Bayesian Analysis of the Convergence Hypothesis in Economic Drowth: A Markov Approach

In this paper, we apply a Markov mixture model to capturing complex dynamics in the cross section distribution of per capita income across countries and examining the convergence hypothesis in economic growth. A Markov mixture model is estimated by a Bayesian procedure via the Gibbs sampler. With the Markov mixture model, 120 countries are clustered into several income classes, and effects of this clustering on tests of the convergence hypotheses are examined. The Markov mixture model also gives us the probability of transition from one income class to the other as well as the probability for each country to belong to an income class at each time period. Our study supports convergence across countries within each income class, but reject it if all the 120 countries are treated as one class.