David Deming: What's Happening to Young Workers, and What College Should Do About It

Ben Zweig: College graduates are now, for the first time in recent memory, more likely to be unemployed than the average worker. How did that happen and what does it mean for higher education? My guest today is David Deming, Professor of Economics at Harvard, Dean of Harvard College. He also writes a great Substack called Forked Lightning and hosts a fantastic podcast called The Context Window. Here's our talk. Welcome to the Economics of Work. I'm excited to welcome David Deming, Professor of Economics at Harvard, Dean of Harvard College, and author of one of my favorite Substack, Forked Lightning, and host of one of my favorite podcasts, The Context Window. So David, welcome to the show.

David Deming: What a kind introduction, Ben. Thank you for having me. I'm excited for the conversation.

Ben Zweig: No, I'm really liking it. I liked Forked Lightning. I can never get that out. Forked Lightning, yeah. But it's great. It's very succinct and to the point. Love it.

David Deming: Thank you. Life is short. You gotta get to the point.

Ben Zweig: Yeah, yeah. They say like, if you're giving a speech, better to be short and good than short and bad, than long and good, than long and bad. So, there you go.

David Deming: Sounds about right to me.

01:02 Is on-the-job training disappearing?

Ben Zweig: Okay, so we hear a lot about like how on-the-job training is kind of diminishing due to AI automating entry-level tasks. That's like a common take. And I wanted to hear your perspective on like, one, is that happening? And if it's a real thing, then like, what is, how do colleges respond to that?

David Deming: Yeah, okay. So it's interesting question to start off with, Ben, because the reality is we don't actually measure on-the-job training or on-the-job learning very well. And so, there's a lot of talk about it happening, and it probably is happening for reasons we can get into. But it's surprisingly hard to just say, well when I worked at this company, I used to get trained 15 hours a week, and now I'm getting trained five, because a lot of the training is apprenticeship, you know? So you're kind of with people who've been around longer than you, soaking the information in by osmosis, rather than like, I'm in school now. So it's a lot harder to measure the extent of mentorship. I do think one reason, let's just say, one reason to think there's less of it happening is because of the shift to remote work, you know? And so it's a lot harder to mentor the person next to you, when they're not actually next to you, they're next to you on a screen, they're not physically next to you. And I know, I'm sure you can think of times early in your career, I know I can, where like I either strategically or unintentionally popped in to talk to a more junior, sorry, a more senior colleague, and get some feedback from them, or watch how they behave in a way that's not really captured in a job task taxonomy or anything like that, but is really important to helping the junior people become senior one day. And I think that's probably not happening as much online, because you go into a meeting, and then when the meeting's over, you hit end meeting, and there's no kind of water cooler conversation, there's no learning by osmosis that's happening afterward, it's all in the meeting, and there's no like recap, hey, here's why I did this way. And so I think there's a lot of common sense reasons to think it's happening less than it did before. And that does have important implications for education and training, because we were counting on employers to do a lot of that training and maybe now it's not happening as much.

Ben Zweig: Yeah, yeah, interesting. So, I mean, if we take this like remote work is driving that like reduction on the job training, like that's very plausible. It sort of rings true it's much harder to like ask a quick question if you have to like set up a 30 minute meeting for it's easy to just like pop in and ask. But I would think that would sort of get figured out with like selection into the types of companies where people want training. Like, so when people are young, they locate in cities for the most part they're more likely to go into a company where there's like more in-person work and young people like working in person.

David Deming: They do. More than everyone else.

Ben Zweig: So one kind of sub question is like, can the market just kind of figure that out? To the extent that young people sort of select into these training opportunities, or is there like a fundamental like mismatch that creates like too much friction that like this sort of training paradigm unravels? Another is like we hear about it happening due to AI and are those kind of like correlated but independent phenomenon or phenomena, or is there actually like an AI effect where you hear of like the investment bankers not, or the consultants not like aligning fonts in a PowerPoint deck and that is like the apprentice type work. So yeah, a couple of questions in there.

04:24 The murder mystery behind new grad unemployment

David Deming: Yeah, so let me first take the question of attribution, which is a really hard one. So we know that like it's a fact that the unemployment rate for new college graduates is higher than it has been in a long time and is elevated relative to the unemployment rate of other groups and has been since the pandemic, basically. So this is a fact, it's like 5.6%, the overall unemployment rate is something like 4.3%. So for the first time in recent history that I'm aware of, young college graduates have a higher unemployment rate than the overall population. So these are just facts and the question is what's caused it and people say, look, it's like a murder mystery. So you could have remote work, that's one thing. The other thing is AI, that's the thing that CEOs and other people are talking about. And then there's this kind of like moment we were in where right after the pandemic, labor was really scarce, the job market was tighter than it had been in a generation and there were help wanted ads everywhere. So people were hoarding talent, keep kind of maybe quote unquote overhiring. And so part of it is just if I'm a CEO, I don't wanna lay people off. So instead I just say, I'm not gonna hire anyone new, I'm gonna freeze employment. And so that kind of low hire, low fire environment tends to disadvantage young people. And so maybe it's just working itself out. So all those three things are in a mix and there's probably other ones too, the change in interest rates. And the truth is like, we're never really going to figure that out in some sense of like, well, it's 20% this and 40% of this. We're always gonna argue about it, and I'm sure they're all interacting. And in some ways, it's not really an answerable question because in some firms it might be 100% due to remote work and some of it might be all AI. And some companies you see this decline in junior hiring when they aren't even using AI, so it can't be AI. So it's gonna be a mix depending on the situation. The way I like to think about it is that they really all interact together in an important way. So part of the thing that happened with remote work was when your work is remote, everything's recorded, right? So not just on Zoom, but also like you have a Slack thread where you catch up afterwards and you have like a CRM that becomes much more important for tracking customer engagement when you're not actually physically next to the person that you're sharing the account with or whatever. So everything gets digitized and codified. And that was all happening before Generative AI came out, but it turns out the digitizing and codifying work is the first step to making it automatable. And so what you see is a kind of, you could call it fortuitous if you're an employer and like unlucky if you're a college grad, like this confluence of these circumstances where you're codifying work in a way that facilitates automation much faster. So it's not one or the other, it's really that they're working together to create this moment. And similarly, like the labor hoarding thing I talked about, well, if you actually are having a hard time finding people to hire and now you can hire remotely, you might say, well, I used to only hire people in the city of Boston, now I'm gonna look in Philadelphia, maybe I'm gonna look in Mumbai too, because if we remote, you're remote. And so it expanded the labor pool for firms and caused them to say, actually, we're not gonna hire junior people, we're gonna outsource it to something else. So they were just more open in this post pandemic moment to different ways of doing things, disruptions to the old order of hire a bunch of young college grads, work them like dogs for a few years and train them up, that kind of model got disrupted for reasons that were unrelated to AI, but AI magnified the impacts of it. That's kind of the story that I would tell that makes it all hang together.

