CEO pay ratios: audit the employee population first
A CEO pay ratio can fall because the employee population changes even when no continuing worker receives a raise. Rebuild the denominator’s population and methodology before describing the ratio as evidence of improving pay fairness. The arithmetic is simple; identifying the compensation measure and comparable workforce is the harder research task.
Establish the disclosure population
The SEC’s pay-ratio rule overview describes the relevant US disclosure and issuer exemptions. Item 402(u) specifies the population and measurement requirements, while staff calculation guidance discusses reasonable estimates and sampling. This article concerns that disclosure, not every country’s wage-gap statistics or the distinct pay-versus-performance calculation.
Record the selected employee-identification date, consolidated subsidiaries, countries, employment types, exclusions and consistently applied identification measure. A third-party worker’s treatment depends on the actual definition and employment circumstances, not simply whether the worker enters the company’s premises. Do not assume an outsourced workforce is included in the same way as direct employees.
Hypothetical composition change
Assume five non-CEO employees have annual total compensation of 30, 40, 50, 90 and 100 thousand. Their median is 50. CEO annual total compensation is 5 million, giving a ratio of 100 to one. All amounts and workers are hypothetical, and assume the simplified population is complete.
Now assume the two lowest-paid roles are outsourced and the remaining employees’ compensation stays unchanged. Assume the new workers are employed and paid by an unaffiliated third party and fall outside the relevant population under the stated facts. The remaining compensation amounts are 50, 90 and 100 thousand, with median 90. The ratio falls to approximately 55.56 to one despite unchanged pay for every continuing employee and the CEO.
| Hypothetical population | Median compensation | CEO compensation | Ratio |
|---|---|---|---|
| Original five employees | 50,000 | 5,000,000 | 100:1 |
| Remaining three employees | 90,000 | 5,000,000 | 55.56:1 |
This is a composition demonstration, not a claim that outsourcing is always permitted to exclude a worker or that any real company manipulated its ratio. The facts must establish the worker’s actual status under the applicable rule.
Separate identification from final compensation
A company can use a permitted consistently applied measure to identify a median employee and then calculate the required annual total compensation. Those are related steps, not necessarily the same payroll number. Record both measures and avoid comparing a salary-only identification figure with a total-compensation CEO numerator.
Check selected-date changes, permitted annualization, exclusions and whether the same median employee is retained under the relevant conditions. Do not independently annualize seasonal workers or convert all part-time employees into full-time equivalents merely to create a seemingly comparable statistic. Any analytical supplement must be clearly labeled and should not impersonate the disclosed ratio.
Build the population bridge
Use opening and closing headcount, acquisitions, disposals, outsourcing, geographic mix, employment categories, exclusion reasons, selected dates and methodology changes. Add numerator changes separately: equity awards, bonuses and a CEO transition can move the ratio even if the workforce is constant.
Observe whether a ratio change is explained by genuine pay changes, population composition or measurement choices. Public reports may not disclose the employee distribution required for an exact decomposition. In that case, identify plausible mechanisms as hypotheses and retain an unexplained residual rather than inventing a precise percentage contribution.
The final memo should report the disclosed ratio, the supported population changes and limits on comparison. It can evaluate transparency and consistency without claiming that one ratio measures employee welfare, productivity or fairness across businesses with fundamentally different workforces.
A simple control is to sort the hypothetical remaining compensation amounts again rather than averaging them. The average of 50, 90 and 100 thousand is 80 thousand, while the median remains 90 thousand; substituting the average would create a different ratio.
FAQ
Does a lower ratio prove ordinary employees were paid more?
No. Population and CEO-compensation changes can move the ratio without continuing-worker raises.
Is the median employee the average employee?
No. The median is a position in the ordered distribution, not total compensation divided by headcount.