Research on the question of regional income convergence has gone through two phases over the past two decades. The first generation of regional convergence studies began to appear as growth theorists turned their attention away from international analyses of country growth patterns having discovered the region as a new unit of analysis (Barro and Sala-i-Martin 1991). This change in scale had a key advantage of increasing (in some cases substantially) the number of cross-sectional observations available for model estimation and hypothesis testing. While the scale of the analysis shifted, these first generation studies relied on the same underlying theoretical and empirical frameworks used in the international literature. In the second phase, the underlying geographical dimensions of the data in convergence studies began to attract attention (Rey and Montouri 1999). This was reflected in several developments. The first saw the increasing application of the methods of spatial econometrics and spatial data analysis to regional case studies. These applications have generated abundant evidence that the spatial effects of dependence and heterogeneity tend to be the rule rather than the exception in practice, and as such their consideration should form a crucial component of empirical analysis. Thus the second generation of regional convergence studies is those characterized by concerns with spatial effects.