Bequest motives of older people in rural China: from the perspective of intergenerational support

Abstract

China, a society long characterized by traditional Confucianism and the practice of filial piety, is undergoing dramatic socioeconomic development and erosion of traditions. In this rapidly changing context, the motives behind rural parental bequests have not been well studied. This paper draws on a survey conducted in rural Anhui province of China and uses hierarchical linear models to examine whether three kinds of support from children are associated with older parents’ bequest motives. We find that while instrumental support accords with an exchange model of motivation, financial transfer is consistent with an altruistic motivation for parental bequest plans. Offspring gender is strongly associated with parental planning of bequests; family division is a strong correlate of parents’ bequest plan and reflects exchange motivation. These findings have important implications for research on traditional culture and an evolving social security system.

Introduction

China, with its long tradition of filial piety in the context of Confucianism, is undergoing dramatic socioeconomic and demographic changes. Traditionally, the practice of filial piety in the patrilineal system requires that as parents raise their children, they potentially deplete their own resources to benefit their children’s well-being. Adult male children, in return, are expected to be primary care providers for their older parents due to the absence of a social security system and the depletion of parental resources through ‘serial division of the family’ (Cong and Silverstein 2012). The motivation for such intergenerational exchange is embedded in altruism and the norms of reciprocity that maintain familial solidarity over the life course (Silverstein et al. 2002). However, with rapid socioeconomic development and modernization, the strong culture of filial piety has weakened dramatically (Chen and Silverstein 2000; Zhan and Montgomery 2003; Chou 2011).

Meanwhile, China is undergoing unprecedented population aging due to the nation’s rapidly increasing life expectancy and its steep fertility decline. According to the 2010 census, there were 119 million people aged 65 or older, including 66.67 million in rural areas (PCO 2012). For rural older people aged 60 or over, 47.74 % list family member support as their primary income source, while 41.18 % of all older people rely on income from their own labor (PCO 2012). Even if older people rely on their family members, they still work actively to earn money. They may receive income from their household workshops or small businesses (Li and Tracy 1999). In rural areas, parents may divide the family property early and live apart from their adult children in order to accumulate money and support themselves before they lose the capacity to work (He 2008). In this way, they can have accumulated savings and personal property upon their death.

Bequests, as a way of downward intergenerational property transfer, are gradually playing a role. According to a survey of 500 respondents implemented in 2007 in rural Hubei and Liaoning provinces, 71.14 % of rural parents want to leave as large a bequest as possible to their children (Yin 2010). Another survey conducted around 2005 indicated that despite China’s high poverty levels, 43 % of females expected or received inheritance, and there was no urban–rural difference (Nauck 2010). Property transfer via bequests by parents affects not only children’s behavior in supporting their parents, but also affects these children’s family members. As people become more affluent, some die with relatively large bequests, and there are more disputes over these bequests (Li and Ma 2010).

Whether the motives behind bequests by older parents are altruistic or can be viewed as an exchange is unclear because there are differences among nations, regions, traditions, beliefs, etc. As China’s traditional family patterns and the practice of filial piety are eroding under dramatic socioeconomic change, and as rural parental bequests and related problems are attracting more attention from both the public and academia, we can ask whether older Chinese parents are still as altruistic as their traditional image in planning their bequests. What factors are associated with the bequest motives of older parents in this traditional family-centered but now rapidly modernizing society? What support behaviors by children are correlated with parents’ decisions concerning the distribution of their bequests? In this paper, we examine the motives behind rural older parents’ bequests from the perspective of children’s support for their parents.

