How accurate is food-for-work self-targeting in the presence of imperfect factor markets? Evidence from Ethiopia

Effective targeting of transfers is a key issue in public policy to combat poverty. Much faith is presently placed in self-targeting mechanisms such as public employment schemes supported by food-for-work transfers. Where targeting errors have been observed, these are usually attributed to mismanagement of key operational details, such as the project's wage rate. Using a unique data set from rural Ethiopia, we demonstrate that targeting errors may also have structural causes in some low-income countries. We hypothesise that imperfect factor markets generate a predictable dispersion across households in reservation wage rates that breaks down the unconditionally positive relation between income and shadow wages on which the theory of self-targeting public employment programmes rests. Our results confirm that the inaccuracy of FFW targeting stems from the fact that, in rural Ethiopia, higher income households are endowed with more labour per unit of land or animal. Due to poor factor markets in land and livestock these labour-abundant households have lower marginal labour productivity on farm, thereby depressing the reservation wage rates they find acceptable for FFW participation.