46th Congress of the European Regional Science Association: "Enlargement, Southern Europe and the Mediterranean", August 30th - September 3rd, 2006, Volos, Greece
In this paper, we use the Beveridge relationship to address the effectiveness of the matching process, that brings workers searching for jobs together with employers searching for workers. For a fixed matching technology, the curve yields a negative relation between the unemployment rate and the rate of vacancies. Movements along a curve reflect adjustments over the business cycle. In a recession vacancies are closed, and workers enter the unemployed. Shifts of the curve are more important here, as they point to structural change. For example, an outward shift of the curve indicates an in-creased mismatch, perhaps due to a deterioration in human capital of the unemployed or changes in the unemployment benefit system, which affects the willingness of the un-employed to fill out vacancies. Empirical estimates rely on panel data. A sample of 180 regional labour markets is em-ployed, and the sample period runs from 1993 to 2004. The regional labour markets are seperated on the base of flows of the job commuters and correspond to travel-to-work areas. Due to common or idiosyncratic shocks, however, the cross sections are not inde-pendent. Instead, they are tied together to some extent, and the spillovers account for spatial effects. As these patterns can have an impact on the correlation between unem-ployment and vacancy rates, the results of OLS regressions are eventually biased. Thus the Beveridge curve is efficiently estimated by a spatial procedure, where regional de-pendencies are taken into account. No previous paper has investigated a similar broad regional dataset so far. The eigenfunction decomposition approach suggested by Griffith (1996, 2000) is used to identify spatial and non-spatial components in regression analysis. As the spatial pat-tern may vary over time, inference is conducted on the base of a spatial seemingly unre-lated regressions (spatial SUR) model. Due to this setup, efficient estimates for the Beveridge relationship are obtained. Time dummies are used to control for shifts in the curve. The empirical results provide some indication that the degree of job mismatch has increased over the recent periods.