Work In Progress
Unconditional Cash Transfers: \A Bayesian Meta-Analysis of 50 Randomized Evaluations in 26 Low and Medium Income Countries
Using Bayesian meta-analysis methods, we estimate the impact of unconditional cash transfers on nine primary outcomes in a sample of 50 randomized evaluations from middle and low income countries. We find strong and positive average treatment effects per dollar transferred on consumption, food security, and income. Critically, given theoretical and policy debates on the topic, we do not find evidence that cash discourages work. We document important heterogeneity along four key dimensions. First, there are important differences by disbursement schedule. Though both stream and lump sum transfers generate increases in consumption and income, the impact of streams flows through food consumption and labor supply while lump sum transfers tend to facilitate asset accumulation. Second, treatment effects across the board show no sign of dissipation over time. Third, effects generally tend to be larger for samples that are less poor to begin with. Fourth, and in line with earlier work, we find slightly diminishing marginal returns to transfer size. Benefit-cost analyses that define benefits as the net present value of the predicted treatment effect on consumption yield ratios between 83% and 414%.