analysis, simulations of visibility indices should
be used with care and, if possible, checked against real world experience.
This is especially true if calculations of frequency of occurrence of the
indices are also made. One fruitful approach is to combine model simula-
tions with historical photographs from the VIEW program network. This
approach has been used with some success for estimating visual impact of
coal development in southern Utah (BLM 1980).
3.8 The Effect of Definition, Measurement, and Presentation of Air
Quality Conditions Upon Benefit Measures
Before economic techniques can be applied to obtain monetary
measures of value for changes in air quality, great care must be taken to
define, measure, and present the alternative air quality levels. Un-
measured changes, or lack of control in many of the experimental factors,
will influence and bias the valuation process. These factors include the
choice of method, location and timing for measuring air quality; the type
of pollutants or impacts to be measured; and when used, pictorial presenta-
tion factors, including the choice of lens types (135mm versus 50mm), the
amount of clouds or snow cover in scene, whether the scene"is mountainous
or flat terrain, and the like. This section discusses the impact upon
benefit meausrement of variation in these factors.
Two problems should be considered when comparing alternative levels
of visibility aesthetics. First, all influencing factors other than the
air quality measure of interest should be held constant or accounted for in
3-30
-------
the analysis. Second, the most appropriate characteristics should be
chosen for measuring and describing the attributes of alternative air
quality levels that influence individuals' behavior at the affected site.
These problems must be addressed, because if unmeasured influences affect
an individual's utility this will be reflected in the monetary measures of
benefits and will inadvertently create inaccuracies.
The need for accuracy and control is demonstrated in Figure 3.6,
which demonstrates that as some measure of visibility aesthetics increases,
the total utility of visibility aesthetics to an affected individual also
increases. Fl and F2 represent two levels of a factor that might influence
the valuation process, such as location, scenic content, or the type of
pollutant measure used, etc. For discussion purposes assume F is the
amount of cloud cover in pictorial representations of two alternative
visibility levels A and B.
Utility
(F = F2)
E
D
Utility
(F = Fl)
0 = Visibility
Aesthetics Measures
Figure 3.6
The Effect of Varying Definition and Presentations Upoa Utility Valuation
(F represents influencing factors such as location, pollution measure,
time of day, scene character: -^ics, etc., where Fl, F2 represent
alternative levels or sites.)
J-31
-------
At cloud cover level Fl, the change in utility as visibility
aesthetics change from level A to level B is the distance CD. The first
problem would occur if level A is presented to the individual with cloud
cover Fl, and level B is presented with cloud cover F2. Then the dif-
ference in utility would be measured as CG, which inadvertantly measures
both changes in utility from visibility aesthetics (CD) and the changes in
utility from cloud cover (DG). This problem can be avoided by controlling
as many extraneous elements as possible and by statistically accounting for
changes in uncontrollable factors.
The second problem of choosing the appropriate level of the in-
fluencing factors is more difficult. In Figure 2.8 the choice of Fl yields
a change in utility of CD for changes from visibility levels A and B, while
the choice of F2 yields the much larger change EG. This problem can be
reconciled through the use of accepted standards, or by varying the levels
of extraneous influences and testing for and measuring their influence.
The choice of standards, which is not often clear-cut, is discussed earlier
in this chapter.
Both problems may be treated concisely in the following mathematical
model for an individual:
V = f(Q, F,) + E
(1)
Where V, Q, F are as in Figure 2.10 and E are other unaccountable
or random influences. Our concern is with 3V/30 over the Q range of in-
terest. If the F influence varies and is not accounted for, one measures
3_V
3Q
F=F1
3F
0=B
where
_3V
3Q
F=F1
is the partial derivative of U with respect to Q given F is at the Fl
level, etc.). As long as this second term is non-zero the analysis is
biased.
3V
9V
F=F1 T 3Q
F=F2
The second problem amounts to a case where
Here it is probably most appropriate to choose a standard if applicable,
such as scene content, pollutant measure, time of day for measurements,
etc., or to determine 3V/3Q based upon the joint probability distri-
bution of Q and F, P(Q, F). In the case where Q takes on two values,
A and B, and F takes on N levels Fi, where i = 1,2..N, then AV/AQ may
approximate 3V/3Q as:
3-32
-------
P(F = Fi, 0 = B)* f(B,Fi) -
i = 1
(2)
N
/(B-A)
£ P (F = Fi, Q = A)* f(A,Fi)
i = 1
where for example, P(F = Fi,Q = B) represents the joint probability of
F = Fi and Q = B.
Because f(Q,F) is often estimated in some variant of its dif-
ference form:
(3) V = f(AQ, AF) + E
equation 2 would then need to be suitably modified. In most cases it
would be appropriate to simply control as many influences as possible
because P(F,Q) may be unknown.
3-33
-------
CHAPTER 4
PRACTICAL APPROACHES TO MEASURING ECONOMIC BENEFITS ASSOCIATED
WITH CHANGES IN VISIBILITY AESTHETICS
Actual Market Versus Contingent Market Approaches
This chapter discusses practical approaches that have been de-
veloped to measure the economic value of changes in visibility. Two sets
of approachesbidding methods and residential property value studiesare
examined in detail because they are well established in theory and have
been the predominant methods applied in air quality benefit analysis to
date. The chapter discusses the theoretical basis of these methods, how
they are applied, their strengths and weaknesses and when they may be
relied upon for doing visibility benefit analysis. Because of the com-
plexity of valuing a non-market public good, there is a great deal of
underlying measurement error in both types of approaches. Many of the
major strengths or weaknesses of an approach relate to the ability to
minimize these measurement problems.
Special methods are necessary for valuing visibility, as with most
environmental goods, because of its special status as a "public good":
the use of visual aesthetics is non-exclusive and there are no existing
markets from which prices and demand curves may be directly obtained.
Given this absence of direct market data, the method used to assess the
value of visual aesthetics must rely upon one of two general classes of
valuation approaches: (1) methods that indirectly use actual market data
regarding the relationship between private markets and public goods,
combined with special analytic methods, to infer implicit prices and the
demand for the public good; and (2) methods that establish values directly
through the use of hypothetical, or what is known as "contingent market"
situations posed to survey respondents.
The first group is labeled "actual market approaches" because these
methods attempt to use existing market data in cases where the selection
of a market good may vary with visibility levels, such as the choice of
residential location. Property value studies, most of which use the
hedonic price technique to estimate an implicit price for air quality,
fall into this category. This type of approach presupposes that indivi-
duals respond in a predictable manner to environmental conditions that
they encounter in places where they live, work, and recreate. For ex-
ample, it is assumed the individuals prefer to live in neighborhoods where
breathing the air is not damaging to their health and prefer to visit
parks where the air is clear. This approach further assumes that the
intensity of these preferences is revealed by individuals' behavior and
their demand for associated market goods, e.g., how much more individuals
pay for homes in neighborhoods with clean air reveals how much they value
-------
clean air, and that the degree to which vacationers change their travel
plans when visibility deteriorates at a particular park reveals how much
they value visibility. Successful application of this approach also
requires that technical measures of pollution concentrations or visibility
levels are reasonable representations of the environmental attributes that
individuals value and that these measures can be used to identify that
part of an individual's behavior attributable to the component of environ-
mental quality being studied.
Although residential property value studies are currently the pre-
dominant variation of the market approach, recent work has also used
commercial property, wages, recreation travel patterns, and expendi-
tures in an attempt to identify willingness to pay for air quality. These
developing approaches are introduced in Section 4.4.
The second group, "contingent market approaches," attempts to
elicit values through surveys of how respondents think they would behave if
a proposed visibility change were to occur in hypothetical situations. The
bidding methods are in this group. In contrast to the market approaches,
non-market approaches do not attempt to infer values of a component of
environmental quality from observation of individuals' actual behavior in
response to a change in environmental quality. Instead they ask the
individuals themselves to predict how they would behave or have behaved in
response to a change in environmental quality. This approach presupposes
that a particular change in environmental quality can be described to the
respondents, usually with photographs and verbal descriptions, in a way
that corresponds to what their perceptions of the actual experience would
be. For example, it is assumed that a photograph of the Grand Canyon
obscured by pollution will elicit a response that corresponds to what the
response to the actual situation would be. This type of approach also
assumes that individuals are capable and willing to predict accurately
their response behavior to a hypothetical situation that they may or may
not have ever actually experienced.
Bidding methods represent the predominant contingent market ap-
proach currently in use to estimate benefits of visual aesthetics. Other
contingent market approaches being investigated include some types of
travel cost methods, ranked attributes and voting schemes. These de-
veloping approaches are introduced in Section 4.4.
4-2
-------
4-2 Bidding Methods
4-2.1 Introduction to Bidding Methods
The term bidding methods is used to describe a set of survey
approaches typified by the iterative bidding approach. These approaches
are often referred to as contingent market approaches because the values
received are contingent upon the hypothetical market established in the
survey instrument. The iterative bidding approach was first pioneered for
outdoor recreation demand analyses by Davis (1963). Since then it has
been used to estimate values relating to wildlife and recreation, risks of
nuclear power plant accidents, power plant siting, television demand,
beach use, urban noise, water pollution, strip mining impacts, and for air
quality aesthetics and health impacts, among others. *
The use of the iterative bidding technique to estimate values
associated with changes in air quality was initiated by Randall et al.
(1974) in the Four Corners area. Additional studies followed including
Brookshire et al. (1976) and Rowe et al. (1980) also in the Four Corners
area; Brookshire et al. (1979) in the Los Angeles South Coast Air Basin;
and EPA (1981), for estimating values related to the Grand Canyon and
Southwest parks for respondents in selected cities across the United
States.
In bidding techniques, respondents are given information on
current and proposed or potential alternative levels of visual quality at
a particular site. They are also given hypothetical markets which des-
cribe how payments are to be made or received by the respondents related
to changes in the air quality level. Next they are asked to bid their
maximum willingness to pay (WTP) or minimum willingness to accept compen-
sation (WTA) to incur or prevent the change. If a degradation of visi-
bility is proposed, respondents might be asked their WTP to prevent this
change or their WTA to allow the change. Similarly, for a proposed visi-
bility improvement, respondents are asked their WTP to fund the change or
WTA to forego the change. For example, respondents might be presented the
following hypothetical situation: "If visibility were to change from level
A to a higher level B," and a market mechanism, "using a surcharge on your
utility bill which would go to a fund to prevent visibility degradation,"
and in an attempt to receive a measure of the consumer surplus associated
with that change, "what is the maximum you are willing to pay each month on
your utility bill to achieve the increase in air quality from level A to
level B?" Because the valuations may vary with small nuances in the
application of this technique, it is imperative that it be carefully
designed and monitored.
A survey of 18 applications of the iterative bidding technique can be
found in NUREG (1979).
4-3
-------
4.2.2 Theoretical Basis
The bidding method approach is closely tied to the basic theory
of consumer valuation. Individuals are perceived as having the ability to
make well defined, rational decisions regarding trade-offs among alterna-
tives in order to maximize their own utility. The questions regarding
WTP and WTA are attempts to directly reveal the equivalent and compensating
surplus measures associated with the proposed air quality change, as shown
in Table 4.1. The measures are typically equivalent and compensating
surplus measures because the respondent must accept either the current
visibility level or the new or proposed visibility level (as shown in
Figure 2.8 in Section 2.2.3).
Table 4.1: Correspondence Between Willingness to Pay Measures and Consumer
Surplus Measures
Proposed Change
Increased Environ- Degraded
Questionnaire Approach mental Quality Environmental Quality
WTP Approach CS ES
WTA Approach ES CS
Use of bidding games requires that certain assumptions be met.
First, that individuals value certain aspects of changes in air quality
and that these aspects can be determined by researchers and appropriately
conveyed to respondents. It must be further assumed that respondents
understand the proposed changes and can, and will, accurately determine
what their behavior would be under the alternative situations. Abstract
economic assumptions are not needed if comparison and aggregation of
benefits is to take place according to some simple classification scheme-
such as by mean responses of different income class. Generally cross-
sectional analysis of bidding method responses and their determinants is
undertaken. This effort requires more specific assumptions about how
visibility enters the utility function and the use of specific forms of
the utility function, such as those presented in Section 2.5.
Property Rights in Bidding Methods
The correct choice of welfare measures in the application of the
iterative bidding approach also depends upon whether or not the individual
thinks he has, or is made to think he has, a right, referred to as property
4-4
-------
rights, to his preferred state of environmental quality and whether the
current state is believed to be better or worse than the alternative
state. This will influence the acceptability of the hypothetical situa-
tion to the respondent. Considering an air quality degradation, a respon-
dent who believes he has a property right to the higher level of air
quality may feel that the liability to compensate those affected or to fund
improvements lies with the polluters. Consequently he may reject the
alternative markets as inequitable. As shown in NUREG (1980), four alter-
native situations can be consideredthe maximum willingness to pay to
avoid a "bad" (WTPB), or to acquire a "good" (WTPG), and the minimum
compensation necessary for the individual to accept a "bad" (WTAB) or
forego a "good" (WTAG). The approach must account for the desirability
of the proposed change and the implied or believed property rights of the
individual, as shown in Table 4.2.
Table 4.2: Appropriate Iterative Bidding Measure of Welfare Changes
Property Right
Preferred State Less Preferred State
Proposed State is
Less Desirable WTAB WTPB
Proposed State is
More Desirable WTAG WTPG
WTPB and WTAG are equivalent measures.
WTPG and WTAB are compensating measures.
4.2.3 Application
Bidding methods have two principal components: the survey pro-
cedure and the survey instrument. "Survey procedure" refers to where
the survey takes place, sampling procedures and how questions are pre-
sented to respondents. "Survey instrument" refers to what is commonly
called a questionnaire, or the paper containing the questions to be
answered by the respondents.
The design of the survey instrument, which is the most important
element of any bidding method, is greatly influenced by the survey pro-
cedure employed. Typically, personal interviews, guided interviews or
questionnaires are used. In the personal interview, each respondent is
questioned individually by the enumerator. This allows the most flexi-
bility in question design and enumerator/respondent interactions, clari-
4-5
-------
fication of definitions, explanation of concepts, and the like. They are
also the most expensive. Guided interviews involve one or more subjects
who read and answer some of the questions on their own, in the enumerator's
presence. In the questionnaire procedure, the survey instrument is
completed by the respondent alone. These require the most amount of
design control and limitations. Other aspects of the survey procedures
are discussed in Section 4.2.6 below.
The concepts and procedures of bidding method survey instruments,
and the hypothetical situations involved, may be unfamiliar to respondents,
and therefore great care must be taken to insure that the survey instru-
ment is concise and well defined, yet fully informative. Survey instruc-
tions which are ambiguous or have ill-defined sections will decrease the
ability of respondents to accurately respond to the proposed situation.
On the other hand, survey instruments which are too long often have poor
response rates. The respondents who do complete the instrument may do so
from a feeling of obligation, but they may not feel obligated, or they may
not be able, to continue responding accurately (response fatigue). There-
fore, a carefully defined survey instrument is central to the success
of this approach to valuing air quality.2
Bidding method survey instruments typically have the following
sections: (1) a general introduction and statement of purpose; (2) intro-
ductory non-valuation questions; (3) scenario development and market
definition; (4) bidding or valuation questions; (5) evaluation of problem
bids; and (6) special options, closing questions and remarks. The remainder
of this section examines the purpose and types of approaches used in each
of the six sections of the questionnaire. Sample questionnaires from
actual experiments and suggested wordings for Water Resource Council
projects may be found in Appendix 1.
General Introduction and Statement of Purpose
The introductory statements serve to identify the research
organization, the purpose of the survey, why the respondent has been
chosen, and how long the survey will take. These comments should be
truthful and to the point. The wording and the enumerator must start out
and remain neutral throughout interview. Enumerators should appear not to
have a vested interest in anything but the accuracy of the results. The
introduction should convey to respondents that accurate answers are of
value to the research project and ultimately to the respondents themselves
so as to increase their incentive to respond accurately; it should also
any government-sponsored research all questionnaires administered to
more than 9 individuals are subject to Office of Management and Budget
(OMB) procedures for questionnaires to insure they are not offensive
or unnecessary. OMB questionnaire guidelines and procedures can be found
in OMB circular 40.
4-6
-------
initiate rapport and provide accurate information so that the respondent
will not later on feel mislead as to the intent of the questionnaire or
the necessary level of effort or time. A sample introduction might read
as follows:
Good morning/afternoon. My name is
(present ID). I am from the Western Policy Research
Institute in Denver, Colorado. We are doing research
to examine the trade-off between recreation, industrial
development and air quality in this area. In order to
accurately perform this research, we need a cross-
section of opinions. Would you be willing to spend ten
minutes to answer some questions about your recreation
patterns and air quality in this area?
Introductory Non-Valuation Questions
Introductory non-valuation questions generally gather useful
socioeconomic characteristics of the respondents, establish existing
behavior, cr simply determine whether the respondent is appropriate to be
included in the sample. For example, a survey of how air quality at
recreation sites will alter recreation location patterns may first need to
establish existing recreation location patterns and link them to existing
air quality levels. Again, questions should be limited only to those
necessary to evaluate responses and implement benefit assessment models.
Sensitive, threatening or embarassing questions, such as education and
income, should not be asked early on, since those concerning the accuracy
of subsequent responses (Locander 1977). Questions such as On a 1 to 5
scale, are you an environmentalist or developer? or Are you a member of an
environmental organization? may have merit in the analysis, but if asked
early on the questionnaire may tend to create a situation where the respon-
dent later feels obligated to bid differently from how he might have bid
had he not been asked these questions.
Scenario Development and Market Definition
Scenario development and market definition is the most critical
aspect of bidding method survey instruments. This section must care-
fully present information on alternative situations, payment procedures
definitions of visual impairment and the like. In essence, before respon-
dents can reveal the value they associate with proposed visibility changes
they must clearly understand, as journalists say, "the who, what, where,
4-7
-------
how, when, and why" of the story. To do so researchers must first deter-
mine how air quality attributes relate to physical measures of pollution,
what attributes of air quality are important to individuals and how best
to convey these attributes to respondents. This is an area where econo-
mists must interact with social and physical scientists (see Chapter 3).
Before a contingent market survey is performed it may also be desirable
to first conduct a survey that determines those attributes of the potential
impacts of importance to affected individuals (for an example see US NFS
1980). If all important attributes of air quality conditions can be
defined and presented in sufficient detail, all respondents will be able
to accurately perceive the changes that are being proposed.
Verbal or written descriptions of conditions alone may be satis-
factory where alternative scenarios can be clearly defined by the enumerator
and understood by respondents. This is possible for many activities and
environmental amenities such as wildlife stocks, but descriptions alone
are not satisfactory for the characterization of air quality conditions.
On-site presentation of conditions, while most accurately reflecting one
level of air quality are also the most difficult to control and expensive
to administer. The conditions, including visual range, contrast, cloud
cover, and the like, must be continuously monitored to accurately account
for variations in the valuations received (see Section 2.4 above). As a
result, pictorial representations of alternative air quality scenarios
using photographs, slides, and motion pictures coupled with verbal descrip-
tions and maps have become the standard mechanism for displaying alternative
scenarios of air quality. The techniques used for pictorial presentations
are described in Chapter 3.
The scenario development should include some or all of the fol-
lowing characteristics:
1. A description of the elements of the air quality for the situa-
tion in question and how it is proposed to change, including
location, intensity, duration, frequency, time of day, plume
blight or regional haze, etc. (each of these is relevant to EPA
promulgated regulations concerning significant changes in air
quality). Air quality is a multifaceted, ever-changing environ-
mental good. The more clearly characteristics and changes can
be mutually understood, the better the evaluation of the
benefit analysis will be. It should also be made clear whether
an impact is to occur at one site or across the whole region.
2. Some indication of why and how these changes affect the indivi-
dual respondent, such as visibility aesthetics in the recreation
experience or at his place of residence, the health effects,
etc. Changes in visibility and health aspects of air quality
may occur independently but may not be believed to be indepen-
dent by respondents, and therefore it is important to key the
respondent on the appropriate aspect of interest in order to
isolate its effects and benefits.
4-8
-------
3. Descriptions of why current or potential air quality levels may
occur. This includes automobile exhaust, better emission
control devices on large new or old power plants in the region,
and the like.
4. How changes might take place, how they will be paid for, and
when and how they will be implemented, such as a monthly
utility bill increase to pay for addition emission control
devises, or a state fund to increase mass transit financed
through increased automotive sales taxes.
To be successful the scenario development must be informative;
clearly understood; realistic by relying upon established patterns
of behavior and legal institutions; have a uniform application to all
respondents; and, hopefully, leave the respondent with a feeling that the
situation and his responses are not only credible but important. The
pictorial representations of alternative scenarios, if used, should il-
lustrate the important air quality attributes and problems relevant to
the study area, including discoloration and visual range in directions of
interest. Current and alternative situations should be clearly differen-
tiated.
The frequency of occurrence of alternative levels of air quality
has emerged as an important concern in visibility benefit analysis. In
the past, alternative scenarios have been presented with each depicting
one level of air quality as "typical." However, the value an individual
places upon improving typical visibility from some level D to a higher
level C may depend upon how often D can still be expected to occur in the
new situation as well as how often even higher levels of visibility, B and
A, can be expected. The damage to an individual who goes to a national
park the only day it is totally polluted may not be offset by the value to
another individual who goes on the only perfect day. There may also be
benefits from reduced uncertainty as to which air quality level will be
encountered.
There are several methods to present and evaluate the effect of
the frequencies of encountering visibility levels other than the typical
level described in any particular scenario. First, one might present some
form of probability distribution for each level of air quality in each
scenario (5 days of A, 20 days of B, and 5 days of C per month) where the
distribution is changed across respondents. Unfortunately, this may
overload the respondent with more information than he can retain and
respondents may end up only concentrating upon the mean levels. (This is
a testable hypothesis.) Second, respondents may simply be told to compare
a situation where B is typical, with some A and some C, but no D, to a
situation where C is typical, with some A, B, and D, or some suitable
combination.
4-9
-------
Perhaps the most appropriate method to account for the impact of
different frequencies of alternative levels of visibility is to estimate
the benefits associated with changes to each level, as in the second
approach described, and then insert these values into a joint prob-
ability distribution function of the expected rates of occurrence of the
alternative visibility levels and visitation rates by time of day and day
of year. This could be done in the estimation process after the survey.
For example, if, in one scenario, 50 percent of visitors or residents could
expect with certainty an air quality change for some level C to level B,
25 percent could expect change from C to A, and 25 percent could expect no
change, then the appropriate value of achieving this scenario, if the
individual were risk neutral, would be .5 * (value of C to B change) + .25
* (value of C to A change).
The following is a sample scenario development from the South Coast
Air Basin study (Brookshire et al. 1979).^
Here are three photographs representing average
levels of visibility for the three different
regions of the Los Angeles Area shown on this map.
Picture A represents poor visibility; Picture B
represents fair visibility; and Picture C represents
good visibility.
Public officials are strongly considering the
possibility of trying to reduce the levels of
emissions throughout the Los Angeles Area. Such
action could require additional funds which might
be generated by (a monthly charge, an extra charge
in your utility bill) for as long as you live in
the Los Angeles area. These funds will be used to
help finance air quality improvements in the Los
Angeles area. (Page 140.)
The aesthetic valuation component of the above scenario development for
areas with poor visibility continues with:
The Los Angeles area has some very beautiful
background scenery. Because of automobile and
industrial emissions, there is a haze which
reduces and distorts the ability to see this
scenery. This means that many people have to
leave Los Angeles and travel long distances to be
able to enjoy views which could be visible from
their homes if these emissions were reduced.
lists in parentheses represent alternatives which are randomly
preselected to test their influence on the valuation process.
4-10
-------
As indicated by the map, you live in an area which
has been classified as having poor air quality
relative to the rest of the Los Angeles area.
Picture A represents the visibility level which
typically occurs in your area. I am only in-
terested in how you value being able to see long
distances.
If the level of emissions could be reduced in the
Los Angeles area so that visibility conditions
would be represented by Picture B instead of A,
not only in the B area but also in your area, and
if the air would be cleaned up to this level in
(2, 10) years, would you pay (a monthly charge, an
extra charge in your utility bill) of ($1, $10,
$50) for as long as you live in the Los Angeles
area? (Page 141.)
Bidding or Valuation Questions
The next section of the bidding procedures attempts to elicit
values for alternative levels of air quality or for changes in air quality.
Respondents may be presented with different bidding schemes, including the
iterative bidding process, selection of intervals, or the selection of
specific alternatives. The iterative bidding procedure involves a situa-
tion where, perhaps, an increase in air quality is proposed from some level
C to higher levels B and A. Answers are elicited in terms of yes or no to
questions expressed in the form, Would you pay amount X...? A yes answer
leads the enumerator to raise the amount and repeat the question, perhaps
repeatedly until a no answer is obtained. At this point the amount is
reduced until a yes answer is again obtained. The largest amount that
elicited a yes answer is recorded as the maximum willingness to pay. WTA
approaches substitute "willingness to accept X" for "willingness to pay X"
and lowers rather than raises the amounts upon receiving yes answers and
vice versa for no answers. Respondents tend to catch on quickly to the
iterative bidding procedure and cut it short by stating their maximum WTP
or minimum WTA for a proposed level or change in air quality. The one
drawback to this iterative procedure is that it requires the personal
interview format, which is the most expensive to administer.
Alternative approaches to the iterative bidding procedure include
simply asking the respondents their maximum WTP or minimum WTA and presenting
dollar intervals to choose from, dollar options to choose from, or a
fill-in-the-blank procedure. These non-iterative bidding approaches are
more amenable to the guided interview and mail questionnaire surveys but
may suffer from increased measurement error because the most precise bid
possible may not be elicited.4
^Mail questionnaires with interval choices have been applied repeatedly
for wildlife valuations (see Hammack and Brown 1974 and Brookshire et al.
1977). The later study found no statistically significant difference between
the mean bids between a personal interview iterative bidding approach and
an interval approach mail questionnaire.
4-11
-------
The bidding questions may address alternative levels of visibility
or changes in the level of visibility. For example an enumerator might ask,
"If you were to still spend (N) days per year recreating at [selected park]
what would be the maximum entrance fee you would be willing to pay per day
if visibility were as depicted in situation [A,B,C]?" In this case the
question would be asked for all three levels with the difference in the
maximum willingness to pay across levels representing the value (welfare
measure) of changes in air quality.
Questions which directly elicit the value of changes in visibility
are asked slightly differently. For example, if some air quality scenario
C is characterised as the existing level an enumerator might ask, "Would
you be willing to pay $X per month on your utility bill to finance emission
control which would result in air quality level A (a higher level) rather
than C?" Using the iterative bidding approach the amount $X is again
varied until the maximum WTP or minimum WTA is determined for the proposed
change. There is a pitfall in this approach to valuing changes in air
quality that must be avoided. If a respondent does not perceive C as being
the current level of air quality, but rather perceives some lower level, D,
as being the current level, care must be taken to insure the bids received
do not include the value of the change from level D to C as well as from
level C to A.
There are several other aspects to the process of eliciting bids,
each with variations in application. The type of payments are typically
accounted for in dollars per unit time; however, expenditures of time,
travel or moving costs are also potential units of measures convertable to
dollar equivalents. Payment procedures must also be considered. Respon-
dents may be presented with one of many schemes as to how actual payments
would be made if the plan were implemented, including everyone paying
equally, everyone paying equally at the mean value estimated for the
population, paying their bid, paying a proportion of the total depending
upon their own bid, and the like. The first two approaches have been used
more frequently.
Payment procedures should be designed to be viewed by respondents
as fair and ethically sound. Bids should also be presented as payments,
not contributions. Respondents should have a clear understanding that if
they were not willing to pay a certain price they would have to go without
the good. Wording such as "would you be willing to pay..." may be mis-
interpreted as an appeal for voluntary contributions.
"Payment vehicle" refers to the method through which payments are
to be made. The most prevalent have been user charges, such as entrance
fees, license, and purchase fees; utility bill increments; and sales
taxes. The payment vehicles should be carefully pretested and include a
neutral vehicle, such as a trust fund devoted entirely to providing the
good. Vehicles which generate emotional responses on related issues should
be avoided.
4-12
-------
Finally, additional information may be given to increase respon-
dents' understanding of the market and increase the accuracy of their
bids. This may include tables which indicate what a monthly payment sums
to over several years, or how each 1 percent of sales tax relates to
dollars spent by income bracket.
Two sample procedures are presented below. The first, from Brook-
shire et al. (1976), is an example of an iterative bidding procedure.
First, let's assume that visitors to GCNRA (Glen Canyon National
Recreation Area) are to finance environmental improvements by
paying an entrance fee to be admitted into the recreation area.
This will be the only way to finance such improvements in the
area. Let's also assume that all visitors to the area will pay the
same daily fee as you, and all the money collected will be used to
finance the environmental improvements shown in the photos.
Q6. Would you be willing to pay a $1.00 per day family fee to
prevent Situation C from occurring, thus preserving Situa-
tion A? $2.00 per day? (increment by $1.00 per day until a
negative response is obtained, then decrease the bid by 25
cents per day until a positive response is obtained, and
record the amount).
Q7- Would you be willing to pay a $1.00 per day fee to prevent
Situation B from occurring, thus preserving Situation A?
(repeat bidding procedure).
The second is an example of a value selection approach.
Clearly, everyone desires cleaner air; however air quality improve-
ments entail considerable expenditures. Let us propose a mechanism
to continue to finance improvements in visibility. Suppose the
method used to finance continued emission control would be through
additional monthly charges on your utility bill and everyone would
pay the same amount.
Ql. With this in mind what is the maximum monthly amount you would
be willing to pay each month to attain the improved visibility
level B rather than the current level C (enter #1 next to the
highest amount in the table below).
Q2. What is the maximum amount you would be willing to pay each
month to expect the improved visibility level A rather than
the current level C (enter #2 next to the highest amount in
the table below).
$ .00 $ 3.00 $ 10.00
$ .50 $ 4.00 $ 15.00
$ 1.00 ^^ $ 5.00 $ 20.00
$ 1.50 $ 6.00 $ 25.00
$ 2.00 $ 7.00 $ 50.00
$ 2.50 $ 8.00 S 75.00
$ 9.00 S100.00
4-13
-------
Evaluation of Problem Bids
In many instances, respondents to bidding method questionnaires
misrepresent their valuations by giving false zero bids, refusing to give
finite bids, or simply refusing to respond to questions posed to them.
These bids, or rejection of the bidding process, do not reflect the true
valuation of changes in air quality and must be distinguished from "true
bids" for use in policy analysis. In many cases problem bids related to
questionnaire design can be eliminated through careful pretesting. In both
the pretest and survey respondents should be queried as to the intent of
any problem bids by presenting alternatives as to why these bids may have
been given. Sample responses and interpretations of zero bids are provided
in Table 4.3. Similar responses for infinity bids could be developed. If
the received bid is truly a rejection of the procedure, alternative ap-
proaches may be attempted.
Special Options, Closing Questions and Remarks
At this point in the survey special options and closing questions
may be presented to respondents. Special options may include attempts to
validate or analyze the values previously received, to separate out aesthe-
tic values from health values, or simply to ask special policy questions
related to the specific site. Final questions may be less restrictive in
content or more open ended, such as the respondent's environmental stance,
his income, or whatever is deemed necessary to insure accurate modeling and
evaluation of responses.
Table 4.3: Interpretation of Zero Bids
Response Alternative
The change in air quality is not
significant.
It is unfair to expect the victim of
the damage to have to pay the costs
of preventing the damage.
The methods of payment are not
satisfactory.
The money collected would probably
be used ineffectively.
Other (please specify).
Interpretation
True zero bid.
Rejection of bidding game.
Vehicle rejection.
Rejection of bidding game
or payment vehicle.
Source: Blank et al. (1977).
4-14
-------
4-2.4 Strengths of the Bidding Questionnaire Approach.5
Contingent market valuation techniques, including bidding methods,
offer bright prospects for analyzing any important issue or value associated
with air quality. As long as a credible situation and market mechanism can
be developed and the survey instrument is carefully designed and pretested,
the researcher can obtain reliable and defensible value estimates.
The strengths of this approach are many, but probably the most
important is that the technique is both well grounded in economic theory
and flexible in application. The approach can be applied to obtain
separate activity, option and existence values, both on-site and at alter-
native locations, while it is virtually impossible to employ market
approaches to elicit option and existence values for changes in air
quality. Carefully designed survey instruments can be applied to evaluate
the lasting or temporal effects of plume blights and of regional haze at
both national parks and in urban centers as long as these impacts can be
conveyed to respondents in a manner they can understand and relate to.
Impacts at any one site or across an entire region can be examined with
bidding methods. Separate aspects of air quality, such as aesthetics and
health, may also be individually valued.
Both potential increases and decreases in air quality may be
examined before a change has occurred. The contingent market approaches
allow the design of trade-offs which are credible but may never even occur
in actual markets due to legal or political constraints. In essence the
approach is not limited by the appropriateness of available data or other
problems in market data.
Another strength is that data are generated that conform to an
economic model, eliminating the necessity to develop a model with limiting
and special assumptions so it may be applied to some existing data set.
This often eliminates special prior theoretical assumptions which are
sometimes necessary to apply market data to visibility benefit models.
Bidding methods also virtually eliminate the worry about complications of
market imperfections and transactions costs, which in many cases may
overwhelm the application and accuracy of market approaches. Another
strength of these approaches is their ability to allow the researcher to
account explicitly for, and to investigate, the impact of property rights
structures.
^Other useful reviews of the strengths of air quality contingent market
approaches are found in Brookshire et al. (1979). It should be noted
that many of these strengths and weaknesses apply to all non-market, or
contingent market, approaches.
4-15
-------
4.2.5 Weaknesses of the Bidding Approaches
Bidding methods have several important problems. The most im-
portant is what sociologists and survey psychologists deal with in inter-
view studiesdesigning questions so as to minimize perception errors and
biased responses. The second problem area is desiging and implementing an
effective survey procedure, and determining the appropriate functional
relationships to evaluate responses. Due to the unique importance of
questionnaire design issues to the application of bidding methods, they are
discussed at length below. Discussion and directions for survey procedures
and data analysis are provided in Section 4.2.6.
Perception errors and biases may be inherent in the survey instru-
ment design. Bidding methods ask respondents to reveal consumer surplus
measures for hypothetical situations often not faced in a market place.
Therefore, only perceptions, or estimates of their "true" values and
behavior within the context of the hypothetical situation and market, as
they perceive it, can be reported. The reported values may well reflect
the respondents "true" values, estimated with a great deal of uncertainty,
and subject to the influences inherent in the design of the survey in-
strument. These influences decrease the accuracy of responses and may
yield potential biases in the valuation process. Minimizing and eliminat-
ing these problems greatly enhance the usefulness of contingent market
approaches to the researcher. The following discussion covers five par-
ticular types of problems that may occurhypothetical bias, strategic
bias, information bias, contingent market rejection and problem bids, and
other problem biases. Tests for the existence and magnitudes and actions
that may be taken to minimize their impacts on the accuracy of the valua-
tion process are also presented.
Hypothetical Biases
The term hypothetical bias refers to any inaccuracies or biases
in respondents' answers because, they are not revealing actual behavior
but rather stating intended behavior in a hypothetical situation. As many
economists have pointed out, if respondents believe a situation to be
fictional, their responses often will also be fictional and therefore will
not reflect their true preferences and behavior.
The design of bidding method survey instruments has the potential
for hypothetical bias problems. Respondents are presented hypothetical
scenarios and asked how their intended time and budget allocations would
change if the scenarios occurred. There are several reasons for inaccurate
responses to these hypothetical questions. First, respondents may be
unable to visualize themselves in the hypothetical situation with which
they are presented and cannot accurately determine what their actual
behavior would be. For example, a non-hunter might have great difficulty
with a question such as, If you desired to hunt elk and could expect a 40
percent chance of successfully bagging an elk, would you be willing to pay
$X_ for the hunting license? because the situation would not be realistic or
relevant to him.
4-16
-------
Second, people may tend to be more conservative and to more
carefully account for time and budget constraints when deciding to allo-
cate real dollars and time than when responding to hypothetical situations.
It may well be the case that a survey instrument can obtain consistent
positive values of some amount, say $5, for nearly anything if the respon-
dents do not believe they will actually have to pay the stated amount, or
if the situation lacks credibility. For example, how credible is it that
park entrance fees will be increased to pay for environmental improvement
at regional power plants, or how credible is it that utilities will raise
all consumers' monthly utility bills by a lump sum amount to pay for
increased emission controls?
The question of hypothetical bias is whether the contingent
answers are an accurate reflection of what would occur if the defined
market did exist. There is evidence that the hypothetical bidding method
survey instrument does generate inaccurate and biased responses. One study
(Bohm 1971) compared the effects of several alternative bidding and payment
schemes upon the valuation procedure for public television. In several of
the alternatives respondents were actually given money to spend and pre-
sented with several different payment schemes, such as all pay equally or
paying your stated bid. Another group was asked how much they would be
willing to pay rather than having to actually pay for public television.
These WTP bids were statistically significantly larger than the bids of
those who actually paid. In another study, Bishop and Heberlein (1979)
found that stated willingness to pay for goose hunting licenses (WTP^)
was much smaller than actual cash offers accepted to forego hunting (WTAB).
These findings must be tempered with the expectation that WTAB would
exceed WTP^ measures. A more appropriate comparison would have been to
compare the same measure in actual and hypothetical situations. In a third
effort Brookshire, Randall and Stoll (1980) found that if respondents felt
either the situation or their bid lacked credibility they tended to de-
crease their bids. Irf summary, hypothetical biases and inaccuracies seem
to exist in bidding methods, but the size and direction is unknown and is
probably unique to each application.
Another indicator of the weakness of the hypothetical approach is
the often high variability of the values received. In several air quality
studies (Brookshire et al. 1976; Randall et al. 1974; and Brookshire et
al. 1979) less than 10 percent of the variation in the bids received could
be accounted for by socioeconomic variables or by changes in the levels
of visibility.
Biases and inaccuracies in hypothetical survey instruments can be
minimized if situations which are realistic and credible can be developed
by researchers and accepted by respondents. Sociologists and public
opinion researchers have built up a substantial body of literature con-
veying methods of making survey instruments more reliable, including the
use of concrete elements of realism and credibility (see Crespi 1971; and
Erskin 1972). Psychologists Ajzen and Fishbein (1977) have shown that
to maximize the relationship between stated intended behavior and actual
4-17
-------
behavior, the hypothetical situations should correspond to actual situa-
tions and behavior in terms of action, context, target, and time frame.
Also, knowing respondents beliefs is the best aid in predicting behavior.
A well designed and pretested iterative bidding survey instrument should
attempt to meet these conditions. The major problem area is the context:
if the contingent market context does not relate to real world situations,
uncertainty and measurement error may increase. The problem here is that
the real world often does not have effective markets in which preferences
for air quality may be reflected in the manner that individuals may desire
for any specific application, such as emission control funds at a national
park.
Biases and large measurement error in hypothetical markets may not
be any more serious than in market approaches if the hypothetical scenario
is credible. The accuracy of values that are reflected in market data
is also subject to limited information when individuals make choices and is
confounded by market imperfections and transaction costs. In fact, in a
carefully designed contingent market, respondents may state intended
behavior which is based upon more, and perhaps more accurate, information
than that available in real markets.
Strategic Biases
Strategic biases occur when respondents systematically overbid or
underbid relative to their true values. Because these bids are not equal
to their "true" values, they are biased estimates of value. There are two
common incentive structures for over or underbiddingattempting to "free
ride," and attempting to influence the study results to an outcome more
favorable to a position the respondent supports.
"Free riding" refers to the concept that a consumer believes he is
better off by not paying for the provision of a public good while at the
same time enjoying its consumption because others have paid for its pro-
vision. 6 An individual attempting to free ride would bid zero even when
visibility changes have a positive value for him.
An individual may overstate or understate his true values in an
attempt to strategically affect the outcome to a position he supports. For
example, a very strong supporter of air quality improvements may overbid
his true preferences to bias the results of the study in his favor, while a
developer may underbid for the same reasons.
The ability to effectively strategically bid can be tied to the
proposed payment scheme.7 For example, respondents may be more inclined
to attempt to free ride under a scheme where everyone is to pay his
free rider problem dates back to the seminal work on public goods
by Samuelson (1954) and respresents the classical argument of why markets
fail to provide public goods, as well as why valuation methods are ex-
pected to reveal biased values.
7See Kurz (1974) and Bohm (1971). Several articles on related topics can
be found in the 1977 issue of Public Choice.
4-18
-------
stated amount rather than if everyone is to pay an equal share of the
total costs. Bidding may also be affected by the credibility of the
payment scheme. If respondents do not believe the payment scheme will
actually be instituted, but that these results will be used for some cause
(respondents must ask themselves who plans to use this and why), there is
an incentive to bias the results to their advantage.
A number of studies have examined the effect of payment schemes
on the bids. Bohm examined six alternative payment schemes for a publicly
provided commodity (public television) that were designed to elicit stra-
tegic responses. Bohm's analysis concluded that there was no significant
difference between any of the payment schemes except when no formal
payments or decisions were to be based upon the results. In this case the
bids were significantly highera problem more closely related to hypo-
thetical bias, discussed earlier. Scherr and Babb (1975) examined the
desire to free ride using three payment schemes to reveal valuations for
two public goodsa concert and a library fund. No evidence was found to
reject the hypothesis that any scheme inhibited free rider behavior.
In a study of aesthetic damages from coal mining in Kentucy, Randall et al.
(1978) found that three different payment mechanisms, which theoretically
should have yielded equivalent mean bids, yielded significantly different
results. Finally, in several laboratory experiments, Vernon Smith (1977,
1980) found that free riding and strategic behavior can be avoided by the
choice of vehicles and decision making processes.
Two air quality studies in which iterative bidding techniques were
used have examined the potential for strategic behavior. In the first,
Brookshire et al. (1976) posited that if a respondent desires to effec-
tively influence the study valuations to a position that reflects his
value, he must possess a great deal of information as to the average bid
of other respondents, the strategic behavior of other respondents, and the
sample size. The study asserted that the existence of strategic bids
will result in a flattened distribution of actual bids relative to the
normal distribution. No formal empirical tests were conducted.
In the second such study. Rowe et al. (1980a) attempted to test
specifically for strategic bias. While many other problems were detected,
only obvious strategic bids were detected. In this study, half the re-
spondents were provided false information about the mean bid of other
respondents before they bid. All respondents were then asked their
environmental stance after bidding, the hypothesis being that environ-
mentalists would be more apt to attempt to increase the mean bid artifi-
cially by overbidding their own "true" value, while developers would act in
just the opposite manner. No statistically significant strategic in-
fluences were found after deleting obvious problem bids, such as respond-
ents who were openly giving false answers or were not cooperating; deleting
all bids greater than ten standard deviations from the income adjusted
means, as such bids are suspect because with most any statistical distri-
bution these values would occur less than once per several hundred observa-
4-19
-------
tions; and after deleting bids determined to be free riders (using Table
4.3 above ).** Next, approximately one third of the respondents in the
sample were provided information on the sample mean bid after they had bid.
Only one respondent altered his bid and that was to the supposedly lower
overall bid.
The evidence to date indicates that strategic behavior related
to the free rider problem may be detected and that behavior aimed at
affecting the overall valuation procedure is not a statistically signifi-
cant influence in the bidding methods when obvious problem bids are
eliminated.
Information Biases
Information may be explicitly and implicitly conveyed to respon-
dents, which, given their uncertainty about their true valuations, may
influence and bias the revealed valuations. For example, in Rowe et al.
(1980a) selected respondents were told a supposed mean value received thus
far that was very low. Individuals receiving this information bid signi-
ficantly less than the other respondents. This suggests that if indivi-
duals are given sufficient information, whether it is or is not valid, and
their true bid exceeds the stated mean bid they may be influenced to
believe their bid is "incorrect" or they will exhibit a form of the classic
free-rider behavior and bid a different amount than their maximum WTP. The
South Coast Air Basin study (Brookshire et al. 1979) also found instances
when the order of the presentation altered the bids in a statistically
significant manner.
Iterative bidding questionnaires are potentially subject to
another type of information bias induced by the suggested starting bid.
The procedure asks for yes and no answers to questions like, Would you be
willing to pay x....? In situations where the respondent is uncertain of
his valuation or has a great deal of problems with the contingent market,
the starting bid may have a strong informational content as to his "ex-
pected" bid. For example, starting values of $10 and $100 may suggest
different expected final bids are to be received. A pretest for starting
bid bias can provide very useful information about the quality of the
contingent market process.
A second problem related to the starting point is that if the
bidding process is methodically implemented the respondent may become bored
and cut the process short at a value other than his maximum WTP, though in
most iterative bidding surveys it has been found that many respondents
quickly adapt to the process and cut it short by jumping to their final
values.
°The study also noted that in the situation depicted the respondents were
to pay for most or all of the abatement costs as well as receive most or
all of the benefits. If respondents, as a group, were paying a minority
of the costs but receive all of the benefits, they may bid strategically
to increase the total paid by the paying majority (possibly in another
state) to increase the unpaid for benefits they receive.
4-20
-------
The same starting point information bias may occur in interval or
numerical selection versions of the bidding technique described. In these
cases, the clustering and range of intervals or dollar values to choose
from may suggest and influence the final values elicited.
Only five iterative bidding studies to date have reportedly tested
for starting point information biases (Rowe et al. 1980, Brookshire and
Randall et al. 1978, Thayer 1981, Randall et al. 1978, and Brookshire et
al. 1979). Rowe et al. (1980a) found starting point bias for the valuation
of air quality in the Farmington, New Mexico, area. The South Coast Air
Basin study (Brookshire et al. 1979) also found a starting bid problem in
about one-sixth of the alternatives bidding situations. No other studies
of visibility using bidding methods have tested for starting point bias.
The exact nature of the starting point bias tests were not reported in two
of the other three studies and cannot be evaluated.
Other influences may be imparted by the enumerator. All surveys
are affected by the interpersonal dynamics of the enumerator/respondent
relationship. The critical element is that the enumerator must remain
neutral and appear interested only in the accuracy of the results. A
carefully designed and pretested questionnaire will test for and
minimize each of these information biases.
Contingent Market Rejection and Problem Bids
Bidding questions for changes in air quality are not always well
received by respondents for many reasons. These may include rejection
of the hypothetical scenario, rejection of the payment mechanism (utility
bill, sales taxes) or payment scheme (all will pay equally versus the
individual will pay his own valuation), rejection of the implied property
rights or liability rules presented in a situation or, rejection for moral
and ethical reasons. Payment vehicle problems are a classic example of a
contingent market rejection that may occur. Respondents may object to the
use of utility bill increments, payroll deductions, or sales taxes. Both
the South Coast Air Basin study (Brookshire et al. 1979) and Rowe et al.
(1980a) found that for about one-third of the bidding situations the
payment vehicle was a statistically significant influence on the bids
received for visibility changes.
Rejection and protest bids have varied from 20 percent (Brookshire
et al. 1976) to 50 percent (Randall et al. 1979, Rowe et al. 1980, and
Brookshire and Randall 1979) for specific applications of the bidding
technique. In these cases, respondents' true values remain unknown and
unaccounted for. If this non-cooperation is not random across respondents,
it will bias the valuation process.
4-21
-------
Another reflection of problems in the bidding method scenario
developments has been that the WTA payment bids are often considerably
higher than WTP bids compared to what theoretically is expected from income
effects alone. For visibility studies Rowe et al. (1981) found the ratio
of WTA to WTP bids ranged from 4 to 14. Other non air quality bidding
technique studies have resulted in WTA measures of up to 12 times larger
than WTP measures.^ This is an area of particular concern for the future
development of the bidding method procedure.
Attempts to obtain compensating surplus measures for proposed
environmental degradation have met with the most protest bids. In willing-
ness to accept compensation approaches, respondents are asked their
minimum willingness to accept payment for decreasing air quality. Res-
pondents may believe they have a right to the higher preferred level of
air quality and reject being "bought off," either for their own sake or
for the sake of the community as a whole. It may also be the case that
they are giving WTAB bids based not upon their own ability to pay (income)
but rather upon the ability to pay of the suggested payee, such as a
utility company. In these cases respondents often reject bidding or give
very large and infinite bids. It is also possible that the compensation
may simply need to be very large to offset losing the wealth associated
with the perceived ownership of the higher level of air quality.
Respondents who believe that they have a right (referred to as
property rights) to the current level of air quality may also reject the
concept of increased payments, for example on their utility bill, to
prevent degradation of air quality, often seeing this as a "victim pays"
concept and feeling that the liability to fund pollution control lies with
the polluter. Rejection of this type occurs considerably less often, since
many respondents realize that the cost of utility-financed emission control
devices are often passed on to consumers.
Other Potential Problems and Biases
Other biases and problems are, and will continue to be, revealed
as more studies are undertaken using the bidding techniques. One such
potential problem is that respondents may be willing not only to make or
take payments representing the dollar value of a proposed change in
visibility, but they may also desire to change locations, rates of parti-
cipation in activities, timing, or take other mitigating action simul-
taneously.
Few bidding methods specifically allow for these simultaneous
activities, although there have been recent efforts to incorporate this
information (Thayer 1981; Blank et al. 1978). Another way to minimize this
impact is to validate results through the use of other approaches, such as
the household production function approach or hedonic approaches.
"See NUREG (1980) for a survey of the bidding methods in several alterna-
tive applications. Rowe and Blank (1981) have examined several arguments
for the large differences in WTP and WTA empirical measures and the
implications for measure selection and application of benefit estimation
techniques.
4-22
-------
Another potential problem, particularly in urban locations, is the
inability to separate out aesthetic from health values associated with
changes in air quality. An example of how this might be minimized by
carefully defining the good and focusing the respondent's attention can be
found in the South Coast Air Basin study (Brookshire et al. 1979).
4.2.6 Survey Procedures and Post Survey Data Analysis
Survey methods used to obtain and analyze data can be as influen-
tial upon bidding method results as is the choice of economic benefit
measures and their application and have, heretofore, not received sufficient
attention. Survey procedures entail a great deal of effort and planning,
as illustrated in Table 4.4. First, questionnaires must be formulated and
designed, as has been described above. Interviewers must be carefully
selected, trained and quality controlled to insure that they are effective
and neutral in their approach. Interviewers who reveal strong attitudes
about air quality to respondents or interviewers who are timid or have
presentation problems can inadvertently bias, or worse, actually falsify
the results they obtain. The survey effort must also consider the sampling
procedures to be used; costs and timing limitations, interview procedures
and coordination; the establishment of trouble telephone lines, contacts
with local officials and the like; and finally, handling and analyzing the
data received.
This section provides discussion and guidance for selected issues
which influence the effectiveness and costs of the bidding method or
other contingent market survey procedures, and samples used to estimate
benefits.10
Table 4.4: Steps in Survey Research
1. HYPOTHESIZINGdeciding what it is you want to study.
2. DESIGNINGestablishing the procedures and methods to use.
3. PLANNINGfiguring materials and personnel needed.
4. FINANCINGarranging support for the survey.
5. SAMPLINGchoosing which people are to be interviewed.
6. DRAFTINGframing the questions for use in the field.
7. CONSTRUCTINGplanning the format of the questionnaire.
8. PRE-TESTINGdetermining whether the questions elicit the data desired.
9. TRAININGteaching interviewers how to gather information correctly.
10. BRIEFINGshowing interviewers how to use the questionnaire.
11. INTERVIEWINGsecuring data from respondents.
12. CONTROLLINGseeing that the interviewing gets done.
13. VERIFYINGassuring that the collected data are accurate.
14. CODINGpreparing the data for analysis.
15. PROCESSINGorganizing data mechanically or electronically.
16. ANALYZINGinterpreting the data.
17. REPORTINGsharing the new knowledge.
Source: Backstrom and Hursh (1976).
''-'Thorough reviews of survey methodologies may be found in Bouchard
(1976); Backstrom and Hursh (1975); Hansen, Hurwitz and Madow (1960); and
Lansing and Morgan (1974). Costs cited are based upon selected research
studies at Abt Associates in 1979-80.
4-23
-------
Survey Type
The first and most important action in designing the survey pro-
cedure is the selection of approach from among personal interviews, guided
interviews, telephone questionnaires and mail questionnaires. The selec-
tion of approach will influence the instrument design, as discussed above,
and the cost of the effort and perhaps the resulting samples. The study
of visibility aesthetics generally necessitates the more expensive personal
interview or guided interview approaches. For personal interviews the
field effort costs vary greatly, from $50 to $150 and up per interview,
depending upon length, geographic location and whether or not individual
respondents are preselected. The advantages of a personal interview are
an increased amount of interaction and clarification between respondents
and enumerators, less demands upon the format and presentation of the
survey instrument and that respondents can be interviewed whenever a
mutually acceptable time can be arranged.
Guided interviews of one or more respondents simultaneously often
reduce costs by up to one half depending again on geographic location and
the type of sample. Because many respondents usually must start the
survey simultaneously, a random selection process of respondentswho may
be forced to wait momentarily or come back at a later timemay be con-
siderably hampered. For example, if only respondents with a strong in-
terest choose to stay or return, the results will be potentially non-
random and biased. The actions of some respondents in the process also
may influence and bias other respondents through the transmittal of
information clues which suggest appropriate answers.
The implementation costs of telephone surveys are on the order of
one half or less of the costs of personal interviews. While random selec-
tion procedures can be easily applied, the response rates (self-selection
to continue the questionnaire) is considerably diminished with lengthy
telephone questionnaires. The rapport between respondent and enumerator
is also often considerably reduced as is the ability to use any props to
present alternative scenarios. This diminishes the effectiveness of
surveys that are concerned with intended behavior based upon incremental
changes in air quality: A verbal description of "a little more polluted"
or "visual range of about 40 miles" is much less effective than the
typical visual aids. This generally eliminates using telephone surveys
for air quality demand studies. Telephone surveys are best used to gather
perspectives, beliefs, and actual or intended behavior in well-defined and
familiar situations.
The cost of mail questionnaires is typically one fourth or less of
those of personal interviews unless elaborate materials are also required.
Picture sets, for example, may drastically increase the cost of presenting
alternative levels of air quality in mail questionnaires. Mail question-
naires are also notorious for having the lowest response rates and a
4-24
-------
greater need for validation. Mail questionnaires require very careful,
simple formatting, and they quite obviously rule out the type of interac-
tion that comes with the iterative bidding procedure. While they have been
successfully used to evaluate wildlife stocks (Hammock and Brown 1974, and
Brookshire et al. 1977), we are not aware of a validated attempt to use
this approach to compare alternative scenarios of visibility aesthetics.
Sampling Procedures
Most survey research efforts face the problem of obtaining a random
sample that is also a representative cross-section of the affected popula-
tion. This may have been an especially important problem area for several
previous bidding method surveys which have experienced high rates of
unwillingness to participate by the initial randomly selected respondents
(see Chapter 6). Responses cannot be obtained from all individuals who
might be affected by a proposed project because both the cost of doing so
and the volume of data generated would be overwhelming. Consequently, a
sample of the population must be selected from which to infer characteris-
tics of the population as a whole. Generally, a "probability sample" is
designed where every individual has a known, although perhaps purposefully
unequal, probability of being selected and where the resulting sample is a
representative cross-section of the population. Such sampling must employ
a predetermined yet completely random sample to insure the validity of the
results and subsequent statistical analysis.
The probability sample must determine the number and kind of
respondents, the probability any respondent would be included, the error
involved from interviewing a sample rather than the whole population, and
the degree of confidence in estimates made from the sample. One of three
types of samples are usually designed and implemented. Where all indivi-
duals are considered to be alike (homogeneous) a simple random sample may
be implemented. Where there are several different groups and where the
individuals within a group are homogeneous, a stratified random sample,
which attempts to insure that a minimum number of respondents within each
group sampled, is performed. These approaches are aimed at sampling a
particular random individual. Cluster samples are aimed at interviewing
selected individuals from a randomly selected cluster of individuals, such
as a household. A cluster sample simplifies efforts by determining the
probability of being sampled and drawing actual samples for clusters of
individuals, such as a household, rather than for all individuals in the
population. Because household data are readily available and by randomly
preselecting households and the members of households to be interviewed a
random survey can be implemented with the cluster approach at a lower cost
than for a simple random survey.
To implement a sample plan for urban locations, basic population
characteristics are often needed, as are census tract information or local
directories and a random selection procedure. For example, to interview
members of selective households, information on the number and location of
4-25
-------
residences is required to calculate the probabilities of sampling any one
household. A random selection procedure must next be implemented, such as
selecting 100 sets of random blocks, households on the block and individuals
within the household. The interviewer is then sent to the northeast (or
other randomly selected corner) of the first block, turns right (or left)
and interviews the selected individual in the first selected household.
If the interview cannot be completed, a callback time is established. The
interviewer then proceeds to the second selected block, etc.
Similar procedures must be implemented to obtain random samples at
recreation areas. For example, obtaining a simple random sample with
uniform probabilities of sampling an individual will depend upon how
interviews are solicited. If interviews take place at camp sites, it is
imperative to know the volume of use at each area. If interviews take
place at park entrances, then more interviews will need to be conducted at
high volume periods than at other times. As noted by Backstrom and Hursh
(1976):
In summary, the cardinal rule in sample drawing is to leave
as much as possible to chance. When the researcher allows
flip-of-the-coin philosophy to dictate his every move, he is
assuring himself (and his critics) that whatever happens is
completely devoid of his personal biases. In sampling, it
is well to admit that almost anything you "decide" to do,
rather than letting chance decide for you, is a contami-
nation of the survey. (Page 64.)
Sample Size Determination and Verification
The necessary sample size is determined by a combination of
inputs. These include the error limits and variances associated with the
expected responses, the statistical tests for which the data are to be
used and the expected completed response rate. These inputs are determined
in part from the analysis planned and from the pretest of the survey
instrument. Many previous visibility benefit surveys have employed samples
of insufficient size for complete analysis.
A rule for every survey is that it must be pretested in the manner
and location it will actually be applied. This identifies problem questions,
error variance estimates, and response rates. The pretest sample size
should be sufficient to determine these values and to test the robustness
of the approach.
With estimated response rates and the calculated number of completed
interviews required, the total number of interviews, the number of inter-
viewers and the days in the field that will be required can be calculated
in order to appropriately plan the actual survey effort.
4-26
-------
A potential bias that has occurred in surveys concerning environ-
mental issues is that of self-selection by the respondents. Many respon-
dents, upon learning the intent of the survey effort, refuse to partici-
pate. If this participation rejection, or self-selection procedure, is
non-random, so may be the results. For example, which individuals are
more likely to allow enumerators to enter their homes, or are more likely
to return and/or wait at a group interview site once the intent of the
questionnaire is known?ll Those most likely to have an interest, some
who feel obliged, and a few with time on their hands? The possibility of
higher response rates by those with a special interest may bias the sample
if this effect is not tested and accounted for.
