SQUIRES: EX-VESSEL PRICE LINKAGES 



the results indicate that the cod and haddock price 

 linkage does not run from New Bedford to Glouces- 

 ter. If a Scheffe interval is not used, then the New 

 Bedford cod and haddock prices do lead those of 

 Gloucester. Therefore, with this caveat, New Bed- 

 ford auction market monthly round ex-vessel cod 

 and haddock prices lead the prices of Gloucester and 

 Boston, and Boston prices may lead those of 

 Gloucester. In any case, it appears that the New 

 Bedford auction market dominates the formation of 

 round ex-vessel prices for cod and haddock. 



The empirical results for yellowtail and winter 

 flounder of Table 2 also contradict the null hypoth- 

 esis that monthly fresh round ex-vessel prices for 

 both species are formed first in New Bedford. In- 

 stead, the findings indicate that pricing feedback ex- 

 ists between both New Bedford and Gloucester and 

 between New Bedford and Boston. These conclu- 

 sions must be tempered by the significant Q-test 

 statistics for several relationships. 



These conclusions lead to a second null hypothesis 



between the prices of New Bedford and Gloucester, 

 New Bedford and Boston, and possibly between 

 Gloucester and Boston rests with a spurious rela- 

 tionship. Although New Bedford is the most impor- 

 tant flounder port by landings in New England, New 

 York City is even more important on the eastern 

 seaboard by volume of consumption. New York 

 City's Fulton Fish Market is primarily a wholesale 

 market without substantial landings. Much of the 

 New England flounder harvested is sent to Fulton 

 on consignment without an ex-vessel price being 

 established in New England. The Fulton Fish 

 Market also begins much earlier in the morning than 

 New Bedford's auction market. Thus the apparent 

 feedback among the ex-vessel yellowtail and winter 

 flounder prices in the New England ports is prob- 

 ably due to their following of the wholesale prices 

 set in the Fulton Fish Market. 



Table 3 presents the results for pollock. As with 

 the other species, consistent results are obtained for 

 different lag lengths. Again, the null hypothesis dic- 



Table 2.— Direct Granger causality tests for monthly fresh round ex-vessel yellowtail and 



winter flounder prices. 



Yellowtail flounder 



Winter flounder 



Direction 1 



Lags 2 F-test 3 Q-test 4 



Direction 1 



Lags 2 F-test 3 Q-test 4 



NB 



NB 



G 



G 



NB 



NB 



B 



B 



G 



G 



B 



B 



->G 



->G 



->NB 



->NB 



->B 



->B 



->NB 



->NB 



->B 



->B 



->G 



->G 



8 

 14 



8 

 14 



8 

 14 



8 

 14 



8 

 14 



8 

 14 



2.05* 



2.27* 



1.96* 



2.48* 



2.02* 



1.76* 



0.75 



2.83 



2.275 



2.50 s 



1.48 



1.27 



20.19 



19.42 



23.56* 



26.78* 



10.56 



9.89 

 17.63 

 21.83* 



9.41 

 12.74 

 16.86 

 18.82 



NB 



NB 



G 



G 



NB 



NB 



B 



B 



G 



G 



B 



B 



->G 



->G 



->NB 



->NB 



->B 



->B 



->NB 



->NB 



->B 



->B 



->G 



->G 



8 

 14 



8 

 14 



8 

 14 



8 

 14 



8 

 14 



8 

 14 



7.52* 

 8.71* 

 2.88* 

 2.97* 



11.87* 

 7.66* 



23.57* 

 5.25* 

 3.87 5 

 4.37 5 

 4.15 s 

 3.56 s 



5.68 



1.78 



16.98 



22.19* 



18.45 



7.92 



9.94 



3.99 



12.43 



4.70 



18.16 



17.34 



'Variable abbreviations are B (Boston), G (Gloucester), NB (New Bedford). 



2 J indicates J months lagged. 



3 Null hypothesis that past values of the causal variable do not significantly affect current values of the 

 dependent variable. An asterisk indicates rejection of the null hypothesis at the 5% level. 



"Null hypothesis that regression residuals are white noise. An asterisk indicates rejection of the null 

 hypothesis at the 5% level. 



5 F-test statistic is significant at the 5% level, but not significant at the 5% level when a Scheffe interval is used. 



to be tested on the yellowtail and winter flounder 

 price linkages between Gloucester and Boston. Since 

 this test is also an unplanned comparison, a Scheffe 

 interval is required. Again, the strict test results in- 

 dicate that neither port's prices lead the other, nor 

 that feedback exists. 

 The most probable explanation for the feedback 



becomes significant only if it exceeds in magnitude ((a-l]F h )' k , 

 where F is the b  100% critical value for F (a-1, N-a) and N is the 

 number of observations. See Snedecor and Cochran (1976, p. 271) 

 for more details. 



Table 3.— Direct Granger causality tests for monthly fresh round 

 ex-vessel pollock prices. 



Direction 1 Lags 2 F-test 3 Q-test 4 



B >G 8 4.34* 12.54 



B >G 14 4.99* 10.15 



G >B 8 5.28* 18.27 



G >B 14 4.77* 22.96* 



'Variable abbreviations are B (Boston), G (Gloucester), NB (New Bedford). 



2 J indicates J months lagged. 



3 Null hypothesis is that past values of the causal variable do not significantly 

 affect current values of the dependent variable. An asterisk indicates rejec- 

 tion of the null hypothesis at the 5% level. 



"Null hypothesis that regression residuals are white noise. An asterisk in- 

 dicates rejection of the null hypothesis at the 5% level. 



441 



