This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire

Intergenerational Coresidence of Older Adults in Japan: Conditions for Cultural Plasticity

Andrus Gerontology Center, University of Southern California, Los Angeles.

Address correspondence to Emiko Takagi, Andrus Gerontology Center, University of Southern California, 3715 McClintock Avenue, Los Angeles, CA 90089. E-mail: takagi{at}usc.edu

	Abstract

TOP Abstract Methods Results Discussion References

 
Objectives. This study investigated individual-level conditions and prefecture-level contextual factors that enable and/or restrict intergenerational coresidence arrangements between older parents and adult children.

Methods. We implemented multinomial logistic regression within a hierarchical approach to compare traditional and nontraditional styles of intergenerational coresidence arrangements. Individual data (N = 3,565) came from the Nihon University Japanese Longitudinal Study of Aging. We supplemented these with information on socioeconomic and welfare characteristics of Japanese prefectures.

Results. Whereas the traditional pattern of coresidence was primarily a value-driven arrangement, nontraditional coresidence was both a value-driven and a need-driven arrangement for older parents with limited physical abilities. Relatively strong economic conditions at the prefecture level were enabling factors for coresidence, whereas the greater per capita utilization of local welfare resources had both enabling and inhibiting effects on coresidence arrangements.

Discussion. The differences between the two types of coresidence are consistent with a shift of intergenerational living arrangements in Japan from a preventive arrangement to a contingent arrangement for older parents in need. We suggest that intergenerational family traditions contain a great amount of plasticity to accommodate societal modernization by adapting to the changing cultural and socioeconomic contexts of the society.

JAPAN is one of the very few developed countries that maintains a relatively high rate of coresidence between older parents and adult children. In Japan, 58% of people 60 years of age and older lived with at least one of their children in 2001, a figure 3 to 10 times greater than that found in comparably developed Western societies such as the United States (17%), Germany (15%), and Sweden (5%; Cabinet Office, 2001). Researchers have viewed the traditional form of intergenerational coresidence in Japan—whereby older parents reside with their eldest son and his family—as a demonstration of filial piety, an ideology that requires children to care for aging parents (Koyano, 1996). However, both the rate of intergenerational coresidence and the belief in filial piety are in decline. The percentage of multigenerational households has been halved over the past two decades, dropping from 50% to 24% (Cabinet Office, 2005). Studies have also indicated increasingly negative attitudes toward the norm of filial piety in Japanese society (e.g., Koyano, 1996, 2003; Ogawa & Retherford, 1993). However, Japan remains an anomalous case within the community of economically advanced countries in that expectations for being in extended-family households still predominate (Maeda & Ishikawa, 2000).

Many scholars have argued that the norm of filial piety, although ebbing, still remains an important consideration in living arrangements. They have also suggested that the social and economic changes accompanying the rapid postwar modernization of Japan have rendered multigenerational households impractical for many families (Kono, 2000; Ogawa & Retherford, 1993; Sodei, 1995). How is a persistent cultural norm favoring intergenerational coresidence reconciled with structural pressures that have led to its decline? Historical evidence suggests that intergenerational coresidence in Japan has begun to shift from a normatively driven "preventive" approach with advance preparation for expected age-related needs of older parents (Hashimoto, 1996) to a "contingent" approach, in which older parents and adult children become reunited after the need arises (J. W. Brown et al., 2002). With this contingent approach, families modify, but do not violate, the tradition of filial piety by delaying the timing of coresidence (Kawabe & Shimizu, 1994; Kono, 2000). This shift presumably diversifies intergenerational coresidence arrangements to include both long-term traditional and intermittent nontraditional household forms.

Rates of intergenerational coresidence, however defined, also differ across geographical areas of Japan. Coresidence tends to be more prevalent in the northeast region than in the southwest region of Japan, reflecting the divergent regional cultures of coresidence (Ministry of Health, Labour and Welfare, 2000; Shimizu, 2004). Multigenerational households tend to be more common in the more traditional and less developed areas, which include areas with worse socioeconomic and housing conditions and fewer welfare resources to support older residents in need (Izuhara, 2000; Shimizu, 2004).

Given regional diversity in the prevalence of intergenerational coresidence and variation in the motivations for intergenerational coresidence (protective and reactive), we contend that drawing conclusions based on gross national statistics about coresidence rates and cultural norms in Japan risks overgeneralizing about a nation that has undergone numerous social changes over the past half century. With this in mind, we examined how the larger socioeconomic, cultural, and welfare contexts shape coresidence patterns in contemporary Japanese society and determine the strength with which filial values are expressed in the practice of intergenerational coresidence. Our analysis of a nationally representative sample of older individuals in Japan included potential predictors of intergenerational coresidence at both the individual and regional levels. In particular, we studied the traditional Japanese arrangement in which coresidence with children is life long, as well as a relatively new type of intergenerational living arrangement in Japan (i.e., the reunification of a child with parent in a shared household later in life in response to family contingencies; Cherlin, 1994; Martin & Tsuya, 1994).

