The Advisors' Blog

– Broc Romanek, CompensationStandards.com

Here’s the intro from this blog by Willis Towers Watson:

We’ve recently been hearing from some companies that they’re reluctant to employ statistical sampling in calculating their CEO pay ratios based on the notion that their calculation will not be as accurate as if they had used actual data for their entire workforce. This thinking draws an analogy to political polling, which is based on statistical sampling and has gotten a bad rap lately because certain polls have failed to accurately predict the outcomes of recent elections. Thus, the logic goes, if polling (and statistical sampling) do not lead to accurate or valid reflections of the true result, why would a company ever want to use this approach to disclose a number as highly charged as the CEO pay ratio?

In reality, however, the entire exercise of compiling the CEO pay ratio is based on estimates, regardless of whether a company uses statistical sampling. This is true because the Securities and Exchange Commission (SEC) does not require companies to calculate Summary Compensation Table (SCT) total compensation for all employees to identify the median employee. As a result, every company will be using a less-than-100%-accurate methodology to arrive at an estimate of the median pay, which is, of course, a permissible approach under the final SEC regulations.

This makes the calculation of the CEO pay ratio different from a political election in that there will never be an indisputable outcome. This may create concerns that some could challenge the approach used to compile the ratio as improper, a particular concern with respect to the plaintiff’s bar. This article is designed to help illustrate how companies can consider addressing those risks and makes the case that using statistical sampling actually can help companies defend themselves on this issue.