Population heterogeneity and growth

I investigate the robustness of the link between growth and heterogeneity in a population along ethnic, linguistic, religious, and socio-economic class lines using cross-country data covering the period 1960–92 and 72 countries. In addition to 21 distinct explanatory variables extensively used in the empirical growth literature, I consider several fragmentation and polarization indices capturing the heterogeneity in a population to deal with measurement uncertainty, and utilize Bayesian Model Averaging techniques to address uncertainty issues regarding model specification. My results indicate that while substantial data evidence favors the inclusion of population heterogeneity as a theory when proxied by fragmentation indices, it does not support the inclusion of this theory when proxied by polarization indices. Furthermore, the religious fragmentation index has the strongest evidence in favor of its inclusion as opposed to ethnic and linguistic fragmentation indices and income inequality.