Abstract

While the literature shows that elderly parents may use bequests to reward children who provide them with time support, there is limited evidence on whether younger, less needy parents base their intended bequest division on alternative forms of support from children. This study uses data from the June 2018 Singapore Life Panel and focuses on a sample of 4,125 adult children and their middle-aged and older parents. From family fixed-effects estimation, I find that parents intend to leave larger bequest shares to coresident children and to children who provide greater material support. Parents also intend to bequeath more to children in whom they confide frequently, while children in whom they confide rarely receive more bequests only if they provide greater material support. The results suggest that parents may interpret physical and emotional proximity to children as signs of filiality for which they may reward them, while detached children may earn such rewards through material support. This study demonstrates the existence of coresidence-for-bequest and money-for-bequest exchanges between adult children and their middle-aged and older parents. These exchanges may translate into future caregiving-for-bequests when parents become elderly, and may thus have broader implications for both individual and societal well-being.

Introduction

Many parents condition their bequests on children's actions (Bernheim et al. 1985; Brown 2006; Groneck 2017; Horioka et al. 2018). Light and McGarry (2004) reported that 25% of American mothers who plan to divide their estate unequally intend to give more to the child who takes care of them. Meanwhile, Francesconi et al. (2015) documented that the proportion of American parents with wills that exclude some children has more than doubled between 1995 and 2010, from 16% to 35%. Such unequal division is more prevalent among parents who have limited contact with some of their children.

While the literature has predominantly addressed how bequest division depends on time support from children to elderly parents in the United States, there is a lacuna on how intended bequest division may depend on other forms of children's support to younger parents in Confucian societies. Such a gap may be because of a combination of cultural factors and data limitations. Monetary transfers from adult children to parents are relatively uncommon in the United States (McGarry and Schoeni 1997). Moreover, the intended share going to each child (intensive margin) is usually not reported in surveys, such that the literature has relied predominantly on indicators of whether a child is a beneficiary (extensive margin). Indeed, there is limited research focusing on various types of transfers or on differences in transfers within families (Haider and McGarry 2018).

This article examines whether parents' early intended bequest share to a child varies with the child's coresidence and material support (i.e., monetary and in-kind transfers). I exploit a large-scale data set of middle-aged and older (but not yet elderly) Singaporeans, the Singapore Life Panel (SLP), which contains rich information on parents' intended bequest division and support from each child, as well as the socioeconomic characteristics of parents and children.

Singapore is an interesting country to study as it promotes the Confucian values of filial piety and also has one of the highest per capita incomes in the world (Tan 2012). This study, therefore, can shed light on how aging parents' exchanges with their adult children may operate in the many rapidly developing Asian societies that share similar values. A child's coresidence with parents may help predict future caregiving (Cong and Silverstein 2015; Fu 2019; Konrad et al. 2002). Similarly, material support from a child may be symbolic in providing familial networking, even if one does not need the extra money per se (Hwang 1987; Rosenzweig and Zhang 2014; Xie and Zhu 2006). Thus, middle-aged or older parents who are not yet in need of caregiving may interpret coresidence and material support as indicators of filiality for which they may intend to reward children.

Data from the SLP reveal that 87% of parents with two or more children intend to leave a bequest to at least one child (i.e., to some or all of their children). Among such families, 25% of parents intend to leave unequal bequests, while 30% of children may expect unequal bequests. Descriptive statistics from a child-level sample—with one observation per child, but several children per family—reveal that parents may intend to leave larger bequests to children who coreside with them and to those who provide them with greater material support. Nevertheless, such patterns may be driven by unobserved factors, such as family generosity, which may influence both bequests and support. Therefore, I employ family fixed-effects (FE) models to estimate the associations and also perform a battery of sensitivity analyses. The findings across all models corroborate and suggest that parents leave larger bequest shares to children who provide them with greater support.

I also attempt to shed light on some of the potential channels driving such results. Parent–child pair emotional closeness may simultaneously encourage parental bequest and support from the child. I thus use the frequency that parents confide in each child as a proxy for parent–child closeness (Chipps and Jarvis 2016; Connidis and Campbell 1995; Rittenour et al. 2007) and find that parents intend to leave larger bequests to children in whom they confide often. Parents still respond positively to coresidence support from children, while the intended bequest share is more responsive to material support provided by children in whom parents rarely confide. These results suggest that parents may reward children on the basis of physical and emotional proximity, while detached children are rewarded for their material support.

I find no evidence that parents leave greater bequests to children who are of lower socioeconomic or health status. In fact, parents bequeath more to children who own a house. There is also no evidence that parents' liquidity constraints or contemporaneous time support from children play a role in driving the associations between bequests and coresidence or material support. Finally, I see no evidence that parental bequests increase with a child's material support in one-child families, suggesting that such families may operate on different premises or that individuals other than children may not be credible as beneficiaries.

Taken individually, each piece of evidence is not sufficient to establish or rule out the existence of specific bequest motives. Together, however, the weight of the evidence supports the notion that some middle-aged and older parents with two or more children base (at least part of) their intended bequest division on physical and emotional closeness to children. The results suggest that parents may interpret such closeness as signs of filiality for which they intend to reward children, while they may rely on material support from detached children to make such rewards.

Contextual Framework

Existing economic studies on bequest division have tended to focus on time-for-bequest exchanges, in which children provide care or attention to elderly parents. In addition, most previous research on the subject has focused on U.S. families. This study examines how younger parents' bequest intentions relate to coresidence and material support from children by exploiting a unique data set on middle-aged and older Singaporeans. The study builds on insights on filial piety in Confucian societies to set the backdrop behind the interpretation of coresidence and material support as potential signs of filiality, for which parents may reward children.

Background

Economic Studies on Bequests

Parents may leave bequests because of egoistic, altruistic, or strategic exchange motives. Under egoism, bequests are luxury goods that parents care directly about (Kopczuk and Lupton 2007; Liu et al. 2019), so that bequests may increase with parental wealth but should not be affected by children's socioeconomic status or actions. Under altruism, parents care about children's well-being (Becker 1974; Laitner and Juster 1996; Laitner and Ohlsson 2001), so that larger bequests are expected to be allocated to the least well-off children when parents care about all their children equally.

