Giving Unto Others: Private Financial Transfers and Hardship Among Families with Children

Abstract

Prior research shows that financial assistance from family and friends is an important source of support for families with children. However, research on financial transfers has largely focused on the recipients of transfers. In this study, using longitudinal data from the Fragile Families and Child Wellbeing Study (n~16,000 person-waves), we examine the association between the provision of financial assistance to family and friends and material hardship. Results from pooled regression and fixed effects models indicate that providing financial transfers is associated with an increased risk of hardship. The most economically disadvantaged groups, single mothers, those in the bottom income tertile, and black mothers, are the most likely to experience hardship after giving a transfer. These findings have important implications for understanding why families may have difficulty meeting basic and essential needs, and how social networks may exacerbate the challenges of escaping poverty and establishing economic self-sufficiency.

Despite substantial progress over the past 50 years in the fight against poverty (Fox et al., 2015; Wimer et al., 2013), poverty rates in the United States remain troublingly high (DeNavas-Walt & Proctor, 2015; Short, 2015), especially among families with children. Many families experience material hardship, or an inability to meet basic or essential needs such as purchasing food or housing (Nelson, 2011; Short, 2005). Material hardships are common (Neckerman, Garfinkel, Teitler, Waldfogel & Wimer, 2016), measure real deprivations (Federman et al. 1996), and have been linked with outcomes such as depression, poor health, and child behavior problems (Gershoff, Aber, Raver & Lennon, 2007; Heflin & Iceland, 2009; Yoo, Slack & Holl). Understanding the roots of families’ inability to meet basic and essential needs is critical for understanding how to further reduce poverty and hardship.

One understudied potential contributor to material hardships suffered by families involves their networks of kin and non-kin. Long lines of research support the notion that these networks are critical for allowing low-income families to survive and get by in the face of chronic shortages of resources, especially when public safety nets may be inadequate or declining (Edin & Lein, 1997; Halpern-Meekin, Edin, Tach & Sykes, 2015; Seefeldt & Sandstrom, 2015; Stack, 1974). Families may support each other strategically knowing that support networks often operate reciprocally such that support given today can be expected to yield potential sources of support given back tomorrow (Offer, 2012). Similarly, families may support each other due to shared norms or notions of kin, without the expectation of reciprocity (e.g. an incarcerated family member; Braman, 2007). Research on kin networks and financial transfers in kin and non-kin networks has largely focused on the recipients of support (e.g. Couch, Daly & Wolf, 1999; Fingerman, Miller, Birditt & Zarit, 2009; Hurd, Smith & Zissimopoulos, 2011). We know little about the provision of financial support to others, and even less about how providing support may affect the material well-being of the givers of transfers.

Using longitudinal data from the Fragile Families and Child Wellbeing Study (FFCWS), this article investigates whether the provision of private financial transfers is related to an increased risk of the experience of material hardship. Although we cannot examine why people provide transfers, these data are especially useful as they provide us with longitudinal information on both the provision of private financial transfers and material hardship. These data also provide us with a large, representative sample of low-income families in large urban areas. Low-income families’ networks may be in greater need of financial support than those of higher income families, especially when public safety nets are unavailable or insufficient. We focus on families with young children, as the experience of material hardship may be particularly detrimental to children’s development (Gershoff et al., 2007; Heflin, London & Scott, 2011; Zilanawala & Pilkauskas, 2012). Specifically, we examine the following questions: 1) Is providing private financial transfers linked with material hardship among families with young children? 2) Do the associations vary by type of material hardship (food, housing, bill, utility or medical)? And 3) are there differences in the associations by household income level, race/ethnicity or relationship status? To the extent that provision of financial support is harmful to the families that give, and if lower-income or more financially precarious populations are most at risk, this research will help us understand low-income families’ ability to establish self-sufficiency.

Background

Private Financial Transfers and Material Hardship

Why might the provision of private financial transfers (PFTs) be linked to material hardships? Both altruism theory and reciprocal exchange theory suggest that families might provide transfers even if it is detrimental to their own wellbeing. Altruism theory posits that family members provide financial transfers to aid kin because of intrinsic or normative values around supporting kin (Becker, 1974). Thus, concern for one’s own kin may lead families to provide PFTs even when it may increase their own risk of experiencing material hardship. Reciprocal exchange theory suggests that private financial transfer provision functions as an exchange (Bernheim, Shleifer, & Summers, 1985). In this framework, individuals may be obliged to reciprocate PFTs because of previous transfers, or may provide transfers even if it has some detrimental impacts on economic wellbeing, because they except to receive support later if needed.

Although reciprocal exchange theory emphasizes the exchange nature of the relationship, these exchanges are inherently social and shaped by social influence and norms (Blau, 1964). This is particularly clear in work on kinscripts, a related perspective that examines how family dynamics are shaped by shared beliefs, contexts, and histories (Stack & Burton, 1993). According to research on kinscripts, and consistent with reciprocal exchange theory, individuals may be compelled to provide assistance even when they cannot afford the expense because of cultural norms, family relationships, and an expectation to prioritize broader family wellbeing.

Research evidence supports both the altruism and reciprocal exchange perspectives (Light & McGarry, 2004). In Stack’s (1974) foundational ethnographic study of a disadvantaged African American community, families adapted to a lack of resources through large and complex support networks based on friendship and family. Stack’s work highlighted that exchanges were an integral part of daily life that allowed families to cope with poverty, but that also created hardship. In this framework, financial and in-kind exchanges that members of the community relied upon for help also served as poverty traps that limited economic mobility. Low-income individuals often attended to the needs of other family members to the detriment of their economic wellbeing. Because financial and in-kind exchanges were fundamental to social life, it was difficult for families to put their own needs ahead of the needs of the community.

