Journal of Cross-Cultural Gerontology 13: 241–264, 1998. © 1998 Kluwer Academic Publishers. Printed in the Netherlands.
241
Intergenerational living arrangements in Turkey IS ¸ IK A. AYTAC ¸ New England Research Institutes Inc., Watertown, Massachusett, USA
Abstract. Despite the growing research on the living arrangements of elderly we know very little about intergenerational relations in Turkey. This research is a step towards this goal. The purpose is to assess whether urbanization and economic development at the aggregate level and modernity and secularism at the individual level result in a decline in coresiding and living nearby older family members in Turkey. Results obtained using multinomial logistic estimates are supportive of the hypotheses. There is a significant variation in coresidence and in the reasons for coresiding with older family members by regional development, urbanization, individual modernity and religiosity. Keywords: Intergenerational living arrangements, Elderly, Turkey, Middle East
Introduction Population aging, the result of economic, social and demographic changes, has led to a growing interest in the wellbeing of older adults by both researchers and policymakers. While there is a growing body of research on intergenerational relations in non-Western countries (Casterline et al. 1992, 1993; DaVanzo & Chan 1994; Martin 1989, 1990; Ofstedal 1994), little is known about the intergenerational relations in Turkey, a predominantly Muslim country. Turkey is a highly complex, ethnically diverse, heterogenous society experiencing rapid changes as a result of legal and educational reform, secularization, economic growth and industrialization. While these changes have played an important role in Turkish society, their impact as a whole has been very uneven. Resulting chiefly from government policies that favored certain regions and sectors, and from the resistance of traditional local elites to reforms, the differential impact of the various reforms and social change in Turkey has led to a society that is characterized by both traditionalism and modernism (Abadan-Unat 1986; Meric 1991; Ozbay 1990; Timur 1972). Using the recent nationwide Turkish Family Structure Survey, the primary objective is to examine intergenerational living arrangements in Turkey. The purpose is to assess whether urbanization and economic development at the aggregate level (ranging from economically less developed to more developed regions), modernity at the individual level (especially increased levels of
242
I¸SIK A. AYTAC ¸
education), and increased levels of secularism result in a decline in coresiding and living nearby older family members.1 Since there is limited information on the characteristics of older family members (discussion below), this research primarily focuses on the effects of the respondents’ characteristics (i.e., married men who have older family members). It is therefore a modest attempt at providing a crossectional snapshot of intergenerational relations in Turkey. Despite these limitations, it is the most recent nationwide research and will provide baseline information on intergenerational relations in Turkey. Background. Historically in Turkey, both because of Islamic precepts and traditional cultural values, older adults – parent, parents-in-laws and older aunts and uncles – have been taken care of by family members, regardless of whether the elderly lived in an extended family household or a nuclear family household. Adult children, especially sons, have always been seen as security for old age (Meric 1991; Timur 1972). Although in 1988, less than one percent of the elderly live in nursing homes (Turkish Family Structure Survey), recent demographic changes may challenge these traditional values. Even though at present Turkey has a young population (one third is under the age 15), declines in the fertility rate will lead to population aging. The total fertility rate has declined from 5 children per woman during the early 1970s to 2.7 in 1993. It is even lower for women with secondary or more education, who average 1.7 children (Hancioglu 1993; Toros 1993). Approximately 60% of Turkish families are nuclear, 19% are patriarchal extended, 13% are transient extended, and 8% are dissolved families or nonfamily households (Timur 1972). Nuclear families are more common among the landless, the more educated and among those who have migrated. However, nuclear families in Turkey often function like extended families and the results obtained from case studies indicate the persistence of close social ties, where emotional bonding and social contact among Turkish families remains very strong (Duben 1982), forming ‘functionally extended families’ (AbadanUnat 1986). Those who migrate to urban areas continue to receive emotional and material support and have frequent social contact from family members who remain in rural areas (Duben 1982). They often receive food from their relatives in rural areas (Ozbay 1985). Theoretical framework. The transition from a traditional society to a ‘modern’ society often changes family structure and family relations in fundamental ways. This is because the various processes of modernization – industrialization, urbanization, secularization, expansion of mass education and mass media, market economy, increasing incomes and female labor force partici-
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
243
pation – undermine the family-based mode of production in which the family serves all basic functions.2 In analyzing how the intergenerational relations in Turkey are influenced by social, economic, and demographic changes, the main question is to what extent the experience of Turkey – a developing Muslim country – will resemble the experience of the developed societies that have completed their demographic transition, and how will it compare with other developing countries with different cultural and religious backgrounds.3 According to demographic transition theory, there are three stages in the demographic transition (Albert & Cattell 1994). The first stage, where both fertility and mortality are high, is followed by the second where mortality declines but fertility remains high resulting in rapid population growth. In the third stage, fertility declines as well, resulting in population aging, thus, an increased concern for the well being of the elderly. Basically, most theories argue that during the transformation from an agricultural to an industrial society the burden of care for the older adults shifts from families to the state or other formal organizations or to the older adults themselves. This occurs because the fall in fertility and rise in life expectancy that accompany development mean that the ratio of older adults needing support to adult children capable of giving it changes dramatically. The traditional means of support may be overwhelmed by the demands placed upon them (Cowgill & Homes 1972; Davis & van den Oever 1981; Ogawa & Retherford 1993; Smelser 1959). In addition, the shift from a rural-agricultural economy to an urbanindustrial economy, together with the increasing levels of education that accompany it, means that parents and children more often differ in profound ways, increasing the potential for intergenerational conflict. The rise of wage labor, the separation of work from family life, and increasing education may erode the economic basis for parental authority and lead to a more individualistic perspective. Norms of filial care of elderly parents also weaken as a consequence of the development process (Ogawa & Retherford 1993). Education is by far the most influential force in the modernizing process, as was the case during the changes that transformed the European cultural value system (Lesthaeghe & Surkyn 1988; Caldwell 1980). In Europe, education diminished the influence of traditional religious beliefs, as well as traditionalism in marriage. This resulted in a definite ideational change – one that fosters postmaterialism, the quest for perfection (such as demanding better contraception, greater educational opportunities for daughters, and preference for a ‘good divorce’ over a bad marriage) and a disposition for independence. Education was found to be the strongest positive correlate of individualism.
