Targeting the “poorest of the poor”: A multi-scale approach

Abstract

Understanding that poverty presents multidimensional and multi-scale deployments, a methodology is proposed that allows to identify those poorest families among the poor in the State of São Paulo for cash transfer purposes. We propose a method to combine the vulnerability of the family with the other scales, resulting in a more refined classification of the situation of poverty, which is more difficult to be biased by the selfreported register. We added a georeferenced context-based component to differentiate among families with same vulnerability but living in different levels of intraurban and municipal conditions. At the municipal scale, we developed two sub-indexes through Principal Component Analysis (PCA): a) the Municipal Context Sub-Index, and b) the Municipal CadUnico Sub-Index. These sub-indexes were then integrated. On the scale of the Census enumeration area, we considered aspects that can aggravate the poverty condition, such as physical environmental risks and socioeconomic deprivation. On the family-scale, we consider the score defined by the São Paulo Multidimensional Poverty Index (MPI). The final multi-scale classification of families results from the sum of the scores obtained at the family level, the Census enumeration area and the municipality. The indexes were validated internally and externally. We simulated a scenario of hypothetical cash transfer programs using the current tool (MPI-S) and the proposed methodology to benefit up to 10,000 and 100,000 families in the State of São Paulo. We implemented the solution in Power BI tool to allow automatic updates of the indexes to be made in the developed “dashboard”. Results allow us to evaluate the impact of both approaches. The current approach, based only in the multidimensional poverty of the family is more distributed, with a wider geographic distribution, sometimes benefiting only one family in some municipalities or a great number of families in one municipality. The proposed approach tends to concentrate the assistance with greater percentages of families in the CadUnico by municipality. Besides aiding on cash transfer decision, the results obtained on the context and the intensity of municipal poverty can be used for programs aimed at leveraging economic development in the different administrative regions of the State of São Paulo.