My Brother’s Keeper? The Association Between Having Siblings in Poor Health and Wealth Accumulation

Abstract

When confronted with the economic costs of addressing a serious health problem, many American households do not possess the ability to deal with the crises on their own and may turn to family members for help. Using longitudinal data from the Panel Study of Income Dynamics, we examine if the level of wealth held by individuals is related to the health problems of their siblings. We find evidence that having a sibling who has experienced a health problem decreases the amount of wealth that some families have. The research has implications for the existing literatures on altruism and kin networks, as it sheds some light on the nature of altruism that prevails in U.S. families and on how kinship networks matter. Because of its focus on the consequences of health problems, the research also has implications for public policy discussions about the health care system and social insurance more generally.

Keywords

wealth debt kinship health

Introduction

Much of the literature establishing family connections in well-being and within family transfers has focused on either the parent–child or husband–wife dyad. The sibling tie, which is the family relationship of longest duration, has tended to receive much less attention. In part, this omission reflects a larger failure to develop normative frameworks of family relations that extend beyond partnership and parenting (Walker, Allen, & Connidis, 2005). Additionally, this omission reflects the complexity of the sibling tie itself, which Allan (1977) noted is often inactive or dormant until a time of need presents itself. In recent years, scholars and policy makers have been transfixed by the low levels of wealth of many American households and by the racial wealth gap (Oliver & Shapiro, 2005; Scholz, Sheshadri, & Khitatrakun, 2006). Given that participation in kin networks may impinge on wealth accumulation by diverting funds that could be saved toward needy relatives (Stack, 1974), or by restricting the amount of time that individuals have to spend in income-earning activities, examining family connections may be helpful for understanding wealth accumulation. The political philosopher Swift once opined that “choosing the right parents turns out to matter more than some had thought,” in his discussion of ways that economic advantage or disadvantage can be produced within families (Swift, 2005, p. 257). We wonder if having the “right” siblings might be equally important.

Our research takes advantage of the unique features of a nationally representative data set, the Panel Study of Income Dynamics (PSID), to examine connections among siblings. We examine individuals’ health circumstances and the extent to which being ill may provide a justification for tapping into one’s kin networks. Additionally, we examine the relationship between an individual’s wealth and the health of his or her siblings. Although the existing sociological research on inter- and intragenerational kinship ties has revealed much about the determinants of kinship configurations, work exploring how kinship networks matter has been more limited. Our research aims to contribute to filling this void by analyzing ways that kin attributes such as having a sibling in poor health might relate to an individual’s outcomes. Our article addresses the following research questions:

Research Question 1: Is wealth lower for individuals who have low-income siblings with health problems?

Research Question 2: Is there evidence that the presence of siblings with health problems affects Blacks and Whites differently?

To proceed toward answering these questions, we first review findings from the existing literatures concerning kin networks and health; along with empirical evidence regarding interfamily transfers. Then we discuss the theoretical perspectives that inform our analysis. Following this, we discuss our data set and explain how the PSID is uniquely suited to allow us to answer our research questions. An analysis of family-level panel data from the PSID that examines association between household wealth and the presence of any siblings with serious health conditions follows. The article concludes with a discussion of the implications of our findings.

Background

Why pose our research questions? Participation in kin networks has been shown to involve transfers of time between households in addition to transfers of funds. There is a large sociological literature documenting individuals’ participation in kin networks (Eggebeen & Hogan, 1990; Heflin & Pattillo, 2002; Hofferth, 1984; McAdoo, 1978; Raley, 1995; Schoeni, 1992; Spreitzer, Schoeni, & Rao, 1996; White & Riedmann, 1992). For example, in her classic book, All Our Kin, Carol Stack tells the story of how a $1,500 inheritance to the Waters family, initially hoped to be to be used as a down payment on a house, was instead spread over their family network to pay for funeral expenses, winter coats, and utilities. Similarly, in a study of middle-class African Americans, McAdoo (1978) finds that 92% of her sample felt obligated to help a less fortunate relative. This scholarship provides a foundation for thinking about a relationship between the circumstances of a relative, like a sibling, and an individual’s ability to accumulate wealth. Specifically, if some U.S. families are providing assistance to relatives, one might ask how this affects their prospects for accumulating wealth.

As shown in Table 1, national-level data indicate that households in which a family member has a chronic disease are more likely to receive financial help from relatives than other households.¹ Almost 13% of households in which the head or wife had been diagnosed with a chronic disease, such as cancer, lung disease, or had had a heart attack or a stroke, report having received help from their relatives in 2005. Among households in which neither the head or wife had a chronic disease, the proportion receiving help from relatives was 30% smaller.² These data suggest a correlation between having an illness and receiving transfers from one’s kin. Similarly, the PSID measure of receipt of help specifically for health care–related purposes indicates that the proportion of households receiving such help among those in which the head or the wife had a chronic disease is almost three times that of the percentage of households where neither the head nor the wife had a chronic disease (3.4% compared with 1.3%). Although the PSID data covering the receipt of help do not allow one to determine which relative provided the help, the findings are consistent with research by Connidis (1994). Connidis argued that there is solidarity among siblings when poor health is at issue. Using data from a survey of 528 respondents age 55 or older, Connidis (1994) asked about three specific types of sibling-to-sibling support: (a) help during an illness, (b) financial assistance, and (c) “help with other things.” The study found that about 10% of the respondents who had a sibling reported receiving help from that sibling when ill and that about 3.4% of respondents reported receiving money from their sibling during the survey year.