Closely related to this study is the literature on how bequest division relates to children's actions. Bernheim et al. (1985) suggested that parents with at least two children may operate under a strategic motive by which bequests are used to reward children according to their actions. The share of bequest is, therefore, expected to increase with the services provided by children. Indeed, evidence from probate records and surveys indicate that parents wish to leave more to children who care for them (Francesconi et al. 2015; Horioka 1984; Horioka et al. 2018; Light and McGarry 2004; Norton and Taylor 2005).

A few studies have examined the relationship between actual bequest division of unmarried parents and late-life or end-of-life care in the United States. Brown (2006) analyzed such relationships using a sample from the Asset and Health Dynamics Among the Oldest Old (AHEAD) survey of parents aged 70 or older with two or more living children and a will or life insurance policy, and found that children who help with caregiving are more likely to be a beneficiary. Similarly, from the Health and Retirement Study (HRS) exit interviews, Groneck (2017) found that children who provided end-of-life care to parents are more likely to receive bequests and also receive higher amounts.

Past research has focused on how bequest division depends on time support from children to unmarried elderly parents in the United States. However, there is very limited evidence on how intended division may depend on other forms of support from children to younger married and unmarried parents in Asian societies. While inter vivos transfers in the United States tend to flow from parents to adult children, the reverse is true in Confucian societies, where children are expected to be filial and be the panacea of support to older parents (CHARLS Research Team 2013; Ho 2015, 2019; Mason and Lee 2018; McGarry and Schoeni 1997; Oliveira 2016).

Filial Piety in Confucian Societies

Confucian ideals bind children to obey and serve their parents, with filial piety or xiào considered as the highest of all virtues. Within Confucian teachings, filiality has been the dominating concept that regulates the relationship between children and parents (Ng et al. 2002). Despite industrialization leading to a lower prevalence of care from children in modern Asia, respect from children is still highly valued by parents and may even contribute toward their psychological well-being (Cheng and Chan 2006). Interestingly, Silverstein et al. (1996) found that too little or too much support from children may both be detrimental to American parents' well-being: too little support may be inadequate in meeting parental needs, while too much may make a parent feel patronized. Such results were not seen in Hong Kong, where parental well-being increases monotonically with support from children. Cheng and Chan (2006) have argued that filial responsibility in the West may mean that adult children help with their parents' actual needs, whereas filial responsibility in the East may imply that they demonstrate their devotion to their parents.

In traditional Chinese societies, filial piety involves reciprocal behaviors in the parent–child relationship, which may involve the exchange of inheritance from parents with the expectation of absolute obedience, respect, and care from the adult children (Das Gupta et al. 2003). In modern Asia, such reciprocal transactions may entail “favor” and “gifts,” rather than “goods” and “services” (Chen et al. 2007). A child's coresidence with parents may help predict prospective caregiving (Cong and Silverstein 2015; Fu 2019; Konrad et al. 2002). Similarly, material support from a child may be symbolic in providing familial networking, even if one does not need the extra money (Hwang 1987; Rosenzweig and Zhang 2014; Xie and Zhu 2006). Given the relative importance of showing respect to parents in Confucian societies, parents may interpret coresidence and material support as indicators of filiality for which they may intend to reward children through bequests.

This study contributes to the literature in several ways. First, the article focuses on the case of an Asian society in which the notion of filial piety is strong. Second, the unique data set used enables the study of parental intended bequest division to each child while parents are younger and healthier, rather than the elderly individuals studied previously. Thus, the study sheds new insights on whether parents think strategically decades ahead of death and prior to the onset of caregiving needs. Third, to the best of my knowledge, this study is the first to show a positive relationship between parental bequests and material transfers from children. It adds to the traditional time-for-bequest exchanges literature by showing that money-for-bequest exchanges may constitute relevant signals for younger parents, especially with regard to detached children. Finally, the article examines the importance of emotional closeness on bequest division intentions and on its interaction with coresidence and material support.

This study focuses on self-reported intended bequest division to each child as opposed to late-life or end-of-life bequest division captured in wills or life insurance policies. Studying parental intentions may shed light on early bequest motives unclouded by other considerations, such as the loss of cognitive awareness as one gets older or guilt over unequal bequests just prior to death (Bernheim and Severinov 2003; McGarry 2001; Severinov 2006; Sloan et al. 1997). Early parental intentions on family exchanges may have important welfare implications for aging parents, especially concerning whether their expectations are eventually met. This study provides a first step toward understanding such intentions.

Hypotheses

This study posits that coresidence and material support may demonstrate filial piety for which parents may reward children through their bequest division.

Coresidence

Parents may view a child's coresidence as a sign of future filiality, as coresidence with parents may help predict future caregiving and, thus, warrant some rewards from parents.

  • Hypothesis 1: Parents intend to leave larger bequest shares to coresident than to noncoresident children.

Furthermore, if coresidence reflects (prospective) support from children:

  • Hypothesis 1a: Parents are more responsive to coresident children who are homeowners.

  • Hypothesis 1b: Older parents are more responsive to coresidence than younger parents.

Material Support

Material support may be symbolic in displaying family ties, even if one does not necessarily need the extra cash. Mauss (1950/2002) has argued that solidarity is achieved through the social bonds created by gift exchange. Therefore, there may be money-for-bequest exchanges within the family.

  • Hypothesis 2: Parents intend to leave larger bequest shares to children who give greater material support.

Although money-for-bequest exchanges have not been explored in the published literature, money-for-money exchanges—such as loan repayments—are well-known. Such a loan repayment motive may be tested by comparing the responsiveness of liquidity-constrained parents (“illiquid”) to that of non-liquidity-constrained parents (“liquid”). Illiquid parents are financially needy and, thus, more likely to borrow from and repay children (possibly using bequests), whereas liquid parents do not need to borrow from or repay children. If a loan repayment motive is at play, then the intended bequest share from illiquid parents should be more responsive to a child's material support than that from those who are liquid. Conversely, if material support is symbolic in driving bequest rewards, then liquid and illiquid parents should behave similarly.

  • Hypothesis 2a: Liquidity-constrained and non-liquidity-constrained parents' intended bequest shares will be equally responsive to children's material support.

