Are CEOs rewarded for luck?

Shareholders of the tech giants were not alone in expressing their dissatisfaction at their leaders’ remuneration. In 2021, the ratio of US CEO pay to median employee pay shot up from 192 in the previous year to 245 to 1; while, according to the FT, the CEOs of S&P 500 firms out-earned the average worker by 324 to 1 that year – and a record number of S&P 500 companies failed to attract 50% support from shareholders for their leader’s pay deals. 

Agency-centric models vs the rent-extraction view

During the extended bull market of the 1990s, publicly traded US companies devised stock-option programmes that yielded hitherto-unseen levels of compensation for senior executives – and spawned a parallel rise in academic work on the subject.

Historically, the academic press used an “agency-centric” model of CEO remuneration. Broadly, this holds that the shareholders are the owners of the company and, as the senior execs are there to deliver for the shareholders, they (the execs) need to be incentivised to make decisions in the interest of the shareholders, and this is what they do.

Thus, according to the agency-centric model, the senior execs are (or should be) compensated for how their efforts affect overall firm performance, net of market/sector performance effects that are not attributable to managerial effort, and many of the empirical data findings in the academic literature are consistent with this view. 

But in 2001, in a seminal paper entitled ‘Are CEOs Rewarded for Luck? The Ones Without Principals Are’, Bertrand and Mullainathan questioned the ability of agency-centric models to explain CEO compensation patterns adequately. The authors looked at whether the level of CEO compensation was related to oil-price changes and industry returns, and found that it was, in fact, related to both – but, of course, the CEO did not control these factors: if the oil and gas industry did well, the CEO was compensated accordingly, regardless of their performance. This finding led the authors to conclude that managers were being paid for lucky outcomes (“pay for luck”) and that the agency-centric view of compensation painted an incomplete picture, at best. 

Other papers developed this idea, including Bebchuck and Fried, who in a 2004 study built on the concept of pay for luck, or pay for non-performance, and found that “structural flaws in corporate governance have enabled managers to influence their own pay and produced widespread distortions in pay arrangements.” 

According to Bebchuck and Fried, the CEOs who had lots of power were able to exercise control and influence over their board of directors – and were also the CEOs who were getting excessively compensated. In other words, powerful CEOs control boards and the pay-setting process, wielding their power to “extract rent” (this is also known as the rent-extraction view).

Role of luck

Garvey and Milbourn put an interesting spin on the tale in their 2006 paper on the relationship between luck and executive compensation. Using industry benchmarks, they found there was “significantly less pay for luck when luck is down (in which case, pay for luck would reduce compensation) than when it is up.” 

In effect, executives are rewarded for good luck, but not penalised to the same degree for bad; hence there was a highly asymmetrical relationship in terms of pay for luck. In effect, Garvey and Milbourn argued that the rent-extraction view requires that CEOs are compensated for good luck but not for bad. 

In the last decade, researchers have offered multiple explanations for why managers need to be compensated for industry-wide performance, even though these returns are beyond a manager’s control. One such theory counters that when firms in an industry do well, the “opportunity set” for talented managers increases, and to avoid peer firms from poaching such managers, firms need to compensate them for the industry’s performance. 

Yet another explanation is based on the argument that, when an industry does well, firms in that industry need to be appropriately positioned to take advantage of these industry-wide changes. This requires effort to be expended by their managers, so it is only fair that they are compensated for the industry-wide performance. 

These counter explanations have dented the reliability of prior evidence supporting the rent-extraction view. Its credibility also suffered from recent research by Daniel, Li and Naveen, whose 2020 study argued that Garvey and Milbourn’s (2006) findings regarding asymmetry were not robust, because asymmetry was observed in only 2% of 205 different regression specifications. This meant, according to the authors, that pay for industry and market performance, even when beyond a manager’s control, is not pay for luck.

