Abstract
Given changing employment conditions and childcare policy investments since Making Ends Meet, Kathryn Edin and Laura Lein’s 1997 book, this study explores the childcare arrangements of low-income parents of young children working nontraditional hours and unpredictable work schedules. Using a 2019 nationally representative sample of low-income parents with young children, we analyze nontraditional hour (NTH) childcare usage patterns by family structure, race and ethnicity, and immigration status. Our findings show that fewer than half of the children use NTH childcare, with unpaid family care being the most common arrangement. Family structure significantly influences NTH childcare utilization, and care patterns by race and ethnicity align with those seen in daytime care. Additionally, we found that childcare subsidy use is low overall among parents working NTH; but subsidy generosity is associated with higher rates of NTH paid family childcare. These results underscore the need to expand childcare options during nontraditional hours and increase access to subsidized NTH care for families with low incomes.
- nonstandard hours
- nontraditional hour work
- nontraditional hour childcare
- precarious work
- family structure
- childcare subsidy
- National Survey of Early Care and Education (NSECE)
Almost three decades after the publication of Kathryn Edin and Laura Lein’s classic book, Making Ends Meet (1997), and the passage of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA), the number of low-income families receiving cash welfare assistance in the United States has declined precipitously, single mothers’ employment has risen, and government spending on subsidized childcare has increased. Yet, working families, especially the single-parent families of focus in Making Ends Meet, continue to experience economic precarity and struggle to find stable, affordable childcare to meet caregiving needs and support children’s healthy development. With the increased childcare, transportation, and clothing expenses related to paid employment, the costs of working can offset the benefits of employment. This was true in the 1990s, and it remains so in 2025.
The persistent difficulty of combining employment with caregiving is due in large part to broader labor market realities. Economic productivity has outpaced compensation gains for low-income workers over these three decades, and job quality has deteriorated, characterized by stagnant wages, few fringe benefits, and a growth in precarious employment (Howell and Kalleberg 2019; Henly et al. 2021). In a post-welfare-reform era, low-income mothers are pushed to accept low-paying, low-quality jobs, with uncertain job security and work schedules that conflict with caregiving responsibilities. Since Harriet Presser first wrote about the prevalence of nontraditional work hours and the challenges they pose for families with young children (Presser and Cox 1997; Presser 2003), these schedules have become more concentrated and increasingly involuntary among disadvantaged groups—such as single parents, workers of color, and those with low education or income (Lou et al. 2022; Pilarz and Walther 2025).
At the same time, and despite expanded public childcare support following welfare reform (Magnuson et al. 2007), the childcare infrastructure remains fragmented, expensive, and ill-equipped to address the needs of low-income working families, particularly those seeking care during NTH or to meet unpredictable work schedules. Families with NTH and precarious schedules rely disproportionately on informal childcare with family, friends, and neighbors (FFN) or go without nonparental care altogether (Coley et al. 2014; Henly and Lambert 2005). Although the prevalence of centers and preschool programs has increased over the past three decades (Chaudry et al. 2021), these programs typically operate during daytime, weekday hours (NSECE Project Team 2015). Licensed home-based programs offer somewhat more schedule flexibility than centers; however, their numbers have declined in recent years. Importantly, fewer than one-fifth of eligible families receive financial assistance from the Child Care Development Fund (CCDF), the primary federal childcare subsidy program in the United States (Chien 2022). Over 70 percent of these families are served by centers, primarily during daytime, weekday hours (Office of Child Care 2025). Thus, there is a misalignment between the low-wage labor market, the childcare market, and the work-contingent welfare system in the United States.
The labor market and childcare experiences of low-income families since Making Ends Meet have also been shaped by family circumstances. Household arrangements—including single-, two-parent, and extended-household structures—may influence the kinds of childcare that parents use to meet work demands, especially during NTH. With the increasing prevalence of single parenthood, which may have reduced the availability of family support from within the household, there has also been a steady increase in extended-household structures, especially among economically disadvantaged households of color (Cross 2018). Extended household members may serve as childcare resources, depending on economic constraints and needs, familial expectations, cultural practices, and caregiving alternatives.
Drawing from a 2019 nationally representative sample of households with children under thirteen years of age, this study examines the childcare arrangements of low-income families who work during NTH with at least one child under five years of age. We conduct all analyses by household composition, including consideration of the presence of nonparental adults in the households of single- and two-parent families. We also examine differences in the use of NTH childcare across racial and ethnic groups, as well as immigration status. These demographic factors are related to labor market behavior and family formation (Landale et al. 2011; Smith 2006) and are known to shape parents’ childcare decision-making (Radey and Brewster 2007; Brandon 2004). Finally, we consider whether the generosity of state childcare subsidy programs is related to NTH childcare decisions. Ultimately, this study aims to advance knowledge about how low-income families working NTH—a key feature of today’s labor market—are meeting their childcare needs and how family structure and childcare subsidy policy shape childcare arrangements for families across race and ethnicity and immigration status. The study highlights the need for targeted policy interventions to improve access to affordable childcare during NTH and to promote the economic stability of low-income families.
LABOR MARKET SINCE MAKING ENDS MEET
In the late 1990s, low-income mothers increased their engagement in formal employment, although the labor market they entered provided limited opportunity for economic stability or advancement. In fact, for almost a half-century, across strong and slack economic periods, employment has become increasingly precarious for workers at the lower end of the labor market (Howell and Kalleberg 2019; Henly et al. 2021). The economic recovery after the Great Recession, from 2007 to 2009, was slow to generate even modest wage increases for low-skilled employees (Groshen and Holzer 2021), and employers have continued to embrace business models that favor cost containment and flexible management practices (Howell and Kalleberg 2019; Lambert 2008). As a result, workers, especially those employed in hourly, low-paying jobs, often have limited control over when or how much they work. They experience unpredictable work schedules and fluctuating work hours that are associated with income volatility, caregiving challenges, and poor worker health and well-being (Lambert et al. 2019; Schneider and Harknett 2019; Ananat et al. 2022; Finnigan 2018). While these practices provide flexibility for employers, they represent instability for workers (Lambert 2008).
Nontraditional work schedules—work that takes place outside of daytime, weekday hours—are a common feature of employment in today’s economy. Some working parents (especially dual-earner couples) seek jobs with nontraditional schedules as a strategy to share caregiving and avoid nonparental childcare. However, research indicates that most low-income mothers work nontraditional schedules involuntarily, especially single mothers (Presser 2003; Presser and Cox 1997). National census data indicate that approximately one-third of young children reside in households with parents employed during NTH (Lou et al. 2022). Nontraditional work schedules are more prevalent among workers of color, single mothers, mothers with young children, and workers employed in leisure and hospitality, retail trades, and other service and laborer positions (Lou et al. 2022; Presser 2003). New research by Alejandra Pilarz and Anna Walther (2025) that examines the prevalence of NTH work schedules from 1988 to 2019 finds that White and married mothers and mothers with a college degree have become less likely to work during nontraditional hours, whereas the prevalence of NTH work for Black and unmarried mothers and mothers without a college degree has either stayed the same or increased during this period. They also find that involuntary NTH work has increased, especially among workers of color and single mothers.