244
I¸SIK A. AYTAC ¸
Furthermore, the weakening of traditional ways occurred faster in areas where production was no longer based on the family unit (Lesthaeghe 1983; Caldwell 1980). For instance, during the end of the nineteenth century, in eastern Germany, northern Netherlands, and southern Spain, the more secularized rural areas were also the ones with large farms employing fulltime, wage-earning farm laborers. Overall, during the last century, the social changes in Europe led to a shift from a group orientation to an individual orientation (Lesthaeghe & Surkyn 1988). Social, economic and legal changes in Turkey. Prior to the establishment of the Turkish Republic in 1923, the Ottoman Empire was ruled by the Islamic law (the Sharia – which advocates sex segregation and unequal legal treatment) and the majority of the population received only religious education (Cagirici 1992). During the 1920s, the new republic adopted the Swiss Civil Code and became the first secular Muslim country. The legal reforms included the abolition of the Sharia, Caliphate and polygamy, the granting of equal rights to women and men in divorce and inheritance, the right to education for both sexes, and the establishment of a minimum age at marriage (15 for girls and 17 for boys). Western-style dress became the norm for both men and women and suffrage for women was granted in 1934. These reforms were intended to ‘civilize’ and ‘modernize’ Turkey using the more developed Western societies as a model (Turkdogan 1992).4 One of the most important reforms was educational reform. Until then Turkish was written using the Arabic alphabet (adopted at the time Turks became Muslims) which was not suitable for the Turkish language due to a mismatch of vowels. As a result it took almost two years to learn how to read and write (Mumcu 1986). The reform replaced the Arabic alphabet with Roman alphabet, which reduced the learning time to 3 months. It was during this time that universal education, as one of the most influential forces of modernization, was first extended to rural Turkey. In 1935, while only 14 percent of all villages had schools, by the 1980s all villages had primary schools (Ozbay 1985). However, the lack of middle and high schools in villages and the tradition of keeping daughters tied to the household has resulted in significant gender inequality in educational attainment. For the most part, only males ages 15–19 migrate to urban areas for education past grade school (Yener 1977). Although the gender gap in education persists, the literacy rate for both sexes increased significantly over the last half century. From 1935 to 1985, the male literacy rate rose from 30 to 86 percent, while during the same period, the female literacy rate rose from 10 to 68 percent (Turkish Census 1985).
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
245
Despite the ‘modernizing’ reforms, in place for more than seven decades, Islamic tradition is still very visible in the eastern and southeastern regions, in small towns and rural areas, and in general among the less educated. Today the evidence of traditionalism appears in headcovering worn by women, in arranged marriages and the bride-price practice (although the latter is in decline), attendance at religious schools, and in religious weddings. Arranged marriages are common because they foster family unity and protect property, political linkages and patriarchal authority within extended families (Fox 1975). Using a sample of 754 married women in Ankara in 1966, Fox (1975) reports that close to three fourths of all marriages were arranged marriages. Even though civil marriages are required today, a sizeable proportion of couples only go through the religious ceremony and a large proportion have both civil and religious ceremonies (Magnarella 1973). Among the more religious and traditional households the physical separation of the sexes is also observed – men and women eat separately and sit in different rooms, although women past childbearing age are accorded more freedom (Kandiyoti 1977; Timur 1972). However, signs of modernity can be found in higher educational attainment, the increase in women’s nonagricultural labor force activities, in preferences for non-local television and radio programs, western dress, love marriages and in civil wedding ceremonies. From 1938 to 1976 the number of female civil servants in Turkey increased 19 times (while the number of male civil servants rose sixfold) and 68 percent of these women were graduates of high school or beyond (only 54 percent of males were). Love marriages, which are a fairly recent phenomenon, are closely related to exposure to urban life during early teenage years, education higher than primary school, and age – women marrying at age 20 or later were more likely to have chosen their own husband. Furthermore, Fox reports that marriage type also affects subsequent marital behavior, love marriages being relatively more egalitarian (1975). Turkey, which is approximately 98 percent Muslim, is far from being a homogenous society. Indeed as a rapidly changing society, it is full of dualities. The urban and western parts of the country are the least traditional, at the same time that significant variation exists in educational attainment. While extended families are more common in the east and north, the decline in family authority is greater among the western and urban nuclear families, where the sons are more educated, have higher incomes, and are exposed to city life (Illbas 1991). Furthermore, significant differences are observed in basic demographic characteristics, such as in total fertility rate, infant mortality rate and nutritional status by region, residence, and education (see Appendix).