Table 1.

Proportion of Families Reporting the Receipt of Help From Relatives—by Health Status.

	Among Households Where Neither Head nor Wife Has a Chronic Disease	Among Households Where Head or Wife Does Have a Chronic Health Condition
Receipt of help from relatives	.100 (.0099)	.126* (.0122)
Received help from family or friends for health-care related purpose	.013 (.0024)	.034*** (.0057)

Note. PSID = Panel Study of Income Dynamics. All analyses use the PSID weights. Standard errors in parenthesis.

Source. Authors’ analysis of data from the 2005 PSID.

p ≤ .10. **p < .05. ***p < .01.

National data indicate that the amount of financial assistance that relatives provide to one another is not inconsequential. According to the PSID data, about 10% of households give money to others outside their immediate family in a given year.³ As shown in Table 2, the average amount of financial assistance given is about $567. If we compare this to average family income, this transfer represents about 1% of a family’s after-tax income.⁴ This number implies a foregone or unrealized saving rate of the same amount. Accordingly, in the absence of these interfamily transfers, households that are providing assistance could be saving at a higher rate and accumulating more wealth. Note that the average amount of the transfer among respondents who report giving to siblings is even higher ($2,482 vs. $567, which is the average across all families).

Table 2.

Household Transfer Amount Given.

	Mean ($)	N
All households	566.67 (125.54)	7,863
Conditional on giving	5,436.91 (1313.32)	790
Among those who report having given to a sibling	2,482.38*** (608)	136
Among those that did not give any support to siblings	539.40*** (127.78)	7,732

Note. PSID = Panel Study of Income Dynamics. All data are weighted. Standard errors in parentheses.

Source. Authors’ analysis of data from the 2005 PSID.

***

Indicates the difference between means is statistically significant at the p = .01 level.

Taken together, the descriptive data in Tables 1 and 2 provide empirical grounds for investigating whether one sibling’s health circumstances might be related to the other’s prospects for accumulating wealth. In a later section, we examine the relationship between these two variables directly. Like Connidis’s work, our research can be viewed as an exploration into ways that poor health may influence sibling solidarity. It therefore expands the volume of research on the sibling dyad. Because individuals of all ages can be plagued with health problems, our study examines a broader range of ages than Connidis (1994), including younger and middle-aged households’ health situations.

Theoretical Framework

Like McCann and Roberto (2012), we find an integrative theoretical approach to be a useful heuristic for studying our topic. Specifically, we link ideas from solidarity theory and models of altruism, while also incorporating theoretical insights from the existing literature on the health–wealth nexus, to formulate a conceptual framework to guide our analysis.

Solidarity Theory and Models of Altruism

In the family studies literature, solidarity theory often is applied as a framework for studying intergenerational relationships (Bengtson, Acock, Allen, Dilworth-Anderson, & Klein, 2005).⁵ This theory posits that solidarity is an important component of family relationships and that feelings of affection and warmth typically exist between family members and can motivate behavior. Altruism is another way of thinking about how feelings of affection and warmth might affect an individual’s behavior. In the existing literature on this topic, altruism is characterized as a situation in which an individual has “caring preferences” (Ermisch, 2003). Accordingly, a person is said to be altruistic toward another if the well-being of the former depends at least partially on the well-being of the latter.

For the purposes of this article, we envision a situation in which the altruist attaches some weight to her sibling’s well-being, but following Chiteji and Hamilton’s model, we envision an individual whose concern is actually dependent on the degree to which her intended beneficiary (her sibling) risks falling below some minimal standard of living that is unacceptable to the former (Chiteji & Hamilton, 2005). So, for example, if a sibling has a high standard of living, there may be little reason for the focal individual to attach much weight to her sibling’s circumstances, whereas if the sibling has a low standard of living, the focal individual will be more likely to exhibit substantial concern for the plight of the sibling and may subsequently decide that she needs to take action to help the sibling. This is akin to suggesting that the feelings of affection and warmth that exist between siblings will trigger assistance (monetary or other) largely when the intended beneficiary is economically vulnerable.

Theory of the Health–Wealth Nexus

Because we are interested in questions of health, our research is also informed by theory postulating a connection between health and wealth, which comes from the existing literature on health. This theory suggests that health problems may depress savings through a combination of reducing earnings and increasing expenditures on out-of-pocket medical expenses (Smith, 1999; Wenzlow, Mullahy, Robert, & Wolfe, 2004).⁶ For example, if an individual is required to reduce his or her work hours because he is sick or in order to care for an ill child, the individual’s income would be lowered, thereby decreasing the likelihood that he will have funds left over to save after paying for basic living expenses. Alternatively, if an individual is required to increase his expenditures due to a health problem, for example, because of the need to pay for medical services or goods (like pills) to treat his ailment or that of his child, then the individual will have less money to save (even if the illness does not require him to cut back on working).

Although the existing literature does not examine the relationship between an individual’s wealth level and the health of the individual’s siblings, it is clear that the scholarship that exists covering kin networks provides a foundation for thinking about such a relationship. If individuals are embedded in kin networks, and if they feel solidarity toward their siblings, it is possible that the processes that link bad health outcomes to low wealth for a given individual may operate with respect to the focal individual’s wealth and her sibling’s health. Specifically, if an individual assists an ill sibling by taking time off from work to help care for the sibling or by providing financial assistance to the sibling, then the nonsick individual’s wealth level would be affected by her sibling’s poor health. This suggests that there may be spillover effects onto other family members if some U.S. households experience difficulty paying their medical bills, or if the adults within them reduce their work hours or have to stop working entirely because of illness. These individuals may turn to siblings when their own resources are exhausted, or when they need help caring for themselves or someone ill within their household.

