Byron Auguste on Rewiring the U.S. Labor Market | Conversations with Tyler

Hello everyone, and welcome back to Conversations with Tyler. I'm here with Byron Auguste, who is president and co-founder of Opportunity at Work, a civic enterprise which aims to improve the US labor market. He also served for 2 years in the White House as Deputy Assistant to the President for Economic Policy and Deputy Director of the National Economic Council. He worked at McKinsey Company for many years, and he was elected principal and also director. Prior to McKinsey, he was an economist at LMC International, Oxford University, and the African Development Bank. He is author of the 1995 book The Economics of International Payments Unions and Clearing Houses. He has an undergraduate degree in economics from Yale and a DPhil in economics from Oxford University, PhD as a Marshall Scholar. Welcome.

Thank you, Tyler. Happy to be here.

So, so much of your current work is about labor markets and training. How much does advancing information technology make training much harder? Are we even keeping pace with technological advance?

Well, I don't think technology makes training harder. I mean, it should make it easier. I mean, there's so many more ways to deliver information, to engage with information, and It's funny because a lot of times I hear people talk about, "Oh, in a digital economy, things are going to be so much harder," but I actually think in many ways it's easier. I mean, think about what a— like someone who's a janitor even or can become sort of a building manager with digital technology. Working it on an iPad is easier than all the physical tools you need. So I think it makes training easier.

But if someone has to just teach me a new physical task, isn't that easier than me having to learn a new piece of software? Even when they changed my Gmail, I got upset, right? It's disorienting.

But I don't see why that would be because a new piece of software is designed typically for people to use. Like, you know, if it's— you know, it might be a little bit complicated, but if it's Salesforce or whatever kind of software you're using, it's designed for humans to use and it could be better for humans to use, but I just think it's not— no, I don't think there's a— fundamental distinction between learning digital tools and learning physical tools. And learning digital tools, you can learn them— you're much more likely to be able to learn them anywhere. I mean, with physical tools, like let's say in textiles or design things, you need to kind of be in a center where there's a lot of other artisans. And with digital tools, you know, you can be more of anywhere.

But jobs where you have to learn a new digital tool, they'll typically require at least a 4-year college degree. So if learning that stuff is so easy, why should the jobs require the college degree?

Well, I think the word require is doing a lot of work there, Tyler. I mean, 75— by that definition, 75% of admin assistant jobs require a bachelor's degree even though two-thirds of admin assistants don't have a bachelor's degree. So when you have a couple million people doing a job without a degree and that job requires them doing a degree, maybe we should reevaluate what requires means. I don't talk about those jobs requiring a degree. Companies often screen out people. They preemptively prescreen people who don't have a degree, but that's not the same as saying it requires it.

Sure, but there's a probability distribution. Wouldn't you agree that people with a college degree would be considerably more able to learn new pieces of software than people without, even though a few million of those without certainly have done it?

So I think what you're describing there is you're describing two distributions and you're saying that on average, the— someone with a bachelor's degree is more likely to have that skill set or learn it than someone without a bachelor's degree. But I heard averages over. Somebody said that. So if you look at the overlap of the distributions, you've got— if you've got millions of people that you're excluding just because your arbitrary definition of them means that on average, that's kind of dumb, isn't it? I mean, it's like prejudice, right? It's prejudging somebody based on a statistical distribution even though they quite likely could have those skills.

But labor markets are in general based on statistical discrimination, right? So what's the metric we should use for deciding which measures of statistical discrimination are dumb and which are smart?

Well, labor markets look like statistical discrimination when you look at them at the market level, but at the level of the point of hiring, The question is you have to typically— particularly when you're hiring on a digital platform where you might have hundreds of jobs, hundreds of people apply for your jobs, you do have to shortlist people. So there is some form of screening based on some sort of signal that is required just from a transaction cost standpoint. But these days, why not? There are so many better ways to do it than a bachelor's degree. 25 years ago when you needed to just have like a keyword search. Fine. But now you could— I mean, you could post the training that you do for your own people once they have the job and see who could do the first 3 modules of the training and give them an interview. There are so many different ways. But the point is you should screen in based on skills, not screen out based on pedigree. That's the fundamental issue.

How will we get or retrain more young men to be willing to train to be nurses or do elder care?

Well, that's a great question because we need a lot more nursing, a lot more elder care. In elder care, I think paying people more would be a good start. I mean, that typically pays very poorly. So I think that should not be overlooked because it's still a market. It's both ways. I think secondly though, there is a— having more pathways and ways to kind of engage. I mean, most people move through the job market based on the skills they have, the experiences they have. And so similar to— if there's not a lot of men in the early stages of care work, it actually is harder for them to move in. Just like if there's not a lot of women in the early stages of construction work, it's harder for them to move in even though there should be a lot more women in construction and a lot more men in care.

The quality of the police in America seems to be quite low in many jurisdictions. How do we fix that? How much of it is just paying them more? How much is training? How much is culture?

Well, it's not an area where I'm an expert, but it does seem like a lot of it has to do with culture, like organizational culture. While I'm not an expert on police, I think I have had a lot of experience with the culture of different organizations. One thing that's so striking is how much cultures— organizational cultures tend to replicate themselves, how much you can come in with a very different kind of attitude and aspiration and you end up norming to the norms of the organization, and I think that probably is applicable in police departments.

At the margin, if you're trying to retrain someone, would you rather they had higher IQ or higher conscientiousness? Which is more scarce?

Well, it depends in part on what you're training them for.

Sure, but on average, people who are 25 and over—

I think IQ is overrated and conscientiousness is underrated for the most part because People can learn new things, but learning, as you know, requires— learning can be uncomfortable. First of all, it requires realizing you don't really know a thing, paying attention to where, "Oh, I know— I sort of know this," but looking at the finer points. So yeah, I think conscientiousness probably matters more on the whole.