246
I¸SIK A. AYTAC ¸
Hypotheses. Turkey may not follow the development path taken by Europe. The Islamic value system in Turkey may prevent a shift to an ideational system similar to the Western one emphasizing individuation as opposed to group solidarity. Thornton and Fricke (1987) report that while there are striking similarities in family change between the West, China and South Asia, there are also important differences that can be attributed to culturally-specific family structure and kinship patterns. More research on developing countries with different cultural and religious backgrounds is needed in order to assess the implications of increasing education, industrialization, urbanization and concomitant social change on the position of the elderly in the family. Some of these ideas can be tested in Turkey, where industrialization, urbanization, and secularism have developed unevenly thereby enabling us to compare living arrangements with elderly across individuals with various levels of exposure to the development process. Based on small area studies there is some support for national variations in family relations, yet to date no systematic nation-wide research has been conducted on intergenerational relations. This research aims at filling this gap. Using the newly released Turkish Family Structure Survey (1988), we will analyze the impact of economic development, urbanization, individual modernity, and religiosity on intergenerational relations in Turkey. The purpose is to assess whether the traditional emphasis on family values and filial piety are being replaced by the Western values. Even though Turkey has not completed its demographic transition, among its more modern, urban, and secular population we expect some aspects of family relations to resemble, more or less, the post-transitional societies, while in others we expect to find strong Islamic values maintaining an influence over kinship ties. Thus it is hypothesized that: H1. Those who live in the more developed regions of the west or the south, or urban areas, who are more educated, and who are not orthodox Muslims will be less likely to coreside with older family members. On the other hand, those who are more traditional, that is, who live in the less developed regions and in rural areas, who have lower education, and who are more orthodox Muslims will be more likely to live with older family members. Despite the trend to ‘westernize’ and the resultant changes in living arrangements, in a Muslim country such as Turkey, where there is strong emphasis on the family, it is expected that families will continue to have close relationships. We speculate that the modernization process in the West, which is rooted in Protestant values of individualism, is different from the modernization process occurring in a Muslim setting where family values and providing support to others are strongly emphasized. Following this we hypothesize that:
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
247
H2. Even though economic development and individual modernity may decrease the likelihood of living together with older family members, it will not have the same effect on living nearby older family members – that is, even if individuals may opt for privacy they will continue to have close family ties by living closeby.
Data and methods Data. The data come from the 1988 Turkish Family Structure Survey (TFSS), conducted by the Turkish Government Planning Office and the State Institute of Statistics. Questionnaires were given to 18,210 respondents, 12,065 from urban areas and 6,145 from rural areas. Since the sex of the respondents is very skewed (93% men and 7% women), and the representativeness of the female respondents is questionable, this research is limited to male respondents. Unmarried respondents (2% of the sample) were also excluded. Furthermore, because we analyze living arrangements with the older family members, the sample is further restricted to married men who have older family members. The resulting sample size is 8639. It is important to note that ‘elderly’ in the Turkish Family Structure Survey refers to all older family members. In responding to the questions on living arrangements with the elderly, although most of the respondents will refer to their parents and parents-in-laws, they may also include other older family members (such as aunts and uncles). This is not a problem in the case of Turkey because traditionally all older family members are taken care of by the family. The main interest here is to what extent the population aging in Turkey will affect the well-being of the older adults in the population. Therefore, the focus is on intergenerational relations in Turkey rather than the parent-child dyad. While the data are well suited to this research, they have been underutilized. This research is the first to analyze the present state of intergenerational living arrangements in Turkey. Dependent variables and estimation Living arrangements. The first dependent variable, the living arrangements of married men with older family members is estimated using a multinomial logit model with outcomes of living together, living nearby, and living far away (the comparison group). Living nearby an older family member is based on the respondent’s subjective response as the survey does not provide information on the distance or travel time.
248
I¸SIK A. AYTAC ¸
Reasons for coresiding. Even though the survey does not provide information on the characteristics of the older family members, which would aid in assessing whether coresidence occurs due to respondents’ or the older generations’ need, there is a general question which asks the reasons for coresiding. The respondents were asked whether they coreside due to tradition and religious obligations or due to the needs of the older family members.5 Multinomial logistic regression is used to contrast those who do not coreside with those who coreside due to tradition versus those who coreside due to the older family members’ needs. Independent variables Region. As discussed earlier, the social, economic, and demographic changes in Turkey have not been homogenous throughout the country. In order to capture this aggregate-level of modernization, we use indicators for regions ranging from the most developed to least developed: west, south, central, north, and east regions.6 The eastern region is used as the comparison category. Location. Two indicators are used to contrast residence in rural areas (30%) with metropolitan areas (28%) and other urban areas (41%). The first distinguishes the three major metropolitan areas in Turkey which have experienced rapid urbanization (Istanbul, Ankara, and Izmir) and is coded 1, 0 otherwise. The second, distinguishes other urban areas, coded 1 if urban, 0 otherwise. Education. Increasing levels of educational attainment was one of the key government policy goals for Turkish society. The TFSS provides information on the degrees completed. We use indicators to distinguish between college graduates, high school graduates, junior high graduates, and those with less than junior high school degree (comparison group). Religiosity. The TFSS contains information on segregation of the sexes emphasized in Islam and still practiced by orthodox Muslims in Turkey. An indicator is used to tap whether respondents practice physical segregation of the sexes, coded 1 if males and females sit in different rooms when they have guests, 0 otherwise. Employment status. Employment status may affect intergenerational living arrangements in several ways. First, those who are not employed may be more likely to coreside with elderly due to their need for a residence. Second, for those who are employed, employment may impose restrictions on location of residence, although the self employed may have more flexibility in choosing their business location. Furthermore, some of the self employed may be in
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
249
a family business which will then enhance coresiding or living nearby older family members. In order to distinguish between various affects of employment status, two indicators are used to compare those who are employees and those who are self employed from the not employed.7 Financial resources of the elderly. Even though the TFSS does not provide detailed information on the older family members, two measures provide useful information on the financial situation of the older generation. First, the respondents were asked who owned the title to their residence. If a member of the older generation owned the house, it was coded as 1, 0 otherwise. Second, the respondents were asked who owned the majority of the family wealth. If an older family member owned it, it was coded as 1, 0 otherwise. Even though these variables do not provide the information on the amount of the resources, they are useful indicators for measuring the relative power of the older family members.8 Control variables. The models are estimated net of various variables that are known to affect living arrangements. These are as follows: Respondent’s age is measured in years and in order to control for nonlinearities a squared term for age is also included into the models. Presence of young children are measured as 1 if there are children less than 12 years of age in the household, 0 otherwise. Unfortunately, the survey does not allow us to break it down into smaller age groups of children. Finally, migration may affect proximity to kin and the living arrangements with kin. In order to estimate the models net of the migration effect, an indicator is used to distinguish migration status, coded 1 if migrated, 0 otherwise. Results Table 1 presents the mean and standard deviations for the dependent and the predictor variables for married men. As the results show 22 percent of our sample coresides with an elderly and 48 percent live nearby an elderly family member. Table 2 presents the multinomial logistic coefficients and the exponent of the coefficients for coresiding, living nearby, and living far away from elderly. It was hypothesized that those who live in the relatively less developed regions and those who are more traditional will be more likely to coreside with the elderly compared to their more modern counterparts. The results are supportive of this hypothesis. As expected there is a strong regional effect on the likelihood of coresiding with elderly. Those who live in the eastern, the least developed region (the comparison group) are more likely to core-
250
I¸SIK A. AYTAC ¸
Table 1. Means and standard deviations for all the variables Mean
S.D.
