FISHERY BULLETIN: VOL. 86, NO. 2 



particular, not all of the estimated coefficiencts 

 have the desired or expected sign. For example, 

 the coefficient for tonnage in the 1981 estimate is 

 negative and that for the square of tonnage is 

 positive. In contrast, a positive coefficient for ton- 

 nage was expected since vessel cost should in- 

 crease with vessel tonnage, and the coefficient for 

 tonnage squared was expected to be negative be- 

 cause vessel costs are believed to increase at a 

 decreasing rate. 



The incorrect signs could be the result of multi- 

 collinearity between vessel characteristics. Aux- 

 iliary regressions, though, do not indicate severe 

 multicollinearity. The R^ values are <0.35. If 

 multicollinearity is a problem, its major effect is 

 on the estimated relationships between cost and 

 individual characteristics, the estimated elastic- 

 ities, and the previously discussed Chow tests. 



Multicollinearity is not thought to be a problem 

 for several reasons. First, Chow tests in the pres- 

 ence of multicollinearity usually result in the ac- 

 ceptance of the null hypothesis of the equality of 

 regression equations or that differences in the es- 

 timated coefficients are statistically insignificant 

 (Maddala 1977). Second, multicollinearity does 

 not pose a problem for estimating capital stock 

 and investment; the estimated equations still per- 

 mit estimation of the conditional mean.^^ 



11 For additional information on the problems of multi- 

 collinearity with respect to analysis and prediction, see Kmenta 

 (1971). 



The equations for the real or deflated value of 

 capital were used to estimate total capital stock 

 and investment for the New England fleet be- 

 tween 1965 and 1981. The real capital stock equa- 

 tions were applied to vessel data available from 

 the Northeast Fisheries Center, and total capital 

 stock was calculated as the sum of the stock over 

 all vessels. Aggregate net investment was calcu- 

 lated as the difference in total capital stock be- 

 tween consecutive years. 



Prior to estimating total capital stock and in- 

 vestment, it was necessary to define vessel type 

 by gear to avoid double counting which might 

 occur since vessels frequently switch port and 

 gear. Three categories of gear types were estab- 

 lished: 1) otter trawl, 2) scallop dredge, and 3) all 

 others, which include lobster trawl and pots, her- 

 ring gear, harpoons, etc. Vessels were assigned a 

 gear type based on a plurality of days at sea. Esti- 

 mates of capital stock and investment are pre- 

 sented in Table 5. 



Several limitations should be considered when 

 evaluating the estimates. First, there is the previ- 

 ously discussed problem that estimates of capital 

 stock may be biased or overestimated. Second, 

 estimates are only for New England vessels or 

 those which are believed to be homeported in New 

 England. Third, the capital stock and investment 

 series pertain to New England and not the United 

 States; that is, estimates should not be inter- 

 preted as net changes in the capital stock and 

 investment in U.S. fisheries. Fourth, under- 



Table 5. — Estimates of real capital stock and investment in New England harvest- 

 ing sector, 1965-81.1 



'Real Investment and capital are deflated with producer pnce Index for tieavy machinery, 

 1967 = 100. 



346 



