Abstract

Recent work suggests that part of the racial gap in wealth is explained by racial differences in network poverty. In this article, data from the 2007 Survey of Consumer Finances and the 2005 and 2007 Panel Study of Income Dynamics (PSID) are used to demonstrate that middle- and upper-income blacks are more likely to provide informal financial assistance than their white counterparts. Further, a lagged model using the PSID finds that this difference in financial assistance can account for part of the racial gap in wealth. An empirically useful definition of negative social capital is developed to illustrate how obligations of group membership can have stratifying consequences for individuals.

Introduction

The racial gap in wealth is dramatic, persistent and has important consequences for individual life chances and group-level differences beyond economic stratification and inequality (Oliver and Shapiro 2006; Shapiro 2004; Conley 1999). Recent studies suggest that a substantial percentage of the racial wealth gap can be accounted for by the fact that middle- and upper-income blacks are more likely to have friends and relatives in poverty than their white counterparts. This argument, made most recently by Heflin and Pattillo (2006) and Chiteji and Hamilton (2002), rests on the assumption that middle- and upper-income blacks are more likely than whites to provide informal financial assistance to friends and family, thereby reducing the capital they have to invest in wealth-producing assets. According to Portes (1998), this is an example of negative social capital, whereby excess claims are made on group members: the pressures placed on individuals by virtue of being part of a group – in this case a black kin network – result in negative consequences for the individual, here in the form of reduced economic capital.

Despite a robust literature analyzing rates of informal assistance by race, to date there has been no systematic test for race differences in the provision of informal financial assistance by income. Given the importance of this question to both material explanations of the racial wealth gap and our theoretical understanding of how membership in groups can have negative, stratifying consequences, differences in the provision of informal financial assistance between blacks and whites by income must be examined. Consequently, this article has two empirical aims. First, to test whether middle- and upper-income blacks are more likely to provide informal financial assistance than middle- and upper-income whites and, second, to determine if the provision of informal financial assistance by middle- and upper-income blacks can account for differences in the relative wealth holdings of these black and white families.

The article proceeds as follows. I begin with a discussion of negative social capital, ultimately refining the conception offered by Portes (1998) into a definition that is logically consistent with the field's current use of social capital and that can be usefully operationalized for empirical investigations. I then offer a framework for how to consider the provision of informal financial assistance by middle- and upper-income1 blacks as both an example of negative social capital and a mechanism of racial stratification and inequality. Following a review of the relevant motivating literature, I present the data and methods used and the results of the empirical analysis. I conclude with a discussion of the material implications of the findings and argue for the use of negative social capital as a construct in the study of social stratification.

Defining Negative Social Capital

Recent decades have seen the rise of “social capital” as a conceptual tool and popular construct in social science research and beyond. The term has been increasingly employed in sociology as both a dependent and independent variable of inquiry. The term has also migrated across the social sciences (Glaeser, Laibson and Sacerdote 2002; Putnam 2000) and found its way into lay vernacular (see, for example, Brooks 2010). Given the concept's power and popularity as a heuristic device, it is not surprising that some have built on the term to create newly specified but related concepts, including negative social capital, a theoretical descendent derived by Portes (1993, 1998) and applied across the field (see, for example, Wacquant 1998). Although the term is in circulation, there exists no general definition of negative social capital or systematic way to situate it vis-à-vis its theoretical antecedent.

The logical first step in pursuit of a consensus definition of negative social capital is to define social capital. Unfortunately, there are competing and conflicting notions of what this concept encompasses. The natural point of departure in sociology is Bourdieu, who defines social capital as “the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance or recognition” (Bourdieu 1986:248; also see Bourdieu and Wacquant 1992:119). Coleman is credited as among the first to employ the term in empirical research. In describing the role of the family in the reproduction of inequality, Coleman develops a working definition of social capital that, like Bourdieu, also stresses its instrumental nature: “social capital is defined by its function. It is not a single entity, but a variety of different entities having two characteristics in common: They all consist of some aspect of social structure, and they facilitate certain actions of individuals who are within the structure” (Coleman 1990:302; see also Coleman 1988). From the perspective of political science, Putnam uses social capital to refer to “features of social organization such as networks, norms, and social trust that facilitate coordination and cooperation for mutual benefit” (1995:67).

For Bourdieu, Coleman and Putnam, social capital is a capacity of actors and their networks, one that is inherently neutral and can be mobilized for both positive and negative ends.2 For each, social capital is a characteristic of groups that describes a capability of actors within these groups. All agree this capacity can be used for purposes both positive and negative. Yet each author focuses on the instrumental or positive aspect of social capital for individual actors, only speaking of the negative aspects of social capital in the aggregate, describing its role in the reproduction of inequality across groups.

This view of social capital as a fundamentally positive force for the individual is best illustrated in the definition offered by Portes in his widely cited review and discussion of the concept: “the consensus is growing in the literature that social capital stands for the ability of actors to secure benefits by virtue of membership in social networks or other social structures” (Portes 1998:6). Later in the same article, Portes introduces negative social capital as a concept to capture the “negative consequences” of social capital for the individual. But this definition is problematic. How are we to understand “negative consequences” stemming from an “individual's ability to secure benefits by virtue of group membership”? Do these negative consequences result from having social capital or are they more basic consequences of group membership?

The necessary condition for social capital is membership in a group. I argue that the nature of the group – and the individual actor's specific place in that group – yield potentially two different types of social capital, labeled as (positive) social capital and negative social capital, respectively, that manifest in a variety of ways. Portes and Sensenbrenner (1993) actually make this point in an earlier work, noting that group membership has both positive and negative consequences for the individual. If social capital is currently used to connote the positive consequences, or “benefits,” of group membership for the individual, then negative social capital can be employed to connote the negative consequences, or “liabilities,” of group membership for the individual. Current uses of negative social capital focus on phenomena at the community (Wacquant 1998) and state level (see, for example, Paldam and Svendsen 2000) – not the individual. Given the rich literature that has emerged detailing the way in which possessing or lacking (positive) social capital can help explain the perpetuation or disruption of stratifying processes, it is instructive to consider how the converse of social capital, instead of just the negation or lack of it, can help us to understand the processes of stratification and inequality. In other words, we often study how group membership can improve individual life chances and, by extension, how not being part of the group can hurt an individual's life chances; yet we rarely consider the potential negative ramifications of group membership, that is, how membership in a group can hurt one's life chances.

Building on Portes' development and discussion of the concept (Portes and Sensenbrenner 1993; Portes 1998), I offer a formal definition of negative social capital that is logically consistent with sociologists current use of social capital:

(Positive) Social Capital: The ability to secure benefits by an individual actor, and the positive consequences experienced, by virtue of membership in social networks or other social structures (Portes 1998).

