Grandparents Caring for Grandchildren in China and Korea: Findings From CHARLS and KLoSA

Abstract

Objectives.

To provide an overview of the prevalence and profiles of grandparents providing childcare to grandchildren in 2 East Asian countries, China and South Korea, characterized by similar demographic developments and a shared cultural background but having very different contemporary institutional and socioeconomic circumstances.

Method.

We apply logistic models to analyze pilot data from the China Health and Retirement Longitudinal Study (CHARLS) and data from the Korean Longitudinal Study of Aging (KLoSA; Wave 2). Our analytic sample comprises 772 Chinese respondents and 4,958 Korean respondents aged 45–79.

Results.

The proportions of grandparents providing childcare to grandchildren differ considerably between China (58%) and South Korea (6%). Still, the determinants of grandparents’ involvement in childcare (e.g., age, geographic proximity) are fairly similar in both countries. However, financial support from adult children to grandparents is found to be significant in China only, whereas Korean grandparents exhibit a greater propensity to care for their (employed) daughters’ children than for their sons’ children.

Discussion.

Our analysis suggests that in South Korea, patrilineal considerations may begin to lose some of their importance in shaping downward functional solidarity between generations and that instead (grand-)children’s actual needs, particularly those related to maternal employment, receive more attention. We find no such evidence in our Chinese sample.

Childcare, East Asia, Grandparenting, Intergenerational support.

Grandparenthood and grandparenting in Asia receive growing attention (e.g., Mehta & Thang, 2012). Our brief report adds to this literature by providing an overview of the prevalence and profiles of grandparents providing childcare to grandchildren in China and South Korea (hereafter, Korea). These two countries are particularly interesting cases because population aging has been projected to take place more rapidly in China and Korea than elsewhere, confronting both countries with massive challenges for their future economic growth and social welfare (cf. Yoon, 2013; Zhang, Guo, & Zheng, 2012). Moreover, population aging and related changes in social conditions, embedded in a more general process of modernization, are likely to challenge traditional aspects of East Asian family life, such as filial piety and patrilineality. However, despite similar demographic developments and a shared cultural background, important institutional and socioeconomic differences remain. Although, for example, China’s current development (in terms of growth in gross domestic product, rate of urbanization, etc.) is more dynamic, Korea’s achieved level of economic modernization is still much higher (despite its substantially lower female labor force participation rate; e.g., Cooke, 2010). Because grandparents’ involvement in families is based on the interplay between such contemporary circumstances and traditional family values and norms (as well as changes therein), it seems worthwhile taking a cross-nationally comparative perspective on grandparent-provided childcare in these two countries. Our empirical investigation draws on data from the China Health and Retirement Longitudinal Study (CHARLS) and the Korean Longitudinal Study of Aging (KLoSA).

Background

Despite concerns about an erosion of filial piety by modernization, it continues to play a major role in contemporary East Asian societies (e.g., Cheung & Kwan, 2009; Lin & Yi, 2011). For example, one married child (preferably a son) is still expected to fulfill the filial obligation of living with the parents (e.g., Silverstein, Cong, & Li, 2006). As a consequence, in the 1990s, about 70% of East Asian grandparents lived with a grandchild (Yasuda, Iwai, Yi, & Xie, 2011), making them an easily accessible and reliable source of care (e.g., Chen, Short, & Entwisle, 2000).

However, preferences for traditional living arrangements are changing in both China and Korea (e.g., Kim & Rhee, 1997; Zhang, 2004) and coresidence rates in East Asia have been declining, particularly in Korea (cf. Yasuda et al., 2011). Likewise, recent investigations point to the emergence of mutual exchange dynamics in families in China, where “economic development [. . .] has altered the meaning of filial piety from an unconditional duty to support one’s elderly parents to a form of support that is to some degree conditional on parents’ earlier support to children” (Cong & Silverstein, 2008, p. 7). Similarly, Lee and Bauer (2010) report evidence suggesting that Korean adult children often pay grandmothers for childcare. Although financial support for elderly parents remains an important element of filial piety, intergenerational “time-for-money” exchanges contribute to maintaining the symmetry in the relationship between grandparents and adult children, particularly if the former have taken primary responsibility for the care of grandchildren. This is important because rather than providing extensive childcare, traditional grandparental roles in East Asia tend to be related to fostering intergenerational bonds and extending family heritage (e.g., Cong & Silverstein, 2012b; Maehara & Takemura, 2007).

