MULTIDIMENSIONAL CHILD POVERTY IN CHILE: Dimensions That Matter for Child Development

The negative consequences of child poverty on well-being in the short and long terms have been well documented, as well as the social costs that it implies. There is a broad consensus both on its multidimensional characteristic and the importance of eradicating it. However, there is less consensus regarding which dimensions of child poverty should be considered to conceptualize and measure child poverty and what the best social policies for its eradication are. This research aims to identify the child poverty indicators that are the best predictors of children’s development, to contribute to the debate regarding the conceptualization and measurement of child poverty. We use data from the Longitudinal Survey of Early Childhood in the middle-high-income country of Chile, where income and multidimensional poverty has reduced over time but is, however, more prevalent for children. Using machine learning techniques, we identify, from among the indicators used to measure multidimensional poverty in Chile and also considering other indicators relevant to measure child well-being, the best predictors of children’s development (measured as language development). We found that the best predictors of children’s development are parental education followed by access to a computer. On a second level, access to the internet and easy access to water predicted higher language development while overcrowding and harsh parenting were associated with lower levels of language development. Going beyond the indicators used by the official MPI, these results show new material and immaterial indicators as relevant for children. These indicators should not only be considered at the moment in the design of a multidimensional poverty measurement method with a focus on children, but they should also be considered in the current social protection system for children and adolescents in order to prevent the negative consequences of poverty in the mid and long terms.