Negative Social Capital: The pressure on an individual actor to incur costs by virtue of membership in social networks or other social structures.

One important distinction made evident in this conception is how (positive) social capital and negative social capital are experienced by the individual actor. Positive social capital comprises largely an “ability” or “potential” of the actor granted by virtue of group membership. The actor can often choose if and how to activate (positive) social capital. Negative social capital, by contrast, is more readily understood as a pressure to incur some sort of negative personal cost as a requirement of membership in the group. Whereas individuals who fail to utilize (positive) social capital do not jeopardize their membership in the group, those who fail to submit to negative social capital may face sanctions and, in some cases, forfeit group membership.

Yet if negative social capital is meant to encompass the potential liabilities of group membership for the individual, does every negative consequence of being black, such as labor market discrimination, qualify as negative social capital? No. Here, it is important to distinguish between a group and a category. Demographic categories such as race, sex and class are classifications of individuals, not groups. Groups are networks that an individual must be able to leave, even if entry was not voluntary (e.g., family), in that they can sever ties and surrender all rights and benefits associated with membership as well as any obligations and responsibilities.3 Only when the negative consequences experienced by the individual are a direct result of the group's structure, norms, expectations or actions can the case be considered negative social capital. Negative consequences imposed on all in-group members by an out-group – for example, discrimination of African Americans by whites in the labor market, or some other external influence – do not qualify as negative social capital.

Portes (1998) identifies four categories of negative consequences of social capital: 1) exclusion of outsiders, 2) excess claims on group members, 3) restrictions on individual freedoms and 4) downward leveling norms. Of these categories, the latter three are directly compatible with a definition of negative social capital that focuses on the pressure felt by an individual to incur costs by virtue of his or her membership in a group. In illustrating how groups make excess claims on individual members, Portes describes the expectation in ethnic networks that those with resources provide assistance to group members, which serves to inhibit individual wealth accumulation and mobility, an example that motivates the current investigation and will be explored in detail below. Further, Portes' description of how tight social networks can enforce conformity and thereby restrict individual freedoms is consistent with a conception of negative social capital as a pressure placed on individuals to incur costs in exchange for group membership. In this instance, the cost is a loss of personal autonomy.

Similarly, in his discussion of downward leveling norms, Portes notes that if group bonds are premised upon a shared (present or historical) experience of adversity or subordination by another group, individuals may feel pressure to submit to continued subordination and avoid engaging in actions that run counter to this group narrative. For example, an individual who achieves success in a labor market that is believed by group members to be unattainable due to discrimination threatens to undermine a source of group solidarity. Here, too, an individual feels pressure to incur personal costs by virtue of his or her membership in the group.

Notably, the first category of negative consequences of social capital outlined by Portes – exclusion of outsiders – is less consistent with the definition of negative social capital provided above. Exclusion of outsiders concerns the inability of outside individuals to enter the group and, in turn, access economic resources controlled by the group. This does not qualify as a distinct category of negative social capital because it is a negative consequence of social capital for individuals outside of the group. Although the requirement to exclude outsiders may be negative for in-group members – for example, by pressuring individuals to abstain from interacting with potential mates or business partners – this may be better understood as an example of how group membership restricts individual freedoms.

Black Kin Networks and Informal Financial Assistance

To illustrate how groups can make excess claims on members, Portes (1998) cites studies of minorities in inner city neighborhoods for whom mutual, informal assistance among kin is commonplace and necessary for survival. A potential negative consequence of this expectation of informal assistance is that it prevents those with financial resources from investing their money for personal advancement; those who experience a windfall or gain steady employment are expected to use their resources to benefit their broader social networks, not to invest in their own wealth and asset accumulation. In this light, the expectation to provide informal financial assistance to friends and family can be experienced as negative social capital for minorities with resources.

The most well-known example of this phenomenon is from Carol Stack's All Our Kin (1974), an ethnographic account of the kin networks of inner-city minorities. Stack describes instances in which a poor black family receives a modest financial windfall, sometimes due to an unexpected inheritance. As word of the family's new resource quickly travels, members of their kin network – friends and relatives – immediately begin making demands on the money. The family has little choice but to surrender their bounty in support of the immediate needs of those close to them. In the process, a lump sum that could have been used for mobility investments such as a vehicle or a down payment on a home is soon whittled away to nothing. Stack characterizes these support structures as systems of reciprocity; McAdoo (1978) goes further, calling these systems a form of “kin insurance” whereby those with means help today with the understanding that they, too, will likely require help in the future (see also Uehara 1990).

Yet what if the expectation of providing financial assistance extended to those members of black kin networks who are not poor? If the expectation of providing informal financial support to disadvantaged friends extends beyond the walls of the black ghetto to the doorsteps of those who have achieved middle-class occupations, then middle-income blacks may be financially hampered relative to whites in their ability to accumulate and invest capital. Given higher rates of poverty and material deprivation among blacks, the expectation that those with means will provide for disadvantaged friends and family may lead to a one-way flow of resources from the middle-income to the poor. This flow of dollars would theoretically reduce the capital available to black middle-income families to invest in wealth building and asset accumulation, thereby contributing to the racial gap in wealth.

Kin Networks and the Race Gap in Wealth

The race gap in wealth is well established (Oliver and Shapiro 2006; Shapiro 2004; Conley 1999). Some estimate that African Americans hold only 16 cents for every dollar in wealth held by whites (Wolff 1998). Traditional explanations for why blacks have less wealth than whites with the same income center on historical patterns of exclusion and isolation. For example, discrimination in the labor and housing markets results in blacks having less wealth to transmit across generations than whites. Further, blacks are less likely to own homes – the most common source of wealth – and those who do are more likely to own homes in areas that have predominately minority residents and so do not enjoy the same appreciation and wealth-building as white homeowners. But what role does providing informal financial assistance to friends and family play in the race wealth gap?

Given historical patterns of exclusion, existing structural inequalities and systematic spatial segregation (Massey and Denton 1993), middle-income blacks are more likely to have friends or family in poverty than middle-income whites. The existing literature clearly documents that blacks have poorer social networks than whites. Heflin and Pattillo (2006) find a significant racial difference in the financial capacity of kin networks – middle-income blacks are more than twice as likely as middle-income whites to have a poor sibling, net of other individual and family characteristics. Analyzing data from the Panel Study of Income Dynamics (PSID), Chiteji and Hamilton (2002) find a higher incidence of poverty among kin of middle-income black families than middle-income white families: 36 percent of parents of middle-income blacks are below the poverty line, compared with only 8 percent of parents of middle-income whites. Siblings of middle-income blacks are four times more likely to be in poverty than siblings of middle-income whites. Further, given that middle-income blacks are also more likely to live closer to low-income neighborhoods than middle-income whites, it is also likely that their nonrelative friend networks are poorer (Massey and Denton 1993; Pattillo 2000). But is there a relationship between network poverty and household wealth?

