• 7. Support For Aging Parents From Daughters Versus Sons
• Chapter
• University of Michigan Press
• pp. 167-195
• View Citation
• Additional Information

7

SUPPORT FOR AGING PARENTS FROM DAUGHTERS VERSUS SONS

Martin King Whyte and Xu Qin

The Chinese family system is famous for its strong emphasis on patrilineal ties and obligations. As noted in chapter 1, for most of Chinese history married sons have been expected to coreside with their aging parents, while daughters have been expected to marry out into other families and communities. In the countryside rules of village and lineage exogamy often prevailed, so that “marriage out” meant moving to another, often distant, village. After they had married out, daughters retained emotional ties with their natal families and periodically visited their parents, but they were not expected to contribute to the old-age support of the latter. Rather, upon marriage the filial obligations of daughters were redirected to the support of their husbands’ parents. Thus support for aging parents came predominantly from sons and daughters-in-law, with married daughters playing little role in most instances. Much of the dynamic tension within extended Chinese families was seen as originating in the difficult struggle of young brides to establish themselves among strangers, for the combination of arranged marriage with village exogamy forced women to reside among people she had often never met before the day of the wedding. Only over the years and through the birth of children (preferably sons) might the terrified bride be transformed into a proud matriarch, only to have her place questioned by a new generation of brides in need of retraining, the wives of her sons.¹ Patrilineal family units had such a powerful place in popular consciousness and in the social structure that divided loyalties (between one’s own parents and one’s husband’s) were strongly discouraged.²

Partly because of this crucial difference in the expected roles of sons and daughters, the birth of a son was usually celebrated much more elaborately as a “big happiness” event than the birth of a daughter (referred to as “small happiness”). Customary phrases such as “the more sons the more wealth” and “raising sons to guard against old age” contrasted with the terms used to describe daughters: “a married-out daughter is like spilled water” (jia chuqu de nuer, po chuqu de shui), and a daughter is an “unprofitable commodity” (pei qian huo). Brides who produced daughters but no sons were blamed for this serious fault, and if the family had enough resources she might be displaced in the husband’s favor by a concubine in the quest for sons to continue the family line. In extreme circumstances newborn daughters might be abandoned or killed, a fate much less likely to occur to newborn sons. What was at stake was not simply the posterity of the family, but the ability of the parents to survive once they could no longer support themselves. Parents without any sons were faced with the prospect of trying to entice a young male to reverse the usual pattern. Instead of bringing a “stranger” bride into his parental home, he would have to marry as a stranger into his wife’s family and agree that future children (or at least male children) would bear his wife’s surname, rather than his own. Given the overwhelming emphasis on patrilineal ties, it was often difficult to find a male willing to accept a compromised status in his wife’s home.³ Other solutions to the problem of old-age support existed, such as adopting a son from another family, but the same patrilineal culture obstructed these options as well.

Despite the dramatic changes that have swept China in recent decades, as reviewed in chapter 1, available evidence indicates that the patrilineal, patrilocal nature of social life still prevails in the countryside.⁴ As a result, old-age support in Chinese rural areas continues to depend primarily on sons, with married daughters playing only a marginal role. For example, one survey conducted in a rural area in Shandong Province in 1992 found that there was a strong gender difference in support for older parents.⁵ Similarly, research conducted on Taiwan indicates that financial support to aging parents from sons remains more important there than support from daughters. One recent study on that island concluded that “daughters’ support still remains very supplementary.”⁶ (We will present our own data on this question comparing Baoding and urban Taiwan in chapter 10.)

The continuing gender disparity in old-age support in rural China is a major factor explaining the problems of enforcing official family planning targets in Chinese villages. It is still the case that if rural parents give birth to daughters but no sons, they face the prospect of having nobody to care for them and support them in old age. This is so because most rural localities have no substantial public source of old-age support available. The patrilineal basis of old-age support is the primary reason why rural parents in large numbers continue to try desperately to have a son.⁷

In China’s cities the situation has changed in many ways. Available evidence indicates that married daughters are now expected to, and do, contribute to the support of their own parents as well as their husbands’ parents.⁸ In addition, the majority of urban retirees receive pensions, as noted in chapter 2, and if so they do not have to rely entirely or even primarily on grown children in their old age. For these reasons, among others, urban parents are less likely than their rural counterparts to feel that it is essential to have at least one son, and enforcement of the “one child policy” has been relatively successful in China’s large cities. At the same time, as noted in chapter 2, it is still two to three times as likely that aging parents will live with a married son as with a married daughter. In other words, the traditional patrilineal bias of Chinese kinship has not disappeared entirely in urban China. However, the evidence presented in chapter 6 indicates that coresidence is not as important as it presumably once was, in terms of providing a basis for old-age support. Daughters (as well as sons) may provide multiple kinds of support to aging parents, even when they do not live under the same roof. Anecdotally, urban Chinese sometimes claim that they prefer to rely on a married daughter, since reliance on a son also implies reliance on a daughter-in-law, and the mother-in-law/daughter-in-law relationship is traditionally prone to conflict. The goal of the present chapter is to go beyond these contradictory indicators to investigate the current situation in regard to the gendered roots of old-age support in Baoding. Do grown and married sons still contribute the major share of the care and support that parents receive from their children in that city, even if daughters help out to some degree? Or do grown and married daughters now contribute as much or even more than do sons?

Relations with Parents—the View from Their Children

The data from our Baoding surveys yield a number of ways to look at the issue of how much daughters versus sons participate in the care and support of their aging parents. As the analyses in chapter 5 indicate, parents and their children do not always agree precisely on how much each child provides. In the present analysis we will examine first a variety of indicators of the sorts of ties and exchanges that Baoding grown children report having with their parents, and whether the extent of those ties and exchanges varies depending upon the gender of the child. Subsequently we will focus more narrowly on financial support given to parents, and in that analysis we will switch to the perspective of the parents: Of those parents receiving some financial support from grown children, what is the relative contribution received from daughters versus sons?⁹

The Baoding child questionnaire contained a wide range of questions about relations between grown children and their parents, questions that we will use to see whether the patterns for daughters differ from those for sons. We focus first on some of the questions dealing with attitudes toward traditional filial obligations that were included in our child questionnaire. The primary issue of interest here is whether there is a difference in filiality between grown sons and daughters, particularly in regard to obligations toward their own parents (rather than their in-laws). In making this comparison we exclude cases of families with daughters but no sons, and those with sons but no daughters. This exclusion is adopted because we assume that gender specialization in relations with parents will only be revealed when siblings of both genders are available. This exclusion reduces the number of child interview cases from 753 to 648, a modest reduction that means that 86% of all the Baoding parents whose children we interviewed had at least one child of each gender.¹⁰ Since we are particularly interested in whether daughters, after marriage, have weaker ties with their parents than do sons, in general we will also restrict our attention here to cases of married children.¹¹ That restriction reduces our child sample further to 523 cases.