If you think about the value of a typical 4-year college degree, say from a large state university in the United States. How much of that value to you is learning and how much is signaling or something else? So Bryan Kaplan thinks it's about 80% signaling. I disagree with him, but that's a number, right?

Because you think— how much do you think is signaling and how much do you think is learning? Are you saying the alternative is learning at college or are you saying something different?

It depends a great deal on the choice of major, but I think 60 to 80% of it is value of learning. There's definitely some signaling, there's some networking, there's some consumption thrown in there. Those would be my numbers. Yeah.

Well, so I thought about this a lot and— because the college wage premium has been pretty stable, right? People who have college degrees do make more than those without. But I think there's something else besides learning in college and the signaling, which is if you look at how these studies work, you're typically measuring an income gap, let's say, 10 years later and we've got a two-tiered labor market in this country. So you started this question with like a typical bachelor's degree, but there's no typical bachelor's degree as you know by topic, but also by school. I mean a college degree isn't one thing. It's like 18,000 different things. So that's the first point. So as we talked about before, there's a huge overlap in the distribution of actual skill, actual competency, IQ, everything you want, conscientiousness, all the rest of it. If you picture someone coming out of let's say a flagship state university versus somebody coming out with most community college degrees or who didn't finish college but who has basically the same kind of characteristics, the first one, the one coming out of college is likely going to go into entry-level sales, entry-level management, entry-level technology and from then on, their employers will invest in developing their human capital as part of their employer's business model and competitive strategy. Whereas the second person, same— who starts with the same skillset, same attributes, will probably be put into entry-level retail warehouse, factory floor, and for the most part, they will be trained for safety compliance and efficiency and not for career development and so forth. So after 10 years of that, well then you're going to have a massive difference in wage outcomes that in the particular scenario I'm talking about really don't have to do with what they learned and they have to do with the initial signaling. But we got to recognize that the fact that people are placed in jobs and kept from moving into jobs that they have the skills for— we mostly learn on the job. That's where most of our human capital is developed. So I think we've got it backwards when we think about a skills gap being driven by education. The skills gap is a result of the opportunity gap that we've imposed. Not entirely, but the implications for the college wage premium is that there's 3 things. There's signaling, there's learning in college, but there's also what are you allowed to learn on the job as a result.

Now you think conscientiousness is quite important. I agree with that. I also see in the literature from personality psychology that Big Five personality traits, they don't change that much over time. So if you're conscientious when you're young, You're probably still pretty conscientious when you're old. So if we have an America with, say, 37% of people get 4-year college degrees, the ones who don't on average are less conscientious. Why think that 10 years later, given their lower conscientiousness, they can be retrained when they either couldn't get through college, wouldn't finish, didn't even start, whatever the case may be?

Well, I think there's 2 problems with what you just said, Tyler. One is you slipped in the on average. Whereas a person is not an average. A person is an individual. So— which is a very big deal because like I don't know what the average level of conscientiousness is for college graduates for SNOT, but like let's assume that it might be higher on average. That absolutely doesn't mean— well, what it means is there's tens of millions of people who don't have college degrees who would rank very highly on conscientiousness. For example, if you start at college and you're a first-generation college student and then your mother who's providing for your younger siblings gets sick and you drop out of college to earn money to support your family, are you less conscientious than the more affluent person who stayed in college? I don't think so. So in other words—

That's not why most people don't finish, right? It isn't whether or not you finish college the best test of conscientiousness we have?

Absolutely not.

What's the best one we have then?

I think if you're an employer looking to hire someone, the best test of conscientiousness is how they do at the job they're in. So what happens—

Well, ex post, but ex ante.

Well, ex ante—

Because tautologically, the people who do well do well, right? But ex ante, what's the best measure of contribution?

So if college— college right now is a signal of many things, but I think the statistics would say it's a huge signal of affluence and generational wealth. Like being able to not only be prepared for the test to go into selective college,— and being able to complete it, having the social capital, having the positions. If the position is that the people who complete college never make mistakes and therefore they're the most conscientious, well, I disagree. I think the people who complete college make lots of mistakes, but they tend to be the ones that have more cushion, more fallback, more shots on goal. There are millions of people who, yeah, they aspire to go to college. I mean, when we talk about people who are skilled through alternative routes or stars,— that don't have bachelor's degrees, about 40% of them actually do have college credits, but they didn't finish for a lot of reasons. For example, they had to work and they had to work a lot and if their work schedule doesn't correspond to their class schedule, then maybe they can't get the classes they need and they basically would have to go 6, 7 years. They run out of money. They might just plain run out of money. There's all sorts of family instability. So in other words, I just don't agree with the view that says that the reason most people who don't complete college don't complete it is because of some Animal House behavior. I disagree. It's mainly because of the economic and social conditions and what they have to do to manage like a totality of life. Then finally I'll say most people in college right now are not like 19, 20-year-olds, right? Most people in the college are actually working learners. That's the typical college student today. So they already come to this with children, with responsibilities, so I just think that our picture of what— college is is very different than from 50 years ago. I mean, it should be different than 50 years ago. It hasn't caught up.

You think it's a kind of market failure that the sheepskin effect is so pronounced in wages? The sheepskin effect meaning someone who quits just short of finishing earns much less than someone who finishes.

I think it's an enormous market failure, yes.

But why don't people then start companies or for that matter nonprofits and hire up all the people who almost finished because they would be way underpriced? But I don't see anyone doing that.

Yeah, there are people—

I'm not doing that. Are you doing it?

