COMPSTAT 2022: Start Registration
View Submission - COMPSTAT2022
Title: Clustering Italian regions on the basis of bivariate income and consumption distributions Authors:  Francesca Condino - University of Calabria (Italy) [presenting]
Antonio Irpino - University of Campania Luigi Vanvitelli (Italy)
Rosanna Verde - University of Campania Luigi Vanvitelli (Italy)
Abstract: In an economic framework, modeling income and consumption characteristics simultaneously can be of considerable relevance. Moreover, it could be of interest to identify homogeneous regions in a country in terms of economic behaviour. With this aim, we propose to jointly model income and consumption data through the copula approach and use the obtained bivariate density functions as descriptors of regions for clustering analysis purposes. In particular, considering data from the Survey on Households Income and Wealth (SHIW) by the Bank of Italy, the bivariate distributions at the regional level are obtained. The Jensen-Shannon divergence can be usefully employed to measure the discrepancies across density functions, as it allows us to take into account marginal and copula effects. The Italian regions are then partitioned in clusters by using a dynamic clustering algorithm, a non-hierarchical iterative algorithm, based on the optimization of an adequacy criterion that measures the fit between clusters and their prototypes. It can be shown that the divergence of all considered objects can be decomposed into two quantities, one relating to the heterogeneity present in the clusters and the other reflecting the discrepancy across clusters, according to Huygens' theorem.