We examined the explanatory power of income differentials model which explains internal migration in post-war Japan. Using prefectural data for 1955-95, we performed lots of cross-sectional regression analysis every year based on (1) two types of regional division (10 regions and 36 regions), (2) three kinds of dependent variables (net migration rates, in-migration rates and out-migration rates) and (3) four kinds of dependent variables (differentials of per capita distributed income, per capita personal income, per capita non-personal income, and per capita employer's income).
The results of our analysis are summarized as follows.
First, regional division affects the results of internal migration in postwar Japan to some extent. It affects little when net migration rates are explained, but especially in the regression analysis of in-migration rates of 36 regions the values of Β estimate show negative sign instead of positive sign of 10 regions during two equilibrating periods of internal migration.
Second, differentials of per capita distributed income explain the net migration rates greatly, but those explain the in-migration rates and out-migration rates of 10 regions weakly, and in-migration rates of 36 regions shows negative sign during two equilibrating periods of internal migration.
Third, we can use differentials of per capita distributed income as the proxy variables of per capita personal income, because differentials of per capita distributed income have the same explanatory power as differentials of per capita personal income.
Fourth, income differentials model has the most explanatory power in the case of (1) long distance migration in which people move out of their local labor markets completely and (2) net migration in which out-migration offsets in-migration.