| 7.2.2 Quantifying
Benefits |
| |
| Just as identifying the benefits of a regulation requires qualitatively describing the first and second functional relationships presented in Figure 7-1, quantifying benefits requires calculating the effects that changes in emissions have on environmental service flows. This involves quantifying changes in emissions, using fate and transport models to estimate the corresponding changes in ambient concentrations of pollutants in environmental media, and then estimating dose-response or concentration-response relationships to translate these changes in ambient concentrations into quantitative changes in environmental damages. |
|
| 7.2.2.1 Quantifying
Health Benefits |
|
| In general, quantifying health benefits involves six steps. These steps include determining the following: |
|
| 1. the dose-response relationship for each health effect, |
| |
| 2. total exposure in the absence of the regulation, |
| |
| 3. the number of baseline cases for each quantifiable health effect, |
| |
| 4. total exposure with the regulation (for each regulatory alternative), |
| |
| 5. the number of cases for each quantifiable health effect with the regulation (for each regulatory alternative), and |
| |
| 6. the number of cases avoided as a result of each alternative. |
| |
| Each of these steps is described below. |
|
| Step 1: Determining the Dose-Response Relationship for Each Health Effect |
|
| A dose-response relationship is an estimate of risk per unit of exposure to a pollutant. For cancer assessments, dose-response has typically been modeled as a linear no-threshold relationship; that is, every unit of exposure contributes equally to aggregate risk. For example, 100 units of human exposure result in a given amount of risk, regardless of whether there is a one-time exposure of 100 units to one person, ten exposures of one unit each to ten different people, or one-time exposures of one unit each to 100 different people. However, the Agency has been moving toward the use of nonlinear health effects models. These models may become more prevalent in future analyses. For noncancer health effects, dose-response relationships may include a threshold and/or may be nonlinear with respect to the level of exposure. |
|
| Step 2: Determining Baseline Exposure |
|
| Human exposure to a pollutant is a function of ambient concentrations of that pollutant in environmental media. Determining baseline exposure requires estimating two parameters. First, analysts must identify the exposed populations (e.g., occupational groups, consumers of particular products, the general population living in a particular area) and the number of exposed individuals in each group. Then one needs to determine the level, duration, route, and frequency of exposure (e.g., 100 ppm time-weighted average exposure for an 8-hour day, 50 days per year). |
|
| At this point in the discussion, it is important to note that individuals may take action to reduce their exposure to harmful substances. For example, individuals may purchase bottled water to avoid exposure to contaminants found in tap water. Such averting behaviors affect the size of the exposed population. Therefore, any exposure assessment should take into account individual actions designed to reduce exposure. However, inclusion of such behavior is usually, at best, only implicit in the exposure assessment. As a result, in the remainder of this section, the discussion of averting behavior in an economic analysis is limited to the use of averting expenditures as a proxy for individual WTP to avoid the health effects expected to result from human exposure to environmental contaminants (see Section 7.2.3.1). |
|
| Step 3: Determining the Number of Baseline Cases for Each Quantifiable Health Effect |
|
| The number of baseline cases for each quantifiable health effect is the product of the baseline number of people exposed, the amount (level, duration, and frequency) of baseline exposure, and the dose-response relationship; that is, the baseline number of cases for each exposed population equals: |
|
| Number exposed x Baseline exposure x Dose-response relationship. |
|
| Step 4: Determining Exposure after the Regulation (for Each Regulatory Option) |
|
| Each regulatory option may result in a reduction in the exposure level, a reduction in the exposed population, or some combination of the two. Therefore, there are two dimensions to characterizing post-regulatory exposure. First, it is necessary to estimate the impact of the option on exposure levels. For example, a regulatory option may reduce (but not eliminate) the exposure to the pollutant from a particular exposure pathway. Second, it is necessary to estimate the expected post-regulatory level of exposure for that exposure pathway. |
|
| Step 5: Determining the Number of Cases for Each Quantifiable Effect with the Regulation (for Each Regulatory Option) |
|
| In this step, Step 3 is repeated, using the post-regulatory estimates of exposure (number of people exposed; frequency, duration, and level of exposure) derived in Step 4. Multiplying the new exposure estimates by the dose-response estimate yields an estimate of the post-regulatory number of cases. |
|
| Step 6: Determining the Number of Cases Avoided as a Result of Each Regulatory Option |
|
| To determine the number of statistical cases avoided as a result of a regulatory option, the number of post-regulatory cases (from Step 5) is subtracted from the baseline number of cases (from Step 3). The difference is the quantified benefits of the regulation as follows: |
|
Quantified
Health
|
|
| Baseline Cases Resulting from Exposure to Pollutant |
|
|
|
| Post-Regulatory Cases Resulting from Exposure to Pollutant |
|
|
|
|
Estimating the concentration-response relationship
provides an alternative to the six-step process outlined above. Although
estimating individual risk per unit of exposure and then multiplying this
estimate by total exposure is the more accurate method of quantifying
health effects, the data required for such an analysis are often not available.