Ben Zweig: Okay, okay, so we've got remote work as a suspect, partial suspect, we've got AI, but let me circle back to AI. And then we've got this great resignation, overhang talent hoarding thing, which I think also makes sense. I feel like that rings true in a way that I hadn't really thought of affecting young people especially. But AI, I feel like we can also, yeah, let me propose a sort of like 2A and 2B. So one is this idea that it's a sort of seniority bias technological change, that it favors older workers, there's less value for younger workers. Another that I've been a little bit suspicious of, but I'm curious to hear your take on, is that it creates matching friction. So the search and match process, you open a job application, a bunch of people apply, and now job applications, you can do AI job applications and someone can apply to a thousand jobs a minute and employers are getting signal jammed and that hurts people that don't have networks and a reputation, which maybe disproportionately affects young people.

David Deming: Yeah, I think that's possible as well.

Ben Zweig: Okay, so multiple culprits, where does that leave us? Actually, before we get to where does that leave young people, where does that leave colleges? Like what are you guys,

08:58 Why David doubts seniority-biased change

Ben Zweig: You're like under a lot of pressure now, there's your constituents.

David Deming: We are, I hadn't noticed. Yeah, yeah. It seems like everything's fine here. No, you're right. So here's where I think, so again, like it's a, as I said, it's kind of like a murder mystery and all the evidence is circumstantial. You're never going to find some people have alibis, some don't, some evidence patterns match up better. You're never gonna, it's never gonna pass muster in a court of law, we can't prove it's one of these four beyond a reasonable doubt. But where it really matters to try to weigh the evidence and say, is it like 60-40 or 80-20, is when you think about the future. So if you think, if your story is the one you mentioned a second ago, seniority bias technological change, meaning there's something intrinsic to AI as a technology that harms young people more than more experienced people, then you want to think about permanent solutions to fix that. Whereas if you think it's actually just a moment we're in, that it's actually this interaction between the two, but in the long run, it won't be seniority bias, then you wouldn't approach it that same way. And so that's where it really matters. And my personal view maybe I'm sticking my neck out a little bit and this is I do not believe the seniority bias technological change story. I think it's narrowly, it's possible to show it's true in the data right now, but I just don't think that's an equilibrium of the labor market.

Ben Zweig: Yeah, what's the reason?

David Deming: Yeah, yeah. And the reason I don't think that is when you look at the history of technological change, a pattern you see over and over again is the heaviest and most intensive users of the technology early on are the ones who benefit the most in the long run. This is like a universal pattern. And and so AI could be different and whatever, but if you look something like three quarters of global usage of ChatGPT among adults is between people under the age of, between 18 and 35. So like, in only something like less than 10% of people over the age of 45 using AI. So it is a technology that is heavily, heavily, heavily used by the young. And so you've got this idea of like, well who knows how to use agentic work platforms and knows how to like prompt, it's definitely young people, much more than older people. And yet somehow they're the ones who are gonna be harmed. Like it just doesn't really add up. And also like today's junior talent are tomorrow's senior executives. And so if you stop hiring junior people, like that works for a couple of years, but you're eating your seed corn. Like, what are you gonna do in 10 years? Like you can't, you're, to me, I think what's more likely then is that those firms are sowing the seeds of their own obsolescence. That is, if you don't have any AI native young people in your company, eventually you're gonna get out competed by a company that is started by a young person. And then you are going to be like the Kodak of this revolution. Like you're going to be a big company that is now a small company. And so I just don't look, and maybe I'm wrong because predictions are hard, especially about the future, as Yogi Berra said. Yeah. So, but to me, when I look at this, it just doesn't hang together in a way that makes sense to me as again, as a long run equilibrium. I'm not denying that it's happening now. I just think if I'm a young person, I'm investing heavily in getting to know this technology and I'm waiting for my moment. And I'm thinking about maybe starting a company or joining a startup or a smaller company that's more agile. I don't wanna wait my turn at some big company that's not hiring as much of people like me anymore. I'm gonna go try to make something of it. So, and that's kind of what I've told college students here at Harvard. We need to help students get there. So I'm not abdicating responsibility, but I think the future really does belong to the young. I believe that. I actually think things are gonna be much better for them in 10 years than they are right now.

12:13 Will young workers' wages take a hit?

Ben Zweig: Yeah, no, I agree. I mean, I think one point on this is like, the first wave of AI was really about lowering barriers to adoption, where you didn't have to be like native or like technologically sophisticated to use it. And that's really changing. Now it's like more sophisticated tools. So it's like probably more biased toward like, young digitally native people more than it had been in the first like, iteration of that. But on this idea of like, our companies like sowing the seeds of their own destruction, I can totally buy the argument that like, that can't persist when it comes to like quantities of young people. Supply creates its own demand. There's like, you need, if there's like a surplus of a certain types of work, or it's gonna, market's gonna clear at some point. But it could still affect relative wages. Do you think that is like, yeah, do you think that could be happening?

