Bud Crystal has forgotten more about executive compensation than most of us will ever know. Indefatigable, Bud continues to study topics in the area and recently decided “to have one last look at CEO pay stats,” after the end of the 2014 proxy season. In his own words:

I created a database consisting of every company with sales of $5 billion or more. That gave me a starting position of 516 companies. (I had aimed for about 500 companies). After eliminating cases with missing data and CEOs with tenure of less than one year, I ended up with a database of 434 companies. Here are the key findings:

  • The "usual suspects" are at the top (e.g., CBS), but we have oddball fiscal year players like Oracle. And then there is KKR, which has a total of $88 million of pay for both Henry Kravis and George Roberts combined.
  • I was able to explain 34 percent of the variation in pay using two factors:
    • Company Size, which gives equal weight to sales, total capital (long-term debt plus shareholders' equity) and number of employees; in other words, an income statement measure, a balance sheet measure and a human capital measure.
    • CEO tenure. As mentioned, the final model includes only CEOs with tenure >= 1 year.
    • The probability of either factor occurring by chance was less than 1-in-10,000.
  • I tried to add a third factor of excess total return (actual total return less return on SPX), but the computer rejected it as having no statistical significance. The correlation between excess return and pay at first looked quite good; then I removed Fannie Mae as an outlier, which, in 2013, had a total return of 1,080 percent. Once that was removed, the relationship between total return and pay collapsed, the probability of a chance result increasing to 80 percent.
  • Ditto for whether or not the CEO was a female, where female CEOs, on average, earned 6.5 percent less than their male counterparts but where the probability of a chance result was an unacceptably-high 60 percent. There may be a "War on Women" out there somewhere, but not when it comes to paying female CEOs in large companies.

Bud created a summary spreadsheet, arranged in descending order of the amount by which actual pay exceeds (or is under) the going rate of pay from the pay model, which he might pass along upon request.