Emotional Support Confounders?

Coresidence and material support could act as a proxy for current emotional support from children. Indeed, parents may also reward children for emotional support and base their rewards on alternate signs, such as material support, to assess the filiality of detached children. One tends to confide in those emotionally closest to oneself, possibly because of the perception of a more cohesive relationship (Chipps and Jarvis 2016; Connidis and Campbell 1995; Rittenour et al. 2007). Hence, I exploit the frequency at which parents confide in each child as a proxy for parent–child pair emotional closeness in the empirical analyses.

  • Hypothesis 3a: Parents leave larger bequest shares to children in whom they confide frequently.

  • Hypothesis 3b: Parents leave larger bequest shares to detached children who give greater material support.

Time Support Confounders?

Coresidence and material support could also act as a proxy for current time support from children. To address this, I exploit information on whether parents pay for domestic services and on the frequency of contact with children. For example, if parents who do not employ domestic services benefit from greater housekeeping help from children, then the responsiveness of the intended bequest share from such parents should be greater than that of other parents. Conversely, if coresidence and material support do not act as a proxy for time support, parents who employ domestic services should behave the same as those who do not.

  • Hypothesis 4a: Parents who pay for domestic services respond similarly as those who do not to children's coresidence and material support.

Furthermore, if coresidence and material support do not act as a proxy for time support from children, then controlling for the frequency of contact with each child (which may capture the frequency of time help) in the empirical analyses should not change the results.

  • Hypothesis 4b: The responsiveness of the parental bequest share to coresidence and material support is the same even after controlling for the frequency of contact with each child.

Credible Beneficiaries?

Finally, the strategic bequest motive is conditional on the presence of credible beneficiaries. Thus, the literature on bequest division tends to focus on families with at least two children and a parent who intends to leave a bequest.

  • Hypothesis 5: The positive responsiveness between parental bequest shares and child support is present only in families with more than one child.

To summarize, the main hypotheses test whether the parental intended bequest share to a child increases with the child's coresidence and material support, while complementary hypotheses help address potential confounders such as emotional and time support.

Data

The Singapore Life Panel (SLP) is well-suited to address how parental intended bequest division may depend on coresidence and material support from children of middle-aged or older parents. The SLP is a nationwide monthly internet survey of Singaporeans (and their spouses) aged 50–70 in 2015, when the survey was launched.1 The respondents are, therefore, younger and healthier than respondents of other aging surveys, such as the AHEAD or the HRS exit interviews. Interested readers may refer to Vaithianathan et al. (2021) for further details on the SLP.

A special child module was included in the June 2018 wave of the SLP, when respondents were surveyed on coresidence, monetary and in-kind support to and from each child, intended bequest share to each child and to nonchild beneficiaries, and the demographic and socioeconomic characteristics of each child. The survey also contains detailed information on the respondent parents' demographic and socioeconomic characteristics. There are approximately 8,000 responses (∼80% response rate) per wave.

Sample

For my analyses, I focus on the June 2018 wave of the SLP, so “respondents” refers to survey participants in June 2018, “parent sample” refers to parent respondents, and “child sample” refers to children. The child sample is the reshaped version of the parent sample such that it contains one observation per child, while there may be several children from the same parent. Aside from selected descriptive statistics on parents, all analyses are performed using the child sample. The sample selection and construction are illustrated in Figure A1 in the online appendix. I limit the parent sample to Singaporeans (95% of respondents) aged 50 or older (98% of respondents), which drops spouses who do not meet the SLP eligibility criteria. I also focus on the first respondent in each household, who make up 69% of respondents in the sample. This selection is made because bequest intentions and child support are reported jointly for married couples, and their children would otherwise be double counted when the sample is reshaped at the child level.2

Because this study addresses bequest division and employs family FE models, I limit the parent sample to respondents who reported having at least two children aged 25–49 (39% of respondents) and intending to leave a positive bequest to at least one child (35% of respondents). This aligns with the literature on bequest division and captures the fact that some children may not receive a bequest (Bernheim et al. 1985; Brown 2006; Groneck 2017). The age restriction further ensures that most children would have completed their education and be in a position to provide support to parents. Forty-seven percent of respondents had at least two children; of those, 83% had children all aged 25–49, while 87% intended to leave a bequest to some or all of their children. After dropping missing values, 1,682 parents are included in the parent sample. Then, I reshape the parent sample to construct the child sample used to perform child-level analyses. This yields a sample of 4,125 children aged 25–49. Analyses on families with only one child are addressed separately later in this study.

Descriptive Statistics

This section presents an overview of parental intended bequest division and support from children. Additional descriptive statistics are shown from the parents' and children's perspectives.

Parental Intended Bequest Division

Parental intended bequests are inferred from the following SLP question:

Parents often leave bequests to children, other family members, or religious/charitable organizations. Sometimes, the bequest given to each child will differ for reasons such as: Each child's unique financial needs or socioeconomic status, varying levels of financial or caregiving support given by each child, religious teachings, or the different nature of each parent–child relationship. What percentage of your total bequest would go to each of the following persons/groups?

The intended beneficiaries include each child, other family members (e.g., nieces and nephews), religious or charitable organizations, and others. The question is asked from the point of view of the respondent if he or she is single, and from the joint point of view of the respondent and his or her spouse if married. The vast majority of children (99%) may expect to receive a bequest, while only 1% may expect not to.3 Because nearly all children may expect to receive a bequest, this study focuses on the intensive margin—that is, on the share of bequest that a parent intends to leave to a child, normalized by the sum of bequest shares that the parent intends to leave to all children. For example, if a parent with two children intends to leave 30% to one child, 60% to the other child, and 10% to other beneficiaries, then the bequest shares of the first and second children are one third and two thirds, respectively. Thus, the share of bequest to each child ranges from 0 to 1. From the parent sample, the intended bequest share to each child is .41, corresponding to 41% (Table 1).