In those cases, it may be possible to estimate the extent of health
effects as a function of ambient concentrations of pollutants in the atmosphere.
This process involves the determining the following:
1. the concentration-response
ratio for each health effect,
2. ambient pollutant concentrations
in the absence of the regulation,
| 3. ambient pollutant concentrations with the regulation (for each regulatory alternative), |
| 4. the number of cases for each health effect with the regulation (for each regulatory alternative), and |
| 5. the number of cases avoided as a result of each alternative. |
This process differs from the six-step process described
above only in that the estimation of a dose-response relationship and
exposure for each health effect for each regulatory alternative are combined
into one step—the estimation of a concentration-response relationship.
A concentration-response relationship is an estimate of the number
of health effects associated with a given ambient pollutant concentration
(e.g., number of asthma attacks associated with a particular concentration
of ground-level ozone). If the ambient concentration of the pollutant
is known both with and without the regulation, then a concentration-response
relationship can be used to estimate changes in the number of health effects
resulting from the regulation. Examples of concentration-response
relationships for health effects developed in the literature include Schwartz
(1993), Ostro and Rothchild (1989), and Pope (1989).
7.2.2.2 Quantifying
Ecological Benefits
Quantifying ecological benefits is similar to
quantifying health benefits using concentration-response relationships.
In particular, when the effect of a regulation on the ambient concentrations
of a pollutant can be estimated, and the relationship between ambient
pollution concentrations and environmental service flows is known, then
the ecological benefits of a regulation can be quantified. For example,
the National Crop Loss Assessment Network (NCLAN) developed concentration-response
relationships linking ground-level ozone to leaf damage and reduced seed
size in an effort to determine the effect of ozone on crop yields. Other
sources of concentration-response relationships for ecological damages
include Martin and Banzhaf (1994); Heagle, Spencer, and Letchworth
(1979); Kopp, Vaughan, and Hazilla (1984); Brewer and Ashcroft
(1982); Oshima et al. (1976); Leung, Reed, and Geng (1982);
Foster et al. (1983); Rowe and Chestnut (1985); Clark,
Henninger, and Brennan (1983); and Keller (1985).
7.2.2.3 Quantifying
Benefits through Nonliving Systems
Degradation of many materials, including metals,
masonry, and paints, is accelerated by exposure to pollutants. For
example, corrosion rates for galvanized steel and zinc as well as erosion
rates of carbonate stone have been studied and strongly linked to acidic
deposition. Building stone, which includes limestone and marble,
is also affected by pollutant emissions. Painted surfaces incur
the largest economic losses from pollutant exposures ( Bernabo et al.,
1988). Although the physical links between pollutant emissions and
materials damages are well established, the precise quantitative relationship
between degradation rates and exposures are not. The effectiveness
of preventive maintenance has further complicated the assessment of pollution
damages. Therefore, quantifying the effects of pollutant emissions
on materials and structures may be quite difficult. Nonetheless,
because reductions in materials damages could constitute a significant
economic benefit of an air quality regulation, it is important to describe
these effects to the greatest extent possible, even if only qualitatively.
| 7.2.3 Monetizing
Benefits |
Monetizing the benefits of a regulation involves estimating
society’s willingness to pay (WTP) for quantified changes in environmental
service flows. In economics, WTP refers to the maximum amount an
individual is willing to pay to acquire a benefit. It is measured
as the reduction in income required to return an individual to the level
of utility he or she enjoyed prior to receiving the benefit. An
alternative measure of economic value is an individual’s willingness
to accept (WTA) compensation to forego a benefit. Conceptually,
WTA is measured as the minimum compensation to required for an individual
to achieve the same level of utility he or she would have attained if
the benefit had been realized. From a theoretical perspective, the
appropriate measure (WTP or WTA) depends on the implicit property rights
in the valuation context, and the two measures should be close in value.
WTP estimates have been generally preferred in the empirical economics
literature. The remainder of this section is devoted to describing
the methods used to estimate WTP for environmental improvements.