Working with the framework for thinking about altruism that was exposited above, it is important to remember that the theory suggests that an altruistic individual will assist her sibling only if the sibling is economically vulnerable. That is to say that, for example, if an individual has a rich sibling who can cover his own medical expenses (or hire a home health care aid) without impoverishing himself, the focal individual may attach little weight to assisting that sibling. Taking an extreme example, the intuition is as follows: If the sibling is much richer than the focal individual—say a millionaire—and covering medical expenses only requires the sibling to forego purchasing a yacht or a pair of shoes (or similarly unsubstantial goods), there is little reason for the altruist to worry about reducing her own consumption to make a transfer to her sibling, either by giving of her time or her money. It is worth noting that such a result would be consistent with stratification theory, which suggests that a person’s position in the social structure will be a factor that affects his experiences in life. Because disadvantaged individuals’ experiences may be different from other people’s experience, our empirical work will examine individuals in different income groups.

Data

Our regression analysis draws on data from the PSID. In what follows, we provide some background information about this data set. We also explain how we constructed our sample and discuss some key variables that are part of our analysis.

The PSID is a longitudinal national survey of U.S. families. It has followed families and their offspring since 1968, annually from 1968 to 1997 and biennially since that time. One of the unique characteristics of this data set is that it contains information about adult individuals and their siblings. Among the wide variety of data included in the PSID are data characterizing families’ and individuals’ demographic characteristics, their economic situation, and their health status. As of 2005, the survey had collected data on more than 8,000 families and on more than 60,000 individuals. Although there has been attrition out of the PSID since its inception, the fact that it follows the children of the original 5,000 families over time (along with the families these “children” establish when they are old enough to set up their own households), and the fact that it refreshes the sample with “births,” means that the PSID continues to be representative and thus an excellent source of data for social science research (Fitzgerald, Gottschalk, & Moffitt, 1998).

The PSID has collected extensive information about individual earnings, labor force participation, marital status, and childbearing since its beginning. In 1984, the survey also began to ask a series of questions about wealth, assets, and debt. Because the wealth-related data are collected at the family-level, however, we cannot use individuals as the focal point for our empirical work. Instead, the family is the unit of analysis for the empirical results discussed later. A variety of health data also are collected in the PSID. These data are collected for specific individuals in the family, namely, the heads and wives of the families who are interviewed. Included among this information are data about the onset of several major illnesses, mental health, and health-related behaviors such as exercising.

Our analysis examines families whose heads are age 25 to 54 years at the 2005 wave of the survey. Each of these families has at least one sibling (or sibling family). To answer our research question, we also incorporate data describing these siblings’ situations into our analysis. The siblings were identified using the PSID’s online Family Identification Mapping System (FIMS). Because the PSID collects its health data by asking about the health of the head and wife of each family unit, as mentioned earlier, we have to use the information from the family in which a sibling was a “head” or a “wife” to extract information about the sibling’s health status.⁷ Although FIMS allows a researcher to match individuals to all their siblings, including step-siblings, our analysis only includes full siblings.⁸ We obtain the socioeconomic, demographic, and health variables that are used in our analysis from the 1999 to 2005 waves of the PSID.⁹

There are 8,002 families in the 2005 wave of the PSID. Restricting the sample to those families with heads aged 25 to 54 years reduces the size of the data set to 5,373. We were able to identify the siblings for 4,059 of these families from 1,352 extended families. Because the PSID only follows individuals with the PSID gene, we do not typically have siblings from both sides of the family tree for families that contain both a head and a wife. Instead, we are likely to observe only the wife’s siblings or only the head’s siblings. Pooling across time gives us a total of 16,236 cases of which 14,978 contain nonmissing data.

Unless otherwise noted, all measures included in our regressions come from the 1999 to 2005 waves of the PSID. Because the PSID surveyed families biennially during this time, this time window represents four different waves or calendar years. We chose 1999 as the start period because this is the first year for which the data covering severe health problems were collected. We can observe individuals’ health problems and their siblings’ health in every wave from 1999 to 2005, along with wealth, indebtedness, age, education, marital status, and number of children.

To see how the relationship between siblings’ health and respondent wealth varies at different points in the income distribution, we construct three subsamples with our data based on the distribution of current household income. Essentially, we break our siblings and our main families into low income, middle income, and high income groupings. We define our low-income sample as those with total family income below the 33rd percentile of the income distribution.¹⁰ The middle-income group is defined as those between the 33rd and 66th percentile. Finally, we define our high-income distribution as those above the 66th percentile. It should be noted that although we control for long-run income in our regressions by using the average of household income over the last three observation points, we use contemporaneous household income when we split the sample into the income categories. Conceptually, this approach assumes that although long-run income is the appropriate measure of income to include in a model explaining wealth accumulation, it is current income that is likely to determine whether an individual requires help from a sibling. The sensitivity of our results to this approach is discussed below.

Measures

To conduct our analysis we need a measure of household wealth and some measure of the health of a person’s siblings. The analysis also requires several control variables so that the regressions will be properly specified.

