The summer parental investment gap? Socioeconomic gaps in the seasonality of parental expenditures and time with school-age children

Abstract

Inequalities in parental investments can shape inequalities in children’s outcomes and life chances. Scholars have theorized how socioeconomic status (SES) may moderate how parents use parental investments to respond to the loss of the provision of public schooling during the summer. We investigate the seasonality of SES gaps in parental investments of both money and time in the United States using the 1996–2019 Consumer Expenditure Survey and the 2003–2019 American Time Use Survey. We find SES gaps in parental investments of both money and time during the summer, and that SES gaps in expenditures are larger in the summer than during non-summer months. We find little evidence that any of these gaps have grown substantially over time. Finally, we find evidence that SES gaps in summer paternal investments of time are driven by investments in younger rather than older school-aged children. Our findings contribute to our understanding of the link between public and parental investments in children, address a key mechanism in the debate about the summer learning gap, and provide new evidence on how parents may target investments in children towards the ages when they are most consequential.

1. Introduction

As spring fades into summer, most schools in the United States close their doors and let students out for a summer break. Scholars and educators have taken great interest in how parents structure children’s activities and spend time with children during these summer months, and whether these summer investments in children could be stratified by parental socioeconomic status (hereafter, SES). While SES inequalities in children’s outcomes are caused in part by inequalities in school experiences (Aikens & Barbarin, 2008; Calarco, 2018; Crosnoe, 2009; Murnane & Reardon, 2018; Oakes, 2005; Calarco, 2020), these effects may pale in comparison to the effects of inequalities in children’s out-of-school experiences (Duncan & Murnane, 2011; Kalil, Ryan, & Corey, 2012; Schneider, Hastings, & LaBriola, 2018). Entwisle et al. (2001) used the analogy of school as a “faucet” which provides somewhat (though not entirely) equal resources into the lives of children of all SES backgrounds. When this stream is turned off during the summer, children can only access parental resources, which are distributed less equally.

Existing research has documented a myriad of ways parental investment can shape learning outcomes for children (Bodovski and Farkas, 2008; Greeman et al., 2011; Hsin and Felfe, 2014; Price and Kalil, 2019; Cano et al., 2019; Li and Hamlin, 2019; Kaushal et al., 2011). Moreover, the effects of these investments can go well beyond children’s academic achievement, improving children’s health (Abufhele et al., 2017), reducing behavioral problems (Cano, 2022), and increasing cultural capital (Rivera, 2012; Turco, 2010). Even though summer is relatively “short” compared to the school year, the cumulation of even small SES gaps in summertime parental investments could over time provide significant advantages to children from higher-SES backgrounds (DiPrete and Eirich, 2006). Thus, summertime could reinforce existing inequalities in children’s life chances that stem in part from their position in the “birth lottery” (Blau and Duncan, 1967; Chetty et al., 2014, 2017; Mitnik, Bryant, and Weber, 2019; Song, Zang, Land, and Zheng, 2022).

In this paper, we first explain how both public and parental investments can shape children’s outcomes and how these investments are stratified by parental SES. We then summarize research on the timing of parental investments, particularly focusing on what we do and do not know about how summer parental investments vary by parental SES. We develop five hypotheses about the seasonality of SES gaps in parental investments to answer a number of questions: (1) Do summer parental investment gaps exist? (2) Do they widen during the summer months? (3) Has this gap grown over time? (4) Has this gap grown more rapidly than the gap in parental investments during the school year? And, (5) does the magnitude of this gap differ by the age of children?

We then answer these questions in the context of both parental investments of money and time in the United States using the 1996–2019 Consumer Expenditure Survey and the 2003–2019 American Time Use Survey. While studies frequently address either parental investments of money and time, we jointly investigate both types of investments and, for investments of time, we analyze both maternal and paternal childcare time. Investigating investments of both money and time is important because, on the one hand, money spent on enrichment activities or childcare could potentially crowd the amount of time that parents spend with their children, which would attenuate the effect of SES gaps of parental investments of money on child outcomes. But, on the other hand, SES gaps of both time and money could increase in the summer, which would imply wider SES gaps in child outcomes.

We find evidence of gaps by SES—measured in terms of both parental education and income—in parental investments of both money and time during the summer. These gaps for expenditures on children are larger in the summer than during non-summer months. While the gaps for maternal or paternal time in child care are not wider in the summer than during the school year, the net result (considering both types of investments) is greater SES gaps in parental investments during summer. We do not find evidence that these gaps have grown substantially over time, suggesting increases in intensive parenting practices characteristic of high-SES parents are not especially concentrated during the summer. In examining differences in summer parental investments by the age of children, we find evidence that SES gaps in summer paternal investments of time in children are driven by gaps in investments in younger school-aged children. The net result of our findings is thus is that SES gaps in summer parental investments are larger among parents of younger children than among parents of older children.

This research contributes to several important strands of the literature on parental investments and children’s outcomes. First, it provides new evidence regarding the link between public and parental investments in children (Entwisle et al., 2001; Cascio and Schanzenbach, 2013; Jackson and Schneider, 2022). In turn, this is also relevant to ongoing debates about the summer learning gap and the place of school and non-school factors in student learning (Entwisle and Alexander, 1995; Downey et al., 2004; Alexander et al., 2007; von Hippel et al., 2018; von Hippel and Hamrock, 2019; Passaretta and Skopek, 2021; Workman et al., 2023). Second, it contributes to debates over whether there are socioeconomic gaps in the extent to which parents target investments in children at the ages when may be more consequential (Heckman, 2006; Kalil et al., 2012). Finally, it advances our understanding of one of the mechanisms through which socioeconomic inequality is maintained across generations (McLanahan, 2004; DiPrete and Eirich, 2006; Cooper and Pugh, 2020).

2. Background and hypotheses

2.1. Public and parental investments as a sources of inequality

Scholars studying the mechanisms underlying this intergenerational transfer of SES have focused on the roles of both public and private investments in children. The most central public investment in children that directly shapes their outcomes is schooling. Children from higher-SES families are more likely to attend better-resourced and higher-performing public schools (Aikens & Barbarin, 2008; Crosnoe, 2009; Goldstein & Hastings, 2019) or attend private schools (Murnane and Reardon, 2018), resulting in significant advantages for children from higher-SES families. Even within the same school, children of high-SES families may still receive better teaching and more advantages over children from lower-SES families (e.g., Oakes, 2005; Calarco, 2020).