David Deming: Well, so I think it's important to drill in on this a little bit. So I think probably what is going to come to an end again, I don't know the future, but this is just a guess, is that this older, Luis Garicano and others have talked a lot in economics about what they call knowledge hierarchy models, which is like, if you think about a modern white collar knowledge focus firm, you're like an IT firm or a consulting firm or software, it's like, they're organized in a pyramid based on expertise. So you hire a bunch of junior people at the bottom layer, they solve problems, the routine problems, the harder problems get kicked up the chain. And so it's just like efficient system for sorting more difficult challenges to people with more experience. And that is, I think a pretty accurate high level description of how a lot of firms operate. I wouldn't surprise me if that model itself is going to be disrupted in the next 10 or 20 years. What that means is it's no longer gonna be taken for granted that you're gonna hire 20 college grads every year. And then, we know if you go work at McKinsey or something, they're hiring five college grads and they're gonna keep one of them and they don't care. And most people leave after two or three years and they just kind of cycle through. It might be that model is gone. And that could be pretty harmful to people who are hoping, who think about a McKinsey job as like your birthright if you go to an elite college, like people think that way and that might be bad for them. And if you're bad for their relative wages, because instead of working at McKinsey, you go work at a regional consulting firm or a smaller company and that you're gonna get paid less. Like that could all happen, it probably will happen. I'm saying something a little different, which is in the long run, I don't think it's gonna be net bad for those people. So I do think, I tell my students this all the time, it's a free country, they should do whatever they want. But when you go take a job at a really big company, you are sacrificing learning for pay. Because you're gonna go work in a big company, be a cog in a big machine, and everything you do is gonna get checked 15 times as it goes up the food chain. And you're not really gonna be held responsible for any decisions you make, and you're not going to learn as much. And so that's a perfectly fine decision and sometimes you're in a situation where you do well, so I'm not trying to talk anybody out of it. But one reason to go work in the government, to go work at a small company or start your own company or work at an NGO that's just getting started is there's so much to do and so few people to do it, that you end up getting broad experience and exposure to everything.

Ben Zweig: That's how it is. Go tell your students to apply to work at Revelio Labs.

David Deming: And so you're gonna bring them in, that's exactly right, you're gonna bring them in, and they're gonna do a lot of things and they're gonna get lots of exposure to you, think about how you think, and they're gonna make consequential decisions, they're gonna be held accountable for. And that is the best way to learn. And so I do think, and you see this pattern a lot, like in my podcast, I interview people who've had successful careers, and they often have nonlinear stories like this. They went and did something off the path that was not the narrow path, but they learned a lot, and they learned something about what they were good at and what they liked, and then they managed to broker that into something interesting and kind of made their way up. And so sometimes, again, you might get paid less and have a less prestigious job, but the learning you acquire in the short run is gonna help you in the long run. And so I just think, that's really what I'm saying for young people, is that in the long run, they're gonna be fine, not like, don't worry, your McKinsey job is still safe for you. I don't actually, that's probably not true.

16:16 Does the market value startup experience?

Ben Zweig: Yeah, so, yeah, there's one kind of puzzle with that, that I'm struggling to wrap my head around, because I think intuitively, I agree, that is like, you learn more from a small agile environment. But when people work at McKinsey, they, that's a credential. And maybe some of that is a signal of getting in. But it seems to be at least a popular conception, that if you're at McKinsey, you will like, the market will think that you learned a lot. And so, if that's really not the case, why doesn't the market put more of a premium on like, sort of being in agile, wearing a lot of hats player at small companies?

David Deming: Well, I think it does actually. So there was this study by Paul Gompers and other colleagues at HBS, a few years ago, where they looked at, they used like these resume data, which have some problems as we all know, but they looked at these resume data and followed people who took jobs at big companies, like I don't wanna pick on McKinsey, like whatever, like Bain or something. Yeah, yeah, it's like big companies. And then they also looked at similar graduates from the same programs who went to start a company or work at a small startup. And then they followed their careers over time and they found that even the people who had failed startups, when they went back to a big company, tended to enter at the same or a slightly higher even level of seniority than the people who stayed in the company. And their point was, look, if you start a company, it fails. That's obviously not as good as succeeding, but the market values that learning and will often bring you in a way that holds you harmless. And so it's not as risky as you might think. Now, is that, I mean, that's one study, so I wouldn't take that to the bank, but I guess the point is, even if you start a company and it fails, or you go to a startup and it doesn't work out, or you go clerk for a judge and the judge is a jerk, and you're still gonna learn something and people will look at that and say, wow, that person did something interesting and you're gonna have a bit of story to tell. And so even if you just think about it as signaling, it's not clear to me that the market signal is bad for something like that.

18:23 How colleges can teach through feedback

Ben Zweig: Yeah, yeah, no, that makes a lot of sense. I mean, yeah, do you think that colleges can also be kind of providing an environment where there's more kind of like startup style learning versus cog in the machine style learning? I mean, I'm sure it's a mix, do you think to the extent that relative value is like shifting, do you think educational institutions at large are adapting or should be adapting?

