Economic
Valuation of Fisheries: Nonmarket Studies in the Clark Fork Basin
John
Duffield
Department of Economics
University
of Montana
Missoula, MT 59812
ABSTRACT:
This paper provides an introduction to methods used by economists a value
nonmarket resources. Recent applications to valuing Montana trout stream
fisheries, including the major waters in the Clark Fork Basin are described.
One finding is that the present recreational value of the Montana stream
fisheries is quite large on the order of 3.0 billion dollars. This estimate is
conservative since includes nonangling recreational use and potentially
significant existence (or intrinsic) values. The valuation across streams
varies considerably reflecting differences in both water quality and quantity.
The highest value/mile (annual basis) in the state IS on the Madison River at
$184.000/mue. The Upper Clark Fork value per mile ($7.400) is the second
lowest for the group of 20 major waters compared. Angler characteristics (such
as average distance traveled. fishing technique angler preferences. and mean
trip length) also vary considerably across site and help explain differences
in value per trip. Consistency, reliability and precision of the results are
discussed.
INTRODUCTION
Although fishery and wildlife resources are generally not traded in
established markets, there are often situations where it would be useful to
know the dollar value of these resources. For example, in 1981 an electric
cooperative proposed to build a 144-mega watt hydroelectric facility at on the
Kootenai River below Libby. The project would have substantially altered
Kootenai Falls and a popular fishery, the China Rapids. In deciding whether
society is better off with or without the dam, it is useful to know if the
dollar value of the electricity exceeds the project costs, including the cost
of the foregone recreation (Duffield. 1984). Similar situations arise when
public agencies consider investing in facilities to improve fisheries; then
the question is whether the investment in a hatchery or boat access is
justified by the recreational benefits.
Motivated initially by the need to
evaluate public investment decisions, economists have developed a variety of
methods to measure the value of nonmarket resources. The latter include not
only fish and wildlife, but other public goods such as clean air, visibility
and the value of health. These procedures have been defined and endorsed by
the U.S. Water Resources Council (1983) and are now being applied to a variety
of state and federal policy decisions. A new and potentially important
application of these methods is in evaluating natural resource damage due to
toxic wastes under the Superfund legislation. For example, a group of
economists are currently working on estimating the total damage to the Alaskan
marine and coastal environment resulting from the Exxon- Valdez oil spill.
Relatedly, these methods can be used to guide resource management decisions.
An interesting current project is examining the effect of downstream releases
from Glen Canyon Dam on boaters in the Grand Canyon of the Colorado. The
tradeoff is the value of electricity to meet peak loads versus the impacts on
recreation and the environment.
In 1985, the Montana Department of Fish, Wildlife and Parks
(DFWP) initialed the Montana Bioeconomics Project, a series of nonmarket
valuation studies aimed at evaluating Montana fish and wildlife resources. The
motivation for this paper is to provide scientists in other disciplines with
an overview of economic methods and results concerning Montana's coldwater
stream fisheries.
The general focus in this symposium is, of course, on the
environmental degradation associated with more than 125 years of copper and
silver mining and smelting activities in the Clark Fork Basin. Botanists,
entomologists and fishery biologists have long examined the potential impacts
on biota of water quality degradation in the basin. However, one way to think
of the social science work described here is also as a type of "biotic
response" -but for an organism further up the food chain. Anglers are
attracted by fish and “good fishing" (though there are of course a
large variety of nonpredative motivations associated with fishing. perhaps
especially in the days of catch and release regulations). We are in fact far
from having an integrated model of the basin that begins with sediment
transport and hydrology and ends with the social response. The work described
below is only casually connected to the biological state of a given river.
However, there does seem to be evidence that the depressed fishery in the
Upper Clark Fork is mirrored by depressed (or mainly absent) anglers.
The emphasis here is on a comparison across the twenty
major trout streams in the state. This comparison provides some insights into
the relative quality of the major fisheries in the Clark Fork Basin: the
mainstem Clark Fork, Rock Creek, the Bitterroot and the Blackfoot. Recent work
concerning the validity, precision and reliability of the estimated values is
also summarized. In addition, simpler and more robust measures of potential
values such as angler use per river mile are also compared across the major
streams.
OVERVIEW
OF METHODS
The most obvious economic dimension of recreational activity is the associated
expenditure on travel, lodging, meals, and equipment. Expenditures are of
interest to economists because of the regional and local economic activity
(measured by employment and income) that expenditures generate. However,
expenditures are ~ inappropriate way to value a given recreational resource,
such as a river-based fishery. The value of these resources is instead
correctly measured by the net benefits they provide to society. From the
standpoint of the river recreationist, travel and related expenditures
indicate the cost of a given activity, not the benefit. Benefits are measured
by individual and aggregate willingness to pay (over and above costs) for the
use of the resource. Net benefits are also termed "efficiency"
measures in that they are used in benefit-cost type analysis to identify the
most efficient use of a given resource (i.e., the use that yields the greatest
net benefits to society).
