Bad CEOs must be overpaid, by definition, because they should be fired. And many are. It is an increasingly risky job, thanks to the Sarbanes-Oxley law, and risk is rewarded. Yet it becomes clear that some outstanding CEOs have been underpaid when they are recruited by better offers from other firms, sometimes in other countries.
A great CEO can do miracles, often turning around a stagnant or failing firm. I bought stock in Apple for about $15 a share. Whatever they paid Steve Jobs is fine with me. If you don't own stock in a company, then what they pay anyone isn't any of your business.
If it means anything at all to claim all CEOs are overpaid, it must mean the person making such a claim believes the best CEOs would still eagerly tackle the toughest chores even if they were paid much less. That is what economists mean by the word "rent."
In a chapter I wrote about executive compensation in William Niskanen's fine book, "After Enron," I found no evidence that rent is common in the market for executive talent. There were unplanned windfalls from stock options in the late '90s, but that was simply because nobody could possibly guess how high tech stocks would fly in the early '90s when those options were handed out.
Eisinger thinks: "The key question is this: Are executives being paid what the market will bear, or is the market broken? The answer lies in whether rich compensation leads to good performance." But that, in turn, depends entirely on how you define both compensation and performance.
The Wall Street Journal article relies on Harvard law professor Lucian Bebchuk -- "a critic of the disconnect between pay and performance" -- and Yaniv Grinstein of Cornell: "In the period from 2001 to 2003, top-executive compensation (for the top five executives) amounted to 9.8 percent of the companies' net income, almost double the 5 percent in 1993 to 1995." That might be interesting if stockholders measured performance by FASB earnings rather than by stock prices and dividends.
The Bebchuk-Grinstein paper uses means averages for compensation, adding up all types of pay for executives at S&P 500 companies and dividing by 500. That sort of average cannot be meaningfully compared with a similar average of earnings for many firms because "average" compensation could easily have be exaggerated by a few huge payouts for executives whose companies' stock did unusually well (which is what pay for performance means).
Bebchuck and Grinstein estimate that a mean average of CEO compensation for S&P 500 companies rose from $9.1 million in 1997 to $17.4 million in 2000, then fell back to $9.1 million in 2003. Since the stock market is the relevant measure of performance, pay and performance clearly did move up and down together. That should be the end of this story. But there is more.
The chosen measure of "compensation in any given year" includes a bonus for the previous year and the estimated future value of stock options granted during any given year. Such a melange of past and future pay cannot be intelligently compared with company earnings in "any given year," even if short-term earnings rather than future growth was really what mattered most to stockholders.
"Another way to gauge the impact of executive pay on shareholder value," suggests Eisinger, "is simply to measure how much stock has been given to employees. In 2004, the employees of the median company in the top 200 corporations held 14 percent of the companies' shares, up from 9.9 percent in 1994."
That's just wrong. Many companies famously granted stock options to mid-level employees in the '90s and offered beneficial stock purchase plans. Employee ownership tells us nothing about the top executives' share of the shares. Moreover, many prominent executives own many shares because they or their family members founded the company, not because they were given options or stock. Other executives purchased shares in their companies, sometimes as a requirement of employment.
The evidence intended to prove that executive pay became disconnected from performance in recent years turns out to show the opposite. Irrefragably biased assertions are easily refragable if we ignore the opinions and contest the facts.