Abstract
This paper documents the well-being of low-income single mothers nearly thirty years after welfare reform. Using unique data from a monthly cross-sectional survey of 7,186 low-income single mothers who are currently receiving or recently received Supplemental Nutrition Assistance Program benefits, we consider how mothers are faring today as compared to similar mothers in the early 1990s, those who were part of Edin and Lein’s 1997 book Making Ends Meet. We report mothers’ employment, earnings, and use of the private and public safety net to make ends meet. We find that both single mothers who work and those who do not work rely on a variety of these resources to survive. Despite accessing an array of supports, single mothers experience very high levels of material hardship and back-owed debts. Employed mothers are able to draw on slightly greater economic resources as compared to mothers who are not employed, but they still experience extremely high rates of material hardship, suggesting that welfare reform has not effectively “made work pay” for the most disadvantaged single mothers.
In 1996, President Bill Clinton ended “welfare as we know it” and signed into law the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA), or “welfare reform.” PRWORA ended the entitlement to cash welfare for mothers in need, establishing a new cash welfare system (Temporary Assistance for Needy Families [TANF]) with stricter eligibility requirements, including a work requirement (see Ananat et al. 2026 for more details). Since welfare reform, access to basic cash assistance among families below the official federal poverty line has declined significantly from about 60 percent of families in 1994 to only 15 percent of families in 2020, and the magnitude of annualized benefits to recipients has also declined (Bruch et al. 2026). More broadly, federal and state social safety nets have become increasingly conditioned on work and reliant on temporary programs meant to address economic downturns (Bruch et al. 2026).
Even before welfare reform, when basic cash assistance was an entitlement program under Aid to Families with Dependent Children (AFDC), Kathryn Edin and Laura Lein’s seminal book Making Ends Meet (1997), hereafter MEM, showed that cash welfare was insufficient for the single mothers who relied on it, whom they classified as “welfare-reliant.” Welfare-reliant mothers in MEM experienced high levels of material hardship and had to seek out other sources of income, such as off-the-books work or borrowing from friends and family, to survive. Mothers in MEM who did not receive AFDC and instead relied on the low-wage labor market, deemed “wage-reliant,” were even worse off than welfare-reliant mothers, in part because of the additional costs associated with work such as childcare or health insurance. Welfare reform sought to change that reality: to reduce the costs of work and increase the cost of welfare receipt. For example, the legislation decoupled Medicaid eligibility from cash welfare participation, allowing mothers to qualify for Medicaid even if they did not receive cash welfare. Welfare reform also changed childcare assistance policies, including allowing states to use TANF dollars to help fund childcare. Time limits for TANF receipt were put in place along with work requirements. Similarly, the Earned Income Tax Credit (EITC), which had been expanded in 1993 as part of a reform aimed to “make work pay,” provided a cash supplement to low wage working mothers. Yet post-welfare reform, research shows that low-wage workers still rely on a variety of survival strategies to make ends meet, such as borrowing from friends and family or selling blood plasma to supplement employment (for instance, Wu and Eamon 2007; Edin and Shaefer 2015; Ochoa et al. 2021; Teitler et al. 2004), and many face high levels of material hardship (for example, Rodems and Shaefer 2020).1
In this study, we provide a contemporary descriptive portrait of the economic realities faced by low-income single mothers in 2022 and 2023.2 Akin to MEM, we consider the economic resources available to single mothers and the extent to which these resources are sufficient to avoid material deprivation. Our central question is: nearly thirty years post-welfare reform, how are low-income single mothers faring? We compare single working mothers to those who are not working to shed light on whether welfare reform made working low-income mothers better off as compared to their nonworking counterparts. Our analysis does not compare to welfare-reliant mothers as in MEM, as few mothers receive basic cash assistance today: just 10 percent in our sample. We will show that, despite employing many survival strategies, material hardship is common for both mothers who are working and those who are not, though hardship levels are somewhat higher for mothers who are not working. Thirty years after welfare reform, low-income single mothers are still struggling to make ends meet.
We use data from a national monthly survey of mothers (N = 7,186) who are currently receiving or recently received Supplemental Nutrition Assistance Program (SNAP) benefits. To our knowledge, these data are unique in that they both provide data on a large sample of low-income families (incomes less than 130 percent of the federal poverty line) and include extensive measures of economic resources and material deprivation, many more than are typically collected in large study samples. The low-income nature of our study sample (mothers have household earnings of less than $11,000 on average, and one-quarter have no earnings) and the survey questions allow us to compare the strategies used by mothers just before welfare reform in MEM to the approaches used by mothers roughly thirty years later. However, our study differs in a few ways. MEM’s sample comprised 379 single mothers in four cities in the late 1980s and early 1990s, when the historical, policy, and social contexts were quite different from our 2022–2023 survey period. Mothers in our study sample come from all fifty states and live in urban, suburban, and rural areas. Additionally, all mothers in our study are connected to the public safety net, as they are all currently or recently receiving SNAP. Although many wage-reliant mothers in MEM also received safety net benefits (such as an EITC benefit), only 28 percent received SNAP (or food stamps, as the program was called at the time).3 Finally, although similar shares of mothers in our study and in MEM are employed, there is no comparable welfare-reliant group today. For that reason, we compare employed mothers to those who are not working. In most analyses we combine those who report being unemployed with those who are out of the labor force for parsimony, but we will highlight differences between these two groups where relevant. Many of the welfare-reliant mothers in MEM were seeking formal sector employment, essentially using AFDC as a reservation wage (a wage floor, or the minimum wage they would be willing to accept from work), which is no longer feasible under TANF.4
This study is descriptive and cannot assess any causal relationships. Because we rely on a sample of single mothers receiving SNAP, all mothers in our study have low incomes; thus, we cannot speak to changes in the composition of low-income mothers that may have occurred as a result of welfare reform. Labor force participation at the population level for mothers was also very high during our study period, 72.9 percent in 2022 (akin to the early 2000s) and a historic high of 74 percent in 2023 (United States Department of Labor 2024). Additionally, although our study takes place after the most severe part of the COVID-19 pandemic, when infections and deaths were highest, it also follows the expiration of most major pandemic policies, such as the federal eviction moratorium, the monthly Child Tax Credit (CTC), and expanded Unemployment Insurance. However, the national emergency related to the COVID pandemic was not lifted until May 2023, and a few policies, including SNAP Emergency Allotments, remained in place during our study. If these COVID pandemic policies reduced hardships, single mothers like those in our sample would likely experience an even greater average level of hardship today than what we observe in this study. Nonetheless, by focusing on a national sample of low-income mothers, we can document the economic realities and lived experiences of low-income single mothers nearly thirty years after welfare reform and compare the experiences of these mothers to the mothers in MEM.
