Introduction
Massive rural-to-urban migration and population ageing are reshaping the face of China. The volume of Chinese rural-to-urban migration more than quadrupled from 34 million in 1989 to 147 million in 2005 (Cai Reference Cai2002; National Statistics Bureaus of China 2006), the later total being roughly one-quarter of the Chinese rural population (Goodkind and West Reference Goodkind and West2002). At the same time, the Chinese population is ageing rapidly both absolutely and relatively. The share of the population aged 65 or more years will increase from seven per cent in 2000 to about 20 per cent in 2040 (United Nations Organisation 2005), and it is projected that by 2050, 31 per cent of the Chinese, or 437 million people, will be older people, twice the number of children (Zhang Reference Zhang2007). Given that more than 64 per cent of the Chinese live in rural areas and that many young people move to cities, population ageing is most marked in the country areas (National Statistics Bureaus of China 2001).
Together, the two trends exacerbate the age-structure imbalances of China's rural areas. It is estimated that the number of rural older people with migrant children reached over 18 million in 2000 (Du et al. Reference Du, Ding, Li and Gui2004). As families, especially adult children, play a crucial role in providing various forms of support for Chinese older people (Chi and Chou Reference Chi and Chou2001; Whyte Reference Whyte2003; Xu Reference Xu2001), the exodus of the young (especially young men) from rural areas has led to growing concern over the support and wellbeing of older people left behind (Silverstein, Cong and Li Reference Silverstein, Cong and Li2006). The geographical separation of adult children and elderly parents is a structural barrier and impedes close inter-generational exchange (King and Vullnetari Reference King and Vullnetari2006; Mallee Reference Mallee, West and Zhao2000; Wilson Reference Wilson2002). Older parents left behind may also experience reduced psychological wellbeing through feelings of loneliness and abandonment in the absence of children (Du et al. Reference Du, Ding, Li and Gui2004). In addition, since pension systems have not yet been extended to rural areas and support from local government is not available for most older people in rural China (Xu Reference Xu2001), the combination of out-migration and ageing also exerts pressures on the health care and social insurance systems.
Despite the growing concerns over the support and wellbeing of older people in rural areas, recent studies from China and other Asian countries suggest that the negative impact of the massive out-migration on older adults may have been overstated. For instance, some research has found that with the recent rapid economic and social changes, family roles and the commitment among most young adults to provide old-age support for their parents endure (Bongaarts and Zimmer Reference Bongaarts and Zimmer2002; Keasberry Reference Keasberry2001; Knodel et al. Reference Knodel, Friedman, Anh and Cuong2000; Lin Reference Lin2002; Zimmer and Kim Reference Zimmer and Kim2002; Zimmer and Kwong Reference Zimmer and Kwong2003). The consistency of inter-generational support through such social changes varies by type of support (e.g. economic or instrumental support) (Lin Reference Lin2002). In terms of the psychological wellbeing of older parents left behind, some researchers suggest that the increasing involvement of the remaining parents in farm work, household chores and grand-parenting, in the absence of their migrant children, may provide older people with continuing status and roles, and thus enhance rather than impair their wellbeing (Becker et al. Reference Becker, Beyene, Newsom and Mayen2003; Chen and Silverstein Reference Chen and Silverstein2000). To understand more fully this issue, this study investigated how the out-migration of adult children affected over two years the monetary, instrumental and emotional forms of inter-generational support among rural Chinese older people, and also their psychological wellbeing, specifically depression and life satisfaction.
