According to the research by professors at Texas Christian University, Penn State University and the University of Texas, differences among forecasts and changes in forecasts are separate matters with separate implications. That is counter to previous thought.
The range of differences in analysts' forecasts, known as the level of forecast dispersion, indicates the amount of uncertainty about future earnings. Small differences indicate low uncertainty. Large differences indicate high uncertainty and a greater risk of low earnings. On the other hand, changes in forecast dispersion, resulting from forecasts being revised after a release of earnings information from a company, indicate the existence of information asymmetry. That means that some analysts had more and better knowledge than others.
This view is in striking contrast to earlier research, which used forecast dispersion to examine both uncertainty and information asymmetry, says Dr. Mary Stanford, professor of accounting at the M.J. Neeley School of Business at Texas Christian University in Fort Worth.
"Our evidence allows investors and researchers to interpret levels of dispersion or changes in dispersion around earnings announcements. High levels of dispersion indicate that everybody is unsure about that stock. Whether or not an investor chooses to trade is up to their risk preference, but everybody is facing the same risk," says Dr. Stanford.
"However, a large change in dispersion, indicating high information asymmetry, suggests there are informed and uninformed people in the market, and the uninformed are at a disadvantage," she says.
Of course, analysts on average are better informed than most investors. But some analysts may have more accurate data than others, she explains. The dilemma is that investors can't tell from the forecasts which analysts are in the know and which are not.
"Unless an investor is extremely knowledgeable, a time of big changes in dispersion is not the best time to trade," she says.
The study, "Further Evidence on the Relation between Analysts' Forecast Dispersion and Stock Returns," by Dr. Stanford, Dr. Orie E. Barron of Pennsylvania State University, and Dr. Yong Yu of the University of Texas, is scheduled to be published later this year in the academic journal Contemporary Accounting Research.
They did statistical analyses of the levels of forecast dispersion and changes in forecast dispersion, both quarterly and annually, of hundreds of large companies traded on American stock exchanges from 1986 to 2003. They found that the levels of dispersion and changes in dispersion were clearly revealed to be separate constructs with meanings independent of each other.
This discovery helps explain and reconcile apparent conflicts among the findings of earlier studies.
"Researchers in accounting and finance have long been using forecast dispersion as a measure both of uncertainty and information asymmetry, but those are two different constructs. When you look just at dispersion, you can't know if it's due to uncertainty or information asymmetry," says Dr. Stanford.
"We've shown that researchers interested in uncertainty should examine levels of dispersion, and those interested in information asymmetry should examine changes in dispersion. Now researchers can select the appropriate measure for future investigations and reinterpret the huge body of past research," she says.
The study included only large firms. That's because, to measure uncertainty and information asymmetry, each firm had to be tracked regularly by at least three analysts. The companies actually averaged more than a dozen analysts each, and had an average market value of $10 billion. The inquiry into levels of dispersion studied 650 firms per year. The inquiry into changes in dispersion examined 250 firms per year.
Average traders can use the findings to reduce their risk of losses in a particular stock by refraining from trading when a firm's forecast dispersion levels are high or when the firm is exhibiting substantial changes in dispersion.
"Investors should look at the history of analysts' forecasts, going back maybe eight quarters. When dispersion levels are low, there's low uncertainty about the stock. And when dispersion doesn't change much, investors are more likely to be trading against people with essentially the same information," says Dr. Stanford.
"It's relatively easy to observe dispersion and changes, and now understand what they mean," she says.
RSS