Living arrangements Coreside Live nearby Live far away (comparison)
0.22 0.48 0.30
0.41 0.49 0.45
Reason for coresiding Coreside due to tradition Coreside due to elderly need
0.11 0.11
0.30 0.31
Location Region West South Central North East (comparison)
0.38 0.17 0.21 0.08 0.16
0.48 0.38 0.40 0.27 0.33
Education Primary school or less (comparison) Junior high school High school College
0.66 0.10 0.13 0.11
0.47 0.30 0.34 0.31
Religiosity Sex segregation
0.24
0.43
Employment status Self employed Employee
0.35 0.48
0.48 0.49
Elderly financial power Elderly owns R’s residence Elderly owns the largest share of the family assets
0.14 0.08
0.35 0.27
0.72 0.38 40.01 1729.37
0.44 0.48 11.32 1038.52
Control variables Kids LT 12 Migrate Age Age-squared N
8639
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
251
side with elderly. Furthermore, the effect is strongest for the most developed regions of the west and the south. The results show a strong effect of urbanization on living arrangements (Table 2). Those who live in the metropolitan areas and other urban areas are less likely to coreside with elderly. Turkey shows a different pattern of intergenerational coresidence by location compared to other Asian countries. Martin (1989) reports no effect of urban residence on coresidence of elderly with adult children in Korea, Malaysia, Philippines, and Fiji. On the other hand, using pooled data for Philippines, Taiwan and Thailand, Casterline et al. (1993) report an increased likelihood of coresidence in metropolitan areas. Similarly, Ofstedal (1994) finds greater likelihood of coresidence in large urban areas in Thailand. Contrary to what we observe for other Asian countries, Turkish respondents in rural areas are more likely to coreside with elderly compared to metropolitan areas and other urban areas.9 Interestingly, even though the likelihood of coresidence is lower in the more economically developed regions, the probability of living nearby elderly does not show a similar pattern with the level of economic development (Table 2). Indeed in one of the more developed regions – the south, married men are more likely to live nearby elderly compared to east (0.309, p < 0.01). On the other hand, in the central area which is less developed than the south, the odds of living nearby elderly is lower than the east (−0.239, p < 0.05). These results are supportive of our hypotheses as they show that while the increased level of development leads to lower coresidence (indicating a preference for privacy), there is no effect of economic development on living nearby – different generations residing nearby and maintaining close family ties. Turning to the effect of individual modernity, measured as the level of educational attainment, as expected, there is a negative relationship between education and coresiding with elderly (Table 2). Compared to those with primary school or less education, college educated are less likely to coreside with elderly (−0.550, p < 0.001). Those with high school or lower education, on the other hand, do not coreside any less than those with primary school or less education. It may be that the elderly family members of the collegeeducated respondents have more resources and therefore are less in need of the respondents’ help. While the more educated are less likely to coreside with elderly, they are also more likely to live nearby. Compared to those who have primary school or lower education, those who have junior high or high school education are around 1.4 and 1.3 times more likely to live nearby elderly, respectively. Turning to the effect of religiosity, whether respondents who observe the physical separation of the sexes differ from the more secular respondents, the results
252
I¸SIK A. AYTAC ¸
Table 2. Multinomial logit estimation of living arrangements with older family members Coreside
Live nearby
Coef. (SE)
Exp. (b)
Coef. (SE)
Exp. (b)
−0.523∗∗∗ (0.118)
0.592
−0.091 (0.103)
0.912
South
−0.623∗∗∗ (0.125)
0.536
Central
−0.425∗∗∗ (0.118)
0.653
−0.256∗ (0.105)
0.773
North
−0.451∗∗ (0.146)
0.636
−0.239 (0.130)
0.786
Metropolitan
−1.009∗∗∗ (0.117)
0.364
−0.074 (0.094)
0.928
Other urban areas
−0.769∗∗∗ (0.088)
0.463
−0.512∗∗∗ (0.077)
0.599
0.221 (0.120)
1.247
0.368∗∗∗ (0.095)
1.445
High school
−0.179 (0.112)
0.835
0.230∗∗ (0.081)
1.259
College
−0.550∗∗∗ (0.129)
0.576
0.078 (0.084)
1.082
0.401∗∗∗ (0.085)
1.494
0.054 (0.075)
1.055
0.230∗ (0.108)
1.259
0.212∗ (0.091)
1.236
−0.446∗∗∗ (0.106)
0.640
−0.429∗∗∗ (0.086)
0.650
1.599∗∗∗ (0.118)
4.948
1.134∗∗∗ (0.107)
3.110
1.078∗∗∗ (0.144)
2.938
0.206 (0.140)
1.229
Location Region West
Education Junior high
Religiosity Sex segregation
Employment status Self employed Employee
Elderly financial power Elderly owns R’s residence Elderly assets
0.309∗∗∗ (0.105)
1.363
253
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
Table 2 (continued) Coreside Coef. (SE) Control variables Kids LT 12
Exp. (b)
Live nearby Coef. (SE)
Exp. (b)
0.184∗ (0.084)
1.202
0.031 (0.067)
1.031
Migrate
−0.821∗∗∗ (0.083)
0.439
−1.418∗∗∗ (0.064)
0.242
Age
−0.015 (0.018)
0.985
0.019 (0.016)
1.019
1.001
−0.001 (0.001)
0.999
Age-squared
0.001∗∗ (0.001)
Intercept
0.442 (0.427)
Chi-square df N
2256.47 36 8407
0.977∗∗ (0.369)
∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001
show a strong effect of being religious and coresiding with elderly (Table 2). Married men who practice sex segregation are 1.5 times more likely than their less religious or secular counterparts to live with elderly (0.401, p < 0.001). In addition, as expected there is no effect of religiosity on living nearby – respondents live nearby elderly regardless of whether they are orthodox Muslims or not. Although this paper focuses on the relationship of regional economic development, individual modernity versus traditionalism, and religiosity with living arrangements, we briefly discuss the effect of the other predictor variables on the likelihood of coresiding or living nearby with elderly (Table 2). The self-employed are more likely to coreside or live nearby elderly. As we discussed earlier, some of the self-employed may be in family businesses and others may have more flexibility in choosing their work location. Respondents who are employees compared to not employed are less likely to coreside or live nearby elderly – these respondents may have less choice over where they live, which may be determined by where they work. As anticipated, the financial power of the elderly has a strong positive effect on living arrangements (Table 2). If the elderly person owns the title to the respondents’ residence, the respondents are five times more likely to live
