Merit Pay for Teachers Works and the Evidence Now Proves It

The idea that any business would pay employees based simply on time served, rather than on the quality of work performed, is economically daffy on its face.  It is as offensive on economic grounds as it is on philosophic ones.

On economic grounds, it obviously leads to a severe case of the agency problem: by giving people no incentive to perform quality work, they (the agents) will likely fail to deliver a quality service to the principals (the people for whose benefit the agents are supposed to work).  On philosophic grounds, it violates the meritorian principle of social justice -- namely, that those who have more merit (work harder, more productively, and more diligently) should be rewarded more than those with less merit.

But the special interests who control the American public education business (i.e., education bureaucrats, administrators, and teachers) view education as not a business at all, but instead a magical realm of "public service" immune to all the principles that apply in the rest of the world.  This is a topsy-turvy alternative universe, indeed, in which telling someone that he has a job for life no matter how poorly he serves the consumer, and that his pay will never increase based upon how well he serves the consumer, is somehow a perfect way to get him to...serve the consumer!

Of course, helping the public education bureaucrats maintain their daffy, delusional worldview is the fact that the public education business is a monopoly.  In a monopoly, workers don't have to worry about pleasing consumers simply because the consumers have no power to go elsewhere.

However, recent research by the eminent economist Ludger Woessmann should shake up the complacent, daffy world of American public education just a bit.  Woessmann did a detailed examination of the results of the 2004 PISA (Programme for International Student Assessment) tests of students worldwide conducted by the OECD (Organization for Economic Development).  It is a robust set of data: over 200,000 fifteen-year-old students from 27 countries on four continents were tested on math, reading, and science in the period of 2003-2004.

As Woessmann notes:

Standard economic theory predicts that workers will exert more effort when monetary rewards are tied to the amount of the product they produce. Not only does performance pay stimulate individual effort on the job, it is theorized, but jobs where rewards are tied to effort attract energetic, risk-taking employees who are likely to be more productive.

And, indeed, that is what he found.  Students in countries with performance-pegged pay for teachers significantly outscored the students in systems that didn't allow for merit pay.

In fact, those students in countries with merit-pay compensation schemes do about 15% of a standard deviation higher in science, and a hefty 25% higher in reading and math than those students in countries without merit pay schemes.  A quarter of a standard deviation is equivalent to about one year of learning, which means that students will gain about a year in math and reading learning, and about a half-year in science learning, from being in a merit-based system.

Woessmann, a model of intellectual honesty, is quick to point out that his is an observational rather than an experimental study.  In an observational study, it is inherently more difficult to precisely isolate the causal factors (or as statisticians put it, "rule out the confounding variables") with the same level of confidence as one would in the ideal case of a control group experiment.

So, for example, the correlation between student performance and performance-based pay schemes might be due to (say) Asian cultures being both open to performance-based pay as well as being more supportive of education generally.  To help rule this out, Woessmann ran his analysis again on a continent-by-continent basis ("continental fixed effects"), comparing Asian countries with merit pay to Asian countries without it, European countries with merit pay to European countries without it, and so on.  Not only did the gap between the performances of students in countries with merit pay systems versus students in countries without such systems remain, but it also widened.

Woessmann redid his analysis to rule out a number of other possible explanations.  To rule out the possibility that exceptionally low GDP might be skewing the results (since national GDP is a known correlate of national school performance), he showed that even if you drop the two countries with exceptionally low GDP rates (Mexico and Turkey), the performance gap remains.

To rule out the possibility that merit pay affects the level of education of the teachers in a country and therefore should not be used as a control variable, Woessmann conducted another analysis excluding that variable.  The gap remained.  Woessman conducted another analysis where he excluded one country at a time, to rule out the possibility that one exceptional case (what is called an "outlier") might be causing the gap.  It remained.

Woessman even conducted another analysis excluding two clusters of countries (in Scandinavia and Eastern Europe) where merit pay is prevalent, to see whether the results where inordinately due to those clusters.  The result was that the gap still remained.

Finally, Woessman ruled out the possibility that it is simply the general flexibility of teacher compensation schemes, not those schemes being tied to student performance, which was responsible for the results.  He did this by estimating the effects of varying salary adjustments according to teaching conditions and responsibilities, teacher age and family status, and finally teacher qualifications and training on student performance.  None of the effects were significant in his results.

So Woessmann's observational study quite robustly supports the view that yea, verily, even in the world of education, performance pay does in fact produce better results.  And he notes that the experimental studies that have been done (in India and Israel) on merit pay for teachers also show at least a short-term positive impact of performance-pegged pay.  Indeed, in this regard, Woessman's study dovetails nicely with those experimental studies, because it examines the long-term effects of merit pay.

Also supporting the view that merit pay works is, of course, its consistency with what we know about human motivation in all other areas of work -- namely, that people are incentivized to do better at their jobs when their pay is tied to the quality of their work.

Lest I be misunderstood, let me hasten to add that nobody is claiming that merit pay is the sole answer to fixing the wretched failure that is American education.  First and foremost, we need school choice -- lots of it, and right away.  We need accurate, independent testing of all taxpayer-supported schools done at the national level, so parents choosing a school have an independent source of reliable information upon which to choose.  We need to eliminate the requirement that people who want to teach get a credential from an education department.  And we surely need to get rid of teacher tenure.

Still, merit pay is part of the solution, as Woessmann's timely study has so nicely demonstrated.

Gary Jason is a contributing editor of Libertyunbound.
If you experience technical problems, please write to