The prevalence of remote work arrangements, specifically working from home, has risen steadily over time, increasing from less than 0.5 percent in 1965 to 7 percent in 2019, although it is less common in low-wage occupations (Leith 2024). The rise in remote work can be attributed to technological advancements, cost-saving strategies, and growing public interest in work arrangements that promote work-family balance (Felstead and Henseke 2017; Sullivan 2012). For parents with NTH work schedules, remote work may provide the flexibility to attend to children during times when nonparental care options are limited. However, research addressing remote work for NTH-working parents is scarce, and the general literature on remote work yields conflicting results. On the one hand, some research finds that remote work can reduce work-family conflict, particularly for mothers (Sherman 2020). However, other studies suggest that blurring work-and-care boundaries through remote work can create work-family conflict (Glavin and Schieman 2012), especially for women (Kim et al. 2019). The surge in remote work during the COVID-19 pandemic left many working mothers overwhelmed by work-care demands (Del Boca et al. 2020; Dunatchik et al. 2021).
The current study is unique in its focus on NTH work schedules while also incorporating other aspects of job quality: schedule (un)predictability and remote work. NTH-working parents may be especially sensitive to these job characteristics, and they have received increased attention since Making Ends Meet.
FRAMEWORKS OF CHILDCARE DECISION-MAKING
According to a consumer choice framework, parents make childcare choices to maximize satisfaction on the basis of preferences regarding price, quality, convenience, familiarity, and so forth, subject to budget constraints (Blau 2001; Tekin 2005). From this perspective, nontraditional work schedules constrain childcare options; indeed, there is a dearth of childcare supply during NTH. An accommodation framework (Meyers and Jordan 2006) also views NTH decisions as constrained. It asserts that decisions are shaped by family resources, social interactions, and heuristic processes as parents navigate the daily stressors of their lives (Meyers and Jordan 2006; Chaudry et al. 2010). Additionally, an accommodation perspective assumes multiple motivations drive decisions, beyond just utility maximization.
Family preferences are considered in both consumer choice and accommodation frameworks. Research on parental preferences during NTH is limited, but some studies suggest that parents prefer FFN providers, especially when care is needed overnight and on weekends and when a child is already in a center arrangement during daytime, weekday hours (Schilder et al. 2022; Stoll et al. 2015). These preferences may be shaped by the unique demands of NTH care. For example, NTH care may overlap with mealtimes, bedtimes, and social activities. Parents may seek providers who have prior relationships with their children and who are familiar with their family routines and values (Bromer et al. 2024). Consequently, family members in the household may be preferred childcare resources when care is needed during NTH. Overall, a complex set of interrelated contextual, interpersonal, and individual factors is at play and is intrinsic to childcare decision-making. Although important as background information, our interest is in assessing the associative (not causal) link between NTH employment and childcare arrangements.
CHILDCARE DURING NTH
Children are cared for in a range of settings when parents work. Some families use center-based arrangements, prekindergarten programs, and nursery schools. Others use home-based arrangements, such as licensed Family Child Care (FCC) homes or informal, license-exempt FFN settings.1 National data indicate that there is a dearth of formal childcare programs available to families needing care to meet variable schedules or NTH hours, especially during evenings, weekends, and overnight (Stoll and Lara 2020).2 Formal home-based programs are more likely than centers to provide care during these hours, although it is informal FFN caregivers who provide the bulk of care during nontraditional hours (NSECE Project Team 2015).3 Previous research on the caregiving arrangements of families with nontraditional work schedules indicates that parents who work outside of daytime, weekday hours use less nonparental childcare overall than families with daytime, weekday schedules (Coley et al. 2014). Among families who do use nonparental childcare, studies consistently observe a higher likelihood of home-based care, especially informal or relative care, among parents who work nontraditional hours compared with traditional work hours (Han 2004; Coley et al. 2014). For instance, Wen-Jui Han (2004) investigated the associations between nontraditional work schedules and childcare decisions among a national sample of mothers with children under age three and finds that the most prevalent type of NTH care was paternal care, followed by relative and family childcare.
More recently, researchers have leveraged novel insights from calendar data to capture the childcare arrangements of working parents at different times of day and days of the week. Utilizing calendar data from the 2012 National Survey of Early Care and Education (NSECE), Lisa Gennetian and colleagues (2019) find important income differences: low-income children who use nonparental childcare spend a substantially greater proportion of those hours in care during NTH compared with higher income children (35 percent versus 15 percent). They also find limited overlap between center-based care hours and parental work hours—and this is especially true for low-income families working NTH. Relatedly, Alejandra Pilarz and colleagues (2019) find that the elevated use of center care by higher-income compared with lower-income families can be partially explained by parental employment characteristics, especially NTH. Thus, whereas both lower- and higher-income parents use centers even when they do not align with work schedules, NTH work schedules nevertheless depress the use of centers, especially for low-income children (Pilarz et al. 2019).
In contrast to these studies, we directly examine the types of care that low-income parents use during NTH working times. Understanding the alignment between work and childcare hours is important for understanding whether families are using a particular childcare arrangement to support their employment responsibilities or for some other purpose. Whereas some children whose parents work NTH benefit from daytime early learning environments, parents still need access to caregiving arrangements for their children when they are working. The work-support function of childcare motivates the current study.
FAMILY STRUCTURE AND THE CHILDCARE ARRANGEMENTS OF NTH-WORKING FAMILIES
Research suggests that family structure shapes the childcare arrangements parents use during NTH. Married and cohabiting couples may use tag team parenting for some NTH childcare needs (Kim 2022), while nonparental adults residing in extended households frequently serve as NTH childcare providers (Fuller-Thomson and Minkler 2001). In contrast, single parents living alone may lack these in-home resources and rely more on outside caregivers (National Center for Education Statistics 2021). In the earlier referenced Han (2004) study, children with married mothers who worked NTH were more likely to be cared for by their fathers or mothers, whereas children in single-parent households more often use relative care. Furthermore, Han demonstrates that the transition of married mothers from traditional to nontraditional work hours corresponds with a shift toward paternal care, whereas unmarried mothers transitioning to nontraditional hours use nonparental caregivers, especially relatives.
Analyzing the 2012 NSECE calendar data and restricting the sample to low-income families, Peter Hepburn (2018) considers three kinds of childcare—nonrelative paid home-based, relative (paid or unpaid), and center-based arrangements—and finds that single-parent households use significantly more hours of any nonparental childcare than two-parent households overall, and especially more hours of relative care. Relative care is marginally more prevalent among single parents with NTH work compared with single parents with traditional work hours, and single parents working NTH use less paid home-based care than their counterparts working traditional hours. For two-parent households, the association between NTH work schedules and childcare type is much less pronounced, likely because a second parent cares for the child during a partner’s working hours.
