This paper presents the main ideas and empirical evidence of four major theoretical perspectives on elderly migration: post-retirement developmental perspective, life-course perspective, intergenerational perspective, and economic perspective. Although most of the empirical evidence is from the United States and Canada, most of the ideas discussed in this article may be useful to other countries that may have different levels of development and different cultural traditions. The main points are that these perspectives are essential for achieving a comprehensive understanding of elderly migration, and that the obsession with any one of them is likely to lead to misinterpretation and misunderstanding.

View full abstract

This paper presents an application of multistate life table models to the union formation and dissolution by marriage, divorce, remarriage, death of spouse and ego for the Japanese men and women in the years 1930, 1955, 1975 and 1995. The multistate life table of nuptilaity has a great methodological advantage of taking into account recurrent and reversible demographic events such as marriage and divorce. On the basis of the four-state marital status life tables thus constructed, some useful analyses have been made in regard to the duration of bachelorhood, marriage, widowhood and divorce in addition to attain more accurate estimation of probabilities of marriage, divorce, remarriage and separation from marriage by mortality. Up to the present time, multistate tables have been prepared for the Japanese population by a few researchers, but the present study is the first attempt to include therein the prewar experience, thus enabling a historical comparison over a sufficiently long period of time. At the same time, this study conducts an analysis of the Japanese nuptial life course by decomposing a difference in each indicator of nuptiality between two years into one attributable to mortality reduction and the other to the remaining probability changes in inter-status transition. Over 65 year's time span between 1930 and 1995, great mortality declines produced a substantial expansion in duration of marriage for both men and women, while the effects of the remaining factors, notably divorce and late marriage, were negative (shortening the marriage duration), but relatively minor. Breaking the period 1930-95 into three periods 1930-55, 1955-75 and 1975-95, for the periods 1930-55 and 1955-75, the spectacular declines in mortality overshadowed the other factors of late marriage and divorce by substantially lengthening the duration of marriage. After 1975, mortality decline over 50 is dominant. But for the period 1975-1995, the effect of mortality decline has subsided and become small, whereas the other factors, late marriage and divorce, have instead curtailed the marriage duration and their range of influence has been nearly twice as large as that of mortality for each sex. Particularly interesting are the changes in the duration of widowhood and interactions of the changes in mortality and the other factor of remarriage. The conventional wisdom erroneously conceives that the duration of widowhood for women would have become longer in the postwar period inasmuch as the life expectancy has greatly been extended. The multistate nuptiality tables reveals, however, that the duration of widowhood for each sex was rather surprisingly long, 9.05 years for men and 16.45 years for women in 1930 and it has not been much changed in the postwar years, for example 9.50 yeas for men and 15. 33 yeas for women in 1995. A rough explanation would be that while in the prewar time mortality was rampant in relatively young ages of 20s and 30s, thus giving a higher likelihood of cutting marriage short by death of spouse, in the postwar period early adult mortality has decreased, though, on the other hand, divorces and remarriages have increased instead. Also curious is the result of decomposition that among men the factor other than mortality, that is remarriage, plays a more important role in determining the duration of widowhood, whereas among women mortality reduction contributes always a much more significant part. The prewar and postwar contrast is particularly apparent when comparing their expected life-time marriage probabilities. Table 1 in this paper clearly depicts such a contrast. While the expected marriage probability was lower in children and early adulthood in 1930 than in 1975 for both men and women, but in the middle ages it was much higher in 1930 than in 1975 and 1995. This would indicate that what might be called "marriageable ages" have substantially been

(View PDF for the rest of the abstract.)

