You Make Me Feel Like a Natural Experiment

My guest today is the Nobel Prize winning M.I.T. economist Josh Angrist. It’s extra special to have Josh on the podcast because he was a pioneer in the very approach I built my own academic career around — what we call natural experiments.

ANGRIST: Natural experiments started to attract people like me, partly because it was interesting and fun, and we had the opportunity to actually say something concrete about the world.

Welcome to People I (Mostly) Admire, with Steve Levitt.

When I talk with businesses, governments, and nonprofits these days, there’s a lot of interest in applying ideas from behavioral economics and in running randomized experiments, but nobody outside of academia is talking about natural experiments. Let’s see if a conversation with Josh Angrist can open people’s eyes to the incredible untapped potential of natural experiments.

LEVITT: So, Josh, you and I have both devoted much of our careers as economists to studying natural experiments. And I sometimes struggle when I try to explain to the uninitiated what natural experiments are and why they’re so valuable for understanding the world. I start by talking about a true randomized experiment, and I say, “You’ve got a researcher. He randomly assigns half the people to a treatment group and half to a control group. And because it’s random, we would expect that absent any kind of treatment, the two groups would have identical outcomes. And if we do observe differences in outcomes across the treatment and the control group, it’s plausible to conclude that it’s due to a causal impact of the treatment.” Then I say, “But honestly, when you get down to it, it isn’t actually the randomization per se, that matters. It’s the fact that you’re taking two, otherwise identical groups of people, who end up being treated very differently for arbitrary reasons.” And I say, “So, just imagine now, if you can go out in the world and find two sets of people who you’d say, ‘Woah, those people should look the same and they get treated differently,’ that is essentially giving you a poor man’s version of a randomized experiment.”

ANGRIST: Or a rich man. You’re blessed when you find that. (laughs) I’m thinking about the way I teach it. And the randomized trial that I begin with is a study by a former student of mine, Kyle Greenberg, who is on the faculty at West Point. And it’s a randomized evaluation of allowing people to use electronics in the classroom. Since I forbid electronics in my classroom, you’re not allowed to have your laptop open or an iPad or anything.

LEVITT: That’s interesting. I don’t know that study. What happened?

ANGRIST: At West Point, everybody takes economics. So, every year there’s a large cohort coming through. But they’re broken up into small classes with different instructors and some of them are allowed to use electronics. And some of them are forbidden. Well, it turns out that there’s a big negative effect of electronics in the classroom. And I explained to my students, I’m forbidding them to use electronics, not because I have this quirk or emotional negative reaction, but it’s scientifically grounded. It’s been shown in randomized trials to reduce learning. But actually, that isn’t really true because I’ve always forbidden electronics. And Kyle was my student and the reason he did this experiment was to find out if I was right. So, the causality goes the other way in that case. But I, of course, knew that I would be vindicated by this trial.

LEVITT: Now, you won a Nobel Prize in part for your work using the Vietnam-era military draft lottery as a set up for a natural experiment. But the setup of that natural experiment is more complex than what we’ve been describing so far.

ANGRIST: Well, many people have wondered, what are the long-run consequences of military service? Are they beneficial or harmful for soldiers? Most soldiers historically in the U.S. have been young men and you’re taking people out of the labor market, and you might be slowing their human capital, as we say, on the job experience. That’s on the negative side. Then on the plus side, you get a lot of benefits in the military. Some people get a lot of training — for a long time, most of our commercial airline pilots were trained in the air force. Most people don’t get those kinds of skills, but many people get some skills. And then afterwards they benefit from the G.I. Bill. A very interesting fact — the veterans of World War II actually live longer than people born in the same year who didn’t serve.

LEVITT: Really? That’s interesting.