7.2.3.1 Monetizing
Health Benefits
Nonfatal Illness and Injury (Morbidity).
Economists use a number of approaches to monetize a change in the
number of cases of a particular morbidity effect. The four primary
approaches are cost-of-illness (COI) methods, expressed preference methods,
averting action methods, and hedonic wage and property value methods.
Of these, COI methods are most often employed in economic analyses
of human health benefits. Other methods include risk-risk tradeoffs,
health state indexes, and damage award approaches. The four primary
approaches are examined below.
Cost-of-Illness Approach. Because of the
difficulties associated with generating or using estimates of WTP for
reductions in the risk of nonfatal illness or injury, analysts often prefer
valuation methods based on the avoided costs of illness or injury. The
COI approach to morbidity valuation measures the direct and indirect costs
resulting from a health effect. Direct costs include such things
as the value of goods and services used to diagnose and treat individuals
suffering from the health effect. Indirect costs consist primarily
of foregone productivity measured by lost wages. Total COI is the
sum of direct and indirect costs. Because the COI approach does
not account for the full range of costs associated with an illness or
injury (e.g., pain and suffering are not included), the results of these
analyses should at best be viewed as lower-bound estimates of society’s
WTP for reductions in such risks.
As mentioned above, the theoretically appropriate
measure of economic benefits is society’s WTP to reduce the risk
of a health effect (i.e., society’s WTP to avoid the risk prior to
the actual occurrence of a health effect). It is important to note
that the COI approach measures costs after a health effect has occurred
rather than an individual’s WTP to avoid the health effect in the
first place. In addition, the COI approach measures the costs to
the individual (out-of-pocket costs and lost wages), the costs to the
individual’s employer (in the case of paid sick leave), and the costs
to third-party payers (payment of insured medical expenses or charity
care), but not changes in individual well-being caused by the illness.
There are two primary approaches to measuring
COI: the prevalence approach and the incidence approach. The
prevalence approach measures total costs of a particular illness or injury
in the population for a given year. In contrast, the incidence approach
measures the cost of an individual case of illness from onset through
recovery or death. For assessing the benefits of OAQPS regulations,
the incidence approach to measuring the cost of an illness or injury is
most appropriate because the health benefits of OAQPS regulations are
typically expressed in terms of the number of cases of a particular health
effect avoided as a result of the regulatory action. Use of the
prevalence approach would be appropriate if the effect of a regulation
eliminated a particular health effect.
Expressed Preference Methods. Expressed
preference methods, including contingent valuation (CV) and conjoint analysis,
can be used to elicit an individual’s WTP to avoid a given health
effect. CV techniques are the primary expressed preference method
used to estimate health effects. CV uses surveys to directly elicit
an individual’s WTP to reduce the risk of a given health effect.
In particular, a CV survey asks each respondent about his or her
personal characteristics, attitudes concerning the commodity being valued
(reductions in the risk of an adverse health effect in this case), and
WTP to acquire the commodity. From the survey responses, analysts
can estimate WTP to reduce the risk of an adverse health effect as a function
of personal characteristics and attitudes.
Although they are the most broadly applicable
valuation methods, CV techniques remain controversial because of their
hypothetical nature. In particular, there are two primary criticisms
of CV. First, because an individual will not actually have to pay
the amount he or she indicates in response to the survey, the individual
may have little incentive to respond truthfully. Second, critics
contend that survey respondents often are unable to comprehend the commodity
they are being asked to value; particularly when the commodity is a very
small change in the risk of experiencing a health effect. Proponents
of CV assert that many of the problems raised by critics can be effectively
controlled through good survey design. Although OAQPS generally
does not conduct original CV studies due to time and other resource constraints,
analysts should be mindful of these criticisms when using existing CV
estimates to monetize benefits.
Rather than directly asking an individual his
or her WTP, conjoint methods ask individuals to choose among different
sets of alternatives. Each alternative is broken down and described
according to a common set of attributes, usually including the health
effect of interest and some monetary measure. The levels of these
attributes are varied across alternatives. By analyzing an individual’s
choices across a number of pairs of alternatives, analysts can derive
estimates of the trade-offs the individual is willing to make among the
different attributes (including cost), thereby estimating the individual’s
WTP to acquire a change in an attribute such as reduced risk of a specific
health effect. Although conjoint analyses show promise in valuing
health effects, these techniques are not commonly used for such a purpose.
Averting Action Methods. In the face of
a potential risk, individuals will often take defensive or averting action.
