Investors prone to extrapolation bias before earnings announcements, study finds

The human brain is a funny thing. In its quest for efficiency, it sometimes takes mental shortcuts when trying to understand a situation or make a decision. Investors are not immune from these belief formation impulses, says University of Notre Dame behavioral finance researcher Peter Kelly. He sees evidence of one such unconscious tendency at work in investor behavior surrounding company earnings announcements. His study, “Earnings Announcement Returns Extrapolation,” forthcoming in the Review of Accounting Studies, finds that when past earnings announcement returns were high, individual investors often became overly optimistic about a firm’s future earnings and were more likely to buy stock right before the next earnings announcement.

People tend to assume, or extrapolate, a high level of certainty about future outcomes based on a small sample of past outcomes, according to Kelly, an assistant professor of finance at Notre Dame’s Mendoza College of Business. “You see a stock going up and you think it’s going to continue to go up. You see a stock going down and you think it’s going to continue to go down. There’s a lot of evidence to suggest that this is how people actually form beliefs in financial markets.” Of course, that’s not how stock markets work, as an overview of historical return patterns would tell you. And yet extrapolative beliefs, which give undue weight to recent outcomes, still crop up as some investors try to predict the future. Earnings announcements provide an ideal setting to study the impact of extrapolative beliefs on individual investor behavior. “You extrapolate how good earnings will be and then the earnings actually come out,”  Kelly said, who collaborated on the study with Aytekin Ertan of London Business School, Stephen Karolyi of Carnegie Mellon University and Robert Stoumbos of Columbia University. He and his colleagues theorized that if some investors were extrapolating beliefs from recent earnings announcement returns and using this knowledge to make bets on upcoming earnings announcement returns, they would buy stock right before the next earnings announcement, pushing up the price. To test their theory, the researchers examined data from 78,000 individual investor accounts between 1991 and 1996 at a large discount brokerage firm. They calculated investors’ extrapolative expectations of future earnings announcement returns as a weighted average of its eight past earnings announcement returns. “We see that people are much more likely to purchase in the period right before an earnings announcement if previous earrings announcements returns have been quite hot,” Kelly said. “And you also see some asset pricing evidence here, too, in the sense that stock prices tend to increase a lot more for firms that have had very, very good earnings announcements in the past.” The researchers found evidence of over-extrapolation by some investors, as well. Once the earnings announcements were released, Kelly said, they saw a price reversal, indicating that over-extrapolating investors pushed the price too high and fundamental traders responded. “Our evidence suggests that extrapolation may help explain predictable return patterns around EAs, namely the pre-EA run-up and post-EA reversal,” the researchers wrote. Kelly, who teaches behavioral finance at the undergraduate level, said that individual investors, in general, should be wary of how over-extrapolating can impact pricing. “Since people over-extrapolate, shareholders are going to enjoy a run-up before the earnings announcement,” he said. “Buf if they hold through the earnings announcement, which a lot of them do, then they’re, on average, going to experience a reversal. And that can be costly for them from a trading perspective.” His research focuses on empirical issues in behavioral finance, especially as it relates to investments. He joined the Notre Dame finance department in 2015.