Building on Stack’s influential work, other scholars have suggested that the need to rely on social support networks for assistance may be harmful and limit economic opportunities especially for low-income families (Dominguez & Watkins, 2003; McAdoo, 1978; Nelson, 2000; Offer, 2012; Uehara, 1990). In a study of poor and low-income single mothers, Edin and Lein (1997) noted that “mothers who manage to escape welfare and the $5-an-hour ghetto might have difficulty getting ahead” because of obligations to support friends and family (p.226). These and other studies suggest that kinscripts and shared norms may compel individuals to provide assistance even when it is not in their best interest (Braman, 2007; Mendez-Luck, Applewhite, Lara, & Toyokawa, 2016; Schmalzbauer, 2004). The notion that the financial exchanges that many families rely on to make ends meet can also create hardship for those providing assistance is provocative. Despite these insights from qualitative research, we know little from probability-based samples whether indeed provision of PFTs is detrimental for low-income families’ wellbeing.

Varieties of Material Hardship

First used by Mayer and Jencks (1989), studies of material hardship have increased over the last few decades, yet there is no agreed upon approach to studying material hardship (eg. Ouellette, Burstein, Long & Beecroft, 2004). Research has suggested that it is important to study different types of hardship, as the underlying mechanisms that cause hardship may vary by hardship type (Heflin & Iceland, 2009; Heflin, Sandberg, & Rafail, 2009). Specifically, if different types of material hardship are the result of similar processes, then it is not necessary to separately consider different types of hardships because the effect of PFTs on each type of hardship should be similar. For example, in this framework, the effect of PFTs on food insecurity should be similar to the effect of PFTs on housing problems. However, if this is not the case, and different social processes drive different types of hardship, then PFTs may only be associated with certain types of hardship or may be strongly associated with some hardships and only weakly related to others.

Prior research has also noted that certain types of hardship are more common than others, namely difficulty paying bills and having utilities cut off are more common than other hardships like unmet medical needs or housing hardships (Teitler, Reichman, & Nepomnyaschy, 2004; Zilanawala & Pilkauskas, 2012). As a result, we may expect these two types of hardship to be more commonly experienced when families provide PFTs (essentially there may be a lower threshold to experiencing these hardships given that they are more common), as compared with housing, medical or food hardship. We may also expect bill and utility hardships to be most strongly linked with giving PFTs because in some ways these hardships are less extreme. For instance, families may have more tolerance for having telephone service cut off than for losing their home or going without needed medical care (Nelson, 2011). Following previous research using the Fragile Families and Child Wellbeing data (Pilkauskas, Currie, & Garfinkel, 2012), we analyze aggregate (or summary) measures of hardship (any/none), types of hardship experienced, and the number of hardship domains experienced. In so doing, we consider hardships holistically, examining relationships between PFTs and global hardship experiences as well as specific hardships.

Differences by Income

The link between giving a PFT and material hardship might also vary by household income. Prior research has found that higher income individuals are more likely to provide transfers, and in larger amounts (Altonji, Hayashi, & Kotlikoff, 1997; Cox, 1987; McGarry & Schoeni, 1995). Higher income individuals may provide PFTs more often because they are able to do so without impacting their economic wellbeing. Reciprocity may also play a less important role among higher-income households than among lower-income households, as providing a PFT is less likely to be related to a need to reciprocate either previous transfers or future transfers. Thus, higher-income households may have greater ability to refuse to give PFTs when it would lead to hardship as compared with lower-income households who may have a greater need for future reciprocity. If this is the case, we anticipate that giving a PFT would be more likely to lead to hardship among lower-income households than higher-income households.

Differences by Race and Ethnicity

As noted above, we expect differences in the association between giving PFTs and material hardship by income. But there are also reasons to expect differences by race/ethnicity, as race/ethnicity is closely linked with financial wellbeing. If less well-off income groups are more likely to experience hardship after giving a PFT (either because of ability to pay or because of additional reciprocal obligations), then we would expect that Black and Hispanic families will experience more hardship after giving a PFT as compared to White families given average differences in income.

Moreover, differences in exchange networks and expectations related to exchanges may lead to racial/ethnic differences in the association between giving PFTs and material hardship that extend beyond simple average differences in financial wellbeing. In particular, Black individuals and families are more likely to have disadvantaged social networks, to live in neighborhoods with high concentrations of poverty, and to have incarcerated family members (Braman, 2007; Massey & Denton, 1993; Wilson, 1987, 1996). As a result, Black individuals and families are more likely than their White counterparts to have network contacts that are in need of assistance. For example, middle-income Black families are much more likely than middle-income White families to have poor family members (Chiteji & Hamilton, 2002, 2005; Heflin & Patillo, 2006) and less likely to have network contacts with college degrees (Tigges, Browne, & Green, 1998). Using data from the Panel Study of Income Dynamics, Chiteji and Hamilton (2002) estimate that more than one-third of middle income African Americans have poor parents, as compared to fewer than ten percent of middle income Whites. Importantly, White-Black differences in sibling and parent poverty contributes to racial inequality wealth and asset accumulation (Chiteji & Hamilton, 2002, 2005). Research has also found that Black mothers are more likely than White mothers to give financial assistance to network ties (Radey & Padilla, 2009; Raley, 1995), and that middle-class Black families feel an obligation to help the larger Black community and to help less fortunate relatives (McAdoo, 1978; Patillo, 2007; Shapiro, 2004). In fact, middle- and upper-income Black families are more likely to provide financial assistance to network ties, and this difference in giving explains part of the White-Black wealth gap (O’Brien 2012).

