Parental Expenditures of Time and Money on Children in the U.S.

How much do parents spend on children in the U.S.? While the U.S. Department of Agriculture (USDA) regularly addresses this question, it considers only money expenditures, omitting the sizeable monetary value of parental time. The 2017 and 2019 Panel Study of Income Dynamics offers a unique opportunity to provide a more complete picture. Analysis of this data reveals considerable substitutability between unpaid and paid childcare and generates estimates of average total expenditures that include a replacement cost estimate of the value of parental time. These estimates, constructed for comparability with USDA measures, reveal both higher levels of average parental expenditure and different patterns across household structure and income. These findings challenge public policies that use USDA estimates as a reference point for setting the child support obligations of non-custodial parents and reimbursement rates for foster care. They also undermine many conventional equivalence scales and measures of income/time poverty.


Introduction
Economists typically define family expenditures as monetary outlays, omitting consideration of the imputed value of time devoted to unpaid household services such as childcare. Existing U.S. Department of Agriculture (USDA) estimates of parental expenditures, often used as benchmarks for public policy, apply this narrow definition. Efforts to go beyond it have been hampered by lack of household-level survey data covering both expenditures of money and expenditures of time. In 2017 and 2019, however, the Panel Survey of Income Dynamics (PSID) collected data on both, providing a unique opportunity to develop a more complete picture of substitutability between time and money, average levels of total parental expenditure, and differences in the composition of expenditures by household structure and income.
Estimates of parental expenditures that include the imputed value of unpaid childcare can improve family decisions. Potential parents should have a clear idea of the time, as well as the money, they are likely to spend. Motherhood, unlike fatherhood, typically imposes temporal constraints that reduce earnings and increase economic vulnerability (Kleven et al. 2019;Misra et al. 2007). Mothers are more likely than fathers to become single parents, a transition that increases demands on their time as well as their money. Parenthood is an enormous source of subjective satisfaction, but it is also a valuable contribution to the capabilities of the next generation that yields fiscal benefits (Wolf et al. 2011). Parental expenditures of both time and money represent a costly investment in the future.
Yet public policies in the U.S. implicitly assume that parenting is not work. Many forms of public assistance in the U.S., including the Earned Income Tax Credit, are conditioned on participation in paid employment, which reduces time available for unpaid childcare and often requires out of pocket expenditures on childcare. Official U.S. poverty thresholds vary only by family composition, with no consideration of reductions in the supply of unpaid childcare resulting from paid employment. Conventional equivalence scales weigh the consumption needs of children less than those of adults, ignoring the costly temporal demands than children impose (Folbre et al. 2018). Neither the child support obligations of non-custodial parents nor the reimbursement rates set for foster parents explicitly consider the value of parental time.
The task of approximating this value is not easy. Childcare itself must be defined, either by letting respondents decide what it means, as in the PSID, or asking them to report specific activities during their waking hours, as in the American Time Use Survey (ATUS). Unpaid childcare itself can take different forms. The ATUS tallies both active childcare (such as feeding or bathing a child) and supervisory time when adults report that a child under the age of 13 was "in their care" while engaging in activities other than active childcare. Time expenditures go beyond active and supervisory care because children increase the demand for other unpaid household services, such as housework and meal preparation.
Most imputations of the value of unpaid services are based on replacement cost-what it would cost on an hourly basis to purchase replacement services of comparable quality-making choice of appropriate wage rates crucial. Substitutability between unpaid work and purchased services indicates how respondents view the relative quality of these two sources of supply. Most households probably require a minimum amount of time for unpaid services for which substitutes cannot be purchased, and, likewise, a minimum amount of income that cannot be replaced by unpaid services. The tradeoffs that take place between these thresholds deserve careful empirical scrutiny (Dorn and Folbre 2022).
Exploration of these methodological issues sets the stage for our analysis of pooled data from the 2017 and 2019 PSID. We estimate the value of monetary expenditures and the imputed value of time expenditures based on assumptions comparable to those applied in USDA estimates. Our estimates of total parental expenditures on children are, not surprisingly, considerably higher than those based on money expenditures alone. They also reveal a very different picture of differences among households with children. USDA measures based on the Consumer Expenditure Survey (CE) show that single parent households devote about the same amount of money to children as two-parent households. Because single parent households have less available parental time, our measures show that their average total expenditures are considerably lower than those of two-parent households. The USDA measures show that twoparent low-income households spend less money on children than affluent households, but the additional parental time they devote to children reduces this difference. These empirical findings urge recalibration of thresholds of need built into a number of public policies.