Christina Cross (2018) demonstrates that extended households have steadily increased from 13 percent in 1996 to 17 percent in 2014 and are even more common in low-income populations of color (Cross 2018). Extended households are an important source of economic and social support, including childcare, and especially so in economically disadvantaged families and families of color (Cross 2018; Pilkauskas 2012). Parents may rely on nonparental adults with whom they live to provide unpaid childcare, either as part of a strategy to share resources and responsibilities within the family or as a normative cultural practice. In the current study, we contribute to the literature by examining how family structure (both single- versus two-parent households and nuclear versus extended households) shapes childcare arrangements for our sample of low-income NTH-working parents and whether these relationships are moderated by race, ethnicity, and immigration status.
CHILDCARE SUBSIDIES
Childcare is a significant household expense, on par with rental housing costs in the one hundred largest metropolitan areas (Poyatzis and Livingston 2024). As a proportion of household income, childcare expenses for one child alone make up between 8.9 percent and 16.0 percent of median income, with the financial burden on lower-income families with care expenses being substantially greater than that of their higher-income counterparts (Payatzis and Livingston 2024). Tuition varies greatly by geography, age of child, and type of provider, with private centers charging higher prices than family childcare homes. Some FFN providers, including relatives and nonrelative caregivers, charge tuition, while others provide care without requiring financial remuneration (Powell et al. 2023).
Established as part of welfare reform legislation in 1996, the CCDF is the main federal program subsidizing the childcare expenses of low-income working families. Unlike subsidized early education programs, such as federal Head Start and state and municipal public prekindergarten programs, CCDF subsidies serve low-income parents employed or participating in a qualifying education or training program. In 2019, CCDF subsidies served almost 1.4 million children nationally. This number increased substantially in the first decade after welfare reform; however, program participation has declined steadily since its height in 2006, when 1.8 million children received CCDF childcare assistance. Higher participation in the early years of the CCDF program tracked funding expansions during that time. However, federal funding leveled off at $5 billion by 2001 and stayed relatively stable through 2018, except for a temporary two-year increase from the American Recovery and Reinvestment Act of 2009. Most states supplement federal contributions to their CCDF programs with general revenue and Temporary Assistance for Needy Families (TANF) block grant dollars. However, these state contributions have not been sufficient to offset the flattening of federal funding or the rise in subsidized center-based care. In 2018, CCDF received $2.37 billion in additional federal dollars, and during the COVID-19 pandemic, states received several billion dollars in one-time pandemic stimulus relief payments through the Child Care Development Block Grant as part of the Coronavirus Aid, Relief, and Economic Security Act (CARES Act) and the American Rescue Plan Act of 2021 (First Five Years Fund 2021).
The federal government sets broad CCDF regulatory guidelines, but states have discretion regarding eligibility criteria and administrative processes. Since CCDF’s inception, the federal government has recommended that states reimburse participating providers at 75 percent of the private market rate. However, most states reimburse considerably less, with only four states meeting the 75th-percentile threshold in 2019 (Schulman 2019). Moreover, nationwide, an average of 11 percent of eligible families receive childcare subsidies, and this too varies considerably across states, ranging from 6 to 21 percent (authors’ own calculation).
Alejandra Pilarz (2018) finds that subsidy generosity is associated with an increased likelihood of using center-based care, and a decreased likelihood of using multiple arrangements. Roberta Weber and colleagues (2014) observe that an expansion of subsidy generosity in Oregon led to an increase in the use of center-based care. We might not expect subsidies to increase center use during NTH, given the limited supply of centers offering care during evenings, weekends, and overnight hours. Subsidies may be associated with greater center use for NTH-working families who need early morning care to supplement a daytime center-based arrangement. Moreover, subsidies can reduce the cost of licensed FCC during NTH for low-income families, while also enabling families to pay FFN providers who might otherwise not receive regular payments for their labor. Overall, we know little about whether or if the parameters of state subsidy policies are associated with childcare arrangements among NTH workers. The current study aims to address that void.
We address three primary research questions. First, we explore descriptively the types of childcare arrangements that low-income families who need care during NTH hours use and how NTH care arrangements vary by family structure, race, ethnicity, and immigration status. Second, we assess these relationships in a multivariate regression framwork, controlling for a set of child, family, and community characteristics. Third, using a similar model, we examine whether low-income children have a higher probability of using paid caregivers during NTH in states that have established more generous childcare subsidy programs since welfare reform.
DATA AND METHODS
We first describe the data used to examine childcare arrangements during NTH. We then detail the construction of our primary measures, including parents’ NTH employment, NTH childcare arrangements, family structure, demographic characteristics, and childcare subsidy policy.
Data
We use calendar data and restricted-use household data from the 2019 NSECE. The NSECE is a repeated cross-sectional survey that is nationally representative of households with children under the age of thirteen. We leverage the NSECE calendar data to obtain temporal information on both parents’ employment and children’s care arrangements on a 15-minute basis in a given week. This enables us to comprehensively capture the alignment of parental employment and childcare during NTH. Additionally, the calendar data allow us to distinguish between arrangements used at different times during NTH, including early morning, evening, overnight, and weekends. For analytic purposes, we construct a child-level dataset containing calendar information on childcare arrangements and parental employment, as well as children’s demographic and community characteristics. We define our sample as children with at least one parent working during NTH and household incomes below 200 percent of the 2018 federal poverty level (FPL). An income limit of 200 percent FPL was chosen to target our study to families experiencing economic disadvantage. Low-income parents often have fewer job options, which may lead them to accept NTH jobs, and they also face more limited financial resources for nonparental childcare. Additionally, we aimed to capture families likely to be eligible for childcare subsidies so as to test our third research question.4 We exclude missing values on covariates (100 cases).5 Consequently, the final sample consists of approximately 1,600 low-income children under age five for whom at least one parent works during NTH.
Parental NTH Employment
Following Gennetian and colleagues (2019), we define parental NTH employment to include work-related activities such as work, school, training, and commuting time to and from work or school destinations. We construct a dichotomous variable for each parent indicating whether they worked during NTH.
Parental Employment Characteristics
Following other studies (for example, Gennetian et al. 2019) and the NSECE definition, we define NTH as hours that fall outside of 8 a.m. to 6 p.m. on weekdays and anytime on weekends.6 We further define early-morning weekday hours as 5 a.m. to 8 a.m. and evening weekday hours as 6 p.m. to 1 a.m. Overnight weekday hours extend from 1 a.m. to 5 a.m. Weekends are defined as Friday midnight to 5 a.m. Monday. In addition, we construct a measure of the proportion of NTH when all parents work. Increased proportions suggest the household will have fewer available parental childcare resources during NTH.