Out-migration and elderly parents in rural areas: empirical findings
Although family spatial separation and its implications for older adults have been investigated since the 1960s (e.g. Greenwell and Bengston Reference Greenwell and Bengston1997; Litwak Reference Litwak1960; Smith Reference Smith1998), most research has been in western countries (Hank Reference Hank2005; Lin and Rogerson Reference Lin and Rogerson1995; Rogerson, Weng and Lin Reference Rogerson, Weng and Lin1993). Only in the last decade have studies begun on the relationship between out-migration of adult children and the wellbeing of remaining parents in developing countries. One of the most direct and profound consequences of out-migration is the change in family structure and living arrangements. The large-scale out-migration of young adults is believed to reduce average household size in labour-sending areas (Barker Reference Barker1994; Wang and Li Reference Wang and Li2005). In China, a relatively high proportion of the rural elderly are found to live alone or in ‘empty-nest’ and ‘skip-generation’ households due to large-scale out-migration (Du et al. Reference Du, Ding, Li and Gui2004). Because household size and living arrangements are key variables in family support (Keasberry Reference Keasberry2001; King and Vullnetari Reference King and Vullnetari2006; Wolff and Dimova Reference Wolff and Dimova2006), such a change challenges the traditional role of the family in elderly care. Studies in China show that since community services are underdeveloped in many rural areas, the needs of elderly parents without nearby children are often under-served, putting them at a risk of declining health (Chen and Silverstein Reference Chen and Silverstein2000; Silverstein, Cong and Li Reference Silverstein, Cong and Li2006).
A general theoretical proposition is that out-migration results in diminished informal support resources among the elderly parents who remain. ‘Modernisation and ageing theory’ also points to a relationship between urbanisation with industrialisation and decreased old-age care, partly as a result of the increased geographical distance between family members, and partly because of a decline in family orientation (Aboderin Reference Aboderin2004; Cowgill and Holmes Reference Cowgill and Holmes1972; Kuhn Reference Kuhn, Jatrana, Toyota and Yeoh2005). These propositions focus on one particular aspect of inter-generational support, namely hands-on support or personal care of older parents. Other types of support, such as monetary and emotional transfers, may be affected differently by social changes in general and by out-migration in particular, but are rarely distinguished in these theories. In addition, they tend to overlook the diversity of local conditions and their relation to family forms and support patterns. As Asian cultures emphasise the collective family and filial obligation, the forces of modernisation, urbanisation and population mobility may not erode support as severely as expected. As seen in the following review of studies in developing countries, a more nuanced picture is emerging of the changes in inter-generational support following out-migration.
The out-migration of adult children changes both the type and extent of family support available to the remaining parents. It could significantly reduce instrumental support (i.e. hands-on support with personal care and housework) (Du et al. Reference Du, Ding, Li and Gui2004), but it has been found that the lack of instrumental support is not related to the psychological wellbeing of the older parents left behind (Silverstein, Cong and Li Reference Silverstein, Cong and Li2006). In other words, even with less instrumental support, parents with migrant children can maintain good psychological states. Numerous studies of financial support received by rural elders consistently have reported that most migrant children compensate for their separation from parents by regularly sending back remittances (Du et al. Reference Du, Ding, Li and Gui2004; Keasberry Reference Keasberry2001; Zhang and Wu Reference Zhang and Wu2003). Factors such as the number of children, children's income, parents' socio-economic status and agriculture-based self-sufficiency associate with the financial support received by the parents who remain (Kreager Reference Kreager2006; Zhang and Wu Reference Zhang and Wu2003). Except for a study in rural Albania (King and Vullnetari Reference King and Vullnetari2006), where elderly parents reported an enhanced sense of economic insecurity following the out-migration of their children, migrant children's remittances are believed to transform positively the lives of families and to enhance the material wellbeing of elderly parents (Du et al. Reference Du, Ding, Li and Gui2004; Huang Reference Huang1997; Knodel and Saengtienchai Reference Knodel and Saengtienchai2005; Zhang and Wu Reference Zhang and Wu2003). The positive impact of remittances may outweigh any negative influence of out-migration on the parents' psychological wellbeing (Wolff and Dimova Reference Wolff and Dimova2006). For instance, it has been found in rural China that the value of the financial transfers received from children is inversely related to depressive symptoms among older people (Silverstein, Cong and Li Reference Silverstein, Cong and Li2006).
