Heterogenous Treatment Effect in Microcredit on Those Who Take It Up

Abstract

Randomized experiments are now widely used in analyzing policy implementation or evaluation because researchers can estimate the average treatment effect between the control and treated groups. While, treatment effects are variable in different subgroups, which is interesting for us to explore. The paper applies novel machine learning technology that detects heterogeneous treatment effect through a data-driven approach. Specifically, the paper applies causal forest to a dataset derived from a random experiment did in Morocco by discussing how each different variables contribute to the final result of borrowing.