Rick Hanushek: What Schools Get Wrong

Ben Zweig: The US spends more per student than almost any country in the world, but results have been underwhelming and getting worse. My guest today has spent 50 years trying to understand why. Rick Hanushek is the Senior Fellow at the Hoover Institution at Stanford and one of the founding figures of the Economics of Education. Here's our talk. Welcome to the Economics of Work. I'm excited to welcome Rick Hanushek, Senior Fellow at the Hoover Institution at Stanford. And one of, if not the pioneer of the Economics of Education, really inventing the field as we understand it today. Rick, welcome to the show.

Rick Hanushek: Well, thanks for having me. Obviously, my work in education relates quite directly to the labor markets that you're interested in.

Ben Zweig: Right, I mean, that's the pathway.

00:43 Where does the value of education come from?

Ben Zweig: Yeah, I mean, education is like the breeding ground of workers, I guess. But yeah, that kind of brings me to a question. How much of education is really vocational? And I guess the broader question behind that is that, I think of education as giving us a lot of things, because knowledge, skills, assessment, networks, what do you think the bulk of the value comes from? Do you have a mental model of where the value of education comes from?

Rick Hanushek: Well, my simplest model that I think is most important is that more education makes people adaptable. So that's something that's really important today because we see labor markets changing at lightning speed and people have to adapt. And the people who adapt are the ones with better education because they're sort of trained to think about different worlds. And so that's my basic model. It's not that education makes me better at pounding nails into things or doing other routine tasks. Those are all kind of trainable on the job, but the education that makes people be able to adjust to new circumstances, I think is what counts.

Ben Zweig: Interesting. So it's more the learning how to learn model rather than the knowledge that you obtain.

Rick Hanushek: Yeah, precisely. Precisely. And that's become much more important today. I mean, I can look up almost any fact I want now without even thinking about how to define it very well.

02:32 Learning useful things vs. learning how to learn

Ben Zweig: Yeah. I mean, when I think about education, especially early childhood education, I feel like there's so much that we learn that feels completely useless. I mean, I think to kids, everything feels useless. But even looking back, there's some things that feel really useless, like learning cursive or state capitals or what it's just things that I'm just like, we can't think of anything like better and more useful to learn. And I don't know, I wonder if that's just fine as long as we're learning something. Does it, I mean, I don't know. Yeah. Do you think that what matters is that we are just learning something or do you think we should be focusing our education on things that teach us how to learn, but also teach us useful stuff?

Rick Hanushek: Well, we're going to learn useful stuff. There's a big movement now to move more toward career and technical education, which I'm a little bit skeptical about.

Ben Zweig: But even in primary and secondary or just in higher ed being more vocational?

Rick Hanushek: Well, it's in primary and secondary and largely in high schools. And maybe we don't need to go to college. Maybe if we become good welders, that's going to be a good circumstance. And I'm a little bit skeptical about taking it to that extent. On the other hand, I view education as rather a progression. We start out and we learn how to divide fractions and so forth, things that people seldom actually do. Then we get into geometry and learning about arc tangents and all kinds of things that I couldn't even do today if I wanted to sit down and do. But it sets me up to start specializing in learning fields. And there's a certain knowledge that's required. If I want to become a world class chemist, I have to know what chemistry is today. And I have to be able to understand it, not just look it up. I have to understand it if I'm going to try to think about where it might go in the future.

Ben Zweig: Yeah. So it kind of makes me wonder if we are learning stuff that's generally useful. There's types of information or types of things that we study that build on other things like the long division leads us to arc tangents and all this weird trigonometry that maybe we never use. But we end up getting deeper and deeper and deeper. And then there are other things where maybe every 10th grader should have a welding class that could just be taught to anyone and we could go very broad and just teach a variety of different skills. My intuition is that there's something valuable about going deep into something and pushing at the frontiers of something. But at the same time, there's something a little dangerous about it because then if someone falls behind, they have a harder time catching up. So how do you think about those trade offs to the extent that they are trade offs?

Rick Hanushek: Well I think in the micro version, they are trade offs. I don't think in the macro version that they are in some sense. We're sort of following a general path. We're learning a variety of things. And at some point, people aren't as well prepared for the next step. And so they probably shouldn't specialize or continue along those lines.

Ben Zweig: Well, but if everything we learned were just kind of something that didn't have prerequisites, then falling behind wouldn't really be a phenomenon, right?

Rick Hanushek: Yeah. Well, it depends what you mean by prerequisites. If I require that somebody taking my advanced public finance course in economics and basic microeconomics, that seems to make sense. If I require them to have organic chemistry, it probably wouldn't. And so it's things that build upon other things. It's logical to make people follow a progression. There are other things that don't really build on other things that we've done. I don't know. I'm not a welder, but I think welding doesn't really build upon a lot of other things before it. You learn the skills that you need.