To shed light on the prevalence of unequal bequests, I compute the bequest share deviation from an equal division. This is computed as the difference between the bequest share to a child and the bequest share if the parent equally divided bequests among children. From the previous example, the bequest share deviations are minus one sixth and one sixth, respectively. Nearly a third of children may expect their bequest to deviate from an equal division. Some 12.7% of children may expect higher than equal bequest shares by an average of 12.1 percentage points, while 17% of children may expect lower than equal bequest shares by 7.9 percentage points.

Figure 1 further shows that 7.9%, 2.6%, and 2.2% (12.8%, 2.1%, and 2.1%) of children may expect higher (lower) than equal bequest shares by 10 or less, 10–20, and 20 or more percentage points, respectively. Thus, while the majority of children may expect equal division shares, approximately 20.7% may expect deviations from an equal share by 10 percentage points or less, while 9% may expect deviations from an equal share by more than 10 percentage points.

I also compute the difference between the maximum and the minimum bequest share within a family from the parent sample. Among the 25.1% of parents who intended to leave unequal bequests, the average difference within the same family was 22.1 percentage points. Figure 2 shows that 12.9%, 7.1%, and 5.1% of families had a difference of 0–10, 10–40, and 40 or more percentage points, respectively. Thus, while the majority of parents intended to leave equal bequests, approximately one quarter intended to divide bequests unequally.

Support From Children

This study focuses on a child's coresidence and material support (i.e., monetary and in-kind transfers made over the previous year). Material support is measured for both coresident and noncoresident children and is based on the SLP question:

Not counting shared housing or food, in the past twelve months, what is the value of financial help or gifts that you or your spouse received from your living child and his/her spouse? Financial help or gifts may include payment for education, housing, marriage gifts, bill payments, gifts of food and clothing etc. If [Central Provident Fund] money was used, please include that.4

As Table 1 shows, 47% of children coresided with parents, while 69% of children provided material support. Among the latter, each child provided an average of S$2,920 (about US$2,141). The share of transfers from each child averages 32% of transfers from all children within the same family.

Table 2 reports the mean of bequest shares by coresidence status and whether children gave below or above the median transfer amount. Coresident children may expect greater bequests by 3.4 percentage points than noncoresident children (p < .01). Similarly, children who provided above the median amount of transfers may expect greater bequests by 1 percentage point than those who provided below the median amount of transfers (p < .05).5

Demographic and Socioeconomic Characteristics

Summary statistics for parents are reported from the parent sample in Table 3. On average, parents were 64 years old, and the majority (62%) reported being in good, very good, or excellent health (33% and 5% reported being in fair or poor health, respectively). Only 8% reported having difficulty with any activity of daily living (ADL), such as dressing, walking, or bathing/showering. The majority (79%) of parents reported being married, and, on average, they had 2.5 children. Eighty-nine percent reported being of Chinese ethnicity, with the rest being Malay or Indian. Parents were relatively wealthy, with an average net wealth of around S$1.2 million (about US$880,000), mostly from housing wealth, CPF savings, and financial assets.6

Table 3 also reports summary statistics for children from the child sample. On average, children were aged 35; 51% were male and 53% were married. Children tended to be highly educated, with 81% having postsecondary education. Half of children owned a house.

Empirical Strategy

The main objective is to estimate the associations between parental intended bequest shares and support from children. I present ordinary least-squares (OLS) and family FE models that help control for unobserved factors that may be common among siblings. I also present robustness specifications including an instrumental variable (IV) strategy. All analyses henceforth are conducted using the child sample.

Empirical Models

The following models are estimated:

Ln(BequestShareij*)=β0+β1Coresidentij+β2Transfersij+ψCXijC+ψPXjP+εij.
(1)

The dependent variable is the logarithmic transformation of the bequest share that parent j intends to give to child i: (Bequest Shareij*)=Ln(Bequest Shareij +1).7

Coresidentij is an indicator that takes a value of 1 if parent j lives with child i and a value of 0 otherwise. Transfersij{Any Transferij, Ln(Transfersij*),Ln(Transfer Shareij*)}:

  • (i) Any Transferij  is an indicator that takes a value of 1 if parent j received any material support from child i and a value of 0 otherwise.

  • (ii) Ln(Transfersij*)=Ln(Transfersij +1) is the logarithmic transformation of the amount of material support that parent j received from child i.8

  • (iii) Ln(Transfer Shareij*)=Ln(Transfer Shareij +1) is the logarithmic transformation of the share of material support that parent j received from child i, normalized by the sum of material support that the parent received from all children.

The child's controls, XijC, refer to demographic and socioeconomic characteristics: second-order polynomials for age; being male (yes/no); being married (yes/no); birth order dummy variables; education dummy variables for secondary school, two- or three-year diploma, and four-year undergraduate or postgraduate university degree; and home ownership (yes/no). The parent's controls, XjP, refer to demographic and socioeconomic characteristics and health variables: second-order polynomials for age; being male; being married; number of children; Chinese ethnicity (yes/no); education dummy variables for secondary school, two- or three-year diploma, and undergraduate or postgraduate degree; inverse hyperbolic sine transformation of net wealth; good health (yes/no); at least one ADL limitation (yes/no); and district of residence dummy variables.9

In FE models, εij=μj+ηij, where μj captures the family fixed effect and ηij  is an idiosyncratic error term. The family FE accounts for family characteristics that are common across the child and his/her siblings. These include XjP (inclusive of family size) and unobserved family characteristics, such as family generosity, which may simultaneously affect both bequests and support. The effects of child support on bequests are identified using within-family differences across siblings, and the within-group R2 are reported in the tables below.10

Note that the logarithmic transformation of the dependent variable avoids the issue of collinearity in the outcomes of children within the same family in FE estimation. Section A.1 in the online appendix shows that this issue arises because bequest shares among children in a family sum to 1; the section presents a consistent FE model.

Robustness Specifications

While FE models help control for unobserved family characteristics that are common across siblings, they do not control for unobserved parent–child pair characteristics that are heterogeneous across siblings. For example, a parent may feel closer to a particular child, leading to their leaving a greater bequest to the closer child, who may also give greater transfers to the parent. In this case, there would be a positive correlation between Transfersij and ηij, which would bias the FE estimates upward. In further analyses, I exploit the frequency at which a parent confides in a child to control for parent–child pair confidence, which may help act as a proxy for parent–child pair closeness and mitigate such issues. I also employ an IV strategy following Lewbel (2012), which exploits information contained in the heteroskedasticity of the error terms to construct valid instruments, and which also addresses potential issues of reverse causality (see section A.2 in the online appendix for details). Finally, I perform heterogeneity analyses using the presence of a will, given that reverse causality issues may be more pertinent when the parent has a will as children may be better aware of parental bequest intentions.

