Loading summary
A
One of the things about designing a marketplace is you have to make people want to come and transact in your marketplace as opposed to transacting elsewhere outside of it, or as opposed to trying to transact differently within the marketplace than what you're offering them. And stability is about that. Stability says, if the mechanism matches me to you, if I'm going to be employed by you, then stability means you might not be my first choice, and I might not be your first choice. But the people who I like better than you don't like me. They prefer the people they were matched with. And the people you like better than me, they got jobs they prefer to the one you're offering. So you and I are the best we can get.
B
Welcome to the Work for Humans podcast. This is Dart Lindsley. Some markets are easy to understand. You want a gallon of milk, you you pay for it, then you go home. But a lot of markets are trickier. Matching a candidate and a job isn't like buying a gallon of milk. Dating isn't either. In some markets, money matters a lot less than fit. For example, you can't simply buy your way into a medical residency program or a kidney transplant. And behind every market is a set of rules that determines who gets what, when, and under what conditions. My guest today is Alvin Roth, professor of economics at Stanford and recipient of the 2012 Nobel Prize for his work on the design of matching markets. Markets where both sides care deeply about who they end up with and where a good outcome depends on more than money. One of the things I appreciate so much about Al's approach is that he treats markets not as theoretical constructs, but as things that we can build in the real world. And he's built quite a few himself. But the real world's messy. And in this episode, we talk about one of the messiest parts of market design described in his book Moral Economics. These are markets where two people want to transact, but a third wants to get in the way. Turns out this situation arises all the time, from kidney exchanges to in vitro fertilization, from vaccines to to gambling, even things as prosaic as selling horse meat for food. In this conversation, Al and I talk about the basics of matching markets and what makes the good ones stable. We go deep on what he calls repugnant paternalism, social norms, and the strange ways societies decide what kinds of exchanges are acceptable and what kinds are not. Along the way, we bring it back to work, because organizations are full of matching problems, moral assumptions, and invisible rules that shape who gets opportunities and who Doesn't. Alright, if you enjoy the show, please leave a review and subscribe wherever you listen to podcasts. And now I give you my conversation with Al Roth. Al Roth, welcome to Work for Humans. Thank you. So your work is focused on two sided matching markets, and it's especially those where a good match is about more than money. So we're going to talk about all sorts of different kinds today, but examples are medical residency matching New York City, high school matching, kidney exchanges. Your work on kidney exchanges has saved thousands of lives. This show starts from the premise that people at work are customers and that work as a product. So this framing that we have reframes all companies as being multi sided. So they are all matching, they're all about matching the work that comes from a traditional customer with people who want to do the work. And so today we're going to end up spending most of our time talking about your most recent book, which is Moral Economics, a fascinating topic. Before we get there, I want to talk about market design in general and what you've dedicated your career to, market design. And so the first thing I'm going to ask is, how is market design different from mechanism design?
A
So market design takes seriously the idea that markets are artifacts that people build to coordinate and cooperate and compete with each other. Mechanism design is a theoretical take on that that's been important in economics where we say, supposing we could design the whole economic environment, what would we do? But the difference between mechanism design and practical market design is in mechanism design, the theorist is godlike and controls the whole universe. So in the mechanism design literature, if I'm designing a mechanism, the people who are participating in it have no opportunities other than to participate in it. And they can only participate in the ways that I prescribe. That is, in a theoretical paper in Econometrica on mechanism design, the author defines all the strategies that are available to the players and then talks about how those will work together towards some objective. But in practical market design, what we get to design are marketplaces, which are small institutions within large economic environments, within large markets. And the first lesson of practical market design is that people have big strategy sets, including not coming to the marketplace that you're designing, but transacting somewhere else or doing other things than you anticipated. So we can't take the godlike view that we control everything. On the contrary, we have to design marketplaces that tempt people to come and join them to make their transactions there. And so when you talk about market design, marketplace design, I'll use those two terms interchangeably. But really, I'm always talking about marketplace design. You have to think, what do marketplaces do? What functions do they serve in a larger market? And one of the things they have to do to make the market work well is make it thick. So the marketplace has to attract enough people to get transactions that are desirable. And then once you have a thick marketplace with lots of transactions, you have to deal with congestion, with all those possible transactions, how do you evaluate them? And the marketplace has to help you do that. And then you have to make it safe for people to participate. So marketplace design is a part of economic engineering as opposed to economic theory.
