Balancing external vs. internal validity: An application of causal forest in finance

92 Pages Posted: 20 May 2020 Last revised: 23 Aug 2022

See all articles by Huseyin Gulen

Huseyin Gulen

Purdue University - Krannert School of Management

Candace Jens

Syracuse University - Whitman School of Management

T. Beau Page

Government of the United States of America - Office of the Comptroller of the Currency (OCC)

Date Written: May 12, 2022

Abstract

Answering a causal question with results extendable outside of a narrow sample is challenging. Regression discontinuity design (RDD) provides results with strong internal but weak external validity. Using Monte Carlo experiments, we compare the performance of RDD against causal forest, a non-parametric, machine-learning-based matching estimator, at recovering estimates in panel data. We show causal forest’s observation-level, heterogeneous treatment effects are robust to confounding so bias is low in many settings. Moreover, any potential bias in forest estimates can be bounded. We re-visit a popular RDD design, debt covenant defaults, to show how extendable and heterogeneous causal forest estimates enhance inferences.

Keywords: causal forest, investment, financing, RDD, machine learning

JEL Classification: G32, G31, C50

Suggested Citation

Gulen, Huseyin and Jens, Candace and Page, Beau, Balancing external vs. internal validity: An application of causal forest in finance (May 12, 2022). Available at SSRN: https://ssrn.com/abstract=3583685 or http://dx.doi.org/10.2139/ssrn.3583685

Huseyin Gulen

Purdue University - Krannert School of Management ( email )

1310 Krannert Building
West Lafayette, IN 47907-1310
United States

Candace Jens (Contact Author)

Syracuse University - Whitman School of Management ( email )

721 University Avenue
Syracuse, NY 13244-2130
United States

Beau Page

Government of the United States of America - Office of the Comptroller of the Currency (OCC) ( email )

400 7th Street SW
Washington, DC 20219
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
485
Abstract Views
2,417
rank
84,703
PlumX Metrics