Results

This section presents estimates of the associations between parental intended bequest shares and support from children. Parents intend to bequeath more to children who are physically and emotionally close to them, while detached children get greater bequest shares when they provide parents with greater material support.

Parental Intended Bequests and Support from Children

Table 4 reports results from OLS and FE models. Columns 1 and 4 consider transfers at the extensive margin Any Transfer, columns 2 and 5 consider transfers at the intensive margin Ln(Transfers *), and columns 3 and 6 consider the share of transfers relative to all children in the family Ln(Transfer Share *). The estimated coefficients for other covariates are reported in Table A1 in the online appendix.

Coresidence tends to be positively associated with bequest shares such that Hypothesis 1 is not rejected. A coresident child may expect the share of bequest to be higher by 1.2–1.8% (p < .01). Similarly, material transfers in the previous year tend to be positively associated with bequest shares such that Hypothesis 2 is not rejected. Providing any material transfer to parents increases the intended bequest share by 0.6–2.5% (p < .05). A 10% increase in the transfer amount in the previous year leads to a 0.01–0.03% increase in bequest share (p < .05), while a 10% increase in transfer share relative to all children leads to a 0.25–0.31% increase in bequest share (p < .01). A back-of-envelope calculation based on bequeathable wealth suggests that a coresident child may expect the bequest amount to increase by up to S$127,074, while providing material support may increase the bequest amount by up to S$176,492. Thus, while the estimated average percentages may seem small, there is nevertheless indication that some parents have an inclination to reward support from children.11

From Table A1 in the online appendix, there is also no evidence that parents leave greater bequests to children of lower socioeconomic status. Child education does not seem to matter, while child house ownership tends to be positively associated with bequest (p < .01). There is also evidence that parents intend to leave larger bequests to sons (p < .05), while children from larger families may expect lower bequests (p < .01).12 Sensitivity analyses controlling for child health also reveal no statistically significant associations between bequest division and child health.

Parent–Child Pair Confidence

The SLP incorporates the following question: “Do you confide in your living child about your worries for the future? For example, about financial security and health issues.”13

Define Confideij as an indicator that takes a value of 1 if the parent confides in the child “often” or “sometimes” and a value of 0 if the parent does so “rarely” or “never.” While this variable may also capture whether parents are inclined to worry, across-sibling variation exploited by the FE specification would capture such an issue through ηj.

Approximately 50% of children benefit from their parent's confidence. Parents are slightly more inclined to leave bequests to children in whom they confide frequently, while the latter are also more likely to coreside with and provide material support to parents. I estimate models following the empirical strategies described earlier, but include interaction terms (Coresidentij×Confideij) and (Transfersij×Confideij) and control for Confideij as additional covariates.

Table 5 reports the marginal effects for parent–child pairs in which the parent confides either frequently or rarely. The magnitudes are always greater for children in whom parents confide rarely than for children in whom parents confide frequently. Nevertheless, the differences in responsiveness of parental bequests with respect to coresidence of the two groups are not statistically significant. Conversely, the differences in responsiveness of parental bequests with respect to material transfers from children in whom parents confide rarely and children in whom parents confide frequently are statistically significant at the 5–10% levels in columns 1, 3, 4, and 6. Thus, parents are more responsive toward a child in whom they confide rarely with respect to the extensive margin of transfers and the relative share of transfers provided by the child. Parents also tend to leave larger bequest shares to children in whom they confide frequently than to children in whom they confide rarely. Hence Hypotheses 3a and 3b are not rejected.

To delve into the differences in parental responsiveness, I perform heterogeneity analyses by child gender and coresidence status. It is well-known that Asian parents favor sons, who may receive larger bequests and also provide greater old-age support (Das Gupta et al. 2003; Ho 2019; Jayachandran 2015). Nevertheless, a proportion test reveals no statistically significant differences between the proportions of sons and daughters in whom parents confide. There are also no statistically significant differences in the responsiveness of parental bequests to support from children of different genders (see Table A2 in the online appendix). The gender differences were also statistically insignificant in the sample of families with mixed-gender children. Because the gender of the recipient may also matter (Szinovacz and Davey 2008), I perform further sensitivity analyses on separate samples for male and female respondents, which reveal no statistically significant differences between the responsiveness of bequests to support from sons and daughters.

It is also possible that parents feel closer to coresident children. Nevertheless, the proportion of children in whom parents confide is only five percentage points greater for coresident children than for noncoresident children (p < .05). Interaction analyses performed by including (Coresidentij×Transfersij) as an additional regressor reveal no statistically significant differences in the responsiveness of parental bequests to material transfers from children with different coresidence statuses (see Table A3 in the online appendix). Thus, the results that parents seem to respond more toward material transfers from detached children do not seem to be driven by children's gender or coresidence.

One-Child Families

One can measure the share of bequest going to an only child as a proportion of the total bequests intended for child and nonchild beneficiaries. The average share of intended bequest to an only child is 85%, while only children are more likely to coreside and also provide greater material support.

Table 6 reports the associations between bequest and support in one-child families. The estimated coefficients of Coresident are positive but statistically insignificant. The estimated coefficients of Transfers are always negative, although they are statistically insignificant at the 5% level across all but one specification. Thus, there is no evidence that bequests increase in a child's material support in one-child families. Conversely, from Table A4 in the online appendix, sensitivity analyses using Model 1 and that split the sample into families with two, three, and four or more children yielded qualitatively similar results to those in Table 4. Thus, Hypothesis 5 cannot be rejected.