B
Right. And in some of your writing, you've made the distinction. It's sort of like the difference between physics and mechanical engineering of a bridge. Physics, there was theoretical bridge that's never existed in the world. The materials are perfectly rigid or something. And this is what makes it real in the world and survivable in the real world. Let's take a couple of examples of the kinds of markets that you've designed and what their characteristics were. Let's pick medical residencies. And I say that one because it's close to work.
A
It's very close to work. Let me take a step back to talk a bit about what makes a market a matching market instead of, say, a commodity market.
B
Would you please?
A
So in a commodity market, you don't care who you're dealing with. You're trying to purchase a commodity. And because it's a commodity, because it's the same no matter who you get it from, you don't care who you're dealing with. And if you're buying it, if you're selling it, you don't care who you're selling it to. So if you're buying shares of IBM on the New York Stock Exchange, you don't worry whether I've taken good care of those shares while I own them, and you don't worry about who I am. You just offer a price. And if we can agree on a price, if you can agree with anyone on the price, you make the trade. But the medical residency market is a labor market where people are applying for jobs. And of course, when you hire someone or when you apply for a job, you care who you're dealing with. You hire particular people. You don't hire generic doctors. You interview them and you try to get to know them. And similarly, when you're looking for a job, you're not just looking for any job, you're looking for particular jobs. And different jobs will be Differently appealing. So you're forming a relationship. And that's what makes a market a matching market. If you care who you're dealing with,
B
you set that against something where it's commodity. And a part of what makes a commodity a commodity is that it is undifferentiated. A coffee bean is a coffee bean or a pork belly is a pork belly. Are matching markets characterized by variation in what people want on one side and variation on what people want on the other side? In other words, it's not a commodity.
A
Let me take a step back again. On commodities, not every coffee bean is the same, not every bushel of wheat is the same, but every bushel of hard red number two winter wheat is the same. That is the Chicago Board of Trade. Their job is to specify the qualities of bushels of wheat closely enough that you don't care which one you get by the time it's a commodity, they're all the same. Now, people are not all the same who you might hire. Just as wheat is not all the same. Before the Chicago Board of Trade, you had to inspect wheat before you bought it, but now you don't because it's commodified. But people are different. And you might want high quality workers of some sort, but because people are different, you still have to interview them and look at their credentials before you hire anyone. But you also might want different things. You might be hiring someone to deal with a geriatric patient load or a pediatric patient load. So of course you're looking for different things than someone who has a different kind of medical practice that they're treating. And similarly for doctors, I might be interested in pediatric patients, or I might be interested in geriatric patients, in which case I'd evaluate different jobs differently. I might also care where the job is and what it's paying and how prestigious it is. I might have lots of things that I care about. And you also might have lots of things that you care about.
B
And one of the real world things that you had to deal with is that some doctors want to be placed together.
A
Some doctors are married to other doctors and are looking for two jobs. I mean, that's a prosaic thing, but it changes the mathematics of matching quite a bit. So there's very neat mathematics of single individuals matching to jobs that want to hire. But it turns out to be more complicated when you have to place couples.
B
I would have given up. I mean, maybe not. Maybe that's the sort of thing that would have obsessed me. And my job is to figure out how to make that work. But it's hard enough to do the first thing and then to add on. Now there's couples in the environment. Maybe we'll get into how that works. But before we do that, one of the things that you have often done, and I don't know if you've done it in every one of your designs, is you essentially batch the processing, which is that, let's say, the residencies. It's the end of the medical year and everybody's applying, and so you get this very large pool of people. So in general, especially if the volume of people looking for something is not that great, you need to hold off long enough in terms of the allocation that you have enough in the pool. Is that a common solution?
A
It's a common thing to think about, but you can't always batch process. Not every market lends itself to that. So the nice thing about residency, I mean nice from the point of view of batch processing, is that most residencies start on July 1st. They're dealing with new graduates of medical school. They all want similar kinds of jobs, that you need to do a residency in order to be licensed to practice medicine in the United States. So when you graduate from medical school, you're looking for a residency, and you want to start on July 1. So batch processing is very plausible there. Now, I've also made some interventions in more decentralized markets, like the market for new PhD economists. And that's not batch processing and kidneys. Kidney exchange is somewhere in between. There are people who are hard to match and people who are easy to match. And you can easily match people who are easy to match, but it might take a long time to match people who are hard to match. So you have to go at it over and over before you get matches for hard to match patient donor pairs.