Grandparenting in contemporary Asia is characterized by continuity, rooted in traditional family values and norms, as well as by changes driven by modernization processes (cf. Thang & Mehta, 2012). Although increasing maternal employment has been suggested to have resulted in a greater involvement of maternal grandmothers in Korea (possibly indicating the emergence of a more general shift from patrilineal to bilateral kinship interactions; see Lee & Bauer, 2013), traditional grandparent roles in China have been particularly challenged by two specific developments: the one-child policy and the dramatic increase in internal labor migration since the early 1980s. Raising the “precious single child” appears to have become an “intergenerational joint mission” between parents and grandparents (Goh, 2006; also see Short, Zhai, Xu, & Yang, 2001)—where the role of the latter becomes dominant, though, if young rural parents are forced to look for employment in cities. During the past 3 decades, an estimated 220 million laborers migrated from rural to urban areas, leaving approximately 58 million children behind—among them, nearly one third separated from both parents. This development has created an enormous demand for extensive childcare by grandparents (e.g., Cong & Silverstein, 2012b).

Thus, we expect substantially larger proportions of grandparents to provide childcare in China than in Korea, especially in rural regions.

Although the demand for center-based childcare has increased, parents and grandparents continue to be the most popular source of care in both countries (e.g., Lee & Bauer, 2013; Zhai & Gao, 2010). However, although Silverstein, Cong, and Li (2007) find that grandparents provided childcare to grandchildren in 35% of family setups in rural China, the share of grandmothers caring for grandchildren has been suggested to be only 13% in Korea (Lee & Bauer, 2010). Despite this overall low proportion, 62% of employed Korean mothers with young children rely on childcare provided by grandmothers, whereas only 28% use center-based childcare (Lee & Bauer, 2010). Maternal employment might thus be an important trigger for grandparents’ involvement in childcare—an assumption corroborated by reports of positive associations between daughters’ labor force participation and grandmothers’ childcare provision in both Korea and China (e.g., Chen, Liu, & Mair, 2011; Lee & Bauer, 2013).

We expect this relationship to be stronger in the former than in the latter country, however, because center-based childcare seems more socially accepted and more readily available in China than in Korea (e.g., Sung, 2003; Zhai & Gao, 2010).

Consistent with patrilineal norms, care for grandchildren in China has been shown to be oriented more toward sons’ children (e.g., Silverstein et al., 2007). This can be explained in part by significantly higher proportions of paternal grandparents living close to or coresiding with their children (e.g., Chen et al., 2000), an observation that has also been made in Korea (e.g., Yasuda et al., 2011). However, significant intergenerational support also takes place outside of coresidential households; hence, one needs to account for geographic proximity. Moreover, in China, paternal grandparents’ households have been shown to be a more popular destination for the care of young children than maternal ones (Chen et al., 2011). However, among Korean mothers utilizing kin-provided childcare, more than half rely on maternal grandparents, whereas only slightly more than one third rely on paternal grandparents (Lee & Bauer, 2013). Finally, the provision of grandchildcare was found to be associated with greater upward financial transfers in both countries (Cong & Silverstein, 2008; Lee & Bauer, 2010).

Therefore, we do not expect any significant cross-national differences in the association between grandparent-provided childcare and proximity or upward financial support. We should find, however, a greater propensity among maternal grandparents to provide childcare in Korea than in China.