No, there are actually lots of people doing it. I think it's a small scale because obviously it's easier and it's a quicker path to profits to market to employers people who they already think they want, right? So to market based on pedigree and so forth. Of course employers want skills and in places where skills are very easy to quantify, you do get businesses like HackerRank if you will or— there's a bunch of businesses in tech for example that are really about Assessing people's skills. That's definitely happening for software developers and then there are also businesses exactly like the ones you're talking about that actually hire— they train and hire people and they actually provide outsource services to companies because those companies wouldn't hire those individuals because they don't have degrees but those individuals are great problem solvers. They're great and again, technology is the area where you see it the most but that is happening actually. That's an arbitrage opportunity for sure.

Now you've spoken in Singapore. What does Singapore need to do to support more creative talent?

Well, I think Singapore's gotten a lot of things right in education, and I think one thing that they recognize is that there has been a bit of a conformity. And so I think one of the best things you can do is to widen the aperture in which you think of as talent and to recognize that sort of hybrid skills are so valuable. So I think a lot of people, not just in Singapore but in this country, will tell their kids, "Well, you should do engineering and math," and there's a lot to be said for that, but software development plus art, software development plus music, there's really incredible value to having, if you will, orthogonal sort of skill sets, hybrid skill sets because then you notice things other people don't and I think so much of creativity comes at the margin between disciplines. I also think we'd get a lot more creativity if you had kind of more diversity of talent and in Singapore context, that has one meaning and in our context, that has a different meaning but really two different life experiences combined with the rigors of like learning something well, right?

But there's plenty of ethnic diversity there, right?

Yeah, no, there is.

And they make a point of mixing people in terms of neighborhoods. A lot of Singaporeans will say to me, well, we've chosen high conscientiousness and in doing so sacrificed creativity. They may word it slightly differently, but what do you think of that Singaporean fear? Well, does that mean conscientiousness is then overrated, not underrated?

Well, I'm smiling. Well, it's— I don't think you look at Singapore and say if Singapore maxes on conscientiousness and it went from being an extremely poor country in the 1950s, an extremely vulnerable country to one of the richest in the world. I don't think that speaks poorly for conscientiousness. I'm smiling though because when I— you're alluding to I spoke at Singapore at their jubilee, the 50th anniversary and they invited me out there to speak about kind of economics and technology and I looked up the Gallup ratings on— engagement at work and these international ratings and there's a set of people who are so unhappy at work that they want to sabotage things and there's a set of people who are so engaged at work that they're excited and there's a set of people in the middle. Singapore has like the highest percentage of people in the middle. They're neither sort of lit up nor are they disgruntled and I mentioned that to the crowd and there was a lot of knowing laughter.

What is your very current opinion of the German model for vocational training? I've been hearing for at least 20 years. We all need to do it. But I look at the quality of other parts of German education, K through 12, we would call, seems to be fraying, quality of German infrastructure poor, a lot of bad German decision-making, energy policy, other areas. Are they still actually on the frontier of vocational training, or is that something we should have been recommending 20 years ago?

So I've spent some time in— in Europe on this question. Look, I think with the German system and a lot of the Northern European— Swiss is similar to this— they are very, very good at pathways of education to employment. So in other words, to have multiple pathways, you're in school like upper secondary, you have pretty clear choices as to pathways and you don't have to have what the equivalent of what we would think of as a bachelor's degree but you get a very rigorous technical education and a pathway into well-paid jobs. Now, I think it's really important— so I think that's a positive and remains a positive. I think it's really important to recognize though when we talk about adopting the German system, people here tend to say, "Oh, German companies do apprenticeships. So US companies should do apprenticeships." But it's not just the German companies do apprenticeships. German companies plug into an apprenticeship system and they play their part in it. Upper secondary education, labor council, industry councils, these kind of collective entities do the heavy lifting and they plug into it. So I don't think it's that applicable here and in particular, I don't think the Hanseatic League and the hundreds of years of sort of history that went into the development of that system, you can't just like replicate that here. I think the US system would look different. I think the weakness of the German system and partly why you're seeing this more and more, Tyler, is they actually don't use this kind of system. For let's say you're in your 30s and there's been a change in industrial structure and you need to retool. They don't use the apprenticeship system for that. When I was— just a couple of years before COVID I was over— the OECD and the EU had a session on the future of adult education as part of their apprenticeship week and adult education in the sense of those already in the workplace is not as strong as the apprenticeship system and I think the dynamism of the economy means that just an education to employment model is not going to be enough. So I would like to see an American system leapfrog the German system, be much more data-driven, be much more inclusive, not just about education to employment but about everyone and their pathways through the job. We can see in the data that such a system is possible.

What's the best US movie or TV show about finding the perfect job?

About finding the perfect job.

Or a very good job. Is it Moby Dick?

Do you think that was a very good job?

Well, maybe for the time, but my follow-up question, which I'll toss out now as well, is do you view popular culture as your ally or enemy in what you're trying to do?

Well, right now I would say it's not mainly the ally, but I But back to your question about American novel. I would go for Mark Twain's Connecticut Yankee in King Arthur's Court. He became like a court wizard based on his practical Yankee understanding of things like the almanac and the pattern of solar eclipses. So that's a good example of practical knowledge leading to a pretty great job. I mean, in those days in King Arthur's Court, court magician was right up there.

And television today?

I think television— well, I don't know how much television really has to say about getting a great job. I think one of the worst and least realistic depictions and one of the most influential was this show called Celebrity Apprentice. You may have heard of it. I've heard of it. Which is not— I mean, which was kind of fantasy and not the way it works. So I don't think on the whole television has been— all that helpful. I think though if anything I would— less television and more kind of elite narratives if you will are I think really over-indexed on college and it's not that that's bad advice to go to college because I think I'm very— I think our university system is in many ways just an incredible asset and I think it's a wonderful thing but the idea that the only path to success is through college, I think becomes self-reinforcing.