Definition and Valuation of Unpaid Childcare
The advent of nationally representative time use surveys has called attention to the quantitative significance of unpaid household work, including childcare. In such surveys, work is typically defined as any activity that another person could, in principle, be paid to perform.
Time-diary surveys such as the ATUS usually go into considerable detail, asking respondents to report their activities during the previous day, and coding physical care of children, such as feeding, bathing, and dressing, developmental care such as reading to children or helping them with homework, and logistical/managerial care such as transporting children, accompanying them to doctor's appointments, or arranging services on their behalf. Young children also require supervision. They cannot be left alone without an older child or adult nearby and "on call" in case active care is required (Folbre and Yoon 2007;Folbre 2022).
Unlike time-diary surveys, those based on activity lists, such as the PSID, ask respondents to report approximately how much time they devoted to an activity during a longer time period, such as a week. Time-diary surveys are generally considered more reliable because respondents are more likely to accurately recall activities during the previous day and less likely to succumb to social desirability bias, reporting what seems appropriate or expected rather than what they actually did. Furthermore, diary-based surveys constrain respondents to a 24-hour day, while activity lists that allow respondents to report activities that overlap with one another.
On the other hand, activity lists allow respondents to apply their own definitions of childcare, which likely include responsibilities that constrain their physical location and their choice of activities. These surveys typically yield much larger estimates of childcare time than diary-based surveys, probably because they pick up extensive supervisory care (UNWomen 2021). Unlike most diary-based surveys, the ATUS asks adults living in a household with a child under 13 if a child was "in your care" while they were engaging in other activities (not including active childcare). Analysis of pooled data for the ATUS for 2004-2019 shows that mothers living in a household with at least one child under 13 spent an average of 2.3 hours in active care, but 8.98 hours with a child in her care. The corresponding figures for fathers were 1.23 and 6.07 (Suh and Folbre 2022). As our empirical analysis will show, ATUS estimates of childcare that include "in your care" supervisory time are remarkably close to the more approximate measures of the PSID.
Most estimates of the value of unpaid services on the household level multiply estimates of the amount of time devoted to these services by a wage rate based either on replacement cost (the hourly wage that would be charged by someone hired to provide services of comparable quality) or opportunity cost (typically proxied by provider's actual or potential hourly wage in paid employment). Replacement cost is generally considered a more appropriate choice for national income accounting because it comes closer to measuring actual value added.
Opportunity cost, which may reflect non-pecuniary preferences for different types of work, is an additional factor especially relevant to individual decisions (National Academy of Science 2005).
A previous imputation of childcare time in the ATUS (on the aggregate, rather than household level) utilized a replacement cost approach, applying a vector of wage rates to specific types of childcare, highest for developmental care and lowest for supervisory care, which was pegged at the federal minimum wage (Suh and Folbre 2016). In the absence of disaggregated data on types of childcare, the minimum wage represents a cautious choice for a lower-bound estimate of the value of parental childcare. In the U.S. today, many states have set minimum wages considerably higher than the federal level, and application of these wage rates allows for geographic variation.

