I kind of agree. I had a discussion about this with a friend of mine who was the lead architect at the last company I was at and he made a good point.
Every manager, once promoted to management, needs to make a choice between:
1) Being on the side of upper management
2) Being on the side of the employees they manage
3) Being on the fence, where they neither have the trust of their bosses or their employees.
We came to the conclusion that the best place for a manager to be is on the side of the employees. The reason being is that the healthiest, most productive teams have a manager that the employees trust in. They know they have someone higher up looking out for them and shielding them so they can get things done. If the company isn't one where this is possible, a really good manager will probably leave for greener pastures.
The next best place for the manager to be is on the side of the upper management, because at least this leaves no room for confusion among their subordinates (uggg, I hate that work, but it's appropriate here).
The worst place to be is in the middle. In this position the manager sabotages his own career and therefore his department because his upper management doesn't trust him and his employees don't trust him because they don't know if he has their best interests at heart half the time.
I think Todd Jackson, GMail product manager, said it best, when he said that "[A manager] can either be a shit funnel or a shit umbrella."
ref: http://techcrunch.com/2010/03/14/key-to-gmail/
I propose that we call this new metric a manager's "body count".
After an employee leaves, his compatriots should rate him to be an asset or liability. Those who quit voluntarily and are considered assets are added to a manager's body count. If the manager is on good terms with the employee who should know why the employee left and will have incentive to raise issues for living that are not his fault as a manager.