| The empirical growth literature has traditionally used cross-sectional designs that are vulnerable to endogeneity biases. While finding valid natural experiments for growth outcomes is a challenge, they have considerable potential for helping to understand income differences between nations. This dissertation studies economic growth in oil-rich countries using modern empirical techniques that emphasize quasi-experimental designs. The first chapter examines the effect of a major initial resource discovery in countries that were previously not resource rich. I find positive effects on GDP and education, reductions in infant mortality, and decreases in democracy levels. The second chapter evaluates growth in the manufacturing sector in oil-rich countries during the 1970s oil price boom and subsequent bust. I find a positive relationship between oil prices and manufacturing output and investment. I also find a negative relationship between oil prices and non-oil natural resources and agricultural exports, suggesting that higher oil prices encouraged industrialization. The third chapter studies a recently developed method of comparative case studies called synthetic controls, which uses a data-driven algorithm to construct a counterfactual that is a weighted combination of control units. Using Monte Carlo simulations, we identify sources of bias in two simple empirical settings, and in one case propose a correction procedure that yields unbiased estimates. |
