Abstract

Multi-faceted big push graduation programs have been shown to help households move out of poverty. Governments have increasingly adopted these programs as part of their social protection schemes, often in simplified forms such as livelihood programs that provide self-employment support. However, it remains an open question how to optimize different program components. Using a large-scale randomized controlled trial embedded in Tanzania’s nationwide social protection program that covers more than 1.3 million households, this paper evaluates the impacts of layering public works and a livelihood program that provides business training and grants, on top of regular cash transfers. One year after implementation, the livelihood program increased household income by up to 37 percent relative to cash transfers alone, while public works generated no additional income gain. We find no evidence of complementarities between public works and the livelihood program. We also do not find differential spillover effects on non-beneficiaries across program components. Cost benefit analysis suggests that each dollar invested in layering the livelihood program on top of cash yields three dollars in return.