Many of these observed differences between White and Black families also extend to Hispanic families. In particular, Hispanic families are more likely than White families to be embedded in disadvantaged social networks (Fisher, 1982), have access to fewer strong ties (Small, 2007) and are less likely to receive financial assistance from family and friends (Lee & Aytac, 1998; Sarkisian, Gerena, & Gerstel, 2007). However, other studies offer more mixed evidence. These studies have found that Hispanic families give more assistance to their parents than White families (Lee & Aytac, 1998), but similar levels of assistance to siblings (White & Reidmann, 1992), and when compared with Black mothers, Hispanic mothers give less financial assistance to family and friends (Radey & Padilla, 2009). Overall, there is evidence that Black and Hispanic individuals are more likely than White individuals to have network ties that need assistance regardless of income. We thus test whether there are racial/ethnic differences in the association between PFTs and material hardship.

Differences by Family Structure

We also expect differences in the association between giving a PFT and hardship by family structure, as family structure is closely aligned with economic wellbeing. First, married mothers on average have a higher household income as compared to unmarried mothers. Thus, we anticipate unmarried mothers (both cohabiting and single) to experience more hardship after providing PFTs, as well as more pressure to give PFTs if their social networks are similarly disadvantaged (homophilous; McPherson, Smith-Lovin & Cook, 2001). Second, married households are more likely to accumulate wealth than unmarried households (Garfinkel & McLanahan, 1986) as marriage (and to a lesser extent, cohabitation) is a wealth promoting institution (through economies of scale, improvements in health, or having dual earners; Acs & Nelson, 2002; Becker, 1981; Lillard & Weiss, 1996; Lupton & Smith, 2003; Manning & Lichter, 1996; Smith, 1995; Waite, 1995).

Yet even beyond the link between relationship status and income or wealth, there are reasons to expect differences in the association with hardship as the nature of extended family relationships varies by family structure. A number of studies have demonstrated that marriage is a “greedy” institution and that married families are less likely to engage in family exchanges (both providing and receiving) as compared with single-parent families (Gerstel & Sarkisian, 2006) and have fewer intergenerational ties (Sarkisian & Gerstel, 2008). Bengtson (2001) also argued that weaker marital ties (or greater single parenthood) increases the need for strong extended family relationships, and a number of studies have found differences by family structure in the receipt of private transfers (Gottlieb, Pilkauskas & Garfinkel, 2014; Hao, 1996; Jayakody, Chatters & Taylor, 1993). No research has examined variation in the association between giving of transfers and hardship by family structure. We add to the literature by examining these associations.

METHOD

Data

We used data from the Fragile Families and Child Wellbeing Study (FFCWS; http://www.fragilefamilies.princeton.edu/documentation), a longitudinal study of approximately 5,000 urban births that were randomly sampled between 1998 and 2000 with an oversample of non-marital births (at a ratio of 3 non-marital to 1 marital birth). Mothers were sampled at the time of the birth of the focal child in 75 hospitals in 20 large U.S. cities (with populations over 200,000) and follow-up interviews were conducted when the child was 1, 3, 5 and 9 years old. Ninety percent of the mothers who completed interviews at birth (N=4,898) were interviewed again when the focal child was roughly one year old (N=4,363), 88 percent at the 3-year survey (N=4,231), 87 percent at the 5-year survey (N=4,139) and 76 percent at the 9-year follow-up (N=3,515). The FFCWS provides a unique dataset with which to examine the relationship between provision of PFTs and material hardship, as the survey asks parents about both giving and receiving PFTs, as well as a detailed battery of material hardship items, over multiple waves of data (years 1–9). Although other datasets collect data on PFTs and hardship, they do not provide as complete data on both variables measured at multiple intervals (allowing for an analysis using change models).

Our measures of interest – giving private financial transfers and material hardship – were not measured at the birth of the child so we used data from the 1, 3, 5 and 9-year surveys in our analyses, although in one analysis our covariates were measured at the baseline survey. We pooled the data resulting in 16,242 person-waves. We restricted our sample to be complete on covariates, private financial transfers, and the hardship variables, resulting in a final sample of 15,816 person-waves. Restricting our sample to be complete on covariates resulted in a 2% loss of sample (359 person-observations). Analyses comparing characteristics of the sample of mothers with item missingness as compared to those who had no item missingness showed no differences.

Although there is little item missingness, as is the case with all longitudinal studies, over time, respondents attrite from the study. Analyses of respondents who attrited suggest that they are more economically disadvantaged than the remaining sample. Mothers who attrite had lower income-to-needs ratios, were less likely to have obtained a high school degree, and were more likely to be immigrants. We address how attrition might affect our results in the discussion section.

Measures

Material hardship

Following previous work (Pilkauskas, Currie & Garfinkel, 2012), we created several measures of hardship based on a series of 10 questions that were asked at each survey wave (except in year 3, where one measure of food hardship was left out). The measure of any hardship indicates that mothers experienced one or more material hardships in the last year (1=any, 0=none). We also constructed five dichotomous measures of individual types of hardships (detailed more below): difficulty paying bills, utility cut-offs, food hardship, unmet medical needs, and housing hardship (we also constructed scales for each individual type of hardship that had more than one question and findings were substantively the same). Lastly, we created a measure of the number of domains of hardship that was a sum of the number of individual types of hardship that mothers experienced (range 0=none, 5=all).

All of the hardship questions were preceded with the following prompt: “We are also interested in some of the problems families face making ends meet. In the past 12 months, did you do any of the following because there wasn’t enough money?” Mothers were coded as having difficulty paying bills if “they did not pay the full amount of rent or mortgage” or “did not pay the full amount of a gas, oil, or electricity bill.” If mothers reported that their “telephone service was ever disconnected” or “gas or electricity was turned off, “ they were coded as having a utility cut-off. The food hardship measure included two questions, “did you receive free food or meals?” and “were you ever hungry, but didn’t eat because you couldn’t afford enough food?” Housing hardship was coded as 1 if a mother “moved in with other people even for a little while because of financial problems,” “stayed in a shelter, in an abandoned building, an automobile or any other place not meant for regular housing, even for one night,” or was “evicted from their home or apartment for not paying the rent or mortgage.” Medical hardship was assessed by the question “was there anyone in your household who needed to see a doctor or go to the hospital but couldn’t because of the cost?”