Heflin and Pattillo (2002) attempt to empirically link race differences in the financial well-being of family and friends with race differences in wealth outcomes such as account ownership and homeownership. Their research finds that having a poor relative helps to explain part of the race differences in account ownership, net of a host of individual and family characteristics. Their work is distinct from previous analyses of the relationship between kin networks and the racial wealth gap in that they move beyond studies of intergenerational transfers (e.g., inheritances and in vivo transfers from parents to children) to study how intragenerational (e.g., transfers to siblings) variation in network capacity affects financial outcomes.

Chiteji and Hamilton (2002) also argue that part of the wealth gap between middle-income whites and middle-income blacks can be explained by the greater need of black parents and siblings. They find that parent and sibling economic status accounts for about eleven percent of the black-white wealth gap even after controlling for sociodemographic indicators such as average lifetime income and family characteristics such as parental wealth and bequests. Combining this effect of network poverty with family background characteristics, the authors conclude that “the middle-class black families in this sample suffered about a 27 percent reduction in their wealth relative to white families as a result of the kin networks into which they were born” (Chiteji and Hamilton 2002:24).

These quantitative studies find a correlation between impoverished kin networks and variation in wealth holdings between middle-income blacks and whites. The theorized mechanism by which race differences in network poverty lead to race differences in wealth is differential patterns of informal financial assistance: middle-income blacks are more likely to provide assistance than their white counterparts. To date there has been no test of this proposition.

Stack's work demonstrates that there exists a powerful expectation of mutual financial assistance among low-income urban minorities. Her work has been followed by other qualitative research documenting a similar phenomenon (see, for example, Dominguez and Watkins 2003). Unfortunately, none of these qualitative accounts include a comparison group of poor whites, so it is impossible to conclude that there is any racial difference in the likelihood of relying on friends and family for financial support. There is, however, a theoretical case to be made that blacks are more likely to activate networks for financial assistance, due to historical exclusion from safety net programs and credit products as well as a persistent lack of attachment to mainstream financial institutions (see Caskey 2005). Further, the history of blacks in America is one of discrimination, in the labor market as well as in public support programs. Some argue that blacks have adapted to these structural constraints through the active mobilization of extensive kin networks that serve as social insurance (see Miller-Cribbs and Farber 2008 for a discussion).

Previous quantitative analyses of race and the provision of informal assistance have yielded conflicting results. In their analysis of patterns of informal giving among siblings, White and Riedmann (1992) find that blacks are less likely to exchange assistance than whites. However, when analysis is restricted to unmarried respondents, Raley (1995) finds that blacks are more likely than whites to engage in financial exchange with kin other than parents. Sarkisian and Gerstel (2004) find that blacks are less likely to provide informal financial assistance than whites, although that study restricts informal financial assistance to amounts of more than $200 given to relatives, biasing results against less wealthy respondents and those who provide assistance to nonrelatives that still may be considered “kin” (Stack 1974). Schoeni (1992), however, finds no racial difference in the likelihood of having provided assistance in his analysis of the PSID. Further, Hogan, Eggebeen and Clogg (1993) note that the high rates of co-residency among African Americans may account for findings that blacks are less likely to participate in informal financial exchange.

Additional studies of race differences in informal financial assistance have largely focused on patterns of assistance from parent to child (see also Jayakody 1998), or on patterns of mutual financial assistance (giving and receiving) among low-income households (e.g., Raley 1995; Jayakody, Chatters and Taylor 1993; Haxton and Harknett 2009). Moreover, each of these studies analyzes race differences controlling for income – none look at how race differences in the provision of financial assistance vary with income. We therefore have no direct quantitative evidence that middle-income blacks are more likely to provide informal financial assistance than middle-income whites.

Qualitative research on social capital among urban minorities suggests that advantaged members of the group may actually be less willing to offer assistance than disadvantaged group members. As Smith (2007) documents, blacks who have achieved advantage through securing a low-wage job are often reluctant to assist members of their network secure employment. Further, Newman (1999) finds that blacks with financial resources are often hesitant to provide assistance to needy friends and relatives. And a closer reading of Stack (1974) finds that while families who receive a financial windfall do ultimately use that resource to assist friends and family, they do so after much initial hesitation and reluctance, acquiescing only once they acknowledge that their windfall is not enough to buffer them from financial shock indefinitely and that they are better off paying in to the “kin insurance” network so that they too can borrow in the future. Those who achieve middle-income status, however, may no longer need this type of “kin insurance” and therefore do not feel obligated to assist disadvantaged friends and family.

Yet a number of qualitative researchers have found the opposite. Building on the work of Dawson (1995), Pattillo argues that many members of the black middle class feel a shared fate with the larger black community and profess a sense of “obligation to help the black poor” (2007:97). In a qualitative study of middle-income blacks, McAdoo (1978) found that fully 92 percent of her sample felt some obligation to help a less fortunate relative. Her study finds that upwardly mobile, middle-income blacks feel a strong desire to provide for their kin, an attitude she argues is “not just a structural coping tactic but has evolved into a strong and valuable cultural pattern” (McAdoo 1978:775). More recently, Shapiro (2004) provides qualitative evidence that black middle-class households feel a strong compulsion to give to their friends and relatives. And they act on it. He tells the story of the Kevin and Donna Hays, African American professionals from Los Angeles (Shapiro 2004:99-100):

“The bank of Kevin,” Donna replies when I ask if they ever help relatives with money or other assistance. “The bank of Kevin. I can't think of anybody who has helped us, but he's always helping people…. Countless times, more times than I can remember. Just crazy amounts of money. Kevin fills in the blanks. Family members still owe me three, four thousand dollars. A thousand dollars. Five hundred. Seven-fifty…. We've had good fortune come to us, just incredibly good fortune, and it really overweighs what we have lost in being good-hearted.”

Shapiro asserts that there exists a marked difference between whites and blacks in providing informal financial assistance to friends and relatives, especially, but not uniquely, among middle-income families. He notes that the whites he interviewed were much less open to providing direct cash assistance to kin than middle-income blacks. He further argues that the race difference in the “flow” of money – where “resources tend to flow from parents to children in white families, while money flows from children to parents and other relatives and friends in black families” (Shapiro 2004:101) – accounts for a part of the racial wealth gap. Blacks who secure middle-class occupations and earn higher wages are less able to translate this income into productive, wealth-building assets than whites, as they are more likely to be asked to use that income to support friends and extended family.