Diversity of Coresidence Across Individuals
In almost all nations, intergenerational coresidence is associated with elevated need in the older generation, particularly poor health status (N. Brown, 2003; Logan & Bian, 1999; Raymo & Kaneda, 2003) and lack of a spouse due to widowhood (Martin & Tsuya, 1994; Raymo & Kaneda, 2003). Kin availability also plays a role in determining whether parents and their adult children coreside. In a strong patrilineal kinship system, such as that found in Japan, older parents who have at least one son are more likely to coreside with a child than those who have no sons (Takagi & Silverstein, 2006). Because a possible alternative to coresidence is an arrangement whereby older parents and adult children live separately but close to one another (Martin & Tsuya, 1994), the availability of adult children with close geographical proximity may reduce the need for, and hence decrease the likelihood of, coresidence. Housing assets of older people are positively associated with sharing a residence with children (Logan & Bian, 1999; Raymo & Kaneda, 2003), presumably because such resources enable older parents to accommodate the housing needs of their adult children.

Coresidence is also related to older parents' cultural beliefs about the desirability of coresidence. Older parents who more strongly embrace traditional family values are more likely to coreside with their children than are those who hold less traditional views (Kurosu, 1994; Takagi & Silverstein, 2006). The educational attainment of parents and children also indirectly influence normative attitudes. In general, parents and children with higher levels of educational attainment tend to express more negative attitudes toward traditional family forms and are less likely to live together than those with less education (Logan & Bian, 1999; Rindfuss, Liao, & Tsuya, 1992; Takagi & Silverstein, 2006).

In terms of demographic characteristics, parents of advanced age are more likely to live with their children (Cabinet Office, 2004). In Japan, parental gender has not been consistently related to coresidence, but some studies have indicated that older men are more likely to live with their children than are older women (e.g., Tsuya & Martin, 1992). Older individuals in rural areas also tend to coreside with their children presumably because of the generally more traditional family culture in rural communities (Raymo & Kaneda, 2003; Tsuya & Martin, 1992).

Diversity of Coresidence Across Regions
Studies have argued that family arrangements in Japan vary across not only individual family circumstances but also the social contexts in which they are embedded, including local housing and economic conditions, cultural climate, and accessibility of social welfare resources (Izuhara, 2000, 2004; Kumagai, 1996, 2005; Kurosu, 1994). Few studies, however, have clarified the impact of this diversity across the national level.

Observed differences in intergenerational coresidence across regions of Japan appear to be related to socioeconomic contexts such as economic well-being and housing conditions. Coresidence tends to be less prevalent in areas with fewer economic opportunities because of limited availability of jobs for younger generations (e.g., Thompson, 2003); however, insufficiency of financial resources may encourage coresidence if adult children lack the financial means to maintain their own households. Alternatively, children may migrate to find jobs in more prosperous regions, suppressing rates of coresidence. Also, limited housing options and space restrict the ability to create or maintain intergenerational living arrangements, which generally require more space for two generations to share a household (Izuhara, 2000; Sodei, 1995).

Regional diversity in coresidence can also derive from the cultural climate of an area—the degree to which regional norms are restrictive or coercive. The cultural environment in the northeast region of Japan tends to embrace intergenerational coresidence as the ideal arrangement for older parents, whereas the cultural climates of the northern island (Hokkaid) and the southwest region encourage independent living for older parents and adult children (Shimizu, 2004). These regional differences tend to exist independently of rural–urban differences, suggesting a robustness of regional culture in coresidence decisions (Raymo & Kaneda, 2003).

Finally, the community-level availability of social welfare services for older people is another potential contextual factor that may influence intergenerational living arrangements in Japan. In response to declines of family size and weakening filial piety, there has been a steady increase in societal demand for publicly provided welfare resources for older Japanese people (Peng, 2003). In the past few decades, the Japanese government has undertaken initiatives to increase the resources supplied to the older population in order to achieve a better balance between private and public support for elders (Ishikawa & Maeda, 2000). A major political shift has been to delegate to the local community greater responsibility and authority for service management (Peng, 2003). As a result, the present social welfare system for older persons in Japan is quite variable across subnational units.

Study Purpose and Research Questions
Our study aimed to address three research questions related to variation in coresidence arrangements in the Japanese context. The first aim was to examine the differential determinants of two different forms of coresidence: the traditional form predominantly governed by cultural norms and nontraditional form made in response to exigent personal needs. Thus, to delineate the divergent functions of intergenerational living arrangements, we contrasted the traditional type of life-long coresidence, wherein older parents and adult children live together since the birth of the child, and the nontraditional type of "boomerang" coresidence, wherein parents and children resume living together after some time has passed. The second aim was to investigate variation in coresidence arrangements across regions of Japan that differ in the economic opportunities afforded the working population, housing space, utilization of public services for older adults, and the extent of traditionalism in the cultural milieu. The third aim was to study whether socioeconomic constraints, cultural milieu, public service, and housing availability modify the relationship between older individuals' expectations of filial piety and coresidence with a child.

We hypothesized that life-long coresidence would be more associated with traditional filial expectations of older parents, whereas the boomerang type of coresidence would be more related to the needs of older parents. In our analysis of regional variation in coresidence rates, we used prefectures as the geographical unit. We hypothesized that the likelihood of both types of intergenerational coresidence would be higher in prefectures with more economic and housing resources. We also hypothesized that life-long coresidence would be more prevalent in prefectures with traditional cultural climates and boomerang coresidence would be less prevalent in prefectures with welfare resources generous enough to substitute for family care. Furthermore, because the degree to which filial expectations become manifest in particular living arrangements may be conditioned on structural aspects of the social and economic environment and the wider cultural context within which individuals are embedded (Ogawa & Retherford, 1993), we hypothesized cross-level interactions between personal expectations and contextual characteristics. Filial expectations will be more strongly associated with coresidence in prefectures that better enable shared living arrangements, in a manner consistent with the above discussion with one exception: We expected that stronger filial expectations would enhance boomerang coresidence in the context of more generous welfare provisions that may serve as supplements to family care.

