Cross-National Measures of the Intensity of COVID-19 Public Health Policies

We show in this research note that the complex nature of COVID-19 policy responses means that models trying to identify the effect of individual policies can produce spurious results unless they take into account systematic measurement error. Employing a simulation of the policymaking process, we find that regression analyses of multiple related policy indicators results in spurious inferences due to measurement error. To remedy this issue, we estimate six new indices of the overall intensity of different types of COVID-19 restrictions that incorporate policymaker intentions behind the design of similar policies. These indices are derived from novel granular data on COVID-19 restrictions from the CoronaNet dataset, and we augment this data with the Oxford COVID-19 Government Response Tracker dataset. To gain estimates with uncertainty, we use a Bayesian time-varying measurement model that provides time-varying policy intensity scores from January 1, 2020, to May 1, 2021, for over 180 countries. We show with these measures that regression models of policy scores on important pandemic outcomes are robust to measurement error and fully incorporate uncertainty.