Literature review

Chinese context: filial piety and gender

Confucianism originated in China. In this ethical-moral system, family members are embedded in a network of lifelong mutual obligations toward each other (Schwarz et al. 2010). The interdependence of generations reflects the belief in filial piety. As a central concept in Confucianism, filial piety defines how children should treat their parents, and indoctrinates adult children to provide direct physical, financial, and emotional care for their aging parents (Zhan et al. 2011). Filial piety is of great importance in China and family obligations are fundamental in parent–child relationships (Xie and Zhu 2009; Lin and Yi 2011; Mao and Chi 2011). Parents, especially rural elders who have no formal pension after the depletion of their resources for their sons’ marriages, rely on their adult children as primary care providers in their old age. Chinese policies and laws articulate and stipulate adult children’s obligations to care for their older parents (Cong and Silverstein 2012). However, due to increased modernization, urbanization, and economic development, the traditional practice of filial piety is tending to dissipate, and the willingness and capacity of adult children to care for their older parents have declined. Social, demographic, and economic changes have influenced traditional intergenerational relations (Chen and Silverstein 2000; Zhan and Montgomery 2003).

The Confucian system is characterized by patriarchality, patrilineality, and patrilocality, which means that gender is a determinant of children’s provision of support to their parents. Parents invest comparatively more in male children’s well-being, and expect future economic support from their sons (Zuo et al. 2011). Sons have always been expected to be the ultimate financial providers, whereas daughters have traditionally been married out to their husbands’ families, and their support of parents is more voluntary and more likely to be out of love than from obligation to their parents (Zhan and Montgomery 2003; Cong and Silverstein 2012). With recent fast social change, however, daughters have taken increasing responsibility for the care of their older parents in terms of emotional, instrumental, and financial support (Schwarz et al. 2010; Cong and Silverstein 2012). Some recent researches have even found that daughters provided more support to older parents and were perceived as more filial than their male siblings (Luo and Zhan 2012). In urban areas, parents received more financial support from married daughters than from married sons (Zhan and Montgomery 2003; Xie and Zhu 2009), and daughters were involved increasingly in providing physical care for parents (Zhan and Montgomery 2003). In rural areas, sons are expected to provide fundamental support for their older parents, whereas daughters tend to provide support for their parents through emotional care in daily life (Jin et al. 2006), but the number of adult daughters serving as the main caregivers for their older parents is also increasing (Yao 2005).

Currently, a rural household will usually be divided if parents have more than one son. Parents are obligated to marry off their sons and build them houses (Gao 2005). Cong and Silverstein (2012) refer to this as ‘the depletion of resources in the form of serial division of the family’; that is, parents transfer their resources to sons one by one when the latter get married. To ensure parental support and reduce intergenerational conflicts, the Family Support Agreement emerged, according to which a voluntary contract on providing support to parents is signed at household division between older parents and adult children (Chou 2011). China’s Law of Inheritance stipulates that descendants of both sexes have equal inheritance rights and remain eligible for bequests, although in patrilineal societies, inheritance takes place mainly between male members of the kinship system (Nauck 2010). Traditionally in China, it is most unlikely that married-out daughters receive parental bequests, and daughters take it for granted that their brothers should inherit their parents’ property even though a daughter may offer as much or even more support to her parents (Yang 2013). Married-out daughters receive their transfer as a dowry at marriage; they usually do not obtain bequests from parents and give up their inheritance right after they are married out (Gao 2005). In families without sons, some property will be transferred to daughters upon death. China’s one-child policy is undermining patrilineal norms (Deutsch 2006).

Theoretical models

There are three main theoretical models to explain bequests: the altruism motive model, the exchange motive model (strategic bequest motive), and the accidental motive model (non-expectant bequest) (McGranahan 2006). The accidental motive model originates from the unpredictability of death. According to this model, older parents do not actively make decisions about their bequests. Accidental bequests may mask individuals’ original motives behind their bequest plans (Hurd 2003). Our interest here is in active decisions of older parents concerning their bequests; therefore, we focus on the altruism and exchange motives.

The altruism motive model assumes that parents are altruistic in their bequests (Barro 1974; Becker 1974) whether or not their children reciprocate (Yin 2010). In particular, they leave bequests to their poor or needy children (Hurd and Smith 2002). According to the altruism motive model, parents care about the well-being of their children, and support them according to the latter’s needs. To those children who are less educated, unemployed, or of lower social status, parents offer more economic support (Wilhelm 1996; Kalmijn 2013). If parents expect some of their children to have low income, they tend to leave more bequests to them (Laitner and Juster 1996).

