The Gini coefficient is a popular concentration measure often used in the analysis of economic inequality. Estimates of this index for small regions may be useful to properly represent inequalities within local communities. However, the small area estimation for the Gini coefficient has not been thoroughly investigated. A method based on area level models, thereby avoiding the assumption of the availability of Census data at the micro level, is proposed. A modified design based estimator for the coefficient with reduced small sample bias is suggested as input for the small area model, while a hierarchical Beta mixed regression model is introduced to combine survey data and auxiliary information. The methodology is illustrated by means of an example based on Italian data from the European Union Survey on Income and Living Conditions.
- Hierarchical Bayes, Beta regression, Income inequality, random effects, Markov Chain Monte Carlo, Variance components