Ben Zweig: Yeah. I mean, I think you could kind of make the same case for more academic subjects. I think of even statistics or psychology or intro to philosophy or something that you only kind of learn in college that you don't really need prerequisites for. I don't really think anyone needs calculus before they learn statistics. You could teach statistics to eighth graders, I think. But we don't. It seems like, I don't know, maybe like a missed opportunity.

Rick Hanushek: So I'm not going to defend our current educational progression. We've got a lot of history there that people are trying to figure out how much of the history should we keep and how much should we change. And that's become so upfront because you can't have a conversation about education or the workforce without saying AI. And so that leads people to question what do we need in broad preparation. But I guess I'm arguing that if you take my view that fundamentally what we want people to do is being able to adapt to future things, then it's not so much what field you take or what the particular prescribed preparation is. It's just that you've done something and you've started working through things that you hadn't done before and trying to figure out how you take something you haven't done before and solve new problems.

Ben Zweig: Yeah. Yeah. I think that really tracks. I mean, just as a little bit of a wonky example. So I went to a Jewish high school and we learned Talmud. And Talmud is the most useless content you could imagine. I mean, it's like learning how to sacrifice animals for a temple that was destroyed 2000 years ago. It's like really couldn't be more irrelevant to anything. But it is just like so in the weeds of the minutia of how these decisions came to be. And I think people that get into it feel like, at least anecdotally, feel like it has helped them think. And they end up being at least very good at very good lawyers because it's like all about legal reasoning. But even just thinking beyond that. So I think that's like the most extreme example I could think of something that we learn to learn how to learn.

Rick Hanushek: Yeah. No, no. I think that's absolutely right. And again, as in all things, there's got to be balance. I mean, there's a state of Israel is trying to deal with the balance of how much learning the Talmud versus other things is. And there are people who have strong interests and preferences along these lines that are willing to fight over what their view of the right balance is. There's nothing I think that's easily solved that I can tell you the balance of all these things.

10:48 How AI changes the returns to education

Ben Zweig: Yeah. So you mentioned AIs entering the conversation of education in a lot. In what ways do you think it is meaningful for the future of education? And in what ways is it not? Like is it mostly useful or dangerous for assessment or for actually like educating or? I don't know. I don't know. I mean, what do you think is interesting there?

Rick Hanushek: Well, I think of AI in two different ways. One is how can artificial intelligence enhance the learning experience in schools and make that better? But I also think of how schools then interact with the labor force and how AI is going to affect that. And so let me start with the latter first. My simplistic view of AI in the labor market is that AI is going to complement the activities of the better of the higher skilled, the better people. And it's going to substitute for low skilled people in large sense. Now that's a big generalization that with all your data, you're going to object to various parts of that generalization. But to me, that overviews says very simply that the return to strong basic education is going to go up because the more skilled people are going to be able to do better and the less skilled people are not. And so that's going to make the value of getting the skills that allow you to use AI effectively more valuable. So I think of that as sort of motivation for why we've got to work hard to get our schools working better, which in the US aren't working all that well right now, in my opinion. Yeah.

13:07 Why don't more Americans go to college?

Ben Zweig: Yeah, I could totally buy this line of reasoning that AI increases the returns to education. But the returns to education are already quite high. So why are we still behind where we should be? I mean, do we need more motivation?

Rick Hanushek: Yeah, no. This is a big question here of why I think people aren't going to college more as opposed to there's this current movement of going less. The value of college has seemed to be high and continues to be high for a long time. And yet there's a lot of people who are arguing, well, maybe I won't get anything out of college. Maybe I'll just do something else. So that's actually a puzzle in my mind of why college attendance rates aren't higher, the sort of attendance rates you see in Korea, which are much higher than they are here.

Ben Zweig: What are the attendance rates in Korea?

Rick Hanushek: Well, I mean, like 95% of the people do some post-secondary education.

Ben Zweig: Oh, well. And what is it in the US?

Rick Hanushek: The US is sort of two-thirds or so enter college and a bunch drop out. And so for a cohort, it's going to be less. There's a lot of people doing it other than pure college activities. But it's probably-- I don't know. I've tried to figure this out and there's some definitional things. Probably three-quarters are doing something post-secondary education for a little bit after high school. But then a lot stop pretty quickly.

Ben Zweig: Yeah.

15:05 AI and EdTech in the classroom

Rick Hanushek: Anyway, the other half, I was just going to say, of AI is how does AI affect what goes on in education? And there I think it's really up in the air. We don't know. I believe in simple answers, I guess. My answer is still that the better teachers will be helped by using various AI devices. And the poorer teachers will find AI as sort of substituting for what they do and not helping them. But there's a rush of people that figure out how does AI fit in. I think probably a vast majority, if not all states today, have a commission to look at how AI should fit into schools. And I don't think that answer is completely clear yet that we don't know how best to use AI. Which is sort of, in my mind, an extension of EdTech too. When you live in Silicon Valley, as I do, you run across people that are doing amazingly exciting things with EdTech for education. I mean, they're just wild and exciting. And yet it doesn't capture the schools of the country. You find some schools introducing various EdTech, but we don't see that it's dramatically changing any education around the country.