The exchange motive model assumes that parents are not concerned about their children’s lives and status but try to influence their children’s support behavior by their bequests. Parents urge their children to take more care of them and offer more support to them in their old age (Bernheim et al. 1985). According to this model, parents’ bequests to their children should be positively correlated with the support from their children. This model is tested by assessing the relationship between parents’ bequests (money or other property) and their children’s support (for example, life care) (Izuhara 2009).The more care parents get from a child, the more they are inclined to bequeath to this child. Parents can threaten not to leave any bequest to those children who do not support them in order to receive more attention and care (Clignet 2009).

Traditionally, intergenerational support in China is more likely to be altruistic. Parents take responsibility for nurturing their children, while children take on the duty of supporting their parents in their old age (Zimmer and Kwong 2003). Parents transfer property to their children from altruistic motives, and offer more help to those children who have less income or are less educated (Li et al. 2010). Sometimes parents help support children who are poor with money or property they receive from well-off children (Yang 2013). However, some studies conclude that children support their parents partly in expectation of bequests: only when parents regard bequests as their children’s reward for supporting them might attract their children’s attention and obtain their children’s support (Liao 2010). Parental support is directly associated with inheritance in rural areas (Gao 2005). Hence, bequest motives of Chinese parents are also characterized by exchange.

According to the altruism motive, parents will leave more bequests to a child in need, independently of the amount of support they have received from this child. Exchange theory predicts that bequest allocations will be strongly related to the exchange balance in intergenerational relationships; that is, receipt of financial, instrumental, and emotional support from children should be taken into consideration when parents allocate their bequests. We examine here the weighting of altruism and exchange motives in older parents’ plans for their bequests.

Data and method

Data

The data are from the 2009 survey of “Well-being of Rural Older People Survey in Chaohu, Anhui Province,” which is a follow-up survey; previous rounds had been conducted in 2001, 2003 and 2006. These data have previously been employed by scholars (Silverstein et al. 2006; Cong and Silverstein 2008; Song 2008; Zuo et al. 2011). The survey was originally fielded in April 2001 as a joint project between Xi’an Jiaotong University and the University of Southern California. With a stratified multistage method, adults aged 60 years and older were randomly selected from 72 randomly selected villages within six rural townships in the Chaohu prefecture of Anhui province. The baseline sample of 2001 included 1,800 respondents, but as there was sample attrition due to mortality and migration, new interviewees were added to the follow-up surveys. The present study uses the 2009 cross-sectional data as information on bequests was not collected in the three preceding surveys. In the 2009 survey, 1,571 people aged 60 or older were interviewed, and 1,224 questionnaires were collected, with a response rate of 77.9 %. To analyze bequest motives, we limited the sample to parents with at least one child and excluded 50 older people without any children at the time of interview, which reduced the sample to 1,174 respondents. After further deleting cases with missing data, or data with severe logical inconsistencies, the working sample size for this bequest study was 1,111, with 4,202 adult children all of whom were aged 16 years or older, producing a missing rate of 5.3 % at the parental level. According to Lynch’s (2003) recommended method, we implemented Multiple Imputation and found our results were robust. We present the results based on listwise deletion in this paper.

Variables

Dependent variables

In examining bequest motives, previous studies have indicated the utility of “subjective” measures of expected bequests (McGarry 1999; Hurd and Smith 2002). In this study, we take the bequests specified in the questionnaire answered by older people as two dependent variables: one is whether the respondent plans to bequeath to a child (0: without plan; 1: with plan); the other is how much he/she plans to leave to a child. If they plan to leave a bequest to a child, he/she was asked for the exact amount, or he/she could choose from the following categories based on RMB currency (100 RMB = 12.5 Euros): (1) below 200, (2) 200–499, (3) 500–999, (4) 1,000–2,999, (5) 3,000–4,999, (6) 5,000–9,999, (7) 10,000–19,999, and (8) 20,000 or above. In our analysis, we used the reported exact amount if available. Otherwise, we interpolated an amount by taking the median value of the chosen category. The natural logarithm of this RMB value (+1) was used in our statistical analysis.