David Deming: I think we can do it more and we should do it more. We probably aren't adapting as fast as we need to, but I think there are a lot of great things happening in colleges that are, I think essentially giving people, the way I would describe it then is like, what is the best way to learn? It's to get feedback. Feedback is what aids learning. If you're in the classroom and you're mostly getting A's or you're only getting feedback a couple times this semester and it's from the TA and not the professor, you're sitting there absorbing information, but you're not getting any feedback about whether you actually understood the concepts. It's a slow, inefficient way to learn. But when you go out there and you try to sell a company or acquire a customer or you try to, I don't know, file a legal brief, it's not just about starting companies. It's like when you actually go do stuff in the world, you're getting immediate feedback from the world about whether or not that works. You either land the customer or you don't. And so that feedback is much more consequential and it's high stakes. So you can't be like, oh, everything's fine. Your essay is great. It's like, you can't BS your way out of it, like you either worked or didn't. And so I just think that's a much more efficient way to learn. And I think in colleges, we are doing a lot of, there's a lot of innovative things certainly happening at Harvard, but I think everywhere, trying to put students in situations where they get consequential feedback immediately, either from the world, from their peers and from professors. And so actually, this is a direction that I think we need to push in more. And it's about learning. And I think people talk a lot about the credentialing value of colleges and I'm not gonna deny any of that, but for my purposes, as somebody who's working inside a school, that is just totally relevant to me. I wanna increase the learning that happens in the school. I do think people learn stuff in college. It's not that this diploma signal is worthless, but if it is worthless, what am I even doing? I'm wasting my time and I don't believe I am. I actually have watched students learn and grow and develop skills that I believe they wouldn't develop if in the counterfactual world, they were hanging out in their parents' basement or working a job. I think they actually get things here they wouldn't get elsewhere. And so I think we need to be, to me, the issue is not the signal. It's that we're not probably doing our best work helping students learn as much as they could in the four years they're here. And so I think there's a lot, I mean, it's a hard challenge, but I think we can do better and a lot of it involves more feedback from the real world.

Ben Zweig: Yeah, so what are some examples of ways to elicit feedback in a university setting? Aside from grades. I mean, yeah, are there more real-time feedback mechanisms?

David Deming: I mean, one thing, you said aside from grades, but one thing is making grades consequential. So part of what we worked on here at Harvard is trying to address the issue of grade inflation. And again, it's not, what is a grade? It's just an, it's sort of arbitrary. We could put it in Greek letters or put it on a hundred point scale. It doesn't really matter. It's just communicating some information. And if you walk into a classroom and you know that basically everybody's gonna get an A, you're not gonna try as hard. And so setting the mechanism of grading to make it compatible with people working hard and putting their best foot forward is part of the story. It's not the only part of the story, but it's important. And then also, we have a lot of applied learning experiences at Harvard. Our students do tons of extracurricular and co-curricular things outside the classroom. Those things are getting more and more popular every year. And I think that's a signal that it's a real part of the education we offer here at Harvard. Part of it is like students learning from each other, the consulting club, the environmental club, they get valuable experience. Many of them have real clients. And so they're essentially doing this outside the classroom system, I should say. I think, and there are some efforts to merge those things. Like when I was in college a gazillion years ago, I did this program where I was living in DC and I worked four days a week at the Center on Budget and Policy Priorities, think tank in DC. And then I spent one day a week in the classroom experience was like an all day experience and it was partly academic, but it was also designed to surface learning from the field in the classroom and reflect on it. And we have a lot of classes like that at colleges all around the country where you are having an experience and you're using the classroom to reflect on that experience. So it's like an inversion of the normal course experience. And I think more stuff like that is definitely good. The Northeastern Co-op program is another example of that. So I think college is headed in that direction. I don't think it should be just work.

Ben Zweig: Yeah, like more sort of Northeastern Drexel style programs.

David Deming: But just to say, not because colleges are meant to be job training. I believe in a liberal arts education. I think it's really important to learn a broad skillset, to develop a set of competencies, not just for work, but for life. Like who do I wanna be? What matters to me? What impact do I wanna have on the world? What is truth? What is beauty? Like those things are really important for developing citizens. But I think the point I'm trying to make is not like vocational or liberal arts. The point I'm trying to make is actually it's about maximizing learning. And a great way people learn is through applied learning. So like learning that's hands-on tends to be better than learning that's just theoretical in the classroom. And so I see that as like, even if we could do applied liberal arts learning, we could do applied vocational. These are two different axes.

23:52 Is a PhD worth it?

Ben Zweig: Yeah, it's interesting you say that. Like, I mean, sometimes people say like, "Oh should I do a PhD in economics?" And I was like, "It's probably not a good ROI on your time, but like it'll totally change who you are." Like it is an identity changer. Like, and so like--

David Deming: I would say there's absolutely no question it was a good ROI in my time. And probably yours too. I think I disagree with that.

Ben Zweig: Oh, well, yeah. You will change who you are, and that's not strictly financial. If you incorporate your mindset change and how you grow as a person, then I totally agree.

David Deming: So if the return is more than financial, you're saying fine.

Ben Zweig: Yes, yes. As a purely financial return, it's probably pretty poor.

David Deming: Yeah. I mean, Even then though, I mean, it's hard to run the counterfactual, but I mean, maybe for you it's different. But one thing I talk to a lot of students and young people and they, even if they don't say it, they often want career advice. And I feel pretty hesitant to offer that because just because it worked for me, it's like selecting on the dependent variable, like who knows? So I don't know that I really know what I'm talking about. But one thing that I, in reflecting on my own career, I'm curious to hear if you agree, is I would say that most of the good things that have happened in my career have been some return on an expertise that I developed that was scarce. So like I learned more about some corner of the world than almost anybody else. And then that became valuable in some way. And then people like, oh, he's a decent guy. Let's make him the Dean and whatever. So it's like, it all kind of stemmed from that decision. And I never would have done it if I didn't get a PhD. Like I basically spent 10 years, toiling in obscurity, learning everything there is to know about human capital or whatever, like knowing more than almost anybody about it. And at the time it felt like a totally nerdy and irrelevant thing to do, but like every good thing that's happened has flowed from that.

Ben Zweig: Yeah, yeah, it's a good point. It's funny how like, yeah, you like, I think you had made this case in a previous context. I mean, you would know better, but that there's sort of this like complementarity between like the obscure technical stuff and the social skills. And you really, yeah, you really need both. And they really reinforce each other quite a bit.