Further Analyses

Section A.2 in the online appendix reports further results to shed light on some of the potential mechanisms that may be driving the positive associations between bequest shares and support from children. To summarize key results: (1) results from IV models are similar to those in Table 4; (2) parents bequeath more to coresident children who own a house, which reinforces the notion that parents may use bequests to compensate children for coresidence support, so Hypothesis 1a is not rejected; (3) older parents are more responsive to coresidence support (that may capture more imminent instrumental help), while younger parents are more responsive to material support such that Hypothesis 1b is not rejected; (4) there are no statistically significant differences in the responsiveness between liquidity-constrained and nonliquidity-constrained parents, so that Hypothesis 2a may not be rejected; (5) there is no evidence that parents who do and do not engage domestic services respond differently and the results are robust to controlling for the frequency of contact such that Hypothesis 4a and 4b are not rejected; and (6) the results are robust to controlling for inter vivos transfers of money and time (grandchild care) from parents and to the presence of a will.

Discussion

The main objective of this study was to estimate the associations between parental intended bequest division and support from children in terms of coresidence and material transfers. Overall, the results indicate that bequest shares are positively associated with such support. Furthermore, children in whom parents confide frequently may expect greater bequests, while children in whom parents confide rarely may expect greater bequests when they give greater material support.

While this study did not aim to disentangle between the different bequest motives, a discussion on how the results may link to existing theories is warranted. Standard models of bequest division have predominantly focused on time-for-bequest exchanges. Under an altruistic motive, parents care about the well-being of children (Becker 1974; Laitner and Juster 1996; Laitner and Ohlsson 2001), so that greater bequests are expected to be allocated to the least well-off children, if parents value children equally. Conversely, under an exchange motive, parents may use bequests to reward children according to their services (Bernheim et al. 1985; Cox and Rank 1992). The share of bequest is, thus, expected to increase with the services provided by the children.

As reported, there is no evidence that parents leave larger bequest shares to children of lower socioeconomic or health status. Conversely, child house ownership is positively associated with bequest share, while older parents are more responsive toward coresidence support. There is also no evidence that parents in one-child families positively respond to material support, suggesting that one-child families may operate on different motives or that beneficiaries other than children are not credible outside options. These results seem consistent with an exchange motive, where parents with at least two children intend to reward children for their actions.

Although these findings indicate the presence of exchanges, it does not mean that other motives are not important. For instance, the positive associations between bequest shares and emotional support from children may be a reflection of parent–child pair cohesion. Moreover, Figures 1 and 2 show that many parents intend to leave equal bequests, which could reflect a wish to signal equal love to children (Bernheim and Severinov 2003). The difficulty of disentangling motives for individual decisions is a well-known challenge in this literature (Groneck 2017). The current study remains agnostic on which motive dominates and focuses on testing whether parental bequests respond to higher support from children, which indicates that at least part of parental bequests may be driven by exchanges.

Parents in the current study are still relatively young and healthy, so that care needs may not be relevant until years or decades later. Interestingly, the results suggest that parents may be inclined to reward children who they are physically and emotionally close to, while they reward detached children on the basis of their material support. The findings suggest that some parents could be thinking strategically from early on and rely on support other than caregiving to base (at least part of) their intended bequest division. A child's coresidence may help predict future caregiving (Cong and Silverstein 2015; Fu 2019; Konrad et al. 2002), while material support may be symbolic in providing familial networking (Hwang 1987; Rosenzweig and Zhang 2014; Xie and Zhu 2006). Thus, middle-aged or older parents who are not yet in need of caregiving support may interpret proximity as signals of filiality for which they may intend to reward children, and in the absence of such proximity, parents may rely on material support for such signals.

Conclusion

This article analyzes how parental intended bequest share varies with a child's coresidence and material support using a rich, large-scale data set of middle-aged and older Singaporeans. The findings suggest that parents intend to leave larger bequest shares to coresident children and to children who provide greater material support. Parents also intend to leave larger bequest shares to children in whom they confide often, while they are more inclined to reward children in whom they rarely confide when the latter give them greater material support.

It is well-known that elder U.S. parents may condition their bequest division on late-life or end-of-life care by children. Missing from these discussions is how bequest division may also depend on other forms of support, even when parents are not necessarily in need of such support. This study provides a first step toward understanding the responsiveness of early parental bequest intentions to child support in the form of coresidence and material transfers in Confucian societies.14 Taken together, the weight of the evidence supports the notion that parents may potentially think strategically decades ahead of death and prior to the onset of caregiving needs. In such cases, parents may rely on coresidence and material support as signals of filiality for which they may intend to reward children.

The presence of exchanges in bequest and support from children has important implications for dynamic life cycle decisions. Recent studies have attempted to disentangle saving for bequests and saving for precautionary motives (Boserup et al. 2016; De Nardi and Yang 2014; Kopczuk and Lupton 2007). This study emphasizes the importance of considering family exchanges in terms of both coresidence-for-bequests and money-for-bequests for the analysis of asset accumulation of middle-aged and older individuals. Such family exchanges may eventually translate into caregiving-for-bequests when individuals are elderly and, thus, also have important implications for the eventual asset deaccumulation of the elderly.

As populations age worldwide, understanding the subtleties of family exchanges beyond bequests for late-life or end-of-life care to needy parents is increasingly relevant to help plan for the future. Parental early intentions on family exchanges may have important welfare implications for aging parents, especially with regard to whether their expectations are eventually met. Because unexpected health shocks may affect parents' cognitive abilities, or because parents may be prone to guilt at unequal bequests just prior to death, it may be interesting to measure the extent to which strategic considerations in early parental intentions get dampened by health shocks and guilt as parents age.

Dynamic considerations may also be relevant if support from children has a cumulative effect on parental intended bequests. Such considerations may be addressed in future research with a longer panel or with information on cumulative support. Because this study's sample is relatively young and healthy, caregiving support from children is likely less relevant. It would be interesting to follow up on the respondents as they get older to analyze whether earlier material support from children translates into later care and corroborates the earlier positive responsiveness of parental intended bequests toward material support.

Acknowledgments

I thank Le-Yu Chen, Luca Facchinello, Jisoo Hwang, Haoming Lui, Kathleen McGarry, Costas Meghir, Erik Meijer, and Jean Yeung, as well as attendees at the RES, AASLE, ESPE, HLFPAS, and WEAI conferences and SMU, Monash, CESR USC, and CFPR NUS seminars for their insightful comments and suggestions. This research uses data from the Singapore Life Panel conducted by the Centre for Research on the Economics of Ageing and Centre for Research on Successful Ageing at Singapore Management University.

