Club convergence in European regions

This study investigates the 'club convergence' hypothesis applying the stochastic notion of convergence to groups of European regions. In order to avoid the group selection bias problem, the innovative regression tree technique was applied to select endogenously the most important variables in achieving the best identification of groups on the base of per capita income and productive specialization. Tests on stochastic convergence in each group evidences a strong convergence among the wealthiest regions of the European Union and a trend of weak convergence among the remaining groups, confirming Baumol's hypothesis of convergence.