ANDRIST: The reason they live longer is selection bias, that people who didn’t serve in World War II — there was usually a health-related reason for that. And that’s associated with reduced longevity. Now, I studied the effects of military service in the Vietnam era and the same sort of selection issues come up there. Many people think Vietnam veterans are disadvantaged. That isn’t really true, on average. because when we have a draft, the armed forces are quite picky. And so, they tend not to take men who are high school dropouts or have a criminal record and have low test scores. Anyway, what I’d really like to do when I’m thinking about natural experiments, my ideal model is a randomized trial. If I could somehow go back in time and run a randomized trial where you know, I’m flipping a coin and I’m saying, “You get to serve and you don’t serve.” So, that, of course, is impossible, but there is something that has an element of that called the draft lottery. So, for men born 1944 to ’53, who were at risk of being drafted to serve in the Vietnam-era armed forces, there was actually a lottery over birthdays and men were called in a randomly assigned order.  

LEVITT: And if I remember correctly — they did this on T.V., right, Josh?

Roger MUDD: The Draft Lottery. A live report on tonight’s picking of the birthdates for the draft …

LEVITT (cont’d): This is all on T.V.

ANDRIST: Yeah.

MUDD: Tonight, for the first time in 27 years, the United States has again started a draft lottery.

ANGRIST: So, you can go now, in fact — YouTube has everything. So, you can go and you can find the two minutes of C.B.S. news with Roger Mudd live doing the draft lottery.

MUDD There are 366 numbers to select, one for each birthday in the year plus one for February 29, the leap year.

ANGRIST: So, they just called men in order.

MUDD: And the famous first pick tonight is September 14, the first birthday that now is designated double-zero-one, which means for 19-year-olds born on September 14, that beginning in January, local draft boards will induct those men born on September 14, barring deferments.

ANGRIST: It is the case that if you had a low draft lottery number, it was far from certain that you would serve. So, you might’ve had a deferment. It could have been an education deferment. It could have been health related. It could have been related to family characteristics. Some occupations were deferred, at least for a while. Only about 35 or 36 percent of white men born in 1950 with low draft-lottery numbers serve. So, at some point they have enough men. So, they reached number 195 and they don’t call anybody with a higher number. So, that splits the cohort — a little over half the cohort has birthdays that were called and then everybody else is scot-free, they don’t have to serve if they don’t want to. Now during the Vietnam period, men with high draft-lottery numbers continue to volunteer in large numbers. About 20 percent of the men who faced no conscription risk, volunteered.

LEVITT: If I’m a listener, here, I’m saying, “God, this is complicated. People are deferring. You don’t know if you’re going to get in or not. How are we ever going to learn anything from this mess?”

ANGRIST: It’s not as simple as a clinical trial where everybody is either randomized into treatment or control. But the draft lottery numbers are surely random and they’re unrelated to your family background or your motivation to serve, or your ability, or anything about you besides your service. So, this is what I figured out, and it’s the bread and butter of my research. There is a way to turn that draft-lottery, draft-eligibility effect, which is essentially the difference between a 35percent probability of service among those with low numbers. And a 20-percent probability of service among those with high numbers. So, there’s a way to turn that differential into the actual effects of military service. So, since the difference of the probability of service is 0.15, I take the effect of draft eligibility on earnings, which is maybe $400 later in life. It’s a negative $400.

LEVITT: So, the people who had low numbers, they earned, on average, later in life, $400 less per year.

ANGRIST: Right. Ten years after the Vietnam conscription ended, they earned $400 less. But why do they earn $400 less? Well, there’s really only one reason for that: They were more likely to serve. And therefore, that $400 is driven entirely by the 15 percentage point gaps in the probability of service. And so, to get back to the actual effect of service, I simply divide by 0.15, in other words, I multiply the 400 by about 6. And that gets me to, a fairly substantial earnings gap.

LEVITT: $2,400, so almost 10, 15 percent of annual earnings.

ANGRIST: Yeah. And the econometric method that does that is called instrumental variables. I didn’t invent instrumental variables, but I was one of the cohort of people who saw the power of instrumental-variables methods to illuminate a lot of interesting causal questions.

You’re listening to People I (Mostly) Admire with Steve Levitt and his conversation with Nobel-prize winner Josh Angrist. After this short break, they’ll return to talk about Josh’s research on education in Israel.

*          *          *

LEVEY: Hi, Levitt.

LEVITT: Hello, Morgan.