David Deming: Yeah, they do. And the thing about the expertise is that, and I think this is probably gonna be even more true with AI, is like, you need to learn things that, you need to be above replacement value relative to just asking Claude or ChatGPT or whatever. Like a higher and higher bar to cross. Yeah, and so you're never gonna beat it on breadth because it has encyclopedia knowledge of everything ever recorded, but you can beat it on depth because there are a lot of things that you and I know about economics that are not in the training data. And so like when I ask ChatGPT or Claude for a literature review, if it's on a topic I know nothing about, it's super additive. If I ask it for a literature review on something I know about, it's not that good. It's like undergrad level, it's good, but it's not as good as my knowledge. And so the lesson from that is acquire some deep expertise that beats our robot overlords.

26:58 Can depth beat AI?

Ben Zweig: Yeah, I wonder how persistent that is. So like, you know-- Yeah, I mean, I wonder like you hear about like proving math theorems and whatever like sort of doing things at the frontier of knowledge. And I don't know, like I don't, I don't, I've never seen these theorems, like what do I know? But I wonder if there is a point where even depth is like not a comparative advantage. And like maybe intelligence at large is just commoditized. And then like surely there are other things which compliment raw intelligence.

David Deming: Well, I think maybe it depends a little bit on what you're using the intelligence for, you know? So why is depth valuable? I mean, yes, it's valuable because you can prove, you can make progress on questions in math that so in that sense, you're out competed. But depth is also valuable because it teaches you about what questions to ask, what things are unanswered, what's on the frontier. And like, maybe that's not really an intelligence thing. That's a what people call taste or a judgment thing, you know? Like I read a lot of labor economics papers, which ones catch my interest? I can't really articulate why always, but like that, to the extent that it has some value about above the AI that could persist even if the AI is super intelligent. In other words, depth isn't just value for greater intelligence, it's valuable for context. And what do other people think about this and stuff like that?

28:22 How do you develop taste?

Ben Zweig: Yeah, that's a very good point. I mean, so if we think that like judgment all sorts of like things that humans are uniquely good at are an advantage. Is there, is there, are there ways to, I mean, I guess you're saying depth learning something deeply is one way to like exercise that muscle. Are there other ways to exercise that muscle? Like, like how do we develop in ways that are more useful today than they used to be?

David Deming: How do we develop taste or develop judgment?

Ben Zweig: Yeah.

David Deming: Yeah, I mean, so I don't know the answer, but I'll idly speculate. I think like, so I was thinking about this because there's this discourse around taste and judgment. I was thinking about this. Like, what does this mean? There's something unsatisfying about it to me. Here's how I would say that taste happens. I don't know if this is right, and this isn't scientific. This is just my instinct. So like, do you ever, do you ever like play video games when you were a kid?

Ben Zweig: Not really, but I mean, enough to like know what they do.

David Deming: Dungeons and Dragons or something like that. Okay, maybe not, fine. So I did, all right. And so one of the things about that is that like, if you're gonna like learn the landscape of something, you can either learn it in a way that gets you minimally to where you wanna go. So like, where's the, where's the boss that I have to beat? Or you can like learn all of it and around it and go in like every room and do everything. And if you're just trying to accomplish the goal, definitely the former is better. But when you learn things, de novo, you don't just take shortcuts. Like you don't just accept the Cliff Notes version of the book, you actually read it yourself. You actually go, then you develop a take on things that is slightly different than the average person's take because you've done it yourself. So like 99% of what I would say about Moby Dick is the same thing that everybody else would say about it. But because I read it myself, there's like a 1% tilt or edge that you have on it because you did it yourself. And to me, that's what taste is. Taste is just like I did the work myself and I have a slightly different angle on it as a result. It's not better, it's not worse, it's just different. Like I watched The Odyssey and I didn't talk to anybody else about it. So I have a take that is not exactly replicated on the internet. But if I go and I read the summary first and then I watch The Odyssey, my take is just gonna be derivative. So it's like doing the work yourself, I think is essential to taste. And so that's another, I think, reason why developing expertise is valuable because it gives you your own unique perspective on the world that will be valuable in some way maybe.

30:42 Hot takes and learning in public

Ben Zweig: Would you recommend that someone try to post their hot takes on the internet and try to get feedback from that?

David Deming: Isn't that what we're all doing now? That's like, that's the economy we live in now. Most people post hot takes on the internet and get feedback about it.

Ben Zweig: That's funny, I recently saw that some university, I wanna say Notre Dame, but I'm not sure, just had a new major in being an influencer. And is that getting railed against by all these people, being like, oh my gosh, what have colleges come to? But maybe that actually is a way to get feedback.

David Deming: Maybe, I mean, I do worry that people are sometimes learning the wrong lesson from this. So let's take you and I as an example. We're spitting hot takes at each other right now, having a grand old time and I write a Substack and we do this stuff, okay? And so if you are just seeing that, you might get the mistaken impression that's our primary job, but it's not. Like you've got a company, I'm the dean of the college, like we're doing this stuff in our spare time. Like I do it for the love of the game. Like I'm not trying to make money in podcasting, I'm not charging money for my Substack. I don't really care, I do it because I enjoy it and I want to have a conversation with a community of people and I'm not trying to go viral and it just doesn't matter to me that much. And it's like a hobby, a fun hobby. And I think it, part of the reason I do it anyway, is I want to learn these things too. Like I want to, like I'm gonna have a podcast come out soon about like where I read all of the papers released by the AI labs and try to condense that into like a few lessons and takeaways. And like I did that because I wanted to do it. And then I thought, well, if I have the discipline of having a several thousand people listen to this, then I'm going to do a better job than if I just do it on my own, because it's like a discipline. And then I can share my learnings for what they're worth with other people. So that's not a career, that's a hobby. And, but the thing that's paying the bills for me and for you is not that. It's like some, I'm like running a college and you're running a company.

Ben Zweig: Yeah, but that might be like a good like way to learn. Like that is your training environment. So like to the extent they use it to like learn stuff rather than produce stuff, like maybe that's effective.