Several qualitative studies have revealed that the emotional support provided by migrant children to parents is more in evidence than day-to-day hands-on care. Although far away, most migrant children still provide long-distance emotional support and actively participate in care-giving for their parents (Baldock Reference Baldock2000; Kuhn Reference Kuhn1999). In Asian countries particularly, it is reported that extended family relations are maintained over great geographical distances and that most migrant children maintain a commitment to filial piety and to parental wellbeing (Knodel et al. Reference Knodel, Friedman, Anh and Cuong2000; Knodel and Saengtienchai Reference Knodel and Saengtienchai2005). Thus, even if the spatial dispersion of a family reduces the children's hands-on support of older adults, emotional ties and support are maintained. Taken together, the reported findings suggest that changes in inter-generational support after out-migration vary by type of support, and that the prevailing proposition of declining old-age support may not be warranted. Whereas instrumental support may be diminished by out-migration, monetary and emotional support from migrant children can persist or even be enhanced. As Kreager (Reference Kreager2006) noted, the children's locations may not be a general proxy for support of elderly parents, and those who move away from their home are not necessarily lost to the support network. In addition, support from non-migrant children is neglected in most existing studies, and a holistic approach that sees the non-migrant children as an alternative support resource is rare. As the majority of elderly people in Asian countries still live close to at least some of their children and receive support from them (Beard and Kunharibowo Reference Beard and Kunharibowo2001; Frankenberg, Chan and Ofstedal Reference Frankenberg, Chan and Ofstedal2002; Keasberry Reference Keasberry2001; Knodel and Debavalya Reference Knodel and Debavalya1997), the absence of one or two migrant children may not lead to a collapse of the parents' support network. The present study considered the support from both migrant and non-migrant children and sought a more comprehensive view of the implications of out-migration. Three forms of support that may be affected differently by out-migration were distinguished.
The psychological wellbeing of elderly parents left behind
Very few studies have investigated the direct effect of out-migration on the mental health of older parents. The existing literature generally points to the negative impact of children's out-migration on the parents' psychological wellbeing. Although monetary support from migrant children tends to have a positive impact on the psychological states of older parents (Silverstein, Cong and Li Reference Silverstein, Cong and Li2006), those left behind reported an increased sense of loss, loneliness, isolation, depression and abandonment in the absence of their children (Du et al. Reference Du, Ding, Li and Gui2004; King and Vullnetari Reference King and Vullnetari2006; Miltiades Reference Miltiades2002; Silverstein, Cong and Li Reference Silverstein, Cong and Li2006). The sense of loss and depression might be a result of unfulfilled expectations for closer or more frequent contacts with children (Miltiades Reference Miltiades2002). The negative affect might be explained by the broken cultural norms and inter-generational solidarity between migrant children and older parents (Mjelde-Mossey, Chi and Lou Reference Mjelde-Mossey, Chi and Lou2006; Xu Reference Xu2001). Most of the research to date has been cross-sectional. Without controlling for a parent's initial health, it is hard to tell whether the out-migration of children induces changes in parental wellbeing, or whether the children of elderly parents of certain health status are more likely to move away. In addition, most studies have been small and have relied on qualitative interviews to explore the subjective feelings of parents. Very few attempts have been made to test quantitatively the hypothesised negative relationship between out-migration and parents' mental health. This study addressed the following specific research questions with particular reference to rural China:
Q1: Does the out-migration of adult children affect the monetary, instrumental and emotional support received by older parents, and how?
Q2: Does the out-migration of adult children affect their parents' psychological wellbeing as measured by depression and life satisfaction, and how?
Q3: After controlling for variations in the three forms of inter-generational support, does the out-migration of adult children affect their parents' psychological wellbeing as measured by depression and life satisfaction, and how?
Design and methods
The sample and measures
The data derived from the 2001 and 2003 waves of a longitudinal study ‘The wellbeing of older people in Anhui Province’ that was conducted jointly by the Schools of Gerontology and Social Work at the University of Southern California and the Population Research Institute of Xi'an Jiaotong University, China. Older adults aged 60 or more years in the rural region of Chaohu, Anhui Province, China, were randomly selected to explore how a series of social changes including rural-to-urban migration affect Chinese rural elders. The region was chosen for the relatively high proportion of out-migration among working-age adults. In 2001, the Wave 1 stratified multi-stage sampling selected 1,802 potential respondents from the administrative records of 72 randomly selected villages in six rural townships in Chaohu. Of this number, 1,696 completed the survey, giving a response rate of 94.1 per cent. In 2003, Wave 2 follow-up interviews were carried out with 1,391 individuals, 82 per cent of the original participants. After excluding cases with missing values on the migration status of children, the final analysis sample was 1,237.
