Journal of Law, Finance, and Accounting > Vol 5 > Issue 2

Machine Learning and Predicted Returns for Event Studies in Securities Litigation

Andrew Baker, Stanford University, USA, abaker2@stanford.edu , Jonah B. Gelbach, University of California at Berkeley, USA, gelbach@berkeley.edu
 
Suggested Citation
Andrew Baker and Jonah B. Gelbach (2020), "Machine Learning and Predicted Returns for Event Studies in Securities Litigation", Journal of Law, Finance, and Accounting: Vol. 5: No. 2, pp 231-272. http://dx.doi.org/10.1561/108.00000047

Publication Date: 08 Sep 2020
© 2020 A. Baker and J. B. Gelbach
 
Subjects
Event studies/market efficiency studies,  Asset pricing,  Financial econometrics,  Law and economics
 
Keywords
JEL Codes: K220, G120
Event studies/market efficiency studiesfinancial econometricsasset pricing
 

Share

Login to download a free copy
In this article:
1. Introduction 
2. Prior Literature 
3. The Benefits of Data-Driven Methods for Event Studies Used in Litigation 
4. Methodology 
5. Simulation Results 
6. Further Results 
7. Conclusion 
References 

Abstract

We investigate the use of machine learning (ML) and other robustestimation techniques in event studies conducted on single securities for the purpose of securities litigation. Single-firm event studies are widely used in civil litigation, with billions of dollars in settlements hinging on the outcome of the exercise. We find that regularization (equivalently, penalized estimation) can yield noticeable improvements in both the variance of event-date abnormal returns and significance-test power. Thus we believe that there is a role for ML methods in event studies used in securities litigation. At the same time, we find that ML-induced performance improvements are smaller than those based on other good practices. Most important are (i) the use of a peer index based on returns for firms in similar industries (how this is computed appears to be less important than that some version be included), and (ii) for significance testing, using the SQ test proposed in Gelbach et al. (2013), because it is robust to the considerable non-normality present in abnormal returns.

DOI:10.1561/108.00000047