We document large and persistent spatial dispersion in unemployment rates, vacancies, labor market tightness, labor market flows, and wages for Germany on a granular regional level. We show that in the 1990s differences in inflows from employment to unemployment were the key driver of regional dispersion in unemployment rates while in the 2000s outflows became more important. To account for the documented regional dispersion we develop a spatial search and matching model with risk-averse agents, endogenous separations and unobservable search effort that leads to moral hazard and quantify the relative importance of 4 potential structural driving forces: dispersion in productivity, in the bargaining strength of workers, in idiosyncratic risk components and in regional matching efficiency. Based on region-specific estimates of these factors we then study the resulting policy trade-off between insurance, regional redistribution and efficiency. We design (optimal) region-specific labor market policies that can be implemented using hiring subsidies, layoff taxes, unemployment insurance benefits and transfers financed by social insurance contributions. We find that a move towards an optimal tax system that explicitly conditions on regional characteristics could lead to sizable welfare and employment gains.