254
I¸SIK A. AYTAC ¸
together or three times more likely to live nearby, net of all other factors.10 This is not surprising given the high costs of housing, especially in the metropolitan and other large urban areas (Turkish Census of Population 1985: 171). As DaVanzo and Chan (1994), report higher housing costs increase the likelihood of intergenerational coresidence in Malaysia. Furthermore, if the elderly owns a larger share of he family assets than the respondent, then they are 2.9 times more likely to coreside with them, although there is not a significant effect of assets on living nearby. It may be that those who coreside, in addition to having the benefit of a rent-free residence, will have other benefits through their access to elderly resources more than those who live nearby. Reasons for coresiding. In the section above, we discussed how regional economic development, individual modernity and religiosity affect living arrangements. Next we discuss the reasons why respondents coreside with elderly, that is, the factors that affect this decision. Table 3 presents the multinomial logit estimates where we contrast those who do not coreside with (1) those who coreside due to tradition and religious expectations, and (2) those who coreside because the elderly need their help, support and care. The results show a strong relationship between economic development and the reported reasons respondents give for coresiding with elderly. Compared to those who live in the least developed eastern region (comparison group), those who live in the other regions are significantly less likely to report that they coreside with elderly due to tradition. Note that there is no effect of regional development on the likelihood of coresiding with elderly if the elderly person needs help, with the exception of the north where they are almost 1.4 times more likely to coreside due to elderly need compared to the east. The effect of living in a metropolitan area or in another urban area is similar for the different reasons for coresiding with elderly (Table 3). The strongest contrast is in coresiding due to tradition for metropolitan versus rural areas (−1.376, p < 0.001). The effect of education and religiosity are also similar for the different reasons for coresiding with elderly. Respondents with high school or college education are significantly less likely to coreside with elderly due to tradition and due to elderly need. Not surprisingly, more religious respondents are 1.7 times more likely to coreside due to tradition and 1.4 times more likely to coreside if elderly need their help compared to less religious respondents. Briefly, the effect of the other predictor variables show that self employed respondents are more likely to coreside if elderly need their help but are not more likely to coreside due to tradition (Table 3). Some of the self employed may be in family business and may need to care for the elderly to pursue the
255
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
Table 3. Multinomial logit estimation of reasons for coresiding with older family members Tradition
Need
Coef. (SE)
Exp. (b)
Coef. (SE)
Exp. (b)
−0.837∗∗∗ (0.124)
0.432
−0.070 (0.137)
0.932
South
−1.182∗∗∗ (0.137)
0.306
−0.260 (0.146)
0.770
Central
−0.495∗∗∗ (0.116)
0.609
0.173 (0.133)
1.189
North
−0.749∗∗∗ (0.153)
0.472
0.302∗ (0.154)
1.353
Metropolitan
−1.376∗∗∗ (0.162)
0.252
−0.660∗∗∗ (0.133)
0.516
Other urban areas
−0.401∗∗∗ (0.094)
0.669
−0.433∗∗∗ (0.097)
0.648
−0.235 (0.159)
0.790
0.114 (0.130)
1.121
High school
−0.354∗ (0.151)
0.701
−0.347∗ (0.138)
0.706
College
−0.374∗ (0.174)
0.687
−0.643∗∗∗ (0.168)
0.525
1.684
0.300∗∗∗ (0.093)
1.350
−0.086 (0.117)
0.916
0.338∗ (0.126)
1.302
−0.512∗∗∗ (0.126)
0.599
0.137 (0.129)
1.147
0.646∗∗∗ (0.108)
1.908
0.570∗∗∗ (0.109)
1.769
1.038∗∗∗ (0.118)
2.824
0.592∗∗∗ 0.130
1.808
Location Region West
Education Junior high
Religiosity Sex segregation
Employment status Self employed Employee
Elderly financial power Elderly owns R’s residence Elderly assets
0.521∗∗∗ (0.089)
256
I¸SIK A. AYTAC ¸
Table 3 (continued) Tradition Coef. (SE) Control variables Kids LT 12 Migrate Age Age-squared
Need
Exp. (b)
Coef. (SE)
Exp. (b)
0.407∗∗∗ (0.104)
1.502
−0.152 (0.094)
0.858
0.096 (0.110)
1.101
−0.071 (0.101)
0.931
−0.054∗∗ (0.018)
0.946
0.030 (0.021)
1.030
1.001
−0.001 (0.001)
0.999
0.001∗∗∗ (0.001)
Intercept
−0.971∗ (0.459)
Chi-square df N
1167.26 36 8303
−3.062∗∗∗ (1.098)
∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001
family business. Furthermore, if the elderly are better off financially relative to the householder, the respondents are much more likely to coreside with them. The odds of living with elderly due to tradition is 2.8 times higher if the elderly owns the largest share of the family assets and almost twice as likely if the elderly own the respondents residence. Both indicators of the elderly financial resources also raise the odds of living with elderly when elderly need help by around 1.8 times. Predicted probabilities. Figures 1 through 5 show the predicted probabilities (based on Tables 2 and 3) for living arrangements (living together or living nearby) and for the reasons for coresiding (tradition versus elderly need) by region, urban-rural residence and education. As seen in Figure 1, net of all other variables, the probability of living nearby elderly is much higher than the probability of coresiding. The probability of coresiding with elderly is lowest in the two more developed regions of the west and south and increases for regions that are less developed, reaching a peak for the least developed eastern region. Contrary to what we observe for the regional variation in coresidence, the probability of living nearby elderly is similar for four regions, showing a higher likelihood for the south. Interestingly, the
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
257
Figure 1. Living arrangements by region: Predicted probabilities.
