What Are You Saying? Using <i>topic</i> to Detect Financial Misreporting

TLDR

The study uses machine learning to evaluate whether the thematic content of financial statement disclosures can improve prediction of intentional misreporting. The authors employ a Bayesian topic modeling algorithm to quantify thematic content in 10‑K narratives from 1994–2012. The algorithm yields semantically meaningful topics that predict misreporting, improving detection by up to 59% over models using standard financial and textual variables, and outperforming traditional models in identifying serious revenue recognition and core expense errors.

Abstract

ABSTRACT We use a machine learning technique to assess whether the thematic content of financial statement disclosures (labeled topic ) is incrementally informative in predicting intentional misreporting. Using a Bayesian topic modeling algorithm, we determine and empirically quantify the topic content of a large collection of 10‐K narratives spanning 1994 to 2012. We find that the algorithm produces a valid set of semantically meaningful topics that predict financial misreporting, based on samples of Securities and Exchange Commission (SEC) enforcement actions (Accounting and Auditing Enforcement Releases [AAERs]) and irregularities identified from financial restatements and 10‐K filing amendments. Our out‐of‐sample tests indicate that topic significantly improves the detection of financial misreporting by as much as 59% when added to models based on commonly used financial and textual style variables. Furthermore, models that incorporate topic significantly outperform traditional models when detecting serious revenue recognition and core expense errors. Taken together, our results suggest that the topics discussed in annual report filings and the attention devoted to each topic are useful signals in detecting financial misreporting.