C-suite pay is often linked to performance measures. But is it linked in the right way? By Alex Edmans
How executive pay is calculated attracts great scrutiny nowadays. Are CEO pay packets well-earned or not? Alex Edmans, Professor of Finance at London Business School, has researched the subject intensively, with a special focus on the measures used to calculate executive compensation and the incentives they create.
It might seem that the simplest and most effective mechanism for aligning a CEO’s interests with those of the company’s owners is through stock-based pay – paying the CEO entirely with shares so that they become a co-owner of the company. This makes the CEO accountable for the long-term stock price. But a lively debate has arisen as to whether that’s enough, or whether the CEO should be evaluated on additional performance measures (known as “signals”).
While the stock price measures the value of the firm, the CEO should be rewarded for their value added to the firm. These concepts are different for two reasons. First, the value of the firm might be affected by factors that have nothing to do with the CEO’s value added. For example, the former boss of UK homebuilder Persimmon was paid £110 million, even though Persimmon’s good performance was because low interest rates had boosted the housing market rather than his strong performance. Second, there may be relevant measures of the CEO’s value added over and above the stock price, such as profitability or sales growth.
The use of these additional signals is justified in a 1979 article by Nobel Laureate Bengt Holmström, which introduces the “informativeness principle”. This states that a signal should be included in a contract if it provides information about the manager’s effort above the information already contained in the stock price.
This principle has been influential in practice – CEOs are paid on more than just the stock price. Indeed, these signals include not only other measures of financial performance than the stock price, but also non-financial measures. A recent joint study by Tom Gosling and Clare Hayes Guymer of the LBS Centre for Corporate Governance and PwC shows that 45% of FTSE 100 companies include sustainability measures such as carbon emissions in their executive pay plans.
However, the informativeness principle only states that other signals should be used, not how they should be used. In practice, 60% of large US companies pay their CEOs via ‘performance-vesting equity’, where positive performance signals lead to the CEO receiving more shares. But it is not clear that this is the best way to use these signals.
Professor Edmans, with Pierre Chaigneau of Queen’s University and Daniel Gottlieb of the London School of Economics, thus set out to investigate how boards should use additional signals. Which dimensions of a contract should one or more extra signals affect? Should they affect the number of shares received, or something else? And how should performance affect the contract – should good performance automatically mean more shares?
The researchers start by studying what the contract should look like when the stock price is the only available measure of performance. Counterintuitively, it turns out that it is not to give the CEO shares. The problem with shares is that they reward CEOs for failure – the shares are still worth something even if they underperform.
A more powerful device is stock options, which are worth zero if the stock price falls below a threshold, known as the strike price. The more the stock price exceeds the threshold, the more the option is worth.
The option contract has two important features: the threshold, and the number of options received if this is met. The more options that “vest” (that the CEO receives), the more their pay rises when the stock price exceeds the threshold (the greater the sensitivity of pay to the stock price). Thus, when additional measures of performance are available, they can either affect the threshold or the number of vesting options.
Which should it be? The researchers show that it depends on the information that the signal provides. Some signals directly inform performance; for example, high profits or low carbon emissions include good performance. The authors call this the “individual effect”, because high profits or low carbon emissions are informative in isolation. You automatically know they’re good news about the CEO’s effort and don’t need to look at the stock price to see this.
"The CEO should only be paid anything at all if the stock price is sufficiently high. If it’s low, then they have failed, and we don’t want to reward failure"
But there’s a second, more complex effect of the signal, which the researchers call the “inference effect”. This affects what the board infers about the CEO’s performance from the stock price. In contrast to the “individual effect”, this involves studying the signal in conjunction with the stock price, rather than in isolation.
The inference effect can be broken down into two parts. The first is the “location effect”. Consider a signal of the number of competitors in a firm’s industry. The more competitors in the industry, the lower the company’s stock price will be. If there are few competitors, the stock price will range from 5 to 10, so a stock price of 6 would be considered poor. But if there are many competitors, the stock price will range from 2 to 7, so a stock price of 6 actually indicates a job well done. The number of competitors affects the location of the range of possible stock prices (either 5 to 10 or 2 to 7); hence its name. This in turn affects what the board infers from seeing a stock price of 6; whether it indicates good or bad performance.
Interestingly, the location effect may counteract the individual effect. As we’ve just seen, more competitors should lead to higher pay, because it means that a given stock price is better news about performance (the location effect). But more competitors is an individually negative signal of performance, because it suggests that the CEO has failed to eliminate competition or deter new entrants (the individual effect). The location effect may outweigh the individual effect so that more competitors might lead to more pay – even though you might think it indicates that the CEO has done a poor job in deterring entry. Thus, it may be efficient for apparently “poor” performance to lead to higher pay.
Importantly, the new research shows that both the “individual effect” and the “location effect” should only affect the strike price, and not the number of options. An individually positive signal, such as high profits or low carbon emissions, should increase the level of pay, which is achieved through lowering the strike price. The same is true for a signal suggesting that the location of the stock price range has shifted downwards, such as more competitors.
In practice, however, most companies use performance-based vesting. Here, signals don’t affect the strike price but the number of options the CEO receives – how sensitive their pay is to the stock price. The authors’ analysis shows that this practice is suboptimal. Signals with individual and location effects should affect the level of pay, not the sensitivity of pay to the stock price.
The number of vesting options should in fact depend on signals that have a “precision effect” (where the signal gauges how precise the stock price is as a measure of the CEO’s efforts.) Prof. Edmans explains: “Imagine a signal of low stock price volatility, indicating that the stock price is driven mainly by the CEO’s efforts, rather than random factors. Because the stock price is a precise measure of effort, the CEO’s pay should be sensitive to the stock price – the number of vesting options should rise.”
Performance-based vesting is popular across the business world, but there were no prior theories proposing the conditions under which it is optimal and which performance signals should be used – thus motivating Prof. Edmans and his colleagues to explore this question. Their analysis shows that the number of options should depend only on signals that affect the precision of the stock price as a measure of effort – which only applies to signals such as stock-price volatility. Most performance measures used in practice, such as profits, carbon emissions and peer performance, should affect the strike price rather than the number of vesting options.
Indeed, if high profits, low carbon emissions and more competitors should increase the level of pay, you might think that the simplest way to do this would be to pay a cash bonus. Prof. Edmans explains that this seemingly simple solution is in fact ineffective. “The CEO should only be paid anything at all if the stock price is sufficiently high. If it’s low, then they have failed, and we don’t want to reward the CEO for failure with a cash bonus. Reducing the strike price means that they only benefit from the positive signal if stock price performance is also strong.”
He concludes: “Performance vesting is an area that has been relatively under-researched. As a result, its practice has been driven by shooting from the hip. It seems natural that good signals of performance should increase the number of options the CEO receives, but we show that it is more efficient to reduce stock price. We hope our analysis will help guide practice.”
Alex Edmans is Professor of Finance; Academic Director, Centre for Corporate Governance at London Business School