Rachael Meager

• May 01, 2018
• Speaker: Rachael Meager
  
• Topic: Evidence Aggregation in the Presence of Heterogeneity: Bayesian Hierarchical Models for Research and Policy Decisions
  

Understanding the generalizable impact of social and economic interventions requires aggregating evidence from multiple studies conducted in different settings. However, such aggregation must allow for the possibility of heterogeneity in effects both within and across different contexts. I use Bayesian Hierarchical models to perform this evidence aggregation for both average treatment effects and distributional effects, with a motivating application of the microcredit literature. Several randomized trials of expanding access to microcredit found inconclusive average treatment effects, yet substantial quantile treatment effects on the tails of household outcome distributions, but the extent to which these findings generalize to future settings was not known.

Aggregating the evidence on sets of quantile effects poses additional challenges relative to average effects because distributional effects must imply monotonic quantiles and pass information across quantiles. Developing and applying these new methods, I find generalizable evidence that microcredit has negligible impact on the distribution of various household outcomes below the 75th percentile, but above this point there is no generalizable prediction. Households with previous business experience account for the majority of the impact in the tails.
 
For information about Rachael, please see her site here.