The resource curse revisited: A Bayesian model averaging approach

The evidence for the effects of oil rents on growth is mixed, a result which can be explained with model uncertainty. We address the issue using Bayesian Model Averaging techniques and an updated cross-country data set for long-term growth in the period 1970-2014, including 91 countries and 54 potential growth determinants. We do not find empirical evidence for the existence of a "natural resource curse" in our sample. On the contrary, our results suggest a robust positive effect of oil rents on long-term economic growth. We then introduce interaction terms of oil rents with potential conditions under which oil dependency can lead to sub-standard growth. The results indicate that the positive effect of oil rents may be conditional on the quality of institutions. We test the robustness of our results using a panel data set and find neither a curse nor a positive effect of oil rents on short- to medium-run growth. (C) 2018 Elsevier B.V. All rights reserved.