Income Sources Across Childhood in Families With Nonresident Fathers

Abstract

Unpartnered mothers rely on formal and informal income sources to support their coresident minor children. Building on work focusing on selective populations and shorter time horizons, we describe the family income sources on which U.S. women and their minor children rely for up to 17 years following an unpartnered birth or union dissolution (Panel Study of Income Dynamics 2001–2017; N = 12,369 person-year records from 3,148 children). Using rich description and fixed-effect models, we treat family income as dynamic, mapping change in the share and amount of family income from multiple sources as children age and women gain employment experience; enter new unions; experience changes in eligibility for public support programs; and receive contributions from kin, friends, and other household members. A patchwork of income sources is the norm throughout childhood, with mothers' earnings nearly universal but insufficient as a sole source of family income. Maternal repartnering increases family income through new partner earnings but is accompanied by offsetting reductions in other income sources, particularly from outside the household. In the context of weak institutional support for U.S. families, families with nonresident fathers rely on a complex mix of income sources to make ends meet.

Introduction

Unpartnered mothers rely on formal and informal income sources to support their coresident minor children (Edin and Lein 1997; Edin and Shaefer 2015; Hays 2003). In their careful accounting of low-skilled unpartnered mothers' household economies before U.S. welfare reform, Edin and Lein (1997) demonstrated that few women could meet their monthly expenses with income from earnings or welfare alone; the majority also relied on fathers' formal or informal child support payments and contributions from romantic partners, friends, and family. Even as contemporary women have accumulated more labor force experience and higher earnings than women of earlier generations, separation and divorce continue to trigger an increased reliance on public and private safety nets during the year following union dissolution, particularly for mothers ending cohabiting unions (Tach and Eads 2015). Maintaining this patchwork of support requires substantial maternal resources and is often insufficient to provide economic security (Halpern-Meekin et al. 2015).

Unpartnered mothers' economic vulnerability extends to many children: across all ages, approximately 20% of children live with an unpartnered mother, another 10% live with their mother and a new partner, and many more will have these experiences during childhood (Payne 2019). Child poverty is more closely tied to family structure in the United States than in other developed countries, a distinction largely attributable to national differences in governmental redistribution policy (Brady and Burroway 2012; Heuveline and Weinshenker 2008). The family income penalty associated with unpartnered parenthood is also borne more heavily by Black and Hispanic families than by White families (Williams and Baker 2021). The extent and unequal burden of economic vulnerability among unpartnered mothers underscores the importance of understanding their income dynamics and strategies for making ends meet.

Our analysis of unpartnered mothers' income dynamics speaks to two strands of research on economic vulnerability. Most proximately, we build on research describing cross-sectional or short-term profiles of unpartnered mothers' varied income sources (Edin and Lein 1997; Stanczyk 2020; Tach and Eads 2015) to illustrate long-term family income changes following an unpartnered birth or union dissolution. Our work also advances research showing that unpartnered parenthood is only one among many risk factors for poverty (Brady et al. 2017). Our framework is adaptable to populations confronting risk factors such as unemployment, poor health, or truncated educational attainment to articulate how individuals and families respond to economic insecurity (Brady et al. 2017). Further, it can be used to describe variation in individuals' responses to a particular risk, such as unpartnered parenthood, depending on whether the expected economic penalties to that risk are greater or lower because of factors like structural racism (Cross et al. 2022), gender inequality, and cultural norms about who is deserving of support (Moffitt 2015).

Our approach offers three main contributions. First, it treats family income as dynamic, highlighting changes in the share and amount of income from multiple sources as children age and women gain employment experience, enter new unions, and become (in)eligible for social welfare programs. Second, we emphasize the income contributions from family, friends, and other household members that complement unpartnered mothers' income from earnings, child support, and public transfer income, which are more often the focus of scholarship on unpartnered-parent household economies. Third, we illustrate how life course events alter income trajectories by considering maternal repartnering. We describe how establishing a new union affects total family income and the share and amount from different income sources. Our analysis shows that a patchwork of income sources is the norm throughout child-rearing years, with mothers' earnings a nearly universal component but insufficient as a sole source of family income. It also highlights how maternal repartnering increases family income through new partner earnings, a gain that is partially offset by reductions in other income sources.

We treat minor children as the unit of analysis and describe changes in annual total family income and income sources from the time of either a mother's unpartnered birth or separation or divorce until children reach age 17. Centering our analysis on a focal child allows us to examine how mothers provide for their children financially and how their strategies intersect with family context and child age. We use unpartnered mother to describe mothers' union status at the outset of the observation period and attend to mothers' repartnering over time. We consider six types of family income: mother's earnings, child support from a nonresident father, earnings from a mother's new cohabiting partner or spouse, public means-tested transfers to the mother or her current spouse or partner, gifts or loans from family or friends, and income from household members other than a new spouse or partner. Descriptive analyses and fixed-effect modeling are applied to data from the 2001–2017 biennial waves of the U.S. Panel Study of Income Dynamics.

