The Geography of Unemployment

Abstract

Unemployment rates differ widely across local labor markets. I offer new empirical evidence that high local unemployment emerges because of elevated local job losing rates. Local employers, rather than local workers, account for most of spatial gaps in job stability. I then propose a theory in which spatial differences in job loss arise endogenously, due to the spatial sorting of heterogeneous employers across local labor markets. Labor market frictions induce productive employers to over-value locating close to each other. The optimal policy incentivizes them to relocate to areas with high job losing rates, providing a rationale for commonly used place-based policies. I estimate the model using French administrative data. The estimated model accounts for over three fourths of the cross-sectional dispersion in unemployment rates, as well as for the respective contributions of job losing and job finding rates. Employers' inefficient location choices amplify spatial unemployment differentials five-fold. Both real-world and optimal place-based policies can yield sizable local and aggregate welfare gains.