We include a measure of work schedule predictability. The NSECE questionnaire asks, “How far in advance do you usually know what days and hours you will need to work?” Response categories include one week or less, between one and two weeks, between three and four weeks, and four weeks or more. We classify individuals who respond three weeks in advance or more as having greater predictability. We construct a composite variable reflecting parents’ schedule predictability as a group, indicating little predictability at work if both parents (or one parent for single-parent households) have less predictability. One limitation of this variable is that the question in the NSECE is a general employment inquiry, not limited to work during NTH.
A measure of remote work is not available in the NSECE, so we utilize parents’ occupational information as a proxy for remote work availability. Utilizing the American Time Use Survey data from 2017 and 2018, Titan Alon and colleagues (2020) computed the proportion of individuals in each occupation reporting that they were able to work remotely. Consistent with previous literature (Collins et al. 2021; Kwon 2023), we define an occupation as one where employees have the option to work remotely if at least half of the respondents were able to work from home.7 Similar to the schedule predictability variable, we consider both parents (or one parent in single-parent households) not being eligible for remote work as indicating the absence of remote-work availability.
Childcare Arrangements
We create an indicator that equals 1 if the child uses any NTH care.8 Subsequently, we distinguish between paid and unpaid NTH care. Paid care includes any childcare involving out-of-pocket expenses by parents as well as subsidies. The NSECE classifies childcare categories by hours of care per week and the type of care provider. First, it distinguishes regular care (an arrangement that is at least five hours per week) from irregular care (less than five hours per week). Within regular care, the NSECE classifies arrangements as center-based, paid individual providers, and unpaid individual providers. Center-based care includes Head Start, public prekindergarten, community-based childcare organizations, and any other non-home-based care that is not drop-in or single-activity. Paid and unpaid individual providers are referred to in this study as family childcare (FCC) providers. The NSECE further breaks down paid individual providers by whether parents have a prior relationship with the provider. Although there might be important differences between paid providers with and without a prior relationship, in this analysis, we combine paid individual providers with and without a prior relationship due to sample size constraints. Consequently, a paid FCC provider with no prior relationship, as well as those with a previous relationship, such as relatives, neighbors, and friends, are collectively categorized as paid individual providers. We label this type of care “paid FCC.” Similarly, unpaid individual providers may or may not have a prior relationship with the parent, although our descriptive analysis indicates these arrangements are primarily with relatives. We label this type of care “unpaid FCC.” We create three indicators to identify whether the child utilizes each of the three types of regular care during NTH. We also include an Other category, comprising (1) regular childcare, which refers to care provided for at least five hours per week, that does not fall under center- or home-based care, such as drop-in programs or single activities, and (2) irregular childcare, which refers to care provided for less than five hours per week.
Family Structure
We examine two dimensions of family structure. The first dimension pertains to single parenthood, for which we create an indicator of whether the parent lives with a spouse or a cohabiting partner.9 The second dimension is the presence of extended-household members. We rely on information about the parental or guardianship status of each adult in the household and the presence of grandparents in the NSECE, and categorize children into two groups: those in extended and nuclear families. Due to data limitations, we cannot determine the specific identity of each nonparent adult household member. We assume that whether or not these nonparent adult household members are related, they all have the potential to serve as childcare providers. Among the roughly 6,500 children under the age of five in the NSECE dataset, approximately 5 percent of children live in alternative family structures, including grandparent-only families, multiple-parent families, and families with unknown types. In our main analysis, we exclude these children to mitigate confounds that might arise from the interplay between family complexities, childcare decision-making, and parental labor supply. Given statistical power considerations, we use two separate family structure variables, one indicating single versus two parents and another for the presence of other adults, rather than a single four-category variable.
Race, Ethnicity, and Immigration Status
We create a four-category variable for the child’s race-ethnicity: White, Black, Latino/a, and Other.10 Given the mix of race or ethnic membership in the Other group and its small sample size (11 percent), we refrain from interpreting Other coefficients. Parental immigration status is defined as at least one parent born outside of the United States.
Childcare Subsidy Policy
Prior research uses state income-eligibility thresholds as a proxy for generosity (Ha and Ybarra 2018). In our measure, we capture both eligibility thresholds (that is, who can receive childcare assistance) and coverage (that is, who is actually receiving it), reflecting a multidimensional aspect of welfare generosity (Van Hootegem et al. 2024). We calculate the share of children receiving childcare subsidies among those eligible under the federal thresholds in 2019.11 The generosity measure ranges from 6–21 percent, with a mean and median of 11 percent.12 We categorize states into two groups by their levels of generosity, using a cutoff of 8 percent. States with a generosity level above 8 percent are classified as more generous, while those at 8 percent or below are classified as less generous.
Control Variables
We include a set of child and family and community characteristics as covariates derived from prior research (Hepburn 2018; Morrissey 2008; Pilarz et al. 2019). The child and family controls include child age in months, child gender, parental age, maternal educational levels (high school graduate or below, some college, and bachelor’s degree or higher). We also include the presence of nearby relatives who can provide childcare, the presence of two or more children under age five in the household, the presence of at least one school-age child, the presence of at least one teenage child, and urbanicity. Community variables include the following: a composite variable representing the proportions of individuals living at or below the federal poverty level within the community they inhabit. In the context of the NSECE, community refers to the census tract where each household was sampled, as well as all adjacent census tracts located within a two-mile radius. We include state minimum wage and unemployment rates in 2019, using data sourced from the University of Kentucky Center for Poverty Research National Welfare Data. Additionally, we consider state Head Start and state prekindergarten spending, which may influence childcare decisions.13 Here, we construct a measure of program funding per child for each policy by dividing the state-level funding amount by the population of age-eligible children in each state. See online appendix table A.1.14
EMPIRICAL APPROACH
We employ a logistic regression approach to examine the likelihoods of using various types of care, including any type of childcare during NTH among the full sample (model A), a paid care arrangement during NTH among the full sample (model B),15 and three different types of regular childcare arrangements (center-based, paid FCC, and unpaid FCC) among the subset of respondents who use at least some NTH care (model C).16 We sequentially estimate four sets of models for each dependent variable. The first model includes only controls for child and family demographic characteristics. The second model adds parents’ employment characteristics. The third model adds a set of controls for community characteristics to account for macroeconomic and policy environments. The fourth model tests whether the associations between family structure and the use of NTH care differ by race and ethnicity, incorporating interactions between family structure and race and ethnicity. We also run the final model with an alternative interaction of family structure and immigration status.17
We add subsidy generosity into the models to answer our last research question. We first introduce an indicator for more generous states to examine the main effects of subsidy generosity in the use of NTH care (model 1), and then we incorporate two interaction terms between subsidy policy parameters and family structure (single parenthood and extended family, model 2). For all subsidy analyses, we restrict our dependent variables to paid childcare and the type of childcare utilized during NTH, given our primary interest in the use of childcare that can be purchased through the subsidy program. Given power limitations, we do not examine three-way interactions by subsidy generosity, family structure, and race.
RESULTS
We first present descriptive results on the usage patterns of NTH care arrangements among low-income families who work during NTH. We then present our main findings from the logistic regression analyses.
