Incentives: The fundamental problem in education: an interview with Eric A. Hanushek

Published January 1, 2000

“The fundamental problem in education is that there aren’t any incentives to increase student performance. Nobody’s career is really dependent upon the children doing well.”

Some of the questions that economist Eric A. Hanushek raises about public policy initiatives for improving student achievement bring to mind the Mullah Nasrudin and his story about a group of people who were searching under a street lamp, trying to help a confused man find his key. When the Mullah asked the man where he thought his key might be, the man pointed to a dark area away from the street lamp. “Then why are you looking here?” asked the Mullah. “Because the light is better here,” said the man.

Currently Professor of Economics and of Public Policy at the University of Rochester, Hanushek has been asking tough questions about the effect of school resources on student performance since the 1960s, when he did research for his Ph.D. at MIT on issues raised by the first Coleman Report. A Distinguished Graduate of the United States Air Force Academy, Hanushek had academic appointments at the U.S. Air Force Academy and Yale University before joining the Rochester faculty in 1978.

Hanushek has published numerous articles in professional journals and is the author of several books on topics ranging from public policy on education and retirement income to statistical methods and information-gathering for decision-makers.

A frequent source of testimony on education issues for Congressional committees, Hanushek was a senior economist at the Cost of Living Council during 1973-74 and served as deputy director of the Congressional Budget Office in 1983-1985. He was president of the Association for Public Policy Analysis and Management in 1988-89. In 1997, he was selected to be a member of the International Academy of Education. Hanushek spoke recently with School Reform News Managing Editor George Clowes.

Clowes: How did you become involved in research in education?

Hanushek: The first Coleman Report on the impact of schools on student performance came out in 1966 and caused quite an uproar. There was a large seminar organized by Pat Moynihan and Fred Mosteller at Harvard to try to figure out what this report said and how to interpret it. I got invited to participate in that seminar, and that led to me doing a thesis on the effects of school resources and other things on student performance. It was an exciting time.

No one quite believed the Coleman Report. In fact, I don’t think it was interpreted correctly for a long time. The common interpretation of the Coleman Report was that schools don’t matter. I think that the right interpretation–that has held up subsequently with lots of later work–is that the measured aspects of schools don’t seem to be systematically related to performance, but that there are huge differences across schools.

Clowes: So we’re not measuring the resources that do account for these differences?

Hanushek: Right. It’s how you measure features in schools. What we found is that what makes a big difference–as all parents know–is which teacher your child gets. But it’s just that you can’t predict who’s going to be a good teacher by looking at whether they’ve got a master’s degree, or if they’re experienced, or what have you. That’s what’s come out of all the subsequent work.

What I’ve come to believe pretty strongly is that the most important aspect is the teacher and teacher quality. This comes partly from some of the more recent work I’ve been doing, which has been concentrating on the basic question, “What determines student achievement?” or “How do teacher differences, teacher salaries, and other factors affect student achievement?”

What we’re finding is that there’s a lot of heterogeneity of teachers within any one given school. That means it doesn’t appear as if it’s all a matter of having good and bad principals, with good principals collecting a set of good teachers and bad principals collecting a set of bad teachers. It seems to be much, much more complicated than that, and it doesn’t seem to be systematic at all.

We’ve been doing a large study of school performance in Texas, where they have annual testing of all students in basic subjects. We’ve been able to follow the performance of individual students as they progress through fourth to seventh grade, and of different cohorts as they go through the same grades in the same school. That’s allowed us to sort out a lot of these questions.

One of our fundamental findings is that there are huge differences in teacher quality, even within individual schools. When we look at the measured attributes of schools, we find a small class-size effect in the earlier grades–in fourth grade–but not in later grades. That class-size effect is much, much, much smaller than the variation in teacher quality. It was just dwarfed by teacher quality differences.

We also find that there is an impact of having a first-year teacher. The first year of teaching is kind of rocky, but after that, teacher experience doesn’t seem to make much difference. Whether teachers have masters degrees or not doesn’t make any systematic differences in performance, either.

Clowes: So a teacher’s first year of teaching is critical for the students?

Hanushek: Yes. That comes into play with class-size reduction programs like the one in California. What they did there was to try to hire a whole lot of new teachers immediately. The richer, wealthier suburban districts hired experienced teachers from the inner-city schools, and the inner-city schools went out and hired all inexperienced teachers. And so you can infer that it’s likely that inner-city children got hurt by the class-size reduction program in California–which is just the opposite of what people want to tell you.

Clowes: One of those unintended consequences that come about when people don’t think through what they’re doing.

Hanushek: Exactly. We also made some effort to look at special education programs and the benefits of special education. What we find is that special education programs seem to lead to higher achievement by children who are in special education.

Moreover, what we find is that having more special education children in a school seems to help the regular education children, too. Now, whether they use special education to sort out and segregate some children that are disruptive, or whether it’s something about the resources that are provided, we don’t know. But regular education children seem to be helped–at least according to the way that Texas funds special education and organizes special education.

Clowes: So you need to look at how special education is organized as well as the way it’s funded?

Hanushek: Right. We don’t have all of the details. We’re just guessing on that.

