238
experiment is to make a good generic, in a sense,
by defining it as being composed of any combination of attributes that the consumer most prefers.
On the other hand, one feature of a unique good is
that it cannot exist in any other form. The value of
the good lies in preserving it as it is. For this reason, the use of choice experiments for unique environmental goods or for existence values may be
inconsistent with how a person views a unique
good.
For environmental goods and services that are
not unique, such as many forms of recreational fishing, boating, hiking, camping, and other experiences, choice experiments may prove to be exceptionally valuable since they allow for large numbers
of combinations of site attributes in one experimental design. Because they are designed to be
somewhat generic, the results of choice experiments may be well suited to apply to many sites.
16.5.2 Revealed Preference, or
Indirect, Methods
Indirect methods rely on observing how people behave in markets that are indirectly related to the
ecosystem good or service that is to be valued.
Some difficulties with indirect methods are that
they imply that it is possible to model precisely how
the behavior in the observed markets is related to
the specific policy impacts on the ecosystem goods
and services. Such modeling sometimes requires
heroic assumptions about individual utility functions and other details that are generally unknowable with certainty. A large number of empirical
papers focus on the sensitivity of models used in
indirect methods to alternative specifications of
utility functions, the value of time, and other assumptions. It is generally a good idea to build into
any study using indirect methods analyses that test
the sensitivity of results to alternative specifications. Another limitation of indirect methods is that
it may not be possible to construct a one-to-one correspondence between the activity that is valued and
the value of the economic changes induced by the
proposed policy. An advantage of many indirect
methods is that often the necessary data can be collected from information that is already available
from secondary sources. In contrast, direct methods require implementation of surveys to collect
primary data.
Travel Cost Models
There are a number of types of travel cost models,
ranging from basic zonal travel cost models to more
Measuring Ecosystem Values
sophisticated random utility travel cost models. The
basic concept for all travel costs models is that the
costs associated with travel time and distance to a
potential recreation site in part determine how
likely an individual is to visit that site. However,
there is a great deal of variety in the types of models. Travel cost models allow for inclusion of substitute sites and can be designed to model how a
person chooses from among mUltiple sites with different characteristics. Thus travel cost models are
suited for valuation of recreational use of ecosystems and have been widely used to value recreational experiences, including fishing, hunting, hiking, boating, and swimming (e.g., see Caulkins et
al., 1986; Englin and Mendelsohn, 1991; Boxall,
1995).
The majority of travel cost models are best suited
for estimating the value of adding a new site, rather
than for valuing marginal quality changes in existing sites. Well-designed studies do a reasonable job
at predicting visitation at sites, provided substitute
sites are properly accounted for in the model. Thus,
if the policy issue under consideration involves estimating the value of and predicted number of visits to a proposed new recreational site, a travel cost
model is likely to work well at a reasonable cost.
However, if the policy context is one that explicitly involves decisions that will cause changes in
the quality of existing sites, then travel cost models may not be as sensitive as stated preference
models or some combination of a stated preference
and travel cost model. Freeman (1993) provides a
detailed summary of the types of travel cost models and the advantages and disadvantages of each.
Hedonic Pricing
Hedonic pricing is often used to value environmental amenities (or disamenities such as pollution) in the context of residential home markets
(e.g., Rosen, 1974; Brown and Rosen, 1982;
Palmquist, 1984, 1991). Hedonic wage models are
used to value the contribution of environmental
amenities and disamenities to explain wage differentials (e.g., see Cropper and Arriga-Salinas, 1980;
Smith and Gilbert, 1984; Hoehn et aI., 1987). The
basic principle is that, along with various features
of a house, such as square feet of living space, number of rooms, and proximity to schools, environmental amenities also contribute to purchase consumer decisions. Thus a home in a desirable
location may carry a premium, which is related to
the value of natural amenities. The value of these
amenities can in theory be teased out of market data
on housing purchases if the data sets are rich
experiment is to make a good generic, in a sense,
by defining it as being composed of any combination of attributes that the consumer most prefers.