In Table 7.1 are displayed figures from our child sample on the level of agreement of married (and siblinged) sons and daughters with various questions about filial obligations. The answers to the question most directly relevant to the present inquiry are shown in the first row of the table. There we can see that only a minority of either sons or daughters agrees with the traditional attitude that married daughters have a lesser obligation to support their own parents than married sons. It seems evident not only that daughters expect to share the burden of supporting aging parents, but that most brothers share this expectation. However, this question also revealed the strongest gender disparity in the table, with daughters much more likely than sons to disagree with the traditional view.¹² In other words, in some cases daughters feel obligated to share the burden even if their brothers do not think they should feel so obligated.

For most of the remaining filial attitude questions shown in Table 7.1, there is no significant difference between the views of daughters versus sons. In only two of the other items shown in the table is there such a difference, and those two instances point in opposite directions. A slightly higher percentage of sons than daughters agreed with the statement that “even when children have married and live elsewhere, they should do as much as possible to see that their parents have a happy old age.” However, sons were also more likely to agree with the not-so-filial view that “young people should pay more attention to their careers than to caring for their parents.” On balance these attitudinal items provide support for the view that the feelings of obligation toward parents of married daughters are now much the same as those of their brothers. (See also the results presented in chapter 4, Table 4.5.) But do filial attitudes translate into similar patterns of relations with parents? Several more sets of figures help illuminate this question.

In Table 7.2 we examine the reported frequency of contacts between married children and their parents and parents-in-law.¹³ The first row in the table shows once again the familiar pattern of married sons living with their parents much more often than married daughters—in this specific comparison the differential is about 2 -1/2 1.¹⁴ The responses in the rest of the table come only from non-coresident children, and concern the frequency of visiting parents. The figures in this table make clear that there is no significant difference in the frequency with which non-coresident daughters and sons visit their own parents. However, there is a significant difference in the frequency of contacts with parents-in-law, with daughters in more frequent contact than sons. We can conclude from these figures that in terms of keeping in touch and visiting, married daughters do “double duty” more often than married sons. Daughters are likely to keep in regular touch with both their own parents and those of their husbands, while sons keep in close touch with their own parents but are more likely to leave social contact with their parents-in-law to their wives. Note also that married daughters visit their own parents more often than they visit their in-laws.¹⁵ The picture of contacts revealed by these figures for married women is quite different from that yielded by previous (mostly rural) ethnographic accounts. Those accounts portray women who had married being expected to visit their natal families only occasionally, and mostly for customarily set events, such as during seasonal festivals, weddings, and funerals.¹⁶

Table 7.3 displays the pattern of responses of married daughters and sons to a variety of other questions about the pattern of current relations with parents. As in the previous two tables, the most common pattern is one in which there are no significant differences between married sons and daughters in their reported relationships with their parents currently. However, it is interesting that daughters are slightly more likely than sons to report that after they married their level of emotional support supplied to their parents declined (row 2), and they are significantly more likely to report that their level of financial support to parents declined (row 1).¹⁷ The only other differences shown in the table that are statistically significant hint that sons may be less emotionally close to parents than daughters. A higher proportion of sons than daughters report that they are sometimes or often the target of parental criticism (line 5), and fewer sons than daughters report agreement with the parents on life style and ideas (last line in the table). Once again the dominant pattern favors an interpretation of similarity in relations between married daughters and sons and their parents, rather than the differentiated pattern stressed in the Confucian past.

If we turn our attention now to the four kinds of support of parents we asked about systematically—physical care for parents (i.e., things like help going to the bathroom and getting dressed), help with daily chores, financial support, and providing material goods—the figures in Table 7.4 reveal some surprises.¹⁸ Daughters appear slightly more likely than sons to provide three of these kinds of support (personal care, help with daily chores, and provision of material goods); sons, in contrast, are slightly more likely to provide financial support to parents. Only the tendency for daughters to more often give food and other material gifts to their parents is statistically significant. The bottom two rows in Table 7.4 provide additional information about the last two kinds of material support. The figures in these rows come from follow-up questions asking those married children who provided support to recall the average amount provided to parents, per month in the case of financial support, and per year in the case of material goods provided. When it comes to monthly financial support to parents, sons report providing significantly more than daughters on average (34.3 yuan versus 17 yuan). Daughters provide somewhat more support than sons to parents in the form of food and other goods (252.3 yuan per year versus 229.7 yuan per year), but this difference is not large enough to be statistically significant. Overall, if we were to accept these retrospective estimates as fully accurate, they tell us that married sons on average provide 641.3 yuan to parents per year, while married daughters provide 456 yuan. These figures might be interpreted as indicating that, whereas daughters are now more actively involved in contacts with and support for their parents than was generally the case in the past, when it comes to what matters most—financial assistance—sons still do significantly more. (Recall also that the analysis presented in chapter 5 pointed to a similar conclusion.) This interpretation would also be consistent with the pattern observed in the first row of Table 7.3, with daughters more often than sons reporting that they reduced their financial support to parents when they got married.

Before accepting this conclusion we need to recognize the fact that married sons and daughters differ in a number of ways that affect this comparison. For example, married sons are much more likely than their sisters to live with their parents, and they also are likely to have higher incomes.¹⁹ Once we recognize these differences, we are led to a further question—if we control for other factors, does being a married daughter versus a married son per se affect the various forms of support provided to parents? To answer this question we need to carry out multivariate analysis of our support measures.