Ben Zweig: So why is that? I mean, is there just inertia or just any technology or any change in how education gets done? Yeah, why do we think that education doesn't evolve and adapt more quickly?

Rick Hanushek: There is a lot of inertia that comes from teachers doing what they want to do as they've done for the last 10 years. There is inertia that comes from union pressures to standardize everything. There's pressure that comes from school administrators who like what they're doing and don't want to see much change. And so there's, at some level, there's this inertia. But what I've told a bunch of EdTech firms, and they never listen to me, is that you have to start thinking about who benefits from your product and who doesn't. And instead of selling your app or your technology to Los Angeles Unified School District and having them force it down everybody's throat, you have to be thinking more about allowing for variations in how people use the products, which leads to the standard economist answer that nobody listens to either, is that you ought to provide incentives for performance, not try to tell them what to do. And so those who can use various EdTech or AI products effectively would gravitate to them. And those that can't use them effectively might do more traditional things, but you judge what they're doing in terms of whether students are learning or not, which is such a radical idea in education. The rest of the labor market sort of understands some of these ideas, but education doesn't.

19:16 Could AI standardize testing?

Ben Zweig: Yeah, it's such a puzzle. I mean, I think, as measuring relative performance is hard. You need some form of standardization. So let me hit you with an idea that, of course, I'm not going to pursue, but anyone listening can pursue if they think it's compelling, if you think it's compelling. So you can be the judge on this idea. One thing that I feel like would be a very nice use of AI in education is kind of standardizing assessment. So right now we have some standardized exams. We have the PISA exams, we have AP exams. There's like a handful, but most exams given administered to students are just the teacher writes it, the teacher grades it, and it gets filed away somewhere and no one sees it. But if that were somehow digitized, let's hand wave our way past that problem and give it to AI. I would think AI would be very good at finding questions in one teacher's exam in school A and finding a basically synonymous question in a different teacher in school B. And being able to kind of assess a student in one school with one set of questions and a student in a totally different school with a different set of questions and basically doing the standardization through its own intelligence. So basically giving us the value of standardized exams when in reality every exam is just kind of written in an idiosyncratic way. What do you think? Would that fundamentally change the way that we measure the value of education and who learns what and which teachers are good and so on and so forth?

Rick Hanushek: That's a really tough problem, of course, because what you would then have to do is try to have some design that involves waiting to see whether kids were equivalent, who knew answers, questions in one class to another, could do the same things subsequently because it's not whether they can solve a particular problem. It's their subsequent value of this. I thought you were going in a different direction, which is the way I've seen AI, and that is AI can in fact write exams very easily. And grade exams. So right now we have standardized tests that each state is required to test kids in grades 3 through 8 in math and reading and so forth by accountability standards, and then we have the National Assessment of Educational Progress that tries to measure kids learning across different states with a common test. These tests, as they've been designed, are extraordinarily expensive to produce. And you're talking about $100,000 to write a new question.

Ben Zweig: Really? Why is that?

Rick Hanushek: You have to pretest them. I don't know. I was on the board that oversaw the National Assessment and kept arguing against this for a long time that because my answer to your assessment problem is a little bit different, and that is that we have 2000 fourth grade math questions that we think cover the substance of what we want people to know at the end of fourth grade. We make all those questions public and let people see the questions that are there and refine them for ones that aren't very good. And then you test people with a random selection out of the 2000 questions. And you could very easily, I think, sort of get teachers focused on the right thing. It would never make sense in a world like that to try to memorize the answers to all 2000 questions. You would want to learn the concepts if you're going to do well.

Ben Zweig: Yeah.

Rick Hanushek: So that's my proposal. It ends cheating problems and all this. You give everybody a random selection. This is how we test pilots. The pilot test has a practical test of whether you can take off and land planes. And it also has a written test that is 763 questions that are publicly known that you get a random selection of 50 of them when you come in to take the test. And if you came in on a different day, you get different 50 and so forth.

Ben Zweig: You could even probably do better than random by this more adaptive exam where what you know passes you on to these like...

Rick Hanushek: What's the other... You're getting into another area where I've been having problems understanding why we don't do things. Adaptive testing obviously. But here's where AI could... I think if we sat down with one of the available AI programs right now, we could generate a quick 150 fourth grade math questions in a standardized form without any trouble. And they would match those that are currently given on the national assessment for fourth grade math. We could train AI this morning. In fact, maybe we should stop this discussion this morning and start our testing program. But here's where AI obviously enters. The problem that education has been considering is not that problem. It's the opposite problem. How do we test students if they have access to AI? Because we ask a question and somebody goes to their Apple Watch and gets the answer off their Apple Watch. And so that's the only issue that is being discussed with great energy today.

