Income inequality decomposition using a finite mixture of log-normal distributions: A Bayesian approach - Archive ouverte HAL Access content directly
Journal Articles Computational Statistics and Data Analysis Year : 2016

Income inequality decomposition using a finite mixture of log-normal distributions: A Bayesian approach

Abstract

The log-normal distribution is convenient for modelling the income distribution, and it offers an analytical expression for most inequality indices that depends only on the shape parameter of the associated Lorenz curve. A decomposable inequality index can be implemented in the framework of a finite mixture of log-normal distributions so that overall inequality can be decomposed into within-subgroup and between-subgroup components. Using a Bayesian approach and a Gibbs sampler, a Rao-Blackwellization can improve inference results on decomposable income inequality indices. The very nature of the economic question can provide prior information so as to distinguish between the income groups and construct an asymmetric prior density which can reduce label switching. Data from the \UK\ Family Expenditure Survey (FES) (1979 to 1996) are used in an extended empirical application.
Not file

Dates and versions

hal-01440303 , version 1 (19-01-2017)

Identifiers

Cite

Michel Lubrano, Abdoul Aziz Junior Ndoye. Income inequality decomposition using a finite mixture of log-normal distributions: A Bayesian approach. Computational Statistics and Data Analysis, 2016, 100, pp.830--846. ⟨10.1016/j.csda.2014.10.009⟩. ⟨hal-01440303⟩
66 View
0 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More