For the dependent variables, psychological wellbeing was operationalised as life satisfaction and depression. The life-satisfaction scale was adapted from Diener et al. (Reference Diener, Emmons, Larsen and Griffin1985) and has seven items that ask respondents whether they agree or disagree with statements that indicate content and discontent with their current lives (‘better life than most’, ‘satisfied with life’, ‘interesting life’, ‘best years of life’, ‘life meets expectations’, ‘life is tedious’ and ‘life is tiring’). After coding the items in the same direction, we summed the seven scores, producing a measure that ranged from zero to seven, with higher scores reflecting higher levels of life satisfaction. The scale has been used extensively and proven reliable in large cross-cultural analyses of life satisfaction (Goodwin, Cook and Yung Reference Goodwin, Cook and Yung2001; Pavot and Diener Reference Pavot and Diener1993). It has demonstrated validity in the Chinese population (Goodwin, Cook and Yung Reference Goodwin, Cook and Yung2001; Leung and Leung Reference Leung and Leung1992; Shek Reference Shek2002). The Cronbach's alpha scores for the Chinese version of the scale were 0.79 and 0.84 in 2001 and 2003, respectively.
We measured depression with nine questions from the Center for Epidemiologic Studies–Depression (CES-D) scale (Radloff Reference Radloff1997). Among the nine items, three indicated feelings of positive affect (happiness, enjoyment, pleasantness), two indicated feelings of negative affect (lonely, upset), two indicated feelings of marginalisation (useless, nothing to do), and two indicated somatic symptoms (poor appetite, has trouble sleeping). The respondents rated the frequency with which each symptom had been experienced during the past week (‘0’=rarely or none of the time, ‘1’=some of the time, ‘2’=most of the time). After coding the items in the same direction, we summed the nine scores, resulting in a depression scale ranging from 0 to 18, with a higher score indicating greater depression. The CES-D scale has been used widely among the Chinese in mainland China, Hong Kong and the United States, and has been confirmed as reliable and valid cross-culturally (Cheung, Liu and Yip Reference Cheung, Liu and Yip2007; Mui, Burnette and Chen Reference Mui, Burnette and Chen2001; Zhang and Norvilitis Reference Zhang and Norvilitis2002). The short form CES-D also has satisfactory content and temporal reliability among Chinese older adults (Boey Reference Boey1999; Cheng and Chan Reference Cheng and Chan2005). In the present study, the Cronbach's alpha scores for this Chinese version of the scale were 0.77 and 0.80 in 2001 and 2003, respectively.
Turning to the independent variables, two represented the presence of out-migrant children. A dummy variable indicated whether a respondent had any migrant children (1) or none (0). The second variable is a count of the number of the respondent's children who were migrants. Migrant children were defined as those who lived in different counties or cities from their parents (Goodkind and West Reference Goodkind and West2002). We measured inter-generational support with three variables: (a) monetary support, (b) instrumental support, and (c) emotional cohesion. The three scores considered support from both migrant and non-migrant children. Monetary support was indicated by the total amount (in renminbi or RMB) that the parent had received from all their children during the previous year (n.b. 100 RMB=US$12). Because the distribution of monetary support was highly skewed, we used the logarithm of the RMB amount (+1) to represent this varaible. We assessed instrumental support by the number of offspring from whom the older parent had received hands-on help in either household tasks (e.g. cleaning the house, washing clothes and dishes) or personal care tasks (e.g. bathing and dressing) during the previous 12 months. Offspring referred to son(s), daughter(s), son(s)-in-law and daughter(s)-in-law.