Giving private financial transfers (PFTs)

We constructed a measure of whether mothers reported giving a PFT based on the question: “In the past twelve months, have you given or loaned any money to friends or relatives?” (1=Yes, 0=No). In an extension, we also consider whether the size of the transfer changes the association between PFTs and hardship. If mothers reported giving money, they were asked how much and we constructed a measure of the amount of giving (coded as zero, $1–100, $101–500, $501–1000, greater than $1000).

Moderating variables

We studied whether the association between giving a private financial transfer and material hardship varied by household income, mother’s race/ethnicity and by mother’s relationship status. To study differences by household income, we used a measure of the household’s average income over years 1, 3, 5 and 9 (in thousands). This measure was derived from an imputed household income measure constructed by FFCWS staff that included the greater of the sum of the component parts of income, including both earnings and public transfers, or the single household income variable, to compute household income. We divided the average household income into tertiles where the bottom tertile had a mean income of about $17,500, the middle tertile $36,000 and the top tertile $87,500. In all of the other analyses, we also included household income as a control (either measured at the baseline survey or as a time varying covariate measured at each wave).

Race/ethnicity was coded as non-Hispanic Black, non-Hispanic White, Hispanic and other non-Hispanic race/ethnicity. Mother’s relationship status was coded as married, cohabiting, and single. For the analyses investigating relationship status as a moderator, mother’s relationship status at the birth of the child was used. In all of the other analyses, relationship status was included as a control, either measured at the birth or measured over time as a time varying covariate.

Control variables

In the pooled analyses without individual fixed effects (detailed further in the analytic approach section), we included a number of control variables that are associated with both the odds of giving a private financial transfer and of experiencing material hardship. These included: Mother’s education (less than high school, high school, some college and college or higher), age, immigrant status (foreign born), the number of children in the household, whether the mother lived with both parents at age 15, and city fixed effects (dummies for the 20 sample cities) and dummy variables for survey year as the data span a decade.

In our analyses using individual fixed effects, as noted earlier, we included time varying measures of household income and relationship status. We also included a time varying measure of perceived social support as mothers who have larger support networks may be more able to avoid material hardships, but may also have greater expectations to provide support. Following prior research (e.g. Harknett & Knab, 2007), we constructed an index of social support that was a sum of 6 measures asking mothers whether they had someone who could: loan them $200, loan them $1000, be counted on for emergency child care, provide a place to live if needed, cosign a bank loan of $1000, or cosign a bank loan of $5000.

Analytic Approach

We employed two strategies to examine whether giving private financial transfers was associated with experiencing material hardship. First, we ran regression models with extensive controls, time and city fixed effects (logistic for any hardship and ordinary least squares for the number of hardship domains – Model 1). We pooled the data and the measures of hardship and PFTs came from years 1, 3, 5 and 9. All the control variables for the analyses without individual fixed effects came from the baseline survey, and thus predate the variables of interest with the exception of survey year (which is time varying). These models were all double clustered at the individual and city level to account for non-independence.

Second, we ran individual (or person-specific) fixed effects regressions. The individual fixed effects model allows us to exploit the longitudinal nature of the data to account for fixed personal characteristics that might be associated with both the likelihood of giving a PFT and the likelihood of experiencing a material hardship. In each of the individual fixed effects models we included a control for survey year. In Model 2, we only included a measure of time, and in Model 3 we also included additional time varying measures, including household income, relationship status and social support, as these characteristics change over time and are likely associated with both hardship and PFTs.

The analyses studying differences by income, race/ethnicity and relationship used Model 3 (individual fixed effects with time varying covariates). The models by race/ethnicity were only run for White, Black and Hispanic mothers, as the “other” group sample was too small. We ran models stratified by each group and then ran Chow tests to study whether differences across groups (e.g. Black vs. White) were statistically different.

RESULTS

Descriptive Results

Table 1 describes the full sample and the incidence of material hardship by whether or not mothers gave a private financial transfer. In terms of sample characteristics (in Column 1), the FFCWS sample was relatively disadvantaged; 39% of mothers had less than a high school degree, whereas only 11% had a college degree. Per the design of the study, approximately 1/4 of the sample was married at the birth – or about 1/3 if we look over time. The sample was also very racially diverse, nearly half of the mothers were Black, 26% were Hispanic and 22% were White.

Table 1.

Sample Characteristics by Private Financial Transfer Giving (N=15,816)

	Full Sample		Gave PFT				T-Test
	(1)		(2)		(3)
			Yes (36%)		No (64%)
	% or M	(SD)	% or M	(SD)	% or M	(SD)
Material hardship measures
Any Hardship	45		52		42		*
Hardship Domains (M) 1	0.85	(1.1)	0.93	(1.14)	0.75	(1.06)	*
Difficulty Paying Bills	30		36		26		*
Food Hardship	13		14		13		*
Utilities Cut Off	21		24		20		*
Medical Hardship	6		7		5		*
Housing Hardship	11		12		11		*
Baseline Covariates
Household Income ($, M)	32,385	(31670)	34,200	(31575)	31,380	(31685)	*
Education
Less than high school	39		34		41		*
High school	26		26		25
Some College	25		30		22		*
College +	11		10		11
Relationship Status
Married	24		23		25		*
Cohabiting	36		37		36
Single	39		41		39		*
Mom’s Age	25.2	(6.03)	24.5	(5.85)	25.5	(6.11)	*
Race/Ethnicity
Black	49		59		43		*
Hispanic	26		19		30		*
White	22		18		23		*
Other	4		4		4
Immigrant	15		10		18		*
# of kids in household	1.26	(1.3)	1.25	(1.26)	1.27	(1.32)
Mom lived with both parents at age 15	43		37		46		*
Time Varying Covariates 2
Household income ($, M)	47,790	(50150)	53,350	(53910)	44,685	(47635)	*
Married	34		34		35
Cohabiting	29		30		28
Single	37		37		37
Social Support (M) 3	4.04	(1.84)	4.26	(1.71)	3.91	(1.9)	*
N	15,816	5,675	10,141

Note:

¹This measure is the number of hardship domains.

