SQUIRES ET AL.: JAPANESE AND U.S. SABLEFISH MARKETS 



single month (here, six months). In this case, this 

 modeling procedure is somewhat ad hoc in the sense 

 that long-run market integration is not a maintained 

 hypothesis, but nonetheless, the results provide 

 good insight into the nature of short-run market 

 integration. 



After experimentation with first-, second-, and 

 third-order polynomials with a six period lag length, 

 we estimated Equation (10) with a second-order 

 polynomial like that given in Equation (12).'' Direct 

 tests of asymmetric price responses are provided by 

 F-tests on the polynomical lag coefficients (the d's 

 of Equation (12)) corresponding to {b'j - b'j) in 

 Equation (10). The parameter estimates of Equa- 

 tion (10) with the second-order polynomial lag are 

 reported in Table 3 and the F-test results are re- 

 ported in Table 4. A Scheffe interval* is used to give 

 a more cautious test by providing a larger critical 

 value than that given by an F-test table due to the 

 experimentation and pretesting used to determine 

 the degree of polynomial. 



The significance test results (at a 5% level of sig- 

 nificance with a Scheffe interval) indicate symmetric 

 price responses, that is, the response in the Alaska 

 fixed gear ex-vessel market to rising Tokyo whole- 

 sale market prices does not differ from responses 

 to declining prices. The results are robust to changes 

 in the order of polynomial from first to second to 

 third and to inclusion or exclusion of an intercept 

 term in Equation (10). Because a first-order poly- 

 nomial did not give sensible results, the peak Alaska 

 response to a Tokyo price change is not immediate, 

 and does not continuously decline throughout the 

 price transmission period. 



The distributed lag estimated under the main- 

 tained hypothesis of symmetrical price responses 

 suggests that the peak price response in the Alaska 

 fixed gear ex-vessel market occurs by the end of the 

 third month after a price change in the Tokyo cen- 



'A second-order polynomial gave the most sensible shape to the 

 actual distributed lag recovered from the polynomial lag. More- 

 over, we followed a nested testing procedure for determining the 

 polynomial degree for a given lag length J suggested by Judge et 

 al. (1980). While these results marginally suggested a third-order 

 polynomial, the actual distributed lag (the 6's in Equation (10)) 

 recovered from the polynomial lag (given by Equation (12)) in- 

 dicated a more plausible shape for the second-order polynomial. 

 In any case, the degree of polynomial did not affect the hypothesis 

 test results for asymmetric pricing. Beginning and endpoint con- 

 straints were not used, and to be consistent with the Ravallion 

 approach, an intercept term was not included (which would other- 

 wise imply an unexplained constant relationship). 



'An F-test of linear restrictions using the Scheffe interval ad- 

 justs the confidence region, so that the F-test statistic is signifi- 

 cant only if it exceeds in magnitude [(a - 1) Fj"". where F is the 

 6 • 100% critical value for F(a - 1, T - a), T is the number of 

 observations, and a is the number of restrictions. See Snedecor 

 and Cochran (1976, p. 271) for details. 



Table 3.— Parameter estimates of asym- 

 metric price linkages model. Standard 

 errors in parentheses. 



Variable 



Parameter 

 estimate 



0,05761 

 (0 18720) 



-0.01289 

 (0.16581) 



-0.00136 

 (0.02676) 



-0.00027- 

 (0.00012) 



0.00019 



(0.00012) 



-0.00003 



(0.00002) 



NOTE: Estimates of Equation (10) with second- 

 order polynomial lag structure given in Equation 

 (12) Variable abbreviations are c, (parameters 

 of polynomial lag for deviations from initial price) 

 and d, (parameters of polynomial lag for asym- 

 metric price linkages). 

 ' denotes statistically significant at 5%. 



Table 4. — F-test for asymmetric price responses. 



F-statistic 4.23090 



F-test for overall significance of dg , d, , and dg 

 for asymmetric price linkages. 



tral wholesale market. Moreover, the impact of a 

 Tokyo central wholesale market price change dies 

 out after the fourth month. 



CONCLUDING REMARKS 



In this study, we examined the Tokyo central 

 wholesale sablefish market and the Pacific coast and 

 Alaska ex-vessel fixed gear sablefish markets for 

 several forms of long-run and short-run market in- 

 tegration and segmentation over 1981-86. 



We found that the Pacific coast fixed gear ex- 

 vessel and Tokyo central wholesale sablefish mar- 

 kets are segmented, so that changes in the Tokyo 

 market prices will have no effect, immediate or 

 lagged, on the prices of the Pacific coast market. 

 The Pacific coast market price instead depends only 

 upon its own lagged values and local market condi- 

 tions; the ex-vessel fixed gear markets operate in- 

 dependently of the Tokyo central wholesale market 

 over 1981-86. Pacific coast fixed gear harvesters 

 of sablefish are unlikely to be adjusting their 

 sablefish harvesting patterns in response to changes 

 in Tokyo central wholesale market price and demand 

 conditions. While a limited quantity of Pacific coast 



349 



