Composite indicators of poverty in the small area case

Nowadays there is a widespread agreement that poverty is a multifaceted phenomenon, encompassing deprivations along multiple dimensions, so that income is just one of the welfare indicators among several others that contribute to measure poverty. In this framework, our methodological choice for the measurement of multidimensional poverty responds to the notion that setting thresholds to separate the poor from the not poor is an inherently arbitrary one. The process of obtaining composite and multidimensional indicators is a very data hungry process. Moreover, current surveys on households are often unplanned to produce significant estimates at domains (target subpopulations), which do not coincide with those of interest. For all these reasons the focus of the paper is not only on the definition of indicators but also on their estimation process, with particular attention to the statistical quality of the current estimation of the indicators (quality dimension of the indicators and obviously their accuracy), and on small area estimation techniques. The proposed methods belongs to the family of M-quantile SAE models and are applied to composite indicators such as those derived under the Sen approach.