Podcast: Eric Hanushek and Paul Peterson discuss why merit pay experiments in the U.S. tend not to last very long or work very well.
As education policy churns through fad after fad, merit pay is really hot right now. The U. S. Department of Education asked states to include proposals for implementing teacher merit pay—pay based on classroom performance—in their 2010 applications for Race to the Top (RttT) monies, and many applicants promised action on this front. In Washington, D.C., former schools chancellor Michelle Rhee negotiated a strikingly original merit-pay plan, despite strong union opposition. According to the latest Education Next poll, public support for merit pay gained significant ground over the past year and now outdistances opposition by a 2:1 margin.
Replacing the standardized salary schedule, where the only factors that determine teacher salaries are the number of years on the job and academic credentials, seems a worthwhile goal. In theory, pay-for-performance plans both provide a clear monetary incentive to teachers to find the best way to motivate and instruct their students and, over the longer term, attract and retain those more-effective teachers who wish to work in a field that rewards professionals for the quality of their efforts. But enacting high-quality performance pay plans in the United States is easier said than done. Last year, the Florida legislature enacted one of the more stringent proposals any state has ever attempted—only to have the bill vetoed by Governor Charlie Crist as a way of jump-starting his ultimately doomed bid to become Florida’s first independent U.S. senator. That is not the only time a merit pay bill has seemed on the verge of success, only to founder or be undermined by the need to compromise. In general, merit pay plans are more likely to be symbolic than substantive and more likely to be promised than delivered.
Most often, they are not even promised. Even if one counts the most token of performance pay plans, they are to be found in no more than 500 school districts out of some 14,000 districts nationwide, a mere 3.5 percent of the total.
When new merit-pay plans are proposed, teachers unions often block their enactment or water down their provisions. In Cincinnati and Philadelphia, for example, merit pay policies were blocked just before they were about to be implemented. Denver’s Professional Compensation for Teachers (ProComp) plan, widely heralded as the leading national example of performance pay, awards more money for earning another degree than for demonstrated performance in the classroom. In Houston, merit was defined so broadly that it included an overwhelming majority of the teachers. In Florida, Iowa, and Texas, the legislatures have encouraged local districts to enact performance pay plans. But unions have been able to dissuade local districts from participating in the state-authorized programs. Only a handful of Florida districts participate in merit pay, for example, even though state funds cover the cost of the initiative.
A strong, well-designed merit-pay plan requires more than offering a bonus to high-performing teachers while paying the remainder according to the standard schedule. To be truly effective, pay for performance must mean in education what it does in other industries—salary increases for the successful, and salary reductions, even dismissals, for poor performers. State laws governing teacher tenure in most states make implementation of such plans unlikely.
All of this leads us to measured skepticism about the merit of merit pay, unless it is coupled with school choice innovations hefty enough to instigate sustained competition among schools and school sectors. Only then would local districts have the incentive to both lobby states for changes in state laws and to negotiate tough contracts with teacher unions. Only then would they find it important, if merely to retain their student enrollments, to structure their pay systems so as to attract top-notch employees and give them strong incentives to strive for excellent performances.
But we have covered a lot of ground very quickly. Let’s step back and consider carefully the propositions we have set forth.
Does It Work?
High-quality research on this topic within the United States is sparse and results are mixed. Matt Springer and his colleagues at Vanderbilt released a study recently on a well-designed randomized trial of a merit pay experiment in Nashville. The program involved bonuses of up to $15,000, which would presumably be large enough to affect individual incentives. Yet virtually no effect was seen on test scores (outside of 5th-grade math, an effect that disappeared for those same children the next year). That said, the Nashville study did not examine long-term effects on the composition of the teacher workforce.
The Bloomberg administration in New York City made headlines in late 2007 by announcing a pilot merit-pay initiative, the School-Wide Performance Bonus Program. The New York City Department of Education randomly assigned eligible schools to treatment or control groups, which has enabled scholars to conduct rigorous evaluations. Early results with respect to student achievement are not promising overall, although the program appears to have had a positive impact in schools with fewer teachers (see “Does Whole-School Performance Pay Improve Student Learning?” research). The researchers theorize that the group benefit feature of the merit pay program made it unlikely that it would have an impact on teacher behavior in any but the smallest schools.