One way to test and adjust for self-selection bias is to identify
two to three variables which are highly correlated with the values received
in the pretest. For a study in Class I areas these might include residence
location, environmental stance, recreation rates and income. Next, the
rates at which these are found in the survey may be tested against the
actual population totals. This can be accomplished in some cases with
secondary data. On the other hand, if the sample differs from the general
population by some important characteristics which cannot be determined
with secondary data, such as environmental stance, the results of the
survey may have a bias that is unknown and cannot be corrected. This may
be a problem in bidding method surveys which often have experienced high
rates of unwillingness to participate. It may be possible to offset this
by a very low-cost 4-question random personal interview at a park entrance
or through a city-wide telephone interview. An example at a park might
read as follows:
Hello, my name is and I am with . We are
doing a park user survey which takes about one
minute. Could you please answer the following question.
(If no, attempt to establish a call back time.)
Ql. In what city and state is your residence located?
Q2. Have you recreated at this site in the past?
Q3. How many days have you spent in national parks
this past year?
Q4. Using this card would you please tell me how you
rank yourself on environmental issues? (some
scales)
Thank you very much.
successful technique of increasing response rates at private residence,
suggested by Mark Thayer and applied by Brookshire et al. (EPA 1981), is to
introduce yourself and intent, then ask respondents to step outside their
home to view a display on their front lawn. Respondents' homes will not
be invaded, and with a sufficiently attractive display they are more easily
enticed to cooperate in the survey.
4-27
-------
If response rates to these or other appropriately selected questions
are significantly different statistically from the sample response rates,
the sample valuations should be appropriately adjusted and aggregated.
Post Survey Data Verification and Analysis
Once survey data are obtained, fairly standard statistical prac-
tices for validation, analysis, and reporting must be used. This section
suggests several minimum efforts that should be made.12
First, all results should be reviewed and verified with interpre-
tations and coding schemes checked before and after the coding and data
entry tasks. Misinterpretation of results or errors in coding or data
entry can easily invalidate many hours of well planned survey efforts.
For example, unanswered WTP or WTA questions, which may frequently occur,
will need missing value codes recognized by the statistical analysis
routines. If these values are left blank and, in some programs, inter-
preted as zeros, all subsequent analyses will be invalidated.
In general, data analyses and presentation should follow and
document standard statistical procedures in a consistent manner. Several
types of analyses are almost always reported at least in working papers
for any contingent market analysis. This should include tallys on the
survey procedure, including the total number of respondents contacted,
response or rejection rates for the whole questionnaire, and critical
questions. Mean response rates to the bidding questions and their standard
errors (standard errors of the means) and the number of observations upon
which the means were calculated should be reported. Particularly useful
data in bidding methods are mean bids, zero bids, and rejection bids for
each scenario according to each stratification, such as by income classes,
accompanied by the number of observations and the variances. Next, fre-
quencies, correlations, and cross-tabulations of the bids, socioeconomic
characteristics of the sample, and other variables of interest may be
calculated. These will allow the researcher to check that the sample is
representative of the population, to check for obvious data entry problems,
and initially to evaluate the distribution of responses and correlations of
variables.
Next, tests of specific hypotheses may be carried out, such as the
effect of questionnaire design or socioeconomic influences upon the bids
received. Because many different influences may be reflected in any
particular bid it will be necessary to use multivariate analysis procedures
such as regression analysis to separate out and test the influence of any
one variable upon the bid structure. The depth of analysis that can be
performed using regression analysis varies from ordinary least squares
to simultaneous equation models with econometric tests of all underlying
classical regression model assumptions. At a minimum, regression models
reader may consult most any statistics or econometrics text for
guidance, including Pindyck and Rubinfeld (1976), and Intriligator
(1978).
4-28
-------
explaining the bid distribution (bid functions) should be tied to models
of consumer behavior and specifications of underlying utility functions.
Because the "true" form of the respondents utility function and subsequent
regression models is likely to be unknown, examining the robustness of the
analysis by varying the functional forms is suggested. All regression
results reported should at least include estimated coefficients, the
number of observations, F statistics, t-statistics, and R^'s.
Tests of a hypothesis should be careful to specify whether and why
a test is two-tailed or one-tailed. Many past air quality studies have
been flawed by the incorrect use of two-tailed statistical criteria when
one-tailed tests were appropriate. For example, a null hypothesis may test
that total willingness to pay for air quality is not affected by income.
The alternative hypothesis suggested by economic theory is that total
willingness to pay will increase with increases in air quality or income.
Consequently, a one-tailed test would be appropriate.
4.3 Residential Property Value Studies
Residential property has been the market good most frequently
studied to estimate the value of air quality from actual market data. The
primary hypothesis behind this approach is that residential property
values are influenced by the level of air quality at the residential site.
Recently, similar approaches have been applied to recreational and com-
mercial property and to other types of nonmarket amenities, such as noise
and accessibility.^ Residential property value studies are emphasized
here because to date they are the most successful efforts of this type for
estimating the benefits of air quality.
The use of residential property value differentials to estimate
how much consumers are willing to pay for various levels of air quality
began with the Ridker and Henning (1967) study in St. Louis. It was based
on the supposition that if air quality varies across the area and if
people are willing to pay more for a residence with better air quality,
the amount they are willing to pay will be revealed by the price dif-
ferences between properties that are similar in all respects except air
quality. None of these types of studies have attempted to separate health
effects from aesthetic effects of air quality. Therefore the discussion
in this section refers primarily to air quality, although the potential for
isolating visibility aesthetics will be discussed.
The variations and refinements of the Ridker and Henning approach
that have been developed since 1967 are based largely on the hedonic price
technique. The following section introduces this technique and explains
the procedure in a non-technical manner. However, functional notation and
graphical presentations are necessary to explain this technique and its
strengths and weaknesses. The more technical application details are
presented in Section 4.3.2. Anyone interested in applying this technique
or in acquiring a thorough understanding of its application details will
examples see Wilman 1981; and Palmquist 1981.
4-29
-------
need to master this material. Two variations of the property value
studies that have been used are the residential location model and the
matched pairs technique. These are briefly described in Section 4.3.3.
The strengths and weaknesses of the property value approaches are presented
in Section 4.3.4.
4.3.1 Introduction to the Hedonic Price Technique
Economists usually look at the prices paid for goods and the
quantities purchased to determine how much people value a market good.
With this information a demand curve can be estimated from which the
benefit measures defined in Chapter 2 are derived. With non-market goods,
such as visibility aesthetics, there are no direct prices and quantities,
but in some cases the selection and price paid for a market good, such as a
residential property, may be influenced by visibility levels. In this way,
analysis of air quality levels and residential property prices may reveal
the value that people place on visibility aesthetics. The hedonic price
technique can be used to estimate the implicit price for a non-market good
that is associated with and influences the value of a market good.
Conventional consumer behavior theory postulates that consumers
derive utility directly from goods and services that are each obtained in
homogeneous (identical) units. However, many goods are not purchased in
homogeneous units. For example, different houses have different numbers
of rooms, distances from employment centers, yards, and views. The
hedonic price technique, as developed by Griliches (1971) and Rosen
(1974), attempts to analyze the value of such goods by breaking them down
into the attributes that they embody. Hedonic price theory suggests that
consumers select the particular bundle of attributes that they desire when
purchasing a differentiated market good, such as housing, and that it is
from these attributes that consumers derive utility. In other words, the
household values a residential property because it embodies particular
attributes that the household desires, such as places to sleep and eat, a
pleasant environment, and access to employment and recreation. This
approach is related to Lancaster's (1966) modification of consumer behavior
theory, which is that consumption is an activity into which goods and
services are inputs and from which attributes or characteristics are
derived. The hedonic approach, however, does not suggest that households
produce these attributes themselves, as does Lancaster, but that they
select from among the bundles of attributes contained within available
market goods such as housing.
The amounts of the attributes in a market good determine the
prices that consumers are willing to pay for the market good. If these
attributes can be defined and quantified by a consistent unit of measure,
then by observing different units of a market good with different levels
of attributes, such as houses with different amounts of rooms, and air
quality, the effect of each attribute on the price of the market good can
be determined. This relationship between the price of a market good and
4-30
-------
its attributes is called a hedonic or implicit price function and its
estimation is the first step of the hedonic price technique. The hedonic
price function enables the analyst to statistically hold all attributes but
one constant and to see how the level of one attribute affects the price of
property. It can answer the question of how differences in visibility
would affect prices of property if each residence were identical in every
other respect.
From this hedonic price function a marginal implicit price for each
attribute can be derived. The marginal implicit price of an attribute is
the addition to the price of the good that must be paid to obtain a good
with an additional unit of the attribute. If one of these attributes is a
measure of visibility, deriving a marginal implicit price for visibility
from a hedonic price function would be a first step toward estimating the
willingness to pay for visibility. The next step requires observation of
how marginal implicit prices for visibility change when visibility levels
change either across locations or across time.
The most recently published property value studies that have
estimated the value of air quality as an attribute of residential property
have been based on the theoretical framework developed by Rosen (1974)
and Freeman (1974). The Rosen and Freeman analysis cleared up some of the
problems and confusions in interpretation of property value differentials
in the earlier studies primarily by pointing out that observed differences
in property values attributable to differences in air quality can only be
interpreted as willingness to pay for that difference in air quality under
special and typically unrealistic conditions. However, the procedure that
they both suggested for estimating WTP from a hedonic price function has
been a subject of recent debate. Suggested modifications to this approach
have not yet appeared in published theoretical or empirical work and are
still in formative stages. Therefore the unmodified Freeman and Rosen
framework is presented here with its current shortcomings. The reader
should be aware that recommended changes in the WTP portion of this
approach are likely to be forthcoming.
The following simplified example illustrates a hedonic price
function for residential property. A few attributes of a residential
property a household may value include:
Al = size of house
A2 = distance from work site
A3 = size of yard
A4 = neighborhood crime rate
A5 = quality of neighborhood schools
A6 = air quality at house site
For purposes of this illustration assume that these six attributes are
those that are important to households when selecting a residence and that
these attributes are observable and quantifiable. A list of all relevant
residential property attributes would, of course, be much longer. A linear
form of the hedonic price function for residential property would be:
PV = bO + blAl + b2A2 + b3A3 + b4A4 -I- b5A5 + b6A6
4-31
-------
where:
PV = residential property value
bO = a constant
bl,...,b6 = coefficients of each corresponding attribute
The analyst obtains data for the prices and attributes of the good and
uses regression analysis to estimate the coefficients of the function.
Residential property value, presumably the sale price it could obtain, is
given here as the price of the housing good. This choice and other possi-
bilities are discussed in detail in Section 4.3.2.
From this hedonic price function a marginal implicit price can be
derived for the air quality at the residence. In this case the estimated
value for b6 would be the marginal implicit price for air quality. It
indicates how much additional money the household must pay for each incre-
mental improvement in air quality at the residence. The household cannot
buy improvements in air quality at a given site, but it can choose among
residences that are similar in all respects except the level of air
quality. A linear hedonic price function implies that the marginal
implicit price of air quality is a constant.
A linear form for the hedonic price function, however, will prob-
ably not be appropriate because it is impossible to repackage housing
bundles without additional cost. For example, a linear hedonic price
function implies that the value of two 750-square-foot homes is equivalent
to the value of one 1,500-square-foot home. Though the amount of materials
in each may be the same, converting the two houses into one or the one into
two would be a difficult and expensive undertaking. A nonlinear hedonic
price function implies that the marginal implicit price of air quality and
other attributes is not constant.
The general form for the hedonic price function for property
value, PV, is:
PV = PV (Ai,...,An)
where:
Ai = the quantity of the ith attribute of the property
The marginal implicit price of attribute Ai is the partial
derivative of the hedonic price function with respect to attribute Ai,
3PV/3Ai. This derivative will be a function of Ai and possibly other
attributes unless the hedonic price function is linear. For example,
an increase in air quality where the air quality is poor might obtain a
higher marginal implicit price than an increase in air quality where air
quality is good. Also, a house with large windows might obtain a higher
marginal implicit price for air quality than a house with small windows.
4-32
-------
Examples of a nonlinear hedonic price function and a marginal
implicit price function for air quality are given in Figure 4.1. PV(Q) is
the relationship between property values and air quality taken from
a hedonic price function for property holding all attributes but air
quality constant. The marginal implicit price function for air quality,
which is designated as PV'(Q), is the partial derivative of the hedonic
price function with respect to air quality, 9PV(Q)/9Q. PV'(Q) is thus
the slope of PV(Q). For many nonlinear forms of the hedonic price func-
tion, the marginal implicit price of air quality will be a function of
other property attributes as well as of air quality. PV'(Q) in Figure
4.1b is therefore the relationship between the marginal implicit price of
air quality and the level of air quality holding all attributes but air
quality constant.
The goal of the analysis is to estimate the value of air quality
so that the benefits of an improvement or the damages of a deterioration
in air quality can be evaluated. Identifying a marginal implicit price
function for air quality is, in most cases, just a first step. The second
step in the hedonic approach is to estimate the demand function for air
quality at the residence. This is the relationship between the marginal
implicit price of air quality and the level of air quality at which the
household will choose to locate. It must also account for the effects of
household income, household size, and other household characteristics
and property attributes that influence the household's demand for air
quality.
This demand function is usually called the WTP function because air
quality is a public good and the household is limited in its choice of how
much air quality to consume. For a market good the demand curve tells how
much the consumer will buy at a given price. The WTP curve tells how much
the consumer would be willing to pay for each increment of air quality.
They contain the same information, but in one the dependent variable is
quantity and in the other it is price. The marginal implicit price func-
tion for air quality is a relationship between the price and the level of
air quality, but it is not normally the same as the WTP function because .It
is the result of supply and demand interaction. The prices are established
where the supply of residential site air quality equals the demand for
residential site air quality at each air quality level.
The marginal implicit price for air quality that each household is
paying may be considered one point on its WTP curve. Two possible willing-
ness to pay curves are shown in Figure 4.1b. These are Wi(Q) for household
i and Wj(Q) for household j. Like normal demand curves they are downward
sloping and they intersect the marginal implicit price function at the
marginal implicit price and air quality combination at which the utility
maximizing household will locate. Because each individual household will
typically have an inconsequential influence on the demand and supply of
residential property, it will face the market determined marginal implicit
prices for all property attributes. A household will maximize its utility
by choosing a property with levels of each attribute where its marginal
willingness to pay for each additional unit of the attribute just equals
the marginal implicit price. Hence, household i chooses a residence with
air quality level Qi and household j chooses a residence with air quality
level Qj.
4-33
-------
PV (Q)
PV
>
>,
0,
o
Air Quality
Figure A.la
Hedonic Price Function for Air Quality
PQ
11
o
X -3 PQI
-I 3
M M
H
fH
-------
PQi can be considered household i's marginal willingness to pay for
the ith unit of air quality and as such represents the household's willing-
ness to pay for a marginal (very small) change in air quality. However,
public policy questions typically require estimation of the household's
willingness to pay for non-marginal changes in air quality, for which the
WTP curve must be known.
The WTP functions for air quality are estimated by deriving the
marginal implicit price from the hedonic price function that each household
is paying and taking this as a measure of its willingness to pay for the
air quality level at its residence. By using household income and other
household characteristics to group together households that are similar, a
set of willingness to pay functions can be estimated. The estimated
willingness to pay function is therefore:
PQi = Wi (Qi, Mi, Zil,...,Zin, Sil,...,Sim)
where:
PQi = the marginal implicit price for air quality being paid by
household i
Mi = the ith household's income
Zil,...,Zin = other characteristics of household i
Sil,...,Sim = other attributes of household i's residence
However, because the implicit price function is also influenced by
supply side variables, estimation of WTP requires consideration of the
supply of residential site air quality as well. This will be discussed in
Section 4.3.2.
Consumer surplus measures can be derived from the WTP curves. In
this case, because the individual is able to vary his consumption bundles
but appears to have no property rights to any level of air quality, other
than what he purchases in the market, the compensating variation measure is
appropriate. However, since the error in the WTP estimation will be likely
to greatly exceed the error in approximation when using ordinary consumer
surplus, the OCS measure is typically used. As was shown in Chapter 2 this
is an approximation of the change in welfare, or equivalently, an approxi-
mation of how much households value the change. This is illustrated in
Figure 4.1b. Household i lives at a location with air quality Qi and
household j lives at a location with air quality Qj. Each household's
willingness to pay for an improvement in air quality from Q to Q* at its
locations is shown by the slashed area under each household's willingness
to pay curve. This is the integral of the WTP function between the original
4-35
-------
and the improved air quality levels. It represents an OCS measure of the
benefits to each household of the improvement in air quality, because it
indicates how much the household would be willing to pay for the air
quality improvement in excess of the expenditures it is currently incurring.
The application of the hedonic price technique requires several
assumptions. The first of these is that households are sufficiently
mobile and aware of air quality impacts to cause current property values
to reflect the marginal willingness to pay for the current air quality
conditions. This will ensure that PQi in Figure 4.1b is a point on house-
hold i's marginal WTP curve. For this to occur households must have enough
information about air quality in a neighborhood in which they buy or own a
residence that they can make a reasonable judgment about what the impact of
living in such conditions would be, and there must be enough competition
among buyers that prices reflect the marginal value of the properties to
the households that purchase them. When actual sales data are used in the
hedonic price study this assumption will be valid as long as households
choose optimal (utility maximizing) residences when they move, but trans-
actions costs may keep households from moving when changing conditions have
rendered their current residences non-optimal for them. Even if owner or
appraiser estimated property values reflect current market conditions it
may not be valid to assume that the current market value reflects the
occupant's marginal willingness to pay for a property that has not been
recently purchased. Freeman (1979b) argues that making this assumption
will introduce error but not necessarily bias unless market conditions have
been changing consistently in one direction and households have been
lagging behind in their adjustments to the changes.
The second assumption is that there is continuous variation in
each attribute, Ai, and that all combinations are available so that
consumers can increase their consumption of Ai by finding another resi-
dence that is identical in all other attributes but embodies more Ai. The
observed price will not be equivalent to marginal WTP if the combination of
attributes that would maximize the household's utility given current prices
is not available. Freeman (1979b) points out that violation of this
assumption may be why Harrison and Rubinfeld (1978) found that some high-
income households chose to live in high-pollution areas even though they
had a higher willingness to pay for air quality. He suggests that this may
be because they wanted a combination of low pollution and high levels of
some other attribute, such as proximity to the cultural amenities of a
downtown area, that was not available.
The third assumption is that all relevant attributes must be
observable and quantifiable. The transition from a model that maintains
that a good embodies attributes which are valued to the empirical realiza-
tion of it that permits the estimation of marginal implicit prices for
these attributes is quite difficult. The application of the hedonic price
technique assumes that the price households are willing to pay for a
residence can be fully explained by specific measures, such as the number
of rooms, the distance from a shopping center, the amount of crime in the
4-36
-------
neighborhood, and the amount of air pollution in the neighborhood. This
requires that, for example, a technical measure of air pollution levels be
used to sort out which of the household's decisions and actions are
attributable to air quality conditions. This presumes that the technical
measure is a reasonably accurate reflection of the household's perceptions
of air quality. If all relevant attributes are not accurately specified,
variations in property value that actually result from differences in
attributes that have been overlooked or incorrectly measured may be
inaccurately attributed to the attributes that have been used.
4.3.2 Application
Applying the hedonic price technique to residential property is a
complicated task because data must be put to a use for which they were not
originally collected and because air quality is only one small part of all
the factors that determine residential property values. This section
presents some of the more important theoretical and practical details that
must be considered.
The Hedonic Price Function: Dependent Variable
The dependent variable of the hedonic price function is the price
per unit of the market good, in this case residential property, that is
being examined. It is not entirely clear what price should be used in a
property value study. Freeman (1979a) argues that the value of land would
be the best price to use when trying to isolate an implicit price for air
quality because the complications caused by variations in the housing
structure could then be avoided. However, land and the structure on it
are usually sold as a unit for a price that reflects their combined value,
hence information on property values is more readily obtained.
It is theoretically possible to use either or both cross-sectional
or time series data because air quality can vary across locations and
across time. However, most studies have used only cross-sectional data or
cross-sectional data over a short time period, such as a year or two.
There are two primary reasons for this choice. One is that changes in
property values over time are probably more complex and therefore more
difficult to explain than differences across locations. The second is
that a major data source for many of the studies has been the U.S. Census
of Population and Housing, which is conducted every ten years and is
therefore not a suitable source for time series data especially because
air quality has only begun to be monitored during the past ten to fifteen
years in most major urban areas.
The appropriate measure of property value for a hedonic price
function is the price that the property would obtain if sold during
the study period. However, only a fraction of residential properties are
sold during any given time period. The analyst must therefore choose
4-37
-------
whether to use the subset of the properties that are actually sold during
the study period or whether to use estimates of the market value of all
the properties in the study area. Data on property sales including price
and property characteristics are available from local government, market
research firms, real estate organizations, or financial institutions.
Estimates of the market value of properties that have not been recently
sold can be obtained by professional appraisal of all individual properties
in the study area or from county records kept for tax purposes. The
latter source should be used with caution, however, because political
motives and infrequent reappraisal may lead to inaccurate records of
property value. Census tract averages of the owner-estimated value of the
properties reported by the U.S. Census of Population and Housing are
frequently used because they are readily available, but some information
is lost in averaging the data within a census tract and the census data do
not cover all the variables that might be desired for estimating a hedonic
price function for residential property. The San Francisco Bay Area study
(Loehman et al. 1980) tested for the effects of different levels of aggre-
gation and found that though results are influenced by the level of ag-
gregation, household level analysis may not be necessary.
The above discussion refers to owner-occupied housing, but a
significant fraction of housing is rented. The observable price in the
rental market is the annual or monthly rent. Some studies have estimated
separate hedonic price functions for these two markets though most have
looked only at owner-occupied housing. However, the relationship between
property values and rents can be used to convert either one to the other
so as to combine rental and owner-occupied housing prices if the researcher
chooses to put them into one hedonic price function. Though most studies
have used only owner-occupied housing prices the same sort of conversion
must be made to translate benefit estimates based on property values to
annual benefit figures.
A residence is a long-term asset from which the occupant receives
benefits over many year; therefore, the price that a household is willing
to pay to purchase a house is the present value of this expected future
stream of benefits. Similarly, the annual rent that the household is
willing to pay is equal to the annual benefit expected from occupying
the house. To obtain a present value, future benefits are discounted
because a benefit received next year is worth less than the same benefit
received this year due to the intervening time. For example, consider a
household that is buying a property for which it would be willing to pay
an annual rent of $5,000. Suppose also that the interest rate it could
get by putting its money in the bank instead by buying a house is 10
percent. The present value, in this case the value of the property to the
household, of a perpetual stream of $5,000 in benefits annually when the
discount rate is 10 percent is:
$5,000 = $50,000
.10
In general, where
4-38
-------
r = annual rent (benefit)
R = property value (present value)
i = the discount rate
the approximate relationship between annual rent and property value
is:
r = Ri.
This formula assumes that the annual benefits will continue indefinitely
and unchanged whereas in reality the household's time horizon is finite
and benefits will be offset by increasing upkeep costs as the house ages.
More complex discounting formulas can be used if a more accurate approxi-
mation is required (see Chiang 1974).
Similarly, a benefit estimate for changes in air quality based on
property values will be the present value of the stream of expected future
benefits of the change in air quality. It can be converted to an annual
benefit figure in the same way that property values can be converted to
annual rents, but some assumptions must be made about whether households
expect current air quality conditions to remain unchanged when they
purchase a residence and about how long the new air quality condition will
exist.
The Hedonic Price Function: Independent Variables
The independent variables of the hedonic price function are the
attributes of the market good being examined. Not only must the appropriate
price data for property be chosen, but all the appropriate data to describe
all the relevant property attributes must also be selected.
To estimate the hedonic price function for residential property.
the independent variables must include all the attributes that influence
the value of the property.^ The choice of these attributes has in
practice been very dependent on data availability and has varied signifi-
cantly from one study to the next. These attributes can be separated into
two categories: house structural variables and neighborhood characteris-
tics and amenities. Table 4.5 gives some examples of variables that have
been found to influence property values in studies that have estimated
willingness to pay for air quality from property values.
^ This difficulty may eventually be avoided in some cases with the use of
the repeat-sale technique described by Palmquist (1981). This new
approach observes changes in sale prices of homes before and after an
environmental change has occurred. Because the same homes are being
observed over time the characteristics other than environmental quality
are unchanged and therefore do not need to be considered.
4-39
-------
Table 4.5: Property Attributes Found Significant in Air Quality Studies
House Structure Neighborhood Characteristics
Age of House Property tax
Number of Rooms Percent Non-white
Lot Size Population/Square Mile
Number of Bathrooms Crime Rate
Air Conditioning Percent Business Acreage
Sales Date School Quality
Pool City Services
Fireplace Accessibility to Transportation Facilities
Persons per Room Distance to Central Business District
Square Footage Noise
Proximity to Recreation Facilities
Proximity to Employment
Air Pollution
Source: Loehman et al. (1980) p. 80
The differences between studies in property attributes used in the
hedonic price function may not be as problematic as they first appear,
because many of the attributes are correlated with one another. For
example, the number of square feet and the number of bathrooms in the
house may be slightly different measures of the capacity of the house.
The analyst must be careful, however, to identify all the attributes that
influence property values, because the influence of omitted attributes
will be statistically attributed to the attributes included in the hedonic
price function if they are correlated.
If the analyst is only concerned with separating the effect of air
quality from the other attributes, incorrectly attributing the influence of
the number of bathrooms to the number of square feet in the house will not
be a problem; however, if omitted attributes are correlated with air
quality, the impact of air pollution on property value will be either over
or understated. For example, air pollution typically decreases as the
distance from the central business district (CBD) increases. If the
distance from the CBD is omitted from the hedonic price function, the
negative influence of air pollution on property values will be understated,
because proximity to the CBD (easier access to theaters, stores, employ-
ment) can be expected to exercise a positive influence on property values.
The benefit of being close to the CBD will counterbalance the damage of the
increased air pollution close to the CBD, and the negative influence of
increased air pollution on property values will appear to be smaller than
it actually is. Alternatively, air pollution may be positively correlated
with traffic noise. If traffic noise is omitted, the pollution will pick
up the negative influence of traffic noise as well and be overstated.
4-40
-------
Until now air quality has been discussed as if it were a unique,
measurable attribute of a property or neighborhood, but in fact the
most available measures of air quality are the levels of monitored air
pollutants. In most urban areas these include carbon monoxide, total sus-
pended particulates, oxides of nitrogen, oxides of sulfur, and ozone, as
well as some composite measures such as the Pollution Standards Index. The
analyst must choose the pollution measure or measures that best reflect
the air pollution impacts that influence property values, taking into
account the availability and quality of the pollution data.
A serious problem for the accurate interpretation of the effect of
the selected pollution measure on property values is that many of these
measures are highly correlated with one another. This makes it difficult
to separate the property value impact of one pollutant from the impacts of
the other pollutants, even though each pollutant tends to have a different
kind of impact on people and property. In a situation where different
pollution measures are highly correlated, one measure must be selected as
a proxy. The estimated marginal implicit price for the selected pollution
measure will then to some extent reflect the impacts of all the correlated
pollution measures. There is no way to entirely correct this problem, and
the analyst must check for its seriousness and take this into account in
interpreting the results of the hedonic study.
Most of the studies have chosen one measure of pollution as a
proxy for air quality and have not attempted the difficult task of
separating health effects from aesthetic effects. If the hedonic price
technique is to be used to measure the value of visibility, a measure of
visibility must be developed that is distinct from the other impacts of
air pollution. Measures of visibility are discussed at length in Chapter
3. Though visibility alone has not been used in a hedonic price study,
some pollutants used have more visibility impacts than others and visual
impacts may very well be those most readily perceived and therefore
reflected in property values. The difficulty would be in trying to
separate the visual from the health and other impacts. If a single
pollutant causes more than one type of impact it may not be possible to
statistically separate their effects on property values.
Available pollution data are limited because the number of
monitoring stations in an area is limited and may not correspond to the
property locations being studied. Some of the recent studies (Harrison
and Rubinfeld 1978; and Bresnock 1980) have used air dispersion models
that produce estimates of air pollution levels over a grid of the area
using monitor station and wind pattern data. These models increase the
flexibility of the property values study, but the accuracy of the study
then depends on the accuracy of the dispersion model.
4-41
-------
The Hedonic Price Function: Functional Form
To estimate a hedonic price function, a specific functional form
must be selected.15 An example of a linear hedonic price function was
given in Section 4.3.1, and several of the earlier studies used this form.
However, the hedonic price function does not have to be linear since
it is not possible to repackage housing bundles without additional cost.
Additionally, without any variation in marginal implicit price over the
study area one cannot observe households' responses to different prices of
air quality and therefore cannot estimate a willingness to pay curve for
air quality. Fortunately, both a priori considerations and empirical
evidence support the assertion that nonlinear functional forms give a
superior hedonic price function to linear functional forms.
An important consideration in the selection of the functional form
for the hedonic price function is the' nature of the relationship between
property values and the air quality or pollution measure being used.
Air pollution is an attribute of residential property that has negative
utility for households and therefore has a negative price. Air quality
would have a positive price. It may sometimes be possible to use positive
measures of air quality, such as visual range, but because measures of air
pollution are more readily available and have been used in most of the
property value studies they will be emphasized here. Beyond the expecta-
tion that as air pollution increases property values will fall, the shape
of this relationship is not known and is not predicted by consumer demand
theory. The appropriate functional form for this relationship must
therefore be determined empirically. Transformations of the pollution
data, such as using the reciprocal in order to obtain a measure of air
quality, impose specific functional forms on the pollution and property
values relationship. Therefore, such transformations should not be used
arbitrarily and, if determined to be appropriate, should be used con-
sistently throughout the analysis.
Figure 4.2 shows a possible relationship between air pollution and
property values. The upper quadrant shows that as pollution increases
property values fall at an increasing rate. Therefore the first deriva-
tive is negative and downward sloping. This is the marginal implicit
price function for air pollution shown in the lower quadrant. It implies
that the amount a household will be compensated for each additional unit
of air pollution increases as air pollution increases. Other hedonic
price functions for property value as a function of air pollution might
imply an upward sloping marginal implicit price function. However, all
such hedonic price functions can be expected to be downward sloping and
therefore all the marginal implicit price functions for air pollution will
be in the negative quadrant. WTP functions for a negative attribute, such
More detailed descriptions of functional forms specifically with regard
to hedonic price functions can be found in Freeman (1979a). General
discussions regarding applied econometrics can be found in Pindyck and
Rubinfeld (1976).
4-42
-------
Property
Values
And
Changes
In
Property
Values
PV(AP)
AP
Air Pollution
Wi(AP)
PV (AP)
Figure 4.2
An Example Of The Relationship Between
Property Values and Air Pollution
-------
as air pollution, can be expected to be downward sloping in the negative
range as shown by Wi(AP) and Wj(AP) in Figure 4.2 They can be thought of
as indicating how much a household would be willing to pay to avoid each
additional unit of air pollution at the residence, or equivalently, how
much the household must be compensated to accept each additional unit of
air pollution at the residence.
Most of the more recent studies have used some variation of the
semi-log functional form, in which the dependent variable is replaced by
its log. Figure 4.3 illustrates the implication of a simple semi-log form
on the relationship between property values, PV, and air pollution, AP-
As AP increases, PV decreases at a decreasing rate if the coefficient of
AP is negative.
Harrison and Rubinfeld (1978) found an exponential transformation
of the pollution variable preferable to entering the pollution variable
linearly. The estimated value of the exponent was approximately equal to
2. Brookshire et al. (1979) and Bresnock (1980) had similar results. All
three of these studies selected the semi-log exponential form as the best
fitting functional form with the dependent variable being the log of
property variables and entering the pollution variable raised to some
power.
Several studies have used the log-linear functional form for the
hedonic price function, where both dependent and independent variables are
converted into logs, and one has used the Box-Cox transformation (Sonstelie
and Portney 1977). As well as implying a nonlinear relationship between
property values and air pollution, these and the semi-log functional forms
imply that the marginal implicit price of air pollution depends on the
levels of other property attributes as well as on the level of air pollu-
tion. This is realistic, but it complicates the estimation of willingness
to pay for air pollution. A few studies have chosen to enter the non air
pollution attributes other than linearly, usually logarithmically or
exponentially, depending on the a priori expectations about the relation-
ship between each attribute and property values. The complexity of
the relationship between each attribute and property values makes speci-
fication of the appropriate form for each variable a difficult task that
the empirical studies have only begun to address.
The log-linear and the simple semi-log forms are more restrictive
than the Box-Cox, the semi-log exponential, and the quadratic in that the
first two impose a positively sloping marginal implicit price function
for air pollution whereas with the latter three forms the slope of the
marginal implicit price function is unrestricted. There is not necessarily
any a priori reason to expect either a positively or negatively sloping
marginal implicit price function for air quality because it is the result
of the interaction of both demand and supply. Therefore, a functional
form that does not restrict the slope of the marginal implicit price
function is preferable. Whether this slope will be positive or negative
is determined by the sign of the second derivative of the hedonic price
4-44
-------
Log PV
AP
AP
(a)
(b)
Figure 4.3
A Semi-Log Functional Form
function with respect to air pollution.^ A negatively sloped marginal
implicit price function as shown in Figure 4.2 indicates that as air
pollution increases property values decrease at an increasing rate. This
implies that households must pay more to move from high levels of air
pollution to moderate levels than to move from moderate levels of air
pollution to low levels, other things such as house quality being equal.
Studies that have experimented with different functional forms
have found that coefficients in the hedonic price function are quite
sensitive to changes in functional form. This means that the functional
form chosen for the hedonic price function will influence the benefit
estimates. Uncertainty as to the appropriate form therefore adds risk to
the use of this technique. Evidence from the empirical studies about the
appropriate functional form for the hedonic price function is discussed in
Chapter 6 and is not, at this time, particularly conclusive. This suggests
that the researcher should examine several alternative functional forms
and test the sensitivity of the final benefit estimates to changes in the
form of the hedonic price function-
^Freeman (1979a) derives the second derivative for each of the commonly
used functional forms and, assuming that the coefficient of the air
pollution variable is negative, he predicts under what conditions the
second derivative will be positive or negative. However, there are a few
errors in the table. The second derivative of the semi-log form should
be b2R and the second derivative given for the semi-log exponential form
is correct only if c = 2. For all c, it should read Pc~2 cbR [(c-1) +
cbPc)] and it will be negative if cbPc > -(c-1) as long as c > 0.
When c = 2, the second derivative will be negative if 2bP^ > -1.
4-45
-------
The Hedonic Price Function; Market Segmentation
Estimating one hedonic price function for an entire study area
implies that the housing market can be considered one market. Straszheim
(1974) questions this assumption and presents evidence that in the San
Francisco area there are several segments of the housing market that
produce different hedonic price functions when studied separately.
Freeman (1979a) argues that two conditions must be met for market segmen-
tation to produce separate hedonic price functions: (1) There must be
some barrier to mobility of buyers such as race or income that prevents
some buyers from participating in some segments of the housing market; and
(2) the structure of demand or supply must vary from one segment to
another or the hedonic price functions will be similar even if mobility is
restricted.
If market segmentation does exist, separate hedonic price func-
tions must be estimated for each segment. Empirical evidence tentatively
suggests the existence of market segmentation. Harrison and Rubinfeld
(1978) separated the Boston housing market into low, medium, and high
household income submarkets and found that the hedonic prices for air
quality varied from one submarket to another. Bresnock (1980) separated
the Denver housing market into two submarketslow-income non-white and
high-income whiteand also found significantly different hedonic prices
in the two submarkets. Loehman et al. (1980) used a different criteria and
separated the San Francisco Bay Area into two geographical submarkets,
East Bay and West Bay. These were selected as distinct submarkets because
over 80 percent of the working residents of the West Bay area were employed
in the West Bay area and over 80 percent of the working residents of the
East Bay area were employed in the East Bay area. They also found different
hedonic prices in each submarket. On the other hand, Nelson (1978) found
no evidence of market segmentation between urban and suburban areas in
Washington D.C.
Ignoring market segmentation, if it in fact exists, will distort
the estimation of benefits of an improvement in air quality. Whether true
benefits would be overstated or understated depends on the distribution of
the air quality improvement and on the WTP in each market segment. It is
not clear, however, exactly how the housing market should be segmented.
Uncertainty as to the appropriate functional form for the hedonic price
function makes it difficult to confidently test for market segmentation
because hedonic prices could appear to change from one submarket to another
submarket in the study area due to an incorrectly specified functional form
as well as due to actual market segmentation. Segmentation according to
income and/or race appears to be successful, and separation into self-
contained residential and employment areas seems to make sense, but more
evidence is needed to support the conclusion that the housing market is
actually segmented along any of these lines. Bresnock (1980) suggests that
information about actual housing transactions and real estate market
information linkages could improve the current methods for segmenting the
housing market.
4-46
-------
Estimation of Willingness to Pay
The empirical relationship between the hedonic price function and
the willingness to pay for a particular attribute is probably the most
problematic link in the work that has used the hedonic price technique to
estimate the benefits of a change in air pollution. Therefore it is
important that anyone who is interested in doing a hedonic price study or
in critically reading studies done by others understand this relationship
and its theoretical development.
Figure 4.4 is taken from Rosen (1974). This article and Freeman
(1974) developed the theoretical relationship between the hedonic price
function and WTP upon which most of the empirical work has since been
based. This presentation will use air quality rather than air pollution
because the reasoning is applicable in either case and working with a
negative attribute adds an unnecessary complication. The relation-
ship between property values and the property's air quality attribute is
PV(Q), which is taken from the hedonic price function for property values
holding all attributes except air quality constant. These observed
combinations of property values and residential site air quality levels are
determined by the equilibrium tangencies between the household bid curves
and the supplier offer functions.
The household bid curve shows the prices the household would be
willing to pay for a residence with different levels of air quality.
Therefore the household's utility is held constant along the bid curve and
the household is indifferent between the property price and air quality
combinations traced out by the curve. The position of the bid curve that
represents the highest utility obtainable by any particular household is
determined by household income and tastes and other demand determining
factors. In Figure 4.4a two such bid curves are drawn. Because the PV(Q)
function shows what any household must pay to obtain a residence with each
corresponding level of air quality, household i will locate at point A
where the household i bid curve is tangent to the hedonic price function.
The bid curve represents the highest utility that household i can obtain
and PVi and Qi is the only combination on that bid curve that household i
can obtain in the housing market. Any other point on PV(Q) would represent
a lower level of utility for household i, hence it is in equilibrium at
point A. Similarly, household j locates at point B. It is located at a
higher level of air quality because of having a higher income or stronger
preferences for clean air (possibly someone in the household has poor
health that is aggravated by air pollution).
The supplier offer curve shows the price the supplier would be
willing to accept for supplying a residence with different levels of air
quality. Profits are constant along the offer curve so that the supplier
is indifferent between locations along it. As with the household, the
supplier can only obtain a price that is on the PV(Q) curve. He will be
in equilibrium when supplying the amount of residential site air quality
4-47
-------
Supplier n
Offer Curve
PV(Q)
Air Quality At
The Residence
Figure 4.4a-
The Hedonic Price Function for Residential Property
Dollars
Sn(Q)
WJ(Q)
PV'(Q)
Qi
Qj
Air Quality At
The Residence
Figure 4.4b
The Marginal Implicit Price Function for Air Quality at the Residence
4-48
-------
that puts him on the offer curve that represents the highest obtainable
profits and at the point where the offer curve is tangent to the hedonic
price function. Hence supplier m is at point A while supplier n is at
point B. The different offer curves for different suppliers represent
differences in the resources, material, land, and technology available to
each.
For there to be a trade off between prices and air quality, as is
illustrated by the two offer curves shown in Figure 4.4a, air quality at
the residential site must affect the costs of supplying the residence. It
is clear that this would be the case for an attribute such as the number of
rooms. An increase in the price of the residence that the supplier could
obtain would compensate him for the additional cost of supplying an addi-
tional room at each residence, hence more rooms would be offered at higher
prices. It is not clear, however, that air quality at the residence is
under the control of the residence supplier. Harrison and Rubinfeld (1978)
assume that air quality at the residence is exogenous and thus unresponsive
to changes in the price of residences. Freeman (1979a) points out that
this assumption is only partially valid.
The number of houses of a given air quality can be increased
either by an improvement in air quality over the urban area,
or by increasing the number of houses available in the
region of given air quality. With present institutional
arrangements, the former can be assumed to be unresponsive
to price; but the latter is likely to be somewhat price
elastic [responsive]. (Page 127.)
This is an empirical question to which we do not as yet have a satisfactory
answer. The degree to which supply adjustment is possible will depend on
the time frame in question. Inelastic supply may be a reasonable assump-
tion for the short run.
In Figure 4.4a only two pairs of supplier offer and household bid
curves are shown, but the hedonic price function is actually formed
by the tangencies of such curves representing the entire range of house-
holds and suppliers. These tangencies represent the equilibrium points
where supply equals demand at property prices and residential site air
quality levels that will be stable as long as there is no change in supply
or demand conditions. Figure 4.4b shows the WTP curves for households i
and j, the marginal cost curves for suppliers m and n and the marginal
implicit price function for air quality at the residence. These are the
first derivatives of the corresponding functions in the upper half of the
figure with respect to air quality (i.e., PV'(Q) represents the rate
of change in property values with respect to air quality).
4-49
-------
The functions the analyst must identify to estimate the benefits
of a non-marginal change in air quality are the WTP functions.17 It is
usually postulated that a consumer's willingness to pay for a good is a
function of the quantity of the good and the consumer's income and tastes.
In the case of residential site air quality Freeman and Rosen suggest
that:
Wi = W(Qi, Mi, Zil Zin.Sil...Sim)
where:
Wi = household i's willingness to pay for Qi
Qi = the level of air quality at household i's residence
Mi = household i's income
Zil,...,Zin = other characteristics of household i
Sil,...,Sim = other attributes of household i's residence
can be estimated by regressing the observed marginal implicit price of
residential site air quality or pollution for each household against the
residential site air quality or pollution level, household income, and
other household and property characteristics. Some of the other household
characteristics that might influence WTP are size, age, and education.
Very little work has been done to date to determine whether any property
characteristics other than air pollution significantly influence willing-
ness to pay for air pollution. Most of the studies that have estimated
willingness to pay for air pollution have used a linear or log-linear
functional form.
This procedure is complicated by the fact that the marginal im-
plicit price for residential site air quality is determined by both supply
side and demand side influences. The impact of the supply side must be
accounted for or the analyst cannot be sure that WTP is being accurately
estimated. This identification problem occurs whenever there exists a
simultaneous relationship, such as supply and demand, of which only the
outcome of the interaction is actually observed.
There are two special cases that make the identification of the
WTP function easier. One is if all households have the same income,
tastes, etc., so that they all can be represented by a single WTP curve.
Then the WTP function is the same as the marginal implicit price function.
This is the implicit assumption made by Ridker and Henning (1967) when they
l'Note that the willingness to pay curves in Figure 4.4b are income
compensated because they are derived from bid curves along which utility
is constant. Though these provide theoretically preferable measures of
welfare, in practice the ordinary willingness to pay curves may often
serve as approximations of these.
4-50
-------
estimate the benefits of a reduction in air pollution by taking the change
in marginal implicit price associated with the air pollution reduction at
each property times the number of households affected. This assumption,
however, is probably not valid in most study areas because households
differ significantly in income, size, tastes, etc. If WTP functions are
actually different from one household to another, as shown in Figure 4.4b,
the assumption of identical WTP will overstate the true benefits of a
reduction in air pollution and understate damages of an increase in
air pollution.
The second special case is if the supply of residential property at
each level of air quality is unresponsive to changes in price. This
ignores the possibility that the supply of residential air quality can be
changed by changing the number of residences located in an area with a
given level of air quality. However, the fixed supply assumption may be a
reasonable assumption, especially in the short run, because supply adjust-
ments in the housing market typically take years.
Another possible solution to the identification problem is to
simultaneously estimate both supply and WTP functions. Nelson (1978) takes
this approach and specifies a supply function. He postulates that the
supplier's offer price for residential site air quality is a function of
the level of air quality at the residence, the census tract population
density- and time required to reach 75 percent of metropolitan employment
from the residence. He reasons that suppliers bid a portion of a fixed
total supply of metropolitan land away from other users and that their bids
are higher when air quality is better. Freeman criticizes this specifica-
tion of the offer function because it focuses on the trade off between
industrial and residential land uses while not considering changes in
residential density, which may influence the amount of housing available at
each air quality level. Witte, Sumka, and Erekson (1979) have examined
some of the supplier characteristics that may be important in determining
the supply of housing though their housing market study does not speci-
fically consider air quality. More work needs to be done in this area.
The procedure, as described above, that has been used in the most
currently published hedonic price studies to estimate WTP has been recently
criticized by authors in this field, informally and at professional meetings,
though the discussion has not yet appeared in publication. The argument is
that by estimating a hedonic price function from cross-sectional data of a
single housing market, the analyst has only identified the marginal implicit
price function and thus has only one point on each WTP function. To
illustrate this problem, consider an urban area where there are three types
of households differentiated only by income. Hence all the households will
fall on one of the three WTP curves shown in Figure 4.5. Even if we assume
that the supply of housing is fixed at three different levels of air
quality, Ql, Q2, and Q3, we cannot identify unique willingness to pay
curves from the information contained in marginal implicit price function,
PV'(Q), because all we have are points A, B, and C which may be consistent
with any number of willingness to pay curves. However, if we have a second
marginal implicit price function, PV'(Q)*, from another time period in the
4-5L
-------
same urban area or from another urban area, and if we can assume that the
residents of the second time period or urban area can be divided into the
same three household groups (or are different in ways for which we can
account), each of the WTP curves can be identified because there are now
two points observed on each.
Therefore, this second step of estimating WTP as executed in the
currently published hedonic property value studies, such as those described
in Chapter 6, is probably not yielding reliable WTP functions, but only
modifying the information contained in the estimated hedonic price function.
In most cases the functional form of the hedonic price function probably
dominates the relationship between the marginal implicit price of air
pollution, which is used as the dependent variable in the WTP function, and
the level of air pollution, which is the primary explanatory variable in
the WTP function, because a nonlinear functional form implies that the
marginal implicit price is a specific function of air pollution. New
approaches that use data regarding shifts in the marginal implicit price
function over time or across several different housing markets may event-
ually alleviate this problem.
Marginal
Implicit
Price
QlQl* Q2 Q2* Q3
Q3* Air Quality
Figure 4.5
Identification of Willingness To Pay
4-52
-------
Freeman (1979a) suggests that benefits can be approximated from
the marginal implicit price function to avoid the difficulties of esti-
mating the WTP functions. This is illustrated in Figure 4.6. The marginal
implicit price function, PV'(AP) is known, but the WTP function, Wi(AP), is
not. If air pollution decreases from API to AP2 true benefits to household
i would be given by the area ABFG. The first possible assumption is that
WTP is constant from API to AP2 implying that it is horizontal at point B.
Benefits would then be area ABCG. This assumption would lead to an over-
statement of true benefits and conversely, would understate the damages of
an increase in air pollution. A second possible assumption is that the
marginal implicit price function is the WTP function. As was previously
mentioned this implies that all households can be represented by one WTP
function. Benefits would then be area ABEG which would overstate the true
benefits of the air pollution reduction less than the constant WTP assump-
tion. Both of these assumptions would allow the analyst to set an upper
bound on benefits or a lower bound on damages of a change in air quality.
A third possible assumption is that WTP is zero when air pollution
is zero and that the WTP function decreases linearly from point B to the
origin. In this case estimated benefits would be area ADEG. In the
depicted situation this would overstate benefits less than the constant WTP
assumption. However, depending on the slope of the actual WTP curve
this assumption could also understate true benefits. Though this assump-
tion would not establish an upper or lower bound on benefits or damages it
may be useful in cases where the estimated marginal implicit price function
has a positive slope.
o
Dollars
AP2
API
AP
Wl(AP)
PV (AP)
Figure 4.6
Approximating Benefit Measures From The
Marginal Implicit Price Function
Source: Freeman (1979a) p. 145.
4-53
-------
4.3.3 Variations in the Property Value Approach
There are two variations to the property value approach that have
been used to estimate the value of air quality from the residential
property market. The first is the matched pairs technique developed by
Brookshire et al. (1979) in their study of the South Coast Air Basin of
Southern California. This study has been replicated by Loehman et al.
(1980) for the San Francisco Bay area. This technique merely adds a
preliminary step to the hedonic price technique by selecting pairs of
neighborhoods that are similar except for their air quality.
The other variation is the residential location model developed by
Polinsky and Shavell (1976) and implemented by Polinsky and Rubinfeld
(1977) and by Smith and Deyak (1975). From this model an equilibrium
rental price function is derived that gives the relationship between
housing rents, travel distance to the central business district (CBD),
neighborhood amenities, and household income that will sustain equilibrium
in residential choice. If a specific form for the utility function is
assumed, the estimated parameters of the rent function can be used to
identify the utility function from which benefit measures can be directly
estimated.
Matched Pairs Technique
One of the problems with the hedonic price approach to estimating
the value of air quality is that there are many influences on the price of
property that are much more important than air quality. Ideally, in a
cross-sectional study, the analyst would like to see the price differences
between properties that are identical in every way except air quality.
More realistically, variation in property characteristics can be reduced by
purposefully selecting some properties or census tracts and excluding
others from the study. Many studies have limited themselves to census
tracts that have a certain minimum percentage of owner-occupied houses or
used other criteria to cut down the wide variety of property characteris-
tics that exist in an urban area.
The South Coast Air Basin study went even further with this selec-
tion process by choosing pairs of neighborhoods that were similar in as
many respects as possible except air quality. If it were possible to
select matched pairs that were identical in all respects except air quality
the property price difference between the two neighborhoods would be
entirely attributable to the difference in air quality. This dollar
difference per air quality unit would be the same as the coefficient of the
air quality variable in a linear hedonic price function for property
values, as long as air quality were measured in the same units.
In practice it is probably not possible to find perfectly matched
neighborhood or property pairs so that the price differences will reflect
4-54
-------
other differencs in addition to the air quality differences. This is
supported by the South Coast Air Basin study, which found that the price
differences between the matched pairs were larger than the price differences
attributed to air quality in a hedonic price function estimated for the
matched properties. The matched pairs approach, however, when used to
select properties for a hedonic price study, has the the advantage of
potentially eliminating some of the non-quantifiable differences between
neighborhoods that are not captured in the hedonic price function and can
bias the measured influence of air quality. There are likely to be elusive
property characteristics that are important to households, but are not
easily captured by measures such as the number of rooms and distance from
the CBD. Careful matching of neighborhoods will hopefully hold constant
some of the characteristics that are difficult to quantify. Also, because
the effect of air quality on property values is small relative to other
determinants of property value, reducing the variations in other attributes
will help to isolate the effect of air quality.
Residential Location Model
This model is based on the models of the residential location
decisions in an urban area, which postulate that the household's resi-
dential location choice depends primarily on the distance from the CBD and
the amenities at the residential location. Air quality is one of these
amenities. If the effect of each amenity on the location decision can be
explained, then the demand for each amenity can be identified. What
permits identification of the demand for amenities even in the absence of a
market is that land is differentiated across the urban area by amenities.
The residential location model postulates that the household's
utility is a function of consumption goods, housing services, and ameni-
ties at its residential location. Thus for a city where all workers are
employed in the CBD and all households and residences (for each household
class) are identical, the equilibrium rental price of residential property
is demonstrated to be a function of the distance from the CBD, amenities at
the residential site, household income, and the equilibrium level of
utility.
In application, a specific functional form is assumed for the
utility function from which a specific rental price function is derived.
Households are grouped into classes according to income and assuming that
all households in a given class have identical utility functions the rental
price function is estimated for each class. The estimated rental price
function must also account for differences in the quantity and quality of
housing that do exist. This procedure also assumes that households are
sufficiently mobile to ensure that equilibrium levels of utility are
attained.
4-55
-------
Depending on the specific form of the utility function from which
the rental price function is derived, the estimated parameters of the
rental price function can be used to identify the parameters of the
utility function. When the utility function is known, EV and CV measures
of welfare changes caused by changes in amenities can be directly esti-
mated. The hedonic approach, on the other hand, estimates WTP functions
from which measures of OCS can be obtained. The functional form used to
estimate the WTP function has specific implications about the nature of the
utility function, but it does not impose any one form on the utility
function.
In application, the residential location model has many of the
same difficulties as the hedonic price technique as well as requiring
several additional and more artificial assumptions about the nature of
housing, households and the urban area. Polinsky and Rubinfeld (1978) and
Smith and Deyak (1975) use the Cobb-Douglas form for the utility function.
The rental price function derived from the Cobb-Douglas utility function
is a log-linear form.
4.3.4 Strengths and Weaknesses
The major strength of the property value approach is that it uses
actual market data that reflect what people have actually spent to obtain
various levels of air quality. This is an important advantage over contin-
gent market approaches that ask people what they would spend but do not
require them to part with their money. However, the use of property values
to estimate the value of air quality is limited by what property value
studies can measure and when they can be used. Most importantly, property
value studies are restricted because they rely on data that were not
collected for this purpose. Several specific limitations are delineated in
this section.
When Property Value Studies Can Be Used
Property value studies can be used to estimate benefits of air
quality only under certain conditions. First, there must be fairly con-
sistent spatial variation in air quality. If air quality is the same over
an entire urban housing market its effect on property values cannot be
separated from other influences on property values. Not only must air
quality vary from one residence to another, but the variation must be
stable enough over time that a measure of average air quality varies from
residence to residence. In other words, if air quality varies such that
one day neighborhood A has clean air and neighborhood B has dirty air and
the next day neighborhood B has the clean air and A the dirty, no dif-
ferences in property values between the neighborhoods can be attributed to
air quality. Fortunately there are often fairly stable patterns of air
4-56
-------
quality over an urban area, such that air quality is usually worse in the
central business district (CBD) and in whatever direction from the CBD that
the wind usually blows. However, property value studies may be of little
use for estimating the impact of the regional haze that is homogeneous over
large areas as occurs primarily in the eastern U.S.
A second condition is that differences in air quality impacts
must be perceived by individuals to be reflected in property values. The
damages of poor air quality include health effects, property damage such
as corrosion and dirt, and aesthetic impacts. There will not be a problem
if households do not correctly attribute the impacts to air pollution as
long as households perceive the differences between residences. For
example, if one neighborhood seems dirtier and less lushly vegetated than
another, property values may reflect these differences even if people do
not realize that they are the result of air pollution. However, there may
be impacts that are not perceived at all. For example, carbon monoxide is
a colorless, odorless pollutant that causes acute symptoms only in high
concentrations. However, there may be health effects of long-term exposure
to low concentrations that people do not attribute to carbon monoxide
exposure. Additionally, property value studies presume that individuals'
perceptions of air quality are reasonably correlated with the technical
measures of air quality that are available to the researchers.
A third condition is that people must be sufficiently mobile for
property values to reflect current air quality conditions and for the
the implicit price of air quality to reflect the household's WTP. Consider
the process whereby a change in air quality would cause property values to
change. If air quality in a particular neighborhood is improved property
values would reflect the change only as new people were attracted to the
neighborhood and caused housing prices to go up. If moving costs are high
or for some other reason people are reluctant to move they may tend to stay
at a less than optimal location given their preferences, income, etc.
Hence, there may be a significant lag between changes in a housing market
and the complete adjustment to these changes. It is not clear how serious
this problem is in practice, but the validity of the property value approach
to air quality benefit estimation may be lessened in housing markets that
have been subject to recent and/or rapid change. This is less of a problem
when actual sales data are used, which requires only that buyers choose an
optimal residential location given conditions at the time of the purchase
and that there is enough competition among buyers to force prices to equal
marginal willingness to pay.
A fourth condition is that it must be possible to separate air
quality from other neighborhood amenities for property value studies to
correctly determine price differences attributable to air quality dif-
ferences. Sensitivity tests in several studies have shown that the air
quality coefficient changes significantly if variables correlated with air
quality are omitted. Similarly, correlations among different air quality
measures may impede the attempt to isolate the visibility damages from the
health damages of air pollution in a property values study. However, the
empirical studies have not yet addressed the question of whether visibility
effects can be separated from the other effects of air quality.
4-57
-------
A fifth condition is that the necessary data must be available.
Property value studies cannot always be used to estimate the benefits
or damages to be expected from a future change in air quality. Changes
that have not yet occurred will not be reflected in property values unless
they have been anticipated by households. In sparsely populated areas
there may not be enough private residential properties and therefore not
enough data to estimate a hedonic price function. Improved research tools
may eventually allow WTP functions that were estimated for one area to be
transferred to another area given similar air quailty conditions and
certain assumptions about similarities between the two populations. If
reliable WTP functions are estimated it would also be possible to predict
the impact of air quality changes that have not yet occurred by extra-
polating beyond the current air quality levels in the study area. Such
improvements may eventually ease the restrictions caused by the data
requirements of the approach.
What Property Value Studies Can Measure
Even under ideal circumstances, a residential property value study
can probably measure only certain aspects of air quality benefits. First
of all, residential property value studies can measure only the impacts of
air quality at the residential site. Because most people do not spend all
of their time at home, they will experience the air quality at other loca-
tions, such as their work place, which will not influence the value of
their residence. It is possible to apply the hedonic price technique to
commercial and industrial property and to wages, but research experience
in these areas is at this time quite limited. Future research may soon
reveal additional evidence about the nature and magnitude of air pollution
impacts at the work site, but most currently available evidence concerns
impacts only at the residential site.
Using property value differences to estimate benefits of air
quality usually assumes that prices and quantities of other things are
not affected by changes in air quality. If this assumption is violated,
the change in property values may not reflect the household's full willing-
ness to pay for air quality. People may respond to changes in air quality
in ways that do not affect the value of their residences but affect the
value of other goods. For example, the demand for some outdoor activities,
such as golfing and jogging, may be affected by changes in air quality
causing changes in the prices of goods associated with these activities
that are not reflected in housing values. If this occurs to any great
extent, property value studies can capture only part of the benefits of a
change in air quality.
Property value studies will capture only part of the benefits if
the assumption of weak complementarity, defined in Section 2.5, does not
hold. Most property value studies have assumed that the marginal value of
air quality in a particular neighborhood is zero for those who do not live
there. Under most circumstances, this may be a reasonable assumption, but
there may be times when it is not valid, such as when the view from one
neighborhood is affected by the air quality in another neighborhood.
Counting only the benefits to the residents of the second neighborhood
would then understate the benefits of an air quality improvement there.
4-58
-------
In some cases this limitation can be overcome by enlarging the area con-
sidered is assessing air quailty impacts on the household. However, if
substantial impacts are caused by air quality well beyond the neighborhood
in which the household resides the property value approach may fail to
capture such impacts. This may frequently be the case with visual impacts
because visual quality is not only influenced by the air quality where the
viewer is located but is also influenced by the air quality between the
viewer and objects that are viewed.