METHODS
The following section describes our data and measures.
Data
Our data come from repeated cross-sectional monthly surveys administered through Providers, a free mobile application developed by Propel that helps families manage SNAP benefits.5 Approximately one in four SNAP recipients nationally use Providers in any given month (including users in every state), with most discovering the service through their social networks or SNAP caseworkers. The monthly surveys, available in English and Spanish, are offered to a random sample of users (for no compensation), take an average of eleven minutes to complete, and include a variety of questions about respondents’ financial security.
We use data from surveys completed between April 2022 and March 2023 (excluding September 2022, November 2022, and December 2022, as we do not have surveys for these three months). To focus the analysis on low-income mothers, we restrict the dataset to female respondents with at least one coresident child under the age of eighteen. We further restrict our sample to mothers who do not have a coresident partner (dropping those who are cohabiting or married, about 30 percent of mothers); however, in an extension, we retain coresident partners and find substantively similar results with respect to sample descriptives and hardship outcomes, and in online appendix table A.1, we show the means on our outcomes of interest for partnered mothers.6 Our final sample includes 7,186 total observations over the nine survey months.7 In all our analyses, we present the means for the largest sample available; some questions (like those pertaining to tax refunds) were only asked in certain months, and others have a small number of missing responses.8
Unfortunately, Propel does not collect data on response rates. Thus, to assess the extent to which our data compare with other large-scale nationally representative datasets with high response rates, we compare our sample to the 2019 American Community Survey (ACS) in table A.2.9 For comparability, we limit the ACS to households with a female reference person with at least one coresident child under the age of eighteen and no coresident partner. We conduct two analyses, one restricting to households receiving SNAP and one restricting to households below 100 percent of the federal poverty line to demonstrate comparability with households in poverty. In general, we find that the Providers sample characteristics are comparable to the ACS, with some small variations, although we are limited in which characteristics we can compare across datasets. Other studies using the Providers sample have also shown that these data are comparable to other large national datasets, including the Current Population Survey and the SNAP Quality Control data (Pilkauskas, Michelmore, and Kovski 2024; Pilkauskas, Michelmore, Kovski, and Shaefer 2024; Kovski et al. 2023).10 The average Providers user may be more digitally connected than other parents on SNAP, as they are using an app to track their SNAP, but we cannot assess this with the data we have. We note that most respondents are between the ages of twenty-five and forty-five, a group that is generally more digitally literate than older age groups (Vogels and Anderson 2019).
Measures
Employment, Household Earnings, and Cash Resources
To assess the economic resources of the household, we examine maternal employment, total household income from earnings, and an indicator of cash on hand. We categorize respondents as employed, unemployed (not working but seeking work), or not in the labor force (not able to work or not seeking work). We further distinguish between those who work part- versus full-time; the survey does not ascertain number of hours worked, but respondents self-report if their work is part- or full-time. As noted earlier, in most analyses, we compare employed mothers to those who are not employed, combining those who are unemployed with those who are not in the labor force; however, when large differences are present, we report the differences between the two groups.
Respondents were also asked to categorize their overall monthly household income from earnings into one of eight ranges (from “no income” to “$4,000 or more”), which we converted into a continuous measure using range midpoints (top-coded at $4,000, 1 percent of sample). Unlike most studies, which ask about income or earnings in the last year, our study focuses on household earnings in the last month, which may be more appropriate for very low-income samples in which earnings volatility is high.
Last, respondents reported how much total money they had on hand at the time of the survey (excluding SNAP benefits) and how long they thought that money would last. They selected among six ranges for each question (from “less than $25” to “$1,000+” for the former, and from “1–2 days” to “two weeks or more” for the latter), and we converted these into continuous measures using range midpoints. For the money on hand, we top-coded at $1,000, (1.7 percent selected $1,000) and for days money would last, we top-coded at two weeks (4.8 percent selected this option). For a detailed list of the wording of the survey questions, please see the online appendix.
Public Benefit Receipt
Just as was the case in MEM, another key source of support comes from the government safety net. Respondents were asked to report which of the following benefits they were currently receiving: SNAP; the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC); TANF; Supplemental Security Income (SSI) or Social Security Disability Insurance (SSDI) benefits; unemployment benefits (UI); Veterans Affairs (VA) benefits; state health insurance (Medicaid, Medicare, or Children’s Health Insurance Program [CHIP]); Section 8 (rental housing assistance); other housing support; Low Income Home Energy Assistance Program (LIHEAP); and retirement income (Social Security or survivors benefits). We construct binary indicators of receipt for each form of public assistance. We do not know who in the respondent’s household receives which public benefit; for example, an older adult or a child may be the recipient of survivors benefits.
We also consider benefits mothers received through the tax system. The April and May 2022 surveys featured questions about filing taxes for tax year 2021, and the March 2023 survey included measures about filing taxes for tax year 2022. Specifically, respondents were asked whether they filed taxes and whether they received or were receiving a tax refund. The April and May 2022 surveys also included questions about how large the tax refund was (with seven answer options between “$500 or less” and “$10,000+,” converted to a continuous measure top-coded at $10,000) and whether mothers opted for a refund advance (from a for-profit tax preparer such as H&R Block or Liberty Tax). We report the mean tax refund amount among all filers, as well as among those who received a refund.
Private and Charitable Resources—Hardship Avoidance Techniques
Many families tap into other private or charitable resources to make ends meet, or they engage in other activities that might help them avoid the experience of material hardship. Edin and Lein’s open-ended approach to data collection allowed them to capture a wide range of these activities, including income from work that was against the law, such as selling sex or stolen goods. While we do not have data on all sources of income, we do have indicators of whether a mother borrowed money or used credit to make a purchase. Specifically, we constructed seven binary measures from a question asking if respondents had borrowed from family or friends, used a credit card, took a bank loan, took a payday loan or pawnshop loan, borrowed from a church or other community group, took an advance on a paycheck, or started a GoFundMe or other fundraiser. We also constructed three binary measures indicating if respondents reported that they donated plasma, visited a food pantry, or relied on family or friends for meals.
Debt
In addition to considering resources or approaches to avoiding economic hardship, we also study the extent to which mothers report carrying various types of debt. Debt may indicate the ability to borrow or use credit to avoid financial hardships, or it may be an indicator of economic precarity. Respondents were asked about their financial debts, namely: debt to others; rent or mortgage debt; utility debt; credit card debt; and other debt such as student loans, child support, or municipal debt, including parking tickets or fines. Rent or mortgage debt was separated into rent debt and mortgage debt using questions that identified whether mothers were renters or owned their home. Categorical responses (from “$0” to “$5,000+”) were converted into a continuous measure using range midpoints (top-coded at $5,000). We present the mean debt levels by category among the respondents who carry that specific kind of debt. For example, we report the prevalence of rental debt among respondents who pay rent and calculate mean rental debt among respondents who have a nonzero amount of rental debt. These debt measures were then summed to determine respondents’ mean total debt from all sources and the proportion of respondents with any debt. In table A.3, we provide more detail on the debt measures.