Yet finding that middle-income blacks are more likely than whites to provide assistance and, further, that this difference in giving can explain part of the racial gap in wealth does not make the act of giving negative. Labeling black kin networks as negative social capital for middle-income blacks in one domain – in this instance, wealth stratification – does not negate the numerous potential benefits members yield by being part of the group. These potential benefits of membership in kin networks – including material benefits resulting from in-kind transfers and unquantifiable benefits such as emotional support, solidarity and identity – likely more than compensate for the financial sacrifice of providing assistance. Nevertheless, it is important to consider how expectations placed on group members – in this case middle-income blacks embedded in kin networks – can have negative stratifying consequences. For this reason, the empirical analysis that follows is necessarily limited to the transfer of financial assistance and its role in perpetuating the racial gap in wealth.

Data and Analytic Approach

Data for this analysis are taken from two nationally representative household surveys: the Survey of Consumer Finances (SCF) and the PSID. The SCF is a cross-sectional survey of household finances sponsored by the Federal Reserve Board of Governors and administered every 3 years by the National Opinion Research Center at the University of Chicago. The SCF survey is designed to be nationally representative and includes an oversample of wealthy families. Data for this analysis are taken from the 2007 survey. The PSID is a longitudinal survey of individuals and the family households in which they reside. The survey is administered every 2 years by the University of Michigan's Survey Research Center. The PSID is designed to be nationally representative. Data for this analysis are taken from the 2005 and 2007 main family data.

To test whether race differences in the provision of informal financial assistance vary by income, two logistic models are estimated, one using the 2007 PSID sample and the other using the 2007 SCF. All analyses are limited to a restricted sample of (non-Hispanic) black and (non-Hispanic) white families. The dependent variable for each dataset is a binary indicator for whether a family provided informal financial assistance in the last year. This construct is operationalized differently in each survey. In the PSID, respondents were asked: “In 2006, did you give any money toward the support of anyone who was not living with you at the time, including child support, alimony, money given to parents, and things like that?” Follow-up questions asked the respondent about their relationship to each person who received assistance (up to five). Respondents were also asked if any of the instances mentioned included child support or alimony payments. Respondents were coded as having provided informal financial assistance if they indicated none of the payments made in the last year were to fulfill child support or alimony obligations or if they indicated providing assistance to someone other than a child or spouse/ex-spouse.

In the SCF, the question on informal financial support comes after a series of questions about whether the respondent, or anyone in the family, provided child support or alimony payments: “During 2006, did you (or anyone in your family living here) provide any (other) financial support for relatives or friends who do not live here? Please do not include alimony or child support.” Respondents who answered yes to this question were coded as having provided informal financial assistance.

Models 1 and 2 are logistic regression models predicting whether or not the family provided informal financial assistance. The key independent variables are family income, a binary indicator for whether the respondent is black, and an interaction between the two.4 As family income is positively skewed in both samples, a log transformation is used and the variable is mean centered to reduce multicollinearity between the interaction and main effect. Both models employ a number of control variables, including a binary indicator for the sex of the family head, a series of dummies for marital status, continuous measures of education and family size and a linear and squared term for age.5,6 Wealth is an aggregate measure equal to the sum of all financial assets less the cost of consumer debt. This variable includes the value of all real estate, stocks, bonds, mutual funds, retirement and pension accounts, checking, savings and other transaction accounts (e.g., money market accounts) minus any debts such as credit cards and outstanding mortgage balances.7

The longitudinal nature of the PSID study enables a test of whether the provision of informal financial assistance can explain race differences in wealth. A lagged model is used for this analysis (Models 3 and 4), where the dependent variable is family wealth in 2007 and family wealth in 2005 is included as a ­control. The independent variables can therefore be interpreted as the association between X and a change in family wealth between 2005 and 2007. Including a lagged dependent variable in the model controls for potential sources of omitted variable bias that are associated with wealth in 2005 and do not change over time.

Model 3 predicts change in family wealth with a measure of how much assistance was provided in 2004, collected in the 2005 survey year. Using data from 2005 ensures that the measure of how much assistance the family provided is temporally prior to the measure of family wealth in 2007. This model includes the same controls as those used in the previous analysis of the PSID using data from the 2005 survey year and an additional squared term for family income in 2005 as there is a likely nonlinear return to wealth by income, that is, higher income families are likely to experience higher increases in wealth than lower income families. Model 4 is the same model with the addition of an interaction term between black and the amount of assistance the family provided in 2004. Coefficients are estimated using ordinary least squares (OLS) regression. Robust standard errors are used to account for heteroskedasticity.

Model specific multiple imputation is used to deal with missing data in both samples.8,9 Coefficients and standard errors presented are therefore estimated using the multiple imputation inference program in Stata.

Results

Descriptive statistics for key dependent and independent variables from both the 2007 PSID and 2007 SCF samples are presented in Table 1. Key demographics are similar across the two datasets with black family heads having lower levels of education and income and a higher likelihood of being female and never married than their white counterparts. The difference in wealth holdings by race is dramatic in both the SCF and the PSID samples, reflecting the well-known racial gap in wealth. Averaging across all families, the proportion of black and white families reporting having provided informal financial assistance in the past year appears to be about equal in the PSID, whereas blacks are significantly more likely than whites to report having provided assistance in the SCF.10 Race differences in the amount of assistance provided are not statistically significant in either sample.

Table 1:

Means of Selected Variables in 2007 PSID and 2007 SCF

 PSID (2007)
 
SCF (2007)
 
 Full Sample White Black Full Sample White Black 
Percent black .154 (.006) – – .145 (.007) – – 
Percent white .846 (.006) – – .854 (.007) – – 
Male head .694 (.008) .734 (.008)* .479 (.019) .712 (.009) .741 (.009)* .535 (.026) 
Household size 2.230 (.019) 2.216 (.021) 2.308 (.053) 2.4730 (.026) 2.245 (.028) 2.602 (.077) 
Age 52.303 (.280) 53.024 (.308)* 47.696 (.621) 51.216 (.353) 52.152 (.383)* 45.733 (.876) 
Education 
 Less than HS .149 (.006) .129 (.006)* .256 (.017) .128 (.007) .122 (.007)* .165 (.019) 
 High school .325 (.008) .318 (.008)* .365 (.019) .321 (.009) .318 (.010) .334 (.025) 
 Some college .240 (.007) .236 (.007) .264 (.018) .239 (.008) .230 (.009)* .295 (.024) 
 College .171 (.006) .189 (.007)* .074 (.009) .221 (.008) .237 (.009)* .129 (.018) 
 Graduate/professional .115 (.005) .129 (.006)* .040 (.008) .116 (.006) .121 (.006) .085 (.016) 
Marital status 
 Married .494 (.008) .541 (.009)* .234 (.015) .502 (.010) .534 (.010)* .314 (.024) 
 Divorced .197 (.007) .185 (.007)* .267 (.017) .210 (.008) .206 (.008) .233 (.022) 
 Widowed .115 (.006) .118 (.006) .100 (.012) .107 (.006) .108 (.007) .102 (.018) 
 Never been married .194 (.007) .156 (.007)* .398 (.019) .181 (.007) .152 (.008)* .351 (.024) 
Total family income (in thousands) 72.70 (1.45) 78.51 (1.69)* 40.83 (1.37) 84.48 (1.90) 91.32 (2.19)* 44.40 (2.31) 
Total family wealth (in thousands) 408.34 (29.11) 467.08 (3.43)* 85.91 (11.35) 288.94 (9.28) 327.57 (10.74)* 64.01 (6.27) 
Provided financial assistance .105 (.005) .105 (.005) .101 (.013) .155 (.007) .149 (.007)* .192 (.021) 
Amount given (including zeros) 1808.28 (916.38) 2077.56 (1082.14) 320.63 (78.01) 1162.45 (113.40) 1161.06 (118.82) 1543.11 (557.96) 
Amount given (nonzeros only) 18855.03 (9527.16) 21599.09 (11206.24) 3341.83 (621.83) 8992.84 (902.62) 8954.92 (861.19) 9165.39 (3141.86) 
 PSID (2007)
 
