Converted using QALY (Quality Adjusted Life Years) used for standard cost-benefit analysis in health and infrastructure projects in Sweden, it is worth closing the school if it alone saves 270 lives a day. Since this is almost an order magnitude more than the difference in the daily number of deaths in Covid19 between Norway and Sweden, this can be viewed as a strong indication against school closing.

According to the report (available here) by Martin Andresen and co-authors, school closures cost NOK 1.7 billion a day. Norwegian and Swedish society and school systems are very similar, so the figure is probably comparable to Sweden. Translated by the ratio of Norway’s and Sweden’s GDP (approximately 5/3, Norwegian oil sector excluded, 1/1 exchange rate), it is about SEK 2.8 billion a day.

In Sweden, when investing in healthcare, roads etc., we use a value of one statistical life year (QALY) of SEK 0.5-1.5 million. To be generous to the idea of ​​school closure, I use the upper bound.

2.8 billion is then almost 1900 years of life. If we assume that the age for coronary death is 75 (which is lower than what it actually is, again to be generous to the idea of school closure), you get with the average life expectancy of 82 years that these life years correspond to 270 deaths a day in covid-19.

So if school closures save 270 lives per day then it might be worth it.

Let’s do a comparison of the death rates between the countries. In Sweden, as of time of writing, 2021 had died from Covid-19 (roughly six weeks from first fatality March 11). Meanwhile, the Norwegian death toll has been 191. Scaled up to the Swedish population it corresponds to 363 people. The difference is about 40 deaths a day. And during this period, the schools (except high schools) had been open in Sweden and closed in Norway (they closed March 13). Norway has been stricter in other ways too (public gathering, closures of restaurants etc.), but let’s be generous to school closures and assume the whole difference in deaths is due to school closures. Then still the death numbers are only around a seventh of what motivates school closure.

Note 1: The report also includes missed work by parents, but that is a small part of the cost. So we can also remove it and draw the same conclusion.

Note 2: The report ignores the fact that school closures most likely will affect weaker students, so their numbers are probably underestimation.

Note 3: Their numbers are in par with other studies they cite.

Note 4: It is difficult to estimate the cost of school closures, nothing historically comparable exists. One issue is whether we should use macroeconomic of microeconomic estimates. But since the benefits of school closures (from the report) are an order magnitude smaller than the costs the burden of proof lies on those who want to close the school: show that another calculation gives one seventh of the cost from the Andresen et al report. Alternatively the effect of temporary school closure is so long-lasting that we should put several months' future reduced deaths on its plus account.

Note 5: Adda (2016) is another study discussing school closures based on the regular flu by Adda (2016). While being a very interesting and rigorous study I view it has having less relevance as there school closures are quicker and are based on a different disease.

Finally: Are there such cost estimates for Sweden? If not, it would be of great value.

Daniel Spiro
Department of Economics
Uppsala University