Independent variables

The primary independent variables of interest are intergenerational support from children to parents. Intergenerational support can be classified as financial support, instrumental support, and emotional support. Financial support refers to material help from children to parents, such as money, clothes, and food. Instrumental support refers to life care, including domestic help by children such as washing, cooking, and cleaning for their parents, and daily life care such as clothing, feeding, and bathing their parents. Emotional support refers to parents’ subjective judgment of emotional closeness to a child.

We measure financial support as the total support in money and property from children to their older parent during the last 12 months; the value of property was transformed into an amount of money. If an older interviewee could not provide a specific figure, he/she was asked to choose from given categories. A given category was transformed into an amount of money and its logarithm used in the same way as for the dependent variables.

Instrumental support is measured by two questions: (1) Does the child help you with household chores? (2) Is the child responsible for your life care, including clothing, feeding, and bathing you? If neither help was provided, the variable was set to 0, otherwise it was set to 1.

Emotional support is assessed with a combination of three items, the first is “do you feel close (emotionally) to this child?” 0 (not close), 1 (fair), and 2 (close); the second is “overall how do you get along well with this child?”, 0 (not well), 1 (fair), and 2 (well); the third is “do you think this child is willing to listen to your difficulties and troubles?”, 0 (not willing), 1 (fair), and 2 (willing). We computed an additive scale ranging from 0 to 6 to measure emotional support; its reliability coefficient was 0.94. Other scholars who used previous waves of this dataset also used this method (Silverstein et al. 2006; Cong and Silverstein 2008; Song 2008).

Control variables

In our analysis, we controlled for the confounding effects of children’s and parents’ individual characteristics. Children’s characteristics include age, gender, marital status, educational level, type of residence, whether the child has his/her own house, and whether the child is living separately from his/her parents. Age was squared as a continuous variable, and other variables are taken as categorical variables or as dummy variables.

Parents’ characteristics include gender, age group (0 = 60–74, 1 ≥ 75), marital status, educational levels, health status, savings, whether the parent has his/her own house, and work status. Savings is the amount specified by the parent and his/her spouse, and the natural logarithm of this savings (its value plus one) was used in the regression analysis. Work status falls into three categories in the questionnaire: not working, working part time, and working full time. In the regression analysis, we combine part-time work and full-time work, and turn work status into a binary variable, 0 = not working and 1 = working full time or part time. The definition and descriptive statistics of variables is shown in Table 1.

Table 1.

Definition and descriptive statistics of variables

Variable	Definition and measurement	Children (n = 4,202)		Son (n = 2,217)
		M	SD	M	SD
Whether a child is in parent’s plan for bequestsa	1 = yes	17.7 %		27.3 %
Amount of bequest to a child (ln + 1)	0–12.61	1.35	2.99	2.11	3.54
Amount of bequest to a child (RMB)	0–300,000	1,150.34	7,756.14	1,903.18	10,166.46
Financial support (ln + 1)	0–10.13	5.01	2.45	4.92	2.66
Financial support (RMB)	0–25,000	631.87	1,227.59	728.07	1,421.76
Instrumental supporta	1 = provided	13.3 %		15.6 %
Emotional support	0–6	4.45	1.49	4.28	1.53
Children’s characteristics
The square of actual age		1,868.04	773.82	1,889.06	768.57
Gendera	1 = female	47.2 %
Education (illiteracy)a
Primary schoola	1 = yes	32.8 %		33.2 %
Middle schoola	1 = yes	31.1 %		40.9 %
High school and abovea	1 = yes	8.5 %		12.5 %
Housea	1 = own house	81.2 %		78.5 %
Residencea	0 = rural, 1 = urban	10.3 %		11.0 %
Marital status (married and co-residing with spouse)a
Divorced or separateda	1 = yes	14.9 %		14.7 %
Widowed or unmarrieda	1 = yes	5.8 %		7.1 %
Dividing from the family or nota	1 = yes	46.8 %		68.2 %