Property value studies may be limited in their ability to measure
air quality benefits at mandatory Class I Federal areas and other recrea-
tional areas. Property at a recreational site is often publicly owned and
therefore not subject to the market conditions that would bring prices in
line with benefits. However, even where the private property market is
sufficiently extensive there are problems posed by the fact that most
people spend only a small fraction of their time at a recreational property.
Health effects, in particular, may not concern the individual who will be
in that location for only a few weeks out of the year. Aesthetic effects,
on the other hand, may influence recreational property values significantly,
since poor visibility, for example, may seriously detract from the enjoy-
ment of the recreational experience. Another serious drawback for property
value studies in recreational areas, where hikers, campers and picnickers
make up a large proportion of the area users, is that property value
studies will pick up air quality impacts only on those who own or rent
property.
An example of an application of the hedonic price technique to
recreational property is provided by Wilman (1981). She applies this
technique to recreational rental property on Cape Cod to estimate the
damages of beach debris to the recreational experience. Cape Cod has a
fairly well developed private property market, and the condition of the
beaches has a direct impact on the recreational experience. Under these
circumstances this approach is applicable.
Property value studies, even in recreational areas, will usually
capture only the activity value of the air quality at that location. This
is because property values will reflect how much people are willing to pay
to live, vacation, and to some extent, be able to visit an area under its
current air quality conditions, which represents activity values at that
location and possibly some option value to be able to visit that site in
the future. How much they would be willing to pay to know that somewhere
the air is clean and to have the option to go there at some future time
will not be reflected in any property values.
Because a residential property provides a stream of services over
an extended period of time, the purchase price of the property is deter-
mined by the present value of the expected future stream of benefits.
Therefore the price of the property is not only influenced by current air
quality conditions but is also influenced by the expectations concerning
future conditions. For example, a property value study to determine
4-59
-------
benefits of current air quality in an area surrounding a long discussed and
well publicized proposed coal fired power plant site may be thrown off
track if expectations of future air quality degradation have already begun
to influence property values.
4.4 Introduction to Alternative Approaches For Measuring Economic
Benefits of Visibility Aesthetics
The emphasis of this chapter has been to present the best method-
ologies that are currently being practiced and have been subjected to some
degree of validation either through repetition or by simultaneously ap-
plying other techniques to obtain comparable results. Several alter-
native approaches are and will continue to be developed. This section
briefly introduces a few of these alternatives.
4.4.1 Ranked Attributes/Market Share
This technique has been used to determine characteristics of
demand for market goods. ° It is currently being developed as a contin-
gent market technique to analyze the demand for air quality through a
survey approach where respondents are asked to rank order choices of
alternative sites, for instance Class I area recreation sites. The alter-
natives have different combinations of attributes, including the level of
air quality, which may be depicted in photographs; the market costs, such
as entrance fees; and the locational attributes, such as congestion and
travel distance. By evaluating how a particular site ranks under different
visibility and other site conditions, it is possible to see the changes in
market shares each site will capture, or the changes in the probability a
particular site will be the most desired site, and to evaluate the demand
for a site as a function of these conditions.
A principal advantage of this approach is that the respondent
is not forced to directly put a dollar value on visibility; but rather
reveals the value of visibility indirectly, and perhaps less objectionably,
through the ranking of sites with alternative levels of air quality, prices
and other attributes. The effects upon a site's ranking from changes in air
quality and prices can be traded off to indirectly determine benefit
measures. The approach may also prove to be a useful form of verification
of other methods.
approach has been used primarily to evaluate travel mode-choice
using actual market data (McFadden 1974; and Domencich and McFadden
1975). Douglas Rae at Charles Rivers Associates is currently attempting
to value alternative levels of air quality at Class I areas using the
approach in a study funded by the Electric Power Research Institute.
4-60
-------
Some of the problems with this approach include the potential need
for a great many complicated alternatives to be ranked to obtain any
precision in the estimates. If each alternative has several attributes
simultaneously changing, respondents may be unable to focus upon all of
the changes. If instead, respondents focus upon changes in only one
attribute this will increase measurement error in the evaluation of the
other attributes. Further, if the alternatives are not clearly under-
stood, which may be the case for individuals who have never visited the
alternative sites, the ability to rank alternatives accurately is seriously
impaired. This problem is similar to that of hypothetical bias. The
approach, as developed thus far, is not applicable for the derivations of
option and existence values.
A pretest of this technique was carried out at Mesa Verde during
the summer of 1980. By determining how the probability that a desired site
will be chosen is affected by air quality, prices and entrance fees,
compensating surplus measures of benefits are obtained. This is done by
holding all variables constant and determining how prices must change to
offset air quality changes such that the probability the desired site will
be chosen from all competing sites remains constant. The pretest values,
ranging from $4 to $5 per day per person to preserve clear air versus plume
blight, and from $8.45 to $10.74 to protect against severe regional haze
(using the best fitting statistical specification), are of a similar order
of magnitude to other benefit measures obtained in the Four Corners region.
Although the pretest encountered many survey problems, the technique holds
promise as a useful tool.
4.4.2 Travel Cost and Site Substitution Approaches
The travel cost method has been used extensively to estimate the
demand for benefits of recreation activities, particularly at specific
recreation sites. The travel cost approach, suggested by Hotelling,
initially developed and applied by Clawson (1959) and Clawson and Knetsch
(1966), and refined by a great many authors, ' recognizes that to use the
services of a recreation site, users not only incur expenses of entry fees
and equipment, but must get to the site. The cost, or price, to an indi-
vidual of using the services at a recreation site will vary according to
the travel time and expenses incurred getting to the site. The further
away users are from a'site, the greater the implicit price of using the
site.
The simplest Clawson-Knetsch (C-K) travel cost approach may be
implemented by assuming there is only one site to be considered by re-
creationists and by assuming the supply of services (recreation visits) at
detailed reviews of the travel cost method may be found in Freeman
(1979a), Smith (1975), Dwyer, Kelly and Bowes (1977), and Knetsch (1974)
among others.
4-61
-------
the site is infinite at the travel cost price (travel cost plus entrance
fees and other user costs) for each individual. Next, by establishing visitor
use rates for recreationists from different distances from the site, a
regression analysis of the visitation rates as a function of the travel
cost measure of price and socioeconomic characteristics of the recrea-
tionists from different distances is used to estimate a demand curve for
the site services. This is frequently done at an aggregate level using
visitations at a site for recreationists from different origins. It may
also be performed with micro-level (household), data on recreation behavior
when such data exists.
In many typical applications of the C-K travel cost approach a
demand curve is estimated relating the travel cost price to the actual
number of visits for a particular site at a single point in time. Because
environmental quality does not change at any point in time, these variables
are often not included. Next, the benefit of having versus not having the
park, given current conditions, are calculated as equal to the consumer
surplus derived by recreationists, i.e., the difference between the demand
(WTP) curve and the entry fee and other travel costs. Because our interest
is with the changes in the levels of environmental quality variables rather
than the existence of the site, this simplified approach will not yield the
desired information.
A generalized presentation of the travel cost demand estimation
for the ith individual (or for all individuals from the ith origin) for
each jth site, which includes environmental quality variables and which
may be used to address the benefits from changes in environmental quality
is:
Vij = Vij (Pe, Px, Dij, c, tij, di, Sih, ei, AQ, Mi, Zi) (1)
where:
Vij = number of visits by individual i (or by individuals from i)
to site j
Vij(.) = visitation functional form which may vary across all
j sites and all i individuals
Pe = vector of money prices of entry (possibly zero) to the
various sites
Px = vector of private goods prices
Dij = vector of round trip distances from residence of individual i
to the various sites ]
c = unit cost of travel miles
tij = vector of travel times to the various sites j for individual i
4-62
-------
di = cost of travel time
Sij = on site time for i at site j
ei = on site value of time
AQ = vector of air and other environmental quality measures of the
various sites
Mi = money income of individual i
Zi = vector of other socioeconomic variables
Demand for recreation visits at each site is a function of the travel costs
and entry fees, the money cost of time in travel and recreation, socio-
economic variables, all prices, and environmental quality variables at all
sites. Economic theory will often constrain the ways these variables enter
the Vij(.) function. Distance and unit travel costs are included to
represent the out-of-pocket costs, while travel time and site time are
included to represent the opportunity cose of getting to and staying at a
site. Environmental quality variables at all sites are included because if
these conditions are known in advance by the recreators it is likely that
they will influence the choice of sites and rate of activity.
What is desired is a measure of how changes in air quality affect
demand for recreation at the affected site. If a recreation demand curve
as a function of air quality can be appropriately estimated for a site,
then consumer surplus can be determined in the usual way as the difference
between demand (WTP) and expenditures: The effect of changes in air
quality upon demand, and therefore consumer surplus, provides a benefit
measure of the air quality change.
The strengths and weaknesses of the travel cost approach have
been discussed at length by other authors (see footnote 19). Strengths are
that the technique focuses upon well established behavior for which data
often exist or are easily obtained (for example, through license plate
surveys of actual visitors) and once the demand estimation is performed,
benefit measures are readily obtainable.
Among the unresolved problems in all applications of the travel
cost approach is the measurement of the cost of time. Time spent travel-
ing to, and recreating at, a site involves opportunity costs of producing
the experience in terms of other activities foregone. Yet the time spent
traveling and recreating may be part of the desired experience. Ignoring
the time cost of travel will bias the price elasticity of demand and
WTP estimates. Further, the measures of travel time and travel distance
are highly correlated and complicate the estimation procedure.
Other problems in the travel cost approach include difficulties in
allocating costs for multi-purpose trips; the assumption that changes in
travel costs have the same effect on recreation demand as changes in entry
fees and other costs; handling capacity problems; and defining ail variables
4-63
-------
and correctly specifying the demand relationship. The definition of the
dependent variable is of particular concern because one needs to have a
consistent measure of the use of the recreation facility. Simply grouping
together trips of different lengths, in terms of recreation days at the
site, raises a problem of an inconsistent and somewhat ambiguous measure of
use. Yet, using recreation days is problematic because each day is subject
to different travel costs. An additional problem is that recreation days
or recreation trips may not capture that aspect of the experience that is
actually demanded. A potentially more desirable approach is to determine
the demand for the characteristics of the recreation experience at
a site. Two approaches focusing upon the demand for attributes at a site are
the contingent travel cost and hedonic travel cost procedures discussed
below.
The travel cost approach is limited in benefit analysis because
by analyzing the behavior of actual users it obtains activity values of
users at a particular site and does not obtain option or existence values
by users and non-users. Also, the technique is site specific, which,
without specification of how changes in characteristics of the recreation
site individually and in combination with changes in air quality affect the
demand for recreation at a site, limits the tranferability of results. For
example, as is the case with all dissimilar sites such as most Class I
areas, estimates of the effect of changes in air quality upon the value of
recreation in the Grand Canyon would be very difficult to transfer to any
other site. To be accurate the specification of demand must also account
for changes in environmental conditions at the site of interest and at
competing sites that may affect demand. This can be a rather difficult
task.
As yet there have not been any empirical applications.of the travel
cost approach to the problem of visibility benefit analysis. Two
general types of travel cost approaches could be used to measure the
effects of changing air quality upon demandsurveys obtaining actual
visitation data and surveys obtaining hypothetical visitation data. In the
first, historical travel cost data and actual visitation data at a site of
interest and supporting data for the site and all competing sites could
be obtained, from which the effect of changes in air quality at the site
upon demand for recreation could be determined using a variation of Equa-
tion 1: Vi=Vi(.), for all i given j = the site of interest. This would
yield site-specific air quality benefit measures. Cross-sectional data of
actual visitation rates at a group of sites could also be used to estimate
20
The approach was being attempted as part of the summer 1980 study of
visual aesthetic impacts of proposed mining operations outside of Bryce
Canyon National Park (U.S. NFS 1980). The results have not yet been
released. Devine and Smith (1981) have addressed theoretical issues
associated with applying travel cost methods to visibility benefit
analysis using the household production function approach to consumer
behavior. They illustrate how the standard travel cost model and the
hedonic travel cost approach (discussed in this section) are special
cases of their generalized presentation.
4-64
-------
a fixed form of Equation 1, Vij = Vij(.), for all i, j. This, however,
assumes the variables of interest are the same, and affect visitation in
the same manner across all sites, thus yielding only an average relation-
ship between changes in visibility and recreation demand at all sites
examined.
Using historical or cross-sectional data is problematic because
visibility conditions are often very similar across competing sites and
through time, and are less of a demand-determining factor than are the
unique characteristics of the sites and other factors that may vary through
time or across sites, such as water quantity or quality. Consequently,
estimates of how changes in visibility conditions affect demand are likely
to be inaccurate unless a substantial data base, including measures of all
other important characteristics at each competing site in each time period,
is compiled and used in the demand estimation. Work in this direction has
been initiated to quantify those aspects of the experience valued by
recreationists. (For example see Stankey (1972, 1973); Brown, Driver and
McConnell (1978); Lucas (1964); Lucas and Stankey (1973); Hendee et al.
(1971) among others.)
For these surveys of actual behavior to be effective for visi-
bility benefit analysis, the proposed change in air quality must also have
occurred, have been accurately defined and measured and have been recognized
by recreationists either at the site or across sites. If recreationists
are either unaware of visibility conditions at recreation sites or unaware
of prevailing conditions on their desired recreation days, a function
relating visitation to air quality will underestimate benefits from control
because the air quality elasticity of demand will understate what responses
would have occurred had recreationists possessed full information. Equiva-
lently, this approach will not appropriately measure benefits or
damages to those individuals who recreate at a site with less than their
desired air quality conditions due to the lack of information. Appropriate
implementation of this travel cost approach implicitly assumes that recrea-
tionists have full knowledge of conditions or requires a model of their
expectations. While such assumptions may be plausible in some cases, in
general there is little evidence to support applying them to air quality,
particularly at distant national parks.
These approaches will yield much different benefit estimates if the
changes in air quality have occurred uniformly across all competing sites
in each time period rather than only at the site of interest. For example,
if air quality decreases at only one of many competing sites, indivi-
duals may substitute to the services at other sites. On the other hand, if
all competing sites are affected equally there is little incentive to
substitute across sites. Therefore, one would expect the substitution away
from the site of interest to be greater when only one site is affected than
when an entire region is affected.
4-65
-------
A second approach would be to survey individuals as to their
recreation travel behavior given alternative hypothetical scenarios of
environmental conditions both at the site and at competing sites. This
''contingent travel cost approach" has been successfully applied by Thayer
(1981) to examine the effects upon recreation patterns from geothermal
development in the Jemez Mountains of New Mexico. Applied to air quality
this approach attempts to estimate benefits by determining travel costs
that respondents would incur to recreate at alternative equivalent sites
given increasing air quality degradation at the surrent site. A utility
maximizing individual is assumed to recreate at the nearest outdoor facility
if all areas are equivalent in all other competing aspects and to recreate
only at the next nearest area if there is a significant deterioration at
the closest site. By asking detailed questions on site attributes,
expected activity rates and selected locations coupled with the costs of
substituting locations, the value of damages associated with alternative
levels of air quality can be inferred.
The contingent travel cost approach has strengths in that all
other factors can be held constant by the scenario development so that
changes in recreation patterns are influenced only by the change in
environmental quality at the site of interest. The difficulties are much
the same as contingent market approaches. Uncertainties about alterna-
tives may cause respondents to be unable to estimate their rate of substi-
tution among the alternatives and the lack of alternatives in some situa-
tions may limit the ability to accurately measure benefits of improved air
quality. Finally, as discussed for the bidding methods, the alternative
conditions must be carefully defined and presented to respondents.
Substantial work has been done to determine the demand for recrea-
tion experiences, but little research has been done on the characteristics
that make up those experiences and how changes in these characteristics
affect demand. Brown and Mendelsohn (1980) have addressed this problem by
developing a promising "hedonic travel cost approach". In this approach
recreationists are assumed to be willing to incur different levels of cost
to recreate at different sites because of the different levels of the
characteristics supplied at the sites. The combination of characteristics,
costs, and socioeconomic characteristics of the individual can be examined,
as in the property value approach, to determine the demand for the charac-
teristics .
The hedonic travel cost approach asserts that given fixed trip
costs plus increasing costs associated with increasing expenditures in
distance traveled and time spent, an individual will choose to travel
further away to recreate only if more distant sites provide, all else
fixed, higher levels of desired characteristics. Therefore, by regressing
travel costs from one point of origin to each of several sites on the
characteristics at each site, a hedonic price function, which the authors
4-66
-------
label a value function, of characteristics is estimated (i.e., prices of
characteristics are origin specific, not site specific). For each point
of origin the marginal value of a unit of characteristic is the partial
derivative of the value function as in the hedonic approach. A separate
hedonic value function is estimated for each point of origin to many
alternative sites. The demand for the characteristic across all origins
and destinations is estimated by regressing the average level of character-
istics demanded by different types of recreationists on the characteristics
prices for those recreationists.
Brown and Mendelsohn (1980) apply the hedonic travel cost technique
to data from a survey of Washington state steelhead trout fishermen. The
survey revealed hours and distance traveled, origin and fishing location,
number and length of trips and a 1 to 10 evaluation for each of the
chararcteristics of the fishing site. Two hedonic value functions of
travel costs versus the levels of characteristics at alternative sites
were estimated for each point of originone function for miles traveled
and one function for time traveled. Linear hedonic price functions were
used in this effort, although not required by the method. Next, the
authors estimate the demand for different length trips (one day, two or
three days, etc.) as a function of socioeconomic characteristics of the
recreationists and the implicit prices of characteristics. Finally, the
demand for site characteristics across all sites by origins is determined
by regressing the average level of characteristics for each trip length on
the price of characteristics and socioeconomic variables.
This hedonic travel cost approach could be applied to determine
air quality values but faces many of the same problems of the C-K approach.
In particular, the measure of air quality must be more carefully defined
than by the simple 1 to 10 scales used in the Brown and Mendelsohn study
because such ratings by non-professionals may tend to be inconsistent (see
Chapter 3) and of limited usefulness in benefit-cost analysis for specific
levels of pollution control. It is also necessary to more carefully
define all relevant characteristics of a site, as in the property value
studies, rather than just a few as in the Brown and Mendelsohn study.
Finally, this approach requires the use of a single attributes to visita-
tion demand function which is assumed to be homogeneous across all sites
for each origin. This first requires defining all relevant sites for the
first stage equations. Further, a linear model maintains that there is one
price for each attribute in the market, defined by the chosen array of
relevant sites. A non-linear hedonic price schedule would allow different
prices of the attributes at each site. Nevertheless, the hedonic price
relationship is assumed constant across all relevant sites. In practice,
this may be a serious limitation to the use of this approach in unique
recreation areas where the effect of air quality upon demand and WTP will
vary substantially with the unique scenic vistas and would be difficult to
quantify in a cross site regression analysis. The approach would be most
useful for recreation experiences across sites that are highly similar such
as beach use, inland fishing, or skiing so that benefit measures could be
confidently transferred from site to site.
4-67
-------
4.4.3 Household Production Function Approach^1
This approach may use market data or survey respondents to deter-
mine the rate and location of specific activities; characteristics of the
site frequented, such as air quality and congestion; the activity engaged
in; and the level of expenditures related to the activity. In the survey
approach, alternative scenarios for air quality are presented, and respon-
dents are asked how they will change their rates of activity, location and
expenditures. Much as in the site substitution and travel costs approach,
changes in expenditures are used to generate demand curves and damage
functions.
Because successful analysis of air quality issues with this
approach necessitates potentially complex data on trade-offs and expendi-
tures, this may diminish the accuracy of using surveys for this technique.
Limitations on available secondary data may also limit the market approach
for this technique.
4.4.4 Wage and Salary Differentials
Labor markets have been examined in an effort to derive values
of urban amenities. Similar to the rationale for the property value
studies, this examination of wage and salary differences is based on the
supposition that individuals must be compensated in the form of higher
real wages (higher nominal wages and/or lower cost of living) to accept
disamenities where they live and work and, conversely, are willing to
accept lower real wages to live and work in locations with positive
amenities. Recent studies indicate that progress is being made toward the
application of this approach to the estimation of the value of air quality,
although a substantial amount of research is necessary before it becomes an
accepted "best practice" for estimating visibility benefits.
Similar to the estimation of a hedonic price function for property
values, some wage studies have estimated a labor market opportunities
function from observations of actual wage levels and different amenity
levels across several urban areas. This function is usually estimated
separately for each occupational category. It represents the market
equilibrium wage level and amenity or disamenity combinations that are
available to individuals.
Each partial derivative of the labor market opportunities function
with respect to an amenity or disamenity represents a marginal implicit
price for the amenity or disamenity. It indicates how much additional
wage an individual must give up to obtain additional units of an amenity,
variation on this approach was first attempted for air quality in
Blank et al. (1977); however, numerous methodological problems hampered
the effort. Another application for wildlife stocks can be found in
Eubanks and Brookshire (1980). Ernie Manuel and Robert Horst, at Math
Tech and under contract to EPA, are currently pursuing this approach for
air quality issues with the use of market data.
4-68
-------
such as a pleasant climate or cultural opportunities, and how much addi-
tional wage they would be compensated to accept additional units of a
disamenity, such as crime or air pollution. As with the property value
studies, these marginal implicit prices represent the individual's marginal
willingness to pay only for the amenity level at which he is observed.
This is because the optimal location for the utility maximizing individual
is where the amount of wages he must give up to obtain an additional unit
of each amenity just equals the amount he is willing to pay for that
additional unit. Hence, he is observed at the point where his WTP function
crosses the marginal implicit price function for each amenity. However,
because individuals have different tastes and preferences (i.e., different
WTP functions), the observed wage differences statistically attributable to
air quality differences will not be eqivalent to the individual's willing-
ness to pay for non-marginal changes in air quality.
The labor market opportunities function is determined by indivi-
duals' willingness to supply labor at different amenity-wage combinations
and by firms' willingness to demand labor at different amenity-wage
combinations. Both sides must be considered if willingness to pay for
other than a marginal change in an amenity level is to be identified.
Current studies have only begun to address the complexity of this inter-
action.
One of the primary difficulties faced by these types of studies
is that air quality is only one of many minor influences that determine in
what city an individual will choose to live. To isolate the effect of
air quality the entire location decision must be explained. Property
value studies must explain where a person chooses to locate within an urban
area, but wages are influenced by the overall air quality levels an indivi-
dual encounters in his entire living and working environment, not at any
one site within an urban area. Hence, intercity comparisons are necessary
to see how wages vary with different air quality levels.
Air quality is probably much less important in determining in
which city an individual will choose to reside than in determining where
in a particular city he will choose to reside. For example, households
are probably more willing to move across town to be in a cleaner neighbor-
hood than to move across the country. That this makes it very difficult to
produce reliable air quality benefit estimates from intercity labor market
data is supported by the findings of Smith and Deyak (1976). They found
air quality measures to be insignificant in influencing property values
across different urban areas, even though numerous studies have found a
significant influence of air quality on property values across a single
urban area.
Two examples of recent research efforts using labor market data
are Rosen (1979) and Cropper and Arriaga-Salinas (1980). Rosen estimated
a relationship between individuals' real wages and average measures of
city amenities and disamenities across nineteen major urban areas for
several occupational categories. He used occupational and personal
characteristics from the 1970 Current Population Survey, which provides
data on an individual rather than an average basis, to account for inter-
personal differences in earning capacity. He then added measures of
amenities for each urban area to obtain a relationship between wages and
4-69
-------
each amenity. Included were measures of air and water pollution, climate,
crime, crowding, and market conditions. He used the results to construct
quality of life indexes for each city. He did not attempt to go beyond
estimating the marginal implicit price of air quality, but his results at
that point seem plausible. A measure of suspended particulates was the
most significant pollution variable. A one standard deviation increase in
particulates (apparently 38 yg/m3 this case) was associated with a 1.7
percent to 5.7 percent wage premium. At the mean annual earnings of the
sample, $9,000, this amounted to $153 to $513 per year.
In a theoretical section of the paper, Rosen demonstrates the
joint adjustment of land and labor markets to variations in amenity
levels. He postulates that utility is a function of consumption goods,
residential space, and amenities, from which he derives indifference
curves for the individual showing the trade-offs between rents and ameni-
ties and between wages and amenities that maintain constant utility. He
does not attempt an empirical estimate of these relationships.
Cropper and Arriaga-Salinas (1980) used a somewhat different
approach by developing a model in which proximity to input and output
markets and property tax rates determine the locational pattern of indus-
trial growth and consequently determine the demand for labor. From a
log-linear utility function that allows for the possibility that amenities
may vary across a single city, she derives a labor supply function for
each city, the arguments of which are employment and amenities in each
city. This labor supply function can be used to estimate the coefficients
of the amenities in the utility function, allowing the derivation of
consumer surplus benefit measures for a change in an amenity. The authors
used a simultaneous estimation procedure to estimate the supply and demand
for labor. The advantage of this approach is that both supply and demand
sides of the labor market can be considered, but it requires specifying a
specific functional form for the utility function, which is unknown.
The measure of air pollution used was sulfur dioxide. It was
found to significantly influence the supply of labor for some, but not
all, occupational categories. She cautions that this approach is quite
experimental, but to illustrate how valuations of amenities can be derived
she estimates that the present discounted value of a 30 percent reduction
in sulfur dioxide, calculated for a person earning the median income in
St. Louis in 1960, is between $418 and $489.
4.4.5 Voting Approaches
Many voting and demand-revealing processes could potentially be
used in determining optimal quantities of, or the willingness to pay
for, environmental goods, including air quality. While there has been
little use of these approaches to date for valuing air quality, growing
professional interest in them will undoubtedly lead to such efforts in the
future. In this light, a few of these approaches are very briefly intro-
duced with the recognition that considerable research will be required
before any of these will become best practice for visibility benefit
analysis
4-70
-------
In many of these approaches an attempt is made to have individuals
reveal estimates of the desired quantity of a public good through various
incentives, taxation schemes and group decision-making mechanisms.22 one
such approach, suggested by Portney (1975) for determining desired levels
of for pollution control, is to ask individuals to indicate their preferred
quantity of a public good given estimated costs of several alternatives and
a specified tax share of the total cost. The utility maximizing individual
will choose the quantity where his marginal cost, in terms of tax burden,
equals his marginal return in terms of environmental amenities. This price
(tax share) and quantity combination yields one point in each individual's
WTP or demand schedule. If the costs or tax system, and therefore the
price, is varied across individuals with similar socioeconomic character-
istics the results can be pooled to derive demand curves for the environ-
mental good from which the willingness to pay for changes, and income and
price elasticities of WTP can be determined. It has been shown that under
cetain conditions this approach may also reveal the economically efficient
level of public good provision (see Bowen 1943 and Portney 1975).
Other survey techniques employing the voting and taxation ap-
proaches have been employed for non-environmental goods and services
(Clark 1971, Tullock and Tideman 1976, and Smith 1977). A problem with
many of these approaches for use in visibility benefit analysis is lack of
credibility of the proposed situation as a method to actually determine
air quality levels. These approaches may have all the problems mentioned
for bidding methods and perhaps others associated with the often complex
instructions and payment schemes involved.
Not all voting analyses need be based on hypothetical surveys. In
real world situations, voting results may yield informative data in optimal
quantities and WTP. In these situations voters choose to vote either for
or against a proposition or to vote for one of two candidates with perhaps
different views. These votes represent actual revelations of perferences
just as market data reveal actual expenditures. The value of voting
information depends critically upon how candidate positions and referendums
are worded and how costs are to be paid by voters.
To be useful it must be assumed that successful referendums or
candidates represent positions reflecting those of the median voter to
insure the likelihood of gaining a majority position. Next, if the out-
come of elections or referendums on the same issue can be simultaneously
observed from many different political jurisdictions along with the dif-
ferences in the propositions, outcomes, payment schemes, and socioeconomic
characteristics of the voters, then statistical analysis can determine the
demand for the good in question (see Freeman 1979a for discussion and
application references).
Two problems are inherent in this analysis of actual votes.
First, votes do not reflect intensity of preferences or of WTP- The same
51/49 percent split in votes can correspond to many different aggregate WTP
amounts for the project (sum of benefits minus sum of damages) depending
Freeman (1979a) for a review of voting procedures. Suggested initial
reading on demand revealing processes include Smith (1977), Clark (1971)
and Tideman and Tullock (1976).
4-71
-------
upon the socioeconomic characteristics of the supporters and opponents of
the legislation. It may even be the case that many projects accepted
(rejected) using a majority rule criterion would have been rejected (ac-
cepted) using a WTP criterion of comparing benefits to damages. Second, as
Freeman (1979a) points out:
The approach appears to be of limited applicability to
environmental services and the benefits of pollution control
in particular. The voting procedures of this and the
preceding sections require that the choice of a quantity, a
referendum, or a platform be coupled somehow with an obliga-
tion to share in some way in the cost of providing that
public good. Therefore, these procedures would be capable
of generating information on the demand for pollution
control only in those cases where the costs of pollution
control will be financed from revenues raised within the
political jurisdiction.
Many of the important pollution control problems for which
benefit data are desired do not fit this description, at
least in the United States. If some portion of pollution
control costs is borne by the private sector, then the link
between a vote on quantity and tax share or price is broken.
And the vote cannot be interpreted as revealing anything
about the economic demand for pollution control. Also, the
voting approach would only be applicable where both the
benefits and costs of the pollution control program fall
entirely within the applicable political jurisdiction. If
pollution spills across jurisdictional boundaries, some of
the benefits of pollution control will be realized outside
the jurisdiction. No voting measure could capture these
interjurisdictional spillovers. (Page 104.)
4.5 Can Benefit Measures Be Transferred Across Studies?
It would be desirable to determine if the results from previous
studies could be transferred to the problem at hand before undergoing the
considerable expense and effort of performing a visibility benefit analysis.
Two levels of accuracy are possible in transferring results. First, it may
be possible in some cases to transfer actual or systematically adjusted
results from past efforts to achieve reasonably precise benefit estimates
for the current study, a level of accuracy we call strong transferability.
Second, it may be possible to use previous results to set only ballpark
upper or lower bounds on benefit estimates for the current study as a
starting point for decision making, a level of accuracy we call weak
transferability. Theoretically, conditions can be established which should
insure valid strong or weak transferability of results. Whether these
conditions are met is an empirical question for which there is very limited
data at present.
4-72
-------
Both cases refer to the ability either to transfer the estimated
per capita benefits for a specific change in visibility and aggregating to
total benefits or to transfer the estimated relationship between benefits
for individuals and various levels of visibility or changes in visibility.
This relationship is represented by the benefit function, such as the
WTP or bid function estimated from a property value study or a bidding
method surveys, which gives WTP or bids for visibility or changes in
visibility as a function of the level of visibility or changes in visi-
bility, socioeconomic characteristics of the population, and physical
characteristics of the study area. Using these relationships, total
benefits could be estimated by deriving estimates of individuals' bene-
fits for the expected change in visibility in the current study area by
adjusting the previously estimated benefit function and appropriately
aggregating across all individuals expected to be affected by the change.
Strong tranferability requires either one of two broad conditions
to be met. First, all population characteristics, current and proposed air
quality levels and types of impacts, affected vistas, and land use char-
acteristics must be identical across both sites. In this circumstance
either site would be expected to yield the same benefit estimates. The
alternative condition is that the two study areas are identical except for
definable differences and that accurate and defensible benefit measure
adjustments could be made to reflect these differences. For example, the
residents of one metropolitan community may differ from those of another
metropolitan community previously studied only in terms of incomes and the
distances of their residences from the central business district. If the
impact upon benefit measures of these differences is well known through
repeated empirical verification, the previous study results could easily
and accurately be adjusted and transferred.
Differences in tastes and preferences of the affected populations
may be the most difficult to account for in transferring benefit measures.
Perhaps the easterner who migrated to the west is willing to pay more for
blue sky and clear views than the neighbor left behind. Perhaps the
vacationer in a national park has different preferences for clean air than
the nearby rancher. Willingness to pay for various aspects of air quality
must be known for all of these different types of people before results
from one area can be strongly transfered to estimate the benefits of an air
quality change for a different population.
Another requirement for strong transferability is that the dif-
ferences in population and site characteristics across studies be such that
benefit functions need not be extrapolated beyond the values over which
they have been estimated. The functional form of bid functions or WTP
function may not be appropriate for data values outside the range for which
they were originally estimated. To extrapolate beyond this range may
result in substantial error in the benefit estimates.
Strong transferability requires the completion of a large number of
benefit assessment studies to determine with confidence how differences in
people and air quality affect benefit measures. Also, it requires the
availability of a large number of complete studies from which to choose.
4-73
-------
It would be unlikely to find completely homogeneous conditions
across two study situations where all relevant variables can be measured
and have the same influence upon the benefit analysis. Until more studies
are completed, it will be difficult even to adjust benefit measures to
achieve precise per capita estimates for a new situation. However, it may
be possible to effect a weak transfer of results to establish either an
upper or lower bound on per capita benefits. Suppose the current situation
is nearly identical to a prior study except that the average income of
affected individuals is greater and the proposed improvement in air quality
is very similar but larger. Both of these changes are known to have
positive effects on benefit measures. Consequently, the prior results
could be used to establish a lower bound on the benefits. If these bene-
fits far exceed the costs of the improvement, no further estimate of
benefits may be necessary. On the other hand, an upper bound might just as
easily be established, and if the upper bound is considerably less than
costs, again no further research may be required.
The necessary conditions for weak transferability of results are
that the individual's tastes and preferences can be assumed to be similar;
that the direction, but not the magnitude, of the effect on benefit measures
of the difference in the situations is empirically established; and that
the differences, which will have substantial impact on benefit measures,
can be identified. A serious limitation is where some differences will
increase benefits and some differences will decrease benefits. Weak trans-
ferability cannot be established without estimates of the magnitude of
enough of these impacts to ensure that the benefit measures are either an
upper or lower bound for the new situation.
4-74
-------
CHAPTER 5
VISIBILITY BENEFIT ASSESSMENT GUIDELINES
5.1 Purpose of the Guidelines
This chapter summarizes the various research tasks necessary in a
typical visibility benefit assessment. For several reasons these guide-
lines are not intended to establish formal procedures that should or must
be followed. First, at present there is no formal legislative requirement
to perform visibility benefit assessments for which such a procedure would
be required. Second, the guidelines should remain flexible to incor-
porate what is learned through more research and repeated application.
These guidelines will help the researcher understand the step-by-
step procedures usually necessary to perform a visibility benefit analysis.
The steps discussed in the rest of this chapter are shown on Figure 5-1.
Steps 3 and 4, selecting the appropriate methodology and applying it to
obtain visibility benefit measures, have been the focus of this guidebook.
The remaining steps, many of which do not require economic analyses as
such, are included in the guidelines as an indication of what is required
to perform a complete benefit analysis and to incorporate the benefit
measures into a benefit-cost framework for resource planning. This dis-
cussion is meant to be a fairly brief outline and will refer to more
detailed discussions in other chapters as necessary. Where possible,
these guidelines also provide relevant sources of data and information.
5.2 Guidelines
5.2.1 Step 1 - Problem Formulation
The first analysis step is a nontechnical determination of the
nature and source of impacts as well as any existing public concern which
could affect the analysis. At this stage, initial screenings can be
used to determine whether technical investigations and further expenditure
of funds are warranted. Problem formulation involves identifying and
defining the basic issues under study. The steps in this phase include:
1. Describe the potentially impacted area. Specifically, define
and describe the area or areas that will or could be impacted by changes
in visibility conditions, including location, population (residents or
recreationists or both), degree of urbanization, land use characteris-
tics, etc. Impact areas may include one or more of the following: man-
5-1
-------
FIGURE 5.1
VISIBILITY BENEFIT ASSESSMENT GUIDELINES
Step 1
Problem Formulation
Step 2
Scenario Development
i
Step 3
Economic Method Selection
Application of
Economic Methods
Other
Contingent
Market
Bidding
Methods
Transfer Results
of Previous Studies
i
Step 5
Derive Aggregate
Benefits
Hedonic Other
Approaches Market
Approaches
Costs of
Control
Step 6
BenefitCost
Analysis
Other Costs
"and Benefits
5-2
-------
datory Class I Federal areas, other parks and recreation areas, Indian
reservations, integral vistas from scenic outlooks, all or parts of urban
areas, nonurban areas, or complete regions of the country (such as the Four
Corners region, the Great Lakes region, the Golden Circle of National
Parks, the South Coast Air Basin and the like).
2. Identify sources. Identify sources which will or could be
responsible for changes in visibility levels at impact locations and the
rates of emissions. These typically include urban plumes, power plants,
smelting operations, burning of forest products, agricultural activities,
natural hazes and dusts, construction activities and the like.
3. Define pollution changes. This includes the types and concen-
trations of pollution changes that will or could occur, with what pollutants,
and the potential impacts. Improvements or degradations, plume blight or
haze, and fugitive dust are examples of this type of change. Typical
pollutants of concern include SC>2, NOX, CO, TSP and ozone. Impacts may
include reduction in visual range, discoloration, and health impacts, as
well as agricultural and material damages. The potential intensity,
duration, frequency and timing of impacts should also be identified.
4. Identify legal foundations. Identify any legal foundations
and requirements relevant to the pollutant type, source and impact areas.
These are identified for several reasons. Many types of pollution impacts
are illegal, regardless of the costs or benefits of control, while other
types and levels of impacts are to be allowed only if costs of control
exceed benefits (see Chapter 1). Other regulations dictate the type of
impacts that, at a minimum, must be examined and how benefits should be
analyzed. Consideration of the relevant legal reqirements will assist the
analyst to define the need for benefit analysis and to meet any analysis
requirements. These may include the Clean Air Act and promulgated regula-
tions, Prevention of Significant Deterioration (PSD) programs, state
implementation plans (SIPs) and National Environmental Protection Act
(NEPA) regulations.
5. Determine public perceptions. Public attitudes and percep-
tions concerning actual and potential emission rates and air quality levels
must be determined through literature research or on-site investigation.
This is helpful in defining the aspects of the problem that are considered
important by those who are or will be affected and assists in identifying
any related special issues of public concern.
6. Determine if further analysis is required. Projects whose
"worst case" visibility impacts could easily be determined to be insigni-
ficant in terms of physical variables do not require further visibility
benefit analysis. This can be done using SAI/EPA (1980) physical screening
procedures based upon location of impacts; distance to receptors; and
emission rates per day of particulates, NOX and S02- Individual sources
which do not exceed critical levels established by these screening criteria
will not alter visibility conditions in a humanly detectable manner. If an
impact cannot be humanly detected through observable reductions in visual
5-3
-------
range, increased discoloration, or other measures of visual quality, there
are no visual aesthetic benefits or damages to be measured and further
visibility benefit analysis would be unwarranted (although analysis of
health related or materials related benefits may be desired), because
control of visibility impacts at this level is not required by clean air
legislation. The exception would be if in Task 5 some specific public
concern related to increased emissions were identified and merited separate
consideration.
5.2.2 Step 2 - Scenario Development
If visibility benefit analyses are to be undertaken, it is at this
stage that exact visibility scenarios should be developed for existing
conditions, the future without the proposed change, and alternative futures
with proposed changes in visibility conditions. This process serves to
define precisely the alternative levels of impacts to be evaluated. It is
also necessary to develop more exact technical relationships and impact
definitions to link changes in emission rates to changes in visibility
conditions using source-impact dispersion modeling as a means of linking
costs of control to the estimated benefit measures. The impact definition
formulated here will influence the selection of the economic valuation
methodology.
1. Establish existing air quality conditions. These include
visual range reduction using measures of visual range, light extinction,
or green contrast; discoloration using blue-red ratio, intensity or con-
centration of pollutants; frequency, duration and timing; and types of pol-
lutants and sources. Existing conditions may be obtained from airports or
the National Oceanic and Atmospheric Administration (NOAA), urban moni-
toring facilities often operated by state air pollution control agencies,
and monitoring facilities operated at and around several mandatory Class I
Federal areas by or for EPA, EPRI and private organizations. (See Chapter
3 and EPA 1979.) It is also necessary to establish which of these air
pollution impacts are perceiveable by humans, how they are perceived, and
what the important visual aesthetic characteristics of the obstructed views
are. Dispersion modeling, as suggested in Task 4 below, is required to
establish the links between emission rates and air quality levels.
2- "Future without" scenario. Establish the most probable future
conditions without the proposed change, if any change is expected, in the
same manner as for existing conditions. Defining the expected future
conditions without the proposed change establishes the appropriate baseline
against which to compare and evaluate the impacts of proposed changes
through time.
3- "Future with" scenario. Establish the conditions for one or
more future scenarios in the same manner as for existing conditions.
Precisely define proposed changes in visibility conditions to be evaluated.
5-4
-------
4. Conduct initial physical impact analysis. The level 2 and
level 3 analysis suggested by SAI/EPA (1980) could be used to perform more
detailed impact analysis for physical parameters. This analysis can
account for locations, atmospheric conditions, timing, topographic effects
of plume transport and diffusion, and the effects on the scenic beauty of
the area. Dispersion models for plume analyses, such as PLUVUE, have been
developed by EPA/SAI (1980); ERT, Inc. (Drivas et al. 1980) and Radian
Corporation (Fabrick et al. 1980). At present only the PLUVUE model has
received limited validation as part of the EPA VISTTA program; however, the
other models are generally very similar. Regional haze and long distance
transport of emissions models are even less well developed. Recent research
is also reported in the EPA (1979) and Chapter 3. Since this step is
outside the realm of economics, it requires coordination with atmospheric/
meteorology researchers.
5.2.3 Step 3 - Selecting the Approach to Estimating Economic Benefits
Once enough information has been gathered about the air quality
impact to define the problem and develop alternative impact scenarios, the
approach to estimating economic benefits of visibility can be selected.
The first issue that should be addressed is whether results from previous
studies can be transferred to the current benefit analysis problem at a
level of accuracy sufficient for decision making. Transferring results
will save considerable time and expense in performing benefit analyses, but
this involves meeting several requirements. At present, many more benefit
studies are needed before there will be a sufficient set of studies to be
able to transfer results from them to new problems on a regular basis. Yet
when results can be either strongly transferred to yield precise benefit
estimates, or weakly transferred to establish that costs are either sub-
stantially larger or smaller than benefits, performing new benefit studies
will not be required. (See Section 4.6.)
When prior results cannot be transferred satisfactorily to the
current problem or when the analyst desires to validate these results, a
benefit estimation study must be undertaken. In some instances the analyst
may desire to apply more than one technique as a means of substantiating
results.
Because each study problem will have unique features, the appro-
priate technique must be selected on a case-by-case basis. This section
outlines several important criteria that should be considered when making
this selection, and provides general indications about which technique is
appropriate in which circumstances. The most important considerations will
be the location and the nature of the impact to be evaluated. Other
considerations include data requirements, necessary assumptions, costs, and
timing requirements. Emphasis is placed here on bidding methods and the
hedonic property value approach, since these are the primary techniques
that have been applied in visibility benefit analysis. Many of the selec-
tion criteria also apply to other actual market and contingent market
5-5
-------
Table 5.1
Selection of Visibility Benefit Estimation Method A Summary1
Selection
Criteria
Transfer Results
of Previous Studies
Bidding Methods and
Other Survey Approaches
Property Values and
Other Market Approaches
Major strength
Major weakness
Cost and timing
Locations where impacts can be
addressed
Types of values
Separability of impacts
(health vs. aesthetics, etc.)
Types of impacts that can be
addressed
Evaluation of current vs.
future; impacts
Atmospheric data requirements
Economic and population
data
Reliability of data and results
Assumptions
Lowest cost, quickest results
Lack of prior studies and consistent
empirical analysis
Lowest cost, quickest results
Ac present: weak transferability for
urban areas and Southwest parka
Constrained by prior studies, primarily
activity values
Constrained by prior studies, primarily
not separated
Constrained by prior studies
Both, depending upon prior studies
Accurate data required to match current
situation to prior studies
Affected population characteristics
Depends upon accuracy of prior studies
and match to current impacts
Flexibility in application to
location, type of impact and
values
Reliability of data from hypo-
thetical questions
Highest costs, often takes > 1
year, seasonal problems
All locations, Class I, urban,
rural, other
Activity, option, existence
Benefits are separable
Haze or plume, and rates and
levels of occurence; uniform
or variable
May evaluate both current or
future impacts
Accurate data required for al-
ternative "typical situations"
across area
Affected population characteris-
tics must be cross-checked with
sample characteristics
Some uncertainty due to hypo-
thetical approach
Reliability of data re-
flecting actual behavior
Availability of data for
nonurban locations and re-
quired variations in impacts
Medium costs, usually < 1
year
Urban, or where sufficient
data exists
Activity, some option
Separability not yet
addressed
Haze or plume, and rates
and levels of occurence;
must vary across study area
All alternative Impact levels
should occur In study area
Accurate data required on all
pollution levels at a neigh-
borhood or census tract level
Property values, property use
and characteristics plus pop-
ulation characteristics
Reliable market data, some
uncertainty regarding estima-
tion techniques
1Thls table is to be used only as a
criterion can be applied to travel
viaibility benefit analysis-
See Table S.2.
summary of more detailed comparisons contained in Chapters 4 and 5 of this
costs, wage and salary hedonic techniques, etc., as they are developed for
guidebook. The same
application to
-------
approaches, such as travel cost approaches, and wages and salary analysis.
In general, bidding methods and other contingent market approaches are more
flexible than the hedonic approaches, but when the hedonic approaches can
be applied the data are often more reliable. However, a violation of the
key assumptions behind either technique may render it inapplicable.
The following are some of the criteria that should be considered in
selecting the appropriate benefit estimation technique. These selection
criteria are summarized in Table 5.1.
1. Location of the impact. Bidding methods can be applied to
estimate benefits of visibility changes in all locations, including urban,
rural, and Class I areas. Property value studies and other actual market
approaches can be used only in urban areas or in highly developed recrea-
tion areas or nonurban areas where property or wage and salary markets are
extensive enough to generate the necessary data. Bidding methods, other
contingent market approaches, and variants of the travel cost approach
are probably the logical choice for estimating air quality benefits in
most recreational, wilderness, and sparsely populated areas. The hedonic
technique applied to residential property values will pick up only impacts
at the home site, while bidding methods can pick up air quality impacts
that occur at residential, recreation and work sites.
2. The nature of the impact. The nature of the air quality
impact will be determined by several different aspects of the problem.
The following are some of the types of impacts that need to be considered
when selecting an air quality benefit estimation technique.
Current or future impacts. Bidding methods can be used to evalu-
ate alternative visibility scenarios which are currently being encountered
or which are expected to be encountered in the future, although the future
scenarios will be hypothetical to the respondent unless he has experienced
it elsewhere. Property value studies can capture the valuation for al-
ternative visibility scenarios that have been or are currently experienced
in some portion of the study area. Scenarios of visibility levels that
have not occurred could be evaluated with the hedonic approach only by
transferring results from other locations or by using local data and
extrapolating the estimated WTP functions to the proposed visibility
level.
Impacts from haze or plume. All methods can address both haze and
plume visibility impacts of air quality changes as long as the impact
affects the individuals' behavior being analyzed. This may be a problem
where there are uniform impacts across individuals. For example, an urban
plume or haze may uniformly obstruct a prominent local view or degrade air
quality equally for all households in a community. Property value dif-
ferences within the community would not usually reflect the benefits of
controlling these problems, unless changes in air quality over time were
reflected in property values and this data were obtainable.
5-7
-------
Frequency, intensity, and duration of impacts. Bidding methods
and other contingent market approaches could conceivably test for the
effects of variations in the frequency, intensity, and duration of the air
quality impacts, but the techniques are constrained by the respondent's
limited capacity or desire to absorb information and differentiate among
alternatives. Hedonic techniques could test for the effects of different
frequencies of certain air quality levels if the available data are suffi-
ciently detailed and accurate.
Aesthetic, health, or materials impacts. These three air pollution
impacts often occur together and are often highly correlated. In many
instances it will be difficult to separate benefit measures for the sepa-
rate impacts. With careful survey control and impact definition, the
economic effects of these different air quality impacts can be separated in
the bidding and other contingent market approaches. This separation will
be difficult with hedonic property value approaches unless measures re-
flecting the different impacts are not highly correlated with one another.
Such measures are, however, frequently correlated because many pollution
causing activities, such as burning of fossil fuels, produce several
different kinds of pollutants that each have different kinds of impacts.
3. Type of value for which benefits are to be estimated. Complete
benefit estimates would include activity, option and existence values, but
depending on the goals of the current study the researcher may choose to
examine only one or two of these values. If impacts are at a national
park, activity values must be determined for users of the park. Option and
existence values must be estimated by obtaining benefits for users and
non-users. Benefit estimates for non-users cannot be obtained on-site and
will often rule out most actual market and travel cost approaches. The
same problem also occurs for existence values in urban areas. Bidding
methods can be used to estimate each of these types of values. Other
contingent market approaches can capture activity values, while their
ability to capture option and existence values has yet to be demonstrated.
Hedonic techniques can capture the activity value of air quality at the
property and work sites, and to some extent the option value of the current
air quality level, although separating these would be difficult. Existence
values probably do not affect property values or wages and salaries, and
therefore are not captured by the hedonic technique.
4. Data requirements and reliability. These include both secon-
dary data obtained from outside sources and primary data generated by the
current study. For example:
Atmospheric data. All approaches to benefit-cost analysis re-
quire that emission rates be accurately linked to visibility and other
air quality measures used in the analysis and that indivduals sense the
differences in air quality conditions. In this way the correspondence
between benefits and costs for proposed changes can be maintained. The
hedonic approach further requires accurate estimation of all potentially
relevant current air quality measures at the neighborhood or census tract
level. Survey approaches require that selected typical levels of air
quality can be tied to emission rates and related to respondents in a
manner they can understand and accurately respond to.
5-8
-------
Economic and population data. The survey methods require accurate
economic and population data to establish survey sampling procedures,
to check survey response rates, and to aggregate individual valuations to
totals for the impacted population. This is sometimes a problem in non-
urban or recreation areas. The hedonic technique requires extensive
property use and characteristics data as well as household and neighbor-
hood characteristics data for the entire study area.
Data reliability. The bidding methods generate primary data on an
individual's willingness to pay for various levels and aspects of air
quality; however, these data are subject to potentially substantial in-
accuracies due to the hypothetical nature of the questions, and due to
potential strategic and information biases. Reliable secondary data
for the hedonic approach are usually available in urban areas, but special
procedures and assumptions may be required to make data from various
sources compatible and applicable for the estimation technique.
5. Necessary assumptions. Assumptions used in benefit estimation
techniques can be categorized as to whether they are analyticalused in
the theoretical developmentor operationalused in application of the
procedures. Table 5.2 presents a summary of assumptions used for three
representative techniques; property value studies, bidding methods and the
contingent travel cost approach. If the assumptions used in any technique
are not valid in theory or not met in practice the validity of the results
generated by the technique are weakened.
To transfer results from previous studies it is first required that
the previous study technique and results be defensible. Next, it must be
assumed, or hypothesized, that the individuals, assumptions, and circum-
stances are identical across the two studies or are different in definable
ways and that the range of visual quality for which the benefit measures
have been determined is comparable to the range of visual quality currently
under study.
6. Costs and timing requirements. The costs and amount of time
required to perform a benefit analysis may significantly affect the choice
of estimation approaches. Many efforts will be comparable across approaches,
including model design and initial data analysis. Efforts to obtain and
enter data may vary significantly.
Survey methods require a survey instrument and sampling procedure
design, survey pretests and evaluation, interview training, and an on-site
survey effort. Costs will vary dramatically depending upon location of
the interview site, availability of trained labor, time limitations, and
the length and type of interviews undertaken. One hundred or more com-
pleted interviews using a cluster sample can easily vary from $10,000 to
$50,000 and take from one month to one year just to complete the survey
design to final data receipt tasks. In order to obtain an adequate sample
of users, survey methods may also have to wait until a peak use period in
recreation areas.
5-9
-------
Table 3.2
ASSUMPTIONS IN VISIBILITY BENEFIT ESTIMATION
Estimation Technique
Assumptions
Property
Value
Bidding
Method
Contingent
Tra ve 1
Cost
I. ANALYTICAL ASSUMPTION'S
1. Consumers make choices to maximize utility subject to budget constraints. X
Utility functions are regularly shaped; directly or indirectly Increasing
In visibility, all goods, all activities and income. All individuals
have the same utility function or are different in definable and oeasure-
able ways (education, age, ethnic background, etc.)* Income is assumed
to be constant*
2. Only air quality environmental changes occur (no changes in water quality, X
congestion, etc.)
3. Air quality changes have the following Impacts upon prices.
a* Changes in air quality do not affect any prices.
b. Changes in air quality affect only property prices. X
4. Air quality changes have the following impacts upon rates of activities
and expenditures.
. Activities and expenditures do not change with changes in air
quailty.
b. All activities and expenditures may change.
c. Only housing activity and expenditures change. X
5. Households purchase one unit of property or. if more than one. all are X
Identical.
6. Property markets are pervasive, have no inperfectlons and are in equill- X
brlum so as to reveal utility maximizing choices.
II. OPERATIONAL ASSUMPTIONS
1. Technical measures of air quality or pollution that are used by researchers X
accurately reflect those aspects of visibility impairment to which in-
dividuals react in the given situation.
2. Visual presentation of air quality conditions accurately communicate those
attributes of air quality to which Individuals react In the given situa-
tion, and are accurately correlated to technical measures.
3. Realistic hypothetical market situations can be described by the researcher
and understood by the respondents, and, respondents are capable and willing
to accurately predict their behavior in the described situations*
4. Individuals are not biased by the questionnaire procedure or, if they are,
these biases can be detected and correctly accounted for in the analysis.
5. Consumers have mobility and full information on alternatives and are X
unconstrained in making choices.
6. Weak separability in the utility function is assumed so that only re-
lated expenditures and activities need to be examined.
7. Weak separability in the utility function is often assumed so other types X
of averting behavior and expenditures need not be examined (this parallels
analytic assumptions 3b and 4c).
8. The urban area as a whole can be treated as a single market without in- X
perfections, or the Imperfections can be accounted for.
9. A random sample of individuals representing a cross-section of those
Affected can be drawn.
10. A random sample of properties can be drawn, which exhibit a continuous X
variation over an adequate range of the air pollution measure and
property characteristics to perform statistical analysis.
11. Appropriate function forms can be specified for analysis.
. Bid functions or WTP functions can be specified which correctly X
reflect underlying utility functions.
b. The correct hedonic price function can be specified and the X
correct form is nonlinear.
12. All influences which affect the valuation process can be deteralned
and accurately quantified.
This includes air quality; population, housing, and neighborhood X
characteristics.
b« This Includes air quality and population characteristics.
c. This includes air quality, rates of activities, expenditures, and
population characteristics; and the value of travel time and costs.
5-10
-------
The costs of the hedonic approach will depend upon the difficulty
of gathering the necessary data. In most urban areas market research or
census data tapes may be quickly obtained for a few hundred to a few
thousand dollars. Other local data are often immediately available from
local government agencies. If additional data are needed, primary surveys
or hiring professional real estate appraisers will considerably increase
expenses. The costs of obtaining and organizing the necessary secondary
data for hedonic approaches will typically be less than one-third those
of a survey effort. This will depend in part upon the effort required to
get the data running on the local computer. The time frame of the data
will affect the validity of the hedonic approach, but the implementation
of the study will not be constrained by the same sort of seasonal problems
that can afflict the survey approach; therefore, the hedonic approach can
usually be more quickly implemented. Due to the complexities of this
approach, the modeling and data analysis costs are substantially larger
than for the bidding method approach.
5.2.4 Step 4 - Application of Economic Methods
This section focuses upon the bidding method and property value
approach to benefit estimation because these are the two approaches most
thoroughly developed and commonly used to date.
5.2.4.1 Application of Bidding Methods
In applying the bidding methods, it is neccessary to design the
appropriate benefit estimation model, to finalize the selection of alter-
native scenarios to be evaluated, and to determine how they will be pre-
sented to respondents. This process can be as follows:
1. Selection of benefit measures. At this stage, it is appropri-
ate to determine the benefit measures the survey will attempt to estimate
from the WTPB, WTPG, WTAB, and WTAG approaches. Property right
structures must also be considered.
2. Select alternative scenarios to be evaluated. Usually no
more than four alternatives should be presented to respondents.
3. Select the methodology for depicting the visibility conditions.
Typical approaches include regular photography, overlay techniques, retouch
techniques, artist renderings and computer simulated techniques. Regular
photography is the predominant approach for depicting visibility conditions.
The standard application is to use a 35mm camera with a 135mm lens, which
gives a precise one-for-one image, and Kodak-chrome II film with film speed
of ASA25 and typical exposure of F8 at 1/125 second, processed in batch at
Kodak labs. The pictures are usually 1/3 - 2/3 split between sky and land
to give foreground, middleground and distance features as well as sky
conditions. The direction of the scenes and time of day are important
5-11
-------
considerations because of the influence of backlighting. No-cloud scena-
rios are becoming standard because the measurement of visibility conditions
is more accurate under this condition. Further references can be found in
Chapter 3, Ettenheim (1979), and EPA (1980). The EPA is also currently
developing a twice-a-day historical photo file at several mandatory Class I
Federal areas.
Computer simulation and other techniques are used when the proposed
alternative conditions do not or have not existed, or simply cannot be
captured with regular photography. This technique has been developed and
used by Williams et al. (1979) at Los Alamos Scientific Laboratory.
Several scenes are frequently used, depicting typical vistas and special
landmarks potentially impacted by visibility changes.
4. Determine survey procedures. This must take into consideration
which impacted populations are to be surveyed, where and how they are to be
surveyed, sampling procedures and the design of the survey instruments. As
previously discussed, personal interviews and guided interviews are typical
methods. Survey procedures in recreation areas are discussed in Section
4.2, Brookshire et al. (1976) and Rowe et al. (1980). Examples of survey
procedures in urban areas can be found in Brookshire et al. (1979), EPA
(1981) and recent NFS/EPA/Wyoming work to be released.
5. Design the survey instrument. This is the most critical stage
for successful application of a bidding method or contingent market approach.
It is discussed in detail in Section 4.2.
6. Pretest. All survey instruments and procedures should be
pretested in the manner and at the site for which they are designed. The
procedure and the instrument are then redesigned based upon pretest
evaluations.
7. Implement the survey. The survey should be implemented as
designed to a sample of sufficient size to make the necessary statistical
inferences with the desired degree of precision. (The appropriate sample
size is best determined from the pretest.) Samples of 100 or more for each
major group (recreationists, residents at each separate location, etc.) are
usually required. See Section 4.2.