²Time varying covariates are measured at Years 1,3,5, and 9.

³The measure of social support is an index of 6 items at years 1,3,5 and 9.

^*Chi-Square/T-tests indicate statistically significant differences between giving and not giving PFTs at p<0.05.

Many mothers in the FFCWS sample experienced material hardship. Forty-five percent experienced at least one hardship, the most common being difficulty paying bills followed by having utilities cut off. In Columns 2 and 3, we show differences in hardship and sample characteristics by whether mothers gave a PFT. A little over one-third (36%) of mothers gave a PFT, and across all of the measures of hardship, mothers who gave a PFT had significantly higher incidence of hardship. Interestingly, although mothers who gave a PFT had more hardships, they also had higher average incomes than those who did not give and were also more highly educated (40% had at least some college versus 33% of those who did not give). Mothers who gave PFTs were more likely to be Black and less likely to be Hispanic or an immigrant than those who did not give PFTs.

In Table 2 we show the differences in material hardships by PFT giving and by income, race/ethnicity and relationship status. First we show the percent of mothers who give within each group. There were large differences by income tertiles in the giving of PFTs. Only 28% of mothers in the bottom tertile gave a PFT, whereas 63% of mothers in the middle and 58% of those in the top tertile did likewise. Differences by race/ethnicity also show that Black mothers were more likely to give PFTs (43%) than Hispanic (27%) or White (30%) mothers. Lastly, there were few differences by mother’s relationship status – roughly 1/3 of married, cohabiting and single mothers gave PFTs.

Table 2.

Material Hardship Measures by Private Financial Transfer Giving and Household Income, Race/Ethnicity and Relationship Status

	Income Tertiles						Race/Ethnicity						Relationship Status
	Bottom		Middle		Top		Black		Hispanic		White		Married		Cohabiting		Single
Gave PFT	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No
	28	72	63	37	58	42	43	57	27	73	30	70	33	67	36	64	37	63
Any Hardship	67	55*	57	45*	37	24*	57	49*	45	40*	44	34*	34	24*	57	48*	56	49*
Hardship Domains (M)	1.4	1.0*	1.0	0.8*	0.6	0.4*	1.0	0.9*	0.8	0.7*	0.9	0.6*	0.6	0.4*	1.1	0.9*	1.0	0.9*
(SD)	(1.3)	(1.2)	(1.1)	(1.1)	(.9)	(.8)	(1.1)	(1.1)	(1.2)	(1.)	(1.2)	(1.)	(.9)	(.8)	(1.2)	(1.1)	(1.1)	(1.1)
Difficulty Paying Bills	42	31*	40	29*	29	18*	40	30*	27	23*	32	24*	26	17*	40	30*	38	29*
Food Hardship	27	20*	14	12*	7	5*	14	14	14	14*	14	11*	8	6*	15	14*	16	15
Utilities Cut Off	38	27*	26	22*	13	9*	27	25	21	18*	19	13*	11	9*	28	23*	27	24*
Medical Hardship	8	6*	9	6*	6	3*	6	5*	9	6*	11	6*	5	4	10	6*	7	6
Housing Hardship	21	17*	13	11*	5	4*	12	13	12	10	10	8	4	4	13	12*	15	14
N	1,419	3,694	1,988	3,341	2,269	3,107	3,340	4,358	1,101	3,035	1,025	2,381	1,294	2,579	2,081	3,652	2,302	3,911

Note:

^*Indicates statistically significant differences between giving and not giving PFTS at p<0.05 from t-tests.

Despite differences in the rates of giving PFTs by group and differences in the levels of hardship experienced by each group, we saw a very similar pattern across all groups: mothers who gave a PFT had significantly higher levels of hardship than those who did not give PFTs. For example, mothers in the bottom income tertile were 12 percentage points more likely to experience any hardship if they gave a PFT than if they did not (67% versus 55%). Similarly, mothers in the top income tertile who gave a PFT were 13 percentage points more likely to experience any hardship (37%) than those who did not give a PFT (24%).

Is Giving A Private Financial Transfer Associated with Material Hardship?

In Table 3, we study the association between giving a PFT and the odds of experiencing any hardship, as well as the number of types of hardship experienced (hardship domains) across several specifications. In Model 1, we show the logit and OLS specifications with city and time fixed effects and found that giving a PFT was significantly associated with 44% higher odds of experiencing any hardship relative to not giving a PFT, and 0.17 higher number of hardship domains experienced. The covariates in this model show that a college education, being married, and being an immigrant was associated with less hardship. The results from Model 2 (plus individual fixed effects) and Model 3 (plus individual fixed effects and a few time varying covariates) show similar findings. Although the coefficients were reduced in the individual fixed effects models (in particular for the hardship domains models), which suggests selection into PFT giving and hardship, the findings remained strong and statistically significant. Giving a PFT was associated with higher odds of experiencing a hardship and more hardship domains.

Table 3.

