The Importance of Distinguishing Errors from Irregularities in Restatement Research: The Case of Restatements and CEO/CFO Turnover

TLDR

Research on corporate restatements has expanded, yet existing studies often conflate intentional misstatements (irregularities) with unintentional errors in datasets like the GAO database. The authors contend that distinguishing errors from irregularities can substantially improve the statistical power of restatement studies, especially as error‑related restatements become more common. They propose a simple classification procedure grounded in prior research, extensive review of restatement announcements, and guidance from legal, audit, and SEC sources to label each restatement as error or irregularity. The classification reveals that irregularities are far more likely to trigger fraud‑related class actions, elicit a markedly larger negative market reaction (−14 % vs.

Abstract

ABSTRACT: Research on restatements has grown significantly in recent years. Many of these studies test hypotheses about the causes and consequences of intentional managerial misreporting but rely on restatement data (such as the GAO database) that contains both irregularities (intentional misstatements) and errors (unintentional misstatements). We argue that researchers can significantly enhance the power of tests related to restatements by distinguishing between errors and irregularities, particularly in recent periods when the relative frequency of error-related restatements is increasing. Based on prior research, the reading of numerous restatement announcements, and the guidance that boards receive from lawyers, auditors, and the SEC on how to respond to suspicions of deliberate misreporting, we propose a straightforward procedure for classifying restatements as either errors or irregularities. We show that most of the restatements we classify as irregularities are followed by fraud-related class action lawsuits as compared to only one lawsuit in the group of restatements classified as errors. As further validation of our proxy, we report that the market reaction to the restatement announcement for our irregularities sample (−14 percent) is also significantly more negative than it is for our errors sample (−2 percent). Finally, we demonstrate the importance of distinguishing errors from irregularities by showing the impact it has on inferences about the relation between restatements and CEO/CFO turnover over time.