View full abstract

The purpose of this paper is to discuss the levels of reproduction of daimyo retainers in the last half of Tokugawa period in Japan. Sekiyama (1958) assumes that the samurai class did not reproduce itself. However, this claim has not yet been confirmed because we cannot directly calculate the reproduction rate from genealogies, which are the primary source for the analysis of samurai population. This study approaches this issue through the method of micro-simulation. First I conducted the simulation by positing a stable population at a simple reproduction level (NRR≒1), using a micro-simulation program, called CAMSIM, which had been developed in the Cambridge Group for the History of Population and Social Structure. Most values of parameters needed for the simulation were calculated from the personal histories of daimyo retainers of Uwajima han, which had been submitted to Uwajima han office. Especially, the age-specific mortality rates listed in the life table of west 8 and 9 and north 8 of Coale and Demeny (1983) were selected as the values of the parameters in consideration of infant mortality rate that could be only calculated from the personal histories. The NRR derived from the simulation was 1.02 with west level 8, 1.09 with west 9, and 1.00 with north 9. In this way, the population of more or less simple reproduction level could be simulated. Next I calculated the mean number of sons per married 'family head'. In the simulated population in question, I defined the 'family head' as the surviving eldest son of 'ego' at his death. The number was 2.14 with west 8, 2.11 with west 9, and 2.21 with north 8. The number can be said to be within the range of 2.11-2.21. I also calculated the mean number of sons surviving more than 15 years per 'family head' from the results of the simulation. The number was 1.20 with west 8, 1.31 with west 9, and 1.24 with north 9. The number can be said to be within the range 1.20-1.31. Third I compared them with the equivalents from genealogies of daimyo retainers in eight hans, which are presented in Murakoshi (1991, 1993, 2001). The mean number of sons per married family head of retainers of Uwajima han was 2.13. This number was more or less the same as the number from the result of the simulation (2.11-2.21). Many values of parameters were calculated from the personal histories of retainers of Uwajima han. Therefore I claimed that the result of the simulation should be acceptable. The mean numbers of adult (surviving more than 15 years) sons per family head of retainers of seven hans-Tokushima han, Aizu han, Akita han, Morioka han, Kaga han, Hagi han and Saga han-have been already estimated. The number in Tokushima han was within the range of 1.22-1.26 from the last half of the 18^<th> century to the first half of the 19^<th> century. The number in Aizu han was 1.24 in the last half of the 18^<th> century. The number in Akita han was 1.22 in the last half of the 18^<th> century the number in Morioka han was 1.25 from the first half of the 18^<th> century to the last half of the 18^<th> century. The number in Kaga han was 1.19 in the last half of the 18^<th> century. The number in Hagi han was 1.20 in the first half of the 18^<th> century. The number in Saga han was within the range of 1.21-1.26 from the last half of the 18^<th> century to the first half of the 19^<th> century. In brief, the mean numbers of adult sons in seven hans were within the range of 1.19-1.26. The numbers calculated from the result of the simulation were within the range of 1.20-1.31. The two ranges were found more or less equal. Therefore I concluded that daimyo retainers must have reproduced themselves, rejecting Sekiyama's claim.

View full abstract

Understanding the idiocyncracies of various migration definitions constitutes basic prerequisites for migration research. The purpose of this paper is to compare the data of 'the last moves within five years' with those of 'residence five years earlier', and examine how these definitions differently delineate the same migrantions. The data used for the study stem from the Fourth Migration Survey, which was conducted in 1996 by the National Institute of Population and Social Security Research, Japan. Analyses show the following points; 1) Total number of migrants is larger in the 'last migration' data, since the data of 'residence five years earlier' cannot grasp those who moved out and returned to the same place within five years. 2) An examination by place of the previous residence reveals that as the distance of move increases, the number of migrants by 'residence five years earlier' rises relative to the one by 'last migration'. This tendency would be explained by the characteristics of multiple moves within five years. While the data differences between the two definitions are caused by multiple moves (including return migrations) within five years, distance patterns of these moves seem, as far as the present dataset is concerned, to be predominantly 'long-to-short' in their sequences. Since the data of 'residence five years earlier' grasp the earlier moves and 'the last move' data focus on the more recent, the former tend to outnumber the latter in the longer distance moves, and the vice versa in the shorter movements. 3) The above pattern of migration distance seems to be influenced by the lifecycle-related characteristics of migration, namely long-distance moves for educational advance or job search in the late teens or early 20s, followed by short-distance relocation for better housing. But this pattern is not necessarily observed among older migrants. 4) In term of reasons of move, higher percentages of 'residence with/near parents' and 'marriage/divorce' are observed in return migrations. Some other multiple moves also seem to show idiosyncrasies in their reasons and migrants' attributes. It could be concluded that the relationships between the migration indicators concerned would not remain constant but change in relation to temporal fluctuations in the trends of return migration, reasons of moves, and attributes of migrants, which are shaped by time-to-time socio-economic situations.

View full abstract