Fair500 → Analysis → The pay ratio explained
Since 2018 every US public company has had to publish the ratio of its chief executive's pay to that of its median employee. The rule is more flexible, and the resulting number less comparable, than almost anyone assumes.
The CEO pay ratio is one line in a proxy statement, and it is the most-quoted and least-understood number in American corporate disclosure. This is what it is, where it came from, and what the rule permits.
Section 953(b) of the Dodd-Frank Wall Street Reform and Consumer Protection Act, passed in 2010, directed the SEC to require public companies to disclose three things: the median annual total compensation of all employees except the chief executive, the annual total compensation of the chief executive, and the ratio between the two.
The SEC adopted the implementing rule in 2015, and the first disclosures appeared in proxy statements filed in 2018. Every US-listed company of any size now publishes the figure annually in its DEF 14A.
Both halves use the same definition of compensation, the one already used in the Summary Compensation Table. That means the ratio is not a salary comparison. It includes salary, cash bonus, the grant-date fair value of stock and option awards, non-equity incentive plan payouts, pension value changes and other compensation. For a typical chief executive, equity is the majority of the total; for a typical employee it is usually zero.
The one-line version. Total compensation of the CEO, divided by the total compensation of the employee who sits exactly at the midpoint of everybody else. Across the S&P 500, the median company reports about 196:1.
The rule's requirement is strict about what must be published and notably loose about how the median employee is found. Five choices belong to the company.
A company need not compute full Summary-Compensation-Table totals for every employee, which would be enormously burdensome at a firm with 300,000 staff. Instead it may pick any "consistently applied compensation measure" to rank its workforce: total cash compensation, W-2 wages, base salary alone, payroll records. Only once the median employee is identified does the company compute that one person's full total compensation properly.
Different measures identify different people. Ranking by base salary alone will find a different median employee than ranking by total cash including overtime and bonus, particularly in a workforce with significant variable pay.
Companies may use reasonable estimates and statistical sampling rather than examining the whole workforce. This is legitimate and often sensible, but it means some published medians are point estimates with unstated uncertainty around them.
The company chooses any date within the last three months of its fiscal year at which to take the employee census. For a business with seasonal hiring, that choice matters considerably. A retailer that picks a date before seasonal hiring begins has a different workforce, and a different median, than one that picks a date in December.
A company may exclude non-US employees where they make up 5% or less of the total workforce, and may exclude employees in jurisdictions whose data privacy laws make collection infeasible. The 5% de minimis allowance is small, but it can be applied selectively to the lowest-paid populations.
Once identified, the same median employee may be used for three consecutive years, provided there has been no significant change in workforce or pay arrangements. So a "median employee" figure may reflect a person identified two years earlier, with only their own compensation updated.
Put the five together and the consequence is unavoidable: two companies with genuinely identical workforces can publish different median figures through methodology alone. The ratio is a reliable disclosure about a single company's own reported numbers. It is a weaker instrument for ranking companies against one another than its widespread use as a league table implies.
This is not a scandal, and it is not concealment. The methodology must be disclosed, and companies do disclose it. It is a design tradeoff. The SEC chose flexibility to keep compliance costs manageable, and paid for it in comparability.
Nor is it the largest problem with reading the ratio across companies. That distinction belongs to workforce composition: a global manufacturer's median employee may work in Malaysia, and a restaurant group's may work twenty hours a week. Those differences swamp methodology effects, and they are the subject of a separate piece, because they explain most of what looks extreme in the ratio data.
Three omissions are worth knowing.
With all of the above stipulated, the ratio remains the only mandatory, standardised, annually updated statistic that connects the top of a company's pay distribution to its middle. Before 2018 there was nothing. Executive pay was disclosed in great detail and worker pay was not disclosed at all, so any comparison relied on national wage averages that had no particular connection to the company in question.
Used carefully, within an industry, with attention to the median, and with the timing distortions of equity grants smoothed out, it supports comparisons that were simply impossible a decade ago. That is the use Fair500 tries to make of it.
No. The rule requires disclosure and nothing else. There is no cap, no threshold and no penalty attached to any particular figure. A few US municipalities have experimented with tax surcharges tied to the ratio, but at federal level the disclosure carries no consequence beyond publicity.
Almost always because the chief executive is paid unusually, not because workers are paid unusually well, though occasionally both. Some executives at partnership-like firms take modest disclosed compensation while participating in profits through other structures. Berkshire Hathaway reports 4:1 because Warren Buffett's salary is $100,000. Take-Two Interactive reports about 2:1 because its chief executive is compensated primarily through an external management company rather than by Take-Two directly, so the Summary Compensation Table records only a small residue of the true arrangement. The figure is correctly disclosed and describes very little.
More than it should, and mostly for a reason unrelated to pay policy: equity grant timing. A board that awards a multi-year grant in one year makes that year's ratio spike and the following years' collapse. This is a large enough distortion that Fair500 recomputes every ratio on a three-year average of CEO compensation. The reasoning is here.