I think it's more likely this: calculating the cost of giving an employee a raise is very simple. Calculating the cost of finding a new employee is extremely va

I think it's more likely this: calculating the cost of giving an employee a raise is very simple. Calculating the cost of finding a new employee is extremely variable (how long does it take and how many people are involved?) and also involves metrics that are very hard to measure (how much more would the team lead and the team members have gotten done that week if they weren't interviewing candidates?) Usually if you're debating management, especially non-technical management, solid numbers win out over hand waivey numbers, even if the hand waivey numbers are representing something far more important and the solid numbers are representing something in a very limited context. An example: Tom is a mediocre programmer, but he's worked for the company for 5 years and has a lot of institutional knowledge. He wants an above average salary now. Should we give it to him? The obvious answer would be to let him walk, he's asking for more than he's really worth. Ok, but how much does the equation change if we factor in that it's going to take 3-5 months to hire his replacement, and you'll need to interview about 20 candidates, some of them for multiple hours, not to mention potentially flying them out? Also now Tom is walking with his institutional knowledge, and the bugs that used to go to him are going to someone that doesn't have the 5 years working with the system he does, so whoever has to pick up his slack is going to be slower than Tom is even if they're a better programmer overall.