Material Hardship
We construct several indices of material hardship to reflect respondents’ well-being in various domains of life. Many of the items included in the survey draw from traditional measures of material hardship, such as those collected by the Survey of Income and Program Participation or the Future of Families and Child Wellbeing Study. For some hardship domains, Propel included further questions related to these traditional measures, such as additional measures of food insecurity. Rather than limit our study to those items that are included in other surveys, we opted to be inclusive and report all of the available measures related to material hardship. We include all related measures in the indices and report Cronbach’s alpha (α), an indicator of measure reliability, and we also report means of the individual items.
Food hardship is based on four items (α = 0.73). These include the two-item food insecurity scale developed by the American Academy of Pediatrics (2015), which asks whether respondents worried that food would run out before they had money to buy more and whether food did not last, as well as two additional questions about skipping meals and eating less. Medical hardship is a three-item index (α = 0.60) that measures whether respondents have foregone needed doctor’s appointments, dentist’s appointments, or prescription medicine. Housing hardship is a six-item index (α = 0.55) that includes questions about staying in a shelter, being evicted, experiencing homelessness, experiencing unstable housing, moving because of an inability to pay one’s rent or mortgage, and living in a doubled-up household. Bill hardship is a three-item index (α = 0.38) that measures whether respondents did not pay their full utilities bills, had their utilities shut off, or decided not to pay a bill. Transportation hardship is a single item measure that captures whether respondents missed work, an appointment, or another event due to lack of transportation. Finally, missing needed items is a single item measure that assesses whether respondents had “everything [they] typically need in [their] home right now,” with answer choices of “do not have most things”; “running low on most things”; “have some things, but not other things”; “have most”; and “have everything.” We created a binary indicator of missing some or most needed household items by combining the first three responses. We then combined all of these indicators into an 18-item overall material hardship index (α = 0.73).
Sample Descriptives
In table 1 we present the descriptive statistics for the sample of mothers. Most respondents are between the ages of twenty-five and forty-four (76 percent), with just 6 percent being below age twenty-four and 18 percent being above age forty-five. The sample is ethnically and racially diverse: 38 percent Black, 32 percent White, and 21 percent Hispanic. Education levels are relatively low; 63 percent have a high school degree or less, 26 percent have some college, 7 percent have an associate degree, and only 3 percent have bachelor’s degrees or more. We have representation from all fifty states and variation in urbanicity and rurality: 48 percent of mothers live in urban areas, 21 percent live in the suburbs, and 31 percent live in rural areas. With respect to living arrangements, 7 percent live with a parent, 7 percent live with siblings or other family members, and 4 percent live with friends or roommates. The average respondent has 2.6 children under age eighteen and lives in a household with 3.9 members.
Sample Descriptive Statistics by Employment Status
As shown in table 1, we also report the characteristics of mothers by employment status. We primarily focus on comparing those who are employed to those who are not employed to mirror MEM. Differences between employed mothers and those who are not employed are relatively small; on average, employed mothers are younger, less White, and more likely to have at least a high school degree.
Table 1 also displays the differences between those who are unemployed and those who are not in the labor force. This highlights several differences. Those not in the labor force are much older; 30 percent are 45 or older as compared to only 10 percent of the unemployed mothers. Mothers who are not in the labor force are also much more likely to be White (44 percent) as compared to unemployed or employed mothers (both 27 percent). Single mothers not in the labor force are less likely to live in urban areas than other mothers but have similar levels of education as mothers who are unemployed. Finally, mothers who are not in the labor force have fewer coresident children, perhaps because their children are more likely to have already grown up. In the remaining analyses we primarily divide our sample between employed and nonemployed (unemployed or out of the labor force) mothers, but we highlight any key differences between those who are unemployed versus not in the labor force and provide detailed breakdowns of all outcomes by group in table A.4.
Results
We begin by considering the various resources mothers have, whether from employment, public programs, friends and family, or other private sources. We then examine mothers’ back-owed debts and the material hardships they experience. We also consider the extent to which access to the public safety net is linked with reduced hardship. Throughout our discussion, we highlight how mothers today compare with mothers making ends meet in the early 1990s as documented in MEM.
All our analyses are descriptive in nature (uncontrolled mean differences). In supplemental analyses (available on request) we examined changes over time for all our outcome measures and found little variation over the one-year period; thus, we focus on annual averages. We also plot trends in the employment and hardship measures over time in figures A.1 and A.2.
How Much Do Mothers Work, Earn, and Have in Cash Resources?
For adults of working age, employment is usually the primary source of economic resources. However, most mothers in our sample are not employed. As shown in figure 1, only 44 percent of Providers mothers are employed, while 28 percent are unemployed and 28 percent are out of the labor force.11 Among mothers who are employed, we find an even split between those who are employed full-time (23 percent) and those who are employed part-time (21 percent). In figure A.1, we show trends in employment status over the year of data and find that levels of employment, unemployment, and not in the labor force are relatively flat (with a small spike in employment in June 2022).
Employment Status and Earnings
Source: Providers study, April 2022 through March 2023.
Note: N = 7,186. Sample is restricted to unpartnered female respondents in households with at least one child under the age of eighteen. Mean (standard deviation) monthly earnings is calculated among employed mothers, N = 3,154.
For context, in table A.5, we provide employment estimates using data from the 2022 ACS. Mothers at the population level have much higher rates of employment than those in our study sample, with 80 percent employed in 2022. Even when we restrict the ACS to mothers who received SNAP in the last twelve months, we find higher rates of employment than in our study sample (67 percent, 45 percent full time and 55 percent part time). Mothers on SNAP in the ACS are slightly less likely to be out of the labor force (25 percent) and much less likely to be unemployed (10 percent) than Providers mothers (both 28 percent). This may in part reflect that most mothers in our study are currently receiving SNAP (about 85 percent report receiving it at the time of the survey), as compared to ACS respondents who received the benefit at some point in the prior year but may or may not be currently receiving SNAP.
Figure 1 also reports the mean monthly earnings among employed mothers in our study. Wage-reliant mothers in MEM earned on average $777 per month, which in today’s dollars is about $1,670 per month (inflated earnings from 1992 to 2023), while average monthly earnings among employed mothers in our sample were about $1,200 per month. This suggests that earnings are about 28 percent lower among working mothers today. Although this comparison is rough, employed mothers in our study earn significantly less than MEM’s wage-reliant mothers, who fared worse than their welfare-reliant counterparts on most hardship measures.