Sibling Health

From 1999 to 2005, the PSID collected data about heads’ and wives’ health status to determine whether the head or wife has a severe health problem. The diseases asked about were stroke, cancer, lung disease, heart attack, and emotional problems. We used information on the first four—the physical health measures—to construct a variable that indicates whether the sibling family had severe health problems. Our sibling-health variable combines the information reported for the head and wife of the sibling family in instances in which a sibling is married or cohabitating. We then did further aggregation, this time across all the siblings (or sibling families) of a given respondent to construct a variable that indicates whether any of the respondent’s siblings has experienced a health problem. That is to say we scanned across all sibling families to construct a measure of whether or not our main family has at least one sibling that has experienced a health problem.

Wealth

For our dependent variables in our regressions, we focus on the PSID measure that indicates household net worth including home equity as our primary wealth measure of interest. Net worth represents the total value of all of a household’s assets minus its debts. Our second wealth variable is a measure of net worth that excludes home equity, which is equal to the negative of net worth if net worth is negative and zero otherwise. These two wealth measures that we use represent the broadest measures of wealth in the PSID.

Control Variables

We include a set of standard economic and demographic controls that are known to be associated with wealth accumulation in all our reduced form models: the age of the household head, age squared, long-run income, education level, marital status (married, never married, and divorced, separated, or widowed), number of children, and a dummy variable indicating if the head is Black (0 otherwise). Moreover, we also use the PSID’s health questions to construct a measure of own health. As noted earlier, the existing literature suggests that there is an association between own health and own wealth. We therefore use the PSID’s data covering chronic conditions (discussed above) to measure the health of the head and wife of each of our main families.

Our regressions use a long-run income measure rather than contemporaneous income. This measure looks across 3 years, including the year of analysis and two PSID years prior to that, and it averages the income data reported for the different years in this range. So, for example, long-run income for 2005 would average income from the 2001 to 2005 wave. Our long-run income measure for 1999 necessarily averages data from the 1999 wave of the PSID and the 1997 and 1996 waves as well (although these latter two waves are not used elsewhere). Because the construction of this variable requires us to look at the same family over a number of different years, we imposed the “same head” requirement when constructing our long-run, average-income measure.

We also include controls for parental education and parental poverty because we were concerned that any association that we observed between wealth and sibling health could potentially be confounded by the presence of a common discount rate among adult siblings.¹¹ For example, it has been argued that individuals who grow up in poverty tend to discount the future more than individuals who do not. In one of the most commonly cited references on the topic within economics, Lawrance (1991) finds empirical evidence of an association between poverty and the rate of time preference. If individuals who come from poor backgrounds discount the future more than other individuals do, then one would expect an individual who grew up poor to accumulate less wealth than an individual who did not (because individuals who discount the future greatly will be less inclined to save for the future than other individuals do). Moreover, the same individual might have siblings whose health is poorer than the siblings of other individuals because discount rates also are thought to influence health. A wide body of research within health economics argues that individuals who discount the future greatly tend to value the long-run benefits of health less than other individuals do, and may engage in unhealthy behavior because of this (Fuchs, 1982; Grossman, 2000). Similarly, research within economics and psychology suggests that individuals’ attitudes toward the future may be shaped during childhood by their parents’ levels of education (Webley & Nyhus, 2006).¹²

Table 3 presents descriptive statistics for all the variables included in our models. All variables measuring dollar amounts are measured in 2005 constant dollars. Only variables indicating siblings’ health problems are presented in Tables 4 and 5.

Table 3.

Descriptive Statistics.

	Low-Income Sample (n = 5,449)	Middle-Income Sample (n = 4,687)	High-Income Sample (n = 4,767)
Wealth	$28,698 (1,56,250)	$93,908 (3,79,722)	$376,759 (1,470,565)
Black household	20.6%	11.1%	5.0%
Age of head	38.6 (12.2)	40.8 (9.9)	44.6 (8.2)
Total long-run family income	$40,622 (41,031)	$56,947 (30,020)	$1,12,233 (86,490)
% Divorced, separated, widowed	42.7%	18.0%	5.7%
% Never married	32.0%	19.2%	7.8%
% Married	25.3%	62.8%	86.5%
Number of kids	0.68 (1.09)	0.90 (1.11)	0.99 (1.10)
Years of education of head	12.71 (2.22)	13.40 (2.06)	14.70 (1.98)
Head has serious health problem	11.8%	7.5%	6.8%
Wife has serious health problem	2.9%	5.4%	5.0%
Low-income sibling has serious health problem	7.1%	4.6%	2.4%
Middle-income sibling has serious health problem	19.8%	17.3%	15.3%
High-income sibling has serious health problem	7.3%	7.6%	8.9%
Years of parental education	11.98 (2.53)	11.25 (2.37)	13.00 (2.40)
Parental poverty	22.1%	26.7%	21.3%

Note. PSID = Panel Study of Income Dynamics. Mean (standard errors) or percentages presented. Low-income sample defined as household income below the 33rd percentile of income distribution (<$35,500). Middle-income sample defined as household income between 33rd and 66th percentile of income distribution ($35,500 to $74,000). High-income sample defined as household income above the 66th percentile of income distribution (>$74,000).

Source. Authors’ calculations based on the 1999-2005 PSID.

Table 4.

Model of Individual Wealth and Sibling Health Shocks.