In order to examine the impact of the gender of the child on the four forms of parental support in a multivariate context, we need to include in our analysis a range of variables that we expect may also have an impact on levels of support. The tables that follow utilize the following set of predictor variables, a selection based upon preliminary analyses and the results reported in other chapters in this volume:

Male (gender of child, 1 = male, 0 = female)

Coresidence (child lives with parent = 1; non-coresident = 0)

Education:

College education (college education or higher = 1; otherwise = 0)

Upper middle school (upper middle school or senior polytechnical school = 1; otherwise = 0)

(Junior middle school or below—omitted category)

Rural origins (grew up in rural area = 1; otherwise = 0)

Parental age:

Parent old old (parent 70 or older = 1; otherwise = 0)

Parent mid old (parent 60–69 = 1; otherwise = 0)

(Parent young old: 50–59—omitted category)

Number of siblings

Number of children²⁰

Ln child income (Natural logarithm of the child monthly income)

Ln parent income (Natural logarithm of parent monthly income)²¹

Parent health problems²²

Parent provision of child care help in the past (1 = yes; 0 = no)

Parent provision of domestic chore help in the past (1 = yes; 0 = no)

Parent provision of financial support in the past (1 = yes; 0 = no)²³

In Table 7.5 we display the logistic regression coefficients showing the relationship between these predictor variables and our four measures of whether the child is currently providing the four forms of parental support measured (in each case with 1 = yes and 0 = no).

A variety of interesting results are displayed in this table, results which generally reinforce conclusions reached in earlier chapters. For example, only in regard to the child helping the parents with domestic chores does coresidence between parent and child make a significant difference (line 2; see also chapter 6). The best educated children are more likely to provide assistance with the physical needs of the parents and with domestic chores, but they are slightly less likely to provide cash and gifts in kind to parents (lines 3–4). On the other hand, children who grew up in rural areas are significantly more likely to provide parents with assistance in cash and in kind, but not particularly to provide physical and chore assistance (line 5). The older the parents are, the more likely they are to receive physical, chore, and goods support, but they are not more likely to receive monetary assistance (lines 6–7). The number of siblings and the number of one’s own children do not appear to have a net impact on support provided to parents, except for the curious and unexpected finding that those with many siblings are more likely to provide cash support than those with few siblings (lines 8–9).

Parent and child incomes are important only as they affect the provision of monetary support from children, with the expected pattern visible in which children with higher incomes are more likely to contribute, and parents with higher incomes are less likely to receive, such support (lines 10–11). The level of parental health problems has a significant association only with the likelihood of the child providing assistance with domestic chores; the association with assistance to the parent in meeting physical needs is positive but not statistically significant (line 12). Finally, we see here the pattern also visible in chapter 8 in which past parental provision of support to children is significantly related to current provision of support by the child to the parents (lines 13–15). In this case past parental provision of monetary support has the most general and strongest effect, but measures of past parental provision of child care and chore assistance also show consistently positive associations, which in some instances are statistically significant.²⁴

The result of greatest interest in this table is, of course, that shown in the first row. There we can see that when the other predictors are controlled for statistically, married daughters are more likely than married sons to provide all four kinds of assistance to parents, and in three of these instances the relationship is statistically significant. In other words, when compared with the bivariate relations displayed in Table 7.4, the contributions of daughters emerge even more strongly here. Even in regard to providing monetary assistance, once we control for other predictors, daughters are now revealed as a little bit more likely than sons to provide such assistance to their own parents, rather than less likely.

However, it might still be the case that while daughters are more likely to contribute to the financial support of their parents than sons, the amounts they provide might be less. If that is the case then we would still have parents relying more heavily on sons than on daughters, with the larger contributions of the former outweighing the more token amounts provided by the latter. To see whether this supposition is correct, we need to subject our measures of the actual amounts provided to parents in cash and in kind to multivariate analysis (in this case via ordinary least squares regression analysis). The results of that analysis are shown in Table 7.6.

In Table 7.6 are displayed the linear regression coefficients between the natural log of the average monthly amount of cash sent to the parents and the estimated annual value of the gifts in kind provided to the parents.²⁵ For the most part the coefficients in the table appear reasonable and lead to a picture similar to those seen in the previous table, and for that reason these results will not be discussed further here. Our primary interest, once again, is in the figures in the first row of the table. Those coefficients reveal that there is not a significant difference between married daughters and sons in the average amount of money contributed to the parents each month, once other predictors have been factored out (although the relationship is positive, indicating that sons may contribute slightly more). However, controls for other predictors do not eliminate the association between child gender and the cash value of the food, clothing, and other gifts in kind given to the parents. Even when other factors are controlled for, daughters are significantly more likely than sons to provide a large cash value of such gifts in kind—by an average of 71 more yuan per year contributed. Thus the total monetary value of the support provided by married daughters (cash plus material goods) significantly exceeds that provided by married sons, once other predictors have been factored out.

We have now concluded our examination of what the data from the Baoding child questionnaire tell us about the amount of support provided to parents by married daughters versus married sons. The results are quite clear and consistent. In regard to one form of support, the provision of monetary assistance to parents, there is not a statistically significant difference between the assistance provided by daughters versus sons.²⁶ Children of both sexes often help out monetarily, and neither in likelihood of providing monetary support nor in the average amount supplied do sons do significantly more. In regard to the other three kinds of support we examined, provision of physical care, help with domestic chores, and the provision of gifts in kind, married daughters are significantly more likely than married sons to provide such assistance, and the amounts of in kind assistance are also significantly larger.

This pattern of greater support to aging parents from married daughters is a striking result, one of the most dramatic findings of our entire Baoding research project. According to these data, the traditional attitudinal bias in favor of support by sons has been eliminated in contemporary Baoding, in favor of a view supporting equal obligations by daughters as well as sons. But in the reality of the support actually provided, insofar as we can measure it with our data, the traditional bias has actually been reversed! Now, married daughters are significantly more likely to provide support for their own parents in various forms than are their brothers. Since we have indications in these data that daughters are also more likely to have obligations toward their in-laws than do their brothers, the net result is one in which the overall obligation of support for the elderly falls disproportionately on women. This pattern is one that is often found in other societies. However, that it now occurs in China’s strongly patrilineal culture is striking. We note once again that evidence cited earlier from rural China in the 1990s, and from contemporary Taiwan, does not show a comparable elimination or reversal of the patrilineal basis of parental support.

However, before proceeding to say more about this pattern of findings, we need to see whether parents report the same reality as do their children. Perhaps even if sons and daughters are providing similar amounts of support, their parents still perceive that they are primarily dependent upon their grown sons, a possibility suggested in chapter 5. In other words, perhaps the traditional patrilineal bias of Chinese culture prevents parents from recognizing and giving full value to the contributions they are receiving from their grown and married daughters. The remainder of this chapter is therefore devoted to examining similar data, but reported by our Baoding parents.

The Balance of Monetary Exchanges between Generations

In our Baoding parent questionnaires, we collected information on support of parents in several places in different formats. Much of the analysis in other chapters in this volume (particularly chapters 5 and 10) relies upon a set of questions to parent respondents about whether they received assistance in physical care, domestic chores, finances, and provision of material goods, and if so from whom (and did they provide assistance in any of these four realms, and if so to whom). That set of questions was designed to collect information on all providers (and recipients) of assistance, and not simply grown children. In a separate part of the parent questionnaire we asked respondents to enumerate each of their grown (age eighteen and above) children, and then to report how much cash they received from, and gave to, each such child each month. With access to this exclusive and exhaustive treatment of monetary exchanges with grown children, we rely on data from this latter set of questions for the analyses that follow.

The first question of interest is simply what is the overall balance of monetary exchanges with all children? We computed a measure indicating whether parents received any cash at all from their children, whether they gave any cash to their children, and if they both gave and received cash from children, whether they received more than they gave. According to these calculations, 40.3% of all our Baoding parents neither receive nor give cash in exchanges with any of their children. In other words, a substantial minority of parents interviewed are operating independently of all their children in a financial sense. It is important to stress the size of this group, since it has often been assumed that giving at least token amounts of cash to parents on a regular basis is culturally obligatory in Chinese society. Our Baoding results indicate otherwise. Of the remaining parents, 8.4% only give cash to children but do not receive any, 2.9% both give and receive but are net givers to children, 6.7% both give and receive but are net recipients, and 41.8% receive cash from children but do not give children cash (N = 958). In other words, from the parents’ point of view, the two predominant patterns are generational financial independence and an exclusive upward flow of support to parents.²⁷

In order to examine the parents’ reports on cash giving by individual children (rather than by all children taken together), we need to first take into account the fact that some parents we interviewed had more children than others. The number of adult children ranged from zero to eight, and from our 1002 parents we have intergenerational monetary exchange data on 3219 parent-child dyads. Since it seems likely that financial exchanges for those with many children will differ from those with few children, we do not want to simply pool together all parent-child dyads.²⁸ Instead, for all of those parents with two or more grown children, we randomly select one child and examine the financial exchanges for that randomly chosen parent-child dyad (N = 995).²⁹ In addition, because our primary interest here is in whether, among sets of siblings, married sons give more to their parents than do married daughters, we further select only those parent-child dyads in which the child was married and had at least one opposite sexed sibling. That restriction reduces our number of parent-child dyads to 722.

When we examine the parents’ reports on cash giving by individual children using this random parent-married child data set, we discover that 32.7% of such married children are reported to give or send cash to parents, with the figures for sons and daughters being essentially identical—32.7% and 32.6%, respectively. However, for those children who do provide cash support to parents, there is some tendency for sons to be reported as providing more per month than daughters. The average contribution per married son is 23.8 yuan, which is significantly higher than the 11.6 yuan average per married daughter (N = 710; p < .001).³⁰ These figures can be compared with the averages we computed earlier from the data supplied by the particular children we interviewed. Recall from Table 7.4 that 45.9% of married sons and 42.2% of married daughters claimed they gave cash to their parents, and that the average monthly contributions claimed by child respondents were 34.3 yuan and 17.0 yuan, respectively.

The figures computed from parents’ reports of cash from children compared to the reports of the children interviewed show a tendency toward some combination of parental underestimation or child overestimation that was commented upon in chapter 5. However, on the issue of contributions of daughters versus sons the two sources of data tell the same story. Parents and children alike agree that daughters and sons are about equally likely to provide cash payments to their parents, but that on average the amount of cash provided by sons is roughly double that provided by daughters. There is no tendency visible in these results for parents to report greater reliance on sons over daughters than we would have assumed from the data the children provided.

We also can compute figures on the cases in which parents send cash to their children, rather than receiving support. Although we have seen that a downward flow of funds to the younger generation is much less common than the traditional upward flow to parents, it turns out that when it does occur it is somewhat more likely to involve payments to married sons rather than married daughters. To be specific, in our randomly chosen parent-child dyads, 9.3% of married sons but only 3.3% of married daughters were receiving some cash from parents, and the average payment to sons of 19.1 yuan contrasted with the average payment to daughters of only 2.4 yuan (N = 714).³¹ These figures suggest the possibility that the greater cash contributions by sons to parents may be cancelled out by the greater support that sons receive from their parents. In fact, that is the net result. We constructed a measure of the net flow of cash per month between the parent and a particular child, which was simply the cash provided to the parent minus the cash received from the parent. In our randomly chosen parent-child dyads, the average net flow of cash to parents was 4.6 yuan from married sons and 9.3 yuan from married daughters (N = 706). In other words, sons may provide more but they also receive more, and, on balance, parents actually receive slightly more financial support from married daughters, although the difference is not statistically significant.

The final step in this analysis of financial exchange data provided by parents is to examine in a multivariate context the predictors of these flows in order to see whether the gender of the child has any net influence. In Table 7.7 we present a multivariate analysis using two alternative outcome measures: the natural logarithm of the monthly cash the parent reported receiving from the randomly selected child, and the natural logarithm of the net monthly cash received from that child (money given to parent minus money received from parent).³² The predictor variables we use in this analysis are mostly the same as those we used in Tables 7.5 and 7.6, except that more characteristics of the parent are available and included, and fewer of the child.³³

A number of interesting findings emerge from the figures in Table 7.7. We find the expected pattern in which child cash payments and net payments to parents are greater if the child has a high income, and less if the parent has a high income. Coresidence with the parent is closely related to both size of cash payments from the child and net payments.³⁴ More educated children provide lower cash contributions to parents but, presumably because they also receive less cash from parents, our education measures are not significantly related to net payments to parents. Other predictors, such as the age of the parent, the parent’s health and marital status, and the number of siblings and children of the child, do not have a significant net impact on either measure of intergenerational monetary exchanges.