However, other research suggests these school-based inequalities may be smaller and less significant than inequalities in children’s family environments. Thus, school could actually be “compensatory” and reduce even greater inequalities between children from different family backgrounds (Downey et al., 2004; Raudenbush and Eschmann, 2015). Parenting practices associated with the “concerted cultivation” of cultural and human capital in children (Lareau, 2002; Lareau, 2011) are positively associated with greater cognitive ability and academic achievement among children (e.g., Bodovski and Farkas, 2008; Cheadle, 2008; Greeman et al., 2011; Hsin and Felfe, 2014; Price and Kalil, 2019). Scholars have focused on two broad forms of intensive parenting that have been described as “parental investments”: (1) financial investments parents make to pay for children’s childcare and schooling, enrollment in supplementary and extracurricular activities, and the purchasing of items and materials necessary for the those in and out-of-school activities; and (2) the extent, content, and quality of the time parents spend with their children.

Though recent research finds that parents from all socioeconomic backgrounds generally aspire to engage in this intensive parenting (Ishizuka, 2019), higher-SES families are more likely to do so, plausibly because they have more resources to invest in their children (Mclanahan, 2004; Bennett et al., 2012; Cheadle and Amato, 2011; Hao and Yeung, 2015; Cooper and Pugh, 2020). The result is substantial gaps by parental education and income in parental investments of both time and money. For example, Schneider, Hastings, & LaBriola, 2018 show that households in the top income quartile spend over three times as much on financial investments in their children as households in the bottom income quartile (see also, Kornrich and Furstenberg, 2013; Kornrich, 2016). Likewise, research shows that more educated parents spend significantly more time in childcare than less educated parents (Gauthier, Timothy, & Frank, 2004; Sayer, Gauthier, & Furstenberg, 2004; Kalil et al., 2012; Altintas, 2016; LaBriola and Schneider, 2021), although the gap between more and less educated mothers has closed in recent years (Prickett and Augustine, 2021). While income is perhaps logically the strongest predictor of parental investments of money in children, Cheadle and Amato (2011) find that parental education is the largest correlate of parenting practices associated with the concerted cultivation of children’s abilities.

2.2. The “other faucet”: seasonality in parental investments

Recent research has re-engaged with questions about the seasonality of children’s achievement gaps. Until very recently, a preponderance of research comparing the rate of growth of children’s test scores over school and non-school months suggested that it is over the summer, not during the school year, in which SES gaps in cognitive ability grew most (e.g., Alexander et al., 2007; Downey et al., 2004; Entwisle and Alexander, 1995). However, new research has found issues with score alignment between different tests and changes in test forms that have overstated what is commonly termed the “summer learning gap” (von Hippel and Hamrock, 2019; von Hippel et al., 2018; Workman et al., 2023). Likewise, an alternative research approach relying not on assessing children’s approaches before and after summer, but rather on a differential exposure approach to attending school, also finds that schooling itself does not appear to be associated with socioeconomic inequality in children’s learning (Passaretta and Skopek, 2021). Instead, the preponderance of evidence now suggests that the most significant inequalities occur before children ever begin school (von Hippel et al., 2018; Skopek and Passaretta, 2021).

However, in contrast to the careful and ongoing research on seasonal patterns in children’s achievement gaps, little research has examined whether there is seasonality in socioeconomic gaps in parental investments of time and money. Entwisle et al. (2001) use the analogy of school providing a “faucet” of resources that is turned off during the summer. When the faucet is turned on during the school year, all students gain in ability (somewhat) equally. However, when the faucet of school resources is turned off during the summer, children’s resources for learning and development primarily come from parental investment, which is distributed far less equally than school resources. These investments can take the form of payments for various enrichment activities (e.g., day or overnight camps, supplemental education, sports programs, music lessons) and/or spending more time talking and playing with children, reading to children, and coordinating children’s activities.

SES gaps in summertime parental investments of time and money are of interest because these investments have important effects on child development. Although difficult to establish the causal effects of parental investments, researchers have consistently found that children’s cognitive skill development and academic achievement are positively associated with parental investments of time (e.g., Hsin and Felfe, 2014; Price and Kalil, 2019; Cano et al., 2019; Li and Hamlin, 2019). Likewise, parental investments of money may provide access to a “shadow educational system” of tutoring and extracurricular classes (Park et al., 2016) and/or a more educationally enriching home environment, both of which have linked to children’s cognitive development and academic achievement (Bodovski and Farkas, 2008; Greeman et al., 2011; Kaushal et al., 2011). More broadly, greater parental investments can positively shape children’s health (Abufhele et al., 2017), and the time children spend with parents and the activities they participate in can help children develop cultural capital, which research shows has also powerful effects on hiring opportunities and workplace success (e. g., Rivera, 2012; Turco, 2010).

While each of these investments may not be especially large, each of these small advantages can accumulate over time into sizable differences in children’s outcomes and life chances (DiPrete and Eirich, 2006). Thus, examining how a broad array of parental investments may be patterned by parental SES during both summer and non-summer months can first help us better understand the timing of these important, parental investments. Additionally, parental investments of time and money are one of the mechanisms predicted to create seasonal patterns in children’s achievement gaps (e.g., Alexander et al., 2007). Hence, the “summer parental investment gap”—or lack thereof—may help us better understand the relationship between parental investments and child learning outcomes over the summer.

Finally, understanding the seasonality of parental investments could provide some additional evidence regarding the link between public and private investments. The faucet analogy suggests parents maintain similar levels of parental investment during the summer as they do during the school year. However, there is good reason to expect SES gaps in parental investments in children might widen during the summer, as high-SES households may increase their private investments in children when their children are not receiving public investments (like public school). In support of this link, Cascio and Schanzenbach (2013) found that the introduction of universal preschool programs during the 1990s led higher-SES parents to reduce their childcare expenses. More recently, Jackson and Schneider (2022) found greater public spending is associated with narrower class gaps in parental investments, in part because public spending on public education was linked to decreases in spending among high-SES households. However, researchers have not examined this link by comparing summer months—when children do not receive public resources from schooling—to non-summer months—where they do.

2.3. A summer parental investment gap?

Both qualitative and quantitative research find a stark SES gap in children’s summertime activities. In ethnographic work, Chin and Phillips (2004) observed that children in high-SES families experienced very full summer schedules that included vacations, day camps, educational activities, music lessons, sports, and other enrichment activities and outings. On the other hand, children in low-SES families had fewer opportunities for these activities, even though their parents also appeared to value these experiences and attempted to provide them as possible. Instead, these parents were constrained by a lack of affordable options for their children and less-accommodating work schedules. These findings are corroborated by statistical analyses of the Early Childhood Longitudinal Study (ECLS-K), which show that higher-SES children are more likely to go on trips to historical sites and museums, attend camp, and receive tutoring than are low-SES children (Meyer et al., 2004; Redford et al., 2018).