	Low-Income Sample (n = 5,449)	Middle-Income Sample (n = 4,687)	High-Income Sample (n = 4,767)
Poor sibling models
Poor sibling serious health problem	−$13,840** (6,381)	−$41,504** (16,797)	−$153,753*** (59,072)
Middle-income sibling models
Middle-income sibling serious health problem	−$9,824 (7,527)	−$4,314 (18,206)	$48,877 (41,465)
High-income sibling models
High-income sibling serious health problem	$1,652 (14,438)	$24,212 (32,344)	$71,477 (58,594)

Note. PSID = Panel Study of Income Dynamics. Coefficients (standard errors) presented. All models include controls for parental education, parental poverty, age of head, age-squared, marital status, education level of head, race (1 = Black; 0 = otherwise), long-run income, number of children, and health of head and wife.

Source. Authors’ calculations based on the 1999-2005 PSID.

***

p < .01. **p < .05. *p < .10.

Table 5.

Model of Individual Wealth and Sibling Health Shocks With Race Interaction.

	Low-Income Sample (n = 5,449)	Middle-Income Sample (n = 4,687)	High-Income Sample (n = 4,767)
Poor sibling models
Poor sibling serious health problem	−$18,272** (8,630)	−$41,663** (17,927)	−$153,387** (65,361)
Black * Poor sibling serious health problem	$13,895 (9,704)	$874 (27,551)	−$3,331 (2,14,302)
Middle-income sibling models
Middle-income sibling serious health problem	−$12,410 (8,908)	−$1,581 (19,889)	$56,424 (43,316)
Black * Middle-income sibling serious health problem	$13,702 (11,463)	$30,311 (26,576)	−$228,545 (2,18,605)
High-income sibling models
High-income sibling serious health problem	−$432 (15,526)	$27,292 (33,844)	$76,300 (59,473)
Black * High-income sibling serious health problem	$21,488 (35,220)	−$71,277* (39,756)	−$364,840 (2,59,450)

Source. Authors’ calculations based on the 1999-2005 PSID.

***

p < .01. **p < .05. *p < .10.

Method

Because we are interested in examining if health problems experienced by siblings are related to the amount of wealth that a household possesses, our empirical strategy is to estimate the following equation:¹³

Y_{i t} = α + X_{i t} λ + β Z_{i t} + ε_{i t},

where Y_it equals the amount of wealth held by the ith household at time t. X_it is a vector of individual characteristics at time t associated with household wealth, such as age, race, education, income, marital status, number of children, and own health problems, parental education level and if the head or wife grew up in poverty; and Z_it is a measure indicating if household i has a sibling who reports having a serious health problem.¹⁴ ε_it is a normally distributed error term with a mean of zero and constant variance clustered at the extended-family level. Note that we construct different Z measures to classify siblings by their income category in addition to the presence of health condition and run separate regressions for each “type” of sibling. That is to say, we report results for (a) regressions that examine the effect of having a low-income sibling in bad health, (b) for regressions that examine the effect of having a middle-income sibling with a health problem, and (c) for regressions that analyze the effect of having a high-income sibling who has a health problem. Our models are estimated on a pooled sample using data from 1999 to 2005 using ordinary least squares.

Results

We begin with an exploration of how siblings’ health problems may affect the wealth of individuals in low-income, middle-income, and high-income families, as shown in Table 4. These models, like all those presented, include controls for the age of the head, age-squared, marital status, education level, race, long-run income, the health of the head and wife, the number of children in the family, parental education, and if the head or wife grew up in poverty.

The Relationship Between Sibling Health and Wealth Accumulation

The regressions indicate that among families in all income groupings, there is a negative association between wealth and having a poor sibling who has poor health, although the magnitude of the estimated coefficient varies dramatically.¹⁵ For example, among low-income households, those with a poor sibling in poor health have almost $14,000 less wealth than other households. Among middle-income households, having a poor sibling with health problems is associated with a reduction of almost $42,000 in wealth and this reduction in wealth grows to $154,000 among high-income households.

The regressions analyzing health problems among higher-income siblings provide a nice falsification test for our hypothesis. Although our hypothesis suggests that low-income individuals may turn to their siblings for assistance out of necessity, there is really no reason to expect higher-income siblings to do the same. Results for models of health problems experienced by siblings with income above $74,000 suggest that these siblings’ health problems do not influence the amount of wealth that an individual has.¹⁶ Regardless of whether one examines low-income households, middle-income households, or high-income households, reported health problems among high-income siblings do not appear to be associated with wealth accumulation. This finding is consistent with the hypothesis that high-income siblings are better able to cover their own medical expenses than low-income siblings (to purchase medicine or in-home care, for example), or to support themselves if an illness requires them to stop working, and that the former are less likely to turn to others for assistance.

Analyses by Race

To determine if the patterns found above are consistent across racial groups, we replicate Table 4 with the addition of an interaction term between being Black and our measure of sibling health problems. The results are shown in Table 5. This interaction term typically is not statistically significant. Accordingly, there is no evidence that there are racial differences in the effect of having a poor sibling with a health problem. Additionally, the main effect of having a poor sibling with a health problem remains quite robust to the inclusion of the interaction term.

Thus, results of models testing the presence of racial variation in the sensitivity of wealth holdings to sibling health indicate that there are few differences between Blacks and Whites.