Background

Children living with unpartnered mothers experience greater economic precarity relative to peers with two-parent families or unpartnered fathers. Various indicators point in this direction, some referring specifically to unpartnered mothers and others to unmarried (including cohabiting) mothers. For example, in 2019, 36.5% of minor children living with an unmarried female householder were in poverty, compared with 6.4% of children living with married householders and 16.3% of children with an unmarried male householder (Semega et al. 2020). Children born to unmarried mothers are also more likely than peers born into married-parent households to experience chronic poverty, with nearly 75% of children born to an unmarried mother between 1968 and 1998 experiencing at least three years of family poverty before age 10, compared with 16% of children born to married parents (Asiamah 2021). Children with unpartnered mothers are disproportionately likely to experience income volatility, and this volatility has been growing (Western et al. 2016).

Various factors contribute to the lower and less stable family income of families headed by unpartnered women. Given the United States' employment-based safety net, differences in earned income play an especially important role (Moffitt 2015; Western et al. 2016). Women continue to earn less than men, largely owing to motherhood (Kleven et al. 2019; Kochhar 2020). Unpartnered mothers may be particularly susceptible to low wages given that they are younger and have less education than partnered mothers (McLanahan 2004). Yet, 40% of U.S. children are supported solely or primarily by their mothers' earnings (Wang et al. 2013), and more than 70% of American mothers will be the primary or sole family income earner at some point during the first 18 years of motherhood (Glass et al. 2021).

Unpartnered mothers cannot rely on the risk-pooling available to partnered mothers with a second potential earner in the household (Oppenheimer 1997). Instead, to buffer against poverty and income volatility, they rely on various income sources to supplement earned income, including child support, public transfers, private transfers from friends and family living elsewhere, and shared economies in multifamily and extended households. Maintaining these varied income sources is intensive and burdensome; it requires mothers to expend time, energy, coping skills, and social capital on reciprocity networks of kin, friends, and former partners (Domínguez and Watkins 2003; Edin and Lein 1997; Hays 2003); to pursue missed or lapsed transfers; and to remain aware of and in compliance with eligibility requirements across a variety of public transfer programs (Handler and Hasenfeld 2007; Paik 2021). Next, we review how each of these sources contributes to family income.

Child Support

Approximately 20% of U.S. children are in the formal child support system (Office of Child Support Enforcement 2021; Ruggles et al. 2021). Among receiving families, child support from noncustodial parents represents approximately 12% of custodial parents' personal income overall; among low-income custodial parents, it contributes more than half of family income (Grall 2020). But the fraction of custodial mothers with formal child support agreements declined over the last two decades, from 64% in 1994 to approximately 50% by 2018 (Grall 2020). This drop is partly attributable to low-income parents' declining participation in the Temporary Assistance for Needy Families (TANF) program, which requires participation in the child support enforcement system. Custodial and noncustodial parents also express reservations about potential punitive sanctions on parents who are in arrears on formal support, and some opt for informal child support instead (Edin et al. 2019; Schroeder 2016; Waller and Plotnick 2001). Even among custodial parents with formal child support orders, less than half receive full payment regularly, and the share who report no receipt increased from 24.2% to 30.2% from 1993 to 2017 (Grall 2020).

Informal transfers partially compensate for low formal child support payments and include cash and transfers that are harder to monetize, such as gifts and in-kind support (Grall 2020; Waller et al. 2018). Unlike formal child support, informal transfers from fathers are less regular and taper off as children grow (Nepomnyaschy and Garfinkel 2010; Sariscsany et al. 2019). They also vary more by father involvement (Garasky et al. 2010; Nepomnyaschy 2007), parent relationship quality (Edin and Nelson 2013), and parents' new relationship formation compared with formal support (Berger et al. 2012; Bronte-Tinkew et al. 2009; Tach et al. 2010). Thus, informal support may be a particularly dynamic component of overall child support during childhood.

Public Support

We consider contributions to family income from four public transfer programs: TANF, Supplemental Security Income (SSI), the cash value of benefits from the Supplemental Nutrition Assistance Program (SNAP, formerly food stamps), and the Earned Income Tax Credit (EITC). TANF recipients are subject to a lifetime five-year limit on receipt of federal TANF dollars, although states may use their funds to augment this limit. Most recipients are unpartnered mothers with minor children. Time limits, low benefit levels, and sanctions for recipients who violate work and compliance requirements contributed to a steady decline in TANF caseloads from a peak of approximately two thirds of poor families with children in the mid-1990s to a plateau in the mid-2010s of roughly one quarter of poor families (Center on Budget and Policy Priorities 2021a). We expect that TANF contributes an increasingly small share of family income over the family life course.

SSI is available to severely disabled individuals who have very low family income and assets. The program serves a small share of families with children, including approximately 2% of children and working-age adults, the majority of whom have no other income source. Unlike TANF, SSI has no time limits, but benefits are reduced when recipients report even small amounts of income or assets (Center on Budget and Policy Priorities 2021b). Thus, family income eligibility is likely to fluctuate over time among those families with any other income source.

