Convergencia regional en Europa

This paper generalises the classical test methodology for cross-section ß-convergence analysis to allow the convergence process to vary between different clubs of regions and to account for possible interregional interactions and co-dependence in growth in Europe over the time period 1995-2000. We find clear evidence for club convergence in Europe for the time period of observation. The sample of regional economies belonging to club A converges in an unconditional sense at a speed of 1.5 percent per year and those belonging to club B (regional economies in CEE and Southern Europe) at a speed of 2.4 percent. It is important to emphasise that a speed of 1.5 or 2.4 percent per year, even though in accordance with previous findings of convergence studies, is very small. The study also illustrates that the classical convergence test methodology that has been the work-horse of most previous convergence studies in mainstream economics is ill-designed to analyse regional convergence due to several reasons. In particular, ignoring the presence of spatial error autocorrelation in convergence analysis carried out with cross-sectional data can lead to wrong conclusions, for example, with respect to the assessment of convergence speed. Thus, testing for the presence of spatial autocorrelation (heterogeneity) by means of appropriate diagnostics and implementing alternative specifications of the convergence test equation when needed are crucial issues in income convergence analysis.