LEVEY: So, now’s the time when we answer a listener question. And our listener Raymond is trying to buy a new home in this boiling hot market. He’s submitted several bids, but has had no luck in getting a bid accepted. What can game theory teach us about gaining an edge as a buyer in this confluence of rising rates, lots of competitive buyers — some even making cash offers — inflation, short supply of homes and surging costs of home construction?

LEVITT: So, I don’t actually think game theory is the answer here. This is a problem for regular old economics and the power of incentives. And so, I have two ideas about how to use incentives to your own benefit that might help Raymond. So, the first suggestion I have —  and it might sound counterintuitive — is that I would suggest to Raymond not to have a buyer’s real estate agent. The unusual structure of real estate deals means that a certain share of the price of the property is set aside to go to the real estate agents. And that money’s split evenly between the seller’s agent and the buyer’s agent. But if Raymond doesn’t have a buyer’s agent, then the entire pool goes to the seller’s agent. So, by not having a buyer’s agent, suddenly, it is really important to the seller’s agent that it’s your offer that gets accepted. So, maybe they’ll try to work some magic on the homeowner, or maybe they’ll just try to shift it so that you get the last chance to make an offer so that you get to make the best offer. I will tell you I’ve done this myself. On numerous occasions, I’ve gotten involved in bidding wars in a case where I didn’t have a buyer’s agent and the other people did. I have won every single one of them. I think it is the single most powerful thing you can do in a competitive market to make sure that you are the winner.

LEVEY: And we should say that you wrote about this in the original Freakonomics book, correct?

LEVITT: We did. So, it started out just as academic research. So, I studied real estate transactions and I learned about the incentives of the agents at that time. But then it went from the academy to real life. I actually went and I tried these tools and wow, they’re like magic. There are very few things that I’ve studied and then put into practice that have worked as well as not having a buyer’s agent.

LEVEY: Okay, and what’s your second suggestion?

LEVITT: This one is very different, but still uses economic thinking, which is in the end oftentimes in these competitive markets, there are many offers which are more or less the same in financial terms. And it’s almost arbitrary which of those offers the seller will take. So, what could you do to tip the scales in your direction? This is going to sound crazy, but my suggestion would be, maybe to make cupcakes or blueberry muffins. And give them as a gift to the seller. Or to write a heartfelt note about how much you love the property and how you will maintain the vision of what the current homeowner has created in the house. This might sound crazy, but it’s been shown in economic research that these non-financial elements can be really powerful to people’s thinking. And when they’re essentially indifferent between who’s going to be the buyer, then that little edge, can be psychologically very powerful. The beauty of it is relative to buying a house, making some cupcakes is really easy. So, even if that strategy fails and like I have to say, I have not ever made cupcakes for the people I’ve been trying to buy a house from, so I can’t vouch that it’ll work.

LEVEY: This is more in theory than in practice.

LEVITT: Yeah, but, look — I think it’s a solid theory. And if anybody actually does take the advice and writes a sonnet or cooks cupcakes for someone they’re trying to buy a house from, I would love to hear from you, whether it works or doesn’t work.

LEVEY: Well, Raymond, good luck trying to buy a home. If you have a question for us, our email address is pima@freakonomics.com. That’s P-I-M-A@freakonomics.com. It’s an acronym for our show, Steve and I read every email that’s sent, and we look forward to reading yours.

LEVITT: And if you would like your email read over the air, accompanying your email with cupcakes would probably be quite an effective strategy.

In the rest of our conversation, I want to first cover my all-time favorite Josh Angrist natural experiment, which, believe it or not, has roots in the Talmud. And I also want to talk about the crazy path to Nobel Prize winner.

LEVITT: One of my favorite examples to give when explaining randomized experiments is your research on Maimonides. Could you explain how a rule laid out almost a thousand years ago by a Jewish scholar provided you with a natural experiment?

ANGRIST: So, it’s a little more complicated than the draft lottery, but it’s not very complicated, and it has this great connection with the Bible. Parents and teachers and perhaps students tend to think that it’s good to have smaller classes in college. And of course, as the parent of small children, you might have been wondering whether your elementary school class sizes that your kids were in were too big and you might’ve liked them to be smaller. Now, economists are interested in that because it’s very expensive to reduce class size. Basically, the primary cost of elementary school education is teacher salaries. And if you want to cut class size in half, you’re basically going to have to hire twice as many teachers. So, that’s very expensive.