SCF (2007)
 
 Full Sample White Black Full Sample White Black 
Percent black .154 (.006) – – .145 (.007) – – 
Percent white .846 (.006) – – .854 (.007) – – 
Male head .694 (.008) .734 (.008)* .479 (.019) .712 (.009) .741 (.009)* .535 (.026) 
Household size 2.230 (.019) 2.216 (.021) 2.308 (.053) 2.4730 (.026) 2.245 (.028) 2.602 (.077) 
Age 52.303 (.280) 53.024 (.308)* 47.696 (.621) 51.216 (.353) 52.152 (.383)* 45.733 (.876) 
Education 
 Less than HS .149 (.006) .129 (.006)* .256 (.017) .128 (.007) .122 (.007)* .165 (.019) 
 High school .325 (.008) .318 (.008)* .365 (.019) .321 (.009) .318 (.010) .334 (.025) 
 Some college .240 (.007) .236 (.007) .264 (.018) .239 (.008) .230 (.009)* .295 (.024) 
 College .171 (.006) .189 (.007)* .074 (.009) .221 (.008) .237 (.009)* .129 (.018) 
 Graduate/professional .115 (.005) .129 (.006)* .040 (.008) .116 (.006) .121 (.006) .085 (.016) 
Marital status 
 Married .494 (.008) .541 (.009)* .234 (.015) .502 (.010) .534 (.010)* .314 (.024) 
 Divorced .197 (.007) .185 (.007)* .267 (.017) .210 (.008) .206 (.008) .233 (.022) 
 Widowed .115 (.006) .118 (.006) .100 (.012) .107 (.006) .108 (.007) .102 (.018) 
 Never been married .194 (.007) .156 (.007)* .398 (.019) .181 (.007) .152 (.008)* .351 (.024) 
Total family income (in thousands) 72.70 (1.45) 78.51 (1.69)* 40.83 (1.37) 84.48 (1.90) 91.32 (2.19)* 44.40 (2.31) 
Total family wealth (in thousands) 408.34 (29.11) 467.08 (3.43)* 85.91 (11.35) 288.94 (9.28) 327.57 (10.74)* 64.01 (6.27) 
Provided financial assistance .105 (.005) .105 (.005) .101 (.013) .155 (.007) .149 (.007)* .192 (.021) 
Amount given (including zeros) 1808.28 (916.38) 2077.56 (1082.14) 320.63 (78.01) 1162.45 (113.40) 1161.06 (118.82) 1543.11 (557.96) 
Amount given (nonzeros only) 18855.03 (9527.16) 21599.09 (11206.24) 3341.83 (621.83) 8992.84 (902.62) 8954.92 (861.19) 9165.39 (3141.86) 

Notes: *Indicates difference is statistically significant at p<.05 level. Linearized standard errors in parentheses; full sample includes blacks and whites. Population weights used. PSID=Panel Study of Income Dynamics; SCF=Survey of Consumer Finances.

Focusing on mean rates of giving across all families, however, obscures a real racial difference in rates of giving at higher levels of income. Figures 1 and 2 present descriptive statistics on the provision of informal financial assistance by race and income from the PSID and SCF, respectively. The trend is striking – whereas low-income blacks and whites provide informal financial assistance at comparable rates, higher income blacks appear to be much more likely than whites to report having provided assistance in the last two years. The overall rates of giving and the differences between blacks and whites by income are very similar across these two distinct datasets, providing strong evidence of a real difference in rates of giving between middle-income blacks and whites.

Figure 1.

Provision of Financial Assistance (PSID) (family income in thousands of dollars)

Figure 1.

Provision of Financial Assistance (PSID) (family income in thousands of dollars)

Figure 2.

Provision of Financial Assistance (SCF) (family income in thousands of dollars)

Figure 2.

Provision of Financial Assistance (SCF) (family income in thousands of dollars)

Next, we turn to the multivariate regression analysis. The results of logistic models predicting the provision of informal financial assistance by race and income for each of the samples are presented in Table 2. The model demonstrates that income is positively associated with providing assistance, that is, across all families, increasing income is associated with increased odds of having provided informal financial assistance. Note, however, the positive interaction term between black and family income. This indicates that the odds of having provided informal financial assistance in the last year differ by income depending on whether the respondent is black or white. Similar model specifications across the PSID and SCF reveal that as income increases, blacks are increasingly more likely to have provided assistance in the past year relative to whites, a relationship that is highly ­significant in both the PSID and the SCF samples. As the family income variable is mean centered, the main effect for black can be interpreted as the difference between blacks and whites for those with average income. The models indicate that for those with average income, blacks are more likely to provide assistance than whites.

Table 2:

Logistic Regression of Provision of Informal Financial Assistance

 Model 1: PSID Model 2: SCF 
 Log Odds (β) Log Odds (β) 
Family income (Log) .424*** (.078) .201*** (.043) 
Family income X black .524*** (.119) .272* (.135) 
Black .251* (.108) .596*** (.164) 
Education .050* (.022) .040* (.020) 
Married −.086 (.165) −.041 (.176) 
Divorced .294 (.155) .127 (.174) 
Widowed .211 (.237) −.158 (.237) 
Male .117 (.133) .157 (.145) 
Household size −.179*** (.041) −.085* (.039) 
Total wealth (Log) .409*** (.099) .279** (.089) 
Age .004 (.018) .068*** (.019) 
Age-squared (× .01) −.003 (.017) −.049** (.017) 
Constant −8.342*** (1.385) −8.063*** (1.295) 
6596 3911 
Pseudo R-squared .064 .074 
 Model 1: PSID Model 2: SCF 
 Log Odds (β) Log Odds (β) 
Family income (Log) .424*** (.078) .201*** (.043) 
Family income X black .524*** (.119) .272* (.135) 
Black .251* (.108) .596*** (.164) 
Education .050* (.022) .040* (.020) 
Married −.086 (.165) −.041 (.176) 
Divorced .294 (.155) .127 (.174) 
Widowed .211 (.237) −.158 (.237) 
Male .117 (.133) .157 (.145) 
Household size −.179*** (.041) −.085* (.039) 
Total wealth (Log) .409*** (.099) .279** (.089) 
Age .004 (.018) .068*** (.019) 
Age-squared (× .01) −.003 (.017) −.049** (.017) 
Constant −8.342*** (1.385) −8.063*** (1.295) 
6596 3911 
Pseudo R-squared .064 .074 

Notes: ***p<.001 **p<.01 *p<.05

Standard errors in parentheses. Two-tailed t-tests. Reported pseudo R-squared is average across imputations. PSID=Panel Study of Income Dynamics; SCF=Survey of Consumer Finances.

