Urban Environments and Neighborhood Change: Exploring Urban Sorting Beyond the Featureless Plain

This paper introduces environmental features explicitly into the analysis of urban residential sorting where geographic barriers can mitigate neighbor externalities. Borders between groups in equilibrium will be more stable when supported by barriers. The hypothesis that racial disparity between neighboring tracts is greater when a barrier separates them is tested for Atlanta in 1990 and 2000 and compared to previous results for Chicago. The econometric estimation accounts for spatial dependence in the data. Significant barrier effects are found for certain types of geographical features (e.g., railroads, landmarks). The effect on local racial dissimilarity of the major extension of the mass transit rail lines in Atlanta in the mid-1990s is also estimated. There is little evidence suggesting that the new MARTA construction significantly affected racial dissimilarity in the areas it bissected. Limitations in the analysis and implications for policy and future are also discussed