I wonder if optimizing for votes, as is the day to day business of most policy makers, gets you to the same place as optimizing for profits like in the private sector? If not, studying that could be one way to improve the public sector output. Not that shareholders vote, on average, but you get the idea.