Regional economic convergence in Greece : a stochastic dominance approach

Traditionally, the analysis of regional economic convergence is based on the notions of β-convergence and σ-convergence. However, both of these approaches have several drawbacks and limitations. To overcome these difficulties, we apply a more recent non-parametric approach in the convergence literature, which differs from the other non parametric kernel estimator approaches. This is the stochastic dominance approach, which is traditionally implemented in investment decision making. Stochastic dominance analysis has many advantages, relative to the traditional methods for convergence analysis, as it is based on minimal assumptions and it is not bound to a certain distribution. Additionally, it produces more evidence about the shape and the evolution of the regional per capita income distribution, rather than a simple measure of dispersion as the σ-convergence approach does. Furthermore, it supports statistical significance tests for the identified tendency towards convergence. Thus, stochastic dominance analysis offers the possibility to examine the whole distribution of per capita income and also to test for polarization. In this study, we investigate the existence and the nature of per capita income convergence between the Greek NUTS-III regions. Specifically, our data covers the 1995-2006 period and the 51 Greek perfectures. Moreover, convergence is studied using the traditional parametric approaches, and the results are compared to those of the stochastic dominance approach.