Notes

1

The SLP was initially managed by the Centre for Research on the Economics of Ageing and subsequently by the Centre for Research on Successful Ageing at Singapore Management University.

2

Sensitivity analyses that (1) take the average of the responses between parents (for bequest intentions, child support, and parental variables) and (2) keep all respondents in the sample resulted in quantitatively similar results.

3

Note that the sample captures families who leave positive bequests to at least some children, suggesting that most Singaporean parents intend to leave bequests to all children if they decide to give positive bequests to at least one child. This practice differs from that in the United States, where 35% of parents exclude some children from their wills (Francesconi et al. 2015). Sensitivity analyses including families with one child and families where the parent intends to leave zero bequests to all children yielded similar results as those from the main sample.

4

The Central Provident Fund (CPF) is Singapore’s compulsory savings plan for pensions, health care (for self and immediate family members), education, and housing.

5

A proportion test reveals that there is no statistically significant difference between the proportions of coresident and noncoresident children who make positive material transfers to parents.

6

All variables are from the June 2018 wave except for net wealth, which is imputed from the annual asset module fielded in January–February 2018, and ADL limitation, which is imputed from the one-off module fielded in May 2018.

7

Sensitivity analyses using the logarithm of the bequest amounts (constructed by multiplying parental bequeathable wealth by the share of bequest) yielded qualitatively similar results.

8

Only 30% of children received material transfers from parents in the past year. Sensitivity analyses on the inverse hyperbolic sine (sinh1) transformation of the amount of net financial transfers (i.e., material transfers from the child to the parent minus material transfers from the parent to the child) yielded similar results. The logarithmic transformation allows for the presence of zeros, while the sinh1 transformation allows for nonpositive net transfers.

9

Second-order polynomials in age allow for potential nonlinearity in age effects. Sensitivity analyses excluding age squared yielded quantitatively very similar results. Individuals who have a primary school education or less are the reference category for education.

10

The FE model estimates a mean-deviated model such that the within-group R2 captures how well the explanatory variables account for changes in the outcome across siblings within a family. As the FE model removes the effects of characteristics that are fixed across siblings (such as parental characteristics and family size), the within-group R2 will tend to be lower than the R2 from OLS.

11

Bequeathable wealth is computed as net housing, business, and financial wealth, and noninvested CPF balances (which are bequeathable). Coresidence (providing any material transfers) may increase bequests by up to 1.8% (2.5%) of maximum bequeathable wealth of S$7,059,680, which implies an increase of up to S$127,074 (S$176,492) in the intended bequest amounts.

12

Two-tailed t tests reveal that sons have 0.1 fewer siblings than daughters (p < .05), which may indicate a slight preference for sons in Singapore. Note that sibling sex composition is differenced out in FE models, while the results are robust to controlling for the number of male and female children separately in OLS models. Note further that one may not necessarily give a causal interpretation to the coefficients of birth orders in OLS models as birth order configuration and family size may be jointly determined (Bagger et al. 2021).

13

The SLP does not differentiate between confidence about financial security and health issues.

14

Heterogeneity analyses indicate that the bequest intentions of the non-Chinese also respond positively to support from children. Future research could explore whether such exchanges prevail in other Asian societies.

