More on the statistical chicanery of the AFL-CIO’s artificially inflated CEO-to-worker pay ratio

From Mark Perry.

"The Wall Street Journal reported today on its front page that “The median pay for CEOs of the biggest U.S. companies was $11.7 million in 2016, up from $10.8 million in 2015 and a postrecession record, according to a Wall Street Journal analysis of S&P 500 firms.”

According to the AFL-CIO’s annual report on CEO pay released in May: “In 2016, CEOs of S&P 500 Index companies received, on average, $13.1 million in total compensation, according to the AFL-CIO’s analysis of available data.” Last year, the AFL-CIO reported that “the average CEO of an S&P 500 company made $12.4 million per year in 2015.”

I’ve previously documented the statistical chicanery and legerdemain employed by the AFL-CIO to exaggerate and inflate its “CEO-to-Worker Pay” ratio (see CD posts here and here) that includes:

1. Using a small sample of the highest paid CEOs in America, the AFL-CIO compares the total annual compensation of about 400 S&P 500 CEOs to the average annual pay of about 100 million rank-and-file workers, most of whom don’t work for S&P 500 companies. A more accurate comparison would be of S&P 500 CEO compensation to the average pay of employees of those same companies.

2. Comparing the total compensation of CEOs in the S&P500 (including all fringe benefits) to the cash-only wages of rank-and-file workers (excluding all fringe benefits), resulting in a distorted apples-to-oranges comparison. To be fair, the AFL-CIO should either: a) include fringe benefits for both CEOs and rank-and-file workers or b) exclude fringe benefits and compare only cash compensation.

3. Comparing the total compensation of CEOs who are working full-time and likely putting in 50-60 hour workweeks managing large multi-national corporation to the cash-only income of rank-and-file workers whose average workweek is only 33.6 hours (less than the 35 weekly hours required to be classified as a full-time worker). How about comparing CEO compensation to the compensation of rank-and-file workers (including fringe benefits) who are working the same number of weekly hours as a typical CEO?

4. CEOs of the S&P 500 are typically in their peak earning years, and their average age is about 57 years. In contrast, the 100 million rank-and-file workers considered by the AFL-CIO include workers of all ages, including many young and part-time workers. A more accurate, apples-to-apples comparison would adjust for age and would compare CEO compensation to the compensation of full-time rank-and-file workers in their prime earning years, e.g. workers in their late 50s.

Based on today’s WSJ article, here’s another item to add to the long list of shady statistics used by the AFL-CIO to calculate an inflated CEO-to-worker pay ratio:

5. The AFL-CIO uses average CEO compensation instead of median CEO compensation, which inflates the figure used for the annual compensation of a typical S&P 500 CEO by about $1.5 million ($13.1 million average vs. $11.7 million median CEO compensation for 2016, and $12.4 million vs. $10.8 million in 2015). By using average CEO pay, the figure is unfairly biased upwards by the influence of a small group of very highly paid outlier CEOs. For example, in 2016 Charter Communications CEO Thomas Rutledge was paid $98.5 million, and six other CEOs were paid more than $40 million.

It’s a standard statistical practice that whenever there are extreme outliers (e.g., home prices, household income) in a sample, median values should be used (median home price, median household income) to more accurately reflect a typical or representative value. If the AFL-CIO had used median instead of average CEO pay, its CEO-to-worker pay ratio would have declined from 347-to-1 to 310-to-1 in 2016, and from 335-to-1 in 2015 to 291-to-1. Therefore, in addition to all of the other questionable statistics used by the AFL-CIO, the use of average CEO pay instead of median CEO pay has artificially inflated its CEO-to-worker pay ratio in the range of 12-15% over the last several years.
Bottom Line: As I reported in this recent CD post, when you correct for many of the questionable statistics used by the AFL-CIO (using median CEO pay of the Russell 3000 companies, including fringe benefits for rank-and-file workers, and assuming a workweek for rank-and-file workers that more closely matched the average weekly hours of a CEO) a more realistic statistical approach deflates the CEO-to-worker pay ratio down from the 347-to-1 ratio reported by the AFL-CIO to a ratio in the 40-to-1 to 60-to-1 range. As I concluded in that post (slightly revised):

Perhaps CEO compensation is an issue that deserves attention. But to bring attention to the issue, the AFL-CIO’s annual reports on the CEO-to-worker pay ratio use a bogus statistical methodology that is so flawed, deceptive and distorted that the union group’s yearly gripes about CEO pay really can’t be taken very seriously. It’s pretty obvious that the AFL-CIO’s approach is to artificially inflate the CEO-to-worker pay ratio for publicity purposes and to generate sensationalized media attention by comparing the average (not median) total compensation of a small group of the highest paid CEOs in America to the cash-only income of 100 million rank-and-file workers who work an average of only 33.6 hours per week. It’s a dishonest approach that wildly exaggerates economic reality."