In 1990, Harvard Business School professor Michael Jensen co-wrote an article making the then-bold claim that CEO compensation should be tied to stock price performance. The point, Jensen and his co-author argued, was to better align incentives and ensure that corporations were able to attract “the best and brightest individuals to careers in corporate management.”
In short: Pay up, or lose out.
At the time, Jensen was little known outside academic circles. But — perhaps unsurprisingly — his advocacy for a new model of CEO pay quickly made him a well-known name in business academia and corporate boardrooms, sought after for advice and affirmation. Soon, “aligning incentives” exploded across corporate America.
The results were staggering. According to data from the Stanford Graduate School of Business, average CEO compensation at the largest firms rose from $1.8 million per year in the 1980s — roughly in line with the previous 45 years — to $4.1 million in the 1990s. By the early 2000s, it had risen to $9.2 million. And those numbers are after adjusting for inflation. The majority of that growth came in the form of options and stock grants, just as Jensen had recommended.
But what if Jensen was wrong?
What if CEOs don’t play much of a role in driving stock price performance, and the “aligned incentives” of equity incentive pay don’t change behavior in any way that benefits shareholders?
What if the “best and brightest” — those executives with the most dazzling CVs and track records — don’t perform any better than less credentialed executives?
And what if Jensen’s philosophy produced better outcomes for CEOs and business school graduates — including those from his own school — but not better outcomes for investors or society at large?
Over the past year, we set out to answer these questions. We created a database of approximately 8,500 CEOs and their characteristics, each individually mapped to their respective companies for the duration of their tenure, and pulled company fundamentals from Compustat, stock returns from the University of Chicago's Center for Research in Security Prices (CRSP), CEO tenure and education from BoardEx, and long-form CEO biographies from Capital IQ. We then ran a battery of tests on the new data set, looking for correlation, persistence, and predictive power. We wanted to answer two sets of questions:
Do CEO characteristics predict stock price performance? Do CEOs with MBAs perform better than CEOs without MBAs? Do CEOs with MBAs from the best MBA programs outperform other CEOs? Do CEOs who worked at top consulting firms and investment banks outperform other CEOs? More broadly, are the “best and brightest” better at running companies?
Is CEO performance persistent? If someone was a successful CEO of one company and took over as CEO of a different company, does his or her performance at the first company predict performance at the second company? If a CEO does a good job for three years, does that predict stock price performance over the subsequent three years? More broadly, are some CEOs better than others at driving share price performance?
This research has important implications for investing. There is broad consensus among investors that one should seek out “well-managed” companies. And what better way to assess the quality of management than to examine the chief executive’s resume and record?
This approach makes intuitive sense. Surely it is better to invest in the star CEO who has a record of stunning returns than a schmuck who has underperformed the S&P. Better still if the star was forged in the crucible of Harvard Business School. There is a big market for books about these genius CEOs and how they achieved their success — and what lessons corporate executives and investors should take away from the histories of “great men.”
The siren songs of credentialism and tales of corporate "great men" are seductive. It is the pedagogy by which most college students learn and explain history. But if the data shows that CEO performance isn’t persistent, or if the resume characteristics we commonly associate with quality don’t, in fact, predict performance, are investors making a mistake in spending so much time on management quality?
Do MBAs Make Better CEOs?
In the 1980s, Jensen noticed a big shift in the career choices of Harvard MBAs. In the late 1970s, about 55 percent of graduates chose careers in corporate management, but by the late 1980s, only 30 percent were making this choice.
Jensen was concerned that this meant America’s “best and brightest” leaders were not going to be running America’s largest companies — and that corporate America needed to increase CEO compensation to lure more Harvard MBAs into corporate management careers.
A central premise of business education is that leadership and management can be taught in the classroom. Harvard Business School says its mission is “to educate leaders who make a difference in the world,” where a difference is defined as creating “real value for society.” And so, Jensen’s logic makes sense: Harvard attracts the very best students and, presumably, is good at educating them to be better business leaders, so corporate America should want more Harvard graduates running companies — and this logic should extend to MBA programs beyond just Harvard.
But regression results suggest a different result entirely. We tagged CEOs by the MBA programs they attended, formed monthly portfolios of companies broken down by the business school each CEO attended, and compared the returns of these portfolios to the broader market.
We found no statistically significant alphas — despite testing every possible school with a reasonable sample size. MBA programs simply do not produce CEOs who are better at running companies, if performance is measured by stock price return.
We ran similar regressions controlling for industry and found that — even after controlling for industry — elite MBAs did not produce positive statistically significant alpha. Elite MBAs did perform relatively well as CEOs in healthcare and consumer staples, but relatively poorly in energy and materials businesses, though those results were not statistically significant. Our study is not the only one to come to this conclusion. A study by economists at the University of Hawaii
asked similar questions and found that firm performance is not predicted by the educational background of the CEOs.
