care of the elderly in japan: changing norms and expectations · changing norms and expectations...

Care of the Elderly in Japan: Changing Norms and ExpectationsAuthor(s): Naohiro Ogawa and Robert D. RetherfordReviewed work(s):Source: Journal of Marriage and Family, Vol. 55, No. 3 (Aug., 1993), pp. 585-597Published by: National Council on Family RelationsStable URL: http://www.jstor.org/stable/353340 .Accessed: 04/11/2011 05:16

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

National Council on Family Relations is collaborating with JSTOR to digitize, preserve and extend access toJournal of Marriage and Family.

http://www.jstor.org

http://www.jstor.org/action/showPublisher?publisherCode=ncfr

http://www.jstor.org/stable/353340?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

NAOHIRO OGAWA Nihon University

ROBERT D. RETHERFORD East-West Center*

Care of the Elderly in Japan:

Changing Norms and Expectations

This article analyzes changes in norms of filial care for elderly parents and expectations of old- age support from children, based on a series of national surveys of currently married women of reproductive age in Japan. The analysis suggests that norms offilial care for elderly parents were fairly constant from 1963 until 1986, when a major weakening of norms began. The sudden normative shift after 1986 is conceptualized as a rapid diffusion process, which appears to have been triggered by government efforts to shift some of the burden of caring for the elderly back to families. In contrast to norms of care for elderly parents, expectations of old-age support from children have declined steadily over time, adapt- ing continuously to changes in underlying socioeconomic and demographic conditions. Overall, the findings suggest that government efforts to shift the burden of caring for the elderly back to families may not be successful.

Although a number of studies have analyzed the weakening of multigenerational coresidence patterns in Japan, focusing especially on households in which elderly parents reside with adult children

(Kojima, 1989; Martin & Tsuya, 1991; Morgan & Hirosima, 1983), there have been no detailed studies of concomitant changes in norms of filial care for elderly parents or expectations of old-age support from children. In this study we analyze trends in these norms and expectations, based on a series of national surveys extending back to 1950.

Our data derive principally from two sets of questions about caring for one's elderly parents and receiving old-age care from one's children. These questions were asked of currently married women of reproductive age. Responses to the first set of questions are characterized as reflecting norms of filial care for parents, and responses to the second set of questions are characterized as reflecting expectations of old-age care from children. By norms, we mean socially accepted stan- dards about how people should behave. Expectations, on the other hand, are less con- strained by what other people think. Thus, a woman may bring shame on herself by failing to care for her parents (or, more commonly in Japan, her husband's parents), whereas she does not bring shame on herself by not expecting old-age support from her children. Although the distinc- tion between norms and expectations in this context is not a sharp one, it will prove useful in our subsequent analysis.

We first consider pertinent theories and frame hypotheses to be tested with our data. We then discuss socioeconomic and demographic changes that have affected trends in norms and expecta-

Population Research Institute, College of Economics, Nihon University, 3-2 Misaki-cho 1-chome, Chiyoda-ku, Tokyo 101, Japan.

*Program on Population, East-West Center, Honolulu, HI 96848.

Journal of Marriage and the Family 55 (August 1993): 585-597 585

586 Journal of Marriage and the Family

tions about care of the elderly in Japan. We next undertake a multivariate analysis of the determinants of norms and expectations of care of the elderly, based on individual-level data from our series of national surveys. The concluding section summarizes the findings and comments briefly on their policy implications.

THEORY AND HYPOTHESES

Japan's transformation into an advanced industrial nation has been accompanied by changes in intergenerational relations within the family that, while unique in certain respects, bear similarities to the historical experience of Western industrial nations. This historical experience has been placed in theoretical perspective by authors such as Davis and Combs (1950), Cowgill and Holmes (1972), Davis and van den Oever (1981), and Lee (1984). The gist of most theories is that, as the transformation from an agrarian society to an industrial society proceeds, the burden of caring for elderly parents tends to shift from adult children to the state, to businesses, unions, and other private sector organizations, and to the elderly them- selves.

One reason why this occurs is that, sooner or later, modernization causes a demographic transition from high birth and death rates to low birth and death rates, a shift that in turn causes a major increase in the old-age dependency ratio (mea- sured here by the ratio of the population aged 65 and over to the population aged 20-64). For example, if, under conditions of replacement-level fertility, life expectancy improves from 40 to 75 years and fertility declines commensurately, the old-age dependency ratio rises from about 14 to 30 elderly persons per 100 persons of working age (calculated from Model West female stable populations with these life expectancies and replacement-level fertility; Coale & Demeny, 1983). The large increase in the old-age dependency ratio creates pressure to shift at least part of the dependency burden of caring for the elderly from familial to societal support systems.

A more complex set of reasons has to do with the shift from a peasant-agrarian economy to an urban-industrial economy (Davis & Combs, 1950; Davis & van den Oever, 1981). In a peasant-agrarian economy, production tends to be family-based and unspecialized, with little labor mobility. Sons tend to follow the occupations of their fathers, and children of both sexes tend to learn their economic roles by working along with

their parents. Because parents usually have more accumulated work-related experience and knowl- edge than their adult children in their shared work, parents tend to retain a high degree of respect and authority in their old age. Age and economic position usually coincide as bases of parental authority, so that intergenerational conflict is minimized. Under these circumstances, coresidence of elderly parents and adult children, supported by norms of filial piety, makes both economic and social sense.

In an urban-industrial economy, however, production is no longer family-based, occupations are specialized, and labor is mobile. Jobs are obtained through a labor market that treats individuals as free agents, not members of families. The new separation of work from family life fosters a greater degree of individualism, not only in the labor market but also in kinship relations. Sons tend not to follow the occupations of their fathers, and parents and adult children often reside in different localities, making coresidence impossible. Where coresidence arrangements do exist, parental authority retains a social basis but little or no economic basis, thereby increasing the potential for intergenerational conflict. The potential for such conflict is further increased by the greater individualism of young wives, who are generally well-educated and often work outside the home independently of their families, and by differences between the generations in life style and outlook.