8. Analyze survey results. At this stage, standard analysis of
mean, variance, correlation, cross-tabulation, and regression analysis of
the bids and socioeconomic variables and other data are conducted. Bid
functions relating individual bids to visibility levels, socioeconomic
variables and other study variables may also be estimated. Several alter-
native functional forms must be estimated to test the robustness of the
estimated impacts of variables upon reported bids. Mean values by classi-
fication scheme (income levels, urban, non-urban, etc.) and/or bid func-
tions should be developed for each potential change in scenarios for each
value type (activity, option, existence) estimated for each affected
population group. Finally, tests and verification of the survey procedure
and instrument biases may be carried out and other special questions
evaluated (for details see Section 4.2).
5-12
-------
5.2.4.2 Application of Hedonic Approaches
The application of the hedonic approach encompasses three major
steps beyond the determination of impacted populations and data avail-
ability: the first is final data selection, the second is data organiza-
tion, and the third is statistical analysis.
1- Select final data. The two most important data selection
decisions are which property value and air pollution data to use. Care
must also be taken to ensure that all the important amenities that in-
fluence residential location decisions in the study area have been
identified.
Select property value data. Household-level data are sometimes
available from market research firms, real estate or financial organiza-
tions, or local government. Household-level data are preferable if enough
house structure and household socioeconomic data are also available at the
household level. Examples of typical data needed were given in Table 4.4.
If adequate household-level data are not available the analyst must decide
whether to use U.S. Census data, which will be available for any urban
area; aggregate data from other sources, such as local government, for
census tracts or other small divisions of the study area; or whether to
have professional property value assessments done in the study area. The
latter choice will be considerably more expensive and may be worthwhile
only if equally reliable house structure and household-level socioeconomic
data can be obtained. Many studies have settled for the census data, which
provide census tract averages of property value, house structure and
household socioeconomic characteristics. Census tracts are selected to be
as homogeneous as possible, but some information is inevitably lost in the
aggregation.
Select air pollution data. Most property value studies have used
measures of pollutant concentrations to represent air quality. These have
the advantages of being readily available for most urban areas and of being
reasonably well linked to emission sources. However, pollutant concentra-
tion measures may not be entirely adequate for visibility benefit analysis
because visual impacts are often caused by combinations of pollutants or by
some components of the measured pollutants. For example, Heisler et al.
(1980) found that only the smaller particles that make up about half the
measured TSP contribute significantly to the visible haze in the Denver
area. Alternatively, direct data on aesthetic impacts such as visual range
and discoloration are not as readily available and the links between such
measures and emission sources are less well known. Future research on this
subject may eventually lead to availability of measures of aesthetic
impacts that are adequate for property value studies.
5-13
-------
Which measure of air quality to use will depend on which pollutants
are most prevalent in the study area and which of these are responsible for
perceived impacts. If available, several measures should be used to check
for consistency and to determine the best proxy for the impact being
studied. Usually only one air quality measure can be used in a hedonic
price function due to the typical correlation between measures. There are
not usually air quality monitoring stations in every census tract or even
in every neighborhood in an area, therefore, air quality measures for each
neighborhood must be extrapolated from monitoring station data. This is
often done using isopleth maps developed by local air quality sur-
veillance agencies that trace typical air patterns over an area. Some air
pollution dispersion models are also available though these are still in
developmental stages.
Verify important neighborhood amenities. This will improve
the reliability of the estimated influence of air quality on property
values. On-site observation and/or a small survey of residents might bring
to light overlooked amenities that are influencing property values.
Appropriate data to represent all important neighborhood amenities must be
obtained.
2. Organize data. Data from different sources must be made
compatible, and a study sample of the impacted population must be selected.
This can be done as follows:
Adjust data from different sources for compatibility. Even if
census data are used, additional data must be obtained from other sources.
Transportation, tax, public services, school, and employment data are
available from local governments, although these are not always reported
for the same area divisions. For example, school district boundaries and
census tract boundaries do not usually coincide. These differences must be
reconciled in some reasonable and consistent manner.
Select study sample. A sample must be selected from the impacted
population to limit some of the non air quality differences between proper-
ties. Many studies have eliminated tracts with less than a certain percen-
tage of owner-ocupied houses or more than a certain percentage of business
properties. Others have used more sophisticated statistical analysis
to group or pair neighborhoods or tracts to minimize differences except
air quality. In smaller or less populated study areas this selection
process will be constrained by the smaller number of property value
observations. The sample must also be selected in accordance with standard
sampling procedures to ensure that it is representative of the impacted
population.
3. Perform statistical analysis. The next major step in the
hedonic approach is the statistical analysis, which includes estimation of
the hedonic price function, WTP functions and benefits or damages.
5-14
-------
Estimate hedonic price function. First, the hedonic price function
must be estimated using standard statistical procedures. Two particularly
serious problem areas for the estimation of the hedonic price function are
the question of whether market segmentation exists in the study area and
the question of what is the appropriate functional form. These problems
are discussed in detail in Section 4.3.2.
Estimate WTP. The current procedure for estimating willingness to
pay for air quality is to take the .nar^lnal implicit price for air quality
as an observation of the household's WTP, as described in Section 4.3.2.
Demand determining household and property characteristics are used as
explanatory variables. Household income is probably the most important
characteristic, but other characteristics should be tested to ensure that
welfare measures are not unstable across different specifications of the
WTP function. Also, to assure that the WTP function is identified, some
considerations must be made regarding the supply of residential site air
quality.
Estimate benefits or damages of the changes in air quality.
From the WTP function, the change in consumer surplus for each household or
household group that will result from a change in air quality can be
derived. Alternatively, rough estimates of the change in consumer surplus
can be directly derived from the marginal implicit price function for air
quality, as described in Section 4.3.2.
5.2.5 Step 5 ~ Aggregation of Benefits Across All Affected Populations
Each technique estimates benefits for a typical individual with
certain characteristics for a precise unit of time. Examples include
annual benefits to homeowners segmented by pollution level and income
level, or daily benefits to typical recreators. The total society benefit
is the aggregation of benefits to all affected individuals for as long as
the impacts occur.
1. Aggregate across individuals. Bid curves, WTP functions or
mean values by selected classifications are aggregated across all affected
individuals for each value type (activity, option, existence) for each
potential change in scenarios. Total benefits are usually estimated on a
per year basis throughout the project lifetime. It may also be desirable
to use the Water Resource Council's Principles and Standards approach
(1979), which displays benefits and damages by categories of "region" and
"rest of nation" and specifically separates impacts to low income, un-
employed, or minority individuals to address equity issues of how costs and
benefits are distributed.
5-15
-------
2. Aggregate across time. Future benefits and damages are
often discounted to present values so that all benefits may be aggregated
to a single total benefit measure. Both the appropriate time horizon and
discount rate must be selected. Discounting represents an intertemporal
weighting of benefits for which there is substantial controversy as to the
correct weighting scheme. Arguments range from using a zero rate of
discount to a high rate of discount based upon public and private rates of
return on investments, efficiency arguments, and other related issues (see
Section 2.3). The discount rate used may be determined by legal require-
ments, as is the case for federal government projects; as a decision
variable; or by using existing private market rates and time horizons. In
most cases it will be beneficial to report aggregate benefits using several
alternative rates to demonstrate the impact of selecting discount rates
upon total benefit measures.
5.2.6 Step 6 - Benefit-Cost Analysis
Visibility benefits or damages, which have been the focus of this
guidebook, are only part of the total benefits and costs associated with
changes in the number or rate of operation of power plants, other industrial
sources, prescribed burning and multiple source regional emissions. Any
decision-making process for optimal resource allocation should take into
account all costs of emission control associated with alternative visibility
levels and all other environmental and non-environmental costs and benefits.
The steps in this phase are as follows:
1. Identify emission control costs. Where possible, initial visi
bility scenarios should have been developed so that the related costs of
emission control equipment can also be estimated. ^ Initial estimates for
best available retrofit and control technologies for existing and new
coal-fired power plants are available through the EPA (1980). The Bureau
of Census also publishes an annual Pollution Abatement Costs and Expendi-
tures . The costs of reducing non-industrial emissions, such as from
prescribed burning, must also be considered. For example, the costs to
timber land owners of using alternatives to burning, such as mechanical
treatment of forest residue, should be considered an emission control
costs .
2. Identify other environmental and non-environmental costs and
benefits. In many cases the estimation of other related cost and benefits
may be of substantial importance and may require even more efforts than
for visibility benefit estimation. Changes in air quality will also have
benefits in health, agricultural and materials damage categories. The
importance of considering other benefits is demonstrated in a benefit-
IA theoretically correct measure of the costs of emission controls would
be the decrease in producer and consumer surpluses caused by the additional
costs of producing the good affected by increased pollution control (elec-
tricity, automobiles, etc.) In most cases the change in the producer's total
costs is taken as an approximation, the accuracy of which depends on the
elasticisties of demand and supply of the good. (See Hirshleifer 1976.)
5-16
-------
cost case study for New Haven, Connecticut, which found visibility aesthe-
tic benefits to be up to five times larger than those from reducing acid
rain, agricultural and materials damages; but on the average only one-seventh
the magnitude of health-related benefits (Mendelsohn 1980). Other impacts
that may have to be considered include the effects of higher energy costs,
transportation costs, the social desirability of energy independence, land
use impacts, and the like. A complete analysis may also consider secondary
impacts. For example, increased visibility degradation in a national park
would decrease the well-being of tourists and the rate of tourism. Conse-
quently, local economic revenues would also be decreased. See NEPA regula-
tions for typical multi-objective resource planning impact categories and
methodology for ranking alternative projects; and see Mishan (1976), Sassone
and Schaffer (1978), Abelson (1979), and Maler and Wyzga (1976) for discus-
sion of benefit-cost analysis techniques.
5-1"
-------
CHAPTER 6
CASE STUDIES OF VISIBILITY BENEFIT ANALYSIS
This chapter discusses five visibility benefit studies in which
the bidding method was used and five in which the hedonic property value
approach was used. Although the bidding method has been applied in
several dozen other environmental/resource problems in recent years,
these five are the best known applications to visibility benefit estima-
tion. 1 There have been nearly two dozen property value studies related
to air quality and visibility.2 The property value studies discussed
here represent the most recent applications of the Rosen-Freeman theore-
tical framework, presented in Chapter 4, which is considered to be the
most defensible framework to date.
These case studies are being presented to exemplify actual appli-
cations, typical results, and strengths and problems for each of these two
approaches to assist researchers and policy makers in designing and evalu-
ating potential new visibility benefit studies. The discussion for each
case study follows, as much as possible, the guidelines presented in
Chapter 5. Many of the studies, however, do not follow or perform each
of the steps because they were the first attempts to test the economic
techniques or to establish order-of-magnitude estimates of benefits or
damages related to changes in air quality, and as such the researchers were
not concerned with performing complete benefit-cost analyses.
6.1 Case Studies Using Bidding Methods
The studies examined here are the Four Corners study, by Randall
et al. (1974); the Lake Powell study, by Brookshire et al. (1976); the
Farmington study, by Rowe et al. (1980a,b); the South Coast Air Basin
study, by Brookshire et al. (1979); and the Grand Canyon/Southwest Parks
study, by EPA (1981). With the exception of the South Coast Air Basin
study, all of these studies relate to visual range and discoloration
impacts upon Class I areas and local communities in the southwestern
United States induced by rapid growth in the coal-fired electric-generat-
ing industry. Each study examines at least one alternative level of
impacts, but seldom are the impact scenarios and benefit measures linked
to actual emission levels or control costs; nevertheless, the results
exhibit a high degree of similarity across studies.
'-For a partial review of bidding method studies for a variety of appli-
cations compiled prior to mid-1978, see NUREG 1979.
^For a review of property value studies completed prior to 1978, see
Freeman 1979a.
6-1
-------
A tabular comparison of selected elements of bidding methods is
provided in Section 6.2.7. Sample questionnaires and photographs are
included in Appendix 1.
6.1.1 The Four Corners Study
The Four Corners study, as reported by Randall, Ives and Eastman
(1974a,b), was the first in which a bidding method was used to assess the
economic benefits and damages associated with changes in air quality and
other environmental variables. As such, the study was more concerned with
developing the technique and obtaining defensible order-of-magnitude
damage estimates than with precise definitions and links between visi-
bility conditions, emission rates and control costs, as would now be
required for a visibility benefit-cost analysis. This study was part of a
larger socioeconomic impact assessment of the rapidly expanding coal-fired
electric-generating industry, including power plants and strip mines in
the Four Corners region (southwestern United States) undertaken by New
Mexico State University.
Problem Formulation and Scenario Development
The specific concern of this study was the Four Corners power
plant in Fruitland, New Mexico, and the Navajo mine, which provided low-
energy, low-sulfur, high-ash subbituminous coal. The 1970 power plant
capacity was 2,080 MW. It was expected that some 31,000 acres would be
stripped over the 40-year life of the mine and that 96,000 tons of parti-
culates, 73,000 tons of sulfur oxides, and 66,000 tons of nitrous oxides
would be emitted annually. The emissions had already considerably reduced
local visibility from the natural state of around 160 km down to between
60 and 100 km (Kneese and Williams 1978).
The Four Corners region is sparsely populated and is typified
by desert landscapes best enjoyed from long distances. At the time of the
study (1972) there were no cities of over 50,000 inhabitants in the
region; however, several large Indian reservations, including the Navajo
reservation, and several national parks and monuments are in close proxi-
mity (see Figure 6-1). In subsequent years actual and proposed emissions
in the Four Corners region became one of the driving forces behind the
1977 passage of the Clean Air Act.
6-2
-------
State of Colorado
"oven we«D Man
Son Juan Co
Montezuma Co. i La Plata Co
Figure 6.1. 1978 existing and proposed coal-burning power plants
in the Four Corners region of southwestern USA. Existing: 1 San
Juan Power Plant (2,175 MW); 2 Four Corners Power Plant (660 MW
and 1,000 MW). Proposed: 3 WESCO 1; 4 WESCO 11; 5 Burnum 11;
6 Burnum 1. (Source: Rowe et al. 1980b.)
Three scenarios of past and expected future levels of impact were
developed and supported with photographic representations. Scenario A
"showed the plant operations circa 1969, prior to installation of addi-
tional emission control equipment, producing the historical maximum
emission of air pollutions." (Page 137.) Scenario B "showed an inter-
mediate level of damage...showed the plant circa 1972, after additional
controls had reduced particulate emissions (i.e., the type of emission
most destructive of visibility)." (Page 137.) Scenario C "showed the
plant with visible emissions reduced to zero." (Page 138.) The impacts of
each scenario also included various degrees of reclamation of soil bank
(A-unreclaimed, after strip-mining, B-leveled but not revegetated, and
C-natural), visible transmission lines (A-highly visible, B-visible in
distance, C-natural) and a visible power plant. The alternative scenarios
were depicted with regular photograhy; however, the scenes and their
contents were not standardized.
6-3
-------
Application of the Bidding Method
Bidding methods were selected as the method with the most potential
for estimating benefits and damages associated with the proposed changes
in environmental conditions. The location of the impacts, much of which
were on or over Indian reservations, national parks, state and Federal land
and utility company land, ruled out the property value approaches.
A typical iterative bidding procedure was used. After an introduc-
tion, respondents were shown the pictures representing the alternative
scenarios and asked their maximum willingness to pay for proposed improve-
ments in environmental quality from state A to states B and C (WTP^ = CS
measures). A few respondents were also presented willingness to accept
compensation question (WTA& = CS). Three payment vehicles were used.
A sales tax on all purchases in the Four Corners Interstate Air Quality
Control Region to finance environmental improvements and which would be
required of all citizens was presented to some non-reservation residents.
An electricity-bill vehicle was presented to other residents. In this
case, the respondent was asked his current monthly electricity bill, then
asked to imagine that an additional charge would be added to his electricity
bill and to the electricity bills of everyone who used electricity produced
in the Four Corners Region, even people as far away as Southern California
(page 140). The monies would then be used to repair the environmental
damages. A user fee for recreation services was presented to non-resident
recreationists in the region, and a general fund for environmental protec-
tion was proposed on the Indian reservation. Zero bids were evaluated as
described in Section 4.2.3. Standard socioeconomic questions were also
asked. To capture valuations for the three predominantly affected popula-
tions, pretested personal interviews were completed for 526 residents, 71
on-reservation Indians and 150 non-resident recreators at several parks,
monuments and forests in the region.
Analysis of Results
Randall et al. (1974a) reported the results for the sales tax
approach used with residents. The mean sales tax bid for residents for
an improvement from the worst to moderate conditions was $50 annually, and
from worst to best conditions, $82.00 annually. The effect of higher
incomes on bids was modest, as reported by income elasticities ranging from
.39 to .65 for non-reservation residents. Income elasticities statisti-
cally equal to zero were reported for reservation residents and recrea-
tionists .
Few other results of the bidding procedure are reported in Randall
et al. (1974). Apparently no tests for vehicle bias or other biases were
conducted. Bid functions, which relate individuals' bids to visibility and
socioeconomic characteristic, were not reported. Randall et al. (1974b)
indicates protest bids as a percent of total bids ranging from 21% for
utility bill payment vehicles to 32% for sales tax payment vehicles for
local residents and up to 49% for residents of Indian reservations with
WTPG questions. Protest bids were given by 52% of the resident respon-
dents and 61% of recreationists given WTAB questions.
6-4
-------
Aggregate Benefits Measures
A central component of this study was the aggregation of bids
to derive a public goods bid curve which relates aggregate bids to visi-
bility levels, as proposed by Bradford (1970). (See Section 2.4.4.) The
results of the sales tax payment survey for residents were added to the
user-fee payment survey for recreationists and aggregated to a total
regional WTP- The results of the electricity bill payment approach were
also aggregated separately for all consumers of Four Corners power. This
second aggregation was not reported because the authors felt that "More
faith may be placed in the regional bid since that bid was derived from all
samples of all segments of its relevant population" (page 141), while the
southern California consumers of Four Corners power were never surveyed.
Table 6-1 reports a summary of aggregate results from Randall
et al- In summary, the authors state in a footnote:
We are not yet in a position to present a complete benefit/
cost analysis of the abatement of the aesthetic environmen-
tal damage associated with the Four Corners power plant and
Navajo mine. Preliminary and tentative calculations indicate
that, if our attribution of most of the benefits reported
here to abatement of particulate air pollutants is reasonable,
99.7% abatement of particulate emissions (the current New
Mexico standard for 1975) is economically justified on the
basis of aesthetic considerations alone. Some additional
abatement beyond the 1975 standard may be justified. The
economic benefits from that abatement which has already
taken place appear to far exceed the costs. (Page 145.)
Table 6.1: Aggregate Benefits For Abatement of Aesthetic Environmental
Damagages Associated With the Four Corners Power Plant, 1972
Item
Emission (tons of
particulates per year)
Estimated regional aggregate bid
($ millions per year)
Standard error
($ millions per year)
95% Confidence limits
($ millions per year)
Situation
A B
96,000 26,000
0 15.54
1.24
+2.43
C
0
24.57
1.52
+2.97
6-5
-------
Evaluation Comments
This study was the first to show that bidding methods could be
applied to estimating benefits and damages associated with changes in
environmental quality. In addition, it further established a research
agenda for years to follow.
The study has several weaknesses. Because the scenarios were not
standardized and included impacts relating to visibility, soil bank recla-
mation, visible transmission lines, and visible power plant siting, the
benefits/damages of each type of impact cannot be separately or accurately
estimated. Next, the electricity bill payment vehicle lends itself
to potential strategic bidding. If respondents had believed that all power
users, including Californians, were to pay for improvement, from which all
benefits would be derived locally, they may have bid higher to increase
local benefits paid by non-local power users. Finally, an aggregate bid
curve was estimated across the three levels of environmental impacts, but
little confidence can be placed in estimation procedures based upon three
data points.
6.1.2 The Lake Powell Study
The Lake Powell study, as reported by Brookshire, Ives and Schulze
(1976), was undertaken as part of the NSF-RANN Lake Powell Research Project
conducted at the University of New Mexico.
Problem Formulation and Scenario Development
At the time the study was initiated (1974) the 2,400 MW Navajo
coal-fired electric-generating power plant was being built overlooking Lake
Powell, and several other large power plants were being proposed for the
area. The Navajo plant was to be fully operational in 1976. The Kaiparowitz
power plant, which received so much attention in the early 70's, was to be
located just to the north of Lake Powell and was to be even larger than
the Navajo plant.
The Navajo power plant was expected to have significant environ-
mental impacts. The potential of another large plant was of great concern
to environmentalists and local residents. (An example of a typical Navajo
plant plume as viewed from the neighboring community of Paige, Arizona,
in 1979 is shown on Figure 1-6 of the EPA Report to Congress, 1979.)
The study concerned itself with the "disturbing visual impact of large
power plants with 700 ft. stacks in a predominantly natural recreational
setting... (and) ... the smoke plumes and resultant loss of visibility,
which may obscure the view." (Page 326.) The Lake Powell region has very
few residents but is near several Indian reservations and centered in the
so-called Golden Circle of natural parks in southern Utah and northern
Arizona. Visitation at Lake Powell alone is now approaching two million
visitor-days per year.
6-6
-------
Three scenarios were developed. The first (A) represented existing
conditions. The second (B) represented existing conditions plus a new
visible power plant with no smoke plumes. The third (C) depicted a visible
power plant with smoke plumes. No formal relationships were estimated
between the characteristics of the plume, such as frequency and intensity,
emission rates or control costs.
Application of the Bidding Method
Bidding methods were selected because of the limited resident
population and the central characteristics of the impact area as a national
park recreation area.
An iterative bidding technique, much the same as in Randall et al.
(1974), was employed. Regular and retouched photographs of the alterna-
tive scenarios using the Navajo power plant were presented to Lake Powell
recreators and local residents. (The bidding instrument and stylized
pictures are contained in Appendix I.) Respondents were then asked what
was the maximum entrance fee they would be willing to pay to prevent
construction of another plant (WTP1* = ES measures): first, where only the
plant would be visible from Lake Powell (scenario B), and second, where
both the plant and pollution would be visible (scenario C). All respon-
dents would be required to pay the same daily fee. The questionnaire also
analyzed zero bids and gathered socioeconomic information. Pretested
personal interviews were completed with 19 local residents of Paige,
Arizona, 20 motel visitors,. 22 developed-campsite users and 21 remote-
area campers.
Analysis of Results
Table 6.2 presents the mean bids by sample group for both pro-
posed changesA to B and A to C. The bids of residents are statistically
significantly smaller than for nonresidents. This is due in part to
income effects and because many local residents directly or indirectly
derive their livelihood as a result of the plant's operation and find the
impacts less offensive. The nonresident groups are statistically indis-
tinguishable due to small sample sizes. The authors suggest that the
level of bids "seem quite reasonable when compared to the average family
expenditure of visitors at GCNRA (Glen Canyon National Recreation Area),
which approximates $24 per day for 1973. This figure does not include
expenditures for travel costs nor the cost of recreational equipment."
(Page 339, parentheses added.) The values are statistically significantly
larger to prevent visible power plants and pollution rather than just
preventing visible power plants.
6-7
-------
Table 6.2: Lake Powell Power Plant Siting Values
Benefits in Dollars Per Day Per
Visitor Party
Group
Residents
Remote
Developed
Motels
Population sample means
Estimated aggregate yearly bid
Mean Bid: A to C ($)
1.75
(0.22)
3.38
(0.50)
2.60
(0.41)
3.11
(0.33)
2.77
(0.19)
727,600
(50.270)
Mean Bid: A to B ($)
0.87
(0.20)
2.11
(0.54)
1.08
(0.29)
1.94
(0.48)
1.58
(0.24)
414,000
(62,400)
Standard errors in parentheses.
Source: Brookshire et al. (1976), page 338.
Several other results were also reported. First, about 21% of
the bidders gave rejection or protest bids, many indicating that the
payment vehicle was unsatisfactory. A theoretical model presented sug-
gested that if strategic bias was present the bid would be bimodally
rather than normally distributed.3 The bid distributions were reported
and appear more normally than bimodally distributed, and there were no
extremely large bids encountered.
o
JThis is because all recreators were to pay the same entrance fee, which
would be used to fund environmental improvements. For an individual to
strategically affect the results so they more closely resembled his true
value rather than the sample mean value, he would have to significantly
overbid or underbid relative to his true value (depending upon whether
his true value was larger or smaller than the sample mean). Substantial
strategic bidding would therefore lead to either a bimodal or, at least,
flattened distribution of bids relative to a normal distribution expected
from sampling theory.
6-8
-------
The authors next examined the income effect as it potentially
affects the difference between theoretical CV and EV measures (the study
actually obtained ES rather than EV measures). The authors found that the
measured income effect upon the actual EV bids received theoretically
predicts that CV values if obtained, would have been of almost exactly
equal magnitudes as the EV measures. This is in keeping with the theore-
tical arguments made by Willig (1976) and reported in Section 2.2.4, above;
however, the application of ES and CS measures need not yield equivalent
empirical values (see NUREG 1979; Randall and Stoll 1980; and Rowe and
Blank 1981).
Aggregate Benefits Measures
To calculate aggregate benefits, projected visitor populations
for each survey group were estimated. Visitor-party-days were then esti-
mated by adjusting for length of visit and family size. Next, by applying
the mean entrance fee bid per visitor-party-day to the aggregate visitor-
party-day estimate, total benefits were estimated. "The results indicate
that recreationists in GCNRA in 1975 were willing to pay $727,000 to avoid
situation C, which is the pollution situation; and in order to avoid
the stacks depicted in situation B, they were willing to pay $414,000."
(Page 338.) The present value of these benefits through time using a
35-year horizon and 5% discount rates is $21 million to prevent situation
C and $12 million to prevent situation B. The effect of the income distri-
bution among the sample groups upon the aggregate bid was also examined.
By increasing the income of any group by 10% and decreasing the income of
all others by 10%, no significant influence upon the aggregate bid was
determined. The effect of an income increase or decrease for all groups
was not determined.
The study notes that there were fifteen national recreation areas
within a 100-mile radius of the proposed Kaiparowitz site, so that the
aggregate benefits of control would be considerably larger than those
estimated for the Lake Powell site alone.
Evaluation Comments
The results of this study again indicate that the economic benefits
of preventing visibility degradation and other related environmental
damages are very large, on the order of $10-20 million at one of many
potentially impacted sites. The study also found a high correlation of the
results with the earlier Four Corners study. (A comparison of results is
discussed in The Farmington study and in Section 6.2.6.)
The study has several rather serious weaknesses for application
to policy decision making. First, proposed air quality levels are not
measured or tied to emission rates and therefore are difficult to relate to
costs of emission control. Further, the scenarios are designed and pic-
torially represented to compare natural visibility conditions versus
6-9
-------
the impacts of a new plant and plume; however, as 1979 conditions indicate,
a correct analysis would require that the smoke plume of any new plant
would need to be evaluated as an additional impact to those of the existing
Navajo power plant.
A second problem is that the pictorial representations of alterna-
tive conditions suffer from a lack of consistency across scenarios as to
the scene and the angle from which it is viewed. For example, scenario A
is depicted with two pictures of canyons at roughly eye level, while
scenario C shows two views of the power plant emitting smoke as seen from
the air. The effect of these variations on the benefit measures is unknown.
Direct estimates of the extra impact of visible plumes given that a
new power plant is already visible (B to C) were not obtained and can
be estimated only by the difference between the mean bids for A to C and
for A to B. Because the plant is visible in both scenarios, this value (B
to C) may be considerably different from values received where a plume is
visible but the plant is not (the plant may be behind a hill or, for most
impacted recreators, the plant may be at a considerable distance). Aggre-
gate benefits, which are carefully calculated for the Lake Powell recreators,
cannot be easily extrapolated to other nearby recreation areas, since one
plant cannot be viewed from all 15 sites within the 100-mile radius of the
Kaiparowitz site. Only benefit measures for visibility impacts without a
visible plant would be valid for extrapolation, and these figures cannot be
derived from these results.4
Finally, it nust be noted that the visual analysis of strategic
bias does not constitute a formal or defensible test, since there is no
knowledge of what the underlying bid structure would look like without
strategic bias. The existence of strategic bias might simply amplify the
variance of the bids while retaining the normal bell-shaped distribution.
6.1.3 The Farmington Study
The Farmington study, as reported in Rowe, Brookshire and d'Arge
(1980a,b) and Blank et al. (1977) was funded by the Electric Power Research
Institute to examine in depth how useful non-market experiments are for
establishing values associated with aesthetic experiences.5 The study
attempted to establish the economic value of visibility over long distances
in the Four Corners region. Additional efforts were made to examine and
test for biases in bidding methods. The Farmington study also included the
first attempt to use the household production function approach for examin-
ing how behavior changes in response to visibility changes. The results of
this latter experiment are reported in Blank et al. (1977).
^This discussion is based upon the original research report and is not
consistent with the review and analysis of this study presented in the
South Coast Air Basin Study (Brookshire et al. 1979, page 40).
^The Electric Power Research Institute is in no way responsible for the
validity of the analyses contained herein.
6-10
-------
Problem Formulation and Scenario Development
By 1976 the Four Corners region's air shed was rapidly becoming
one of the most significantly impacted in the county. As previously
noted, the Four Corners power plant had already affected visibility con-
ditions in the Four Corners area, including Farmington, New Mexico, which
is the largest community in close proximity. In 1976 the plant had 2,175
MW of power. Since the Randall et al. study, the San Juan power plant had
begun operations in the region with 660 MW of power and planned to add an
additional 1,000 MW to its capacity. In addition, four coal gasification
plants with a total of 1,750 million cubic feet of syngas capacity had
been proposed for the area (WESCO I and II and Burnham I and II). Each of
these actual and potential emission sources are shown on Figure 6-1. An
example of the visibility degradation from the Four Corners plant is shown
on Figure 1-5 of the EPA report to Congress (1979). It was estimated that
daily regional emissions of 400-500 tons per day of S02 were reducing
typical visibility from natural conditions of 160 km to 100 km and that if
emissions increased to 1,200 tons per day of S02, visibility would be
further reduced to 60 km (Kneese and Williams 1978).
This study was the first in which the bidding method approach was
used to develop scenarios specifically linked to physical measures of
visibility levels and emission rates. Three alternative scenarios were
developed with decreasing levels of visibility and a fourth with a visible
power plant. The first (A) depicted a visual range of about 120 km, or
marginally better than the current "typical" conditions. The second (B)
represented a visual range of about 80 km, and the third (C) represented a
visual range of about 40 km. Scenario D represented a combination of C
plus the visible siting of the power plant. The frequency of the alter-
native scenarios, both in the study period and in the expected future, was
unknown, although some idea of the potential expected future emissions and
resultant visual range was later taken from Kneese and Williams (1978).
Each alternative scenario was depicted with regular photography
using two standardized views from Farmington: one of Shiprock to the
west, which is a prominent landscape feature, and the second of the La
Plata Mountains to the north. The final photographs were chosen from a set
taken at two-hour intervals over a five-day period, and very nearly
represent actual visual ranges of 120, 80 and 40 km (see Rowe et al.
1980a, pages 4-5). The photographs do exhibit variation in time of day,
coloration and cloud cover, which may have influenced the values reported
by respondents (see Appendix 1).
Application of Bidding Method
As in the Four Corners study, bidding methods were selected
because local land ownership and use characteristics did not lend them-
selves to accurate property value analysis. Second, a major objective of
the study was to analyze the validity of the bidding method approach
applied by Randall et al. (1974) and Brookshire et al. (1976) in the same
impact area.
6-11
-------
Respondents were provided variations on the typical iterative
bidding procedure for proposed degradations in air quality from the
highest level (A) to the lower levels (B, C, D). After an introduction
and presentation of the photographs, both willingness to pay questions to
prevent the change (WTP^ = ES measures) and willingness to accept
compensation questions (WTA^ = CS measures) were posed to respondents.
This was the first attempt in a visibility study to test the influence of
the difference between WTP and WTA compensation approaches upon the
valuations received for visibility.
Other attributes of the questionnaire were varied across respon-
dents to examine the influence of bidding method questionaire design.
Starting bids of $1, $5, and $10 were randomly preassigned to respondents.
Electricity bill increments and payroll deduction payment vehicles were
also randomly preassigned to resident respondents. Nonresidents were
presented with a typical user fee payment vehicle. Zero and infinite bids
were evaluated, and socioeconomic characteristics, including environmental
stance, were obtained. Other compensating approaches and an interview
evaluation were incorporated but are not discussed here. A copy of the
questionnaire is included in Appendix 1.
Tests for information and strategic bias were incorporated into
the questionnaire design. First the WTP bidding procedure was structured
so that the individual presumed he would have to pay the average bid, not
his own. Therefore, if an individual's value was above the average and he
desired to increase the magnitude of the aggregate bid, he would have an
incentive to increase his stated bid so that the mean bid would more
nearly represent his own "true" valuation, yet he would not actually be
expected to pay the higher stated bid. Second, to test for information
bias and provide information almost necessary to bid strategically, ran-
domly preselected respondents were provided "supposed" mean bids of prior
respondents, which was, in fact, much lower than the true mean bids.
Those respondents who did not receive this information prior to bidding
were presented the same information after bidding and asked if they
desired to revise their bids. Finally, all respondents were asked after
their WTP bids if they would bid more if the aggregate amount was not
enough to fund the proposed environmental protection.
Pretested personal interviews were administered to over 100 Farm-
ington residents in their homes and 30 recreationists at the nearby Navajo
Reservoir.
Analysis of Results
The most noticeable result is that the WTAB (CS) bids were 4
to 14 times larger than the WTP^ (ES) bids and had a greater variability.
Also, over half of the respondents gave infinity bids or refused to bid
on the WTAB approach, rejecting the concept of selling their "right" to
the higher air quality level. The authors provide a variety of reasons
to explain the large divergence between the WTP and WTA bids. First, the
WTP values were income limited, whereas the WTA values were not. Second,
implied liability rules or property rights may influence the values
6-12
-------
differently. The WTPB approaches essentially represent a "victim pays"
concept, while the WTAB approaches place the liability for maintianing visi-
bility with the power companies and the property right to higher levels of
visibilities with the residents. This WTAB approach presupposes that the
power companies will attempt to buy off the consumer rather than cleanse
the air. (See Section 4.2 for discussion on these influences in the
bidding process.)
The next significant result was that the starting bid significantly
influenced the bids. For each $1.00 increase in the starting bids, the
mean values increased by about $.60. Also, the method of payment signifi-
cantly influenced the size of the bid, with stated bids using payroll
deductions larger than for electricity bill increments.
Information bias was present, but strategic bias was not found.
Those respondents who received false information understating the mean bid
of other respondents prior to bidding consistently bid less than those who
did not receive the information. A test for strategic bias was designed
which asserted that environmentalists who received the information would
increase their bid while conservationists would decrease their bids. After
adjusting for zero rejection bids and very obvious problem bids (bids
greater than 10 standard deviations from the mean), the test failed to
find strategic bias.
The study raised a question as to whether the bidding process
elicits the maximum WTP and/or whether respondents simply attempt to
cooperate with the enumerator. One-third of the respondents who were told
their bid would be too low to support the proposed environmental protection
revised their bid upward.
To examine the importance of the survey instrument design biases
upon a respondent's bidding behavior, bid functions were estimated repre-
senting a weighted average between the "true" bid predicted by socio-
economic variables as derived from utility functions (see Section 2.5)
and the information received via the survey instrument (starting bids,
payment vehicle, and other information). The authors report that for the
WTP bids, the estimated coefficients indicate that the reported bids are
nearly equally made up of the respondents' "true" valuation and survey
information influences. On the other hand, the survey influences were not
statistically significant for the WTA bids. In these cases the difference
between starting bids of $1 and $5, for example, was insignificant relative
to final mean bids ranging from $25 to $100.
A major element of the Farmington study was to specifically test
for comparability with the two previous studies. All three studies are
similar in general design but different in their application. The Far-
mington questionnaire was considerably longer and contained many tests on
the bidding method procedures. The scenarios vary considerably across the
studies. The Four Corners scenarios depicted unmeasured visibility con-
ditions plus unreclaimed soil banks and transmission lines. The Lake
Powell study scenarios differ from the Farmington study due to their
setting at a lake and the focus upon plumes. The time periods of study are
also several years apart.
6-13
-------
Table 6.3: Farmington Study Results (Per Month, Per Family or Unrelated
Individual)
Bid
Residents
ESABa
ESAC
ESACR
ESBC
ESADb
CSAB
CSAC
CSBC
CSAD
Sample Size
93
93
73
93
93
45
35
36
31
Mean Bid
4.75
6.54
7.58
3.53
6.85
24.47
71.44
46.63
113.68
SE of the
Mean Bid
0.386
0.576
0.766
0.462
0.592
7.54
23.14
14.14
37.69
Nonresidents ES bids (per day)c
ESAB
ESAC
ESBC
ESAD
26
26
26
26
3.00
4.06
2.53
4.56
0.77
1.05
0.65
1.11
ES-CS Bids
ESAB ES bid for situation A (120 km)a to B (80 km)
ESAC ES bid for situation A (120 km) to C (40 km)
ESACR ESAC bid revised when ESAC was stated to be too low
ESBC ES bid for situation B (80 km) to C (40 km)
ESAD ES bid for situation A to C with visible power plants
CSAB CS bid for situation A (120 km) to B (80 km)
CSAC CS bid for situation A (120 km) to C (40 km)
CSBC CS bid for situation B (80 km) to C (40 km)
CSAD CS bid for situation A to C with visible power plants
Source: Rowe et al. (1980a)
aESAB, for example, represents a bid to maintain visibility level
A rather than have visibility level B. "A" represents visibility
of 120 km, "B" visibility of 80 km, and "C" visibility of 40 km.
"Visibility picture D depicted a power plant and visibility reduction
to level C.
cNonresident bids were obtained at the Navajo Reservoir, near Farming-
ton New Mexico.
6-14
-------
Accounting for the many differences between the studies, the
empirical results were very similar. Randall et al. (1974) reported
yearly bids for residents of $85 (A to C) and $50 (B to C) per household.
The Farmington results for the most comparable situations were $82.20 and
$57.00. Considering the offsetting influences of the Four Corners results,
including environmental degradation other than visibility and power plants,
and that there was a significant inflation rate between the studies, the
results are remarkably comparable. The results of the Lake Powell Study
for recreationists are also in close agreement with the most comparable
applications of the Farmington study, which is for starting bids of $1.00
comparing best to worst. These results are reported in Table 6.6.
Aggregate Benefits Measures
Aggregate benefit measures for residents were obtained by two
approaches and had nearly equal results (Rowe et al. 1980b). The first
multiplied the mean bids of nine income groups by the projected number of
individuals in each group in San Juan County, New Mexico. The second
applied linear approximated bid curves, again aggregating across the local
population. Nonresident aggregates were determined for Navajo Reservoir
recreationists only based upon Bureau of Reclamation estimates of visita-
tion. The authors report:
When it comes to reporting aggregate benefits, our confi-
dence lies primarily with the ES measures. The CS measures
(and standard errors) are an order of maginitude greater.
In fact, if just one infinity bid is accepted, the CS bids
are immeasurably greater. Using a 10 percent discount rate
over a thirty-five year horizon, the aggregate resident and
recreationist benefit measure is $14.2 million for situation
A to B and $19.2 million for situation A to C. Statistical
confidence intervals have a spread of approximately _+ $1
million in each case. The present value of the plant siting
benefits amounts to $1.1 million. Apparently, the exact
plant siting is of considerably less importance than the
pollution effects of a plant in the general vicinity. (Rowe
et al. 1980b, page 103.)
Evaluation Comments
In this study an effort was made to develop a standardized photo-
graphic presentation of air quality conditions and to link the conditions
depicted in the photographs with actual visibility conditions in terms of
visual range. The rate of local emissions that would produce the depicted
visibility conditions were also roughly determined. In this way, the
study established links between control of emissions and benefit measures
6-15
-------
for local residents and nonresidents. The standard photographs did have
some variation in terms of lighting, time of day and coloration, but
the effect of these on the results is unknown. No efforts were made to
further define the scenario in terms of timing, frequency, location and the
like, which is now recognized as important to useful benefit analysis.
Only activity values were estimated in this study.
The Farmington study has received considerable attention because
it was the first such study to detect a great many problems and biases in
the results. This occurred in part because the survey instrument was
specifically designed to statistically test for the such biases if they
occurred. It has been suggested that these problems resulted because the
good to be valued was poorly specified (EPA 1979). This is probably not
the primary reason. Although many residents felt the degraded visibility
depicted in the photographs was due to wind-blown dust rather than to power
plant emissions, they generally realized that emissions could cause such
degradations. Also, relative to other such studies, the determination and
presentation of visibility conditions was fairly well done.
The problem appears to have been that the contingent market was not
credible or acceptable to respondents which decreased the incentives for
accuracy, making respondents more susceptible to variations in the question-
naire. It was well known that local power was supplied by the city-owned
power plant, not the two large plants operating in the region that supply
power to Southern California; consequently, many residents felt it inappro-
priate to pay to prevent pollution that results from power production for
Southern California. This is also reflected in the very high rate of
rejection or infinity bids for the WTAB questions. The difficult task of
valuing visibility was made more difficult and more susceptible to inaccura-
cies by the problems in the contingent market definition.
The problem in the contingent market was further reflected in the
WTAB questions. About one-half of all respondents rejected these situa-
tions by refusing to cooperate or giving infinity bids; many suggesting
that such a situation was unacceptable. The WTAB bids, which were the
first to be reported for such a study, were considerably larger than the
WTPB bids and the difference was larger than would have been expected from
the Willig (1976) or Randall and Stoll (1980) analyses. This again sug-
gests that great care must be taken to define contingent markets and to
account for property rights. The study also analyzed bid functions expli-
citly derived from utility functions, which related bids to socioeconomic
variables, a step which had not been performed in earlier studies.
A final problem area in this study was the survey procedure which
used a stratified random sampling from a reverse telephone directory
and door-to-door solicitation. A substantial number of potential respon-
dents who were contacted refused to participate (tally sheets were not
kept) yielding a potential self-selection bias as discussed in Section
4.2. The resultant sample size for recreationists and for residents
responding to the WTAB questions was insufficient for thorough statis-
tical analysis.
6-16
-------
6.1.4 The South Coast Air Basin Study
The South Coast Air Basin (SCAB) study as reported by Brookshire,
d'Arge, Schulze and Thayer (1979, 1980, 1981) was funded by the EPA
and operated out of the universities of Wyoming and New Mexico. The study
was specifically designed to perform and compare bidding methods and
property value approaches for measuring benefits from air pollution con-
trol in the same region. Therefore, much of the research methodology was
designed to implement both approaches and eventually to compare the result-
ing valuations.
Problem Formulation and Scenario Development
Severe pollution episodes in Los Angeles extending into surrounding
communities and deserts, otherwise known as the (SCAB), are well known
nationwide as among the worst of urban pollution problems. Numerous
studies have demonstrated the accumulation and movement of the smog in the
area. The most serious effects are reduced visibility and health-related
impacts. The pollutants of primary concern are nitrous oxides, carbon
monoxide, ozone and hydrocarbons and are primarily related to mobile
sources, space heating and unique climatologic and topographic charateris-
tics of the region.
Visibility conditions in the region range from less than 2 miles
visual range upward. A visual range of 30 miles or more would generally
be considered good. If no further pollution programs are implemented,
increased degradation in air quality can be expected as the population
increases.
Three scenarios with subsequent supporting photography were deve-
loped for the bidding method study to analyze the impacts of both increased
and decreased levels of air quality. The A scenario depicted "poor" air
quality as a visual range of about 2 miles. This situation is typical in a
substantial portion of the L.A. area. The B scenario represented "fair"
air quality, typified by a visual range around 12 miles, the predominant
condition in the area. The C scenario depicted "good" air quality as a
visual range of about 28 miles. This level is found mostly in communities
closest to the ocean. No attempt was made to relate visibility changes to
changes in emission rates from any particular source.
The alternative scenarios were depicted using regular photography
(see Appendix A) and maps depicting areas with the different levels of
air quality. The good and fair conditons were depicted with two views
from Griffith Park Observatory in Los Angeles. One view was toward down-
town, about 5 miles away. The second was toward the west showing several
buildings 4 miles away and two sets of hills in the background. The poor
scenario was depicted by similar pictures but at another location because
consistent lighting and color characteristics could not be obtained for
this air quality level at Griffith Park.
6-17
-------
Application of the Bidding Method
AH data and requirements necessary to perform both the property
value and bidding method approaches for determining benefits associated
with changes in air quality were available. The bidding method study did
have one important advantage in that perceptions and values of the inter-
related visibility aesthetics and health affects of air pollution could be
sorted out. The property value approach can only do this implicitly by
attempting to link pollution measures to calculated visibility and health
impacts, but these calculated effects may not reflect the perceptions of
local residents, which limits their usefulness.
The bidding method questionnaire was the most complex and lengthy
to date (see Appendix 1). The questionnaire took respondents through a
nine-step process to determine the individual's value of proposed changes
in air quality using both a WTPG iterative bidding procedure and a
substitution of activities and expenditures procedure. The questionnaire
attempted to break down the total benefits in terms of visibility aesthe-
tics, and acute health and chronic health values. It also used two alter-
native payment vehiclesa monthly charge and an extra payment on utility
billswhere everyone was to pay his stated bid. Alternative starting bids
($1, $10, $50) were randomly selected, as was the period of time required
to clean the air (2 or 10 years).
Several other issues were examined. Approximately one-half of
the respondents were mailed a pamphlet describing the health effects of
air pollution before they were interviewed in order to examine the in-
fluence of information upon reported values. Standard questions evalu-
ating zero bids and obtaining socioeconomic data were also included. A
great deal of other information relating to the individual's allocation of
time and money depending upon air quality levels, household and neighbor-
hood characteristics, and other social and philosophical perceptions was
gathered, but the results were reported for very few of these variables.
The application of the questionnaire was specifically designed
to minimize hypothetical biases. Due to the large variation in air
quality across the SCAB area, respondents were familiar with the alterna-
tive levels being examined. To increase the reality of the situation,
respondents were asked to bid only on proposed improvements from the
current state at their location to the higher levels. Respondents in
locations with existing air quality rated good were asked to bid on their
WTPG for improvements in air quality to a good level for the entire
region.
A total of 345 respondents were eventually interviewed in 12
different communities. Potential respondents were first contacted by
telephone to determine their willingness to participate. Unfortunately,
many respondents who initially expressed a willingness to participate
later refused, and additional interviews were set up using telephone and
house-to-house solicitations.
6-18
-------
Analysis of Results
The monthly mean bids for the total health and aesthetic benefits
related to air quality improvements are reported in Table 6.4. Communities
with existing conditions of poor air quality bid between $11 and $22
to improve local conditions to fair, with an overall average bid of
$14.54. Communities with fair air quality bid between $5.55 and $28.18 to
improve local conditions to good with an overall average bid of $20.21.
Communities with good air quality bid between $17.95 to $67.15 to improve
the whole region's conditions to good, with an overall average bid of
$34.09. On the average, for all proposed changes the aesthetic, acute
health and chronic health components constituted about 34%, 40% and 26%,
respectively, of the total mean bids. One interesting result of the bids
that was not reported but can be determined from the results is that the
mean monthly value for proposed improvements from poor to fair and from
poor to good are almost exactly identical ($0.13 difference).
As reported in Brookshire et al. (1979) numerous survey design
characteristics were found to statistically influence the reported bids.
The payment vehicle was found to be a significant influence in about
one-third of the bidding situations. The starting bid was significant in
about one-sixth of the situations. Information biases caused by the order
in which bids were requested and whether individuals first received a
health information pamphlet were also found. All of these influences were
significant enough not only to influence individual community values, but
to influence the mean values for the whole sample. The authors state that
for vehicle bias the principal problem area appears to be in the aesthetic
bids (see Brookshire et al. 1979, page 89). The proposed period of time to
clean the air also affected the reported bids.
Bid functions were estimated, from which it was possible to deter-
mine the income elasticities of stated WTP, which varied from zero to 0.67.
Most other results relating to rejection rates, protest bids, etc., were
not reported.
Aggregate Benefit Measures
Table 6.5 reports estimated aggregate annual benefits ranging from
.58 billion to .65 billion dollars for a 30% region-wide improvement in air
quality. The exact aggregation procedure is not reported. For comparison
with other studies the present value of these aggregate benefits over a
35-year horizon with a 10% discount rate ranges from 5.6 billion to 6.3
billion dollars. These estimated benefits were not tied to specific
emission control methods and were not related to costs to improve air
quality.
6-19
-------
Table 6.4: South Coast Air Basin Study Bidding Resultsl
Total $ Bid for Local Air Quality
Improvement from Poor to Fair2
Total $ Bid for Local Air Quality
Improvement from Fair to Good2
Community Mean Bid (per month)
(standard error, #OBS)
Community Mean Bid (per month)
(standard error, #OBS)
El Monte
Montebello
La Canada
Total Sample
$11.10
(13.13,20)
$11.42
(15.15,19)
$22.06
(33.24,17)
$14.54
(21.93,56)
Total bid for regional air quality
improvement to good at all Iocations3
Community Mean bid (per month)
(Standard error, #OBS)
Canoga Park
Huntington Beach
Irvine
Culver City
Encino
Newport Beach
Total Sample
$16.08
(15.45,34)
$24.34
(25,46,38)
$22.37
(19.13,27)
$28.18
(34.17,30)
$16.51
(13.38,37)
$ 5.55
( 6.83,20)
$20.31
(23.0,186)
Pacific Palisades
Palos Verdes
Redondo Beach
Total Sample
$67.15
(27.45,20)
$17.95
(3.91,19)
$20.46
(5.33,26)
$34.09
(8.78,65)
^Unadjusted total bids, including aesthetic, acute and chronic health values.
2These results are from Brookshire et al. (1980). There are several unexplained
differences and unreported values from the original work, Brookshires et al.
(1979). The standard errors appear to be reported for the individual observa-
tion.
JThese results are from Brookshire et al. (1979) and report the value for region
wide air quality improvements to good conditions bid by residents of areas with
existing conditons of good. Standard errors are for the mean bid.
6-20
-------
Table 6.5: Bidding Method Estimates of Total Benefits for Air Quality
Improvement in the South Coast Air Basin
(Approximate 30% Improvement in Ambient Air Quality)
Survey Study
Preliminary
Mean Regression
Bid Results
Average ($) bid per house- $29** $26***
hold per month
Annual benefits (selected
areas and groups of the $.65 $.58
South Coast Air Basin
in billions of $'s)
Source: Brookshire et al. 1979, page 134.
**Based on maximum total bid with an adjustment for years to achieve
improvements in air quality.
***Based on maximum total bid equation with an adjustment for the
amount of air pollution information available to the household.
Evaluation Comments
The results of this study indicate that improving air quality in
the SCAB would be very beneficial because of the large population that
would receive both health and visibility aesthetics benefits. This is the
first application of the bidding method to urban pollution problems.
Comparisons with the SCAB property value study described below suggest that
the benefit measures obtained are defensible.
An effort was made to define visibility conditions in terms of
visual range reduction and to increase the control and standardization of
supporting photography, but only a limited effort was made to link visi-
bility conditions to specific rates and sources of emissions. Consequently,
no formal benefit-cost analysis can be performed. An effort was also made
to design the survey instrument carefully to increase the credibility of
the proposed scenarios. Nevertheless, there were still substantial question-
naire design biases, particularly related to bids for visual aesthetics,
for which there was no explanation. Also, in the survey procedure, as in
the Farmington study, rejection rates were not tallied or analyzed.
6-21
-------
6.1.5 The Grand Canyon/Southwest Parks Study
This study, which is a joint effort by the U.S. EPA, U.S. NFS,
the University of Wyoming and the University of New Mexico (EPA 1981)
is the most recent and most concerted effort to link emission rates and
sources to visibility conditions and to economic benefits of control.
The study is also the first attempt to estimate existence values and to
determine values for respondents who are not physically on-site in the
impact area. In this case, the area of concern is the national parks
of the Southwest, of which the Grand Canyon is the most prominent, with
the respondents interviewed in four major metropolitan areas. Three basic
questions were addressed: These were, What is the value of controlling
haze in the Grand Canyon? What is the value of controlling haze in all
Southwest national parks? and What is the value of controlling plumes in
the Southwest Class I areas?
Problem Formulation and Scenario Development
The parklands of the Southwest are considered unique national
treasures. For many people these parks are valuable whether or not they
have visited or ever will visit them; or in other words, there is not
only a value associated with their use but also with their existence.
The enjoyment of many of these parks is particularly related to their
impressive visual aesthetics; therefore, if the ambient air quality is
degraded, so are the values of the parks.
The Grand Canyon is the largest and probably the best known of
these parks, particularly because of its impressive scenic vistas. Un-
fortunately, like many of the parks, it has experienced periodic bouts of
air pollution from man-made (anthropogenic) sources and thus has become the
subject of many controversies concerning air pollution in national parks.
In fact, many assert that the presentation by Gordon Anderson (Friends of
the Earth) to U.S. Congressional members of a series of photographs depicting
severe air pollution in the Grand Canyon was a major influencing factor in
the eventual passage of the 1977 Clean Air Amendment.
Although the national parks of the Southwest are all protected
from air pollution, they are located in an energy-rich region. In the Four
Corners states (Colorado, New Mexico, Utah, and Arizona) there are coal-
fired electric-generating plants with about 10,000 MW of power with another
13,000 MW of power proposed for the 1990's. If all of these plants are
developed without controls as to emission rates or location, severe
visibility degradation could occur. These plants are now subject to the
national goal and visibility regulations to control emissions so as to
minimize adverse visibility impacts that will affect any Class I areas.
Two questions are relevant: What rate and location of emissions will
result in adverse impacts? and What is the economic benefit of these
controls?
6-22
-------
To develop scenarios of impacts, the authors concentrated upon
SC>2 emissions in the area from coal-fired power plants. Emissions of
SC>2 are a major portion of the fine particles which are directly related
to visibility reductions. To estimate the potential impacts of the proposed
new plants on visibility, a survey was performed to determine the maximum
controlled and uncontrolled SC>2 emission rates of each existing and
proposed plant. Increases in SC>2 were then distributed uniformly across
all sites, and the effect on air quality was determined by calculating the
resultant impact measured by changes in a physical measure called "green
contrast" which can be directly linked to visual range. This uniform
distribution must be taken as a weak approximation of the actual distribu-
tion of impacts, since the location of the sources as well as graphic
features, etc., will greatly affect the level of impacts across the area.
Recognizing this, the authors state:
All three scenarios of future SC>2 emissions in the test
region should be recomputed with the use of a long range
transport model, allowing for: 1) the transport of distant
emissions into the region; 2) the chemistry of SC>2 conver-
sion to sulfate fine particulate; 3) the removal by dry and
wet deposition of pollutants affecting visibility; and 4)
the inclusion of smelter and urban emissions.
Three conceptual scenarios were developed. The first suggested no
future visibility deterioration would occur, since smelter SC>2 emission
may decrease more than S02 emissions increase from proposed coal-fired
power plants. A second scenario was developed based upon controlled emis-
sion rates for existing and proposed plants in which emissions would
increase by about 400 tons per day, which again would not significantly
impact visibility. A third scenario was based upon uncontrolled future
emission rates of 3,230 to 3,900 tons per day of S02, which would signi-
ficantly decrease air quality. The controlled and variations upon the
uncontrolled emission scenarios were used as the two basic alternatives for
comparisons in the actual bidding process.
Application of the Bidding Method
Bidding methods were selected because they are the only method
designed thus far to capture existence values (which the authors of this
study include in their definition of preservation values).
Regular photography was used to depict alternative air quality
levels. Respondents were shown five views depicting regional haze. Each
view was depicted with five air quality levelsA (worst) to E (best).
Three of the views were from the Grand Canyon, and one each was from Zion
and Mesa Verde. The photographs, selected from the EPA/NPS regional
visibility monitoring network photo file, exhibited a high degree of
consistency across air quality levels for each view. By using the
6-23
-------
EPA/NPS network, it was possible to obtain precise measurements of actual
air quality conditions at the time of the pictures, which allowed selection
of pictures closely representing the proposed alternative conditions.
Level C pictures were selected to represent "typical" conditons. Levels
A, B, D and E were selected to represent humanly perceptible increments in
air quality. The uncontrolled emissions scenario, and subsequent decreased
air quality, was most closely represented in some cases by the A level
picture and in some cases by the B level picture. This inconsistency in
the assignment of the photographic representations of the uncontrolled
scenario, coupled with the assumption of a uniform emission impact approach,
creates a problem in determining which bids (C to A or C to B) would
appropriately represent a program of uncontrolled emissions in the region
or at any specific location. Air quality levels D and E were included to
evaluate the benefits of even more stringent air quality control programs.
Sample slides for one view used for this study may be found in the back of
the guidebook.
Participants were also asked about their willingness to pay to
prevent a plume from being seen in a Class I area. Two photographs taken
at Hopi Firetower Observation Point in the Grand Canyon were used: one
with a plume and one without, with all other conditions the same. The
authors note the source was believed to be a controlled burn.
Two versions of the survey instrument were implemented: one to
obtain user values and one to obtain preservation values (user and exist-
ence values). After a brief introduction concerning visibility and after
showing the photographs of the Grand Canyon, the respondents were cate-
gorized as to whether they had visited the area in the past or expected to
visit it in the future, or whether they did not expect ever to visit the
area. About one-third of those who had visited or expected to visit the
Southwest parks were given the user value questions and about two-thirds of
the users plus those who never expected to visit the park were given the
preservation value questions.^
Individuals in each group were asked a series of bidding questions
relating to preserving air quality at the Grand Canyon, then were asked
what additional payments they would make to preserve air quality at all
other Southwest national parks as well, and finally were asked what payment
they would make to avoid plume blight. The survey concluded with typical
socioeconomic questions. Park users were asked the maximum increase in
entrance fees they were willing to pay to ensure air quality levels E, D, C
or B would be the conditions they could expect on the day they were at the
Grand Canyon, or any Southwest park (regional analysis) rather than the
°No analysis of respondent uncertainty was undertaken. These researchers
assumed zero uncertainty on the part of the respondent as to future usage
rates so as to eliminate option values. This may lead to some problems,
since respondents who state no expectation of future visitation but who
are uncertain may not be bidding on a zero rate of future visitation. On
the other hand, option value is assumed to be zero for actual users and
those who state they will use the park but may, in fact, be uncertain of
future use. Option value is therefore likely to be included in the
preservation value bids by users, and non-users.
'It seems that an appropriate benchmark would have been established by
first asking the maximum entrance fee respondents would have been willing
to pay given scenario A.
6-24
-------
lowest level A (WTP^ = CS).^ For plume blight, users were asked their
maximum entrance fee increment they were willing to pay to ensure plume
blight would not occur on the days they were at the park (WTPB = ES).