	METHODS

TOP Abstract Methods Results Discussion References

 
Data
For this analysis we employed data from the baseline sample of the Nihon University Japanese Longitudinal Study of Aging (NUJLSOA). The NUJLSOA is a national longitudinal study conducted by Nihon University that focuses on the health and social conditions of the population aged 65 years and older in Japan (University of Southern California/University of California, Los Angeles, Center on Biodemography and Population Health, 2004). The initial wave in 1999 consisted of 4,997 respondents, including 606 proxies, representing a response rate of 74.6%.

Our study focused on the subgroup of the NUJLSOA sample whose information was provided by self-report and who had at least one living child. From the 4,085 respondents who satisfied these selection criteria, we further excluded 520 respondents (12.7% of the selected subsample) who had missing values on relevant variables that could not be reasonably imputed, yielding an effective sample size of 3,565. Comparisons between our analytic sample and the full sample revealed a difference only with regard to health status, such that we found that our sample tended to be in better health than the full sample. This was not surprising given that the majority of respondents excluded from our analysis had severe health problems that had precluded them from answering survey questions. As a diagnostic for detecting sample selection bias, we applied Heckman's two-stage instrumental variable method (Heckman, 1979) and found that the risk of attrition had little effect on the dependent variables of interest. Nevertheless, readers should exercise caution before generalizing our findings to the entire older population of Japan.

Dependent Variable
We represented the outcome variable of this study using three categories, each signifying a particular intergenerational living arrangement. The first consisted of parents who lived separately from their children (non-coresident). The second category consisted of older parents who had lived with an adult child since the child's birth (life-long coresident). The third category consisted of older parents who had begun living with an adult child later in the child's life, representing a reunification (boomerang coresident). Although our data did not allow us to make this distinction, boomerang coresidence could include both older parents who had moved into their adult children's home as well as those whose children had moved in with them. We also note that a small number of parental households (n = 21) had both a child who had never left home and at least one who had previously lived independently and had returned. We classified this hybrid form of coresidence as a boomerang type of household because having at least one boomerang adult child makes this household arrangement nontraditional from the parent's point of view. However, in practical terms, assigning the group to one coresident category or the other was of little consequence to our findings.

Independent Variables: Individual Level
Individual-level predictors of coresidence reflected older parents' demographic characteristics, socioeconomic resources and needs, family composition and availability, and normative attitudes. We operationalized parental age as a continuous variable ranging from 65 to 99. We created dichotomous variables for gender of older parents (1 = female, 0 = male) and rural residency (1 = lives in a rural area, 0 = lives in an urban area).

Socioeconomic resources of older parents included home ownership coded as a dichotomous variable (1 = owns home, 0 = does not own home) and household income assessed with a 13-point ordinal scale ranging from 1 (less than 500,000 Japanese yen, or approximately $4,160 U.S.) to 13 (more than 15 million Japanese yen, or approximately $125,000 U.S.). Due to the substantial proportion of missing values (about 20% of the original data set), we imputed the household income variable within the stratum of parental age and gender using the hotdeck method, which replaces missing values based on other cases with no missing data. We caution that single imputation strategies generally bias standard errors downward, thus increasing the risk of Type I error. Older parents' health status was indicated by their self-reported ability to perform seven instrumental daily activities: preparing meals, leaving home to purchase necessary items, taking care of financial matters, using the telephone, doing light housework, taking a bus or train, and taking medication as prescribed. These were measured with a 4-point scale ranging from 1 (not difficult) to 4 (unable to perform). Due to skewness in this variable, we combined all seven items and created a dichotomous variable (1 = has difficulty in performing at least one item, 0 = no difficulty in any of the seven).

Three dichotomous variables represented family composition and child availability: widowhood (1 = widowed, 0 = not widowed), son availability (1 = has at least one son, 0 = has no son), and presence of child living nearby (1 = has at least one child living in the same town or city, 0 = no child nearby). We note that because of the minimal divorce rate of older people in Japan (3.5% in the 2000 Japanese Census and 1.7% of our sample) and our preference for making inferences about coresidence among widowed parents, we included the divorced/separated with married parents. We included number of children as a continuous variable.

Education of both older parents and adult children was measured on a 4-point scale: junior high school (1), high school (2), vocational school or junior college (3), and 4-year university or graduate school (4). In order to summarize educational achievement of multiple offspring, we calculated the average educational level for all living children of each older respondent. We measured older parents' attitudes toward filial responsibility by the extent of their agreement with the following statement: "A child should be expected to support and take care of his or her aged parents, as the child should feel a sense of gratitude to the parents for raising him/her." Responses were provided on a 4-point scale ranging from disagree (1) to agree (4).

Independent Variables: Prefecture Level
We used characteristics of prefectures to examine the impact of contextual factors on residential arrangements. Somewhat analogous to states in the United States, prefectures are subnational jurisdictions administered by elected governors. Japan consists of 47 prefectures, each containing smaller governmental units of cities, towns, and villages. In our analysis, prefectures reflected the social and economic environments to which their residents were regularly exposed. We investigated four characteristics of prefectures that we expected to influence the context in which intergenerational coresidence takes place: economic well-being, housing space, cultural milieu, and public welfare service utilization by aged individuals.