Figure 2. Living arrangements by education: predicted probabilities.
probability of living nearby elderly is highest in the south with the lowest probability of coresiding, suggesting that there is substitution of coresidence with living nearby. Thus, Turkish families maintain close family ties regardless of whether they coreside or live nearby. Similar patterns are observed by education (Figure 2). While the probability of living nearby is fairly steady for different educational attainment, there is a steady decline in the probability of coresidence by higher levels of education. Figure 3 shows the variation in cultural values by region. The probability of living with elderly due to tradition is lower among the more developed regions, considerably higher in the central and north regions and much higher in the least developed eastern region. Note that in all regions except for the east, the probability of coresiding due to tradition or elderly need are at similar levels or the latter is higher. Not surprisingly, in the least developed eastern region, the tradition outweighs the elderly need – respondents coreside with elderly regardless of whether the elderly need help or not. In addition, in the north, south and the west, the probability of coresiding due to elderly need is higher compared to tradition indicating that even though traditional values are
258
I¸SIK A. AYTAC ¸
Figure 3. Reasons for coresiding by region: Predicted probabilities.
Figure 4. Reasons for coresiding by education: Predicted probabilities.
changing, families are there to help their elderly members when the elderly is in need. Figure 4 shows the predicted probabilities for reasons for coresiding by education. As expected, the probability of coresiding with elderly due to tradition or elderly need is lower among the more educated, although the junior high or high school graduates are more likely to coreside if the elderly need their help than due to tradition. Finally, Figure 5 shows the living arrangements and reasons for coresiding by urban-rural residence. In both urban and rural areas probability of living nearby is high, while coresidence is only high in rural areas. The reasons for coresiding also differs by urban and rural residence. While in rural areas they are more likely to coreside with elderly due to tradition, in urban areas, elderly need is more prevalent.
Conclusion Despite the growing research on living relations of elderly, we know very little about intergenerational relations in Turkey. We take a step towards this
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
259
Figure 5. Living arrangements and reasons for coresiding by location.
goal by examining intergenerational living arrangements in Turkey, a predominantly Muslim country undergoing rapid legal, economic, and social changes. Coresidence may be seen either as loss of privacy or as providing companionship, social support, and financial support (for a detailed discussion on the cost and benefits of coresidence, see Burch & Matthews 1987: DaVanzo & Chan 1994). There is a positive relationship between economic resources and independent living in the USA (Mutchler & Burr 1991; Bresford & Rivlin 1966; Michael, Fuchs & Scott 1980; Pampel 1983). In contrast to many east and southeast Asian countries, where around 60% to 70% of elderly coreside with adult children (Albert & Cattell 1994), the incidence of coresidence with elderly is low in Turkey (around one fifth), while the incidence of living nearby is quite high (around one half), suggesting that Turkish families prefer to have privacy while maintaining close family ties. Our results are supportive of the view that the impact of the processes of development are not the same for all countries, with specific cultures and family and kinship patterns playing an important role in the outcome of intergenerational relations. Similar results are observed for Japan, a much more economically developed society, where two-thirds of elderly coreside with their children. Thus, the impact of economic development on family relations is not unilinear as was suggested by modernization theorists (Cowgill & Holmes 1972). As hypothesized, there is significant variation in coresidence and in the reasons for coresiding with elderly by regional development, urbanization, individual modernity and religiosity. The probability of living with elderly is much higher for rural areas and less developed regions, as well as for the more traditional respondents, specifically those with lower levels of education and the more fundamentalist Muslims. Nevertheless, this does not imply that increased levels of development, urbanization, secularism, and educational attainment mean a decline in family ties. Turkey does not seem to follow the
260
I¸SIK A. AYTAC ¸
path taken by the West. Instead, the results show the importance of family ties in Turkish society as around 70 percent either coreside or live nearby elderly. Data limitations impact this research in several ways. First, our primary focus on the impact of the characteristics of the respondent on living arrangements and on the reasons for coresiding with older family members was inevitable because we have limited information on the characteristics of the older family members. As others have shown it is important to take into consideration both the characteristics of the young and the old generations (Acquilino 1990; Casterline et al. 1991; DaVanzo & Chan 1994; Elman & Uhlenberg 1995; Ofstedal 1994; Wolf & Soldo 1988). Second, with this data set we are unable to factor in information on all available kin and their characteristics. Finally, in a rapidly changing society such as Turkey, cross-sectional data are of limited value for studying trends such as among those who migrate from rural to urban settings. This final limitation has serious repercussions for the broader theoretical relevance of this research. Although, the impact of industrialization, urbanization, and the market economy on family relations have been the focus of major sociological theories (such as modernization, Marxist and neo-Marxian theories), testing these theories with crossectional data is beyond the scope of this research. Despite these limitations, this research is the most recent nationwide study of the present state of intergenerational living arrangements in Turkey and as such will serve as baseline information for future research. In order to ensure the wellbeing of the older generations in the future, we need to understand the determinants of intergenerational living arrangements today and develop appropriate social policies for the future. At present the policy of Turkish government is based on the belief that the family is responsible for the elderly. However, with the rapid increase in women’s labor force participation due to increased levels of education and economic development, families may not be able to shoulder to sole responsibility of elderly care. Our results suggest that given the strong family ties in Turkey, nursing homes for the elderly are not the optimal solution. Policies that enable families to maintain close ties but help alleviate the burden of elderly care may be more appropriate for the cultural setting of Turkish society.
