Controlling for group-level heterogeneity in causal forest

Causal forest is part of a growing class of doubly-robust machine learning based estimators that non-parametrically recovers heterogeneity in treatment effects. However, causal forest’s usefulness is currently limited because the group-level heterogeneity present in many economics settings violates a key assumption of causal forest required for the recovery of unbiased effects. We provide a solution: estimate group-level fixed effects in a regression, create a vector of fixed effects coefficients, and include this vector in the casual forest estimation. Monte Carlo simulations show our solution’s success and the shortcomings of alternatives. Our study greatly increases the number of settings in which unbiased, heterogeneous treatment effects are recoverable