Summary While the stylised fact that, around the world, areas of rapid growth are geographically distinct from those with a poor growth performance, is easily accepted, little effort has been made in the empirical literature to incorporate location into the analysis of regional convergence in terms of economic growth. This paper provides a quantitative evaluation of the role of location in regional convergence, using the argument that growth in each region depends not only on its own characteristics but also on those of the regions that form the neighbourhood to which it belongs. Applying spatial econometrics it is found that, during the 1990s, opposing forces were exerting their influence across the EU: while regions did converge at a rate close to the 2 percent mark, neighbourhoods of regions diverged at an almost equal rate, leaving a net effect of convergence considerably smaller than the one previously reported in the literature.

This article offers a new approach in assessing the presence of regional convergence in income per capita and applies it to data for sixty-five regions of the European Union over the decade leading up to the entry into force of the Maastricht Treaty. Within the framework of distributional dynamics analysis, the author proposes the use of quantitative techniques traditionally employed to inform investment decision making under uncertainty. After investigating the intradistributional dynamics with Markov chains, the author tests for regional convergence using secondorder stochastic dominance. For the sample and time horizon considered, the author finds evidence of regional convergence that is neither fast nor continuous. In other words, the regions in the sample display high persistence in belonging to a certain income group, while subperiods of convergence and divergence in income per capita are discernible. The approach proposed is grounded in economic theory, offers information about changes across the entire distribution, and can be adapted to incorporate location into the analysis.

No abstract

No abstract