LEVITT: And Josh, how come you can’t just say, some kids are in big classes, some kids are in small classes. Why can’t you just look at the difference and know the answer?

ANGRIST: Well, you can, and you should, but it might be misleading. So, it turns out that in the samples that I’ve looked at, which are mostly from Israel — but you would see the same thing in the United States to a large extent — kids in bigger classes actually have higher test scores. It’s kind of contrary cause smaller class were supposed to be better. But typically, class size is not randomly assigned, and classes are bigger in more densely populated and growing areas.

LEVITT: And in the U.S. also, I think often special education kids are put into very small classes.

ANGRIST: Absolutely. So, then you have kids that are identified as needing special education services — well, that usually is associated with low test scores and poor outcomes. So, that makes it very hard to just interpret the naive correlation between class size and achievement. But there is some quasi-experimental variation in class size, and that brings us to Maimonides’ rule. So, in Israel, the maximum class size is 40 in elementary schools. And Victor Lavy and I took advantage of this maximum class size to study class size effects in Israel. If you’re in a fifth-grade cohort and there’s 40 kids in the cohort, you’re in a pretty big class because 40 is allowed. But then suppose on September 1st when the school year starts, another kid shows up, then they have to split that class and now you’re in a class that’s half as large. And essentially what the Maimonides’ rule research design does is it takes the arrival of that marginal kid that pushes you over the corner in the class size, because, all of a sudden, one additional kid causes a big jump in average class size. In this case, the jump is negative — a big decline.

LEVITT: So, my framework would be: The kids who are at a school that have 39 kids in fifth grade are almost identical to the kids who are in a school that had 41 kids in fifth grade on average. And so, the fact that the kids who are in the 41-person class, they get divided into two classes of 20 and 21. They get treated very differently than the kids who were all lumped in with 39. And that’s seems pretty arbitrary and it has the feel of a randomized experiment without — you, Josh Angrist don’t have a control over which kids get assigned, which, but really we’re arguing it’s by luck by circumstance that kids get tossed into these very different outcomes.

ANGRIST: That’s the claim.

LEVITT: And now why is it called Maimonides’ rule?

ANGRIST: Jews know Maimonides as the Rambam, a great Talmudic scholar. The Rambam wrote in the 12th century about various things related to Jewish life. He was essentially interpreting the Talmud. And the Talmud is a great compendium of biblical commentary. It covers a lot of things about how life should be organized, down to very prosaic things like how schools should be organized. And Maimonides said the Talmud says that there’s a maximum class size of 40. Now actually, it turns out he took some liberties because if you go back to the Talmud, the maximum is 50. So, it’s not clear how 50 became 40, maybe in the intervening 600 years between the Talmud and Maimonides, they were due for a reduction in class size. And they’re not too many classes of 40 in the United States today. And even in Israel. Over time class size is falling. I should mention also, the actual effects that we found are that it is beneficial to reduce classes. When class size goes down, achievement goes up.

LEVITT: And it went up a lot, right? was, it was 0.25 standard deviation, which is big for an education intervention.

ANGRIST: Yeah.

LEVITT: Okay, but what’s interesting is that then you went back later, and the results were totally different. What’s that all about?

ANGRIST: One of the lessons here is that as an empirical scholar, you always have to be prepared to be surprised. And your earlier findings might not hold up. Victor and I, with two of our graduate students, redid the Maimonides research design a couple of years ago with more recent data. And we didn’t find any achievement effects. And actually, the way that that second paper evolved is one of my own graduate students had gone back to the original Maimonides’ rule data and he found some problems in what we did there. So, it turns out for example, that it isn’t true that whether there’s a 40th kid or a 41st kid is purely, randomly assigned. There’s too many cohorts with 41 kids relative to 40.

LEVITT: You’ve got 40 kids. You got a lot of incentive to find the 41st.