Our primary interest, once again, is in the figures in the first row of the table for the net effect of whether the child is a son or a daughter. We see in Table 7.7 that the gender of the child does not make a significant difference for either of our monetary exchange measures. We find the same associations that we found at the bivariate level, with sons likely to provide more cash to parents each month but daughters more likely to provide slightly larger net contributions to parents. However, once the other predictors are controlled for, neither relationship is statistically significant. This finding essentially duplicates what we found earlier in this chapter when analyzing data from our child questionnaires (see Table 7.6). Sons appear to be providing more support to parents, but once other predictors are taken into account, sons are not found to be providing a significantly larger sum. To put the matter in other terms, the apparently larger cash contributions from sons are produced in large part by the fact that sons tend to have higher incomes than daughters and are more likely to coreside with parents. But with income and coresidence status controlled, the effect of child gender disappears. In other words, when daughters have high income or coreside with parents, they are also likely to contribute significantly larger sums to support their parents, with no visible tendency to contribute less just because they are daughters. In fact, given our finding that the net contribution of daughters to parents is slightly larger, and that gifts in kind from daughters are significantly larger (see Table 7.6), the total monetary burden on married daughters to support their parents is somewhat greater than on their brothers.

Throughout this chapter we have looked primarily at contributions from sons and daughters to their aging parents. In closing, we briefly look at the other side of the picture. Do parents make greater efforts to assist their sons than they do their daughters? We have already seen that there is some indication that parents are more likely to provide cash payments to married sons than daughters, although such “downward payments” are the exception rather than the general rule. Do we find the same pattern if we examine other kinds of parental assistance and support? In our parent and child questionnaires we obtained data on several kinds of parental assistance to children: help in completing homework, help in getting into the schools they attended, help in getting the first job, help in changing jobs later on, help in finding a spouse, and cash contributions to the wedding of the child. In Table 7.8 we display the results of a comparison of daughter versus son reports of receiving these various kinds of parental assistance.³⁵

We can see from Table 7.8 that there is no significant difference between the assistance that married daughters versus married sons recall receiving from parents in completing homework assignments, getting into school, changing jobs, and finding a person to marry. One curious reversal is apparent in the traditional pattern of preferential treatment of sons. Married daughters report relying somewhat more on parents to obtain their first job assignments than do married sons. It seems likely that this daughter preference is related to the 1970s practice (now discontinued) of allowing a parent to retire early in order to have a child assigned to a job in the parent’s work unit, with daughters using this route to escape from Cultural Revolution-era rural exiles, while sons were seen as more able to endure such exiles.³⁶ However, we do not have detailed enough evidence in our Baoding data to test this supposition.

The strongest differentials visible in Table 7.8 concern two traditional forms of son preference. Married sons are more than twice as likely as married daughters to report they relied on parents for access to housing. That difference is a product of the bias (noted repeatedly in this volume and in Table 7.2 in this chapter) of parents coresiding with a married son much more often than with a married daughter. Coresidence usually occurs when the son lives in parental housing, rather than through the parent moving into housing allocated to the son. This pattern thus exemplifies reliance of sons on their parents. The other large difference appears in the last row of the table, where married sons report parents contributing about 45% more to the costs of their weddings than married daughters report. This gap is again a product of adherence to “traditional” patterns, with Chinese custom dictating that the groom’s parents pay the bulk of wedding expenses. However, in this case it is notable that the average family contribution to the weddings of daughters is far from minimal—3255 yuan (versus 4725 yuan for sons). Since we do not have comparable data from the prerevolutionary period, we cannot be certain that these Baoding figures represent a shift toward more balanced contributions to the weddings of sons and daughters than in the past.³⁷ In sum, there is not a general pattern of preferential investment in the futures of sons over daughters revealed by these figures. Instead we see a pattern in which parental investments seem comparable for children of both sexes except where particular traditional customs are involved (coresidence with sons and the obligation of a groom’s family to pay for the wedding, but greater responsibility to protect daughters).

Conclusions

In the preceding pages we have engaged in a wide-ranging review of the evidence contained in the Baoding survey regarding intergenerational exchanges between parents and their sons and daughters. The findings from this review are mixed, but on balance the major conclusion we have reached is that the traditional pattern of reliance on sons has disappeared. Aging Baoding parents are as much or more likely to rely on married daughters as on married sons, and in several realms daughters provide significantly more assistance. Where sons are providing more assistance, factors such as higher incomes and coresidence with parents, rather than gender per se, provide the explanation. Although we have not reached a situation of fully equivalent relations between parents and their sons versus their daughters, as shown by the persistence of a preference for coresidence with married sons, on balance we are struck by the many realms in which gender does not appear to affect how parents relate to their grown children.

A major implication of the current pattern of intergenerational relations is that while parents may hold a variety of sentimental and traditional reasons for wanting to have at least one son, in Baoding and presumably other major Chinese cities it is no longer essential that they do so. Grown daughters can and do provide the emotional support, help in daily life, and financial assistance that parents feel they need in order to avoid a desolate old age. In terms of the basics of a secure old age, individuals who have one or more daughters but no sons are no longer to be pitied. In fact, anecdotally it is common to hear urban parents now claim that they prefer the care of daughters. As noted at the outset of this chapter, this shift toward reliance on both daughters and sons has a number of important implications, including helping to make the state’s extremely strict family planning policy (essentially a “one child” rule since 1979) somewhat palatable in Baoding and other large cities.

