How Informative Is the Text of Securities Complaints?

Much of the research in law and finance reduces long, complex texts down to a small number of variables. Examples include the coding of corporate charters as an entrenchment index or characterizing dense securities complaints by using variables that capture the amount at issue, the statutes alleged to have been violated, and the presence of an SEC investigation. Legal scholars have often voiced concerns that this type of dimensionality reduction loses much of the nuance and detail that is embedded in legal text. This paper assesses this worry by applying text analysis and machine learning to a corpus of more than five thousand complaints filed in private securities class actions that collectively contain over 90 million words. This analysis demonstrates the richness of the information embedded in these complaints by answering three related questions. First, does the text provide indications of the eventual outcomes in the cases? Second, do text-based variables out perform non-text variables in predicting outcomes? And, finally, does text analysis outperform market participants when used as a basis for picking securities? The evidence suggests that the answer to all three of these questions is yes and these findings have several implications. First, the results suggest that there is meaningful information embedded in text and, relatedly, that excessive reduction of dimensions may produce omitted variable bias. Second, evidence of abnormal returns is consistent with the understanding that it takes time for market participants to absorb complex information. Finally, the ability to predict first-filed complaints more readily than consolidated complaints suggests that the goal of the Private Securities Litigation Reform Act to reduce the race to the courthouse and improve the quality of initial complaints has not been fully realized