Dan Wilson, Vice President, Economic Research, Federal Reserve Bank of San Francisco
Abstract: This paper considers how the spatial distribution of U.S. economic activity may be impacted by climate change in the decades ahead. First, we use recently developed daily weather data at the county level from 1951-2020 to construct annual measures of average and extreme precipitation and the number of days per year in various temperature bins. Second, we exploit the substantial geographic and temporal variation in these data to estimate the longer-run effects of weather on local population, employment, wages, and house prices. In contrast to prior studies inferring longer run effects from short-run effects, we estimate the full dynamic response of economic outcomes to weather shocks using panel autoregressive distributed lag (ADL) models covering 30 years of weather. We show this long-lagged ADL model accounts for both ex-post and ex-ante adaptation to the extent that agents have adaptive expectations. Our historical results point to long-lasting effects of extreme temperature and precipitation on local employment and population. Lastly, we use the estimated panel models to project the spatial distribution of these local outcomes in 2050 using historical and projected county-level weather. The projections suggest that climate change could increase aggregate population reallocation by 27 to 55% depending on the climate scenario, with population and employment generally shifting in economic activity away from the southeast and toward the mountain west.