Job search and hiring in local labor markets: Spillovers in regional matching functions *

• We estimate regional matching functions on NUTS 3 level (2000 - 2010) for Germany.
• Spillovers account for 75% of the elasticity between unemployment and matches of 0.4.
• The Spatial Durbin Error Model fits the data best, suggesting ‘local’ spillovers.
• The results differ only slightly between nine different spatial weight matrices.
• Major labor market reforms have increased the spatial mobility of the unemployed.

In this paper we take a fresh look at the job matching process within local labor markets in Germany. Drawing on smaller geographic units than the previous literature, we estimate regional matching functions on NUTS 3 level for the years 2000 to 2010. The elasticity between matches and unemployment ranges between 0.4 and 0.5 with 75% of this effect being driven by the impact that unemployment has on matches in neighboring regions. The effect of vacancies on matches is substantially smaller but also robustly positive. Bayesian model comparison tests suggest that spillovers from unemployment and vacancies are confined to local labor markets, which are best approximated by geographical distance rather than by present or past infrastructure or commuter numbers. Spillovers from unemployment arise exclusively after a series of major labor market reforms (‘Hartz Reforms') have been implemented between 2003 and 2005, indicating that the reforms have contributed to an increased spatial mobility of the unemployed.