B
And luckily, there's kidney dialysis, and so there's some patience there while we wait for a match.
A
Right, Patients serving two kinds of words there.
B
Yeah, exactly. Exactly. Okay. And so you need enough in there to make a match, and then you need reference variables. So, for instance, let's take the medical professionals. They rank order what they're interested in, and the medical schools rank order who they're interested in. Is that a necessary part of matching markets?
A
No. Preferences are important, but rank ordering is not. So think about a matching market like Uber, where they're matching passengers to drivers, but they have a pretty simple process because they can guess very closely what your preferences are. Your preferences are a car that will pick you up quickly. There's A couple of grades of cars you can choose, but loosely speaking, they're looking for a nearby car on the passenger side and on the driver side, the drivers also have some preferences, but loosely speaking, they're looking for a good ride. Also nearby, they don't want to drive empty for a long way. So Uber, which depends on the technology of smartphones and global positioning satellites, they already have a lot of the information they need for for most of their matching. Now think about Airbnb, though. Airbnb knows you want an apartment. It can know what city it's in. It could ask you and know the neighborhood, but that doesn't tell them which apartment you want. They have to show you pictures of the living room and the bedroom and things like that because your preferences are hard to describe. So they have to show you the apartment so you can choose among them, and that's how you express your preferences there.
B
I'm very pleased to hear that that's going to play a part later in the discussion, which is that with much of the way work is allocated inside companies. It's not like kidneys. Nope, kidneys. We have seven dimensions of matching or something like that. You raised your eyebrows. There's probably 12, but there's a handful of matches. We know how to run the test on them. They are basically yes, no's. The information's great. But what it means is that there are markets where we know it when
A
we see it, and certainly with kidneys, too, there are plenty of continuous variables like age that people have preferences and their surgeons have preferences. There are things like age and there are things like how long is the artery in the kidney that's been recovered that make you want to. If you're talking about deceased donors that make some surgeons reject some kidneys that other surgeons would accept.
B
I see. Okay. Okay. I'm glad it's dirtying up. In other words, I'm glad that the cases don't have to be simple to work, because the messier they get, the more applicable they are to at least the worlds that I've lived in.
A
In market design, we have to work with them as they come. Some of them are messy and some are less messy, and some of the work is. We've started with easier ones. Because when you're developing market design as a thing, it makes sense to start simple.
B
Yes. Especially because you're developing market design to some extent as a discipline. It's good to both start simple and especially because developing it as a discipline. What is stability in a market?
A
Supposing we're making matches. Remember I said to you that one of the things about designing a marketplace is you have to make people want to come and transact in your marketplace as opposed to transacting elsewhere outside of it, or as opposed to trying to transact differently within the marketplace than what you're offering them. And stability is about that. Stability says if the mechanism. Think about the resonant match. If it matches me to you, if I'm going to be employed by you, then stability means you might not be my first choice and I might not be your first choice. But the people who I like better than you like me. They prefer the people they were matched with, and the people you like better than me, they got jobs they prefer to the one you're offering. So you and I are the best we can get. You're my most preferred job that hasn't filled your positions with people they like better than me.
B
Which is interesting, because it not only needs to be stable, I need to believe it's stable.
A
Well, the reason it needs to be stable is you can find out if it's unstable with just a few phone calls. Supposing I produce a match and it matches you to your third choice job. Now you have to accept that job. They're going to send you a contract and you have to take it if the match is going to be one that works. But if it's your third choice job, you only have to make two phone calls to find out whether one of the two jobs that you prefer might prefer you. And if it's stable, then when you call those two jobs, they say, oh, sorry, we actually filled all our positions. And what they mean is we were offered matches with people who we preferred to you, and that's why we didn't offer you a job.
B
What's the difference between stability and no justified envy?
A
There are different explanations for the same mathematics. So some of the matches, we do have an element of compulsory power to them. If we're doing school choice and you live in a city and the Board of Education in that city assigns your child to some school, you may not be able to call up the schools you like better and switch, because the Board of Education may have the power to tell you where to enroll your child, but justified envy in that case would be the absence of what we call stability when you can act on it. That is, you got assigned to your third choice school, you would have preferred your first choice and they would have preferred you. And when that happens, school principals and parents often find ways to subvert the municipal system so that they can get what they want.