We measured emotional cohesion by three questions that assessed the quality of each parent–child relationship that was adapted from the ‘Intergenerational Solidarity Inventory’ (seeMangen, Bengtson and Landry Reference Mangen, Bengtson and Landry1992). The three questions were: (a) ‘Taking everything into consideration, how close do you feel to (this child)?’, (b) ‘How much do you feel that (this child) would be willing to listen when you need to talk about your worries and problems?’, and (c) ‘Overall, how well do you and (this child) get along together?’ For each question, the score ranged from 0 to 2 (‘0’=not at all close not at all/not at all well, ‘1’=somewhat close/somewhat/somewhat well, ‘2’=very close/very much/very well). The additive score for each child ranged from 0 to 6. We took the average score across all children for each parent. The Cronbach's alpha reliability scores for this scale were 0.94 and 0.95 for the two waves.
The control variables were the socio-demographic characteristics of the older adults (age, gender, marital status, education attainment, income, living arrangement, total number of children and functional health difficulties). Age was measured in years, and gender, marital status and education attainment were coded as dummy variables (female=‘1’, married persons=‘1’; received some education=‘1’). Income was indicated by the total income that the respondent (and spouse, if married) had received from work or a pension during the previous year. Because the distribution of income was highly skewed, we used the logarithm of the RMB amount (+1) to represent the income. As the focus of this study was support from children, we coded living arrangement as a categorical variable (‘0’=living alone (reference group), ‘1’=living with offspring (including children, grandchildren, son(s)-in-law and daughter(s)-in-law), ‘2’=living with others, i.e. spouses, parents, siblings or other relatives). A dummy variable captured the change in living arrangement over time (‘1’=those who started to live alone between 2001 and 2003, ‘0’=others).
Health status was represented by functional health difficulties. The respondents indicated their level of difficulty (‘0’=none, ‘1’=some, ‘2’=cannot do it without help) in performing 15 tasks that represented the activities of daily living (dressing or undressing, bathing, walking around the room, getting out of bed, standing up from a chair, going to the toilet, and eating); in performing the instrumental activities of daily living (preparing meals, shopping, doing housework, taking the bus or train, managing money); and in doing activities requiring physical strength, mobility and flexibility (lifting a 10 kilogram bag of rice; climbing one flight of stairs; walking 100 metres; and stooping, crouching or kneeling). The summed functional health difficulty scores ranged from 0 (no difficulties performing any task) to 30 (unable to perform any task). The Cronbach's alpha reliability scores for this scale were 0.95 in both 2001 and 2003. We subtracted the functional health difficulty score in 2001 from that in 2003 to determine the change in functional status over time. All the key variables (i.e. psychological wellbeing, the out-migration of children, inter-generational support, and control variables) were repeated measures at both waves. We used the information of out-migration at Wave 1 to predict the psychological outcomes at Wave 2, while controlling for the baseline socio-demographic variables, inter-generational support and psychological wellbeing. The change in living arrangement and functional health over time were also included in the analysis.
Statistical methods
Descriptive statistics were used to profile the baseline sample, and independent t-tests to examine the differences between the respondents with and without migrant children at Wave 1, their received inter-generational support and psychological wellbeing at Wave 2, and the change in support and wellbeing over the two years. We further carried out two multiple regression analyses to predict both depression and life satisfaction of parents at Wave 2. For both outcome measures, Model 1 had depression (or life satisfaction) at Wave 1, the socio-demographic variables and inter-generational support at Wave 1, and the changes in living arrangements and functional health between the two waves, while Model 2 added the number of migrant children at Wave 1. Controlling for baseline inter-generational support and psychological wellbeing helped obviate self-selection problems by discounting the possibility that the initial support and psychological wellbeing of older adults induced the out-migration of children. The two-stage regressions enabled us to evaluate the unique contribution of the out-migration of children on parental wellbeing over two years, while the effects of the socio-demographic characteristics of the parents, baseline inter-generational support, and the change in living arrangement and functional health over time were specified.
