[U]niversities may well be better off by paying lower salaries to tenured faculty, despite the adverse incentive effects, than paying higher salaries to professors without tenure. In other words, Steve thinks the competitive market for professors is resulting in inefficient contracts, while I believe that, absent any reason for market failure, the labor contracts we observe are likely to be efficient.
But there is a glaring source of market failure that Mankiw overlooks: the non-profit status of universities. There are lots of reasons to think that non-profits tend to choose inefficient paths of least resistance. (Here and here are two cases of interest).
The most obvious reason: In a non-profit, efficient reforms give the people in charge big headaches (see the case of Larry Summers) with virtually zero financial reward. So why rock the boat?