The international evidence on performance pay is more encouraging, including a recent worldwide look that indicates that students learn more in countries with performance pay plans, all other known factors held constant. Ludger Woessman (see “Merit Pay International,” research) looked at 27 Organisation for Economic Cooperation and Development (OECD) countries and found that students in countries with some form of performance pay for teachers score about 25 percent of a standard deviation higher on the international math test than do their peers in countries without teacher performance pay.
If merit pay seems promising (and certainly not harmful), convincing tests of its performance are difficult to undertake within the United States, simply because merit pay proposals typically end up being blocked, co-opted, or diluted by established interests. Admittedly, it is not easy to identify the various instances where merit pay has been proposed but then blocked from enactment, and therefore we cannot provide an explicit enumeration. In all likelihood, most potential proposals are never articulated, simply because likely sponsors regard the cause as hopeless. When 96.5 percent of all districts rigidly follow a standard salary schedule, it takes an energetic and devoted innovator to brave the odds and try to break from tradition nonetheless.
Still, there are several telling examples of established interests blocking merit pay proposals. Governor Mitt Romney proposed merit pay in Massachusetts back in 2005–06, as part of an education budget that included tens of millions in new spending. That proposal went down to defeat; as the Lowell Sun reported, “the Massachusetts Teachers Association [MTA] and United Teachers of Lowell opposed the idea. Catherine Boudreau, president of the MTA, called teacher bonuses ‘inequitable and divisive.’”
Philadelphia tried to institute a pilot merit-pay program in 2000, but later ditched the initiative, “calling it too expensive, too difficult to administer, and a failure at giving teachers useful feedback” according to the Philadelphia Inquirer. Then, in 2006, Philadelphia received a $20.5 million grant from the U.S. government to develop a merit pay program. Said the Inquirer, “At the time, the federal grant was announced with much fanfare—the union would be the district’s partner, officials said, ensuring the plan would succeed where others failed. But the deal fell apart.” The local union abandoned the program in the face of a “surprise $180 million budget deficit,” and the district gave the money to charter schools instead.
Another example comes from Cincinnati. That city’s merit-pay plan proposed in 2002 was overwhelmingly voted down by teachers (1892 to 73), even though it did not base bonuses on student test scores. As Education Week noted, the plan “was based on an extensive evaluation system, which determines whether teachers advance in five career categories…. The evaluations entail multiple classroom observations by fellow teachers and administrators and portfolios that include logs of parent contacts, lesson plans, student work, and more.” Education Week quoted a former associate superintendent of the Cincinnati schools, who blamed the proposal’s failure on the fact that it “would have applied to nearly all teachers, rather than allowing veterans the choice of opting into the new system.”
In Alabama, the state’s “Race to the Top” application originally proposed merit pay and a “new salary schedule that would give more money to math, science and special-education teachers,” but that portion of the application was deleted, reported the Press-Register (Mobile), “after Alabama Education Association leader Paul Hubbert wrote state Superintendent Joe Morton a letter…opposing them”
If special interests fail to block a merit pay program, they may still be able to make it temporary. A Little Rock, Arkansas, performance-pay program lasted only three years and was not renewed by the local school board, despite evidence of positive effects on student achievement in math, reading, and language. Similarly, the Alaska School Performance Incentive Program was canceled after three years.
Special interests are also able to repeal merit pay based on putative budgetary constraints. The state of North Carolina suspended incentive awards to high-performing schools in 2008–09 due to budget problems. Winston-Salem/Forsyth County, North Carolina, suspended its bonus program due to budget difficulties as well.
Unions have been similarly successful at preventing local districts from participating in statewide programs, as the experience in Florida, Iowa, and Texas shows. Florida’s “Merit Award Program” provides state money to local school districts. According to the Florida Department of Education, “Each district will determine an amount equal to at least 5% and no more than 10% of that district’s average teacher salary to be awarded to all of the top performing personnel in the district, regardless of years of experience.” Even though this program involves free money from the state for districts to hand out to teachers, the political forces opposing merit pay were able to prevent 88 percent of Florida districts from participating in 2009.
Similarly, Iowa’s statewide Career Ladder and Pay-for-Performance grant program was passed in 2007, but only 3 Iowa districts, out of 360, bothered to apply. Only 20 percent of Texas districts opted into the District Awards for Teacher Excellence program in 2009–10. In other words, as a result of political opposition, the vast majority of school districts, even in conservative Texas, turned down extra money from the state rather than adopt merit pay.