Under these circumstances, it is natural that both adult children and elderly parents often prefer, if possible, to maintain at least partial inde- pendence by residing in separate households. At the same time, the maintenance of separate households becomes more feasible because of rising in- comes, which enable accumulation of personal savings for old age, and because of private pension programs and social security programs, which arise in response to the socioeconomic and demographic changes just described.

During this modernization process, norms of filial care for elderly parents also weaken. Here we draw on a theory of normative change developed by Retherford (Retherford, 1985, 1987; Retherford & Palmore, 1983). Like other norms, norms of care for the elderly tend to be rooted in the conditions of everyday life, and they therefore tend to change as these underlying conditions change. To some extent, however, norms have a life of their own, especially if they become for- malized in religious doctrine or civil law, as is often true of norms about filial obligations to par-

Care of the Elderly in Japan 587

ents. For this reason, and also because deviation from norms entails social disapproval and psycho- logical costs, normative changes may lag behind changes in underlying socioeconomic and demographic conditions. The faster the underlying conditions change, the greater this lag is likely to be, because it takes time for norms to adjust.

When norms adjust to changes in underlying conditions, the diffusion process is not necessari- ly gradual. It sometimes proceeds in spurts, which can be sparked, for example, by coalesence of ad- vocates of normative change into organized movements, by enactment of new laws, or even by simple advocacy of change by respected leaders of a population latently receptive to such change. Spurts of normative change can be viewed as a process of rapid diffusion of new rules of behavior.

Rapid diffusion of normative change is most likely to be observed in societies with a high degree of social integration, by which we mean widely shared norms and values and good internal communication. A high degree of social integration is typically found in societies that are homogeneous in such characteristics as language, ethnicity, and religion, and that have highly developed systems of communication, particularly radio and television. Once a normative shift begins in a highly integrated society, the diffusion process often proceeds swiftly because the population tends to function normatively as a unit.

It seems plausible that Japan, as one of the most culturally homogeneous and socially integrated countries in the world, could exhibit the pattern just described. We accordingly hypothesize a pattern of change in norms of filial care for elderly parents whereby these norms remain relatively stable for a time, then weaken rapidly across the population (starting with opinion leaders in the higher socioeconomic groups), and then stabilize at a new plateau. On the other hand, we hypothesize that expectations of old-age support from children, being less sensitive to normative constraints, adjust more gradually and continuously to changes in underlying socioeconomic and demographic conditions.

SOCIOECONOMIC AND DEMOGRAPHIC CHANGE IN

POSTWAR JAPAN

How closely does Japan's experience conform to these hypothesized dynamics of change? Since World War II, the pace of socioeconomic and demographic change in Japan has been extraordinar-

ily rapid (Minami, 1986). For example, between 1950 and 1980, real per capita income, often used as a measure of general level of development, increased by more than seven times (Preston & Kono, 1988). Between 1950 and 1990, the proportion living in urban places increased from 37% to 77% (Ogawa, 1986; Statistics Bureau, 1992). Extraordinarily rapid declines in fertility and mortality occurred as well (Hodge & Ogawa, 1991; Otani & Atoh, 1988). Between 1947 and 1990, the total fertility rate (the number of births a woman would have if she experienced current age-specific birth rates throughout her reproductive age span) fell from 4.5 to 1.5 children per woman, and life expectancy increased from 50 to 76 years for males and from 54 to 82 years for fe- males (Ministry of Health and Welfare, 1991).

As a consequence of major declines in fertility and mortality, Japan's population aged rapidly. The proportion of the population aged 65 and over increased from 5% in 1950 to 12% in 1990, and is projected to increase to 25% in 2020 (Ogawa, 1989). By way of comparison, the country with the oldest population in the world in 1990, Sweden, had 18% of its population aged 65 and over (United Nations, 1991). Japan's old-age dependency ratio (ages 65+/ages 20-64) increased from 8 per 100 in 1950 to 17 per 100 in 1990, and is projected to increase to 40 per 100 in 2020, far in excess of what is projected for any other country at that time (Ogawa, 1989).

Household structure has changed in the direc- tion of relatively fewer stem households and relatively more nuclear households (Morgan & Hirosima, 1983; Preston & Kono, 1988). Between 1960 and 1990, the proportion of the elderly aged 65 and over living with their children fell from 87% to 50%, a level still much higher than in other advanced industrial countries (Statistics Bureau, 1992). According to one international comparative survey conducted in 1990, the proportion of the elderly aged 60 and over living in three-generation households was less than 1% in England, slightly more than 1% in the United States, 3% in Germany, but 32% in Japan (Management and Coordination Agency, 1991). These comparative data indicate that changes in coresidence patterns in Japan have lagged behind changes in underlying socioeconomic and demographic conditions.

In Japan, older persons are also less likely to be institutionalized (Gibson, 1992; Maeda, 1990). The proportion of persons aged 65 and over who were institutionalized was only 1.6% in Japan in


1985 but 5% in the United States in 1987 (Martin, 1989). On the other hand, hospital stays of elderly persons are about ten times longer in Japan than in the United States (Health and Welfare Statistics Association, 1988; U.S. Bureau of the Census, 1990), indicating a certain amount of "social hospitalization" (Martin, 1989).

The unusually high incidence of coresidence in Japan partly reflects the persistence of Confucian moral prescriptions about filial obligations to parents. These moral prescriptions, which are reinforced by a "shame culture" characterized by deep sensitivity to social approval, emphasize obligations between individuals in direct lineal descent within stem families (Kendig, 1989; Martin, 1989). Reflecting this traditional morali- ty, more than 60% of those coresiding with parents stated in a recent survey that they had chosen such living arrangements mainly because it was their duty as the eldest son or daughter (Kojima, 1989; Martin & Tsuya, 1991; Morgan & Hirosima, 1983). Customarily, the first in line for this duty is the eldest son, if there is one, and, in fact, more than 80% of coresident households are of this type (Ogawa & Ermisch, 1991). In such households, it is usually the son's wife who plays the role of principal caregiver for her parents-in- law if they need care (Campbell & Brody, 1985). Reinforcing Confucian tradition are legal requirements of familial support of the elderly that still exist in Japan, although they are rarely invoked. In this respect also, Japan is unusual, inasmuch as such requirements have been abolished in most other industrial nations (Gibson, 1992).

