Walsh, Johnson, and McKean
Issues in Nonmarket Valuation 179
nonmarket benefit estimates from 1968-88.
The study represents an update and evaluation
of a previous review by Sorg and Loomis. Their
93 benefit estimates in studies completed from
1968-82 are supplemented with 20 they missed
plus 164 estimates in studies completed from
1983-88. The objective is to provide a range
of benefit estimates for major recreation ac-
tivities in Forest Regions for the 1990 resource
planning program (RPA) of the Forest Service,
U.S. Department of Agriculture. Congress re-
quires that the agency prepare long-run (50-
year) forest plans every five years. As part of
this process, the agency periodically reviews
demand studies applying the contingent val-
uation (CVM), travel cost (TCM), and related
methods to provide an empirical basis for re-
vision of unit-day values. For example, the
Dwyer, Kelly, and Bowes literature review
contributed, in part, to estimation of recrea-
tion values for the 1980 RPA. The exercise has
been controversial because the agency, lacking
a scientific basis for adjustment, has relied on
the concept of reasonable and proper levels for
the purpose intended.
Detailed descriptions and evaluations of the
design aspects of studies completed from 1968-
82 were prepared by Sorg and Loomis on be-
half of the Forest Service 1985 RPA. As might
be expected, many of the early studies were of
dubious quality from the standpoint of being
able to make benefit inferences. Only midway
in the review period did the federal govern-
ment (Water Resources Council 1973, 1979)
issue guidelines on statistical sampling, vehicle
travel cost, travel time cost, substitutes, and
other aspects of experimental design to be used
by new studies. The guidelines clearly were
minimal when judged by the standards of some
of the best studies. Even so, several of the stud-
ies did not meet them in important respects
and, therefore, were of almost no value in es-
tablishing comparable measures of the net
benefits of recreation activities. The consensus
judgment of a panel of evaluators was that
substantial adjustment should be made in the
reported values before presentation of the
summary statistics.
As a result, Sorg and Loomis increased the
reported TCM values by 300∕o for the omission
of travel time, both TCM and CVM values
were increased by 15o∕o for omission of out-of-
state users, and TCM values were decreased
15% for application of the individual obser-
vation approach. They argued that omission
of travel time from the TCM demand function
leads to a downward bias in estimated benefits.
The cost of time spent traveling equals the
difference between the net willingness to pay
for sightseeing benefits enroute and the op-
portunity cost of the time in an alternative
activity. Similarly, omission of out-of-state
users tends to understate the number of visits
to most resource-based sites at relatively higher
travel costs. If they travel further than in-state
users, the upper limit of travel cost in the de-
mand curve will be understated and benefits
will be biased downward. The individual ob-
servation approach uses trips per participant
as the dependent variable. While this is sta-
tistically more efficient than the zonal ap-
proach, it omits the effect of travel cost on the
probability of participating. The resulting de-
mand curve is often less elastic which results
in overstating recreation benefits of activities
when the probability of participation decreases
significantly at higher travel costs.
Table 1 illustrates the resulting summary
statistics for the recreation use categories of
the Forest Service. The 287 estimates of net
economic value per day reported by 120 out-
door recreation demand studies from 1968 to
1988 are adjusted for method as in Sorg and
Loomis and are in third-quarter 1987 dollars.
Mean value of the estimates is $34 per day
with a 95% confidence interval of $31 to $37
and a range of $4 to $220. The median is $27.
These values are shown for each activity along
with output of the agency. Average benefit of
activities ranges from $12 to $72 per day with
the highest values reported for hunting, fishing,
nonmotorized boating, hiking, and winter
sports. This approach assumes that the socio-
economic characteristics of users and the qual-
ity of study sites are sufficiently similar that a
common pattern of consumption applies to
each. Ideally, the distribution of values would
be approximately normal with a few outliers
at both the high and low ends. Given a suffi-
cient number of studies, the solid core of val-
ues in the middle would be the most reason-
able estimate.
A number of problems should be considered
before analysts could reasonably apply this in-
formation to policy decisions. First, for most
recreation activities, an insufficient number of
studies have been completed to obtain reason-
able estimates of value by this method. Even
where there are a large number of studies, the
frequency distribution is often skewed with the