Preservation value respondents were asked the maximum they would be willing
to pay on their monthly utility bills to preserve visibility at the Grand
Canyon, then for all Southwest parks at air quality level C rather than at
level B (WTPB = ES). A similar plume blight question was also asked. To
avoid starting bid bias, respondents were asked to check the appropriate
value on a list of values (point selection) for each proposed air quality
change; however, there is no evidence that this procedure does not also
bias results by the selection and clustering of value points provided.
Finally, standard zero bid evaluation questions were incorporated into the
questionnaire. No other tests for survey design influence were incorporated.
Pretested questionnaires were applied to a stratified random
sample of residents of four metropolitan areas in the summer of 1980.
Four hundred fifty preservation questionnaires were completed: 115 in
Albuquerque, 127 in Los Angeles, 110 in Denver and 98 in Chicago. One
hundred sixty-six user value questionnaires were completed; 61 in Albuquer-
que, 60 in Los Angeles and 45 in Denver. The questionnaires were admi-
nistered through a random door-to-door solicitation (see footnote 11, page
4-27 of this guidebook).
Analysis of Results
Mean bids of all respondents in the four cities are reported
in Table 6-6. A noticeable feature of the user value bids is the apparent
increasing returns to scale from improvements in air quality. The authors
attribute this to a Dubos effect which asserts that as the environment
deteriorates, people care less and less about future deterioration, or, in
other words, people put a special value on pristine environmental conditions.
The preservation value questionnaire was undertaken to obtain
both user and existence values. Thus, by comparing the reported preserva-
tion values with user values the importance of existence values could be
derived. Unfortunately, the bids are obtained for non-comparable scenarios:
A to B, A to C, etc. in the user questionnaire, a WTPG approach; and for
C to B in the preservation questionnaire, a WTP^ approach. It is not
necessarily the case that the A to C bid minus the A to B bid should equal
a B to C bid due to non-linearities in value functions, or that a B to C
bid would equal a C to B bid due to the effects of changes in implicit
liability rules. Nevertheless, making these assumptions and the assumption
that the user values and preservation values received are both accurate
measures of benefits, the preservation values (especially the existence
values) are found to overwhelm user values. For example, from the reported
bids to protect only the Grand Canyon, we calculate the following: The
average user attended the park .283 days per year and would be willing to
pay an extra $1.08 per recreation day to have conditions C versus condi-
tions B (calculated as the A to C bid minus the A to B bid).
6-25
-------
TABLE 6-6
GRAND CANYON/SOUTHWEST PARKS STUDY RESULTS
Bid Type
USER QUESTIONNAIRE
Grand
WTPG
Canyon Protection
for A to B
A to C
A to D
A to E
Mean Bid
(user values using a per day park
based upon 166 maximum number of
1.68
2.76
3.90
5.15
S.E. 's of
the Mean*
entrance fee
observations)
.126
.137
.200
.663
Additional for Regional Protection
WTPB for C to B 4.29
Additional for Plume Avoidance
WTPB 4.10
.717
.425
PRESERVATION QUESTIONNAIRE (user value plus existence value using a
monthly utility bill increment based upon
450 maximum number of observations)
Grand Canyon Protection
WTPB for C to B 5.38
Additional for Regional Protection
WTPB for C to B 4.58
Additional for Plume Avoidance
WTPB 3.53
.785
.701
.467
Source: EPA (1981)
Calculated assuming reported S.E.'s are for the distribution of values,
not the mean value, and assuming all of the observations are complete.
6-26
-------
This amounts to an average $.305 user value per year. On the other hand,
the mean preservation bids for C to B were an extra $5.38 per month on an
individuals utility bill or $64.56 a year. The authors assert that
preservation bids of users and non-users are not significantly different.
Therefore, the implied existence value is 210 times as large as the cal-
culated user values. Using the same approach, existence values are 11
times as large to preserve the remainder of the region, and 21 times as
large to prevent plume blight in the Grand Canyon. The fact that this
ratio is so much larger for the Grand Canyon as compared to the Southwest
parks suggests the importance of its role as a unique national treasure.
The authors examine zero bids and find between 0 and 20% rejection
bids depending upon the scenario; although, how these bids are handled is
not reported. Other rejection bids are not reported. No tests for other
instrument biases were conducted. A linear bid fucntion analysis of
preservation bids finds that only age consistently influences the bids.
The estimated income elasticity, when significantly different from zero, is
very small; ranging up to only .05. In fact, the estimated bid functions
explain very little of the deviations in the reported bids across respon-
dents (R2 = .05).
Aggregrate Benefit Measures
Bids from the preservation value questionnaires are aggregrated
to total benefits for residents of six Southwestern states (California,
Colorado, Arizona, Utah, Nevada, and New Mexico; and for the entire conti-
nental United States as reported in Table 6-7.
Table 6.7: Aggregate Benefit Measures From the Grand Canyon/Southwest
Parks Study
DOLLARS (in Millions)
Annual (Present Value @ 10%, 30 yrs)
BENEFITS FOR PRESERVING VISIBILITY IN:
Southwest U.S.
The Grand Canyon $470 ($ 4,430) $3,370 ($ 31,770)
The Region (Grand Canyon, Mesa
Verde and Zion) $889 ($ 8,380) $5,760 ($ 54,300)
Avoidance of Plume Blight in
Grand Canyon $373 ($ 3,516) $2,040 ($ 19,23U)
Source: U.S. EPA (1981)
6-27
-------
Benefits are related to emissions by estimating that complete
decontrol of projected regional power plant emission of SC>2 in 1990 would
decrease visibility by the same amount or more than is depicted in the
preservation value questionnaire scenario. Thus, one can interpret the
regional potection figure of almost 9 billion dollars per year as the
benefits of power plant SC>2 controls projected to be in place in the
region by 1990.
Evaluation Comments
This study attempted to use the state-of-the-art in contingent
market methods to address two issues beyond the valuation of a unique
environmental resource, both of which have heretofore received limited
attention. These were the relative magnitude of user values and existence
values and the cumulative impact of regional energy growth upon regional
air quality. The study has also undertaken the most detailed effort to
link emissions to air quality, to photographs, and finally to economic
valuations. By doing so, the study provides results of great interest.
Specifically, preservation values of the Grand Canyon and other Southwest
parks are very large and may well exceed all emission control costs in the
region.
Limitations in the state-of-the-art suggest that the study may, how-
ever, have tried to do too much. For example, on the technical non-economic
side, the researchers were forced to use an assumption of uniform distribu-
tion of impacts across the region due to limitations in dispersion modeling.
Second, due to limited historical photograph files, the researchers were
unable to obtain photographs depicting the same rates of change in air
quality for each scenario at each site. In other words, the change in air
quality depicted between any two air quality levels, say C to B, is not
consistent across sites. These problems weaken the comparison of the
benefit estimates to cost estimates at each site or across the region, but
the discrepancies are not so serious as to eliminate ballpark comparisons
for broad policy considerations.
The bidding method questionnaire was desigend to minimize many
potential problems and, to its credit, received a very high response rate.
Further, it demonstrated that existence values are important and may over-
whelm user values in importance. This is not unexpected. Even if each
individual's existence value is small, the ratio of users to non-users
suggests that aggregate existence values may exceed the aggregate user
values. What is unexpected is that even for users the implied existence
values are 11 to 200 times as large as their user value. In other words,
the results suggests that user values are insignificant to the valuation of
the Grand Canyon and other Southwest parks, even to users. One problem
which may be contributing to this result is the lack of attention given to
the respondents' uncertainty about future use (see Footnote 6 above).
Consequently the preservation bid includes user, existence and option values
related to prospective future user benefits. This issue of uncertainty
related to future use will need greater attention in any study addressing
future use values or existence values.
There are several other areas of concern with the bidding method
results and their interpretation. First, there is a lack of consistency in
approaches between the user and preservation (user plus existence values)
6-28
-------
questions. They are different types of questions (WTPG, WTPB) over
different sets of impacts ( A to B, A to C, etc. vs. C to B). Next, are
the findings, based only upon a comparison of mean bids, neither the rate
of park use, nor distance from the site affects preservation values. A
more in-depth analysis of bids as a function of distance, socioeonomic
characteristics, use rates, etc., all at once may have found otherwise.
The authors' assertion of a Dubos effect may be the case, but is subject to
question. The rate of increase in the user value bids is not significantly
different from zero. A closer examination of the photographic character-
istics (Table 3, p. 23) shows that the air quality conditions (measured by
green contrast) on average also increase at a small but increasing rate,
further diminishing support for the Dubos effect.
The aspect of the bidding method results which seems to have the
most limited credibility is the assumption that all households across the
country would pay monthly increments in local utility to fund environmental
improvements in the Southwest. The empirical finding that most households
would be willing to pay such a substantial amount (about $160 per year) to
protect the Grand Canyon and the Southwest from haze and the Grand Canyon
from plume blight combined with the approach may lead to questions about
the accuracy of the results. Finally, extrapolating to United States
totals from results of Albuquerque, Los Angeles, Denver, and Chicago,
while ignoring the heavily populated mid-east, eastern seaboard, and the
South, is particularly dangerous.
6.1.6 A Comparison of Visibility Related Bidding Method Studies
Table 6.8 compares 19 elements of the five bidding method studies
previously described. The studies were conducted sequentially. Four of
the studies were concerned with the benefits of pollution control in the
Southwest where many Class I areas have been impacted by power plant
development, and one study was concerned with benefits of urban pollu-
tion control (SCAB). The number of interviews completed in each study was
sufficient for defensible statistical analysis; however, the sample size
for subgroups in the Lake Powell study and for recreationists in the
Farmington study were too small to do much analysis beyond reporting mean
bids. The percentage of interviews that was usable was very high in each
study, ranging from about 70% up. What is critical but infrequently
reported in the studies is the exact sampling procedures employed and the
number of individuals contacted who refused to participate. If the non-
participation ratio is high, as was the case in many studies, there may
have been self-selection problems that biased the randomness of the survey.
Future studies will need to plan and document survey efforts more carefully
to insure that the benefit analysis is defensible.
The scenario development (element 6) has evolved from general
scenarios presented in Randall et al. (1974), which typified air quality
variations from "worst" to "best" and lump several impacts together, to
specifically defined scenarios designed to separate out visibility aesthe-
tics, power-plant siting, and health-related impacts.
The length of bidding method questionnaires, measured by the
number of questions (element 9) has been continually increasing. This
represents a trend to include time and expenditure substitution data to
6-29
-------
TABLE 6.8
A SUMMARY COMPARISON OF BIDDING METHOD VISIBILITY STUDIES
i
Ul
o
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
Design Elements
Time
Place
Respondents
ducted (percent usable)
Usuable Interviews by
Respondent Group
Scenarios
Length (/ Questions)
Value Types
Question Type
Bid Scale
Payment Vehicle
Mean Recreation Income
Mean Resident Income
Income Elasticities
of Bids: a-Reaidents
b-Non-Resldents
WTB Bid Comparisons
a. Best to worst-average
yearly bid for resi-
dent households (i of
Four Corners Study
1972
Four Corners
Res Idents/Recreat Ionia ts/
Indians
1,099 (69%)
526/150/71
Emissions, Strip Mining,
Transmission Lines
A-Worot
B-Mlddle
C-Best
Nit
Activity
WTPC. WTAB
Iterative Bidding
User Fees /Electricity
Bill/Sales Tax
NR
NR
0.39 to .65
.09 to .16
$85.00
(526,4.31)
Lake Powell Study
1975
Lake Powell
Res ident a /Recreat ionls t s
104 (79*)
19/63
Air Quality and Power
Plant
A-No Plant
B-Plant, No Plume
C-Plant with Plume
9
Activity
WTPB
Iterative Bidding
User Feea
$22,000
$16,000
NR
-
Farmlngton Study
1977
Farmlngton, NM
Resldents/Recreat Ionls ts
130 (92Z)
93/26
Air Quality and Power
Plant
A-Visibllity 120 km
B-Vtsibility 80 km
C-Visibillty 40 km
D~C + visible plant
33 to 49
Activity
WTPB WTAB
Iterative Bidding
Utility Bill/Payroll
Deduction/User Fee
$11,900
$15,200
0.25 to 0.36 (ES Blda)
0.0
$82.2
(93,9.10)
South Coast Air Basin Study
1978
South Coaat Air Basin
Residents
345 (NR)
345 (NR)
Air Quality and Health
A-Poor visibility 2 miles
B-Falr visibility 12 miles
C Good visibility 28 miles
79 to 450
User
WTPG, HTAB
Iterative Bidding
Utility Bill/Monthly Pay-
ment to Conaervatlon Fund
-
$8/800 to $36,242
.00 to .67
$245.20
(48,55.3)
Grand Canyon/
Southwest Parks Study
1980
Four Urban areas
evaluating the Grand
Canyon and South-
western parks
Recreationists and
Non-Recreationlsta
NR
166 / 450
Air Quality '
A-worst to E-Best
with C - typical for
Haze, A-none, B-plume,
for plume analysis
NR
User, Presentation
WTPB
Point Selection
Electricity Bills/
User Fees
$25,300 to $30,800
$19,000 to $30,570
.00 to .05
See Text
respondents, standard
error)
-------
TABLE 6.8 (Continued)
I
OJ
15b
c
d
16.
17.
18.
Design Elements Pour Corners Study Lake Powell Study
. Best to Moderate-Average $50.00
yearly bid for residents (526,3.02)
households (tiof respon-
dents, standard error)
. Middle to worst-Average
yearly bid for residents
households (Sot respon-
dents, standard error)
. Best to worse-Average - $ 2.95
dally bill ($1976) (19, 0.2)
Recreatlonlsts with $1.00
starting bid (
-------
complement the reported bid; however, if response fatigue and inaccuracies
(as well as costs) are to be minimized, this trend should be reversed by
careful modeling, selection and definition of alternative scenarios and
survey instrument design.
Elements 8 and 9 refer to the type of values examinedactivity.
option and existence, and the type of WTP or WTA measures used. WTPB (an
ES measure) and WTPG (a CS measure) measures have become the most fre-
quently used. The two applications of the WTAB measure both incurred
over 50 percent rejection or protest bids, indicating they did not present
a realistic, credible or acceptable situation to respondents. Activity
values only were obtained in all studies except the Grand Canyon/Southwest
Parks study.
Most applications used the classic iterative bidding approach
(element 10); however, the more recent efforts have been directed toward
investigating alternative approaches. Payment vehicles (element 11)
were most always user fees for recreationists and electricity bills for
local residents. Both of these have an element of reality, since they
directly relate to the recreation experience or to the consumer who would
eventually pay for emission control.
Tied to the payment vehicle is the question of whether an individual
has the ability to gauge the impact of the payment vehicle procedure or of
his bid upon annual, or lifetime, expenditures. Does the individual recog-
nize the total impact of a small change in sales tax, or increases in
monthly utility bills and does the recognition of these impacts influence
bidding? Only one air quality bidding method study (SCAB) has provided
selected respondents with "life tables" of what a monthly bid amounts to
over a number of years. Unfortunately, no results were reported of
how presentation of these tables influenced mean bids.
In all of these studies income has been found to be statistically
correlated with reported bids. Mean income for recreationists, residents
and the calculated income elasticity are reported in elements 12-14. In
each case reported, the income elasticity was less than one for residents
and very nearly equal to zero for recreationists; therefore, large income
changes would have been necessary to significantly change mean bids and the
aggregate benefit measures.**
Element 15 reports the mean bids for up to four situations in
each study. Because each of the proposed changes in air quality and the
related impacts varied greatly across the studies and were in different
periods of time, the figures cannot be precisely or statistically compared;
therefore, best, worst and moderate rates of impacts are used for the sake
of comparison. Recalling that the SCAB results were one-third visual
°The income elasticity for the Lake Powell study (Brookshire et al. 1976)
was not reported but can be calculated indirectly from equation 5.11, page
342. This was not done here, because the equation combines values for
both residents and recreators; this overstates the income elasticity,
because residents had lower incomes but also felt less of an impact,
because many derived their income directly or indirectly from the plant
and many are acculturated to its impacts.
6-32
-------
aesthetics and two-thirds health related, while health effects were minimal
in the other studies, the results are remarkably similar for each of the
most comparable situations. Annual resident household values for the
visibility component for best to worst is in the $80 range for each study;
$60 to $80 for best to moderate; and $40 to $60 for moderate to worst. The
average daily entrance fee increment bid by recreators is also consistent
across studies. If other factors, such as inflation and impacts components,
were to be adjusted, the values would be even more comparable across
studies. This suggests that with a sufficient number of studies for a
variety of locations and impacts, and with more work on how changes in the
characteristics of the scenario affect valuations, bidding method results
will be at least weakly transferable across sites at some future date. It
should be noted that the results of the Grand Canyon study can only be
carefully compared to the others because it also estimates existance
values.
Element 16 shows if the studies formally tested for and found
design biases affecting the results. Both the Farmington study and the
South Coast Air Basin study tested for and found biases. The problem in
the Farmington study can be identified as a hypothetical bias problem which
manifested itself through increased measurement errors and biases. No
explanations are provided as to why the biases were found in the SCAB
study, particularly when the alternative scenarios are so well known to
local residents.^ Several bidding games for other goods have tested for
biases and and have not found them (Brookshire et al. 1978; Randall et al.
1979; and Thayer, 1981). While all applications of the bidding methods may
not yield these biases, all applications to visibility benefit analysis
that had formally designed and statistically implemented bias tests found
biases. In some cases the biases were not large, but the results indicate
that respondents do have trouble putting value upon visibility and that
they are influenced by the survey and questionnaire design (see Rowe et al.
1980a).
Elements 17-19 report total yearly and present value figures and
summary comments. The total value figures vary greatly across studies
because the size of the impacted population also varies substantially and
because several studies examined the impacts only upon selected subsets of
the totally impacted area (all but the Grand Canyon and SCAB studies).
Future studies will need to more carefully define and calculate aggregate
benefits to increase the defensibility of the analysis.
6.2 Hedonic Approach Case Studies
The five property value studies in this section are the Washington
D.C. study, by Nelson (1978); the Boston study, by Harrison and Rubinfeld
(1978); the Denver Study, by Bresnock (1980); The South Coast Air Basin
study, by Brookshire et al. (1979; 1980); and the San Francisco Bay Area
study, by Loehman et al. (1980). They are all based, at least in part, on
the Rosen-Freeman theoretical framework for applying the hedonic price
^Brookshire et al. (1980) reports that biases were not found in the SCAB
study (page 14) even though the original research reports clearly show
starting bid, information and payment vehicle biases were found (Brookshire
et al- 1979, pages 80-103) for subsets of the bids from which it can be
inferred that these biases for the total bids were also present.
6-33
-------
technique to residential property values to estimate the WTP for air
quality at the residential site. This approach relies on the initial work
with hedonic prices by Griliches (1971) and the fairly extensive housing
market studies represented by studies such as Kain and Quigley (1972).
It would be useful for public policy purposes to know what the
benefits of reducing urban air pollution would be. Of particular concern
is that national ambient air quality standards are being violated for some
pollutants in many urban areas around the country. The national primary
standards, listed in Table 6.9, are based on the health effects of the
various pollutants. National secondary ambient air quality standards
address protection of the public from all adverse effects of air pollutants
including aesthetic impacts. As shown in Table 6.9, secondary standards
differ for only sulfur dioxide and particulates.
Although health impacts of urban air pollution have been the primary
regulatory concern, there is evidence that urban residents are concerned
with both effects^ and researchers have recognized that property value
studies can measure them both. However, none of these studies have at-
tempted to separate aesthetic and health impacts. It may be possible to
link specific pollutant concentrations to visibility impacts, such as
visual range or discoloration, but in the urban areas that have been
studied the different available pollution measures have tended to be highly
correlated with one another. The property value approach cannot be used to
distinguish whether residents are concerned with health effects or aesthetic
effects or both, if these effects are highly correlated, and especially if
both effects are caused by the same pollutant. Hence, the pollutant chosen
for a study using the hedonic price technique is generally considered a
proxy for all air pollution impacts in that study area.
In each of these studies a hedonic price function for residential
property values has been estimated. Because the list of variables used in
this function is lengthy and varied for each study, the variables used in
each study are listed in Table 6.10 in Section 6.2.6. Some of the studies
have reported the mean marginal implicit price of air pollution from this
function. This tells how much a unit reduction in air pollution would add
to the market value of the property. When the hedonic price function is
nonlinear this price is not constant; therefore, it is usually evaluated at
its mean value for comparison purposes. In each of the studies, the
marginal implicit price of air pollution for each household is used as the
dependent variable in the WTP function from which benefit estimates of a
reduction in air pollution (or damages of an increase) can be derived.
Some of the studies have derived such benefit estimates for an expected or
potential change in air pollution.
This review emphasizes the data used, the relationship estimated,
the results obtained, and when applicable, the scenario used to develop
benefit estimates in each of the studies. The hedonic price technique
requires application of the appropriate statistical techniques, particular-
ly regression analysis, the details of which are complex and beyond
the focus of this review.
10For instance, the SCAB study (Brookshire et al. 1979 and 1980) found that
approximately one-third of the Los Angeles area residents' stated WTP
for an improvement in air quality was attributed to aesthetic effects.
6-34
-------
Table 6.9: National Ambient Air Quality Standards
Pollutant
Concentration Limit^
micrograms per parts per
cubic meter (ug/m^)c million (ppm)"
Averaging Time
carbon monoxide
ozone6
hydrocarbons
nitrogen dioxide
sulfur dioxide primary
secondary
particulate matter primary
secondary
10,000
40,000
160
160
100
80
365
1,300
75
260
150
9
35
0.12
0.24
0.05
0.03
0.14
0.05
8 hours3
1 hour3
1 hour3
3 hours3
1 year
1 year
24 hours3
3 hours3
1 year^
24 hours3
24 hours3
Source: 40 CFR, Part 50 (1980)
3 not to be exceeded more than once annually
b geometic mean
c 1 ug/m^ = .1 mg/hm^
d 1 ppm = 10 pptm = 100 pphm
eThe ozone standard was lowered from .08 to .12 in 1978.
f Primary and secondary standards are the same unless otherwise noted.
6.2.1 The Washington B.C. Study
Study Area Description and Data Selection
For the first application of the estimation procedure suggested
by Rosen (1974) and Freeman (1974) to obtain demand and supply equations
for urban air quality, Nelson (1978) studied the Washington D.C. Standard
Metropolitan Statistical Area (SMSA).
U.S. Census data from 1970 for property values and for several
house and neighborhood characteristics were used. For neighborhood access-
ibility and amenities the author used travel time to 75% of metropolitan
6-35
-------
employment, travel time to the CBD, crimes per million dollars of property,
property tax rates, public school expenditures per student, and whether the
census tract was adjacent to a scenic river. Air pollution measures used
were average monthly geometric mean in micrograms per cubic meter (yg/m-')
of total suspended particulates (TSP) and average daily arithmetic mean in
parts per million (ppm) of oxidant during summer.
Census tracts with at least 200 single-family owner-occupied units
were selected. These included 231 urban and 235 suburban tracts. Eleven
tracts that did not fit the general pattern of property values and charac-
teristics were eliminated. These had unique features that caused property
values to be unusually high: five were in Georgetown and three were near
Capitol Hill.
Hedonic Price Function Estimation
Separate hedonic price functions were estimated for TSP and
oxidant. Both TSP and oxidant were found to have significant negative
influence on property values, though the TSP coefficient was somewhat
stronger. All the variables except school quality and crime rates were
found to be statistically significant in influencing property values.
Travel time to employment was found to be a preferable measure of access-
ibility than travel time to the CBD.
Linear, semi-log, inverse semi-log and log-linear functional forms
for the hedonic price function were tested. Though results did not vary
greatly, the log-linear form was chosen as the best fit. Because the
second derivative of the log-linear form with respect to air pollution will
always be positive, this functional form constrains the marginal implicit
price for air pollution to be an increasing function of air pollution.
This means that the marginal implicit price becomes a smaller negative
value as air pollution increases. (See Section 4.3.2.)
The final hedonic price function explained 88% of the variation in
property values. When divided into urban and suburban tracts, significantly
different hedonic price functions were not found, and hence there was no
evidence of market segmentation along these lines. The results indicated
that the mean marginal implicit price of TSP was $69 per pg/m^ and of
oxidant was $15 per .001 ppm, which means that a one yg/m^ reduction in
TSP was associated with an average property value increase of $69 and a
.001 ppm reduction in oxidant was associated with an average increase of
$15 in property values.
WTP Function Estimation
This study is unique in that both willingness to pay for air
quality and supply of air quality functions were estimated. The author
hypothesized that WTP is a function of air quality, median family income,
6-36
-------
and median persons per unit, and that supply is a function of air quality,
population density, and travel time to 75% of metropolitan employment. The
air pollution measure was transformed to an air quality measure by taking
the reciprocal of the TSP variable. This had the effect of making WTP and
supply functions of a good (air quality) rather than a bad (air pollution).
The expected inverse relationship between the marginal implicit price and
the level of air quality was obtained, indicating that willingness to pay
for air quality decreases as air quality increases. The results indicate
that the supply is somewhat inelastic (unresponsive) to the marginal
implicit price of air quality. The other four studies discussed here have
assumed that it is perfectly inelastic.
Benefit Estimates
Benefit figures for a change in air pollution in the Washington
D.C. area were not estimated. However, from his WTP function it would be
possible to derive benefit estimates for a reduction in TSP, and possibly
other pollutants, expected to be achieved by a specific program, which then
could be compared with the costs of the program. Whether the TSP variable
was correlated with other pollution measures would have to be considered to
determine whether the estimated function represented willingness to pay for
a reduction in some or all of the air pollutants then found in Washington
D.C. or for a reduction in TSP only.
Evaluation Comments
The choice of the log-linear functional form for the hedonic price
function was justified by the assertion that the slope of the marginal
implicit price function is not constrained by this form because it depends
on the ratio of property value (PV) to air pollution (AP). When PV = aAPb
it is true that the marginal implicit price function is given by b'(PV/AP).
However, because there is an inverse relationship between PV and AP, as AP
increases PV/AP will always decrease. Also, because of the inverse rela-
tionship between PV and AP, the parameter, b, will be negative. A negative
constant times a decreasing number results in an increasingly small nega-
tive number, hence the marginal implicit price function will always increase
as air pollution increases. The same conclusion is obtained by taking the
derivative of the marginal implicit price function with respect to air
pollution, which is the second derivative of the hedonic price function.
For all PV > 0, AP > 0, and b < 0 as is expected, the second derivative
will always~~be positive. This conclusion would not hold for b > 0. Hence,
the choice of the log-linear form has restricted the slope of the marginal
implicit price function for air pollution to being positive.
To estimate willingness to pay for air quality, the air pollution
variable was transformed into a measure of air quality by taking the
reciprocal of the particulate measure, which imposes a relationship between
air quality and air pollution that may not be correct. We can expect an
inverse relationship between air quality and air pollution, but the exact
form of this relationship is unknown. If such a transformation was appro-
priate, it should have been used for estimating the hedonic price function
as well-
6-37
-------
The specification of the supply function for air quality at the
residence used in this study has been the subject of controversy. Freeman
(1979a) suggests that it would be more appropriate to concentrate on
changes in housing density rather than on the tradeoff between residential
and other land uses that Nelson emphasizes in his discussion. Some housing
market studies have estimated supply functions for housing attributes.
Witte, Sumka, and Erekson (1979) estimate the supply of three housing
attributesdwelling quality, dwelling size, and lot sizeas functions of
the quantity of the attributes and supplier characteristics that are
expected to influence the supplier's efficiency and experience. More work
needs to be done to verify the factors that appear to influence supply of
housing attributes and to identify factors that may not have as yet been
considered.
6.2.2 The Boston Study
Study Area Decription and Data Selection
Harrison and Rubinfeld (1978) studied the Boston SMSA. The
air pollution measures used were annual average parts per hundred million
(pphm) nitrogen oxides (NOX) and annual average milligrams per hundred
cubic meter (mg/hm^) TSP- More than half of both pollutants were
produced by stationary sources which tended to be in the central city.
The means of these measures were 5.55 and 6.31, respectively. These were
close to the national primary ambient air quality standards, indicating that
violations were occurring in some part of the study area. NOX approxi-
mately ranged from 3 to 9 pphm across the area. NOX and TSP were highly
correlated with one another. These measures were generated by an air
pollution dispersion model for zones that were larger than census tracts
but that covered the entire metropolitan area, so that pollution estimates
were available for tracts that were not located near monitoring stations.
They selected 1970 U.S. Census data for property values and for
several house and neighborhood characteristics. To represent neighborhood
accessibility and amenities, they used access to radial highways, distance
to employment, crime rates, property taxes, student/teacher ratios, whether
the tract was adjacent to the Charles River, percentage of non-retail
business acreage, and percentage of land zoned for large lots.
Census tracts with no housing or that were dominated by an insti-
tution were eliminated from the sample, leaving 506 tracts. Each town
in the area was made up of several tracts, and some data, such as property
tax rates and land use, were for the entire town rather than for the
individual tract.
6-38
-------
Hedonic Price Function Estimation
Two pollution measures were used in separate hedonic price
functions due to their high correlation with one another. Both pollutants
were found to have a significant negative influence on property values.
The coefficients of the non-pollution variables were virtually the same in
the two equations, which the authors cited as evidence that households are
averse to pollution in general rather than to any particular pollutant.
Both of these equations explained a little more than 80% of the variation
in property values.
The semi-log functional form was chosen as slightly superior to
the linear form. Possible nonlinearities in the pollution variables
were explored, and NOX^ and TSP^ were chosen as the specifications of
the pollution variables with the best fit. This estimated functional form
implied that at the mean value of NOX and the other variables, the mar-
ginal implicit price of air pollution became an increasingly large negative
value as NOX increased; however, the estimated TSP equation implied that
at the mean value of TSP and the other variables the marginal implicit
price of air pollution became an increasingly small negative value as TSP
increased. As discussed in Section 4.3.2, either a positive or a negative
slope may occur, but obtaining contradictory results in the same study area
for pollutants that are highly correlated is difficult to explain.
Unlike the results of the Washington D.C. study, the age of the
house was found to be insignificant, and the crime and school variables
were found to be significant. The land use variables were not significant.
Harrison and Rubinfeld reported that at the mean value of NOX and the
other variables, the marginal implicit price was $1,613 for a one pphm
reduction in NOX.
WTP Function Estimation
Willingness to pay for a reduction in NOX was estimated as a
function of NOX and median family income. The authors assumed that the
supply of residential site air pollution was perfectly inelastic with
respect to price, which allowed them to estimate only a WTP function, as
discussed in Section 4.3.2. Tests of this assumption indicated that it may
not have been entirely valid, but they argued that alternative procedures
did not greatly change their final benefit estimates.
Linear and log-linear forms of the WTP functions were estimated
with the log-linear giving the best fit. The pollution coefficient was
allowed to vary across three income groups: less than $9,500, between
$9,500 and $13,000, and over $13,000 median annual household income. A
positive relationship was found between income and WTP. For all groups,
willingness to pay for a reduction in NOX was an increasing function of
air pollution. This is the expected direction of the relationship because
6-39
-------
an increasing willingness to pay for a reduction in NOX (a positive price)
is equivalent to a decreasing willingness to pay for NOX (a negative
price). Harrison and Rubinfeld, as well as the subsequent authors pre-
sented here, used the absolute value of the marginal implicit price as
the dependent variable in the WTP function rather than the actual negative
marginal implicit price.
Benefits Estimates
The Clean Air Act mandated that by 1978 new cars were to be pro-
ducing 90% less emissions of NOX, carbon monoxide, and hydrocarbons than
in 1970. Since the average life of an automobile is ten years, most of the
cars on the road in 1990 will have been produced after 1978. Harrison and
Rubinfeld used the NOX concentrations predicted by the air pollution
dispersion model based on expected 1990 automobile emission characteristics
to estimate average household benefits of attaining the federal emissions
standards. These are given by the formula:
n f N0x70i
AB = E (HHi) \ Wi dNOxi
n
I
where:
AB = average annual benefits per household
HH^ = number of households in tract i
r = the discount rate
N0x70i = 1970 NOX level in tract i
N0x90i = 199° N0x level in tract i
The integral is the area under the WTP curve for tract i between 1970 NOX
levels and 1990 NOX levels. When the single WTP function that was
estimated for all households and a 10% discount rate were used, this figure
was $83 per year. When the three WTP functions estimated for the different
income groups were used, the average benefits were $93 per year for low-
income households and $71 per year for high-income households, because
low-income households tended to live in more polluted areas.
The authors point out that to the extent that the WTP equations for
an NOX reduction reflect damages associated with non-automobile pollution,
other than NOX, the benefit estimate for auto emission controls may be over-
stated even though the calculations assumed no reduction from non-automobile
6-40
-------
sources of NOX. On the other hand, other automobile emissions, such as car-
bon monoxide, and impacts that may not be reflected in housing prices may
not be fully reflected in the NOX equation, and therefore the true bene-
fits would tend to be understated. The authors did not venture to guess
which of these effects would predominate.
Evaluation Comments
The mean marginal implicit price for NOX and the benefit estimates
derived from the WTP functions seem to be quite feasible, but there appears
to be a problem with the hedonic price function using TSP. The marginal
implicit price function for TSP is given by the first derivative of the
hedonic price function with respect to TSP, which evaluated at the mean
values of the TSP and the property value variables results in a marginal
implicit price of over $1,000,000 for a one mg/hm^ reduction in TSP.
This is an unrealistic result for which there is no obvious explanation.
The authors report that there is evidence of heteroscedasticity in
their WTP function. This might be caused by omitted variables. They
report that a few housing attributes in addition to air pollution were
statistically significant in explaining willingness to pay for a reduction
in air pollution. They argue that the benefit estimates were not greatly
changed by the inclusion of other housing attributes and therefore used the
WTP function with only NOX and income to derive benefit estimates.
However, under different circumstances the influence of other attributes
may have a significant effect on benefit estimates and the possibility of
complement and/or substitute relationships between air pollution and other
housing attributes should be more fully explored.
The use of an air pollution dispersion model to obtain data on air
pollution concentrations in each neighborhood increases the researcher's
flexibility in choosing the neighborhoods or tracts to be used in the
study, because actual monitor station data is usually quite limited.
However, the accuracy of the benefit estimates then depends on the accuracy
of the dispersion model. Progress is being made in the development of such
models, but it may be premature to place much confidence in currently
available air pollution dispersion models for urban areas.
6.2.3 The Denver Study
Study Area Description and Data Selection
Bresnock (1980) performed a hedonic price study of property values
in Denver County, which is the city center of the Denver SMSA. Denver has
a serious air pollution problem, fondly called "the brown cloud," with
carbon monoxide (CO), ozone, and TSP frequently exceeding national ambient
6-41
-------
air quality standards. Approximately 80% of the total air pollutants and
over 90% of the CO are believed to come from mobile sources. This study
concentrated on CO, which, during the second worst day in 1978, had an
8-hour average concentration mean of 121 part per ten million (pptm) and
ranged from 18 pptm to 239 pptm across the study area.
Census tract level 1977 data from the Denver Planning Office, which
included average sale prices of owner-occupied homes, and census tract
averages of several house structure variables were used. Amenity and
accessibility variables used were percentage of land used for transporta-
tion, business, and utilities; number of public school children per acre;
crime rates on persons and on property; distance to CBD; and access to
major streets and highways. Tracts with less than 20% owner occupancy or
predominantly nonresidential, open space, or high residential density
composition were eliminated, leaving 109 census tracts for the study sample.
The CO data were obtained from a pollution dispersion model deve-
loped by Systems Applications, Inc. (SAI) that predicted CO concentra-
tions for each square mile based on monitor station readings and wind
patterns on the second worst day in 1978. These grids were then matched to
the census tracts in the study sample.
Hedonic Price Function Estimation
Linear and semi-log functional forms were tested, and though
the differences were inconclusive, the semi-log form was selected because
a nonlinear form was considered theoretically superior. Different ex-
ponents for the pollution variable were examined with a value of 2 chosen
as providing the best fit. For the values of the estimated coefficients
and the range of the pollution variable in this study, this functional form
implied that the marginal implicit price of air pollution became an in-
creasingly large negative value as air pollution increased. All the
coefficients were significant except those for average square feet of lot
size and school children per acre. The final hedonic price function
explained 90% of the variation in average property sale prices.
The hedonic price function was tested for market segmentation by
separating the sample tracts into a greater than 40% minority and average
income less than $12,000 group and a less than 40% minority and average
income more than $12,000 group. These two hedonic price functions were
found to be significantly different.
At the mean, a one pptm reduction in CO was associated with a $67
increase in property values; however, for the two separate hedonic func-
tions the mean marginal implicit prices for a one pptm reduction of
CO were $83 for high-income areas and $79 for low-income areas.
6-42
-------
Willingness to Pay Function Estimation
Following the same procedure as Harrison and Rubinfeld, Bresnock
assumed supply of residential site air pollution was fixed at all locations,
Willingness to pay for a reduction in CO was hypothesized to be a function
of CO levels, median family income, number of people over age 65 per acre,
percentage of housing financed via conventional loans, autos per capita,
percentage change in family income 1970-1977, average number of persons per
unit, and average appreciation rate of property value for 1976-1977.
This was a more extensive specification of the WTP function than
that used in the other studies, and all the variables were found to signi-
ficantly influence WTP except number of the elderly and change in income
variables. A log-linear functional form was chosen as a better fit than
the linear form. This function explained 83% of the variation in willing-
ness to pay, whereas the Boston study's equation explained only 64%.
Benefit Estimates
The Colorado State Implementation Plan (SIP) is hoping to achieve
national ambient air quality standards by 1982. Predictions cf 1982 CO
levels for the same weather conditions as the second worst day in 1978, but
with the emission reductions to be achieved by the SIP, were generated by
the SAI Denver model. Average annual household benefits from the single
WTP function were $258 for this reduction in CO, using a 10% discount rate
and assuming a perpetual stream of benefits (see Section 4.3.2). When
estimated separately for the two segments, this became $230 for the high-
income group and $374 for the low-income group. This was because the
low-income group tended to be living in areas of higher pollution, which
outweighed the fact that the high-income group was willing tc pay more for
a given reduction in air pollution.
Evaluation Comments
Probably the most important question concerning the Denver study is
the use of CO as the air pollution measure. Carbon monoxide is a colorless
and odorless pollutant and although the CO pollution problem in Denver is
well publicized it is unlikely that residents are aware of the day-to-day
carbon monoxide levels in their particular neighborhoods. The property
value approach relies not on awareness of the overall pollution levels in
the area but on awareness of neighborhood-to-neighborhood variations in
pollution. Pollution that is perceived by residents in the Denver Area is
actually a mixture of several individual pollutants produced from several
different sources and from chemical reactions in the atmosphere. The
author justifies the use of CO as a proxy for air pollution in Denver
because it was highly correlated with the coefficient of haze, which is a
6-43
-------
measure of what people see but which had not been projected for the 1982
plans. CO was also the pollutant that had been most successfuly modeled in
the air pollution dispersion models used for this study. Using CO as a
proxy for air pollution was therefore a reasonable choice, but it would be
interesting to see results of the same procedure using a measure of a
pollutant that is more readily noticed by the average person.
There is currently some uncertainty about the accuracy of the air
pollution dispersion models such as the one that was used to obtain data on
census tract CO levels for this study. These techniques are still being
developed, and though improving, are not be entirely reliable for urban
pollution dispersion modeling.
Additionally, the author used data for CO levels on one particular
day. Most other studies have used an average measure of pollution for the
study period, because households are expected to be more concerned with
their average exposure to harmful or unpleasant pollution than with their
exposure on a particular day. A measure of pollution on a particular day
is only appropriate for a hedonic price function for property value if it
represents a typical or average pattern of pollution dispersion across the
study area.11
The specification of the WTP function was more extensive than that
of previous studies, and the author reported than there was no evidence of
heteroscedasticity. This seems to be an improvement over Harrison and
Rubinfeld's WTP equation, although only additional household characteris-
tics and no additional property attributes were included. Some instability
in the estimated function, such as the coefficient for autos per capita
switching from positive and significant in the linear willingness to pay
function to negative and significant in the log-linear form, indicates that
more possibilities in the specification of the willingness to pay function
need to be explored.
To test for market segmentation the sample was divided into two
subgroupshigh-income, white and low-income, minorityand a separate
hedonic price function was estimted for each. That the coefficients in
each of these separate functinos were significantly different from one
another was taken as evidence of market segmentation, but the same result
could have resulted from an incorrectly specified hedonic price function.
As discussed in Section 4.3.2, if the correct functional form can be
specified there would be no need to estimate separate hedonic price func-
tions for each subgroup unless the structure of supply and demand were
different in each subgroup. If they were different, this would imply the
need for separate and unique functional specifications of both funcational
form and explanatory variables for each subgroup.
author has later explained that the geographical variation in
CO on the second maximum day closely matched the pattern of annual
average TSP across the study area, indicating that the measure of CO on
the second maximum day was an appropriate proxy for average pattern of
air pollution for this study. (Bresnock 1981)
6-44
-------
6.2.4 The South Coast Air Basin Study
Brookshire, Thayer, Schulze, and d'Arge (1979, 1980) used the
South Coast Air Basin (SCAB) of Southern California to obtain and compare
the results of a bidding method survey and property value study for the
same area.
Study Area Description and Data Selection
Southern California has a well defined and well publicized air
pollution problem. Both mobile and stationary sources are responsible, and
national primary ambient air quality standards are frequently exceeded.
Photochemical smog, largely ozone, and its precursors, NOX and hydro-
carbons, are the predominant pollutants. The SCAB study used both
NC-2 and TSP measures. Annual averages of these pollutants for 1975
ranged from 6.2 to 13.0 pphm and 67 to 130 pg/nH, respectively, across
the study area. That California has led the nation in automobile emission
standards was considered evidence that residents are aware of the air
pollution damages and are willing to incur some costs to reduce the problem.
Household level data on property sales prices for owner-occupied
single-family residences and other characteristics of the property were
obtained from Market Data Center in Los Angeles for homes sold between
January 1977 and March 1978. Data on property taxes, public expenditures,
crime rates, school quality, weighted distances to the major employment
centers, and the distance to the beach were used to represent the access-
ibility and amenities of each community. Community boundaries did not
always coincide with census tract boundaries, with communities sometimes
covering part of one or two tracts. Census data were adjusted to match
community boundaries.
Pollution measures for N02 and TSP were obtained from the
California Air Resources Board. The 1975 average of the daily hourly
average maximum for each pollutant was used. These were interpolated from
data obtained from monitoring stations so that the locations of the stations
influenced the selection of the community study sample.
Seven community pairs were selected to minimize differences other
than air quality. Communities were rated as having poor, fair, or good air
quality, with approximately a 25%-30% difference in average pollution
concentrations between each category. They were the first to use this
matched pairs technique, as described in Section 4.3.3, to remove some of
the non air quality differences between properties. The study of a small
6-45
-------
number of paired communities or tracts was possible because the household
level data provided a large number of observations within each community or
tract. It is not possible to be quite so selective when using census tract
averages. All 719 sales of owner-occupied single-family residences in each
of the 14 selected communities during January 1977 to March 1978 made up
the final study sample.
Before a hedonic price function was estimated the authors examined
the property value differences between the matched communities. If the
communities were perfectly identical except for air pollution, the property
value differences would reveal approximately the same information as a
linear hedonic price function; that is, how much households are paying to
avoid each additional unit of air pollution. However, as expected, the
linear and nonlinear hedonic price functions that were subsequently
estimated provided considerably smaller estimates of the marginal implicit
price of air pollution, indicating that the communities were not perfectly
matched.
Hedonic Price Function Estimation
A hedonic price function was estimated for each pollution variable.
Both had the expected negative coefficients and were statistically signi-
ficant. Both equations explain 88% of the variation in property prices.
Linear and semi-log exponential functional forms were tested with the
latter selected as the better fit. The estimated exponent for the N02
variable and for the TSP variable in separate hedonic price functions
was 2. For the range of the pollution variables in this study both of the
functions imply that the marginal implicit price of air pollution became an
increasingly large negative value as air pollution increased. Each one
pphm reduction in NC>2 was associated with an increase in property values
of $2,010 at the mean value of N0£ and the other variables. The TSP
equation implied a marginal implicit price of approximately $300 per
IJg/m3 at the mean value of TSP and the other variables.^
Because the community sample used for this property value study
did not match the survey sample that was eventually used, this hedonic
price function was later re-estimated for six of the seven community
pairs. The results of both estimations are reported here because the
authors did not estimate a WTP function the second time. They decided that
they could not accept the assumptions implicit in this methodology for
estimating WTP in their comparison of the results of the property value and
bidding method approaches.
12The price for TSP was not reported by the authors but was roughly
estimated using the reported coefficient and assuming a mean property
value of $100,000 and a mean TSP level of 100 ug/m3.
6-46
-------
The most important difference in results was that the best fitting
functional form for the second hedonic price function was that in which
the pollution variables were entered in logarithmic form rather than
squared. This implies that the marginal implicit price for air pollution
became an increasingly small negative as air pollution increased, which was
not consistent with the first hedonic price function. From this second
hedonic price function, the mean marginal implicit price for a change from
poor to fair air quality was $46 per month, and the mean marginal implicit
price for a change from fair to good air quality was $59 per month. These
are roughly equivalent to saying that a 1 pphm reduction in NC>2 will
increase property values by $1,840 and $2,360, respectively, and a one
Mg/nH reduction in TSP will increase property values by $184 and $236,
respectively.^ These are reasonably close to the marginal implicit
prices obtained by the first hedonic price funtion.
WTP Pay Function Estimation
Willingness to pay for a reduction in air pollution was estimated
for N02 and for TSP from the first hedonic price function. Supply of air
pollution was assumed to be fixed at each location, and community level
data were used because household level income data were not available.
Willingness to pay for a reduction in air pollution was hypothesized to be
a function of N02 or TSP levels and community average household incomes.
As expected, WTP was found to increase as pollution increased and as
incomes increased. Similar results were obtained for both linear and
log-linear forms and for both pollutants.
Benefit Estimates
From the log-linear WTP function, average household benefits for
an improvement from poor to fair or from fair to good air quality, both
approximately 25%-30% reductions in air pollution, were estimated at $44
per month from the N02 equation and $49 per month from the TSP equation
using a discount rate of about 10%. This means that the average annual
benefits of reducing air pollution by 25%-30% would be between $528 and
$588 per household.
Evaluation Comments
An important question regarding the results of the SCAB study
concerns the difference in best fitting functional form obtained after the
sample was reduced to twelve communities. The community pair, Whittier and
Orange, was dropped because bidding method surveys were not conducted in
those communities. This reduced the number of properties from 719 to 634,
capitalized prices were computed from the reported monthly prices
assuming a 10% discount rate and a 3 pphm N02 or a 30 ng/m3
change from poor to fair and from fair to good air quality.
6-47
-------
which is still a more than adequate number for which to estimate a hedonic
price function. The authors report that with the same functional form,
semi-log exponential using the log of property values and air pollution
squared, the air pollution coefficient was no longer statistically signi-
ficant. The air pollution coefficient became significant when the air
pollution variable was entered in log form rather than squared.
Such instability caused by a fairly minor change in the sample is
disconcerting. Judging from the average property values and characteristics
levels reported for Whittier and Orange these communities fell between the
upper and lower extremes of these variables in the sample. The difference
between the two functional forms is significant in that the first implied a
negatively sloping marginal implicit price function for the range of the
variables in the sample whereas the second implied a positively sloping
marginal implicit price function. In both cases the authors report that
the results are not very much different than those obtained with a linear
form, but they offer no explanation of the apparent instability of the
nonlinear functional form. It might be reasonable to obtain such vari-
ability across different urban areas, but not across slight modifications
of the sample within one urban area.
It is interesting to note that the property value differentials
obtained from the second hedonic price function, $46 and $59 per month for
a 25%-30% reduction in N02, only slightly exceed the average household
benefits derived from the WTP function estimated from the first hedonic
price function ($42 to $44 per month) for the same reduction in NC^.
These both significantly exceed the average stated willingness to pay
obtained from the bidding method survey, $15 and $20 per month for the same
reduction in air pollution of 25%-30% in fair and poor areas. Which, if
any, of these estimates is closest to the true value that households place
on such an improvement in air quality is not known.
This study has made an important contribution with the development
of the matched pairs technique for selecting property values- Although
alone it is of limited usefulness, as discussed in Section 4.3.3, in
conjunction with the estimation of a hedonic price function it may help to
eliminate some of the non-quantifiable differences between properties. As
previously mentioned, this selection process was possible because of the
availability of household level data.
6.2.5 The San Francisco Bay Area Study
Study Area Description and Data Selection
Loehman, Boldt, and Chaikin (1980) studied the San Francisco Bay
area (SFBA) in a replication of the SCAB study. Air pollution is less
severe in the San Francisco area than in Southern California, although
photochemical oxidants are the primary pollutants in both areas. The
results indicate that benefit estimates are not directly transferable, but
this question needs further study.
6-48
-------
Household level data were obtained from the Market Data Center for
owner-occupied single-family properties sold during 1978. Household,
census tract, and city level data sets were developed so that the effects
of different levels of aggregation could be examined.
Data for NC>2, TSP, ozone and the Pollution Standards Index
(PSI) were used. A unique ozone index was developed by multiplying the
number of days on which the national primary standard was exceeded times
the average daily maximum for July through September for 1979 and 1978.
This variable included information about ozone concentration maximums and
frequency of occurrence. Also, the PSI was adjusted by multiplying the
1977-78 average PSI value times the number of days that were not rated
good.
Tracts with no property sales or with bad or missing data were
eliminated from the sample. Also, tracts dominated by unincorporated
areas, institutional facilities, very low or very high residential density,
and low owner occupancy were eliminated. The remaining tracts represented
six tract types which were subject to stratified sampling proportional to
population in each group. Forty-two tracts were thus chosen, and the
household sample consisted of all houses sold in these tracts. The house-
hold sample included 2,570 observations.
Hedonic Price Function Estimation
The first hedonic price function estimated was identical to the
one estimated for the SCAB study, except that initially, the major amenity
variable was changed from the distance to the beach to elevation, because
view was considered more important than the accessibility to the beach in
the San Francisco area. The results of the otherwise similar hedonic price
function indicated that it explained about 10% less of the variation in
property prices than it had explained in the SCAB study. Also, the N0£
coefficient was significantly positive, but this was attributed to the fact
that N02 is not a problem pollutant in the SFBA. TSP and the ozone
measures had significantly negative coefficients, as expected, yet PSI was
insignificant.
Concluding that the hedonic price function specification has to
be adapted to each study area, the researchers specified a new function.
A great deal of data were available, and due to the typically high correla-
tions between variables, factor analysis was applied to group the variables
into factors within which the correlation was maximized and between which
it was minimized. Then a few variables from each of the groups were chosen
as proxies for each group. The new specification explained 87% of the
variation in property prices.
The logarithmic form of property prices was used as the dependent
variable. The pollution variable and other variables expected to have a
negative influence on property prices were squared while the others were
entered in logarithmic form. The authors explain that this allowed the WTP
functions for the "bad" and "good" attributes to have the expected slopes.
6-49
-------
The East Bay and West Bay areas were found to produce signifi-
cantly different hedonic price functions. These two areas are separated by
the San Francisco Bay, and over 80% of the work trips originating in each
area are made to destinations within the same area. The ozone index and
the adjusted PSI variable were selected as the air pollution proxies for
the final hedonic price functions. Different results were obtained when
the tract and city level data were used, indicating that the level of
aggregation will make a difference.
Willingness to Pay Function Estimation
Household level data on incomes and age of buyer were obtained from
savings and loan data, and to ensure consistency, the hedonic price func-
tion was re-estimated with the savings and loan property price and attri-
butes data. Willingness to pay for a reduction in air pollution was then
estimated as a function of ozone, household income, and age of buyer.
A log-linear functional form was used for the WTP function. For comparison,
WTP was also estimated with city level data, because none of the other
studies have used household level data. In both cases, willingness to pay
for a reduction in air pollution was found to increase with pollution
levels and with income, as expected.
Previous studies have used the actual property value to derive the
marginal implicit price of air pollution that is taken as an observation of
the household's WTP, but this study used the property value predicted by
the estimated hedonic price function given each property's attributes.
For example, with a semi-log exponetial hedonic price function:
log PVi = a + bAPi2
where:
PVi = actual property value of residence i
APi = actual air pollution at residence i
The marginal implicit price function is given by:
dPVi
dApi = 2b-APi-PVi
Most authors have used the estimated value of the coefficient, b, and the
actual air pollution and property value at residence i to compute the
marginal implicit price of air pollution at residence i. However, the
6-50
-------
authors of this study chose to compute the predicted property value at
location i, PVi, from the estimated hedonic price function given b and the
actual level of AQi. The PVi and b were used along with AQi to compute the
marginal implicit price of air pollution at residence i.
Benefit Estimates
The preliminary results of this study indicated that annual
benefit estimates derived from the household level WTP function for a 30%
decrease in the ozone index ranged from $.13 to $436.37 per household, the
higher figures in the more polluted areas. From the city level data,
average annual benefits of a 30% decrease in the ozone index were found to
range from $.12 to $337.10 per household with an overall average of $82 per
household. This average benefit figure for a 30% reduction in pollution is
smaller than that obtained by the SCAB study, but pollution levels are
higher in Southern California, and the benefit estimates are based on a
different pollution measure.
Evaluation Comments 14
This study provides the most thorough examination of potentially
significant property attributes of any of the studies discussed here. This
is important because air pollution is a relatively unimportant attribute in
determining property values. Therefore the more carefully all significant
attributes are identified the more confidence can be placed on the influence
attributed to air pollution.
One problem with the explanation of the hedonic price function
specification for this study is that vacancy rates, percentage of new
houses, and percentage of land available for development were added as
supply side variables. These variables could possibly be interpreted as
neighborhood characteristics, but the inclusion of variables other than
attributes of the good in the hedonic price function contradicts the theory
of hedonic prices as presented by Griliches (1971) and Rosen (1974). The
hedonic price function is supposed to separate the price of the property
into the amounts attributable to each attribute of the property. Supplier
and household characteristics enter the suppliers' offer and households'
bid curves. The interaction of these functions results in the hedonic
prices that are observed in the market, but as the hedonic price theory is
currently applied, the supplier and household characteristics do not
directly enter the hedonic price function. (See Sections 4.3.1 and 4.3.2.)
The functional form for the hedonic price function was selected
in order to obtain the expected negatively sloped WTP for "good" attri-
butes and positively sloped WTP for a reduction in "bad" attributes.
This reasoning makes sense, but it is not the appropriate justification
review is based on a preliminary report of the study results.
6-51
-------
for the selection of functional form. The appropriate functional form for
the hedonic price function is one that does not constrain the slope of the
marginal implicit price with respect to the quantity of an attribute, not
one that constrains this slope to be either positive or negative. (See
Section 4.3.2.) In fact, the functional form that was selected does
allow the marginal implicit price functions for both bad attributes
and good attributes to have either positive or negative slopes as would not
be the case if all attributes were entered in log form or if all attributes
were entered squared. However, there are other more general functional
forms, such as Box-Cox, that would also allow such flexibility.
The authors argue that it is preferable to use the estimated
property value rather than the actual property value to compute the mar-
ginal implicit price of air pollution at each residence due to the simul-
taneity of the determination of the hedonic price function and the marginal
implicit price function. However, the marginal implicit price function is
mathematically derived from and entirely determined by the hedonic price
function. They are not simultaneous equations in the usual sense. It may
also be that statistical bias results from the use of the estimated
value of PVi in this functional form.15 Their procedure does not appear
to be an improvement over the use of actual property values to compute
marginal implicit prices.
The air pollution measures used were average concentrations times
the number of days that exceeded certain levels. Though these measures
make sense in that they consider both concentration levels and frequency of
occurrence, they make the results of this study difficult to compare with
national air quality standards and with the results used by other studies.
6.2.6 Comparison and Review
It is extremely difficult to compare the numerical results of
these studies, because they cover different study areas with different air
pollution measures; however, it is striking that all of the studies have
found that a measure reflecting levels of one or more of the problematic
air pollutants in each study area has a statistically significant nega-
tive relationship with property values. This is strong evidence that
people are willing to pay higher prices for homes in locations with better
air quality and that the results of these preferences are reflected
in observable differences in property values. The Boston study found
that at mean values of the NOX measure and other variables in the hedonic
price function, a 1 pphm reduction in NOX was associated with a $1,613
increase in residential property values in 1970 dollars. The SCAB study
found that a reduction in N0£ from poor to fair conditions, an approxi-
mately 3 pphm reduction, was associated with an average increase in pro-
perty values of approximately $5,520; and a reduction in N02 from fair to
^v. Kerry Smith has suggested that due to Jensen's inequality, which says
that the expected value of a convex function is not equal to the convex
function of the expected value (see Mood, Graybill and Boes 1974), PVi
will be a biased estimate of E(PVi).
6-52
-------
good conditions, also about a 3 pphm reduction, was associated with an
approximately $7,080 increase in property values in 1977-1978 dollars.1^
These results are very much in the same vicinity. The SCAB study also
obtained mean marginal implicit prices of from $184 to $300 per Pg/m^ TSP,
while the Washington DC study obtained $69 per yg/m^ TSP in 1970 dollars.
Table 6.10 lists all the variables used in the hedonic price func-
tions in each study. They vary from one study to the next due to different
data availability and different study area characteristics. Because each
of these studies has found some variables that significantly influence
property values which have not been used in some of the other studies, it
is unlikely that early studies that examined a limited number of variables
specified the hedonic price function entirely accurately. All of the
studies have explained 80-90% of the variation in property values, but air
pollution is a small influence compared with many other variables, and
therefore other minor influences that have been overlooked may be affecting
the influence attributed to air pollution if they are correlated with air
pollution.
More work needs to be done to verify that no important location-
specific amenities that may be correlated with air pollution have been
overlooked in these studies. In particular these studies have looked
at school quality; crime rates; accessibility to major scenic and recrea-
tional amenities, such as rivers or the ocean; access to employment; land
use; and access to the CBD; but we do not know if these and air quality are
the only neighborhood amenities that need to be considered. It is possible
that proximity to parks, availability of public transportation, proximity
to grocery stores, and many other considerations are as or more important
as air quality in the household's residential location decision. The SFBA
study has explored many more possible influences than the other studies and
has found several of them to be significant.
Table 5.11 summarizes several important aspects of each of the
studies including pollution measures used, mean marginal implicit prices,
and WTP estimates for the scenario considered by the study.
Evidence concerning the appropriate functional form for the
hedonic price function is quite inconclusive. The same forms were not
tested in each case, but some studies found that the best fitting form
was one that implies that the marginal implicit price of air pollution in-
creases as air pollution increases, while others found the opposite.
Because the marginal implicit price function is determined by the inter-
action of supply and demand for air pollution, there is no reason to expect
it to be either positively or negatively sloped, or even to expect it to
have the same slope in all study areas, but the sometimes contradictory and
frequently inconclusive results are somewhat disconcerting. Harrison and
Rubinfeld tested for and found that benefit estimates were sensitive
figures were estimated from fhe reported monthly rent differentials
assuming a 10% discount rate and a perpetual stream of benefits.