We assessed the economic well-being of each prefecture in 1999 using an index of its financial power (zaisei-ryoku shisu in Japanese), calculated as the ratio of total financial income to total financial demand (Development Bank of Japan, 2006). If the financial power index exceeds the value of 1, one can assume the prefecture to have more financial resources than it requires for its basic needs. This index represents the amount of tax income relative to outlays collected from individuals and local businesses by the prefecture. Thus, larger values of the index indicate a stronger and more vital economy as well as greater employment opportunities.

We also included an indicator of the housing conditions available to families in each prefecture. Because coresidence arrangements generally require larger dwelling units to accommodate multiple generations in a single household (Logan & Bian, 1999; Sodei, 1995), we included a variable representing the average floor area in square meters per household in each prefecture (one square meter = 10.76 square feet) as retrieved from the 2000 Japan Census (Statistical Bureau at Ministry of Internal Affairs and Communications, 2006).

We represented the normative climate using data from the 1998 National Family Research of Japan to calculate the average prefecture-wide response to the following statement: "Children ought to live with their parents when their elderly parents can no longer live by themselves." Responses were scored on a 4-point scale, ranging from 1 (disagree) to 4 (agree). The 1998 National Family Research of Japan is a national survey conducted by the National Family Research Working Group at the Japan Society of Family Research and contains data from 6,985 residents of Japan aged between 28 and 77 years old (National Family Research of Japan, 2004). We used data from this general population survey to represent prefecture-level attitudes toward coresidence for several reasons. First, the survey measures attitudes across all age groups, not just older individuals. This is a useful property of the data because intergenerational coresidence is an arrangement formed by two age-discrepant generations. Second, the sample was large enough to provide reliable estimates of average attitude within each prefecture. Third, the attitude question related directly to the issue at hand, namely intergenerational coresidence.

To measure availability of public welfare services directed toward older residents in each prefecture, we used an index that represents the utilization of old-age welfare services at the prefecture level. This measure, the local caregiving power index (chiiki-kaigo-ryoku in Japanese), provides an estimate of the coverage of local welfare support for older residents in the community (Sumitomo Life Insurance Research Institute, 2001). This index represents annual days of utilization of three types of elder care services (home care service, day care service, and short-stay service) per 100 individuals aged 65 years and older in each prefecture.

Statistical Model
This study aimed to investigate ways in which individual-level and prefecture-level characteristics are associated with the likelihood of two types of intergenerational coresidence. To do this, we employed a multinomial logistic regression using hierarchical generalized linear modeling (Luke, 2004; Raudenbush & Bryk, 2002). Using this approach, we contrasted life-long and boomerang coresidence with non-coresidence using a logit function in which the likelihood of each was estimated within a two-level structure of individuals (Level 1) and prefectures (Level 2). We note that our statistical approach employed a random effects model in which intercept and slope coefficients of individual-level predictors varied across prefectures and were then predicted by prefecture-level characteristics.

We incorporated into our model the strategy of grand-mean centering all predictor variables (Luke, 2004) so that random intercepts described the likelihood of each type of coresidence for the "average older person" in the sample. This centering method has the advantage of holding constant the characteristics of older people that may vary across prefectures, thus allowing for the unique identification of the influence of prefecture characteristics.

We note that the interpretation of numerical estimates from random effects models does not necessarily describe average effects in the population. In our application, we made inferences with respect to an "average" individual—that person who was at the mean of all variables (Muramatsu & Campbell, 2002). An alternative approach to random effects is population-averaged estimation, in which the emphasis is on identifying average effects within a population of interest. In such an approach, the nonindependence of individuals due to clustering is considered a nuisance to be statistically removed (Hu, Goldberg, Hedeker, Flay, & Pentz, 1998). In practice, however, the distinction between random effects and population-averaged approaches is negligible when there is little variation in the random effects. In our research, variation in the random effects—though significant—was relatively small. In converting random effects estimates to population-averaged estimates with a formula suggested by Hu and colleagues, we found that each method produced almost identical estimates, indicating that the effects for the average person in the population were the same as the average effects in the population. We present our results from a random effects model for heuristic reasons because the principal goal of our research was to explain regional variation in coresidence patterns.

At Level 1, we estimated within-prefecture effects with characteristics of individuals as predictors:

where _mij represents the predicted log odds that person i in prefecture j is in coresidence category m (life long or boomerang) relative to the reference group (non-coresidence), b_0j(m) are random intercepts representing the overall adjusted log odds of type m coresidence within the J prefectures, and b₁ to b_z represent fixed effects of Z individual characteristics (IC).

At Level 2, we estimated the effects of prefecture-level characteristics on Level 1 intercepts to test for sources of contextual variation in the prevalence of each type of coresidence:

where for each of m residential contrasts, g₀₀ represents the constant, g₀₁ to g_0k are the effects of K prefecture characteristics (PC), and u_0j represents the residual term.

We estimated cross-level interactions to test our cultural plasticity hypothesis by first allowing effects of filial responsibility at Level 1 to vary across prefectures and then predicting these random effects by prefecture characteristics. In this equation, our interest was in identifying how prefecture attributes modify the influence of personal filial attitudes on coresidence outcomes. Thus, to test for cross-level interactions we estimated both equations above and the following Level 2 equation:

where b_1j(m) are prefecture-specific effects of filial responsibility on each of the m residential contrasts, g₁₀ represents the constant, g₁₁ to g_1k are the effects of prefecture characteristics, and u_1j represents the residual term.

