I think the math is more complicated than this makes it seem. The missing bits are that many employees will stay even though they are underpaid and many people

I think the math is more complicated than this makes it seem. The missing bits are that many employees will stay even though they are underpaid and many people will still leave even if they are given retention raises. If an employer gives a raise, that is a reoccurring expense they might not have had to pay; it might decrease the chance an employee leave from 25% to 10%, which means it is an expense they didn't have to pay 85% of the time (75% they would have stayed anyway and 10% chance they will leave anyway). 100% of the time the company will have to pay the expense, though. The few people you do retain from the raises are likely going to be people who will leave for a better job at some point anyway, even with the raise. After a number of years of giving retention raises, you will likely lose most of the people you retained for a short time with the raise, while you continue to pay the increased salary to people who would have stayed anyway. Over a longer period of time, it becomes almost pure cost without a ton of benefit. There is obviously a balance here. You want to give the right raises of the right amount to retain your best talent without spending more money than you have to. If the pay is bad enough, you will lose all your best talent. A good HR department would know these things. They will know industry standard retention rates and know if the company is above or below the average. They will talk with managers to assess the quality of the people leaving and why, and adjust raises and bonuses as needed. While this sucks as employees, it isn't irrational.