Because women are the primary caregivers for elderly parents, changes in the status of women have also had an impact on care of the elderly. Educational attainment of women has been rising rapidly in Japan over the last few decades (Nakata & Mosk, 1987). In 1955, only 5% of women in the relevant age groups were enrolled in junior college or university, compared with 15% of men. By 1990 these figures had risen to 37% and 35%, respectively, so that the enrollment ratio of women slightly surpassed that of men (Ministry of Education, 1991). However, within the higher- education category, women are more concentrat- ed in junior colleges than are men. There has also been a major shift in female employment, away from farming and unpaid family work to paid employment outside the home (Ogawa, 1987; Ogawa & Hodge, in press). Among married women aged 20-49, the proportion working as paid employees rose from 13% to 42% between

1963 and 1990 (Ogawa & Ermisch, 1991; Shimada & Higuchi, 1985). During the same period, the proportion working as family workers de- creased from 13% to 10%. These changes in women's education and work outside the home have tended to weaken obligations to elderly parents (Maeda, 1990).

Legal changes have also weakened these obligations. In 1948, the Civil Code was revised to reflect more closely a philosophy of individual rights suited to the requirements of a modern economy (Kendig, 1989). The new Civil Code, which supplanted the old Meiji Civil Code, did away with primogeniture inheritance and the legal power of family heads over arranged marriage (Sano, 1958). These Code revisions were not only a response to past changes in gender roles, but also contributed to further changes in gender roles (Campbell & Brody, 1985).

Yet another factor tending to weaken obligations to care for elderly parents is increased public support for the aged in the form of pension and medical coverage. In 1961, the government estab- lished a universal pension and health care system. Japan's recent pension replacement rates (defined as the ratio of pension to previous earnings, taking into account how long the recipient paid into the system before retiring) are now comparable to those in other industrial nations (Clark, 1990). Although Japan's outlays for social security are still low compared with those in other advanced industrial nations (less than half of that in many Western European nations), they have grown substantially over time. The proportion of gross domestic product allotted to the social security system rose from 5% to 12% between 1965 and 1987 (Social Insurance Agency, 1991). The main reason why the proportion of gross domestic product allotted to social security is comparatively low is that Japan's social security system has not yet matured; in other words, it still pays out comparatively modest benefits because current beneficia- ries did not pay much into the system due to the recency of the program.

Although the evidence suggests that support of the elderly within families is increasingly based on personal needs and motives rather than blind con- formity to traditional norms and values (Campbell & Brody, 1985; Maeda, 1990; Sakamoto, 1975), coresidence patterns and supporting norms and values have nevertheless changed slowly during the postwar period, compared with the rapidity of underlying socioeconomic and demographic change (Kuroda & Hauser, 1981; Maeda, 1990).


Comparisons with other developed countries support such a conclusion. As suggested earlier, one could plausibly argue that this is an instance of normative change lagging behind rapid changes in underlying socioeconomic and demographic conditions, a hypothesis which we address in our analysis below.

TRENDS IN NORMS AND EXPECTATIONS OF CARE FOR THE ELDERLY

Our analysis of changes in norms and expectations of care for the elderly draws on nationally representative survey data from the various rounds of the National Survey on Family Planning, conducted by the Population Problems Research Council of the Mainichi Newspapers approximately every other year since 1950. Respondents are currently married women below age 50. Although the most recent survey round added women who were not currently married, we confine our attention to currently married women in order to maintain comparability with earlier survey rounds.

The various survey rounds were based on stratified multistage cluster samples. Cities, towns, and villages throughout Japan were first stratified by population and other characteristics. Respondents were then randomly chosen from the resident reg- isters in each primary sampling unit. (For a detailed discussion of sampling methods, see Hodge & Ogawa, 1991.)

Norms of Filial Care of Elderly Parents

Since the seventh survey round in 1963, the Mainichi surveys have included a question about filial care of elderly parents: "What is your opinion about children caring for their elderly parents?" (original question in Japanese). The concept of care is broad, including financial, physical, and emotional care, and is not restricted to coresidence. The seven response categories for this question are good custom, natural duty as children, unavoidable due to inadequacy of public facilities and old-age pensions, not a good custom, do not know, other, and no answer. "Good custom" is considered a stronger pro-care response to this question than "natural duty" because it repre- sents a stronger affirmation of the Confucian ideal of filial piety. As already mentioned, we view this question as reflecting norms of filial care for elderly parents.

Table 1 reveals changes in the distribution of

TABLE 1. DISTRIBUTION OF RESPONSES ABOUT CARING FOR ELDERLY PARENTS (BY PERCENTAGE)

Response Categories Number

Good Natural Not Good Of Year Custom Duty Unavoidable Custom Other Cases 1963 39 41 9 3 8 2,970 1965 38 43 8 3 8 3,021 1967 35 42 10 3 10 3,051 1969 31 43 9 4 13 3,082 1971 28 46 10 4 12 3,112 1973 29 45 11 4 12 3,022 1975 26 49 9 3 13 2,935 1977 28 47 9 4 12 2,884 1979 25 49 10 4 12 2,890 1981 29 51 6 3 11 3,025 1984 22 55 6 4 14 2,682 1986 17 58 7 5 14 2,503 1988 21 44 13 7 15 2,493 1990 20 30 22 12 15 2,606

Note: "Do not know" responses are included in the "other" category. "No answer" responses are omitted from the table.

responses over time in the various surveys taken between 1963 and 1990. Between 1963 and 1986, the proportion of women who responded "good custom" declined, and the proportion who responded "natural duty" increased. The combined percentage responding "good custom" or "natural duty" remained approximately constant at about 75% to 80%. The proportions responding "unavoidable" and "not a good custom" remained low and stable. A major shift in pattern occurred after 1986. Between 1986 and 1990, the proportion responding "good custom" remained stable at about 20%, but the proportion responding "natural duty" fell steeply from 58% to 30%. The combined percentage for these two categories, which previously had changed little between 1963 and 1986, fell from 75% to 50%. After 1986, there were compensating increases in the proportions in the remaining categories. The proportion responding "unavoidable" rose from 7% to 22%, and the proportion responding "not a good custom" rose from 4% to 12%.

