This should be the sort of thing that the managers above notice, I wonder why they never seem to. Tracking turnover within a team should be a trivial query in any HR system.
I've been friends with the (competent) managers above a bad manager, and discussed a number of times how bad this manager was. They agreed with my assessment, and yet did nothing about it. I suspect this was partly because they underestimated how much of a disastrous effect this manager had and figured that the unpleasantness of firing him combined with the work of finding someone else would not be worth it. But mainly I think it was because they were 'friendly' with this bad manager (on account of having lunch and meetings together as managers), and because he was 'one of them'.
Maybe if managers are kept from become too much of a tribe some of this could be avoided (assuming that there is competence at some higher level, of course)?
Managers are like this x10. You go to meetups - they go to MBA school. You post on forums - they play golf and go to strip clubs. Rationality goes out of the window when tribal loyalty comes in.
I believe that they made the wrong call. But then again I'm biased, disgruntled, and all the rest of it.
There are many signs of bad managers, I am sure we all have them, but my number one is, thinking their title makes them right and beyond question. Got enough of those around to choke a horse. Worse it is usually a symptom that comes from top down and is hard to fix.
And yes - this should be trivial to measure. Hiring a good employee costs X. Of those, 1/Y is a great one. The cost of losing a great one is XY. XY is usually significantly higher than a year's salary.
For the same reason, I also think managers should be rewarded for creating great employees and giving them to the broader organization.