In table 2 we consider monthly household earnings (which can include not only the mother’s earnings but also those of other adults in the household). Monthly household earnings are very low (on average less than $900); 26 percent of mothers report having no household earnings, 19 percent have household earnings of less than $500 per month, and another 19 percent have household earnings between $500 and $999. Only 12 percent of mothers report living in households with monthly earnings of more than $2,000 per month. For context, average household size in our study sample is 4 people, and the 2022 official poverty threshold for a family of four was $27,750 ($30,000 in 2023) (Assistant Secretary for Planning and Evaluation, US Department of Health and Human Services, n.d.). If we assume monthly household income was similar over the prior twelve months, nearly all study participants are living below the poverty line.
Employment, Earnings and Cash Resources by Employment Status
Table 2 also shows monthly household earnings by employment status. Perhaps unsurprisingly, only 4 percent of those who are employed reported no earnings for the last month (which may be the case if a mother only recently became employed) as compared to 44 percent of mothers who are not working.12 Average monthly household earnings are also higher for mothers who are working, about $1,360 per month versus $510 for those who are not working. Approximately 12 percent of mothers in our study live with other family members (their own parents or siblings) who may also contribute to household earnings, which is why this average is slightly higher than their own reported earnings.
The lack of cash resources available to mothers in our study is further illuminated when they were asked about how much money they had on hand at that time. On average, mothers reported having a total of $100 on hand and that the cash they had would last about 3 days. Employed mothers had on average $118 on hand as compared to not employed mothers who had about $33 less—$85 on hand. Together, these figures suggest that the average mother in our sample is living in poverty and has little cash on hand, regardless of employment status. Although Edin and Lein did not ask about cash on hand, the detailed budgets they did collect revealed that after expenses, mothers had little to no cash left over each month, similar to our findings here.
What Public Benefits Did Low-Income Mothers Access?
Low-income mothers accessed many public benefits to help make ends meet. In figure 2, we report the overall share of single mothers accessing different types of benefits on average (panel A) and by employment status (panel B). About 86 percent of mothers in our sample reported receiving SNAP. Although this may seem low given that mothers were recruited from an app that is designed to help families track their SNAP benefits, the app also provides coupons, which may keep former SNAP users on the app, and others may be checking to see if benefits were reinstated. The next most commonly accessed benefit is public health insurance: 58 percent of mothers reported receiving Medicaid, Medicare, or Children’s Health Insurance Program (CHIP) benefits. Unfortunately, we do not know who is covered by the different health plans, whether the mother herself, her child(ren), or both.13 Rates of food stamp and Medicaid receipt were very high among welfare-reliant mothers in MEM (95 percent and 100 percent, respectively), whereas only 28 percent of wage-reliant mothers also received food stamps and few had either Medicaid or employer-provided health insurance. In panel B, we show SNAP and Medicaid receipt by employment status. Akin to MEM, we find higher food stamp usage among mothers who were not working (90 percent) but rates are still very high for those who are employed (81 percent), likely as a result of our sampling approach (recruiting mothers using a food-stamp tracking app). We see similar rates of Medicaid use by employment status, although at lower rates than in MEM.
Public Safety Net
Source: Providers study, April 2022 through March 2023.
Note: Sample is restricted to unpartnered female respondents in households with at least one child under the age of eighteen. N = 7,186 (all); 3,154 (employed); and 4,032 (not employed).
Nineteen percent of mothers accessed WIC and 16 percent accessed SSI or SSDI. When we look at differences by employment status in panel B, we see that those who are not employed are far more likely to receive SSI or SSDI, nearly 24 percent as compared to only 7 percent of employed mothers. Differences are even more striking when we divide the group of mothers who are not employed between those not in the labor force and those who are unemployed. Here we find that 40 percent of mothers who are not in the labor force receive SSI or SSDI as compared to only 8 percent of unemployed mothers (shown in table A.4). Recall that mothers who are not in the labor force are on average older, which may explain higher rates of receipt of disability insurance, although again we do not know who receives SSI or SSDI (the mother or a child) nor can we distinguish SSI from SSDI. (We also find that mothers who are not in the labor force are much more likely to receive retirement income, 5 percent as compared to 1 percent of employed or unemployed mothers.) SSI or SSDI receipt is much higher in our study than in MEM (16 percent versus 6 percent), likely reflecting the fact that SSI payments to both children and adults (ages eighteen to sixty-four) have increased significantly over this time period (Social Security Administration 2023).
In terms of housing assistance, 14 percent of mothers had some form of Section 8 housing assistance (housing choice vouchers or other rental assistance), and another 5 percent received another unspecified form of housing support. When we consider employment status, we see that mothers who are not working are more likely to receive both Section 8 (17 percent) and other housing assistance (6 percent) than mothers who are employed (12 percent and 4 percent, respectively).
As noted earlier, relatively few mothers receive TANF today, only 10 percent. When we separate employed mothers from mothers who were not employed, we find that rates of receipt of TANF are far higher among those who are not employed (15 percent) as compared to those who are employed (only 4 percent). In comparison to MEM, in which 56 percent of study participants received AFDC, rates of cash welfare receipt in our study are far lower. This change is not surprising given the changes to the social safety net under PRWORA, which ended the legal entitlement to assistance determined by need and allowed states to divert the block grant to other uses (Edin and Shaefer 2015). Today, only about 22 percent of TANF dollars go to basic assistance (Center on Budget and Policy Priorities 2022).
Of the remaining public benefits, only 7 percent of mothers (6 percent employed and 9 percent not employed) received Low Income Home Energy Assistance Program (LIHEAP) support, which helps low-income individuals with the payment of utilities. Two percent of mothers reported receiving retirement or survivors benefits and only 0.2 percent received VA benefits. Notably, only 0.6 percent of mothers overall and 1 percent of unemployed mothers received UI. This is striking given that 28 percent of mothers reported being unemployed, indicating that most mothers who were unemployed were unable to access UI.14
Tax benefits or refunds are another key source of income support to low-income families (Hoynes and Schanzenbach 2018; Halpern-Meekin et al. 2016), helping families pay off debts, pay bills, and augment savings (Tach and Sternberg Greene 2014; Mendenhall et al. 2012; Abbott and Tach 2026). In table 3, we show tax-filing behavior and refund receipt during tax season 2022 (April and May) and 2023 (March) by employment status. Our data only include refund amounts for tax season 2022. Overall, we see that 75 percent of single mothers filed taxes in 2022 but only 52 percent filed in 2023. This difference is likely driven by the temporary expansion to the CTC in 2021 which allowed parents to receive a refundable tax credit even if they had little or no earnings, which was not the case in 2022 or before (see Abbott and Tach 2026 for more details).15 There are large differences in tax filing when we look at heterogeneity by employment status. In 2022, 91 percent of employed mothers filed taxes whereas only 62 percent of mothers who were not employed filed taxes. In table A.4, we show differences between those who are not in the labor force and those who are unemployed and find that 75 percent of unemployed mothers filed taxes as compared to only 50 percent of those who are not in the labor force. Differences in tax filing are even more pronounced in 2023; 76 percent of employed mothers and only 33 percent of mothers who were not employed filed taxes. It is important to note that our measure of employment reflects current status, whereas our tax filing measures are about the prior year. For that reason, some currently employed mothers may not file because they did not have work history in the prior year.