These changes suggest a two-stage normative shift: first a gradual shift from "good custom" to "natural duty," and second a sudden shift from "natural duty" to "unavoidable" and "not a good custom." In the second shift, the percentage responding "unavoidable" increased much more than the percentage responding "not a good custom." If we combine "good custom" and "natural duty" into a single "traditional" category, and "unavoidable" and "not a good custom" into a single "modern" category, it is evident that norms remained highly traditional between 1963 and 1986, and that a major shift from traditional to


modem occurred after 1986. The rapid shift from traditional to modem (pri-

marily from "natural duty" to "unavoidable") appears to have been triggered by highly publicized major cutbacks in social security benefits, reflecting government concern about rapid population aging and burgeoning costs of the social security system (Martin, 1989; Ogawa, 1989). In the 1970s there was little governmental or public awareness of population aging. Interest in the subject picked up in the 1980s. The country's first conference on this topic was held at Nihon University in 1982. At about this time the problem of population aging began to receive attention in the mass media. Also in 1982 the report of the top-level Long-Term Outlook Committee of the government's Economic Planning Agency listed population aging, along with internationalization and matura- tion of the economy, as one of the three major challenges for 21st-century Japan (Martin, 1989).

In 1983, the government abolished the free medical service program for those aged 70 and over after 10 years of existence, in order to rein in increasing costs. Since 1984, all insured persons must cover 10% of their medical costs. In 1986, pension benefits were drastically reduced, by about one-third for those not grandfathered under the earlier rules. A series of White Papers issued by the Ministry of Health and Welfare, starting in 1985, addressed the question of the optimal blend of public and family support for the elderly. In 1986, a policy statement on the National Long- Term Program, relating to the so-called "society of longevity," was adopted by the Cabinet. This policy statement was especially important because it triggered a series of policies and programs related to the society of longevity at both national and local levels. The concept of the society of longevity was further elaborated in various government documents published in subsequent years. For example, the 1986 and 1987 White Papers on Health and Family Welfare advocated a gradual shift from public support to home care for the elderly. And, in 1989, the government an- nounced its "Golden Plan" to promote home care for the elderly by expanding social services such as home helpers and respite care facilities (Maeda, 1990). Concomitant with these government pronouncements and policy changes, public awareness of the problem of population aging increased tremendously over the decade, fueled by gloomy media coverage.

The precise mechanism by which the highly publicized cutbacks in social security and the re-

duced role of the government in providing care to the elderly triggered a shift in opinion about filial care from "natural duty" to "unavoidable" is not entirely clear. Perhaps the "natural duty" response became less palatable when it became apparent that the reality of this natural duty had suddenly become considerably more burdensome.

These normative changes, including the occur- rence of a sudden shift after 1986, are consistent with our earlier hypotheses about the effects of modernization on norms of care for the elderly. Because the pace of socioeconomic and demographic change has been extremely rapid, so that normative change tends to lag behind socioeconomic and demographic change, and because the Japanese population is highly homogeneous and socially integrated in the sense described earlier, the phenomenon of sudden onset and rapid diffusion of normative change is more likely to be observed in Japan than in most other national populations.

Expectations of Old-Age Supportfrom Children

Since 1950, the Mainichi surveys have also included a question about women's expectations of old-age support from their children: "Are you planning to depend on your children in your old age (including adopted children, if any)?" (original question in Japanese). Again, the concept of

TABLE 2. DISTRIBUTION OF RESPONSES ABOUT

OLD-AGE DEPENDENCE ON CHILDREN (BY PERCENTAGE)

Response Categories Number

Never Thought Do Not of Year Expect About It Expect Cases 1950 65 16 20 2,404 1952 51 18 31 2,754 1955 54 22 24 2,595 1957 53 19 28 2,693 1959 48 20 32 2,604 1961 43 25 32 2,441 1963 35 16 50 2,751 1965 36 16 48 2,817 1967 33 17 50 2,781 1969 30 18 52 2,848 1971 26 19 55 2,888 1973 27 20 54 2,767 1975 27 22 52 2,705 1977 27 21 53 2,648 1979 25 20 56 2,651 1981 1984 1986 1988 19 22 59 2,356 1990 18 20 62 2,319

Note: In the 1981 survey the question was not asked, and in the 1984 and 1986 surveys the question was asked but with incomparable response categories.


care or dependence is broad and is not restricted to coresidence. The response categories are expect to depend on children, do not expect to depend on children, and never thought about it. Before the seventh survey round in 1963, there was one additional response category, "would like to depend on children but does not seem possible," which we have grouped with "do not expect to depend." In some surveys, this question was asked only of women with at least one child. To maintain comparability, we imposed this re- striction on all of the surveys when analyzing the data.

Between 1950 and 1990, the proportion of respondents who expected to depend on their children declined from 65% to 18% (see Table 2). For the most part this decline was rather gradual. The proportion who did not expect to depend on their children increased from 20% to 62%. This increase was also gradual, except for a sharp jump between 1961 and 1963. The jump was probably due mainly to the establishment of universal pension and health care systems at that time. The proportion responding "never thought about it" remained fairly stable at about 20%.

Although the trend in the decreasing proportion responding "expect to depend" in Table 2 coin-

cides fairly closely with the decreasing trend in the percentage responding "good custom" in Table 1, it is not true that those who responded "good custom" to the normative question invariably responded "expect to depend" to the expectation question. In a supplementary tabulation (not shown), the proportions of those responding "good custom" who also responded "do not expect" were 46% in 1963 and 52% in 1990. The correlation between the responses "good custom" and "expect to depend" ranges from 0.05 in the 1965 survey to 0.15 in the 1979 survey. To the extent that expectations of old-age support from one's own children are a harbinger of the children's norms of filial obligations to their elderly parents, these figures suggest that the proportion in the "good custom" category could decrease further in the future.

Overall, these results indicate a marked decline in the proportion of women who expect to depend on their children in their old age, consistent with our earlier hypotheses about the effects of modernization on expectations about old-age support from children. There is no evidence of a lag in adjustment to underlying socioeconomic and demographic conditions, a point to which we shall return later.

