School Closures and NPIs during the Great Influenza Pandemic in Sweden
Final Report Abstract
As the Corona pandemic has unfolded, governments throughout the world implemented various non-pharmaceutical interventions (NPIs) – and one of the most prominent measures has been school closures. However, school closures are highly controversial given that they potentially deprive children of their right to education, with potentially more adverse effects in vulnerable populations. While the effects of such measures on disease spread materialise in the short term, the effects on human capital accumulation can be observed decades later. In order to be able to compare these immediate and very long-term effect within a unified frame-work, the project has studied measures taken in Sweden during the 1918-20 influenza pandemic, combining existing and purpose-built datasets of exceptional detail, coverage, and accuracy. Collecting data on the implementation of school closures in each one of Sweden’s 2,500 school districts at the time, and combining them with individual-level data on exposure, mortality, educational attainment and earnings, the project studies the effects of school closures across a wide range of outcomes and time intervals using difference-in-differences. One central finding of the project is that a causal identification of effects is challenging, since the probability of a school closure is extremely responsive to surges in influenza infections; a finding which was possible solely due to the fine-grained mortality data generated for the project. To address this challenge, the project has developed a design that can be used to estimate effects of closures even in the presence of extreme reverse causality. Comparing mortality outcomes across fast-and slow-closing schools before and after closures, it is estimated that peak influenza mortality may be reduced by half by a fast closure. Exploiting detailed information on earnings and educational attainment – as well as exact date and parish of birth – for the entire population, we also estimate the impact of school closures on these later outcomes, considering outcomes up to 50 years after the pandemic. In this part causal identification is more straightforward since children who were in school by 1918 may be compared to younger cohorts who were not yet enrolled. The central finding from this analysis is that there are no discernible effects on educational attainment or earnings from these school closures. This may to some extent be due to their relatively short duration, averaging 20 days. In addition to these core findings, the project has generated a number of valuable datasets that are of exceptional quality and can be linked to various existing data sources. The project has also led to several spin-off studies, which either explore the same pandemic from different angles or examine general mortality during the same historical period.
Publications
-
The 1918–1919 Influenza Pandemic in Economic History. Oxford Research Encyclopedia of Economics and Finance. Oxford University Press.
Karlsson, Martin; Kühnle, Daniel & Prodromidis, Nikolaos
-
School Closures, Mortality, and Human Capital: Evidence from the Universe of Closures during the 1918 Pandemic in Sweden. CEPR Discussion Paper No. 18399.
Dahl, C. M., Hansen, C. W., Jensen, P. S., Karlsson, M. & Kühnle, D.
-
Pollution and Mortality: Evidence from early 20th-Century Sweden.
Haylock, M., Karlsson, M. & Orbrizan, M.