ANGRIST: Exactly. So, it turns out that the manipulation of enrollment is not a big deal, but the findings in the new data were quite different. And we did work very hard to figure out why, but we couldn’t figure it out.

LEVITT: You raise an important point, which is the style of research that you and I do is better at describing the facts of the world, than having really necessarily compelling mechanisms underlying it, right? So, you set out to understand whether smaller class sizes affect kids’ achievement. The design doesn’t say, “Does it affect achievement because teachers give more attention to some particular kid or because kids feel safer?” That’s not the nature of what we do. So, we can’t always understand.

ANGRIST: We don’t always have a mechanism.

LEVITT: Yeah, exactly.

ANGRIST: That’s true. Some people use that fact to criticize our kind of work. I always feel like, ultimately, that’s really no different than clinical trials in medicine. There are lots of things that have therapeutic value for poorly understood reasons. So, social science is like that, but not always, Steve. If you go back to the draft lottery story, we got to the point where we said people who were forced into the military end up earning less as a result of their service 10 years later. I do have an explanation for that. One is that there’s a lasting health effect. Another is an economic mechanism that has to do with the fact that these are young men. We know that, among young men, earnings rise very steeply when you’re young. So, they rise at about 10 percent a year when you first enter the labor market, and then they tend to flatten out. So, I can use that fact to calibrate the earnings losses from Vietnam-era military service. So, if you say, “The actual reason why military service reduces my earnings is that I entered the labor market later, and therefore I start two-years later on that experience profile,” that turns out to fit the actual earnings losses from Vietnam-era military service very well. And eventually, veterans should catch up. And that turns out to be visible in the data.

LEVITT: Yeah, that makes a lot of sense. I certainly didn’t mean to imply that we never have mechanisms, but we don’t necessarily have mechanisms.

ANGRIST: We don’t necessarily. And, you know, sometimes we’re happy to have the answer.

LEVITT: Now, I want to go back to the second Maimonides paper for a minute, because I just think about the politics of our profession. So, you went yourself and you gathered new data and you found a result which contradicted it. And I think people’s reaction to that was, “Wow, that’s interesting. I’m glad Josh went back and got that data.” I think if someone else had gone back — essentially replicated your study and your data, found a different result — all of a sudden there’d be hand ringing and complaining. “Oh, did Angrist and Lavy rig their data?” Do you think that’s right? Don’t you think it’s true — the fact that you’ve reversed your own result saved you an enormous amount of headache in terms of reputation and fighting in the profession?

ANGRIST: I don’t know, um — I always tell my students, an empiricist is just always on the defensive. You can never prove that your findings are right and other people and you, yourself, can falsify your findings. And given that, it’s better that you’re also very critical of your own work and you try to figure out what the problems are before somebody else does.

LEVITT: I certainly know a lot of researchers who I think, faced with the reality that a really important result that had gotten them a lot of credit was going to be reversed, might not have been so eager to go ahead and publish it.  

ANGRIST: Victor and I weren’t happy. I can’t say that it made our day.

LEVITT: I think if someone else had done it, I suspect they would have pointed to the small errors that you had made around the edge of 40 and 41.

ANDRIST: They might have.

LEVITT: And they wouldn’t have wanted to make a big deal out of it because, wow, if I can bring down the Nobel Prize-winning Josh Angrist —

ANGRIST: Well, I didn’t have a Nobel Prize at the time.

LEVITT: I think you’ve saved yourself a lot of time and heartache by doing it yourself.

ANGRIST: I think that’s a fair point, but I don’t really mind that. Sometimes I say, “A good empiricist never has a restful night.” And it is true that I wake up sometimes four or five in the morning and I think, “Oh, my God, there’s something wrong with this or that thing.” I know there’s this tendency that if things work out, and everything looks clear, you kind of accept those findings, but you need to be critical of everything.

LEVITT: So, when you started doing economic analysis of natural experiments in the 1990s, you were a real pioneer. It just wasn’t the way economists analyzed the world. In hindsight, it seems completely obvious that natural experiments should be central to the economics toolkit. But why do you think the generations before you just couldn’t see that?

