Excess mortality data avoid miscounting deaths from under-reported pandemic-related deaths and other health conditions left untreated. This paper analyses cumulative excess mortality, which is a more robust measure (particularly early in the pandemic) than the case counts or counts of COVID-19 deaths typically used in epidemiological studies. Definitions and data measurement issues around excess mortality are surveyed, considering data quality and comparability both internationally and within the US. An analytical review of the related spatial literature, mostly at the county level, addresses choice of functional form, the exclusion of key controls, and types of biases including endogeneity. A state-level study offers a useful, complementary perspective to county-level studies. This is the first state-level analysis of excess mortality across the 51 US states, showing, for the three waves of the pandemic, the effects of racial composition, age structure, population density, poverty, income, critical care capacity and other structural features, and of pandemic-related governance actions and political allegiance. The apparently strong positive correlation between the Democrat vote share in the 2016 election and high excess mortality in the first wave of the 2020 pandemic is an illusion: with the relevant socio-economic controls, it reverses strikingly. In all three waves of the pandemic, conditional on the other risk factors, states with higher Democrat vote shares experienced lower excess mortality. The same conclusion holds for a cross-section analysis for the entire year from the arrival of the pandemic in the US. This is consistent with spatial studies at the county level, linking partisan allegiance with private attitudes, behaviour and COVID-19 deaths.

This is joint work with Janine Aron.