Bayesian Model Averaging
Home | Papers | Software | Courses

iii. Robust Growth Determinants. Gernot Doppelhofer and Melvyn Weeks (2008)

This paper investigates the robustness of determinants of economic growth in the presence of model uncertainty. Several studies in the empirical growth literature, characterized by a large number of models and relatively few observations, have used Bayesian Model Averaging (BMA) techniques to address model uncertainty. Benchmark BMA uses linear regression models with independent normal sampling and homoscedastic errors. In contrast to these restrictive assumptions, this paper allows a priori for the presence of heteroscedastic errors due to parameter heterogeneity and outliers. The paper finds that inference and policy analysis about growth determinants are significantly affected by allowing deviations from the normal benchmark BMA. We identify outlying observations -- most notably Botswana -- in explaining economic growth in a cross-section of countries.

Paper links:
Cambridge Working Papers: