Targeting Transfers through Restrictions on Recipients

Abstract

Our objective is to design transfer programs that maximize some social welfare function. Following the optimal income tax formulation, we assume that a random process endows individuals with characteristics that influence welfare, such as ability to earn income, medical condition, or unmonitorable income. The goal is to maximize the utility of a randomly chosen individual. (Other individualistic social welfare functions would lead to equivalent results.) If the random characteristics were observable, a first best outcome could be achieved by making them the basis for lump sum transfers. In practice, however, we can observe only indicators of these characteristics, such as earnings, medical expenditures, or consumption patterns. The challenge for policy is to design an efficient second best transfer program based on indicators. In this paper we argue that 1) An optimal transfer program in general must sacrifice productive efficiency to target efficiency. This is done by imposing restrictions on the choices made by intended beneficiaries. 2) A program that incorporates restrictions-such as means-tested in-kind transfers, commodity-specific taxes and subsidies, and even ordeals (i.e., the imposition of deadweight costs to qualify for a transfer)-will perform better than programs that rely solely on income taxes and cash transfers.