Conditional Associations of Multidimensional Poverty Indicators in Argentina: A Graphical Representation

GARCIA ARANCIBIA RODRIGO; GIRELA IGNACIO

Buenos Aires

Congreso; LVI Reunión Anual de la AAEP; 2021

Asociación Argentina de Economía Política

Considering a multidimensional poverty index (MPI) for Argentina, in this paper we explore the conditionaldependency relationships between indicators of dimensions commonly included in a MPI, by means of graphicalmodels. Specifically, we propose an Ising graphical model with covariates to represent the MPI as an undirectedgraph. Using the Encuesta Permanente de Hogares (EPH) from Argentina for the last quarter of 2020, we definethe poverty dimensions and their indicators. First, we estimate a graph for non-monetary indicators given theusual monetary poverty index. We find that the conditional dependencies between the non-monetary indicatorsdepend strongly on monetary poverty, and even more so when we consider extremely poor households in termsof income. Additionally, from the joint graphical analysis with monetary and non-monetary indicators, we findthat monetary poverty is more associated with indicators of employment, social protection and education. Onthe other hand, a greater conditional association between the indicators of housing quality and those of thedimension of access to basic services (sanitation, water and cooking fuel) is revealed