Wall Street analysts still exuberant in earnings projections

March 21, 2008

University Park, Pa. — Wall Street analysts still forecast earnings per share (EPS) growth two times the level of historic GDP growth even after a landmark billion-dollar settlement aimed at rooting out analyst bias, according to new research co-authored by a professor at Penn State's Smeal College of Business.

"Wall Street analysts basically do two things — recommend stocks to buy and forecast earnings," says J. Randall Woolridge, professor of finance at Smeal and co-author of the report. "Previous studies suggest their stock recommendations do not perform well, and now we show that their long-term earnings per share growth rate forecasts are excessive and upwardly biased."

Woolridge and co-author Patrick Cusatis, assistant professor of finance at Penn State Harrisburg, examined analysts' long-term (three to five years) and one-year ahead annual growth rate estimates for all companies from 1984 to 2006. Their findings show that, for both types of forecasts, analysts consistently project EPS growth rates much higher than actual growth and that firms rarely meet or exceed their projected EPS growth rates.

"Actual EPS growth surpassed forecasted long-term growth in only two instances," says Woolridge. "And both of those times represent earnings recovery periods following recessions. For the rest of the 20-year period we examined, analysts significantly overestimated EPS growth.

Over the entire time period, analysts' long-term forecasted EPS growth averaged 14.7 percent, but companies only averaged actual long-term EPS growth of 9.1 percent.

On one-year forecasts, analyst projections fared a little better, but they were still overly optimistic. The average forecasted one-year EPS growth rate was 13.8 percent compared to the actual rate of 9.8 percent.

"A significant factor in the upward bias in long-term earnings rate forecasts is the reluctance of analysts to forecast negative EPS growth," says Woolridge.  On this issue, the researchers find that, on average, 31.1 percent of all companies experienced negative earnings growth over successive three-to-five year periods, but analysts' project negative EPS growth less than one percent of the time.

Further evidence of upward analyst bias is found when comparing growth forecasts with actual GDP. Historic GDP growth over the past 40 years has averaged 7.4 percent while analysts project long-term growth at an average rate of 14.7 percent.

Analyst bias may be the result of several factors, according to Cusatis and Woolridge. The first explanation is based on career concerns or conflicts of interest.

"Analysts are rewarded for biased forecasts by their employers who want them to hype stocks so that the brokerage house can garner trading commissions and win underwriting deals," the researchers write.

This conflict of interest should have been squelched by former New York Attorney General Elliot Spitzer's investigation and the $1.5 billion payment made by U.S. investment firms in the 2003 Global Analysts Research Settlements (GARS). However, Cusatis and Woolridge find that GARS had no affect on analysts' growth-rate forecasts, which have remained around their historic levels of about 15 percent.

A second explanation for the bias is the fact that analysts only follow stocks that they recommend and therefore do not issue forecasts on stocks they don’t like. Finally, the bias may be the result of analysts becoming attached to the companies that they follow and, as a result, losing objectivity.

"If analysts systematically believe that they follow companies that are superior to others, they will be reluctant to issue negative earnings forecasts," the professors write. "Since they are only projecting the companies they follow, and not the market, the end result is a strong upward bias on earnings projections."

"The Accuracy of Analysts' Long-Term Earnings Per Share Growth Rate Forecasts" is forthcoming.


(Media Contacts)

Last Updated March 19, 2009