	RESULTS

TOP Abstract Methods Results Discussion References

 
We first present the characteristics of the sample (Table 1). Somewhat more than half of the older parents (58%) lived with their children. About 45% lived with at least one child as a life-long partner, and 13% lived with at least one child who had previously lived independently (i.e., boomerang coresidence). Slightly more than half (57%) were women, somewhat more than one third (38%) were widowed, and 38% had at least a high school education. The average age of the sample was 75 years. Respondents had an average household income between 1.5 million Japanese yen (about $12,500 U.S.) and 3 million Japanese yen (about $25,000 U.S.), the large majority (90%) owned their home, and about 60% lived in an urban area. In terms of family composition, older parents in our sample had 2.6 children on average, the majority (78%) had at least one son, and slightly fewer than half (46%) had at least one non-coresiding child living in the same town or city. Adult children's average level of educational attainment corresponded to a value signifying more than high school but less than a college graduate. Data were available from individuals in all 47 prefectures, averaging 77 respondents per prefecture with a range from 21 to 287.

Table 3 presents logit coefficients and converted odds ratios from the multinomial logistic regression. Results for the main effects are shown in the first two equations. We note that because individual-level characteristics were controlled at Level 1 as mean-centered variables, random intercepts at Level 2 were already adjusted for compositional differences between prefectures. In other words, contextual characteristics predicted the log odds of coresidence when individual-level variables in each prefecture were held constant at their respective grand means.

Cultural variables affected the likelihood of life-long coresidence. Parents who more strongly endorsed traditional attitudes toward filial responsibility were more likely than those who less strongly endorsed these attitudes to live in life-long coresident arrangements, as were respondents with less education when compared to those with more education. Rural residents, typically more traditional than their urban counterparts, were 40% more likely to live in life-long arrangements with their children.

Characteristics of the children also affected likelihood of coresidence. Parents with more children, less educated children, and those without children living independently in the same town or city were more likely to coreside with children in a life-long or a boomerang arrangement. Parents with at least one son were 75% more likely than those parents without a son to live in a life-long arrangement.

Examining the influence of the four contextual variables, we observed that elderly residents of prefectures with more spacious housing on average were more likely to coreside with children on a life-long basis (67% more for each standard unit change in housing space). Older residents living in prefectures with stronger economies were more likely than those in weaker economic contexts to engage in both types of coresidence. The effect was substantially higher for life-long coresidence, with a 58% increase for every standard unit increase in financial power index. The availability of local welfare resources tended to reduce the likelihood of both types of coresidence. Living in a more service-rich prefecture was linked to a reduction in life-long coresidence and boomerang coresidence (a standard score increase in availability indicated 16% and 13% lower residence, respectively). The normative climate of a prefecture had no influence on the likelihood of either type of coresidence.

We introduced cross-level interactions between values of filial responsibility and the four contextual variables in the last two equations of Table 3. We found two such interactions, both with respect to boomerang coresidence. Living in a more traditional normative climate increased the strength with which older parents' values toward filial responsibility predicted their likelihood of reconstituting a shared household with a formerly independent child. Predicted probabilities based on model estimates demonstrated this interaction at three levels of normative climate as shown in Figure 1. The second interaction tested was with respect to public service availability. The relationship between traditional filial values and living with a boomerang child was increasingly positive for older parents residing in prefectures with greater availability of public services. Figure 2 shows these increasing slopes based on predicted probabilities.

View larger version (12K): [in this window] [in a new window]   Figure 1. Predicted probability of boomerang coresidence by older parent's normative attitude and cultural climate of prefecture. Predicted probabilities were calculated for an older resident with average values of individual and prefecture characteristics. SD = standard deviation

View larger version (13K): [in this window] [in a new window]   Figure 2. Predicted probability of boomerang coresidence by older parent's normative attitude and welfare resource availability of prefecture. Predicted probabilities were calculated for an older resident with average values of individual and prefecture characteristics. SD = standard deviation

	DISCUSSION

TOP Abstract Methods Results Discussion References

 
This study examined the extent to which norms, needs, and resources guide the intergenerational coresidence of older people in Japan. To address this issue, we took an innovative approach that disaggregated coresidence into permanent and contingent types of shared living arrangements and examined the role played by the larger ecological context within which coresidence arrangements are entered. In general, we conclude that (a) motives for intergenerational coresidence in Japan are diverse and in some instances deviate from traditional forms, and (b) filial piety is a malleable cultural script that has different manifestations depending on the contextual characteristics that enable or inhibit the formation of multigenerational households.

Divergences between the two types of coresidence fell into predictable patterns. Although the two forms of coresidence are similar in form, they appear to be differently prescriptive as judged by the greater strength with which filial attitudes predict the more traditional arrangement. It is interesting to note that the less traditional boomerang type of coresidence was more strongly associated with traditional values when the parent lived in prefectures with more traditional normative climates. One possible interpretation of this suggests that more traditionally minded parents tend to reunite with their children when social norms are more regulative regarding the role of children. That such regulation takes place only at the intersection of personal values and contextual milieu can be understood by action regarding coresidence, whereas the life-long arrangement requires the lack of action.