Where markets exist for recreational resources, the
observed relationship of prices and quantity of use (the economic demand
function) can be used to measure net benefits. For example, in many European
countries, fishing rights are owned by the adjacent landowner. In Norway
prices on the best Atlantic salmon waters are on the order of $200 per day.
There are a few examples of fee fisheries in Montana, including Nelson's
Spring Creek near Livingston where the charge is currently $40 per day for the
summer season and $20 in the winter. The fee on Red Rock River south of Dillon
is now $45 per day. However, in an area where free public access characterizes
most fishing opportunities, fees on restricted access sites may only indicate
the value of the quality differential between the fee fishery and open access
sites (Stoll, 1988).
In the absence of extensive markets for Montana's
recreational fisheries, nonmarket valuation methods are used to measure net
willingness to pay. The two most widely used approaches are the travel cost
method and contingent valuation. The travel cost approach is based on
observations of how participation varies with the cost of travel to a site.
For example, per capita visits to the Madison River are much higher from
nearby communities such as Ennis or Bozeman than from Billings or Salt Lake.
The key assumption of the model is that individuals would react to a fee for
use at the site in the same way that they respond to the cost of travel. In
this way a demand curve relating price (entrance fee) and quantity demanded
(total trips) is derived. There are a variety of methodological issues related
to this approach including choice of functional form and the value of travel
time. For a discussion of these issues see Dwyer, Kelly and Bowes (1977) and
Duffield, Loomis and Brooks (1988).
The contingent valuation approach is
more straightforward in that participants are surveyed and asked directly what
they are willing to pay for use of the resource. Generally a hypothetical
situation that involves a payment is described to the respondent and her
valuation response is given contingent on accepting the situation -hence the
term, contingent valuation. Hypothetical payment vehicles in previous studies
have included entrance fees, increases in monthly electrical bill, changes in
taxes, increases in travel costs, and contributions to trust funds.
Methodological issues associated with contingent valuation include
choice of question format and welfare measure. Cummings. Brookshire and
Schultze (1986 and Mitchell and Carson (1989) provide recent surveys of this
literature. 6)
There has been
an increasing use of these methods to value outdoor recreation in the United
States. In a 1978 1iterature review, Dwyer identified 15 such studies while
Walsh, Johnson and McKean (1988) in a recent review discuss a total of 120
studies completed in 1968-1988. The majority of these studies (82 percent)
have been concerned with hunting and fishing uses of wildlife, while
comparatively little attention has been given to so called nonconsumptive uses
such as wildlife viewing, photography or nature study.
The discussion to this point has been in terms of nonmarket
resource values related to direct recreational use. As first proposed by
Krutilla (1967), there are additionally indirect or intrinsic values
associated with the preservation or existence of a given resource, such as a
unique natural environment or species. These values, generally termed
existence values, are independent of personal use and may involve motives such
as altruism, concern for other species or a desire to protect resources for
future generations. Existence values are evidenced by donations to
organizations such as the Nature Conservancy or the World Wildlife Fund.
Randall and Stoll (1983) have described a total valuation framework for
estimating both direct use and existence values for a given resources. A
recent review (Butkay and Duffield, 1990) identified only five such studies of
wildlife resources including estimates for bald eagles, whooping cranes, and
bighorn sheep.
APPLICATION
TO MONT ANA STREAM FISHERIES
This section provides an overview of recent application of both travel cost
and contingent valuation to Montana trout stream fisheries. These estimates
are limited to direct angling use and are therefore conservative for excluding
existence values and the value of other direct recreational uses such as
boating or general shoreline activity. The intent here is to provide insight
into methods and basic findings. Detailed discussion of the theoretical
motivation, data, methods and results are available in the referenced studies.
CONTINGENT
VALUATION
A contingent valuation study of seventeen major Montana trout streams was
initiated in 1986 (Duffield and Allen, 1988). A total of 2672 questionnaires
were mailed to resident and nonresident license holders who where known to
have fished 111 given water in the previous year. A total of 2171 completed
questionnaires were received for a response rate of 81 percent; this is a very
high response for a mail survey.