6-!j3
-------
Table 6.10: Variables in the Hedonic Price Function
Washington B.C. Boston Denver SCAB SFBA
Accessibility of Neighborhood
distance to CBD
distance/time to employment
access to highways
distance to beach
X
X
X
X
X
X
X
X
X
Environmental Characteristics of Neighborhood
carbon monoxide
nitrogen oxides
suspended particulates
ozone
temperature
pollution standards index
X
X
X
X
X
X
X
X
X
X
X
X
House Characteristics
% over 30 years old
age of house
% new houses
house condition index
lot size
living area
rooms/unit
# bedrooms
# bathrooms
sales date
pool
fireplace
central air-conditioning
no toilet
parking on site
persons/unit
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
6-54
-------
Table 6.10 (Continued)
Washington D.C. Boston Denver SCAB SFBA
Socioeconomic Neighborhood Characteristics
% owner occupied
% black X
% Spanish
% lower status
populaton density
median years of school
% zoned large lots
% below poverty level
% white
housing density
Taxes and Public Services in Neighborhood
public safety expenditures
public school children/acre
school expenditures/student X
school test scores
student/teacher ratio
property tax rates X
composite tax rate
Neighborhood Amenities and Disamenities
% transportation, commercial,
utilities acreage
% business, commercial, industrial
acreage
% non-retail business acreage
% available for development
% precluded from development
vacancy rate
crime rate X
adjacent to river X
view present
elevation
slope
Property Values
median census tract market X
value
average census tract sale
price
individual property sale
price
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
6-55
-------
Table 6.11
Marginal Implicit Prices and Willingness to Pay Measures In Property Value Studies3
Study
(Time)
Washington, D.C.
(1969-70 Data)
Boston
(1970 Data)
Denver
<^ (1977-78 Data)
1
Ul
SCABC
(1975-78 Data)
SFBA
(1979-80 Data)
N.R. - Not Reported.
aResults based upon
Hedonlc
Price Functional
Forms Examined"
Linear, Semi-Log,
Inverse Semi-Log,
Log-Linear*
Linear, Semi-Log,
Seal-Log Exponential*
Linear, Semi-Log
Semi-log Exponential*
Linear, Semi-Log,
Semi-Log Expoential*
Linear, Semi -Log,
Exponentials Log-
Linear*
Semi-Log
Exponential*
Pollutant
Measures
TSP (ug/m3)
oxldant (.001 ppm)
NO (pphm)
TSP (mg/ha3)
CO pptm
TSP (ug/m3)
NO (pphm)
Same
Ozone Index
PSI Adjusted
those presented by authors. Conversion factors
with asSL'mDtlona .
Magnitude of
Pollutant
Examined
Mean; Range
(standard
deviation)
N.R.
6.31; 3-9
(1.16)
5.55; N.R.
(1.50)
121; 18-239
(55.5)
N.R.; 6.2-13.0
N.R.; 67-130
Same
90; N.R.
(81.6)
15.9; N.R.
(9.1)
are: lyg/m^ 1
Marginal
Implicit
Price
at Means
$69/ug/m3
$157.001 ppm
$l,613/pptm
$l,000.000b/
mg/hm
$67-8J/pptm
$300b/Mg/m3
$2,010/pphm
$184-$236b/Mg/m3
$l,B40-$2,360b/
pphm
N.R.
n:g/hm^; 1 ppm - 10
Willingness
To Pay Scenario
None reported.
Reduce automobile NO emission
from 1970 levels to expected 1990
levels based upon new automobile
emissions standards.
SIP to meet National Ambient Air
Quality Standards by 1982.
25-30Z Improvement In regloa-
wlde sir quality.
30X decrease in ozone Index
region-wide.
pptm 100 pphm.
Average
Annual Willingness
Pay Per Household
$83/y ear /household
($71 high income; $93 low
Income)
$258/year /household
($230 high Income; $374
low income)
$528-$588/year/
household
Not estimated
$82/year/household
cResulta reported In the upper and lower sections correspond to the first and second hedonlc price functions that were estimated for SCAB.
"Functional form of hedonlc price function used for benefit estimation denoted by *.
-------
to the use of different functional forms of the hedonic price function.
The uncertainty about the correct functional form of the hedonic price
function indicates that future property value studies should be careful to
test alternative forms and check the sensitivity of their benefit estimates
to the form chosen.
Table 6.12: WTP Elasticities
Pollution Income
Washington D. C.a
Boston
Denver
SCAB
SFBA
+.81
+.90 to
+.84 to
+.87
+1.00
.97
.98
+.80
+.80
+.63 to
+1.15
+.28
.90
aNelson uses 1/p the WTP function, but in the log-linear form this only
changes the sign of the elasticity.
log WTP = log a + b log 1/p + c log Y
= log a + b log 1 - b log p + c log Y
Table 6.12 reports the air pollution elasticity as defined by the
percentage change in willingness to pay for a reduction in air pollution
with respect to a one percent change in air pollution, and income elasticity
as defined by the percentage change in willingness to pay for a reduction
in air pollution with respect to a one percent change in income. In the
log-linear form of the WTP functions these are given by the estimated
coefficients of the pollution and income variables.' As shown in Table
6.12, the estimates of elasticity of willingness to pay for a reduction in
air pollution with respect to air pollution are all somewhat less than or
equal to 1. This means that a 10% increase in pollution would be associated
with a somewhat less than 10% increase in willingness to pay for a reduction
in air pollution.
WTP elasticities, as defined for Table 6.12, are not reported in all
of the studies. Some of the studies reported elasticities as if the WTP
function were a demand function. When a demand curve is estimated with
price as the dependent variable, i.e., when willingness to pay is estimated,
the dependent variable must be changed to quantity to obtain the usual
elasticities.
(1) log PRICE = log a + b log QUANTITY + c log INCOME
can be rewritten
(2) log QUANTITY = log a* - 1/b log PRICE + c/b log INCOME
When equation 1 is estimated, the usual price elasticity, % -QUANTITY/
% ^PRICE, is given by -1/b, and the usual income elasticity, % ^QUANTITY/
% -INCOME, is given by c/b. However, these are not very useful in the
analysis of the willingness to pay for a public good, because the consumer
does not have the flexibility in choice of what quantity to consume that
he has for private goods.
6-57
-------
The estimated income elasticities of WTP vary somewhat more. All
the studies but one found that willingness to pay for a reduction in
pollution is income inelastic, which means that a 10% increase in income
would cause less than a 10% increase in WTP. The one that is most divergent
is from the SFBA study, the only one that used household level data in the
WTP function. The Denver study was the only one to estimate different
income elasticities for different income groups and found that the high-
income group had the higher income elasticity. This is consistent with the
suggestion that air quality is a luxury good, because lower-income house-
holds would be willing to spend less of a 10% increase in their incomes on
air pollution reduction than would higher income households.
The consistency of the results of the estimation of elasticities of
willingness to pay for a reduction in air pollution should, however, be
viewed with caution. It may be more the result of the consistency of the
estimation procedure than of the similarity of WTP across areas. Four out
of five of these are based on a hedonic price function estimated with the
same functional form. Also, the growing skepticism about the information
obtained in the second step of this procedurespecifically, when the
derivative of the hedonic price function with respect to air pollution is
used as the dependent variable in the WTP functionindicates that much
more work in this area needs to be done. The somewhat comparable mean
marginal implicit prices obtained by different studies for the same pollutant
is promising. The one direct test of transferability (the SFBA study)
indicated the hedonic price function for property values was definitely not
transferable and that, not surprisingly, willingness to pay to reduce air
pollution will be different when air pollution conditions are different.
6.3 Comparison of Property Value and Bidding Method Benefit Measures
The measures of aggregate benefits to be achieved from emission
controls may be influenced by the selection of the property value or
bidding approaches. To date, the bidding approaches and other non-market
approaches give the only reliable estimates of visibility-related benefits
in Class I or recreation areas. In urban areas both approaches may be
applied; yet, in general, the question still remains, Do the benefit
estimation methods yield comparable results?
It is extremely difficult to compare the results of the bidding
method studies and the property value studies, because the economic good,
visibility or air quality, is defined differently in each type of study.
In several bidding studies the respondents were asked to value nebulously
defined best to worst cases. In other bidding studies more concrete
definitions of visual range reduction in kilometers were used. Property
value studies have used pollutant concentrations. This causes a problem,
since there is a limited ability to compare the visual effects of parts per
million of ozone or S02 to kilometers reduction in visual range and most
especially to the public's perceptions of these reductions.
6-58
-------
Given the difficulty in making comparisons, the benefit estimates
derived from the two approaches do exhibit a degree of comparability.
Because each of the studies uses different measures and levels of air
pollution, we group them into generalized worst, best and moderate condi-
tions for comparison. Several of the bidding method studies obtained
annual benefit estimates for a comparison of worst and best scenarios
of about $80 per household for non-metropolitan areas. The SCAB study
bidding method study obtained larger estimates of about $250 per household
per year for comparisons of worst to best, of which about $80 was bid for
the visibility aesthetics component and the rest for health-related changes.
Benefits from property value studies are difficult to compare
because the pollution measures and proposed improvements greatly vary
across studies. In the more severely polluted areas of Denver and Southern
California, values of around $250 to $500 per household were estimated for
30% to 50% improvements in air quality. As demonstrated in the SCAB
bidding method study, a significant portion of these benefits may be health
related, so that the aesthetic portion is likely to be in the vicinity of
$100 to $200 per household per year, which is somewhat larger than the
bidding method results. A comparable result of most bidding method and
hedonic property value studies is that the measured willingness to pay for
visibility changes was consistently found to be inelastic with respect to
income.
Because of differences in time periods, pollutant measures,
location, etc., more exacting comparison or statistical tests across most
studies would be spurious; however, the SCAB study simultaneously applied
both the bidding method and hedonic property value approaches for the
specific purpose of comparing the results. The SFBA study is replicating
this experiment, but their bidding method study results are not yet avail-
able.
The results of the SCAB comparison for an approximately 30% region-
wide improvement in air quality yielded bidding method estimates up to 62%
as large as of the property value estimates. An average bid of $312 per
household per year as opposed to an average WTP derived from the property
value study of $504 per year was probably the smallest difference in
results between the two approaches.
Results varied greatly with different functional forms for the
property values study and with different analysis techniques for the
survey, with the bidding method results tending to be significantly smaller
but within the same order of magnitude as the property value results.
However, in Brookshire et al. (1980) the authors revised this comparison
due to their lack of confidence in the procedure for estimating WTP func-
tions from the hedonic price function for property values. (See Section
4.3.2.) As an alternative, the authors hypothesize that property value
differences will exceed the true WTP for a given improvement in air quality
as was illustrated in Figure 4.1. To test this hypothesis they re-estimated
the hedonic price function for property values to more closely match the
community sample used for the bidding method survey and from this equation
6-59
-------
computed the property value difference associated with an approximately 30%
reduction in air pollution. The average property value differences were
$46 to $59 per month for a comparable improvement in air quality. The bids
then did not contradict the hypothesis that true WTP would be less than the
property value differences. A more rigorous test for comparing the two
approaches will require additional research on the estimation of reliable
WTP for air quality functions from property values.
6.4 Visibility Benefits in Benefit-Cost Analysis
The important question for visibility benefit analysis is, Are
these aesthetic benefits of a sufficient magnitude, relative to costs, to
be influential in a benefit-cost analysis of emissions control? A complete
comparison of benefits and costs would be a study unto itself and has not
been the focus of this guidebook. Further, it is especially difficult to
compare benefits and cost measures from different studies that do not
evaluate the same rate and type of emission control. Nevertheless, the
aggregate benefit measures from case studies just reviewed may be roughly
compared with estimates of control costs to answer the above question.
One source of estimated coal-fired electric-generating power plant
control costs is provided by ICF Inc. (1980). This effort identified
individual power plants that may have been required to install additional
emission control devices to meet EPA requirements regulating visibility
impacts in Class I areas, and calculated the required reduction in emis-
sions and costs of control. Up to 12 plants were identified that may have
had to install additional emission controls based upon several criteria of
maximum allowable visibility impacts in neighboring mandatory Class I
Federal areas potentially attributable to the plants.
Subsequent EPA interpretations of the regulations and analysis
determined that only one plant (the Four Corners plant) would need to
undertake additional control efforts to meet the regulations and that such
efforts were already planned in the New Mexico State Implementation Plan
(EPA 1981). With this caveat in mind, the ICF estimates of potential
control costs for several coal-fired electric-generating plants in the
Southwest may be compared with prior benefit studies in the Southwest to
address our question. An additional caveat is that the comparisons will
be very rough because none of the conditions or impact areas exactly
correspond across the benefit and cost studies. The Grand Canyon/Southwest
Parks study was to address the issue of comparative costs and benefits but,
is not yet available.
ICF identified five power plants in the Southwest potentially
impacting visibility in neighboring mandatory Class I Federal areas beyond
the defined critical levels. Two of these are in the Four Corners area and
affect Mesa Verde National Park (see Table 6.13). Degraded visibility is
said to occur when either visual range is reduced by more than 5% or
discoloration is greater than 5% (are measured by a blue/red ratio) from
6-60
-------
TABLE 6.13
COSTS OF CONTROLLING VISIBILITY IMPACTS IN CLASS I AREAS IN THE SOUTHWEST
Source
Nava jo
Four Corners
Mohave
Comnnche
liayden
San Juan
State
Arizona
New Mexico
Nevada
Colorado
Colorado
New Mexico
Affected
Class I Area
Grand Canyon
Mesa Verde
Joshua Tree
Rocky Mountains
Flat Tops
Mesa Verde
Distance Current
to Worst Case
Class I Visual Range
Area Reduction
97.7 km
62.7 km 5.75Z
183.0 km
82.0 km
64.0 km
49.8 km
Capital
Cost of
Controlling
Emission
That Affect
Visual Range*
(million)
NR
$57.5
NR
NR
NR
NR
Current
Worst Case
Blue/Red Ratio
.825
.850
.915
.930
.930
.940
Capital
Cost of
Controlling
Emission
That Affect
Discoloration**
(million)
$44.4
45.1
41.8
14.3
10.5
8.6
Source: ICF Inc. 1980, see also EPA (1981) for analysis and discussion.
*Estlmated costs of controlling worst case visual range reduction In affected Class I area to 5 percent. Costs
based upon controlling 502 emissions using a combination of scubbers, changes In electrostatic preclpltators and
changes In coal type.
**Estlmated costs of controlling worst case plume related blue/red discoloration ratio as close to 95 percent as
presently obtainable. Costs based upon controlling NC>2 emission by Installing over-fire or curtain air ports
over or around the boiler burner nozzle and Installation of low-NOx burners.
-------
man-made sources. The estimated capital costs of controlling emissions so
that these plants do not exceed these standards in Class I areas range from
$8.6 million at the San Juan plant to $102.6 million in the Four Corners
plant (columns 6 and 8 in Table 6.11), with a maximum impact on utility
rates of 3.1% (see IGF 1980, page 5-3). EPA subsequently determined that,
at the present time, NOX emissions cannot be reduced to achieve a
substantial improvement in visibility and is planning to delay imposition
of NOX controls "until such time as more effective with respect to
visibility, controls are available" (EPA, 1981).
The IGF estimated costs of control for plants in the Southwest to
meet the above criteria exceed the aggregate visibility benefit measures
(reported in Table 6.6) for both the Four Corners and Farmington benefit
studies conducted in the same area. These differences appear even greater
considering that the ICF estimated control costs appear to be for much
smaller changes in air quality than those ranging from 3.3% to over 100% in
these benefit studies.
Upon closer examination these benefit-cost discrepancies are not so
great. First, small changes in air quality at distant national parks may
also result in large changes in air quality in the communities and other
recreation sites (non-Class I areas) much closer to the emission sources
and examined by the benefit studies. Second, most all aggregate benefit
measures reported for the two Four Corners studies are for only a portion
(10-15%) of the areas impacted by changes in emission controls. Further,
these studies principally examined visibility aesthetic related activity
values. Initial unreleased results of the Grand Canyon/ Southwest Parks
study suggest preservation values may exceed activity values. Therefore,
the benefits reported are only a portion of the benefits received for a
portion of the impacted population.
Accounting for these factors it is possible that visibility aesthe-
tics benefits of control exceed the costs of control. The estimation and
addition of health, agriculture and materials damage benefits would further
support the probability that benefits may exceed the costs of control.
Clearly, these comparisons are very rough, but they do suggest that visi-
bility benefits are of a substantial magnitude to significantly influence a
benefit-cost determination of optimal levels of emissions control. This
should provide a strong incentive to push forward in considering and
measuring visibility benefits, more carefully related to proposed changes
in air quality, when making possible decisions on the appropriate level of
emissions control and air quality in the southwest.
The large benefit measures obtained for urban areas also suggests
that air quality related aesthetics and health benefits are substantial and
that more work is needed to specifically relate these benefits to the costs
of control to help determine the optimal level of air quality control in
these locations as well as in national parks.
6-62
-------
6-5 Applicability of Benefit Estimation Techniques for Prescribed
Burning
Analysis of benefits from air pollution control has typically been
concerned with continuous or frequent emissions from point sources of
pollution, such as power plants and smelters, or from multiple sources,
such as with urban and regional haze. Prescribed burning, which has the
temporary side effect of producing airborne pollutants and visual degra-
dation, may also be affected by Clean Air Act regulations because it is a
humanly induced (anthropogenic) source of pollution that sometimes occurs
in or near mandatory Class I Federal areas. Prescribed burning is some-
times viewed as a unique issue with regard to air quality regulations and
one for which the usual analysis does not apply because it causes pollution
infrequently and has many beneficial results. However, from an economics
perspective, prescribed burning may be analyzed in the same benefit-cost
framework used to analyze other anthropogenic sources of visibility degra-
dation.
Prescribed burning is the land management tool of burning an area
under controlled conditions for forestry and wildlife management, hazard
reduction, grazing and other land management objectives. It is most widely
used in timber producing areas to eliminate forest residue left from
harvesting operations that would be fuel for high intensity wildfires,
to remove underbrush before harvesting, and to protect the habitat of the
trees for future harvest. Alternatives include mechanical treatment,
chemical treatment, or changed harvesting systems that leave less residue
or remove underbrush. Prescribed burning is also used in wilderness and
other forest areas to maintain the natural ecosystems. Such prescribed
burning produces benefits similar to those of natural wildfires by main-
taining the normal wildlife habitat without some of the more damaging
effects of uncontrolled wildfires. An alternative that is sometimes used
is to allow naturally ignited fires to burn under controlled conditions.
There are no other currently documented practical alternatives to pre-
scribed burning for maintaining natural ecosystems (EPA 1979).
A benefit-cost analysis of prescribed burning would help in deter-
mining that amount of burning which would be economically optimal. Essen-
tially the same procedure would be used as for deciding the amount of
pollution to be allowed from a power plant. This would be to measure the
gains in consumer and producer surplus for those who benefit from prescribed
burning and the losses in consumer and producer surplus for those who incur
damages. Cramer and Pickford (1973) summarize this tradeoff for prescribed
burning.
The idea of permitting open burning of unwanted forest
residue, thereby placing great amounts of combustion
product? in the air, may be unacceptable to a lot of
people. But permitting these residues to accumulate
thereby inhibiting the reproduction of needed trees and
at the same time increasing the risk of a severely dam-
aging fire or of a smoke episode at some chance time
6-63
-------
is also quite unacceptable. Accomplishing the residue removal job
by methods other than burning may also have environmental and
economic consequences that are unacceptable to a lot of people. So
we must examine the possibility that the carefully regulated use of
fire may be less unacceptable than other alternatives in some
circumstances. (Page 239.)
There are several specific costs and benefits associated with
prescribed burning that must be considered in a benefit-cost analysis.
Among the costs are the operational costs of setting, monitoring and
controlling the burn; the temporary visual aesthetic impacts; the temporary
loss of land for recreation and wildlife in cases where the area was
previously available for these purposes; and possibly some health impacts
from the smoke. An important benefit of prescribed burning is the removal
of forest residue to reduce the risks of substantial damage due to periodic
uncontrolled wildfires upon air quality, recreation, wildlife and commercial/
private use of the affected lands. Additional benefits include the reduc-
tion of underbrush, the maintenance of the natural ecosystem, and the
maintenance of a beneficial environment for future timber harvesting and
for wildlife. A complete analysis would also consider the costs of
alternatives, such as mechanical treatment of residuals, which can be used
to achieve the same goal; however, in some instances, such as the preser-
vation of natural ecosystems, there are few practical alternatives. Users
of prescribed burning, if forced to reduce or stop the practice, would be
expected to use some alternative procedures to at least partially compen-
sate for the reduction in prescribed burning.
It has been suggested that emissions from prescribed burning are
less than what would occur in its absense and are a minor source of pol-
lution.
In comparison with other sources, prescribed burning
accounts for less than 2 percent of the particulate
produced by all urban/industrial and rural/agricul-
tural sources. Forest fires are more serious, pro-
ducing possibly five times the particulates from pre-
scribed burning. (Murphy, Fritschen, and Cramer, 1970,
page 4.)
This implies that solely in terms of air quality, the benefits of pre-
scribed burningreduced damages from wildfires because they occur less
oftenoutweigh the costsperiodic visual aesthetic damage and possibly
health impacts caused by the smoke from prescribed burning.
The visibility benefit analysis techniques described in this
guidebook are readily applicable for quantifying the visual aesthetic impacts
and other costs and benefits of prescribed burning. The attributes of the
prescribed burning or wildfire, such as frequency, duration and intensity,
can be considered part of the necessary scenario development in much the
6-64
-------
same manner as for other sources of visual impacts. Because most such im-
pacts would be located in nonurban areas, contingent market or travel cost
and site substitution methods would probably be the most applicable benefit
estimation techniques. The contingent market approaches, such as bidding
methods, could be used to estimate damage to recreation, property, businesses,
and wilderness resulting from prescribed burns and wildfires. Travel cost
models could be used to estimate visitation rates and recreation benefits
in affected areas before, during, and after periods of burns (prescribed
and wildfire) and could then be used to calculate the benefits lost during
and after the burn. Unfortunately, with the travel cost approach it would
be difficult to separate air quality impacts from other impacts of such
burning, such as loss of wildlife. This approach also will not capture the
damages to local residents, to those passing through the area, or to those
using impacted recreation areas without prior knowledge that the air
quality conditions in the area would be affected by nearby burning.
6-65
-------
GLOSSARY
Terms are defined in relation to their use in this guidebook.
Activity Value: A measure of value assigned to a good or service when it
is in use. Also referred to as user value.
Bid Curve: A graphical representation of the quantities of a good that
the consumer would be willing to purchase at each price of the good,
with each price and quantity combination yielding the same utility.
Bid Function: A function relating stated WTP or WTA measures for visi-
bility changes or levels to the visibility change or level, socio-
economic characteristics of the respondents and other explanatory
variables
Bidding Methods; A direct survey approach in which respondents are given
information on a hypothetical situation and are asked through a
structured procedure to estimate WTA and WTP measures.
Compensating Surplus (CS): A consumer surplus measure that is the change
in income that, given a new quantity of a good, offsets the change in
utility induced by the quantity change. The individual has the
new quantity of the good but remains on the old indifference curve.
Compensating Variation (CV): A consumer surplus measure that is the
change in income that, given a new price of a good, offsets the change
in utility induced by the price change. The individual faces the new
price but remains on the old indifference curve.
Consumer Surplus: The difference between what a consumer would be willing
to pay rather than do without each unit of a good and what the consumer
actually pays for each unit of the good.
Demand Curve: Same as Ordinary Demand Curve.
Discount Rate: An interest rate used to calculate the present value of
future benefits or costs.
Equilibrium: In reference to the market for a good, it means that a price
has been established at which the supply of the good equals the demand
for the good. Equilibrium is the condition which when, or if, attained
is stable as long as supply or demand conditions do not change.
Equivalent Surplus (ES): A consumer surplus measure that is the change in
income that, given the old quantity of the good, yields the same
change in utility as that induced by the quantity change. The indi-
vidual has at the old quantity of the good but is on the new indif-
ference curve.
G-l
-------
Equivalent Variation (EV): A consumer surplus measure that is the change
in income that, given the old price of a good, yields the same change
in utility as the price change. The individual faces the old price
but is on the new indifference curve.
Existence Value; The value assigned to the simple existence of a good,
such as visibility aesthetics at a site.
Expenditure: The actual amount a consumer spends for a good or service.
Expenditure Function; Shows the change in expenditures that will maintain
a designated level of an individual's utility when prices and/or
environmental quality change.
Externality; Interaction between economic agents (i.e., firms or house-
holds) that brings costs or benefits to those agents without their
consent in the absence of market determined compensation.
Hedonic Price; The implicit price of a non-market good or characteristic
that is implied by the prices of associated market goods.
Hypothetical Bias; Inaccuracies or biases in survey responses occuring
when a hypothetical situation is unfamiliar or unbelievable to respond-
ents and results in misstatements of the responses.
Income Compensated Demand Curve (ICDC); The relationship between price of
a good and quantity demanded when, for a given price change, the
consumer's income is adjusted so that his utility remains constant.
Income Effect: That part of a change in the purchases of a good that can
be attributed to the change in purchasing power that has resulted from
a change in the price of the good.
Indifference Curve: A graphical representation of a set of consumption
alternatives each yielding the same utility.
Information Bias: Variations in bidding method responses introduced by
variation in the type of information provided in the survey instrument.
Market Segmentation: Occurs when the market for a good, such as housing,
is made up of several distinct submarkets with unique supply and
demand conditions in each.
Matched Pairs Technique; A procedure for selecting properties to be used
in a hedonic price study by which neighborhood pairs are selected
that are similar in all respects except the attribute to be priced.
Non-market Good: A good, such as a public good, that is not exchanged in
a market and for which there is no directly observable price.
Normal Good: A good for which the quantity demanded at any price increases
as income increases.
G-2
-------
Offer Curve: A graphical representation of the quantities of a good that
the supplier would be willing to supply at each price of the good,
with each price and quantity combination yielding the same profit.
Option Value; A value assigned to the option to consume a good or service
at a stated price at some future date.
Ordinary (Marshallian) Demand Curve (ODC): A curve showing the quantity
of a good or service that a utility maximizing consumer or consumers
with a given income level will demand at each price.
Payment Scheme; The method by which actual payments would be determined
if a proposed plan were implemented. For example, all respondents pay
the mean value, or all respondents pay their stated value, etc.
Payment Vehicle: The method by which payment for something is made,
such as a user fee or a tax.
Price Elasticity: A measure of how responsive the quantity demanded or
supplied is to a change in the price of a good. It is the percentage
change in quantity divided by the percentage change in price.
Public Good: A good that is non-exclusive; that consumption by any one
individual does not preclude consumption by other individuals.
Since it can be provided to additional individuals without limiting
its availability to others, the marginal cost of providing the
good to additional individuals is zero.
Rejection Bias: Zero bids, infinity bids or bids which misrepresent true
valuations because respondents reject some or all of the bidding
method procedure.
Residential Location Model: A model that explains the household's
residential location decision based on distance from central business
district and neighborhood amenities.
Starting Point Bias: Inaccuracies in bidding method responses because the
starting bid suggests information as to the expected final bid,
thus influencing respondents to give bids other than their true
value.
Strategic Bias: Inaccuracies in responses due to respondents systema-
tically over or understating true values in an attempt to influence
survey results.
Substitution Effect: That part of a change in the purchases of a good
that can be attributed to substitution among goods that has resulted
from a change in the price of the good relative to the prices of other
goods.
Survey Instrument: Any type of survey form containing descriptions and
the questions to be answered by respondents.
G-3
-------
Utility: A qualitative measure of an individual's well being.
Utility Function: The relationship between an individual's utility and
the quantity of goods and services consumed.
Vehicle Bias: In bidding methods, variations in responses due to varia-
tions in the payment vehicle used.
Weak Complementarity: Two goods are weakly complementary if when the
quantity demanded of one is zero, the marginal utility for changes in
the other good is zero.
Willingness to Pay (WTP): The maximum dollar amount an individual
is willing to pay to promote an improvement or prevent a degradation
in environmental quality.
Willingness to Accept (WTA); The minimum dollar amount an individual
is willing to accept in compensation for a proposed degradation, or to
forego a proposed improvement in environmental quality.
EPA AIR QUALITY REGULATION TERMS
Terms are defined in EPA (1980c).
Adverse impact on visibility means, for purposes of S307, visibility
impairment which interferes with the management, protection, preser-
vation, or enjoyment of the visitor's visual experience of the Federal
Class I area. This determination must be made on a case-by-case basis
taking into account the geographic extent, intensity, duration,
frequency and time of visibility impairment, and how these factors
correlate with (1) times of visitor use of the Federal Class I area,
and (2) the frequency and timing of natural conditions that reduce
visibility. This term does not include effects on integral vistas.
Best Available Retrofit Technology (BART) means an emission limitation
based on the degree of reduction achievable through the application of
the best system of continuous emission reduction for each pollutant
which is emitted by an existing stationary facility.
Federal Land Manager means the Secretary of the department with authority
over the Federal Class I area or, with respect to Roosevelt-Campobello
International Park, the Chairman of the Roosevelt-Campobello Inter-
national Park Commission.
Fugitive emissions means those emissions which could not reasonably pass
through a stack, chimney, vent, or other functionally equivalent
opening.
G-4
-------
Integral vista means a view perceived from within the mandatory Class I
Federal area of a specific landmark or panorama located outside the
boundary of the mandatory Class I Federal area.
Mandatory Class I Federal Area includes all international parks, national
wilderness areas, and the national memorial parks exceeding 5,000
acres and national parks exceeding 6,000 acres.
Natural conditions includes naturally occurring phenomena that reduce
visibility as measured in terms of visual range, contrast, or color-
ation.
Reasonably attributable means attributable by visual observation or any
other technique the state deems appropriate.
Significant impairment means, for purposes of S303, visibility impairment
which, in the judgment of the Administrator, interferes with the
management, protection, preservation, or enjoyment of the visitor's
visual experience of the mandatory Class I Federal area. This deter-
mination must be made on a case-by-case basis taking into account the
geographic extent, intensity, duration, frequency and time of the
visibility impairment, and how these factors correlate with (1) times
of visitor use of the mandatory Class I Federal area, and (2) the
frequency and timing of natural conditions that reduce visibility.
Visibility impairment means any humanly perceptible change in visibility
(visual range, contrast, coloration) from that which would have
existed under natural conditions.
Visibility in any mandatory Class I Federal area includes any integral
vista associated with that area.
G-5
-------
Bibliography
Bibliography entries are coded according to their content. Most
entries are also annotated so the researcher can easily determine content
and compile references on the desired topic area. The reference codes
are:
1. Consumer behavior theory, welfare economics and other economic
theory.
2. Property value studiestheory and application.
3. Bidding method and studies related to contingent market approach-
theory and application.
4. Travel cost site substitution studies, household production
function, wages, voting and other techniques.
5. Economics of environment and natural resources.
6. Benefit-cost analysis and environmental control cost studies.
7. Modeling, measuring, and monitoring visibility and human per-
ceptions .
8. Survey theory, statistical theory, mathematical modeling.
9. Legislative documents and interpretations.
-------
Bibliography
Code
Abelson, Peter. 1979. Cost Benefit Analysis and Environmental 6
Problems. Westmead, Farnborough, Hants., England: Saxon
House.
Overview of benefit cost analysis with applied core
studies on soil conservation, sand mining, airport location
and noise, and property prices and amenity values.
Ajzen, I., and Fishbien, M. 1977- "Attitude-Behavior Relations: 3, 8
A Theoretical Analysis and Review of Empirical Research."
Psychological Bulletin 84: 888-918.
Discusses, among other issues, conditions under which
responses to questionnaires concerning intended behavior will
accurately predict actual behavior.
Anderson, R.J., Jr., and Crocker, T.D. 1971. "Air Pollution and
Residential Property Values." Urban Studies 8 (October):
171-180.
One of the earlier property value studies which found a
significant negative relationship between property values and
a measure of sulfur oxides and suspended particulates in
Washington B.C., Kansas City, and St. Louis.
1972. "Air Pollution and Property Values: A Reply."
The Review of Economics and Statistics 54 (November):
470-473.
A response to the criticisms by Freeman (1971) of the
interpretation of empirical results presented in Anderson and
Crocker (1971). The authors defend the use of the observed
relationship between property values and air pollution to
estimate air quality benefits.
Anderson, Robert A. 1980. "Visibility Valuation in Federal 2, 3,
Class I Areas." USEPA Order //W-4207-NASX. (Draft.) 4, 6
A short overview of benefit techniques that may be
applied to visibility benefit estimation in Class I areas.
B-l
-------
Code
Appel, David. 1980. "Estimating the Benefits of Air Quality 2
Improvement: An Hedonic Price Index Approach Applied to the
New York Metropolitan Area." Doctoral thesis, Rutgers State
University, New Brunswick, New Jersey.
Estimates a hedonic price function for residential
property in the suburban New York metropolitan area, with a
measure of suspended particulates used as an environmental
attribute. Particulates were found to have a significant
negative influence on property values comparable in magnitude
to that found in earlier studies.
Arrow, Kenneth J., and Fisher, Anthony C. 1974. "Environmental 1, 5
Preservation, Uncertainty, and Irreversibility." Quarterly
Journal of Economics 88: 312-319.
Introduces the quasi-option value concept and shows that
the expected benefits of an irreversible decision should be
adjusted to reflect the loss of options that the decision
entails.
Atkinson, Scott E., and Lewis, Donald H. 1976. "Determination 5, 6
and Implementation of Optimal Air Quality Standards."
Journal of Environmental Economics and Management 3: 363-380.
Backstrom, Charles H., and Hursh, Gerald D. 1974. Survey 8
Research. Chicago, Illinois: Northwestern University Press.
A concise, readable handbook on basic survey research
procedures.
Baumol, William J., and Oates, Wallace E. 1975. The Theory of 1, 5
Environmental Policy. Englewood Cliff, New Jersey: Prentice-
Hall, Inc.
Rigorous theoretical demonstration in a general equili-
brium framework that marginal taxes set equal to the marginal
damages caused by a negative externality, such as air pollution,
will provide an economically optimal solution.
1979. Economics, Environmental Policy, and the Quality
of Life. Englewood Cliff, New Jersey: Prentice-Hall, Inc.
Discusses the problem of maintaining environmental
quality from a historical, institutional and policy point of
view. Central theme is that rational, effective environ-
mental policies can be developed by applying the principles
B-2
-------
Code
of economics, and in particular by using the pricing system,
that will provide incentives for consumers, business and
government to protect the environment at a minimum cost to
society.
Becker, Gary S. 1965. "A Theory of the Allocation of Time.
The Economic Journal 75 (September): 493-517-
Bergstrom, R.W.; Babson, B.L.; Latimer, D.A.; Holman, Hoi-Ying;
and Wojcik, M.A. 1980. "Comparison of the Observed and
Predicted Visual Effects Caused by Power Plant Plumes."
Paper presented at the Symposium on Plumes and Visibility,
Grand Canyon, Arizona, November 10-14.
Berry, Brian J.L., and Bednarz, Robert S. 1975. "A Hedonic
Model of Prices and Assessments for Single-Family Homes:
Does the Assessor Follow the Market or the Market Follow the
Assessor?" Land Economics 51 (February): 21-40.
Examines the differences between market values and
assessed values of residential properties in the Chicago area
using hedonic price equations to account for the characteris-
tics of the property. Air pollution measures are used, with
mixed results.
Bishop, R.C., and Heberlein, T.A. 1979. "Measuring Values of 3, 4
Extra Market Goods: Are Indirect Measures Biased?" American
Journal of Agricultural Economics 61 (December): 926-930.
Discusses problems in travel cost (TC) and contingent
market (CM) experiments. In an experiment where actual cash
offers for goose permits are compared with TC and CM esti-
mates, all TC and CM estimates substantially overestimated or
underestimated cash offers. Cash payments were not compared
to CM and TC estimates.
Blackwell, H.R. 1946. "Contrast Thresholds of the Human Eye.
Journal of the Optical Society of America 36: 642-643.
Blank, F., Brookshire, D.; Crocker, T.; d'Arge, R.; Horst, R.; 3, 4
and Rowe, R. 1977. "Valuation of Aesthetic Preferences: A
Case Study of the Economic Value of Visibility." Research
report to the Electric Power Research Institute, Resource and
Environmental Economics Laboratory, University of Wyoming.
B-3
-------
Code
The original research report on the Farmington study,
with results reported for both the bidding method study and
the household production function approach.
Blank, F., and Rowe, R.D. 1981. "Deviations Among Empirical 1, 3
Consumer Surplus Measures." Paper presented at the Western
Economic Association Meetings, San Francisco, California.
Examines several arguments as to why empirical ES and CS
measures have been observed to diverge beyond what is pre-
dicted by Randall and Stoll (1980).
Blumenthal, D.L.; Richards, L.W.; Macias, E.S.; Bergstrom, R.W.; 7
Wilson, W.E.; and Bhardwaja, P.S. 1980. "Effects of a
Coal-Fired Power Plant and Other Sources on Southwestern
Visibility (Interim Summary of EPA's Project VISTTA)."
Paper presented at the Symposium on Plumes and Visibility,
Grand Canyon, Arizona, November 10-14.
Bockstael, Nancy E., and McConnell, Kenneth E. 1980. "Calcu- 1, 3
lating Equivalent and Compensating Variation for Natural
Resource Facilities." Land Economics 56: 56-63.
Examines one argument as to whether empirical ES and CS
may diverge more than predicted by Willig (1976) and Randall
and Stoll (1980).
Bohm, P. 1971. "An Approach to the Problem of Estimating Demand
for Public Goods." Swedish Journal of Economics 73 (March):
94-105.
1972. "Estimating Demand for Public Goods: An Experi-
ment." European Economic Review 3: 111-130.
Evaluates the impact of eight alternative payment schemes
upon stated willingness to pay for public television.
Bouchard, T.J. 1976. "Field Research Methods: Interviewing,
Questionnaires, Participant Observation, Systematic Observa-
tion, Unobtrusive Measures." In Handbook of Industrial and
Organizational Psychology. Edited by M.D. Dunnett. Chicago,
Illinois: Rand McNally.
B-4
-------
Code
Bowen, Howard R. 1943. "The Interpretation of Voting in the
Allocation of Economic Resources." Quarterly Journal of
Economics 58: 27-48.
Bradford, D.F. 1970. "Benefit-Cost Analysis and Demand Curves
for Public Goods." Kyklos 23: 1145-1159.
Examines the benefits of public good provision. A demand
like "aggregate bid curve" is developed as a useful tool and
integrated into benefit-cost analysis.
Bresnock, Anne E. 1980. "Housing Prices, Income and Environ-
mental Quality in Denver." Paper presented at the American
Economic Association meeting, Denver, Colorado, September.
A hedonic property value study of Denver; discussed in
Section 6.3.3 of this guidebook.
1981. Assistant Professor. Department of Economics, San
Diego State University, San Diego, California. Personal
communication.
Brookshire, D., and Crocker, T. 1979. "The Use of Survey 3
Instruments in Economic Valuations of Environmental Goods."
In Assessment of Amenity Resource Values. Washington, D.C.:
U.S. Department of Agriculture, Rocky Mountain Forest and
Range Experiment Station.
Discusses the benefits and costs of using the survey
instruments approach for valuation purposes relative to
alternatives. The core of this paper is contained in the
following reference, pages 29-45.
Brookshire, D.; d'Arge, R.; Schulze, W. ; and Thayer, M. 1979. 2, 3
Methods Development for Assessing Air Pollution Control
Benefits.EPA-600/6-79-0016.Washington, D.C.: USEPA. Vol.
2: Experiments in Valuing Non-Market Goods: A Case Study of
Alternative Benefit Measures of Air Pollution Control in the
South Coast Air Basin of Southern California (February).
A report of the South Coast Air Basin Study of willing-
ness to pay for air quality comparing a bidding method and a
hedonic property value approach discussed in Sections 6.2.4.
and 6.3.4.
B-5
-------
Code
1980. "Valuing Public Goods: A Comparison of Survey and 2, 3
Hedonic Approaches." Resource and Environmental Economics
Laboratory, University of Wyoming. (Mimeograph.)
A short summary with some revisions of the South Coast
Basin study, discussed in Sections 6.2.4 and 6.3.4 of this
guidebook, comparing the results of the bidding method and
hedonic price approach.
. 1981. "Experiments in Valuing Public Goods." In Advances 2, 3
in Applied Microeconomic Theory. Edited by V.K. Smith.
Greenwich, Connecticut: JAI Press. (Forthcoming.)
A somewhat shortened presentation of the 1979 South Coast
Air Basin study by the same authors.
Brookshire, D.; Crocker, T.; d'Arge, R.; Eubanks, L.; Randall,
A.; Rowe, R.; and Stoll, J. 1977. "Methodological Experi-
ments in Valuing Wildlife Resources." Final Report for U.S.
Fish and Wildlife Service, Phase I. Laramie, Wyoming:
Resource and Environmental Economics Laboratory, University
of Wyoming.
Examines the application of iterative bidding techniques
for activity and option values related to western wildlife
studies. Travel cost and site substitution approach is also
examined.
Brookshire, David S., and Eubanks, Larry S. 1978. "Contingent
Valuation and Revealing the Actual Demand for Public Environ-
mental Commodities." Paper presented at the Public Choice
Society meeting, New Orleans, Louisiana, March.
Brookshire, D.; Ives, B.; and Schulze, W. 1976. "The Valuation
of Aesthetic Preferences." Journal of Environmental Economics
and Management 3 (December): 325-346.
Reports the results of the Lake Powell study discussed at
length in Section 6.2.2 of this guidebook.
Brookshire, D., and Randall, A. 1978. "Public Policy Alterna-
tives, Public Goods, and Contingent Valuation Mechanisms."
Paper presented at the Western Economic Association meeting,
Honolulu, Hawaii, June 20-26.
B-6
-------
Code
Brookshire, David S.; Randall, Alan; and Stoll, John R. 1980.
"Valuing Increments and Decrements in Natural Resource
Service Flows." American Journal of Agricultural Economics
61 (August): 478-488.
Discusses an application of the contingent market bidding
method approach to the estimation of values for fish and
wildlife stocks in the West.
Brown, Gardner, Jr., and Mendelsohn, Robert. 1980. "The Hedonic-
Travel Cost Method." Final report prepared for Division of
Program Plans, U.S. Department of Interior, Seattle, Washing-
ton, University of Washington.
Applies a hedonic approach to value site attributes
across many competing sites, thus deriving average implicit
prices for the attributes.
Brown, Gardner, M., Jr., and Pollakowski, Henry 0. 1977-
"Economic Valuation of Shoreline." The Review of Economics
and Statistics 59 (August): 272-278.
Application of the hedonic property value approach to the
valuation of water-related open space.
Brown, P.J.; Driver, B.L.; and McConnell, C. 1978. "Opportunity
Spectrum Concept and Behavioral Information in Outdoor
Recreation Resource Supply Inventories: Background and
Application." Rocky Mountain Forest and Range Experiment
Station General Technical Report 55. Fort Collins, Colorado.
Cesario, F.J. 1976. "Value of Time in Recreation Benefit 1, 4
Studies." Land Economics 55: 32-41.
Cesario, F.J., and Knetsch, J.L. 1976. "A Recreation Site
Demand and Benefit Estimation Model." Regional Studies 10:
97-104.
Chiang, Alpha C. 1974. Fundamental Methods of Mathematical
Economics. New York: McGraw-Hill, Inc.
An introductory text of mathematics in economics.
B-7
-------
Code
Chipman, J.S., and Moore, J.C. 1980. "Compensating Variation,
Consumer's Surplus, and Welfare." American Economic Review
70: 933-949.
Advanced theoretical presentation of conditions under
which compensating variation measure can be used to rank
alternative projects and how EV and CV can be derived from
observable demand curves.
Cicchetti, Charles J., and Freeman, A. Myrick III. 1971. 1
"Option Demand and Consumer Surplus: Further Comment."
Quarterly Journal of Economics 85 (August): 528-539.
Shows that when there is a non-exclusive use of a good,
depletable supply, and uncertainty in supply and demand,
option values will be positive for risk averse individuals
and should be computed and added to traditional user value
consumer surplus benefits measures.
Cicchetti, Charles J., and Smith, V. Kerry. 1976. The Costs of 1, 5
Congestion. Cambridge, Massachusetts: Ballinger Publishing
Company.
Develops a theoretical model of the impacts of congestion
on users of a common resource and presents an application of
the model for estimating the impacts of congestion on an
individual's wilderness recreational experience.
Clark, E. 1971. "Multipart Pricing of Public Goods." Public 1, 3
Choice 11 (Fall): 17-33.
Clark, R.; Hendee, J.C.; and Cambell, F. 1971. "Values, Behavior and 7
Conflict in Modern Camping Culture." Journal of Leisure Research.
3: 143-159.
Clawson, M. 1959. Methods of Measuring Demand for and Benefits 5
of Outdoor Recreation. Reprint No. 10. Washington, D.C.:
Resources for the Future, Inc.
Clawson, M., and Knetsch, J.L. 1966. Economics of Outdoor
Recreation. Baltimore: Johns Hopkins Press.
Cornsweet, T.N. 1970. Visual Perception. New York and London:
Academic Press.
B-8
-------
Code
Craik, K.H. 1979. "The Place of Perceived Environmental Quality
Indices (PEQIs) in Atmospheric Visibility Monitoring and
Preservation." Proceedings of the Workshop in Visibility
Values, Fort Collins, Colorado, January 28-February 1.
General Technical Report WO-18. Washington, D.C.: U.S.
Department of Agriculture.
Cramer, Owen P., and Pickford, Stewart G. 1973. "Factors 6
Influencing Smoke Management Decision in Forest Areas."
Proceedings of Air Quality and Smoke from Urban and
Forest Fires International Symposium, October 24-26.
Crespi, I. 1971. "What Kinds of Attitude Measures are Predic- 3, 8
tive of Behavior?" Public Opinion Quarterly 35: 327-34.
Cropper, M.L., and Arriaga-Salinas, A.S. 1980. "Intercity Wage 4
Differentials and the Value of Air Quality." Journal of
Urban Economics 8: 236-254.
Estimates labor supply and demand functions across
several cities from which values of air quality are derived.
Discussed in Section 4.4.4 of this guidebook.
Currie, J.M.; Murphy, J.A.; and Schmitz, A. 1971. "The Concept 1
of Economic Surplus and Its Use in Economic Analysis." The
Economic Journal 81: 741-799.
A complete review paper of all aspects of economic
surplus accepted by the profession as of 1971.
Davis, R. 1963. "Recreation Planning as an Economic Problem." 5
Natural Resources Journal 3: 239-249.
de Nevers, N. 1975. The Application of Cost-Benefit Analysis to 6, 7
Air Pollution Control Regulation: A General Discussion with
Application to the Example of Sulfur Dioxide Control at the
Huntington and Emery Powerplants. Salt Lake City: Utah
Power and Light Company.
1979. "Local Visibility and Power Generation in Central 7
Utah." Paper presented at "View of VisibilityRegulatory
and Scientific," Air Pollution Control Association Specialty
Conference, Denver, Colorado, November 26-27.
B-9
-------
Code
Devine, Hugh, and Smith, V. Kerry. 1981. "Visibility Benefit
Analysis and Travel Cost Recreational Demand Models."
(Mimeograph.)
Using the household production function approach to
consumer behavior, demand models for recreation are developed
and related to site attributes. This approach is tied to the
travel cost models and the hedonic travel cost model.
Deyak, T.A., and Smith, V. Kerry. 1974. "Residential Property
Values and Air Pollution: Some New Evidence." Quarterly
Review of Economics and Business 14 (4): 93-100.
Estimates a property value function based on a residential
location model across 100 U.S. cities using SMSA data; finds
a significant negative relationship between property values
and a measure of suspended particulates.
1978. "Congestion in Participation in Outdoor Recreation: 4, 5
A Household Production Function Approach." Journal of
Environmental Economics and Management 5: 63-80.
Diamond, P.A., and McFadden, D.L. 1974. "Some Uses of the
Expenditure Function in Public Finance." Journal of Public
Economics 3: 3-21.
Uses expenditure functions to examine the dead weight
burden of taxation, optimal commodity taxes, and criteria for
indivisible public investments financed by lump sum taxation.
Domencich, Thomas A., and McFadden, Daniel. 1975. Urban Travel 1, 4
Demand: A Behavioral Analysis. Amsterdam: North Holland
Publishing Company.
Drivas, P.J.; Bass, A.; and Heinold, D.W. 1980. "A Plume Blight
Visibility Model for Regulatory Use." Symposium on Plumes
and Visibility, Grand Canyon, Arizona, November 10-14.
Dwyer, J.F.; Kelly, J.R.; and Bowes, M.D. 1977. Improved
Procedures for Valuation of the Contribution of Recreation to
National Economic Development. WRC Research Report No. 128.
Urbana, Illinois: Water Resources Center.
B-10
-------
Code
Guidelines for the use of economic benefit measures and
techniques applied to recreation experiences. Not speci-
fically related to visibility.
Erskin, H. 1972. "The Polls: Pollution and its Costs." Public 3
Opinion Quarterly 36: 120-135.
Ettenheim, George P.,Jr. 1979. Considerations in Visibility 7
Monitoring Measurements. Altadena, California: Meteorology
Research Inc., July.
Eubanks, Larry S., and Brookshire, David S. 1980. "Household
Production and Non-Market Valuation." Paper presented
American Economic Association meetings, Denver, Colorado,
September.
Flagg, A.T. 1976. "Methodological Problems in Estimating
Recreational Demand Functions and Evaluating Recreational
Benefits." Regional Studies 10: 353-362.
Freeman, A. Myrick III. 1979a. The Benefits of Environmental 1, 2,
Improvement: Theory and Practice. Baltimore and London: 4, 5
Johns Hopkins University Press for Resources for the Future,
Inc.
A thorough and sometimes technical discussion of the
economic approach to understanding and identifying the
benefits of environmental quality. The property value
approaches are given a lengthy review.
1979b. "Hedonic Prices, Property Values and Meauring
Envrionmental Benefits: A Survey of the Issues." Scandi-
navian Journal of Economics 81: 154-173.
An overview of the theoretical underpinnings of the
hedonic price technique as it is applied to property value
data to obtain willingness to pay for air quality similar to
the presentation in Section 4.3.3 and in Freeman (1979a).
The article includes responses to some of the criticisms of
the approach that have been raised.
Freeman, A. Myrick III, and Haveman, Robert H. 1977. "Congestion,
Quality Deterioration, and Heterogeneous Tastes." Journal of
Public Economics 8: (October) 225-232.
Fromm, G. 1968. "Comment." In Problems in Public Expenditure
Analysis. Edited by S.B. Chase, Jr. Washington, D.C.:
Brookings Institute.
B-ll
-------
Code
Gibson, John G., and Anderson, Robert W. 1975. "The Estimation 4
of Consumers' Surplus from a Recreational Facility with
Optional Tariffs." Applied Economics 7: 73-79.
Goodwin, Susan Ann. 1977. "Measuring the Value of Housing 2
QualityA Note." Journal of Regional Science 17 (April):
107-115.
Presents the locational variables that were found to
significantly influence monthly rents in the Boston area.
These included an air pollution index and an employment
accessibility index.
Griliches, Zvi, editor. 1971. Price Indexes and Quality Change. 1, 2
Cambridge, Massachusetts: Harvard University Press.
Hammock, Judd, and Brown, Gardner, M. 1974. Waterfowl and 3, 5
Wetlands: Toward Bioeconomic Analysis. Baltimore: Johns
Hopkins University Press for Resources for the Future,
Inc.
Hansen, Morris H.; Hurwitz, William N.; and Madow, William G.
1960. Sample Survey Methods and Theory, New York: John
Wiley and Sons, Inc.
Sampling techniques and statistical distribution tests
and variance calculations are presented.
Harrison, D., Jr., and Rubinfeld, D.L. 1978. "Hedonic Housing
Prices and the Demand for Clean Air." Journal of Environmen-
tal Economics and Management 5 (March): 81-102.
A hedonic price study of property values in the Boston
area, presented in Section 6.3.2.
Hause, John C. 1975. "The Theory of Welfare Cost Measurement.
Journal of Political Economy 83: 1145-1182.
Heisler, S.L.; Henry, R.C.; Watson, J.G.; and Hidy, G.M. 1980.
The 1978 Denver Winter Haze Study. Volume I: Executive
Summary. Environmental Research and Technology Inc. West-
lake Village, California, for Motor Vehicle Manufacturers
Association.
B-12
-------
Code
Hendee, J.C., and Burdge, R.J.; 1974. "The Substitutability Concept:
Implications for Recreation Research and Management." Journal
of Leisure Research. 6:155-162.
Hendee, J.C.; Gale, R.P.; and Cotton, W.R., Jr. 1971. "A Typology
of Outdoor Recreation Activity Preferences." Journal of
Environmental Education. 3:28-34.
Henderson, A. 1941. "Consumer Surplus and the Compensating
Variation." The Review of Economic Studies 8: 117-121.
Henry, R.C. 1979a. "Psychophysics and Visibility Values in
Proceedings of the Workshop in Visibility Values." Edited
by Fox et al., Fort Collins, Colorado, January 28-February
1. USDA General Technical Report WO-18.
. 1979b. "The Human Observer and Visibility-Modern Psycho-
physics Applied to Visibility Degradation." Paper presented
at "View on Visibility-Regulatory and Scientific," Air
Pollution Control Association Specialty Conference, Denver,
Colorado., November 26-27.
Hicks, J.R. 1941. "The Rehabilitation of Consumers' Surplus."
The Review of Economic Studies 8: 108-116.
1944. "The Four Consumers' Surplus." Review of Economic
Studies 11 (Winter): 31-41,
1945-46. "The Generalized Theory of Consumer's Surplus."
Review of Economic Studies 13: 68-74.
1956. A Revision of Demand Theory. Oxford: Clarendon
Press.
Hirshleifer, Jack. 1976. Price Theory and Applications.
Englewood Cliffs, New Jersey: Prentice-Hall, Inc.
A textbook of intermediate level economic theory and
practice.
Hori, H. 1975. "Revealed Preferences for Public Goods."
American Economic Review 65: 978-991.
B-13
-------
Code
Husar, R., and White, W. 1976. "On the Color of the Los Angeles 7
Smog." Atmospheric Environment 10: 199-204.
Husar, R.B. et al. 1979. "Trends of Eastern U.S. Haziness Since 7
1948." Paper presented at the Fourth Symposium of Atmospheric
Turbulence, Diffusion and Air Pollution, Reno, Nevada,
January 15-18.
Hylland, Aanund, and Zeckhauser, Richard. 1979. "The Efficient 1, 4
Allocation of Individuals to Positions." Journal of Politi-
cal Economy 87: 293-314.
. 1979. "The Impossibility of Bayesian Group Decision 1, 4
Making With Separate Aggregation of Beliefs and Values,
Econometrica 47: 1321-1336.
ICF Incorporated. 1980. "Preliminary Assessment of Economic
Impact Visibility Regulations: Draft Report." Washington,
D.C.
Intriligator, Michael D. 1978. Econometric Models, Techniques
and Applications. Englewood Cliffs, New Jersey: Prentice-
Hall Inc.
James, T.H., editor. 1977. The Theory of the Photographic
Process. 4th Edition. New York: Macmillan & Company.
Kain, John F., and Quigley, John H. 1972. "Housing Market
Discrimination, Homeownership, and Savings Behavior."
American Economic Review 62: 263-277.
Kneese, Allen W., and Bower, Blair T. 1972. Environmental
Quality Analysis; Theory and Method in Social Sciences.
Baltimore and London: Johns Hopkins University Press for
Resources for the Future, Inc.
Kneese, A. and Williams, M. 1978. "Environmental Aspects of
Resources Policy for a Regional Setting." Department of
Economics, University of New Mexico, Albuquerque. (Mimeo-
graph.)
Knetsch, J. L. 1963. "Outdoor Recreation Demand and Benefits."
Land Economics 39: 387-396.
B-14
-------
Code
1974. Outdoor Recreation and Water Resource Planning.
Water Resources Monograph No. 3. Washington, B.C.: American
Geophysical Union.
Knetsch, J.L., and Cesario, F.J. 1976. "Some Problems in
Estimating the Demand for Outdoor Recreation: Comment."
American Journal of Agricultural Economics 58 (August):
596-597.
Krutilla, John V.; Cicchetti, Charles J.; Freeman, A. Myrick III;
and Russell, Clifford S. 1972. "Obserations on the Economics
of Irreplaceable Assets." In Environmental Quality Analysis:
Theory and Method in Social Science. Edited by Allen Kneese
and Blair T. Bower. Baltimore and London: Johns Hopkins
University Press for Resources for the Future, Inc.
A case study of Hells Canyon showing preservation values
exceeding opportunity costs and that inclusion of option
values may have increased preservation benefits.
Krutilla, John V., and Fisher, Anthony C. 1975. The Economics
of Natural Environments: Studies in the Valuation of Com-
modity and Amenity Resources. Baltimore and London: Johns
Hopkins University Press for Resources for Future, Inc.
Kurz, Mordecai. 1974. "An Experimental Approach to the Deter-
mination of the Demand for Public Goods." Journal of Public
Economics 3: 329-348.
Lancaster, Kelvin J. 1966. "A New Approach to Consumer Theory."
Journal of Political Economy 74 (April): 132-157.
Lansing, John B., and Morgan, James N. 1974. Economic Survey
Methods. Ann Arbor: University of Michigan Press.
Comprehensive analysis of survey procedures and analysis
for economics.
Latimer, Douglas A., and Daniel, Terry C. 1979. "Preliminary
Results of a Study of Human Judgements of Visial Air Quality,
Paper presented at the Air Pollution Control Association
Conference on Visibility, Denver, Colorado, November.
B-15
-------
Code
Latimer, D.; Daniel, D.A.; Hogo, H. 1980. "The Effect of
Atmospheric Optical Conditions on Perceived Scenic Beauty.
Symposium on Plumes and Visibility, Grand Canyon, Arizona,
November 10-14.
Layard, P.R.G., and Walters, A.A. 1978. Microeconomic Theory.
New York: McGraw-Hill Book Company.
Lea, S.E.G. 1978. "The Psychology and Economics of Demand."
Psychological Bulletin 85: 441-466.
Lind, R.C.; Arrow, K.J.; Corey, G.; Dasgupta, P.; Sen, A.; Stauffer,
T.; Stiglitz, J.E.; Stockfish, J.; and Wilson, R. 1982. Dis-
counting for Time and Risk in Energy Policy. Baltimore and
London: Johns Hopkins University Press for Resources for the
Future, Inc.
Locander, William; Sudman, Seymore; and Bradbury, Norman. 1976.