Korean grandmothers have been suggested to provide care for their working daughters’ children not to support female autonomy but, on the contrary, to maintain traditional sex role ideology (see Lee & Bauer, 2013). Our empirically derived hypothesis regarding differences in Chinese and Korean maternal grandparents’ provision of childcare is therefore consistent with the theory of structural lag (Riley, Kahn, & Foner, 1994). Although China is experiencing faster changes today, the lag between traditional norms and the practical requirements of its achieved level of socioeconomic development seems larger in Korea. An important reason for this is that childcare is still considered a private (i.e., women’s) rather than a societal task, even if the mother is a full-time employee (e.g., Sung, 2003). Along the same lines, but more generally, Kyung-Sup and Min-Young (2010) have argued that Korean women’s individualization has occurred primarily as a matter of practicality rather than ideational change and that institutions of modernity (such as welfare states, firms, and schools) have become increasingly ineffective in helping to alleviate sex-based familial burdens and dilemmas.

Method

We draw on the data from the CHARLS and KLoSA (cf. Lee, 2010). The currently available Chinese data are from the 2008 CHARLS pilot survey, which was conducted in two provinces: Gansu, a poorly developed western inland region, and Zheijang, a prosperous coastal province. The pilot sample is representative of the non-institutionalized population aged 45 and older in these provinces. Interviews were conducted with 2,685 individuals from 1,570 households, which corresponds to a response rate of 85% (for further details, see Zhao, Strauss, Park, Shen, & Sun, 2009). CHARLS is closely modeled on KLoSA, whose baseline wave with 10,254 respondents from 6,171 households was conducted in 2006. The data are representative of non-institutionalized Koreans aged 45 and older in 15 large administrative areas. Our analysis is based on 8,688 reinterviews conducted with baseline respondents in 2008 (the retention rate in this follow-up survey was 85%). Although resulting in a somewhat smaller sample, using KLoSA’s Wave 2 allowed us to link grandparents’ characteristics to those of specific children, which was not possible in the study of Lee and Bauer (2010) based on KLoSA’s Wave 1.

Our analytic sample consists of all respondents who reported having “any grandchildren who are not adults” (CHARLS) or any “living grandchildren” (KLoSA), respectively, and who were aged 45–79 at the time of the interview (we applied this upper age limit to exclude grandparents—36 in CHARLS, 723 in KLoSA—with barely any chance of having a grandchild sufficiently young to be covered by KLoSA’s childcare question; see later discussion). Note that in CHARLS, only one partner served as the “family respondent” for the entire household, whereas in KLoSA, each partner provided information. This resulted in a sample of 772 Chinese and 4,958 Korean respondents (the latter living in 2,924 households). Compared with the CHARLS sample, Korean respondents are, on average, 5 years older, more likely to be female, but significantly less likely to work for pay, to be in good health, or to coreside with an adult child (see Table 1 for descriptive sample statistics).

Table 1.

Descriptive Sample Statistics (Unweighted)

Variables	China (%, n = 772)	South Korea (%, n = 4,958)
Dependent variable
Provision of grandchild care	57.6	5.6
Characteristics of grandparents
Age, years
45–59	42.9	18.7
60–69	38.7	42.5
70–79	18.4	38.8
Sex: female	48.1	60.2
Partnership status: lives without partner	25.4	22.6
Employment status: works for pay	62.2	34.3
Self-assessed health: good or better	45.6	25.7
Received financial support from children	53.5	58.3
Proximity
Coresidence	31.0	7.1
Close	17.9	21.4
Medium	28.6	41.4
Long distance	17.1	28.8
Residence: rural	54.4	39.5
Characteristics of adult children
Sex: female	39.1	51.3
Partnership status: lives without partner	2.5	4.5
Employment status: works for pay	84.6	66.3

Note. Unweighted data from the 2008 China Health and Retirement Longitudinal Study (CHARLS; pilot survey) and the 2008 Korean Longitudinal Study of Aging (KLoSA; Wave 2).

