What do you think? Leave a respectful comment.

New Study Gauges Teachers Impact on Students’ Lifetime Earnings

Replacing a bad teacher with an average or a good one has measurable economic benefits such as boosting a student's lifetime earnings by hundreds of thousands of dollars, according to a new study done in part by Harvard University economist Raj Chetty. Ray Suarez and Chetty discuss the study's findings.

Read the Full Transcript


    And finally tonight, putting a price on the value of good teachers. A large and new study addresses just that.

    Ray Suarez has the story.


    The debate over testing in schools, and whether students' scores adequately reflect a teacher's performance, has been raging for well over a decade. Now a new study has tracked more than two-and-a-half million students over two decades.

    It found test scores are indeed a good gauge for evaluating student performance. And the study found replacing a bad teacher with an average or a good one can translate into a huge economic difference. Combined, the students could earn hundreds of thousands of dollars more over their working lifetimes.

    We look at the study and the response it's stirred with Harvard economist Raj Chetty, one of its three authors. And we hope to be joined by Randi Weingarten, president of the American Federation of Teachers, the second largest teachers union.

    Professor Chetty, how do you measure student performance over such a long period of time? Some of the people in your study are in their late 20s now. And how do you determine it's the teachers who made the difference?

  • RAJ CHETTY, Harvard University:

    The way we do that is, we track these students from elementary school, so we know what teachers they had in elementary school, and then we look at how they're doing as adults.

    And the way we measure teacher quality is look at the average impacts that teachers had on the students' test scores, the students that they were teaching. So, for example, a teacher who systematically raised her students' test scores year after year, we would call that teacher a high-value-added teacher.

    And a teacher who didn't do as well might be an average- or low-value-added teacher. And so the question we then ask is, suppose you, by chance, were assigned to a high-value-added teacher when you were in fourth grade, and I was less lucky and only got an average-value-added teacher. Are we doing differently as adults? Are we more likely to go to college if we have a higher value-added teacher? Are you more likely to earn a higher salary and so on?

    And we find remarkable impacts. Students who were assigned to high-value-added teachers earn more. They're more likely to go to college. They're less likely to have a teenage pregnancy, and they do better on a number of dimensions that we're able to study.


    But there are so many variables involved. How do you exclude things like the level of education in the home, the overall achievement of an individual school, what the individual student was doing before they got to that grade and got a standardized test?


    That's an excellent question.

    In fact, what makes this topic so difficult to study and has been a real challenge in this area, as in the social sciences more generally. So, our approach in this study is to use sort of random variation in the set of teachers who happen to be around when you arrive in a school.

    So let me give an example. Suppose, Ray, that you are a year older than me and we both are in the same school and you arrive in fourth grade in 1995, when Mrs. Smith, who let's say is a high-value-added teacher, is teaching that grade in that school.

    And you have Mrs. Smith and you end up doing well and scoring high on your tests. Now I come along in 1996, and let's says Mrs. Smith has gone on maternity leave or is not around for some reason. There is a lot of teacher-switching that we see. Now I don't have the luck to be taught by Mrs. Smith.

    And so what we find in the data is that such students are — they score lower and then they don't do as well as adults. So the type of variation that I'm describing there is basically random, because the students who happen to get to grade four when a good teacher was there vs. the next year, when a good teacher might not be there, that is essentially random variation. So, those students are comparable, and you are able to control for all those factors that you mentioned.


    If we take your study at face value, do its conclusions risk flattening, simplifying what it is that makes a good teacher, limiting it to achievement on standardized tests?


    I think the main message of our study is that standardized test score impacts can be a useful input into evaluating teachers, but by no means are we saying that test scores are the end-all and be-all of how teachers should be evaluated.

    We think that they're one aspect of what should factor into the formula. One would also want to use things like principal evaluations or maybe even student evaluations or other measures of teacher quality. But I think there's some useful data here that could be very useful in improving teacher quality.


    Unfortunately, Randi Weingarten, of the AFT is stuck in New York City traffic, always an occupational hazard at this time of day.

    But one thing that teacher unions generally have been keeping an eye on is school systems' attempts to pay for performance, to try to create a formula around their students' achievement that would reward them with cash. After your study, is that justified, do you think?


    I think teachers who are high-value-added and are raising test scores are having tremendous benefits for their students.

    So, for example, a teacher who is in the top 5 percent, an excellent teacher, we calculate generates about $250,000 or more of additional earnings for their students over their lives in a single classroom of about 28 students.

    Now, I think it makes sense to try to reward teachers who are doing extremely well. They're providing a great service to the economy. I think it's important to recognize teachers who are having such great impacts. Whether the best policy to raise teacher value added is merit pay or better teacher training or some other sort of tool is less clear.


    You know, a lot of the schools with the worst performance in national standardized tests are also schools in poor neighborhoods, where you also have high teacher turnover.

    If you identify high-value-added teachers, as you call them, how do you get them to stay in a school that needs a lot of help?


    That's a great question.

    High turnover is indeed a problem, because one of the things that we see in the data is that teachers' value-added grows as they become more experienced. So the first time a teacher is teaching, it's quite natural they're learning on the job and they don't do quite as well as after they have a few years of experience.

    And so in these lower-income neighborhoods, where you have a lot of teacher turnover, that's a further reason that those children are not getting as great opportunities as we think they should be.

    How you reduce turnover, I think, paying teachers bonuses, especially if they are doing very well, possibly increasing teacher salaries, providing more support so that the classroom environment in which they're teaching is more constructive, easier to teach in, I think all of these things could potentially be very helpful.


    Is — could it also be concluded from your study that it ought to be easier to fire ineffective teachers? And I'm really sorry the union leader isn't here with us right now when I'm asking this question.

    But is that part of your conclusion?



    I think — you know, let me make an analogy here. Suppose you are managing a baseball team, say, the Boston Red Sox, and you're trying to do as well as you can. You have players with different batting averages. One approach you might take is to bring the hitting coach out and try to raise the batting averages of the players you have.

    But I think it also makes a lot of sense — and this will make sense to sports fans — that, on occasion, you might decide to let some of the players with lower batting averages go, and try to get somebody else who might do better. And so I think it makes sense to use a combination of those tools.

    Here, I think the stakes are even much bigger. We're talking about the future of our children, rather than winning a baseball game. So I think it does make sense to consider those policies seriously.


    Professor Chetty, thanks for joining us.


    Thank you. My pleasure.