Somewhere, in the last few months -- alas, I don't recall where -- I read about a company that rates its managers on how much their teams accomplish, divided by the size of the team. They claimed, plausibly I think, that this system effectively discouraged such fiefdom-building.
That's an awful recipe for overworked employees within a short time, unless the formula was only one small part of how managers were ranked. Anonymous management satisfaction surveys getting factored in would be a good start.
I've always wondered if a very broad range of metrics gathered at all times, and a constantly changing formula picked at the last moment, coupled with an explicitly-written policy that if everyone hit the selected high mark then everyone would get rewarded Lake Woebegone-like, would avoid people working towards the formulae.