SNAP reaches substantially more families than TANF and SSI, serving 42 million people and approximately 84% of eligible households in 2017 (USDA Food and Nutrition Service 2019). The program is particularly important for families headed by unpartnered mothers (Shaefer and Gutierrez 2013). More than 40% of program expenditures are directed to children aged 6 or younger, and use patterns are heterogeneous. A recent study of SNAP recipients in Virginia documented that SNAP enrollment decreased each year after a child's birth and that the earliest enrollees were also the most persistent (Heflin et al. 2020). Relative to TANF, SNAP uptake is more sensitive to economic shocks, such as the Great Recession or the COVID-19 pandemic. Thus, for many children with unpartnered mothers, SNAP may be short term or episodic rather than a permanent feature of family income.

The EITC is administered through the tax system and is tied directly to employment. Low- to moderate-income workers with qualifying children are eligible with some restrictions. The refundable credit accrues from the first reported earned dollar, rises to a maximum before being phased out at higher income levels, and is typically paid in a lump sum with individuals' annual tax refunds. Over the last two decades, the EITC has emerged as the most important source of public support for many low-income families (Moffitt 2015).

Informal Transfers From Kin and Friends

In monthly income accounts, informal transfers from kin and friends represent a frequent but proportionally small share of family income in single-parent households. Halpern-Meekin et al. (2015) reported that among lower income working families in 2007, private cash transfers to unpartnered-parent households contributed approximately $130 per month, or roughly 7% of total monthly income. Income from kin and friends similarly contributed, on average, 5% to 7% of monthly income to unpartnered low-income mothers in the early 1990s, with approximately half of women receiving such income (Edin and Lein 1997). Among unmarried mothers whose children were born in large U.S. cities shortly after welfare reform, nearly two thirds reported receiving income support from friends and family during their child's first year (Teitler et al. 2004).

Although families headed by unpartnered parents—particularly low-income families—clearly receive income transfers from their private social networks, it is less clear whether such transfers are routine. Research on perceived social support has considered whether individuals have family or friends who would provide money in an emergency or to support a substantial investment, such as a rent deposit (Harknett 2006; Harknett and Knab 2007; Harknett and Hartnett 2011). The implicit expectation in this literature is that private transfers occur in response to occasional emergent needs rather than as a strategy undergirding long-term financial stability. Thus, informal social network transfers may be an irregular but critical component of family income, and the likelihood of receiving such occasional income may change over the life course.

Other Household Members' Income

Unpartnered mothers may also rely on other household members' income. This income might include earnings or transfers from working-age children who remain in the household or from other kin, such as the mother's parents or adult siblings with whom she and her children coreside (Harvey et al. 2021; Pilkauskas et al. 2020). Coresidence with household members other than one's spouse or partner and minor children is a frequent strategy unpartnered mothers use to reduce housing costs and share living expenses (Mollborn et al. 2011; Pilkauskas 2012). However, coresidence does not necessarily mean that household members pool all of their income for common use (Eickmeyer et al. 2019). Our main analyses include other household members' income as a source of mothers' family income, but results are organized to allow the reader to disaggregate this component from the total.

Maternal Repartnering

Women's entry into new romantic unions—measured here by entry into cohabitation or remarriage—may trigger changes in the share and amount of family income contributed from various sources. A new partner's income will increase total family income, all else being equal. But the addition of this new income source may lead to reduced contributions from other sources. For example, Berger et al. (2012) reported that mothers' transitions into new romantic partnerships are associated with declines in child support from nonresident fathers. The addition of a coresident partner or spouse may contribute to family income ineligibility for means-tested income transfers. Women may also pull back on paid employment, and friends and family may similarly reduce their support if a new partner's financial contributions lessen family income needs. The enduring expectation of men as primary breadwinners might further reinforce reductions in public transfers, mothers' earnings, and support from family and friends (e.g., Townsend 2002). For example, repartnering appears to trigger a gendered division of household and market labor, with women reducing their time in paid work and increasing their time in unpaid household labor (Bianchi et al. 2000; Sullivan et al. 2018).

Alternatively, repartnering may lead to complementarities in other income sources. Women's earnings may increase if they can work more hours or pursue a higher-paying job with a second parent figure in the household to provide childcare. A new partner's own network of family and friend relationships also may provide a new source of informal transfer income. Little empirical evidence on these processes is available, and our understanding is limited.

Analysis Plan

Data and Sample

We use data from the 2001–2017 U.S. Panel Study of Income Dynamics (PSID). The PSID, the world's longest-running active household panel study, launched in 1968 with a sample of 4,802 families to study family income change. The study has a genealogical design. Children born to or adopted by original PSID householders become PSID respondents themselves upon establishing their own households in adulthood. Immigrant refreshers in 1997 and 2017 maintained the sample's population representativeness. (We exclude families in the 2017 immigrant refresher because no retrospective information on family income is available prior to 2016.) The PSID includes an oversample of mostly Black families with low income (income at or below 200% of the federal poverty level) in 1967. Interviews with a single respondent per family were conducted annually until 1997 and have been conducted every other year since then.

