Regional income convergence in the United States: 1970--200

This dissertation presents an empirical analysis of regional income convergence in the United States in the period 1970--2000. The objectives of the research are to analyze the trend of inter-state income convergence and measure the effects of growth factors in regional economic convergence. This research is based on the theories that are drawn from the studies of economic growth and convergence and of regional income inequality. Neighborhood Disparity Index (NDI), a spatial inequality measure, is believed to be used for the first time in measuring a convergence. Spatial statistical techniques are employed to analyze the role of space in regional income convergence. Based on the data adjusted with the state-level deflators, this research finds that there was an overall sigma convergence over 1970--2000 despite fluctuations from the late 1970s to the late 1980s. Absolute beta convergence also occurred in this period, but the convergence speed was less than 2 percent per year. There was significant movement of the states in cross-state income distribution. Regarding growth factors, it is found that the growth rate of state per capita personal income is negatively related to the initial income level, initial income inequality, and the extent of initial state fiscal control, and is positively related to the growth of private capital stock and the growth of the proportion of college graduates in the population. It is also found that spatial price differentials and spatial dependence impact the analysis of regional income convergence. Without using the state-level deflators, regional income inequality is underestimated or overestimated in the study period. The extent of under- and over-estimation varies with inequality measures. Three conventional inequality measures including standard deviation, coefficient of variation, and the Gini coefficient, are highly correlated with Moran's I statistics, a measure of spatial dependence, but NDI has much weaker correlation with spatial dependence, suggesting that NDI is less impacted by spatial dependence and thus is a better measure for regional income inequality than the conventional inequality measures.