In terms of parental need, the boomerang household appears to be more common among parents who have poor health and who are widowed. This is to be expected, as the reunification of parents and children represents a form of coresidence that is peculiarly modern, suggesting the decoupling of nuclear households in modern nations (Goode, 1963; Parsons & Bales, 1955) with punctuated bursts of support on an as-needed basis (Riley & Riley, 1993). Because life-long coresidence is culturally prescribed by norms of filial piety, older parents in such households will represent a mix of healthy and unhealthy parents. Examined another way, living with parents when one becomes widowed may make the even the most reluctant child remain in the survivor's household when compared to the relative freedom granted that same child living with married parents. The contrast between the two types of coresidence arrangements implies that the etiology of coresidence is important to consider in contemporary Japanese society, where family behavior is increasingly motivated by the point-in-time emergence of need and less so by prescriptive obligations regarding lifestyle (e.g., Koyano, 2003; Pimentel & Liu, 2005).

There were similarities between the two arrangements as well. For instance, both types of coresidence tend to accommodate widowed parents. In addition, both living arrangements are more common among parents with more children, and less common among parents who have children living nearby. The sustenance of traditional cultural practices requires that a sufficient level of socioeconomic resources be available. This was particularly reflected by housing conditions. As past studies have suggested (e.g., Logan & Bian, 1999; Takagi & Silverstein, 2006), an older parent's home ownership seems to be a critical prerequisite for coresidence. At the contextual level, larger amounts of housing space enable families to live together on a life-long basis—an arrangement more likely to entail the possibility of having three generations in the same household. In addition, a stronger economy enhances both types of coresidence, mostly the life-long type. It is possible that healthier economies produce work opportunities for younger families that keep them (or allow them to be) coresident with their parents. Consistent with modified modernization theory (Inglehart & Baker, 2000), our results suggest that socioeconomic development provides some members of the younger generation the resources that enable them to fulfill their perceived cultural duties.

More generous local welfare policies at the prefecture level produce a reduction in coresidence. At first blush, this seems to support a substitution hypothesis whereby public services "crowd out" the demand for coresidence (Kono, 2000). However, at the same time, our analysis also shows that traditional values at the individual level are more likely to be expressed in boomerang coresidence in prefectures with greater welfare provisions for older adults. This suggests that formal services and traditional orientations are complementary with respect to coresidence decisions, which is reflected in the principles underlying the Japanese-style welfare system that emphasizes the importance of partnerships between families and the public welfare sector in the support of older individuals (Goodman & Peng, 1996). In this respect, local welfare resources in Japanese communities may implicitly reinforce the contribution of family support for needy older individuals. At the same time, the partnership between welfare service and family implied in this study signifies the complementary relationship between formal and informal caregivers that researchers have observed in other national contexts such as the United States and some European countries (Daatland & Herlofson, 2003; Muramatsu & Campbell, 2002).

There are several limitations to our investigation that merit discussion. First, the cross-sectional nature of the data limits our ability to make causal inferences. The study on which our analysis was based offers the possibility of longitudinal analysis. However, because relatively few respondents changed their living arrangements over the 2 years between interviews, the analysis of household transitions was not feasible. Therefore we focused only on the first wave of data that had the added advantage of providing a larger and more representative sample than a longitudinal analysis would have provided. We acknowledge several potential problems of our cross-sectional analysis: (a) We almost certainly underestimated lifetime risk of living with a boomerang child because we were incapable of detecting past episodes that had ended; (b) our contemporaneous predictor variables may not have captured well the precipitating causes of residential arrangements, particularly those arrangements of longer duration; and (c) predictor variables may have been endogenous with respect to living arrangement and may themselves have changed in response to household conditions. Clearly, future research on this topic must include prospective longitudinal designs to better establish the causal ordering of variables.

Second, we note that we omitted from our analyses several characteristics of adult children, such as their health and marital status, because they were not distributed well enough to produce convergence in our multilevel models. We also did not have available direct measurements of children's values and economic needs. However, if we accept that educational level of children is a proxy for their socioeconomic status and filial attitudes, our results are consistent with expectations; the likelihood of each type of coresidence was lower among parents with more highly educated children. This suggests that the historical reduction in multigenerational households may have its roots in the changing values among the younger generation in Japan. Finally, we note that our contextual measures may merit further examination in future research. For instance, we originally considered housing costs to be another important contextual factor in producing intergenerational coresidence (Mutchler & Burr, 2003); however, the average cost of housing was highly correlated with other prefecture variables in our analysis, particularly the indicator of economic well-being (r =.9). Consequently, our measure of overall economic well-being virtually subsumed the cost of housing in its purview. Thus, the high cost of housing in prefectures with more robust economies may help explain why older residents in such prefectures are more likely to live with children who tend to find local employment opportunities but who are financially unable to live independently.

Population aging and the emergence of the welfare state demand that modern societies with strong cultural preferences for intergenerational coresidence reassess the role of children as a source of support for older people. This involves modifying expectations for traditional living arrangements and finding a comfortable balance among preferences, opportunity, and social welfare services (Hashimoto, 1997). Our study demonstrates that intergenerational family arrangements and fulfillment of older parents' preferences hinge upon not only personal circumstances and cultural beliefs but also on socioeconomic and public resources available at the community level. We conclude that traditional values are not necessarily inconsistent with modern bureaucratic systems of care and economic development in Japan and indeed may, under certain circumstances, be strengthened by them, albeit in somewhat modified form.

