NEW YORK - When it comes to earnings estimates, close agreement among analysts isn't necessarily a good thing. "It makes me nervous when analysts are that sure of themselves," says Joseph Kalinowski, equity strategist at Thomson Financial/IBES.
Close agreement among analysts should draw more suspicion when valuations are high in a particular sector and analysts are revising their estimates upward in lockstep. "In that situation, if a company starts to miss or gives downward guidance," Kalinowski explains, "it can all come crashing down."
On the other hand, too much divergence among analysts can signal a problem. For example, 2001 earnings-per-share (EPS) forecasts for Lucent Technologies (nyse: LU - news - people) range from 40 cents to -$1.25 per share. Given Lucent's 3.4 billion common shares outstanding, this represents a difference of $2.9 billion in projected losses.
How do investors know when variation between earnings estimates is too great, too little, or just right? Kalinowski uses a statistic known as the coefficient of variation (CV), which measures how similar earnings forecasts are to each other. Generally, a small CV indicates strong agreement among analysts and a large CV indicates strong disagreement.
Kalinowski suggests that investors compare a company's current CV against its historical CV, or against the industry average. "When the CV is in line with and above either of these numbers, earnings surprises tend to have less effect on the price of a company's stock," explains Kalinowski.
Full story at Forbes.com