The Relationship Between Sibling Health and Debt Accumulation

A closely related question to the one explored above is whether debt holdings might be affected by sibling health. Arguing that an individual might be willing to incur debt to help a poor sibling can be thought of as noting that anyone who faces a decision about whether to help a sibling can base his decision about whether he can afford to provide assistance on his permanent income (not just his current income). That is to say, just as he can borrow against his future income to contend with unexpected expenses associated with a shock to his own health, as standard economic theory already acknowledges, an individual could also borrow against future income to obtain the funds he needs to help a relative contend with a health problem. Knowing whether individuals behave this way could allow for a more nuanced interpretation of the wealth result. Because net worth measures the value of total assets relative to the value of total debt, it can be lower for individuals who have siblings in poor health for one of two reasons: either (a) because the individual draws down on his stock of savings (or liquidates some assets to help a sibling who is in poor health) or (b) because the former borrows to help the latter.

To investigate this question, we replicated the analysis presented above with two different measures of indebtedness. The first is a measure of noncollateralized debt. Although this debt category can include amounts owed as credit card debt, for medical bills, legal bills, student loans, and loans from relatives, it is generally treated by researchers as a measure of credit card debt. In addition to the debt data, we used the PSID wealth variable that was analyzed in the previous section to construct a second measure of indebtedness—negative net worth. This measure has been used in the literature as a measure of asset deprivation (Haveman & Wolff, 2005). It also can be viewed as a measure of indebtedness however, since it identifies families who have accumulated enough debt to make the value of their debt exceed the total value of their assets (Chiteji, 2007).

For each measure of indebtedness, we ran the same set of nine models shown for wealth in Table 4, substituting tobit models for ordinary least squares.¹⁷ The only statistically significant finding occurs when we examine the indebtedness of low-income households. For models of negative net worth, we find that having a poor sibling with a serious health problem is associated with an approximately $12,000 higher indebtedness (p = .061). For both debt measures, we find that having a middle-income sibling with a health problem was marginally associated with higher levels of debt ($3,173 for noncollateralized debt, p = .088; $8,325 for negative net worth, p = .076) among low-income households. These findings are consistent with the hypothesis that households may borrow against future income to help siblings with medical problems. Although at first glance one might expect low-income households to be the least likely type of household to borrow to help a sibling because their income may not be expected to grow much over time, our finding that they may take on debt to help a sibling while other households do not is understandable. Low-income households may be less able to draw on preexisting savings when a sibling is in need than other households. Because they are less likely to have reserve funds due to their low levels of wealth, borrowing may be an easier way for them to provide help. Middle-class households and high-income households would be in a different position. They are more likely to have substantial savings that they can tap into to help a sibling in need.

Sensitivity Analyses

One issue that bears exploring is if the use of current income to define the income groups (low income, middle income, and high income) has any consequence for the results. To test the sensitivity of our results to this decision, we replicated Table 4 using our long-run measure of household income to define income groups. Conceptually, this analysis assumes that small fluctuations in annual income may mean that a household only belongs to an income classification for a short time and their behavior is better predicted by smoothing their income. However, when we examine results using our long-run measure of income to classify households, we find that results are largely consistent to those shown in Table 4. Low-income households whose siblings have health problems have lower levels of wealth than those whose siblings do not have health problems (without regard to the income classification of the sibling). Additionally, we find that having a low-income sibling with a health problem is associated with lower levels of wealth among both middle- and high-income households. Thus, the results are at least as strong as those shown in Table 4 when using the long-run income measure to assign class status.

Might our sibling variable simply be picking up the effect of a shared extended-family trait not captured by parental education or by growing up poor? The finding that the effects of sibling health problems depend on the economic status of the sibling in poor health suggests that the regressions are not simply capturing some kind of extended-family background effect such as a shared environment while growing up. If the sibling variable reflected the influence of a shared extended-family trait, then there would be no reason for its effect to vary depending on whether the sibling turned out to be high income or not as an adult. Yet we find that bad health matters when the sibling is low income, but not when he or she is higher income. The most reasonable interpretation of this outcome is that when high-income siblings experience a health problem, they do not need to seek assistance from a sibling, whereas less fortunate siblings can require help to contend with the problem.

Discussion

The association between siblings’ health and own wealth is particularly interesting to explore in light of recent discussions about health care. The United States has the benefit of having one of the highest life expectancy rates in the world, indicating that the health of its population is better, on average, than that of many other countries. Yet this does not make Americans exempt from having to struggle with health problems. People fall ill. Some have to take sick days from work as a result, and some are hospitalized within their lifetime. In fact, data from the 2011 U.S. Statistical Abstract indicate that in 2008, 118 per 1,000 people were admitted to the hospital in the United States (p. 119). It is perhaps no wonder then that discussions in both the popular press and the policy arena suggest that many families worry about getting sick and that others are saddled with medical bills (Doty, Edwards, & Holmgren, 2005; Heavey, 2008; Himmelstein, Warren, Thorne, & Woolhandler, 2005; Zeldin & Rukavina, 2007). Given the economic costs that illness can impose on a sick individual, it is not unreasonable to expect that some American households might not possess the ability to deal with the crisis on their own and that they might turn to a sibling for help. Accordingly, there may be spillover effects on other family members. This fact could explain the association that we find between having a low-income sibling in poor health and low levels of wealth.

As noted earlier, there are multiple pathways through which a sibling’s health might affect another’s ability to accumulate wealth. Siblings may need help paying medical bills, buying medicine, covering their health insurance premiums, or they may need help covering basic living expenses if their poor health has forced them to reduce their work hours. Alternatively, they may need a family member to help care for them, be it through visits to a hospital or extended care at home. All present possible pathways through which an individual might be affected by having a sibling who is in poor health. We leave it to future research to construct tests that distinguish between them.

