The Washington PostDemocracy Dies in Darkness

How long is too long to be CEO?

(Yuriko Nakao - REUTERS)

If Jeff Immelt indeed steps down sometime before the expected 20-year mark as CEO of General Electric, his tenure will be unusual. Not for being shorter than average — but actually, for being far longer. 

The Wall Street Journal reported Tuesday that there have been talks for him to do so. Immelt, who started the job four days before September 11, 2001, has had a rocky tenure that spanned the financial crisis, during which the company's finance arm came under immense pressure and lost its vaunted AAA credit rating. Still, his 13 years at the top exceed the 9.7 years that are now average for CEOs of Standard & Poor's 500-stock index companies. And even that number is particularly long: 9.7 years is the highest average tenure, according to a recent report by the Conference Board, since 2002. 

Two decades, after all, is long for anyone to keep such a demanding job, however well compensated. As the Journal story cited, the nature of a global CEO job like Immelt's requires such a brutal schedule that it's hard to imagine doing it for five years, much less 10 or 20.

So if 20 years is too long, and 9.7 years is the longest average we've seen in years, what's actually the right tenure for a CEO? Is there one? And is there a point at which the experience, confidence and relationships that come with being in the job for a long time can start to hurt rather than help companies?

Some evidence suggests that CEOs who hold the job for longer than a decade can be good for shareholders. A Fortune survey from 2012 found the sweet spot to be between 10 and 15 years. In examining 100 public companies with CEO departures, it found that brief tenures of five years or less showed an equal mix of positive and negative returns for shareholders. Around the seven year mark, positive returns began to outweigh negative ones; and between 10 and 15 years was "ideal."

Meanwhile, yet another study by Booz Allen Hamilton from back in 2002 found that CEOs who were in the job more than 10 years tended to have greater median annual rates of return than those with shorter tenures. Still, the Booz article also raised an important point: "There is a chicken-and-egg question here — does good performance lead to longevity, or does longevity lead to good performance?" Moreover, the consulting firm's study also showed that in the second half of the tenures of long-serving CEOs, returns for shareholders sank. Those who served more than 10 years had median annual returns of 5.9 percent in the first half of their tenure, compared to -0.9 percent in the second half.

So maybe there is a downside to sticking around too long. Yet another intriguing recent study backs that up. It looked at how CEOs' performance fared over time — not just with shareholders, but with employees and customers. It found that a far shorter tenure of just 4.8 years is actually optimal.

Here's why. CEOs' relationships with their employees, unsurprisingly, grow stronger as years go by. Yet their relationships with customers, after improving for several years, start to weaken over time, leading to worse performance. The researchers explained it this way in the Harvard Business Review last year: "As CEOs accumulate knowledge and become entrenched, they rely more on their internal networks for information, growing less attuned to market conditions. And, because they have more invested in the firm, they favor avoiding losses over pursuing gains. Their attachment to the status quo makes them less responsive to vacillating consumer preferences."

Of course, whether or not Jeff Immelt — or any CEO — has stayed in the job long enough is ultimately up to the company's board of directors. (In Immelt's case, the Journal reports that they remain strongly in his corner.) But if boards and CEOs get too cozy over the years, directors could be less likely to ask the hard questions of whether the leader is being too cautious or too internally focused to make the right decisions for shareholders.

Another recent study by the executive search firm Spencer Stuart found that companies that replace about one board member a year on average had market-beating performance.  If the average U.S. board has 10 or 11 members and that optimal level of board turnover actually happens, a CEO who sticks around for 10 years would have an entirely new set of people on the board than those who first elected him or her to the job — and this new group might be less hesitant to make a change. Perhaps the best question, then, isn't what's the right tenure for CEOs, but what's the right length of service for the people who make the decisions about their jobs.

Read also:

Boards don't know their next CEOs

Average pay for Washington-area CEOs more modest than national peers

Like On Leadership? Follow us on Facebook and Twitter.