Over at Ta-Nehisi's place, a guest-poster makes the following observation:
Work environments hospitable to continual innovation tend to have relatively low barriers to entry, and relatively low barriers to exit. Schools invert that. Many have extensive up-front credentialing requirements, forcing novice teachers to invest substantial time and money at the beginning of their careers, before they can even decide whether they are indeed well-suited for the job. Early career teachers tend to get the least desirable assignments, and to be paid barely enough on which to live. On the other hand, most compensation packages are grossly back-loaded, offering lock-step seniority raises and substantial retirement benefits. So it's tough to get in the door, and once you do, leaving entails abandoning the rewards for which you've already labored before you can enjoy them. That's crazy.
Naturally, I am prone to believing that this is true. Which, naturally, makes me suspicious at how sensible this sounds. After all, I could tell a different story: how teachers freed from the fear of being fired are more willing to innovate. I'm sure I could track down at least one case where that's actually true.
How to resolve those competing stories? I think by looking at other industries. Why did the unions damage GM so badly? I'd argue that it was because GM was in an industry where barriers to entry were high, and barriers to exit were also high--it's not easy just to shut down your billions of dollars worth of capital plant. In industries that are competitive, the unions don't necessarily damage their firms (but also don't get their workers nearly as much in the way of wage boosts, work rules, and job protections.)
You can observe roughly the same pattern in schools. In suburban schools, where parents will exit underperforming systems, the teachers' unions don't matter all that much. In systems where exit is very difficult--particularly urban systems, where the teachers need to be better than average just to overcome the harships their students started out with--they're a huge barrier to reform. In other words, exit/entry at the employer level can substitute, if imperfectly, for those things at the teacher level.
So the thesis does seem to hold, but only with some caveats. I nonetheless concur that in an era when the schools clearly need change, the ability to cling to your job like a rabid abalone should not be something that teacher compensation systems select for--yet that's one of the biggest benefits the job offers. This does, indeed, seem crazy.