Parents characteristics		Parents (n = 1,111)		Parents (n = 1,050)
Plans to bequeatha	1 = yes	41.2 %		38.9 %
Total of planned bequest	0–302,000	4,350.81	15,652.36	4,018.42	14,987.8
Gendera	1 = female	52.3 %		52.7 %
Marital statusa	1 = with spouse	64.1 %		64.2 %
Educationa	1 = with education	32.8 %		32.5 %
Age groupa	0 = 60–74 1 ≥ 75	37.9 %		37.5 %
Working or nota	1 = working	57.2 %		57.1 %
Amount of savings	0–150,000	2,116.6	7,027.9	2,123.00	7,145.60
Savings (ln + 1)	0–11.92	3.26	3.86	3.27	3.86
Housea	1 = own house	52.9 %		52.0 %
Health status (not good)
Faira	1 = yes	45.7 %		46.2 %
Gooda	1 = yes	28.3 %		28.8

^aThese are dummy variables

Statistical methods

We used hierarchical linear models (HLM) in the analysis for the following two reasons; on one hand, some characteristics of different children of older parents may be correlated; on the other hand, children’s behavior may be affected by their parents’ characteristics. We used a multilevel random effect models to distinguish effects of older parents and those of children.

For the binary-dependent variable of whether an older parent plans to bequeath to a child, the multilevel model was as follows:

$$\text{Level}1:log\frac{p}{1-p}=B_{0j}+B_{1\mathrm{j}}{X}_{\mathrm{ij}}+e_{\mathrm{ij}}$$ 1

$$\text{Level}2:B_{0\mathrm{j}}=C_{00}+C_{01}{Z}_j+u_{0\mathrm{j}}$$ 2

In Eq. (1), p is the probability that a parent plans to bequeath to a child, X _ij is the child’s feature, i is the child’s code, and j is the parent’s code. Considering the hierarchical structure of the data, the intercepts in Eq. (1) are established using Eq. (2), where Z _j are the variables at the parent level and are assessed using parents’ characteristics, with C ₀₀, C ₀₁ to be estimated. We treat u _0j as a random variable and assume that it has a normal distribution N (0, ${\mathit{\sigma }}_j^2$).

For the continuous variable of the bequest amount a parent plans to leave for a child, the multilevel model was as follows:

$$\text{Level}1:y=B_{0\mathrm{j}}+B_{1j}{X}_{\mathrm{ij}}+e_{\mathrm{ij}}$$ 3

$$\text{Level}2:B_{0\mathrm{j}}=C_{00}+C_{01}{Z}_{\mathrm{j}}+u_{0\mathrm{j}}$$ 4

In Eq. (3), y is the natural logarithm of the bequest amount a parent plans to leave for a child, and the other variables and coefficients are the same as in Eqs. (1) and (2).

Results

Descriptive results

The descriptive statistics are shown in Table 1. Of the 1,111 interviewed older parents, 57.2 % are still working full or part time, and 41.2 % plan to leave a bequest to at least one of their children. The average savings amount is 2,116.6 RMB (265 Euros). For those parents who plan to bequeath to children, the amount of planned bequest averages 4,350.8 RMB (544 Euros), almost double the average savings. Of the 4,202 children, 17.7 % (745 persons, 605 males, and 140 females) are in their parents’ plan for a bequest. Their average financial support to parent is 631.8 RMB (79 Euros), and 13.3 % of children provide instrumental support, while the emotional support averages 4.45 (SD = 1.49) on a 6-point scale, indicating that parents feel a relatively high emotional closeness to their children.

As bequests in China are largely gender-based, we carried out an analysis without daughters, as shown in Table 1. After deleting cases without sons, we are left with a sample of 1,050 older parents, with 2,217 sons. For this sample, 38.9 % plan to leave a bequest to at least one son. The average savings amount is 2,123.0 RMB (265 Euros). For these parents who plan to bequeath to children, the amount of planned bequest averages 4,018.4 RMB (502 Euros). Of the 2,217 sons, 27.3 % (605 sons out of 2,217) are in their parent’s plan for a bequest. Their average financial support to parents was 728.0 RMB (91 Euros), and 15.6 % of sons provide instrumental support, while the emotional support averages 4.28 (SD = 1.53) on a 6-point scale, indicating that parents feel a relatively high emotional closeness to their sons.