We also look at differences in teacher salaries across districts, and we ask: What happens if one district pays more than other districts for teachers? We find that there is some effect of salaries on movement, but it’s much less than the effect of the student body characteristics on movements.

Basically, teachers are moving to teach higher-achieving, higher-income white children. Teachers seem to be following the characteristics of the student body more than following higher salaries. As far as we can tell, teachers are not looking for better working conditions in terms of smaller classes and things like that, which don’t seem to affect moving.

The other thing we find is that salary differences aren’t very related to student achievement. What that says is that if you raise salaries you can in fact usually get a larger pool of teachers to choose from, but that schools don’t systematically choose the better teachers.

Clowes: Is that because of the teacher pay structure?

Hanushek: It’s because of teacher compensation and the structure of management in schools–neither is really paying attention to student performance.

Then we look at another question: What about students moving across schools? What we find is that every time a student moves, that student takes an achievement loss. That includes the normal structural moves in schools, like moving from sixth grade to seventh grade and changing from elementary school to junior high. When you do that, you lose ground. We find that schools with more turnover in them hurt the students that don’t move. So if you’re in a high turnover school, that hurts you, even though you don’t move.

Now, some moves are voluntary and some are somewhat involuntary. Some people have to move for economic reasons, financial circumstances, job loss, or whatever. But if somebody moves to a different district at a time when not everybody is moving–so it looks like a voluntary move–those children tend to make up for any achievement loss within the next year. Those moves look like moves to get to better schools.

What really hurts is if you move within the same district. If you’re churning around in the same district, then you’re in trouble.

Clowes The President has said that our tax dollars shouldn’t go to support what doesn’t work. Since he’s currently forcing Congress to support a major effort to reduce class sizes, what evidence is there that this is a policy that does work?

Hanushek: I don’t think there’s much evidence. The one thing we have is some of the results out of Project STAR, but I think Project STAR has not been interpreted correctly for policy purposes.

There are two aspects. One is the way they implemented the whole experiment. There’s a lot of uncertainty there: They lose a lot of children, they have children changing across treatment groups, they don’t test a lot of the children, and they have new children coming in all the time. Less than half of the children who start the experiment in kindergarten are still there in third grade.

But if you ignore all of that, what you find is that it appears that if you’re in a small kindergarten or maybe a small first grade, your achievement is higher than if you’re in a large kindergarten or possibly first grade. But after that, there’s no added gain in achievement over time from being in a small class. What that seems to be saying is that there’s a certain one-time effect of being in small classes if it’s done early. That doesn’t support policies of reducing the size of sixth-grade classes and things like that.

If you look at the way that the California class-size reduction program was implemented, they started with smaller second-grade classes, and then maybe first-grade, and then third-grade, and so on. Even if you took STAR literally, that doesn’t seem to be the way to do it: It’s starting on grades that don’t seem appropriate from the STAR evidence.

But balanced against the results of Project STAR is a lot of econometric evidence, international evidence, even national aggregate evidence that says that reducing class size is unlikely to improve student performance. You have one result from Project STAR but you also have a lot of other evidence on the effects of class size that has developed over the years. For example, we’ve had reductions in class size over the last 25 years but we’ve had essentially no change in student achievement.

As far as I can tell, a large number of educators were surprised at this whole push on class size because they didn’t believe that was the way to go. Now it’s become a very political thing. It looks much more like a political issue about hiring teachers than a student achievement issue.

Clowes: If we hire more teachers to do the same amount of work, wouldn’t teacher productivity go down?

Hanushek: That’s another complicated issue. Education output has essentially been constant for 25 years, as far as we can tell from the data from the National Assessment of Educational Progress. So what you’ve had is roughly constant performance, but you’ve had rapidly increasing expenditures. From that, it looks as if productivity is going down, and it looks as if it’s going down more than in other service sectors where you think that they have the same problems of labor intensity and so forth.

But it’s not entirely surprising that we don’t see student performance going up when we spend more because none of our policies and organization in schools pays attention to output or performance.

The fundamental problem in education is that there aren’t any incentives to increase student performance. Nobody’s career is really dependent upon the children doing well. Pay, hiring, and everything else is essentially independent of how well somebody does teaching and how well the school does at increasing student performance.

Clowes: Would spending more money improve performance?

Hanushek: In my view, there’s not a very close relationship between spending and student performance and so we have no way of saying what we can expect if we put any amount of spending into a school. We know that some schools will spend money well, we know that others will spend it badly, and we know that on average, we don’t get much effect.

Clowes: What other strategies should policy-makers pursue to improve student achievement?

Hanushek: In my opinion, we have to do better at making sure we have high-quality teachers in all classrooms. The way you’re going to do that, I think, is having incentives in the system that relate to student performance.

Basically, if you want to improve student performance, you have to pay attention to student performance. Right now, we say “We want to improve student performance,” but then we pay attention to all kinds of other things but not to student performance.

We have to get better incentives in schools to increase student performance. Now, there’s an intermediate position that says along the way you also have to do much better at measuring the value-added by schools and teachers. We’re not very good at doing that.