Arnold Kling  

Against Merit Pay for Teachers

The Latest Nobel... Financial Dysfunction: Who Sha...

That would be my position.

A government-run system of teacher compensation, based on test scores, would in some ways be the worst of all worlds. It would create incentives for teachers to "game" the system. It would give too much weight to a noisy indicator of performance. As a result, it would do little or nothing to improve accountability or to reward better teachers.

I think that the only solution is to get the government out of the business of providing education. A few days ago, I finally got around to watching the documentary Waiting for Superman. That film could be interpreted as making an even more passionate statement in favor of getting government out and letting private firms compete.

Comments and Sharing

COMMENTS (14 to date)
Roger Sweeny writes:
Using test scores as the determinant of teacher pay misses what economist Friedrich Hayek called “tacit knowledge” and “local knowledge.” Many factors affect student test scores, meaning scores are a noisy indicator of teacher performance. People closest to the teacher, including peers, principals, and parents, have more information about teacher quality than what can be obtained by remote administrators relying on test scores.

The first two statements are true but the third is false. In fact, "peers, principals, and parents" have very little other information about teacher quality.

Principals are "supervisors" only in a legal sense. They have almost no direct contact with teachers teaching. Assistant principals and department heads have more but it isn't much. Maybe a few "observations" a year, observations that are usually scheduled in advance, and are really more in the nature of a dog and pony show.

Principals and administrators probably will find out if a teacher is regularly coming in drunk or if he sits all period and has the students fill out worksheets, but beyond that, meh.

Parents, too, don't find out much. Again, the news of a truly bad or abusive teacher will make it home and be believed but otherwise, it's a crap shoot. Does the parent discount stories of a bad teacher? Does she give too much credence to complaints about a good teacher? What makes it to the principal is an incredibly noisy signal.

Peers don't know too much either. It's a cliche that once the classroom door closes, each teacher is in his or her own world. Very few teachers see other teachers at work and very few know much in the way of facts about what other teachers do or how students react.

(The above is based on my experience as a teacher in an ordinary high school in the Boston area, and on talking to other teachers in eastern MA.)

Thomas DeMeo writes:

Are we letting go of something important by systematically removing these few institutions where different elements of society interact with each other and are forced to work things out? I agree it is inefficient on many levels, but will we still have any community?

kyle8 writes:

What about merit pay in a purely private system? wouldn't the same incentives apply?

I suppose the personal oversight by those running the school and their greater discretion in hiring might compensate for that. But not if the teacher union remains powerful.

Mr. Econotarian writes:

1) Testing is important. It is too important to be performed by the teachers educating the students, it should be done by non-interested outside third parties to avoid gaming/cheating.

2) Teachers should report to someone who is truly their manager. That someone should be free to hire/fire & set wages. Currently they "report" to a principal who has none of those abilities.

3) Whomever this school manager is, they should report to a multi-school district manager who should be free to hire/fire & set wages for their reports.

4) Some teachers will fail, and school managers will deal with them. Some school managers will fail, and district managers will deal with them.

5) The position of school manager and chief disciplinary officer for the school should be separated, as one deals with teachers, and the other deals with students.

Mike Rulle writes:

Interestingly, there is little to no incentive pay in private secondary schools. However, what they do have and exercise is the power to eliminate underperformers by their own standards. As one gets more senority, one gets more pay and percs----but they do not get anything approaching tenure, either formally or informally.

I think it is more analogous to old private partnerships. More or less equality of pay as long as you are good enough to stick around. These schools are completely unemcumbered by government induced requirements. Hence,consumer satisfaction is high.

My understanding is this tends to also be true in parts of the public school system where one has to apply to get in---NYC has many of these schools besides the obvious well known ones. Again, creating competition to get in seems to result in creating competition for the schools themselves to perform well.

Complicating matters is that many suburban towns love their school system. I think the reason is ultimately they deliver the "goods"--their kids get into good colleges. The last thing they would want is competition from other towns which would limit spaces for their own kids. It would be treated with all the warmth and love as forced busing was in the 1970s. These parents tend not to care about costs because they don't pay the full share. Certainly their real estate taxes pay, but not all who pay taxes use school services.