Table 3:

Lagged OLS Regression Predicting 2007 Family Wealth (PSID)

 Model 3 Model 4 
Total wealth 2005 (Log) .708*** (.010) .706*** (.010) 
Total amount given 2004 (log) .003* (.001) .005** (.002) 
Total amount given 2004 X black  −.007* (.003) 
Black −.029*** (.008) −.024** (.008) 
Family income 2005 (log) .007 (.004) .007 (.004) 
Family income 2005 (log)-squared .002*** (.0002) .002*** (.0002) 
Education 2005 .005** (.002) .005** (.002) 
Married 2005 −.020 (.011) −.020 (.011) 
Divorced 2005 −.047*** (.011) −.047*** (.011) 
Widowed 2005 −.065*** (.017) −.064*** (.017) 
Male 2005 .003 (.010) .003 (.010) 
Household size 2005 −.005* (.003) −.005* (.003) 
Age 2005 .002 (.001) .003*** (.0003) 
Constant 3.433*** (.130) 3.456*** (.130) 
6522 6522 
R-squared .576 .577 
 Model 3 Model 4 
Total wealth 2005 (Log) .708*** (.010) .706*** (.010) 
Total amount given 2004 (log) .003* (.001) .005** (.002) 
Total amount given 2004 X black  −.007* (.003) 
Black −.029*** (.008) −.024** (.008) 
Family income 2005 (log) .007 (.004) .007 (.004) 
Family income 2005 (log)-squared .002*** (.0002) .002*** (.0002) 
Education 2005 .005** (.002) .005** (.002) 
Married 2005 −.020 (.011) −.020 (.011) 
Divorced 2005 −.047*** (.011) −.047*** (.011) 
Widowed 2005 −.065*** (.017) −.064*** (.017) 
Male 2005 .003 (.010) .003 (.010) 
Household size 2005 −.005* (.003) −.005* (.003) 
Age 2005 .002 (.001) .003*** (.0003) 
Constant 3.433*** (.130) 3.456*** (.130) 
6522 6522 
R-squared .576 .577 

Notes: ***p<.001 **p<.01 *p<.05

Robust standard errors in parentheses. Two-tailed t-tests. Reported R-squared is average across imputations. PSID=Panel Study of Income Dynamics; OLS=Ordinary Least Squares.

To better understand the net association of the interaction and main effects, Figures 3 and 4 graphically present the predicted probability of having provided informal financial assistance by race and income for each sample. The predicted probability of having provided assistance is calculated allowing income and race to vary while holding all other variables at their means. This display illustrates a clear difference between blacks and whites in the relationship between family income and the likelihood of having provided financial assistance: the predicted probability of having provided assistance increases linearly with income for blacks whereas the predicted probability of giving remains relatively flat for whites across income. The relationship is remarkably similar across both datasets, providing strong evidence that there is a real difference in the provision of financial assistance by race: as income increases, blacks are increasingly more likely than whites to report having provided financial assistance to friends and family. Middle-income blacks are significantly more likely than middle-income whites to report having provided informal financial assistance in the past year.11

Figure 3.

Probability of Providing Assistance (PSID)

Figure 3.

Probability of Providing Assistance (PSID)

Figure 4.

Probability of Providing Assistance (SCF)

Figure 4.

Probability of Providing Assistance (SCF)

But how does this difference in giving by race and income influence patterns of wealth accumulation? Model 3 tests for this association by using a lagged ordinary least squares (OLS) model to analyze the PSID. In this model, family wealth in 2007 is predicted using family-level covariates from 2005 including family wealth, income, ­education, marital status and a measure of how much assistance the family provided in 2004. Using 2005 wealth to predict 2007 family wealth holdings enables us to interpret coefficients as representing an association with the change in family wealth from 2005 to 2007. The coefficient on the assistance term is positive, indicating that across all families, the amount of money given is positively associated with an increase in family wealth. Yet does this positive association hold for both blacks and whites?

Model 4 explores this question by including an interaction between the amount of assistance provided and whether the family is black. The coefficient on this interaction is negative and significant, indicating a real difference between blacks and whites in the effect of giving on the change in wealth.12 Although the magnitude of the coefficient on the interaction is greater than the main effect for amount given, the net negative wealth effect of giving for blacks is not statistically different from zero. Therefore, this model suggests that the provision of informal financial assistance is associated with an increase in wealth for whites and not for blacks, thereby contributing to the racial gap in wealth, at least among those who provide informal financial assistance.

Discussion and Conclusions

Previous quantitative studies demonstrate that blacks have more economically disadvantaged social networks than whites and that this network disadvantage is associated with race differences in wealth, net of lifetime income, parental wealth and a host of other factors. Existing qualitative research suggests that an explanation for this association is informal financial assistance: low-income blacks are more likely to ask friends and family for assistance than low-income whites and middle-income blacks are more likely than their white counterparts to provide financial assistance to friends and family. This article presents the first attempt to quantitatively test this argument. Analysis of data from two ­nationally representative surveys reveals that middle-income blacks are more likely than middle-income whites to report having provided assistance to friends and family in the past year. Further, the above-mentioned analysis provides compelling evidence for the proposition suggested by Heflin and Pattillo (2002, 2006), Shapiro (2004), Portes (1998) and others that black/white differences in the provision of informal financial assistance to friends and families contributes to the race gap in wealth. Whereas giving appears to be associated with a positive wealth effect for whites, there is no such positive effect for blacks. This study therefore provides concrete evidence for how having impoverished social networks can account for part of the race gap in wealth, illustrating a mechanism absent in previous work connecting impoverished kin networks with reduced wealth holdings.