Ben Zweig: Interesting. I feel like that's kind of an obvious one. I don't know, maybe I'm missing it, but I feel like you just can't evaluate with internet in the room. You need old school blue books and that's that and handwriting. But I don't know, is there any other position to have on it?

Rick Hanushek: Or you give them a Chromebook that's disconnected from the internet. I don't actually have to do the cursive that you and I learned to do in school. But that's the obvious thing, what you do. And there's a question about whether there is such a thing as a take-home exam anymore or how you do that. So there's some adjustments that have to be made. But that's what people are spending more time on, then how can we use this innovation to in fact enhance what we're doing?

27:21 Paying teachers for performance

Ben Zweig: Yeah, interesting. Okay, so let's say we have that and we can evaluate well. And I know that we're not in that paradigm, but let's say we were to get to that paradigm and we have good evaluation of students and what they know and what they know in subsequent classes and can trace that. Then do you think if we had great evaluation, would we have better education or better teachers, better education, or what still needs to be solved once we have good evaluation?

Rick Hanushek: Well, today we have a reasonable evaluation in these state tests, but we don't rate teachers on the basis of how well their students are learning. So if we wanted to get better education, in my mind, the secret is having more effective teachers. And how do you get that? Well, maybe we want to pay more effective teachers more than less effective teachers and try to encourage other people to come into teaching as opposed to judging what they learned in school as opposed to what the kids are learning in their classrooms.

Ben Zweig: Yeah. So it is shocking how standardized the pay is for teachers, like how tight that distribution is.

Rick Hanushek: Well, it's set in most districts. The single salary schedule says you look up how much graduate education does the teacher have and how many years of experience, and I can tell you her salary.

Ben Zweig: Why is that? Yeah. Why is there not more performance pay or just discretionary pay? I think evaluating based on results makes sense. But I also, I mean, let me know if you disagree, but I kind of feel like everyone knows who the good teachers are anyway. So even if it were sort of set in a more discretionary way, that would probably be like positive on its face.

Rick Hanushek: Well, I agree with your overall view that few parents are confused about whether their teacher is really good or isn't. And you have lots of people queuing up to try to make sure their kid gets the right teacher, at least in suburban areas where the principals pay attention to the community or something.

Ben Zweig: Have you ever measured that directly, like the relationship between perceived quality teachers and their sort of measured performance?

Rick Hanushek: Sure. There are some measures of that. And we have two different kinds of evidence. One is if you measure their objective performance on the basis of how much do students gain in terms of achievement over the year, value added of a teacher in the lexicon of education. And you can compare that to first, the subjective rankings of raters who come in and rate teachers. Those seem to be pretty highly correlated. But more shaky.

Ben Zweig: How high are we talking? Like 0.8? 0.4? I don't know.

Rick Hanushek: I think sort of at the individual level, the correlation is probably 0.6 or something.

Ben Zweig: Oh, that's high.

Rick Hanushek: But the key here is not the overall correlation. The only thing that matters if you're running a school is who are really good teachers and who are really bad teachers. And then there's a lot of confusion in the middle. And so you don't care about the confusion in the middle. What you want to do is try to get more of those really effective teachers and not have so many really poor teachers. And there, the subjective information is much tighter. It's pretty obvious who the really good teachers are and who the real turkeys are.

32:16 The districts that broke the pay mold

Rick Hanushek: But there's a second bit of information that I find also pretty startling. There are a few school districts in the country, I'll name three in a minute, where they have broken the pattern of paying by degrees and experience and have gone more to effectiveness. And in one of those is Dallas, Texas. And Dallas, Texas has a rating system for teachers that includes student survey information.

Ben Zweig: Oh, wow.