TABLE 3. DISTRIBUTION OF THE SAMPLES ON THE PREDICTOR VARIABLES USED IN THE LOGIT REGRESSION ANALYSIS OF NORMS ABOUT CARING FOR ELDERLY PARENTS (BY PERCENTAGE)

Year and Survey Number 1963 1965 1971 1975 1977 1979 1981 1986 1988 1990

Predictor Variables (7th) (8th) (1 lth) (13th) (14th) (15th) (16th) (18th) (19th) (20th) Wife's age

20-29 25 26 22 23 22 21 18 15 13 14 30-39 44 44 39 42 42 44 43 45 45 39 45-49 31 30 39 35 36 35 39 41 42 47

Wife's education Junior high school 55 55 48 37 37 32 30 19 15 13 Senior high school 41 41 46 53 52 57 53 56 56 56 Junior college or other 4 4 6 10 11 11 17 25 30 32

Husband's occupation Farmer 18 18 13 8 9 9 4 4 2 2 Self-employed 20 18 20 19 19 20 22 22 18 17 Blue-collar 20 24 22 26 23 26 26 26 28 26 White-collar 41 41 45 47 49 45 48 49 53 56

Current residence Urban 58 62 69 72 75 76 79 83 84 80 Rural 42 38 31 28 25 24 21 17 16 20

Currently living with parent(s) Yes 39 38 35 30 29 32 32 32 29 33 No 61 62 65 70 71 68 68 68 71 67

Annual household income Low 39 23 33 36 22 35 31 49 31 24 Middle 42 50 43 40 44 39 27 32 36 41 High 19 27 24 24 34 26 42 20 33 34

Number of cases 2,632 2,722 2,801 2,477 2,519 2,473 2,583 2,209 2,078 2,217 Note: The samples used for the logit regression analysis of expectations of old-age support from one's own children are

somewhat smaller (because the analysis of expectations was restricted to women with at least one child) and yield slightly different percentages. The differences between the two sets of percentages are too small to justify inclusion of two separate tables of percentages.


DETERMINANTS OF NORMS AND EXPECTATIONS OF CARE FOR THE ELDERLY

To augment the trend analysis, we also undertook an analysis of the determinants of norms and expectations regarding care of the elderly. To sim- plify the analysis, we dichotomized the response variables. The response variable for norms of filial support of elderly parents was coded 1 if "good custom" or "natural duty" and 0 otherwise. The response variable for expectations of old-age support from children was coded 1 if "expect" and 0 otherwise.

An appropriate method for analyzing the determinants of a dichotomized response variable is logit regression. The determinants, or predictor variables, included in our logit regressions are wife's age, wife's education, husband's occupation, urban-rural residence, whether or not the woman is currently living with her own or her husband's parents (excluded from the analysis of norms because of simultaneity, or two-way causa- tion), and annual household income. Wife's age is included because one's views about care of the elderly are likely to change as one approaches old age. The remaining predictor variables (wife's education, husband's occupation, etc.) are standard indicators of level of modernization, which, ac-

cording to the theory sketched earlier, affects norms and expectations about care of the elderly. The set of predictor variables does not represent a definitive list of determinants but instead reflects what is available from the Mainichi surveys. All of the predictor variables are categorized and represented by dummy variables in the regressions.

Annual household income is treated different- ly from the other variables, in that the income cutting points that define income categories change from one survey to the next. Cutting points were chosen in order to retain roughly equal numbers of cases in each of the three income categories. (Precisely equal numbers were not possible because of precoded responses.) Had this not been done, then, for example, the "low" category as defined in 1963 would have been virtually empty in 1990, because of the enormous increases in in- comes during the intervening 27 years.

The distributions of the samples on each predictor variable are shown in Table 3. Over suc- cessive surveys, the sample population became older, more educated, more urban, less likely to be living with parents, and more likely to have a white collar husband than a farmer husband. The trends in percentages in each household income

TABLE 4. ADJUSTED PERCENTAGES OF RESPONDENTS WHO CONSIDER THAT FILIAL CARE OF ELDERLY PARENTS IS EITHER A "GOOD CUSTOM" OR "NATURAL DUTY"

Year and Survey Number 1963 1965 1971 1975 1977 1979 1981 1986 1988 1990

Predictor Variables (7th) (8th) ( lth) (13th) (14th) (15th) (16th) (18th) (19th) (20th) Wife's age

20-29 72** 69** 70* 68** 70** 67** 78** 79 67 49 30-39 77 79 70** 70** 74** 71" 81* 77 61** 49 45-49a 80 83 76 81 81 76 85 76 69 50

Wife's education Junior high schoola 81 83 73 77 75 72 78 74 65 59 Senior high school 79 81 74 77 76 72 82* 77 67 49** Junior college or other 75 77 71 72 79 74 85* 78 64 46**

Husband's occupation Farmera 86 84 79 79 77 81 83 81 78 50 Self-employed 83 84 79 76 80 77 84 81 65 57 Blue-collar 78** 78 73 74 77 69** 81 74 65 45 White-collar 76** 79 71* 76 76 73* 83 77 66 49

Current residence Urban 78 80 71** 74** 76** 72 81** 76 64** 49 Rurala 76 81 79 80 82 75 89 81 73 51

Annual household income Lowa 77 82 69 73 77 74 82 75 65 50 Middle 77 79 72 75 77 74 83 77 67 52 High 79** 79 77** 77 77 70 83 78 65 46

Total Adjusted percentageb 78 80 73 75 77 73 83 77 66 49 Observed percentage 80 80 74 74 75 72 81 76 65 49

Number of cases 2,632 2,722 2,801 2,477 2,519 2,473 2,583 2,209 2,078 2,217 aReference category for each variable in the underlying logit regression. bObtained by setting all predictor variables at their means in 1990. *.01 < p < .05. **p ? .01.


category are meaningless because of the changing cutting points that define income categories.

