Local Labor Markets? Evidence from a Spatial Job Search Model on Large Scale French Micro-Data

We extend the wage-posting model of Burdett and Mortensen (1998) to space and propose an estimation method for job mobility costs without any prior assumption about the latter or local distributions of job offers. Our structural model introduces job mobility costs, spatial coupling and arbitration between pairs of geographical areas on a forward-looking basis. Costs are defined in a broad sense and hamper job mobility, thus benefiting from remote job opportunities. Distributions of wages within each geographical area and transitions flows are used for identification. Then, we present an application to low-skill workers at the employment zone level using a large administrative dataset (comprehensive DADS) exploited for the first time. We find that geography play an important role, with the existence of local labor markets. First, the spatial barrier bites at very short radius, less than fifty kilometres (thirty miles). Second, job mobility costs are high in average, between e70,000 and e100,000, from 16% to 21% of the reservation utility and represent about 10%-15% of the job value. Third, the latter are highly heterogeneous and asymmetric across locations, from e9,000 to e170,000. Fourth, space costs around e1,100 per kilometre of employment displacement, in consistency with a gravity model