Rick Hanushek: And it turns out that student survey information, which is not heavily weighted in their evaluation, but student survey information is highly correlated with value added, not with how funny the teacher is or so forth. It's with the learning gains that teachers are getting. So I think we have today three school systems in the large that are doing something different. One is Washington, D.C., which when Michelle Rhee was the chancellor way back in 2010, put into place a system that had a very good evaluation system of teachers that partly used test scores if they were available, but mainly used outside ratings, according to a given rubric of what the teacher is doing in class and so forth. People who do really well in terms of effective instruction over a two year period got very, very large bumps in their base salary. It wasn't a one year bonus. It was an increase in their base salary. People who did very badly for two years in a row got fired. And in fact, Washington, D.C. Fired a large number of teachers over the 2010 period and so forth. Washington, D.C. Has showed better performance on state tests than over time than the other 35 big city school districts that are evaluated according to the national assessment test. And it's shown that it's worked. Dallas, Texas did something slightly different that I think is actually better in many ways. They put in place an evaluation system that again was pretty sophisticated based upon the rubrics that outside raters used. Outside raters came in 10 times a year on unannounced schedules and so forth, something that's pretty odd in education. And they completely eliminated the old single salary schedule for teaching and paid everybody according to their effectiveness rating. It turns out after doing that, the district did better than you would predict by comparison districts. And then they did something even more than that. They had now information on how effective individual teachers were. They offered bonuses to teachers to move into their most disadvantaged schools, the schools with the lowest performance. But lots of districts have done that, offering bonuses. But most districts say, "Well, if you come to this school, we'll give you $5,000 more a year." Dallas didn't do that. They said, "If you're the best teacher, most effective teacher, we'll give you a $10,000 bonus. If you're the next most effective teacher, we'll give you an $8,000 bonus. If you're the next most effective, we'll give you a $6,000 bonus." So that the size of the bonus depended upon how effective the teacher was. They found out two things. One is effective teachers responded to incentives and were willing to move to disadvantaged schools. And secondly, within two years, the worst schools in the district came up to the district average within two years by, in fact, trying to provide incentives for good teachers to be there and deal with the kids that needed them.

Ben Zweig: Yeah, that's really wild. I mean, it seems so obvious in some ways that if someone's performing well, you want to incentivize them to stay, take on more work, have them take on harder work. If someone's doing poorly, you want to cut them. So I think that is very obvious for the private sector. I mean, like, blindingly, obviously. Why is it like...

Rick Hanushek: The surprising part, Ben, the surprising part is not that people responded to incentives. At least you and I have been taught that incentives matter. And if you identify what you want, you can get there with a set of incentives often. What's surprising is I gave you Washington, D.C., Dallas, and currently Houston, which got taken over by the state and used that to completely change their pay system along the lines of Dallas. The guy who invented the Dallas system is actually now the superintendent in Houston.

Ben Zweig: Oh, nice. Okay, cool. But yeah, why can we only count them on one hand?

Rick Hanushek: So I've given you three. There are 13,500 school districts in the country. And these are not unknown cases. The data are there. People can see the data. They've been made public in various ways. And yet, 13,497 school districts are not jumping at picking up these examples. That's the really, I think, startling and important part here is that school districts, to the extent that they say, "Yes, we want higher student performance," don't really have strong incentives to follow up on that when, in fact, there are other things that they might want to do, make everybody happy and comfortable and so forth. And they don't go through what turns out to be currently some difficult political fights to get people to agree to do something different.

39:51 Who resists performance pay?

Ben Zweig: So why do they... Like if we take these examples of Dallas, Houston, Washington, D.C., I understand it's better the students have benefited. It's better for the students. Who loses? I mean, if you were some entity, some player in Dallas, who's worse off? Who's fighting against this being implemented?

Rick Hanushek: Well, there are losers. Ineffective teachers become the losers.

Ben Zweig: Do ineffective teachers have a lot of political power?

Rick Hanushek: You don't have much sympathy for these people, but they're real people and they have jobs.

Ben Zweig: But that seems like a, I don't know, is that like a strong political block?

Rick Hanushek: Well, I mean, frankly, yes, it is a strong political block because the one thing that we know over time is that teachers' unions resist having differential pay and having these differential evaluation systems. And they have been consistent and effective in their messaging. Over a long period of time, they say, "Ben, you are right that Sally Smith is not a very effective teacher." And everybody knows that. And the principal knows that the teachers know it, the parents know it, that Sally is not holding her own there. But if we look, if the district does something with Sally, they're coming after you next, then. And they will be after you if we allow this to happen and we can't control it. And that's part of it. The other part is, we have seen, and I won't go into the full example, but that superintendents of schools in districts where they're sort of pushed to use more performance evaluations have resisted arguing, "Well, it's tough to make these personnel decisions." I don't like to make bad decisions and tell somebody that they should find another job. And so they would rather not do that if they can avoid it. So there's a lot of inertia and resistance there that...oh, but it can be broken down too. There's a bright side to this whole story.

42:40 Texas's statewide teacher incentives

Ben Zweig: Whew, it was getting dark.

Rick Hanushek: The current commissioner in the state of Texas, state commissioner, is a really thoughtful person named Mike Morath. And he was on the school board in Dallas when they changed their system. And he, with looking at some of the data and others, including myself, have tried to help them see the data that this is effective. As commissioner, he worked with the state legislature and some other groups, outside of groups that saw this, to develop an incentive system that said to districts, basically, if you put in place a reasonable evaluation system that's related to student learning, and then identify effective teachers and pay those teachers more, particularly pay them more if they work in disadvantaged schools, the state will pick up the extra cost for the first few years. You don't have to do this district. If it's too hard for you, don't bother. But if you want to do this, the state has grant funds that can be used to support change. At the beginning of last school year, half of the, two-thirds of the school districts in Texas applied for this grant. Now, I'm not sure if they were all eligible, and there's a criteria about what makes them eligible and so forth. And we have yet to see how they actually perform with this grant. But districts respond to incentives. So they agree with their teachers and their school boards that in fact, they should apply for these grants, and they do.