In line with these concerns, my LBS colleague Alex Edmans, together with Xavier Gabaix of Harvard University and Dirk Jenter of the London School of Economics and Political Science, called for more rigorous empirical evaluation of the rent-extraction view. In a 2017 paper, they observe that the recent theoretical contributions make it clear that shareholder-value models can be consistent with a wide range of observed compensation patterns and practices, including the large increase in executive pay since the 1970s. Edmans et al noted the challenge now was to confront these new models more rigorously with the data, explore their limitations, and contrast them with the rent-extraction models.

Our research setting: the US Tax Cuts and Jobs Act of 2017

Together with Martina Andreani of LBS and Atif Ellahie of the David Eccles School of Business at the University of Utah I set out to re-examine the agency-centric vs rent-extraction views of CEO compensation by investigating the impact on firms of the US Tax Cuts and Jobs Act of 2017 (TCJA). 

The TCJA was one of the biggest tax changes in the corporate history of the United States. Before it came into force, US companies paid 35% of their profits in taxes to the government. (Internationally, this rate is comparatively uncompetitive; in the UK, for example, the rate prior to the government’s autumn 2022 reform was 19%).

The impact of the Act was a perfect setting for our investigation for several reasons, but first let’s recall what the competing views say.

Agency-centric models predict that CEOs should not be rewarded for one-off windfalls that (i) relate to past transactions; (ii) are unrelated to firm-specific, industry or market-wide performance; and (iii) have no impact on future firm profitability or investment opportunities.

By contrast, rent-extraction models predict that CEOs facing weak scrutiny of their pay will be rewarded for gains that are unrelated to managerial effort, but not penalised for losses that are beyond a manager’s control. 

The Act was highly conducive to our purposes for two main reasons. First, Donald Trump’s Bill, which exceeded 400 pages in length, was passed into law in barely three months, so it was difficult for firms to predict its final form and adjust their behaviour accordingly in advance (the US Securities and Exchange Commission allowed firms one transition year to recognise the tax effects). Second, it introduced major reforms, including a significantly lower corporate income tax rate (reduced from 35% to 21%) that led some firms to report one-off transitional gains and others to report losses. 

The context meant we were able to conduct a quasi-natural experiment and focus on CEO compensation for one-time tax-gains and tax losses in companies by asking this question: If a tax gain accrues to a company as a result of the reform, does the CEO deserve compensation for it? 

Another favourable aspect of the TCJA is that the one-off tax gains and losses were heterogenous and substantial. For example, Berkshire Hathaway received a net tax benefit of $29 billion; whereas Citigroup suffered a net tax loss of $22.6 billion. Roughly equal numbers of firms reported increases and decreases in net earnings. And, whereas for TCJA-benefitting firms aggregate profits increased from $79 billion to $315 billion, the TCJA converted aggregate profits of $160 billion to a loss of $71 billion for other firms. This made the Act an ideal setting to check the asymmetry in CEO compensation to tax gains and losses. 

It’s key to note that the tax windfalls that we examine are one-off, not the recurring benefits of a reduced corporate tax rate in the future; thus, the tax windfalls we study do not affect the future profitability of the firm. These windfalls arose from companies’ past activities, over a long period of time; hence were not something that the current CEO had brought about.

The agency-centric model says that the CEO should not get compensated for this – but the rent-extraction model holds that the CEO will be compensated for tax gains: although company performance is unrelated to managerial effort, it is “free money” and the weakly monitored CEOs will have a share of it. In other words, it is pay for luck.

One-time winners and losers

The TCJA required remeasurement of deferred tax assets (DTAs) and deferred tax liabilities (DTLs) in the balance sheet, with one-time gains and losses from this remeasurement reflected in the income statement. This was key to our research design because differences in the dates when companies had to recognise a profit in their income statement relative to when taxes had to be paid on that profit meant the TCJA created one-time winners and losers.

