236
are difficult to measure because an extrapolation
must be made from the observation of particular
choices made by the consumer to general conclusions about preferences. This implies that the researcher must be able to hypothesize the functional
form and underlying theory that connect the observed market behavior with the unobservable and
still hypothetical welfare change that would result
from the policy scenario under consideration. As
well, direct methods allow the researcher to hypothesize assumptions about the respondents' preferences and then test these assumptions to verify a
respondent's motives. Revealed preference methods (also known as indirect methods, because they
indirectly estimate demand for a nonmarket good
based on some observed demand for a related market good) do not easily allow for these tests. Finally, arguments in favor of direct methods include
the fact that there is a real distinction between private market goods and public goods, which make
revealed preference methods harder to apply, especially in the case of public goods, which are often of national rather than local scope.
Contingent Valuation
The contingent valuation method (CVM) uses carefully worded surveys to gather primary data that is
analyzed by an appropriate econometric model to
predict what people would be willing to pay (or
willing to accept in compensation) for a specified
change in the price or the quality of an environmental amenity. The CVM has been shown to work
especially well for recreational use values for goods
with which respondents have had prior experience.
The term contingent refers to the fact that the valuation of the good is contingent on the hypothetical assumption of a plausible market and method
of payment for the good. A simple CVM study
might consist of a detailed description of a good,
followed by a question asking the respondent
whether he or she would be willing to pay a given
price for provision of the good. The yes or no responses to a range of prices can be used to establish a cumulative probability distribution of a representative individual's probability of answering
yes. The area beneath is the expected value of a yes
response and thus the average willingness to pay
(WTP) for the environmental change in question.
This example is based on a dichotomous choice
model CVM, since respondents are given the option to say yes or no to given prices. Contingent
valuation can be accomplished with other models
as well, including those that ask follow-up questions after the first response.
Measuring Ecosystem Values
The benefit to individuals is measured in terms
of their maximum willingness to pay to achieve the
change, under the assumption that the respondents
would not be willing to pay more than the benefit
that they would receive. Statistical models are used
to estimate mean WTP for an individual from the
survey sample. These estimates can then be aggregated over the relevant population to determine an
estimate of benefits generated by a nonmarket
amenity. The interested reader is referred to
Mitchell and Carson (1989) for a discussion of the
theoretical underpinnings of CVM and practical
suggestions on how to perform CVM. Freeman
(1993) provides a good review ofthe economic theory regarding the use of WTP to measure individual and aggregate benefits of nonmarket amenities.
As of 1993, no less than 1500 studies and papers
using the CVM were documented by Carson et al.
(1993), many of them applied to recreational use
value. Since then, much of the theoretical and
methodological advances have focused on applying CVM to measure nonuse values. One of the
most well-known examinations of the CVM was
that carried out by the National Oceanic and Atmospheric Administration (NOAA). At the time of
the Exxon Valdez spill at Prince William Sound,
Alaska, in March 1989, the U.S. government had
considered CVM and valuation estimates from
CVM as acceptable to determine compensation levels in liability cases.
The Exxon Corporation, which was responsible
for the Exxon Valdez oil spill, hired two consulting
economists to expose any reasons why CVM
should not be used to estimate lost passive use values. 5 The CVM was, and still is, considered the
only method acceptable for measurement of nonuse
values. The stakes were high, since estimates of lost
nonuse values would greatly increase the amount
that the corporation could be held liable for. The
stakes involved with the state of Alaska's lawsuit,
consulting income, the attack on the academic literature supporting the theoretical basis of CVM,
and academic reputations, precipitated a heated debate that divided the discipline. At the time, it was
not uncommon to find Exxon representatives in attendance at academic conferences ready to criticize
any application of CVM. The potential for public
spectacle substantially boosted attendance at these
sessions as the lines between theory, practice, academics, and politics became blurred.
