Static and Dynamic Decompositions of Income Inequality in Brazil between 1980 and 2000

Jeronimo O. Muniz, University of Wisconsin at Madison

This article uses the methodology suggested by Cowell and Jenkins (1995) and Mookerjee and Shorrocks (1982) to decompose two generalized inequality measures: the Theil index, and the mean log deviation of income. The decomposition separates the total inequality observed in Brazil in 1980, 1990 and 2000 into three income classes: poor, average and rich. This decomposition exercise shows how population and income shares of each subgroup have changed over time, calculates individual inequality measures for each subpopulation and indicates which share of total inequality is due to inequality within and between these three groups. The dynamic decomposition shows what share of variation in inequality is attributable to the allocation of people in different population partitions, what share is due to variations in the relative income between partitions, and what share is due to changes within partitions.