This research on the SES gap in children’s summertime activities suggests—but does not show—that there are also large SES gaps in parental investments of money and time during the summer. Moreover, this work generally does not assess whether these SES gaps in parental investments are larger during the summer than during the school year. One notable exception is research by Gershenson (2013) that examined seasonality in parental investments of time (but not money) and found evidence that, while time spent managing children’s activities decreases during the summer, SES gaps in this time widen by a moderate amount during the summer.

We first examine whether and by how much higher-SES families invest more money and time in their children during the summer than lower-SES families (that is, whether higher-SES families “keep the faucet on” during the summer). Given prior research, support for this hypothesis would not be surprising, but our test of it also provides a measure of the size of the gap between high and low-SES parents.

Hypothesis 1. SES gaps in parental investments exist during the summer.

We then examine whether these gaps in parental investments are larger in the summer than during the school year, which might reflect higher-SES parents increasing their parental investments (i.e., “turning up the faucet”) to compensate for school being out.

Hypothesis 2. SES gaps in parental investments are greater in the summer than during the school year.

2.4. Changes over time in the seasonality of parental investments

Research has documented how, compared to several decades prior, parents on average spend more money on and more engaged time in childcare with their children, an increase largely driven by higher-SES parents (Kornrich and Furstenberg, 2013; Kornrich, 2016). Schneider, Hastings, & LaBriola (2018) documented how rising income inequality is responsible for widening class gaps in parental investments of money in children, both because rising inequality has afforded higher-income families more money to spend on their children and because it has influenced higher-SES parents to spend an increasing share of their income on investment goods for children.

The research on widening SES gaps in time use is more mixed. One study of not-yet-school-aged children observed widening SES gaps between 1988 and 2012 in time-intensive parenting behaviors that are associated with greater school readiness—including reading to and telling stories to children and teaching children basic reading and math skills (Kalil et al., 2016). However, a study of kindergarteners in 1998 and 2010 found significant but narrowing SES gaps in parental engagement in learning activities both inside and outside of the home (Bassok et al., 2016). More generally, gaps in parental childcare time by parental income and education widened through the 2000s (Ramey and Ramey, 2010; Altintas, 2016), but appear to have stalled and even narrowed since (Cha and Park, 2020; Prickett and Augustine, 2021).

Nevertheless, given widening SES gaps in parental investments of money in children, SES gaps in total parental investments in children have likely widened. We examine whether there has been a parallel widening summer SES gaps in parental investments in children over time. Moreover, widening SES gaps in parental investments in children over time could reflect increased investments by high-SES parents during the summer to compensate for school being out. We thus examine whether SES gaps in parental investments of children are disproportionately widening during the summer.

Hypothesis 3. SES summer gaps in parental investments have widened over the past several decades.

Hypothesis 4. SES summer gaps in parental investments have widened more than SES non-summer gaps in parental investments over the past several decades.

2.5. Parental investments by child age

Finally, there may also be variation in the seasonality of SES gaps in parental investments by child age. One highly-cited model of child development suggests returns on investment in children’s abilities, whether via direct investment in their human capital (Heckman, 2006; Cordero-Coma and Esping-Andersen, 2018) or via moving to more advantaged neighborhoods (Chetty et al., 2016), are largest when children are younger. If higher-SES parents target their increased investments in children during the summer at the ages when they are most consequential (e.g., Kalil et al., 2012), then we may see wider SES summer gaps in parental investments towards younger school-aged children.

On the other hand, much of the narrative around concerted cultivation has focused on higher-SES parents investing in activities like extra tutoring, SAT prep courses, and resume fillers that will help their children get into an elite college (e.g., Cooper, 2014), and these investment activities are more likely to occur when children are in middle and high school. This leads to two plausible competing hypotheses that we can examine:

Hypothesis 5A. SES summer gaps in parental investments are larger for younger school-age children than older school-age children.

Hypothesis 5B. SES summer gaps in parental investments are larger for older school-age children than younger school-age children.

3. The current paper

This paper advances current research in several key ways. First, despite the growth in parental spending on activities designed for children’s enrichment both during and outside the school year, we are not aware of any other work examining the seasonality of SES gaps in parental investments of money. Previous studies that document differences by SES in children’s summer activities have relied on binary measures from the ECLS-K of whether or not certain pre-selected activities occurred at all (e.g., Meyer et al., 2004; Redford et al., 2018). Our analysis of the amount of spending on these activities is advantageous in that it captures all spending within the enrichment categories regardless of the specific activity, provides a rough proxy for the quality of those activities (to the extent that quality is associated with cost), and is available across many years.

Similarly, although there has been a great deal of research on parental time use, little research explores its seasonality, with the exception of Gershenson’s (2013) analysis noted above.¹ Our research advances Gershenson (2013) in several ways. First, we separately examine the time investments of mothers and fathers. Mothers perform substantially more childcare than fathers (Bianchi 2000, Altintas, 2016; Sayer, Bianchi, & Robinson, 2004), and SES gaps in childcare time are wider among mothers than among fathers (Schneider, Hastings, and LaBriola 2018, Cha & Park, 2020; Schneider, Hastings, & LaBriola, 2018). Further, ethnographic research suggests that mothers are primarily responsible for coordinating children’s enrichment activities during the summer (Lareau 2011). As such, the seasonality of SES gaps may differ between mothers and fathers. Second, we more than double the observations of Gershenson (2013) by extending the analysis through 2019. This also allows us to see if SES summer gaps in parental childcare have widened recently, consistent with rising anxiety among high-SES parents about their children’s outcomes (e.g., Cooper, 2014).

Finally, more broadly, we examine each ceteris paribus hypothesis separately for outcomes of parental investments of expenditures on children’s enrichment and of time through childcare. Both types of investment in children’s lives are understood to contribute to children’s achievement and long-term outcomes in adulthood, yet there are also potential trade-offs between the two. Money spent on enrichment activities or childcare could potentially reduce the amount of time that parents spend with their children. However, it is also possible that SES gaps of both time and money increase in the summer, which would widen inequalities. As such, any broad conclusions about the seasonality in SES gaps in parental investments in children should account for both parental investments of time and money.

4. Data

We test these hypotheses in the context of the United States using publicly available microdata from two large, nationally-representative, long-running surveys: the Consumer Expenditure Survey (CEX) and the American Time Use Survey (ATUS).² Our analysis focuses on households with school-age children (ages 6–17) who are likely to be impacted by the seasonality of schooling. Below, we describe each dataset in detail, with descriptive statistics provided in the Supplement.

