Does the location and the institutional background matter in convergence modelling of the EU regions?

The paper focuses on the testing of regional income convergence of NUTS2 EU regions during the 2000–2011 period with the emphasis on conditional income convergence using both non-spatial and spatial approaches. In order to examine the fact that the per capita GDP growth rate in one region may be associated with the growth rate in neighbouring regions, our empirical part begins with the Exploratory Spatial Data Analysis (ESDA). This preliminary analysis based on the ESDA tools has suggested the appropriateness of the spatial aspect in income regional convergence modelling. Following these results, we proceed with the estimation of the beta-convergence models for purpose of verifying various absolute and conditional convergence hypotheses. The estimation of the income convergence models confirmed that the spatial dependence among regions does matter. We have also employed the dummy variable corresponding to the former political orientation of the regions in order to test the conditional convergence hypothesis assuming that it controls for different country specific factors. The incorporation of the dummy variable enabled us to assess the speed of convergence and the half-life time separately for two groups of regions under the consideration (post-communist or not). Almost all hypotheses of the analysed convergence models were confirmed. The inclusion of spatial effects to the models always led to worsening of the convergence characteristics. Overall, the results of our analysis imply that convergence process is not determined only by a region’s initial income and other specific factors but also essentially by its neighbourhood region’s growth performance.