Grants vs. credits for improving the livelihoods of ultra-poor: Evidence from Ethiopia

Reaching the ultra-poor and enhancing graduation have long been challenges in many social protection programs. This paper compares the behaviors and performances of most vulnerable, ultra-poor and risk-averse households, who have been granted cash transfers for livelihoods investment, with the behaviors and performance of credit recipients, who are relatively better-off and willing to take credit risks. Using data from the Ethiopian pilot program, we tested if freely provided cash is used less efficiently as the sunk cost hypothesis portrays. Our data revealed that credit recipients indeed perform better than grant recipients. However, when we control wealth and other household characteristics, grant-based investments perform better than credit-based livelihood investments. Grants were allocated more likely to the planned investment than credits and the performances of the former is higher than the latter, both with and without controlling the intensive knowledge supports provided to grant recipients. The result is consistent and robust across different estimation approaches. This implies that the sunk cost hypothesis is not an important disincentive in livelihood grants. We concluded that livelihood grant (asset transfer) not only helps to reach the excluded ultra-poor but also to improve the effectiveness and productivity of rural livelihood investments. We further explained the possible reason for the superiority of grant over loan and its implication on graduation and program costs.