Limited Information Bayesian Model Averaging For Dynamic Panels With Short Time Periods PDF Download

Do growth spells in Africa end because of bad realizations of the same factors that influence growth spells in the rest of the world, or because of different factors altogether? To answer this question, we examine determinants of growth spells in Africa and the rest of the world using Bayesian Mode Averaging techniques for proportional hazards models. We define growth spells as periods of sustained growth episodes between growth accelerations and decelerations and then relate the probability that a growth spell ends to various determinants including exogenous shocks, physical and human capital, macroeconomic policy, and sociopolitical factors. Our analysis suggests that determinants of growth spells in Africa are different from those in the rest of the world. The majority of the identified robust determinants have a distinct impact in only one of the two samples: initial income, terms of trade, exchange rate undervaluation and inflation, influence spells only in the world sample, while openness and droughts seem to only affect Africa. In addition, a few common determinants - proxies for human and physical capital and changes in the world interest rate - have very different marginal effects in the two samples.