4.1. Consumer expenditure survey

We use the 1996–2019 CEX to conduct the first empirical analysis of SES gaps in spending on children’s enrichment activities by season. The CEX Interview surveys capture measures of parental investments and are collected quarterly from each household for four consecutive quarters. Crucially, although each survey collects a full quarter (3 months) of expenditures, each expenditure is associated with the particular month that the expense occurred. It is plausible that, in some cases, the month in which the investment expense occurred is not the same month as when the child actually experiences that investment. This could occur if, for example, parents pay for larger expenses like school tuition or summer camp in advance, or conversely, if parents are able to smooth out large one-time payments over several months. To the extent that parents pay for summer investments in children outside the summer, or vice versa, this will downwardly bias the coefficients in our models reflecting seasonal differences in expenditures on child investments. Thus, class gaps in the investments purchased by parents that children actually experience during the summer may actually be wider than our models report.

Given our focus on school-age children, we exclude families with children under 6 because we only know each family’s total expenditures in a given category and not to which child the expenditure was directed. Altogether, our expenditure analysis is based on observations from 40,399 households in 282,230 household-months.

4.1.1. Measures of parental investments of money

We examine expenditures across a broad range of activities that include extracurriculars (e.g., fees for recreational lessons, instruction), paid childcare (e.g., babysitting, nannies, and daycare centers), and schooling-related expenses (e.g., summer school, school books and supplies, and tuition). Following existing work, we focus on the combined measure of these expenditures and divide by the number of children in the household to generate per-child expenditure measure (e.g., Amorim, 2021; Hastings and Schneider, 2021; Jackson and Schneider, 2022; Kornrich and Furstenberg, 2013; Kornrich, 2016). Using the combined measure is advantageous in that it captures the breadth of expenditures that may each give small accumulated advantages to some children over others, and it captures shifts in spending between investment categories that are likely to occur between summer and non-summer months (e.g., schooling to enrichment activities). Dollar amounts are adjusted to 2021 real dollars using the CPI-U-RS.

4.1.2. CEX independent variables

We operationalize SES in terms of both household income and parental education. For education, we code whether the most educated parent in the household has obtained a Bachelor’s degree, while for income we divide the sample into quintiles.³ A binary “Summer” variable indicates whether an expense took place in summer (June and July) or non-summer (September-April) months. We exclude May and August because these months typically include both in-school and out-of-school periods.⁴ We control for factors that might confound the relationship between SES and parental investments in children: age of the oldest parent, race of each parent (non-Hispanic white, non-Hispanic Black, Hispanic, and non-Hispanic other), average parental work hours per week, whether or not the children’s grandparents live in the home, the number of children aged 6–11, and the number of children aged 12–17.

4.2. American time use survey

We use data from the 2003–2019 American Time Use Survey (ATUS) to analyze SES gaps in parental investments of time by season. Each observation contains time use data from a 24-hour time diary of the respondent, alongside information on their SES, family structure, and demographics. As with the expenditure analysis, we focus on parental respondents who are living with at least one child under age 18 and who do not live with a child under age 6. We drop respondents for whom household income is missing. Additionally, the household income of ATUS respondents who filled out their final CPS survey before October 2003 was coded under a scheme that does not accurately allow us to separate middle- and high-income households, so we drop these households as well. The resulting ATUS analysis sample contains 16,687 mothers and 11,534 fathers.

4.2.1. Measures of parental investments with of time with children

The main dependent variable is the total amount of time a parental diarist reports performing childcare of household children as their primary activity. This includes time in basic care (care of infants, general care of older children, and medical care of children), playing, teaching (supervising children, helping them with homework, and reading to or talking with children), and management (including coordination of extracurricular activities and travel related to childcare). This operationalization is consistent with measures of parental investments of time used in previous research (Altintas, 2016; Bianchi, Robinson, & Milkie, 2006; Kalil, Ryan, & Corey, 2012; Schneider, Hastings, & LaBriola, 2018). We also report results that use an expanded measure of investments of time in children that additionally accounts for secondary childcare, which ATUS defines as care for children under age 13 that is done while the respondent is doing something else as a primary activity. While this time is certainly important, we focus on primary childcare time as our preferred measure of parental investments of time in children.

4.2.2. ATUS independent variables

We use the same control variables in our ATUS analyses as in our CEX analyses, with two modifications. First, the variables for age, race, and education reflect the parent filling out the time diary and not both parents. Second, we control for the number of minutes the respondent worked on the diary day instead of weekly work hours, to more directly account for tradeoffs between work and childcare time.

5. Plan of analysis

We estimate linear regression models to examine the seasonality of parental investments in children. In the equations below, we bold the coefficients that test the relevant hypotheses.

First, we test Hypothesis 1 using only observations from summer months. Formally, we estimate:

$$\mathit{\text{ParentalInvestments}}={\beta }_0+{\mathbf{\beta }}_{\mathbf{1}}SES+{\beta }_2\mathit{\text{Controls}}+{\mu }_{\mathit{\text{year}}}$$ (1)

where SES is our categorical variables for SES, Controls is a vector of individual and household level controls, and μ_year are a set of indicator variables for each (i.e., year fixed effects) included to net out overall time trends in parental investments. β₁ is a vector of coefficients representing the estimate for each SES category (either the income quintiles or having a Bachelor’s degree). We examine income and education SES measures in separate models, but the results considering them together are substantively similar and presented in the appendix.

To test Hypothesis 2, we analyze all observations from summer and non-summer months, and include a dichotomous variable for summer and an interaction between summer and SES:

$$\mathit{\text{ParentalInvestments}}={\beta }_0+{\beta }_1\mathit{\text{Summer}}+{\beta }_{2}SES+{\mathbf{\beta }}_{3}SES\times \text{Summer}+{\beta }_4\mathit{\text{Controls}}+{\mu }_{\mathit{\text{year}}}$$ (2)

The estimate of β₃ informs Hypothesis 2 by capturing the seasonal differences in parental investments by SES. In the expenditure analysis, we have multiple months of data for each household, so we can also employ household-level fixed effects (+ μ_household) that control for the main effect of other time-invariant household characteristics. While the fixed effect absorbs the main effect of SES, the interaction term may reintroduce time-invariant confounders associated with both SES and parental financial investments. We account for this possibility by including interactions between SES and each control variable (time invariant or not), which we also include as control variables. For the fixed effects models we exclude 4.6% of households that changed their education level or income quintile, so that the SES x Summer interaction terms can be interpreted as only the effect of a change in the season (the results including them are nearly identical).

