This thesis applies so-called Bayesian model averaging (BMA) to three different economic questions substantially exposed to model uncertainty. Chapter 2 addresses a major issue of modern development economics: the analysis of the determinants of pro-poor growth (PPG), which seeks to combine high growth rates with poverty reduction. Vietnam is an interesting example for such an analysis because this country is a showcase for effective policies of PPG. However, it is not clear which factors have contributed to which extent to Vietnam’s PPG. Chapter 3 analyzes whether foreign direct investment (FDI) has been beneficial for productivity growth in Latin America (LA). FDI has surged in LA since the mid 1990s. BMA allows accounting for the major shifts in the regional composition of these inflows, for the varying types of and motives for FDI, and for differing local conditions within LA. Finally, Chapter 4 goes a step further and investigates not only the robustness of different types of growth determinants. It also introduces Weighted-Average Least Squares (WALS) as a new model averaging technique, which is theoretically and practically superior to BMA. By comparing the estimation results of WALS and BMA and conducting various robustness checks, we analyze the importance of various 'fundamental' growth determinants such as geography, institutions, fractionalization and culture or religion.
|Qualification||Doctor of Philosophy|
|Award date||31 Oct 2008|
|Place of Publication||Tilburg|
|Publication status||Published - 2008|