The perceived quality of each institution appeared to have no correlation with stock price returns. Northwestern led with an alpha of 0.58 percent per month. Stanford eked out a barely positive alpha of 0.03 percent per month. Harvard and Wharton had negative alphas of -0.15 percent and -0.19 percent, respectively, per month. While these rankings likely occurred by sheer chance, they do nothing to support Jensen’s thesis.
Lastly, we looked at how CEOs who had previously worked at investment banks and elite consulting firms performed. If Jensen’s core thesis were true, we would expect CEOs with these elite credentials to outperform the market.
We thus formed monthly portfolios for bankers and consultants. As we did with MBAs, we then ran industry-controlled Fama-French three-factor regressions. The result: Neither bankers nor consultants produced statistically significant alphas. We also back-tested portfolios designed to favor ex-bankers and consultants and found no significant edge (though consultants had a statistically insignificant edge on bankers).
This suggests that the “best and brightest” do not have a statistically significant edge when it comes to managing public companies. An elite pedigree — the type of pedigree favored by headhunters and corporate boards — is not predictive of superior management. One of the central rationales for Jensen’s campaign (increasing CEO pay by tying it to share price performance) appears, in retrospect, to have little empirical support. These credentials, however, are significantly overrepresented in the CEO biography database. The elite credentials thus benefit the individual, but there is little evidence that these credentials benefit shareholders.
It’s unclear precisely why the evidence suggests that highly credentialed CEOs from our most elite MBA programs and their funnel careers, like banking and consulting, appear to add no measurable value to shareholders. However, we found wisdom in a saying of the oldest living CEO, a 100-year-old billionaire from Singapore who still goes to work every day to mentor his son in leading the firm. His son, Teo Siong Seng, said, “My father taught me one thing: In Chinese, it’s ‘yi de fu ren’ — that means you want people to obey you not because of your authority, not because of your power, or because you are fierce, but more because of your integrity, your quality, that people actually respect you and listen to you.”
Bloomberg shows that “there is no education data available” for the 100-year-old CEO, Chang Yun Chung, so we cannot vet his educational credentials — but we suspect he did not obtain an MBA.
Is CEO Performance Persistent?
CEOs’ educational credentials might not predict success, but do their track records?
Gregg Lowe is a star CEO when measured by his consistency at generating shareholder value. He took over at Freescale Semiconductor in June 2012. Three years later, he managed the sale of Freescale to NXP Semiconductors for nearly four times the share price when he took over. In September 2017, he took over as CEO of Cree. The stock is up almost 60 percent since he became CEO, versus a flat S&P 500 over the same period.
But for every Gregg Lowe, there is an example on the other side.
Brian Woolf led Cache from 2000 to 2008, earning investors a 5.8 times return on their investment over those eight years. He was hired five years later to be CEO of Body Central. The press release announcing his hiring cited his excellent leadership at Cache. But by January 2015, Body Central had closed every single one of its stores and investors had lost almost all their capital.
These two opposing examples highlight key questions about corporate management: Is CEO performance persistent? Are some people simply better managers, able to consistently generate high returns for investors? And can we identify these people based on an examination of their track records?
We tested these questions empirically. We looked first at performance persistence within companies: whether a CEO’s early track record predicts later performance. We filtered our database for CEOs who had tenures longer than six years at one company. We then separated each CEO’s return record into two buckets: Years 1–3 and Years 4–6. To eliminate the confounding effect of different market conditions, we adjusted returns by the performance of the S&P 500 index. Within each of the two time-horizon buckets, we sorted the CEOs’ market-adjusted returns into quartiles. We then identified the CEOs who were above the median twice and those who were in the top quartile twice.
To interpret the results, it’s important to consider what we would expect to see by random chance. Similar to a coin flip, 50 percent of CEOs should be above-average performers, and 50 percent of CEOs should be below-average over each three-year time horizon. And just as a coin has a 25 percent chance of landing on heads twice in a row (50 percent x 50 percent), we should expect 25 percent of CEOs to be above average in both Years 1–3 and Years 4–6, purely by chance. But if we were to believe that CEO performance is always persistent, we would expect to see 50 percent of CEOs perform above average in the two successive three-year periods (50 percent x 100 percent).
A similar calculation would apply to the top-quartile CEOs. If every CEO randomly has a 25 percent chance of being in the top quartile over a three-year period, then we should expect about 6 percent of CEOs to be in the top quartile in two successive three-year periods (25 percent x 25 percent) due to luck alone. If, however, there is full performance persistence, then 25 percent of CEOs would be in the top quartile in Years 1–3 and Years 4–6.
The actual CEO performance results line up very closely with what we would expect to see by chance. Twenty-five percent of the 2,420 CEOs in our database had above-average performance in two successive three-year periods, and 7 percent of CEOs had top-quartile performance in two successive three-year periods.
The above figures are clear: There is almost no persistence in CEO performance. The observed number of CEOs in each category is indistinguishable from what we would expect if the process were entirely random. These results held when controlling for industry and the Fama-French factors.
Visualizations of the underlying return data tell the same story. Figure 3 below compares Years 1–3 returns to Years 4–6 returns. If the performance between these two periods were perfectly correlated, the data would form a 45-degree line.