Although not definitive, the regression results presented above are suggestive of the presence of family connections between adult siblings in times of need. Our findings are consistent with Allan’s (1977) conclusion that sibling ties are often inactive or dormant until a time of need presents itself. We view our work as a part of a continued conversation about adult siblings’ connections to one another, one that pushes the literature on the family to continue to move beyond parent–child connections and connections between husbands and wives, in order to consider sibling ties when attempting to understand family relationships.

Additionally, a connection between one sibling’s health problems and a different sibling’s economic circumstances provides researchers in the literature covering saving behavior with much food for thought. That literature typically has focused narrowly on the relationship between one’s own characteristics (including one’s health) and one’s wealth. Yet the finding that there is an association between an individual’s wealth and her siblings’ health—in addition to the individual’s own health problems—necessarily challenges this literature to think more broadly about the “family” and the range of family members whose circumstances may affect a person’s savings, showing how family studies research can inform scholarship outside the discipline.

We would be remiss if we failed to note that our work also has implications for discussions about the transmission of advantage or disadvantage within the family. As noted by Swift (2005), it should surprise no one that the family sometimes hinders the achievement of equality of opportunity. Families of means routinely pass along advantages to their offspring—by making financial transfers to them, by investing in them during childhood, and through other activities that are “constitutive” of the family. In fact, as Shapiro (2004) writes, “Americans highly value two cherished but contradictory notions: equal opportunity and a family’s ability to pass along advantages to their children.” (p. 36) Although our research does not address questions about how families transmit advantage or engage in social positioning, it does suggest one avenue through which the circumstances of one’s extended family may disadvantage another individual. Hard-working individuals who have siblings who fall ill may find themselves unable to save as much as they desire, if they feel obligated to provide assistance to the sick sibling. In a society with insufficient social insurance protections for all, this problem will be particularly acute.

Conclusion

The economist Adam Smith (1854/2000) once wrote, “How selfish soever man may be supposed, there are evidently some principles in his nature, which interest him in the fortune of others.” (p. 3) Such concern for others’ situations certainly applies to the family, where bonds of affection and responsibility often tie individuals to one another. Drawing on sociological research on the presence of kinship ties and solidarity among family members, this article has explored the hypothesis that there may be spillover effects onto others when a sibling falls ill. Specifically, we examined if there is evidence of assistance within kin networks by examining if household wealth is related to the health of siblings. The article finds evidence that having a sibling who has experienced a health problem is associated with a decrease in the amount of wealth that some families have.

The research has implications for the existing literatures on kin networks, as it sheds some light on the nature of altruism that prevails in U.S. families and on how kinship networks matter. Moreover, it expands the existing literature on ties between siblings, and offers a deeper understanding of the connections between health problems and families’ saving behavior that may benefit policy researchers and policy makers. Two of the biggest policy concerns about the present configuration of the U.S. health care system are that it leaves many people uncovered and that the costs of medical services are rising faster than the rate of growth of prices for other goods and services in the economy. Knowing the extent to which prime-aged households are affected by health problems, the ways they are affected, and the extent to which families are required to help each other buffer shocks may help policy makers make informed choices about how to improve the present system.

Footnotes

Appendix

Table B1.

Model of Negative Net Worth and Sibling Health Shocks.

	Low-Income Sample (n = 4,545)	Middle-Income Sample (n = 4,054)	High Income Sample (n = 4,118)
Poor sibling models
Poor sibling serious health problem	12044.97* (6425.611)	−111.0147 (10753.93)	−47295.14 (33560.56)
Middle-income sibling models
Middle-income sibling serious health problem	8324.93* (4696.249)	7078.357 (5576.462)	−18049.86 (12086.59)
High-income sibling models
High-income sibling serious health problem	12948.31 (8264.812)	9193.201 (8058.685)	−8103.947 (13914.56)

Note. Authors’ calculations based on the 1999-2005 PSID; coefficients (standard errors) presented. All models include controls for parental education, parental poverty, age of head, age-squared, marital status, education level of head, race (1 = Black; 0 = otherwise), long-run income, number of children, and health of head and wife.

***

p < .01. **p < .05. *p < .10.

Acknowledgements

The authors would like to thank participants at the PSID’s SES and Health across Generations and Over the Life Course workshop for their insightful comments.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Chiteji wishes to acknowledge the financial support of a Skidmore College SUN grant award, made possible by NSF Advance Grant #0820080. Dr. Heflin acknowledges the support received from the University of Missouri’s Summer Research Fellowship.

Notes

References

Allan

(1977). Sibling solidarity. Journal of Marriage and the Family, 39, 177-184.

Bengtson

Acock

Allen

Dilworth-Anderson

Klein

(2005). Sourcebook of family theory and research. Thousand Oaks, CA: Sage.

Chiteji

(2007). To have and to hold: An analysis of young adult debt. In Danziger

S. H.

Rouse

C. E.

(Eds.), The price of independence: The economics of early adulthood (pp. 239-258). New York, NY: Russell Sage.

Chiteji

Hamilton

(2005). Family matters: Kin networks and asset accumulation. In Sherraden

(Ed.), Inclusion in the American dream: Assets, poverty and public policy (pp. 87-111). New York, NY: Oxford University Press.

Connidis

(1994). Sibling support in older age. Journal of Gerontology, 49, S309-S317.

Doty

M. M.

Edwards

J. N.

Holmgren

A. L.

(2005, August). Seeing red: Americans driven into debt by medical bills (Vol. 10). New York, NY: Commonwealth Fund.