ANGRIST: Well, everything that sort of works out as a good idea, ex-post, it sort of seems obvious. “Oh, why didn’t everybody do that?” I remember when I got to Princeton in the fall of ’85, you had this sense that the really smart people did theory and empirical work was maybe looked down on a little. There was maybe some selection away from empirical work by people who might’ve been able to make substantial empirical contributions. What happened is that natural experiments started to attract people like me, partly because it was interesting and fun. And we had the opportunity to actually say something concrete about the world. There was something going on at Princeton in the 1980s that was really very exciting. There was a cohort of graduate students, many went on to very distinguished careers — Bob Lalonde, was a big influence on me and we were very excited by a couple of questions. One is, do we really need randomized trials to answer causal questions? The question that motivated Bob’s work was, do you really need a randomized trial to estimate the effect of training programs? Something like a government training program for disadvantaged workers — for workers who lost their jobs. And Bob had a wonderful thesis where he showed quite convincingly that without a randomized trial, you weren’t very likely to get the right answer. And I remember reading Bob’s work and thinking, this is what I want to do. And the draft lottery — very quickly I understood that this was in the same domain. It had the same sort of methodological flavor. And I just got very excited about that.

LEVITT:  So, it wasn’t purely by chance that you were studying Israeli schools, despite the fact that you grew up in Pittsburgh, but your backstory is, I have to say, an incredibly unusual one for a Nobel Prize winner. You were certainly not on the Ph.D. path early in life. Do I actually remember correctly that you dropped out of high school at one point?

ANGRIST: Yeah, in a manner of speaking. I wasn’t a very good high school student. I didn’t like school at all. I figured out that the graduation requirements in Pittsburgh were minimal. And som I was able to finish high school with a diploma in 11th grade. I was 16.

LEVITT: And what did you have in mind?

ANGRIST: I wanted to go work.

LEVITT: What did you end up doing?

ANGRIST: I worked as a busboy. I worked as a telephone solicitor. I also would work in the summer as a camp counselor for mentally-handicapped people. But I realized that there wasn’t really much of a future in this. And my friends also were going to college, the ones who had graduated on time. So, I thought I should go to college.

LEVITT: You got into Oberlin —

ANGRIST: I was lucky. Somebody like me probably wouldn’t get it into Oberlin today. I had decent S.A.T. scores. I had also done some writing, so I was writing short stories. My stories weren’t great, but they were probably good enough to show that I had the potential to do college work.

LEVITT: So, you got into Oberlin. And something turned around because you ended up writing a senior thesis that blew people’s minds, right?

ANGRIST: Well, I don’t know if “blew people’s minds,” but by the time I was a senior at Oberlin, I was doing good college work. And I was into economics big time. And I did a senior thesis on sample selection bias, which is the problem of not observing wages for people who don’t work. And I was very lucky that Oberlin’s honors program brought in an outside examiner. And my examiner was Orley Ashenfelter who, is a distinguished, labor economist from Princeton. And Orley came and spent like three days at Oberlin, and we hit it off. And he wrote me a letter and he said, “When you finish college, if you want to go to graduate school, you can come to Princeton.” So, that was very lucky for me.

LEVITT: Okay. So, this wouldn’t sound like that outrageous of a story if this is where it ended, but Orley said you can come be a Ph.D. student at Princeton.

ANGRIST: But I didn’t.

LEVITT: What did you do instead?

ANGRIST: I went to Israel.

LEVITT: You went to Israel. Okay, and what happened there?

ANGRIST: So, when I was a junior in college, I spent spring break in Israel, where I connected with a good friend of mine from high school, Mike Drescher. And I was very captivated by Israel. My mother’s parents had seven siblings who all moved to Israel in the 1930s, fleeing pogroms in Europe — Lithuania. And I just connected with it in a very powerful way. And I said, “I want to come back.” And Mike and I, we agreed that after we got out of college, we would spend some time in Israel. So, I enrolled as a master’s student at Hebrew University.

LEVITT: But things didn’t go well, right?

ANGRIST: No, I didn’t do very well at Hebrew University. It’s a difficult program. It was in Hebrew. I was learning Hebrew. So, Mike and I decided we’re going to quit what we’re doing. We’re going to become Israeli citizens, and we’re going to join the army.