Merit Pay in Name Only
When interest groups succeed in diluting or co-opting a merit pay plan, the plan ends up rewarding teachers mostly or entirely for inputs (e.g., professional development, graduate degrees, national certification) rather than for outputs (test scores, graduation rates, or even supervisor assessments).
One example is Arizona’s Classroom Site Fund (CSF), a mandatory statewide program that involved a couple of new taxes. Districts had to “allocate forty per cent of the monies for teacher compensation increases based on performance and employment related expenses, twenty per cent of the monies for teacher base salary increases and employment related expenses and forty per cent of the monies for maintenance and operation purposes.”
According to a 2010 report from the Arizona Auditor General, out of 222 districts receiving CSF funding, the auditor could identify only 29 “with strong performance pay plans that did a good job of linking teacher performance pay to student achievement.” The report noted that “allowing districts the freedom to determine performance pay goals can help gain district and teacher buy-in,” but that such freedom “has also led to inconsistent performance pay plans and to situations in which teachers receive similar performance pay for significantly different levels of effort and related performance results.”
One example from the auditor’s report deserves to be highlighted:
One district awarded performance pay to eligible employees if freshman students’ algebra test scores increased by at least 10 percent between a pre- and post-test. The actual increase in test scores was almost 90 percent. Since the pre-test is given to freshman students who have never been exposed to algebra and the post-test is given to them after receiving a full year of algebra instruction, it should be expected that scores would increase significantly more than 10 percent.
In other words, algebra teachers were being rewarded merely for getting students to learn 10 percent more about algebra than they knew before studying that subject at all. This is not a high hurdle to clear.
Denver’s ProComp program has been heralded as a political and policy success. Then Senator Barack Obama said, “Cities like Denver have already proven that by working with teachers, this can work, that we can find new ways to increase pay that are developed with teachers, not imposed on them and not just based on an arbitrary test score.” But the Denver ProComp program may be less than meets the eye. For one thing, it exempts teachers hired before January 1, 2006, from having to join, which means that the vast majority of teachers whose pay depends on seniority rather than on merit are able to keep their old pay structure in place. And if older teachers opt to enter the ProComp program, they keep their old base salary; the ProComp program merely offers them a chance for bonuses on top of that old salary.
The ProComp program also rewards the old definition of “merit” more immediately and to a greater extent than it does anything that improves student achievement. The largest monetary award is for earning a graduate degree: a $3,300 permanent salary increase plus a tuition or student loan subsidy of $1,000 per year for up to four years. By comparison, teachers receive a one-time award, not a bump up in base salary, of up to $2,403.26 if their students exceed “district expectations” for student growth.
Moreover, as Paul Teske, a principal evaluator of the ProComp program, noted in the Christian Science Monitor, bad teachers face no penalty under the ProComp or similar merit-pay programs: “I guess your salary stays low, and maybe that sends the message that you should look at another career. But ProComp doesn’t directly address that.”
The federal Teacher Incentive Fund (TIF) provides grants to school districts that promise to develop merit pay programs “for raising student achievement and for taking positions in high-need schools.” Currently, the Department of Education’s website lists 33 TIF grantees, including some small districts and a few major city districts. But these programs may also end up being diluted or co-opted.
For example, the TIF program in Charlotte-Mecklenburg (North Carolina) includes substantial bonuses for professional development, working at hard-to-staff schools or in hard-to-staff subjects, and for taking on leadership roles. To the extent the program involves student achievement, it bases awards on “student learning objectives” as “created by individual teachers, with the approval of site-based administrators”; these objectives “will be measured by a combination of existing assessment instruments, and teacher designed tools,” as well as by state standardized tests. The superintendent of Charlotte-Mecklenburg schools recently announced a plan to bring performance pay to the entire district.
Perhaps it is desirable to have teachers receive more professional development, work in hard-to-staff schools or subjects, and assume leadership roles, but these are inputs, not student outcomes. The bulk of the Charlotte-Mecklenburg TIF program, like many such programs, is MPINO—merit pay in name only.
Some locales have diluted the merit pay concept by making the bonuses to teachers small and setting the bar for receiving the bonuses low, thereby converting merit pay into something approximating an across-the board pay raise.