David Deming: It's definitely effective. And it also, at least for me, I don't know if you, it's like keeps me sane. Like, I think of the job I'm doing, like running a college, like running a college, it's impactful, but I didn't get into this to do that. Like I like ideas. And to me, this is like the way I start, I can feel like myself again is to like talk about research and have conversations with people like you. So this is like pure consumption value for me.

Ben Zweig: Yeah, no, I feel the same way. Like I actually feel like I mean, I feel like I'm like an academic economist in my soul, but I'm like running a company because I feel like it's the, it's a undervalued way to contribute to advancing the science of labor markets. That is like through data. Yeah.

33:29 Do good academics make good CEOs?

David Deming: Wait, can I ask you a question, Ben? So if you were to estimate like the correlation between the skills required to be a good academic economist and the skills required to run a company, what is that correlation? Like in the population?

Ben Zweig: It's a good question. So if you'd asked me five years ago, I would have said very low, but like now, because I didn't really understand what it is to be an academic, really, when I was like doing my PhD, I thought you just like think really hard and like write papers and do research and stuff. I now that I'm like working with a lot of academics, like, and delivering them data and stuff, I realized that like to be an effective academic researcher, you need to like, you need to hustle. You're like, you're kind of like selling stuff. You're finding you're like dealing with vendors and like you're like facilitating schedule. You're trying to like manage teams. You're an entrepreneur in a way. It's like actually quite a lot closer than I would have guessed.

David Deming: So does that mean you think most of you talk to a lot of academics. Do you think most of them would actually be good CEOs of startups?

Ben Zweig: I think the really good academics would be good CEOs of startups. But that's not to say the smartest academics. So, yeah, like, yeah, I'm trying to think of like, who's a really good example here? Yeah, like, I mean, yeah, I'll tell you like, I mean, yeah, you brought up Luis Garicano. Like I just spoke to him for the podcast and like, he's like, got a lot of things on his plate. He's like managing a lot of stuff. He was like he was a politician and like he's like really like trying to like promote ideas and like deal with tension there that I think is like impressive. But then on the other side, I feel like I think about Erik Brynjolfsson as like someone who just like, like, I feel like his superpower is he like, he runs a lab basically in a way that like, I feel like he's common in physics, but uncommon in economics, where it's just like he's got like a team that is like aligned on a mission that's all like digging into stuff. And I'm just like, he's like a manager, essentially. And like, it's very impressive. So those are the two that come to mind, but in very different ways.

David Deming: Now those are good examples and both of them are doing really interesting work, but are not only doing academic work, you know? So Erik has this company, WorkHelix Luis is like, he was a member of parliament and so it's like, and that is, I have to say, one of the best, it's something I wish I had known because there's some pretty hard times when you're going through grad school and you're like, is this really worth it? I'm so isolated, I'm doing, but like once you become an academic, it's like the best part of it is it's not one job. Like I could be dean for a few years, then I'll go back to doing research, I can write a book, I can, whatever, I can do lots of different things and still have the same job. And so it keeps you, it can keep you refreshed if you're willing to engage and like to me, the most, like this job I'm doing is completely, I would say in terms of daily activities, almost totally unrelated to the job of being an academic. And so it means I'm basically learning a new career on the fly and that in itself is enjoyable. Like I'm learning how to do things.

36:45 What it takes to run Harvard College

Ben Zweig: Do you think they kind of reinforce each other? Like, do you think you were a better dean from having been an academic or are they kind of unrelated?

David Deming: So the way I think about this, I've said this to a few people and I really think it's true, is it like the Venn diagram of people who can get tenure at a top research university and run anything, it's like tiny, you know? So like, I'm probably a well below average as a manager, but far above the average of academics who could be managers, you know what I mean? So like that's a kind of good place to be in because it's like, you don't have to be that good to do a good job because the replacement value isn't that high, you know? So I think it's like, it's not just a different skill set, it's a different mindset, you know? Like a big part of being a good leader, certainly in academia, but anywhere I think, is like not wanting to take credit for what you've done and actually spending most of your time and energy trying to elevate other people and activate other people. Whereas being an academic is very much about like building your own portfolio, your own identity. It's a very, I don't say self-absorbed because it's a very self-focused enterprise. It's like the business is me and I got to make a name for myself. And so you just have to totally invert the way you think. Being the smartest guy in the room is actively harmful if you are trying to lead something. People like this guy, you know? So you actually hold your, try to speak last, like try to I try a lot to like make the people who I work with empowered. So it's like, I don't want them to feel like they always have to check with me to do anything. I want them to have some room to run and like feel empowered to do their best work and support them when they, you know? So that's just not the mindset you're in as an academic a lot of times.

Ben Zweig: Yeah, do you think you're bringing your mindset from parenting to being a dean?

David Deming: Maybe, maybe. I mean, I don't want to make it seem like the people I work with are kids. Like a lot of them are older than me and more distinguished in various ways. But I think about it, and that's another thing is like, people say, hey, what kind of management training did you get? And the answer is like none. Like I have no idea what I'm doing. So that means I need to be kind of humble about this. Like the Harvard College has more than 300 employees and there's 7,000 students. There's like a lot of management and I don't have a lot of experience doing that. And that means I have to rely on people who are career academic leaders, people like the Dean of Students, Tom Dunne, who's like been doing this for a long time. And like, I can't really, not only is it inefficient for me to tell them how to manage, it's actually like hubristic. Like he knows much more than I do about it. And so part of this, I'm learning from the people who at least nominally work for me. Like, because they have so much to offer. So it requires a kind of humility, I think humility to be done well. It's like, I don't really know everything. I know my little corner, you know?

39:23 Management vs. leadership

Ben Zweig: Yeah. So from like, so you've like studied management and you've like done management. And like now that you've kind of done both, like how would you characterize like, what is management while standing on one foot?

