?
Instrumental variables analysis of COVID-19 vaccination in Moscow
This article examines the effect of COVID-19 vaccination on the disease dynamics in Moscow using an instrumental variables (IV) approach. Based on weekly data on cases, deaths, and mobility, I compare ordinary least squares (OLS) results with two-stage least squares (TSLS) estimates, addressing endogeneity in vaccination and behavior. The TSLS estimates imply that the effect on new cases is about ten times larger than suggested by OLS, indicating a substantial downward bias in naïve models. Moreover, the two doses of Sputnik V have distinct effects: the first mainly lowers infection risk, while the second notably reduces mortality among infected individuals. These findings underscore the importance of high vaccination coverage, clear public communication, and integration of behavioral dynamics as critical elements of public health policy in large urban settings.