Descriptive Results
Table 1 reports statistics for the main variables in our multivariate models (family structure, subsidy generosity, race and ethnicity, immigration status, and job characteristics) and our outcomes related to NTH childcare arrangements. About half of the children live in two-parent households without other adults, whereas about a quarter come from two-parent, extended households. Single-parent, nuclear households account for 13 percent, followed by single-parent, extended households at 11 percent. Over 85 percent of families live in states where the share of children receiving childcare subsidies (among those eligible under the federal thresholds in 2019) is above 8 percent. See online appendix figure A.1 for detailed state breakdown.
Weighted Descriptive Statistics on Key Measures
Also reported in table 1, the sample consists of 40 percent White children, 32 percent Latino/a children, and 16 percent Black children, with the remainder comprising other racial and ethnic groups. Approximately 27 percent of the children come from immigrant families. Black children represent more than a quarter of children in single-parent households, whereas they account for around 10–18 percent in two-parent households. Latino/a children are most likely to be found in extended households, regardless of whether they are in single-parent or two-parent families. Both two-parent households and extended households are more likely to have immigrant backgrounds compared to single-parent and nuclear households, respectively. This suggests that the living arrangements of immigrant families compared to nonimmigrant families may be more adaptable to NTH employment since extended family members or spouses may be available to provide childcare.18
Parental employment characteristics that are indicative of job quality vary significantly by family structure (see table 1). Over 90 percent of two-parent households report having some schedule predictability at work, whereas only about 60 percent of single parents experience the same level of predictability. Additionally, nearly all single parents lack remote work options, whereas more than half of two-parent households are estimated to benefit from such flexibility. Descriptive statistics for all control variables are in the online appendix table A.1.
Patterns of NTH Care Utilization Overall and by Family Structure
Table 1 also reports descriptive statistics for NTH childcare for the full sample and by family structure. Among the full sample of low-income children under age five with at least one parent working during NTH, slightly more than two-fifths (43 percent) use any form of NTH care. About 15 percent utilize paid NTH care, and approximately 30 percent use unpaid NTH care. Among the subsample who use NTH care, the largest share relies on unpaid FCC (48 percent), followed by paid FCC (19 percent) and center-based care (17 percent). Although not shown in table 1, the remaining children use other types of NTH care arrangements.
The descriptive statistics reported earlier and in table 1 reflect our analytic sample of low-income parents who work at least some hours during nontraditional times. We note that there are distinct patterns in the use of childcare among our analytic sample compared to low-income parents in the NSECE who work solely during traditional hours. For example, in supplemental analyses of the NSECE data that includes traditional hour workers (not shown in table 1), we find that FCC is more prevalent among children with NTH-working parents compared to those with parents working traditional hours. Specifically, among children from low-income backgrounds whose parents work traditional hours and use some form of nonparental care, 10 percent are in paid FCC (compared with 19 percent for the NTH analytic sample) and 31 percent in unpaid FCC (compared with 48 percent for the NTH analytic sample). In contrast, 38 percent are enrolled in centers (compared with just 17 percent for children with NTH-working parents).
As shown in table 1, NTH care is more common among single parents, as expected from prior research. Specifically, 30 percent of children in two-parent households with no other adults and 40 percent in two-parent extended households report using NTH care. In contrast, about three-fourths of children in single-parent households report using some form of NTH care, regardless of whether extended family members are present. Although unpaid FCC is the most prevalent type of NTH care across all family structures, children in extended households are more likely to use unpaid FCC than nuclear households, with those in single-parent, extended households being the most likely at 68 percent (see online appendix figure A.2). Use of paid FCC ranges from 14–22 percent, with those in single-parent households being slightly more likely to use paid FCC. Children in single, nuclear families are the most likely to use center-based care (27 percent), followed by those in two-parent extended families (18 percent).
Patterns of NTH Care Utilization by Race and Ethnicity and Immigration Status
Table A.2 presents descriptive statistics by race and ethnicity and immigration status. Overall, Black children show the highest rate of using any type of NTH care at 55 percent, followed by White at 42 percent and Latino/a children at 39 percent. Across all racial and ethnic groups, unpaid FCC is most common (see online appendix figure A.3). For paid FCC, Latino/a children (30 percent) are the most likely to use that arrangement, followed by Black (24 percent) and White (8 percent) children. In contrast, White children account for the largest share of center care (20 percent), followed by Black (17 percent) and Latino/a (9 percent) children. Regarding immigration status, unpaid FCC is most prevalent for both children with immigrant backgrounds (40 percent) and native-born children (50 percent). Immigrant children use paid FCC (26 percent) as next most common, with center care (13 percent) least often. In contrast, native-born children use center care (18 percent) and paid FCC (17 percent) at comparable rates. See online appendix figure A.4.
Patterns of NTH Care Utilization by Timing of NTH
Childcare times within NTH may imply distinct parental preferences and care needs (Pilarz et al. 2019). Thus, we investigate the timing of different types of care arrangements (see online appendix figure A.5). Early morning care (56 percent) was most common, followed by weekend (45 percent) and evening care (45 percent). Only 12 percent of NTH care users use overnight care. NTH center-based care is almost exclusively used in the early morning. These results reflect the reality that centers may have only been available to families who worked primarily during daytime, weekday hours but started their shifts early in the morning. The use of paid FCC is relatively consistent across times, ranging from 7–10 percent, except for overnight care (1 percent). While overnight care was least common, 9 percent of NTH care users relied on overnight unpaid FCC.
Regression Results
We report logistic regression results by use of any NTH care and paid NTH care in table 2 and type of NTH care for subsample that uses NTH care in table 3. We then describe the associations between state-level childcare subsidy generosity and NTH care use in table 4.
Logistic Regression Results on the Use of Any NTH Childcare and Paid NTH Childcare
Logistic Regression Results on Each Type of NTH Childcare
Childcare Subsidy Generosity: Logistic Regression Results on Paid NTH Childcare and Each Type of NTH Childcare
Any NTH Childcare
The left panel in table 2 presents the estimated odds ratios (ORs) and robust standard errors from the logistic regression models examining how demographic factors of interest—household structure, race and ethnicity, and immigration status—are associated with the use of any NTH childcare for children under age five with at least one parent working NTH in low-income families. We hypothesize that single parenthood is associated with an increased use of NTH care. As expected, in model 1, controlling for child and family characteristics, the OR for single parenthood is greater than 1 and statistically significant. Specifically, the odds of using any type of childcare during NTH for children in single-parent households are approximately 7.8 times greater than those in two-parent households, holding constant child and family characteristics. As additional controls are introduced from model 1 to model 3, the magnitude of the OR for single parenthood decreases from 7.79 to 2.62. This suggests that some of the associations between single parenthood and the utilization of NTH care are explained by factors such as parent employment and community characteristics. The OR for extended households is statistically significant in model 1 but becomes statistically insignificant with additional control variables, indicating that the presence of other adult household members is not associated with the use of NTH care. The race and ethnicity and immigration variables also are not statistically significant in the full models.
