Abstract:The paper analyzes the contribution of human capital and of the labour augmenting component to the observed levels of per capita GDP using a random coefficient finite mixture model. Our approach deals parsimoniously with three important problems of the empirical literature (parameter heterogeneity, omitted variable bias and departures from the normality assumption) and provides additional insights for the interpretation of the determinants of economic development. More specifically, we identify five clusters of countries in which the heterogeneous impact of human capital and of the labour augmenting component on per capita GDP depends on differences in latent variables (i.e. cultural and institutional factors, quality of the educational system). Our results seem to find support for theoretical hypotheses arguing that these latent variables are crucial to address talents to economically productive activities and to increase returns to schooling.