For the contingent valuation
question, anglers were asked a question in the so- called dichotomous choice
format. In this format the response is “yes” or “no” as to willingness
to pay a specific offer amount. This general format was first utilized by
Bishop and Heberlein (1979). In the current application, anglers were asked to
identify total trip expenditures for their most recent fishing trip, and then
were asked "would you still have made the trip if your share of the
expenses had been (dollar amount) more?” (yes or no). The dollar amount was
varied randomly from $1 to $500 across respondents. It may be noted that an
alternative question format is to ask an open-ended "what is the largest
amount you would be willing to pay before you would discontinue use... This is
a much harder question to answer and generally results in lower respondent
participation rates.
For purposes of illustration, the response for the Missouri
River (Holler to cascade section) is tabulated in Table I for the sub sample
of 156 respondents whose most recent trip was to this river. As one would
expect, respondents were much more likely to be willing to pay low offer
amounts than high amounts. For example, 24 of 26 respondents asked to pay $1
to $5 said yes, while only 1 of 27 was willing to pay an amount of $150 to
$500. Respondents also provided information on their socio-economic
characteristics and trip characteristics. The relationship of the probability
of paying to the offer amount and selected variables can be estimated using a
logistic functional form. Maximum likelihood procedures were used in the SPSSX
statistical package. The estimate for the Missouri is:
In (P/(I-P) = 2.55 -1.63 LDOLAMT + .323 LRGCOT (1)
(8.75)
(-5.96)
(3.77)
+ .902 LINCOME (2.24)
where:
p = probability of a yes response
LDOLAMT
= log of dollar bid amount
LRGCOT
= log of number of large trout caught this trip
LINCOME
= log of reported household income
(t-statistic
in parenthesis)
The estimated coefficients are highly
significant and the signs are consistent with a priori expectations. The more
successful the trip and the higher the income of the participant, the more
likely the respondent will be willing to pay a given amount, The estimated
relationship of bid amount and the probability of a yes for the Missouri River
sample is plotted in Figure
1, One measure of the average net willingness to
pay is given by the area under the curve in Figure 1 up to the maximum and
asked ($500), This truncated mean for the example is estimated at $63 dollars
per trip, A variety of other measures can also be computed from the
relationship in Equation 1 such as the median of the distribution of
willingness to pay values. There is an ongoing debate among economists as to
which of these measures is most appropriate.
For comparison
purposes, the response for 146 Madison River anglers is also shown in Table
1.
At a given bid amount, the proportion of Madison River anglers wining to pay
is much higher; for example, only 19 percent of Missouri River anglers faced
with offers in the range of $35 to $50 said yes, while 68 percent of Madison
River anglers would pay this amount. This relatively higher probability is
shown in Figure 2. The resulting estimated truncated mean value for the
Madison is therefore also much higher at $228 compared to $63 for the
Missouri. Estimated values per trip for 17 rivers using the contingent
valuation method are shown is Table
2.
TRAVEL
COST MODEL
A travel cost model application to 49 specific rivers or tributaries and 28
lakes in Montana was begun in 1985 (Duffield, Loomis and Brooks). Data was
collected through two separate surveys administered by Montana Department of
Fish, Wildlife and Parks (DFWP). The DFWP annual angler pressure survey (where
1500 to 3000 license holders are sampled monthly) was utilized to obtain basic
origin- destination information. A total of 36.000 surveys were mailed in 1985
with a response of 54 percent or 19,271 surveys. A supplemental phone survey
of 2000 licensed anglers was conducted in September and October of 1985 with a
response rate of 75 percent. The latter survey provided detailed socioeconomic
and trip expenditure data. Only the stream results are discussed here.
The basic relationship in the travel cost model is between
per capita participation from a given zone and the associated travel cost to a
given site. Origin zones at the county, county group, state and regional level
were defined at increasing distances from the river sites. A total of 836
origin-destination pairs were identified for the stream model; a scatter plot
of trips per capita versus distance for this sample is shown in Figure
3. The
basic relationship is for decreasing visitation with higher distance. The
statistical relationship of visitation and other explanatory variable was
estimated using an ordinary least squares estimate of a linear regression
model:
In(TRIPSij/POPij) = -1.615 -1.798 In(RTDISTij) (2)
(t-statistic)
(-2.96) (-50.13)
+.389 In(SUMTRTj) -4.43 In(A VYRSFi)
(7.25) (-4.32)
where:
lRIPSij
= stream fishing trips from origin i to site j
POPi
= origin i's population
RmISTij
= round trip road distance from i to j
SUMTRTj = sum or trout catch at j
AVYRSFi = average years fished by anglers in origin i
Equation
(2) is based on a total of 727 observations with complete information. The
model provides a good fit to the data and has high explanatory power (adjusted
r-square is. 782). All reported parameters are highly significant. The sign on
distance and trout catch is as hypothesized; the sign on the fishing
experience variable does not have an obvious interpretation.
