A Congressional Budget Office report released in October reports that the majority of wealth created between 1979 and 2007 was captured by the top 1 percent of all earners. Their income grew by 275 percent, with far more modest gains of 18 to 65 percent garnered by all other households. The report comes at a time when Occupy Wall Street demonstrators have positioned themselves as representatives of the “99 percent” — that segment that has not shared in the largess enjoyed by the top 1 percent.
One proposed explanation is that the high salaries of star athletes and entertainers skew these figures. But a recent analysis of tax data by economists at Williams College shows that of the top 1 percent, close to 60 percent are managers or executives, not athletes or superstars. “That is strong evidence that much of the inequality over that last few decades really is driven by top executives at these firms,” Professor Jerry Kim says.
“Salaries rose much faster than the market value of companies in the past decade,” Professor Bruce Kogut says. “The puzzle, then, is why has pay for performance rewarded executives so much more handsomely in the last decade than ever before?” Kim and Kogut, working with Jae-Suk Yang, research fellow at the Sanford C. Bernstein & Co. Center for Leadership and Ethics, address this question precisely, inspired by a recent study suggesting that friends of obese people have a significantly greater likelihood of themselves becoming obese than people with fewer obese friends. Just as in studies on obesity that measure whether people weigh more or less than a body mass index, the researchers calculated high and low pay by noting if CEOs were paid more or less than a long-term trend line that relates pay to market value. “If obesity could spread through a social network then one could ask: does high pay — or in our term, being fat cats — spread through networks of CEOs?” Kim asks. “More specifically,” explained Kim, “could friendship or other ties at high-paying firms serve as a channel for high compensation?”
The researchers anchored the baseline pay on firms’ market values, labeling those CEOs paid above what is predicted by the market value of the firm as overpaid and those paid below predicted market value as underpaid. They then identified which factors increased the likelihood of a CEO going from an underpaid state to an overpaid state.
The researchers noted that the Internet boom, when pay skyrocketed because options became much more valuable, created far larger compensation differences among CEOs than before, unleashing a contagion over the next decade. The boom inflated expectations and norms about pay, which spread through social networks. Firms outside the dot.com realm started benchmarking pay through their connections to the dot.com, so executive pay rose, and benchmarking continued — and continues — to drive pay up and up. (The researchers’ simulation suggests there are some constraining forces that keep pay from spiraling completely out of control, as when CEOs get fired or move to other jobs.)
To reach this conclusion, Kim, Kogut, and Yang investigated three kinds of social connections that could influence pay:
Board interlocks: When firms share board members, information passes between those boards. “A member sitting on two or three boards sees one firm hand out raises; it is natural for her to convey to the other firms that raises represent fair pay,” Kim explains. Would highly interlocked boards reveal patterns of spreading of high pay?
Compensation peer groups: Firms have a comparison set of firms they view as similar to them that serve as a reference for CEO pay. If one bank suddenly gave its CEO a pay raise, would that influence other banks to raise the pay of their CEOs?
Education networks: Alumni of the same school may think of themselves as similar to each other. If one sees a fellow alumnus receive a large pay increase, could that change what she believes her value to be?
Using data from Compustat’s ExecuComp database to analyze compensation since 1992 and simulate future compensation patterns, the researchers found evidence that peer connections matter most: when a peer moved from an underpaid state to an overpaid state, the likelihood of their connections moving from an underpaid to an overpaid state also increased. Education connections were the second strongest, while the much vaunted power of board interlocks were negligible.
Taken together, these findings challenge the idea that CEOs’ outsize pay is primarily a function of outsize talent and suggest that growing income inequality even at the very top of the distribution deserves attention as a driver of the increased share of income going to CEOs. “It’s very difficult to determine the value of any single individual’s contributions to firm performance. Industry-wide and firm-level norms for fair pay are driven by looking around and saying, if so-and-so person is worth X then I’m worth Y,” Kogut says. “CEO pay appears to be determined through norms and social comparisons rather than just market-determined.”
Jerry Kim is assistant professor of management at Columbia Business School.
Bruce Kogut is the Sanford C. Bernstein & Co. Professor of Leadership and Ethics in the Management Division and director of the Sanford C. Bernstein & Co. Center for Leadership and Ethics at Columbia Business School.
Read the Research
"Executive Compensation, Fat Cats and Best Athletes"