The Peter Principle Is Real. Someone Finally Measured It.

The Peter Principle, the idea that people rise until they reach a job they are bad at, was a joke in a 1969 book. In 2019 three economists tested it against payroll data and found it holds.

The study

Benson, Li and Shue used transaction-level data from a sales performance software vendor covering 131 United States firms, 38,843 sales workers and 1,553 promotions into management, between 2005 and 2011.

Sales is a good testing ground because individual output is measured precisely, which removes most of the ambiguity that makes this question hard elsewhere.

What they found

Firms promote on sales. Doubling a worker’s relative sales performance raised their monthly promotion probability by 0.074 percentage points, a 32% increase on a base rate of 4%. In this data, doubling sales means moving from roughly the 50th to the 67th percentile, so this is not an extreme comparison.

And high pre-promotion sales predicted worse management. Doubling a manager’s pre-promotion sales corresponded to a decline in manager value added such that each subordinate’s sales fell 6.1%. With a typical span of five reports, that is nearly a third of a person’s output lost.

The effect was, if anything, stronger for managers assigned to a team other than their own, which rules out the explanation that they simply inherited a weaker team.

The thing firms overlook

Sales collaboration experience, measured as the number of colleagues sharing credit on transactions, positively predicted manager value added. It was not consistently rewarded in promotion decisions.

In other words the signal that predicts good management was visible in the data and largely ignored, while the signal that predicts bad management was the one being rewarded. Under a counterfactual policy maximising expected managerial quality, average quality would be 30% higher.

The caveat the coverage drops

The authors are careful in a way most summaries are not. They write: "We caution against interpreting these results as evidence that firms have mistaken beliefs or behave inefficiently."

Promoting top performers may be an efficient way to motivate everyone else. If the promotion is the prize, awarding it on measured output keeps the tournament credible. The finding is that managerial fit and incentive design pull in different directions, and firms appear to trade one against the other deliberately.

Consistent with that, firms whose managers supervise larger teams put less weight on sales and more on collaboration.

What to do with it

• If you are being promoted for individual output, ask what evidence exists that you will be good at the new job
• If you are promoting, collaboration history is a better signal than performance and is usually sitting unused in your data
• Being excellent at the work is not evidence of being able to lead people doing it

This is the structural version of the argument in why most people become managers before they are ready. The evidence on what management quality is actually worth sits in does management actually matter, and the career fork itself in the software engineer guide.