One major puzzle about the emergence of a dual pattern of reliance on grown daughters as well as sons remains. As we noted at the beginning of this chapter, evidence from the Chinese countryside in the 1990s and from contemporary Taiwan indicates that in those settings predominant reliance on sons is still the rule. Why has this primary dependence upon sons disappeared in Baoding? Unfortunately, this is not a question we can answer definitively with data from only one city at one point in time. However, we suspect that the shift is explained by the distinctive features of social institutions developed in urban China during the Mao era, rather than by more recent developments.

Although theories of modernization tell us that economic development may weaken patrilineal biases,³⁸ evidence exists that the shift toward reliance on daughters as well as sons predates China’s recent economic reforms. Less systematic data from urban China in the 1970s indicate that already at that time daughters as well as sons were providing support to aging parents.³⁹ Obviously, the fact that a more economically developed Chinese society, Taiwan, does not show the same breakdown of primary reliance on sons suggests that it is not how rich or developed a society is, per se, that produces sharing of the support burden by daughters.

We suspect that the breakdown of primary reliance on sons can be traced to the fact that cities in the PRC (both in the Mao era and to a considerable extent today) have certain institutional features that set them apart from contemporary rural areas in the PRC, and from rural and urban areas in both contemporary Taiwan and in mainland China before the revolution. In these other Chinese settings, corporate family property that is passed down patrilineally and family-based firms provide a central basis of family security and old-age support.⁴⁰ Even in contemporary Taiwan, family-based enterprises are the predominant form of business organization. In this sort of setting, parents manage property and enterprises for the benefit of the entire family and pass this management and control on to their sons, while daughters have no share and tend to marry out into the families of strangers.⁴¹ Sons, in return for their share in the family-based property and enterprises, provide support and care to their parents in their old age. (See chapter 9 for a more extended discussion of the contrasting histories and institutions of Taiwan and China.)

This family property- and enterprise-based nexus of intergenerational relations was eliminated by the socialist transformation of the economy of urban China during the 1950s. Family-run enterprises disappeared from urban China for thirty years, and they are only now beginning to gradually reemerge as a result of China’s reforms.⁴² Similarly, property inheritance within families became insignificant as a source of the well-being and opportunities for the younger generation, as success in life came to depend overwhelmingly on pleasing authorities within bureaucratic schools and work units (aided, to be sure, by the family’s efforts to develop personal ties with bureaucratic gatekeepers).⁴³ Families in urban China came to depend not on property management and family-based enterprises, but simply on pooling of the earnings and pensions of individual members and balancing these against the consumption needs of all. For most urban Chinese families in the mid-1990s, this was still the general case.

Given this situation, parents in urban China faced a fundamentally altered calculus in their quest for old-age security. With transmission of property and family businesses eliminated as a basis for family continuity and security, a variety of other kinds of parental support and assistance to children have to provide the basis for filial bonds of grown children with aging parents (see the discussion in chapter 8). However, these other forms of assistance (providing financial help, assistance with child-care and domestic chores, and so forth) can as readily be provided to daughters as to sons. And the evidence presented in this chapter indicates that daughters are as able as sons to reciprocate for this assistance by providing care and support to their own aging parents.

Socialist institutions in the Chinese countryside did not so thoroughly eliminate the corporate property and enterprise basis of rural families as was the case with urban families. Although rural families relied to a considerable extent on earnings from collectivized labor to meet their needs, a residual family corporate enterprise remained in the form of the family’s private plot, household sidelines, and family-reared animals, as well as in their housing. In many locales the family’s cash incomes depended primarily upon these family-based economic activities, rather than on their labor in the collective fields. Even earnings from collective labor were turned over to family heads, rather than paid to each individual earner.⁴⁴ Villages also continued to be organized in terms of traditional patrilineages, structures upon which the socialist organizational forms of communes, brigades, and teams were superimposed. In this setting there was no strong impetus to break down the tendency for parents to work to bind their sons to them, and to let their daughters leave and be replaced by daughters-in-law. Patrilocally organized families with corporate property thus remained dominant in the Chinese countryside, unlike the situation in the cities. They provided a structural basis for the rapid reemergence of family-based entrepreneurial activity in the altered conditions of the reform era.⁴⁵

Thus it seems likely that it was the socialist transformations of urban institutions carried out in the 1950s that provided the basis for the erosion of predominant reliance of parents on grown sons. The result, as we have seen, is a pattern in which both daughters and sons are relied on for support in old age, and in which in some respects daughters provide more support than do sons. Daughters are also involved in helping to support their in-laws, so in urban China today we see a pattern that is common in many other societies, in which grown women bear a disproportionate share of the burden of caring for and helping to support aging relatives.

What is not clear is whether this pattern of sharing of the support burden between sons and daughters will be an enduring pattern. Could the reform-induced loosening of the restrictions on private property and family-based firms eventually produce a revival of patrilineally based corporate families, thus encouraging a swing back to predominant reliance on sons for old age support? Such a scenario seems quite unlikely. After more than a decade of urban reforms, our survey in Baoding found no evidence of any return to traditional patterns. Furthermore, since the small birth cohorts of the “one child policy” era are just beginning to enter adulthood and marry, it seems rather more likely that the ability to rely on a daughter, if a son is not available, will be reinforced in the future, rather than weakened. Shared reliance on daughters as well as sons, while a product of socialist conditions, seems likely to endure in the altered world of reform era China.

NOTES

¹ The best general account of these tensions is still Margery Wolf, Woman and the Family in Rural Taiwan (Stanford: Stanford University Press, 1972).

² However, the elite in rural areas placed less emphasis on the exclusive claims of patrilineal kinship than did ordinary villagers, with ties through wives often used to bolster the family’s status. See the discussion in Rubie Watson, “Class Differences and Affinal Relations in South China,” Man 16 (1981): 593–615.

³ However, research in Taiwan early in the 20th century indicates that there were some localities in which enough males married into their wives’ homes to overcome the stigma involved. See the discussion in Burton Pasternak, Guests in the Dragon (New York: Columbia University Press, 1983). Where such “uxorilocal” marriages were not common, the most likely context for them to occur was generally a third or fourth son of a family of limited means marrying into a more prosperous family with daughters but no sons.

⁴ See the discussion in Martin King Whyte, “Revolutionary Change and Patrilocal Residence in China,” Ethnology 18 (1979): 211–27.