This chart reveals no discernable relationship for CEOs in general. Historical performance does not appear to predict future performance, at least as measured by share price returns. The fourth-quartile results look very similar to the first-quartile results, implying that negative performance is no more persistent than positive performance.
This is not an intuitive finding — and academic studies suggest that this is not how boards think, particularly when it comes to firing bad performers. A 2015 study
found that CEOs are often fired after bad firm performance caused by factors beyond their control, a finding in conflict with the standard economic theory of rational expectations. Boards are far more likely to fire CEOs when the industry is having trouble broadly, attributing to a person what is in fact an exogenous economic shock.
We then looked at CEOs who have run multiple companies to see if their performance at the first company predicted outcomes at the second. Headhunters and corporate boards often look for CEOs with a track record of creating value at another company when choosing whom to hire. But if past performance doesn’t predict future results, then they might be looking at an irrelevant variable.
To investigate this hypothesis, we filtered our database for CEOs who have headed multiple companies. We then tagged each CEO-company pair with whether it was the first, second, third, or fourth company the CEO has headed. Because very few CEOs have headed more than two companies, we separated the data into two buckets: Company 1 and Company 2 (ignoring observations beyond the first two companies). To eliminate the confounding effect of different market conditions, we adjusted returns by the performance of the S&P 500 index. Within the two company buckets, we sorted the CEOs’ market-adjusted returns into quartiles. We then identified the CEOs who were above the median twice and those who were in the top quartile twice.
As before, if outcomes were completely random, we would expect 25 percent of CEOs to be above-average performers across two companies (50 percent x 50 percent). Similarly, we would expect 6 percent of CEOs to be top-quartile performers across two companies (25 percent x 25 percent) by sheer luck. The actual CEO performance results are below.
Once again, there is little to no persistence in CEO performance from one company to the next. Again, these results held when controlling for industry and the Fama-French factors. The proportion of CEOs who continue to perform well is in line with what we would expect under random conditions.
The media is filled with depictions of visionary CEOs who have a record of generating extraordinary returns. An inordinate amount of journalistic effort is directed at dissecting their lives. What was his childhood like? What is his morning routine? What are his management principles? Does he use PowerPoint? The huddled masses clamor for these details in hopes of grabbing a piece of the star’s genius for themselves.
The cult of the CEO is difficult to resist. Management, after all, is a team sport. Just as a quarterback can control the team's offense, the theory goes, so too can CEOs control their large public companies. If enterprise value has soared, it is because the CEO is a genius visionary. If multiples have compressed, it is because the CEO is an arrogant fool. And since the CEO is the key determinant of the company’s future, virtually any level of CEO compensation is justifiable.
But beneath the mountain of CEO profiles are base rates that are virtually indistinguishable from randomness. The focus on the “great man” theory of corporate management may lead to persistent errors. For investors favoring stocks with strong past-performing CEOs, the base rates suggest this is like betting on heads because the last two coin flips came up that way. If they pay up for this “quality,” it’s worse than that.
A Harvard Business Review article
, “The Art and Science of Finding the Right CEO,” lists “proven track record” as a top, “obvious” criterion for selecting CEOs. But, to quote Sherlock Holmes, “There is nothing more deceptive than an obvious fact.”
Journalists, investors, and boards are placing excessive emphasis on CEO pedigrees and track records. In a world that is feedback-rich, stochastic, and “fat tailed,” the simple narrative of the “great man” does not appear to have much quantitative merit — rather, it seems like yet another cognitive bias in the vein of those discovered by Daniel Kahneman.
Of course, we cannot prove that CEO credentials don’t have an effect on share price. It’s impossible to prove a negative — what statisticians call a null hypothesis. We are simply pointing out that there is no convincing evidence in favor of rejecting that null hypothesis. U.S. companies adopted Jensen’s ideas without any data suggesting that incentive pay would actually result in better stock price performance — and no evidence to suggest Jensen’s thesis was correct has emerged in the 29 years since that great experiment began.
An effective counterargument might be that share price return is not a good metric for CEO performance, that stock price is simply not within the control of the CEO, being driven to a large extent by factors like changes in investor sentiment and macroeconomic conditions.
This thesis would, however, imply that incentive compensation tied to stock price is ineffective. That is a thesis — unlike the elitist premise that pedigree predicts performance — that is supported by data. An S&P Global Report
found no link between CEO pay and stock price performance. “Despite wide acceptance of executive pay-for-performance, we find no evidence that high levels of total incentive compensation (performance-based cash plus stock and stock option awards) result in higher-than-average shareholder returns,” the authors wrote.
Abraham Lincoln famously said, “I claim not to have controlled events, but confess plainly that events have controlled me." Very few of our CEOs are willing to make a similar confession about the share prices of the companies they run.
But if there is no evidence that stock returns are attributable to CEOs, then what justification is there for their stratospheric pay? How much longer will investors and boards be fooled by randomness and hollow credentialism?