A very good friend of mine is a mid-level manager in a public sector agency (he works for the state). While getting ready to counsel a lackluster employee one day, he ran cost/benefit analyses on the employee in question, and then on other employees, ones who were doing well. The results intrigued him, so he expanded the study to a large subset of the people who report to him.
He found that, without fail, his highest-paid people provided the best return for their salary...and not just overall, but on a per-dollar basis too.
I told him "Good. Those people are where you want to invest your payroll dollars."
He started as though something had struck him. "You're right," he finally said.
It sounds as though your friend may be overpaid.
It's long been recognized in the software industry that the developers in the top 10% are several times as productive as the average programmer, while their salaries are not several times as high. So you get much better returns by having a small number of very good people than a large number of low-paid drones