16.5 Techniques to Measure Nonmarket Economic Values
237
to make recommendations about whether estimates
based on the method were valid and reliable enough
to be used for applied work and damage assessments in liability claims. Kenneth Arrow and
Robert Solow, two Nobel laureates in microeconomic and macroeconomic theory respectively,
headed the panel. Since neither had been involved
in CVM studies previously, they were considered
objective chairs. After a lengthy review, the panel
concluded the "contingent valuation method can
produce results that appear to be consistent with assumptions of rational choice" (Arrow et al., 1993,
p. 4604). They asserted that "contingent valuation
studies convey useful information (that) can produce estimates reliable enough to be the starting
point of a judicial process of damage assessment
including lost passive use values" (Arrow et al.,
1993, p. 4610). Thus the federal government continues to accept CVM today for assessing lost environmental values in damage liability cases in the
United States.
While the hearings focused on the theoretical basis of CVM, it was also clear that methodological
issues made a significant impact on the reliability
of CVM. The panel recommended that, although
estimates of existence values derived through CVM
studies can, in theory, be reliable enough, the
method must be conscientiously applied to meet
rigorous standards. Otherwise, the numbers CVM
produces could be essentially meaningless. The
panel noted that, since no other generally accepted
means of determining existence values exists,
CVM takes on an "added importance in the light
of the impossibility of validating externally the results of CV studies" (Arrow et aI., 1993, p. 4603).
For this reason, a system of internal checks on the
findings is especially important to build into any
study. The aftermath of the NOAA hearings resulted in a set of recommendations for future CVM
studies. And while current research appears to indicate that some of these recommendations may be
open to debate as methods are refined, it has become standard for CVM studies performed since
then to explicitly address these recommendations.
Choice Experiments
Choice experiments are a relatively new valuation
method, but are based on the same theoretical economic and behavioral principles regarding choice
as contingent valuation. In terms of the modeling
approaches used for each method, contingent valuation can be regarded as a simple version of a
choice experiment. Since both methods are based
on random utility models, some researchers have
successfully combined contingent valuation and
choice experiment response data to enhance the interpretive value of statistical findings (Boxall et al.,
1996).
Choice experiments were developed from conjoint analysis, which is used in marketing to determine the attributes of a product likely to result in
the largest market share for the product. Conjoint
analysis is based on surveys that required potential
consumers to rank alternative options in terms of
combinations of features in a good or service. The
difference between conjoint and choice experiments is that conjoint analyses ask a respondent to
reveal their preferences by ranking alternatives. A
choice experiment presents the respondent with
several paired groups of selected profiles of a good.
Each profile is composed of several attributes. For
each pair, the respondent chooses which of the two
options is preferred. Since the respondent is given
several sets of paired profiles, it is possible to statistically generate the worth of each attribute to the
individual. From these "part-worths," it is possible
to calculate the value of any potential combination
of attributes. Thus the value of any combination of
attributes is defined as the sum of the partial values of each of the component parts. This is a rather
simplistic description of a choice experiment.
This area of current valuation research potentially holds much promise. One particularly interesting use of choice experiments is in the development of joint models using bQth choice experiments
and revealed preference models, such as in
Adamowicz et al. (1997), and using contingent valuation, such as in Boxall et al. (1996). In both of
these joint approaches, the use of choice experiments appears to enhance the value of the information from the method with which it is paired.
