A major challenge faced by governments in low-income countries is to permanently alleviate households from poverty. Recent evidence suggests that the “Targeting the Ultrapoor” (TUP) program pioneered by BRAC in Bangladesh has proven effective and portable across settings (Banerjee et al., 2015; Bandiera et al., 2017). As a result, there has been growing interest in the program and up to 30 governments had been piloting variants of TUP by the end of 2015 (CGAP, 2015).
The TUP program consists of a large asset transfer (typically livestock) combined with asset-specific training; this brings up the question whether beneficiaries could do better with an equivalent valued cash transfer. Neoclassical theory suggests positive results as cash allows them to choose the highest return activities. However, it might lead to negative results if markets are imperfect, or cash is subject to informal claims, or irrelevant behavioural biases that for in-kind asset transfers. Any of these results can inform the design of social protection programs in low-income settings.
This project uses a large-scale randomised control trial to compare the classic TUP design to a modified design where beneficiaries have the choice of cash instead of an asset-skills bundle. The study is set within four of the poorest districts in Southern Punjab (where 10 million of Pakistan’s 40 million poorest households reside).