Tax Use and Taxes as a Source of Support
Although differences in tax filing are large over time and across employment status, almost all mothers who filed taxes reported getting a refund in both years (93 percent in 2022 and 96 percent in 2023) and differences by employment status are small. Among mothers who received a refund, they got an average of about $5,400 in 2022, with employed mothers receiving on average $1,000 more than those who are not employed. Finally, mothers were asked if they received a tax refund advance, which is a short-term, high-interest loan from a for-profit tax preparer; 31 percent reported that they did so (33 percent of employed and 28 percent of not employed respondents).
What Other Resources Do Mothers Tap into to Make Ends Meet?
Next, we consider sources of private or charitable support, or what Edin and Lein called “economic survival strategies.” Some of these “resources,” like visiting a food bank or selling blood plasma, are stigmatizing and may come at physical or time costs that disincentivize use (Fong et al. 2016; Kissane 2003) and are in contrast to other programs like the EITC that mothers find affirming of their labor (Halpern-Meekin et al. 2016). Nonetheless, these resources likely help families make ends meet and avoid more severe material hardships. These resources are shown in figure 3. Panel A shows private and charitable resources for the full sample of mothers and panel B shows differences by employment status.
Use of Private and Charitable Resources
Source: Providers study, April 2022 through March 2023.
Note: Sample is restricted to unpartnered female respondents in households with at least one child under the age of eighteen (N = 7,186). Sample for plasma question is smaller as this question was not included until May 2022 (N = 6,438 all; 2,823 employed; and 3,615 not employed).
The most common source of private support mothers relied on was borrowing money from friends and family to cover expenses (48 percent). This is comparable to the 46 percent of mothers in MEM who reported doing the same. Some mothers also used credit cards to cover expenses (19 percent), but only 56 percent of mothers reported having a credit card. Even though 86 percent of mothers in our study were receiving SNAP, 20 percent also reported using food pantries and 19 percent reported relying on friends and family for meals. Others covered expenses by taking out payday or pawnshop loans (8 percent), receiving money from churches or charitable groups (4 percent), or taking out other loans (3 percent) or paycheck advances (4 percent). Prior research suggests that low-income individuals donate plasma to make ends meet (Edin and Shaefer 2015; Ochoa et al. 2021). We find that 4 percent of mothers in our sample had donated plasma in the last month, although this question did not ask mothers if they did this specifically to cover expenses; some plasma donation may have been done for nonfinancial reasons.
Differences by employment are shown in panel B. Employed mothers were far more likely to rely on private funding mechanisms (using a credit card, using a pawnshop, taking out a paycheck advance or a bank loan) than mothers who are not employed. Mothers who were not employed were far more likely to rely on charitable support like borrowing from friends and family, relying on them for food, and using a food pantry.
How Are Mothers Faring?
Thus far we have discussed resources, income, public support, and forms of private support that likely help mothers make ends meet. We now consider how mothers fared during this time. We begin by examining their debt levels and then turn to their experience of material hardship. Although debt may be considered a resource, for example a home mortgage that may come with a real asset, the debts mothers reported in our study were largely bad forms of debt, such as back-owed rent or credit card debt. For this reason, we treat these debts as measures of material ill-being.
Panel A in figure 4 shows the share of single mothers with debt by type of debt and panel B shows the share by employment status. In panel C we show the amount of debt among those with positive debt (that is, excluding zeros) and then in panel D we show the same analysis by employment status.16 Seventy-five percent of mothers in our sample pay rent and slightly over half of those mothers (55 percent) have some rent debt, owing an average of $1,385 in back-owed rent. Prevalence of rental debt by employment status does not vary, but mothers who are not employed owe more ($1,307 for employed mothers versus $1,462 for mothers who are not employed). Only 5 percent of Providers mothers pay a mortgage, but similarly, we see that about half of these mothers (51 percent) have back-owed mortgages, owing an average of about $2,000. Here we find that mothers who are not employed (panel B) are far more likely to report a back-owed mortgage (59 percent) than employed mothers (45 percent), but the amount of debt (panel D) is similar conditional on having past due mortgage debt. Debt to others is relatively common (60 percent, owing an average of about $1,100), as is back-owed utility debt (75 percent, owing an average of about $800) and other debt (49 percent, owing an average of nearly $3,000). Rates are similar for mothers who are employed and those who are not employed.
Prevalence and Amount of Debt
Source: Providers study, April 2022 through March 2023.
Note: Sample is restricted to unpartnered female respondents in households with at least one child under the age of eighteen. Rent debt is calculated among renters (75 percent of the sample), and the share with rent debt is among those with that debt (55 percent). Mortgage debt is calculated among those with a mortgage (5 percent of the sample), and credit card debt is calculated among those with a credit card (56 percent of the sample).
Only 56 percent of mothers reported having a credit card (64 percent and 49 percent of employed and not employed mothers, respectively), but among those with credit cards, most (79 percent) have some credit card debt, and they owe about $1,800 on average. Employed mothers are more likely to have credit card debt (82 percent versus 75 percent) but, conditional on having any credit card debt, they have similar levels of debt as mothers who are not employed (both approximately $1,800). If we tally across all forms of debt, 94 percent of the mothers in our study reported having some debt, an average of $4,750 of total debt, nearly two times the debt reported in MEM (about $1,000, or $2,400 in today’s dollars). This may in part be driven by higher access to credit cards and payday loans in the current period than in the late 1980s, when the MEM data were collected, which may allow mothers to accrue more debt. Mean total debt is consistent irrespective of employment status.
In table 4 we turn to mothers’ experiences of material hardship. Material hardships measure the incidence of concrete adversities that are likely to arise as a consequence of insufficient income. We present both summary measures of each form of hardship, as well as detailed breakdowns of the specific hardships experienced. In figure A.2 we show these measures over time. Most of the measures of hardship are relatively consistent over the year, although there appears to be a slight time trend with most hardships (such as food hardship, not paying utilities, and evictions) peaking in October 2022.
