Convergence Tools and Mixture Analysis

In this paper we employ traditional beta, sigma, gamma-convergence tools to assess growth behaviour of groups of countries in connection with mixture densities. In the presence of mixtures, the standard methods to assess convergence are likely to be misleading. With respect to some recent papers dealing with issues of clustering through mixture densities of GDP, we assess what happens when the densities are multivariate, so that they contains traditional growth determinants. Our results show that, taken as global measures, convergence is an uncommon feature, whereas focusing on the individual components of the mixture we find pockets of convergence. Additionally, we focus attention on the classic determinants of growth and how these variables behave over time and within groups. Here the findings suggest that the behavior of countries within components is driven by similarities in the behavior of some of the growth determinants and that the well known result of a changing shape of income distribution from 1960 to 2000 is strongly related to changes in relationships among these variables began during 1980s.