Ben Zweig: Yeah. I mean, at least that gives performance paid to the best teachers. It doesn't quite solve the problem of not cutting poor teachers. I mean, if we're thinking about this as like a union phenomenon, where the unions are protecting the sort of vulnerable segments of their own population, that's like one part of it. And if it is just like discomfort of the superintendent they work with these people, they know them, they live close together maybe it just like is hard for them because they're all part of the same community. I mean, I would think that could possibly be disentangled with looking at the behavior of schools that are not part of a union, like private schools or charter schools, where they don't need to engage with the union. I mean, can those things be disentangled?

Rick Hanushek: Well, a little bit. I mean, and that certainly is the case. Now, Texas, I gave you these examples in Texas. And my examples are weighted toward Texas. Texas does not allow collective bargaining. And so that doesn't mean there's no unions. The unions work with school boards to try to get what they want. But there's no...

Ben Zweig: What do they do if they're not bargaining? I mean, just communicating?

Rick Hanushek: There's individual contracts. And there... I think districts can develop contracts for all the employees, just like private firms. We have now a limited number of private firms that have big unions involved in them. They still have contracts and they still know the value of keeping employees and not having high turnover and so forth. Districts could in fact do that. It's just kind of alien to what we've become accustomed to in schools. So it's sort of thinking about why have US firms been so productive in a worldwide sense? We allow for lots of flexibility and we have lots of companies that have started to think about what incentives make sense for what we want to do. And yet when we go to the public sector, we think that's somehow very different, that we can't think of operating a public sector. So charter schools in fact have taken advantage of that flexibility and very few of them are unionized. Some of them, a few of them are. And they're showing that you can in fact get very good outcomes in their charter schools. We don't know as much about private schools, frankly, because private schools basically don't release data on what they're doing. Charter schools have the advantage of having the flexibility but being required to be responsible for student outcomes.

Ben Zweig: Yeah, I mean, one time when I was, I remember when I was in eighth grade, I went to like evaluate a high school. So I spent a day there, I spent a half a day, I don't know one of the freshmen or sophomores like took me around and like and then I sat in on a couple classes and one of the classes was just taught by this teacher that was so terrible, just like really abysmal, like had no control of the class. Like was boring and slow and like everyone knew that she was really terrible. And then after the student who was who is walking me around said like, Oh, how was that? I was like, Well, this teacher was really bad. But since everyone knows she's bad, I'm sure she'll be fired by the time I would join. And everyone started cracking up. I mean, they thought it was the craziest thing because she had been there for like 40 years or something. And, and they were just like, wow, you're like, really don't get it. And of course, I didn't get it. I was like a kid. But yeah, I mean, that's just like one anecdote that doesn't happen even in private schools. But there's one other thing I mean, I think we've been talking a lot about like, cross sectional variation in teacher quality.

49:27 Why teaching lost its prestige

Ben Zweig: And I wonder if there's something just happening more secular over time. Also, I mean, there was, there was this survey, I'm sure you've seen it. That I'm forgetting the specifics of it. But it was saying that for the first time in a long time, parents considered their kids becoming a teacher is not something they'd be proud of, or something like it was no longer considered a respectable occupation. And that is, is very sad in, in a lot of ways. And I wonder if that is a byproduct of this widespread perception that there's a lot of bad teachers out there and doesn't signal this like prestige. In a way where we're just sort of flooded we've sort of flooded the system with people that are not high status in a way that we see. I mean, what do you make of this just like secular decline in the prestige of education?

Rick Hanushek: Well, that's part of it. But my aunt, who is no longer with us, unfortunately, but my aunt was valedictorian of her college class and went into teaching, became a principal in Cleveland Public Schools, and was an example of this extraordinary woman who went into teaching. Today, I doubt that you can find a single college valedictorian who goes into teaching. What has changed? Well, when my aunt was thinking about a career and a work life, she could be a teacher or a nurse. And that was it. Today, women can be doctors, lawyers, business executives. And that has first taken all the college valedictorians away from teaching. But it's also led to this prestige matter. And so part of it is schools traditionally have not paid as much as alternative jobs. It's actually pretty close if you measure what people know, if you if you have test information about how effective people are in learning by test, you find that the difference isn't that much. It's just that we get people with lower measured skills going into teaching more often than other places.

Ben Zweig: Right. That makes sense. I mean, I can't help but think this is like this has something to do with like, Baumol's cost disease in a way, where education has been a low productivity sector.

