Spatial differences in unemployment rates are large and persistent. Our objective is to develop a quantitative theory of these differences. Considering the local unemployment rate as an (equilibrium) outcome between workers and firms, we document facts on both sides of the labor market. First, we confirm recent findings that differences in separation rates (unemployment inflows) are more important than differences in job-finding rates (unemployment outflows) in accounting for local unemployment differences. Second, we present novel facts on the differences in job-creation and job-filling rates. We document that local labor markets with lower unemployment rates are tighter, have lower vacancy filling rates, and higher average vacancy duration. These facts constitute empirical regularities in the United States, Germany, and the United Kingdom. On the theory side, we demonstrate that a Diamond-Mortensen-Pissarides model with endogenous separations quantitatively accounts for all the documented facts on the geography of job creation and job destruction. Furthermore, we demonstrate that this model not only matches labor market differences across space but also over the business cycle.