Every year for more than a decade, various think tanks in Canada, the United States, and the United Kingdom produce a “bombshell” study where they state that the Chief Executive Officers (CEOs) of the largest corporations earn many times the income of the average worker. For example, the Canadian Centre for Policy Alternatives (CCPA) published a report that depicted what it called a “new gilded age” because of the finding that the highest-paid CEOs in Canada made 246 times what the average worker made. In the United States, the Economic Policy Institute (EPI) made the same point, finding a higher proportion of 344 times in its latest report. In all the reports, there is an attempt to imply that these differentials are outrageous and that they are unlinked with market fundamentals.

There is no way around it. No reason to mince words. These studies are designed to generate outrage. They are conceived to find exactly what the authors want to find (a high ratio) even if the statistics produced in no way justify the outrage or the claim of a disconnect with market fundamentals.

In 2019, I went through studies such as those of the CCPA and the EPI to understand their methodology and found that they take five important shortcuts to generate the results they obtain and make the incorrect inferences they ended up making.

First, they typically (but not always) take the “salary” of the average worker. Salary, however, is not the same as compensation, which is broader. Compensation includes salary, health insurance, life insurance, retirement benefits, and paid vacation days. This is an important sleight of hand. In Canada, when you switch to the total compensation of the average worker, you find that the ratio falls by more than 10 percent.

Second, they also compare with all workers. But the CEOs they pick are from a small share of the top firms. These firms are large and recognizable, and generally offer higher wages than the average firm. Shifting to the total compensation of workers in the top firms (those of more than 500 employees) shrinks the ratios of CEO compensation to average worker compensation by 24 percent.

Third, the selection of the top of top firms is misleading. The top firms are generally very different from all other large firms. They are more likely to be engaged in international competition, and in sectors with high turnover in firm composition. So, what happens if we extend the composition to a larger set of the “top firms”? In Canada, there are data that allow us to expand the comparison of the CEOs of the top 100 firms to the CEOs of the top 1000 firms. Using the average compensation of CEOs in the top 1000 firms shrinks the ratio relative to the compensation of the average worker in large firms by 81 percent.

Fourth, they fail to point out that CEOs are often fired from their positions. The turnover is huge: 42 of the top 100 Canadian CEOs in 2007 were out by 2008. By 2017, only 15 remained. In the United States, Steven Kaplan and Bernadette Minton found a similar rate of turnover since 2000. Most of the turnover is not due to CEOs just retiring or being hired elsewhere, but due to disappointing performances and board decisions to fire them.

Fifth, they assume that we should judge the performance based on stock market valuations. But that is incorrect since the valuations are the realized outcomes and not the causal effect of a CEO. For example, imagine that a firm expects the market for its goods to contract in the years to come. The board of administrators is concerned about bankruptcy and it hires someone to avoid the fulfillment of that fear. The CEO manages to accomplish this and the firm shows zero earnings growth instead of negative earnings growth. CEOs are hired and fired according to whether they meet expected performance. The EPI and CCPA studies always fail to acknowledge this simple fact of financial economics.

So, what determines CEO pay? Supply and demand. In the last decades, as a result of a more globalized economy, large firms saw increasing competition. Whereas national firms used to be worried mostly about other domestic players, they now have to worry about competitors all around the world. The cost of managerial mistakes in the face of this competition is also greater. The demand for extremely rare managerial skills has thus increased in order to avoid these mistakes. This is why we observe that more than 60 percent of corporate executives now have graduate degrees (compared to less than 10 percent in the 1930s) in pure sciences, mathematics, engineering, and statistics. They are also older – reflecting greater acquired experience – than in the past.

Simultaneously, the supply of such skills has not kept up with demand. The time needed to accumulate skills and experience is considerable and few people want to have the downsides of being a CEO. That life is stressful and less conducive to family formation. There is a strong reputational aspect to the craft. Errors can follow a person for a long time, forever tainting a reputation. Few people want to follow that path given these trade-offs. Moreover (and somewhat counterintuitively), while CEOs of top firms are smart people, they are probably not the smartest people in terms of cognitive abilities. But they score higher on non-cognitive abilities such as autonomy, self-discipline, resilience, and impulse control. These non-cognitive skills are complementary to the cognitive skills. The problem is that the combination of such complementary skills is rare, resulting in only a small pool of possible candidates. The supply is growing slowly.

Demand is increasing faster than supply, which results in growing CEO pay. It’s basic economics.

To be sure, we can argue that one component of demand is tied to the non-managerial skills of a CEO – namely political skills. Boards of administrators – especially in my native Canada – often pick CEOs based on their ability to lobby governments for favors and privileges. For example, Canada’s aeronautics industry frequently receives government subsidies. The CEOs of firms in that industry are selected in part based on their ability to request government aid. I deem this to be a problem, as the outcome of such hires is lower living standards for taxpayers and consumers. But it is not the same problem as the one that the CCPA and EPI highlight.

So, let’s stop faking up outrage by engaging in performative statistical construction.