Publicly traded firms often issue earnings forecasts that miss the mark, and a new study from the University of Iowa suggests CEO overconfidence is the reason for the poor predictions.

The study, “CEO Overconfidence and Management Forecasting,” finds that CEOs who are overconfident are more likely to voluntarily issue earnings forecasts. Those forecasts tend to be excessively optimistic—and more likely to be wrong.

The researchers say their findings can help answer a question that analysts have asked for years: Why do CEOs issue earnings forecasts to begin with when a miss can significantly reduce a firm’s stock price and call into question the CEO’s credibility?

“Overconfident managers are generally more optimistic about the future of the firm, and also overestimate their ability to influence earnings and underestimate the probability of unexpected events, such as fluctuations in the business cycle,” says Paul Hribar, professor of accounting in the UI’s Henry B. Tippie College of Business and study co-author.

He says the pressure could even prompt CEOs to manage earnings to avoid missing their own forecasts when they realize they’ve been too optimistic, creating long-term problems for the firm.

Whether it’s ego or hubris or something else, Hribar says the study doesn’t venture into the CEOs’ psychological motivations. But it definitely suggests they hold beliefs that are too optimistic and feel that they predict the future with greater precision than they actually do.

The study looked at the stock prices of more than 2,100 U.S. firms and the confidence level of the 3,300 CEOs who led them. The researchers measured confidence by looking at the value of a CEO’s options when they were exercised because more confident CEOs will hold onto their options on the assumption their value is going to increase. The researchers also used media coverage of each CEO as a measuring stick, looking at how many stories used words like “confident” or “optimistic” to describe a potentially overconfident CEO, as opposed to “cautious” or “reliable,” which suggest one who is more conservative.

The study found that between 36 and 51 percent of CEOs were classified as overconfident each year, and that the higher a CEO was on the confidence scale, the more likely that CEO was to issue an earnings forecast—and also, the more likely that forecast was to be wrong.

The paper, co-authored by Holly Yang, assistant professor of accounting at the Singapore Management University, will be published in an upcoming issue of Contemporary Accounting Research.