Rick Hanushek: And that's part of it. I think that is part of it. I think also, the way we manage teaching and the single salary schedule, make it hard to go into teaching. It's very expensive to be a teacher today because you for the most part have to get a state certification that requires you to take a bunch of very dreary college courses and to have a very restricted college experience to meet the certification requirements in a lot of states. That makes it very expensive for people because they've done things that have limited value outside of teaching and outside of getting the certificate. So we shoot ourselves in the feet by some of the ways that we manage teaching and sort of keep people out who might wanna be in teaching but don't see it as where they wanna go.

Ben Zweig: Yeah, do you know if that is a global phenomenon or more US specific? Declining prestige.

Rick Hanushek: Low prestige is pretty much a global phenomenon in many places. I think the one exception might be throughout much of East Asia where teachers are still revered and still respected. I have a couple of PhD students from Korea from long ago that still send me cards on teacher day.

Ben Zweig: That's so nice.

Rick Hanushek: It's very nice but I think a fairly common phenomenon and it's related to the fact that women are going into lots more occupations and jobs today and that means that well, teachers might not stack up against a good pediatrician or something.

55:37 Tracking and peer effects

Ben Zweig: Yeah, for sure. I mean, yeah, I also wanna ask a little bit about like the student perspective because one question I thought a lot about over the years is like tracking and the idea of whether the variance in abilities or like aptitudes in a classroom matter and whether it's better to be kind of at the top or the bottom or the middle. Yeah, how important is tracking?

Rick Hanushek: Well, I'll use that word the way we use in the US which means how do we mix up classes basically and there's not very big decisions made until maybe high school when people start choosing courses. In other countries, say Germany where I've spent a lot of time, they make decisions in the fourth grade about what you're going to study and how your education is gonna go. Tracking to them is dividing people into different schools and different entire subject matter areas. Early tracking is clearly not a good thing in my opinion. I've looked at this with some co-authors across countries. What happens, countries that have early tracking end up expanding the variation in performance and learning of kids over time because when you choose at fourth grade to where somebody goes, kids whose parents are not as adept at teaching end up with kids in the lower tracks and the more vocational tracks and so forth, kids like your three daughters where you're gonna teach them a lot and particularly if they don't have the most effective teachers, you're gonna make up for those teachers. By fourth grade, it's clear they're on the college path. And so you get very large differences of kids with early tracking and it doesn't make any difference in terms of overall achievement at the end. It's just that you increase the variance.

Ben Zweig: Yeah, so it kind of seems like an insane way to restrict mobility. I mean, that seems just on its face a bad thing but I guess I really meant like more like basic advanced classes for the same.

Rick Hanushek: In the US, any tracking comes later when you're choosing courses and who goes into the AP courses and not. There's been a lot of work on the effect of peers on performance and it appears that there's some positive effect of peers. Most of the studies I think come out that way. The prevailing view in schools is that, well, we should avoid separating people into different courses because the poor kids benefit from having smarter kids in the class but the smarter kids aren't affected. I think that's more ideology than research evidence because I think there is some advantage to having smarter peers.

Ben Zweig: Yeah.

Rick Hanushek: I don't know the current, it's been back and forth in different views on whether it's good to be the big fish in the small pond or the whatever. And I don't know that there's a clear answer there but I think it does make a difference if you can offer advanced classes to those who would benefit from it. So for example, for a while California was moving toward having eighth grade algebra to set people up for other courses. And then the argument became, well, this isn't good because we're hurting kids who aren't prepared for it. So in order to get more equity in the system, we will ban eighth grade algebra and just have everybody go to sort of normal intermediate level math in eighth grade and then they'll do algebra ninth grade which is sort of the wrong way to get equity in my opinion of trying to hold down the top in order to get everybody more equal as opposed to bringing up the bottom.

1:00:55 Teaching mixed classrooms, with help from AI

Ben Zweig: Yeah it kind of makes me think, I used to be a professor at CUNY and in the CUNY schools, there's a very wide variance of like aptitude and ability in these classes. And I was teaching intro to microeconomics or something. And like, I really struggled with who to teach to. Do I teach to the top? Do I teach to the middle? Do I teach to the bottom? Yeah, I mean, before I reveal what I did, I mean, what do you do? I mean you taught in places where there's like, I think a lower variance, but I imagine every instructor has to make that call.

Rick Hanushek: Well, you try to deal with both ends at the same time. I mean, in my view you try to intrigue the top learners with some sort of advanced questions and stuff and then you try to make sure that everybody gets to the basics. I should come back. I mean, now that you brought that up, not at the college level, although it could be, but at the K to 12 level, one of the obvious uses of AI is to sort of more individualize instruction.

Ben Zweig: Right, so the tutor, yeah.