B
The question that I want to explore is whether some markets are simply unstable, they cannot be stabilized. And it's related to the question about who essentially starts the bidding. And so, for instance, there might be a difference in outcome between a worker proposing and firm proposing algorithm. And I think that the difference is just who starts. And I don't really understand that.
A
So we have for the resident match and for school choice, we have helped doctors and municipal school systems implement algorithms that can be described as taking those preference lists that people have submitted and causing one side to make proposals and the other side to make acceptances. Rejections. Those problems are pretty symmetric. So whenever we talk about student proposing, we could also talk about school proposing or doctor proposing and residency program proposing. Those differences turn out not to be so important in large markets.
B
Okay. But there is this question of whether or not some are just unstable. You just can't do it well.
A
So one of the places you can talk about is how people deal with congestion. And in a lot of labor markets, firms deal with congestion by making exploding offers, by making you an offer before other people have made you an offer. So that's flattering. Now, I'm the employer, and I call you up and I say, dart, we like the look of your resume. Your interview went well. We'd like to make you an offer. And you say, great. How long can I think about this? And I say, well, why don't you decide while we're on the phone and you're thinking, I have another interview tomorrow and another interview next week. I'd like to know what my options are before I have to tell Al whether I want to work for him. But my incentives might be, you could get a job at one of those places and you would like it better than my job. So I'm trying to pin you down now. So that causes instabilities depending on how much you feel obliged to actually show up at my job once you've told me that, they'll take it over the phone. Yeah.
B
And so anybody who's ever had a student thinking about colleges and they've gotten that early acceptance that comes in a year before the actual application process or something, that's an example of an exploding offer. And it's something that they're doing because it's a way for them to get an edge on other competing schools and
A
other competing students on both sides. So if you're talking about college admissions, there's a couple of different kinds of admissions that are early. One is called binding early decision. And that's one where the high schools help the colleges enforce it. You can only apply to one place, and if they accept you, you have already promised to attend. That's the binding part. Then there's other things that are called early action, and some of them are called single application early action, that is the student in the high school agree that I'm applying to your college early, and I'll only apply to one place early. But if I get accepted, I can still apply to other places later. And then there's just other kinds of early action. I mean, there's a variety, but that's right. I mean, many competitive colleges, ones that reject a lot of the applicants, fill up to half their class size with people admitted early.
B
I think it's important here to talk about objective, a good matching market. Everybody is in their best position or as close to best as possible. And ideally, nobody is left out by accident.
A
Well, yeah, absolutely. You never like to leave people out by accident. The same goes for insulting people. You should never insult anyone by accident.
B
Accident. Yeah, I've gotten that advice before. I'm still working on it. But there's a difference between matching people to what they want and matching people to what is best for them. And so a lot of this is based upon my best guess of what I really want. An example is I applied to graduate schools. I got the school I wanted, and why did I want that school? I wanted that school because of a particular writer who was a professor there. I got there, he was on sabbatical for my whole graduate degree. So first of all, I lacked the information to make a good choice. But it easily could have been that there was another school that was better for me as a writer that I didn't know about. And so it seems like a shortcoming of this kind of matching.
A
So there's lots of markets where you don't know. Well, they're experience goods. Right. If you apply to a PhD program in economics these days, it takes six years to get a PhD. It's very hard to fully guess how you'll enjoy the third of those six years and what job you'll get at the end. So often we make choices with incomplete information, not just on PhD programs and writing programs, but on spouses. Life is full of incomplete information, but you still, you marry who you think you want to marry, and then you discover over the course of children and grandchildren, whether you were a good match or not.
B
Okay, that makes a lot of sense to me, which is that this is the best we can do. At this moment in time, given the information we have, we're going to hit a stable market. That means everybody gets the best thing that they could get. And then we start again. We're going to learn what comes next.
A
That's right. So marriages. There are also divorces, but lots of marriages last a long time. And of course, there are countries in which you get a lot of advice about who to marry from. Maybe your parents. There could be arranged marriages, as in South Asia. In medical situations, for kidneys, a lot of the decisions are made by surgeons, not by patients. Prescription drugs. You need a medicine for what ails you, but for many drugs, the law requires you to get a doctor's prescription because we think you should get advice on what medicines to get.
