Spatial Filtering and the Speed of Income Convergence


In this work I try to assess the robustness of the determinants of income growth differences at the subnational level in the period from 1995 to 2010 by using a modified version of an extensive new global dataset collected by Shleifer et. al. (2014). In order to do so I apply a method that combines spatial eigenvector-filtering and Bayesian Model Averaging which was put forward by Crespo-Cuaresma and Feldkircher (2013). Furthermore I am depicting subnational within country inequality patterns through exploratory data analysis and find an inverted-U-shaped relationship between spatial inequality and economic development as initially proposed by Kuznets (1955). My BMA statistics suggest that a substantial part of the speed of income convergence has to be ascribed to spatially correlated spillovers. A corollary of this econometric result is, that gains from spatial spillovers are likely to be higher when subnational entities are more homogenous and display a lower degree of inequality.

MSc Thesis, WU Vienna