Monday, 27 November 2017

CEO pay is not all about CEOs' performance, or company performance

While I was away overseas CEO pay was back in the news, mostly courtesy of the continuing fallout from Theo Spiering's $8.32 million salary-plus-benefits package announced back in September. In the New Zealand Herald, Helen Roberts (University of Otago) argues for greater transparency in CEO pay:
We are continually told seven figure sums are needed to retain top executives, without any substance or proof that it needs to be that high.
The reality is that it is the independent third-party remuneration advisers who set the expectations. Compensation consultants use median pay levels from the previous year to determine the median pay level for the current round of contracts; as pay levels increase the median pay level also goes up, driving all CEO pay levels up in that industry.
So the decisions are effectively being made based on the recommendations of only a few.
This becomes a never-ending cycle of artificially inflated salary packages, irrespective of company performance or any parity with pay for salaried workers- companies are effectively being held to ransom.
She then goes on to talk about how loosely CEO pay is related to actual company performance (read the whole article, it's interesting). However, there is a key point about CEO pay that is missed from Roberts's discussion, and also from arguments in favour of high CEO pay, such as this earlier article by Jim Rose, who focused more on superstar effects and essentially argues that if CEOs weren't earning their large salaries, they wouldn't keep their jobs.

That missing point is that the market for executives is a tournament (which I have written about earlier, also in the context of CEO pay). In tournaments the winner is not only paid for their own performance, but paid a high bonus as an incentive for those lower down (e.g. the next tier of executives, in the case of CEO pay) to work harder.

Tournament effects were first described by Sherwin Rosen and Ed Lazear in the early 1980s. In labour markets where there are significant tournament effects at play, workers are paid a 'prize' for their relative performance - maybe a raise or a promotion. The tournament 'winner' only needs to be a little bit better than the second best worker in order to 'win' the tournament, and claim the prize.

However, if winning the tournament is mostly about luck rather than good performance, then the prize needs to be very large in order to incentivise the workers to work hard to 'win' (otherwise, if the prize is small, why work hard if winning comes mostly down to luck?). The large role of luck in performance could be argued to be true of top executives (the tier below CEOs), where their performance can only be measured by metrics that they probably have only small positive influence over (and are more driven by economy-wide factors, especially in the case of large companies). [*] So, because companies want to incentivise their (non-CEO) top executives to work hard, ensuring that the CEO pay is a large step up is one way to do so. [**]

So, the focus on the lack of clear relationship between CEO pay and company performance, and calls for increasing transparency of CEO pay setting, are at least a little misplaced. Unless we first disentangle the incentive effects that are directed at other top executives.


[*] I say positive influence here, because I'm sure that a really bad executive can have considerable negative influence on a company's performance, but it isn't at all clear to me that for a broad range of competent executives, there is much to choose between them.

[**] I do wonder how vulnerable this theory is to the extent of internal vs. external appointments as CEO, since it seems to rely on internal appointments being the norm. On the other hand, the threat of external appointments could increase the incentive effects for internal top executives, since they would have to compete on performance with potential hires from outside the company.

Read more:


  1. Good summary. Tournament point much harder to make than the share price effect of CEO accidentally deaths

    1. Harder to make, and harder to measure. And I suspect that no company would want to try to defend high CEO pay by actually saying "we're using it to incentivise the executive VPs to work harder".