To evaluate a firm’s standing and formulate profitability projections, sell-side equity analysts use public information, including financial statements, publications, and peer forecasts. This last tool elicits a form of social learning that shapes the industry experts’ outlooks in perhaps unexpected ways. A working paper reveals that analysts observe each other, from their past mistakes to demographic attributes, for a market evaluator climate influenced by the dynamics of peer social learning.
Examining the learning behaviors of nearly 5,000 analysts from 1984 to 2017, Miami Herbert Business School’s Alok Kumar, Ville Rantala, and Rosy Xu uncover that analysts influence each other through portfolio-wide forecast errors and revisions. When making determinations on a given firm, an analyst looks beyond the focal firm to the other companies in his or her portfolio and considers recent peer forecast errors on these other firms. For example, if last quarter’s peer forecasts for the other firms were systematically higher than actual earnings, an analyst adjusts towards a more pessimistic earnings forecast to correct for the observed deviation. Likewise, systematic lower forecasts prompt an adjustment towards more optimism. In this manner, the paper captures the inverse relationship between an analyst’s relative optimism and peer recent forecast errors.
But analysts do not only adjust for colleagues’ mistakes; they may also imitate their actions, such as in the case of bold forecasts – the more daring projections that deviate from the consensus and one’s own previous forecasts. “A bold forecast means that an analyst takes an independent stance,” Rantala says. “They are likely to capture attention because it shows that one of your colleagues is doing something different from what everybody else is doing.” The observing analyst becomes more likely to issue a bold forecast when peers have recently done the same for other portfolio firms.
The study also captures a one-third overlap among portfolios of analysts that cover the same firm. Hence, by observing peer forecasts on their portfolios as a whole, the equity researchers tap into peer analyses for the two-thirds portions that do not overlap. Observations from the entire portfolio “may allow analysts to discover hidden correlations between firms or industrial links that competitors don’t see,” Rantala says. Yet the forecasters may also miss information from companies outside of an analyzed set. Using the airlines industry as an example, he illustrates: “Peer learning from Delta may influence views on United, but the analyst may miss learning from American if the latter is not in their set of covered firms.”
Beyond these pros and cons, learning among peers extends into the context of demographics. The study shows that analysts respond more to colleagues who share similar personal characteristics, such as ethnicity or gender. For example, men learn more from the men in their field, and women from the women. Through participation in the same earnings calls or by following institutional investor rankings, “analysts are most of the time cognizant of who their peers are,” says Rantala, pointing out that research in social psychology and sociology documents a tendency among people to interact with others who are similar. The finding suggests that peer learning does not only occur as a strategic impulse for market assessment, but also as a cultural dynamic.
Nevertheless, learning from peer forecasts and their outcomes with a portfolio-wide viewpoint – even if influenced more by those who share personal characteristics – mostly leads to more accuracy. The aggregate peer learning effect guides analysts closer to actual realized earnings 67 percent of the time. Evidently, social learning plays an important role for those tasked with formulating equity market predictions. The recommendations that investors receive carry with them the impacts of peer observations, which generally enable the transmittal of more sound advice.
Source:
Kumar, Alok and Rantala, Ville and Xu, Rosy (November 14, 2019). Social Learning and Analyst Behavior. Available at SSRN: https://ssrn.com/abstract=3487103 or http://dx.doi.org/10.2139/ssrn.3487103