Material Hardships
The most common hardship mothers reported was not having needed household items, like household products or cleaning items; 70 percent of mothers (67 percent of employed and 72 percent of not employed) said this was the case. The second most common type of hardship mothers reported is bill hardship (65 percent overall, 62 percent employed, and 66 percent not employed), and in particular not paying utility bills in full (52 percent overall). That so many mothers report both bill hardship and a lack of needed items suggests these mothers have real challenges making ends meet and may reflect the fact that most public assistance programs do not address these needs. For instance, SNAP cannot be used to buy household supplies, diapers, menstrual products or cover over-the-counter medical expenses, meaning families must find other ways to finance those expenses; Edin and Lein found that the cost of these items can constitute a small yet significant share of single mothers’ budgets.17 Although LIHEAP helps some low-income households with heating and utilities, LIHEAP is a block grant and states set their own eligibility criteria.18 Only 7 percent of mothers reported receiving this assistance, and estimates suggest that less than one-fifth of eligible households receive LIHEAP nationally (National Energy Assistance Directors Association 2021). In table 5, we look at levels of hardship experienced by mothers who are receiving various forms of public assistance.19 We find little evidence that mothers who receive LIHEAP are less likely to report utility bill hardships or fewer utility cutoffs than mothers overall, perhaps because experiencing these hardships leads mothers to apply for LIHEAP.
Material Hardships by Public Assistance Receipt, Full Sample
Food hardship was the third most common hardship, with few differences in prevalence by employment status. Sixty-two percent of mothers reported experiencing at least one of the four food hardship measures. Specifically, 43 percent of mothers worried that the food they had would run out before they would have money to buy more, 36 percent said the food they bought did not last, 33 percent ate less, and 26 percent skipped meals.20 In MEM, 28 percent of mothers reported being unable to buy needed food in the previous 12 months, rates that are far lower than for the mothers in our study. The higher rates of food insecurity in our study likely reflect the fact that mothers in our sample are all recipients of food stamps (either currently or in the recent past); if mothers apply for SNAP when they have food needs, that might explain the higher prevalence of food insecurity.21 The higher rates of food insecurity today may also reflect the limited receipt of basic cash assistance as compared to the mothers in MEM.
The high rates of food insecurity are especially notable because two COVID pandemic food-related policies were still in place in many states during our study period: Pandemic Electronic Benefit Transfer (P-EBT) and Emergency Allotments (EA). P-EBT provided additional food assistance to low-income families who qualified for free or reduced-price school meals. Thirty-five states still offered P-EBT during the 2022–2023 school year (and all but two offered P-EBT during summer months). EA allowed states to provide SNAP-eligible households with the maximum SNAP benefit for their household size regardless of income, and although ten states had ended the EA by the time our study began, most still had the policy in place through February 2023, when Congress ended the program (Dasgupta and Plum 2023). Some research finds that the EA and P-EBT reduced food insufficiency (Schanzenbach 2023; Richterman et al. 2023; Bauer 2020); thus, food insecurity would likely have been higher if these policies had not been in place.
About one-third of mothers reported experiencing at least one form of medical hardship (not seeing a doctor, not seeing a dentist, or going without a needed prescription), a rate that was consistent across employment status. Twenty-one percent of mothers in MEM reported that they or their family members were unable to see a doctor when needed, and rates were far higher for wage-reliant mothers (39 percent) as compared to welfare-reliant mothers (7 percent). Wage-reliant MEM mothers had great difficulty securing health insurance from the government or their employers, whereas rates of public health insurance in our sample do not differ by employment status. Lack of needed dental care was especially common in our sample, as nearly one-quarter reported this hardship. In table 5, we see little evidence that having Medicaid, Medicare, and CHIP is linked with fewer medical hardships, but this may in part reflect the idiosyncratic nature of medical needs as compared with, say, food needs; conditional on having a medical need, Medicaid would likely prevent an unmet medical hardship (Currie and Chorniy 2021).
Transportation insecurity is experienced by 28 percent of mothers in our study. Interestingly, this is one of the few hardships where differences by employment status are quite pronounced; 22 percent of employed mothers as compared to 34 percent of mothers who were not employed experience transportation insecurity. No US safety net program directly targets transportation costs, although some states use TANF dollars to pay for reimbursement of work-related travel expenses or for contracting transportation services. Notably, our estimates are similar to those in nationally representative surveys of adults (with more extensive measures of transportation insecurity) that also find that about 25 percent of adults face transportation insecurity (Murphy et al. 2022).
Finally, we find that 25 percent of mothers reported experiencing some form of housing hardship. This indicator of housing hardship does not include traditional measures of housing hardship that often focus on the ability to pay one’s rent or mortgage. Instead, our measures mostly capture more extreme forms of housing hardship, such as homelessness or eviction. Doubling up, another indicator we include, is a common precursor to homelessness (Skobba and Goetz 2014) but is not always considered to be a housing hardship, as families double up for a variety of reasons, not only economic needs (Harvey and Dunifon 2023). Doubling up is significantly less prevalent in our sample (10 percent of respondents) than in MEM (23 percent), which is somewhat surprising given that the share of children living in doubled-up households has increased dramatically since the 1990s (Harvey et al. 2021; Amorim and Pilkauskas 2023). The lower rates of doubling up among Providers mothers may be due to high attachment to safety net programs that disincentivize doubling up (such as SSI, public housing, or SNAP). In table 5 we see doubling up, as well as all other housing hardships, is far less prevalent for mothers receiving Section 8 housing.
When we consider differences by employment status, we find that mothers who are not working have higher rates of housing hardship (29 percent) than employed mothers (21 percent). The main drivers of this difference are higher rates of eviction, homelessness, and unstable housing for mothers who are not employed.22 Nine percent of mothers in MEM reported being evicted in the last year, rates that are comparable to our study. However, our survey question only asks about the last 30 days, whereas MEM asked about the last year. Were eviction rates measured over the year, we would likely find much higher rates in the Providers sample.
CONCLUSION
In this paper we provide a contemporary portrait of the economic realities faced by low-income mothers and document the resources they rely on to make ends meet. Our goals are descriptive, so we make no causal claims. Rather, we aim to describe how low-income single mothers are faring thirty years after the large welfare reforms of the mid-1990s and compare their survival strategies to those of the mothers documented in Edin and Lein’s study of women just before said reforms. To do this, we use a large, nationally representative sample of mothers who are currently receiving or recently received SNAP. By focusing on a low-income population of single mothers, we study a group of mothers who are in many ways like the mothers in MEM. We examine a comprehensive set of economic survival strategies, considering employment, safety net programs (including tax refunds), and private and charitable sources of support (such as food from friends or food banks). These measures are rarely available in large-scale datasets and, to our knowledge, not available in any other large sample studies of low-income families. We also consider whether these resources are sufficient by studying mothers’ experiences of back-owed debts and material hardship.
