Age-Disaggregated Intra-Metropolitan Migration in Tokyo Metropolitan Area

Beginning in the mid-1950s, when Japan began to show a high rate of economic growth, the migration system in Japan was characterized by a drastic in-flow of people from the non-metropolitan areas to a few leading metropolitan areas, such as Tokyo, Osaka and Nagoya. This trend, however, has gradually changed since the 1970s in the wake of the oil crisis. One prominent pattern is a kind of return migration called a “U turn” or “J turn”.<br> During the past two decades, the trend of migration within Tokyo Metropolitan Area has also experienced a great change. Watanabe (1978a) argues that intra-metropolitan migration system of this region in the 1970s is composed of two main migration flow patterns, in-migration from non-metropolitan areas and centrifugal out-migration from Tokyo City to its periphery. The former is attributed to younger migrants, 15 to 19 years old, looking for jobs or entering universities. The latter, which is fueled by the former, involves relocation within the Tokyo Metropolitan Area by pepole 20-29 years of age who tend to be single or young married couples with no children, and out-migration to neighboring prefectures by people 30 years of age and over. This centrifugal out-migration is often related to marriage or some change in living arrangement that necessitates a move. Watanabe's (1978a) study further attempts to explain the intra-metropolitan migration system in Tokyo in terms of the progress of household members through the typical stages of the life cycle.<br> The purpose of this paper is to examine the patterns of intra-metropolitan migration by age group in the Tokyo Metropolitan Area, with reference to changes in migrants' life cycles. The 1980 population census contains data on regional migration flows during 1979-1980 except for flows of less than 50 people. Age-disaggregated migration flow data are unpublished without age-disaggregated total in-migration, out-migration, and intra-areal migration for each area. This paper therefore attempts the recovery of migration streams by age group from published aggregated census data, using entropy maximizing technique (Chilton and Poet, 1973). This method makes it possible to make the most accurate estimate of age-disaggregated migration OD data, subject to the constraints that honor the column and row totals for each age group, and the total of migrants moving from origins to destinations. Numerical examples of this method are given in Figure 3. Ultimately, the raw data for the present analysis are converted to a three-way data matrix composed of 49 origins in Tokyo, 157 destinations in Tokyo Metropolitan Area including Tokyo and the four neighboring Prefectures of Saitama, Chiba. Kanagawa and Yamanashi, and 11 age groups (see Figures 1 and 2).<br> In order to clarify the fundamental flow patterns of the complicated infra-metropolitan migration system in the Tokyo Metropolitan Area, Tucker's (1966) three-mode factor analysis is applied to the estimated age-disaggregated OD migration data. The procedure of three-mode factor analysis is shown in Figure 4. The three modes are summarized into factors one by one; last, the information contained in the data is collected into the three-mode factor score matrix-the core matrix.<br> The major findings of the three-mode foctor analysis can be summarized as follows:<br> 1) Age group factors of migrants and one-mode factor scores<br> Factor analysis of age groups produced only two factors, AGEFAC 1 and AGEFAC 2, accounting for 98.0% of the total variance. Table 1 shows the age groups with factor loadings after varimax rotation. In this case, one-mode factor scores matrix corresponds to 2 (age group factors)×49 (origins)×157 (destinations). The distributions of the dyads composed of origins and destinations with high one-mode factor scores are plotted in Figure 5.<br> The first age group factor, AGEFAC 1, is correlated with migrants 0-14 years of age and those 30 and over.