Regarding the interactions between family structure and race and ethnicity and immigration, model 4 in the left panel of table 2 shows that Black children in extended households have a higher probability of using any NTH care, with a probability of 55 percent compared with around 40 percent or less for all other groups (see online appendix figure A.6). Thus, the role of extended household members as childcare providers during NTH is likely most pronounced for Black families. Interactions between family structure and immigration status were not statistically significant, as presented in the online appendix table A.3.
In terms of job characteristics, we anticipate that work schedule predictability is positively associated with the use of NTH care, whereas remote work availability is negatively related to the use of NTH care. As the left panel of table 2 (Any NTH Care) indicates, we do not find a statistically significant association between work schedule predictability and using NTH care. However, as expected, parents with some remote work availability are less likely to use any type of care during these hours compared to those with limited availability, perhaps because this employment flexibility allows them to provide childcare at home.
Paid NTH Childcare
We hypothesize that two-parent households and extended households are less likely to use paid NTH arrangements than single-parent and nuclear households. As expected, and shown in table 2 (model 3 under Paid NTH Care), the odds of using paid care (relative to unpaid care or no care) are almost three times as high for single-parent households as for two-parent households. This may be due to the caregiving role of a partner, which can reduce the need for paid childcare in two-parent households. Single parents may also have greater access to childcare subsidies, enabling them to afford paid childcare, although the overall receipt of subsidies in the sample is quite low.19 As expected, living in an extended household structure reduces the odds of paid care. Specifically, the odds of using paid care for extended households are 43 percent lower than for households with no other adults present, likely due to the availability of household members as unpaid caregivers (see table 2, model 3 under Paid NTH Care). Similar to the results for any NTH care, we find no statistically significant differences in the use of paid NTH care by race and ethnicity or by immigrant background.
Regarding interactions between family structure and race and ethnicity and immigrant background, in general, the lower use of paid care in extended households was not conditioned by race and ethnicity (table 2, model 4 under Paid NTH Care) or immigrant status (table A.3); however, for single- versus two-parent households, the overall pattern showing a greater probability of paid care for single parents disappeared for Black children. That is, two-parent Black households (18 percent) have a similar probability as their single-parent counterparts (20 percent) of using paid care (see figure A.7).
For job characteristics, similar to the use of any NTH care, work schedule predictability does not show a statistically significant association with the use of paid NTH care (table 2, right-hand panel under Paid NTH Care). However, as expected, parents in occupations with remote work availability are statistically significantly less likely to utilize paid NTH care, and this relationship is robust across models 1–3. Parents who are able to work from home during NTH may be less inclined to seek paid arrangements and instead try to simultaneously juggle paid employment with caregiving responsibilities in the household.
Type of NTH Care
To assess associations with the type of NTH regular care—center-based care, paid FCC, and unpaid FCC—we narrow our primary sample to children who use at least some form of NTH care. We note that very few centers are open during NTH, except for early morning hours, which limits the interpretation of these results. Accordingly, only a small percentage of low-income children in our sample, roughly 7 percent, are in centers during NTH, and almost all of this care is during early morning hours. Despite the limited availability of center-based care during NTH, we nevertheless observe important differences in use by family structure and race and ethnicity.
Table 3 reports the results for center-based care (left-hand panel), paid FCC (center panel), and unpaid FCC (right-hand panel). For center-based care, models 1–3 indicate that low-income children in single-parent households are statistically significantly more likely to use NTH center-based care compared to those in two-parent households in each model. For extended households, the OR is less than 1, indicating a lower likelihood of using centers; however, the associations do not remain statistically significant once additional control variables are included. Given that center-based care is primarily early morning care, it is possible that, in two-parent and extended households, the second parent or other adult household member can watch the child during these hours, while single-parent households and nuclear households have fewer care options within the household and instead rely on early morning center hours when available.
We also observe an interaction by race and ethnicity (but not immigrant status) with single parent households (see table 3 model 4 under Center-Based Care) for center-based care, although we interpret these interactions cautiously given small cell sizes. Figure A.8 in the online appendix shows that the positive association between single parenthood and the use of NTH center-based care is primarily observed among White children, with no significant difference for Black and Latino/a children. One possible explanation is the higher enrollment of daytime center-based care among White children (42.5 percent versus 21.3 percent for Black children and 21.6 percent for Latino/a children). With greater numbers of Black and Latino/a children in the sample, we might have observed a statistically significant difference by family structure for Black and Latino/a children as well. In addition, White parents were somewhat more likely to work during early mornings when NTH centers are most available, although this difference is small (67 percent compared with 62 percent).
For paid FCC (table 3, center panel Paid Family Childcare), the ORs for extended households from models 1–3 are consistently statistically significant and less than 1, indicating that children in extended households are less likely to use paid FCC arrangements compared with those from nuclear families. In contrast, the ORs for extended households using unpaid FCC from models 1–3 (right-hand panel under Unpaid Family Childcare) are consistently greater than 1 and statistically significant. Model 3 shows, for example, that extended households have a three times greater odds of unpaid FCC care than nuclear households, underscoring again the important role that extended household members likely play as unpaid caregivers. Interestingly, Latino/a children are more likely to use paid FCC than White children, while there are no racial and ethnic differences, or differences by immigrant background, in the use of unpaid FCC. Lastly, we do not find any meaningful moderation by race and ethnicity in the associations between unpaid and paid family NTH care and family structure. Notably, the interaction term for paid FCC between extended households and immigration background, as shown in table A.3 in the online appendix, reveals that children from immigrant families living in extended households are less likely to utilize paid FCC during NTH compared with those without other adults in the household. This finding suggests that extended household members may be serving as caregivers to these families.
Lastly, table 3 indicates that parents’ job characteristics appear to influence the types of NTH care children use. First, parents with some schedule predictability are more likely to use centers but not paid or unpaid FCC. The association is greater than 1 and robust across all models for centers. This finding suggests that parents with predictable work schedules are less likely to scramble to put care together at the last minute and can therefore access more formal arrangements. Additionally, for remote work availability, the association is negative and statistically significant for paid FCC but not associated with the use of centers or unpaid FCC. This supports the earlier finding, shown in table 2, that low-income parents with access to remote work have lower odds of using paid childcare.
Childcare Subsidies
We examine whether state-level subsidy policy generosity is related to NTH care usage among low-income NTH-working families and whether these relationships are moderated by family structure. In these models, we restrict our dependent variables to paid NTH childcare and the type of childcare utilized during NTH. Table 4 reports the results for two models, the model with all control variables (model 1) and the full model with interaction terms (model 2) for each of the dependent variables. As reflected in table 4, results suggest that living in a state with a greater proportion of eligible families receiving subsidy does not increase the likelihood that low-income families working NTH use paid care during NTH, nor is subsidy generosity related to the type of care used during NTH. Importantly, however, the interaction between subsidy generosity and single parenthood is statistically significant. Specifically, low-income children from single-parent households in more generous states are more than three times as likely to use paid FCC than those in less generous states (see table 4, model 2 under Paid Family Childcare). A state’s subsidy generosity may help single-parents afford paid FCC, while two-parent households may have more difficulty meeting subsidy eligibility requirements or may prefer to rely on a second parent for NTH childcare and forgo the subsidies for which they may be eligible.