Single mothers in our study had limited resources from employment. More than half (56 percent) of mothers were not currently working (unemployed or not in the labor force) and 26 percent reported no household earnings from employment. Although employed mothers fared better economically than the mothers who were not employed, average monthly earnings were only $1,200, well below the poverty line ($30,000 in 2023 for a family of four), and household-level earnings were not much higher.
Strikingly, employed mothers in our study earned almost 28 percent less than the average wage-reliant mother in MEM. There are several reasons why this might be the case. Edin and Lein argued that AFDC provided mothers with a wage floor, giving them the ability to wait for a job that paid more than basic cash assistance and to opt for employment that paid slightly better wages, even if those wages did not lead mothers to fare better than the mothers on AFDC. This argument suggests that today, without AFDC, single mothers must opt to take lower paying jobs. This theory is consistent with research findings suggesting that, on average, welfare reform failed to make work pay. Although women moved into the labor force, mothers, and particularly those with the lowest levels of education, fared worse economically, even after accounting for transfers (Ziliak 2016). In addition to welfare reform, the labor market has changed significantly since the 1990s, including increasing wage bifurcation, with jobs at the bottom of the income distribution growing at a faster pace (Autor and Dorn 2013); increasing employment precarity (Kalleberg 2009); and more income volatility (Gennetian et al. 2019). The federal minimum wage has not been increased in 15 years, despite inflation. Together, these trends likely explain the lower earnings of the mothers in our study as compared to 30 years ago. Finally, it could be the case that this difference simply represents differences in samples between MEM and the current study.
Since the early 1990s, the public safety net has also increasingly relied on refundable tax credits like the EITC. Yet even if we include the tax credits with earned household income, we still find that mothers are living well below the poverty line. Overall, we estimate that single mothers in our sample have average household earnings of about $16,000, including $10,600 in earnings and $5,400 in tax credits in 2022, an amount boosted by the temporary CTC expansion and other pandemic policies. Employed mothers fare slightly better, receiving approximately $22,200 ($16,350 + $5,830) in household earnings as compared to $11,000 for mothers who are not employed ($6,125 + $4,850). Although unusually generous tax credits in 2022 added nearly 35 percent to household earnings—helping to make work pay—we find little evidence to suggest that these low-income single mothers are faring any better economically today.
Given low incomes from earnings, mothers unsurprisingly relied on additional support from both the public and private safety nets. Public supports included SNAP, Medicaid, WIC, SSI, and public housing, while private supports included borrowing from friends and family, using food pantries, relying on friends and family for food, and using high-interest credit like pawnshops or payday loans. Overall, employed mothers accessed fewer public safety net benefits than mothers who were not employed (especially SSI and TANF), but they still used many public programs. Additionally, both employed and not employed mothers relied heavily on private safety nets, although mothers who were not employed relied more on charity, friends, and family, whereas employed mothers were more likely to use high-interest loans (credit cards or pawnshops) to make ends meet. Although we characterize the private safety net as a resource, these sources of support can come with future costs, such as repaying loans or gifts or providing assistance later, and can be taxing to family and friend networks (Halpern-Meekin et al. 2015). Similarly, use of charitable support comes with costs, as these sources of support are associated with high levels of stigma in addition to often taking significant time to access (Kissane 2003). Receipt of public benefits is also associated with stigma, although the perception of stigma varies significantly by benefit, while tax benefits are widely seen as less stigmatizing (Halpern-Meekin et al. 2016). We do not have measures of child support receipt, a form of support that played an important role for the mothers in MEM and a resource linked with improved economic outcomes (for example, Pilarz and Cuesta 2025).
We also find no evidence to suggest that mothers today are faring better in terms of back-owed debt and material hardships. In addition to earning less than the wage-reliant mothers in MEM, mothers in our study also have higher levels of debt, roughly twice the amount of debt as the single mothers in MEM. Hardship rates were similar across wage-reliant and welfare-reliant mothers in MEM. Mothers who are employed fare slightly better in terms of back-owed debts and material hardships—but just barely. More than 60 percent of both employed and not employed mothers report having food insecurity, missing needed household items, and an inability to pay bills. About one-third of mothers experience medical hardship and one-quarter face housing and transportation difficulties. To put these estimates in context, nationally representative estimates (pooled data from 1996 to 2011) find that about 13 percent of households with children face food insufficiency, 14 percent face a medical hardship, and about 22 percent are unable to meet expenses, rates that are far lower than those we found in our study (Rodems and Shaefer 2020).
Today, unlike in MEM, we have so few welfare-reliant mothers that direct comparisons are not possible. However, by comparing mothers who are employed to those who were wage-reliant in MEM, we find that employed mothers today are faring worse economically, have higher levels of debt, and have very high rates of material hardship despite relying heavily on both public and private safety nets to make ends meet. Our study suggests that thirty years after welfare reform, at least for highly disadvantaged mothers, welfare reforms did not appear to make work pay.
Our study has some limitations. Although we try to make our study sample as comparable as the mothers in MEM, it is not possible to make the samples the same. Nor can we know the extent to which the composition of low-income mothers has changed over time due to welfare reforms or other policy and market-based changes. In addition, although our study took place after the worst of the COVID pandemic, the pandemic period had not yet been formally ended at the time of our study. Some of the pandemic-era policies were still in place (such as the EA), likely affecting the hardship rates we observe. While our survey was offered in both English and Spanish, we may have missed mothers who were not able to take the survey in one of those two languages. Because mothers were all connected to the SNAP program, our study may represent somewhat more stable low-income mothers, as those who are undocumented or not connected to social services may have been even more vulnerable than the mothers we studied here. Additionally, mothers in our sample are using a smartphone app and thus must have a minimum level of technological savvy. Nonetheless, analyses comparing our sample of mothers to other large, nationally representative samples with high response rates suggest that mothers in our study are highly comparable on the measures we can compare (see Pilkauskas, Michelmore, and Kovski 2024; Pilkauskas, Michelmore, Kovski, and Shaefer 2024).
