BLOOMINGTON, Ind. -- In a challenge to prevailing wisdom that CEO and board chair positions should be held by two different people as "best practice," new research indicates that the roles should be split only when there is a performance problem, and then only through a "demotion strategy" that keeps the CEO but brings in an independent chair, as an overt signal to reverse course.

This is the primary finding of a study from the Indiana University Kelley School of Business that is the first to identify the distinct performance consequences of three approaches to CEO-chair separation: apprentice, departure and demotion.

The findings buck current recommendations of legislators, governance experts and analysts seeking to establish separate CEO-chair roles as standard practice. These efforts have gained some traction: Since the 2002 passage of the Sarbanes-Oxley Act, the number of S&P 500 companies dividing the roles has dramatically increased to 43 percent from 25 percent.

Matthew Semadeni. Indiana University

"Companies shouldn't undertake a separation process because they think they should, or because other companies have, but instead take a studied approach that looks at current performance, determines how a change will affect overall performance, and then separate only if necessary," said co-author Matthew Semadeni, associate professor of strategy at the Kelley School.

No link between 'CEO duality' and company performance

According to the researchers, there is no relationship between "CEO duality" -- having one person serve as chair and CEO -- and a company's performance. Further, there is no evidence that poor performance alone triggers a move to divide the responsibilities between two people. Taking action when a company is performing poorly may trigger a turnaround, but it depends on the approach used. However, changing the model during a period of solid returns will likely cause a reversal of fortune.

"Dividing leadership roles when performance is high is like taking medicine when you're not sick: It won't work and may cause other problems," Semadeni said.

While it may seem prudent in the current economic climate to split the CEO and chair roles, co-author Ryan Krause, Ph.D. candidate in strategy at the Kelley School, said the more relevant questions are when and how to do so, because timing and approach affect the outcome.

"Too often, the decision is made out of the blue, without any real planning or consideration of current performance, and ultimately it doesn't change anything," he said.

To better understand the underlying motivations for changing CEO duality and the potential outcomes, Krause and Semadeni looked at 309 firms in the S&P 1500 and Fortune 1000 indices that separated CEO-chair positions from 2002 to 2006. Each used one of three distinct separation strategies:

• Apprentice, in which a sitting CEO-chair relinquishes the CEO title but remains chair, creating a succession event.
• Departure, in which a firm fills the two positions with two new individuals, creating a CEO succession event and a governance change.
• Demotion, in which the CEO retains that title and a new chair is appointed, creating only a governance change.

Measuring firm performance with stock returns and analyst ratings, the Kelley researchers found that companies following a demotion separation strategy subsequently experienced significantly larger performance reversals than firms that separated roles in either an apprentice or a departure situation. In fact, they found the demotion separations reversed the prior year's performance by 150 percent.

Demotion = most corrective action

"Although the demotion strategy carries some risk, it is the most corrective option when used in cases of poor performance because it imposes independent oversight on the CEO and provides the best opportunity to change course," Krause said. "It is an unambiguous signal to the CEO that the firm needs to be fixed and the CEO's only job is to provide a solution."

The demotion strategy was used recently at Chesapeake Energy, which had performance and leadership issues. Such a strategy can be beneficial if the company wants to hold onto an otherwise valuable CEO, as in Chesapeake's case. However, the authors caution that using such a strategy when a CEO is performing well can send mixed signals to leadership, employees and the market -- and turn the company's performance in the wrong direction.

In contrast, the other two strategies don't produce much change. With a departure strategy, the company starts over entirely with two unknown individuals and thus, performance could go in either direction. Keeping the chair but bringing in a new CEO -- as with an apprentice approach that allows a new CEO time to develop skills -- creates no change at all.

"Certainly policy and governance groups are pushing companies to have separate CEO and board chairs, but our findings prove that it is not the right move for all companies," Krause said. "Boards that do this under the wrong conditions can send their company off a cliff, so our caution to them, in the simplest terms, is 'if it ain't broke, don't fix it.'"

"Apprentice, Departure, and Demotion: An Examination of the Three Types of CEO-Board Chair Separation" adds to the current literature on CEO duality and the nature of the working relationship between a CEO and board chair. Future investigation into boardroom processes will better identify how CEO-chair separation can be used as a tool selectively and with care.

The paper is forthcoming in the Academy of Management Journal.