While Covid-19 takes most of its death toll among the elderly, it also affects future generations through disrupting the education system. Unesco (https://en.unesco.org/covid19/educationresponse) publishes a map of the school closure around the globe over the Covid-19 period. In the spring of 2020, at the peak of the first wave, most schools were closed around the globe, affecting as many as 1.5 billion children. Most European countries have tried to avoid shutting down schools again during the second Covod wave, although some are under heavy pressure these days. In Italy, for instance, about half of the children in school age are currently following distance learning programs.
In the United States, the situation varies significantly across states and school districts. In October, New York was the first big city in the country to reopen its public schools. Yet, it did not last long. On November 18, Mayor de Blasio announced that schools would shut down again following the new surge in test positivity rate. Schools are still totally or partially closed in Latin America and in India. Unesco estimates that about 950 million children continue to be affected by school closures.
Experts on education are studying the effect of this unique event. While we have some information about the immediate effects of learning losses, it is more difficult to foresee the long-run effects of school closures. Are children going to catch-up? Or will learning losses mark their entire lifetime? Will the effect be homogeneous in the population? Or will it be stronger for the more fragile segments, exacerbating income inequality? In lack of data, we must resort to models based on past behavior and events to forecast the future evolution.
Wealthier parents can hire tutors
In an ongoing collaborative project involving Francesco Agostinelli (University of Pennsylvania), Giuseppe Sorrenti (University of Amsterdam), and Matthias Doepke (Northwestern University), we study the unequal effects of a prolonged disruption of in-class activity. We argue that the disruption goes beyond the lack of access to professional instructors. Another important effect works through peers. Public schools are an important vehicle of socialization. They allow – sometimes even force – children from very different upbringing and socio-economic backgrounds to meet, sit together, and pal up with each other.
School closures tend to restrict children’s interactions within smaller inner circles, such as their families or the block where they live. Using the Add-Heath data, a representative dataset of high school children in the United States in the 1990s, we find that even in a country where socio-economic segregation across schools is notoriously large, public schools promote equalization through peers. For instance, segregation would increase dramatically if children interacted only with peers living in the same block rather than with all those attending the same school. Many studies, including ours, find significant peer effects on the cognitive and noncognitive development of children.
Restrictions on peer interactions compound with other gaps for disadvantaged families. First, wealthier parents can hire tutors. Each of us academics receive countless requests from parents offering generous private tutoring deals. Second, and perhaps more important, well-educated parents can cover for school instructors by spending more time with their children. Time diaries suggest that parents have quadrupled the time spent on active child care relative to normal times.
Parents interfere with how their children select friends
However, both the quantity and the quality of the potential support differ greatly across families. A study by Adams-Prassl, Boneva, Golin, and Rauh documents a large heterogeneity across occupations in the extent to which parents can work remotely and be available to their children. These occupations are strongly associated with income and education. According to our calculations, the family income gradient (i.e., the difference between poorer and richer families) in terms of time spent on active child care went up by a factor of four relative to pre-Covid time.
Finally, our earlier study («It Takes a Village: the Economics of Parenting with Neighborhood and Peer Effects,» NBER WP27050, April 2020) shows that parents more or less consciously tend to push their children to hang around with children who do well in school and to stay away from those who exhibit disruptive behavior. Parental interference with how their children select friends is especially strong in unequal communities.
This pattern – that we already observe in normal times – is likely to become stronger when children return to school more different in terms of cognition and behavior. In other words, if Covid separation exacerbates learning gaps, better-off parents will try harder to keep their children away from the more problematic kids.
Effects of the shock are unequal
Our study considers the compounded effect of these factors. We model Covid as a negative shock that reduces the productivity of the learning technology. We infer the size of the shock from earlier events such as natural disaster and the effect of summer school closure in regular years.
While this shock hits all children, three forces makes its effect unequal. First, during Covid time children stop meeting in schools and are restricted to interact with other children living in the same block. Second, families differ in their ability to cope with the lack of in-person instruction – with differences reflecting the possibility for parents to work from home during Covid. Third, peer effects are influenced both before and after the school closure by the tendency for parents to push their children to stay away from «rotten apples.»
The results are striking. For children in tenth grade of the US school system (i.e., the second year of high school), we predict learning losses relative to a normal year that increase significantly as we move down the income ladder. These range between insignificant losses for the children of the 10% richest families to losses above 70% for the poorest 10% families. We can also forecast the extent to which schools can remedy for these losses after reopening (assuming they don’t take any special measures). We find that a large portion of the unequal losses persists until the end of the high-school cycle. On the one hand, the negative effects on cognition spill over to the children of richer families (e.g., through slowing down instruction), who suffer moderate learning losses in the order of 10% relative to an undisrupted high-school cycle. For the children of the 10% poorest families, the losses are in the order of 35%, namely, schools makes up for only 50% of the initial losses over three years after reopening.
Targetting disadvantaged families and groups
Some caveat are in order. First, the results are preliminary and we are working on validating them with further data. Second, the quantitative effects above refer to one year of disruption of in-school learning. This is realistic in some countries, though it might be an overstatement for many European countries. Finally, we ignore the effect of private tutoring that might allow richer families to further mitigate the effects of the Covid shock. Considering this factor would increase inequality even further.
In spite of these considerations, it is clear that the health crisis will leave a mark on future generations potentially accelerating the trend toward growing inequality and declining social mobility. The awareness of this problem calls for robust government intervention with policies in support of public education and human capital formation especially targeted to disadvantaged families and groups