We caught up with one of the authors, LBS professor Freek Vermeulen, who wrote the essay with his colleague Dan Cable, by phone while he was in Germany on a ski trip with his family. Our conversation with Vermeulen, which has been edited for length and clarity, is below.
So can you sum up your argument for those who haven’t seen the piece?
For most CEOs -- and actually most top senior executives -- their pay depends to a very large extent on some measure of performance. And I mean a very large extent in comparison to the rest of us. Most of us get a fixed salary and maybe a Christmas bonus. But for CEOs, it’s actually nothing unusual for 60, 70 or 80 percent of their remuneration to be dependent on performance. But we have quite a bit of research on the effects of that performance-based pay, and it isn’t very pretty. We know from research that it doesn’t have a very positive effect.
Before you get into the research, what's the core of what you’re saying?
We know from this research that [performance-based pay] isn’t that effective. Others have been saying we need to change the measures of performance. But based on what we know from research, we say no. Actually, what you should do is something radically different. What you should do is not pay them for performance at all, but give them a fixed salary.
So how do you back that up?
There's research that shows that for humans in general, performance-related pay often backfires. That seems slightly counterintuitive. But if a very large component of someone’s pay is dependent on performance -- and hence there’s a lot at stake -- people seem to freeze.
That’s not true in routine tasks, such as for people who work in factories. But when you ask people to do a very creative task, they fail to come up with creative solutions simply because there’s so much at stake. Of course the job of the CEO is not a routine task. It is about finding novel solutions, and innovations, and so on. Therefore, we say the job of a CEO is not suited to performance-related pay.
You also say that research has shown performance-based pay can actually backfire, and lead to cheating.
We have lots of studies on CEOs that does indeed show that [performance-related pay] leads to fraudulent behavior, or leads to overly risky behavior. There are even studies on companies whose CEOs' pay is very much based on performance, and we see more product recalls. The negative effects seem to permeate the whole organization, starting from the top down.
What kind of reaction have you gotten from the essay?
One response is about the size of CEO pay. It’s a very valid question -- do they get paid too much? -- but that’s not what we’re talking about here. People have been confusing that a bit, though I think we were very explicit about it.
The second type of response we’ve been getting is people saying the research you’re basing your argument on is not based on studies of CEOs. First of all, that’s not entirely true. A number of the studies based on unethical behavior or fraudulent behavior is done on CEOs. But much of the research we cite is indeed based on humans. I think that is part of the problem. People somehow seem to think that what applies to humans does not apply to CEOs.
Almost like they’re seen as a different breed?
CEOs are human. And actually that’s what we do see in this research. They do react to incentives. In fact I think we should treat CEOs like other humans and give them a fixed salary because their behavior is very human. If we give them wrong incentives, their behavior will also be wrong.
The notion of "pay for performance" has become almost biblical in the belief systems of most boards and human resources departments. Shareholder activists are hugely critical when a company has too much cash compensation. How would you begin to try to change their minds?
The most prominent reaction to the piece is "you’re throwing out the baby with the bathwater." Yes, how we currently pay CEOs distorts behavior--I haven’t read many people deny that. … But then they say you just have to give them better incentives and better measures. Usually then people say we have to tie their pay to long-term performance.
I just have to ask two questions, then. Please, define for me "long term?" And the second question: Please define performance? Yes, long-term performance is what the CEO is supposed to do. But how on earth can you measure this? And because CEOs are human, we know that any imperfect measure we’re going to use is going to distort their behavior. No matter what you do, you’re going to get it wrong.
Good CEOs really don’t need it. We have something called intrinsic motivation and it’s very powerful. Pay for performance is what we call extrinsic motivation. But pay for performance also destroys that intrinsic motivation.
Pay for performance is also no longer a concept reserved just for executives. Many people in professional or middle management office jobs get some portion of their pay in bonuses. Do you not think some small percentage of pay based on the variability of performance is a good idea?
People have been thinking 'you’re putting forward a very radical proposal, to make it completely fixed, to stir debate a bit.' Often that’s true, and I take an extreme point of view to argue one side of the debate and come out somewhere in the middle.
In this case I actually quite mean it in saying let’s make it 100 percent fixed. Even if only 30 percent of pay is flexible, you’re still going to get it wrong. You’re still going to distort behavior and your measure of performance is still going to be imperfect. It's still going to end up destroying part of your intrinsic motivation. So this time I actually quite mean it. Make all of it fixed.
So how do you measure or manage performance, or motivate CEOs? Can you really find enough CEOs who just innately want to do the job well for its own sake?
If a board member said that to me -- "perhaps he's not intrinsically motivated?" -- I'd say "get someone else." I really mean that. If you have someone who will only do a good job if there's a really large bonus, you really need someone else.
How then do you really measure a CEO’s performance? Every measure you're going to have is some approximation. Of course you’re going to have to try and measure things. Measures should be indicators, and they should help you make a judgment call of whether your CEO is doing the right thing or not. But the measure is not the real thing. That’s usually the problem. Once you have a measure, people start to optimize it.