Next, we test Hypotheses 3 and 4 by examining whether differences have grown over time. We estimate the following model using only summer observations:

$$\mathit{\text{ParentalInvestments}}={\beta }_0+{\beta }_1\mathit{\text{Year}}+{\beta }_{2}SES+{\mathbf{\beta }}_{3}SES\times \text{Year}+{\beta }_4\mathit{\text{Controls}}$$ (3)

where the estimates of β₃ show how much the gap widened over time, all else being equal. Although this equation models linear growth, we find substantively similar results with more complex growth trajectories (e. g., higher-order polynomials) and we also present non-parametric plots of these trends over time.

We then test if the SES summer gaps in parental investments have widened more than SES non-summer gaps in parental investments over the past several decades. For this test we use all observations and include interactions of Summer × Year and Summer × SES and a triple interaction of SES × Year × Summer. Evidence of a widening gap more in the summer than in the school year would be supported by a significant and positive coefficient for the triple interaction.

$$\mathit{\text{ParentalInvestments}}={\beta }_0+{\beta }_1\mathit{\text{Year}}+{\beta }_{2}SES+{\beta }_3\mathit{\text{Summer}}+{\beta }_{4}SES\times \mathit{\text{Year}}+{\beta }_{5}SES\times \mathit{\text{Summer}}+{\beta }_6\mathit{\text{Year}}\times \mathit{\text{Summer}}+{\mathbf{\beta }}_{\mathbf{7}}SES\times \mathbf{\text{Year}}\times \mathbf{\text{Summer}}+{\beta }_8\mathit{\text{Controls}}$$ (4)

Finally, we test Hypotheses 5A and 5B, which ask if the seasonality of SES gaps in parental investment in children differs between households with younger and older school-aged children. To do this, we focus on families for whom the children are either all ages 6–11 or all 12–17. We split the data between ages 11 and 12 because it reflects the middle of the distribution of children’s ages, and because it generally corresponds to the end of elementary school and the beginning of middle school in the U.S. education system (we found similar results with a slightly higher or lower cutoff). We cannot determine whether investments are directed toward older or younger school-aged children for families with children in both age categories, so these families are excluded for this last test. We re-estimate the models corresponding to Eq. 1, but now additionally include an interaction between SES and a variable indicating whether all household children are younger (ages 6–11) or older (12–17).

$$\mathit{\text{ParentalInvestments}}={\beta }_0+{\beta }_1\mathit{\text{Childs}}\prime Age+{\beta }_{2}SES+{\mathbf{\beta }}_{3}SES\times \text{Child}Age+{\beta }_4\mathit{\text{Controls}}+{\mu }_{\mathit{\text{year}}}$$ (5)

All of our models include sampling weights. The time-use models use survey replicate weights that account for sampling variance, and the expenditure models include standard errors clustered by households.

6. RESULTS

6.1. Summer gap

Table 1 shows the analyses testing Hypotheses 1 and 2. Models in Columns 1, 4, and 6 estimate the differences in parental investments during summer months by SES, while the remaining columns estimate the seasonality of these differences.

Table 1.

Regression Models of Parental Investments.

	Expenditures			Maternal Time		Paternal Time
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Summer	Full Year	Full Year, FE	Summer	Full Year	Summer	Full Year
Education Models
BA or higher	138.6 *** (7.49)	102.5 *** (3.65)		19.7 *** (4.55)	112 *** (2.02)	7.37* (3.55)	12.3 *** (1.55)
Summer		−8.25 *** (1.68)	81.8 (61.7)		−25.9 *** (2.57)		−8.75 *** (2.39)
BA or higher x Summer		41.1 *** (7.89)	41.1 *** (8.44)		6.48 (4.43)		−5.97 (3.67)
Income Models
Bottom quintile	−16.8 (10.3)	−12.5 * (4.96)		−9.85 (6.34)	−4.27 (3.27)	−1.36 (5.29)	−4.95 (2.85)
2nd quintile	30.7 *** (8.33)	−23.1 *** (2.94)		−7.72 (5.92)	−1.35 (2.73)	−3.20 (4.75)	−2.68 (2.58)
4th quintile	26.3 *** (7.96)	35.5 *** (3.87)		6.27 (5.78)	6.41 (3.35)	4.64 (5.02)	6.56 ** (2.47)
Top quintile	181.1 *** (12.8)	141.8 *** (5.65)		9.76 (5.98)	10.0 ** (3.64)	1.58 (4.42)	8.40 ** (2.61)
Summer		0.88 (6.06)	91.8 (62.3)		21.1 *** (4.02)		−9.62 ** (3.60)
Bottom quintile x Summer		−2.51 (8.15)	−9.93 (12.6)		−3.30 (5.92)		2.87 (5.50)
2nd quintile x Summer		−7.26 (6.80)	−15.6 (9.64)		−4.98 (6.07)		−0.15 (5.19)
4th quintile x Summer		−7.67 (7.75)	−11.7 (9.05)		−0.90 (6.47)		−3.04 (5.50)
Top quintile x Summer		42.7 *** (12.2)	48.8 *** (12.8)		−0.68 (6.95)		−7.19 (5.18)
Observations	56,420	282,230	269,226	3,290	16,687	2,216	11,534

Standard errors in parentheses.

^*p < .05,

^**p < .01,

^***p < .001

Education and income models are run separately. Models with both measures together are in the Appendix. Education reference group is families without a B.A. Income reference group is the middle income quintile. Household controls and year fixed effects are included in all models. The fixed effects model also includes interactions between summer and each control variable.

As predicted by Hypothesis 1, the models show SES gaps during the summer. In the education model in Column 1, families where at least one parent has a BA spend—on average and all else being equal—about $139 per month more (in 2021 dollars) during the summer than a family without a BA. Differences by income are especially large between the top income quintile and all others, as families in the highest income quintile spend $181 more monthly than those in the middle quintile and nearly $200 more than those in the bottom quartile.

Models in Column 2 examine whether these SES gaps in parental expenditures are larger in the summer than during the school year (Hypothesis 2). “Non-summer” is the reference category, so the SES coefficients simply indicate that SES gaps exist during the school year as well as in the summer. To examine the seasonality of this gap, we focus on the interaction terms. We find that families with a BA increase their spending in the summer by $41/month more than those without a BA, while families in the highest income quintile increase their spending by $43/month relative to the middle income quintile. Our household fixed-effect models that examine Hypothesis 2 (Column 3) provide nearly identical estimates of the seasonality of SES gaps.

We illustrate these gaps with the predicted parental investments of money using the coefficients from Column 2 (setting all other covariates to their mean) and plotting these values and their confidence intervals in the first column of Fig. 1. The figure highlights the large overall differences by education and between the top income quintile and other income groups, as well as the increases in investments during the summer for the more-educated and highest-income households.