⁵ See the evidence presented in Hongqiu Yang, “The Distributive Norm of Monetary Support to Older Parents: A Look at a Township in China,” Journal of Marriage and the Family 58 (1996): 404–15. See also William Parish, Shen Chongling, and Chang Chihsiang, “Family Support Networks in the Chinese Countryside,” Hong Kong Institute of Asia-Pacific Studies working paper, 1996.

⁶ Quotation from Yean-ju Lee, William L. Parish, and Robert J. Willis, “Sons, Daughters, and Intergenerational Support in Taiwan,” American Journal of Sociology 99 (1994): 1037. See also Te-Hsiung Sun and Yin-Hsing Liu, “Changes in Intergenerational Relations in the Chinese Family: Taiwan’s Experience,” in Tradition and Change in the Asian Family, ed. Lee-Jay Cho and Moto Yada (Honolulu: East-West Center, 1994).

⁷ In much of rural China in recent times a de facto two-child limit has been enforced. Simple probabilities indicate that roughly one-quarter of all parents will end up with two daughters, and thus facing the possibility of not having any grown son to support them in old age unless they can get around the official limits. At the same time, most rural families no longer believe that the more sons the better. Indeed, some research evidence suggests that many rural families do not want more than one son. However, they also cannot contemplate survival with no sons. See the discussion in Susan Greenhalgh, Zhu Chuzhu, and Li Nan, “Restraining Population Growth in Three Chinese Villages, 1988–1993,” Population and Development Review 20 (1994): 365–95.

⁸ See the evidence presented in chapter 5 of the present volume. For earlier evidence see Martin King Whyte and William L. Parish, Urban Life in Contemporary China (Chicago: University of Chicago Press, 1984), chapter 7.

⁹ We switch in midstream between data provided by children and by parents because the child questionnaire contains fuller data on multiple forms of child contacts with parents, while data on money received from each grown child is available only from the parent questionnaire.

¹⁰ In fact, visual inspection of a large number of tables with and without this exclusion of single gender sets revealed very similar patterns.

¹¹ The ethnographic literature on Chinese families in earlier times does not portray unmarried daughters as providing less support to their parents than unmarried sons. Indeed, if anything there may be a tendency for unmarried daughters to do more than their brothers to support parents, perhaps because brothers are seen as having decades to repay their obligation to parents, while daughters will only do so up until their marriages. See Janet W. Salaff, Working Daughters of Hong Kong (Cambridge: Cambridge University Press, 1981).

¹² The full distribution on this item was as follows: for married sons 3.6% strongly agreed, 30.2% agreed, 59.4% disagreed, and 6.8% strongly disagreed; for married daughters only 1.2% strongly agreed, 11.9% agreed, 73.8% disagreed, and 13.1% strongly disagreed.

¹³ The figures presented here are the sum of the first two categories of a six category measure: daily or almost daily, several times a week, several times a month, several times a year or less, once a year or less, and no contact.

¹⁴ Keep in mind that this comparison involves only those sons and daughters who were married and had at least one opposite-sexed sibling.

¹⁵ In most respects the pattern of our findings in regard to visits with parents and parents-in-law is much the same as that revealed in another survey conducted in two of China’s largest cities, Shanghai and Tianjin. However, in that survey there was a slight but nonetheless statistically significant tendency for non-coresident sons more often than daughters to see their parents daily or almost daily. See Fuqin Bian, John R. Logan, and Yanjie Bian, “Intergenerational Relations in Urban China: Proximity, Contact, and Help to Parents,” Demography 35 (1994): 1037.

¹⁶ Of course, our traditional image is based upon patterns in Chinese villages prior to the revolution, and as noted earlier, village exogamy remained the general rule at that time. As a result, brides often lived at some distance from their natal families, and distance alone would have prevented frequent contacts. In contemporary Baoding, in contrast, most grown children live close at hand, as noted in earlier chapters. Thus, keeping in regular touch with parents or in-laws who live separately is much easier than was the case in Chinese villages of an earlier era.

¹⁷ We are uncertain how much importance to give to these disparities. The analysis presented in chapter 6 hinted that answers to these two questions could reflect anxieties about whether filial obligations are being met more than actual reductions in support to parents after marriage.

¹⁸ These are the same items analyzed in chapter 5. However, because our sample in the present chapter is restricted to married children with opposite-sexed siblings, the specific results differ somewhat.

¹⁹ For the subset of cases of grown children we are examining here (N=523), the mean monthly income of sons is 458 yuan and the mean monthly income of daughters is 337 yuan.

²⁰ These two items are included and expected to have similar effects. Those with more siblings might be less likely to provide support to their parents, or to provide less. Those with more children might be less willing and able to help support aging parents.

²¹ A natural logarithm transformation was used in these two measures to correct for the skewness of the distribution of incomes. In computing these two measures, the figure of 1 yuan was added to the reported income before taking the log, in order to avoid the problem caused by the fact that the natural log of zero is indeterminate.

²² The parent health problem scale was constructed from a complex series of questions we asked each parent. We asked first whether they were currently suffering from any ailments, and for those who said yes, we asked them to specify up to eight such current ailments. For each enumerated current ailment, we then asked them to report whether it was producing no inconvenience, some inconvenience, or major inconvenience for them. The resulting scale is a 4-point scale in which 0=no current ailments reported, 1 = one or more ailments reported, but none of them causing any inconvenience; 2=one or more ailments reported, and suffering some inconvenience as a result, but no major inconvenience; and 3=one or more ailments, and major inconvenience being suffered. (See the general results for the parent sample on this index, as reported in chapter 2.) We also ran the same analyses using an alternative health measure, an index of physical limitations (also used in chapter 2). In general the results were much the same, but the parental health problems index showed a higher level of association with the parental support measures used in the current chapter, so we use this measure of parental need for physical assistance in the present analysis.

²³ These last three measures all refer to past assistance provided by the parent(s) to the respondent during adulthood. They are included in the model as a result of Jieming Chen’s analysis in chapter 8, which shows that these measures of past parental assistance are significant predictors of current levels of support from the child to the parents.

²⁴ It is curious, however, that past parental provision of monetary support is not very predictive of current child provision of monetary support to parents, while past parental provision of childcare assistance is.

