Adaptation and the Mortality Effects of Temperature Across U.S. Climate Regions

We study heterogeneity in the relationship between temperature and mortality across U.S. climate regions and its implications for climate adaptation. Using exogenous variation in temperature and data on all elderly Medicare beneficiaries from 1992 – 2011, we show that the mortality effect of hot days is much larger in cool ZIP codes than in warm ones and that the opposite is true for cold days. We attribute this heterogeneity to historical climate adaptation. As one adaptive mechanism, air conditioning penetration explains nearly all of the regional heterogeneity in heat-driven morality but not cold-driven mortality. Combining these results with projected changes in local temperature distributions by the end of the century, we show that failure to incorporate climate heterogeneity in temperature effects can lead to mortality predictions that are wrong in sign for both cool and warm climates. Allowing regions to adapt to future climate according to the degree of climate adaptation currently observed across climates yields mortality impacts of climate change that are much lower than those estimated without allowing for adaptation, and possibly even negative.