|
the sum of net willingness to pay and the
<br />amount actually spent on the good. Since
<br />the amount actually spent is part of the
<br />cost of participation, the benefits (i.e., the
<br />net willingness to pay) are that amount in
<br />excess of what people actually spent.
<br />To estimate the changes in consumer
<br />surplus resulting from changes in stream-
<br />flow in the single-site format, the follow-
<br />ing single-site pooled time-series cross-
<br />section travel cost model (equation [1])
<br />needs to be estimated. This model will be
<br />used to estimate the demand for trout fish-
<br />ing along the North Fork of the Feather
<br />River. Because individual observation data
<br />were not available, a zonal TCM model is
<br />used. The zonal form of the TCM utilizes
<br />counties of visitor residence as the "zones"
<br />of visitor origin. All visits from a given
<br />county are aggregated together as one ob-
<br />servation. Thus, there are as many obser-
<br />vations as there are counties visiting the
<br />site. This compares with the individual ob-
<br />servation TCM model in which the num-
<br />ber of observations equals the number of
<br />individuals visiting the site.
<br />Since many site quality variables, such
<br />as total fish catch, are available only on a
<br />seasonal or yearly basis, estimation of a
<br />coefficient on site quality must usually be
<br />performed using multi-site cross-sectional
<br />data; that is, observing how recreationists
<br />respond to differences in site quality across
<br />sites (Vaughan and Russell 1982). How-
<br />ever, the application of BCA to value
<br />changes in site quality often involves
<br />changes in quality at just one site. Per-
<br />forming this analysis requires knowledge
<br />of the visitors' response to changes in qual-
<br />ity at just that site. Since time-series data
<br />are rarely available, the possibility of es-
<br />timating the visitors' response to quality
<br />over time at the study site is eliminated.
<br />However, for this project, five years of data
<br />for individual sections of the North Fork
<br />of the Feather River were available. There-
<br />fore, it was possible to estimate a single-
<br />site demand equation incorporating a site
<br />quality variable.
<br />For the recreational site, the following
<br />simultaneous system is specified:
<br />TRIPS;,/POP;, = f(TRVCOST;,, INC,,,
<br />FISHCATCH„
<br />SUBS;) + u;,, (1)
<br />FISHCATCH, = f(FLOW„ TRIPS;,
<br />/POP;,) + v,,, (2)
<br />where:
<br />i = 1, ... , n are the number of visitor
<br />origins.
<br />t = 1, ... , T years.
<br />TRVCOST;, is the transportation and
<br />time cost of traveling from origin i to
<br />the specified site in year t.
<br />INC,, is average household income in
<br />origin i in year t.
<br />FISHCATCH, is a river quality vari-
<br />able at time t.
<br />SUBS; is the price of substitute fishing
<br />site available to origin i.
<br />u,, and v;, are random disturbance
<br />terms.
<br />FLOW, is a cubic feet per second of
<br />flow in year,.
<br />CASE STUDY
<br />The study river is the North Fork of the
<br />Feather River in Northern California, up-
<br />stream of the Oroville Dam. The visitation
<br />data were collected by the California De-
<br />partment of Fish and Game with funding
<br />provided by Pacific Gas and Electric Com-
<br />pany. The data were collected using a short
<br />on-site survey for the years 1981-1985. The
<br />survey recorded such things as county of
<br />angler origin, composition of fish catch,
<br />hours fished, and fishing equipment used.
<br />The raw data were compiled by the De-
<br />partment of Fish and Game in an aggregate
<br />form by county of origin (i.e., the individ-
<br />ual anglers were not asked to state their
<br />seasonal number of visits). As a result, the
<br />zonal TCM model must be used for this
<br />study.
<br />The anglers' creel, the number of fish
<br />kept by the angler, is incorporated into the
<br />model as the fishing quality variable. The
<br />level of creel is available for each of the
<br />six separate sections of the river for each
<br />of the five years of the study. Therefore,
<br />river section specific pooled time-series
<br />cross-section regressions, that include the
<br />creel variable for each of the river sections,
<br />can be estimated. Unlike the purely cross-
<br />sectional case, where a quality coefficient
<br />can usually only be estimated with multi-
<br />site data, the quality coefficients can be es-
<br />timated separately for each river section.
<br />J. Loomis and J. Cooper
<br />25 Ir-;_
|