"An Investigation of Interview Threat and Response Distortion.
Journal of the American Statistical Association 71: 268-275.
Loehman, Edna; Boldt, David; and Chaikin, Kathleen. 1980.
"Measuring the Benefits of Air Quality Improvements in the
San Francisco Bay Area. Part I: Study Design and Property
Value Study." Draft final report. Menlo Park, California:
SRI International, September.
A hedonic property value study of the San Francisco area
as part of a replication of the South Coast Air Basin study,
discussed in Section 6.3.5 of this guidebook.
Lucas, R.C. 1964. "Wilderness Perceptions and Use: The Example
of the Boundary Waters Canoe Area. Natural Resources Journal.
3:394-411
Lucas, R.C., and Stankey G.H. 1973. "Social Carrying Capacity
for Backcountry Recreation." In Outdoor Recreation Research:
Applying the Results. Forest Service General Technical Report
NC-9, St. Paul Minnesota.
Maler, K. 1974. Environmental Economics: A Theoretical Inquiry. 1, 5
Baltimore and London: Johns Hopkins University Press.
Advanced theoretical treatise on economics of the environ-
ment .
B-16
-------
Code
Maler, K., and Wyzga, Ronald E.. 1976. Economic Measurement of 1, 5,
Environmental Damage: A Technical Handbook. Paris: Organi- 6
zation for Economic Cooperation and Development.
Discusses general valuation techniques and economic
benefit measures for a broad class of environmental damages.
Malm, William; Leiker, Karen; and Molenar, John. 1979. "Human 7
Perception of Visual Air Quality." Paper presented at the
Air Pollution Control Association meeting, Denver, Colorado,
November.
Malm, W.C.; Kelley, K.; and Molenar, J. 1980a. "Human Percep-
tion of Visual Air Quality (Regional Haze)." Symposium on
Plumes and Visibility, Grand Canyon, AZ, November 10-14.
1980b. "Human Perception of Visual Air Quality (Layered
Haze)." Symposium on Plumes and Visibility, Grand Canyon,
Arizona, November 10-14.
Marshall, Alfred. 1920. Principles of Economics. 8th edition.
London: Macmillan & Company.
McFadden, Daniel. 1977. "Quantitative Methods for Analyzing
Travel Behavior of Individuals: Some Recent Developments."
Institute of Transportation Studies Working Paper No. 7704.
University of California, Berkeley.
Mendelsohn, Robert. 1980. "An Economic Analysis of Air Pollu-
tion from Coal-Fired Power Plants." Journal of Environmental
Economics and Management 7 (March): 30-43.
Performs a benefit-cost analysis for coal-fired power
plants in the New Haven, Connecticut area. Optimum rates of
emission control are estimated drawing upon a Gaussian
dispersion model; dose response rates; estimated benefits for
acid rain, materials, agricultural and visibility aesthetics;
and estimated costs for control technologies.
Middleton, W.E.K. 1952. Vision Through The Atmosphere. Toronto:
University of Toronto Press.
Mikesell, Raymond, F. 1977. The Rate of Discount for Evaluating 5, 6
Public Projects, Washington, D.C.: American Enterprise
Institute for Public Policy Research.
B-17
-------
Code
A useful treatise which addresses the basic issue con-
cerning the social rate of discount and the application to
several types of public projects. Issues addressed include
the opportunity cost of capital, taxation, future generations,
exhaustible natural resources and inflation.
Miller, Jon, R. and Lad, Frank. 1981. "Uncertainty, Irreversibility, 1
and Quasi-Option Value: A Bayesian Approach." Paper presented at
the 1981 Western Economic Association Meetings, San Francisco,
California.
Contrary to Arrow and Fisher (1974), the paper, using
a complete Bayesian framework, demonstrates that with a
proper treatment of uncertainty in the face of irreversibility,
development decisions with quasi-option demand are not
necessarily more conservative than decisions without the
quasi-option. The paper also introduces reneg opportunities
and reneg values.
Mishan, E.J. 1972. Economics for Social Decisions; Elements of
Cost-Benefit Analysis. New York and Washington: Praeger
Publishers.
Non-mathematical discussion of the basic rational and
concepts of cost-benefit analysis, useful to readers with
basic knowledge of economics.
1976. Cost-Benefit Analysis; An Introduction. New York
and Washington: Praeger Publishers
Mood, A.M.; Graybill, F.A.; and Boes, B.C. 1974. Introduction
to the Theory of Statistics. 3rd edition. New York:
McGraw-Hill Book Company.
Muellbauer, John. 1974. "Household Production Theory, Quality,
and the 'Hedonic Technique.'" The American Economic Review
64 (December): 977-994.
A theoretical discussion that criticizes the use of
household production approach to develop price indexes that
account for quality changes and concludes that household
production theory will be useful in this respect only if the
househhold production function is nonjoint and has constant
returns to scale, which will not typically be the case.
B-18
-------
Code
Mumpower, J.; Middleton, P.; Dennis, R.L.; Stewart, T.R.; Veirs,
V. 1980. "Visual Air Quality Assessment: Denver Case
Study." Symposium on Plume Visibility, Grand Canyon,
Arizona, November 10-14.
Murphy, James L.; Fritschen, Leo J.; and Cramer, Owen P. 1970.
"Slash Burning: Pollution Can Be Reduced." Fire Control
Notes 31 (3): 3-5.
Muth, Richard S. 1966. "Household Production and Consumer
Demand Func]ions." Econometrica 34 (July): 699-708.
Nelson, Jon P- 1978. "Residential Choice, Hedonic Prices, and
the Demand for Air Quality." Journal of Urban Economics 5:
357-369.
Presents the Washington D.C. Study that is discussed in
Section 6.3.1.
Niskanen, William A., and Hanke, Steve H. 1977. "Notes: Land
Prices Substantially Underestimate the Value of Environmental
Quality." The Review of Economics and Statistics 59 (August):
375-377.
Points out that most property value studies to derive the
benefits of air pollution control have overlooked the effects
of a change in air quality on property tax revenues and have
therefore underestimated the benefits of a reduction in air
pollution.
Palmquist, Raymond B. 1981. "Measuring Environmental Effects on
Property Values Without Hedonic Regressions." Journal of
Urban Economics. (Forthcoming.)
Describes the repeat-sale technique that uses differences
in sale prices of the same homes before and after a change in
environmental quality to estimate the value of the change.
Applies it to a change in noise levels after the construction
of an interstate highway through a neighborhood in Seattle.
Patinkin, Don. 1963. "Demand Curves and Consumer's Surplus."
In Measurement in Economics: Studies in Mathematical
Economics and Econometrics in Memory of Yehuda Grunfeld.
Edited by Carl Christ et al. Stanford, California: Stanford
University Press.
B-19
-------
Paulson, N.R. 1979. "Some Perspectives on Visibility and Land
Management." Proceedings of the Workshop in Visibility
Values, Fort Collins, CO, January 28-February 1, U.S. Dept.
of Agriculture General Technical Report WO-18.
Pearse, P.H. 1968. "A New Approach to the Evaluation of Non-
Priced Recreational Resources." Land Economics 44 (February):
89-92.
Applies a variation of the travel cost method to estimate
consumer surplus valuations.
Pendse, Dilip, and Wyckoff, J.B. 1974. "Reports and Comments:
Scope for Valuation of Environmental Goods." Land Economics
50 (February): 89-92.
Pindyck, Robert S., and Rubinfeld, Daniel L. 1976. Econometric
Models and Economic Forecasts. New York: McGraw-Hill Book
Company.
Intermediate level theory and application of econometrics
text.
Polinsky, Mitchell A. 1972. "Probabilistic Compensation Criteria."
Quarterly Journal of Economics 86 (August): 407-425.
Polinsky, Mitchell A., and Shavell, Steven. 1975. "The Air
Pollution and Property Value Debate." The Review of Eco-
nomics and Statistics 57 (February): 100-104.
A criticism of the theoretical justification offered by
Anderson and Crocker (1972) for their property value study of
St Louis (Anderson, Crocker, 1971) and an initial description
of the residential location model introduced in Section
4.3.4.
1976. "Amenities and Property Values in a Model of an 1, 2
Urban Area." Journal of Public Economics 5: 119-129.
Presents the theoretical residential location model that
was introduced in Section 4.3.4 in an effort to explain the
relationship between amenities and property value with
emphasis on the effect of different levels of household
mobility.
B-20
-------
Code
Polinsky, Mitchell A., and Rubinfeld, Daniel L. 1977. "Property 1, 2
Values and the Benefits of Environmental Improvements:
Theory and Measurement." In Public Economics and the Quality
of Life. Edited by Lowdon, Wingo, and Evans. Baltimore and
London: Johns Hopkins University Press for Resources for the
Future, Inc.
Presents a theoretical discussion of the residential location
model that was introduced in Section 4.3.4 and describes its applica-
tion in the St. Louis metropolitan area.
Pollak, R., and Wachter, M. 1975. "The Relevance of the House- 1, 4
hold Production Functions and Its Implications for the
Allocation of Time." Journal of Political Economy 83:
255-277.
Presents a theoretical argument that applications of the
household production approach that make use of implicit
prices of characteristics require that the household produc-
tion function have constant returns to scale and be nonjoint;
similar to the conclusions of Muelbauer (1974).
Portney, Paul R. 1975. "Voting, Cost-Benefit Analysis, and 4, 6
Water Pollution Policy." In Cost-Benefit Analysis and Water
Pollution Policy. Edited by Henry M. Peskin and Eugene
Seskin. Washington, DC: The Urban Institute.
Rae, Douglas A. 1980. "Visibility Impairment at Mesa Verde 4
National Park: An Analysis of Benefits and Costs of Controlling
Emissions in The Four Corners Area." A Charles Rivers
Associates, Inc. Interim Report for the Electric Power
Research Institute, December.
Applies the ranked attribute approach to visibility
benefit analysis at Mesa Verde. See Section 4.4.1.
Randall, A.; Ives, B.; and Eastman, C. 1974a. "Bidding Games 3
for Valuation of Aesthetic Environmental Improvements."
Journal of Environmental Economics and Mangement 1: 132-149.
The original bidding method application to visual aesthe-
tics, better known as the Four Corners study. See Section
6.2.1 of this guidebook.
1974b. "Benefits of Abating Aesthetic Environmental
Damage from the Four Corners Power Plant, Fruitland, New
Mexico." Bulletin 618. New Mexico State University Agri-
cultural Experiment Station.
B-21
-------
Code
Randall, A.; Grunewald, 0.; Johnson, S.; Ausness, R.; and 3, 6
Pagoulatos, A. 1978. Estimating Environmental Damages from
Surface Mining of Coal in Appalachia: A Case Study. EPA-600/
2-78-003. Washington, D.C.: USEPA, January.
Comprehensive and integrated examination of the benefits
of surface mine reclamation. Bidding methods are used to
estimate aesthetics-related benefits.
1978. "Reclaiming Coal Surface Mines in Central Appalachia: 3, 6
A Case Study of the Benefits and Costs." Land Economics 54:
472-489.
A summary of the report on the same topic, listed above.
Randall, A., and Stoll, J. 1980. "Consumers' Surplus in Commodity
Space." American Economic Review 70 (July): 449-455.
Discusses and calculates the error in measurement incurred
by using the ordinary consumer surplus measure when ES or CS
measures are appropriate. See Chapter 2 of this guidebook.
Ridker, Ronald G., and Henning, John A. 1967. "The Determinants
of Residential Property Values With Special Reference to Air
Pollution." The Review of Economics and Statistics 49 (May):
246-257.
First attempt to use property value data to estimate the
benefits of air pollution control based on the observed
differences in property values with respect to air quality.
They found a statistically significant relationship between a
measure of sulfur oxides and property values in St. Louis,
although their interpretation of property value differentials
as directly reflecting benefits of pollution control was
later criticized.
Roberts, E.M.; Gordon, J.L.; Hoase, D.L.; Kary, R.E.; and Weiss,
J.R. 1974. "Visibility Measurements in the Painted Desert
through Photographic Photometry." Bulletin No. 47- Denver,
Colorado: Dames and Moore Engineering.
Robinson, E. 1970. "Effect on the Physical Properties of the
Atmosphere." In Vol. 1: Air Pollution. Edited by A. Stern.
Academic Press.
B-22
-------
Code
Rosen, Sherwin. 1974. "Hedonic Prices and Implicit Markets: 1, 2
Product Differentiation in Pure Competition." Journal of
Political Economy 82 (January/February): 34-55.
A technical discussion of the equilibrium conditions in a
market for a differentiated product. The theory of hedonic
prices and potential applications to estimating demand for a
characteristic of a product are consistant with Freeman
(1974) and have been the basis of many property value studies
that have estimated willingness to pay for air quality, as
described in Section 4.3.3 of this guidebook.
. 1979. "Wage-based Indexes of Urban Quality in Life." In 1, 4
Current Issues in Urban Economics. Edited by Peter Mieszkowski
and Mahlon Straszheim. Baltimore and London: Johns Hopkins
University Press.
Estimates a hedonic price function for wages across
several cities from which values of air quality are derived.
Discussed in Section 4.4.4 of this guidebook.
Rowe, R.; d'Arge, R.; and Brookshire, D. 1980a. "An Experiment
on the Economic Value of Visibility." Journal of Environmen-
tal Economics and Management 7 (March): 1-19.
Summary article on the Farmington study focusing upon
biases in the bidding method. See Section 6.2.3 of this
guidebook.
. 1980b. "Environmental Preference and Effluent Charges."
In Progress in Resource Management and Environmental Planning.
Edited by Timothy O'Riordin and R. Kerry Turner. London:
John Wiley & Sons, Ltd. Series in Environmental Economics.
Summary of the Farmington study; focuses upon aggregate
benefits and policy applications for emissions control. See
section 6.2.3 of this guidebook.
Rowe, Robert D. and Blank, Frederick M. 1981. "Deviations
In Empirical Consumers' Surplus Measures." Paper presented at
the Fifty-Sixth Conference of the Western Economics Association,
San Francisco, California.
Addresses a broad spectrum of arguments as to why
empirical ES and CS measures have substantially deviated and
the implications for measure selection and application of
benefit estimation techniques.
B-23
-------
Code
Samuelson, P.A. 1954. "The Pure Theory of Public Expenditures."
Review of Economics and Statistics 36 (November): 387-389.
Sandier, Todd, and Smith, V. Kerry. 1976. "Intertemporal and 1
Intergenerational Pareto Efficiency." Journal of Environ-
mental Economics and Management 2 (February): 151-159.
Efficiency conditions are derived for both public and
private goods which provide benefits over time.
. 1977. "Intertemporal and Intergenerational Pareto 5, 6
Efficiency Revisited." Journal of Environmental Economics
and Management 4: 252-257-
Sassone, Peter G., and Schaffer, William A. 1978. Cost-Benefit
Analysis: A Handbook. London: Academic Press.
A thorough, step-by-step handbook discussing the philosophy
and application of benefit cost analysis.
Scherr, Bruce A., and Babb, Emerson, M. 1975. "Pricing Public
Goods: An Experiment with Two Proposed Pricing Systems."
Public Choice 23: 35-48.
Analyzes bidding behavior for a concert and library fund
under three alternative payment schemes. They could not
reject that any scheme inhibited "free rider" behavior.
Also, mean bids under the three schemes were significantly
different.
Schmalensee, R. 1972. "Option Demand and Consumer's Surplus:
Valuing Price Changes under Uncertainty." The American
Economic Review 62: 813-824.
Theoretical paper which shows that consideration of
alternative project site uses and the consideration of risk
preferences may lead to both positive or negative option
values for a specified project.
Schulze, W., and d'Arge, R. 1977. "On the Valuation of Recrea-
tion Damages." Paper presented at the American Economic
Association, December.
B-24
-------
Code
Silberberg, Eugene. 1972. "Duality and the Many Consumer's 1
Surpluses." American Economic Review 5: 942-952.
Sinden, J.A. 1974. "A Utility Approach to the Valuation of 1, 4
Recreational and Aesthetic Experiences." American Journal of
Agricultural Economics 56 (February): 61-72.
Sinden, J.A. 1979. Unpriced Values. New York: John Wiley and
Sons.
Small, Kenneth A., and Rosen, Harvey S. 1981. "Applied Welfare
Economics with Discrete Choice Models." Econometrica 49
(January): 105-130.
Analyzes methods of applied welfare economics in cases of
discrete choice using advanced theory.
Smith, Barton A. 1978. "Measuring the Value of Urban Amenities."
Journal of Urban Economics 5: 370-387.
Derives a measure of land value premiums from differences
between land prices at different locations and using data
from Chicago; finds a significant negative relationship
between this location premium and air pollution in a hedonic
price function.
Smith, V. Kerry. 1975. "The Estimation and Use of Models of the 1, 4
Demand for Outdoor Recreation." In Assessing Demand for
Outdoor Recreation. Washington, D.C.: U.S. Department of
Interior, Bureau of Outdoor Recreation.
1976. The Economic Consequences of Air Pollution.
Cambridge, Massachusetts: Ballinger Publishing Company.
Pulls together several articles on economic approaches
to quantifying the damages of air pollution with emphasis on
property value studies; includes the results of a cross-city
property value study that was based on the residential
location model.
Smith, V. Kerry, and Deyak, Timothy A. 1975. "Measuring the
Impact of Air Pollution on Property Values." Journal of
Regional Science 15: 277-288.
B-25
-------
Code
Similar to the intercity study by Deyak and Smith (1974)
but with the addition of tax and public service characteris-
tics of each city and using central city data. The air
pollution measure was not statistically significant in
explaining property value differences across cities. This
study is also presented in Smith (1976).
Smith, Vernon L. 1977. "The Principle of Unanimity and Voluntary 3, 4
Consent in Social Choice." Journal of Political Economy 85:
1125-1139.
. 1980. "Experiments With a Decentralized Mechanism for 3, 4
Public Good Decisions." American Economic Review 70
(September): 584-599.
Reports on a laboratory experiment demonstrating the
possibility of designing decentralized mechanisms for optimal
public good decisions which exhibit minimal or no free rider
behavior.
Sonstelie, Jon C., and Portney, Paul R. 1977. "Gross Rent and a 2
Reinterpretation of the Tiebout Hypothesis." (Mimeograph.)
A property value study of San Mateo County, California.
The pollution variable was found to be negative and statis-
tically significant.
Stankey, G.H. 1972. "A Strategy for the Definition and Manage- 5, 7
ment of Wilderness Quality." In Natural Environments;
Studies in Theoretical and Applied Analysis. Edited by J.V.
Krutilla. Baltimore: Johns Hopkins Press.
. 1973. "Visitor Perception of Wilderness Recreation 7
Carrying Capacity." Forest Service Paper INT-142. Ogden Utah.
Straszheim, Mahlon. 1974. "Hedonic Estimation of Housing Market
Prices: A Further Comment." The Review of Economics and
Statistics (August): 404-406.
Argues that there is evidence of distinct submarkets in
the housing market in San Francisco and that hedonic property
value studies based on data that is pooled across an entire
urban area may be obscuring the true supply and demand
structure.
B-26
-------
Code
Thayer, M. 1980. "Contingent Valuation Techniques for Assessing
Environmental Impacts: Further Evidence." Journal of Environ-
mental Economics and Management 8 (March): 27-44.
Uses a contingent market questionnaire to examine recrea-
tion site substitution and changes in expenditures related to
potential impacts of geothermal energy development in the
Jenez Mountains, New Mexico.
Tideman, Nicolaus T., and Tullock, Gordon. 1976. "A New and
Superior Principle for Collective Choice." Journal of
Political Economy 84 (November/December): 1145-1159.
Trijonis, J. 1979. "Visibility in the Southwest: An Explora-
tion of the Historical Data Base." Atmospheric Environment
13: 833.
U.S. Congress. Senate. 1977. The Clean Air Act as Amended
August 1977. 95th Congress, 1st Session, Committee Print,
Serial No. 95-11.
U.S. Department of Agriculture. 1977- Forest Service. Outdoor
Recreation: Advances in the Application of Economics.
Proceedings of a National Symposium, March.
U.S. Department of Commerce. 1975. "Surface Observations."
Meteorological Handbook. No. 1. National Oceanic and
Atmospheric Administration, Silver Springs, Maryland.
U.S. Environmental Protection Agency. 1975. Guidelines of
Reclassification of Areas under EPA Regulations to Prevent
the Significant Deterioration of Air Quality. June.
. 1978. The Development of Mathematical Models for the
Prediction of Anthropogenic Visibility Impairment^EPA-45013-
78-110 a, b, c.
. 1979. Protecting Visibility: An EPA Report to Congress. 7, 9
EPA-450/5-79-008.
. 1980a. Draft Interim Guidance for Visibility Monitoring.
Las Vegas, Nevada: Environmental Systems Monitoring Laboratory,
July.
B-27
-------
Code
1980b. Proposed Guidelines for Determining Best Available
"Retrofit Technology for Coal-Fired Power Plants and Other
Existing Stationary Facilities. EPA-450/3-80-009, March.
. 1980c. Visibility Protection for Federal Class I Areas.
"45 FR 80084-80095. December.
Final rulemaking for Phase I protection of clean air in
Class I areas authorized under the Clean Air Act as amended
1977-
. 1980d. Workbook for Estimating Visibility Impairment.
EPA-450/4-80-031.
. 1980e. User's Manual for the Plume Visibility Model
(PLUVUE). EPA-450/4-80-032.
. 1980f. Criteria for the Identification of Integral Vistas.
Draft.
. 1980g. Assessment of Economic Impacts of Visibility 6, 9
Regulations.EPA-450/2-80-084, November.
Revision of the IGF (1980) report on the costs of the
1980 visibility regulations. Points out only one power
plant, the Four Corners plant, would be affected; however,
this plant will no longer be affected once it reduces emis-
sions in accordance with the New Mexico State Implementation
Plan.
. 1981. The Benefits of Preserving Visibility in the
National Parklands of the Southwest. Forthcoming.
U.S. National Park Service. 1980. Results of the NPS Visitor 3
Survey Conducted at Bryce Canyon National Park, Summer 1980.
September.
U.S. Nuclear Regulatory Commission 1980. The Visual Aesthetic 3, 6
Impact of Alternative Closed Cycle Cooling Systems. NUREG/CR-
0939, PNL-2952.
B-28
-------
Code
U.S. Water Resources Council. 1979. "Procedures for Evaluation 6, 9
of National Economic Development (NED) Benefits and Costs in
Water Resource Planning (Level C) and Proposed Revisions to
the Standards for Planning Water and Related Land Resources."
18 FR Part 704. May 24.
Vickerman, R.W. 1974. "The Evaluation of Benefits from Recrea-
tional Projects." Urban Studies 11: 277-288.
Walther, Eric G. 1980. "Making Visibility Degradation Visible." 7, 9
Environmental Impact Assessment 1: 78-80.
Walther, E.G., and R.M. Newburn. 1980. "Standard Visual Range
Measured in the NPS/EPA Regional Visibility Network from
December 1978 through August 1979." University of Nevada,
Visibility Research Center, Las Vegas, Nevada.
Wennergren, E.B., and Fullerton, H.H. 1972. "Estimating Quality
and Location Value of Recreational Resources." Journal of
Leisure Research 4: 170-83.
Wieand, Kenneth F. 1973. "Air Pollution and Property Values: A
Study of the St. Louis Area." Journal of Regional Science 13
(April): 91-95.
Argues that in a property value study of the St. Louis
area, rent per acre is a better measure of property price
than total rent. Found that with the same data used by
Ridker and Henning (1967), the coefficients of the pollution
variables were not significantly negative when the dependent
variable was changed to rent per acre.
Williams, M.D. 1979. "Plume Blight Visibility Modeling with a
Simulated Photograph Technique." Paper presented at "View on
VisibilityRegulatory and Scientific," Air Pollution
Control Association Specialty Conference, Denver, Colorado,
November 26-27.
Willig, Robert D. 1976. "Consumer's Surplus Without Apology."
American Economic Review 66 (September): 587-597.
Discusses and calculates the measurement error incurred
by using the ordinary consumer surplus measure when EV and CV
measures are appropriate. See Chapter 2 of this guidebook.
B-29
-------
Code
Wilman, Elizabeth A. 1981. "Hedonic Prices and Beach Recrea-
tional Values." In Advances in Applied Microeconomic Theory.
Edited by V.K. Smith. Greenwich, Connecticut: JAI Press.
(Forthcoming.)
Uses the hedonic technique to estimate recreational value
of beaches to renters of vacation property on Cape Cod. Marginal
implicit prices for several beach quality attributes are derived,
and supply and demand equations for beach quality are estimated.
Witte, Ann D.; Sumka, Howard J.; and Erekson, Homer. 1979. "An
Estimate of a Structural Hedonic Price Model of the Housing
Market: An Application of Rosen's Theory of Implicit Markets."
Econometrica 47 (September): 1151-1173.
Implicit prices for housing attributes (air quality is
not used) are derived from a hedonic price function and used
to estimate bids and offer functions for the housing attributes.
Wyszecki, G., and W.S. Stiles. 1967- Color Science. New York
and London: John Wiley & Sons.
B-30
-------
APPENDIX 1
Sample Bidding Method Questionnaires and Picture Sets
Page
A. The Lake Powell Study A-l
B. The Farmington Study A-3
C. The South Coast Air Basin Study A-14
D. The Grand Caynon/Southwest Parks Study A-37
E. Water Resource Council Suggested Contingent
Market Valuation Techniques to Estimate
Recreation Benefits
A-41
-------
A. THE LAKE POWELL STUDY
BKOOKSlIlRi:, 1VES AND SCIIULZE
UUinK '-I'M- lt»r l-tlm-KK f KC. rt -u lunlnt <" :>vn-*nj
for Ah.ilntrnl -! A*-%(Wt !< l.nvl rnnnvnt*! ()«>nr
Good Hornlnf/A
ThU
tion.
1. Mow »*ny MB
2. Wh«t ! th«
1. Vh«r« «r« yo
(c)
4. If you don't aln^. couU you flm»m9 Indlcat* which of [K« followlni brcckcti yovr Immllj Intom* f«llt Into:
_ _ _ 0 - 4,»« _ 20.000 - ?*.»M
_ ' J.OOO - t.91» _ 21.000 - J».«M
__ ^^.lO.OOO - 1*.»9« _ 30.0OO - 41. 1«
15,000 - 1».W _ 50.000 .«-
TH«r« .r«
to « it
of Ch« «n«lron*«nt. If thr
Uk«. If »if pollutloo L«
llttl* »ok«;
ubitant1*1IT
rir»i. Uc1
to W «dktt
will b* u»rd to (In
t h« ount.)
S/d«y
8.
-------
Silimlioii It.
Sim:ili
-------
B. THE FARMINGTON STUDY
QUESTIONNAIRE - BIDDING GATE
Estimation of Aesthetic Damages of Air Pollution from Coal Fired Power Plants
Good morning/afternoon. My name is (PCESENT ID). I an from the
Department of Economics at the University of Wyoming. We are doing research
under a grant from the Electric Power Research Institute which is funded by
the power plants across the nation. What v;e are attempting to do is measure
the value of better visibility associated with increased emission control
from coal-fired electric generating plants. In order for us to do this
would you be willing to answer some questions about visibility in this area?
PART I
1. JOINT QUESTIONNAIRE SINGLE QUESTIONNAIRE .
2. Have you recently completed any other questionnaire for the University
Of Wyoming? yes no. (IF YES AND NOT A JOINT QUESTIONNAIRE THANK
THEM FOR THEIR TII1E AND EXCUSE YOURSELF.)
3. Sex: male female.
A. Marital Status: single married other (SPECIFY).
5. Number of persons in your household/group: .
6. Occupation: Name of Employer: .
71. Are you a resident of the Four Corners area? resident nonresident.
ii. (FOR NONRESIDENTS)
A. Where are you staying? a) lodge or motel b) local residence
c) developed or semi-developed campground d) other.
B. What state and county are you frosi?
C. Are you on vacation ______ or business ?
8. How long have you been in the Four Corners area? .
9. Do you have regular access to an automotive vehicle? yes no.
10. If children are living with you what is the age of your youngest child?
yes children no children age of younpest child.
Thank you.
As you may know, there are presently two power plants in the Four Corners area.
These plants use coal to generate electric pox^er. The need for maintaining and
perhaps increasing the nation's energy resources is represented by these plants.
Coupled with further cr.ergy development, however, is the problem of stack
emissions from power plants. In order to study the possibility of overspending
or underspending on emissions control by power cotrpanica the Electric Power
Research Institute has commissioned this effort. The purpose is to determine
A-3
-------
QUESTIONNAIRE - BIDDING GAME
how individuals are affected by visibility changes induced by power plant
emissions.
(PRESENT THE PHOTOGRAPHS) These photographs are designed to show varying
degrees of visibility. Column A depicts a situation with high visibility,
while B and C show decreasing levels of visibility due to increased emissions
and dust in the air. Column D depicts the visible stacks and emissions of
the plants. Uhile the pictures may include some windblown dust, they repre-
sent typical levels of visibility that could occur if the power plants
reduced their emission control. Each column represents the same visibility
for two different views: Shiprock to the west and the San Juan Mountains
to the north.
PART II. EQUIVALENT VARIATION
Clearly, everyone desires cleaner air, however, air quality improvements
entail considerable expenditures. Let us propose a mechanism to continue
to finance improvements in visibility.
(FOR, RESIDENTS) Suppose the method used to finance emission control will be
through ( ) additional monthly charges on your utility bill ( )
additional payroll deductions (CHECK MUCH) and everyone would have to pay
the sane amount.
(FOR NONRESIDENTS) Suppose the method used to finance emission control will
be through additional fees to parks, campgrounds and a surtax on gas, food
and lodging. (ALSO SUBSTITUTE DAYS FOR MOUTIIS.)
(DO EITHER 1,2,3,4 or 3a,l,2,4 SEQUENTIALLY. CHECK THE ONE USED.)
la) With this in mind, would you be willing to pay $ (USE STARTING BID
OF $1,5, .or 10) per month (extra on your utility bill) (as a payroll
deduction) to have visibility as depicted in Column A rather than as
depicted in Column B? (GO THROUGH BIDDING PROCESS.)
b) (PEPEAT CHANGING COLUMN B TO COLUMN C.)
c) (PEPEAT COMPARING B AND C.)
d) Similarly the power plants and their
ORIGINAL BIDS
emissions may be out in the open, such
la. $
as a combination of Columns C and D
depicts, or hidden from view, for
A-4
b. $
c. $
d. $
-------
QUESTIONNAIRE - BIDDING GAME
Instance behind a mesa, and the emissions level could be as in Colurai A.
(REPEAT BIDDING GAME COMPARING THE COMBINATION OF C AND D WITH A.)
2. Did you bid zero because you believe that: (HAND RESPONDENT THE AUXILLIARY
SHEET WITH ALTERNATIVES. MORE THAN ONE ANSWER IS ACCEPTABLE BUT NOT
ENCOURAGED. RECORD ANSWER HERE:) .
3a. Suppose you were told that the average amount people were willing to pay
in other studies of similar circumstances was:
$1.25 to have typical visibility as depicted in Column A rather than
as depicted in Column B
$1.50 to have typical visibility as depicted in Column A rather than
as depicted in Column C
$ .25 to have typical visibiltty as depicted in Coluinn B rather than as
depicted in Column C.
$1.75 (IF 3a IS DONE FIRST READ Ic) to have visibility is depicted in
Column A rather than as depicted in the combination of Columns
C and D.
3o-. i) G>ven this information, what would you be willing to pay to have
visibility as in A rather than
as in B?
ii) Given this information, what
would you be willing to pay to
iv. $
have visibility as in A rather
REVISED BIDS
3b. i. $
ii. $"
iii. $
than as in C?
iii) Given this information, what
would you be willing to pay to
have visibility as in B rather than as in C?
iv) Given this information, what would you be willing to pay to have
visibility as in A rather than as in C + D?
3c. Why did/didn't you change your bids?
A-5
-------
QUESTIONNAIRE - BIDDING GAlffi
A. If your bid to change the visibility from that depicted in C to that
shown in A actually becomes the average bid in this area, an insufficient
amount of money would be collected to actually finance the changa. Would
you be willing to raise your bid? no
(IF YES) To what?
A-6
-------
QUESTIONAIHE - BIDDING GATE
PART III
As an alternative to controlling emissions, the power companies could simply
compensate or pay people for any damages inflicted upon them Ossociated with
reduced visibility. This compensation could be accomplished by having the
power companies:
(FOR RESIDENTS) reduce utility rates for the people in the area affected.
(FOR NONRESIDENTS) subsidize reductions in lodging expenses and entrance fees
to parks, reservoirs, and lodging in this area.
(DO EITHER 1,2,3,4 QR 3a.l.2.4 ' SEQUENTIALLY, CHECK THE ONE USED.)
la. Would you be willing to accept a $ (USE STARTING BID OF $1,5, or 10)
reduction per nonth (in your utility bill) (in park, reservoir and
lodging fees) to allow typical visibility as in Column B rather than
as in Column A?
b. (REPEAT a. CHANGE COLUMN B TO COLUliK C)
c. (REPEAT a. CHANGING B TO C AND A TO B.)
d. Again the power plants and their emissions
may be out in the open, such as in
ORIGINAL BIDS
Column D with emissions level C or
la. $
b. $
c. $"
d. $
hidden as in Column A with almost
no emissions. (REPEAT a. CHANGE B
TO COMBINATION OF C + D.)
(IF ZERO BID)
2i. Did you bid zero because you believe that (HAND RESPONDENT THE AUXILLIARY
SHEET. MORE THAN ONE ANSWER IS ACCEPTABLE BUT MOT ENCOURAGED. RECORD
ANSWER HERE:)
_^ _
(IF INFINITE BID)
2ii. You have said you could not be compensated enough to accept the level
of emissions depicted in Column . Could you be compensated by a lump
sum sufficient to relocate to a similar location in Mew Mexico, Colorado,
A-7
-------
QUESTIONNAIRE - BIDDING GAME
Arizona, or Utah? yes no.
If yes, what would be sufficient compensation: (BID DOWN BY
10% INCREMENTS.)
If no, what would you do if situation B (c) were the case? .
(ATTEMPT TO CONVERT THIS TO A $ FIGURE.)
3a. Suppose you were told that the average payment people were willing to
accept, found in other studies of similar circumstances was actually:
$2.25 per month to just compensate them for damages associated with the
reduced visibility as depicted in Coluim B rather than A.
$2.50 per month to just compensate them for damages associated with
the reduced visibility aa depicted in Column C rather than A.
$ .25 per month to just compensate them for damages associated with
the reduced visibility as depicted in Column C rather than B.
$3.75 per month to just compensate them for damages associated with
the change in visibility associated with the combination of
Columns C + D, rather than A.
3b. i. Given this information, what now would be your bid to allow
visibility as in Column B
rather than as in Column A?
ii. (SUBSTITUTE C FOR B ABOVE)
REVISED BIDS
3b.
i. $
ii. $"
iii. $
iii. (SUBSTITUTE C FOR B AND B FOR A ABOVE)
iv. (SUBSTITUTE D FOR B ABOVE)
3c. Why did/didn't you change your bid?
A. As another way to estimate compensation, what is the minimum amount of
recreational activity of your choice, such as a trip or outing, that you
would accept to compensate you for any damages inflicted upon you due to
the reduced visibility of Colum B?
Column C?_
Column D?
(IF THE RESPONDENT HAS TROUBLE ANSWERING, GIVE HIM AUXILLIARY SHEET WITH CHOICES
IN EACH CASE ASK WHAT IS THE VALUE TO THIS?)
A-8
VALUE OF RECREATION
i. $
ii. $
iii. $
-------
QUESTIONNAIRE - BIDDING GAME
PART IV.
QUESTIONS 1, 2, AND 3 REQUIRE THE RESPONDENT TO HAVE THE AUXILIARY SHEET
1. If you don't mind and for the purpose of this study only, could you please
give me the letter corresponding to the category into which your annual
income would fall?
2. Could you please give me the letter corresponding to your educational
background?
3. Could you please give me the letter corresponding to your ethnic
background?
4. Have you contributed either time or money to any environmental cause or
organization recently? yes no. IF YES, about how much?
For the following questions, would you respond as best as possible with either:
not at all, to some degree, or greatly.
5. In your opinion, how much do you believe the development of the local power
plants has positively affected the business of your employer?
not at all to some degree greatly
6. How has the power plant development positively affected your career?
not at all to some degree greatly
7. Do you think a typical daily visibility similar to Column C, if caused
by emissions, would:
a) shorten your life? not at all, to some degree, greatly
b) require you to spend more money on drug items or doctors' fees?
not at all, to some degree, greatly
c) make it harder to do your work? not at all, to some degree,
greatly
d) make your life less pleasant? not at all, to some degree*
greatly
8. On a scale of 1 to 5, where 1 represents conservative, 3 moderate, and
t
5 liberal, where would you place yourself politically on a national level?
9. On a scale of 1 to 5, where 1 roproaeuta coiiaurvatl oniat and 5 developer,
where would you place yourself envirunroo.ntnlYy c.iv a nut-liAknl. le-.-Jol7
A-9
-------
QUESTIONNAIRE - BIDDIIIG GAME
PART V.
COMPLETE THIS SECTION IKMDEIATELY AFTER THE INTERVIEW BUT APART FROM THE
INTERVIEWEE'S PRESENCE. YOU, THE INTERVIEWER, ARE TO AN SITE R THE QUESTIONS,
neatness of house interior: sloppy, fairly neat, did not see.
neatness of house exterior: sloppy, fairly neat, did not see.
appr .ximate cost of home: $ . ADDRESS OF HOUSE:
Write a couple of sentences on your impression of the interviewee's veracity
and the extent to which he had his wits about him.
Does the interviewee seem to have vision problems?
If yes, why?
Did the interviewee lose patience? At what point did this loss of patience
first become noticeable?
NAIffi OF INTERVIEWS R_
Date of interview
Location of interview
Time of interview
A-JO
-------
-------
A-12
.i,.-^.-, '..'._->. ._-
- ,:f, i .
V'~./i>^v"
- j
«r^. .
'"-r'*t- >.
,- X 'T,
A..
-------
r
,' .'"'..'.--"" -" "," *
A-13
-------
C. South Coast Air Basin Study
This section presents the basic survey instrument employed in the
South Coast Air Basin Study. As discussed, the initiation point for
the survey instrument was either acute health effects or aesthetic effects.
Three basic areas existed (good, fair, bad). Thus the following combina-
tions existed for survey instrument types.
to B (Aesthetic)
to B (Acute)
to C (Aesthetic)
to C (Acute)
to C (Aesthetic)
to C (Acute)
(Aesthetic)
(Acute)
1.
2.
3.
4.
5.
6.
7.
8.
Format
Format
Format
Format
Format
Format
Format
Format
for
for
for
for
for
for
for
for
A
A
A
A
B
B
C
C
area
area
area
area
area
area
area
area
moving
mov ing
moving
mov ing
mov ing
moving
to C*
to C*
The structure of the different combination was identical.
1 is presented for illustrative purposes.
Comb inat ion
A-14
-------
Table A.I
IMDOOR ACTIVITY AND COST LIST
I
i'
Ln
Activity
Indoor Spectator Event!
Indoor Tennis
Raquelball, lUndball
Table Tuinls
Bow] Ing
Indoor Gardening or
Flxlnp, up House
Central Exercise
Organizer! Shorts Evonta
R c -i d i r ^
Te it >M 3 Ion
Mov Itt
Club Acl Ivtt las.
Orfi^nlzn clous
Indlvldu.il Sports
Sw 1 in 1 n^
Visiting Neighbors or
Fr IcnJa
Other (specify)
V1
Hours
Per Week
A
B
C
0
Tinea
Per Week
A
B
C
D
Location
(Hap Grid)
A; B
C
0
Miles
Traveled
A; e , cj P
Direct
Costs
A
B
C
0
1
i
I '
1
1
i
Z Day
Equipment
Replacement
Costs
laportincc
-------
Table A.2
OUTDOOR ACTIVITY AND COST LIST
Activity
Outdoor Spectator
Sports
Tennis
BlUng
Be.ieh Activities
Centre! Ererclse
Flslilnv;
Su lr\.T\lnz
Sa 11 Inj
Joi'gln,'/'..'.jlklng
HobbUs, Arts & Crafts
Outdoor Gardening or
Ftxln,- up Mouse-
Col £
Hiking
Canplng
V1
Organized Sports Events '
Individual Sports
Events
Other (spcclfv)
Hours
Per Week
A
B
C
D
1
Tinea
Per Week
A
B
C
D
Location
(Map Grid)
A
B
C
D
Miles
Traveled
t.
B
C
D
Direct
Costs
A
B
C
D
X Day
Equlpoenc
Replacement
Costs
Importance
-------
GAME FOR A AREA MOVING TO B, AESTHETIC Questionnaire #
Interviewer //
INTRODUCTION
1. In a typical week how much day and night leisure time do you have available?
This includes both weekdays and weekends. By leisure, I mean the time you
do not spend eating, sleeping, or working to earn a living. hours
2. Has air pollution influenced where you have chosen to live? Yes[ ] No[ J
3. Has air pollution influenced where you have chosen to work? Yes[ ] No[ ]
4. Would you classify the air quality in the area where you live as:
Good[ 1 Fair[ ] Poorf ]
5. Would you classify the air quality in the area where you work as:
Good[ ] Fair[ ] PoorI ]
5a. What is your occupation?
6. Are you aware of any health hazards due to air pollution? Yes[ ] No[ ]
7. How long have you lived in the Los Angeles area?
8. How much, longer do you plan to live in the los Angeles area?
9. Do you think automobile emission standards should be: Increased[ ]
Decreased[ ] Kept the Same[ ]
ADMINISTER INDOOR AND OUTDOOR ACTIVITY AND COST LISTS (TYPICAL WEEK) CHECK
THE TOTAL TIME CONSTRAINT
Bidding Game for Residents of Area A
Here are three photographs representing average levels of visibility
for the three different regions of the Los Angeles Area shown on this
map. Picture A represents poor visibility; Picture B represents fair
visibility; and Picture C represents good visibility.
Public officials are strongly considering the possibility of trying to
reduce the levels of emissions throughout the Los Angeles Area. Such
action could require additional funds which might be generated by (a
monthly charge, an extra charge in your utility bill) for as long as
you live in the Los Angeles area. These funds will be used to help
finance air quality improvements in the Los Angeles area.
-------
Aesthetic
The Los Angeles area has some very Beautiful background scenery.
Because of automobile and industrial emissions, there is a haze which
reduces and distorts the ah'i'lity to see this scenery. This means that
many people have to leave Los Angeles and travel long distances to be
able to enjoy views which could be visible from their homes if these
emissions were reduced.
As indicated by the map, you live in an area which has been classified
as having poor air quality relative to the rest of the Los Angeles
area. Picture A represents the visibility level which typically occurs
in your area. I am only interested in how you value being able to see
long distances.
If the level of emissions could be reduced in the Los Angeles area so
that visibility conditions would be represented by Picture B instead of
A, not only in the B area but also in your area, and if the air would
be cleaned up to this level in (2, 10) years, would you pay (a monthly
charge, an extra charge in your utility bill) of C$1, $10, $50) for as
long as you live in the Los Angeles area?
[ ] DO FOLLOWING ONLY IF CHECKED
LIFE TABLE: Here is a table that might help you. It shows the total amount
you would pay for as long as you live in Los Angeles for various amounts of
monthly payments.
RECORD MAXIMUM BID
Would you consider moving to a new location in the los Angeles area
if air quality were like Picture C everywhere? Yes[ ] No[ ]
IF YES:
Where would you move? (GRID LOCATION ON MAP)
A-18
-------
Bidding - Aesthetics and Acute Health.
The questions I asked earlier were only concerned with, your perception
of visibility. This section deals with, not only visibility but with.
short term health effects which may be aggravated by air pollution.
Some pollutants in high concentrations cause eye irritation for many
individuals. Studies have shown that about half of the population
experiences eye irritation under conditions represented by Picture A;
about onefourth experiences eve irritation under conditions represented
by Picture F; while no one experiences- these irritating effects when
conditions are represented by Picture C.
Since you reside in area A, which has been classified as having poor
air quality, there is reduced visibility as well as irritating health
effects as compared with F. If the level of emissions could be reduced
in the Los- Angeles area so that visibility and irritating health effects
were represented by Picture F not only in the F area but also in your
area, and if the air would be cleaned up to this level in (2, 10) years,
would you pay a (a monthly charge, an extra charge in your utility bill)
of CSTART FIDDING WITH PREVIOUS MAXIMUM BID) for as long as you live
in the Los Angeles area?
[ ] DO FOLLOWING ONLY IF CHECKED
Life Table: Here is a table that might help you. It shows the total amount
you would pay for as long as you live in Los Angeles for various amounts of
monthly payments.
RECORD MAXIMUM BID
REVISIONS IF NECESSARY
Would you consider moving to a new location in the Los Angeles area
if air quality were like Picture C everywhere? Yes[ ] No[ ]
IF YES:
Where would you move? (GRID LOCATION ON MAP)
-19
-------
Substitutions
Aesthetic; Original Position A; Movement to B
NON-ZERO BIDS:
You stated that you would pay $ per month for as long as you live
in the Los Angeles area if visibility improved from A to a condition
like that shown in B, if this could be accomplished in (2, 10) years.
If you actually paid this amount in order to help finance air pollution
control programs-, you would have les-s money to spend overall. However,
because you said you would pay some amount of money, you are indicating
that clearer air is something you value. Consequently, when conditions
like B are achieved, even though you have paid money to help improve
the visibility, this does not mean that your standard of living is
worse than before because now you will be living and recreating in a
less polluted area.
If the visibility conditions were to improve from A to B in your area,
would the improved conditions change the pattern of your leisure
activities? This could be changes in time per week, location, or
frequency. Yes[ ] No[ ]
IF NO, GO TO NEXT BIDDING GAME
IF YES, THEN:
A. Administer indoor and outdoor activity and cost list
B. Check time constraint
ZERO BIDS:
Although you told me that you would not pay anything to have visibility
improve throughout the Los Angeles area to like that shown in Picture B,
would the improved conditions change the pattern of your leisure
activities? This could be changes in time per week, location, or
frequency. Yes[ ] No[ ]
IF NO, GO TO NEXT BIDDING GAME
IF YES,.THEN:
A. Administer indoor and outdoor activity and cost lists
B. Check time constraint
A-20
-------
Substitutions
Aesthetic + Acute; Original Position A; Movement to K
NON-ZERO BIDS:
With the extra information on possible short term health, effects when
conditions are like A, you said that you would pay $ per month.
for as long as yon lived in the Los Angeles area if conditions im-
proved from those associated with Picture A to conditions shown in
Picture B, and if this coold be accomplished in (2, 10) years. As
before, Because you said you would pay some amount of money, you are
indicating that clearer air is something you value. Consequently, when
conditions like A are achieved, even though you have paid mqnay to help
improve the visibility and to lessen short term health effects, this
does not -mean that your standard of living is worse than before because
now you will be living and recreating in a less polluted area.
If conditions improved so that Picture B were representative of the
entire area, with no visibility problems or irritating effects, would
the improved conditions change the pattern of your leisure activities?
Yes[ ] No[ ]
IF NO, GO TO NEXT BIDDING GAME
IF YES, THEN:
A. Administer indoor and outdoor activity and cost lists
B. Check time constraint
ZERO BIDS:
Although you told me you would not pay anything to have visibility
conditions and short term health effects improve throughout the area
to like those shown in Picture B, would the improved conditions change
the pattern of your leisure activities? Yes[ ] No[ ]
IF NO, GO TO NEXT BIDDING GAME
IF YES, THEN:
A. Administer indoor and outdoor activity and cost lists
B. Check time constraint
A-21
-------
Substitutions
Aesthetic + Acute + Chronic; Original Position A; Movement to K
NON-ZERO BIDS:
Given the information that continued exposure to levels of air pol-
lution like those shown in Picture A could actually reduce your life
expectancy, yoa said you would pay $ per month for as long as you
lived in th~e Los- Angeles- area if conditions improved from those in A to
those shown in E, and if this- could Ke accomplished in C2, 10) years.
Once again, I would like you to thinR of this expenditure as leaving
you as well off as Before yoa paid the -money, since you are now living
and recreating in a less- polluted area.
If conditions- improved so that Picture E were representative of the
entire area, with no visibility problems or short and long term health
effects-, would the improved conditions- change tr>e pattern of your lei-
sure activities Yes[ ] Nb[ ]
IF NO, PROCEED TO GENERAL INFORMATION SECTION
IF YES, THEN:
A. Administer indoor and outdoor activity and cost lists
B. Check time constraint
ZERO BIDS:
Although you told me that you would not pay anything to have visibility
or short and long term health effects improve throughout the Los
Angeles area to like those shown in Picture E, would the improved con-
ditions change the pattern of your leisure activities? Yes[ ] No[ ]
IF NO, GO TO GENERAL INFORMATION SECTION
IF YES, THEN:
A. Administer indoor and outdoor activity and cost lists
B. Check time constraint
A-22
-------
Bidding - Aesthetic, Acute Health, and Chronic Health Effects
The quality of the air may also affect your long term I.ealth. There is
evidence that high concentrations of emissions as represented in
Pictures A and B have lasting effects- upon the respiratory and circul-
atory systems in addition to eye irritation and reduced visibility.
Evidence shows- that, on the average, people who live in areas with con-
centrations like those in Picture A can expect a reduced lifespan of up
to 2 years compared with people who live in conditions represented by
B, and up to 3 years when compared with people who live in conditions
represented by Picture C.
If the level of emissions could be reduced in the Los Angeles area so
that visibility, short and long term health conditions would be repre-
sented by Picture B instead of A, not only in the B" area but also in
your area, and if the air would be cleaned up to this level in..f2, 1Q)
years, would you pay (a monthly charge, an extra charge in your utility
bill) of CSTART BIDDING WITH PREVIOUS BID) for as long as you live in
the Los Angeles area?
I ] DO FOLLOWING ONLY IF CHECKED
Life Table: Here is a table that might help you. It shows the total amount
you would pay for as long as you live in Los Angeles for vaiious amounts of
monthly payments.
RECORD BID
REVISION IF NECESSARY
Is there some other payment scheme besides (a monthly charge, an extra
charge in your utility bill) that you would prefer? Yes[ ] No[ ]
IF YES:
What would it be?
Would you consider moving to a new location in the Los Angeles area
if air quality were like Picture C everywhere? Yes[ ] No[ ]
IF YES:
Where would you move? (GRID LOCATION ON MAP)
i-23
-------
GENERAL INFORMATION SHEET: WOULD YOU PLEASE FILL OUT THE FOLLOWING?
1. Age
2. Sex: Male] J Female! ]
3. Marital status: Single! ] MarriedJ 3
A. Number of persons in your household?
5. Your education: years. Highest degree obtained:
High School! ] College! 1 Vocationall ] Advanced! J
6'. Address of employment:
7. Location of employer(s) (GRID LOCATION ON MAP)
8. Are there any environmental Hazards associated with your job, such as
noise, health, or sight? YesI 1 Nof ^ IF YES: What are these hazards?
9. What is the percentage of your work time indoors? %
10. In a typical work week how much time do you spend on the job? _ hours
11. If you received our pamphlet last week, did you read:
[ ] 0-5 pages
[ ] 5-10 pages
[ ] more than 10 pages
If you live in area A or B:
How much would you pay for this same house (apartment) today if it
were located in an area where the air pollution levels were like those
shown in Picture C? $ _
If you live in area C:
Do you believe that any part of the value of your home is because you
live in a relatively unpolluted part of Los Angeles? Yes[ ] Nof ]
IF YES: How much (% or dollars) _
If you live in area B or C:
Would you consider moving if the air pollution problem were as bad as
A throughout the entire area? Yes[ ] No{ J
IF YES: Where would you most likely move? (GRID LOCATION ON MAP)
A-2 A
-------
How much do you think it would cost to clean up air pollution in the Loa
Angeles area to a condition like that shown in Picture C?
$
If all citizens were billed equally, how much do you think it would cost
each person in order to achieve conditions- like that shown in Picture C?
$
A-2 5
-------
1. Characteristics of home:
Living Area: square feet
Number of Rooms-:
Number of Bedrooms
Number of Bathrooms:
Other Rooms: (PLEASE CHECK}
[ ] Den
[ ] Family room
[ ] Dining room
[ ] Enclosed porch
[ ] Attic
[ ] Basement
% Easement finished
T~] Utility room
[ ] Other
Scenic View: YesI ] No[ ] IF YESs Specify_
Number of Stories: (INCLUDE BASEMENT)
Remodeled: Yes[ ] No[ ] Don't know[ ]
IF YES: Specify previous style and date _
2. Equipment: (PLEASE CHECK)
[ ] Dishwasher
I_] Disposal
[ ] Central Air Conditioning
[ ] Trash Compactor
[ ] Central Heating
Pool: Yes[ ] No[ ]
IF YES: Circle whether heated, enclosed, or other (specify)
Fireplace: Yes[ ] No[ ]
Age of home: years (when constructed)
A-26
-------
IF YOU LIVE IN AN APARTMENT, GO TO QUESTION 4
3. a). Year of purchase:
b) Could you please indicate what the purchase price was: $_
c) What are your monthly payments: $
d) What do you feel your home is worth, in today's market? $
e) What are your property tax payments- per year? $
f) How long have you been living in this house? years
GO TO QUESTION 5
A. a) How long have you Been living in this apartment: years
b) Would you indicate your monthly rent? $
5. What are your insurance payments per year? $
6. What do you pay monthly for general upkeep around your home Capartment)?
7. Why have you chosen to live in this area? RANK IN ORDER OF IMPORTANCE,
WHERE ONE IS MOST IMPORTANT. CHOOSE TOP FIVE.
[ ] Attractiveness of area in general
[ ] Close to work
[ ] Close to recreation activities
[ ] Close to friends
[ ] Close to schools
[ ] Close to services
[ ] Close to transportation routes
[ ] Air quality
[ ] Affordability of home
[ ] Low crime rate
[ ] Prestige of area
[ ] Quiet neighborhood
[ ] Other
8. What are your average expenditures per month for food: $
9. What are your average expenditures per year for clothing? $
10. Please mark the box corresponding to your annual household income.
[ ] 0-$5,000 I ] $30,000-$35,QOO
[ ] $5,000- $10,000 I ] $35,000-$40,000
[ ] $10,000-$15,000 f ] $AO,000-$50,000
[ ] $15,000-$20,000 I ] $50,000-$60,000
[ ] $20,000-$25,000 I 1 $60,000-$SO,000
[ 1 $25,000-$30,000 I ] Over $80,000
A-27
-------
1. Do you own or share in the ownership of a motor vehicle? YesJ J NoJ; J
Type of VehicleCs) Model Tear
_Model Year
Model Year
2. How many licensed drivers are in your family?
3. How many miles per gallon do you get for each car in the city?
mpg
mpg
mpg
How many miles do you and your family typically travel in your automo-
Bile per week? miles
4. How many hours do you and your family spend in a typical week commuting
to:
Work or school hours
Shopping hours
Recreational activities hours
5. Do you participate in a car pool? Yes | ] No [ ]
6. How much do you spend each month ons
a) Gasoline costs $
b) Maintenance costs $
c) Public transportation fares $
d) Insurance payments $
7. Did you take a vacation within the last year where you were away from
home for more than 4 days? Yes[ ] No[ J
IF YES: About how much were your expenditures on this trip? $
A-28
-------
1. Have you ever had any of the following? (PLEASE CIRCLED
a) High blfcod pressure ell Asthma
hi Heart trouble fj Chronic nervous trouble
c} Stroke gj Cancer
d) Chronic bronchitis hj Tuberculosis
2. Have you ever' had trouble with the following? (PLEASE CIRCLE)
i) Pain in the heart or tightness or heaviness in the chest
j) Trouble breathing or shortness- of breath
k) Frequent headaches
1) Constant coughing or frequent heave chest colds
m) Frequent eye irritations
n) Allergies
o) Nose and throat irritation
3. Are any of these (the above) conditions aggravated Cor made worse) by
heavy air pollution? YesI ] Nol ]
IF YES: Which ones? (LIST LETTERS CIRCLED)
4. Do you suffer from any other diesases which could be made worse by poor
air quality? Yes[ 1 No[ ] Specify
5. Do you or any member of your family have any physical disabilities which
limit your activities? Yes[ 1 No[ ].
6. Would conditions like those in Picture C, if they occurred" over the
entire area?
a) Make your life more
pleasant Not at all[ ] To some degree[ ]
Greatly[ ]
b) Require you to spend
less money on drug
items or doctor's Not at all[ ] To some degree[ ]
fees Greatly[ ]
c) Make it easier to Not at all[ ] To some degree[ ]
do your work? Greatly[ ]
7. Do you enjoy doing your leisure activities more during the ^ay or during
the night? Day Night Makes no difference
8. Do you smoke? Yes[ J No[ ] IF YES: How many packs per day?
9. Do you take medication regularly? Yes[ ] Nol ]
IF YES: Monthly expense on this medication? $ /month
A-29
-------
10. How much do you spend monthly on medical problems associated with, air
pollution effects? $ /month.
11. How. much, do you spend yearly for doctor's fees? $ /year
12. How much, do you spend yearly on medical and life insurance?
$ /year
13. Have you purchased any items- to reduce your exposure to air pollution
(such as carbon filters)? YesI 3 Nol ]
IF YES: What items?
A-30
-------
1. Which, one. of these statements applies to you? (CHECK. ONE}
I ] I have not Been hothered by air pollution.
[ J I have been Somewhat Bothered By air pollution.
[ ] I have Been Bothered quite a lot By air pollution.
2. Do you Believe that air pollution in Los Angeles: (CHECK ONE)
[ ] Has Become worse since you have lived here.
[ ] Has stayed aBout the same since you have lived here.
[ ] ,Has gotten Better since you have lived here.
3. What do you think should Be done aBout air pollution? (CHECK ONE)
["] Ignored
[ ] Reduced
4. Please rank the following proBlems in terms of importance (most to
least) as the major issues facing the community. (CHOOSE TOP FIVE)
[ ] Juvenile delinquency J ] Nuclear energy
[ ] CommunicaBle disease I ] Alcoholism
[ ] Unemployment [ ] Water pollution
[ ] Air pollution [ ] Energy
[ ] Car accidents [ ] Congestion
[ ] Crime [ ] Other
5. Do you believe that air pollution in the Los Angeles area:
[ ] Cannot Be reduced Below its present level
[ ] Can be reduced below its present level
[ ] Can be almost completely eliminated
6. What do you think the words "air pollution" mean to most people in the
Los Angeles area? Do they mean:
a) Frequent bad smells in the air Yes[ J No! J
b) Too much dirt and dust in the air Yes! ] Nol J
c) Frequent haze or fog in the air Yes[ ] No[ ]
d) Frequent irritation of the eyes YesI ] No[ ]
e) Frequent nose or throat irritation Yes[ ] No[ ]
f) Other YesI ]
7. Have you read or seen anything in the newspaper recently about air
pollution? YesI ] Nol ]
8. When you read the newspaper, do you generally choose to read articles
on air pollution? YesI ] No[ ]
A-31
-------
9. Do you consult the daily air pollution index before engaging in any
activities? YesJ J NoJ J
IF YES: What kind of activities?