References

Bagger, J., Birchenall, J. A., Mansour, H., & Urzua, S. (
2021
).
Education, birth order, and family size
.
Economic Journal
,
131
,
33
69
.
Becker, G. S. (
1974
).
A theory of social interactions
.
Journal of Political Economy
,
82
,
1063
1093
.
Bernheim, B. D., & Severinov, S. (
2003
).
Bequests as signals: An explanation for the equal division puzzle
.
Journal of Politcal Economy
,
111
,
733
764
.
Bernheim, B. D., Shleifer, A., & Summers, L. H. (
1985
).
The strategic bequest motive
.
Journal of Political Economy
,
93
,
1045
1076
.
Boserup, S. H., Kopczuk, W., & Kreiner, C. T. (
2016
).
The role of bequests in shaping wealth inequality
.
American Economic Review: Papers & Proceedings
,
106
,
656
661
.
Brown, M. (
2006
).
Informal care and the division of end-of-life transfers
.
Journal of Human Resources
,
41
,
191
219
.
CHARLS Research Team
. (
2013
).
Challenges of population aging in China: Evidence from the National Baseline Survey of the China Health and Retirement Longitudinal Study (CHARLS)
(Technical report).
Beijing, China
:
Peking University
.
Chen, S. X., Bond, M. H., & Tang, D. (
2007
).
Decomposing filial piety into filial attitudes and filial enactments
.
Asian Journal of Social Psychology
,
10
,
213
223
.
Cheng, S. T., & Chan, A. C. (
2006
).
Filial piety and psychological well-being in well older Chinese
.
Journals of Gerontology, Series B: Psychological Sciences and Social Sciences
,
61
,
262
269
.
Chipps, J., & Jarvis, M. A. (
2016
).
Social capital and mental well-being of older people residing in a residential care facility in Durban, South Africa
.
Aging & Mental Health
,
20
,
1264
1270
.
Cong, Z., & Silverstein, M. (
2015
).
End-of-life co-residence of older parents and their sons in rural China
.
Canadian Journal on Aging
,
34
,
331
341
.
Connidis, I. A., & Campbell, L. D. (
1995
).
Closeness, confiding and contact among siblings in middle and late adulthood
.
Journal of Family Issues
,
16
,
722
745
.
Cox, D., & Rank, M. R. (
1992
).
Inter-vivos transfers and intergenerational exchange
.
Review of Economics and Statistics
,
74
,
305
314
.
Das Gupta, M., Zhenghua, J., Bohua, L., Zhenming, X., Chung, W., & Hwa-Ok, B. (
2003
).
Why is son preference so persistent in East and South Asia? A cross-country study of China, India and the Republic of Korea
.
Journal of Development Studies
,
40
(
2
),
153
187
.
De Nardi, M., & Yang, F. (
2014
).
Bequests and heterogeneity in retirement wealth
.
European Economic Review
,
72
,
182
196
.
Francesconi, M., Pollak, R. A., & Tabasso, D. (
2015
).
Unequal bequests
(IZA Discussion Paper No. 8813).
Bonn, Germany
:
Institute for the Study of Labor
.
Fu, C. H. (
2019
).
Living arrangement and caregiving expectation: The effect of residential proximity on inter vivos transfer
.
Journal of Population Economics
,
32
,
247
275
.
Groneck, M. (
2017
).
Bequests and informal long-term care
.
Journal of Human Resources
,
52
,
531
572
.
Haider, S. J., & McGarry, K. (
2018
).
Parental investments in college and later cash transfers
.
Demography
,
55
,
1705
1725
.
Ho, C. (
2015
).
Welfare-to-work reform and intergenerational support: Grandmothers' response to the 1996 PRWORA
.
Journal of Marriage and Family
,
77
,
407
423
.
Ho, C. (
2019
).
Child's gender, parental monetary investments and care of elderly parents in China
.
Review of Economics of the Household
,
17
,
741
774
.
Horioka, C. Y. (
1984
).
The applicability of the life-cycle hypothesis of saving to Japan
.
Kyoto University Economic Review
,
54
(
2
),
31
56
.
Horioka, C. Y., Gahramanov, E., Hayat, A., & Tang, X. (
2018
).
Why do children take care of their elderly parents? Are the Japanese any different?
International Economic Review
,
59
,
113
136
.
Hwang, K. (
1987
).
Face and favor: The Chinese power game
.
American Journal of Sociology
,
92
,
944
974
.
Jayachandran, S. (
2015
).
The roots of gender inequality in developing countries
.
Annual Review of Economics
,
7
,
63
88
.
Konrad, K. A., Kumnemund, H., Lommerud, K. E., & Robledo, J. R. (
2002
).
Geography of the family
.
American Economic Review
,
92
,
981
998
.
Kopczuk, W., & Lupton, J. P. (
2007
).
To leave or not to leave: The distribution of bequest motives
.
Review of Economic Studies
,
74
,
207
235
.
Laitner, J., & Juster, F. T. (
1996
).
New evidence on altruism: A study of TIAA-CREF retirees
.
American Economic Review
,
86
,
893
908
.
Laitner, J., & Ohlsson, H. (
2001
).
Bequest motives: A comparison of Sweden and the United States
.
Journal of Public Economics
,
79
,
205
236
.
Lewbel, A. (
2012
).
Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models
.
Journal of Business & Economic Statistics
,
30
,
67
80
.
Light, A., & McGarry, K. (
2004
).
Why parents play favorites: Explanations for unequal bequests
.
American Economic Review
,
94
,
1669
1681
.
Liu, H., Song, C., & Zhu, S. (
2019
).
Intentional bequest motives and the choice of annuity
[Unpublished manuscript].
Department of Economics, National University of Singapore, Singapore
.
Mason, A., & Lee, R. (
2018
).
Intergenerational transfers and the older population
. In Hayward, M. D. & Majmundar, M. K. (Eds.),
Future directions for the demography of aging: Proceedings of a workshop
(pp.
187
216
).
Washington, DC
:
National Academies Press
.
Mauss, M. (
2002
).
The gift: The form and reason for exchange in archaic societies
(Halls, W. D., Trans.).
London, UK
:
Routledge
. (Original work published 1950)
McGarry, K. (
2001
).
The cost of equality: Unequal bequests and tax avoidance
.
Journal of Public Economics
,
79
,
179
204
.
McGarry, K., & Schoeni, R. F. (
1997
).
Transfer behavior within the family: Results from the Asset and Health Dynamics Study
.
Journals of Gerontology, Series B: Psychological Sciences and Social Sciences
,
52B
(Special issue),
82
92
.
Ng, A. C. Y., Phillips, D., & Lee, W. K. (
2002
).
Persistence and challenges to filial piety and informal support of older persons in a modern Chinese society: A case study in Tuen Mun, Hong Kong
.
Journal of Aging Studies
,
16
,
135
153
.
Norton, E. C., & Taylor, D. H. (
2005
).
Equal division of estates and the exchange motive
.
Journal of Aging & Social Policy
,
17
(
1
),
63
82
.
Oliveira, J. (
2016
).
The value of children: Inter-generational support, fertility, and human capital
.
Journal of Development Economics
,
120
,
1
16
.
Rittenour, C. E., Myers, S. A., & Brann, M. (
2007
).
Commitment and emotional closeness in the sibling relationship
.
Southern Communication Journal
,
72
,
169
183
.
Rosenzweig, M. R., & Zhang, J. (
2014
).
Co-residence, life-cycle savings and inter-generational support in urban China
(NBER Working Paper 20057).
Cambridge, MA
:
National Bureau of Economic Research
.
Severinov, S. (
2006
).
Bequests as signals: Implications for fiscal policy
.
Journal of Public Economics
,
90
,
1995
2008
.
Silverstein, M., Chen, X., & Heller, K. (
1996
).
Too much of a good thing? Intergenerational social support and the psychological well-being of older parents
.
Journal of Marriage and the Family
,
58
,
970
982
.
Sloan, F. A., Picone, G., & Hoerger, T. J. (
1997
).
The supply of children's time to disabled elderly parents
.
Economic Inquiry
,
35
,
295
308
.
Szinovacz, M. E., & Davey, A. (
2008
).
The division of parent care between spouses
.
Ageing & Society
,
28
,
571
597
.
Tan, C. (
2012
). “
Our shared values” in Singapore: A Confucian perspective
.
Educational Theory
,
62
,
449
463
.
Vaithianathan, R., Hool, B., Hurd, M. D., & Rohwedder, S. (
2021
).
High-frequency internet survey of a probability sample of older Singaporeans: The Singapore Life Panel
.
Singapore Economic Review
,
66
,
1759
1778
.
Xie, Y., & Zhu, H. (
2006
).
Do sons or daughters give more money to parents in urban China?
Journal of Marriage and Family
,
71
,
174
186
.
This is an open access article distributed under the terms of a Creative Commons license (CC BY-NC-ND 4.0).

Supplementary data