Substitutability and Outsourcing
Most parents develop caring relationships and child-specific skills that make some portion of their childcare time irreplaceable by market substitutes. Children in the U.S. today typically rely on a combination of parental care, care by other family members or friends, and paid care. Relatively little is known about specific "care packages," because the Consumer Expenditure Survey (CE) does not collect data on time use, the ATUS does not collect data on expenditures, and neither collects data on assistance received from family members and friends.
The ATUS does include measures of family income, and analysis of its relationship to unpaid work finds little correlation (Frazis and Stewart 2011). This does not imply lack of substitutability on a more disaggregated level: individuals in high-income households may engage in different kinds of unpaid work than those in low-income households.
However, it is difficult for most households to compensate for a low level of market income by providing services for themselves because they lack the resources and the skills to do so. Few people can build their own houses, fix their own cars, or produce their own phones or computers. Certain activities of household production may be more amenable to market substitution than others. Aguiar and Hurst (2005) find extensive substitutability between money and time devoted to meal provision. Substitution with respect to another activity of household provisioning-the care of children and family members experiencing sickness, disability or frailties of old age-has received little attention. Time devoted to care provision is quantitatively far more important than food preparation for households that include dependents in need of such care. The need for supervision, combined with the need for flexibility to provide active care when needed, represents a significant constraint on maternal employment. Suzanne Bianchi observed long ago that time devoted to active childcare differs remarkably little across employed and non-employed mothers, considering the temporal demands of most paid jobs (Bianchi 2002). According to a recent analysis of pooled data from the ATUS for the 2004-2019 period, employed mothers in a household with at least one child under 13 spent an average of 1.52 hours in active childcare, while those who were not employed spent an average of 3.13 hours-a substantial difference in percentage terms, but, in absolute terms, only 1.61 hours (Suh and Folbre 2022). By contrast, these employed mothers spent 2.83 hours on children in their care, while those who were not employed spent an average of 7.85 hours. It seems quite likely that paid or donated childcare substitutes more strongly for parental supervisory time than for active paternal childcare. Importantly, the PSID provides a way of examining the relationship between a measure of parental childcare time that includes supervisory care and parental expenditures on purchased childcare.

Data
The Panel Study of Income Dynamics (PSID) is a nationally representative panel study inaugurated in 1968 that interviewed families annually until 1997 and biennially thereafter. It has collected information on paid work and housework hours from family members since 1968; in 2017 and 2019 it included an activity list that asked respondents to report time use in a typical week devoted to eight non-employment activities (housework, personal care, shopping, childcare, adult care, education, volunteering, leisure) for themselves and their spouse/partner (when applicable). The survey also collected information on family expenditures, wealth, and income.
We restrict our attention to families with at most two adults in the family and at least one child under 18, where the second adult is the spouse/partner of the reference person, in order to compute total hours devoted by all adults in the family to unpaid household services including childcare. Our analysis is disaggregated by the age of youngest child, dividing families into six categories: families where the youngest child is aged 0-2, 3-5, 6-8, 9-11, 12-14, and 15-17 years. All adults in the families included must be between 20 and 60 years of age. Of the total sample of 19,176 households in 2017 and 2019, around 3,994 partnered two-adult families and 1,804 single-adult families satisfy our restrictions. The latter are primarily single-mother families  Table   1). Childcare time is identical for women in both the PSID and ATUS, supporting the likelihood that the stylized question on time use in the PSID picks up the effect of "in your care" responsibilities. However, men report greater time on childcare in the ATUS than in the PSID (11 hours compared to 8 hours), which accounts for some of the discrepancy in time spent on all household services (26 versus 19 hours for men in the PSID). Men may be more likely to report in-your-care time when they are specifically prompted to report it in the ATUS. In general, the PSID appears to undercount men's unpaid work, implying that our measure of unpaid household services in the PSID is an underestimate. Another limitation of the PSID time use data lies in lack of differentiation among types of childcare. In particular, we cannot distinguish between time spent on activities that are considered more likely to promote child development (reading or playing with children) and other less-intensive responsibilities, such as in-your-care time.
Research based on the ATUS shows that young children in households with high levels of maternal education spend less time with parents but receive more developmental care (Flood et al. 2022). Our estimates of variations in PSID childcare time across categories of family income are subject to the caveat that unpaid childcare quality cannot be taken into account. 3 This problem is less severe for expenditures on paid childcare since these are likely to correlate with quality.