I agree with a start slow method and see how it goes. Vouchers paid for by ones personal real estate taxes is not a bad place to start. Like so many things, it is easier to arrive at a theoretical outcome considerably better than we have, but so much harder to politically implement.

We have the further problem of Unionization combined with the Bureaucratic instinct to expand. Wisconsin and NJ are remarkable examples of virtually inconsequential cuts causing outrage (perhaps faux) among the Unions. One thing is certain. We do not need the Federal Government involved----but we went down the road before---the Department of Education will never be eliminated.

PrometheeFeu writes:


Schools that do a better job of attracting better teachers will eventually become known as good schools which will attract more students and greater profit. Conversely, schools that have lots of bad teachers will tend to loose money and run out of business.

GIVCO writes:

The headline is misleading unless you mean that you're against ALL merit pay for teachers.

@Roger Sweeny

That sounds wrong. A normal school has formal events, informal coffee klatches, gossip networks, etc. where parents, administrators, and peers exchange information. Maybe its not perfect information, but its useful information and equivalent to what many businesses use for their merit bonuses.

Harold Black writes:

You need to read Francis Tarkenton's piece in the Wall Street Journal asking what would happen if the NFL emulated how teachers are compensated.

Seth writes:

It seems like the primary bad assumption is test-based accountability. The discussion on alternatives to this seems very light. It's just assumed that test-based accountability is the best thing we have.

At events with other parents we trade info on teachers to learn who the better ones are (and we also weigh the opinions of those other parents). We talk to administrators. We talk to our children's teachers.

I'm not sure about this, but I think in private schools, administrators tend hold onto the teachers that parents highly recommend because that's what keeps those parents coming.

Basically, what I'm saying is that most businesses operate on their ability to attract customers and not on an average standardized test score. McDonalds and Burger King would be indistinguishable on a standardized test. But, there are still good reasons folks choose one over another at any given time.

Charles R. Williams writes:

In a school everyone knows who the best teachers are and who the worst teachers are. The problem with merit pay is that administrators who would make decisions about rewarding merit are themselves unaccountable. A second problem is that teachers themselves often do not have an objective viewpoint about their own performance. Using student tests to identify high performing teachers is ludicrous for many reasons.

Merit pay for teachers will work only if administrators are truly accountable to parents and if teacher pay is not disconnected from the market.

ciro curbelo writes:

I agree with Givco. The title should be changed to "The Economic Weakness of centrally-planned Merit Pay for Teachers.

There is nothing wrong with merit-pay per se. There is a lot wrong with trying to do it in a centrally-planned, government-run governance structure.

Also, it strikes me that the arguments apply not only to merit pay, but also to teacher evaluation and performance management.

Customer "feedback" in the form of voting with your feed is the only force that can keep boards, senior and mid-level managers, and teacher managers in line and focused on doing what is necessary to attract and retain students.

Without this you get poorly designed and implemented evaluation and compensation systems that are easily gamed (think Atlanta, Denver, NYC, Houston to name a few).

This is why "implementation with fidelity" is such a buzzword in education today. Because unaccountable central planners are divining human capital systems that senior and middle managers have no desire to implement.

siredge writes:

Absolute scores can be gamed. Freakanomics demonstrated how that can be caught: check the students after they are done with the next teacher. Give the students a different teacher, then see how they do in subsequent tests. This would work better if students rotated teachers two or three times per school year instead of just once, but even that would be enough to detect cheating once a teacher has participated for 2 or 3 cycles.

Roger Sweeny writes:


The information that gets passed around says almost nothing about how much students are learning.

"Everyone knows" about the exceptionally good or the exceptionally bad teacher, but no one really knows about the 95% in the middle.

Otto Maddox writes:

Oh hell, the administrators are in bed with the faculty and politics would determine merit pay.

Comments for this entry have been closed
Return to top