What "Overpaid" and "Underpaid" Might Mean.
Over on Concurring Opinions, Sarah Lawsky asks what we mean when we say that judges (and others) are underpaid, or that law professors (and others) are overpaid. She asks for an explanation of how markets fail in these sorts of cases, and what kind of an argument could justify appealing to the "intrinsic value" of various kinds of services in order to ground the claim that, regardless of whether markets are functioning well or not, someone is overpaid or underpaid.
There are two ways that we can easily make sense of such a claim. The first relies on transaction costs and externalities, and suggests that there is a market failure in such cases. The second relies on the idea that there are multiple possible efficient outcomes, and that we can make normative judgments between those outcomes. I'll summarize them here. (Application of these concepts to judges and law professors will be left as an exercise for the reader. :-) )
1. Externalities and Transaction Costs.
The first way someone can be overpaid or underpaid is if their work produces uncompensated externalities. Once we recognize externalities, and the notion that there's a difference between the social value of someone's product and the value to the buyer of someone's product, this is really easy to see.
The ultimate limiting case is the case of a mafia hit man. The hit man is obviously overpaid: he causes so much harm to the victim, and to the society, that even if the market between hit men and people who want to have people killed is perfectly efficient, the inability to make the hit man internalize the externalities means that he's getting far more money than the worth of his services. This is a market failure: there are too many transaction costs to allow the rest of us to, say, pay him to do something less injurious with his time (like sit around doing nothing, even).
For a less extreme case, consider, for example, the case of a lawyer in one's least favorite side of the Tort Wars (someone who defends tobacco companies, if one is a liberal, say, or a personal injury lawyer if one is a conservative). Again, the idea is a mismatch between social value and the incentives of the people who are doing the paying. For some reason, those doing the paying have a lot of money, and they're willing to spend it to do something horrible to the rest of us, and transaction costs are such that the market can't get some Coasean solution.
Suppose, for example, that a lawyer for the tobacco companies produces one million dollars worth of benefit for the tobacco companies, and is paid precisely that (by the tobacco companies). Suppose also that the lawyer's defense of the tobacco companies causes one dollar worth of harm to each of 10 million smokers. In a world without transaction costs, collective action problems, etc., the lawyer could be paid to stop working for the tobacco companies. Such costs exist, so the market fails. The lawyer is overpaid: he's paid a million bucks to produce nine million dollars worth of social harm.
Likewise (although slightly less obvious -- and I'm a little less sure of this), suppose that someone's job produces a lot of positive externalities. An easy case here is a private school teacher (I specify the teacher as being employed by a private school to make the question of who is doing the paying nice and simple). His services might be worth $50,000/year to a school and parents, but he might be producing many times that in social gains from having an educated population, etc. We can, I think, say that the teacher is underpaid in the sense that the market fails to capture all the value that the teacher produces.
2. Division of the Gains From Cooperation
Even without market failures, we might think that people can be overpaid or underpaid relative to some coherent normative standard if there are multiple efficient divisions of the surplus from cooperation. Suppose, for example, that Sam would be willing to take an underwater basket-weaving job for $50,000/year, and that Mary, the employer, would be willing to pay $100,000/year to hire an underwater basket-weaver. Any salary in [50k, 100k] will be efficient, and consistent with a well-functioning market.
But that doesn't mean the choice from that range is exempt from normative criticism! One of the many big ideas from Rawls's Theory of Justice is the idea that what we're doing, when we consciously arrange our economic system in the interests of justice, is establishing a fair division of the gains from cooperation. And I think that's intuitive and important. Given that our transaction produces a surplus, and given that efficiency considerations don't dictate how that surplus is to be divided, why can't we say that some divisions of the surplus are unfair? That is, there may be multiple pareto-optimal distributions, such that we have to appeal to non-market values to determine what the distribution should be.