Fig. 1.

Predicted Parental Investments by SES and Season. Predicted values are based on models in Table 1.

We next look at whether SES gaps in parental investments of time are wider during the summer, and whether the size of SES gaps in parental investments of time vary between summer and the school year. We look at maternal time use (Columns 4 and 5 of Table 1) separately from paternal time use, given substantial differences in the amount of time that mothers and fathers perform in child care.

Models in Column 4 reveal wide education-based gaps in maternal investments of time in children during the summer. All else equal, mothers with at least a Bachelor’s degree spend 20 min more of daily childcare time than do other mothers of school-aged children. Looking at income-based gaps, the coefficients reflect a gradient of increased child care time with increased income. Although no quintile is statistically different from the middle quintile, there is a statistically significant difference between the top and bottom income quintiles (19.6 min, p < .001). Column 5 shows that mothers spend about 20 fewer minutes doing primary childcare during the summer than during the school year.⁵ However, the coefficients on the interaction terms for education × summer and income × summer are small and not statistically significant.

Turning to fathers, we see a similar story. During the summer, those with at least a Bachelor’s degree perform about 7 more minutes of primary childcare time per day than those without, net of observable characteristics (Column 6). However, we do not find income-based gaps in paternal childcare time during the summer. We also do not find widening of gaps by education or income in paternal childcare time during the summer (Column 7).

The time use results are visualized in Columns 2 and 3 of Fig. 1, highlighting several broad features: first, consistent with previous research, mothers spend far more time doing child care than fathers. Second, both mothers and fathers spend less time doing primary child care in the summer. Third, gaps in childcare time exist by parental education for both mothers and fathers, but these are similarly sized across seasons.

In sum, these results support Hypothesis 1 that there are significant SES gaps in parental investments in children during the summer. The results provide partial support to Hypothesis 2—while SES gaps in parental investments of money widen during the summer, there is no evidence of seasonality in SES gaps in parental investments of time. However, given that the time use gaps also do not decrease during the summer, the net result is that when considering money and time use together, SES gaps in parental investments widen during the summer, at the same time that equalizing public investments in children in the form of schooling are no longer present.

6.2. Change over time

Our next set of analyses examines whether these gaps have changed over time. First, Fig. 2 presents non-parametric lowess plots of the unconditional average parental investments in summer and non-summer months over the period of the analyses, for each education level and the bottom, middle, and top income quintiles. Looking at the top panel of the first column, summer expenditures appear to have grown somewhat for families where at least one parent has a Bachelor’s degree, and decreased slightly among other families, leading to an overall widening gap in summer expenditures. Similarly, looking at the bottom panel of the first column, we see an increase in summer expenditures for the top income quintile, and a slight decrease for the bottom quintile, also leading to a widening gap. However, within SES groups, summer and non-summer parental expenditures of money on children roughly appear to move together over time.

Fig. 2.

Parental Investments over Time by Season.

Table 2 shows regression results more rigorously testing Hypotheses 3 and 4. Models in Column 1 examine the change over time in summer parental investments of money, with a focus on the SES × year interaction terms. Consistent with Fig. 2, coefficients are larger for higher SES categories. The differences by education are not statistically significant, but the interaction terms with income show a widening over time—the middle income quintile increased their summer spending by an additional $5 more each year than the bottom income quintile, and $2.50 more than the 2nd income quintile, while there was not a statistically significant difference between the middle and top income quintiles. In the models in Column 2 we include non-summer observations and focus on the triple interaction of SES × summer × year. Again, the significant coefficients are for the lower income quintiles—the bottom income quintile decreased their spending by $3 less ($2.30 less for the 4th quintile) each year in summer vs. non-summer relative to the middle income quintile.

Table 2.

Regression Models of Parental Investments Over Time.

	Expenditures		Maternal Time		Paternal Time
	(1)	(2)	(3)	(4)	(5)	(6)
	Summer	Full Year	Summer	Full Year	Summer	Full Year
Education Models
BA or higher x Summer x Year		1.37 (1.23)		0.61 (1.07)		−1.20 (0.73)
BA or higher x Year	1.79 (1.25)	0.48 (0.60)	−0.87 (0.98)	−1.48 *** (0.43)	−1.46 * (0.65)	−0.22 (0.35)
BA or higher x Summer		23.6 (14.4)		2.24 (10.8)		5.30 (7.01)
Summer x Year		0.21 (0.26)		−1.03 (0.58)		0.38 (0.56)
Income Models
Bottom quintile x Summer x Year		−3.03 * (1.45)		−0.34 (1.41)		1.00 (1.29)
2nd quintile x Summer x Year		−2.31 * (1.15)		−2.26 (1.32)		−0.10 (1.12)
4th quintile x Summer x Year		−1.17 (1.23)		−2.24 (1.37)		−0.028 (1.14)
Top quintile x Summer x Year		−0.62 (2.02)		−1.10 (1.49)		−0.17 (1.17)
Bottom quintile x Year	−5.05 ** (1.59)	−1.88 * (0.83)	−1.19 (1.31)	−0.85 (0.59)	0.93 (1.10)	0.042 (0.61)
2nd quintile x Year	−2.50 * (1.21)	−0.31 (0.40)	−2.90 * (1.22)	−0.63 (0.62)	0.37 (1.03)	0.48 (0.55)
4th quintile x Year	−1.92 (1.34)	−0.67 (0.59)	−3.52 ** (1.20)	−1.19 (0.63)	0.0093 (1.02)	−0.016 (0.49)
Top quintile x Year	1.48 (2.18)	2.22 * (0.86)	−2.88 * (1.36)	−1.62 * (0.64)	0.22 (1.04)	0.41 (0.53)
Bottom quintile x Summer		32.7 * (13.8)		−0.51 (12.4)		−6.52 (11.8)
2nd quintile x Summer		19.6 * (9.47)		15.1 (11.7)		0.85 (10.1)
4th quintile x Summer		5.91 (11.9)		18.9 (12.8)		−2.08 (11.3)
Top quintile x Summer		49.6 * (20.7)		9.10 (14.1)		−5.65 (10.6)
Summer x Year		2.40 * (1.06)		0.54 (0.97)		−0.29 (0.77)
Observations	56,420	282,230	3,290	16,687	2,216	11,534

Standard errors in parentheses.

^*p < .05,

^**p < .01,

^***p < .001

Only interactions terms shown, but all models include main effects of each term as well as all controls. Education and income models are run separately. Models with both measures together are in the Appendix. Education reference group is families without a B.A. Income reference group is the middle income quintile.

