How U.S. Parents Adjusted Their Time Use During the COVID School Closures

In March 2020, in response to the COVID-19 pandemic, U.S. schools nationwide moved from in-person to remote instruction. The suspension of in-person learning meant the loss of about 30 hours of childcare time every week, which meant parents had to seek out alternative childcare arrangements. These school closures were long lasting, with many schools not reopening for permanent on-site instruction for about a year.

How exactly did this loss of childcare time affect parents' use of their own time? Enghin Atalay, Ryan Kobler, and Ryan Michaels of the Federal Reserve Bank of Philadelphia address this topic in their paper, "School Closures, Parental Labor Supply, and Time Use." Specifically, they study how parents' personal preferences, available technologies, and other constraints influenced how and to what extent they adjusted their time use during the period of remote school instruction.

When U.S. schools first shifted to remote instruction, it was widely thought that the school closures would greatly impact parents' time spent working.¹ But the subsequent literature did not show a dramatic reduction in the number of hours parents worked.² This unexpected finding prompted Atalay, Kobler, and Michaels to revisit and extend this research, with the goal of better understanding how parents adjusted their time use.

They modeled the relationship between parents' work time and the type of schooling instruction using "foot traffic" at school campuses (based on SafeGraph's mobile phone location data) and the Bureau of Labor Statistics' (BLS) Current Population Survey data on hours worked.³ Next, they analyzed the BLS's American Time Use Survey (ATUS) data to better understand what factors influenced how parents adjusted their time spent on many activities in addition to work. Finally, they interpreted their empirical results through the lens of simple theories of parental time allocation.

Using their simplest model specification, which relates work time to the type of schooling instruction, they find that the school closures had little impact on parents' time working. However, extensions of this simple model yield quite different and conflicting results.

First, they extended their model to account for nationwide trends in parental work hours over time - and not just differences in hours worked related to the type of schooling instruction. They find that time at work increased by 2.5 hours for female parents when in-person learning resumed, compared with 1.6 hours for male parents. These results are about the same for parents across levels of educational attainment, except for college-educated fathers, who did not adjust their work time at all.

They then extended their model to account for long-standing differences in parental work hours across geographical areas - and not just differences in hours worked related to areas' schooling formats in the pandemic. This specification leads to a strikingly different result: Changes in schooling format had no significant effect on work hours. The authors suggest the very simple explanation that counties where schools reopened relatively late also tended to be counties where parents generally worked less even before the pandemic.

These findings show that the results from the simpler models are likely to overstate the true effect of school closures. The authors contend that the true labor supply response is likely to be a very small share of the 30 hours per week of on-site childcare time provided by the reopened schools, and thus "parents must have adjusted time use on other margins so as to both attend to children and supply labor."

For additional insights, the authors turned their attention to the ATUS. Their analysis of the ATUS data corroborates their initial finding that school closures led to relatively little change in hours worked. The ATUS data also reveal surprisingly little change in any other category of time use: Parents barely adjusted their amount of leisure or home production during the school closures.⁴

But how did parents manage to sustain their work hours if they didn't trim their time spent on these other activities? One answer revealed in the ATUS data is that some parents were able to work from home while simultaneously supervising children. This access to telework, the authors find, "was likely one means by which some parents insulated their schedules from pandemic disruptions." In particular, college-educated parents spent six more hours per week working from home while simultaneously caring for their children. However, among those without a college education, there was no change in hours spent teleworking, which can be explained by lower-skilled workers generally having fewer telework opportunities.

The ATUS data also reveal that nonparental care was a second means of adjusting to school closures. Specifically, the ATUS provides a measure of time that individuals over the age of 60 spend with children who are not their own. (Many of these survey respondents, the authors suspect, are grandparents.) The data show that this form of nonparental care increased in areas with virtual schooling. The effect was significant for respondents without a college degree, who allocated up to four more hours per week for the care of others' children.⁵

Finally, the authors incorporate teleworking and nonparental care into simple theories of parental time allocation. They show how their empirical results can be used to assess key properties of these theories, such as parents' willingness to substitute nonparental care for their own time with children. Armed with these insights, the theories can then be applied to study new topics - for instance, to evaluate how childcare subsidies would affect parental time use.

Atalay, Kobler, and Michaels's findings help to inform public policy and advance our understanding of parental time use. The authors hope that, given the small sample size of the ATUS data and the relative scarcity of data on nonparental care time, their paper will also stimulate efforts to expand data collection in these areas.