The market-research company FactSet reports that for each quarter over the past five years, an average of 72 percent of companies in the S&P 500 beat earnings estimates. Past research, including by University of Pennsylvania’s Scott Richardson, University of California at Irvine’s Siew Hong Teoh, and Boston University’s Peter D. Wysocki has found that analysts’ forecasts become more pessimistic and thus beatable as the quarter end approaches, but an unaddressed question is how this walk-down affects clients. If analysts revise their forecasts downward each quarter to placate managers, wouldn’t this confuse the investors who ultimately pay for their services?
According to Chicago Booth’s Philip G. Berger and Washington University’s Charles G. Ham and Zachary R. Kaplan, analysts walk down forecasts by suppressing positive news from quarterly forecasts, not by issuing misleading negative revisions. When analysts have positive news, they will often revise the share price target upward or state explicitly that they expect companies to beat earnings estimates, while leaving the quarterly forecast unrevised. Suppressing positive news leads to beatable forecasts—behavior that benefits corporate executives but carries important implications for both the individual investors who rely on these predictions and researchers studying investor expectations.
When securities analysts receive updated information after issuing a quarterly forecast, they have three options: revise the current-quarter earnings forecast; issue an alternative forecast signal, such as a revision to the share price target or future-quarter earnings; or issue no additional forecast.
By not disseminating all information through the current-quarter earnings forecast, which is widely available through commercial databases, analysts provide an advantage to investment clients who have paid for access to the full breadth of their research product. “Analysts convey information in ways that enable them to be of service to clients, who they care about, and, at the same time, to avoid displeasing corporate managers, who they also care about,” Berger says. “Non-clients, who rely on earnings forecasts because they do not have access to the whole of an analysts’ work product, end up with skewed information, but this is not a primary concern for the analysts’ business.” The researchers demonstrate that a simple strategy based on buying companies expected to beat earnings, using share price target revisions and the text of reports, yields significant abnormal returns, suggesting the market does not see through the analysts’ strategy for conveying information selectively.
Previous research, including that of Stanford’s Maureen McNichols and University of Waterloo’s Patricia C. O’Brien, demonstrates that analysts issue forecasts selectively—not always revising them to fully incorporate the latest information. But the prior research has not been able to disentangle whether such omissions occur because analysts intentionally withhold information or unintentionally omit it, perhaps because they lack (or misunderstand) the new information. Berger and his colleagues solved this problem by focusing on analysts who published—in ways other than updating the current-quarter earnings forecast—information about the same company at the same time they chose not to update the current-quarter earnings forecast.
The researchers find that analysts were more likely to issue alternative forecast signals rather than revising current-quarter earnings forecast reports for positive news, whereas they were more likely to revise their current-quarter estimates downward for negative news.
By showing that analysts’ options for responding to new information have distinct implications for the expected value of future earnings, the research design allowed the authors to shed light on the motives for—and consequences of—the decision to include or omit new information in the current-quarter earnings forecast. It did so by considering whether analysts choose forecasts to revise depending on the positive or negative direction of the news.
Using data from Thomson Reuters’ I/B/E/S forecast compiling system, the researchers analyzed 847,471 analyst reports from the first quarter of 1999 through the third quarter of 2016, representing 8,860 analysts and 7,933 companies. They find that analysts were more likely to issue alternative forecast signals rather than revising current-quarter earnings forecast reports for positive news, whereas they were more likely to revise their current-quarter estimates downward for negative news. This suggests analysts may be responding to incentives—such as continued access to top corporate executives—that encourage them to issue forecasts that managers will meet or beat.
Analysts were also less likely to revise their current-quarter earnings forecast when doing so would have moved them away from the consensus, an indication of herding behavior. And they issued alternative forecast signals in lieu of updating their current-quarter forecasts more frequently when the quantity of news was low, consistent with there being costs to analysts and the users of their reports in making such updates.
The work that analysts produce is complex and has multiple distribution outputs. Because analysts issue a variety of forecasts—including ones that don’t focus on current earnings, such as analyst reports, share-price target revisions, and revisions to future-quarter earnings forecasts—revising a subset, rather than all, of their reports can reduce processing costs for themselves and their clients. It also allows analysts to cater to managers’ preferences for current-quarter estimates that companies can beat.
The researchers’ findings support the idea that the high value that analysts place on access to management along with the low value that institutional clients attach to earnings-forecast accuracy may motivate analysts to selectively decrease the flow of information into the current-quarter earnings forecast. In addition to potentially influencing individual investors who rely on these reports, such behavior has critical implications for academic researchers who use analysts’ current-quarter forecasts to gauge investor expectations.