For example, in the case of groundwater contamination, averting
action might include purchases of water filters, bottled water, and other
alternative water supplies. In the case of air pollution, averting
actions may include such things as avoiding going out of doors. These
types of observable behavior can provide analysts with information about
an individual’s WTP to avoid specific health risks; however, more
generally they provide information about the costs of these behaviors,
their relation to the source of such risk, and the magnitude of cost savings
that would result from controlling the source.
As noted above, examining averting behavior provides
the analyst with information regarding individuals’ WTP to avoid
exposure to a source of risk. Averting actions taken in response
to risk may also enter the regulatory analysis in the risk assessment
to the extent that these actions reduce the overall exposure of the population
to harmful pollutants. If the exposure assessment of a regulatory
analysis takes into account changes in averting behavior between the baseline
and control scenarios, then the cost savings attributable to the reduced
need for averting action can be considered one component of the benefits
of the rule. When examining expected reductions in averting behavior,
the analyst must be careful not to double count. In particular,
analysts must estimate the benefits to those who, because of the rule,
change their averting behavior separately from those who experience a
reduction in exposure (and subsequent risk) without changing their own
averting behavior. To the former, averting expenditures measure
potential cost savings resulting from the rule. To the latter, averting
expenditures provide an estimate of WTP to avoid exposure.
Averting action methods are often preferred to
expressed preference methods because they are based on actual behavior;
however, measuring the benefits and costs of averting actions is often
difficult. First, to determine an individual’s WTP to avoid
a specific health effect by examining his or her averting behavior, it
may be necessary to determine the benefit the individual expects to receive
as a result of his or her averting actions. Second, averting actions
are typically taken to avoid a health outcome by reducing the individual’s
exposure to a harmful event. Therefore, WTP measures estimated based
on averting actions are usually employed as estimates of WTP to avoid
exposure and not as measures of WTP to avoid a specific health effect.
Therefore, using averting action methods to value specific health
outcomes of an OAQPS rulemaking may be difficult.
Hedonic Methods. Two common applications
of hedonic methods are used to value changes in health risks: hedonic
wage models and hedonic property value models. Hedonic wage models
are based on the premise that, all else equal, an individual working in
a risky occupation will require higher compensation for his or her labor
than will the same individual working in a less risky occupation. By
examining wage differentials among individuals working in various occupations,
analysts can estimate a worker’s WTA for exposure to higher levels
of risk on the job.
Hedonic property value models are based on the
theory that the price of a residential property is equal to the discounted
present value of the lifetime residential value of the home. The
price of the home can then be estimated as a function of its structural
characteristics and the characteristics of the surrounding community.
By estimating housing prices over a range of properties, each with
varying structural and community characteristics (including air quality),
analysts can infer a household’s WTP for each of these characteristics.
For a good example of a hedonic study designed to estimate the value
of avoiding environmental health risks, see Mendelsohn et al. (1992).
Fatality. The benefits of OAQPS regulations
can also include reductions in the risk of premature death. The
economics literature discussing the value of changes in fatality risks
is rather extensive and provides a relatively strong basis for monetizing
benefits when the number of deaths avoided as a result of a regulatory
action can be calculated.
Value of a Statistical Life (VSL). Monetary
estimates of changes in fatality risk are often expressed in terms of
the VSL. The term “value of a statistical life” is easily
misinterpreted and should be carefully described when used in benefits
analysis. In particular, VSL refers to the WTP for reductions in
the risk of premature death aggregated over the population experiencing
the risk reduction; that is, VSL refers to the sum of many small reductions
in fatality risks. For example, if the annual risk of death is reduced
by 1 in 1,000,000 for each of 2,000,000 people, then two statistical lives
are saved each year as a result of the risk reduction measures. If
each individual is willing to pay $5 for the risk reduction of 1 in 1,000,000,
then the value of each statistical life saved is $5 million.
The basic assumption underlying the VSL approach
is that equal increments in fatality risks are valued equally. For
example, a reduction in the risk of death of 1 in 1,000,000 is valued
the same whether or not the original fatality risk was 1 in 100,000 or
1 in 1,000,000. This assumption is generally defensible if the level
of risk prior to the change is small (usually 1 in 100,000 or less) as
is usually the case for fatality risks resulting from environmental hazards.
Because economic theory maintains that the marginal utility of risk
reduction is an increasing function of the level of baseline risk, this
assumption may not be valid for risks greater than 1 in 100,000. For
similar reasons, the VSL approach is only appropriate for marginal changes
in the risk of death and should not be used to value more significant
changes. Because changes in individual fatality risks resulting
from environmental regulation are typically very small, the VSL approach
is usually acceptable for OAQPS benefits analyses.