It seems to me there's good causal evidence that if you finish 4-year college, you will end up healthier for the rest of your life. Do the alternatives to college give you that same benefit?

Well, so whenever we talk about the statistics of college, I think, and even setting aside causality, right? Right, it's causal.

Clearly there's a selection effect, right? But even causal.

Yeah, yeah, yeah. Even setting aside those measurement issues, I think— Tyler Cowen. It's very different to say, "How does it make sense to act in the system as it exists today?" and "How should the system be?" Right? That's different because for an individual person, they can't change the system. So yes, absolutely. I think there's a lot of benefits to graduating from college and if you are in a position to graduate from college from a sort of a quality college, you definitely should try to do that. My advice. Even if you want to be a plumber, like maybe you should do that first because then society as it exists will both allow you to be a plumber and allow you to do other things if you decide you want to do other things later. But that's different than saying that we should organize a system in which something that costs tens of thousands of dollars and many years of your life is a prerequisite to having a decent job and a decent living. I think that's a terrible way to organize a society, which is different than saying Given how we've organized a society, what should an individual choose?

What has gone wrong with American men?

How long have you got? I mean, it seems like it's quite a list.

What's your most basic explanation? Like life expectancy between American men— and you can just do white American men if you want— white American men, white Australian men. It's a huge gap pre-COVID. Social indicators for a significant chunk of American men. You can divide it up different ways, but yeah, they're way down. What happened? What's your simplest theory?

Well, I don't know. It is very striking because I think as far as I know, there's only two places in the world where such a large part of the population is seeing reduced life expectancy. Well, with COVID now.

Putting aside COVID, all this is still happening, right?

Yeah, yeah. But I'm saying pre-COVID, yeah. I mean, Russia is one place, right? Where it's sort of deaths of despair, sort of drinking, and then in much of of the United States and look, I think the math says that if you were to anchor it in life expectancy, the issue is deaths of despair, right? It's suicide. It's—

But why is despair up here, right? It's not the worst country in the world.

It's not and I think it's— but, you know, I think there's a lot of evidence that it's how you— thought your life would be relative to what it is. I mean, I think if you look at the— over time, if this American dream anchor point is sort of a life where your kids live better than you and you live better than your parents, that's something that certainly for white Americans, north of 90% were experiencing if you were born in say 1940 and now it's a coin flip, right? As if you were born in 1980, it was like 50— you're age 30. It looks like about half of Americans were living better than their parents and about half were living worse and presumably when we see the 2020 data, it will have gone underwater. So we're the first generation in which Americans and certainly white Americans are living worse in material terms and in terms of what they— their life possibilities where more than half are living worse than their parents. I think that is a profound shock and I think that on top of it, given so much of it is economics and gender expectations being what they are, that probably does hit men harder.

So you think we should try to lower our expectations for significant portions of American society or raise them?

Well, we could do better. Like, we could— you could lower your expectations, but you could also do better in your outcomes, which it would include. So look, the kinds of issues we're talking about are disproportionately concentrated in white men without bachelor's degrees. And I think if you look at what was possible for white men without bachelor's degrees, if they work hard, were conscientious, they could make a lot of progress in their lives before. And more and more of those pathways where if you didn't have a college degree, but you kind of like you know, you kept your nose clean and you worked hard. I mean, again, for white men, that usually worked out. And I think now it often doesn't. And a big part of it is this, this, uh, just insane levels of pedigree bias where we attribute so much value to having a bachelor's degree as opposed to actually doing the job. And that doesn't just affect white men. I mean, it affects like Many, but I do think that that shift in reality and that relative to expectations has been a significant part of it.

So there's two versions of your hypothesis that I can see. One is that it would be socially better to value the college degree less for reasons you just stated amongst others. The other is that even at the margin, even privately, we're way overvaluing the college degree.

Oh, yeah.

If it's the latter, you should be extremely optimistic, right?

Well, yeah. It is both, but it's definitely the latter. So in other words, because you say privately we're overvaluing the college degree, again, if you unpack that at the level of a hiring transaction, absolutely, in most cases, employers overvalue a college degree. But you got to thin slice it. I'm not saying that doctors shouldn't have medical degrees. Right? Or perhaps even that, you know, you need a civil engineering degree to be building a bridge. But we are requiring bachelor's degrees for jobs in sales, in general management, in, you know, in like IT operations where you can easily see whether someone has the skills or not and you're like, you know, millions of employers for millions of jobs, you're adding these degree requirements. So I'm saying something Yes, at the individual level, I think we're overvaluing degrees and we are undervaluing the skills of people who are skilled through alternative routes of starts. But at a system level, it's incredibly destructive math. So if 70% of new jobs in the 10 years to COVID were ones that employers said, "Oh, you need a bachelor's degree," but only 40% of American workers have a bachelor's degree, well then obviously that math doesn't work. Absolutely self-harm at a system level. So it is a market problem, but it's a collective action problem too.

