Analysts Might Herd, But They Don’t Trample InvestorsPublished: September 08, 2011 in Knowledge@Emory
When the dot.com bubble started to burst in March 2000, those who’d been disbelievers in the high-flying sector pounced. In his widely watched annual letter in April 2001, Warren Buffett wrote that “it was as if some virus, racing wildly among investment professionals as well as amateurs, induced hallucinations in which the value of stocks in certain sectors became decoupled from the values of the businesses that underlay them.” In the bubble’s post-mortem, some suggested that what Buffett described as a virus might better be explained in terms of “herding”—the tendency for different agents making individual decisions to nonetheless take similar actions at roughly the same time. The theory was that many stock analysts had talked up dot.com stocks to keep pace with their peers, thereby stoking demand that ultimately resulted in excess volatility when the bubble burst.
But in a paper recently published in the Review of Financial Studies, Narasimhan Jegadeesh at Emory University’s Goizueta Business School argues that any ill effect from herding by analysts is usually blunted because investors are well aware of the tendency. In a paper titled “Do Analysts Herd? An Analysis of Recommendations and Market Reactions,” Jegadeesh and colleague Woojin Kim of Korea University analyzed stock price movements and analyst recommendations for some 5,714 firms between 1993 and 2005. They found that investors seem to pay much more attention to analyst recommendations that go against the grain. That contrasts with changes that move toward the consensus viewpoint, which don’t drive as much action by investors. Stock buyers, in other words, are careful not to get trampled by the herd.
“The market is smart enough not to blindly follow analysts,” says Jegadeesh, who is the Dean’s Distinguished Chair of Finance at Goizueta. “The market seems to understand that analysts have some incentives to herd, and the market responds appropriately.”
There are two broad reasons that individuals might herd, Jegadeesh says. The first stems from information. An earnings announcement with positive news about a company’s revenue or net income might prompt many analysts to up their earnings estimates for the firm at around the same time. Analysts who rationally incorporate the latest financial results from the same information source will come to similar conclusions about the company’s direction. But a second reason for herding is not so benign. Individuals might herd because they are responding to incentives that encourage imitation. Previous research, for example, has suggested that analysts with low abilities often issue earnings forecasts that mimic those of superior analysts, all in hopes of winning more compensation.
“If agents act similarly only because of common information, then their actions result from rational and optimal use of information, and they are not influenced by extraneous incentives or biases,” the researchers write. “On the other hand, if agents herd for any other reason, then their actions differ from the action that one would take if he or she uses available information rationally without any influence of extraneous incentives or biases . . . Non-information-driven herding could introduce noise in prices, and contribute to excess volatility that many view as undesirable.”
Analysts typically rate stocks as “strong buy,” “buy,” “hold,” “sell,” or “strong sell.” To investigate whether analysts herd when they make changes among these five recommendation levels, the researchers tracked 71,555 upgrades and downgrades issued by analysts on the stocks of 5,714 firms between 1993 and 2005. The sample included all firms that had at least two active recommendations from analysts during the time period, including at least one revision. The 6,588 analysts in the sample came from a variety of brokerages ranging from one-analyst shops to large players like Merrill Lynch and Morgan Stanley. The researchers looked companies’ “abnormal returns,” meaning the change in the stocks’ value relative to changes in the value of the broader index, around recommendation changes.
Overall, the researchers found that analyst upgrades corresponded with a 1.97 percent abnormal returns on the day of the upgrade. Analyst downgrades corresponded with a 3.22 percent decline in abnormal returns on the day of the downgrade. The impact was significantly different when recommendation changes moved toward the consensus view relative to when they moved away from the consensus view. In the case of upgrades, the average abnormal returns was 1.75 percent when recommendations moved toward the consensus and 2.15 percent when the upgrade moved away from the consensus. The effect was more dramatic with downgrades. The average abnormal returns were –2.4 percent when the downgrade moved toward the consensus and –4 percent when the downgrade moved against the consensus.
The study also offers information about which analysts are more likely to herd, and when they are more likely to do so. There is no difference, for example, in the herding tendencies of all-star analysts, lead analysts, and other analysts, the researchers found. Analysts who make frequent revisions to their recommendations, however, were less likely to herd. And analysts were less likely to herd when evaluating stocks where there was a broad range of recommendations. Analysts from large, reputable brokerages were more likely to herd, the study suggests.
Regulation also had an influence. In the fourth quarter of 2002, ten major brokerages struck a global settlement with state and federal officials that erected barriers between the investment banking activities of brokerages and their stock research departments. In the time following the settlement, analysts were less likely to herd. This suggests that currying favor with potential investment banking clients might have contributed to herding incentives prior to 2002, the researchers write. Even so, the primary take-away as to the deleterious influence of herding on market prices is that the practice isn’t that worrisome. The study found that a large part of the stock price response to recommendation revisions occurred on the day of the change, and market prices continued to reflect information in the revisions for up to six months into the future. If the revisions had been driven by an impulse to herd that pushed investors away from a true valuation of the stock based on the fundamentals of the company, the study results would have found corrections in market prices at some point following the changes, Jegadeesh says.
“Media accounts and some academic papers have suggested that analysts’ herding tendencies could introduce noise into prices because the market could potentially overweight the common mistakes of the herd,” the study authors write. “However, our results indicate that the market anticipates analysts’ tendencies to herd, and the market price reaction on the revision date accounts for such herding tendencies. Therefore, we doubt that herding by analysts when they make recommendations would have any destabilizing effect on prices.”
The study suggests that downgrades that go against the consensus amount to higher-profile acts that are more likely to significantly move the stock price. In general, analysts seem to be more reluctant to stand out from the crowd when they convey negative information, although it depends on who they work for, the research suggests. Analysts from less prestigious firms might be more willing to offer such recommendations because they are “underdog” players who want to attract attention in the market, Jegadeesh says. Such downgrades might be less common from large brokerage houses where analysts seem more risk averse and are more cautious about offering recommendation changes that could possibly make them look bad. But whatever the motives of analysts, the study suggests that commentators should not overstate the importance of herding when looking for the cause of market ills.
“It’s not what analysts were saying that drove the tech bubble,” Jegadeesh says. The study didn’t set out to explain what caused the dot.com bubble, but Jegadeesh points out that “it could be that investors were looking at their neighbors who were buying those stocks and saying ‘I want to do this’ . . . sometimes things turn out better than we expect, sometimes things turn out worse.”