To give an example of this effect, suppose a company has generated a profit of $1 million in the tax year ending 31 December 2017 and recognises the profit in its 2017 financial statements, but the government allows the company to defer payment of taxes on its 2017 profit for two years. 

That means, for tax purposes, the profit is treated as though it was earned in 2019. So, in 2017, the company shows a $1 million profit and recognises a deferred tax liability (DTL) in its accounts of $350,000, reflecting the tax payments at 35% of its profits that need to be made in two years. But then, with little prior notice, the government reduces the tax rate to 20% from 2018 onwards, bringing the company’s tax bill on its 2017 profits down to $200,000. The difference between the DTL recognised in 2017 ($350,000) and the DTL needed as of 2018 ($200,000) is a one-time tax gain of $150,000. This results from the company’s past activities, but is recognised in its 2018 accounts as a $150,000 profit. So, should the CEO get a slice of this $150,000? 

The results

Our sample consisted of 2081 relatively large firms over the seven-year period from January 2013 to December 2019. These firms paid their CEOs an average compensation of $6.3 million, of which their fixed compensation (salary) constituted a relatively small proportion, totalling a mean average of around $0.8 million. The difference is made up of discretionary bonuses and incentive plans. 

Our findings show that, upon TCJA passage, the CEO compensation increased substantially more for firms reporting large windfall gains when compared to firms reporting small windfall gains. Specifically, the CEOs of firms in the top quartile of gain-reporting firms increased their compensation by 8.3% more than firms reporting in the bottom quartile. This 8.3% equates to around $330,000 in additional compensation for CEOs of firms with large tax gains. Consistent with this reward being one-off for transitional tax gains, it is reflected only in the discretionary component of CEO pay. 

In contrast to the rewards seen for reporting large tax gains, we find no evidence that CEOs were penalised for reporting tax losses. CEO compensation was unaffected by the magnitude of the tax windfall losses reported by firms.

These findings are consistent with the predictions of the rent-extraction model that CEOs would be paid for lucky windfall gains, but not penalised for windfall losses. Also supporting this model, we find that pay-for-luck occurs mainly for CEOs whose pay is poorly scrutinised because the firm has a weak board of directors, has few analysts following its activities, has more transient investors or attracts scant media interest. 

It is also significant that the pay for tax gains is much stronger for domestic firms (ie; those whose operations are headquartered in the US) than for international firms. This is because the latter are typically exposed to greater pay scrutiny and, as they pay taxes in multiple countries, their benefits from the reduction in US tax rates are limited.

We also found that the compensation was only given to the CEO (and, to a lesser extent, the CFO): other senior executives and ‘ordinary’ employees did not share in the tax windfall gain. This is consistent with the view that the CEO wields managerial power over the governance of the firm; such that they are able to extract greater compensation than they merit purely as a reward for effort – which is again consistent with the pay-for-luck view of CEO compensation.

The results also show that CEO pay for tax gains is not a reward for their lobbying efforts or support of Donald Trump’s candidacy, whose election victory made the TCJA gains possible. In fact, contrary to this theory, the pay for windfall gains was more likely to be awarded to politically neutral and even Democratic-leaning CEOs than their Republican counterparts.

Lastly, because the compensation given for tax windfall gains does not improve company performance in the future, it is hard to see these payments as tending to better align CEOs’ interests with those of shareholders, or yielding direct benefits to shareholders.

Because the additional compensation was not due to any managerial effort or tending to lead to better future performance, we can conclude that the pay associated with the tax windfall gains is more consistent with the rent-extraction view of CEO compensation than the agency-centric models. In other words, it is not compensation that is in the interest of shareholders in any way.

Of course, the findings are specific to a particular context; hence they may not be generalisable to other settings. Nonetheless, the fact that CEOs behave in this way in one context strongly suggests that the level of CEO compensation is not always optimum from the point of view of shareholders.

Lakshmanan Shivakumar is the Lord David Sainsbury of Turville Professor and Professor of Accounting at London Business School.