Results
Table 1 profiles the older adults at the baseline. The mean age of the analysis sample was 70 years (standard deviation (s.d.)=6.9), and the majority were female (52%) and married (61%). The average respondent had four adult children, of whom about one-third (1.4) were migrants. Around one-half (51%) of the respondents lived with their offspring, and 19 per cent lived by themselves. The sample characteristics are comparable to the national population in terms of the mean age, gender ratio, marital status and average household size (National Statistics Bureaus of China 2001), but a larger proportion of the respondents lived alone and a smaller proportion lived with children (Zeng and Wang Reference Zeng and Wang2003). These differentials are consistent with the large-scale out-migration of adult children. The average functional health difficulty score for the respondents was 4.7 out of a possible 30 (s.d.=6.2). On average, each participant had received instrumental support from one child, and received 1,061 RMB (approximately US$135) from all children during the previous 12 months. The respondents reported a relatively high mean score for emotional cohesion with children (4.2 of a possible 6, s.d.=1.2). The average depression and life-satisfaction scores at the baseline were 6.2 (s.d.=4.0) and 4.3 (s.d.=2.1), respectively.
Socio-demographic profile of the sample in 2001
Note: s.d.: standard deviation.
To address the first two research questions, we conducted a series of independent t-tests to examine whether older adults with and without migrant children in 2001 differed in their received support and psychological wellbeing in 2003, and whether there had been a difference in the change in support and psychological wellbeing between 2001 and 2003 (Table 2). The results show that those who had migrant children at Wave 1 received significantly more monetary support, and less instrumental support by 2003, and they had a significantly lower level of depression. The two groups did not differ in mean emotional cohesion with their children or life satisfaction at follow-up. These initial results suggested that older parents with migrant children in 2001 were advantaged in terms of monetary support and psychological outcomes in 2003. When the changes in support and psychological wellbeing over time are compared (the lower panel of the Table 2), however, it was found that having migrant children at baseline associated with less increase in monetary support. The two groups were no different for change in psychological wellbeing.
Inter-generational support and psychological wellbeing of older rural Chinese parents with and without migrant children in 2001
Note: s.d.: standard deviation.
Significance levels: (1,235 degrees of freedom) * p<0.05, *** p<0.001.
Taken altogether, the bivariate comparisons revealed that older parents with migrant children in 2001 were less depressed at baseline than those without migrant children, which explained why the former group had a significantly lower level of depression at Wave 2, but not the change in depression over time. Turning to the lower increase over two years in monetary support among parents with migrant children, an unexpected result, it turns out that those with migrant children at baseline received significantly more monetary support at that time than those having no migrant children.Footnote 1 This suggests that older parents with more monetary support at baseline were more likely to have children that had migrated, but it is possible that one or more children migrated long before the Wave 1 survey, thus even if out-migrations were associated with a significant increase in monetary support, the magnitude of the change over two years may not have been great.
To disentangle the relationships between out-migration, inter-generational support and the parents' psychological wellbeing, two-step multiple regressions were estimated. The dependent variables were the parents' depression and life-satisfaction scores at Wave 2, the independent variable was out-migration intensity at Wave 1 (measured by the number of the parent's children that were migrants), and the control variables were baseline depression or life satisfaction and inter-generational support. The results are shown in Table 3. Beginning with the regressions of depression, Model 1 revealed that older parents with higher income were healthier and less depressive at baseline and had a lower level of depression at follow-up. Starting to live alone and experiencing more functional health difficulties over time also associated with more depressive symptoms at follow-up. Receiving less instrumental and emotional support from children at baseline predicted a significant increase in the parents' depression at follow-up. Model 2 added the number of migrant children that a parent had at baseline, which significantly increased the association with the parents' depression at follow-up (ΔF (degrees of freedom 1, 1,171)=5.71, p<0.05). In other words, when controlling for the baseline socio-demographic variables and inter-generational support, there was a positive relationship between the number of children who had migrated and the parents' depression. Although the bivariate results suggested that older parents with migrant children at baseline had relatively fewer depressive symptoms at follow-up, the multivariate regressions suggest that inter-generational support buffers the negative impact of out-migration. When the effect of the inter-generational support was discounted, the out-migration of children led to more depressive symptoms among older parents.
