Gini Indices by Quantile Range

• Activity Number: 181
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Chaitra Nagaraja*+
• Companies: Fordham University
• Keywords: Gini index

Abstract

The current global economic turmoil has brought forth many philosophical debates on what constitutes a fair society. In these discussions, the concept of income inequality has surfaced as a key issue. To measure relative inequality, the Gini index is often used for studying income distributions. Of particular interest is examining Gini indices for subpopulations defined by quantile partitions (e.g. lower, middle, upper class). For example, the Occupy Wall Street protestors in 2012 made popular the idea of a 99% against the 1% based on income disparities. However, dividing the data into subgroups based on income ranking creates dependent data within those partitions even if the data were originally collected independently. As a result, we cannot apply the fact that the Gini index, computed for independent and identically distributed data, belongs to the class of U-statistics. Consequently, we cannot use the asymptotic properties and confidence interval construction methods from the existing theory for Gini indices. This talk is focused on computed the individual Gini coefficient estimates and respective standard errors using resampling methods.