The collective action problem version of your hypothesis seems much more plausible to me. So as you know, putting aside pandemic times, employers are demanding not only more 4-year degrees, but master's degrees or more for a lot of jobs. So the market has decided in a big way to move in the other direction. So the level of market value—

Sorry, have you interviewed the market? How has the market decided that? I think the institutional factors are really, really important to understand. So it's not because— the US labor market is not a spot market for ditch digging. It's not a market in that sense. It is where years, decades of investment in human capital, if you will, are monetized. It's a market where—— and like healthcare, it's a market where kind of the ultimate use is not necessarily by the person hiring or training. And even more so, it's a market where self-concept plays a big role. So for example, in the HR kind of field, that was a field 30 years ago where if you were good, conscientious, people person, manager, someone might say, "Hey, you should look into a stint in HR." The HR department had people who had worked in operations on the factory floor, in finance, just all over. It was a very diverse group. Now, it's a field that over the last 30 years has become the most credentialized. Now, that's the front end of screening. So, hey, we're hiring for an important job. Well, my job is important and I have to have a bachelor's degree for it, so probably it has to— have a bachelor's degree. There's a tremendous amount of cut and pasting going on and just without thinking about what the role requires. So I think you can't ignore this kind of hysteresis effect, if you will, right? There's a path dependency here that's taken us far off course and it's not just about some sort of kind of like straight free market decisions.

But we have such a diversified economy, right? So many millions of employers. They're willing to do a lot of weird things. You see that in the tech sector. And to think this problem they can overcome because they're just massively failing at the level of millions of individual decisions.

Well, just to be clear, I think it can be overcome. I mean, I'm working full-time to overcome it. So to your point, I'm very optimistic about solving the problem and for 3 main reasons. The biggest reason isn't even the one you're talking about. The biggest reason I'm optimistic about solving it is that you've got 70 million stars, half the workforce, who are busting their tails to learn new things, to actually figure out what it would take to have the skills for the job. I don't know if you saw this. There was a Harvard Business Review— the big survey, Harvard Business School survey and Accenture, I think, that actually looked at attitudes about digital transformation. Senior executives said that one of the biggest barriers to digital transformation is that their frontline workers did not want to learn new skills. But they also surveyed frontline workers who off the charts wanted to learn new skills for digital transformation and saw as the problem that they weren't being given any guidance on what skills to learn. I think you see that that's absolutely true of college. Think about the information problem facing someone working in a frontline job, say making $15 an hour, who wants to do better, wants to earn more, wants to be able to support their family more. Half the people will say, "Oh, you need to go to school." So you need to go 4 years, 5 years, 6 years, and then see what happens. And the other half of the people are like, oh, you should apply for a job. If you say go to school, well, where do I get the money? And by the way, even if I can get scholarships or student loans or whatever, like, who supports my kids? Who pays the rent? Okay, that's one problem. And then it's going to be years before you know. But then if you apply for a job, right, you're not going to know that much more either. You might not get the job. But then the question is, why didn't you get the job? You don't get feedback from the interview. You don't know if you've been screened out. We've worked with people who said, "You can't get me a job at this company. I've applied 4 times. They rejected me." But guess what? They hadn't rejected them. They had screened them out. They had not evaluated their skills and found them wanting. They had screened them out without considering them. And since we partner with this company to say, "Hey, in our system, you just do it based on skills and attributes, you don't screen people out, they actually got the job. And so there is a market failure in the actual workflow of hiring that is massive. And I agree with you, we should be very optimistic about being able to overcome that. And I can tell you lots of employers have been moving in that direction for the last 5, 10 years because they see the old direction is not working for them.

If it's privately profitable to deviate from this practice, Why not run a for-profit rather than a nonprofit?

Well, there are for-profits that are taking pieces of this and I think it's great. I think if I were doing this as a for-profit, I would be sort of skimming, right? I would be taking the kind of the folks who are absolutely overwhelmingly kind of should have the job today. It's just like not even a close call and creating some sort of arbitrage there. But I think there's a much larger opportunity. I think no one entity, no one person, no one organization, not even one sector can solve this problem alone. I really think it's a multi-sector problem. So the question isn't how can a nonprofit solve it because a nonprofit can't solve it. The question is what's missing and fundamentally I believe and Opportunity at Work believes what's missing is an actual set of valid signals that— so you screen in signals as opposed to screen out signals. You cannot interview everybody. But if you're screening out because on average you think people with bachelor's degree are better, so therefore you're screening out millions of people who can do the job, well, that's a terrible system. An individual employer, it might not— they can't necessarily change that whole system because the system works through hiring platforms and lots of other things, and the employers that have the biggest ability to change it don't need to change it as much because they're winning the poaching wars, right? It's the companies that are losing the poaching wars that are actually really sucking wind on And they don't have the power to change that system, but they can change their practices, and we're working to help them.

As you know, more and more top universities are moving away from requiring standardized testing for people applying. Is this good or bad from your point of view?

Well, I think it's really too early to tell because the question is—

But you want alternative markers and not just what kind of family you came from, what kind of prep you had. So if you're just smart, why shouldn't we let you standardized test? Yeah, I think alternative markers are key.

Look, I think it depends— this is actually a pretty complicated issue and I've talked to university administrators and admissions people and it's interesting that the variety of different ways they're trying to work on this. But I will say this. If you think about something like the SAT, when it first started,— like I'm talking about in the 1930s essentially, it was an alternative route into a college because the way like— and it was because it was started with the Ivies, right? It was started with James Conant and Harvard and the Ivies and the Seven Sisters and the rest and then it gradually sort of moved out. But the problem they were trying to solve back in the '30s was that up until that point, the way you got into say Dartmouth is The headmaster of Choate would write to Dartmouth and say, "Here's our 15 candidates for Dartmouth," and Dartmouth would mostly take them because Choate knew what Dartmouth wanted. But then you had the high school movement in the US where between 1909 and 1939, you went from 9% of American teenagers going to high school to 79% going to high school. So now suddenly you had high school students applying to college. They were at Dubuque normal school in Iowa. Like how does Dartmouth know whether this person was— so it wasn't— so the people from Choate didn't start taking the SATs, but the SAT, even though it was a pretty terrible test at the time, it was better than nothing. It was a way that someone who was out there not in the normal feeder schools could distinguish themselves and I think it is a very valuable role to play. As you know, Tyler, the SAT does to some extent still play that role, but it also— because now that everybody has had to use it, it also is something that can be gamed more and test prep and all the rest of it. So—

But it tracks IQ pretty closely and a lot of Asian schools way overemphasize standard testing, I would say, and they've risen to very high levels of quality very quickly.

