Nancy Flanagan asked a very good question in the comments of Sunday's post that got me thinking.
In the original post I asked what we should incentivize when paying teachers and discussed the difficulties in trying to make this decision. Indeed, some basketball teams are finding that incentivizing certain outcomes encourages players to do things that actually hurt the team. In other words, it's crucial to carefully think through what is incentivized and, in some cases, it may be better to incentivize means rather than ends. I postulated that the simplest solution in basketball is to just base bonuses on wins but that education has no such bottom line measurement.
Nancy suggested rewarding teachers who earn national board certification, and I noted that this is a measure of good a teacher is -- not how well they performed in a given year. To which she responded:
"Are you trying to make a distinction between a teacher's general ability to increase student learning, and year-to-year test data, which will fluctuate? Which of these is more important to incentivize?"
And I think I finally have an answer. It's more important to incentivize performance, but it's more important to reward quality. Let me explain.
Two chief complaints exist about the way teachers are paid:
1.) Better teachers aren't paid any more -- which isn't fair
2.) There's no incentive for teachers to work harder or do better -- which doesn't fit with economic theory
The problems are related, but different -- and we can attempt to fix both of those problems in different ways. Measuring the quality of a teacher is easier than measuring the performance of a teacher. Studies using sophisticated value-added measurements show large fluctations in performance year-to-year. It's not really plausible that teachers really skip around that much in terms of effort, quality of lesson plans, or anything else they can control. Much of the fluctuations must be due to a combination of environmental factors (different kids and different classes click better with different teachers) and measurement error in test scores. So simply rewarding teachers for their test scores in a certain year relies on some combination of teacher quality, performance, and chance. In other words, it's not really satisfying our desire to pay better teachers more.
On the other hand, it's much easier to determine who is a "better" teacher over time -- through observations, test score results, national board certification, etc. So it makes more sense to boost the pay of teachers who we've determined are better because we think they're good rather than because we think they did a good job this year.
At the same time, this doesn't solve the conondrum with teachers not having an (economic) incentive to push their kids to achieve. In order to fix this, incentives based on yearly performance need to exist. Of course, measuring yearly performance is incredibly difficult -- but that's another topic for another time.
So, in short, my answer to Nancy's question is" "yes," and "both."