Table 4 shows adjusted percentages of women who responded that filial care of elderly parents is either a "good custom" or "natural duty." These adjusted percentages were derived from underlying logit regressions (one for each survey). The nature of the adjustment is that, in each panel of percentages corresponding to a particular predictor variable, all other predictor variables (i.e., the dummy variables representing the categories of those predictor variables) are held constant at their mean values in the 1990 sample. Although separate logit regressions were run for each survey, control variables are always held constant at their means in the 1990 survey in order to increase comparability of results for the various survey years. The variable "currently living with parents" is excluded from these regressions because of problems of simultaneity (two-way cau- sation).

The adjusted percentages by age were fairly constant over time until 1986, but then fell sharply, consistent with the simple trend data shown earlier in Table 1. Cross-sectionally, age had a statistically significant effect on the adjusted percentage until 1986, with younger women somewhat less likely than older women to view filial care favorably. After 1986, age has no significant effect on this adjusted percentage. Not only did the adjusted percentages fall sharply after 1986, but also the differences among them disappeared.

Before 1981, wife's education had little effect on norms about filial support of the elderly. In the early surveys, those with more education were slightly less likely to view filial care favorably, but the differences by education were not statistically significant. In the 14th (1977) through 18th (1986) surveys the differential reversed, with those with more education viewing filial care slightly more favorably, but again the differences by education were not statistically significant except in 1981. After 1986, the adjusted percentage responding "good custom" or "natural duty" for those with high school or more education fell especially rapidly, so that by 1990 the percentages for women with this level of education were sig- nificantly lower than those with less than a high school education. These findings suggest that more educated persons were in the vanguard of the normative shift that appears to have occurred after 1986, although not by very much. In the typical diffusion pattern, opinion leaders and trend

setters tend to have higher socioeconomic status, but, in culturally homogeneous Japan, norms are evidently shared to the point that followers are not far behind the trend setters.

The adjusted percentages by husband's occupation show that women with husbands with blue collar or white collar occupations usually favored filial care somewhat less than those with husbands who were farmers or self-employed, but in most instances the differences by occupation were not statistically significant. None of the differences were statistically significant after 1979. The adjusted percentage who viewed filial care favorably declined sharply after 1986 in all occupa- tional categories.

Urban women were somewhat less likely than rural women to view filial care favorably, and this urban-rural difference is statistically significant in about half the surveys. The differences are rather small, however, and virtually nonexistent in the 1990 survey. Income has no consistent effect.

The findings that emerge from Table 4 are that the effects of the predictor variables are small cross-sectionally, and the percentages who view filial care favorably declined sharply after 1986 in all categories of all predictor variables. These findings accord with our hypothesis that, in a homogeneous country like Japan where norms tend to be shared throughout the population, major normative shifts can occur and, when they do occur, they often do so across the board. This does not mean that the predictor variables in Table 4 are without effect, only that the effects are observed longitudinally. The effects tend not to show up much in cross-sectional observations of the population.

The lack of substantial cross-sectional effects of the predictor variables explains why the trend in the adjusted percentage obtained by setting all predictor variables at their means in 1990 hardly differs from the trend in the observed unadjusted percentage (see the two rows labeled "Total" in Table 4). The lack of cross-sectional effects im- plies that compositional changes in the samples, with respect to categories of the predictor variables, have virtually no effect on the trend in the percentage who favorably view filial support of the elderly, except insofar as the compositional changes contribute to building latent receptivity to a population-wide normative shift. This is also true in another set of logit regressions (not shown) where the dependent variable is 1 if "good custom" and 0 otherwise. Although the adjusted percentages decline more gradually in this


TABLE 5. ADJUSTED PERCENTAGES OF RESPONDENTS WHO EXPECT TO DEPEND ON THEIR CHILDREN IN THEIR OLD AGE

Year and Survey Number 1963 1965 1971 1975 1977 1979 1988 1990

Predictor Variables (7th) (8th) (11th) (13th) (14th) (15th) (19th) (20th) Wife's age

20-29 9** 12** 1l** 7** 9** 9** 15" 12** 30-39 16** 17** 9** 15* 15* 14** 14** 14** 40-49a 22 26 20 26 31 25 22 20

Wife's education Junior high schoola 29 37 25 29 28 30 20 20 Senior high school 19** 22** 14** 20** 22** 18** 18 17 Junior college or higher 11** 13"* 10** 12** 15** 14** 15 14"

Husband's occupation Farmera 43 37 26 29 39 28 34 25 Self-employed 23** 26** 16** 21 24** 19" 21 18 Blue-collar 26** 27** 15"* 18** 24** 19* 19* 18 White-collar 13** 16** 13** 17** 17** 16** 15* 15

Current residence Urban 16** 19** 12** 16** 19** 16** 15* 15* Rurala 22 25 23 29 25 27 26 23

Currently living with parent(s) Yes 18 23* 15 21** 24* 21** 22** 18 Noa 17 19 13 17 19 16 15 16

Annual household income Lowa 23 25 15 21 24 19 20 18 Middle 18** 19** 13 17* 20 18 18 18 High 14** 18** 14 17 18" 16 15 13*

Total Overall expected proportionb 17 20 14 18 20 18 17 16 Observed percentage 33 35 25 25 25 23 18 17

Number of cases 2,361 2,505 2,568 2,279 2,207 2,026 1,910 2,013 aReference category for each variable in the underlying logit regression. bBased on means observed in 1990. *.01 <<p<.05 **p?<.01.

case, they also decline across the board, and compositional changes have virtually no effect on the trend.

The determinants of expectations of old-age support from children are somewhat different (Table 5). There is a strong cross-sectional effect of age on expectation of support, but this effect does not change appreciably over time. The lack of change indicates that the effect of age is a life- cycle effect, whereby women are more likely to expect old-age support the closer they get to old age. Within age categories, the percentages hardly change over time, so there is no longitudinal effect. Education has a sharp negative effect on expectation of support, but, again, the adjusted percentages show little decline over time, except for those with least education (junior high school only). Husband's occupation also has a substantial cross-sectional effect. Women with husbands who are farmers have the highest expectations of support, followed by self-employed and blue collar, followed by white collar. The differences by husband's occupation tend to diminish over time and by 1990 are no longer statistically significant.

