Household Characteristics and the Distribution of Income In Italy: An Application of Social Distance Measures

The Lorenz criterion of preferable distributions fails to distinguish adequately between convergence to the global mean and clustering around local means. This concern has motivated independent work by Wolfson, and Esteban and Ray on the notion of polarization. In this paper I build on this recent work by providing a new method that characterizes changes in the entire distribution, rather than focussing only on dispersion. In particular, the approach proposed offers a new decomposition method of within‐ and between‐group components that differs from the classical method of additively decomposable inequality indices. The new method can monitor which factors modified the entire distribution, where precisely on the distribution these factors had an effect, and what determined the variation in the level of social distance between groups or geographic areas. Summary statistics of the observed movements and of distance between and divergence among the estimated and the counterfactual distributions are provided as well as a new index of social distance. The new method is then applied to Italian data on income distribution between 1987 and 1995.