Does Giving Private Financial Transfers Predict Material Hardship? OLS and Individual Fixed-Effects Regressions

	Model 1: Pooled Regression			Model 2: Individual Fixed Effects			Model 3: Individual Fixed Effects
	Any Hardship		Hardship Domains	Any Hardship		Hardship Domains	Any Hardship		Hardship Domains
	Logit		OLS	Logit		OLS	Logit		OLS
	β (SE)	OR	β(SE)	β (SE)	OR	β(SE)	β (SE)	OR	β(SE)
Gave PFT	0.37** (0.05)	1.44	0.17** (0.02)	0.18** (0.06)	1.20	0.05* (0.02)	0.25** (0.06)	1.28	0.07** (0.02)
Household income	−0.01** (0.00)	0.99	−0.00** (0.00)
Education
Less than high school	0.74** (0.15)	2.09	0.29** (0.07)
High school	0.55** (0.14)	1.74	0.15** (0.05)
Some college	0.76** (0.13)	2.14	0.25** (0.06)
Relationship Status
Married	−0.42** (0.07)	0.66	−0.21** (0.04)
Cohabiting	0.12* (0.05)	1.12	0.05+ (0.03)
Age	0.00 (0.00)	1.00	0.00 (0.00)
Race/Ethnicity
Non-Hispanic Black	0.05 (0.09)	1.05	−0.08+ (0.05)
Hispanic	−0.14 (0.10)	0.87	−0.12* (0.05)
Other	0.07 (0.13)	1.08	0.04 (0.06)
Immigrant	−0.28* (0.14)	0.76	−0.17* (0.07)
# of kids in household	0.07** (0.02)	1.07	0.03** (0.01)
Mother lived with both parents at age 15	−0.22** (0.06)	0.80	−0.11** (0.03)
Household income							−0.01** (0.00)	0.99	−0.00** (0.00)
Married							−0.12 (0.10)	0.89	−0.09** (0.03)
Cohabiting							−0.05 (0.07)	0.95	−0.04+ (0.02)
Social support							−0.20** (0.02)	0.82	−0.08** (0.01)
Constant	−0.75** (0.17)	0.47** (−4.30)	0.67** (0.09)			0.71** (0.04)			1.14** (0.05)
Observations	15,816	15,816	8,353	15,816	8,353	15,816
R-squared			0.090			0.013			0.035
Number of individuals	4,600	4,600	2,271	4,600	2,271	4,600

Note: All non-time varying covariates are measured at the birth of the child. All models include measures of time not shown. The models without individual fixed-effects also include city fixed-effects and are double clustered at the individual and city level.

^**p<0.01,

^*p<0.05,

⁺p<0.1

Do the Associations between PFTs and Hardship Vary by Type of Hardship?

In Table 4 we explored the association between PFTs and different types of hardship, using the model specification in Model 3 in Table 3 (with individual fixed effects and time varying covariates). We found that giving a PFT was associated with increased odds of bill (odds ratio 1.25) and food hardship (odds ratio 1.25), whereas associations with the other types of hardships were not statistically significant. Coefficients for other types of hardship were in the expected direction, but did not reach conventional levels of statistical significance. Our results suggest that PFTs are most strongly associated with interruptions in the payment of routine household items, like those for food or bills, and less associated with more severe forms of hardship like being unable to see a doctor.

Table 4.

Logit Regressions (Odds Ratios) Regressing Type of Material Hardship on Giving Private Financial Transfers

	Hardship Type
	Bill		Food		Utilities Cut-Off		Medical		Housing
	β (SE)	OR	β (SE)	OR	β (SE)	OR	β (SE)	OR	β (SE)	OR
Gave PFT	0.22** (0.06)	1.25	0.22** (0.09)	1.25	0.11 (0.07)	1.11	0.06 (0.11)	1.06	0.07 (0.08)	1.07
Observations	7,932		4,170		6,793		2,594		4,356
Number of individuals	2,150		1,139		1,846		706		1,197

Note: Standard errors in parentheses. All models include individual fixed-effects, measures of time, social support, income and relationship status.

^**p<0.01,

^*p<0.05,

⁺p<0.1

Do the Associations between PFTs and Hardship Differ by Income, Race/Ethnicity or Relationship Status?

To investigate whether these findings differed by income, race/ethnicity and relationship status, in Table 5, we ran the analyses stratifying by group. We found that the association between giving PFTs and hardship was concentrated among the most traditionally economically disadvantaged groups (lower income, single mothers, Black mothers). First, we found that the association between giving a PFT and hardship was concentrated among those in the bottom income tertile. Giving a PFT was associated with 59% higher odds of experiencing any hardship among the bottom income tertile but was not associated with higher odds of hardship for the middle and top tertiles. Similarly, giving a PFT was significantly associated the number of hardship domains (β= 0.20). Chow tests confirmed that mothers in the bottom income tertile were distinct from those in the middle and top tertiles.

Table 5.

Individual Fixed-Effects Models Stratified by Income Tertiles, Race/Ethnicity and Relationship Status

	Any Hardship						Hardship Domains
	Logit						OLS
	β(SE)	OR	β(SE)	OR	β(SE)	OR	β(SE)	β(SE)	β(SE)
Income Tertiles
	Bottom		Middle		Top		Bottom	Middle	Top
Gave PFT	0.49** (0.11)	1.59	0.12 (0.10)	1.07	0.17 (0.11)	1.15	0.20** (0.05)	0.01 (0.05)	−0.04 (0.03)
Observations	2,892		3,196		2,265		5,113	5,329	5,377
Number of individuals	796		868		607		1,527	1,537	1,536
Race/Ethnicity
	Black		Hispanic		White		Black	Hispanic	White
Gave PFT	0.32** (0.08)	1.38	0.22+ (0.13)	1.24	0.09 (0.16)	1.09	0.08* (0.04)	0.07 (0.06)	−0.00 (0.05)
Observations	4,565		2,158		1,360		7,695	4,133	3,406
Number of individuals	1,233		598		367		2,196	1,255	970
Relationship Status
	Married		Cohabiting		Single		Married	Cohabiting	Single
Gave PFT	−0.07 (0.15)	0.94	0.17+ (0.10)	1.18	0.45** (0.09)	1.57	−0.06 (0.04)	0.03 (0.05)	0.15** (0.04)
Observations	1,453		3,250		3,641		3,873	5,730	6,210
Number of individuals	391		883		995		1,122	1,680	1,798

Note: All models include individual fixed-effects, time and social support. The income tertile models also include relationship status, the race/ethnic models include relationship status and income, the relationship models include income.

