The authors apply the method to the differences between the Brazilian income distribution and those of Mexico and the United States, and find that most of Brazil's excess income inequality is due to underlying inequalities in the distribution of two key endowments: access to education and to sources of nonlabor income, mainly pensions.