Rick Hanushek: A better set of sort of readings and plan for what people are working on based upon their performance and having EdTech help to identify and put together learning plans for kids at different levels so that you can in fact try to meet what they're doing. The Houston district that is in this reform agenda led by their superintendent, Mike Miles, has a day set up so that some of the fast learners are in specialized classes at the same time that some of the slow learners get sort of more remedial classes and sort of shuffles people around within the day to try to meet that. That's sort of a crude level of I think one of two or three categories of faster and fastest. But there's a place where there have been some, I think, successful cases and some lesser successful cases on providing individual instruction based upon AI and individual plans. Part of the difference is how do the individual plans interact with teachers? Because it's patently obvious to me, I think, and coming out of the pandemic, that having real live teachers around is really good and a benefit as opposed to entirely hybrid instruction or remote instruction, that teachers are still valuable in this modern age. We've found that in the pandemic. And I think that's partly the case with individualized instruction is how much do people go off on their own with their own little work plan in front of a computer or how much do they do interacting with teachers?

Ben Zweig: Yeah, yeah, that makes a lot of sense to me. I mean when I was teaching, I think I became very aware of what was more taxing for me, for myself like what would be sustainable? And I think different teachers get their energy from different places, you know? So I think some really like to help others like get knowledgeable and some just like to talk about the things they like to talk about. So I was more in the latter category, I'm sorry to say, but I just liked talking about the more interesting topics that I think probably skewed toward the top end of the distribution and eventually gravitated toward more electives. But it is an exhausting job you have to like find out what's sustainable. So at least at the college level, I mean, I don't know. It's probably a lot more exhausting in K through 12.

Rick Hanushek: Right, I think you pointed to an important issue there in that teaching is really a complicated job and a difficult job to do it well. And different people have different solutions to how to make it work. And some people can't make it work, because it's too difficult for them.

Ben Zweig: Yeah, you have to be good at a lot of things.

Rick Hanushek: Yeah.

Ben Zweig: Yeah.

Rick Hanushek: That probably comes down to how do you evaluate the teachers and so forth, because it's not a cookie cutter thing. There's no simple way for uniformly good pedagogy.

1:06:25 Is homework dead?

Ben Zweig: Yeah, yeah, for sure. So I have one final question for you. And thanks so much for being so generous with your time. Homework, I wanna leave with homework. Is homework dead or is homework live and well? I mean, should it be dead? Yeah, what is the status on the understanding of the field and your understanding of the future of homework?

Rick Hanushek: Homework is dying, probably not completely dead yet. Homework, I think, conceptually and empirically, is a good thing to the extent that you get students thinking about what they did in class and trying to put it together in some useful way. If you're around the Palo Alto schools, the discussion about homework is, are our kids asked to do too much? Is there too much homework? So there's lots of debate about it. I think it's dying because of, in part because of AI, that I can get AI to solve a lot more math problems than I personally can solve today. And so it's hard to have simple homework. It's hard to get away from the fact that if I asked one of the LLMs to write a one paragraph introduction to the world of work, they would, in fact, produce a pretty good one paragraph. Discussion of that, so it's hard to get people in some ways to do homework. On the other hand, I guess what we're moving to is homework helps the willing. And so you have homework and the willing gets something out of it and the unwilling don't get much out of it. You have then trouble if you wanted to actually grade homework, but who wants to do that? You want to use that as a learning device and try to incentivize students to use it as a learning device.

Ben Zweig: Yeah, I mean, yeah, homework is, even if it's not mandatory, anyone can do it if they want to do it. And now there's no limit to finding problem sets or things to learn with AI, like you mentioned. So maybe that is good. If it's at least there as an option.

Rick Hanushek: Well, what we're gonna see in both schools and individuals is that some people will take advantage of what's there and some people won't. And then it has a long tail because people then try to take their skill set to the labor market. And what we see in the labor market is highly differentiated rewards based upon this tail of the distribution.

Ben Zweig: Yeah, yeah. So would you make a prediction then that we'd see increasing inequality due to, I mean, I guess we could get that for a lot of reasons, but aside from a prediction, do you think that doing away with homework would causally lead to more inequality because you have this stretching out of the tail and different--

Rick Hanushek: Yes, I think likely. I think that likely, just as I think AI itself is going to push to widen the distribution of incomes and rewards, I think this all builds on each other, the skills that you get in school that allow you to interact with devices like artificial intelligence, but allows you to interact in work circumstances pays off.

Ben Zweig: Yeah, okay. Well, thank you so much. I mean, this is so interesting hearing this perspectives and chatting with you about so many different things. Thank you, thanks for being part of the show.

Rick Hanushek: Well, it's always wonderful to be with you and I hope we can do this again at some point in the future. I might run out of things to say, but it's always good to talk.

Ben Zweig: Yeah, thank you so much. The Economics of Work is brought to you by Revelio Labs, workforce data for research and benchmarking. If you enjoyed, please remember to rate, review, and subscribe. It'll help us reach other curious listeners. I'm your host, Ben Zweig. Our producer is Cole Wagner. Thanks for listening.