²⁵ As with our income measures, a logarithmic transformation is applied to compensate for the skewness of these measures. Once again the value of 1 yuan was added before transformation in order to keep in the analysis cases in which zero was supplied.

²⁶ To be sure, this conclusion is produced by the fact that two predictors, high child income and coresidence with parents, which are more typical of sons than of daughters, are controlled for in the models utilized in Tables 7.5 and 7.6. Both of these predictors are associated with higher levels of financial support for parents. What the multivariate analysis tells us is that it is high personal income and coresidence with parents that is predictive of high levels of financial support provided, not the gender of the child.

²⁷ These percentages differ somewhat from the calculations presented in chapter 2 because they are based upon separate questions dealing with each child, rather than general questions about financial support given and received. However, the general picture they provide, of parental independence and reliance on children predominating, is the same. The mean total cash received from all children per month per parent is 95.90 yuan; the mean cash given to all children per month per parent is 27.04 yuan; and the net amount of cash received per month per parent after deducting cash given to children is 69.74 yuan. If we examine the exchanges for each individual parent-child dyad (N=3219), rather than for each parent with all children as a group, then 57.3% of children neither give to, nor receive money regularly from, their parents, 5.4% only receive money, 34.8% only give money, .5% receive more than they give, and 1.8% give more than they receive. Of course, these calculations are based only on questions concerning cash exchanges, rather than gifts in kind. If the latter were factored into the picture no doubt there would be fewer parents operating independently.

28 In fact the results presented in Table 7.5 suggest a curious pattern in which those with large numbers of sibilings are more likely than others to send parents cash contributions.

²⁹ Our thanks to Jieming Chen for constructing the data commands in order to form the full parent-child dyad and the random parent-grown child data sets that are used in the analyses which follow.

³⁰ In this and subsequent averages, we consider all children on whom we have data, and not only those who sent (or received) money. In other words, these averages include values of “0” for those cases in which no cash was given (or received).

³¹ These two contrasts were not strong enough to surpass the p=0.05 level of statistical significance.

³² A logarithmic transformation of these outcome measures is carried out to compensate for the skewed distribution of values on both measures. For the cash given to parents we added 1 yuan before the transformation to avoid taking the log of zero; for the net transfer we added 4001 yuan before the transformation to obtain a positive distribution—a procedure required because the minimum value of the net cash measure was -4000 yuan (in other words, one parent claimed to regularly give 4000 yuan to the randomly selected child). The effect of this manipulation was to produce a much wider span of values in the net cash than in the child to parent cash measure, so three decimal places are shown in that column of Table 7.7.

³³ Note that we include in Table 7.7 a measure of whether the parent was of rural origin, rather than the child. We also include a measure of whether the parent is a party member and a simple measure of parental education (with 1=upper middle school and above, and otherwise=0).

³⁴ Note that this finding appears to contradict the conclusion in chapter 6 that coresidence does not make a significant net difference for the provision of financial support from grown children to parents. From Table 7.5 as well as the analysis in chapter 6 we know that whether a child lives with the parent does not have a net impact on whether the child will contribute money to the parent or parents. However, for married children with opposite sexed siblings, at least, the amount of money given per month is nonetheless higher for coresident children.

³⁵ The figures in the first six rows of the table represent the number of instances in which the child said they depended entirely (wanduan kao) on the parents in that particular realm. The results are much the same whether we use the responses to these questions provided by parents or by the children we interviewed. As in earlier analyses, we restrict our attention here to cases of children who are married and have siblings of the opposite sex.

36 The practice of taking early retirement in exchange for the work unit providing a job for one’s child, already noted in chapters 3–4, was referred to as the dingti system. In such cases, traditional concern about female virtue and the hazards posed by distant rural assignments may have counterbalanced preference for promoting the economic success of sons. See the discussion in Martin K. Whyte and William L. Parish, Urban Life in Contemporary China (Chicago: University of Chicago Press, 1984), 222. However, one study of rates of return from rural exile did not find a significant gender difference. See Xueguang Zhou and Liren Hou, “Children of the Cultural Revolution: The State and the Life Course in the People’s Republic of China,” American Sociological Review 64 (1999): 12–36. On the general movement to send urban educated youths to the countryside in the decade after 1968, see Thomas Bernstein, Up to the Mountains, Down to the Villages (New Haven: Yale University. Press, 1977).

³⁷ Traditionally the major expenses for the groom’s family involved the equipping of a bride price payment to the bride’s family and the staging of an elaborate wedding feast. The bride’s family was not expected to pay for an elaborate wedding feast, and their primary expense was the equipping of a dowry for their daughter. Expenditures on dowries were large among the upper classes but often minimal among the poor.

³⁸ See William J. Goode, World Revolution and Family Patterns (New York: The Free Press, 1963).

³⁹ See Whyte and Parish, Urban Life in Contemporary China, 222. See also Deborah Davis-Friedmann, Long Lives: Chinese Elderly and the Communist Revolution (Cambridge: Harvard University Press, 1983).

⁴⁰ See the discussion in Myron Cohen, House United, House Divided (New York: Columbia University. Press, 1976).

⁴¹ See the discussion in Susan Greenhalgh, “Sexual Stratification: The Other Side of ‘Growth with Equity’ in East Asia,” Population and Development Review 11 (1985): 265–314.

⁴² In our Baoding sample such firms are virtually invisible. As noted in chapter 4, only 3.4% of the grown children we interviewed were either self-employed or working in private firms of any kind.

⁴³ The most important form of property that some families possessed during the prereform era was the privately owned housing that a minority of urban families was able to hold onto. However, the restrictions of bureaucratic socialism prevented families from employing their housing as a money-earning asset. Other forms of family property, such as bank accounts and consumer durables, similarly could not be put to entrepreneurial use and could be readily divided among children (including daughters) via inheritance.

⁴⁴ Housing was also in almost all areas privately owned and inherited, unlike the situation in China’s cities, where housing rented from work units and city housing offices was the general rule.

⁴⁵ See the discussion in Martin Whyte, “The Social Basis of China’s Economic Development,” The China Quarterly 144 (1995): 37–57.

57