Before turning to the regression analyses, we examined the connection between parental wealth and their bequest plan. As indicated in Table 2, 54.1 % of parents with savings plan to leave bequests, whereas only 31.5 % of parents without savings have such a plan. Further, 67.5 % of those with their own house plan to leave bequests, much higher than among the 11.7 % of parents who do not own their house.

Table 2.

Frequency distribution of bequest by saving and house

	Whether a parent plans to bequeath children			χ2 (Chi squared)
	No	Yes	Total
(1) Savings
No	433 (68.5 %)	199 (31.5 %)	632 (100 %)	57.36***
Yes	220 (45.9 %)	259 (54.1 %)	479 (100 %)
(2) House
No	462 (88.3 %)	61 (11.7 %)	523 (100 %)	356.38***
Yes	191 (32.5 %)	397 (67.5 %)	588 (100 %)

*** p < 0.001; ** p < 0.01; * p < 0.05; ⁺ p < 0.1

Regression results including both sons and daughters

Financial support from children

The financial support a parent received from a child is negatively correlated with the parent’s plan to bequeath to the child, as shown in Table 3. Thus, parents plan to leave a bequest and a relatively larger bequest to those children who do not provide them with financial support, suggesting that parents’ bequest is not due to an exchange motive from the financial support perspective. This is further supported by the variable of children’s house ownership. In rural China, it is important for a family to have its own house, or it will be regarded as poor. Table 3 shows that parents are more inclined to bequeath to those children who do not have their own house. Parents are more likely to bequeath a larger amount to children with a rural household registration type and to unmarried or widowed children. All of this indicates that parents tend to leave bequests to children who are not well off. Providing less financial support to a parent may be due primarily to the child’s disadvantaged economic status. Thus, the arrangements of rural parents’ bequests are altruistic from the perspective of financial support by children.

Table 3.

Multilevel regression model of parents’ bequest motive

	Plan to bequeath to a child or not Model 1 (yes or no)	Amount of bequest to a child Model 2
Fixed effect
Intercept	−1.854***	2.488***
Financial support	−0.042*	−0.064**
Instrumental support	0.443***	0.293*
Emotional support	−0.031	−0.008
Children’s characteristics
Gender	−2.043***	−1.912***
Age	−0.000	−0.000+
Education (illiterate)
Primary school	0.109	−0.156
Junior high	0.102	−0.010
Senior high or more	0.062	−0.001
House	−0.701***	−0.915***
Residence	−0.314+	−0.315+
Marriage (married and co-residing with spouse)
Divorced or separated	−0.154	−0.127
Widowed or unmarried	0.311+	0.413*
Dividing from the family or not	−0.045	−0.257*
Parents’ characteristics
Gender	−0.322*	−0.241+
Marital status	−0.029	−0.023
Education	−0.112	0.039
Age group	−0.136	−0.111
Savings	0.107***	0.099***
House	2.214***	1.801***
Working or not	0.186	0.095
Health status (not good)
Fair	0.267+	0.209+
Good	0.372*	0.348*
Random effects
Level 2: in intercept	2.429***	2.105***
Level 1: within-person	–	4.939
Deviance	–	19,727.57
The number of children	4,202	4,202
The number of parents	1,111	1,111

*** p < 0.001; ** p < 0.01; * p < 0.05; ⁺  p < 0.1

Instrumental support from children

Parents’ bequest plans are strongly correlated with the instrumental support a child has provided. Children who offered life care to their parents are more likely to receive a bequest, and it will be larger, which accords with the exchange motive. Emotional support is not (statistically) significantly related to parents’ bequest motive.

Some control variables are associated with the odds of parents’ bequest plan. On the parents’ side, parent’s gender, parents’ savings, having a house, and health status are correlates. Parents with savings, their own house, and in good health are more likely to leave a bequest to children than those without savings. On the children’s side, compared to daughters, sons are more likely to receive bequests, and these will be larger.

Regression results including only sons

As inheritance is determined by children’s gender and bequests are mainly left to sons, we discuss below parents’ motives for bequests to their sons and include only sons in the regression analysis shown in Table 4. We include parents with at least one son, and the sample size is now 1,050, with 2,217 sons.

