Is results-based aid more effective than conventional aid? Evidence from the health sector in El Salvador

Results-based aid (RBA) models link funds to outcomes, rather than paying for inputs. Despite their theoretical appeal and recent adoption by donors and multilateral development banks, there is limited empirical evidence supporting this form of aid for national governments. We estimate the effects of a RBA model using a natural experiment in El Salvador, where the same community health intervention was implemented in 98 municipalities using one of three financing models. The Salud Mesoamerica Initiative funded fourteen municipalities with a RBA model that partially conditions funds on the attainment of externally measured maternal and child health targets. Fifty-four municipalities funded inputs using conventional aid and thirty had national funds. Using a difference-in-difference approach and national health systems data we find that preventive health services increased by 19.8% in conventional aid municipalities and by 42% in RBA municipalities compared to national funds, suggesting that the results-based conditionality roughly doubled aid effectiveness. Effects are driven by increases in maternal and child preventive services incentivized by the RBA model. Rather than diverting resources from other populations, we find that the expansion of health services under RBA also benefited men and the elderly, not explicitly incentivized by the results model. While data on final health outcomes are not available, our results on proxy measures point to potential improvements in population health. The effects appear to have been driven by a more rapid expansion of health infrastructure and qualified personnel by motivated national authorities.