Decomposition of rank-dependent measures of inequality by subgroups

The purpose of additive subgroup decomposition is to study the relationship between overall inequality and inequality within and between population subgroups defined by variables like gender, age, education and region of residence. As opposed to the inequality measures that are additively decomposable, the so-called generalized entropy family of inequality measures, the Gini coefficient does not admit decomposition into within- and between-group components but does also require an interaction (overlapping) term. The purpose of this paper is to introduce an alternative decomposition method that can be considered to be a parallel to Lerman and Yitzhaki's (1985) elasticity approach for decomposing the Gini coefficient by income sources, which means that the elasticity of the Gini coefficient with respect to various income components is treated as the basic quantities of the decomposition method. Thus, rather than decomposing the Gini coefficient or any other inequality measure into a within-inequality term, a between-inequality term and eventually an interaction term, the basic quantities of the introduced method are the effects of marginal changes in variables that are used to specify the population subgroups