We next examine how parental investments of time in children during the summer and school year have changed over time. The remaining panels of Fig. 2 show lowess plots charting maternal and parental childcare time from 2003 to 2019. We see some evidence of an increase in maternal childcare time for the middle income quintile in the summer months, but otherwise the gaps do not show any consistent trends.

Looking at childcare time in summer months, the coefficients by parents with a BA are negative for both mothers (Column 3) and fathers (Column 5), but statistically significant only for fathers. We also see a significant increase in summer childcare for mothers in the middle income quintile relative to the other groups, but not for fathers. There is no significant difference between summer and school months in how these SES gaps in paternal childcare time have changed (Models 4 and 6).

Taken together, both the most flexible visualization of the over-time trends in lowess plots of Fig. 2 and the linear models of Table 2 provide little evidence of a consistent change over time in summer gaps in parental investments of money or time.

6.3. Differences by age

Our last analyses examine whether summertime SES gaps in parental investments in children vary by the age of children. Column 1 of Table 3 presents the regression results predicting parental expenditures of money. Looking at the SES × Ages 12–17 coefficients, although we find no significant differences by age of child for more and less educated families during the summer, the interaction terms decrease with income. This is visualized in Fig. 3, which shows the predicted summer expenditures by SES and age of child in the first column. We see that—as already shown in the analyses of Hypothesis 1—families with a BA and those in the upper income quintiles already spend more on children regardless of age. In addition, across every SES category families are predicted to spend more on younger children. However, high-income families in the top income quartiles spend especially more on young children (~$350 month).

Table 3.

Regression Models of Summer Parental Investments by Age of Child.

	(1)	(2)	(3)
	Expenditures	Maternal Time	Paternal Time
Education Models
BA or higher x Children ages 12–17	−3.67 (18.9)	−5.23 (9.46)	−24.5 ** (8.56)
BA or higher	147.9 *** (12.9)	18.1 * (9.09)	17.6 * (7.75)
Children ages 12–17	−51.5 *** (4.92)	−42.4 ** (13.5)	−9.40 (9.97)
Income Models
Bottom quintile x Children ages 12–17	11.8 (18.2)	8.72 (13.6)	−12.5 (11.9)
2nd quintile x Children ages 12–17	8.21 (15.3)	6.15 (12.3)	−8.92 (9.60)
4th quintile x Children ages 12–17	−29.5 (17.8)	28.6 * (13.3)	−24.2 (12.7)
Top quintile x Children ages 12–17	−58.2 (31.1)	−20.9 (16.5)	−38.0 *** (11.1)
Bottom quintile	−23.7 (12.3)	−8.18 (12.9)	10.7 (8.94)
2nd quintile	−38.1 *** (8.84)	−7.55 (12.2)	3.61 (7.59)
4th quintile	49.9 *** (11.2)	−12.2 (10.6)	19.8 (10.4)
Top quintile	232.4 *** (23.3)	24.0 (15.1)	25.6 ** (9.74)
Children ages 12–17	−50.2 *** (13.7)	−52.0 ** (16.5)	−3.89 (10.2)
Observations	42,040	2,440	1,653

Standard errors in parentheses.

^*p < .05,

^**p < .01,

^***p < .001

Education and income models are run separately. Models with both measures together are in the Appendix. Education reference group is families without a B. A. Income reference group is the middle income quintile. Household controls and year fixed effects included in all models.

Fig. 3.

Predicted Summer Parental Investments by Age of Child Predicted values are based on models in Table 3.

Models 2 and 3 of Table 3 display the results of regression models testing whether SES gaps in maternal and paternal investments of child care time during the summer differ between households with younger and older children. The BA × older children coefficients are negative in both models, but statistically significant only for paternal childcare time—in the summer, fathers in college-educated families spend 25 more minutes doing primary childcare with younger children than older children than do fathers in other families. The income × age coefficients do not show any strong patterns: the 4th income quintile × Children ages 12–17 coefficient is statistically significant and positive (relative to the middle quintile) for mothers, while the top income quintile × Children ages 12–17 is statistically significant and negative for fathers.

Columns 2 and 3 of Fig. 3 show these SES gaps in childcare time during the summer by the age of household children. Notably, while there are apparent education- and income-based gaps in summer childcare time among mothers and fathers with household children aged 6–11 (the blue bars), these gaps are much less apparent among fathers with household children aged 12–17 (the red bars). In sum, the evidence shows more (but not overwhelming) support for Hypothesis 5A over 5B. SES gaps in summer parental expenditures appear somewhat wider and more consistent for families with younger children.

6.4. Robustness and alternate models

We estimate a number of alternate models and summarize the results here. All models described below are available in the supplemental appendix.

First, we re-estimated all models without any of the household controls (Supplemental Tables A3–A8). These models showed slightly larger SES gaps, consistent with the fact that we are no longer holding “all else equal” through the controls.

Second, we re-estimated all models using our SES indicators of education and income in the same model (Tables A9–A11). These models show similar or slightly smaller SES gaps, which makes sense given the strong correlation between family income and education.

Both income and education are distinct dimensions of SES, and scholars have extensively examined how both are important drivers of inequality in parental investment in children (Cheadle and Amato, 2011; Hao and Yeung, 2015). However, these measures are also strongly correlated, and thus we can run the risk of overcontrolling (Grätz, 2022). In general, these models of parental investments show similar or slightly smaller SES differences than in our main model, which is consistent with the fact that we are now estimating differences by education for families with the same income (or vice versa).

Third, for the expenditure models, we show we find similar results without dividing by the number of children (Tables A12–A14); and in the fixed effects models, we find nearly identical results even after leaving in the small portion of households whose education or income quintile changed. (Table A15).

Fourth, we estimated models that use several alternate measures of parental time investments in children (Tables A16–A25): primary childcare time per household child; splitting up primary childcare time into basic care and management of children on one hand, and play and teaching with children on the other; and additionally accounting for secondary childcare time, or time spent doing another activity while in care of a child under 13. In general, we find similar results when considering primary childcare time per kid as when considering primary childcare time, indicating that our main results are robust to considering economies of scale that are present in providing childcare for multiple children. When comparing childcare time spent in basic care and management to time spent in play and teaching, we find that summer SES gaps in the former category are larger than in the latter category, but for both categories, we find little evidence of seasonality in SES gaps. Finally, when using a measure of primary and secondary childcare time, we do see increases in this more encompassing measure of time during summer months, consistent with parents performing more secondary childcare time during the hours when school would have been in session. However, SES gaps in this measure of childcare time also appear to be similar in summer and non-summer months.