The PSID includes 10,764 children born in 1986–2012 and observed in at least one survey wave between 2003 and 2017. Before sample exclusions, 67% of children had a coresident father at birth, and 33% had a nonresident father. We exclude 783 children who were observed in only one wave, 393 children with missing data on their mother's or father's coresidential status, and 6,049 children who did not meet the condition of living with their biological mother and apart from their biological father in at least two waves. From this sample, we generate a child-wave file that includes all survey waves in which children were living with their biological mother and apart from their biological father (N = 14,396 child-waves). We lose additional cases owing to item-specific missingness on key variables,¹ resulting in a final sample of 12,369 child-waves from 3,148 children. Of these, 8,493 child-waves are from the 2,048 children in our sample without a coresident father at birth, and 3,876 child-waves are from the 1,100 children who experienced parental separation. Each child contributed two to eight records, given the biennial interview schedule.

Measures

Our research questions require repeated measurement of family income disaggregated into various sources with regular periodicity in a sample with robust longitudinal response rates. The PSID is well suited to these requirements. During the PSID interview, the respondent (usually a householder) reports on all income sources in the prior calendar year. Income component data are reported with relatively low item nonresponse relative to other national surveys, and PSID staff impute this information as needed (Duffy 2011). High wave-to-wave response rates (89% to 95% in survey years for the analytic sample) enable longitudinal analyses over a long time horizon relative to contemporaneous national studies of family income (Panel Study of Income Dynamics 2019).

We include six types of annual income: mother's earned income, such as salary, wages, and tips; spouse's/partner's earned income; any child support; means-tested cash and near-cash transfers to the mother or her spouse/partner, including TANF, SSI (including amounts the mother or partner received for themselves and amounts they received on behalf of another recipient), SNAP, and EITC; financial help from friends or family; and income from other household members. The last category includes income from others in the family unit, such as children who have not yet moved out, as well as total income in other family units that share the same household as a child and the child's mother. We convert income from each year to 2000 dollars and top-code at the 99th percentile. With the exception of the EITC, this measure of family income is pretax and excludes unearned income sources (e.g., investment income), which are infrequent in our sample.

Survey respondents report all income sources except EITC for the calendar year before the interview year. The EITC refund amount is estimated on the basis of past-year earned income for the tax filing unit in the family household using TAXSIM (version 32), a publicly available program to calculate federal and state tax liability from individual-level data (https://users.nber.org/∼taxsim/taxsim32/). We use the estimation methodology that Kimberlin et al. (2015) developed for use with the PSID. We include EITC refunds as transfer income rather than as a component of mothers' earnings because tax filing units in the household may include other income earners.

We generate a dummy variable for parental union status at birth, differentiating children whose parents were not married or cohabiting in the year of their birth from those whose parents were married or cohabiting but later separated. Among the latter group, a second indicator further differentiates early and late separations: those occurring before child age 4 versus at age 4 or older (approximately the median age of separation for sample children with a resident father at birth). For the 51% of sample children whose parental union status at birth is not directly observed or from whom it cannot be directly inferred through parents' marriage histories, we use information provided in the PSID Family Composition File (Fomby 2020) about the first union observed after the child's birth to assign union status at birth. That is, if at the first observation, the mother was living in a coresidential romantic relationship that she entered before the child's birth, we assume that the child's parents were together at the child's birth. If she is in a coresidential romantic relationship that she entered after the child's birth or if she is unpartnered, we assume that the child's parents were not together at the child's birth. This strategy allows us to identify parental union status for more than 99% of children eligible for our sample.

We create a count of time spent without a coresident father. For children without a coresident father at birth, this variable is measured by child age assessed at the beginning of the calendar year for which the mother's income is reported. For children with a coresident father at birth, this variable is measured by years between the parents' separation and the beginning of the calendar year for which the mother's income is reported. In assessing years separated for children with multiple observed father exits, we include only the period after the most recent exit. For most children, we use the month and year of the father's exit from the household to precisely measure the time since the parents' separation. For those missing the father's exit date, we assume that the father's exit occurred at the midpoint between the date of the last survey wave in which the father was observed coresident with the child and the date of the first wave in which the child is observed without a coresident father.

To assess the impact of mother's repartnering on the receipt of other types of income, we account for whether the mother is either married or cohabiting with someone other than the child's biological father and the focal child's age at the start of the mother's repartnership.² Our indicator variable includes three categories: not repartnered, the current repartnership began early in the child's life (before age 4), and the current repartnership began later in the child's life (at age 4 or older). We assess partnership status in the wave before the year for which income is reported. For example, in the context of our fixed-effect models (described later), the mother's repartnered status at the 2015 interview is used to estimate how income in the calendar year 2016 (reported at the 2017 interview) deviates from income averaged over all the child-years an individual child contributes to the analysis. If information is not available from the prior survey wave, we create repartnership status in a given year from reported move-in and move-out dates of current and former partners.

Finally, we generate a time-varying indicator of whether the focal child's mother had additional children following entry into the sample. This variable captures any children born to or adopted by the mother following the focal child's birth or their biological parents' separation. These children are likely biologically related to a new partner and half-siblings of the focal child. Table A1 in the online appendix shows mean values for time spent without a biological father and shares experiencing early and late parental separations, early and late maternal repartnerships, and new siblings for our full sample and by mother's union status at birth.

