16 valuation techniques partly because of its long

16 / Valuation Techniques
partly because of its long history in forestry economics, but mostly
because it mimics common empirical techniques used elsewhere in
economics. Analysts have tended to look favorably on the travel cost
approach to natural resource valuation because it is based on actual
behavior rather than verbal responses to hypothetical scenarios.
Individuals are actually observed spending money and time, and their
economic values are deduced from their behavior. In appropriate
circumstances, travel cost models can often be applied without enor-
mous expense.
The greatest disadvantage of travel cost and other indirect techniques is
that they cannot be used unless there is some easily observable behavior
that can be used to reveal values. In addition, travel cost models can be
technically and statistically complicated. Data must be employed to
statistically estimate increasingly sophisticated econometric models that
take into account sample selection problems and nonlinear consumer
surplus estimates. In addition, the resulting estimates sometimes have
been found to be rather sensitive to arbitrary choices of the functional
form of the estimating equation, the treatment of the value of an indi-
vidual's time, the existence of multiple stops during the travel period,
and the recognition of substitute sites. Finally, the travel cost approach
requires that the analyst be in a position to correlate environmental
changes with the behavior of visitors. Example 4 provides an illustra-
tion of the method, taking these caveats into account.
Random Utility Models
Random utility models are conceptually linked with the travel cost
models in that they seek the same sorts of values and use the same sort
of logic. However, random utility models provide a different structure
in which to model recreational demand, one which focuses attention on
choices among substitute sites for any given recreational trip instead of
the number of trips taken to a given site. These models are especially

Nonmarket Valuation: Indirect Techniques I 17
Example 4: Travel Cost Model
Scott et al. (1997) used the travel cost method to estimate willingness to pay for
upland bird hunting in Benton-Franklin Counties in eastern Washington state.
Valuation data were obtained from the Washington State Department of Fish and
Wildlife (Upland Game Division) and the "1991 Washington Survey of Fishing,
Hunting, and Wildlife Associated Recreation." The authors apportioned travel by
upland bird hunters to Benton-Franklin Counties into five zones based on state
averages: those that travelled less than 25 miles to their hunting site, between
25 and 50 miles, between 50 and 100 miles, between 100 and 250 miles, and
over 250 miles.
The authors estimated that the average cost per small game hunter in 1991 was
$193, and assumed that this cost varied in proportion to distance travelled to
Benton-Franklin Counties. Given the latter assumption, they estimated an
average cost per zone by multiplying the average cost of $193 by the ratio of the
median distance in each zone to the average distance travelled. Using this esti-
mation of travel costs, willingness to pay for hunting shrub-steppe dependent
game birds was estimated for the individual hunting zones and then aggregated
across zones to obtain a willingness-to-pay estimate of $3.2 million in annual
recreational benefits.