The Geography of job creation and job destruction

Spatial differences in labor market performance are large and highly persistent. Using data from the United States, Germany, and the United Kingdom, we document striking similarities across these countries in the spatial differences in unemployment, vacancies, and vacancy filling, job finding, and separation rates. The novel facts on the geography of vacancies and vacancy filling are instrumental in guiding and disciplining the development of a theory of local labor market performance. We find that a spatial version of a Diamond–Mortensen–Pissarides model with endogenous separations and on-the-job search quantitatively accounts for all the documented empirical regularities. The model also quantitatively rationalizes why differences in job-separation rates have primary importance in inducing differences in unemployment across space while changes in the job-finding rate are the main driver in unemployment fluctuations over the business cycle. • Unemployment differences across local labor markets are large and persistent. • In high unemployment locations, tightness is lower and vacancy filling is faster. • A spatial DMP model can match all spatial facts of worker and job flows. • OJS reconciles differences of job finding across space and business cycles.