This article investigates how appropriation cost and government expenditure would affect on income disparity from empirical aspects. Empirically, we use corruption perceptions indices and Gini coefficients of 129 countries or regions to estimate appropriation cost and income disparity respectively, and design regression analysis by different groups, which is determined by different levels of corruption perception indices. Regression results reveal that for countries with high appropriation cost, government expenditure has a significant negative coefficient on income disparity, whereas the coefficient becomes positive in countries with low appropriation cost. Moreover, this article provides with evidence that there is a U-shape relationship between income disparity and per capita income in the high appropriation cost sample, as well as the whole sample and remaining subsamples support the Kuznets Hypothesis.