Opinion: We Should Pay More for the Best Teachers

A version of this essay appeared on Matthew Yglesias’ Slow Boring, a site dedicated to offering pragmatic takes on politics and public policy. David Broockman and Josh Kalla recently published a paper in which they tested the electoral impact of “moving to the center” on a range of issues. In theoretical terms, the main import of this paper is to establish that adopting a more moderate view improves your electoral performance only if the more moderate view is more popular. In most instances it is, but that’s not always the case. They find that Democrats mostly benefit from moving to the center on cultural issues (affirmative action, crime, admission of asylum seekers, girls’ sports) but are generally better off maintaining conventional progressive positions in areas like Social Security and health care. (Note that this is different from the populist formula of saying that Democrats will benefit from moving further left on economic issues. It’s just that a basic defense of the existing safety net is very popular.) I wrote about these findings broadly in the New York Times, but there is one very interesting exception to the culture/economics pattern that’s worth talking about in detail: performance pay for teachers. On this issue, they tested three positions: - Liberal: It should be difficult to fire teachers, and their pay should primarily depend on how long they have worked as a teacher. - Conservative: Being a teacher should be like any other job, where better teachers get paid more and worse teachers can be easily fired or get paid less. - Centrist: It should be difficult to fire teachers, but better teachers should get paid more than worse teachers. In contrast to the cultural issues, where moving to the center helps Democrats, this all sounds a little tedious and technical. But Broockman and Kalla find that attributing the centrist position to a Democratic candidate has a larger impact on that candidate’s vote share than moving to the center on girls’ sports teams, on asylum, on gender transition surgeries for minors, or on prison sentences for property crimes. The impact is bigger than embracing an “all of the above” energy policy rather than one that is renewables-only. In fact, the only issues that had a larger impact on candidate evaluation in their survey were affirmative action in college admissions and racial targeting of small business loans. That means that this wonkish technical question of teacher compensation is actually a bigger deal in the eyes of voters than a lot of inflammatory cultural issues. Why is this shift so high-impact? Not only is the liberal position very unpopular (it’s just as unpopular on these other topics), even the moderate position doesn’t have much support. Over 60% of voters agree with the stated conservative position. The authors also find consistently in this experiment that voters with conservative views on a given topic appreciate it when Democrats move to the center on that issue (and vice versa for voters with liberal views when evaluating Republicans). The upshot is that there is a lot of electoral upside to offering the pretty tepid take that teacher compensation should be based on performance rather than pure seniority, even without any compromise on basic job protections. It makes sense for politicians to say things that voters agree with, but you want to be careful about committing yourself to policies that will catastrophically fail if implemented. I think it is reasonably likely that a candidate could win an election by promising price controls on groceries, and that’s especially true if they did it while maintaining broadly moderate vibes. The problem is that if a state actually imposed price controls on groceries, it would immediately suffer from severe shortages. And when that happened, the voters would not say “Ah, my bad, I didn’t think this through.” Voters, sadly, are not very self-aware and they would blame the politicians who caused the shortages. Conversely, while voters were deeply skeptical of New York City’s congestion pricing plan, once it was implemented, we saw the same political dynamic that we previously saw in Oslo, Stockholm, and London: Voters liked the reduction in traffic volume. Voter skepticism of congestion pricing seems to be driven by skepticism that it will work, but when they see that it works, they like it. I don’t want to downplay the pure politics here, though. Lots of people in the Democratic Party camp are spinning their wheels furiously in a quest for big exciting new ideas, often embracing proposals that make almost no sense on the merits. It turns out, though, that there is an incredibly large political upside to just saying “Hey, we should get schools to give raises only to their best teachers so we’re making sure they don’t leave the profession.” Consider all the genuine excitement that DOGE initially aroused in some quarters before crashing and burning as a total failure: Lots of people are very sincerely interested in the question of improving the performance of the public sector, and putting out plausible ideas to do this — like paying teachers based on demonstrated skill rather than pure seniority — is (in the eyes of the voters) a genuinely exciting big new idea. On the merits, to the extent that there’s a problem here, it’s that translating the words “better teachers get paid more” into an actual policy is a nontrivial task. You don’t want to reward the teachers whose students do the best on year-end tests, because that’s just encouraging teachers to take on the easiest assignments in classrooms full of smart kids with no problems. You also don’t want to reward the teachers whose students like them best, because the point of school is to make kids do boring stuff like learn math. So you need an evaluation system and that’s going to be hard to get right. The good news is that I surveyed the evidence and, while there is a wide range of program designs and a wide range of findings among researchers, nobody is finding that performance-pay programs backfire and make things worse. The entire research debate is between enthusiasts who claim large benefits and skeptics who say the benefits are small or null. I think the existence of skeptical findings — and a general sense that the impact of all education policy changes is pretty small — has created a permission structure for Democrats to backslide from Obama-era efforts at education reform in order to avoid conflict with unions. That’s why the polling is relevant, though. The political upside to embracing the commonsense view that pure seniority-based pay doesn’t make sense is large. And it’s also just clearly the case that pure seniority-based pay doesn’t make sense! Designing and implementing a great teacher-evaluation system is hard, but “be at least as good as strict seniority” is easy because the strict seniority system is clearly bad. There are a lot of studies of merit-pay systems. If you want a summary, there’s a meta-analysis looking at a few dozen of them that concludes that “the effect of teacher merit pay on student test scores is positive and statistically significant (0.043 standard deviation).” It also features the boring but important caveat that the effect “varies by program design and study context” — the details matter. You can also look at the most negative studies. Here’s a program out of Nashville that had no effect, a 2013 null-effect study of a New York incentive-pay program, and another bonus program that didn’t work in part because half of eligible teachers didn’t know it existed. I think that for understanding the recent political history of teacher compensation in the United States, the most important paper is “Taking Teacher Evaluation to Scale: The Effect of State Reforms on Achievement and Attainment” by Joshua Bleiberg, Eric Brunner, Erica Harbatkin, Matthew Kraft, and Matthew Springer. It studies how the Obama administration took advantage of the state/local budget crisis inf…