Geographic Dispersion of Economic Shocks: Evidence from the Fracking Revolution: Reply

摘要

Measuring the geographic spillovers from an economic shock remains a challenging econometric problem. In Feyrer, Mansur, and Sacerdote (2017) we study the propagation of positive shocks from the recent boom in oil and gas production in the United States. We regress changes in income per capita on new energy production per capita within increasingly larger geographic circles. James and Smith (2020) proposes instead a single regression of county income per capita on energy production from successively larger donuts around the county. This method controls for production outside of the circle of interest and is likely the appropriate estimation method for estimating the impact of within-county production. Their results suggest that FMS overestimates the impact of new production. We show that we can incorporate similar controls using our basic estimation method and that (unlike James and Smith) these controls do not significantly change our results. To explore these differences, we perform simulation exercises which show that the James-Smith estimation method is biased downward with the heterogeneous population distributions across counties that we observe in the data.