	Acknowledgments

 
This research was partially supported by Grants AG021656 and AG021609 from the National Institute on Aging. We would like to thank the Nihon University Center for Information Networking for use of the Nihon University Japanese Longitudinal Study of Aging data, as well as the Social Science Japan Data Archive, Information Center for Social Science Research on Japan, Institute of Social Science at the University of Tokyo for providing the data from the National Family Research of Japan in 1998. We especially would like to thank Dr. Yasuhiko Saito at Nihon University for his assistance in accessing the data. We also appreciate the thoughtful comments and suggestions provided from the anonymous reviewers of the Journal of Gerontology: Social Sciences.

	Footnotes

 
Decision Editor: Kenneth F. Ferraro, PhD

Received for publication September 22, 2006. Accepted for publication April 13, 2007.

	References

TOP Abstract Methods Results Discussion References

 

• Brown, J. W., Liang, J., Krause, N., Akiyama, H., Sugisawa, H., Fukaya, T. (2002). Transitions in living arrangements among elders in Japan: Does health make a difference? Journal of Gerontology: Social Sciences, 57B,S209-S220.[Abstract/Free Full Text]
• Brown, N. (2003). Under one roof: The evolving story of three generation housing in Japan. In J. W. Traphagan & J. Knight (Eds.), Demographic change and the family in Japan's aging society (pp. 53–71). New York: State University of New York Press.
• Cabinet Office. (2001). Koreisha no seikatsu to ishiki ni kansuru kokusai hikaku chosa [International comparison of life and opinions of older people]. Retrieved August 8, 2006, from http://www8.cao.go.jp/kourei/ishiki/h12_kiso/html/0-1.html.
• Cabinet Office. (2004). Korei-shakai hakusho [Annual report on the aging society]. Tokyo: Gyosei.
• Cabinet Office. (2005). Korei-shakai hakusho [Annual report on the aging society]. Retrieved October 14, 2005, from http://www8.cao.go.jp/kourei/whitepaper/w-2005/zenbun/17index.html.
• Cherlin, A. (1994). The Japanese family in comparative perspective. In L. J. Cho & M. Yada (Eds.), Tradition and change in the Asian family (pp. 421–433). Honolulu: University of Hawaii Press.
• Daatland, S. O., Herlofson, K. (2003). ‘Lost solidarity’ or ‘changed solidarity': A comparative European view of normative family solidarity. Ageing and Society, 23,537-560.
• Development Bank of Japan. (2006). Chiiki deta beisu [Local data base]. Retrieved February 10, 2006, from http://www.dbj.go.jp/japanese/download/database/main.html.
• Goode, W. J. (1963). World revolution and family patterns. New York: Free Press of Glencoe.
• Goodman, R., Peng, I. (1996). The East Asian welfare states: Peripatetic learning, adaptive change, and nation-building. In G. Esping-Andersen (Ed.), Welfare states in transition: National adaptations in global economies (pp. 192–224). Thousand Oaks, CA: Sage.
• Hashimoto, A. (1996). The gift of generations: Japanese and American perspectives on aging and the social contract. New York: Cambridge University Press.
• Hashimoto, A. (1997). Designing family values: Cultural assumptions of an aging society. Japan Quarterly, 44,(4), 59-65.
• Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47,(1), 153-162.
• Hu, F. B., Goldberg, J., Hedeker, D., Flay, B. R., Pentz, M. A. (1998). Comparison of population-averaged and subject-specific approaches for analyzing repeated binary outcomes. American Journal of Epidemiology, 147,694-703.[Abstract/Free Full Text]
• Inglehart, R., Baker, W. E. (2000). Modernization, cultural change, and the persistence of traditional values. American Sociological Review, 65,19-51.
• Ishikawa, H., Maeda, D. (2000). Development of long-term care for elderly people in Japan. In D. R. Phillips (Ed.), Aging in the Asia-Pacific region: Issues, policies, and future trends (pp. 133–157). New York: Routledge.
• Izuhara, M. (2000). Changing family tradition: Housing choices and constraints for older people in Japan. Housing Studies, 15,89-110.
• Izuhara, M. (2004). Negotiating family support? The ‘generational contract’ between long-term care and inheritance. Journal of Social Policy, 33,649-665.
• Kawabe, H., Shimizu, H. (1994). Japanese perceptions of the family and living arrangements: The trend toward nuclearization. In L.-J. Cho & M. Yada (Eds.), Tradition and change in the Asian family (pp. 121–133). Honolulu: University of Hawaii Press.
• Kono, M. (2000). The impact of modernization and social policy on family care for older people in Japan. Journal of Social Policy, 29,181-203.
• Koyano, W. (1996). Filial piety and intergenerational solidarity in Japan. Australian Journal on Ageing, 15,51-56.
• Koyano, W. (2003). Intergenerational relationships of Japanese seniors: Changing patterns. In V. L. Bengtson & A. Lowenstein (Eds.), Global aging and its challenge to families (pp. 272–283). New York: Aldine de Gruyter.
• Kumagai, F. (1996). Unmasking Japan today: The impact of traditional value on modern Japanese society. Westport, CT: Praeger.
