Does Mobile Money Improve Livelihoods? A Bayesian Approach to Evidence Aggregation

Using microdata from 9 experimental studies, I examine the causal effect of mobile money services on poverty-related outcomes in developing countries, and evaluate whether these effects generalize across contexts. I leverage a range of Bayesian hierarchical models to conduct a meta-analysis of field experiments that provide access to mobile money accounts in South Asia and Africa. I show that the average effects of mobile money on most financial outcomes are small and uncertain, and that these results hold across a range of specifications. Contrary to the popular observational and policy literature, I estimate that in experimental settings effects are least likely to generalize for remittance outcomes. I also demonstrate that the results from more simple meta-analytic models - which ignore intra-study heterogeneity - tend to overestimate the effects of mobile money. Estimating various Bayesian models of distributional effects, I find little evidence to support the claim that mobile money harms households or businesses.