Chow tests find that the bottom income quintile is significantly different (p<0.05) from the middle and top quintiles. No differences by race/ethnicity are statistically significant. Cohabiting and single results are statistically different from married results.

^**p<0.01,

^*p<0.05,

⁺p<0.1

Differences by race/ethnicity revealed that the association between giving and hardship was concentrated among Black mothers, and to a lesser extent, Hispanic mothers. Here we found significantly higher odds of experiencing hardship (and more domains of hardship) among Black mothers (odds ratio 1.38) and Hispanic mothers (odds ratio 1.24) who gave a PFT but not for White mothers. However, Chow tests did not find any statistically significant differences across race/ethnic groups (possibly as a result of insufficient power).

Lastly, the analyses by relationship status showed that single mothers were most likely to experience hardship when giving a PFT. We found for both single and cohabiting mothers who gave PFTs the odds of experiencing any hardship were significantly higher (odds ratio 1.57 and 1.18, respectively), although the association was only marginally significant for cohabitors. For the number of hardship domains, the association was only significant for single mothers (β= 0.15). Chow tests found that single and cohabiting mothers were statistically different from married mothers.

Supplemental Analyses

To test the strength of the observed associations, we ran a number of additional analyses. First, we examined whether the association between giving a PFT and hardship differed by the amount of money transferred. These findings are available in Appendix Table 1. We found that hardship was concentrated among giving at the lower levels – less than $100 and $100–499. Giving at the higher levels – $500–999 or greater than $1000 – was not associated with increased odds of hardship. Additional analyses examined differences by group (race/ethnicity, relationship status, income) and found the same associations: lower levels of giving were more strongly associated with hardship.

Appendix 1.

Regressing Material Hardship on Giving Private Financial Transfer Levels

	Any Hardship		Hardship Domains
	Logit		OLS
	β(SE)	OR	β(SE)
Giving Levels
<$100	0.40** (0.10)	1.49	0.11** (0.03)
$100–499	0.25** (0.08)	1.28	0.07** (0.02)
$500–999	0.06 (0.12)	1.06	0.05 (0.04)
$1000+	0.12 (0.13)	1.13	0.04 (0.04)
Observations	8,125	15,509
R-squared			0.036
Number of Individuals	2,232	4,586

Note: Standard errors in parentheses. Models include individual fixed-effects, measures of time, social support, income and relationship status.

^**p<0.01,

^*p<0.05,

⁺p<0.1

Prior literature has emphasized the reciprocal nature of exchanges within low-income communities and social networks. Because we also had data on whether or not mothers had received transfers, we ran models including time varying measures of receipt of a transfer as a control. Including this measure did not alter the findings.

We also ran a number of additional sensitivity analyses. First, we ran Model 1 including a number of time invariant measures (impulsive behavior, depression, health status, and a measure of public program use). The substantive results were unchanged. Second, we tested the inclusion of several additional time varying covariates in the models with individual fixed effects. These included alternative measures of income (poverty-to-needs ratio, equivalized household income and mother’s earnings), measures of public assistance receipt (Medicaid, Supplemental Nutrition Assistance, Temporary Assistance for Needy Families, Public housing/Section 8 voucher), a measure of mother’s employment, and asset ownership (home and car ownership). Again, inclusion of these covariates (individually or all together) did not change the findings.

Because measures of household income might also include PFTs, we ran the analyses that tested differences by income tertile using alternative economic specifications, including maternal earnings, income-to-needs ratio and equivalized income. The findings were largely the same with one exception: in the earnings model, the middle tertile also showed significant links between transfers and hardship. We opted to retain household income as our main measure, as opposed to say mother’s earnings, as material hardship is largely a household level measure (e.g. eviction or utilities disconnection), but differences across specifications were small.

Additionally, transfers may occur within or between households. To examine whether the association varied if transfers were limited to those that went outside of the household, we ran an extension where we limited the sample to mothers who were not doubled up (living with additional adults beyond the nuclear family). This resulted in a large decrease in sample size. Notably, by limiting the analysis to those mothers who were not doubled up, we excluded mothers who were particularly disadvantaged (those who were doubled up had an average annual income of $39,000 as compared to $53,000 for those who were not doubled up). In effect, this analysis excluded those who were most likely to experience hardship when giving a PFT. Nonetheless, the findings were robust for the any hardship analysis and positive but not statistically significant for the number of hardship domains analysis.

Families may also give PFTs more than once. To examine whether repeated giving over time was more strongly associated with material hardship, we ran a model where the independent variable was the number of waves of giving from years 1–5 and the outcome was hardship at year 9, and another model that examined the number of waves of giving from years 1–9 with hardship outcomes at year 9. Both models were very consistent, showing that an increase in the number of waves a mother gave PFTs positively and significantly predicted hardship.

As noted earlier, the data span a decade, including the dot com bubble and part of the Great Recession, which prior research has linked with both private financial transfer receipt and material hardships (Garfinkel & Pilkauskas, 2016; Gottlieb, Pilkauskas & Garfinkel, 2014; Pilkauskas, Currie & Garfinkel, 2012; Pilkauskas & Garfinkel, 2016). We tested the inclusion of time varying city level unemployment rates and the findings were again unchanged.

DISCUSSION

Our study contributes to a large body of research that examines social support among low-income families. We extend prior research by focusing on the provision of financial transfers, rather than receipt of financial transfers, and by using longitudinal data on families with young children to examine links with material hardship. We also move beyond previous literature by studying links with different types of hardship and by examining differences by income, race/ethnicity and relationship status.

