Comment on ‘Government mandated lockdowns do not reduce COVID-19 deaths: implications for evaluating the stringent New Zealand response’

In ‘Government mandated lockdowns do not reduce COVID-19 deaths: implications for evaluating the stringent New Zealand response’ (New Zealand Economic Papers, 2020), Gibson claims that ‘Lockdowns do not reduce COVID-19 deaths’ on the basis of an instrument variable linear regression on county-level cross-sectional data in the United States. Here we argue that Gibson’s analysis is not robust. In particular, Gibson (i) neglects the spatio-temporal heterogeneity in the spread of COVID-19 in the United States, namely that spread was from well-connected urban counties to more isolated rural counties; (ii) selects cross-sections at arbitrary times from what is an on-going spatially heterogeneous dynamical process, introducing bias that he fails to control for; and (iii) makes a choice of instrument variable (political affiliation) that is correlated with the heterogeneity (and therefore the bias) and that could plausibly influence the output variable in his regression independently of the explanatory variable.