Unstandardised regression coefficients predicating the depression of older rural Chinese adults in 2003
Notes:1. Reference category is living alone. 2. With 1, 1,171 degrees of freedom. SE: standard error.
Significance levels: (1,235 degrees of freedom) * p<0.05, ** p<0.01, *** p<0.001.
Similar two-step regressions to predict the life satisfaction of parents at follow-up were estimated. The results are shown in Table 4. Model 1 indicates that having more children, living with offspring, and having higher life satisfaction at baseline predicted the parents' sense of life satisfaction at follow-up. Having fewer functional health difficulties at baseline and relatively little decline in functional health over two years also associated significantly with higher life satisfaction at Wave 2. Unlike the model for depression (for which instrumental support from children at baseline was also protective), only a high level of baseline emotional cohesion with children enhanced the parents' life satisfaction at follow-up. As before, Model 2 added the number of migrant children at baseline. This variable was negatively related to the parents' life satisfaction at Wave 2, but with only marginal significance (p=0.06).
Unstandardised regression coefficients predicating the life satisfaction of older rural Chinese adults in 2003
Notes:1. Reference category is living alone. 2. With 1, 1,171 degrees of freedom. SE: standard error.
Significance levels: † p<0.1, * p<0.05, ** p<0.01, *** p<0.001.
Discussion
Large-scale internal migration in many developing countries presents a special challenge for the support of the rising number of older people in rural areas. While a few recent papers have investigated the consequences of out-migration on the family support and health of the older parents left behind, very little is known about the implications for their psychological wellbeing. Most previous findings have been limited by cross-sectional designs, which cannot directly test the causal relationship between out-migration and parental wellbeing. This study has examined the associations between the out-migration of adult children, inter-generational support and the psychological wellbeing of older people in rural China. We have investigated how the out-migration of adult children affected the monetary, instrumental and emotional support received by the parents, and their depression and life satisfaction. Consistent with previous findings (Du et al. Reference Du, Ding, Li and Gui2004; Keasberry Reference Keasberry2001), it was found by the follow-up survey that, compared to those who had all their children living nearby, older adults with migrant children received significantly more monetary support but less instrumental support. The higher monetary support may be largely attributed to remittances from migrant children. These money transfers can be regarded as the migrant children's compensation to their parents for their inability to provide direct personal care. Even so, the evidence that out-migration associated with increased monetary support and reduced instrumental support needs to be interpreted with caution. It was also noted that parents with migrant children already had more monetary support and less instrumental support when the children migrated, although it is possible that the differences at baseline arose because out-migrations occurred before the baseline survey. It is also possible that older adults at baseline who had more monetary support and were healthier (and so received less instrumental support) were more likely to have children who migrated. The data did not enable us to evaluate the competing explanations, but nevertheless it is clear that older parents with migrant children enjoyed more monetary support and received less instrumental support from their children.
In this study, the elderly parents with migrant children at baseline reported at follow-up a mean score for emotional cohesion with their children that was similar to the score for those without migrant children. This implies that the family relations can be maintained over great geographical distances. Although family spatial dispersion may negatively affect the hands-on support for the parents left behind, emotional ties and social exchange can be maintained between older parents and migrant children. This finding suggests that Chinese migrant children remain actively involved in care-giving for their older parents even when they move away, but again there are other possible explanations. While emotional cohesion between migrant children and their parents may decline in the absence of children, the increased monetary support for parents may buffer the effect. It should be noted that the measure used was of the average emotional closeness between the parents and all their children, which might obscure a differential in the closeness to migrant and non-migrant children.
We also tested how the out-migration affected the psychological wellbeing of the older parents left behind. Contrary to previous qualitative findings (e.g. Du Reference Du, West and Zhao2000; Du et al. Reference Du, Ding, Li and Gui2004; King and Vullnetari Reference King and Vullnetari2006; Miltiades Reference Miltiades2002; Murphy Reference Murphy, West and Zhao2004), it was found that when compared to those without migrant children, older parents with migrant children had a significantly lower level of depression after two years. Further examination revealed, however, that parents with migrant children had less depressive symptoms at baseline. The multiple regression analyses confirmed that when the positive effect of inter-generational support was taken into account, those with more migrant children tended to have a significantly higher level of depression and a lower level of life satisfaction. In other words, given no continued inter-generational support, out-migration impaired the psychological wellbeing of the parents.
