The purposes of this paper are, first, to analyze population distribution changes 1965-95 in Japan by four categories-the existing metropolitan area; new suburbs; new metropolitan area; and non-metropolitan area-by using the Hoover index, and, second, to compare the results obtained for Japan with those for the US presented in Nucci and Long (1995) and Long and Nucci (1997). To clarify the temporal changes in population distribution, the delineation of metropolitan areas needs to employ a floating principle rather than a fixed one, and the definition of metropolitan areas should be similar to that of Standard Metropolitan Statistical Areas (SMSA) used in the US. To meet these conditions, the Standard Metropolitan Employment Areas (SMEA) advocated by H. Yamada and K. Tokuoka are used in this article.
In Japan, during the period 1965-75, there occurred a considerable population concentration due to a marked population increase in the old suburbs located in the core regions (such as Kanto, Tokai and Kinki) and to a large population decrease in non-metropolitan areas in the peripheral regions (such as Hokkaido, Tohoku, Hokuriku, Chugoku, Shikoku and Kyushu). During the period 1975-85, the difference between the population change rates of the core and the periphery levelled off. However, in the peripheral regions, the main category of population growth was the old central cities and depopulation in non-metropolitan areas continued. In the core regions, the category showing the largest rate of increase shifted from the old suburbs to the newly-added suburbs, implying population deconcentration from the old SMEAs. During 1985-95, deconcentration proceeded further in the core portion of the country. Meanwhile, in the peripheral regions, whereas the category showing the largest rate of increase finally shifted from the old SMEAs to the newer suburbs, non-metropolitan areas returned to recording a heavy population decrease.
With respect to the 1965-95 period as a whole, the population increase in the newer SMEAs created in the non-metropolitan areas was negligible. Bearing in mind the depopulation of the non-SMEAs, this means that we cannot regard the overall trend of population redistribution as being counterurbanization.
With regard to the Japan-US comparison, population deconcentration in the old metropolitan areas is common to both countries, but the following remarkable differences can be noted. First, in the US, the contribution of the new metropolitan areas created in the non-metropolitan areas to urbanization is more conspicuous than that of the new suburbs added to the old metropolitan areas. Second, whereas a clean break with the past in terms of population redistribution was found in the US, such a phenomenon was not observed in Japan. Third, in Japan, population growth in the core regions and depopulation in the peripheral regions are found throughout the period, whereas in the US, both types of region take on different change aspects and the spatial patterns of population redistribution are far more complicated.