Method

We begin with a descriptive analysis of family income sources stratified by maternal union status at birth. We present unconditional average annual income from each source, income conditional on receipt, and the share of children's families receiving each income source in each year after an unpartnered birth or union dissolution. We then describe the cumulative share of children in families receiving income from various sources—for example, family income only from the mother's earnings versus a combination of income sources. Descriptive analyses are based on repeated cross sections representing the share of children at each age or year since separation in families with a given income mix. The data are weighted to be representative of families present in the United States in 1997 (the year of the most recent immigrant refresher whose members are included in the analytic sample), using the last observed longitudinal individual-level weight for a child.

We next use within-child fixed-effects linear regression models to summarize the magnitude and statistical significance of changes in the amounts and proportions of total annual income from each source across child waves. We estimate models pooled over mothers' union status at birth with a linear specification of time since a mother's unpartnered birth or union dissolution to track the duration of living apart from a biological father.³ For children born to partnered parents, we interact this duration variable with our indicator for early versus late parental separation to account for group differences in family income trajectories. To assess the relationship between repartnering and income, we include our indicator for repartnership, again differentiating between early and late repartnerships to assess how the relationship between child age and income interact with maternal repartnering. Finally, we include our time-varying measure of whether the focal child's mother had additional children to address complexities around support for children in the household who may be biologically related to a current partner.

Fixed-effect models estimate the share of income from each source averaged over all child waves and describe deviations from that within-child average as a function of the time-varying covariates in the model. We cluster standard errors at the mother level to account for families with multiple children. Because each child serves as their own control, these models remove all time-invariant attributes that vary between children and that may be related to the outcome. Therefore, we do not add controls for stable characteristics, such as child race/ethnicity or gender.

Weighted sample characteristics are provided in the online Table A1. We present these characteristics stratified by maternal union status at birth to illustrate group differences that may contribute to divergent baseline income and income trajectories between the two groups net of the time-varying characteristics included in fixed-effects models. In supplementary results discussed later, we estimate random-effect models with the time-varying variables included in our fixed-effect models and time-invariant demographic characteristics shown in Table A1.

A reproducibility package to generate the results summarized below is publicly available at https://doi.org/10.6077/k6q3-1667.

Results

Descriptive Patterns

Figures 1 and 2 present unconditional means for each income source in each year in area graphs to visually summarize the fraction of family income attributable to each source over time. As a benchmark for comparison, the dotted horizontal line in each figure represents the federal poverty threshold in 2000 for a family with one adult and two children before means-tested public transfer receipt ($13,874; U.S. Census Bureau 2000). Mothers' income is the biggest single source of total income, but other sources together contribute substantially to family income over childhood. Consistent with the more disadvantaged positions of mothers who are unpartnered at birth (Figure 1) relative to those separated following birth (Figure 2), income levels are lower overall for children born to an unpartnered mother—especially earnings and child support, which together stabilize above the poverty threshold only after children reach age 10.

Tables 1 and 2 describe these income trajectories from birth to age 17. In Table 1, for each year following birth to an unpartnered mother, the first column gives the average total income. The remaining columns present average unconditional amounts from each income source, the share of children receiving each source, and the average amounts conditional on any receipt. We include other household members' income in estimates of unconditional total family income. As noted earlier, the amount in column F can be subtracted from the total to obtain an estimate of family income excluding these contributions. Table 2 provides the same information for the years after a union dissolution for children born to partnered parents. Here, we collapse the last three postseparation years (15–17) because of the small number of observations for each year.

The first three rows in Table 1 describe family income for children one year after birth to an unpartnered mother. Average family income in that year combined across the six sources is $24,593. Mother's earnings contribute an unconditional average of $10,241 to that total, 82% of children have a mother with earned income in that year, and mothers' earnings average $12,424 conditional on having earnings (column A). Columns B through F may be interpreted the same way, and the unconditional means (first row) in each cell A–F sum to average family income.

We highlight four themes. First, in all years, 75% or more of children reside in families in which mothers contribute earned income, regardless of maternal union status at birth. Yet, in most years, mothers' unconditional average earned income represents only 40% to 60% of all family income, indicating reliance on other sources. Thus, a mother's earned income is both nearly universal and nearly universally insufficient as a family income source following an unpartnered birth or union separation. This pattern is consistent with recent work on the prevalence and lifetime likelihood of mothers' status as breadwinners (Glass et al. 2021) and with classic work describing mothers' challenges in gathering many strands to make ends meet (Edin and Lein 1997).

Second, the amount and contributing share of each income source changes with time since unpartnered birth or separation. Some income sources become more frequent and conditional means increase with years since unpartnered birth or separation. Among children born to unpartnered mothers (Table 1), mothers' conditional earnings increase by 77% between years 1 and 17 after the child's birth. Over the same interval, the share of children in families receiving new partner income increases from 15% to 28%, and the conditional amount received increases by 40%. The share of children with family income from other household members also rises from 17% to 70%, but the conditional mean income drops by half, likely reflecting changes in the sources of other household members' income. In the earliest years following an unpartnered birth, mothers may live in multigenerational or other doubled-up households to pool income with other adults with years of labor force experience. In the outgoing years, analyses (available upon request) suggest that household income comes largely from adolescent or young adult children who remain in the family unit and contribute earnings from part-time or entry-level employment.

