With the advent of population aging, most demographers would probably expect such a rapid change of age structure to have favourable consequences for children and troubling ones for the aged. Fewer children should mean less competition for resources in the home as well as greater availability of social services earmaked for children. The sharp rise in the number of elderly persons should put enormous pressure on resources directed towards the old ages, such as medical care facilities, nursing homes, and social security funds. Samuel Preston argues to the contrary that the very increase in the number of the elderly has contributed to their imporoved well-being, while the decrease in the number of children has served to worsen their relative position. This is an anti-Malthusian postulation. The present paper is an application of Preston's thesis to the situation of Japan where culture and socio-political traditions are widely different from the ones in North America. Surprisingly, Preston's ideas seems to be applicable even to the case of Japan. The analysis presented in this paper can basically support the hypothesis that the very fact of increasing size of population works favourably for enhancing the well-being of its membership. The study first dealt with the three-year comparisons of Japanese expenditure data by age on the basis of special tabulations of the Social Survey for Health and Welfare Administration. In the second part, an attempt has been made to retabulate and analyze the latest available income data from the National Survey on Family Income and Expenditure collected by the Statistics Bureau. The results of both surveys seem to generally support the relevance to Preston's effect.

View full abstract

This study examines the effects of sib size and birth order on the coresidence of never-married youth with their parents in Japan, where eldest sons are often expected to live with their parents after marriage. Logistic regressions are performed using the data from 1982 National Fertility Survey (Single-Youth Survey) conducted by the Institute of Population Problems in Tokyo. While sib size has a significant and negative effect on prenuptial coresidence among both sexes, eldest-child status has a significant and positive effect among males only. The results support the hypotheses about more pressure toward daughters to stay home before marriage and to leave home after marriage ; more pressure toward eldest children to stay home before and after marriage ; and more pressure from crowding toward children of a larger family to leave home before marriage. Logistic regressions are also performed on the determinants of postnuptial coresidence plans. While the effect of eldest-child status is highly significant among males, that of sib size is not among either sex. This seems to suggest that the normative pressure toward eldest children about the postnuptial coresidence may be stronger than the pressure toward them about the prenuptial coresidence which can be affected by the opposing pressure from crowding.

View full abstract

In this paper, the entire age patterns of Japanese mortality (all and selected causes) which are described by means of mathematical representation called parameterized mortality schedules were discussed. A large number of mathematical expressions (functions) have been proposed and fitted to mortality, and the results have been widely applied to data smoothing, interpolation, comparative analysis, and forecasting. The Heligman-Pollard model which is one of the parameterized models used in United Nations, International Institute for Applied Systems Analysis (Austria) and Colorado University (A. Rogers) was fitted to Japanese mortality. Fit was almost good except for the years near 1900 and 1945, and after 1980. This means that this model may have to be slightly modified for Japanese mortality. The history of mortality decline during the 1895-1985 period was described by history of the eight parameters defining the Heligman-Pollard model. Furthermore, new parameterized cause-specific mortality models were presented for mortality from malignant neoplasms, heart disease and cerebrovascular disease. These kinds of models have many problems awaiting solution and no useful results have been reported yet. The work reported in this paper is only the first step in a sequence that should make operational the use of parameterized model schedules in mortality analysis. Finally, our future research plans for parameterized cause-specific mortality schedule using both the idea as stated above and another one were addressed.

View full abstract

Prefectural Life Tables have been constructed periodically by the Statistics and Information Department, Ministry of Health and Welfare, in every 5 years since 1965. Because of the lack of stability in age-specific death rates for small areas, life tables for municipal units are not successfully developed. In order to construct municipal life tables, it is crucial to stabilize age-specific death rates for small areas. For this purpose, it has been shown that Bayesian statistics provides a very useful tool. A study group, of which the authors are the members, was organized by Prof. Yukio Suzuki, emeritus professor for university of Tokyo, to apply the Bayesian approach for the first time to construct life tables in Japan. As the product of the group, "Municipal Life Tables 1985" was published in April 1989 by the Health and Welfare Statistics Association. In this paper, we first discuss methodology in applying Bayesian statistics to the construction of life table for small areas. Secondly, the effects of a Bayesian approach are shown in comparison with ordinary approach. Using the municipal life tables, regional death levels are also analysed to show new evidences. In 'Municipial Life Tables 1985', we used data of 5 years between 1983 and 1987. Every prefecture is divided into two districts:urban and rural, and death probabilities of the urban/rural district are used as informative prior to estimste death probabilities by sex and age class for each municipality within the district. As the result of applying Bayesian approach, these death probabilities are stabilized remarkably, and in almost all of the municipalities, the standard deviations for expectation of life at birth are within 0.5 years. It is also shown that the effectiveness of Bayesian statistics in reducing these standard deviations is especially remarkable for small municipalities. In pursuit of real death level in various areas, it is proved that municipalities in some specific areas enjoy long expectation of life and others not. This kind of new evidence is one of those which the municipal life tables alone provide.

View full abstract