David Deming: Harder than it looks. I'll tell you that, Ben. Yeah, yeah. No, it's, I mean, so I would think of, so there's management and there's leadership. Okay, and these are different things. And my job for whatever reason requires both. So leadership is like setting the strategic direction, giving speeches, shaking hands. It's almost political, you know? So like, I am a visible administrator in the life of Harvard College students. And like, I walk around and they see me and I have, there's like a self presentation and like a owning that you have to do that is very leadership oriented. And like just setting a good example and setting the strategic direction for the school and rallying people and all that stuff. And then the management is like, okay, we've got a lot of work to get done. Like who's gonna do what and in what, and how are we gonna organize people to get that work done most effectively? And that's what I've studied a lot more is the management side of it, just to be clear. And I think that's a really important skill. I mean, I always knew it was important. I always studied it, but I think two things. One is I'm living it so you can see how much it matters to have a good manager. And second, I think like with AI agents, we're all managers now literally all managers now. And so that skill of you could think of it like in some of the work that we've done, that you alluded to, we think about it as essentially allocative efficiency. So you've got a bunch of resources. How do you organize them to their most productive uses? Yes, you can think about it like you got three workers and you have three different jobs you wanna do, who's gonna do what? And there's some like productivity schedule of each of them. And it's like, so we have these assessments we've built where we actually ask people to do things like that. And it turns out being good at assigning workers to their best use in these artificial lab settings is correlated with your ability to like run a retail store in South America or like be a manager in a real world task. And so that's kind of what I did in my, are still doing in my research life is like building assessments of this managerial skill. And I think it really, living it has made me believe more in that approach, not less.

41:28 Can management be taught?

Ben Zweig: Is that something that is like being taught in some way? Like, I mean, that happens naturally, but to the extent that people are trading tasks and like allocating amongst themselves, or maybe there's like some manager who's like doing the allocation, like and sort of like doing this reconfiguration, like is that teachable?

David Deming: I don't know, I think it is. I think it's like anything, like I think most things are teachable, but also there's variation among people and how good they are at baseline at it and how fast they learn. I think that's like anything, you know? So my sense is like, you probably can teach it, but it probably matters whether you're naturally oriented towards this type of thing. I think a big thing that is like largely outside of our assessment that I think matters in the real world is this feedback thing again. So like, how open are you to adjusting what you do based on real world feedback? So like you come in, like when I started as Dean, like I really didn't know who to trust. I didn't know who was good in their job. And it was like, how attentive am I to, okay, I tried some strategy, either worked or didn't, and I'm willing to adjust. Am I willing to try assigning people to different roles to see if it works? Am I willing to mess around with formats and then really pay attention to when the orientation is humming and when it's getting bogged down? And I think a good manager knows when to stay above the fray and knows when to really dig into the weeds. So in my Dean job, like I found that, at least initially and maybe for the long run, I've had to really dig into the details of things, not so much because I have a, I mean, I might have a view about how it should be done, not because I want to micromanage people, but I really need to understand the context in which people are operating in order to make good high level decisions about who should be doing what. And so I spent a lot of time, both with staff and with students, just like, there's a problem, tell me about it. Tell me more about that. Tell me, and just like digging in till I feel I really understand what's going on. Yeah. And obviously it's not sustainable to do that forever, but you have to do it enough that you know like what the heck is happening out there in this big organization you're trying to run. So I came away from that with a feeling that like a good manager is not just only doing the like high level keeping the trains moving, they also strategically dig in on certain things so that they can understand the texture of it. And you have to know when to do each one.

Ben Zweig: Yeah, interesting. That kind of makes me think of what you said earlier about like, how you need to like sometimes dig in and be like a frontier expert in some domain. Because like maybe because then you see the whole like Dunning-Kruger cycle where like, you know like how hard and complicated things can be. Like when things get really, really deep. So maybe just having gone through that experience can like allow you to decide like when you can like abstract away from a certain problem or when you need to like drill into some like bottleneck.

David Deming: Yeah, no, I think that's well said. And that's where I think the PhD training does come in because if you think about what you got to do to write a dissertation, like you literally learn more about some smaller corner of the world than anybody else. And then you have to zoom out at the end and explain to everybody else why they should care about this thing that only you know about. So you have to go in really deep and then zoom back out. And I've found that like I have this, I've been, I've trained myself over the years that there's this like, just like little thing in my head. If I like 99% understand something, but 1% of it still doesn't make sense to me, I just have to know. So like, wait, explain that to me again, like what's going on? And I think that is extremely useful mental discipline. Like you're not satisfied with understanding 95% of the issue, you need to really dig in, but then you can't live your life that way, you know? So you've got to again, be able to zoom out. So I don't know, maybe I was born this way, but I don't think so. I think probably that the experiences I had doing research helped me in that way.

Ben Zweig: Yeah I mean there's like, there's like contradictory takes a little bit of like, is PhD good? Do they make good managers or bad managers? I mean, maybe they're not so contradictory to the extent that like the training is like causally like associated with like good management, but like correlation wise, like just selection wise is like selects people who are bad at management.

David Deming: Well, that's the thing I think is where, I think that's spot on, like what you're worried is that, I don't know that a PhD makes you a worse manager. I just think the kinds of people who select and are looking to getting PhDs are often, I don't wanna say bad managers, it's people who are excellent individual contributors. It's people who like, because what the thing about the PhD job that is so different than anything else, it's like the blank screen phenomenon every single day of your life. Like nobody is looking over your shoulder telling you what to do. You can do anything you want at any day. And some people find that liberating and some people find it crushing. So you're selecting for the people who find it liberating. Like, I don't want a boss, don't tell me what to do. I'm doing my own thing, only I know. And so it's like that type of person is often not oriented toward management, not always.

46:25 How young people feel about AI

Ben Zweig: Yeah, yeah. I mean, okay, so thanks so much for your time. I have one final question for you. I think you're like super generous for your time. Yeah, yeah, no, I mean, we could go on for ages, but I don't wanna hold you. I know you got a lot to do. But I mean, one thing I wanna, I just gotta ask you, I mean, now that you're like, Dean of Harvard, it seems like you're also, I feel like you're also taking on this role of like this communicator of what AI means for young people. Like, this is a moment of AI. And like, young people hate AI, they're nervous, they're anxious, and like it's like a bad time for them. And like, how do you find that? Like, what are young people really like feeling? And like, what is the essence of that? And how are you able to like communicate to them how to navigate this like technological frontier?