In all years, 74% to 90% of children's families receive public support. This receipt reflects widespread eligibility within our sample of low- to moderate-income families for federal tax refunds through the EITC program. Informal transfers from kin and friends become less frequent over time but never diminish completely. For families continuing to receive this income source, conditional means are similar across years, suggesting that private transfers continue to contribute nontrivially to family income for an increasingly selected subset of families. The share of children in families receiving child support fluctuates in the earliest years after birth to an unpartnered mother but then stabilizes at around 30%, with conditional means of approximately $3,000 in most years. In sum, family income sources remain complex and varied throughout childhood. Patterns of change are similar for children born to partnered parents (Table 2).

Third, family income profiles by union status at child's birth are distinct. Total family income in all years is lower in families in which a child's father was nonresident at birth. In the first year, for example, family income is close to $25,000 for children whose mothers were unpartnered at birth compared with nearly $42,000 for children whose mothers were separated or divorced from the child's father. The disparity in family income between the two groups narrows in the outermost years, with a gap of roughly $11,500. Mothers who dissolve their unions are somewhat more likely to be employed and to have higher earnings in all years than mothers who are unpartnered at the child's birth; indeed, the conditional average annual earned income of a mother partnered at birth in the first year after separation exceeds the earned income of mothers who were unpartnered at birth 17 years later ($28,598 vs. $22,018). (These estimates are unadjusted for age, educational attainment, or labor force experience.)

Relative to children whose mothers were unpartnered at birth, those whose mothers were partnered also have a higher frequency of child support receipt and larger payments conditional on receipt. Over time, previously partnered mothers more often have earnings of new cohabiting or married partners compared with mothers who were unpartnered at birth, and the conditional average annual earned income of mothers' new partners is higher among children whose mothers were previously partnered. Children with formerly partnered mothers also have a slightly higher frequency of family income potentially available from other household members, particularly in the years just after separation. These children's families also have lower receipt of income from public support and informal transfers. However, among those who receive this income, conditional averages are similar to those of children whose fathers were nonresident at birth.

Fourth, the frequency of receiving household income through mothers' repartnering covaries with other family income components. For children in both groups, as the share of children with household income from mothers' new partners increases, the share receiving income from public or informal transfers declines. Mothers who were partnered at birth also fluctuate in their employment and their unconditional and conditional annual earned income as the share with income from repartnerships increases. However, these patterns may be unrelated and may simply reflect changes with the passage of time.

Figure 3 displays the proportion of children whose family income comes from an increasingly complex combination of sources in each year since birth (panel a) or parental separation (panel b). Each row presents data by year since birth or separation. The values in the shaded horizontal bars describe the share of children whose income is derived from an increasing mix of sources. Within each row, values in the horizontal bars may be summed to describe the cumulative share of children with family income on a continuum from less varied to more varied sources. For simplicity, these results pool income from private transfers and other household members. Table A2 in the online appendix provides the estimates from Figure 3 in tabular form, as well as two sets of analogous estimates that exclude the EITC and other household members' income, respectively. The last column of Table A2 shows the small percentage of children whose mothers have no reported income from earnings, child support, or transfers and live with other household members who provide the only family income source.

The first shaded bar in Figure 3 describes the share of children receiving income only from the mother's earnings. In the first year after the birth of a child with a nonresident father (panel a, first row), no one in our sample relies exclusively on mother's earnings. Although an exclusive reliance on mothers' earnings increases over the focal child's life course, in no year do more than 7% of children born to an unpartnered mother have her earnings as the sole source of family income.

Two percent of children in their first year (panel a, first bar in the first row) have family income that includes child support only or a combination of child support and mother's earnings. The third bar shows that 5% of children have family income from a new partner's earnings alone or in combination with child support and mother's earnings. Together, the first three bars describe family income sources for 7% of children. These sources represent family income derived entirely from the earnings of one, two, or three adults connected to a child as a biological parent, adoptive parent, or stepparent (including mothers' cohabiting partners as stepparents and regarding child support payments as contributions from fathers' earned income). These combined sources fully account for total income in 7% to 21% of families in the years following an unpartnered birth. This figure is higher in the columns of Table A2 that exclude other household members' potential contributions from the denominator.

The remaining bars in Figure 3 absorb profiles where family income is at least partly from sources other than parents' and partners' earnings. Public support, provided largely through the EITC, contributes to family income among 45% of children born to an unpartnered mother. Cumulatively, 52% receive all family income from a combination of earned income and public support in their first year. The remaining children (48%, last bar) have family income derived at least partly from informal transfers from non-coresident kin and friends or from other household members. As children age, the overall share whose family income comes only from parent or partner earnings increases and the share including any amount of public support, kin and friend transfers, or other household members' income declines until 15 years after birth.

Panel b of Figure 3 displays results for children born to partnered parents. In the years after separation, the fraction of children whose family income comes exclusively from biological parents, adoptive parents, or stepparents is slightly higher than that of children born to unpartnered parents. This fraction (the sum of shares in the first three bars in each row) ranges from 12% to 29% across years. The general pattern of complexity in income sources is similar to that for children born to unpartnered mothers, with a substantial share of children across years receiving family income from sources beyond parents' earnings.

