Child vs. Household MPIs in Colombia: Do they Identify the Same Children as Multidimensionally Poor?

Despite increased recognition of the importance of measuring child poverty from a multidimensional perspective, most countries with National Multidimensional Poverty Indices (MPIs) have used the household as the unit of identification and used the results of national MPIs to define the levels of child poverty in a country. This assumes that resources are shared equally among all household members, and ignores possible intra-household inequalities. Given the lack of knowledge about whether Household MPIs properly identify children who are multidimensionally poor, this article aims to compare the results of a child-specific MPI exercise and a household-specific MPI exercise and identify the dimensions and individual characteristics that explain this gap. To fulfil this objective, we computed a Child MPI for Colombia and compared the results with the Colombian National MPI. In addition, we estimated probit and biprobit models to identify the determinants of being a multidimensionally poor child under both measures. The results of the analysis reveal three main findings: (1) there is a mismatch between the two measures; (2) the deprivation profiles of multidimensionally poor children are different depending on which MPI is used to classify them as poor; and (3) children who are multidimensionally poor according to a Child MPI have different individual and family characteristics compared with children who are classified as poor according to a Household MPI. These three main findings reveal that it is necessary to analyse child poverty using an MPI that captures individual deprivations.