Tradeoffs Between Time and Money in the PSID
The PSID data make it possible to explore the impact of outsourcing on time use, distinguishing between weekly hours on unpaid services that are not directly child-related and those that are. Non-child related expenditures and non-directly child related time devoted to housework and shopping appear to be complements rather than substitutes.  Multivariate analysis provides further confirmation of this contrast. Panel A of Table 2 regresses non-child family expenditures on housework, conditioning on the number of children, age of youngest child, metropolitan area, and region (specification 1); in specifications (2)   These patterns likely reflect the effect of women's paid employment hours. Table 3 focuses on two-adult partnered-couple families with children, while distinguishing between women's and men's paid employment hours among the controls. Non-child expenditures continue to have a zero (though insignificant) relationship with housework. A $1000 increase in annual expenditures on childcare is associated with a 0.8 hour/week reduction for women, suggesting that much of the 0.9 hour/$1,000 reduction in specification (3) of Table 2 is driven by reductions for women. A one-hour increase in their partner's paid work increases unpaid childcare by similar amounts (about 0.3 hours/week) for both women and men; however, an increase in their own hours of paid work decreases unpaid childcare by a slightly greater amount for women (0.6 hours/week) than for men (0.4 hours/week).

Parental Expenditures of Time and Money on Children
In order to construct a measure of total average expenditures on children including the room house is assumed to have 3 bedrooms). We concur with the USDA that using the average cost of an additional bedroom is a "conservative estimate of housing expenses on children because it does not account fully for the fact that some families pay more for housing to live in a community with preferred schools or other amenities for children" (Lino et al. 2017, p. 9). The results we obtain (see Figure 4)

Variation by Household Composition
In their estimates of money expenditures, the USDA reports similar expenditures on children for two-and single-adult families. For instance, in 2015, total family money expenditures on a child from birth through age 17 were estimated to be $172,200 for single-parents and $174,690 for married-couple households with before-tax income below $59,200 (Lino et al. 2017:13).
Money expenditures on children are only slightly lower for single-parent households, even though the latter expenditures are likely to be underestimates (non-custodial parents may also make purchases on behalf of a child that are not captured in the CE data). Because single-parent families have lower incomes, childcare expenditures form a greater share of total expenditures for these households. This raises an important question: how do total expenditures on children in single-parent households compare to those in dual-parent households?
Two-adult families have a larger overall time budget than single adult families, especially for supervisory care of children. Therefore, we hypothesize that family time devoted to children exhibits much greater differences between two-adult and single-adult families than money expenditures. Figure 6 supports this hypothesis. 9 Families with two partnered adults spend significantly more average time on childcare than single adult families (see Appendix Figure A.7 for similar patterns for average housework time). (Note that to the extent that the PSID undercounts men's childcare time, the estimates for two-adult families are likely to be underestimates.) Across most child ages, partnered mothers spend less time on childcare than single mothers, though much more than partnered fathers. Single parents devote more time to childcare than partnered mothers (with the exception of those with a child 12-14), but not enough to compensate for lack of a partner (with the exception of those with a child 15-17). On average, single-adult families spend 7 weekly hours less than two-adult families on childcare, or about 9% less than the total time spent by two-adult partnered families in the sample.