Several of the findings deserve more investigation. Firstly, one interpretation of the results is that out-migration is a self-selection process from the parents' side. It was found at both baseline and follow-up that older parents with migrant children had more monetary support and better health conditions than those without migrant children, which suggests that healthier and more resource-rich older parents were more likely to have migrant children, or to enable them to move out. This finding demonstrates the importance of controlling for the initial status of older parents when examining the impact of out-migration. As has been shown, the evidence from bivariate and multivariate tests of change over time can be very different. Secondly, the findings point to the important role of continuous inter-generational support in maintaining the psychological wellbeing of the older parents who remain behind. Inter-generational support may moderate the effect of out-migrations on parental wellbeing. The results have shown that although the out-migration of children directly impairs the psychological wellbeing of the parents, such a negative influence can be buffered by increased monetary support for the parents and by continuing emotional cohesion between the generations. The lack of a difference in the change in psychological wellbeing between the parents with and without migrant children may be through the continued inter-generational support whether from migrant or non-migrant children. Thirdly, and of considerable interest, this study has found that the negative influence of out-migration on the psychological wellbeing of older parents was not as significant as expected. For one thing, adding the out-migration term in Model 2 did not significantly augment the explained variance in psychological wellbeing, nor did it greatly change the regression coefficients of the already-entered predictors. In other words, the out-migration of adult children did not significantly disturb the relationships between the other predictors of psychological wellbeing.
In this study, the measures of support were based on the information about all the children of their parents. This may partly explain the non-significant change over time in the three forms of support for parents. Many previous studies have focused only on migrant children, but the present findings indicate that the attributes and behaviour of both migrant and non-migrant children need to be considered. The presence of non-migrant children and the support provided by them may buffer the negative impact of their siblings' migrations on elderly parents. This may particularly apply in Asian countries where the majority of elderly people still live close to at least some of their children (Beard and Kunharibowo Reference Beard and Kunharibowo2001; Frankenberg, Chan and Ofstedal Reference Frankenberg, Chan and Ofstedal2002; Keasberry Reference Keasberry2001; Knodel and Debavalya Reference Knodel and Debavalya1997; Silverstein, Cong and Li Reference Silverstein, Cong and Li2006). In our sample, around 50 per cent of the respondents lived with at least one offspring.
Some limitations of the findings should be noted. Firstly, older adults in this study were treated as a homogeneous group. It is possible that older parents with certain attributes (e.g. gender, marital status, health conditions and socio-economic status) might be more vulnerable when facing the changes subsequent to children's out-migration. The explanations given by this study thus need to be interpreted with caution. In addition, several attributes of migrant children – gender, duration of migration and distance from parents – may also affect the inter-generational relationship and the psychological outcomes for the parents. These factors deserve further investigation.
Despite this limitation, this study has provided valuable insights into the complex relationships between children's out-migration, changing inter-generational support patterns, and their implications for the psychological wellbeing of older parents. It has also demonstrated the importance of a longitudinal design and of considering baseline support and health outcomes when studying the impacts of out-migration on the parents left behind. Given the growing magnitude of rural-to-urban migration in many developing countries, the care and wellbeing of the older parents left behind will continue to be the issues of great importance. In China particularly, one may expect more rural elders to be living with fewer children in the future, as a consequence of the one-child fertility policy and of changing family structures in the face of the country's dramatic social changes. If that is the case, even a low intensity of adult children's out-migration may create significant pressures on old-age support. More concerted effort is required to deepen our understanding of the longitudinal effects of out-migration on various aspects of the ageing experience, particularly in the rural population.
Acknowledgements
This research was supported by grants from the Fogarty International Center of the National Institutes of Health (R03TW01060-01) and the USC US-China Institute. The authors would like to thank Shuzhuo Li for his leadership as the Co-director of the larger research project on Chinese rural elders, and Zhen Cong for her contribution to data collection.