The findings should not be interpreted to mean that participation in kin networks and providing financial assistance offers to no benefits to middle-income blacks. Other benefits of group membership, such as in-kind transfers and emotional support, may well outweigh the costs associated with financial transfers. Yet it is instructive to consider the price individual actors pay to maintain membership in groups, in this case literally in dollars but in other instances in emotion, time, behavior or status. It may be the case that blacks who have achieved middle-income status have the stability or security to provide support to their poorer kin, or it may be that they are more willing to provide financial assistance than other types of support, such as a job referral, since a financial transaction involves no third party and therefore no potential of reputational harm, a fear expressed by informants in Smith's (2007) study. How individuals negotiate and react to the expectations and obligations placed on them by group members is a potentially fruitful and important avenue of research, particularly in instances where fulfillment of obligations to the group has stratifying consequences for the individual and, of course, by extension the group itself in the aggregate.

Sociology has a long tradition of theorizing and analyzing the positive benefits of group membership for individual actors. Indeed, in the study of social pathologies, sociologists favor explanations that emphasize an individual's lack of embeddedness in a social network or social structure over individual neuroses (Durkheim [1897] 1979). Therefore, it would appear the prescription for society's ills is to ensure every individual is rooted in groups and networks.

But group membership comes with a series of expectations and obligations that may also have negative consequences for the individual actor. And enduring these negative consequences is likely required of those who wish to be in the group or to enjoy the full range of benefits associated with membership. On balance, it is likely that in many circumstances the benefits of group membership – everything from a sense of belonging to the more instrumental benefits of group membership often described as (positive) social capital – outweigh the costs. When the reverse is true, actors would be likely to leave the group. But the ties that bind some groups – in this case, kin networks – may be so strong and the benefits of membership so fundamental that the negative consequences for individuals must be quite severe before severing ties with the group is even considered.

Social science research must move beyond analyses of in-group/out-group exclusion and conflict to more systematically examine the constraints and expectations group membership places on individual actors. These processes have consequences for individual behavior that in turn have implications for the stratification of social groups.

Notes

1
In the current investigation, the terms “middle-income” and “upper-income” are used principally as a heuristic tool; the analysis presented in this study purposefully employs a continuous measure of income. For the sake of brevity, hereafter the term middle-income is used, although the arguments and empirical analysis presented are in no way constrained to those who fall under a certain income level.
2
In his work on collective efficacy, Sampson, too, stresses the neutrality of social capital: “resources or networks alone (e.g., voluntary associations, friendship ties, organizational density) are neutral – they may or may not be effective mechanism for achieving intended effect” (Sampson, Morenoff and Earls 1999:635).
3
Brubaker (2003) makes a similar argument.
4
Using interaction terms to assess group differences may be problematic in logistic regression models due to potential unequal residual dispersion for the two groups (Williams 2009). To test for this possibility, heterogenous choice logit and probit models were estimated using oglm and hetprob commands in Stata with option “hetero” specifying black. These models found no evidence of differential dispersion for the black and white groups. Therefore, the correction for unequal residual dispersion is unnecessary and results reported are for standard logistic regression models. As an additional check, predictors of residual dispersion were also identified using the stepwise approach outlined by Williams. In the PSID sample, age, education, black and male were identified as being significant predictors of residual dispersion. Accounting for these four variables in the heterogenous choice model did not substantively change the results of the analysis; the interaction between black and family income remained positive and significant at the p<.05 level.
5
To generate comparable models across the PSID and SCF, a number of theoretically relevant predictor variables available in the PSID have been omitted. In separate analysis not presented here, models using the PSID additionally control for whether either respondent's or their spouse's parents are poor, whether parents are living and whether siblings are living. In addition, these models control for metropolitan status and include state fixed effects. Inclusion of this fuller list of variables did not change the substantive magnitude of results presented here.
6
The squared term for income was not statistically significant in Models 1 and 2 and was therefore removed from the analysis. A squared term for income is included in Models 3 and 4. In addition to total family income, separate analysis used an adjusted measure by diving income by the square root of household size. Use of this measure produced the same substantive results in all four models.
7
For the PSID, the summary net worth variable is used; for the SCF, the wealth variable is constructed by the author to mirror the information captured in the PSID summary measure.
8
Analysis of the PSID and SCF samples using listwise deletion yields the same substantive results.
9
Model-specific imputation is used for all analyses of the PSID and SCF. Five imputations are constructed. Dependent and independent variables are used to construct the imputations, but imputed values for dependent variables are dropped before conducting analysis. Sensitivity analyses demonstrate results are robust to using 10, 20 and 100 sets for imputation.
10
The higher rates of overall giving in the SCF is likely an artifact of question wording as the SCF stressed reporting whether the household head or anyone in the household provided assistance whereas the PSID asked only about the behavior of the household head.
11
Alternative model specifications included an interaction between black and family wealth. The coefficient for this interaction does not achieve statistical significance in either sample, although it is in the predicted direction (positive).
12
Alternative model specifications included a three-way interaction between black, family income in 2005 and total amount given in 2004. The coefficient for this interaction does not achieve statistical significance.

References

Bourdieu
P.
1986
.
‘The Forms of Capital
.’ Pp.
241
58
in
Handbook of theory and research for the sociology of education
 , edited by
Richardson
John G
.
New York, NY
:
Greenwood Press
Bourdieu
P.
Wacquant
L. P. D
.
1992
.
An Invitation to Reflexive sociology
 .
Chicago, IL
:
University of Chicago Press
.
Brooks
David
.
2010
.
The Lean Years
.
Brubaker
R.
2003
. “
Ethnicity without groups
.”
European Journal of Sociology
 
43
:
163
189
.
Caskey
J. P.
2005
. “
Reaching out to the unbanked
.” Pp.
149
167
in
Inclusion in the American Dream: Assets, Poverty and Public Policy
 , edited by
Sherraden
Michael
.
New York, NY
:
Oxford University Press
Chiteji
N.S
Hamilton
D
.
2002
. “
Family connections and the black-white wealth gap among middle-class families
.”
The Review of Black Political Economy
 
30
:
9
28
.
Coleman
J. S
1988
. “
Social capital in the creation of human capital
.”
American journal of sociology
 
94
:
95
121
.
Coleman
J. S
1990
.
Foundations of social theory
 .
Cambridge, MA
:
Harvard University Press
.
Conley
D.
1999
.
Being black, living in the red: Race, wealth, and social policy in America
 .
Berkeley, CA
:
University of California Press
.
Dawson
M. C
1995
.
Behind the mule: Race and class in African-American politics
 .
Princeton, NJ
:
Princeton University Press
.
Durkheim
E.
1979
. “
Suicide: a study in sociology [1897]
.”
Trans. John A. Spaulding and George Simpson
 . Ed.
George
Simpson
.
New York, NY
:
The Free Press
.
Dominguez
S.
Watkins
C
.
2003
. “
Creating networks for survival and mobility: Social capital among African-American and Latin-American low-income mothers
.”
Social Problems
 