A-32
-------
1. If you received our health, pamphlet, what do you think about it?
[ ] Did not read [ J Made me more concerned ab.out
[ ] Very informative health, effects
[ ] Hard to understand J ~] Had no influence on me
[ ] Scientific mumbo-jumbo
2. Here is a list of words and phrases. Select two which describe how you
feel about your participation in this survey.
[ ] Stimulating
[ ] Just tolerable
[ ] A waste of time
[ ] Educational
[ ] Boring
[ ] An invasion of privacy
[ ] Interesting
[ ] Kind of fun
[ ] Hard to take seriously
3. Here is a different list of words and phrases. Select two which describe
how you feel about the questionnaire.
[ ] Relevant
[ ] Credible
[ ] Likely to influence air quality control
[ ] Unrealistic
[ ] Pretty flakey
[ ] Unlikely to have any effect on air quality control
[ ] Irrelevant
4. Finally, here is another list of words and phrases. Select one from
each column to describe how you feel about your answers to the question-
naire.
Column 1 Column 2
[ ] Quite accurate [ ] A fairly good guide for
valuing air quality
[ ] There was no way I could
come up with accurate [ ] A good guide for valuing
answers. air quality.
[ ] Accurate in a "ball park" [ ] A poor guide for valuing
kind of way. air quality.
THANK YOU FOR YOUR COOPERATION.
-i-J i
-------
Figure 4.4a
(Good)
Photograph Depicting Observation Paths for "Good" Visibility
n
---: -? v. -^. T. -', ..-,... . . _
g*'....!.."' -it.?-f^;,mmtr "v'^-4;-.'1-:- ^y^B^^-^^,;' «>'g>^
I
A-34
-------
Figure A.4b
(Fair)
Photograph Depicting Observation Paths for "Fair" Visibility
" '
A-3 5
-------
Figure A,4c
(Poor)
Photographs Depicting Observation Paths for "Poor" Visibility
A-36
-------
D. GRAND CANYON STUDY
(Questionnaire and Complete Picture Set Not Yet Released)
A-3 7
-------
GRAND CANYON/SOUTHWEST PARKS STUDY
HOPI POINT EAST (A.M.)
(This set of slides is attached in the back
of the Guidebook)
HOPI POINT WEST (A.M.)
HOPI POINT WEST (P.M.)
Grand Canyon Photograph Board
Figure 1
A-38
-------
HOPI POINT EAST (A.M.)
(fron Grand Canyon)
SHIPROCK 9AM
TRUMBULL MT.
(from Zion)
9AM
Regional Photograph Board
Figure 2
-------
GRAND CANYON PLUME BOARD
Figure 3
A- 40
-------
E. WATER RESOURCE COUNCIL SUGGESTED CONTINGENT MARKET
VALUATION TECHNIQUES TO ESTIMATE
RECREATION BENEFITS
This section reproduces the Water Resource Council's guidelines
on contingent market estimation techniques as contained in the Federal
Register (1979).
Federal Register / Vol. 44. No. 102 / Thursday. M;iy 24, 1979 / Proposed Rules
3Q229
(3) Contingent Valuation (Sun-ey)
Methods, (i) Overview. (A) Contingent
valuation methods (CVM's) obtain
estimates of changes in NED benefits as
a result of changes in the quantity of
recreation consumed by directly asking
individual recreationists. Vjk is an
individual household's willingness to
pay (WTPj for changes in quantity, Q, of
recreation at site j. Individual values
may be aggregated across the
population of the market area for facility
j by summing V,* for all n households in
the market area:
7J-
(B) Contingent valuation methods
consist of designing and using simulated
markets to identify the value of
recreational amenities just as actual
markets would, if they existed. Three
basic steps are involved: (1) the analyst
establishes a market to the respondent;
(2] he permits the respondent to "use"
the market to make "trades" and
establish prices or values which reflect
the respondent's individual valuation of
the recreation opportunities "bought" or
"sold"; and (3) the analyst treats the
values reported by the respondent as
individual values for the recreation,
contingent upon the existence of the
market. The respondent's bids are used
with the data contained in the market
description (step 1) to estimate the
aggregate value of the recreation being
studied.
(C) Contingent valuation methods are
particularly appropriate (1) for
evaluating projects likely to be one of
several destinations visited on a single
trip, and (2) where a project results in a
relatively small change in quality of
recreation at a site. Contingent value
results may be adversely affected unless
questions are carefully designed and
A- 41
-------
J0230
Federal Reviser / Vol. 44, No. 102 / Thursday. May 24. 1970 / Proposed Rules
pretested to avoid several possible
kinds of response bias. Several
techniques are available for obtaining
the individual bids, which are the basic
data for CVM.
(ii) Iterative Bidding Formats. (A)
Iterative surveys are distinguished f'ro.i
other contingent valuation methods in
that, following establishment of the
market and a complete description of
the recreational good, service, or
amenity to be valued, the respondent
reacts to a series of values posed by the
enumerator. The respondent answers
"yes" or "no" to questions asking if he is
willing to pay the stated amount of
money to obtain the stated increment in
the recreational good. The enumerator
iteratively varies the value posed, untii
the highest amount which the ,
respondent is willing to pay is identified.
This amount is treated as the
respondent's bid for the specified
increment in the recreational good.
(B) Iterative bidding techniques are
most effectively applied in personal
interview surveys. Mail survey formats
have also been used in research studies.
These typically ask the respondent to
react ("yes" or "no") to a small number
of different but specified values stated
in iterative questions and, finally, asked
an open-ended question: "Now, write
down the maximum amount you will be
willing to pay. S ." At the
present time, mail survey applications of
the iterative bidding technique have not
been adequately tested and cannot be
rcommended.
(C) The recreation facilities to be
evaluated will be described in quantity.
quality, time, and location dimensions.
These should be hypothetical in the
sense that they do not precisely describe
features of actual sites or proposed
projects, but should be described with
sufficient precision that the respondent
has adequate information on which to
base a valuation. To permit estimation
of regional models, quantity, quality.
and location dimensions should be
varied and the iterative bidding exercise
repeated. Verbal descriptions should be
precise, and in addition, wherever
practicable, the pertinent aspects of the
facilities should be displayed or
depicted using nonverbal stimuli (e.g.,
photographs, artist's drawings, motion
pictures, scale models, etc.).
(D) The good to be valued is "the right
to use * * * (the recreation facility).
* * ' for one year." When the Rood is
defined in this way, the V,k obtained are
annual measures of the individual's
willingness to pay for the existence of a
given increment or decrement in
recreation opportunities which would be
provided at site j. Where there are
compelling practical reasons to use a
bidding format which defines the good
in some other terms (e.g.. day of use,
trip, etc.), sampling and aggregation
procedures must be designed so that the
annual values are calculated.
(E) The institutional rules pertaining
to the hypothetical market will be
described in sufficient detail that the
respondent knows his rights and the
rights of all others in the market. These
rules should be realistic and credible.
They should place the respondent in a
role, and encourage market behavior
with which he is familiar. They should
be of a kind generally viewed as just,
fair, and ethically sound. They should be
nonthreatening: in particular, formats
which threaten the respondent with a
welfare shock that he may view as
unfair should be avoided.
(F) Vehicles should be carefully
pretested. At the pretest stage, always
include a neutral vehicle, e.g. "the
money collected will be placed in a trust
fund and devoted entirely to providing
* * * [the good]."
(G) The respondent should be
provided price or value information and
asked "would you buy?," with the clear
understanding that "if no, then you
would go without" The question "would
you be willing to pay . . ?" should be
avoided since this may be interpreted by
some respondents as an appeal for
voluntary contributions. Instead the
question must be worded so as to
suggest the pragmatic, if harsh, "take it,
or leave it" atmosphere of the
marketplace.
(H) Depending on the answer, "yes"
or "no," the price or value is varied
iteratively and the question is repeated
until the respondent's point of
indifference between the money and the
good is identified. Early iterations may
change the price widely until the
enumerator senses that he is
approaching the respondent's
indifference point Then, iterative price
variations will become finer.
(I) The starting price quote (called
"starting point") will vary across
respondents. The particular starting
point assigned to a given respondent
will be chosen randomly.
(J) The method of payment (called the
payment vehicle) should be specified.
Payment vehicles which may. in and of
themselves, generate an emotional
reaction should be avoided: emotional
reactions to a specific payment vehicle
introduce a confusing element into the
bid data. Vehicles based on tax
increments, utility bill increments, and
hunting or fishing license fee increments
may introduce such problems.
(K) General formats for iterative
bidding questions are presented below.
(Specific examples follow.) The question
form must be specific to the particular
measure of value to be elicited from the
respondent. WTP formats should always
be used. However. WTP formats may be
incremental (willingness to pay for an
increment in a desired recreational
opportunity) or decremental (willingness
to pay to avoid a threatened decrement
in a desired recreational opportunity).
The incremental version has two major
advantages: it is the theoretically correct
measure and, since it confronts the
respondent with the (hypothetical)
chance to pay for a desired good, it is
not likely to provoke an offended
reaction on the part of the respondent
The decremental version, which asks
the respondent to pay to avoid a change
he does not want may seem unfair or
morally offensive to at least some, and
thus may elicit biased or otherwise
unreliable value estimates. The
incremental version is preferred.
wherever it is credible.
(L) The incremental version may not
be credible where the real world
experience is typically one of
decrements rather than increments. To
determine the value of remaining
"unspoiled natural environments" in a
world where these are fast disappearing,
questions asking "if a new unspoiled
natural recreation environment could be
created and the right to use it would
cost. * * *, would you buy?" may be
rejected as fantasy by some
respondents. In these circumstances, it
may be necessary to resort to
decremental formats. Since reasonable
doubts can be raised, a priori, about the
efficiency of WTP formats in this
circumstance, the following precautions
are essential: take great care to design
the best (i.e., most consistent and
plausible, and least offensive) formats;
and pretest at least two differenct
formats, in order to permit statistical
testing for differences in the
performance of alternative formats.
(M) General examples of the WTP
formats are:
WTP incremental: "If you had the
opportunity to obtain * * * (describe
an increment in recreation
facilities]. * * [describe hypothetical
market rules and payment
vehicle] * * * would you
pay * * * [starting price] * * *? Yes
(pay) . Or would you refuse to pay,
and make do without * * * [the
increment]? No(pay) . Reiterate
with new prices until the highest price
eliciting a "yes" response is identified.
decrementai (example I):
[describe a decrement in
A-42
-------
Federal Re»istor / Vol. 44, No. 102 / Thursday. May 24. 1979 / Proposed Rules
30231
recreation facilities) * * * will occur
unless * * * (describe market rules and
payment vehicle]. Would you
pay * * * [starting price] * * * to
avoid * * * [the decrement]?
Yes(pay) . Or would you refuse to
pay, and thus permit * " * [the
decrement]? ."
WTP decremental (example 2):
" * * * [describe a recreation facility
which is currently available to
respondent] * * * is currently
1 available * * * [describe existing
market rules, existing payment vehicle,
and existing price]. Unless * * * [the
existing price] * * * is
increased, * * * [describe a decrement
in recreation facilities] * * * will occur.
Would you pay * * * [starting price,
which is some increment over the
existing price] * * ", in order to
prevent * * * [the decrement]?
Yes(pay) . Or wouid you refuse to
pay, and thus permit [the decrement]?
No(pay) ."Reiterate * * *
(N) Since som<" respondents may bid
only zero amounts to WTP questions, it
is important to identify which zero bids
represent true zero valuations and
which, if any, represent a protest against
the rules or payment vehicles used in
the iterative bidding format. Check
questions should always be used to
probe "zero" responses to WTP formats,
e.g., "Did you bid zero because: (Check
one}. .
(1) You believe * * * [the stated
increment] * * * would be worth
nothing to you?
(2) You believe * * * [the
vehicle] * * * is already too high?
(3] You believe * ' * [the statement
increment] * * * would be of value, but
you do not think it is fair to
expect * * * [the respondent's class of
citizen e.g. hunting license holders,
utility customers] * * * to pay for it."
(O) Given this'kind of question.
answers (2) and (3) are "protest"
responses, addressed not to the value of
the good but to some element of the
question format. Protest bids should be
recorded but eliminated from
calculations to estimate values. In
pretests, formats that elicit more than 15
percent protest responses should be
discarded, since a high incidence of
protest bids may be an indication that
some nonzero bids are also distorted.
(iii) Noniterative Bidding for/note. (A)
Noniterative bidding formats are
adaptable to implementation with mail
surveys. There arc two kinds of
noniterativc formats: (1) close-ended
formats, which ask respondents to react
("yes" or "no") to a single stated value.
arid (2) open-eiidcd formats, which ask
the respondent to write down the
maximum amount he would be willing
to pay. A variant of the second kind of
format asks the respondent to either
select his maximum WTP from a list of
stated discrete values or write down his
maximum WTP. Noniterative bidding
formats are not likely to be as reliable
as iterative formats.
(B) The use of noniterative mail
survey formats is generally discouraged.
It is permissible only for recreation
benefit analysis of small'projects. Mail
survey formats will, insofar as is /
practicable, retain the basic attributes of
the personal interview formats
described above. Survey instruments
should be printed using a process that
permits reproduction of color
photographs and, if appropriate, other
nonverbal stimuli.
(C) The sample should be divided,
using open-ended bidding formats with
one half and close-ended formats with
the other half. The bids obtain using
these two kinds of formats should be
analyzed to determine if the format
influences the results obtained, to a
significant degree. Examples of these
formats are presented below.
(D) Open-ended: Due to pressures of
population growth and economic
development, 10 miles of trout stream
such as that shown in photograph
are likely to be converted to
other uses (e.g., a reservoir] and thus
lost for trout fishing. Assume the only
way to preserve this 10-mile stretch for
trout fishing is for trout fishermen to
agree to buy an annual pass to fish in
that stream segment. The money
collected would pay for preservation of
the stream section. If the stream
segment was miles from your
home, and you could expect to catch
trout in a typical day's fishing
there, what is the maximum amount you
would pay for the annual fishing pass?
Dollars per year
(E) Closed-ended: (information
presented does not change; the final
question reads:] * * * and an annual
fishing pass costs S (assign dollar
amounts randomly to respondents].
Would you buy one? Yes No
(iv) Use Estimation with C-VM's. (A)
All of the contingent valuation
procedures described thus far generate
annual value estimates directly, instead
of generating, first, values per user day
and, then, estimates of expected user
days. The "annual value estimation"
procedure is superior, in that it is more
reliable, it automatically corrects for the
economic influence of existing
recreation opportunities, and it is better
adapted to estimating activity and
existence values, where both are
important.
(B) However, if for some reason
values per user day are desired by
analysts, contingent valuation formats
can be designed to estimate such values.
Questions are worded in terms of a
day's activity rather than an annual
valuation. Great care must be taken in
the case of proposed increments to
determine the respondent's valuation of
a day at the proposed site, given the
continued availability of existing sites.
Estimates of use may be made either by
collecting such information as part of
the survey or by other approved
methods.
(C] To collect use information in the
survey, proceed as follows:
{!) For decrements in recreation
opportunities, ask: how many trips the
household made (/) last year and (ii) in a
typical year, if last year was unusual for
any reason: how many days did the trip
last; and how many household members
participated in each trip.
(2) For increments, ask: (/) the same
information as for decrements, but
pertaining to existing recreation sites
similar to the proposed increment. Then,
if the proposed increment (described
with verbal and nonverbal stimuli) were
made available: [if] "how many trips,
how long, and how many family
members" for the proposed increment:
and [iii] "how many trips, how long, and
how many family members" in total for
both the existing and proposed sites.
(v) Using Contingent Valuation
Methods. Contingent valuation methods
can be used to develop value estimator
models or to estimate recreation
benefits for a specific proposed project.
Th:.^w two uses arc discussed below.
A-43
-------
INDEX
Abclson, Peter, 3-17
Absolute thresholds, 3-6
Abe. Associates, 4-23
Activity value, 2-2, 2-222-23
Acrosola, and pollution, 3-17
Aesthetics
hedonle price technique and, 4-30
measuring economic benefits of, 1-11-6, 4-604-62
residential property value studies and, 4-294-60
Age, and perception, 3-8
Aggregation, 5-155-16
discount rate choice In, 2-242-25
discount rate disadvantages In, 2-252-26
cross Individuals, 2-232-24, 5-15
Inequities In distribution of benefits In, 2-24
locatlonal distribution of impacts In, 2-24
shortcomings of, 2-232-24
across time, 2-242-26, 5-15
total benefit measures with, 2-222-26
weighting scheme used In, 2-24
Agricultural activity. In visibility Impairment, 3-11
Air pollution, see Pollution
Airport visibility data. 3-13. 3-14, 3-21, 5-4
Air quality
national standards for, 6-34, 6-35
in urban areas, 1-2
values In benefit-cost analysis of, 2-2
willingness to pay for housing and, 2-9
Air quality benefit analysis
activity values, options values, and existence
values In, 2-2, 2-222-23
actual market versus contingent market
approaches In, 4-14-2
benefit measures and effects of, 3-303-33
benefits measures transferred across studies In,
4-724-74
bidding methods used In, 4-34-22
central business district distances In property
studies and, 4-404-41, 4-55, 4-57
. definition, measurement indices and presentation
Of air quality levels and, 3-303-33
federal clean air legislation and, 1-2
guidelines In, 5-15-17
hedonle approaches used In, 5-125-13
hedonle travel cost approach In, 4-67
household production function approach in, 2-28,
4-68
Batched pairs technique In, 4-544-55
prescribed burning In, 6-636-65
public goods In, 2-2
ranked attributes/market share technique in, 4-60
4-61
residential location model In, 4-554-56
residential property value studies and, 4-29, 4-36
4-37. 4-404-41, 4-424-44, 4-474-53, 4-584-60
travel cost and site substitution methods in, 4-61
4-67
visibility benefits In, 6-606-62
wage and salary differentials In, 4-69, 4-70
tion and Transport In the Atmosphere), 3-30
AJxen, I.. 4-17
American Petroleum Institute, 3-13
Anderson, Cordon, 6-22
ArrUga-Sallnaa. A.S.. 4-69, 4-70
Arrow, Kenneth J., 2-2, 2-24
Atmospheric data
pollutant gases and, 3-10, 3-11
selection of estimation methods and, 5-8
visibility benefit analysis and, 3-13-3, 3-10
Babb, Fmerson N., 4-19
Backstroo, Charles H.. 4-23. 4-26
Baumol, William J., 2-24
Backer Gary S.t 2-27
Benefit-cost analysis
Activity value In, 2-2
air quality values In, 2-2
benefits considered In, 1-41-5
concepts of benefits In, 2-12-2
costs considered In, 1-5, 2-1, 5-16
decision-making process and, 1-5
economic benefit measures in, 2-12-33
factors considered in, 1-41-5
guidelines to use of, 5-165-17
monetary measures of benefits in, 2-32-9
need for, 1-11-5
option value in, 2-2
overview of, 1-51-7
public goods in, 2-2
quasl-optlon value in, 2-2
teps In, 1-51-7
utility concept in, 2-1
value of techniques In, 1-5
Benefits
activity value In, 2-2, 2-222-23
aggregation across individuals in, 2-232-24
aggregation across time for, 2-242-26
aggregation to total measures In, 2-222-26
in air quality control, 2-222-23
concepts of, 2-12-2
consumer surplus as measure of, 2-3, 2-4
demand curve approach as measure of, 2-42-9
discount rate for future, 2-242-26
economic measures of, 2-12-33
existence value in, 2-2, 2-222-23
expenditures as measure of, 2-3, 2-4
Income as measure of, 2-3
Indifference map used in measure of, 2-10, 2-11
inequities In distribution of, 2-24
locational distribution of Impacts of, 2-24
marginal utility In measure of, 2-10
monetary measures of, 2-32-9
non-monetary valuation of, 2-3
option value In, 2-2, 2-222-23
refined consumer surplus monetary measures of, 29
2-22
technical consumer behavior models in, 2-262-33
use of term, 2-1
utility concept in, 2-1
utility map framework in, 2-12-12
values considered in, 2-2
Bergstrom, R.W., 3-30
Bidding methods, 4-1, 4-34-29
application of, 4-54-15, 5-115-12
biases In, 4-184-21, 4-214-23, 4-27
bidding or valuation questions used In, 4-114-13
case studies of, 6-16-33
choice of survey type used In, 4-244-25
closing questions and remarks used in, <4-14
comparison of property value case studies with, 6-58
6-60
comparison of visibility related studies in, 6-29
6-33
contingent market approach of, 4-2
contingent market rejection and problem bids In, 4-21
4-22
criteria for selection of, 5-6, 5-7, 5-8
data reliability In, 5-9
evaluation of problem bids In, 4-14
Farmlngton study with, 6-106-16, A-3A-10
Four Corners study in, 6-26-6
free riding In, 4-18, 4-19, 4-20
general Introduction and scatexent of purpose In, 4-6
Grand Canyon/Southwest Parks study with, 6-226-29
hypothetical biases In, 4-16-.-16
information biases In, 4-204-21
1-1
-------
introductory non-valuutlon questions used In, 4-7
Lake Powell study with, 6-66-10
post-survey data verification and analysis in, 4-23
4-29
pretest In, 5-12
property rights in, 4-44-5
regression analysis used in, 4-284-29
ample size determination and verification la, 4-26
4-28
sampling procedures in, 4-254-26
scenario development and market definition in, 4-7
4-11
South Coast Air Basin study with, 6-176-21
special options In, 4-14
steps in survey research in, 4-23
strategic biases In, 4-184-20
strengths of, 4-15
survey Instrument in, 4-54-14, 5-12
survey procedures and pose survey data analysis In,
4-5, 4-234-29
theoretical basis for, 4-44-5
use of, 4-4
use of term, 4-3
weaknesses of, 4-164-23
zero bids In, 4-14
sec also Iterative bidding technique
Bishop, R.C., 4-17
Blackwell, H.R., 3-7
Blank, Frederick M., 4-14, 4-22, 4-68, 6-9. 6-10
Blumenthal, D.L., 3-30
Bockstael, Nancy E., 2-20
Boes, D.C., 6-52
Boom, P., 4-17, 4-184-19
Boldt, David, 6-48
Boston study, 6-33, 6-386-41
benefits estimate in, 6-406-41
comparison and review of hedonlc case studies with,
6-526-58
evaluation comments on, 6-41
hedonlc price function estimation in, 6-39
housing market segmentation in, 4-46
study area description and data selection in, 6-38
WTP function estimation In, 6-396-40
Bouchard, T.J., 4-23
Bowen, Howard R., 4-71
Bowes, H.D., 4-61
Bradford, D.F., 6-5
Bresnock, Anne E., 4-41, 4-44, 4-46. 6-33, 6-41, 6-43,
6-44
Brightness
light intensity measures with, 3-15
perception of visibility and, 3-5
Brown, Gardner, Jr., 4-66, 4-67
Brown, Gardner M., 4-11, 4-25
Brown, P.J., 4-65
Brookshlre, David S., 2-27, 4-3, 4-10, 4-11, 4-13, 4-15,
4-17, 4-19, 4-20, 4-21, 4-23, 4-25, 4-27, 4-44, 4-54,
4-68, 5-12, 6-1, 6-6, 6-7, 6-9, 6-10, 6-11, 6-17,
6-19, 6-20, 6-21, 6-32, 6-33, 6-34, 6-35, 6-59
Bryce Canyon National Park, 2-3, 4-64
Bureau of Land Management (8LM), 3-27, 3-30
Bureau of the Census, 5-16
Burning
benefit estimation techniques in, 6-636-65
costs and benefits from, 6-64
use of, 6-63
visibility impairment from, 3-11
Carbon Monoxide, in Denver study, 6-41--6-44
Cedar Mountain study, 3-21
Chaikln, Kathleen, 6-48
Chiang, Alpha C., 4-39
Cicchettl, Charles J., 2-2, 2-29
Clark, E., 4-71
Class 1 federal areas, see Federal areas (Class 1)
Clawson, M., 4-61
Clean Air Act
Four Corners study and, 6-2
national goals under, 1-11-2
prescribed burning under, 6-63
visibility benefits analysts and, 5-3
visibility protection provisions of, 1-2
Cloud cover, and perception, 37
Cluster samples
bidding methods and, 4-25
costs of Interviews in, 5-9
Coal production, 3-27, 3-30, 6-26-6
Colors
light intensity measures with, 3-14, 3-15
memory, 3-27
perception of visibility and, 3-5
photography and, 3-273-28
Combustion, and visibility Impairment, 3-10, 3-11
Compensating surplus (CS)
quantity change measures with, 2-172-18
reference level of welfare on, 2-202-21
selection factors in, 2-22
In simple utility map approach, 2-27
size difference measures with, 2-20
Compensating variation (CV)
consumer surplus comparisons with, 2-21--2-22
consumer surplus measures of price changes with, 2-17
In expenditure function approach, 2-29
Income effects on, 2-17
In residential location model, 4-56
election factors in, 2-22
ize difference measures with, 2-192-20
weak complementarity measures and, 2-33
Computer simulations
In bidding methods, 5-11, 5-12
of visibility impairments, 3-27, 3-28
Consumer behavior theory
consumer surplus and,. 2-262-33
hedonic price technique and, 4-30
Consumer surplus
s benefits measure, 2-3, 2-4
Comparisons among measures of, 2-192-22
Compensating surplus measures In, 2-172-18
definition of, 2-10
equivalent surplus measures in, 2-172-18
expenditure function approach to, 2-292-31
household production function approach in, 2-27
2-29
Income compensated demand curve (ICDC) in, 2-122-14,
'income effect on, 2-12. 2-14
Indifference map with, 2-10, 2-rll
Marshall's theories on, 2-10, 2-13
price changes and, 2-162-17
quantity changes and measure of, 2-172-19
reference level of welfare in, 2-202-21
refined measures of, 2-92-22
election of measures in, 2-22
simple utility map approach In, 2-262-27
Ize comparisons in, 2-192-20
technical consumer behavior models and, 2-262-33
utility map framework of, 2-102-12
weak complementarity measures in, 2-312-33
willingness to pay measures In, 4-4
see also Ordinary consumer surplus
Contrast
light Intensity measures with, 3-15
in photography, 3-28
Cornsveet, T.N., 3-15
Costa
Analysis of, see_ Air Quality Benefit-cost analysis
bidding methods survey instrument choice and, 4-24
definition of, 2-1
determination of visibility impairments and 1-4
emission control, 5-16
of estimation methods, 5-95-11
power plant emission controls and, 6-62
Cralk, K.H., 3-12
Cramer, Oven P., 6-636-64
Crespl, I., 4-17
Cropper, M.L., 4-69, 4-70
CS, see Compensating surplus (CS)
CV, see Compensating variation (GV)
O'Arge, R., 6-10, 6-17, 6-45
Davis, R., 4-3
Demand curves
as benefits measure,'2-42-9
consumer surplus concept and, 2-12
difficulties in estimating data for, 2-9
1-2
-------
explanation of. 2-4. 2-5
Income compensated (ICDC), 2-122-14, 2-15
Income effect on, 2-12
rccrvjtlou ex.itopl.es of, 2-62-8
DeNevurs, N., 3-21
Denver study. 6-33, 6-416-44
benefit estimates In, 6-43
comparison and review of hedonlc case studies vlth,
6-526-58
valuation comments on, 6-436-44
hedonlc price function estimation In, 6-42
housing market segmentation In,' 4-46
particle llcht-absorptlon In, 3-17
study area description and data selection In, 6-41
6-42
WTP estimation In, 6-43
Devlne, Hugh, 4-64
Deyak, Timothy A., 4-54, 4-56, 4-69
Diamond, P.A., 2-29
Donenclch, \homaa A., 4-60
Orlvas, P.J., 3-14, 5-5
Driver, B.L., 4-65
Dust, and visibility Impairment, 3-11, 3-21, 3-26
Eastman, C., 6-2
Economic benefit analysis, see Benefit-cost analysis
Education, and perception, 3-7, 3-8
Emissions
Four Corners study of, 6-26-6
Grand Canyon/Southwest Parks study of, 6-226-29
Lake Powell study of, 6-66-10
Environmental Protection Agency (EPA)
atmospheric technical Information from, 3-1
benefit analysis reference documents from, 1-8, 3-4
existing air quality conditions and facilities of,
5-4
flexibility of impairment determination In, 1-4
Grand Canyon/Southwest Parks study of, 6-1, 6-22
light intensity models of, 3-14
photo file of, 5-12
PLIWJE model and, 3-30
power plant emission laws and, 660, 6-62
standard visual range of, 3-13
survey methods used In, 4-27
VIEW program of, 3-21
visibility Indices froa, 3-12, 3-13
visibility monitoring guidelines of, 3-18, 3-30
visibility protection laws and, 1-2
VISTTA prograa of, 3-30
EPA, see Environmental Protection Agency (EPA)
Equivalent surplus (ES)
quantity changes and, 2-172-18
reference level of welfare on, 2-202-21
election factors for, 2-22
In simple utility map approach, 2-27
size difference measures with, 2-20
Equivalent variation (EV)
consumer surplus comparisons with, 2-21
consumer surplus measure of price changes and, 2-17
in expenditure function approach, 2-302-31
In household production function approach, 229
Income effects on, 2-17
in residential location model, 4-56
selection factors In, 2-22
size differences with, 2-192-20
Erekson, Homer, 4-51, 6-38
Erskln, H., 4-17
ERT, Inc., 5-5
ES, see Equivalent surplus (ES)
Ettenhelm, George, P., Jr., 5-12
Eubanks, Larry S., 2-27, 4-68
EV, see Equivalent variation (EV)
Existence value, 2-2, 2-222-23, 6-286-29
Expenditures
as benefits measure, 2-3, 2-4
consumer surplus measures with, 2-292-31
demand curves with, 2-4, 2-62-9
fabrlck, 5-5
Fsralngton study, 6-1, 6-106-16
aggregate benefits measures In, 6-15
analysis of results In, 6-126-15
application of bidding method In, 6-116-12
comparison of bidding method studies with, 6-296-33
estimated cost of controls in, 6-62
evaluation coamonts on, 6-156-16
photography used In, 6-11, 6-156-16, A-2, A-ll
A-13
problem formulation and scenario development In, 6-11
questionnaire-bidding game used In, A-3A-10
Federal areas (Class 1)
Clean Air Act on goals for, 1-11-2, 1-5
Farolngton study on, 6-106-16-
flexibility of impairment determination in, 1-21-4
Four Corners study on, b-26-6
Integral vista protection in, 1-2
Lake Powell study in, 6-66-10
multiple source regional hazes and urban plumes in,
1-2
prescribed burning In, 6-63
residential property value studies applied to, 4-59
type of pollution In, 1-2, 1-3
visibility monitoring In, 3-173-18
Film, in photography, 3-28
Fires, and visual impairment, 3-26; see also Burning
Flshbcin, M., 4-17
Fisher, Anthony, 2-2, 2-24
Fog, in visibility Impairment, 3-26
Forestry activities, in visibility impairment, 3-11
Four Corners study, 6-1, 6-26-6
aggregate benefits measures in, 6-5
analysis of results of, 6-4
application of bidding methods in, 6-4
comparison of bidding method studies with, 6-296-33
EPA regulations and, 6-60
estimated cost of controls in, 6-62
evaluation comments on, 6-6
iterative bidding technique used in, 4-3
problem formulation and scenario development in, 6-26-3
Freeman, A. Myrlck III, 1-5, 2-2, 2-17, 2-18, 2-22, 2-23,
2-29, 2-31, 2-32, 4-31, 4-36, 4-37, 4-42, 4-45, 4-46,
4-47, 4-49, 4-51. 4-53, 4-61, 4-71, 4-72, 6-1, 6-35, 6-38
Free riding, in bidding .methods, 4-18, 4-19, 4-20
Frltschen, Leo J., 6-64
Caseous emissions
atmospheric chemistry of, 3-10, 3-11
scattering of light by, 3-16, 3-17
visibility impairment from, 3-10
Glen Canyon National Recreation Area (GCNRA), 6-7, 4-13
Grand Canyon/Southwest Parks study, 6-1, 6-226-29
aggregate benefit measures In, 6-276-28
analysis of results in, 6-256-27
application of bidding method in, 6-236-25
benefit estimates In, 6-62
comparison of bidding methods studies with, 6-296-33
valuation comments on, 6-286-29
problem formulation and scenario development in, 6-22
6-23
survey instruments used in, A-37A-40
Graphics, In visual environment presentation, 3-263-29
benefit measures and effects of, 3-30
in bidding methods, 5-115-12
factors affecting reproduction In, 3-27
goal analysis of results In, 6-256-27
application of bidding method In, 6-236-25
benefit estimates In, 6-62
comparison of bidding methods studies with, 6-296-33
valuation comments on, 6-286-29
problem formulation and scenario development in, 6-22
«-23
survey instruments used In, A-37A-40
Graphics, in visual environment presentation, 3-263-29
benefit measures and effects of, 3-30
in bidding methods, 5-115-12
factors affecting reproduction In, 3-27
goal of, 3-27
memory colors In, 3-27
problems encountered in, 3-273-28
selection of, 5-115-12
Craybill. F.A.. 6-52
Crillches, Zvl, 4-30, 6-34, 6-51
Guidelines In visibility benefit analysis, 5-15-17
aggregation of benefits across all populations In
5-155-16
1-3
-------
application of economic methods In, 5-11--5-15
benefit-cost analysli In, 5-165-17
bidding methods applications In, 5-115-12
coats and timing requirements In, 5-9--5-11
data requirement!! and reliability In, 5-85-9
hedonlc approaches In, 5-135-15
necessary assumptions in estimation methods used in,
5-9
problem formulation In, 5-15-4
purpose of, 5-1
scenario development in, 5-45-5
selection of estimation method used In, 5-55-11
Hammock., Judd, 4-11, 4-25
Hansen, Morris II., 4-23
Harrison, D., Jr., 4-26, 4-40, 4-44. 4-46, 4-49, 6-33,
6-38, 6-40, 6-43, 6-53
Hause, John C., 2-22
Haveman, Robert H., 2-23
Haze
Grand Canyon/Southwest Parks study of, 6-226-29
perception of visibility and, 3-5
photography of, 3-28
selection, of estimation method for, 5-7
visibility impairment and, 3-26
Heberleln, T.A., 4-17
Hedonlc price technique, 4-304-37
Ir pollution related to property values la, 4-42
4-44
applications of, 4-374-53, 5-125-13
assumptions in application of, 4-364-37
benefit estimate for air quality in, 4-39
Boston study of, 6-386-41
Box-Cox transformation in, 4-44
case studies for, 6-336-58
comparison and review of case studies using, 6-52
6-58
comparison of property value case studies with, 6-58
6-60
coses of estimation methods used In, 5-11
data selection in, 5-125-13
Denver study with, 6-416-44
dependent variable In, 4-374-39
example of use of, 4-314-36
functional t'orm of, 4-424-45
Independent variables In, 4-394-41
log-linear functlthods used in, 5-11
data selection In, 5-125-13
Denver study with, 6-416-44
dependent variable in, 4-374-39
example of use of, 4-314-36
functional form of, 4-424-45
Independent variables in, 4-394-41
log-linear functional form in, 4-44
narglnal implicit price in, 4-31
market segmentation in, 4-46
natched pairs technique in, 4-54
residential location model in, 4-55
residential property in, 4-314-36, 4-374-53
San Francisco Bay area study in, 6-486-52
selection criteria for, 5-7, 5-8
simple semi-log form In, 4-44, 4-45
South Coast Air Basin study in, 6-456-48
Washington, D.C., study In, 6-356-38
willingness to pay estimates in, 4-474-53
Hedonlc travel cost approach, 4-664-67
Helsler, S.L., 3-17, 5-13
Hendee, J.C., 4-65
Henderson, A., 2-12
Hennlng, John A., 4-29, 4-50
Henry, Ronald C., 3-1, 3-9, 3-17, 3-27
Hicks, J.R., 2-12, 2-17
Hlrschleifer, Jack, 2-4, 5-16
Ho rat, Robert, 4-68
Hotolllng, Harold, 4-61
Household data, in samples, 4-254-26
Household production function
consumer surplus measures with, 2-272-29
Farnlngton study with, 6-10
visibility benefit analysis with, 4-68
Housing market
hedontc price function In property value studies and, 4-46
willingness to pay and air quality conditions and,
2-9
Humidity, and visibility Impairment, 3-26
Hursh, Gerald D., 4-23, 4-26
Uurvltz, William H., 4-23
Husar, R.B., 3-17, 3-21
ICF, Inc., 6-606-62
Identification thresholds, 3-6
Income
aggregation across individuals and, 2-24
as benefits measure, 2-3
la bidding method studies, 6-32
consumer surplus measures of price changes with, 2-16
2-17
demand curves with, 2-42-9
Income compensated demand curve with, 2-122-14, 2-15
indifference maps with, 2-10, 2-11
reference levels of welfare change and, 2-202-21
In simple utility map approach to consumer surplus,
2-262-27
In weak complementarity measures, 2-31--2-33
Income effect. In consumer surplus measures, 2-12, 2-14,
2-17, 2-18
Incremental thresholds, 3-6
Indifference maps, in benefits measures, 2-10, 2-11
Inflation rates, and discount rates for benefits In
aggregation, 2-25
Institute for Behavioral Science, 3-8
Integral vistas, 1-2
Interest rates, and discount.rates for benefits In
aggregation, 2-25
Interviews
biases of Interviewers In, 4-23
In bidding methods, 4-54-6, 4-11, 4-13
costs of, 5-9
sample size determination and verification la, 4-26
4-28
selection factors for use of, 424
self-selection bias in, 4-27
Intrlllgator, Michael D., 4-28
Iterative bidding technique
application of, 4-3
bidding or valuation questions used in, 4-114-12
comparison of bidding methods studies with, 6-30, 6-32
Farmlngton study with, 6-12
Four Corners study with, 6-4
Lake Powell study with, 6-7
welfare changes on, 4-44-5
Ivea, B., 6-2, 6-6, 6-7, 6-9
Janes, T.H., 3-27
tain, John F., 6-34
Kalparowltz power plant, 3-27, 6-66-10
Kelly, J.R., 4-61
Kneese, A., 6-2, 6-11
Knetsch, J.L., 4-61
Kodachrome film, 3-28
Koshmleder's Law, 3-16
Knit Ilia, John V., 2-24
Kurz, Hordecal, 4-18
Lad. Frank, 2-2
Lake Powell study, 6-1, 6-66-10
aggregate benefits measures In, 6-9
analysis of results in, 6-76-9
application of bidding method in, 6-7
comparison of bidding methods studies with, 6-296-33
evaluation comments on, 6-96-10
problem formulation and scenario development In, 6-6
6-7
questionnaire used in, A-l
Lancaster, Kelvin J., 2-27, 4-30
Lansing, John 8., 4-23
Larcom, Lucy, 3-5
LaClmcr, Douglas A., 3-5, 3-7, 3-8, 3-12, 3-13, 3-14. 3-15,
3-16, 3-21, 3-27. 3-28
Layard, P.R.G., 2-16
Legislation
flexibility of Impairment determination in, 1-21-4
Integral vistas under, 1-2
monitoring In, 3-173-18
see also Clean Air Act
1-4
-------
Light
airborne part Ic 1m and scstterlng of, 3-16, 3-17
gases and scattering of, 3-163-17
visibility Impairment measures with, 3-12, 3-26
Light-absorption coetflclcnt, 3-16, 3-17
Light-cxttnctlon coefficient, 3-16
Light-scattering coefficient, 3-16, 3-17
Llnd. R.C., 2-23, 2-24
Locandcr, William. 4-17
Lochoan, Edna, 4-38, 4-40, 4-46, 4-54, 6-33, 6-48
Los Alamos Scientific Laboratory.(LASL), 3-27, 3-28, 5-12
Lucas, R.C., 4-65
HcConncll, C., 4-65
McConncll, Kenneth E., 2-20
HcFadden, Daniel L., 2-29, 4-60
Madov, William C., 4-23
Hall questionnaires, and costs, 4-244-25
Maler. K., 2-.29, 2-31, 5-17
Malm, W.C., 3-5, 3-7, 3-8, 3-9, 3-12, 3-13, 3-14, 3-16,
3-27, 3-28, 3-29
Manuel, Ernie, 4-68
Marginal Implicit price, 4-31
for residential property, 4-324-33, 4-48, 4-50
Marginal utility
in benefits measures, 2-10
In Indifference maps, 2-10
Marshall, Alfred, 2-10. 2-12
Hatched pairs technique, 4-54
in South Coast Air Basin study, 6-456-48
Memory colors, 3-27
Meteorological range
definition of, 3-13
perception measures and, 3-13
visibility and, 3-21
Mendelsohn, Robert, 4-66, 4-67, 5-17
Middle ton, W.E.K., 3-12, 3-16
Hlddleton's Lav, 3-16
Mlkesell, Raymond P., 2-23, 2-24, 2-25
Miller, Jon R., 2-2
Mlshan, E.J., 2-23, 2-24, 5-17
Modeling of visibility, 3-293-30
Monitoring, visibility, 3-173-26
EPA guidelines for, 3-18
federal regulations and, 3-17
natural conditions and, 3-213-26
need for, 3-173-18
programs for, 3-21, 3-223-24
technology used in, 3-183-20
Mood, A.M., 6-52
Morgan, James N., 4-23
Huellbauer, John, 2-27
Muopower, J., 3-8, 3-13
Murphy, James L., 6-64
Muth, Richard S., 2-27
National Center for Atmospheric Research JNCAR), 3-8
Rational Environmental Protection Act (NEPA), 5-3, 5-17
National Oceanic and Atmospheric Administration (NOAA),
3-21, 5-4
National Park Service (NFS), 2-3, 4-8
standard visual range of, 3-13
VIEW program of, 3-21
Nelson, Jon P., 4-46, 4-51, 6-33, 6-35, 6-38
Newburn, R.M., 3-21, 3-23, 3-24
New Haven, Connecticut, cost case study, 5-17
Nitrogen dioxide
light absorption by, 3-17
visibility Impairment from, 3-10
Nuclear Regulatory Commission (NUREC), 2-21, 4-3, 4-5,
4-22, 6-1. 6-9
Oates, Wallace E., 2-24
Option value, 2-2. 2-222-23
Ordinary consumer surplus (OCS)
correctness of, 2-122-14
deficiencies In, 2-9
demand curves with, 2-4, 2-62-9
selection of measures In, 2-22
site difference measures In, 2-192-20
Overlay techniques In photography, 3-29, 5-11
Osone, and visual Impairment, 3-10
Palmqulst, Raymond B., 4-29, 4-39
Particles, airborne
light absorption by, 3-17
visibility uiui!.ur«:.s Cor, 3-12, 3-163-17
visibility impairment from, 3-10
Patlnkln, Don, 2-17
Paulson, N.R., 3-9
Payment schemes
comparison of bidding method studies with, 6-30, 6-32
contingent market rejection from, 4-21
effect of, on bids, 4-19, 4-21
hypothetical biases In, 4-17
scenario development with, 4-12
Perception
cloud cover and, 3-7
content of scene and, 3-7
direct measures of, 3-12, 3-133-14
Interpretations of, 3-13-3
memory colors In, 3-27
neterologlcal range used In, 3-13
psychological factors in, 3-7
relative importance of environmental clues In, 3-8
scenic beauty and, 3-7, 3-8
standard visual range used in, 313
textures in, 3-9
thresholds in, 3-6
of visibility, 3-53-6
visibility impairment from, 3-4
Visual Air Quality Index (VAQI) for, 3-133-14
Photography, 5-115-12
color problems In, 3-273-28
effects of benefit measures of, 3-30
In Farmlngton study, 6-11, 6-156-16, A-2, AllA-13
film used by, 3-28
loss of contrast in, 3-28
modification techniques in, 3-29
in South Coast Air Basin study, A-34A-36
in visibility impairment, 3-3
Physical factors, in visibility impairment, 3-3
Plckford, Steward C., 6-636-64
Plndyck, Robert S., 4-28, 4-42
Plumes
Grand Canyon/Southwest Parks study of, 6-226-29
identification threshold for, 3-6
lake Powell study of, 6-6, 6-7, 6-10
perception of visibility and, 3-5
photography of, 3-28, 3-29
selection of estimation method for, 5-7
visibility impairment from, 3-3
PLUVUE model, 3-30, 5-5
Pollcymaklng, and benefit-cost analysis, 1-5
Pollnsky, Mitchell A., 2-23, 4-54, 4-56
Pollak, R., 2-27
Pollution
atmospheric chemistry of, 3-10, 3-11
particle light-scattering by, 3-17
property values related to, 4-424-44
visibility Impairment from, 3-3, 3-10
Population data
discount rates for benefits in aggregation and, 2-25
selection of estimation method and, 5-9
Portney, Paul R., 4-44, 4-71
Power plans
Farmlngton study on, 6-106-16
Four Corners study In, 6-26-6
Grand Canyon/Southwest Parks study of, 6-22
Lake Powell study on, 6-66-10
visibility benefits In benefit-cost analysis and,
6-606-62
Prescribed burning, see Burning
Price
attributes in market good and, 4-304-31
consumer surplus measures of changes in, 2-162-17
demand curve methods with, 2-42-9
expenditure function approach to consumer surplus and,
2-292-31
hedonlc, see Hedonlc price technique
income compensated demand curve (ICDC) for, Z-I22-14
2-15
indifference saps with, 2-10, 2-U
Batched pairs technique far, 4-544-55
reference level of wi-H'ure changes and,' 2-21
in residential location models, -»-55>-56
weak complementarity measures In, 2-312-33
1-5
-------
Probability sample. In bidding methods, 4-25
Property values
bidding molhods and, 4-44-5
consumer iiuri'Lus measure of visibility and, 2-21
housing market segmentation from hedonlc price function
In, 4-46
South Coast Air Basin study of, 6-176-21
see also Residential property value studies
Psychological factors
perceptual thresholds and, 3-7
visibility Impairment analysis and, 3-2, 3-3, 3-4
Fsychophyslcs
light intensity measures In, 3-14
visibility Impairment analysis with, 3-1, 3-3
Public goods, In benefit-cost analysis, 2-2
Quality, air, see Air quality
Quantitative measures
atmospheric science and, 3-1
consumer surplus measures of, 2-17--2-19
Quasl-opclon value In benefit-cost analysis, 2-2
Questionnaires
In bidding methods, 4-5, 4-6, 4-11
cost of use of, 4-244-25
In Farmlngton study, A-3A-10
in Lake Powell study, A-l
Office of Management and Budget (OMB) procedures for,
4-6
selection factors for use of, 4-24
In South Coast Air Basin study, A-17A-33
Qulgley, John H., 6-34
Radian Corporation, 5-5
Randall, S., 2-19, 2-20, 4-3, 4-17, 4-19, 4-21, 6-1, 6-2,
6-4.
6-5, 6-7, 6-9, 6-11, 6-15, 6-16, 6-29, 6-33
Random samples, In bidding methods, 4-254-26
Ranked attributes/market share technique, 4-604-61
Raylelgh scattering, 3-163-17
Recreation areas
Four Corners study of, 6-26-6
Grand Canyon/Southwest Parks study of, 6-226-29
hedonlc travel cost approach to, 4-664-67
inpact of congestion on, 2-29
Lake Powell study in, 6-66-10
ranked attributes/market share techniques in, 460
residential property value studies in, 4-59
sampling procedures for, 4-26
travel cost method for benefits of, 4-61
willingness to pay for visibility demand curve in,
2-62-8
Regression analysis in bidding methods, 4-284-29
Rental price, in residential location model, 4-56
Residential property value studies, 4-1, 4-294-60
actual market approaches of, 4-1, 4-2
assumptions In use of, 5-9, 5-10
benefit estimate for air quality price In, 4-39
benefits measured with, 4-584-60
Boston study with, 6-386-41
case studies for, 6-356-58
central business district distances and air pollution
In, 4-404-41
conditions for use of, 4-564-57
Denver study with, 6*-416-44
functional form In, 4-424-45
hedonlc price technique used in, 4-314-36, 4-37
4-53
hypothesis behind, 4-29
independent variables In, 4-394-41
Batched pairs technique in, 4-544-55
price as dependent variable In, 4-374-39
rental market price in, 4-38
residential location model in, 4-554-56
San Francisco Bay area study In, 6-486-52
(election criteria for, 5-6, 5-7, 5-8
South Coast Air Basin study with, 6-456-48
strengths and weaknesses of use of, 4-56
variations in techniques used in, 4-544-56
willingness to pay estimates In, 4-474-53
Retouch techniques In photography, 3-29, 5-11
Ridker, Ronald C., 4-29, 4-50
Rosen, Sherwln, 4-30, 4-31, 4-47, 4-69, 6-35, 6-51
Rowe, R., 2-27, 4-3, 4-19, 4-20, 4-21, 4-22, 5-12, 6-1,
6-3, 6-9. 6-10, 614, 6-15
Rubinfuld, Daniel L., 4-28, 4-36, 4-40, 4-42, 4-44, 4-49,
6-33, 6-38, 6-40, 6-43, 6-53
Sampling procedures, In bidding methods, 4-254-26
Samuelaon, P.A., 4-18
Sandier, Todd, 2-24
San Francisco Bay area study, 6-33, 6486-52
benefit estimates in, 6-51
comparison and review of hedonlc approach studies with,
6-526-58
evaluation comments on, 6516-52
hedonlc price function estimates in, 6-496-50
housing market segmentation in, 4-46
study area description and data selection in, 6-48
6-49
HTP estimation In, 6-506-51
Sassone, Peter 0., 1-5, 2-23, 2-24, 5-17
Scattering of light, 3-163-17, 3-21, 3-26
Scenarios
air quality on, 4-8, 4-94-10
bidding methods and, 4-74-11, 5-11
bidding methods comparisons on, 6-29, 6-30
characteristics of, 4-84-9
establish existing conditions in, 5-4
in Farmlngton study, 6-11
in Four Corners study, 6-3
"future with," 5-4
"future without," 5-4
Grand Canyon/Southwest Parks study, 5-23
Iterative bidding procedure in, 4-13
in Lake Powell study, 6-7
payment procedures in, 4-12
sample procedures in, 4.-13
in South Coast Air Basin study, 4-104-11, 6-17
value selection approach in, 4-13
in visibility benefits analysis, 5-45-5
Schaffer, William A., 1-5, 2-23, 2-24, 5-17
Scherr, Bruce A., 4-19
Schmalensee, R., 2-2
Schulze, W.. 6-6, 6-7, 6-9, 6-17, 6-45
Sex differences. In perception, 3-8
Shavell, Steven, 4-54
Sllberberg, Eugene, 2-21
Slnden, J.A., 2-3
Smith, V. Kerry, 2-24, 2-29, 4-54, 4-56, 4-61, 4-64,
4-69, 6-52
Smith, Vernon L., 4-19, 4-71
Soil dust, and visibility, 3-11, 3-21, 3-26
Solar angle, 3-7
Sonstelle, Jon C., 4-44
South Coast Air Basin (SCAB) study, 6-1, 6-176-21
aggregate benefit measures In, 6-19, 6-21
analysis of results of, 6-19, 6-20
benefit estimates in, 6-47
bidding methods used In, 6-59
comparison and review of hedontc case studies with,
6-526-58
comparison of bidding methods studies with, 6-296-33
evaluation comments on, 6-21, 6-476-48
hedonlc price function estimation in, 6-466-47
Information biases in, 4-20, 4-21
instrument design in. 4-23
iterative bidding techniques in, 4-3
matched pairs technique in, 4-544-55, 6-456-46
6-466-47, 6-48
payment vehicle problems In, 4-21
problem formulation and scenario development in, 6-17
property value approach in, 6-33, 6-456-48
scenario development In, 4-104-11
tudy area description and data selection in. 6-45
6-46
urvey Instrument used in, A-14A-36
WTP function estimation In, 6-47
Standard visual range, 3-13
Stankey, C.H., 4-65
State implementation plans (SIPs), 1-2, 5-3
Stiles, U.S., 3-5, 3-15
Stoll, J., 2-19, 2-20, 4-17, 6-9, 6-16
Straszhelm, M.ihlon, 4-46
Sulfur dioxide, in .ilr quality studies, 4-70
Sumka, Howard J., 4-51, 6-38
1-6
-------
Survey Inxtrumcnt
bidding or valuation questions on, 4-11--4-13
choice of. 4-244-25
co«ti of use of. 4-244-25
design of, 4-54-6, 5-12
evaluation of problem bids In, 4-14
general Introduction and statement of purpose la,
4-64-7
hypothetical biases In. 4-164-17
Introductory non-valuation questions In, 4-7
cenarlo development and market definition In, 4-7
4-11
ectlons of, 4-64-14
special options, closing questions und remarks In,
4-14
zero bide In, 4-14
Systems Applications, Inc. (SAI), 5-3, 5-5, 6-42
Taxation, In visibility benefit analysis, 4-71
Telephone survey costs, 4-24
Textures, In perception, 3-9
Thayer, M., 4-21, 4-22, 4-27, 4-66, 6-17, 6-33, 6-45
Thresholds, perceptual, 3-6
bsoluce, 3-6
concent of scene and, 3-7
definition of, 3-6
Identification, 3-6
Incremental, 3-6
light Intensity measures and, 3-16
Psychological factors In, 3-7
visibility Impairments and, 3-4 '
In Visual Air Quality Index (VAQI), 3-14
Tldenan, Nlcolaus, T., 4-71
Travel cost methods, 2-9, 4-614-67
assumption In use of, 5-9, 5-10
generalized presentation of, 4-624-63
hedonlc travel cost approach and, 4-664-67
problems with, 4-634-64
election criteria for, 5-7
fltrcngths and weaknesses of, 4-63
visibility benefit analysis with, 4-644-65
Irljonls, J., 3-13
Tullock, Cordon, 4-71
Urban areas, and air quality, 1-2
User value, see Activity value
Utah coal production, 3-27, 3-30
Utility, In benefit-cost analysis, 2-1
Utility maps
benefits measures with, 2-102-12
consumer surplus measures with, 2-262-27
Value selection. In bidding methods, 4-13
VIEW (Visibility'investigative Experiment In the West)
benefit analysis and, 3-30
visibility monitoring with, 3-21, 3-223-24
Visibility
in Class I federal areas, 1-2, 1-3, 1-5
demand curve examples of, 2-62-9
human perception of, 3-53-6
nap of existing visual range In, 3-21, 3-25
natural conditions and, 3-213-26
property rights and consumer surplus measures of,
2-21
Visibility benefit analysis
ir quality values considered in, 2-2
atmospheric science applied to, 3-13-3
benefit measures tranferred across studies and, 4-72
4-74
ease studies of, 6-16-65
EPA reference documents on, 1-8
guidelines In, 5-15-17
household production function approach In, 4-68
overview of, 1-51-7
quantitative measures used In, 3-1
ranked attributes/market share acthod In, 4-604-61
steps in, 1-51-7
travel coat method applied to, 4-644-65
voting approaches to, 4-704-72
wage and salary differentials method In, 4-684-7C
see also Air quality benefit analysis
Visibility Impairment
atmospheric chemistry and, 3-10
combustion In, 3-10. 3-11
definitions of, 3-12
direct measures of human perception in, 3-12, 3-13
3-14
factors affecting, 3-33-5, 3-103-11, 3-213-26
Indices of, 3-123-17
light Intensity measures In, 3-12, 3-143-16
meteorological range in measure of, 3-13
modeling used in, 3-293-30
monitoring of, 3-173-26
natural sources of, 3-21, 3-26
optical property measures used In, 3-12, 3-16317
physical factors In, 3-3
pictorial graphic displays of, 3-2£3-29
pollutant gases in, 3-10, 3-11
psychological factors In, 3-4
sequential process used in, 3133
sources of, 3-93-11
standard visual range for, 3-13
use of term, 3-43-5
Visual Air Quality Index (VAQI) for, 3-133-14
Visibility legislation, 1-2
flexibility of Impairment determination in, 1-21-4
Integral vistas under, 1-2
monitoring emphasis In, 3-173-13
VTSTTA (Visibility Impairment due to Sulfur Transforma-
tion and Transport in the Atmosphere), 3-30
Visual Air Quality Index (VAQI)
disadvantages of, 3-14
human perception of visibility measured on, 3-12,
3-133-14
light Intensity measures on, 3-16
perception differences on, 3-73-8
Voting approaches, 4-704-72
problems In use of, 4-714-72
taxation used in, 4-71
Wachter, M., 2-27
Wage and salary differentials, in visibility benefits
analysis, 4-684-70
Walters, A.A., 2-16
Walther. Elc C., 3-21, 3-23, 3-24
Washington, D.C., study, 6-33, 6-356-36
benefit estimates of, 6-37
comparison and review of hedonic studies with, 6-52
6-58
evaluation comments on, 6-376-38
hedonlc price function estimation In, 6-36
study area description and data selection in, 6-35
6-36
WTP function estimation in, 6-366-37
Water hazes, and visibility impairment, 3-26
Water Resource Council, 2-24, 5-15
guidelines from, A-41A-43
Weak complementarity. 2-312-33, 4-58
White, M., 3-17
Williams, M., 6-2, 6-11
Williams, M.D., 3-14, 3-27, 5-12
Wllllg, Robert D., 2-19, 2-20, 6-9, 6-16
Willingness to accept compensation (WTA), in bidding
methods, 4-3, 4-4, 4-22
Willingness to pay (WTP)
aggregation across individuals and, 2-24
air quality standards and housing and, 2-9
as benefits measure, 2-32-4
In bidding methods, 4-3, 4-4
bidding or valuation questions on, 4-11413
Boston study with, 6-396-40
consumer surplus measures and, 4-4
demand curves with, 2-4, 2-62-9
Denver study with, 6-43
elasticities of, in hedonic case studies, 6-576-58
hedonlc price technique with, 4-31, 4-33, 4-35, 4-47
4-53, 5-15
Identification of. In hedonic price technique, 4-50
4-52
Information biases In bidding methods and, 4-20
4-21
positive Income effect and, 2-17
1-7
-------
in residential property value studies, 4-29
San Francisco Bay area study with, 6-506-51
South Coast Air Basin study with, 6-47
voting approaches and, 4-70, 4-71, 4-72
Washington, D.C., study with, 6-366-37
weak complementarity measures with, 2-33
Wllman, Elizabeth A., 4-29, 4-59
Wlttc, Ann D., 4-51, 6-38
Wyszeck.1, C., 3-5, 3-15
Wyiga, Ronald E., 5-17
Zero bids, 4-14
I-B
------- |