Search

The article presents an empirical test of interregional input-output models. It estimates Japanese production in the year 1963. According to the author, substitution of 1963 technology for the 1960 base-year technology decreases errors of estimation for most industries, but substantial errors still occur for some industries as a result of shifts in interregional shipments. Therefore, only the point estimate gravity trade model results are presented in this article. The author states that all tests made so far indicate that the row coefficient model estimates Japanese regional production less accurately than column coefficient or gravity trade models. The method of handling foreign trade in the model may contribute to the instability of row coefficients for Japan. For any of the three models, a separate estimation of interregional shipments of foreign exports and imports may be desirable because of the importance of foreign trade in Japan. Finally, tests reveal no discernible difference in the overall predictive ability of the column coefficient and the point estimate gravity trade models. Both models give reasonable estimates of Japanese regional production.