Table 4.

Multilevel regression model of parents’ motive of bequest to sons

	Plan to bequeath to a son or not Model 1 (yes or no)	Amount of bequest to a son Model 2
Fixed effect
Intercept	−1.899***	1.938***
Financial support	−0.034+	−0.033
Instrumental support	0.434***	0.428**
Emotional support	−0.021	−0.009
Children’s characteristics
Age	−0.000	−0.000
Education (illiterate)
Primary school	0.227	0.186
Junior high	0.095	0.017
Senior high or more	0.159	0.031
House	−0.756***	−1.182 ***
Residence	−0.787**	−1.042***
Marriage (married and co-residing with spouse)
Divorced or separated	−0.216	−0.148
Widowed or unmarried	0.145	0.587*
Dividing from the family or not	−0.322**	−0.635***
Parents’ characteristics
Gender	−0.389*	−0.473*
Marital status	0.215	0.306
Education	−0.218	0.018
Age group	−0.303+	−0.477*
Savings	0.114***	0.149***
House	2.256***	2.878***
Working or not	0.120	0.009
Health outcomes (not good)
Fair	0.388*	0.414*
Good	0.729***	1.000***
Random effects
Level 2: in intercept	2.327***	6.162***
Level 1: within-person	–	3.498
Deviance	–	10,661.04
The number of sons	2,217	2,217
The number of parents	1,050	1,050

*** p < 0.001; ** p < 0.01; * p < 0.05; ⁺ p < 0.1

We see in Table 4 that emotional support is not a predictor of parent’s bequest plan, while these plans are still strongly correlated with instrumental support from sons. One noteworthy change concerns whether a son has divided his family from his parents. The odds and amount of bequest increased for those who had not divided from their parents. This may be due to the traditional family division practice where, for parents with more than one son, older sons usually divide their family from their parents after marriage. An agreement will usually be reached upon family division about parental support, and inheritance is generally directly related to obligations to parents (Zheng 2005). A son who does not divide from his parents will usually inherit the parents’ house and other bequests as well.

Discussion and conclusion

China is now confronted with rapid aging and a dramatic increase in the number of older people in rural areas. Traditional practice of filial piety and gender norms is strongly influenced by rapid modernization and socioeconomic development, and so is intergenerational exchange. Individuals’ motives for bequests reflect both children’s needs (altruism) and older people’s own desire for support while alive (exchange theory) (Kim et al. 2013). In this paper, we find a mix of both altruistic and exchange motives in bequest plans.

Our study indicates that children’s financial support is negatively correlated with the parent’s plan of whether to bequeath to the child, and children without their own houses are more likely to be bequeathed. This is largely consistent with Li et al. (2010), and Yang (2013) whose studies show that parents are inclined to transfer more wealth to children who are economically disadvantaged. Our finding shows that in terms of economic exchange, Chinese parents are still altruistic in their bequest plans.

Our results also show that a parent’s plan to bequeath to a child is significantly correlated with instrumental support the child provided to that parent. Parents’ bequest motives are of the exchange type with respect to instrumental support from their children, which indicates that older parents need help with personal care and household chores. On one hand, children’s care enhances parents’ satisfaction with life when parents’ functioning declines and their self-reliance is reduced (Yao 2006). On the other hand, sick or disabled parents desire most instrumental support from their children, especially when rural young adult children migrate to cities for work (Huang 2012).

Given their disadvantaged economic status, rural older people have limited ability to buy old age care from the market (Shi 2012), and with limited availability of rural medical care, these older people are frequently in poor health (Lei et al. 2014).Therefore, those who provide instrumental support will be repaid in parental bequests, proving an exchange motive behind parental planning of their bequests.

Emotional support from children is not statistically significant. One possible reason may be that there is little emotional communication between parents and children. In China, emotional comfort of older people comes mainly from their spouses (Zeng 2010), and older males, especially widowed older males, are almost excluded from the outside world when it comes to emotional communication, even with their children (Song 2008). Further, parents may be reluctant to admit to emotional distance from children, especially to interviewers, as such children may be regarded as unfilial, which in turn reflects parental failure.

