Strategic Disclosure Misclassification
Carnegie Mellon University – Tepper School of Business
Stephen A. Karolyi
Carnegie Mellon University – Tepper School of Business
At this year's SALT New York conference, Wences Casares, the chairman of XAPO, and Peter Briger, the principal and co-chief executive officer of Fortress Investment Group discussed the macro case for Bitcoin. Q2 2021 hedge fund letters, conferences and more XAPO describes itself as the first digital bank of its kind, which offers the "convenience" Read More
University of Minnesota – Carlson School of Management
May 11, 2016
When SEC disclosure guidelines are not bright-lined, do firms use discretion strategically? To answer this question, we apply modern machine learning techniques to characterize disclosure misclassification by public companies. We find that 17% of disclosures are misclassified, and those concerning material definitive agreements, executives or directors contracts and turnover, and delisting notices are most commonly misclassified. Using EDGAR search traffic data, we provide evidence that misclassification successfully reduces investor attention. Through attention, misclassification leads to a significant and persistent impact on absolute market reactions. For misclassified filings, search traffic is 45% lower and absolute market reactions are 79 bps smaller. Finally, we find that managers strategically misclassify disclosure if the news is negative and when market attention is high.
Strategic Disclosure Misclassification – Introduction
The objective of the Securities Exchange Act of 1934 was to standardize, regulate, and enforce public disclosure of relevant corporate information. This objective was accomplished, in part, by requiring companies to file annual 10-K and quarterly 10-Q financial statements. For changes in operational or financial conditions that occur between periodic reports but would still be of interest to investors, the Securities and Exchange Commission (SEC) also mandated 8-K filings, or “current reports”.
This disclosure regime evolved to the point that public companies pursuant to it issued 327,775 filings in 2003 alone and has led to concerns that investors with limited attention may not be able to discern material events from immaterial ones.1 The SEC has taken steps to alleviate investors’ search costs by categorizing disclosures into groups based on their content. This categorization effort has resulted in a classification-based system of disclosure|current reports now include content-based classifications, which ostensibly help investors filter current reports based on their needs. On the other hand, if classification is not subject to bright-lined rules, such a system may create the scope for opportunistic discretion (Dye, 2002; Gao, Sapra and Xue, 2016).
In this paper, we study the 8-K classification scheme to determine if firms strategically misreport the classification of disclosures concerning material changes in operational or financial conditions to avoid investor attention. However, our empirical examination of strategic disclosure misclassification is subject to several empirical challenges. Most importantly, identifying misclassified disclosures is difficult because the “correct” classification for each disclosure is a latent construct and not subject to bright-line rules. To overcome this challenge, we employ novel machine learning techniques to classify 8-K lings based on their textual content. That is, we develop a latent classification system based on textual similarity and identify misclassified filings based on deviations between realized classifications and our latent classification. With this novel measure, we explore misclassification motives, complementarities with other strategic disclosure choices, and the market consequences of misclassification. Our evidence suggests that (i) a significant fraction (17%) of 8-K filings are misclassified, (ii) managers strategically misclassify negative news, particularly when market attention is high, and (iii) misclassification successfully reduces investor attention and absolute market reactions. These results have important implications for the design of the SEC’s disclosure system, investor information acquisition behavior, and corporate disclosure policy.
Because we rely on deviations from realized 8-K filing classifications, our approach depends on the existing classification system of 8-K filings. By the early 2000s, 8-K lings were mandatory for twelve distinct item classifications (i.e. news categories) and companies had a maximum of fifteen days to report an event.3 On August 23, 2004, the Securities and Exchange Commission (SEC) mandated new disclosure requirements for Form 8-K. These new requirements expanded the number of items separately classified from twelve to twenty-six and reduced the maximum lag between the event date and reporting date to four business days (Lerman and Livnat, 2010). With this new disclosure regime, the SEC attempted to meet investor demand for timely, relevant, and reliable information and to reduce securities fraud.
By increasing the number of distinct items and creating a new topical format, the SEC reduced ambiguity over what events should be disclosed on a mandatory basis. Despite these changes, the SEC did not remove the catch-all “Other Events” item, now known as Item 8.01, which suggests that, unlike categorized items, the SEC did not anticipate investor demand for information events that may be filed under this category. As investor demand for information contained in filings is related to market reactions, events led under the Item 8.01 category are likely to be met with smaller market reactions. Consistent with this conjecture, (Lerman and Livnat, 2010) finds that Item 8.01 filings have lower absolute market reactions, on average, than other item types, which makes it the ideal target classification for misclassified filings.4 Therefore, we focus our identifying variation on misclassified filings that have a realized classification of Item 8.01.
See full PDF below.