For example, the Texas Educator Excellence Grant (TEEG) program began in 2006–07 and ended after the 2008–09 year; it was funded at approximately $100 million per year. After analysts at the National Center on Performance Incentives (NCPI) reported no positive effects on student test scores, the Dallas Morning News declared the program a failure. NCPI report coauthor Lori Taylor speculated that “one possible cause of the program’s failure was that bonuses were relatively small and were given to most teachers at each school—about 70 percent—so that the incentive for individual teachers to push for higher scores was ‘relatively weak.’ In addition, campuses that qualified already had to be higher performers, so it was difficult to register much improvement.”
The same thing seems to be happening in Houston, where a merit pay program has existed since January 2007. The district announced financial awards totaling $40.4 million in 2010. The district’s webpage notes, “in all, 15,688 HISD employees received performance pay [in 2010], ranging from $25 to $15,530. That’s 88 percent of eligible HISD employees.”
Minnesota’s oft-heralded “Q Comp” program offers yet another example of a “merit pay” program that ends up as an across-the-board pay raise. As the Minneapolis Star Tribune recently reported, “In 22 school districts whose Q Comp practices were analyzed by the Star Tribune in 2009, more than 99 percent of teachers in the program received merit raises during the preceding school year. Only 27 of the roughly 4,200 teachers eligible did not get a pay raise.”
The New York City School-Wide Performance Bonus Program mentioned above may also have been undermined by its structure. Some 180 schools were eligible in the 2007–08 school year for a collective $14 million in bonuses, or $3,000 per union teacher, if they met test score goals established by the district. In a key factor that enabled the plan to draw union support, committees composed of a principal, a person of the principal’s choosing, and two union representatives were allowed to decide how the bonuses should be distributed at any given school. Researchers identified a number of drawbacks to the program design, including the possibility that bonuses based on school-wide improvements weaken the incentives for individual teachers to increase their efforts.
Making Merit Pay Work
The prospects for merit pay are not promising, despite both theoretical and empirical reasons for expecting that the programs would produce positive results for students. Our findings are consistent with the theory that school districts are not primarily educational institutions where policies are organized around maximizing student achievement. Instead, they are best understood, at least when it comes to compensation policies, as political entities shaped by powerful interest groups, including organized groups of employees.
Viewed in that light, it is unsurprising that public school systems have relatively little interest in authentic merit-pay programs. If some teachers could earn improvements in their wages and working conditions from their own efforts rather than from the efforts of their organized representatives or affiliated politicians, then more-effective teachers would have little reason to support the unions financially or politically. Their interests would be at odds with those of less-effective teachers. In short, the single salary schedule by which almost all public school teachers are paid is essential to the financial and political power of established interests.
One way to diminish the power of established interests and permit the adoption and implementation of merit pay is to expand choice and competition in education. If students choose their school, those schools have incentives to adopt and implement policies and practices that will improve their quality and attract students as well as the resources they generate. If merit pay systems help attract and motivate effective teachers, schools in a more competitive environment will have incentives to adopt those systems. They are more likely to design and maintain merit pay systems in a sensible way, since their revenue depends on it.
Schools that already compete for students appear more open to including merit pay in their personnel policies. According to University of Washington’s Daniel Goldhaber and his colleagues, charter schools are more likely than traditional public schools to use merit pay. Michael Podgursky, professor of economics at the University of Missouri, looked at data from the 1999–2000 Schools and Staffing Survey and found that when school administrators were asked whether they used salaries to reward “excellence,” only 6 percent of traditional public school administrators answered yes, while “the rates for charter (36 percent) and private schools (22 percent) were much higher.” Even those charter and private schools without a formal performance-pay plan are typically able to offer higher salaries to teachers they hope to retain and, as important, to readily dismiss teachers deemed ineffective.
Attaching continued employment and level of compensation to job performance is something that frequently occurs among private enterprises in competitive markets. The difficulty with merit pay in education is that it attempts to simulate a market-based practice in a nonmarket environment. None of the forces that cause organizations to seek effective merit pay systems, or to maintain and alter them effectively over time, exist in public education.
Imposing merit pay on an unwilling education system is like trying to get kids to eat their vegetables when the kids are 25 years old and stronger than their parents. No matter how nutritious green beans may be, powerful adults who don’t want to eat them can usually keep them off their plates and can almost always keep them out of their mouths.
Stuart Buck is a doctoral fellow in education reform at the University of Arkansas. Jay P. Greene is professor of education reform at the University of Arkansas and a fellow at the George W. Bush Institute.