Among the remaining predictor variables, urban women have substantially lower expectations of support than rural women, but the adjust-

ed percentages do not change greatly over time. Women currently living with parents have somewhat higher expectations of support than women not living with parents, but the adjusted percentages do not change much over time. Household income has a negative effect on expectation of support. Again the adjusted percentages change only slightly over time, but it must be remem- bered that the cutting points that define income categories are not fixed, so that trends in these percentages are not very informative.

The overall picture that emerges from Table 5 is that the predictor variables have fairly substantial cross-sectional effects, but the adjusted category-specific percentages who expect old-age support from their children change little, if at all, over time. This means that the substantial decline in the overall observed percentage who expect support (see the next-to-last row in Table 5) is due almost entirely to compositional changes, insofar as, over time, the population became older, more educated, more white collar, more urban, and less likely to be living with parents. Indeed, when composition is held constant by setting all predictor variables at their means in 1990, the overall percentage of women who expect support hardly changes.

There is no indication in Table 5 of any lags or


sudden shifts in expectations. The table suggests that, when a woman moves into a new socioeconomic category, she quickly adjusts her expectations to coincide with the expectations that are typical of that category. The table therefore suggests that expectations, in contrast to norms, tend to adapt continuously to changes in underlying socioeconomic and demographic conditions, in ac- cordance with our initial hypothesis.

CONCLUSION

During the past 40 years, Japan has undergone rapid socioeconomic development, which has gone hand in hand with a demographic transformation accompanied by rapid population aging. A number of previous studies have focused on related changes in patterns of coresidence of adult children living with their elderly parents. In this study we have analyzed trends and determinants of concomitant changes in norms and expectations of care for the elderly. In doing so we have made use of a remarkable and unique data set.

The analysis of normative change indicates the presence of a type of cultural lag, whereby changes in norms of filial care for elderly parents have lagged behind changes in underlying socioeconomic and demographic conditions. In homogeneous, highly integrated societies such as Japan, these norms tend to be widely shared. When they change with a lag, latent receptivity to change can build up in the population, with the result that normative change may proceed swiftly once it begins. This is precisely the pattern observed in the Mainichi survey data, which show little change in norms up to 1986, followed by a sharp, population-wide decline in the percentage who view filial support of elderly parents as a good custom or natural duty. This normative shift appears to have been precipitated by the rather sudden emergence of government concern about the problem of population aging, by concomitant policy changes aimed at containing rapidly rising social security costs, and by extensive mass media coverage of these developments.

Although our analysis assumes that norms of filial care for elderly parents are fundamentally anchored in underlying socioeconomic and demographic conditions, it should not be inferred that norms have no influence on behavior. Once norms of filial care for the elderly shift, it seems reasonable to expect that women will become psychologically more comfortable with depar- tures from culturally prescribed stem family liv-

ing arrangements, so that there will be feedback effects on coresidence patterns. In this article, however, we have not investigated this feedback.

The findings about expectations of old-age support from children show a different pattern. Expectations of old-age support have declined steadily over time and show little evidence of lags or sudden shifts. Within categories of socioeconomic and demographic predictor variables, the adjusted percentages expecting old-age support change little over time, consistent with the hypothesis that expectations of old-age support from children adjust continuously to changes in underlying socioeconomic and demographic conditions. The large decline in the overall observed percentage expecting old-age support occurs because of compositional shifts of population into socioeconomic categories characterized by lower expectations of old-age support from children.

A policy implication of our findings is that current government efforts to shift more of the burden of care of the elderly from government back to families may not be very successful. Further doubts about the success of these efforts are raised by recently reported data showing that a substantial majority of Japanese people prefer a high-cost/high-welfare social security system over a low-cost/low-welfare system. According to these data, the proportion favoring a high- cost/high-welfare system was 56% in 1978 and 57% in 1992 (Opinion Survey Department, 1992). If the public proves willing to assume the additional tax burden necessary to finance a high- cost system, the government may ultimately opt for such a system.

It should be borne in mind that our data and conclusions are based on survey responses from women. A similar analysis for Japanese men (not possible with our data) might look quite different, because elderly Japanese men are more likely to depend on their wives for care, whereas women, who live longer than men and are more likely to be widowed, must rely more on children. Moreover, because of the dominant patrilocal residence pattern, the elderly parents who require care are normally the husband's parents.

NOTE

We thank the Population Problems Research Council of the Mainichi Newspapers of Japan for making available data from the various rounds of the National Survey on Family Planning. We also thank Hal Kendig, Kazuko Koshitani, Toshio Kuroda, Peter Laslett, Linda G. Martin, Andrew Mason, Michio Ozaki, and William H. Sewell for helpful comments, and Rikiya Matsukura


and Hideyuki Kujirai for computer programming assis- tance. This work was partially supported by a research grant from AFLAC Japan (American Family Life Assurance Company of Japan).

REFERENCES

Campbell, R., & Brody, E. M. (1985). Women's changing roles and help to the elderly: Attitudes of women in the United States and Japan. The Gerontologist, 25, 584-592.

Clark, R. (1990). Japanese retirement systems. Home- wood, IL: Irwin.

Cowgill, D. O., & Holmes, L. D. (Eds.). (1972). Aging and modernization. New York: Appleton-Century-Croft.

Coale, A. J., & Demeny, P. (1983). Regional model life tables and stable populations (2nd ed.). New York: Academic Press.

Davis, K., & Combs, J. W. (1950). The sociology of an aging population. In Proceedings of the Eastern States Health Education Conference, March 31-April 1, 1949: The Social and Biological Challenge of Our Aging Population (pp. 146-170). New York: Columbia University Press.

Davis, K., & van den Oever, P. (1981). Age relations and public policy in industrial societies. Population and Development Review, 7, 1-18.

Gibson, M. J. (1992). Public health and social policy. In H. Kendig, A. Hashimoto, & L. Coppard (Eds.), Family support to the elderly: The international experience (pp. 88-114). Oxford: Oxford University Press.

Health and Welfare Statistics Association. (1988). Kokumin eisei no doko 1988 nen [Trends of National Health for 1988]. Tokyo: Author.