Our binary dependent variable is coded 1 if respondents reported to have “spent any time taking care of their grandchildren last year” (CHARLS) or “during the past 12 months, took care of any of their grandchildren younger than 10 years old” (KLoSA), respectively, and 0 otherwise. Note that we do not know the grandchildren’s exact age. However, although the childcare questions are not identical, excluding adult grandchildren in CHARLS and restricting our sample to respondents younger than 80 years of age should result in fairly comparable measures across the two surveys. For the multivariate analysis, we used standard logistic regression, estimating robust standard errors for the Korean sample to account for the clustering of partners in the same household. Results are presented as odds ratios with 95% confidence intervals.

Control variables cover grandparents’ and children’s characteristics. We account for grandparents’ basic sociodemographic characteristics (age, sex, partnership status, employment status, and self-assessed health) and take into consideration whether grandparents received financial support from any of their children in the past year (CHARLS records amounts of ≥100 yuans [~ US$ 16] only, whereas KLoSA only counts support from non-coresident children). Unfortunately, we could not properly identify the child who provided financial assistance. We also control for grandparents’ geographic proximity to the selected child (see later discussion) by introducing dummy variables indicating coresidence and close, medium-, or long-distance residence. Note that these categories cannot be compared directly across surveys: in CHARLS, respondents were asked where the child lives (e.g., same or different village, county, province), whereas KLoSA’s question referred to travel times. Finally, we distinguish grandparents living in rural areas from those in urban areas. In the CHARLS sample, all respondents from Gansu were defined as “rural,” whereas this applies to all KLoSA respondents living outside metropolitan areas.

If grandparents had cared for grandchildren from more than one child, we used information on the adult child, for whose child (i.e., grandchild) the greatest frequency of caregiving was reported (i.e., only one parent–child dyad is considered). If this information was missing, we selected the youngest adult child; if a respondent reported to have grandchildren but had not looked after any of them, we also used the information on the grandparents’ youngest adult child. CHARLS and KLoSA allow accounting for adult children’s sex, partnership status, and employment status.

Results

The most striking difference between China and Korea is the great difference in proportions of grandparents having provided any childcare: 58% in China versus 6% in Korea (where the proportion of households in which at least one of the grandparents has cared for a grandchild is 9%; see Table 1). Qualitatively, this finding corroborates previous evidence suggesting very distinct levels of grandparent-provided childcare in the two countries, underlining the importance of investigating potential differences in its determinants.

Overall, however, we do not find systematic cross-national differences in the associations between grandparents’ (adult children’s, respectively) characteristics and the probability of caring for grandchildren (see Table 2). Looking at sociodemographic characteristics, we observe the lowest odds of providing childcare in the 70+ age group. Grandmothers are more likely to report having cared for a grandchild, but the coefficient is significant only in the Korean sample (probably because male CHARLS family respondents also report on their female partner’s childcare). Chinese grandparents without a partner and gainfully employed Korean grandparents exhibit a lower propensity to provide care, whereas grandparents’ self-assessed health appears to be unrelated to grandchildcare in both countries.

Table 2.

Results of Multivariate Logistic Regression for Grandparents’ Provision of Childcare

