Insider Herding: How Information Flows Within The Firm
Washington State University
University of Tennessee, Knoxville – Department of Finance
In an effort to understand the information content and trading behavior of informed corporate insiders, we explore similarities in insiders’ trading behavior as evidence of herding and a proxy for information flows. We observe that corporate insiders cluster trades around those of insiders at their firm. We also document that trades are more profitable when insiders cluster trades together, especially when following those of CEOs or CFOs. Our results are consistent with intentional insider herding resulting from information sharing. Overall, we offer empirical evidence of daily herding among equity traders and provide insight into how information flows within the firm as insiders trade collectively.
Insider Herding: How Information Flows Within The Firm – Introduction
Do corporate insiders herd together when they trade? Answering this question provides insight into the flow of information between corporate insiders. We perform the first investigation of insider herding and explore the information flows between corporate insiders through similarities in trading behavior. Existing research documents that corporate insider trades exhibit superior stock performance (Jaffe, 1974; Seyhun, 1986; Lin and Howe, 1990; Seyhun, 1998; Lakonishok and Lee, 2001). While aggregating all insider trades often masks the extent of information used by insiders, Lorie and Niederhoffer (1968), Cohen et al. (2012) and Biggerstaff et al. (2015) use patterns of insider trading to identify abnormalities in trading behavior. These time-series patterns in insider trades more clearly identify informed trading.
Our study bridges herding and insider trading literatures by exploring similarities in insiders’ trading behavior as evidence of herding among corporate insiders and as a proxy for information flows within the firm. Academic researchers study herding by equity traders but have largely focused on changes in quarterly holdings by institutional traders (Lakonishok et al., 1992; Wermers, 1999; Sias, 2004; Brown et al., 2013; Koch, 2014). However, Sias (2004) defines herding as “a group of investors following each other into (or out of) the same securities over some period of time.” We investigate whether insiders at a firm herd together when they trade and extend the herding literature by exploring daily herding behavior of investors with information pertaining to a firm’s internal operations – corporate insiders (Lakonishok and Lee, 2001; Sias and Whidbee, 2010; Cohen et al., 2012). We investigate similarities among insiders’ trading patterns and show insiders herd when they trade by documenting daily insider trading patterns that identify informed trades and information sharing among corporate insiders.
We posit at least three plausible explanations for the herding behavior by insiders. First, insider herding could result from private information channels, where relatively more informed insiders privately share information by tipping other less informed insiders and coordinate trading to mutually benefit from the shared information. While insider collusion to mutually profit from information is not in accord with theoretical models, recently researchers suggest alternative predictions. Early models suggest insiders with shared private information should compete aggressively to profit from their tradable information without benefiting other insiders (Holden and Subrahmanyam, 1992; Back et al., 2000). However, Indjejikian et al. (2014) and Ahern (2015) find rational explanations for insiders to share tradable information to jointly profit from private information. Since insider trading in a group reduces the conditional probability of being caught, insiders trading together can find safety in numbers (Sah, 1991; Glaeser et al., 1996).
On the other hand, herding could also result from insiders independently obtaining and trading on related information signals, similar to investigative herding (Lakonishok et al., 1992). In this case, we would not expect insider trade disclosure to influence herds resulting from independent insider decisions based on correlated information sets available to coworkers.
Alternatively, insider herding could result from uninformed insiders viewing the Form 4 insider trade disclosures of insider peers and following their trades in an information cascade (Bikhchandani et al., 1992). Insiders could disregard their own noisy information set and trade with the herd based on information deduced from the public disclosure of other insider trades. We seek to determine if and how corporate insiders herd by examining insider trading patterns to assimilate existing literatures on herding and insider trading.
To understand the motivation for insider trade clusters, we use network-centric analysis to identify trades most likely to be informed within insider trading clusters and discern the relative timing of insider trades.1 We find that 22% of insider purchases occur on the same day of at least one other insider purchase, and likewise, 21% of insider sales occur on the day of another insider sale.2 When we compare the conditional probability of insiders trading around other insider trades, we observe that insiders are 164% (30%) more likely to purchase (sell) shares of stock when other insiders also purchase (sell) shares on the same day. However, insider sales do not predict insider purchases, suggesting herding does not result from restrictive insider trading corporate policies. In addition to limiting our sample to non-routine trades (Cohen et al., 2012), we further control for trading around earnings announcements, where corporations frequently restrict insider trading. Our results suggest recent trading by a peer insider is an economically and statistically significant determinant of insider trading.
Next, we explore whether insider herding represents informed trading activities. We show purchases of insider herds are both contrarian and informed, following poor past performance and preceding superior future performance. We observe that insider purchases followed by another insider’s purchase within two days obtain four percent higher abnormal returns over the subsequent three months, which is twice as large as the abnormal returns following insider purchases that are independent of other insider purchases. Further, we find larger market reactions following insider herds than following independent insider trades. Moreover, we find no evidence of reversals, which makes herding based upon trading fads, characteristics and reputation unlikely explanations for our results (Sias, 2004).
See full PDF below.