Experimental Design
A Randomized Controlled Trial will be conducted among 250 highly vulnerable refugee households in Kyangwali refugee settlement, Uganda. These 250 households were referred to the partner NGO as being particularly vulnerable households, who the NGO has enrolled on a first-come-first-serve basis to their 6-month unconditional cash transfer program. These 250 households will be randomized into 2 groups: ‘Cash’, and ‘Graduation’.
Due to NGO's “first-come-first-serve” enrollment of households into the cash transfer program, there exist many more households that are equally vulnerable as those eligible for the cash transfer, that were not selected for the cash transfer. A subsample of 125 of those households will be chosen to form the ‘No Cash’ comparison group. These are similarly vulnerable households that do not receive a cash transfer. Note, I do not randomize between who receives a cash transfer, and who does not. The NGO determines which households receive cash or not. However, the “first-come-first-serve” enrollment is quasi-random, and hence the ‘No Cash’ comparison group is the best comparison group that can be achieved in a fair and ethical manner. To compare the ‘No Cash’ group to the other two groups, a Regression Discontinuity Design will be deployed based on the household’s Vulnerability Score.
Households from the ‘No Cash’, ‘Cash’, and ‘Graduation’ groups will undergo seven rounds of surveys: 1 baseline, 3 phone surveys after 6, 12, and 18 months, an endline survey after 2 years, and two post-program phone surveys after 2.5 and 3 years (compared to the start date of the program).