In sum, we find that nearly thirty years after welfare reform, low-income single mothers are not faring very well. They are merely surviving, not thriving. Despite extensive use of the public safety net and private supports, mothers in our sample still experience extremely high levels of material hardship. Although employed mothers are doing marginally better than those who are not employed, many are still struggling, particularly those who would likely have received AFDC if not for welfare reform. When taken in conjunction with the changes in the labor market and policy landscape that have occurred since, welfare reform and the ensuing work supports do not appear to have successfully improved the material well-being of low-income single mothers.
FOOTNOTES
↵1. Several studies also document the high levels of material hardship among families during the COVID-19 pandemic, especially in 2020 (Karpman et al. 2020; Karpman and Zuckerman, 2021; Bauer 2020; Schanzenbach and Pitts 2020; Shaefer et al. 2020; Enriquez and Goldstein 2020) and 2021 (Horowitz et al. 2021; CBPP 2021; Cooney et al. 2022).
↵2. We use the term single mother to reflect language used in MEM. We exclude mothers who are married or living with a romantic partner. Although single implies unpartnered, we cannot assess whether mothers have nonresident romantic partners.
↵3. Food stamps became SNAP in 2008, and several small changes were made to the program, which resulted in greater take-up of SNAP in the last decade and a half (Bruch et al. 2026).
In addition, there are other minor differences across our studies. For example, Edin and Lein focus on expenditures and on the amount of money received by mothers from various forms of assistance. We do not have these dollar measures in our study. Similarly, Edin and Lein ask mothers about hardships experienced over the last twelve months, whereas our study asks mothers only about the previous month.
↵4. Edin and Lein do not categorize mothers into the formal labor market categories we use in this study (employed, unemployed, not in the labor force) but they note that some welfare-reliant mothers were seeking work, and others worked in the informal labor market. Thus, the welfare-reliant mothers may have included those who we would consider unemployed today, as well as those who were out of the labor force.
↵5. In August 2024 the app Providers was renamed Propel.
↵6. The online appendix for tables A.1 through A.7 and for figures A.1 and A.2 can be found at https://www.rsfjournal.org/content/12/2/57/tab-supplemental.
↵7. Analytic samples for each survey month are as follows: April 2022 (N = 748); May 2022 (N = 767); June 2022 (N = 915); July 2022 (N = 737); August 2022 (N = 952); October 2022 (N = 666); January 2023 (N = 870); February 2023 (N = 826); March 2023 (N = 705).
↵8. Missing responses are driven by individuals not completing the survey, as progression through the survey requires a response (except for skip patterns or qualitative responses). About three-quarters of respondents who start the survey complete the survey. Because demographic information is at the end of the survey, we cannot assess whether those who do not complete the survey differ from those who do, but analyses restricting to only those who complete the survey are substantively similar to the statistics reported here.
↵9. The ACS is a nationally representative survey of the US population collected by the US Census Bureau. Annual samples are approximately three million households. We use data from IPUMS US (Ruggles et al. 2024).
↵10. The Providers data have also been used in other published studies and research briefs (Pilkauskas and Cooney 2021; Michelmore and Pilkauskas 2023; Pilkauskas and Michelmore 2021).
↵11. Forty-four percent of mothers in MEM were employed. We offer this as a point of information, but because Edin and Lein did not randomly sample mothers, we do not know if this is representative of low-income mothers at the time.
↵12. Differences between unemployed mothers and those not in the labor force are small, although those who are not in the labor force have slightly higher monthly household earnings ($430 versus $590, respectively).
↵13. Our question asks respondents “Which benefits do you receive right now?” so we do not know if mothers are reporting about themselves or their children. Estimates suggest that 81 percent of poor children and about 46 percent of poor nonelderly adults receive Medicaid (Burns et al. 2025).
↵14. UI was expanded during the pandemic to increase eligibility, duration of the benefit, and the amount of the benefit, but this ended in fall 2021.
↵15. In March 2021, Congress passed the American Rescue Plan Act (ARPA), which temporarily expanded the CTC by providing a larger benefit to families, extending the benefit to families with no or very low earnings, and delivering half the credit as a monthly stipend for six months (July through December 2021) and the other half of the credit at tax time (February through April 2022). ARPA also expanded the Child and Dependent Care Credit (CDCC), a tax credit to help defray the costs of childcare or caring for a family member, and it was made significantly more generous and refundable (as it is not normally a refundable credit). Thus, refunds during tax season 2022 were higher than normal given these two credits, as well as the ability to claim any Economic Impact Payments (stimulus) that mothers had not received.
↵16. Greater detail on debt is available in table A.3.
↵17. Research on the Providers sample that asked parents to report on how they were spending the CTC found that about 5 percent of parents reported using the CTC for household items like toiletries and cleaning supplies, and this was even more prevalent among those with no earnings (9 percent) (Pilkauskas, Michelmore, Kovski, and Shaefer 2024; Michelmore and Pilkauskas 2023).
↵18. The ARPA significantly increased funding for LIHEAP, which ended in September 2022.
↵19. In table A.6, we show hardship by public safety net program use and by employment status. In table A.7, we show the rates of hardship by use of the private and charitable safety net. We find that those who are using these safety nets (like food from friends) are faring far worse in terms of material hardship than mothers in our study overall.
↵20. Note that when we instead use the two-item food insecurity scale commonly used in research and endorsed by the American Academy of Pediatrics (2015), we still find that 45 percent of mothers experienced food insecurity. Figure A.2 shows food hardships over time, and we see the lowest rates around tax season (March and April).
↵21. Look-back periods can affect recall on food insecurity, which may also explain differences between mothers in MEM and Providers mothers (Livings et al. 2023).
↵22. Federal eviction moratoria ended in October 2021, but many states and localities continued to have moratoria in place long after the end of the federal moratorium. Another pandemic policy was the COVID Emergency Rental Assistance (ERA) program to help low-income renters pay their rent and utility bills. The ERA program allowed states to provide assistance through September 2025; however, many states closed their ERA programs to new applicants in 2022 as they ran out of funding for the program (National Council of State Housing Agencies, 2024).
- © 2026 Russell Sage Foundation. Pilkauskas, Natasha V., and Kevin Bruey. 2026. “Making Ends Meet Thirty Years Later: How Single Mothers Survive on Low Incomes.” RSF: The Russell Sage Foundation Journal of the Social Sciences 12(2): 57–82. https://doi.org/10.7758/RSF.2026.12.2.03. The authors thank Propel for access to their Providers data. We thank Samiul Jubaed for his assistance in preparing the data and Nicole Kovski and Katherine Michelmore for help with data comparisons. We thank Kathy Edin for her extensive feedback, as well as the editors of this double issue and the conference participants. Direct correspondence to: Natasha Pilkauskas, at npilkaus{at}umich.edu, 735 South State Street, Ann Arbor, MI 48109, United States.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
REFERENCES
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.