October 3, 2010
Incentives to Teachers and Students
Incentives to Teachers and Students-Becker
The traditional public school system had a very weak incentive structure. Since students automatically went to neighborhood schools, schools did not have to compete for students. The teachers unions long ago eliminated merit pay, and made teachers' pay determined almost entirely by degrees and years spent teaching. Finally, in this traditional system, students had little incentive to work hard, especially when they did not expect to get much education, and could count on always getting promoted.
All this is beginning to change for the better, despite fierce opposition from teachers unions to virtually every important reform. School vouchers, and especially the charter school movement, is bringing competition to the traditional neighborhood school. No longer can public schools automatically have a captive audience of all the school age children in their neighborhood. Charter schools are expanding as rapidly as allowed by local and state restrictions that have been due mainly to lobbying of teachers unions.
Fortunately, restrictions on charter schools, and the even greater limits imposed on the scope of school vouchers, are breaking down, in large part because parents are becoming vocal, and are also voting with the feet of their children. In addition, careful evaluations of charter schools in randomized experiments that compare the performance on standardized tests of students in charter schools to performance by students in regular public schools generally show that charter school students do significantly better (see, for example, studies by Josh Angrist of MIT and co-authors of charter schools in and near Boston).
Some progress has also been made in finding ways to better motivate students, even in traditional public schools. A large and mainly well designed experiment of poor rural children in Mexico that began in the 1990s, originally called Progresa (now called Oportunidades), showed that even young children from uneducated families could be better motivated by financial incentives. Progresa gave monthly stipends to mothers of children in the program if they attended school regularly and did decently. School attendance did rise for families participating in this program, and students stayed in school longer.
Experiments are now ongoing in American public schools by my colleagues Steve Levitt and John List, along with Roland Fryer of Harvard, on motivating students by giving financial rewards directly to students rather than to parents. Although the studies are not yet completed, I would anticipate that students, especially older students, could be well motivated by monetary and other rewards for good performance.
Not surprisingly, teachers unions fight hardest against reforms that change the way teachers are paid, especially when they introduce incentives for teachers to perform more effectively. Teachers' pay should tend to depend on education and experience, but it should also be sensitive to their revealed effectiveness as teachers. Teachers unions have for many decades fought merit pay that allows principals and other administrators to determine "merit" by their evaluations of how well teachers perform. The unions have claimed that this approach to merit would give higher pay not to teachers who do the best teaching, but to teachers who ingratiate themselves into the good graces of the administrators. Perhaps that argument has some merit in the traditional school system where administrators of neighborhood schools have some monopoly power, but it is far less compelling as administrators of public schools increasingly have to compete for students by offering better education.
In any case, the call now is for merit pay based on more objective criteria, such as students' performance on standardized tests. Since students differ greatly in how well prepared they are for particular classes and subject matter, it is crucial to design a merit pay system that ties pay not to the absolute level of performance, but to the increment in performance added by different teachers. It may also be desirable to try to reduce "teaching to the test", whereby teachers only emphasize materials that are included in the tests used to determine merit pay, although teaching to the test is valuable if doing well on the "test" requires knowledge of the important principles.
I do not want to minimize the difficulty of getting a well-designed system of merit pay, and Posner discusses many such difficulties. However, my colleague, Derek Neal, has proposed an attractive system of merit pay, where student performance at the end of the year is compared to that of students who perform about as well at the beginning of the year. Essentially, teachers then get a bonus that depends on the percentile ranking of their students at year-end compared with the performance of the comparable students. Other education specialists have different ways to determine merit pay that may be worth considering.
The disgraceful reaction of the LA teachers union to publication by the LA Times of the database that gives performance scores of the students of 6000 elementary school teachers is indicative of how teachers unions feel toward rewarding better teachers. The support by Arne Duncan, the US Secretary of Education, of the newspaper's publication of this information is highly commendable. Not only should such information be published and publicized, but they should also be used to design a system where merit plays a sizable part in the monetary compensation of teachers.
The Value-Added Teacher Reform Program—Posner
There is widespread concern that elementary and secondary school education in the United States is deteriorating and is now inferior to that of many other countries, as measured for example by high-school graduation rates and college attendance rates. Proposals for reform fall into two main classes: increasing competition in the provision of educational services; improving the quality of public school teachers.
