Table 1 Policy summary

Background

Studies have not reached a consensus about the effectiveness of school closures on reducing the spread of COVID-19, partly because of analytical challenges. To estimate the causal effects strictly in Japan in spring 2020, we compared the number of cases between municipalities with open schools and municipalities with closed schools, both of which were selected so that they are similar in terms of dozens of features (for example, past number of cases; social, economic and political factors; and NPIs other than school closures such as stay-at-home warnings and business closures).

Main findings and limitations

On average, the numbers of cases per 100,000 residents in municipalities with closed schools were not statistically significantly smaller than those in municipalities with open schools. Because both groups are comparable except for school closure status, the difference between them should only be attributed to school closure status. Thus, we did not find any evidence that school closures significantly decreased the spread of COVID-19. It is possible that school closures lowered COVID-19 cases outside Japan and/or after spring 2020, in particular in the presence of new variants of SARS-CoV-2.

Policy implications

Our null results concerning the supposed benefits of closing schools suggest that policymakers should be cautious when considering similar policies in the future, especially given the substantial costs to the well-being of both children and parents. Our recommendation is that governments should monitor SARS-CoV-2 infection rates and school closures at a granular level (for example, municipality or school district) in real time (for example, daily) to inform their policy decisions.