Bayesian Causal Forests & the 2022 ACIC Data Challenge: Scalability and Sensitivity

TL;DR

This paper introduces flexBCF, a scalable implementation of Bayesian Causal Forests, applied to the 2022 ACIC Data Challenge, demonstrating improved calibration of uncertainty intervals and analyzing sensitivity to modeling choices.

Contribution

We developed flexBCF, a scalable and flexible version of BCF, and evaluated its performance and sensitivity in a large-scale causal inference challenge.

Findings

flexBCF improved uncertainty interval calibration

Point predictions were robust to modeling choices

Sensitivity to propensity score estimation was analyzed

Abstract

We demonstrate how Hahn et al.'s Bayesian Causal Forests model (BCF) can be used to estimate conditional average treatment effects for the longitudinal dataset in the 2022 American Causal Inference Conference Data Challenge. Unfortunately, existing implementations of BCF do not scale to the size of the challenge data. Therefore, we developed flexBCF -- a more scalable and flexible implementation of BCF -- and used it in our challenge submission. We investigate the sensitivity of our results to the choice of propensity score estimation method and the use of sparsity-inducing regression tree priors. While we found that our overall point predictions were not especially sensitive to these modeling choices, we did observe that running BCF with flexibly estimated propensity scores often yielded better-calibrated uncertainty intervals.