Acknowledgments This research was completed while I¸sık Aytaç was a postdoctoral fellow at the Population Research Center at the University of Chicago. An earlier version was presented at the annual meeting of he Population Association of America, San Francisco, California, April 1995. Support for this research was provided
INTERGENERATIONAL LIVING ARRANGEMENTS IN TURKEY
261
by National Institute of Aging Training Grant No. T32 HD07302 to the University of Chicago. The data are provided by the Turkish Republic Prime Ministry State Planning Organization and Department of Statistical Institute. I would like to thank to Fay Booker for data preparation, Linda Waite, Bruce Rankin and two anonymous reviewers for helpful comments. Notes 1. The survey asks about living arrangements with ‘older family members’. In the Turkish context, older family members will include primarily parents and parents-in-laws, though occasionally, respondents may have included older family members, such as aunts and uncles. 2. Throughout the paper the term modernization is used to signify the various social processes of urbanization, industrialization, mass education, secularization, and economic development. It is not used to reflect the perspective of modernization theory per se, which assumes that all countries follow the same linear developmental path (Cowgill & Holmes 1972). 3. Comparing living arrangements in different countries is not without difficulty. De Vos and Holden (1988) report that different measures of living arrangements across countries may produce different results in cross-national comparisons. 4. Recently there has been a resurgence of Islamic fundamentalism in Turkey, which is primarily urban-based though it has rural backing (Gocek & Schuman 1993). The Islamic Prosperity Party (Refah Partisi) received around 20 percent of votes during the 1995 national elections. We do not expect this movement to have much of an impact on the results of this research which uses a survey conducted in 1988. 5. Another response category under the reasons for coresiding with elderly included coresiding due to the respondent’s needs. Due to a small proportion of answers in this category (around two percent), they were excluded. 6. The massive development project undergoing in south eastern Turkey did not have an effect on the area prior to this survey. 7. In some circumstances, there may be a possible endogeneity between employment status and living arrangements with older family members. That is, not only will employment have an affect on living arrangements, but living arrangements may also affect employment status. For instance, once one starts to coreside with an older family member, they may discontinue working for they may no longer need to work for financial reasons or they may need to take care of the older family member. For our sample of married men, we do not anticipate endogeneity to be an issue. Due to traditional gender roles in Turkey, men will most likely continue to work as the household breadwinner, and will be very unlikely to actually do the physical care of the older family members, especially if they are married. Therefore, for our sample, any effect of living arrangements on employment may be for the wives of the respondents, but not for the employment of the married male respondents. 8. It is important to note here that due to the tax laws, some parents transfer the ownership of their residences to their adult children early on, in order to avoid the high taxes adult children will have to pay when they inherit property. Therefore, the measure we use here should be considered as a very conservative measure of elderly financial power as some elderly may have the power even if the title of a property is under their children’s name.
262
REFERENCES
9. De Vos (1990) also reports no urban and rural difference in extended family living in six Latin American countries. 10. In our sample, the respondents’ residence is owned by the elderly for 37 percent of the coresider households, 52 percent of those with elderly nearby, and for 11 percent of those with elderly far away.
Appendix Table A1. Basic demographic characteristics by region, residence and education Total fertility ratea
Infant mortality
Height for ageb
Region West South Central North East
2.0 2.4 2.4 3.2 4.4
42.7 55.4 57.9 44.2 60.0
1.6 3.7 5.1 5.2 12.7
Residence Urban Rural
2.4 3.1
44.0 65.4
3.7 9.2
Education Less than primary Primary and above Less than secondary Secondary and above
4.2 — 2.4 1.7
68.0 43.6 — —
11.6 — 3.2 0.8
a Rate for women age 15–49 years. b Percentage of children under five years who are classified as undernourished, measured
as below three standard deviation units from the median of the NCHS/CDC/WHO international reference population. SOURCE: Ministry of Health [Turkey], Hacettepe University Institute of Population Studies, and Macro International Inc., Turkish demographic and health survey 1993 (pp. 25, 81, 115) Ankara, Turkey, 1994.
References Abadan-Unat, N. (1986). Women in the developing world: Evidence from Turkey. Denver, CO: University of Denver. Acquilino, W. S. (1990). The likelihood of parent-adult child coresidence: Effects of family structure and parental characteristics, Journal of Marriage and the Family 52: 405–419. Albert, S. M. & Cattell, M. G. (1994). Old age in global perspective: Cross-cultural and cross-national views. New York: G. K. Hall & Co.