The literature indicates that empirical estimates
of WTP for risk reduction are sensitive to whether the risk is borne voluntarily.
In general, it is believed that risk-averse individuals have a greater
WTP for marginal risk reductions than those who choose risks voluntarily.
Because most VSL studies are based on wage compensation data, the
results are applicable to changes in voluntary risks, and the population
over which these values are estimated is often not representative of the
population affected by a regulation; however, because, as noted above,
fatality risks from both involuntary environmental hazards and occupational
fatality risks tend to be quite small, using VSL estimates developed using
wage compensation data is reasonable.
In the past, most EPA analyses have used point
estimates of VSL derived from the economics literature. There are
two alternatives to this approach. First, analysts could apply a
range of VSL values found in the existing literature to develop upper-
and lower-bound estimates of the total value of lives saved as a result
of a regulation. This approach, however, often gives equal weight
to very high and very low estimates that may be found infrequently in
the literature. Alternatively, analysts could use distributions
of VSL values rather than point estimates or ranges. Distributions
give quantitative weights to the likely accuracy of the different VSL
estimates based on the frequency with which they are found in the existing
literature.
In assessing the benefits of the CAA, analysts
conducted a meta-analysis of VSL estimates found in the literature and
fit the results to a Weibull distribution to characterize the range of
possible benefit values ( EPA, 1996a). This approach revealed that
the majority of VSL estimates found in the literature tend to cluster
in the range of $3 million to $7 million, with a central estimate of approximately
$5 million. A review of methods used by EPA in valuing changes in
mortality risks conducted for OP by Chestnut, Mills, and Alberini
(1997) notes that there is an insufficient empirical basis to justify
the disparity in VSL estimates chosen across programs within the Agency.
Although this report does not recommend specific VSL estimates to
be used in EPA economic analyses, it does suggest that the central tendency
of $5 million revealed in the CAA benefits analysis may provide the best
starting point.
Value of a Statistical Life Year (VSLY). An
alternative method of expressing reductions in mortality risk is the VSLY.
For example, if a regulation is estimated to save one statistical
life among a population of working adults whose average life expectancy
is 40 years, then the regulation would result in 40 life-years extended.
In general, there are two methods for valuing the number of life-years
extended:
- applying results from studies in which the WTP for a risk reduction
is estimated as a function of age; and
- annualizing VSL estimates using an appropriate discount rate
and average life expectancy.
|
Both of these approaches have some fundamental shortcomings. In particular,
when deriving values of risk reduction as a function of age, the type
of risk and the size of the marginal change in risk used to estimate the
value of a risk reduction must closely match the risk reductions resulting
from the regulation. Annualizing VSL estimates does not provide
an independent estimate of VSLY but simply rescales the VSL estimate.
Current research does not suggest a definitive method for estimating
VSLY that is sensitive to such variables as current age, latency period,
life-years remaining, and the social value of different risk reductions.
7.2.3.2 Monetizing
Ecological Benefits and Benefits to Materials and Structures
Once changes in ecological service flows have
been identified and quantified, analysts can use a number of valuation
methods to monetize the benefits of a regulation. The four primary
methods for monetizing ecological benefits are
- hedonic property value models,
- travel cost models,
- expressed preference methods, and
- market models.
|
Table 7-1 shows the categories of benefits for which each of these valuation methods
is most applicable. Each of these methods is briefly described below.
Hedonic Property Value Models. As
noted above, hedonic pricing theory maintains that the housing market
functions as a market for environmental quality insofar as environmental
quality is a characteristic of the property being purchased or of the
community in which the property is located. Although often used
to assess WTP to avoid environmental health risks , hedonic property value
models can be used to estimate the value individuals and households place
on the perceived amenity and recreation benefits provided by the property.
In addition, hedonic methods could conceivably be used to estimate
housing damages related to pollutant exposure; however, these methods
have not been used for this purpose to date.
Travel Cost Models. Travel cost models
have become a common method in the economics literature for estimating
the benefits of environmental improvements to recreators. For example,
these models can be used to estimate anglers’ WTP to reduce toxic
contamination of a water body that otherwise would be subject to a fish
consumption advisory (see, for example, Montgomery and Needelman,
1997). Most often, travel cost models take the form of a discrete-choice
model known as a random utility model (RUM) in which a recreator’s
choice among a set of unique recreation sites is modeled as a function
of the
|
![[logo] US EPA](http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
