Does Temperature Affect Productivity? Evidence over the Short and Long Run

This paper studies the relationship between temperature and productivity across space and time. First, using six separate datasets—the annual Current Population Survey (CPS, 1963-2015), the monthly CPS (1994-2014), the Census (1900-2015), the Quarterly Census of Employment andWages (QCEW, 1990-2014), the Longitudinal Employer-Household Dynamics (LEHD, 1997-2014), and the County Business Patterns (1987-2014)—I find no association between temperature and productivity. These results are robust to different functional form, different time steps, different measurements of temperature, different geographic aggregations, and heterogeneous treatment effects by industry and occupation. Second, I build and estimate a general equilibrium spatial sorting model where temperature affects the locational choice of both households and firms and affects productivity differently for each industry: agriculture, manufacturing, services, and business. Third, I use the model to estimate a counterfactual change in the distribution of temperature to quantify the potential costs of climate change allowing for the endogenous reallocation of workers across metro areas. JEL: O44, J31, Q51, Q54, R11