FISHERY BULLETIN: VOL. 87. NO. 1 



Fishery Service (U.S. Department of Commerce 

 1981). Since in this study we wish to investigate the 

 value of success by fish species, we chose samples 

 of fishermen who preferred one of the three species 

 of fish (bluefish, summer flounder, or weakfish). 

 These species are of considerable importance to 

 managers developing plans for fisheries along the 

 Atlantic coast and are fairly similar with respect to 

 mode, sites, and season. Since the number of obser- 

 vations for a given fishing site is generally quite low, 

 and reduced further by focusing on specific fish 

 species, pooling individual observations over sites 

 was necessary in order to have enough interviews 

 for statistical validity. Our sample sizes of anglers 

 for bluefish, summer flounder, and weakfish are 270, 

 161, and 57 respectively and comprise sites from the 

 Florida east coast to New York State. These data 

 are pooled within a covariance statistical framework 

 (i.e., with intercept and slope dummy variables) thus 

 allowing the testing for differences across target 

 species. 



Although the survey contains a large and useful 

 set of economic information on marine recreational 

 fishing, the data provided are by no means ideal for 

 an application of the travel cost method. Certain 

 enhancements to the travel cost method could not 

 be performed due to lack of data.^ In addition, 

 adjustments to travel distance and income were 

 needed given the nature of the survey instrument.'' 



The actual survey questions providing the data base 

 can be found in Table 1. 



A final point about the data base concerns the fish- 

 ing success measure. Since trip frequency repre- 

 sents activity over the past year, ideally one would 

 like a measure of fishing success to be reflective of 

 the last year and thus reflect ex ante or expected 

 fishing success. Unfortunately, the survey provides 

 no longitudinal information on individual anglers. 

 The measure of success is only for the day of the 

 interview and may not have been typical and, there- 

 fore, inconsistent with the fisherman's past be- 

 havior.'' We are forced to assume that ex post fishing 

 success is a proxy variable for ex ante (expected) suc- 

 cess. Travel frequency, distance, and fishing success 

 thus reflect long-run equilibrium adjustment by the 

 fishermen.* The empirical significance of fishing suc- 

 cess reflects on both the importance of success to 

 fishermen and the closeness of success realizations 

 versus expectations. 



EMPIRICAL MODEL 



Trip demand for the ith fisherman is specified as 

 a long-linear equation of either of the following 

 forms: 



In Qi = bo -t- bi In P; -i- bg In Sj 

 + bg In Ij -(- bZ -I- ej 



(3) 



'Two refinements that are noteworthy, but could not be incor- 

 porated into the analysis due to the lack of data, include time costs 

 and multiple site substitutions. It has been argued that time spent 

 travelling as well as time spent at the recreational site reflects op- 

 portunity costs and should be included as part of the price of the 

 fishing trip (Wilman 1980). The survey provides no information 

 on travel nor visitation time. 



Multiple fishing sites can provide an opportunity to construct 

 prices for recreational substitutes and, thus, include these vari- 

 ables in the statistical estimation of the demand curve. See Samples 

 and Bishop (1985) and Vaughn and Russell (1982). Unfortunately, 

 no information on the angler's point of origin (e.g., ZIP code or 

 area code) was available on the tabulated survey available to us 

 so as to construct accurate distance (and price) measures for sub- 

 stitute sites. 



'Since travel distance is a proxy for travel costs associated vrith 

 fishing, travel distance from a permanent home to the fishing site 

 might overstate travel costs for those individuals who were part- 

 year residents of the area, vacationers, or on business. For part- 

 time residents and those on business, the distance from last night's 

 accommodation rather than home was used as the appropriate 

 measure of travel distance. For vacationers, who comprised around 

 one-sixth of the sample, one-half of the distance from home was 

 used as their fishing travel cost. 



Adjustments for the income variable included 1) assigning 

 midrange values since respondents were asked for their income 

 category rather as an actual dollar amount and 2) dealing with miss- 

 ing data since the income question appeared on a follow-up tele- 

 phone survey for which the response rate was approximately half 

 that of the field survey. Missing observations were handled by the 

 zero-order approach whereby means replace missing values (Mad- 

 dala 1977). Since income is an exogenous control variable and not 

 central to the valuation calculations, these procedures were felt 

 to be acceptable. 



In Pj = ao -t- aj In Qj -i- a.^ In S; 

 -I- a3 In Ij -(- aZ -I- V; 



(4) 



where P, Q, S, I > 0; and 



Qi = the number of site-specific fishing trips 

 (including the day of the survey) made 

 in the last 12 months (Table 1, question 

 16), 

 Pj = round-trip cost in dollars to the site 

 from either home or last night's accom- 

 modation (Table 1, question 18, as mod- 



^An attempt was made to improve the success measure by focus- 

 ing only on fishermen for whom the fishing success on the day of 

 survey could be considered normal. This was done by utilizing a 

 satisfaction level variable (Table 1, question 23) and eliminating 

 those observations whose satisfaction was very high or very low. 

 By eliminating those individuals with extreme satisfaction, it was 

 felt that those individuals for whom the day's catch was not nor- 

 mal (or what was expected), would be eliminated from the sam- 

 ple, (jnfortunately, the filter did not distinguish perfectly, and, in 

 addition, reduced the sample to unacceptably low levels in part 

 because satisfaction is measured on the follow-up telephone survey 

 which had a lower response rate. The statistical results using this 

 filter were less significant and, thus, the approach was abandoned. 



"The implication of these potential errors in measurement is that 

 the coefficient of success will be underestimated to a degree de- 

 pending on the ratio of the variance of the error in measuring true 

 success over the variance of observed success. 



226 



