A spatial approach to EU regional economic convergence : a comparison between parametric and non-parametric analysis.

The economic convergence among European regions over the period 1980-2006 is analysed in the first place by using a conditional β-convergence model and a distance-based weight matrix and secondarily by a spatially-conditioned stochastic kernel approach. A Spatial Autoregressive model which identifies two spatial regimes and spatial dependence finds that the convergence process is affected by polarization into two clusters defined both on a geographical and economic criterion, which converge at different rates towards different steady states. A similar result is then reached through a non-parametric analysis of the income distribution dynamics. These results confirm the hypothesis that a methodology which uses spatial econometric techniques is needed. They also suggest some implications for EU Regional Policy that should be taken into account.