Hodge, R. W., & Ogawa, N. (1991). Fertility change in contemporary Japan. Chicago: University of Chicago Press.

Kendig, H. (1989). Social change andfamily dependency in old age: Perceptions of Japanese women in middle age (NUPRI Research Paper Series No. 54). Tokyo: Nihon University Population Research Institute.

Kojima, H. (1989). Intergenerational household exten- sion in Japan. In F.K. Goldscheider & C. Goldscheider (Eds.), Ethnicity and the new family economy: Living arrangements and intergenerational financial flows (pp. 163-184). Boulder, CO: Westview Press.

Kuroda, T., & Hauser, P. M. (1981). Aging of the population of Japan and its policy implications (NUPRI Research Paper Series No. 1). Tokyo: Nihon University Population Research Institute.

Lee, G. R. (1984). Status of the elderly: Economic and familial antecedents. Journal of Marriage and the Family, 46, 267-275.

Maeda, D. (1990, October). Declining family care vs. expanding role of public services: Study of social and legal aspects of the Japanese experiences. Paper presented at the United Nations International Conference on Aging Populations in the Context of the Family, Kitakyushu, Japan.

Management and Coordination Agency. (1991). Rojin no seikatsu to ishiki: Dai sankai kokusai hikaku

chosa kekka hokokusho [Life and perceptions of the elderly: Report of the results on the third international comparative survey]. Tokyo: Chuo Hoke Shuppan.

Martin, L. G. (1989). The graying of Japan. Population Bulletin, 44, 1-42.

Martin, L. G., & Tsuya, N. 0. (1991). Interactions of middle-aged Japanese with their parents. Population Studies, 45, 299-311.

Minami, R. (1986). The economic development of Japan: A quantitative study. London: MacMillan Press.

Ministry of Education. (1991). Statistical bulletin of education. Tokyo: Daiichi Hoki Shupan.

Ministry of Health and Welfare. (1991). Latest demographic statistics 1990-1991. Tokyo: Institute of Population Problems.

Morgan, S. P., & Hirosima, K. (1983). The persistence of extended family residence in Japan: Anachronism or alternative strategy? American Sociological Review, 48, 269-281.

Nakata, Y., & Mosk, C. (1987). The demand for college education in postwar Japan. Journal of Human Re- sources, 23, 377-404.

Ogawa, N. (1986). Internal migration in Japanese postwar development (NUPRI Research Paper Series No. 33). Tokyo: Nihon University Population Research Institute.

Ogawa, N. (1987). Sex differentials in labour force participation and earnings in Japan. In United Nations ESCAP (Ed.), Women's Economic Participation in Asia and the Pacific (pp. 305-332). Bangkok: United Nations ESCAP.

Ogawa, N. (1989). Population aging and its impact upon health resource requirements at government and familial levels in Japan. Ageing and Society, 9, 383-405.

Ogawa, N., & Ermisch, J. F. (1991). Family structure, home time demands and the employment patterns of Japanese married women. Unpublished manuscript.

Ogawa, N., & Hodge, R. W. (in press). Patrilocality, childbearing, and labor supply and earning power of married Japanese women. In J. F. Ermisch & N. Ogawa (Eds.), The family, the market and the state in aging societies. Oxford: Clarendon Press.

Opinion Survey Department. (1993, September 27). Survey on the aging society: A report. Mainichi Newspaper (morning edition), pp. 14-15.

Otani, K., & Atoh, M. (1988). The social consequences of rapid fertility decline in Japan. Paper presented at the IUSSP Seminar on Fertility Transition in Asia: Diversity and Change, Bangkok.

Preston, S. H., & Kono, S. (1988). Trends in well-being of children and the elderly in Japan. In J. L. Palmer, T. Smeeding, & B. Torrey (Eds.), The vulnerable (pp. 277-307). Washington, DC: Urban Institute Press.

Retherford, R. D. (1985). A theory of marital fertility transition. Population Studies, 39, 249-268.

Retherford, R. D. (1987). An integration of two fertility analysis frameworks. Genus, 43, 1-18.

Retherford, R. D., & Palmore, J. A. (1983). Diffusion processes affecting fertility regulation. In R. A. Bulatao & R. D. Lee (Eds.), Determinants offertili- ty in developing countries (Vol. 2, pp. 295-339). New York: Academic Press.


Sakamoto, Y. (1975). Study of the Japanese national character [English summary]. Annals of the Institute of Statistical Mathematics, Part 5, Supplement No. 8. Tokyo.

Sano, C. (1958). Changing Values of the Japanese Family. Westport, CT: Greenwood Press.

Shimada, H., & Higuchi, Y. (1985). An analysis of trends in female labor force participation in Japan. Journal of Labor Economics, 3, 355-374.

Social Insurance Agency. (1991). Outline of Social In-

surance in Japan 1990. Tokyo: Japan International Social Security Association.

Statistics Bureau. (1992). 1990 Population Census of Japan. Tokyo: Japan Statistical Association.

United Nations. (1991). World Population Prospects 1990. New York: Author.

U.S. Bureau of the Census. (1990). Statistical Abstract of the United States 1990. Washington, DC: U.S. Government Printing Office.

A comprehensive collection

of family issues research.

Contemporary Families: Looking Forward, Looking Back Edited by Alan Booth Penn State University

This publication contains outstanding research on family issues in the 1980s with trends for the next decade. Authoritative information on families that is a must resource for you.

Order Today! Only $23.95 for non-members $19.95 for members and orders of 10 or more. Price includes postage and handling for U.S. orders only. Canadian orders must add 7% GST (#R-123-830-465). Foreign orders add $2 per book.

classroom teiixi .../. . . . . . . ......... . . ....... . . . ....... ... . .

k .../.

"A valuable resource for family studies. It brings together the work of leading scholars and presents impressive testi- mony about the progress of family studies today."

Graham B. Spanier Univ. of NE-Lincoln

0

,,r l,

National Council on Family Relations 3989 Central Ave. N.E. #550 Minneapolis, MN 55421 Phone: 612/781-9331 FAX: 612/781-9348

care of the elderly in japan: changing norms and expectations · changing norms and expectations...

Documents