Close
Matthew Gentzkow with Amy Finkelstein & Heidi Williams
Abstract
Mortality rates vary substantially across the US. In this paper, we use mortality outcomes of migrants in the Medicare population to understand the origins of these differences. We decompose areas' overall four-year mortality rates into two components: (i) the health capital stock of their residents, including the effect of genetic endowments and prior health investments, and (ii) the immediate causal effect of local factors such as the quality and quantity of health care, climate, and pollution. To identify the causal effects, we apply a novel strategy to adjust for unobserved differences in health capital that may be correlated with migrants' choice of destination. We find that these causal effects vary substantially across areas, with a standard deviation at least one third of the overall cross-are standard deviation of mortality. We find that they are only imperfectly correlated with average mortality, so that the share of cross-sectional variation they explain is relatively small. Finally, we show that areas with more non-profit and high-quality hospitals, more primary care doctors and specialists per capita, and lower temperature, homicide, and accident rates all tend to have more favorable causal effects.