On its website, the AFL-CIO, the largest federation of labor unions in the United States, has a page called Executive Paywatch that is meant to demonstrate just how much corporate executives’ pay dwarfs the compensation of the average worker. On this page, the AFL-CIO reports that the total pay of the CEOs of America’s largest corporations was, on average, 373 times larger than the earnings of an average American worker in 2014, and 335 times larger in 2015. These are striking ratios that are meant to bolster the AFL-CIO’s message: The top executives of America’s corporations are vastly overpaid, and most American workers are woefully underpaid.

For that reason, it may come as a surprise that the AFL-CIO’s calculations grossly understate just how much money executives make. While the AFL-CIO’s calculations are for CEOs at S&P 500 companies, our analysis of data for the 500 highest-paid senior executives (not all of whom are CEOs) from the ExecuComp database, which is maintained by Standard & Poor’s, suggests that the Executive Paywatch ratios are far too low. Data on these executives’ actual take-home pay, which is published, as required by law, in companies’ annual filings with the Securities and Exchange Commission (SEC), show that in 2014, senior executives made 949 times as much money as the average worker, far higher than the AFL-CIO’s ratio of 373:1.

What accounts for this huge discrepancy? The SEC is to blame. The AFL-CIO—as well as most economists and reporters—relies on the SEC-sanctioned estimates of senior-executive stock-based pay that each company reports in the Summary Compensation Table of the proxy statement that it files annually with the SEC, the federal agency whose mission is to protect investors. Curiously, that figure, the “estimated fair value” of executives’ stock-based pay, or EFV, is featured much more prominently in companies’ SEC filings than the figure indicating their actual realized gains, or ARG, on stock-based pay. ARG is a figure that permits the calculation of senior executives’ actual take-home pay—the number they report in their personal-income tax filings with the Internal Revenue Service.

The AFL-CIO is by no means alone in using EFV when it could and should be using ARG. For example, the annual compensation figures for hundreds of CEOs published earlier this year by The New York Times and The Wall Street Journal report total compensation based on the EFV measures of stock-based pay. (The Atlantic has reported on statistics based on similar estimates, too.) The problem is that, on average, for the highest-paid executives, pay estimates that include EFV are far lower than those that include ARG, especially when the stock market is booming. We aren’t the first to point out discrepancies like this, and even such a mainstream institution as the Federal Reserve is aware of them. Earlier this year, for example, two researchers at the Richmond Fed published a report demonstrating how the two measures of pay can diverge.

In 2014, the latest year for which ExecuComp has full data, the average total EFV compensation of the 500 highest-paid executives named on SEC filings (as ranked by total EFV compensation) was $19.3 million, with 62 percent of that attributable to stock-based pay. But the average ARG compensation of the 500 highest-paid executives (as ranked by total ARG compensation) was $34.3 million, with 81 percent of that coming from stock-based pay. As we document in detail in a report issued late last month, total ARG compensation was greater than total EFV compensation in every year from 2006 to 2014, with the largest differences occurring when prices on the stock market were on the rise.

Executive Pay: Two Different Measures, Wildly Different Results

Average annual compensation of the 500 highest-paid executives in the S&P 500 ExecuComp database as ranked by total EFV compensation or total ARG compensation, 2006-2014. (Authors’ calculations; ExecuComp data retrieved on August 4, 2016)

Year to year, ARG compensation rose and fell significantly with the S&P 500 Index, whereas EFV compensation remained relatively flat. The correlation of ARG with stock-price movements makes intuitive sense because an executive’s ARG depends on the number of shares that he or she receives from exercising options and the vesting of awards, which is multiplied by the company’s stock price on the particular days that he or she chooses to exercise options and on which award-vesting restrictions are met. So, in any given year for any given executive, EFV, which is based on newly vested options and awards, may be higher than ARG. But when a company’s stock price rises rapidly, ARG will tend to outstrip EFV.

For the highest-paid executives, ARG on stock-based pay makes up the majority, and in some cases the vast majority, of their total compensation. Other components of executive pay, such as salary and bonus, are typically relatively small percentages of their total pay. The graph below shows the difference between ARG and EFV for both stock options and stock awards, with all four measures drawn from the compensation data for the 500 highest-paid executives ranked by total ARG compensation. The graph also makes clear the sensitivity of ARG, but not EFV, to stock-market booms and busts.