Our findings show that families with young children provide private financial transfers to friends and family even when it is detrimental to their own material wellbeing. On average we find that mothers who gave a PFT have a predicted level of hardship that is 4 percentage points higher than those who did not give a PFT (48% versus 44%). This suggests that giving PFTs is not limited to families who can afford to do so, but that many families are giving transfers when it is not in their best financial interest. This finding is in keeping with prior qualitative literature that has documented, at least among low-income families, many families suffer economic hardships because of an obligation to support extended kin or friends (e.g. Edin & Lein, 1997; Stack, 1974).

Why might families provide transfers even when it is detrimental to their material wellbeing? Although we could not examine this directly, altruism, reciprocal exchange and kinscript theories support our findings. If families feel obligated to provide assistance to kin, and kin are in need, then a PFT may be provided even if it means going without some basic necessities. Or if in the past, families received assistance from kin or friends who now require assistance, transfers may be provided because of feelings of obligation or reciprocity. In the analyses that also controlled for receipt of a PFT, we found no change in the association between giving a PFT and material hardship. The dynamics of reciprocity are likely to be influenced by the experience of your network (if someone experiences a shock) and may also be influenced by the receipt of other forms of non-cash assistance (say child care). Research that can more fully test the potentially reciprocal relationship of the giving of PFTs, or other predictors of giving PFTs, is an important next step.

Giving PFTs was more strongly associated with bill and food hardships as compared with housing, utility or medical hardship. This finding is in keeping with prior research that suggests that the underlying mechanisms differ for various types of hardships (Heflin, Sandburg & Rafail, 2009). What is less clear is why bill and food hardships are most strongly associated with giving PFTs as compared to other hardships. One reason may be because these hardships are less extreme than other hardships. Families may be more willing to experience these hardships in exchange for providing financial assistance to friends and family. One might also argue that these hardships are easier to cause, or are more common, but the descriptive statistics suggest that although difficulty paying bills is the most common material hardship, many more families experience having their utilities cut off than food hardship. Thus, it does not appear that these are simply the most common types of hardship, but rather it suggests there may be different underlying processes between giving PFTs and specific types of hardship.

In an extension that explored differences by amount of giving, we also find that the associations between giving transfers and hardship are strongest for families giving smaller amounts of money. In particular, we find that giving less than $500 a year is most strongly associated with hardship. This suggests that the families who are in the most precarious financial positions, those already near hardship prior to giving a PFT, are the ones who experience hardship when they give to their friends and family. In contrast, those who give larger amounts are presumably more able to give and do not experience the same level of economic distress.

The results that examined differences by income, race/ethnicity and relationship status also support the notion that it is the families who are most vulnerable in terms of economic wellbeing who experience hardship after giving a PFT. We find that it is the more economically disadvantaged groups, low-income, Black and single mothers, who experience higher levels of material hardship when they provide PFTs. For example, among mothers in the bottom income tertile, 63% of those who gave a PFT are predicted to experience a material hardship as compared with 53% of those who did not. In comparison, we find no significant associations between giving a PFT and hardship for any of the other groups.

Our study has some limitations. First, we cannot say anything about causality. Although individual fixed effects models control for time invariant unobservable characteristics, they cannot account for unobserved time varying characteristics that may be associated with both giving a private financial transfer and material hardship (like community dynamics or other labor market forces). We tested the inclusion of a number of additional time varying covariates and the city local unemployment rate and found it did not change our results, but there may be other macroeconomic measures, or individual time varying measures, that impact hardship and PFT giving. Fixed effects models also cannot rule out the possibility of a reverse relationship, where experiencing a hardship leads to greater giving; however, we believe this to be unlikely. Similarly, because PFTs and hardship were measured at the same time, we cannot account for the fact that mothers may have experienced a hardship themselves and therefore are less likely to provide a transfer. To the extent that is true, we likely underestimate the associations.

Second, our sample is not generalizable to the total US population. Nonetheless, the longitudinal nature of our data allowed us to exploit individual fixed effects to better examine associations between PFT giving and material hardship, something that is not possible in most datasets. Additionally, as a result of the oversample of non-marital births, the sample is very economically and racially diverse, allowing for an examination of differences by group. Third, attrition may affect our findings. Mothers who attrite are more economically disadvantaged and slightly more likely to be Hispanic. Given that the association between PFT giving and material hardship is strongest among more disadvantaged mothers, our findings may underestimate the effect of PFT giving on material hardship, though of course this remains unknown. Last our measures of transfers were limited. Because we did not know who received the transfer, we could not distinguish within household transfers from those between households. Many studies have shown that families, and in particular low-income families, also provide transfers in the form of doubling up, child care assistance or transportation help (e.g. Pilkauskas, Garfinkel & McLanahan, 2014; Uttal, 1999). Nor do we know about why transfers were given. Future research that can also incorporate other forms of support, and distinguish between and within household support, and consider why transfers were given, would be a useful next step.

Despite some limitations, this paper suggests that economically vulnerable families, especially those with the lowest incomes, are providing PFTs even when it is linked with poorer material wellbeing. This has implications for thinking about how families may be able to escape poverty and establish self-sufficiency. Families are usually embedded in homophilous communities, where mothers’ networks are similarly economically disadvantaged and are likely to frequently be in economic need. We do not know why families give transfers, be it because of inadequacies in the public safety net, obligations of reciprocity or norms, these families appear to be experiencing real deprivations as a result of providing PFTs. Research that can better understand the mechanisms through which PFT giving is leading to material hardship would be a fruitful next step. Future research should also consider how public policies, like those related to incarceration or the strength of the safety net, might exacerbate families’ needs to provide transfers and consider how we might mitigate material hardships.