50
:
111
35
.
Fernandez-Kelly
M. P
.
1995
. “
Social and cultural capital in the urban ghetto: Implications for the economic sociology of immigration
.” Pp.
213
47
in
The economic sociology of immigration,
  edited by
Portes
Alejandro
.
New York, NY
:
Russell Sage Foundation
Glaeser
E. L
Laibson
D.
Sacerdote
B
.
2002
. “
An economic approach to social capital
.”
The Economic Journal
 
112
:
437
58
.
Haxton
C. L
Harknett
K
.
2009
. “
Racial and Gender Differences in Kin Support: A Mixed-Methods Study of African American and Hispanic Couples
.”
Journal of Family Issues
 
30
:
1019
40
.
Heflin
C. M
Pattillo
M
.
2002
. “
Kin effects on black-white account and home ownership
.”
Sociological inquiry
 
72
:
220
39
.
Heflin
C. M
Pattillo
M
.
2006
. “
Poverty in the family: Race, siblings, and socioeconomic heterogeneity
.”
Social Science Research
 
35
:
804
22
.
Hogan
D.
Eggebeen
D.
Clogg
C
.
1993
. “
The structure of intergenerational exchanges in American families
.”
American Journal of Sociology
 
98
:
1428
58
.
Jayakody
R.
1998
. “
Race differences in intergenerational financial assistance: The needs of children and the resources of parents
.”
Journal of Family Issues
 
19
:
508
.
Jayakody
R.
Chatters
L. M
Taylor
R. J
.
1993
. “
Family support to single and married African American mothers: The provision of financial, emotional, and child care assistance
.”
Journal of Marriage and the Family
 
55
:
261
76
.
Massey
D. S.
Denton
N. A
.
1993
.
American Apartheid: Segregation and the Making of the Underclass
 .
Cambridge, MA
:
Harvard University Press
.
McAdoo
Harriette P.
1978
. “
Factors Related to Stability in Upwardly Mobile Black Families
Journal of Marriage and Family
 .
40
:
761
76
.
McAdoo
H. P.
1980
. “
Black mothers and the extended family support network
.” Pp.
125
44
in
The black woman,
  edited by
Rodgers-Rose
L
.
Beverly Hills, CA
:
Sage
Miller-Cribbs
J. E
Farber
N. B
.
2008
. “
Kin Networks and Poverty among African Americans: Past and Present
.”
Social Work
 
53
:
43
51
.
Center National Poverty
,
University of Michigan.
2010
.
Facts About Poverty. Available at: http://www.npc.umich.edu/poverty
Newman
K. S.
1999
.
No shame in my game: The working poor in the inner city
 .
New York, NY
:
Russell Sage Foundation
.
Oliver
M. L
Shapiro
T. M
.
2006
.
Black wealth, white wealth: A new perspective on racial inequality
 .
New York, NY
:
Routledge
.
Paldam
M.
Svendsen
G. T
.
2000
.
Missing social capital and the transition in Eastern Europe
 .
The Aarhus School of Business, Department of Economics, Faculty of Business Administration
.
Pattillo-McCoy
M.
2000
.
Black picket fences: Privilege and peril among the Black middle-income
 .
Chicago, IL
:
University of Chicago Press
.
Pattillo
M. E
.
2007
.
Black on the block: The politics of race and class in the city
 .
Chicago, IL
:
University of Chicago Press
.
Portes
A.
1998
. “
Social capital: its origins and applications in modern sociology
.”
Annual review of sociology
 
24
:
1
24
.
Portes
A.
Sensenbrenner
J
.
1993
. “
Embeddedness and immigration: Notes on the social determinants of economic action
.”
American journal of sociology
 
98
:
1320
50
.
Putnam
R. D.
1995
. “
Bowling alone: America's declining social capital
.”
Journal of Democracy
 
6
(1)
:
65
78
.
Putnam
R. D.
2000
.
Bowling alone: The collapse and revival of American community
 .
New York, NY
:
Simon & Schuster
.
Putnam
R. D.
2007
. “
E pluribus unum: Diversity and community in the twenty-first century the 2006 Johan Skytte Prize Lecture
.”
Scandinavian Political Studies
 
30
:
137
174
.
Raley
R. K.
1995
. “
Black-white differences in kin contact and exchange among never married adults
.”
Journal of Family Issues
 
16
:
77
.
Royston
P
.
2004
.
Multiple imputation of missing values. Stata Journal
4
(3)
:
227
41
.
Sampson
Robert J
Morenoff
Jeffrey D
Earls
Felton
.
1999
. “
Beyond Social Capital: Spatial dynamics of collective efficiency for children
.”
American Journal of Sociological Review
 
64
:
633
60
.
Sarkisian
N.
Gerstel
N
.
2004
. “
Kin support among Blacks and Whites: Race and family organization
.”
American Sociological Review
 
69
:
812
37
.
Schoeni
R. F
.
1992
.
Another Leak in the Bucket?: Public Transfer Income and Private Family Support
 .
University of Michigan, Population Studies Center
.
Shapiro
T. M.
2004
.
The hidden cost of being African American: How wealth perpetuates inequality
 .
New York, NY
:
Oxford University Press
.
Smith
S. S.
2007
.
Lone Pursuit: Distrust and Defensive Individualism Among the Black Poor
 .
New York, NY
:
Russell Sage Foundation
.
Stack
C. B.
1974
.
All our kin: Strategies for survival in Black communities
 .
New York, NY
:
Harper & Row Publishers
.
Uehara
E.
1990
. “
Dual exchange theory, social networks, and informal social support
.”
The American Journal of Sociology
 
96
:
521
57
.
Wacquant
L. J.D
.
1998
. “
Negative social capital: State breakdown and social destitution in America's urban core
.”
Journal of Housing and the Built Environment
 
13
:
25
40
.
White
Lynn K.
Riedmann
Agnes
.
1992
. “
Ties among Adult Siblings.” Social Forces
71
:
85
102
.
Williams
Richard
.
2009
. “
Using heterogeneous choice models to compare logit and probit coefficients across groups.” Sociological Methods and Research
37
:
531
59
.
Wolff
Edward N.
1998
. “
Recent Trends in the Size Distribution of Household Wealth
.”
Journal of Economic Perspectives
 
12
:
131
50
.

Author notes

The author would like to acknowledge Sara McLanahan, David Grusky, Martin Ruef, Katherine Newman, David Pedulla, Sarah Brayne, and participants in the Princeton Empirical Seminar, the Harvard-Manchester Summer Workshop on Inequality, and the Princeton Advanced Workshop in Social Policy for their thoughtful comments and the National Science Foundation for financial support. Direct correspondence to Rourke L. O'Brien, 107 Wallace Hall, Princeton, NJ 08544, USA.