Growth Empirics in Panel Data Under Model Uncertainty and Weak Exogeneity

This paper considers panel growth regressions in the presence of model uncertainty and reverse causality concerns. For this purpose, my econometric framework combines Bayesian model averaging with a suitable likelihood function for dynamic panel models with weakly exogenous regressors and fixed effects. An application of this econometric methodology to a panel of countries over the 1960–2000 period highlights the difficulties in identifying the sources of economic growth by means of cross-country regressions. In particular, none of the nine candidate regressors considered can be labeled as a robust determinant of economic growth. Moreover, the estimated rate of conditional convergence is indistinguishable from zero. Copyright © 2014 John Wiley & Sons, Ltd.