Understanding the Impact of Microcredit Expansions : A Bayesian Hierarchical Analysis of 7 Randomised Experiments

Bayesian hierarchical models serve as a standard methodology for aggregation and synthesis, used widely in statistics and other disciplines. I use this framework to aggregate the data from seven randomised experiments of expanding access to microcredit, assessing both the general impact of the intervention and the heterogeneity across contexts. The general impact on household profits is small, with a posterior mean of 26 USD PPP per year, and an impact of zero lies well within the central 50% posterior credible interval. Standard pooling metrics for the studies indicate 65-95% pooling on the treatment effects, suggesting that the site-specific effects are informative for each other and for the general case. Further analysis incorporating household covariates shows that the cross-study heterogeneity is almost entirely generated by heterogeneous effects for the 27% households who previously operated businesses before microcredit expansion. A cautious assessment of the correlations between site-specific covariates and treatment effects using a Bayesian Ridge procedure indicates that the interest rate on the microloans has the strongest correlation