The donor problem: An experimental analysis of beneficiary empowerment

Donors often rely on local intermediaries to deliver benefits to target beneficiaries. Each selected recipient observes if the intermediary under-delivers to them, so they serve as natural monitors. These recipients may, however, withhold complaints to 'thank' the intermediary for selecting them. Furthermore, the intermediary may distort selection (for example, by picking richer recipients who feel less entitled) to reduce complaints. We design an experimental game representing the donor's problem. We compare two institutions. In one treatment the intermediary selects recipients. In the other selection is random - as by an uninformed donor. In our data random selection dominates delegation of the selection task to the intermediary. Selection distortions are similar but intermediaries divert more when they have selection power (correctly anticipating that gratitude for selection will reduce complaints). Our results identify a problem in combining selection and delivery tasks. The insights are also applicable to social funds, decentralisation and participatory projects.