Eggebeen

Hogan

D. P.

(1990). Giving between generations in American families. Human Nature, 1, 211-232.

Ermisch

(2003). An economic analysis of the family. Princeton, NJ: Princeton University Press.

Fitzgerald

Gottschalk

Moffitt

(1998). Analysis of sample attrition in panel data: The Michigan Panel Study of Income Dynamics (Technical Working Paper, No. 0220). Cambridge, MA: National Bureau of Economic Research.

10.

Fuchs

(1982). Time preference and health: An exploratory study. In Fuchs

V. R.

(Ed.), Economic aspects of health (pp. 93-120). Chicago, IL: University of Chicago Press.

11.

Gittleman

Wolff

(2004). Racial differences in patterns of wealth accumulation. Journal of Human Resources, 34, 193-227.

12.

Grossman

(2000). The human capital model. In Culyer

A. J.

Newhouse

J. P.

(Eds.), Handbook of health economics (Vol. 1-A, pp. 348-408). New York, NY: Elsevier.

13.

Haveman

Wolff

(2005). The concept and measurement of asset poverty: Levels, trends and composition for the U.S., 1983-2001. Journal of Economic Inequality, 2, 145-169.

14.

Heavey

(2008, August 20). Consumers face rising medical debt. Retrieved from http://www.reuters.com/article/2008/08/20/us-medicaldebt-idUSN1932186920080820

15.

Heflin

Pattillo

(2002). Kin effects on Black-White account and home ownership. Sociological Inquiry, 72, 220-239.

16.

Himmelstein

D. U.

Warren

Thorne

Woolhandler

(2005). Illness and injury as contributors to bankruptcy. Health Affairs, 24, w63-w73.

17.

Hofferth

(1984). Kin networks, race, and family structure. Journal of Marriage and the Family, 46, 791-806.

18.

Hurst

Luoh

Stafford

(1998). The wealth dynamics of American families, 1984-1994. Brookings Papers on Economic Activity, (1), 267-337.

19.

Lawrance

(1991). Poverty and the rate of time preference. Journal of Political Economy, 99, 54-77.

20.

McAdoo

H. P.

(1978). Factors related to stability in upwardly mobile Black families. Journal of Marriage and the Family, 40, 761-776.

21.

McCann

B. R.

Roberto

K. A.

(2012). Relational systems: How older women with chronic health problems construct close relationships. Journal of Family Issues, 33, 559-583. doi:10.1177/0192513X11414796

22.

Michaud

P. C.

van Soest

(2008). Health and wealth of elderly couples: Causality tests using dynamic panel data models. Journal of Health Economics, 27, 1312-1325.

23.

Oliver

Shapiro

(2005). Black wealth/White wealth: A new perspective on racial inequality. New York, NY: Routledge.

24.

Raley

R. K.

(1995). Black-White differences in kin contact and exchange among never married adults. Journal of Family Issues, 16, 77-103.

25.

Schoeni

R. F.

(1992). Essays on private interhousehold transfers and the family (Doctoral dissertation). University of Michigan, Ann Arbor.

26.

Scholz

J. K.

Sheshadri

Khitatrakun

(2006). Are households saving optimally for retirement? Journal of Political Economy, 114, 607-643.

27.

Shapiro

(2004). The hidden costs of being African American: How wealth perpetuates inequality. New York, NY: Oxford University Press.

28.

Smith

(2000). The theory of moral sentiments. New York, NY: Prometheus Books. (Original work published 1854)

29.

Smith

J. P.

(1999). Healthy bodies and thick wallets: The dual relation between health and economic status. Journal of Economic Perspectives, 13, 145-166.

30.

Spreitzer

Schoeni

Rao

K. V.

(1996). Tracing intergenerational relations through reports of transfers of time and money: A comparative analysis of African-Americans, Hispanics, and Whites. International Journal of Sociology and Social Policy, 16(12), 11-34.

31.

Stack

(1974). All our kin: Strategies for survival in a Black community. New York, NY: Harper & Row.

32.

Swift

(2005). Justice, luck and the family. In Bowles

Gintis

Osborne-Groves

(Eds.), Unequal chances: Family background and economic success (pp. 233-256). New York, NY: Russell Sage Foundation.

33.

U.S. Census Bureau, Statistical Abstract of the United States: 2011 (130th Edition) Washington, DC, 2011; Table 171: Hospital Use Rates by Type of Hospital: 1990-2008 <http://www.census.gov/compendia/statab/>

34.

Walker

A. J.

Allen

K. R.

Connidis

I. A.

(2005). Theorizing and studying sibling ties in adulthood. In Bengtson

V. L.

Acock

A. C.

Allen

K. R.

Dilworth-Anderson

Klein

D. M.

(Eds.), Sourcebook of family theory & research (pp. 167-190). Thousand Oaks, CA: Sage.

35.

Webley

Nyhus

(2006). Parents’ influence on children’s future-orientation and saving. Journal of Economic Psychology, 27, 140-164.

36.

Wenzlow

Mullahy

Robert

Wolfe

(2004). An empirical investigation of the relationship between health and wealth using the Survey of Consumer Finances (Russell Sage Foundation Working Paper). New York, NY: Russell Sage Foundation.

37.

White

L. K.

Riedmann

(1992). Ties among adult siblings. Social Forces, 71, 85-102.

38.

Zeldin

Rukavina

(2007). Borrowing to stay healthy: How credit card debt is related to medical expenses. New York, NY: Demos.