It just seems like a good thing to do. Tyler, I think there's a much bigger problem. I mean, to me, the bigger problem is this. There is a greater demand for higher education and rightly so. But if you look at the number of quality seats in higher education in this country, they have grown very little in a period where we've gone from almost no one had a college degree. By the way, high schools were to get the jobs of the future, right? High schools were not meant as a— were not developed as a way to get into college. High schools were developed as a way to get into a good job, right? Jobs of the future at that time. So now that in 1983 with The Nation at Risk, we suddenly decided that, "Oh, the point of high school is for everyone to go to college." Like no one had told high schools up until that point. So now that that's the case, you basically— the basic problem is if you keep the number of college seats constant and you demand that everybody try to go to college and in fact if businesses to some extent weaponize college degrees by saying like, "Oh, if you don't have a college degree, you can't get into a decent job," well then suddenly you've got huge demand for college degrees and you've got basically mostly fixed supply, right? So what happens as an economist? Well, what happens is you create this horrible zero-sum bottleneck in the talent development system and in the signaling system. So yeah, you can debate what should be the signals that allow people to go to college, but the fact of the matter is We should have like 50% more, maybe twice as many high-quality seats in college. That's the problem. So affirmative action and testing and issues like this become life or death issues because we put way too much weight on that bottleneck. What we really need is more quality college seats and we need many more valid alternative routes and that should be not just for people who are 18, 19, 20.— people in their 20s, people in their 30s, people in their 40s, people in their 50s because jobs are changing all the time and we need all the talents. It is absolutely not the case that people stop learning. So that's what we need. So I think a lot of this is a second-order distraction because what we actually need is more of the good thing, not just different ways of screening people out.

If you're agnostic on standardized testing, what is it you think we should be measuring more as an admissions input?

What I'm saying about testing is that—

It doesn't have to be a test. It could be any kind of measurement. There must be something we should measure more. What is that something?

I think it's more of a signal pluralism that I'm arguing for. In other words, I don't believe— There's so many talents and there's so many different needs and skills, so the idea that we should all norm on a single way to evaluate talent at every level is kind of counter—

We can measure a bunch of things more, right? You don't simply want more discretion for admissions officers, right? Yeah.

But for example, I think— but I think there should be— for example, explicitly, you should be able to— first of all, the college and vocational should not be nearly as separate as they are. You should think about vocational as applied learning and there should be way more of it in college. Northeastern University, for example, has done a very good job of that and I think college should also be much more seen as skills acquisition along the way. You think of a Western Governors University, which is an example of filling a mark— it's a nonprofit, but it's been growing by I think 10%, 15% a year since it was founded and it's competency-based education where every level of learning can be validated. So in principle— You know, the bachelor's degree is the ribbon and bow at the end of the process.

Genomics doesn't seem to work yet practically, but should we use it when it does? Wouldn't it be a good way to identify people who couldn't go to college but are very talented?

I think there's lots easier ways than genomics, like a lot more direct ways to recognize people.

So again, ex post, you see who's done well, but ex ante, what's better when it works?

Yeah, but think about the data we have right now. Opportunity at Work, when we say— as we got started, because we're trying to create a market change, right? So we're thinking about talent category creation, right? Of people who are skilled through alternative routes, 70 million people with STARS. The fact is you can see what skills people have most easily by what jobs they do today, right? One of the things that's really powerful is that there's a huge overlap in the skill set required for many low-wage jobs and those for many middle-wage jobs, right? So like a second-level call center representative has a lot of the skills required for enterprise sales, for example, but enterprise sales might pay twice as much money. So if you think about formal education or training as more like topping up the skills you learn on the job, You have a much more efficient path to source people by the millions into jobs you want. So that's a much more direct path. There's about 50 gateway jobs that are lower-wage jobs where the skills required for them basically match the skills for middle-wage jobs. So the 70 million stars, 30 million of them have skills for a job that pays 50% more than the jobs they're in today. And you'll say, "Well, why doesn't the market allow them to move into those jobs?" The answer is it partly does. Like that is we have a database of 130 million job transitions and that's exactly how stars and for that matter people with bachelor's degrees tend to move into higher wage jobs. They move into higher wage jobs with an overlapping, highly overlapping skill set to the jobs they're in. So like we see like a much more direct impact than genomics or teaching third graders to code or any number of things you might also think might matter. A much more direct effect is instead of these movements happening at random, why don't we put some streetlights around them? Why don't we put some traffic— and actually treat these accidental pathways up through the labor market as intentional ones? I think that can be the basis of an American system, if you will, that's much more related to data, much more driven by work as opposed to just waiting. I think that would be a far superior— or rather— not as a replacement for higher education, because again, pluralism matters. It's creating a wide variety of ways that people can succeed.

Let's say we had an AI system that predicted labor outcomes pretty well, and maybe it wasn't transparent. We didn't even understand how it worked, right? That's quite possible with algorithms. Should we use that a lot, or is that a problem or an issue?

