In light of the financial crisis of 2008 and the subsequent outrage over the continuing bonuses paid to many of those deemed responsible for it, many people wonder how incentives really affect CEOs and Wall Street executives. Corporate boards generally assume that very large performance- based bonuses will motivate CEOs to invest more effort in their jobs and that the increased effort will result in higher- quality output. But is this really the case?
Through experiments we conducted, all of which can be found in much more depth in my new book, The Upside of Irrationality, we discovered that using money to motivate people can be a double-edged sword. For tasks that require cognitive ability, low to moderate performance-based incentives can help. But when the incentive level is very high, it can command too much attention and thereby distract the person's mind with thoughts about the reward. This can create stress and ultimately reduce the level of performance.
A few years ago, before the financial crisis, I was invited to give a talk to a select group of bankers. The meeting took place in a well-appointed conference room at a large investment company's office in New York City. The food and wine were delicious and the views from the windows spectacular. I told the audience about different projects I was working on, including the experiments on high bonuses in India and MIT. They all nodded their heads in agreement with the theory that high bonuses might backfire-until I suggested that the same psychological effects might also apply to the people in the room. They were clearly offended by the suggestion. The idea that their bonuses could negatively influence their work performance was preposterous, they claimed.
I tried another approach and asked for a volunteer from the audience to describe how the work atmosphere at his firm changes at the end of the year. "During November and December," the fellow said, "very little work gets done. People mostly think about their bonuses and about what they will be able to afford." In response, I asked the audience to try on the idea that the focus on their upcoming bonuses might have a negative effect on their performance, but they refused to see my point. Maybe it was the alcohol, but I suspect that those folks simply didn't want to acknowledge the possibility that their bonuses were vastly oversized. (As the prolific author and journalist Upton Sinclair once noted, "It is difficult to get a man to understand something when his salary depends upon his not understanding it.")
Somewhat unsurprisingly, when presented with the results of these experiments, the bankers also maintained that they were, apparently, super special individuals; unlike most people, they insisted, they work better under stress. It didn't seem to me that they were really so different from other people, but I conceded that perhaps they were right. I invited them to come to the lab so that we could run an experiment to find out for sure. But, given how busy bankers are and the size of their paychecks, it was impossible to tempt them to take part in our experiments or to offer them a bonus that would have been large enough to be meaningful for them.
Without the ability to test bankers, Racheli Barkan (a professor at Ben-Gurion University in Israel) and I looked for another source of data that could help us understand how highly paid, highly specialized professionals perform under great pressure. I know nothing about basketball, but Racheli is an expert, and she suggested that we look at clutch players-the basketball heroes who sink a basket just as the buzzer sounds.
Clutch players are paid much more than other players, and are presumed to perform especially brilliantly during the last few minutes or seconds of a game, when stress and pressure are highest.
With the help of Duke University men's basketball Coach Mike Krzyzewski ("Coach K"), we got a group of professional coaches to identify clutch players in the NBA (the coaches agreed, to a large extent, about who is and who is not a clutch player). Next, we watched videos of the twenty most crucial games for each clutch player in an entire NBA season (by most crucial, we meant that the score difference at the end of the game did not exceed three points). For each of those games, we measured how many points the clutch players had shot in the last five minutes of the first half of each game, when pressure was relatively low. Then we compared that number to the number of points scored during the last five minutes of the game, when the outcome was hanging by a thread and stress was at its peak. We also noted the same measures for all the other "non-clutch" players who were playing in the same games.
We found that the non-clutch players scored more or less the same in the low-stress and high-stress moments, whereas there was actually a substantial improvement for clutch players during the last five minutes of the games. So far it looked good for the clutch players and, by analogy, the bankers, as it seemed that some highly qualified people could, in fact, perform better under pressure.
But -- and I'm sure you expected a "but" -- there are two ways to gain more points in the last five minutes of the game. An NBA clutch player can either improve his percentage success (which would indicate a sharpening of performance) or shoot more often with the same percentage (which suggests no improvement in skill but rather a change in the number of attempts). So we looked separately at whether the clutch players actually shot better or just more often. As it turned out, the clutch players did not improve their skill; they just tried many more times. Their field goal percentage did not increase in the last five minutes (meaning that their shots were no more accurate); neither was it the case that non- clutch players got worse.
At this point you probably think that clutch players are guarded more heavily during the end of the game and this is why they don't show the expected increase in performance. To see if this were indeed the case, we counted how many times they were fouled and also looked at their free throws. We found the same pattern: the heavily guarded clutch players were fouled more and got to shoot from the free-throw line more frequently, but their scoring percentage was unchanged. Certainly, clutch players are very good players, but our analysis showed that, contrary to common belief, their performance doesn't improve in the last, most important part of the game.
Obviously, NBA players are not bankers. The NBA is much more selective than the financial industry; very few people are sufficiently skilled to play professional basketball, while many, many people work as professional bankers. As we've seen, it's also easier to get positive returns from high incentives when we're talking about physical rather than cognitive skills. NBA players use both, but playing basketball is more of a physical than a mental activity (at least relative to banking). So it would be far more challenging for the bankers to demonstrate "clutch" abilities when the task is less physical and demands more gray matter. Also, since the basketball players don't actually improve under pressure, it's even more unlikely that bankers would be able to perform to a higher degree when they are under the gun.
Could all this mean that sometimes we might actually behave less rationally when we try harder? If that's so, what is the correct way to pay people without overstressing them? One simple solution is to keep bonuses low-something those bankers I met with might not appreciate. Another approach might be to pay employees on a straight salary basis. Though it would eliminate the consequences of over- motivation, it would also eradicate some of the benefits of performance-based payment. A better approach might be to keep the motivating element of performance-based payment but eliminate some of the nonproductive stress it creates. To achieve this, we could, for example, offer employees smaller and more frequent bonuses. Another approach might be to offer employees a performance-based payment that is averaged over time-say, the previous five years, rather than only the last year. This way, employees in their fifth year would know 80 percent of their bonus in advance (based on the previous four years), and the immediate effect of the present year's performance would matter less.
Whatever approach we take to optimize performance, it should be clear that we need a better understanding of the links between compensation, motivation, stress, and performance. And we need to take our peculiarities and irrationalities into account.