In this study, we evaluate how well various systems for identifying and targeting assistance to the poorest of the poor actually identify the poorest. Firstly, we consider the methods used to identify households eligible for participation in assistance programs administered by the Indian government. Secondly, we evaluate Participatory Rural Appraisals (PRAs) as a mechanism to identify exceptionally poor households. Finally, we investigate whether additional verification of information gathered in PRAs improves targeting. For each method of targeting, we examine whether the households identified by that process are more disadvantaged according to several measures of economic well-being than households which were not identified. We conclude that PRAs and PRAs coupled with additional verification successfully identify a population which is measurably poorer in various respects, especially those which are more readily observed. The standard government procedures, however, do not appear to target the very poorest for assistance. Based on this sample, households targeted for government assistance are observationally equivalent to those that are not.
July 18, 2009