Assistance to poor households when income is not observed: targeted in-kind and in-cash transfers

Abstract

Policymakers frequently design self-targeting programs or target poor areas to assist poor families when income is not observed. Self-targeting schemes take advantage of differences in participation costs in assistance programs across households. Geographic targeting assumes that transfers are solely determined by the region of residence: to receive the benefits, households not initially present in the targeted areas must relocate away from their original place of residence and live with the poor, which entails a cost that can also be interpreted as a participation cost in the assistance program. The paper shows that a combination of in-kind and in-cash transfers tied to the consumption of a publicly provided private good targeted to the poor becomes very useful when non-poor households have different participation costs. By distinguishing users and non-users of public facilities additional in-cash transfers can be directed to the poor more effectively. The paper demonstrates that the publicly provided good ends up being undersupplied, and the distortion becomes less important as participation costs rise. More importantly, it shows that this kind of redistributive program dominates a pure in-cash scheme.