A limitation of this particular model is the absence of a
variable measuring the price and availability of substitutes.
Net benefit estimates at the site level were estimated by
integrating equation 2 for every origin-destination up to the maximum observed
distance for the given site. (It can be shown that this approach is
analytically equivalent to the usual approach of estimating the relationship
of incremental travel cost and total visitation. ) The implicit intercept on
the quantity axis at a zero site price was set at the observed trip level
following Gum and Martin (1975). Distance was convened to travel cost based on
estimated variable costs of 22.4 cents per mile derived from angler reported
expenditures and trip distances. Additionally travel time was valued at
one-fifth the reported wage rate or 4.6 cents per mile. The latter is based on
a contingent valuation estimate of Montana angler willingness to pay to
shorten travel time and is somewhat conservative compared to the U.S. Water
Resources Council (1983) standard recommended procedure of one-third the wage
rate.
Travel cost model estimates of value per trip for 20 major
Montana trout streams are provided in Table 2.
Other
Fishery Economics Results
In addition to estimating baseline values by stream, a number of other aspects
of fishery economics have been examined for Montana fisheries. Duffield and
Allen 1988) undertook a market segmentation analysis to identify angler types.
Cluster, analysis based on angler reported reasons for fishing yielded four
distinct angler experiences or subgroups. These were consistent with the
theory of angler specialization described by Bryan (1979) and included
occasional users, generalists and a specialist category. Estimated trip values
for the subgroups based on contingent evaluation varied markedly from $7.56
per trip for the occasional user to $91.03 and 117.07 for two generalist
groups and $170.28 for specialists. These findings provide further insight
into why average values vary across rivers; the more highly valued rivers are
attracting a greater share of the specialized anglers.
There are also interesting differences in values across
activities and trip qualities. For example, in a detailed contingent valuation
study of Rock Creek anglers (Duffield. 1989) it was found that the average
float angler trip was worth about 50 percent more than the average trip for
fishing from the bank on this stream. Similarly, trips with chances of
catching more trout or more large trout were more highly valued. The value of
increased catch or higher success can also be inferred from the travel cost
model reported above. Loomis (1989) applied this model to estimate the cost of
lowered fishing quality on the Gallatin and Upper Yellowstone if a proposed
wilderness area in the Gallatin National Forest was logged. Loomis estimated
the foregone recreation benefits associated with reduced trout catch. The
latter in turn was due to sedimentation that lowered sustainable fish
populations.
VALIDATION,
RELIABILITY AND PRECISION OF ESTIMATES
VALIDATION
OF ESTIMATES
One indication of the validity of the estimated values presented in Table 2 is
to determine the consistency across methods. The mean value for the 17 streams
based he CVM approach is $126.69 while the mean for the TCM approach for the
same set of streams is quite similar at $121.69. It should be noted that in
the original report, a variety of specific CVM and TCM models and welfare
measures were mined. The values reported are for two specific measures that
appeared to be superior (the travel cost model using reported angler costs and
the observed trip intercept and the CVM model using the logistic mean
truncated at the maximum bid level of $500).
The ranking of rivers by TCM value is provided in Table
4.
Generally the listing is consistent with a notion of which are the
"best" fishing streams by rcputation. The Madison, Upper
Yellowstone, Rock Creek, Big Hole, Gallatin and Bighorn are in the top ten,
while the lower valued streams include the Clark Fork, Swan, Flathead and
Kootenai. The TCM values are closely related to the average round trip
distance anglers actually drive to fish the river. It makes sense that the
better fisheries will have a larger spatial market. For example, the average
trip to the Madison is 1146 miles, while the average trip to the Upper Clark
Fork is 227 miles round trip. This indicates that the Upper Clark Fork is,
comparatively, a fishery of only local importance. The Madison is a river of
regional importance or a "destination" fishery.
Generally
speaking, the higher valued streams have a greater share of fly fishermen,
have more anglers saying it is their favorite stream and draw anglers spending
more lime per trip al the given stream (Table
4).
RELIABILITY
OF ESTIMATES
The reliability of the methods is indicated by examining whether similar results are obtained in repealed applications. The CVM current trip question again asked of anglers on one of the sites, Rock Creek, in a 1988 survey (Duffield, 1989). The simple bivariate logit models estimated on the 1986 and 1988 data were very similar, with a predicted probability of a yes response al $250, $500 and $2000 bid levels being .170, .107 and .039 in the 1986 model and .164, .090 and .024 in the 1988 model. The mean values when the two models were truncated al the $500 bid level were very similar al $113 for 1986 and $118 for 1988. Corrected for inflation, the estimates would be even more similar.