DISCUSSION
Despite policy efforts since the publication of Making Ends Meet, including the expansion of public childcare assistance, low-income working parents continue to face a precarious labor market and must navigate the complexities of a fragmented childcare system that is poorly suited to their budgets and work schedules. Using calendar data from the 2019 NSECE, this study aims to understand how low-income NTH-working parents meet childcare needs given these challenges, and whether and how informal family resources and public childcare subsidies support their efforts in a post-welfare reform era.
We observe that slightly more than half of low-income children under age five for whom at least one parent works NTH do not use regular childcare during NTH. Unpaid FCC is the most common care arrangement for children who are in childcare during NTH. We also observe differences by timing, with most NTH care occurring during weekday early mornings, followed by weekends. In our multivariate findings, single-parent households show an increased use of NTH care (and paid NTH care) compared to those in two-parent households. This is consistent with previous studies (Han 2004) and may be attributed to the absence of a partner, as two-parent households often employ a tag team approach to balancing work and family responsibilities (Kim 2022). Single-parent households are also more likely to use centers than two-parent households, particularly among White children. It is important to interpret this finding in context: the prevalence of center use in this sample is extremely low for both single- and two-parent households given limited NTH center availability. Nearly all (90 percent) of the NTH care provided in centers is during early morning times and used by children who are also in centers during regular daytime hours. Children in single-parent households who use centers during the day may have the greatest need for extended care during these early morning hours. In contrast to those in two-parent households, these children may face a lack of family resources during early mornings (such as no second parent present in the household), prompting single parents to seek centers that offer early morning hours. Since more White children are already enrolled in daytime centers compared to children of color, the positive association is stronger for White children. Overall, these findings highlight the need for policy to incentivize centers to extend care, both during early morning and evening hours, which would allow children in center-based care during daytime hours to experience fewer disruptions in care arrangements and avoid multiple arrangements.
Second, we found that Black children in single- and two-parent families were equally likely to use paid FCC, distinct from the patterns for White and Latino/a children. Latino/a children are less likely to utilize center-based care and more likely to rely on paid FCC compared with White children. These patterns generally mirror those observed in daytime childcare usage (Fram and Kim 2008; Radey and Brewster 2007), suggesting that while the context for NTH care may differ from that of daytime care, racial and ethnic factors remain important to consider in reaching diverse families.
Third, the presence of extended household members appears to have opposite effects on paid and unpaid FCC. Residing in an extended household reduces paid FCC while increasing unpaid FCC. It is plausible that coresidential nonparent adults play a role in childcare provision at no financial cost when at least one of the child’s parents is working NTH. This reinforces the notion of extended living arrangements as an adaptive strategy and an accommodation for the competing demands of work and care, especially in the context of economic precarity (Cross 2018; Henly 2002; Pilkauskas 2012). However, our data do not allow us to identify temporal relationships between decisions regarding living arrangements and employment. Single parents may have decided to double up with relatives to facilitate NTH employment. Conversely, for single parents already living with relatives, the availability of care provided by these relatives may have motivated them to accept NTH jobs. Additionally, the literature on living arrangements and maternal labor supply has documented that living with or close to grandparents increases maternal employment because grandparents serve as childcare resources (Compton and Pollak 2014; Kwon 2023). Exploring whether this relationship holds during NTH would be an important topic for future research.
Fourth, among low-income parents working NTH schedules, significant differences in NTH care utilization patterns emerge by schedule predictability and a proxy for remote work. Families with greater schedule predictability are more likely to use center-based care, as these centers offer less flexible hours (NSECE Project Team 2015). Conversely, remote work opportunities may enable parents to care for their children while working during NTH hours. However, we are unable to assess whether such blurred boundaries between work and home mitigate work-family conflicts or exacerbate them (Kim et al. 2020). Notably, those with greater remote flexibility are slightly more likely to work on weekends (54 percent versus 46 percent), indicating that their jobs may require weekend work while allowing for remote arrangements.20 Overall, these findings indicate that job characteristics, in addition to NTH, are important to assess when considering the childcare needs of families with young children.
Finally, what are the implications of our study findings for public policy? We found some evidence that subsidy generosity is associated with the use of NTH paid family childcare arrangements for single parents in this low-income sample. However, there was no association between center care use and subsidy generosity, as has been found in prior studies of daytime care arrangements (Pilarz 2018; Weber et al. 2014). This could be the result of two interacting forces. First, there are few centers that serve children with NTH care needs, especially evenings, overnights, and weekends, restricting many NTH-working families to the informal childcare market. At the same time, states are increasingly using subsidy dollars to pay for centers, with only 5 percent of subsidies nationwide paying for license-exempt home-based childcare (Office of Child Care 2025). Thus, the subsidy program, as it operates in most states, does not serve the NTH childcare market very well. Given that the informal sector is shouldering the care burden during NTH, states might more actively recruit license-exempt providers and the families that use them into the subsidized care system. Directing subsidy dollars to these caregivers could improve the economic circumstances of providers, parents, and the extended family systems that are disproportionately helping parents manage work and care responsibilities. Subsidizing these arrangements could also improve care quality for children if providers invest subsidy dollars in their programs. It might also be fruitful for states to provide incentives to centers and licensed family homes to extend hours to evenings, weekends, or even overnights when demand for NTH care during these times is sufficiently high to make hour extensions feasible (Henly and Adams 2018). Such efforts have rarely been undertaken or evaluated, although the 2014 reauthorization of the Child Care and Development Block Grant instructed states to increase the supply of NTH childcare (Sandstrom et al. 2019). More research is necessary to understand the conditions under which parents would seek center-based and licensed home-based childcare during different nontraditional times, and the types of incentives that would successfully compel providers to extend their services to a wider range of hours.
Parental employment conditions may also help explain why we did not see a robust relationship between subsidy generosity and childcare arrangements. As Harriet Presser (2003) points out, many individuals working NTH—especially, low-income NTH workers—do so because they are unable to find standard hour employment, contributing to high job turnover. Job instability is a predictor of short spells on childcare subsidies (Henly et al. 2017). Therefore, low-income parents in our sample who work NTH may struggle to take advantage of childcare subsidies due in part to job instability. The irony of these findings is that families with the greatest need for stable, quality childcare are also those with the lowest earnings and the most unstable NTH jobs that prevent them from accessing childcare supports. Despite decades of increases in women’s employment and childcare reforms, neither paid employment nor public support has resolved their work-family conflicts.