The preceding analysis provides a rich descriptive portrait of children's family income but has two significant limitations. First, each person-year describes a different set of individuals, capturing children of a given age irrespective of the survey year. Thus, the tables provide pooled estimates of family income at a given age over 17 years and nine survey waves and cannot be interpreted as an individual-level longitudinal explication of early life course income dynamics. Thus, we cannot discern how time and changing circumstances contribute to increasing or decreasing reliance on specific income sources as children age. Second, children observed only a few years after parents' separation are likely to be heterogeneous with regard to age at parental separation compared with children whose parents separated 15 or more years previously; conversely, youth who experienced an early parental union dissolution may be more highly selected and hold distinct underlying income profiles compared to children whose parents separated later.

Fixed-Effect Model Results

To estimate individual-level effects of time and parental repartnering on children's family income, we use ordinary least-squares fixed-effect models estimating the share and amount of family income from each source. For children born to partnered parents, interaction terms between years since separation and early and late separation (before child age 4 vs. older) account for potential heterogeneity in income trajectories following earlier versus later union dissolution. For children whose mothers repartner, separate dummy variables for early and late repartnership (before child age 4 vs. older) account for potential interactions between repartnering, child age, and income.

Table 3 summarizes results from models estimating the share of family income from each source. The dependent variables range in value from 0 (no income from this source) to 1 (all income from this source). Coefficients represent the estimated deviation from the within-person average associated with a one-unit change in the independent variable. Multiplying coefficients by 100 allows their values to be expressed as percentage-point changes.

For children born to an unpartnered mother, the main effect of time describes the association of child age in years with changes in each outcome. For children born to a partnered mother, the effect of time since separation is captured in the main effect for time and its interaction with the timing of the separation, either early (before child age 4) or later. For example, for children born to an unpartnered mother, each additional year of child age is associated with a 0.5-percentage-point decline (−0.005 × 100) in the share of family income from public support (Table 3, row 1, column 4). Child age is also associated with a 0.2-percentage-point annual decrease in the share from kin and friends and a 0.8-percentage-point increase in the share from other household members. The main effect of time is nonsignificant for mother's earnings, child support, and new partner's earnings. Results are statistically equivalent for children born to a partnered mother (i.e., interaction terms are statistically nonsignificant) with one exception: the change in the share of income from child support decreases by an additional 0.2 and 0.4 percentage points each year for children who experienced an early or late separation, respectively.

Maternal repartnering is consistently associated with changes in the share of income across sources. For mothers remaining in a repartnership formed before child age 4, the mother's earnings share is 8.6 percentage points lower in each year than in the years she is unpartnered, a difference that is almost entirely offset by a new partner's earnings (8.5 percentage points). Shares of family income from child support (1.5 percentage points), public support (4.2 percentage points), and kin and friends (1.3 percentage points) are also lower in years in an early repartnership, but the share of income from other household members is 7.1 percentage points higher. When mothers remain in a repartnership formed at child age 4 or later, the share of income from the new partner's earnings is 16.8 percentage points higher than in their unpartnered years and 8.3 percentage points higher than in early repartnership years (p < .01; contrast not shown). For years when mothers are in early compared with later repartnerships, shares of income are statistically equivalent for all other sources except other household members' income, which is higher in early repartnership years (p < .01; contrast not shown).

Table 4 presents estimated deviations in income amount from the person-centered mean for total family income and each component. For children born to an unpartnered mother, each additional year of child age increases total family income by $602 (coefficients rounded to the nearest dollar). This increase is the net gain from increased mother's earnings ($303), partner's earnings ($104), and other household members' income ($296), and decreased public support ($52) and private transfers ($29). Results are statistically similar for children born to partnered mothers regardless of the timing of separation, except that income from child support decreases by an additional $103 each year in families where children experienced separation at age 4 or later relative to families where children had an unpartnered mother at birth.

Years in which a mother remains in an early repartnership are associated with a substantial increase in total family income ($5,173). This gain comes largely through the new partner's earnings ($3,580) and other household members' income ($3,863), but it is partially offset by lost child support ($336) and income from family and friends ($184). Thus, early repartnership is associated with a strong net gain for total family income but comes at the cost of some private transfer income and child support. When a mother remains in a repartnership formed at or after child age 4, total family income is even higher ($9,779) and comes entirely through the new partner's earnings ($10,089), with some loss of child support ($277) and private transfers ($116). Differences in the amount of family income, the partner's earnings, and other household members' income by repartnership timing are statistically significant (p < .01; contrasts not shown).

Table A3 in the online appendix summarizes the results from Tables 3 and 4 when EITC is excluded from public support income. These results are generally similar to the estimates based on including EITC funds as transfer income. Online Table A4 summarizes random-effects models that include the time-invariant covariates presented in Table A1. Relationships between the variables of interest remain mostly similar in magnitude, direction, and statistical significance, with one exception: the main effect of time on predicted values of the amount of total income and mothers' earnings are approximately 50% smaller in magnitude than that observed from the fixed-effects models.