The Importance of Accounting for Stock-Based Pay

ARG from stock options and stock awards compared with EFV of stock options and stock awards for the 500 highest-paid executives as ranked by total ARG compensation, 2006-2014. Note: Executives with $1 billion or more in total ARG compensation are excluded from these computations. (Authors’ calculations; ExecuComp data retrieved on August 4, 2016)  

EFV tends to understate ARG because its estimation methods, which seek to value stock-based pay in advance of realized gains, fail to capture actual stock-price increases and the timing of realized gains. In the case of stock options, the mathematical models, known as Black-Scholes-Merton, that are used to generate EFV assume that stock-price changes are distributed such that one should expect lots of small changes but very few large changes in stock prices. EFV is recorded as compensation in the year in which shares in an option vest, using the grant-date stock price in the estimation model. But, in timing the exercise of the options and generating ARG, executives are on the lookout for large stock-price increases, which will be driven by some combination of innovation, speculation, and manipulation. Black-Scholes-Merton-type models capture none of these three drivers of the stock market, whereas top executives can to some extent influence all of them.

In the case of stock awards, EFV uses the grant-date stock price times the number of shares in the award, recorded in the year that the award vests. But ARG depends on the vesting-date stock price and, in some cases, the ability of the company to hit a performance target such as earnings per share. That performance metric, which can be influenced by executive actions, often triggers the vesting date or determines the number of shares that vest, or both.

Indeed, the point of executive stock options and stock awards is to give executives an incentive to take actions that increase the company’s stock price. Take for example the much-discussed case of the drug company Mylan and the price-gouging of EpiPen users. The profits from these price increases have driven up Mylan’s stock price, and with it senior-executive pay. The reports that in 2015 the pay of Mylan’s CEO, Heather Bresch, was $18.9 million have outraged the public. But that’s an EFV number. In fact, in 2015 Bresch’s total ARG compensation was $44.4 million, of which 72 percent came from stock-based pay.

An even more dramatic example of the difference between EFV and ARG is the total compensation of John C. Martin, the CEO of Gilead Sciences, a drug company that has been in the public spotlight since 2013 for the pricing of its Hepatitis C medicines, Sovaldi and Harvoni. Based on the sales of these drugs, Gilead’s profits soared from a two-year total of $5.7 billion in 2012 and 2013 to $30.2 billion in 2014 and 2015. Gilead’s stock price tripled between 2014 and 2015, boosted by not only its pricing of Sovaldi and Harvoni but also by $15.3 billion in stock buybacks.

In a 2015 Huffington Post op-ed entitled “Gilead’s Greed That Kills,” the economist Jeffrey Sachs wrote that Martin “took home a reported $19 million in [2014] compensation—the spoils of untrammeled greed.” But that figure, which is recorded as Martin’s total 2014 pay in the Summary Compensation Table of Gilead’s SEC filings, is based on the EFV measures of his stock-based pay, not the ARG measures. In 2014, according to those same filings, Martin actually took home $192.8 million, and in 2015, when his total EFV pay was reported as $18.8 million, his actual take-home pay was $232 million. From 1996 through 2015, as Gilead’s CEO, Martin’s total EFV compensation added up to $209 million, but his total ARG pay was just over $1 billion, of which 95 percent came from stock-based compensation.

Accurate knowledge of how much executives like Bresch and Martin got paid raises the more fundamental question of how they and other senior executives manage to realize these enormous gains. A focus on ARG compensation reveals the overwhelming importance of stock-based pay in total compensation, which, in turn, affects the way executives run companies: They can take actions that drive up stock prices, even if temporarily, simply for the sake of increasing their own realized gains.

Throughout the American economy, senior executives have become enamored with stock buybacks as a potent means of manipulating their companies’ stock prices. Over the past decade, net corporate stock issues (that’s new shares issued minus share repurchases, plus shares retired in merger-and-acquisition deals) have drained more than $4 trillion from all U.S. non-financial corporations. The 459 companies in the S&P 500 that were publicly listed between 2006 and 2015 did $3.9 trillion in buybacks, equal to 54 percent of their net income. That’s on top of the $2.7 trillion that these companies distributed to shareholders as dividends, representing another 37 percent of net income. Through stock buybacks of this magnitude, executives effectively participate in the looting of the corporations they run. In some cases, the stock-price increases last just long enough for the executives to exercise their options, have their awards vest, and sell the acquired shares.

Enabling all this is the SEC, the U.S. government agency that is supposed to regulate the stock market to prevent manipulation of stock prices. In fact, since 1982, under Rule 10b-18, the SEC has not only enabled but even promoted stock-price manipulation. That’s in addition to sanctioning a mismeasurement of executive compensation that obscures the ways in which senior executives benefit personally from the manipulative boosting of their companies’ stock prices.

Given that the SEC requires companies to report their executives’ ARG, how did it come to be that EFV became the go-to metric for corporate leaders’ pay? Ironically, the SEC’s mismeasure of executive pay evolved largely as the result of the agency’s attempts, starting in the early 1990s, to make companies’ reporting of executive compensation more rigorous. In response to growing concerns about soaring executive pay in the late 1980s and early ‘90s, the SEC in 1994 replaced the proxy statement’s Cash Compensation Table, in which the only data reported for the company’s highest-paid executives were salary and bonus compensation, with the Summary Compensation Table, which contained data on many additional components of pay. That year, the SEC also had companies report ARG from the exercise of stock options (but not the vesting of stock awards) for the executives named on proxy statements, but that information wasn’t required to be in the new table. At this point, EFV played no role in the filings.