The U.S. government reacted by creating the
NOAA panel to review the critiques of CVM and
5They used the term passive use values to refer to existence values and nonuse values.
are difficult to measure because an extrapolation
must be made from the observation of particular
choices made by the consumer to general conclusions about preferences. This implies that the researcher must be able to hypothesize the functional
form and underlying theory that connect the observed market behavior with the unobservable and
still hypothetical welfare change that would result
from the policy scenario under consideration. As
well, direct methods allow the researcher to hypothesize assumptions about the respondents' preferences and then test these assumptions to verify a
respondent's motives. Revealed preference methods (also known as indirect methods, because they
indirectly estimate demand for a nonmarket good
based on some observed demand for a related market good) do not easily allow for these tests. Finally, arguments in favor of direct methods include
the fact that there is a real distinction between private market goods and public goods, which make
revealed preference methods harder to apply, especially in the case of public goods, which are often of national rather than local scope.
Contingent Valuation
The contingent valuation method (CVM) uses carefully worded surveys to gather primary data that is
analyzed by an appropriate econometric model to
predict what people would be willing to pay (or
willing to accept in compensation) for a specified
change in the price or the quality of an environmental amenity. The CVM has been shown to work
especially well for recreational use values for goods
with which respondents have had prior experience.
The term contingent refers to the fact that the valuation of the good is contingent on the hypothetical assumption of a plausible market and method
of payment for the good. A simple CVM study
might consist of a detailed description of a good,
followed by a question asking the respondent
whether he or she would be willing to pay a given
price for provision of the good. The yes or no responses to a range of prices can be used to establish a cumulative probability distribution of a representative individual's probability of answering
yes. The area beneath is the expected value of a yes
response and thus the average willingness to pay
(WTP) for the environmental change in question.
This example is based on a dichotomous choice
model CVM, since respondents are given the option to say yes or no to given prices. Contingent
valuation can be accomplished with other models
as well, including those that ask follow-up questions after the first response.
Measuring Ecosystem Values
The benefit to individuals is measured in terms
of their maximum willingness to pay to achieve the
change, under the assumption that the respondents
would not be willing to pay more than the benefit
that they would receive. Statistical models are used
to estimate mean WTP for an individual from the
survey sample. These estimates can then be aggregated over the relevant population to determine an
estimate of benefits generated by a nonmarket
amenity. The interested reader is referred to
Mitchell and Carson (1989) for a discussion of the
theoretical underpinnings of CVM and practical
suggestions on how to perform CVM. Freeman
(1993) provides a good review ofthe economic theory regarding the use of WTP to measure individual and aggregate benefits of nonmarket amenities.
As of 1993, no less than 1500 studies and papers
using the CVM were documented by Carson et al.
(1993), many of them applied to recreational use
value. Since then, much of the theoretical and
methodological advances have focused on applying CVM to measure nonuse values. One of the
most well-known examinations of the CVM was
that carried out by the National Oceanic and Atmospheric Administration (NOAA). At the time of
the Exxon Valdez spill at Prince William Sound,
Alaska, in March 1989, the U.S. government had
considered CVM and valuation estimates from
CVM as acceptable to determine compensation levels in liability cases.
The Exxon Corporation, which was responsible
for the Exxon Valdez oil spill, hired two consulting
economists to expose any reasons why CVM
should not be used to estimate lost passive use values. 5 The CVM was, and still is, considered the
only method acceptable for measurement of nonuse
values. The stakes were high, since estimates of lost
nonuse values would greatly increase the amount
that the corporation could be held liable for. The
stakes involved with the state of Alaska's lawsuit,
consulting income, the attack on the academic literature supporting the theoretical basis of CVM,
and academic reputations, precipitated a heated debate that divided the discipline. At the time, it was
not uncommon to find Exxon representatives in attendance at academic conferences ready to criticize
any application of CVM. The potential for public
spectacle substantially boosted attendance at these
sessions as the lines between theory, practice, academics, and politics became blurred.
The U.S. government reacted by creating the
NOAA panel to review the critiques of CVM and
5They used the term passive use values to refer to existence values and nonuse values.