How Accurate is Food-for-Work Self-Targeting in the Presence of Imperfect Factor Markets? Evidence from Ethiopia

Abstract

Both the research and policymaking communities have shown considerable interest in identifying mechanisms that might improve the targeting of transfers to needy beneficiaries. Self-targeting mechanisms such as food for work have attracted particular attention because of their purported ability to induce self-selection out of the pool by the relatively well-off, thereby enabling concentration of transfer resources on the neediest subpopulations with minimum expenditure on administration. This paper calls attention to an important oversight in the existing literature on self-targeting: its dependence on households facing a parametric wage. Since imperfect factor markets plague many of the settings, such as rural Ethiopia, in which FFW is widely used, this oversight matters. Using a unique data set in which rural Ethiopian households declared reservation wage rates for FFW participation, we find significant errors of both exclusion and inclusion in any public works employment scheme. These errors appear perfectly consistent with a simple structural model that allows for household-specific valuation of labor as a function of relative factor endowments and human capital characteristics, with an added premium for public project participation that depends on one's history of past participation.