Contingent valuation focuses on the willingness
to pay (or accept compensation for) a specific quality or quantity change for one well-specified good,
whereas a choice experiment defines the good in
terms of a number of characteristics and then focuses on the trade-offs between combinations of
characteristics. Choice experiments have most
commonly been used in marketing studies to determine what combination of characteristics of a
new product would receive the greatest share of the
market for the product. The technique is most well
suited for goods that are very well defined and very
familiar to consumers. For example, a choice experiment is particularly well suited to determine
what bundle of options for long-distance telephone
services are best for specific markets or to determine what bundles of cable TV channels should be
targeted to specific markets. The nature of a choice
237
to make recommendations about whether estimates
based on the method were valid and reliable enough
to be used for applied work and damage assessments in liability claims. Kenneth Arrow and
Robert Solow, two Nobel laureates in microeconomic and macroeconomic theory respectively,
headed the panel. Since neither had been involved
in CVM studies previously, they were considered
objective chairs. After a lengthy review, the panel
concluded the "contingent valuation method can
produce results that appear to be consistent with assumptions of rational choice" (Arrow et al., 1993,
p. 4604). They asserted that "contingent valuation
studies convey useful information (that) can produce estimates reliable enough to be the starting
point of a judicial process of damage assessment
including lost passive use values" (Arrow et al.,
1993, p. 4610). Thus the federal government continues to accept CVM today for assessing lost environmental values in damage liability cases in the
United States.
While the hearings focused on the theoretical basis of CVM, it was also clear that methodological
issues made a significant impact on the reliability
of CVM. The panel recommended that, although
estimates of existence values derived through CVM
studies can, in theory, be reliable enough, the
method must be conscientiously applied to meet
rigorous standards. Otherwise, the numbers CVM
produces could be essentially meaningless. The
panel noted that, since no other generally accepted
means of determining existence values exists,
CVM takes on an "added importance in the light
of the impossibility of validating externally the results of CV studies" (Arrow et aI., 1993, p. 4603).
For this reason, a system of internal checks on the
findings is especially important to build into any
study. The aftermath of the NOAA hearings resulted in a set of recommendations for future CVM
studies. And while current research appears to indicate that some of these recommendations may be
open to debate as methods are refined, it has become standard for CVM studies performed since
then to explicitly address these recommendations.
Choice Experiments
Choice experiments are a relatively new valuation
method, but are based on the same theoretical economic and behavioral principles regarding choice
as contingent valuation. In terms of the modeling
approaches used for each method, contingent valuation can be regarded as a simple version of a
choice experiment. Since both methods are based
on random utility models, some researchers have
successfully combined contingent valuation and
choice experiment response data to enhance the interpretive value of statistical findings (Boxall et al.,
1996).
Choice experiments were developed from conjoint analysis, which is used in marketing to determine the attributes of a product likely to result in
the largest market share for the product. Conjoint
analysis is based on surveys that required potential
consumers to rank alternative options in terms of
combinations of features in a good or service. The
difference between conjoint and choice experiments is that conjoint analyses ask a respondent to
reveal their preferences by ranking alternatives. A
choice experiment presents the respondent with
several paired groups of selected profiles of a good.
Each profile is composed of several attributes. For
each pair, the respondent chooses which of the two
options is preferred. Since the respondent is given
several sets of paired profiles, it is possible to statistically generate the worth of each attribute to the
individual. From these "part-worths," it is possible
to calculate the value of any potential combination
of attributes. Thus the value of any combination of
attributes is defined as the sum of the partial values of each of the component parts. This is a rather
simplistic description of a choice experiment.
This area of current valuation research potentially holds much promise. One particularly interesting use of choice experiments is in the development of joint models using bQth choice experiments
and revealed preference models, such as in
Adamowicz et al. (1997), and using contingent valuation, such as in Boxall et al. (1996). In both of
these joint approaches, the use of choice experiments appears to enhance the value of the information from the method with which it is paired.
Contingent valuation focuses on the willingness
to pay (or accept compensation for) a specific quality or quantity change for one well-specified good,
whereas a choice experiment defines the good in
terms of a number of characteristics and then focuses on the trade-offs between combinations of
characteristics. Choice experiments have most
commonly been used in marketing studies to determine what combination of characteristics of a
new product would receive the greatest share of the
market for the product. The technique is most well
suited for goods that are very well defined and very
familiar to consumers. For example, a choice experiment is particularly well suited to determine
what bundle of options for long-distance telephone
services are best for specific markets or to determine what bundles of cable TV channels should be
targeted to specific markets. The nature of a choice