A reform aimed at improving the quality of public school teachers that has received a good deal of attention lately is the "value added" method of evaluating teachers' performance, now being used in the public schools of Los Angeles. See "Grading the Teachers: Value-Added Analysis," www.latimes.com/news/local/teachers-investigation/. The evaluation begins with determining the average improvement of a student, say between the end of third grade and the end of fourth grade, and then comparing the student's actual improvement, all as measured by performance on standardized tests. If his or her improvement is above the average improvements, the teacher is rewarded, for example with a bonus; if below, the teacher can be counseled in an effort to improve the teacher's performance. The Secretary of Education, Arne Duncan, supports value-added teacher evaluations (and informing parents of the teachers' scores); the teachers' unions oppose it as a step toward merit-based pay of public school teachers or even elimination of tenure.
The main objection to the program is that the value added by a teacher can't really be measured. The reason is that much may influence a student's performance (including his year-to-year improvement) besides the teacher, including fellow students, conditions at home, and the student's intelligence and application. These other factors could in principle be controlled for, but actually to do so would probably strain the ability of public school bureaucracies to devise and administer sophisticated statistical measurements. The alternative would be to assume that differences across students average out. But unless classes are very large and students are assigned to teachers randomly, differences in average performance are unlikely to be statistically robust.
The value-added methodology is, moreover, very difficult to apply beyond elementary school. When students have more than one teacher at a time their progression from year to year is the result of a team effort, and it is difficult to identify the contribution of each teacher. Even if improvement (or lack thereof) is measured on a subject-by-subject basis, the existence of complementarities between subjects (math and science, for example, or history and social studies) means that a teacher in one subject can influence student performance in another. And the complementaries can be subtle: an excellent English teacher may inspire her students with enthusiasm for school in general, stimulating them to improve their academic performance in unrelated courses.
Even with all these difficulties acknowledged, the granting of bonuses to teachers who receive above-average value-added evaluations would have some good effect on teachers' motivation. But of course the money has to come from somewhere, and the benefit may not equal the cost. It is doubtful, moreover, that value-added evaluations, even when publicized (as the Los Angeles Times did recently with the L.A. public schoolteacher evaluations, causing a good deal of commotion and, it seems, the suicide of one teacher), have much effect on bad teachers, either by causing them to improve or by easing them out of the system. The methodology is too crude (and likely to remain so) to provide a solid basis for censure, self-criticism, instituting a system of merit pay, or ending teacher tenure. Tenure has of course bad effects, whether it's tenure in public or private schools, in public or private universities, in the federal judiciary, or unionized workplaces: it encourages slacking off, selects for people who have a high degree of leisure preference, and leads to retention of poor performers. At the same time, however, it is a form of compensation valued by many; were it eliminated in public schools, the schools would have to pay higher, maybe much higher, salaries, which hardly seems feasible in today's economic climate (which is likely to be tomorrow's too).
So value-added evaluation of public schoolteachers, while ingenious (despite its limitations) and growing in popularity, does not seem to be the answer, or even a major part of the answer, to dissatisfaction with American education. Competition is more promising. Two forms should be distinguished: charter schools, which are public schools (that is, publicly financed and tuition free) that are however managed outside the normal public school system in order to enable and encourage experimentation; and means-tested vouchers, which are scholarships that a student can use to attend a private (including parochial) school. About a million students are enrolled in charter schools, and 200,000 other students receive vouchers enabling them to attend private schools. Home schooling is another alternative to public schools, and an important form of competitive education, but it is not feasible for students from poor homes because their parents (often just a mother) rarely have enough education for them to able to teach their children.
The charter schools have turned out to be a mixed bag. There is excess demand for them, which is some evidence that they are superior to public schools. But studies of drop-out rates and other measures of quality indicate that while some charter schools are indeed better than public schools, many are worse; there is as yet no convincing evidence that on balance they are superior to public schools. There is more evidence that vouchers improve educational performance, though it is not conclusive. See "Is School Choice Enough?" www.city-journal.org/2008/forum0124.html. Vouchers enable poor students to attend established schools with a proven record of quality (many of these are Catholic parochial schools), so it is not surprising that they are more effective in improving academic performance than conversion of existing public schools to quasi-private status, which is the character of the charter-school movement.
Teachers' unions are more fiercely opposed to vouchers than to charter schools, which is a vote in favor of vouchers! Private schools have greater freedom from regulation and are less likely to be unionized than charter schools (although charter schools, too, are generally not required to bargain collectively with their teachers), and they are numerous and established and can expand to accommodate increased demand.
I favor vouchers, but they are no panacea. Obviously, basic education is an important social good. But even bad schools provide that. How much value good schools can add to the skills and knowledge of students who now attend bad schools is uncertain. Maybe most students who attend bad schools have limited aptitude and motivation because of low IQ, poor physical or mental health, peer-group pressures, a bad family environment, or effects of popular culture. How far such impediments to academic performance can be remedied by teachers, however skilled, and at what cost, is unclear to me.