7. Discussion and Conclusion

In this paper, we first examine whether or not higher-SES families invest more money and time in their children during the summer than lower-SES families; then, we examine whether or not these (expected) gaps in parental investments are larger in the summer than during the school year. We find gaps by parental education for both parental investments of both time and money during the summer, and we find large gaps by parental income for expenditures on children. These gaps for expenditures (but not time use) are larger in the summer than during non-summer months (including in models with household fixed effects). When considering both forms of investment together, the net result is greater SES gaps in parental investments during summer.

An important limitation of the expenditure analysis already noted is that the data record the month of the expense. However, as discussed in the Data section, we would expect this to downwardly bias the differences between seasons (specifically affecting the tests of Hypotheses 2 and 4 for the expenditure analysis). Thus, our estimates of seasonal differences are likely conservative, and it is quite possible that class gaps in the investments purchased by parents that children actually experience during the summer are wider than our models actually report.

Are these gaps that we do find meaningful? One way to consider the effect sizes is to compare it to other research. For example, Schneider, Hastings, and LaBriola (2018) found that—using the same measure of financial parental investments—the gap in 2014 between children in the bottom quartile and the middle 50% was about $35 and between the middle 50% and the top 10% was about $200 (after adjusting to 2021 real dollars and dividing into months). We can also look to existing literature on the returns to investments in children. Recent evidence from studies of school finance reform suggests that a $1000 in annual spending ($111 if divided over a 9-month period) per pupil is associated with a 0.12–0.24 standard deviation increase in school achievement (Lafortune, Rothstein, and Schanzenbach, 2018, p. 4). By both metrics, our estimated summertime gaps by education and income are comparable to this, while we see based on the interaction terms that this gap further increases by about a third in the summer.

Although we do not find strong evidence of seasonality of SES gaps in parental investments of time, the summertime SES gaps we do observe are consequential. For example, Hsin and Felfe (2014) estimate that an additional hour per week of maternal educational childcare time is associated with 0.014 standard deviation increases in test scores, while Price and Kalil (2019) estimate that a standard deviation increase in mother-child reading time is associated with a 0.80 standard deviation in child reading achievement. It should be underlined that the effects of these investments of money and time are likely to accumulate throughout childhood, potentially providing cumulative advantages to higher-SES children that lead to achievement gaps in adulthood (DiPrete and Eirich, 2006).

Next, we examine the over-time dimensions of the seasonality of SES gaps in parental investments in children. We hypothesized there would be a widening of summer SES gaps in parental investments in children over time, reflecting the rise of intensive parenting practices associated with concerted cultivation. However, we do not find significant evidence that they grew more in the summer over the period of our analysis or that they have widened more over time relative to during the school year. This suggests summer has not been a key source of increases in intensive parenting in the last two decades, though it is possible that summer SES gaps in parental investment grew during the 1980s and 1990s alongside the non-summer months (Altintas, 2016; Kornrich, 2016; Kornrich & Furstenberg, 2013; Schneider, Hastings, & LaBriola, 2018).

Finally, we examined differences in summer parental investments by children’s ages. We find that SES gaps in paternal investments of time appear to be driven by SES gaps in investments targeted to younger school-aged children, though we do not see this pattern for maternal investments of time or parental expenditures on children. Though much of the narrative around concerted cultivation has focused on higher-SES parents investing in activities for older children to help them get into an elite college, some research suggests investments targeted at younger children have greater returns (Heckman, 2006; Chetty et al., 2016), and thus larger SES gaps in parental investments at this earlier stage seem more likely to be consequential for long-term inequalities in children’s outcomes. This is consistent with earlier research that found a “developmental gradient” wherein higher-SES mothers not only spend more with children but also better tailor the composition of their childcare time to meet children’s needs (Kalil et al., 2012).

Our work provides the first analysis of the seasonality of parental investments that includes both time and money. Considering these together is important because investments of time and money can involve tradeoffs, and, empirically, we find some notable differences in our results between the models of expenditures and time use. Taking both into account more fully reveals the extent of inequality in parental investments by SES.

These findings may also have implications for the summer learning gap debate (Alexander, Doris, & Olson, 2007; Downey, Paul, & Broh, 2004; Entwisle & Alexander, 1995; von Hippel et al., 2018; von Hippel and Hamrock, 2019; Workman et al., 2023). Parental investments are often presumed to be a mechanism for summer learning (or lack thereof), yet it is not possible to identify, from the data used here, how parental investment gaps translate into achievement gaps. However, the most recent research to date finds achievement gaps do not grow substantially more in the summer than in the school year. More generally, recent evidence suggests school itself neither greatly exacerbates nor compensates for inequalities in children’s family contexts (von Hippel et al., 2018; Skopek and Passaretta, 2021; Passaretta and Skopek, 2021).

Our findings also suggest that research designs to measure this gap based on differences in test scores before and after summer do not neatly capture a constant effect of the home environment on learning outcomes. As SES gaps in parental investments of money in children are, in fact, larger during summer months, any summer learning gaps would reflect both the loss of investments from school and changes in parental investments that children experience at home. More broadly, to understand the role of parental investments in learning, future research is needed that can link comprehensive measures of parental investments (measured across seasons) to children’s achievement—data that currently does not exist to our knowledge. Future research could also focus more on how inequalities in parental investments before children are of school age are associated with children’s achievement in their first year of schooling.

Our findings also specifically provide new evidence regarding the link between public and parental investments in children. When the “faucet” of public investments is turned off, higher-SES families respond by turning up some of their own parental investments in a way that lower-income and less-education families do not. More work is needed to understand how public expenditures in summer (e.g., summer school; day camps; recreation leagues) might directly affect private investments in children. To the extent that high-SES parents respond to decreases in public investments during the summer by “turning up” their private investments in their children, this may imply that increases in public investments in children during the summer may lead high-SES parents to “turn down” their investments in children. This has important policy implications, as it would reduce inequality in the total amount of resources that children receive and potentially help to narrow the summer learning gap. Our results suggest the “turning up” part is relatively modest—in reality, we find that the other faucet for children from SES families “stays on” throughout the summer. Thus, it is also possible that policies to create non-school, targeted summer programming could reduce inequalities, since it is the time of year when children’s schedules—especially for those from lower-SES families—may be the most flexible.

Parenting is a long-term, year-round endeavor, and there is no “summer break” for parents’ investments in children. Scholars have already noted how increased inequalities in resources can increasingly lead to “diverging destinies” in children’s outcomes. (McLanahan, 2004; Cooper and Pugh, 2020), in part through different parental investments of time and money. Understanding the timing of these investments throughout the year is essential for understanding the processes that lead to stratification in children’s life chances and later-life outcomes.