Distribution dynamics of regional GDP per employee in unified Germany

We investigate to what extent convergence in production levels per worker has been achieved in Germany since unification. To this end, we model the distribution of GDP per employee across German districts using two-component normal mixtures. While in the first year after unification, the two component distributions were clearly separated and bimodal, corresponding to the East and West German districts, respectively, in the following years they started to merge showing only one mode. Still, using the recently developed EM-Test for homogeneity in normal mixtures, the hypothesis of just a single normal component for the whole distribution is clearly rejected for all years. A Posterior analysis shows that about half of the East German districts were assigned to the richer component in 2006, thus catching up to levels of the West. The growth rate of a mover district is about one percentage point higher than the growth rate of a non-mover district which had the same initial level of GDP per employee. -- Regional convergence ; distribution dynamics ; mixture models ; Germany