During the 1990s, it was the business-run Financial Accounting Standards Board (FASB) that was interested in EFV because of the rise of stock options as a means of compensating a broad base of professional, technical, and administrative personnel, especially at technology companies such as Intel, Microsoft, and Cisco. The asset managers of pension funds and mutual funds were concerned that as thousands or even tens of thousands of a company’s employees exercised their options, the dilution of the company’s shareholdings would have a negative impact on the market prices of stocks in the funds’ portfolios. To help guide their decisions concerning the buying, holding, and selling of shares, the asset managers asked the FASB to come up with an up-front estimate of the cost of all these broad-based stock options. In calculating these estimates, the FASB favored Black-Scholes-Merton option-pricing models, which were devised by economists as part of the movement to create financial derivatives.

Then, with the bursting of the stock-market bubble in 2001 and the bankruptcy at the end of that year of Enron, transparency in corporate accounting and reporting took on a new urgency, leading Congress to pass the Sarbanes-Oxley Act of 2002. As part of this effort, the SEC, relying on the FASB’s accounting standards, beginning in the 2006 fiscal year required companies to use EFV for the stock-based compensation of all their employees, to be recorded in their income statements contained in an annual filing to the SEC known as a 10-K.

However, it was the granting of stock options to a broad base of employees that led asset managers to demand an up-front fair-value estimate of their dilutive impact. Senior executives were receiving lots of stock options, but if those had been the only options concerned, the FASB would probably not have implemented EFV for reporting stock-option compensation expenses in 10-K filings. Given the origins of EFV and the availability of ARG measures, it was a mistake for the SEC to use EFV to report senior executive pay in proxy statements.

Yet armed with EFV measures, in 2006 the SEC revamped the Summary Compensation Table it had introduced in 1994, describing it as “the cornerstone of the SEC's required disclosure on executive compensation.” For the first time, the SEC required companies to provide a figure indicating executives’ total compensation, which appears in the Summary Compensation Table and includes EFV measures of stock options and stock awards. As a result, anyone who wants to get an SEC-approved figure on a named senior executive’s total compensation can pluck the number from the Summary Compensation Table, without necessarily noting that it includes EFV measures.

Meanwhile, as part of these revisions, the SEC also required companies to report the ARG from the vesting of stock awards (in addition to the ARG from the exercise of stock options, which since 1994 companies had been already required to disclose). But the SEC made the unwise decision to place this ARG data in the peripheral Option Exercises and Stock Vested table instead of in its “cornerstone” Summary Compensation Table, where the EFV measures appear.

Why did the SEC privilege EFV over ARG in constructing a total compensation figure? The SEC declined to comment for this article, but perhaps it was because some of its key personnel had an intellectual commitment to the Black-Scholes-Merton models taught in business schools and economics departments and implemented on Wall Street. In 2007, Chester Spatt, then the SEC’s chief economist and director of economic analysis, gave a speech at the Wharton School about accountants’ increased reliance on EFV. Spatt asserted that the reliability of modeling methods for the valuation of employee stock options was “clear from the history of mortgage-backed securities.” As was made painfully apparent in the financial crisis that occurred not long after Spatt’s 2007 speech, the models underlying mortgage-backed securities have the potential, like those generating EFV figures, to wreak economic havoc.

Sadly, the SEC’s rules about corporate disclosure are moving in the wrong direction. Next year, as mandated by the Dodd-Frank Act of 2010, the SEC’s Pay Ratio Disclosure Rule will go into effect, requiring companies to publish the ratio of their CEO’s pay to that of their median employee’s. However, the SEC instructs companies to report these ratios using EFV measures of executive pay, not ARG measures—furthering the systematic underestimation of U.S. income inequality.

Why has the discrepancy between ARG and EFV gone largely unremarked? There are in fact many people who are well aware of the difference, but who have no incentive to speak up and correct this misinformation. When The Wall Street Journal and The New York Times published their lists of the purportedly highest-paid CEOs earlier this year, was there one among the hundreds of named CEOs who wrote a letter to the editor, revealing what he or she had really been paid? Or were there any CEOs who complained that they were omitted from the list or, if included, would have ranked much higher if the highest-paid CEOs had been scored by actual realized gains?

These executive-compensation figures are so large that they can appear abstract, even meaningless, to the ordinary worker. When someone is already making tens of millions of dollars, what's the significance of tens of millions more? What most people do not realize is that the corporate behavior that results in this level of extreme wealth is largely responsible for the decades-long erosion of the American middle class. America’s severe and persistent income inequality is not just about how much these executives have been able to secure as compensation. It is more fundamentally about how senior executives have been able to boost their stock-based pay, while the average American worker faces wage stagnation.