Paper short abstract:

A new regression-based decomposition method is developed to estimate the contribution of regional factors to total income inequality. The method is applied to the Chinese urban data for the reference year of 2007 and compared to more traditional regression-based decomposition methods.

Paper long abstract:

A new regression-based decomposition method is developed to estimate the contribution of regional factors to total income inequality. Unlike the previous regression-based decomposition methods, it enables attributing total income inequality to five parts: pure regional effect, inter-regional differences in coefficients, inter- and intra-regional differences in characteristics, and an unexplained part. It follows that our approach has advantages over the existing regression-based decomposition methods in that the effects on total inequality of differential returns to income determinants across regions can be calculated, and the contributions to total inequality of the intra- and inter-regional distributions of income determinants can be distinguished.The method is applied to data for urban China in 2007 and our decomposition results lead in this application to clearer policy implications than the results based on the Shorrocks method.