Abstract
Tied project aid is often said to suffer from allocative inefficiency, yet it seems to benefit from strong donor support. One reason may be that tied aid is perceived as less corrupt than untied program aid. We present a simple model where perceived aid diversion can induce tied transfers. In a dictator game, we then compare the willingness to make a monetary contribution to Zambia's national health budget (CBoR) with the willingness to make a tied transfer (mosquito nets) to a health-care clinic (KC) in Lusaka. Donors clearly prefer tied aid to untied program aid. First, the mean tied donation to KC (SEK 44) is highly positive. Second, it is significantly higher than the mean monetary transfer to CBoR (SEK 26). Third, the fraction of donors who give at least one net to KC (65%) is significantly higher than the share only giving money to the CBoR (16%). Exit questionnaires suggest that the reason is a fear of corruption and misallocation at the CBoR. Our experiment indicates that reducing developing country corruption could benefit aid recipient countries in two ways, increasing both allocative and productive aid efficiency.