Confounding, spillovers, and interactions influence estimates of social distancing policy effects

Abstract

Understanding the effectiveness of social distancing policies aimed at curbing the spread of COVID-19 and other infectious diseases is a pressing need that a growing number of studies seek to address. Using fine-grained mobility and epidemiological data, we show that widely used methods to estimate the effects of policies are highly sensitive to accounting for the facts that policies are often enacted simultaneously, that they can complement, or substitute for, each other, and that they can spill over as persons move across geographies. Accounting for confounding and spillovers can change conclusions about the direction and magnitude of the effects of social distancing policies, while failing to consider interactions may hamper the design of optimal public policy packages.

Keywords: Social distancing, policy design, spillovers, confounding, interactions

JEL Classification: I10, I18, C18, C23