	China (n = 726)		South Korea (n = 4,878)
	Odds ratio	95% CI	Odds ratio	95% CI
Characteristics of grandparents
Age, years
45–59 (reference)	1.00	—	1.00	—
60–69	1.26	0.86–1.84	0.61**	0.43–0.87
70–79	0.44**	0.26–0.73	0.16***	0.10–0.25
Sex: female	1.30	0.92–1.85	2.22***	1.67–2.95
Partnership: lives w/out partner	0.41***	0.27–0.61	0.80	0.55–1.15
Employment status: works for pay	1.04	0.71–1.50	0.42***	0.29–0.61
Self-assessed health: good or better	0.98	0.69–1.36	1.01	0.73–1.41
Received financial support from children	1.83***	1.31–2.56	1.21	0.81–1.79
Proximity
Coresidence (reference)	1.00	—	1.00	—
Close	0.42***	0.26–0.68	0.27***	0.13–0.57
Medium	0.37***	0.23–0.58	0.07***	0.03–0.15
Long distance	0.24***	0.14–0.39	0.06***	0.03–0.15
Residence: rural	1.28	0.91–1.79	0.75	0.55–1.03
Characteristics of adult children
Sex: female	0.76	0.52–1.10	1.67*	1.17–2.41
Partnership: lives w/out partner	4.54*	1.27–16.09	2.50*	1.22–5.12
Employment status: works for pay	2.40***	1.51–3.82	5.62***	3.37–9.38
Pseudo-R2	.12		.27

Notes. CI = confidence interval; data from the 2008 China Health and Retirement Longitudinal Study (CHARLS; pilot survey) and the 2008 Korean Longitudinal Study of Aging (KLoSA, Wave 2).

*p < .05. **p < .01. ***p < .001.

Turning to upward intergenerational transfers and coresidence, our findings suggest that grandparents receiving financial support tend to be more likely to spend time taking care of a child, albeit the observed association is statistically significant only in China. Intergenerational coresidence is paralleled by significantly higher odds of reporting childcare in both samples (an interaction with the adult child’s sex did not provide any further insights; details not shown). We do not observe any differences between grandparents living in rural and urban areas.

Finally, our results concerning the role of adult children’s characteristics in determining grandparents’ propensity to provide childcare reveal strong evidence for an important role of children’s needs: if the adult child is without a partner or gainfully employed, both Chinese and Korean grandparents are much more likely to care for their grandchild than otherwise. Although not significant, the sign of the coefficient for the adult child’s sex in the CHARLS sample is consistent with patrilineality and previous evidence pointing to bias toward son’s children. Conversely, Korean grandparents appear to be significantly more likely to support daughters than sons. An interaction between adult children’s sex and employment status shows, however, that this is only true, if the daughter works for pay, whereas grandparents are least likely to provide any childcare, if the grandchild’s mother is a daughter who is not gainfully employed (details not shown).

Discussion

The first remarkable result of our analysis is the striking difference between Chinese (58%) and Korean (6%) grandparents’ involvement in childcare. Although, qualitatively, this is consistent with previously reported evidence, the very low proportion of childcare providers among Korean grandparents clearly needs further explanation. One reason is that we were unable to exclude grandparents without a grandchild younger than age 10 from our sample, presumably resulting in an underestimation of the proportion of caregivers. However, the proportion of grandmothers reporting to have cared for a grandchild in our study is just about half of what Lee and Bauer (2010) found in their analysis of KLoSA’s baseline wave (7% vs 13%). This massive decrease apparently results from the fact that in the follow-up wave, which was not supplemented by a refresher sample, a substantial number of grandchildren, being now 2 years older, no longer meet the restrictive age criterion of the childcare question. But even the proportion reported in study of Lee and Bauer (2010) is low, both in comparison to China and in a more global perspective (about 50% of grandparents in the United States and Europe provide some kind of childcare; cf. Hank & Buber, 2009).

A second finding is that despite substantial differences in proportions of grandparents being engaged in childcare activities, the determinants of their involvement are, by and large, fairly similar in China and Korea. One exception is the difference in the association between grandparents’ receipt of financial support and their propensity to provide care. As expected, we observe a positive correlation in China (consistent with a “mutual aid” model of the family; cf. Cong & Silverstein, 2008). Though pointing in the same direction, the respective coefficient is insignificant in the Korean sample. Because Lee and Bauer (2010) suggest that it is more likely to expect a statistically significant association between upward financial transfers and Korean grandparents’ childcare provision if full-time rather than part-time care is considered, the lack of significance in our study may result from a low proportion of full-time caregivers in KLoSA’s Wave 2 sample.