On the other hand, one feature of a unique good is
that it cannot exist in any other form. The value of
the good lies in preserving it as it is. For this reason, the use of choice experiments for unique environmental goods or for existence values may be
inconsistent with how a person views a unique
good.
For environmental goods and services that are
not unique, such as many forms of recreational fishing, boating, hiking, camping, and other experiences, choice experiments may prove to be exceptionally valuable since they allow for large numbers
of combinations of site attributes in one experimental design. Because they are designed to be
somewhat generic, the results of choice experiments may be well suited to apply to many sites.
16.5.2 Revealed Preference, or
Indirect, Methods
Indirect methods rely on observing how people behave in markets that are indirectly related to the
ecosystem good or service that is to be valued.
Some difficulties with indirect methods are that
they imply that it is possible to model precisely how
the behavior in the observed markets is related to
the specific policy impacts on the ecosystem goods
and services. Such modeling sometimes requires
heroic assumptions about individual utility functions and other details that are generally unknowable with certainty. A large number of empirical
papers focus on the sensitivity of models used in
indirect methods to alternative specifications of
utility functions, the value of time, and other assumptions. It is generally a good idea to build into
any study using indirect methods analyses that test
the sensitivity of results to alternative specifications. Another limitation of indirect methods is that
it may not be possible to construct a one-to-one correspondence between the activity that is valued and
the value of the economic changes induced by the
proposed policy. An advantage of many indirect
methods is that often the necessary data can be collected from information that is already available
from secondary sources. In contrast, direct methods require implementation of surveys to collect
primary data.
Travel Cost Models
There are a number of types of travel cost models,
ranging from basic zonal travel cost models to more
Measuring Ecosystem Values
sophisticated random utility travel cost models. The
basic concept for all travel costs models is that the
costs associated with travel time and distance to a
potential recreation site in part determine how
likely an individual is to visit that site. However,
there is a great deal of variety in the types of models. Travel cost models allow for inclusion of substitute sites and can be designed to model how a
person chooses from among mUltiple sites with different characteristics. Thus travel cost models are
suited for valuation of recreational use of ecosystems and have been widely used to value recreational experiences, including fishing, hunting, hiking, boating, and swimming (e.g., see Caulkins et
al., 1986; Englin and Mendelsohn, 1991; Boxall,
1995).
The majority of travel cost models are best suited
for estimating the value of adding a new site, rather
than for valuing marginal quality changes in existing sites. Well-designed studies do a reasonable job
at predicting visitation at sites, provided substitute
sites are properly accounted for in the model. Thus,
if the policy issue under consideration involves estimating the value of and predicted number of visits to a proposed new recreational site, a travel cost
model is likely to work well at a reasonable cost.
However, if the policy context is one that explicitly involves decisions that will cause changes in
the quality of existing sites, then travel cost models may not be as sensitive as stated preference
models or some combination of a stated preference
and travel cost model. Freeman (1993) provides a
detailed summary of the types of travel cost models and the advantages and disadvantages of each.
Hedonic Pricing
Hedonic pricing is often used to value environmental amenities (or disamenities such as pollution) in the context of residential home markets
(e.g., Rosen, 1974; Brown and Rosen, 1982;
Palmquist, 1984, 1991). Hedonic wage models are
used to value the contribution of environmental
amenities and disamenities to explain wage differentials (e.g., see Cropper and Arriga-Salinas, 1980;
Smith and Gilbert, 1984; Hoehn et aI., 1987). The
basic principle is that, along with various features
of a house, such as square feet of living space, number of rooms, and proximity to schools, environmental amenities also contribute to purchase consumer decisions. Thus a home in a desirable
location may carry a premium, which is related to
the value of natural amenities. The value of these
amenities can in theory be teased out of market data
on housing purchases if the data sets are rich