Balancing External vs. Internal Validity: An Application of Causal Forest in Finance
Abstract
Answering causal questions with generalizable results is challenging. Estimators requiring pseudorandomization provide estimates with no bias (i.e., strong internal validity) but limited generalizability (i.e., weak external validity). Theoretically, causal forest, a nonparametric, machine learning–based matching estimator, can provide low-to-no-bias, generalizable estimates even when treatment is endogenous. We empirically compare the performance of ordinary least squares (OLS), regression discontinuity design (RDD), and causal forest at recovering estimates in simulated observational panel data and show the robustness of causal forest estimates to many sources of bias. We revisit a popular RDD setting, debt covenant default, to show how extendable, heterogeneous causal forest estimates can enhance inferences.
This paper was accepted by Tomasz Piskorski, finance.
Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.00109.
