Don Boudreaux links to Bryan Caplan, who has an interesting post up pointing out that we need to consider the dynamic response to a minimum wage announcement. I only call it "interesting" because I'm not entirely sure what I think of it yet, but I think it's worth thinking about.
The point is that if a minimum wage hike is passed in advance firms can gradually adjust through reduced hiring and attrition (if indeed that's the direction they would respond - of course if the monopsony model is correct there may be increased hiring gradually). The point is, by the time of the actual discontinuity in the law everything might have happened already.
The trouble I'm having with this is that I'm not sure what would cause the gradual response. If you were making major transitions to capital-intensive production I could see doing that gradually, but presumably that's not what we're talking about in many of these cases (although I suppose that's an empirical question). If you're not doing the same thing, more capital intensively, why not just pay the same amount of people low wages until the minimum wage goes into effect and then pull up stakes? In the high turnover world of fast food franchises (as in, for example, the Card and Krueger study) that seems very reasonable. What glide path are we really thinking of in the Card and Krueger case??
But as I say - it's interesting. I don't quite know the answer.
This reminds me of a post I had a couple weeks back on the minimum wage noting that we have to follow things forward in time before we pass judgment. Why? Because different monopsony models imply different long run effects. If it's a straightforward market power model then the minimum wage (as long as it's modest) should be good in the long and short run. If it's a fixed labor cost version of the monopsony model it will have good short run and bad long run effects. So you need to look a while out to really get a sense of what you're dealing with.
It's an ex-post dynamic response explanation. Here, Bryan and Don are talking about an ex-ante dynamic response. But it's still the same sort of idea.