• Kumagai, F. (2005). Co-residency living arrangement as a factor to facilitate labor force participation of Japanese women: Community level analysis of Japanese families. International Journal of Sociology and Social Policy, 25,(3), 42-62.
• Kurosu, S. (1994). Who lives in the extended family and why? The case of Japan. In L.-J. Cho & M. Yada (Eds.), Tradition and change in the Asian family (pp. 177–198). Honolulu: University of Hawaii Press.
• Logan, J. R., Bian, F. (1999). Family values and coresidence with married children in urban China. Social Forces, 77,1253-1282.
• Long, J. S. (1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage.
• Luke, D. A. (2004). Multilevel modeling. Thousand Oaks, CA: Sage.
• Maeda, D., Ishikawa, H. (2000). Ageing in Japan: Retirement, daily lives, pensions and social security. In D. R. Phillips (Ed.), Aging in the Asia-Pacific region: Issues, policies, and future trends (pp. 113–132). New York: Routledge.
• Martin, L. G., Tsuya, N. O. (1994). Middle-aged Japanese and their parents: Coresidence and contact. In L.-J. Cho & M. Yada (Eds.), Tradition and change in the Asian family (pp. 153–178). Honolulu: University of Hawaii Press.
• Ministry of Health, Labour and Welfare. (2000). Kosei hakusho [White paper: Annual report on health and welfare]. Tokyo: Gyosei.
• Muramatsu, N., Campbell, R. (2002). State expenditures on home and community based services and use of formal and informal personal assistance: A multilevel analysis. Journal of Health and Social Behavior, 43,107-124.[Medline]
• Mutchler, J. E., Burr, J. A. (2003). Living arrangements among older persons: A multilevel analysis of housing market effects. Research on Aging, 25,531-558.[Abstract]
• National Family Research of Japan. (2004). NFRJ98 gaiyo[Profile of NFRJ98]. Retrieved September 18, 2004, from http://www.waseda.jp/assoc-nfroffice/NFRJ98_profile.htm.
• Ogawa, N., Retherford, R. D. (1993). Care of the elderly in Japan: Changing norms and expectations. Journal of Marriage and Family, 55,585-597.
• Parsons, T., Bales, R. F. (1955). Family, socialization, and interaction process. Glencoe, IL: Free Press.
• Peng, I. (2003). Pushing for social care expansion: Demography, gender, and the new politics of the welfare state in Japan. American Asian Review, 21,(2), 25-55.
• Pimentel, E. E., Liu, J. (2005). Patrilineal coresidence in urban China: A life course perspective. Intergenerational Journal of Sociology and Social Policy, 25,63-91.
• Raudenbush, S. W., Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage.
• Raymo, J. M., Kaneda, T. (2003). Changes in the living arrangements of Japanese elderly: The role of demographic factors. In J. W. Traphagan & J. Knight (Eds.), Demographic change and the family in Japan's aging society (pp. 27–52). New York: State University of New York Press.
• Riley, M. W., Riley, J. W., Jr.. (1993). Connections: Kin and cohort. In V. L. Bengtson, & A. W. Achenbaum (Eds.), The changing contract across generations (pp. 169–189). New York: Aldine Transaction.
• Rindfuss, R. R., Liao, T. F., Tsuya, N. O. (1992). Contact with parents in Japan: Effects on opinions toward gender and intergenerational roles. Journal of Marriage and Family, 54,812-822.
• Shimizu, H. (2004). Kazoku to fuyo [Family and support]. In H. Shimizu, K. Mori, M. Iwakami, & M. Yamada (Eds.), Kazoku kakumei [Family revolution] (pp. 200–206). Tokyo: Kobundo.
• Snijders, T. A. B., Bosker, R. J. (1994). Modeled variance in two-level models. Sociological Methods & Research, 22,342-363.[Abstract]
• Sodei, T. (1995). Care of the elderly: A women's issue. In K. Fujimura-Fanselow & A. Kameda (Eds.), Japanese women: New feminist perspective on the past, present, and future (pp. 213–228). New York: Feminist Press.
• Statistical Bureau at Ministry of Internal Affairs and Communications. (2006). Population census. Retrieved August 11, 2006, from http://www.stat.go.jp/english/data/kokusei/2000/kihon1/00/13.htm.
• Sumitomo Life Insurance Research Institute. (2001). Chiiki kaigoryoku deta bukku [Databook of local caregiving power]. Tokyo: Chuo Houki.
• Takagi, E., Silverstein, M. (2006). Intergenerational coresidence of the Japanese elderly: Are cultural norms proactive or reactive? Research on Aging, 28,473-492.[Abstract/Free Full Text]
• Thompson, C. S. (2003). Depopulation in rural Japan: "Population politics" in Towa-cho. In J. W. Traphagan & J. Knight (Eds.), Demographic change and the family in Japan's aging society (pp. 89–106). New York: State University of New York Press.
• Tsuya, N. O., Martin, L. G. (1992). Living arrangements of elderly Japanese and attitudes toward inheritance. Journal of Gerontology: Social Sciences, 47,S45-S54.
• University of Southern California/University of California, Los Angeles, Center on Biodemography and Population Health. (2004). Nihon University Japanese Longitudinal Study of Aging. Retrieved December 11, 2004, from http://www.usc.edu/dept/gero/CBPH/nujlsoa/index.htm.

This article has been cited by other articles:

				S. O. LONG, R. CAMPBELL, and N. Chie Does It Matter Who Cares? A Comparison of Daughters versus Daughters-in-Law in Japanese Elder Care Soc. Sci. Jpn. Res., June 1, 2009; 12(1): 1 - 21. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire