Discuss as:

Lavish CEO pay doesn't work as intended: study

The arms race in CEO pay doesn’t help performance or retention, according to a new study.

Outsized CEO pay has been a bone of contention for shareholders and a flashpoint for public discontent, but was always justified with the conventional wisdom that if a company doesn’t pay its top brass top dollar, they’ll go somewhere where they can earn more.

Charles Elson and Craig Ferrere, director and fellow, respectively, at the John L. Weinberg Center for Corporate Governance at the University of Delaware, took aim at this long-held theory in a new paper, as reported by the New York Times last weekend.

"The problem is the standard practice of benchmarking pay to that of peers," the two wrote.

To determine how much to pay a CEO, corporate compensation committees look at how much the chiefs of similar companies earn, which has the result of lumping together all CEO talent into one pool. Elson and Ferrere argued that expertise in management isn’t the same, and isn't as good, as having a deep base of knowledge in one particular industry. 

Many of the skills that make a chief executive successful don’t translate to another company. “The theoretical underpinning of [peer grouping] became the notion of transferability,” Elson said. “That was false. The superstar theory of the CEO failed, and if transferability failed, the peer system has to fail.”

“There’s been a sense for some time that the external view of pay-setting has been a problem in ratcheting up pay," said Paul Hodgson, chief research analyst at corporate governance research company GMI Ratings. 

Elson and Ferrere said benchmarking against what other CEOs make is flawed in several ways that inflate pay packages.

One problem is that the definition of what constitutes a “similar” company can be manipulated to skew pay higher by including larger companies or ones in different industries. Institutional shareholders are “very, very suspicious” of that practice, Hodgson said.

Even without manipulation, Elson said companies should decide how much to pay a CEO based on performance, not how much his or her counterparts earn. “If you don’t have that internal benchmark you’ve made a mistake to begin with,” he said.

Two other elements of peer benchmarking nudge up pay, the researchers wrote. Compensation committees are reluctant to give CEOs any less than the median pay for the peer group. Compensation packages generally range from the 50th to the 90th percentile, which inevitably inflates pay over time. Another unintended consequence of offering above-average pay is that successful CEOs wind up subsidizing their less-talented counterparts, since performance isn’t factored into the equation. 

The researchers drew a link between higher CEO compensation and the trend of chief executives being recruited from outside a company, but said companies perform better over the long term when someone familiar with the industry is at the helm.

“The necessary skills to successfully run a company cannot be acquired besides through actual experience at the company,” they wrote. “The years spent acquiring 'general' experience rather than a firm-­specific expertise were unproductive.”

Companies would do better to look within their own ranks than to cast a wider net, they said, since even a CEO who displays talent running one company usually can’t export those skills to a new corporation. They pointed to earlier research showing that, although CEOs brought in from outside often earned more, the companies they led subsequently performed on par with or worse than those that promoted from inside.

“Each organization has its own culture and if you’re in the culture... you’re often more effective than someone who isn’t,” Elson said.

"It’s always more expensive," Hodgson said. "Very concrete examples show that hiring from the outside often doesn’t work.”

More money and business news:

Follow NBCNews.com business on Twitter and Facebook