Well, as you described it, I think that's a terrible idea, and here's why. Because if you have an AI that can predict outcomes, and as you mentioned, we don't understand it, the way an AI is almost certainly predicting those outcomes is not by some inherent characteristics, it's by machine learning. So in other words, it's looking at the actual outcomes and saying, what's the pattern that generates these outcomes? Well, if the actual outcomes today are generated by all sorts of bias, and most obviously for the purposes of this conversation, this degree bias, right? Well, then the AI will say, oh, well, you're not going to probably be able to You won't get that job because all our data says you won't. But why is it? It's not because of some intrinsic reason. It's because we have a system that's biased against you and the same can be true if you look at the levels of occupational segregation in this country by gender certainly but by race also. For you to believe that that's like an actually optimal situation, I mean you'd have to have a pretty low opinion of let's say Black people, right? Because like there's a—

But why think an AI program is going to hurt Black people? I mean, isn't in the sense that we can get an AI program?

No, I'm not saying the AI program is going to hurt Black people. I'm saying, and I'm not saying it's going to, the AI program is going to hurt. I'm saying that the current systems, right? And let's sort of, if we were to keep it on stars versus people with bachelor's degrees, people who have the skills to do the job, but can't get the job. There's millions of such people. They're being screened out not by AI, they're being screened out by a set of human systems, by the way, mostly designed by accident inadvertently with respect to their talents for the most part. And so that is what's doing it. What AI is doing, machine learning AI, is turbocharging the failures of the human system. I'm not blaming the AI to be very clear. I think AI could be very helpful. Extremely helpful. I think to apply machine learning to human learning, for example, I think that could be very powerful. I could see in the future you could have a business model that had an AI patch. You have 85% of the skills for the job. The AI can do the other— help you do the other 15%, can sort of teach you as you go, can validate when you have the skills. It could be metered, and then it goes away and you have all the skills. So AI can be incredibly helpful, but it's got to be trying to solve the right problem. I mean, this is the biggest thing about AI. What is the problem you're trying to solve? And secondarily, what are the constraints you're imposing? Because an AI will then go to town. But if you're trying to solve the wrong problem with an AI, it's a disaster. So I'm not blaming the AI. That's the wrong problem statement. A right problem statement is, how do you have people who have tremendous amount of skills but not all of them? And we're not quite sure. If you set AI to solve that problem, well, I think AI could be very powerful.

But isn't there a tension in what you're saying? So you're trying to tell all these employers, If you do what I'm recommending, it will be good for your private profits. And then you're turning around and telling them, well, any algorithm you use that will predict actual outcomes is bad. That's what you shouldn't do.

That's not what I said. Isn't that a very uphill climb? No, I'm not saying that any outcomes you— I'm saying that if you predict whether some— I'm saying garbage in, garbage out is what I'm saying. I'm saying if your process is biased and then you use AI to help you predict the process.

But the word bias is itself biased. If you have an AI program that does predict who will Do well on the job, employers will want to use it. Do well how? Generate profits for the company.

And you think that the companies in their own internal systems that they could have sufficiently good attribution of who develops profits for the company? I think, look, if you're talking about sales, for example, maybe, but most jobs are collaborative jobs where the work product is team product. I think, I mean, probably Google has the most— I mean, I think not probably, definitely has the most sophisticated people analytics team in the world. One of the things— they've concluded a couple of things. One, they've concluded that when you have all the data, a bachelor's degree has zero additional explanatory power. So that's one thing when you apply the most of it. The second thing is the best teams are not just aggregations of the "best individuals in the role." The best teams have dynamics of their own. That include some that are more team players, some that are more like— So I just think it's not that applying AI has to be problematic, it's that applying AI to a problem that's poorly understood and poorly specified is not going to work.

As you know, the state of Maryland now no longer requires a bachelor's degree for a large number of government jobs. It's early on in this, but how do you think that's going so far? What is going well? What is not going well?

Well, as you may know, Opportunity at Work partnered with the state of Maryland to do this and we actually sort of showed them— showed Governor Hogan and his team the data on— there's 1.3 million stars in Maryland, half of whom have the skills for jobs that pay more and they cut across all races and all regions. It's particularly powerful for veterans, for people in rural areas of Maryland. Disproportionately African Americans, also Latinos. So it's a very power— and the state of Maryland had a lot of jobs it couldn't fill. So what it actually did is a combination of removing degree requirements for some jobs where they clearly didn't belong and in other cases, it had to do with jobs that there weren't really formal degree requirements but there also wasn't outreach to STARS.— a big thing that Maryland and Governor Hogan did and he and I were there announcing this together but he really sort of focused on we see your talent. We see your skills. We want it. Like we— and they— we actually— we want your help solving the problems for the state of Maryland. That's really important. I hear— I get kind of annoyed when I hear people about, "Oh, are we going to give someone a job?" A job is not a gift, right? Work is solving problems and STARS are problem solvers. We have a lot of unsolved problems and when you have people who can help you solve the problems and you're keeping them out, that's ridiculous. So Maryland also posted hundreds of these jobs on Stellar Works, which is Opportunity Works kind of STARS talent marketplace and look, it's getting the pipeline going because they've had lots of applications. We offer some additional ways to kind of screen people in. Again, at the end of the day, it's those who hire in Maryland. There's a very formal process. But one thing that's been really gratifying is that many other states have reached out and many other county governments reached out, but also lots of employers. Many employers that we worked with already have seen the Maryland example as being very validating for their work.

Now, you argued before, and I fully agree, that top schools, indeed non-top schools, should let in many more candidates, right? Given that they don't do this, and this argument is at this point not new, I mean, how much should we actually trust these institutions of higher education? Should we respect them? Should we honor them? Should we think they're well-intentioned but somehow just not understanding the argument? Or should we think they're actually just self-interested and elitist and anti-egalitarian?

