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



importance. Beginning in 1984, these inventory 

 holdings were generally 15% of the exports. Before 

 1984, cold storage holdings formed a greater pro- 

 portion of exports, because exports were substan- 

 tially more limited in quantity. Shipping lags are 

 somewhat seasonal, and while no specific informa- 

 tion is available, shipments certainly require only 

 a relatively limited time. 



Asymmetric Pricing and Price 

 Transmission 



When any form of full short-run market integra- 

 tion is absent, price changes in the market of origin 

 are not immediately and fully passed on to the local 

 markets within one time period. Yet, while the cen- 

 tral and local markets may not be fully linked by 

 prices in the short run, if markets are not seg- 

 mented, a weaker form of short-run market integra- 

 tion may still be taking place. Moreover, the short- 

 run response to rising prices emanating from the 

 central market can differ from the response to price 

 declines. This price stickiness produces asymmetric 

 local market price responses to central market price 

 changes. 



While the Ravallion market model does not for- 

 mally incorporate these forms of market integra- 

 tion—incomplete short-run market integration with 

 asymmetric pricing, a modified Wolfram (1971) 

 framework developed by Young (1980) and Ward 

 (1982) does offer a formal model of price formation 

 for examining this possible form of short-run market 

 integration. The modified Wolfram framework pre- 

 sented in Ward uses a finite distributed lag func- 

 tion: 



P,;, = ao« + 2: [6;(Pi,_,.i - P,o) 



-I- (6; - 6;)P';,_j+i] + e,- 



(10) 



where Pn, indicates the central market price in the 

 initial time period and where 



P'v -(Pu-,- Pu-.-,)Zly (11) 



where Z;'_, = 1 if Pi,.i <Pi,_,-r 



= otherwise. 



The estimate of {bj - 6j) provides a direct test of 

 the asymmetric condition, where b'j measures the 



response to a rising price Pj and b'j relates to a 

 declining price Pj. 



Polynomial distributed lags can be substitited into 

 Equation (11) to provide structure, thereby reduc- 

 ing multicollinearity and conserving degrees of 

 freedom. For the case of a second-order polynomial, 



bj = Co -(- cj + c.J' 



and b'j - bj = d\ + dj + d2J'' 



(12) 



where the c's and d's are parameters to be eco- 

 nometrically estimated. The values and standard 

 errors of the 6's can then be recovered from these 

 estimates of the c's and d's. Significance tests on 

 df), di , and do as a group (a linear restriction on 

 Equation (10)) provide direct tests of asymmetric 

 price linkages. 



DATA 



Average monthly market price data for the Tokyo 

 central wholesale market during 1981-86 were ob- 

 tained from the Tokyo Central Wholesale Market 

 Yearbook. These are implicit prices formed by divid- 

 ing monthly total revenues by comparable quan- 

 tities. Most of the sablefish sold in this market are 

 not a homogenous commodity, so that movements 

 of these average sablefish prices may reflect changes 

 in the composition of the product form and quality. 

 However, because we were only able to obtain sim- 

 ple (unweighted) arithmetic average prices formed 

 by linear aggregation, we were forced to assume 

 that different product forms and grades are perfect 

 substitutes for each other. The data are in raw, un- 

 seasonalized form, without any prior seasonal ad- 

 justments or smoothing (which would otherwise con- 

 found the distributed lag relationships). '' 



Alaska and Pacific coast ex-vessel prices tend to 

 be competitively formed. Seattle dominates as the 

 port of export. Many of the prices received by 

 Alaska vessels are formed in a Seattle auction, in 

 which roughly 10-12 processors or brokers bid after 

 the auction has received a call from an Alaska vessel 

 reporting its catch information. Some ex-vessel 

 prices in Alaska and the Pacific coast are directly 

 formed when a vessel lands its catch at a processor. 



^Because these prices are average implicit prices, they may not 

 be equilibrium prices. Spot prices are available, but a problem with 

 spot prices is that the price of the time and day sampled may not 

 be indicative of the entire month. Also, while spot prices are avail- 

 able for different size classes, proportions in the market mix are 

 unavailable. 



345 



