Do lockdowns work? A counterfactual for Sweden
A paper by Benjamin Born, Alexander M. Dietrich and Gernot J. Müller study the case of Sweden – one of the few countries without a lockdown – and use synthetic control techniques to develop a counterfactual lockdown scenario. They find that a lockdown would not have helped much in Sweden.
The authors use a group of European countries to construct a doppelganger that behaves just like Sweden in terms of infections before the lockdown. They use observations for European countries, considering as day 1 the day when the number of infections surpasses a threshold of one infection per one million inhabitants. Since no lockdown was imposed within 13 days from this so-called day 1, they impose that a doppelganger of Sweden behaves like Sweden during these first 13 days in terms of infections and has similar population size and urbanization rate. The (almost) perfect doppelganger of Sweden is a weighted average of the Netherlands (weight of 0.39), Denmark (0.26), Finland (0.19), Norway (0.15), and Portugal (0.01).
With this methodology, the authors find that infection dynamics in the doppelganger since the lockdown do not systematically differ from the observed dynamics in Sweden. The same is true for the dynamics of deaths from Covid-19. They draw the conclusion that a lockdown would not have helped much in Sweden.
Google mobility data show that Swedes adjusted their activities in similar ways as in the doppelganger, although to a somewhat lesser extent. Therefore, voluntary social restraint goes some way in resolving the “lockdown puzzle”. Related to the issue of external validity, the authors clarify that the fact that a lockdown would not have made a big difference for Sweden does not mean that it didn’t work elsewhere.
Link to paper here.
Paola Di Casola*
Posted by Maria Perrotta Berlin,
*The opinions expressed in this post are the sole responsibility of the author and should not be interpreted as reflecting the views of Sveriges Riksbank.