Like, what's our proper mental stance here? Well, you know, I think— the argument that institutions are self-interested is not as much of an insult as some people give it credit. I mean—

I'm willing to take the fully cynical view.

No, but I'm saying as economists, like the idea that if you put a set of incentives in front of an institution, over time that they will respond to those incentives, that's like pretty normal. So let's review the incentives for higher education, right? So we create this system. Where businesses decide that the way to decide whether they would even bother to spend the time to figure out whether somebody could get a decent job and lead a decent life as a result because that's how you lead it in America. You work, right? And you sort of— you advance. So if they say, "Well, this thing that was never invented for that, this college degree and ideally a selective college degree, that's what we're going to base it on." So that's the thing. So it sets this incredible magnetic signal. That everybody has got to charge towards that and then we say, "All right. So that sets off the competition in higher ed. So how do we judge whether you're doing well in higher ed? I'll tell you what. How about this? The more people you can exclude, the better you're doing." So we're saying like everybody needs a college degree and then we're saying you get to the top of the rankings in US News and World Report by excluding as many people as you can, right? So that is a terrible incentive structure. So surprise. Colleges are trying to— right? They're trying to like rise up the rankings. They're trying to be better reviewed. So you get what you ask for. That's the thing. What's mystifying to me, Tyler, is how surprised businesses are when they say, "Oh, colleges aren't giving us the skills we want." It's like, well, they're giving you what you ask for, which is people with bachelor's degrees, right? Then we're surprised when we say, "Oh, the best college is the one that keeps the most out." and then we're saying, "Well, why don't colleges let more people in?" Right? So the problem starts with this combination of an elitist attitude that says like, "Oh, how do you know something's good? It's that not many people can do it." No, that's wrong. I think colleges should be judged by how many people they successfully include and how much progress they help to make and not just in terms of their potential earnings, although that's obviously important, but their what they learn. They might learn that they want to do something completely different. So I think colleges should be judged by that and right now, we're judging colleges in the— almost the worst possible way. So it shouldn't be surprising that many colleges feel the pressure to try to be elite, to see how many people they can get to apply, how many people they can reject and not to add seats because then it would be harder to do that. So I think we need to change the incentive structure for universities, but that's a very different thing than saying we should disrespect college education or university education. I think it's obviously incredibly important and valuable for the country. We need more of it for more people.

But to change the incentive structure, maybe we do need to start calling them actually low status according to the correct moral theory because it's pretty locked in. Top schools stay top schools for decades, maybe centuries, perhaps someday millennia.

Yeah. Well, just speaking personally, I do not consider— you know, a school so great on the basis of, you know, its kind of, you know, huge rejection rate. I think that's terrible. The schools I respect the most are the ones that actually do the most to develop the most people and it's OK if any one school does it a certain way because we have a tremendous amount of diversity in higher education and that's good because we have diversity in people and diversity in jobs and diversity in regions. I think the problem is more when we go to this monoculture and we say there's a single metric by which to judge people, by which to judge universities. I think that's a pretty big mistake.

Now you hire many people. You've worked for McKinsey. What's your favorite interview question?

My interview questions tend not to be like a single question that's the same for everybody, which rightly or wrongly, I tend to be really interested. In what people have done and what they've learned. If someone describes kind of a smooth progression without ever making a mistake or ever learning a thing, I'm not that impressed. I tend to drill down when I get a sign that they learned something to really understand what they learned, how they learned it, and at what cost and how they apply it now. So that tends to be— but it tends to be specific to the— to the situation and to the role, of course, because I would be looking for different things for someone who's doing, let's say, sales or business development versus someone who's doing analytics. Like someone who's doing sales and business development who maybe can't sell me on anything in the interview, I'd be more concerned. But someone who's doing kind of careful research who doesn't sell me on anything in the interview, I'd be okay with that.

And how useful do you think are interviews? And I mean interviews done by you, not done by a mediocre HR person.

Tyler Cowen. First of all, there's a lot of wonderful HR people. They're not all mediocre.

Sure, but the average interview may not be that useful, right?

Yeah, right. It might not be, but actually I'm not sure it's the HR person who's got the more mediocre interviews. I mean, often— I mean, on the one hand, I think hiring managers tend to know better what they need, but they may not be as thoughtful about the interview. But setting that aside. I think interviews can be really valuable. I wouldn't hire a senior executive, for example, without an interview. But for me at that level, what I'm looking for a little bit is anomalies. I would put more weight on their track record, but in the interview, I'd want to understand— you know, you call it production function. Like how did they get that track record? And is that something that is like more at the expense of the others around them or is it something that actually, you know, has sort of built up the success of the team too? And so if an interview doesn't jibe with the track record, like I want to dig deeper. I think interviews may be overrated for jobs where there are or a whole series of interviews might be overrated for jobs where there's a huge number of positions to be filled and the role is like relatively sort of standardized and understood. I think we should have a lot more where people can demonstrate that they're ready for a job and even if they don't get that particular job, they get pulled into another similar job without a whole other round of interviews.

Very last question. What was it like having Charles Goodhart as your advisor for your PhD?

Oh. Well, Charles Goodhart was actually my referee for my PhD, not my advisor. I think he— this was on international payment unions and clearinghouses. So it was central banking, right? Well, I would say it was good because number 1, I got my doctorate, and then number 2, he recommended it to Macmillan and St. Martin's for publication. That's why it was published as a book. It is definitely not my idea that the world needed book on that, but Professor Goodhart thought so.

Payment systems have become a big deal since you wrote that book, much bigger than at the time.

Absolutely. So you were ahead of the curve. I have a hedge against the collapse of the global economy. It will be all the more useful.

Byron O'Geist, thank you very much.

Thank you, Tyler Cowen.