Roland Rathelot, CREST

 "The Geography of Job Search and Mismatch Unemployment"


Can we reduce unemployment by moving job seekers to areas with better job opportunities? To answer this question, we need data on the distance between job seekers and the jobs they apply to. Using novel data from the popular website, we quantify how application probability declines with distance from the job seekers' zip code of residence. 82% of applications are sent to jobs within the same city (Core-Based Statistical Area, CBSA), but only 46% are sent to jobs within the same county.

We build a simple search-and-matching framework in which job seekers have a distaste for distance and use our data to estimate its parameters. Using our model, we find that US unemployment could be reduced by up to 3% by reallocating job seekers across zip codes. This magnitude of mismatch is similar to what we find using data aggregated at the CBSA level. Our evidence suggests that the CBSA is an acceptable definition of a local labor market.