Another difference between Chinese and Korean grandparents is that the latter are more likely to care for their (employed) daughters’ children than for their sons’ children. This might be interpreted as early evidence for patrilineal considerations beginning to lose some of their importance in shaping functional solidarity between generations and that, instead, (grand-)children’s actual needs receive more attention. However, whether our observation really indicates an emerging process of structural change toward greater sex neutrality in East Asian—or at least South Korean—intergenerational relationships cannot be said for sure yet. If this were the case, we would expect this process to unfold very slowly because traditional behaviors (behavioral expectations, respectively) are likely to persist for a long time, even if the social and economic conditions under which they evolved have already changed (cf. Cong & Silverstein, 2012a; Lee & Bauer 2013).

It seems that differences in (grand-)children’s needs, particularly those resulting from maternal employment, are a major force driving the cross-national differences in levels of grandparent-provided childcare. Although we do not find the hypothesized differences between rural and urban areas (which is consistent with similar “non-effects” reported in the studies by Chen et al., 2011, and Lee & Bauer, 2010), labor migration in association with an overall high labor force participation of mothers is likely to be an important factor in explaining the substantially higher demand for non-parental childcare in China. Despite the Chinese government’s substantial investments in childcare facilities, the total demand cannot be satisfied by center-based services alone and grandparents need to step in as a complementary source of care (e.g., Chen et al., 2000; Zhai & Gao, 2010). In Korea, women’s engagement in paid work has increased, but this is still much lower than that among Chinese women (cf. Cooke, 2010). The overall demand for non-parental childcare might therefore be lower in Korea than in China, but because childcare is still widely considered a private rather than a societal task, Korean mothers tend to rely on female relatives for childcare because of the lack of government schemes (e.g., Lee & Bauer, 2013; Sung, 2003). For many gainfully employed Korean daughters, grandparental childcare is thus a necessary prerequisite to continue their working careers after childbirth—and grandparents’ primary role here appears to be one that substitutes rather than complements public day-care facilities (see Hank & Buber, 2009, who provide a similar argument for the Southern European context).

Although other sources suggest a high intensity of care in both China and Korea, if any childcare is provided at all (e.g., Chen et al., 2011; Lee & Bauer, 2010), our data—unfortunately—do not allow the proper investigation of potential cross-national differences in the intensity of grandparent-provided childcare. Other limitations of our data are the lack of precision in the measurement of who provided financial support to grandparents and the missing information on non-coresident grandchildren’s age and sex. However, these limitations are counterbalanced by some unique advantages: unlike Lee and Bauer (2010), we could link Korean grandparents to their children’s characteristics; and different from Chen and colleagues (2011), for example, we were able to include non-coresident Chinese grandparents in our analysis. Finally, KLoSA and CHARLS are designed as long-run panel studies (Lee, 2010), allowing researchers to monitor continuity and change in filial norms and expectations regarding children’s responsibilities toward their parents (e.g., Cong & Silverstein, 2012a; Kim & Choi, 2013) and to further substantiate our assertion of a possible decline in the relevance of patrilineal norms in downward intergenerational support from grandparents to their (grand-)children.

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgments

Pei-Chun Ko is supported by a stipend from the German Research Foundation (DFG). We are grateful to the China Center for Economic Research at Beijing University for providing us with the CHARLS pilot data and to Hyejung Lee from the Korea Labor Institute for providing us with the KLoSA data. We also thank the editor of Journals of Gerontology Series B: Psychological and Social Sciences, three anonymous reviewers, and participants of a SOCLIFE seminar at the University of Cologne for their helpful comments. Pei-Chun Ko contributed to planning and writing the paper; she performed all statistical analyses. Karsten Hank contributed to planning and writing the paper; he supervised the data analysis.

Author notes

Decision Editor: Merril Silverstein, PhD