David Deming: Yeah, well, let me tell you why I'm doing it first, and then maybe I can say, so--

Ben Zweig: By the way, I love that you're doing it. I feel like you are the voice of like here's like what young people need to know, and like, it's very valuable.

David Deming: Well, thank you, Ben, I appreciate that. I mean, I think I've talked to enough young people, but mainly Harvard, but elsewhere too, who do find value in it. And also, they're curious so you say young people hate AI, and some of them do hate it, but not all of them do. And a lot of people are just very, they just really wanna know what's coming next, and how should I think about this big new, and people have an intuition that the ground is shifting out of their feet. And that causes a lot of anxiety, but it's also a huge learning opportunity, and not like an opportunity for me to teach you what the future is, because I don't know either. It's like, we're learning together. And so there's something just intensely, intensely interesting and collaborative about this moment we're in, because it's like, look, I'm an old guy, like, this is your technology, your generation's technology, not mine. So like, the goal here is I'm gonna maybe try to tell you a little bit, but eventually you're gonna be teaching me, because you're gonna learn, this is your world, and not mine, and I'm gonna be a geezer pretty soon. And so I try to approach it from that, like, hey, there's these college students coming, they wanna know about it, I wanna know about it too, let's do it together. And I gave this, like when I started as dean, I gave this speech called Convocation, which happens at the beginning of the year, it's my first year, it's my first big announced speech, really, like my debutante ball at Harvard College. So I decided that I was gonna give, often these speeches are like platitudes, like make the most out of your time here, make good friends, and I was like, I'm not doing that, this is like the first lesson, like I'm gonna do it classically. And I was like, I don't know how this is gonna go, but I'm gonna give it a shot. So I talked about like the research on, I said there's this storm clouds approaching, and everyone thought I was talking about the Trump administration, but I was not, I was talking about AI, I was like, you are worried about whether you're gonna find a job at college.

Ben Zweig: There's no shortage of storm clouds approaching.

David Deming: No, lots of storm clouds. So, and I was saying, look, but here's why I think you should treat it as an opportunity. Not like it's gonna be good for all of you, and everything's great, but like, there's opportunity in this moment. And history tells us that the opportunity is often seized by talented, ambitious young people who are open to a new future. And I went through some of the evidence on that, that like, past bust, technology boom, bust cycles have created the next great companies. If you look at the top 20 S&P companies in terms of market cap, it's like 18 of them are new as of 15 years ago. Like, there's just tons of disruption, you know? So if you went back and you, the example I gave in the speech is like, if you were looking for a good job in 1995 when you graduated from Harvard College, you go to like Bank of America or Procter & Gamble, probably fine companies, or you go work at a startup like Google or Nvidia. These were startups at that time, and now they're huge companies. And so that is, I believe, the kind of moment we're in now. And if you're willing to go out there and take a little risk, there's opportunity in it. It's not like a Pollyannaish message, it's a hopeful framing that I think speaks to people. Because first of all, they want to believe it. They want to believe this is not all bad. And I think it's actually true, it isn't all bad. There are some bad things about AI. But the thing is, like, I'm not trying to be a guru. Like, I don't know what's coming, nobody does. I'm trying to like say, okay, I'm learning this too, let's learn together. And so I think if anything, that's what speaks to young people, is like I'm taking them seriously. They're anxious about it, and I'm engaging with them about it. But I'm not telling them I have all the answers. I'm some self-assured boomer telling them what to do. I'm not doing that, I don't know. But I am saying, here's how I think about it. Here's what I've learned from my experience with this. Like, I didn't get an email address till I was in college. Like, the internet didn't really exist until I was a teenager. So like, I've lived through a few of these things, I'm sure, you know. So it's like, I don't know everything, I know something. So I think it's, I think if anything, it's that openness that is what young people are looking for. I think they're quite tired of being told by older people how to be, but they are interested in learning. And I don't know, I hope that they find it refreshing to see somebody who's past their prime, like me, still trying to learn something new. I hope that's a source of inspiration for them. And I hope they will be that way when they're in my advanced age.

Ben Zweig: I hope so too. I mean, I feel like every generation is like, also doesn't want to hear it from old people.

David Deming: Yeah, do you think they're unique in that way? No, not at all. And like, the other thing is if you read about Harvard in the newspaper, or you're like an alum or a donor or something, and you like hear the way other people talk about college students, you get this impression that it's like college campuses are a war zone and everyone's at each other's throats. And it's just not like that at all. Like I see students every day and like almost all of the conversations I have with them are just ordinary student stuff. Like, what am I gonna major in? Like, what am I what's my job gonna be? My girlfriend just broke up with me, whatever it is. It's like normal stuff. And you just realize they're just people like us and they have hopes and dreams and they're tremendously talented, you know? And I think being around college students and really listening to them, it keeps you young and keeps you interested in the world. Cause you realize like these are the people the things that I think are weird, like the fact that people scroll on our phones all the time are just part of the water to them. It's just totally natural. And that means the world is gonna be different when these people come into power than it is now. And I find that kind of exciting and interesting.

Ben Zweig: Yeah.

David Deming: Yeah.

Ben Zweig: Yeah, no, I do too. It's very exciting times. Well, I hope you're steering them in the right direction or at least like giving them the guidance to steer themselves in the right direction.

David Deming: Maybe they're steering me that's.

Ben Zweig: Yeah, hope they're steering you in the right direction. It sounds like it sounds like they are. Thank you so much for being part of the show. I gotta recommend your Substack, "Forked Lightning" and your podcast, "The Context Window." Really great having you on. Thanks for joining.

David Deming: Thanks so much, I really appreciate it.

Ben Zweig: 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.