With regard to gender, our results again show the institutionally patrilineal pattern of bequeathing mainly to sons, as the gender of a child is a very strong predictor of parents’ bequest plans. However, we also see that daughters almost equal sons as caregivers to parents, and of those in parents’ bequest plans, daughters account for 20 %, although we must admit that most of the listed daughters had parents with no male offspring. It may be predicted that with the maturing of the “only-child generation,” patrilineal inheritance may weaken and inheritance may become more equally distributed even in rural China; that is, as claimed by Deutsch (2006) the one-child policy may be undermining patrilineal norms.

Among sons, family division is strongly correlated with parental bequest. As a deeply entrenched custom, parents with more than one son usually divide the families, which may result in a voluntary Family Support Agreement (Chou 2011). Such agreements usually include items about parental support, as well as inheritance rights. Generally, inheritance rights are contingent on obligations to provide parental support. The son who does not divide from his parents is usually the primary caregiver, and receives the parents’ bequest (Zheng 2005). Rural parents will usually allocate their bequest according to this custom, rather than writing a will. Perhaps this is the reason that Yin (2010) finds that children are more likely to live with their older parents if their parents have a bequest motive, and concludes that children are selfishly motivated in coresidence with parents in China. In our understanding, the correlation between family division and parent’s bequest plan is a reflection of rural customs, with or without a written agreement between parents and sons. In this sense, family division and bequests partially reflect a kind of exchange relationship, rather than pure filial piety.

Our study has several limitations. First, as in other studies based on cross-sectional survey, this study cannot establish causal links between independent variables and the dependent variables. Second, due to the unavailability of data on older people’s wealth and the value of their house, we just use their savings and whether they own a house to measure their wealth. This is not exactly a substitute for the wealth of older people. Third, the survey was conducted in rural Anhui province, where socioeconomic development is ranked at the middle in China. But as China is a vast country characterized by great regional differences, and the practice of filial piety, gender norms, family division may vary dramatically across regions, it is difficult to determine whether our results can be generalized within the rural context of Chinese family life.

Despite these limitations, our study contributes to the literature on bequests, a subject that is seriously under examined in China. Previous studies on intergenerational exchange focus on inter-vivos exchange, but few have touched on rural bequest motives in China. Inter-vivos transfers and bequests appear to follow different logics and have different effects, and therefore need to be studied separately (Kohli 2004). Our findings show that parents exhibit a mixture of motives in bequest planning. In terms of economic exchange, parents show an altruism motive; however, when it comes to instrumental support from children, parents hold an exchange motive. We find gender to be an important predictor of parental bequests, but we point out new trends in the entrenched patrilineal institution of bequest. More importantly, our findings suggest that, as an institutional arrangement in the patrilineal system, family division still dominates and reflects mainly an exchange relationship regarding parental bequests.

China, traditionally claimed to be characterized by reciprocal intergenerational exchange, has been reshaping intergenerational relations in the rural context of rapid socioeconomic development. While self-interest is deprecated in Confucian ethics, the traditional processes of family relations are animated by the self-interested actions of family members (Barbalet 2014). An inter-country comparison of bequest plans reveals strikingly that bequest plans of the Chinese appear to be much more consistent with selfish preferences than those of Americans and Indians, partly due to the unsound social safety nets in China (Horioka 2014). The decline of parental authority and the erosion of filial piety have made support for older persons less secure over time; parents use their bequests in exchange for old age support. Exchange is bidirectional, and whether adult children are willing to accept an exchange offered by parents remains unclear. Whether a relatively large bequest can guarantee generous old age support from children remains to be seen, let alone the generally meager bequests from rural parents. Caputo (2002) indicated that for adult children, providing support to aging parents may be driven by the needs of parents and filial responsibility, rather than expectations of future reciprocity on the part of their parents. Given the rapid socioeconomic development, massive migration of adult children from rural to urban areas in search of jobs, and disadvantaged economic conditions of older rural people, a well-built social security system for older people should be a high priority in the Chinese government’s agenda.