using a single year's observations in both 

 fisheries on the same variables. The results 

 were very encouraging in that there was a high 

 degree of stability in the coefficients and their 

 t ratios. These stable results were obtained in 

 fisheries which, if anything, are notorious for 

 their variability in almost all aspects: biological, 

 economic, atmospheric, and oceanographic. Some 

 results illustrating this stability for the trawl 

 fishery are shown in Appendix 1. 



The New England Trawl Fleet 



The statistical results for the New England 

 trawl fishery were very good. The overall "fit" 

 of the data in the equations was very high, 

 especially when one considers the heterogeneity 

 of this fleet. The equations are rich in informa- 

 tion in that many of the variables about which 

 hypotheses were made were statistically sig- 

 nificant with the right signs. 



Because of the unclear nature of variables 

 discussed, the equations were run using the 

 alternatives for the same variables where pos- 

 sible. This will allow direct comparison of the 

 results. In a sense, we shall permit the data 

 to decide which are better variables. We will 

 briefly run through the results according to the 

 topics covered in the theoretical section. 



The following general production function 

 was established for the New England trawl 

 fleet: 



(3) = /(FT, GRT, HP, CR, AGE, C, PT) 

 where O = output, either total pounds 



or total value, 

 FT = fishing time, either days 



fished or days absent, 

 GRT = gross registered tonnage, 

 HP = horsepower, 

 CR = crew size, 

 AGE = age of the vessel, 

 C = construction, 1 if wood, 



otherwise, 

 PT = homeport dummy variables. 



The equations providing the best results are 

 shown in Table 1. These equations will be 

 discussed below. A more complete set of regres- 

 sions is shown in Appendix Table 1. 



The tests of whether total value or total 

 pounds was the better measure of output in this 



fishery are shown in Problems 1 through 4. The 

 measures of overall fit (R-) are lower in Problems 

 1 and 2, which use total pounds as the dependent 

 variable (0.40 and 0.54), than in Problems 3 and 

 4, which use total value as the dependent 

 variable (0.83 and 0.83). Thus, the fishermen 

 appear to have implicitly taken into account 

 expected prices, expected catch rates, and 

 steaming time to the grounds and made deci- 

 sions as to where to go and what to fish. Hence, 

 relative total revenue appears to reflect the 

 fishing power of New England vessels. The 

 more fishing power, the higher revenues are 

 expected to be. 



The most powerful explanatory variables 

 for either total pounds or total value were the 

 fishing time variables. That is, the more days 

 fished or days absent, the higher the total value 

 and total pounds. On the basis of contributions 

 to the overall goodness of fit, there is no way 

 to choose between these two variables. Our 

 choice, therefore, will have to rest upon their 

 effects on other variables and on the cost of 

 gathering the information. 



In Problem 3, using total value as the 

 dependent variable and days fishing as the 

 measure of fishing time, crew size becomes 

 statistically nonsignificant and negative. In 

 Problem 4, when days absent is used, crew size 

 becomes statistically significant and a very 

 powerful explanatory variable. Days fishing 

 appears to be a less desirable measure of fishing 

 time in that: (1) It is theoretically inferior on 

 economic grounds as discussed previously; 

 (2) it causes other important variables to have 

 the wrong sign; (3) it costs more money to 

 collect this information; and (4) it is probably 

 more subject to error. 



The vessel size variables used were gross 

 registered tonnage (GRT) and horsepower (HP). 

 GRT was the more powerful of these variables 

 as it was statistically significant in all equations 

 and explained a large part of output. HP was 

 not as powerful a variable in terms of its partial 

 correlation coefficient. However, it was statisti- 

 cally significant when total value was the de- 

 pendent variable, indicating that it made an 

 independent contribution to fishing power. 



The variable that indicated the age of a 

 vessel had a negative coefficient and was sta- 

 tistically significant in most cases. There are 

 at least three hypotheses why older vessels 



47 



