Spatial Patterns of German Labor Market: Panel Data Analysis of Regional Unemployment

This chapter is devoted to the investigation of spatial spillover effects of the regional unemployment in Germany. Due to historical reasons, the differences between eastern and western regions of Germany persist over time. We explore the differences in the determinants of the regional unemployment as well as the differences in spatial effects by estimating spatial models. We use panel data for 407 out of 413 German regions (using the NUTS III regional structure) for 2001 through 2009. In order to account for possible spatial interactions between regions, we use a spatial weighting matrix of inverse distances. We estimate static and dynamic models by the maximum likelihood estimation approach, developed by Anselin (Spatial econometrics: Methods and models, Berlin: Springer, 1988) specifically for spatial models and elaborated by Lee and Yu (Journal of Econometrics, 154, 165-185, 2010a; Regional Science and Urban Economics, 40, 255-271, 2010b). We reveal that the unemployment in western regions is more of disequilibrium nature, while the unemployment in eastern regions is more of equilibrium nature. Using System GMM approach, we estimate the extended specification of the dynamic model and find that the unemployment in eastern regions affects both the unemployment in western and eastern regions of Germany, whereas the unemployment in western regions has an impact only on other western regions.