CEO pay has increased 725 percent over three decades, while worker pay has essentially remained flat. A new study challenges a conventional practice corporations use to justify skyrocketing CEO pay, which is that without it, CEOs would leave for competitors. According to the study by the University of Delaware’s Charles M. Elson and Craig K. Ferrere:

It is increasingly apparent that the pay awarded to chief executives is becoming profoundly detached from not just the pay of the average worker, but also from the companies they run. Offsetting the external focus, which is so heavily relied upon today, with internal metrics and internal benchmarking may help to curb the persistent escalation. We hope that if directors are no longer constrained by notions of “competitive” pay, which are driven by the false belief that CEOs are interchangeable, they may have the space to rationalize the upward spiraling pay ratchet and deliver what is more shareholder acceptable compensation.

Company boards rely on a practice where they use loosely defined “peer groups” of supposedly similar companies to set the CEO’s compensation. In reality, few CEOs leave one company for another: Of 1,800 CEO successions between 1993-2005, less than 2 percent had held the position at a competing firm. Their skills, highly specific to the company, are not easily transferrable.

Another issue is the “peer groups” companies use is so loosely defined that it includes firms that are much larger or aren’t in the same industry, much less rivals. In other words, the CEO of IBM is unlikely to jump to AT&T, Ford or Pfizer, even though those companies’ CEOs are included in IBM’s peer group.

A recent example may include Best Buy, which offered its new CEO a three-year compensation package of $32 million, after laying off 2,400 employees this summer. A company spokes woman [defended CEO Hubert Joly’s pay as “in-line with best practice for Fortune 50 companies,” and “is squarely in the mid-range for a CEO of a company the size of Best Buy.”

“It’s a false paradox,,” study co-author Elson told the New York Times. “The peer group is based on the theory of transferability of talent. But we found that C.E.O. skills are very firm-specific. C.E.O.’s don’t move very often, but when they do, they’re flops.”