Variation by Family Income and Maternal Earnings
Since purchased childcare services partially substitute for unpaid childcare time among households with total higher expenditures, estimates of money expenditures on children that omit consideration of the value of parental time likely overstate inequality in total parental spending.
To test this hypothesis while controlling for family composition, we focus on two-adult, partnered couple families. As an extension of this hypothesis, we explore the possibility that maternal earnings have a stronger positive effect than paternal earnings on the purchase of childcare, whether as a result of gender norms or household bargaining. This implies that the high substitutability between expenditures and unpaid parental care time described earlier is largely driven by maternal work hours. Table 4 documents variation in unpaid services across income terciles for partnered two-adult families, conditioning on the number of children and the age of the youngest child. Moving from the bottom tercile to the top tercile is associated with more than a 20-hour weekly reduction in unpaid services devoted to children (an 18-hour reduction in childcare, and a two-hour reduction in housework per person). Adding the imputed value of unpaid household child services to monetary expenditures radically alters the picture of total expenditures on children (see Figure 8). High-income families still appear to devote higher levels of total resources to children, but inequality in resources devoted to children between low-and high-income families is substantially reduced. Rather than costing three times as much for high-income families ($23,200) compared to low-income families ($7,600) (as inferred from average monetary expenditures), high-income families spend only 1.2 times as much as low-income families when the value of unpaid household services is included into the cost of children. More dramatically, it also changes how the cost of children evolves with child age: monetary expenditures are roughly constant or even higher for older children compared to younger children, but when time costs are factored in, the cost of children falls with age. Next, we examine how women and men in two-adult, partnered-couple families with children vary their unpaid work when their own or their spouse's labor income changes (See Table 5). Column 1 suggests that a $1000 increase in woman's own labor income is associated with 0.2 hour reduction per week in unpaid childcare time. There is a smaller associated reduction in housework (0.1 hours per week). Women's higher labor income is also associated with an increase in their partners' unpaid work, though the magnitude of this increase does not fully compensate for the reduction in women's hours. On the other hand, an increase in men's labor income is associated with non-significant or very small effects for both women and men. Source: PSID 2017-2019. Sample restricted to two-adult, partnered-couple families. Metropolitan area and region controls included. Standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01

Conclusion
As far as we know, the estimates provided above are the first empirically based estimates While all imputations of the market value of non-market time are approximate, they provide a more accurate picture than the USDA's existing practice of acknowledging indirect costs by assigning an average childcare cost expenditure to all families. Even parents that spend large amounts on purchased care also provide significant amounts of unpaid care. In addition to documenting a higher magnitude of parental expenditures, our analysis modifies USDA findings in other respects. It reveals a greater economic disadvantage for children in single parent households, and considerable reduction in the economic disadvantage for children in low income relative to higher income households.
Our estimates are subject to several caveats. Valuation of parental time by the effective minimum wage provides only a lower-bound estimate of its value since some portion of this time simply cannot be replaced by a market substitute. Our estimates do not account for differences in the quality of parental childcare, essentially equating supervisory care with active developmental care. Nor do they account for public good effects: dividing time devoted to housework by the number of household members may understate its benefits; likewise, supervision of two children may not require any more time than supervision of one. Dividing household totals for childcare by the number of children has the effect of lowering amounts per child, understating the actual childcare received by children in families with more children (which tend to be lower income families). Also, as aforementioned, the PSID understates paternal hours of childcare relative to the ATUS.
None of these caveats, however, undermine our most important point: parental time represents a valuable expenditure on children that requires consideration. Policies making public assistance contingent on participation in paid employment (such as the Earned Income Tax Credit) should factor in the increased expenses incurred when publicly funded childcare is not available. Parental child support responsibilities should not ignore the value of in-kind contributions of parental care for either custodial or non-custodial parents. Foster parents deserve some recompense for the time they devote to their wards, not just for expenditures on housing, food, clothing and other out of pocket expenses.
Future research on total parental expenditures may be able to take advantage of an effort currently underway by the U.S. Bureau of Labor Statistics to create a synthetic data set by statistically matching observations from the ATUS and the Consumer Expenditure Survey.
Redesign of existing survey infrastructure in the U.S. and elsewhere to provide a unified picture of the relationship between expenditures of time and money would offer even more useful results. In the meantime, the PSID remains a useful source of data for further exploration of this issue.