REFERENCES
263
Bressford, J. C. & Rivlin, A. M. (1966). Privacy, poverty and old age, Demography 3: 247– 258. Burch, T. K. & Matthews, B. J. (1987). Household formation in developed societies, Population and Development Review 13(3): 495–511. Cagirici, M. (1992). Islam dusuncesinde aile ahlaki. In T. C. Basbakanlik Aile Arastirma Kurumu (ed.), Sosyo-Kulturel Degisme Surecinde Turk Ailesi, Vol. 1 (pp. 341–352). Caldwell, J. C. (1980). Mass education as a determinant of the timing of fertility decline, Population and Development Review 6(2): 225–255. Casterline, J. B., Williams, L., Hermalin, A., Chang, M. C., Chayovan, N., Cheung, P., Domingo, L., Knodel, J. & Ofstedal, M. B. (1991). Differences in the living arrangements of the elderly in four Asian countries: The interplay of constraints and preferences, Comparative Study of the Elderly in Asia, Research Reports No. 91–10. Population Studies Center, University of Michigan. Casterline, J. B., Williams, L., Chang, M. C., Chayovan, N., Cheung, P. & Domingo, L. (1992). Living arrangements of the elderly in four Asian countries. Paper presented at the National Institute and Aging, Summer Institute in Aging, RAND, Santa Monica, California. Casterline, J. B., Chang, M. C., Chayovan, N. & Domingo, L. (1993). The density of parentchild co-residence in East and Southeast Asia. Paper presented at the annual meetings of the Population Association of America, Cincinnati, Ohio. Cowgill, D. O. & Holmes, L. D. (1972). Aging and modernization. New York: Meredith Corporation. DaVanzo, J. & Chan, A. (1994). Living arrangements of older Malaysians: Who coresides with their adult children, Demography 31(1): 95–113. Davis, K. & van den Oever, P. (1981). Age relations and public policy in advanced industrial societies, Population and Development Review 7(1): 1–18. De Vos, S. (1990). Extended family living among older people in six Latin American countries, Journal of Gerontology: Social Sciences 45(3): S87–94. De Vos, S. & Holden, K. (1988). Measures comparing living arrangements of the elderly: An assessment, Population and Development Review 14(4): 688–704. Duben, A. (1982). The significance of family. In Kagitcibasi (ed.), Sex roles, family and community in Turkey (pp. 73–99). Bloomington: Indiana University, Turkish Studies. Elman, C. & Uhlenberg, P. (1995). Co-residence in the early twentieth century: Elderly women in the United States and their children, Population Studies 49: 501–517. Fox, G. L. (1975). Love match and arranged marriage in a modernizing nation: Mate selection in Ankara, Turkey, Journal of Marriage and Family 37: 180–193. Gocek, M. & Schuman, H. (1993). Defining the boundaries of Islamic fundamentalism: The case of modern Turkey, A proposal submitted to National Science Foundation. Hancioglu, A. (1993). Introduction. Demographic and health survey: Turkey. Hacettepe University, Institute of Population Studies. Ilbas, Z. (1991). Aile – akrabalik iliskileri ve degismeler uzerine bir arastirma. Aile yazilari 2: Kulturel degerler ve sosyal degisme. Ankara: Aile Arastirma Kurumu Baskanligi Yayini. Kandiyoti, D. (1977). Sex roles and social change: A comparative appraisal of Turkey’s women, Signs 3: 57–73. Lesthaeghe, R. & Surkyn, J. (1988). Cultural dynamics and economic theories of fertility, Population and Development Review 14(1): 1–45. Magnarella, P. J. (1973). The reception of Swiss family law in Turkey, Anthropological Quarterly 46: 100–116. Martin, G. L. (1989). Living arrangements of the elderly in Fiji, Korea, Malaysia, and the Philippines, Demography 26(4): 627–643.
264
REFERENCES
Martin. G. L. (1990). Changing intergenerational family relations in East Asia, The Annals of the American Academy of Political and Social Science 510: 102–114. Meric, U. (1991). Islamda aile. Aile yazilari 2: Kulturel degerler ve sosyal degisme (pp. 423– 444). Ankara: Basbakanlik Aile Arastirma Kurumu. Michael, R. T., Fuchs, V. & Scott, S. R. (1980). Changes in the propensity to live alone: 1950–1976, Demography 17: 39–56. Mumcu, A. (1986). Tarih acisindan Turk devriminin temelleri ve gelisimi. Ankara: Inkilap Kitabevi. Mutchler, J. E. & Burr, J. A. (1991). A longitudinal analysis of households and nonhousehold living arrangements in later life, Demography 28: 375–390. Ofstedal, M. B. (1994). An application of the simultaneous logit model: Coresidence choices of elderly parents and adult children in Taiwan. Paper presented at the National Institute of Aging, Summer Institute in Aging, RAND, Santa Monica, California. Ogawa, N. & Retherford, R. D. (1993). Care of the elderly in Japan: Changing norms and expectations, Journal of Marriage and the Family 55(3): 585–597. Ozbay, F. (1990). Turkiye’ de aile ve hane yapisi: Dun, bugun, yarin. Marmara Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi. Cilt VII, Sayi 1–2. Pampel, F. C. (1983). Changes in the propensity to live alone: Evidence from consecutive cross-sectional surveys, 1960–1976, Demography 20: 433–447. Smelser, N. (1959). Social change and the industrial revolution. Chicago: University of Chicago Press. Timur, S. (1972). Turkiye’de Aile Yapisi. Ankara: Hacettepe University. Turk Aile Yapisi Arastirmasi, 1992. Ankara: DPT, Sosyal Planlama Genel Mudurlugu, p. 64. Thornton, A. & Fricke, T. (1987). Social change and the family: Comparative perspectives from the West, China, and south Asia, Sociological Forum 2(4): 746–779. Toros, A. (1993). Fertility, In: Demographic and health survey: Turkey. Hacettepe University, Institute of Population Studies. Turkdogan, O. (1992). Turk Ailesinin Genel Yapisi. In T. C. Basbakanlik Aile Arastirma Kurumu (ed.), Sosyo-Kulturel Degisme Surecinde Turk Ailesi, Vol. 1 (pp. 29–66). Turkish Census of Population: Social and Economic Characteristics of Population (Genel Nufus Sayimi: Nufusun Sosyal ve Ekonomik Nitelikleri). 1985. T. C. Basbakanlik Devlet Istatistik Enstitusu. Turkoz E., ed. (1985). Family in Turkish society: Sociological and legal studies (pp. 43–77). Ankara: Turkish Social Science Association. Wolf, D. A. & Soldo, B. J. (1988). Household composition choices of older unmarried women, Demography 25(3): 387–403. Yener, S. (1977). 1965–1970 doneminde iller arasi gocler ve goc edenlerin nitelikleri. Ankara: State Planning Organization Publication. Address for correspondence: I¸sik A. Aytaç, New England Research Institutes Inc., 9 Galen Street, Watertown, MA 02472, USA Phone: (617) 923-7747; Fax: (617) 926-8246; E-mail:
[email protected]