Discussion

This work extends prior research on how mothers make ends meet (Edin and Lein 1997; Stanczyk 2020; Tach and Eads 2015). It does so from the child's perspective, following changes in family income from the time of their birth to an unpartnered mother or the dissolution of their parents' marriage or cohabitation. We provide a dynamic perspective on the patchwork of income U.S. mothers gather to support their families in the context of weak institutional support. Our study therefore sheds new light on changes in the mix of income sources as children age and women gain employment experience, enter new unions, experience changes in eligibility for social welfare programs, and rely to varying degrees on support from family, friends, and other household members.

Up to 17 years following an unpartnered birth or union dissolution, nearly all families continue to rely on a complex combination of income sources within and outside of children's households, including public and private transfers. Mothers' income is a critical component of income, composing approximately half of family income in most years. Still, less than 10% of children at any given age rely on mothers' earnings as the sole source of family income. We conclude that mothers' earnings are a nearly universal component of children's family earnings but also nearly universally insufficient to make ends meet. Further, we find that less than 30% of children have family income composed solely of a combination of a mother's earned income, child support from a nonresident father, and a new partner's earnings. The rest have income from a mix of nonparental contributions, including public support, private transfers, and income from other household members.

Our approach highlights the importance of family, friends, and household members to mothers' income strategies for many years following entry into an unpartnered-parent household. Although mothers are more likely to draw on these private sources in the years immediately after an unpartnered birth or union dissolution, they continue to represent a nontrivial portion of family income across childhood. In most years, 40% to 50% of unpartnered-mother families report receiving private transfers from family or friends or contributions from other household members, representing 8% to 20% of average annual income. Rather than being tied to an early life stage, private transfers are threaded into family economies over the long term. Further, a substantial share of families receive private transfer income regardless of whether parents were living together at birth. This perspective offers new insight into private transfers, which have mostly been studied as a strategy for making ends meet (Edin and Lein 1997) in the years shortly following an unpartnered birth (Domínguez and Watkins 2003; Harknett and Knab 2007; Stack 1974). Future research should consider how private transfers sustain other economically vulnerable populations over the long term.

Our approach further illustrates how life course events alter the trajectory of income from various sources by considering the specific case of maternal repartnering. We do so in a fixed-effect framework accounting for age or time from separation and the mother's time-varying repartnership status, as well as time-invariant attributes of children that may be related to family income dynamics. We find that mothers' repartnering increases family income overall through new partners' earnings, which, on average, exceed mothers' when part of the household income mix. Repartnering changes the relative importance of income sources on which families rely and is associated with absolute declines in some income sources. Most importantly, it is associated with declines in child support among children born to unpartnered parents, which likely reflects and shapes father–child relationships. This pattern of decline is consistent with past work linking mothers' new relationships to declines in fathers' contact with their nonresident biological children (Tach et al. 2010). This dynamic framework extends our understanding of how relationship formation affects families' economic stability.

This study has some limitations. First, we focus on one recent historical period in a single national context where unpartnered parenthood is more often a risk factor for child poverty than in other countries. These results may not generalize to other periods or countries. Even within the period and national context examined here, our results may differ across years (e.g., during the Great Recession in 2007–2009) and U.S. states, and our data are not representative of children whose families entered the United States during the years of observation. Second, because we lack a measure of in-kind (noncash) contributions to households from nonresident fathers, other kin and friends, and public assistance (e.g., housing and childcare subsidies), we cannot fully account for changes in contributions, particularly where patterns of cash and in-kind support differ over the life course (Grall 2020; Nepomnyaschy and Garfinkel 2010; Sariscsany et al. 2019; Waller et al. 2018). Finally, and relatedly, we face some limitations disentangling the interplay of child support and public assistance with family composition over time.

U.S. children experience substantial income insecurity, particularly in unpartnered-mother families—an experience that has been on the rise with the shift to an employment-based safety net and increasing precarity in low-wage work (LaBriola and Schneider 2020; Western et al. 2016). Recent data show that 30% of children live with an unpartnered mother or with their mother and a new partner, and many more will do so during childhood (Payne 2019). Relative to other rich countries, the United States provides little institutional support to buffer the effects of job instability and irregular working hours or to protect workers in the event of illness or caretaking needs (Glass et al. 2016; Gornick and Meyers 2003). In this context, unpartnered mothers rely on private networks and spend scarce resources to patch together income from a mix of sources (Edin and Lein 1997; Halpern-Meekin et al. 2015). The lack of institutional support for families takes a toll on parents (Glass et al. 2016), and the networks on which unpartnered mothers rely are also fragile, introducing another layer of uncertainty and strain (Harvey 2022). A recent policy change aimed at alleviating the impacts of the COVID-19 crisis offered a promising reprieve: the temporary child tax credits paid out during the pandemic kept economically vulnerable families from material hardship, and extending them has the potential to drastically reduce poverty (Acs and Werner 2021; Greenstein 2021). Progress has stalled on proposals to continue this critical support, leaving families with the uncertainty and strain inherent in the income patchwork we document here.

Acknowledgments

Authors are listed alphabetically. We gratefully acknowledge helpful suggestions and comments from the journal editor and anonymous reviewers, outstanding research assistance from Yongxin Shang, and generous support from NICHD research grant R01HD088506 and center grants P2CHD041028 and P2CHD044964 and the Cornell Center for Social Sciences Data Archiving and Replication Service.

Notes

References