Screen readers: Two navigational links to follow.Skip to site navigation.Skip to page content.
The University of Iowa News Services
The University of Iowa News Services Home News Releases UI in the News Subscribe to UI News Contact Us

University of Iowa News Release

 

Sept. 1, 2009

Tippie researcher tests investor opinion divergence measures

One investor says buy, another says sell. It's what makes the market work. But how big is the difference between the two investors' beliefs about the stock's value?

Researchers have tried for years to figure that out, and have developed a number of measures in an effort to explain it. Now a researcher at the University of Iowa's Tippie College of Business has developed a two-step process he thinks shows which of those measures are most accurate at determining what is known as investors' opinion divergence.

"I was looking for a true measure of the magnitude of two people's disagreement about the value of a stock," said Jon Garfinkel (left), associate professor of finance.

Researchers are interested in opinion divergence because it might be able to provide clues as to how a stock will move in the future. Opinion divergence is one factor that's thought to affect a stock's risk, and researchers think it could also combine with other factors to prevent a stock from settling at its true price. Divergence is also thought to affect how traders react to a firm's earnings news.

But while researchers have developed several tools to measure opinion divergence, nobody's completely sure how accurate they are. This is where Garfinkel comes in, as the process he developed seems to suss out which of the previous literature's measurements are the most accurate.

To develop his measure, Garfinkel examined limit orders placed with the New York Stock Exchange during three months in 2002. But just getting the data turned out to be a trick in itself. First, he had to convince the NYSE, which is rather protective of its proprietary data, to share it with him.

"But I was very precise in my request," he said. "I told them exactly what I needed and what I planned to use it for, so they wouldn't think I was on a fishing expedition." Still, he said he and his research assistant had to sign non-disclosure agreements before they would release it.

Then he had to actually analyze the data, which was so immense -- measuring somewhere between 100 and 200 GIGs -- that the NYSE delivered it to him on a hard drive.

Garfinkel said the fact the data was proprietary and not available to the public (it still isn't) was key to his research because it was information that no previous researcher had used to address this kind of question. It showed what price investors were asking to trade at, providing a clearer picture of their price targets for the stock.

Garfinkel used the data to calculate the standard deviation across orders of the differences between each limit order price for a stock and the stock's most recent trade price. He then compared his measure to the previously used measures of opinion divergence and found that "unexplained volume" was most accurate.

Unexplained volume begins with a previous day's volume as an indicator of a stock's expected volume on any given day, and compares this with actual volume. If actual volume significantly exceeds what is expected and there's nothing to explain that difference (such as an earnings report or other news story), it suggests investors are trading a lot because they disagree about the stock's value.

His process also found that one measurement frequently used by researchers -- analyst forecast dispersion -- did not fare so well. That measure uses analysts' opinions as a reflection of investor opinion, but Garfinkel thinks it may be a weaker indicator because what analysts think doesn't always align with what investors think.

Garfinkel's research, "Measuring Investors' Opinion Divergence," will appear in a forthcoming issue of the Journal of Accounting Research.

STORY SOURCE: University of Iowa News Service, 300 Plaza Centre One, Iowa City, Iowa 52242-2500

MEDIA CONTACT: Tom Snee, 319-384-0010 (office), 319-541-8434 (cell), tom-snee@uiowa.edu