LEVITT: This was not a light undertaking, because there was fighting going on at that time in Lebanon, right?

ANGRIST: Yeah, it was 1983. So, Israel was mired in Lebanon. Now that I look at it from the point of view of an adult, I see that was a terrible idea. But at the time I didn’t think too hard about it. And Mike and I joined, and we enlisted in an infantry brigade called the Nahal and we aspired to serve in the paratrooper battalion. And eventually I was able to that.

LEVITT: Paratrooper means you jump out of airplanes?

ANGRIST: Well, you’re airborne. You don’t jump out too much, because there’s very little call for airborne infantry parachuting in modern warfare. But you still train, you go to jump school and you do some jumps. And it’s physically very challenging.

LEVITT: And then you saw real combat too, right?

ANGRIST: I wouldn’t say it was combat, but it had its moments. We would go into Lebanon for a few months, and we’d be in a base there and we’d do some things, some ambushes and things, but very little happened to me. I’m lucky. I had friends that weren’t so lucky.

LEVITT: Somehow you got back to Princeton. What — you just called Orley on the phone after a bunch of years as a paratrooper?

ANGRIST: So, I was a soldier for two years. And I got out of the army in 1985. And I was thinking that maybe I should go back to school. So, I called Orley and I said, “You remember you wrote me a letter and everything? Is that offer still good?” He said, “Sure, you can come in the fall.” I said, “Well, how will I fund it?” And he said, “No, no, we’ll take care of you. Don’t worry about it.” Later on, I was joking with Dave Card about this. And I said to David, “I’m a faculty member at M.I.T. I cannot do what Orley did. I cannot like, talk to a guy or a gal, who I think is very talented or well-suited for graduate and say, ‘Hey, come to M.I.T. I’ll take care of it.'” I don’t have the power to do that, unfortunately. And David said, “Well, Orley didn’t have the power to do that either. He just did that.”

LEVITT:  Now at that point, now you’re really on a path to Nobel, from the moment you got to Princeton to today.

ANGRIST: Not exactly. I struggled at Princeton, initially.

LEVITT: Did you?

ANGRIST: Oh yeah. I mean, I’d been out of school for three years. One of them was the Hebrew U experience that I didn’t learn much. And, the math is very high, so I struggled mightily and again, I had to take a tutor for some part of my first year. But I was lucky that I knew that that was where I want to be. Like, I remember just every morning I would ride my bike up to the library and I would think, ‘Oh, I’m so happy to be here. I couldn’t be happier.’

How many examples have there been on this podcast of people who struggled only to later achieve great things? Josh is another great example of someone who looked lost, and maybe even was lost, but found his calling in the end. It’s never too late to make a change. I’m not saying you’ll win a Nobel Prize, but the research says that change is good. If you’re trapped, maybe today is the day to make your move. And if you do, I’d love to hear about it.  If you’re interested in natural experiments, check out my past episode with Dr. Bapu Jena or his podcast Freakonomics, M.D. on the Freakonomics Radio Network.

People I (Mostly) Admire is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and Freakonomics M.D. All our shows are produced by Stitcher and Renbud Radio. Morgan Levey is our producer and Jasmin Klinger is our engineer. We had help on this episode from Alina Kulman. Our staff also includes Alison Craiglow, Greg Rippin, Gabriel Roth, Rebecca Lee Douglas, Zack Lapinski, Julie Kanfer, Eleanor Osborne, Mary Diduch, Ryan Kelley, Emma Tyrrell, Lyric Bowditch, Jacob Clemente, and Stephen Dubner. Our theme music was composed by Luis Guerra. To listen ad-free, subscribe to Stitcher Premium. We can be reached at pima@freakonomics.com, that’s P-I-M-A@freakonomics.com. Thanks for listening.

ANGRIST: My father’s theory of child-rearing is that you should have early dinner. If you have late dinner, people get cranky and hungry.

LEVITT: I’m going to take that to heart. From now on, in the Levitt household, dinner will be early.

ANGRIST: Yeah, you should be eating at 5:30.