Evidence and Drivers of Income Polarization in Italy: A Relative Distribution Analysis Using Recentered Influence Function Regressions

Abstract

The contribution of this paper is twofold. First, it analyzes Italian personal income distribution changes between 1991 and 2020 using the “relative distribution” method. This nonparametric tool uses all distribution information, unlike summary statistics. The analysis documents how distributional changes over time hollowed out the middle of the income distribution and increased household concentration around the highest and lowest deciles, causing polarization to rise. Second, and most importantly, the paper develops a novel approach to quantify the drivers of distributional changes by directly relating the impact of changes in the expected values of the covariates on the relative polarization indices, which code the direction and magnitude of distributional changes. Results show that the household head’s education and employment status increase the income distribution’s polarization over time. When the household’s head is female, foreign-born, or from the Centre/South/Islands, a counter-polarization effect occurs.