This study is not without limitations. It is a cross-sectional study and, therefore, our analytical approach is associational and cannot establish causal relationships. Our findings would also be richer with more information about the relationship status of nonparental adults in the household and how these members contribute to childcare decisions. Another limitation is our inability to differentiate between licensed and unlicensed care in the NSECE survey. Unpaid family caregivers are undoubtedly not licensed, but paid arrangements include both licensed and unlicensed FCC and FFN programs. It would be useful to understand whether and how licensing status matters to parental decisions about NTH care. Prior research suggests that licensed care (as well as center care) is generally of higher quality than unlicensed care or unregulated home-based care (Chaudry et al. 2021). However, conventional quality measures may be less applicable to NTH care, where traditional school-readiness curricula and provider-led activities may be less relevant than care that is individualized to children’s routines and consistent with family traditions and values (Bromer et al. 2024).
By comprehensively considering family structure, race and ethnicity, and immigration status and by attending to childcare use across NTH times, this study expands knowledge of the NTH childcare arrangements of low-income working families with NTH work schedules. Our findings suggest that, despite significant public investments in childcare since Making Ends Meet, these advancements largely bypass families with NTH care needs. Many families use childcare during NTH hours, but scarce childcare supply during NTH hours and subsidy programs that favor center care make it difficult for these families to enroll in the subsidy program. Incentivizing centers and licensed family childcare homes to provide care during NTH when demand warrants it, as well as encouraging states to extend subsidy coverage to the informal childcare sector, would be two steps toward positively supporting the childcare needs of families working NTH. Ultimately, however, labor market reforms, in addition to childcare policy reforms, will be necessary to significantly improve the caregiving challenges faced by families working nontraditional hours.
FOOTNOTES
↵1. Center-based arrangements are typically licensed and regulated by states, while some centers, such as those run by religious institutions or schools, may be exempt from licensure or regulation depending on state law.
↵2. For example, centers rarely offer care during evenings (2 percent), weekends (3 percent), or overnights (6 percent). Formal home-based programs provide care during evenings (16 percent), weekends (23 percent), and overnights (19 percent) (NSECE Project Team 2015).
↵3. The National Survey of Early Care Education (NSECE), one of the most reliable sources of national data on providers, does not include data on licensure. Rather, home providers are defined as listed which approximates licensure and suggests greater formality, and unlisted which approximates license-exempt FFN and considered informal.
↵4. To assess the sensitivity of the results to our 200-percent FPL decision, we also ran our main models for families below 250 percent FPL. This change in sample definition increased the sample size from 1,600 to 1,960 but did not substantially alter our main findings.
↵5. Unweighted NSECE sample sizes are rounded to the nearest 20 according to the NSECE restricted-use data policy.
↵6. We conducted a robustness test using an alternative definition of NTH (any time outside of 7 a.m.–6 p.m. on weekdays), which is detailed in the supplementary analyses section in the online appendix.
↵7. These occupations include management, business, and financial operations; computer and mathematical; architecture and engineering; life, physical, and social science; legal; and arts, design, entertainment, sports, and media occupations. Additionally, we constructed a continuous variable from the proportion of individuals in each occupation who reported being able to work from home, using estimates from Alon et al. (2020). Higher values of this variable indicate greater flexibility in work arrangements for the parent. As an alternative to using a binary remote work indicator, we included this continuous variable in our models as a robustness check. While the direction of the odds ratios remained consistent, the effect sizes were somewhat attenuated.
↵8. We conducted a robustness check using an alternative definition of nonparental childcare. While our main analysis classifies care by nonresident parents as parental care, the robustness check categorizes it as nonparental care. The results (see the online appendix) are not substantially different from our main findings.
↵9. Although we can observe parental or guardianship status, the NSECE does not provide information on the marital status of the parents, so we cannot distinguish between married spouses and unmarried cohabiting partners.
↵10. The Other category includes Asian, Native Hawaiian and Pacific Islander, American Indian or Alaska Native, multiple races, and other races.
↵11. Although not shown in the paper, we ran our models with the generosity measure based solely on state income eligibility thresholds and found that the results were consistent with our main findings.
↵12. The estimated number of eligible children comes from a report by Nina Chien (2022) from the Office of the Assistant Secretary for Planning and Evaluation, US Department of Health and Human Services. The number of children served is obtained from the Administration for Children and Families, US Department of Health and Human Services.
↵13. Data on Head Start funding were retrieved from the Office of Head Start. Information on state prekindergarten spending was drawn from the State of Preschool Yearbook published by the National Institute for Early Education Research.
↵14. The online appendix can be found at https://www.rsfjournal.org/content/12/1/192/tab-supplemental.
↵15. In this model, we aim to compare the odds of using paid NTH arrangements versus unpaid or no care. Since approximately half of the children in our sample do not utilize any form of care, this high rate of nonuse presents a hurdle to or a lower need for accessing it for some families. As a result, restricting our sample to those who use some type of NTH care could introduce selection bias.
↵16. We also conducted a multinomial regression for model C on a categorical variable indicating the primary mode of NTH childcare to compare the likelihoods of using each type of NTH care relative to other types. However, small cell sizes reduce statistical power. Thus, we present results from the logistic regression model, which, while not substantially different from the multinomial model results, exhibits greater statistical significance.
↵17. Nearly half of Latino/a children (48 percent) and one-fifth of Black children (20 percent) in our sample come from immigrant families; immigrant families represent only 9 percent of White children.
↵18. At the same time, if extended household members or spouses are not available to provide childcare, NTH-working parents in immigrant households may have a more difficult time using nonresident relative providers because they are less likely to have relatives living nearby than nonimmigrants. Specifically, 43 percent of immigrant families reported having nearby relatives who can provide childcare, compared with 58 percent of nonimmigrant families.
↵19. In our sample, only 5 percent reported receiving childcare subsidies, significantly lower than the national take-up rate of 16 percent (Chien 2022). This discrepancy may be attributed to misreporting.
↵20. Parents with greater remote work availability are less likely to work during early mornings on weekdays than those with limited flexibility. However, there is no statistically significant difference in their likelihood of working during weekday evenings or overnights.
- © 2026 Russell Sage Foundation. Kwon, Sarah Jiyoon, Julia R. Henly, and Suh Kyung Lee. 2026. “Aligning Work and Care in a 24/7 Economy: The Childcare Arrangements of Low-Income Families Working Nontraditional Hours.” RSF: The Russell Sage Foundation Journal of the Social Sciences 12(1): 192–219. https://doi.org/10.7758/RSF.2026.12.1.08. An earlier version of this study was presented at the Russell Sage Foundation internal conference in June 2024. The authors thank the conference participants for their helpful comments. Research reported in this publication was supported by the Administration for Children and Families under award no. 90YE0285. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Administration for Children and Families. Direct correspondence to: Julia Henly, at jhenly{at}uchicago.edu, 969 East 60th Street, Chicago IL 60637, 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.