[T]he problem with public sector compensation is that there is often very little clarity in terms of whether or not taxpayers are getting a good deal. One of the big reasons right-wingers are so hot for merit pay, based on my limited experience, is that they’re generally pretty comfortable with the idea of at least some public workers making much more than they are making now, provided other workers who’d be willing to work for less because they’re not likely to attract better offers are either paid less or fired.
I agree with Manzi that this still doesn't really tell us whether state workers are overpaid, underpaid, or just-right-paid. I suspect that the answer is probably "both"--adjusting for worker quality, the median government worker is probably overpaid, while in skilled specialties, salaries are probably not attracting as much of the top-flight talent as we'd ideally like. (This is why I have been advocating, futilely, that we make it possible to pay SEC employees multiples of what the President of the United States makes.) But as Manzi, who does this stuff for a living, will undoubtedly tell you, setting compensation is a really hard problem that no one's got a very good handle on. So that's just a suspicion, based on my experience of state bureaucracies, and my best guess at the incentive effects of the current structure.