B
Which brings us to repugnance and repugnance variables. And this brings us to the topic of your book, Moral Economics From Prostitution to Organ Sales. What controversial transactions? You could have said repugnant, but controversial transactions reveal about how markets work. What is a repugnant transaction?
A
Let me digress for just a moment before answering that, because this book, Moral Economics, is my second book addressed to a general audience. And the first one, which we've already talked about a little bit, was called who Gets what and why. And the difference between those two books is the first book is an optimistic, cheerful book about market design that says some markets and marketplaces are broken, but we can fix them. And I talk about markets that I've had the privilege of helping to fix. And this book is about markets that are broken, but it may be hard to fix because we can't agree on what it means to fix them. We don't necessarily agree on what we want those markets to do. And that brings me to repugnant transactions.
B
The way I've been explaining it to people in preparing for this is that there are some transactions where two people want to engage in them, but some third party would like to have control over that transaction for various reasons.
A
Well, often what the third party wants to do is ban the transaction, not allow them to to do this transaction that they want to do. But there are transactions that I might want to engage in, like start a pig farm next to your house that would harm you. Pig farms don't smell good. I'm not studying that because it's easy to understand why you don't. You probably live in a neighborhood that is zoned for residential housing, not for farming. And having a farm next to your house would decrease your enjoyment of your house if it smelled bad. So when I talk About a repugnant transaction. I mean, a transaction that some people want to engage in and other people don't want them to without any obvious, easily measurable harm to themselves, but don't want them to, often for moral or religious reasons. And one of the ways I talk about that, I specify that there's no obvious harm to the people who are objecting is maybe these can even be transactions that they won't know happened unless someone tells them. So a pig farm. You would understand immediately when you woke up in the morning that there was a new neighbor. But a prototypical example of the kind of repugnant transaction I have in mind is same sex marriage. Two people want to marry each other. Some people forcefully feel that that shouldn't be allowed. But you can't tell whether someone is married unless they tell you by, for instance, wearing a wedding ring. So the objections to same sex marriage aren't the same as objections to having a discotheque next to your house that will keep you awake at night. It's a moral or religious objection. And the question is how to deal with that because there are a lot of those objections around and we ban a lot of markets. But you asked me why repugnant was not in the book title and it was in the working title.
B
I bet it was.
A
But then people would say to me, so what's repugnant? And I would say, well, how about same sex marriage? The way I just described it to you? And they'd say, oh, you don't like same sex marriage? And I'd say, no, no, no, no, that's not what I mean by repugnant. It's not that I don't like it or that you shouldn't like it. I mean, some people don't like it. And that's why I started talking about controversial markets.
B
Oh, I get it. It totally makes sense. Yes. You're very balanced, by the way, throughout the entire book on saying this is just the way things are. Are. It's not an opinion piece. And I will tell you that the list of transactions that I developed while I was reading your book and added to, I think, in some cases is incredibly long prostitution, doctor assisted suicide, legalized weed or drugs of any kind, in vitro fertilization, vaccines, payment for blood. We'll get to that one. Abortion. Sale of horse meat in California is illegal.
A
It's a felony.
B
It's a felony. It's a felony. Okay, it's off my list. Well, you can buy it.
A
It's selling it that's illegal.
B
It's Illegal. Okay, okay, good. Indentured servitude, and then plural marriage. Polyamory, gambling, performance enhancing drugs. This list goes on and on. Everybody has an opinion. And weirdly, some things that should be called, that should be repugnant, like boxing. No, it's okay. We're making some money off of that. It's okay if people hit each other in the face. I find it's repugnant in MMA when somebody's on the ground and they're being hit, but I'm perfectly fine when they're up on the ground. So I assume that the reason that you ended at this book, that you arrived at this book, is that when you're not theorizing about markets, you're actually designing markets. This is what you face.
A
Absolutely. So one of the first things you learn as a market designer is there are ideas that sometimes seem like a good idea to economists that don't seem like a good idea to everyone. And incidentally, that's so pervasive that it's too important to be left to the philosophers. Economists and the general public needs to understand how we decide which markets we support and which we try to ban. And part of trying to ban markets is that not just markets need social support, but bans need social support. So bans that don't have social support give rise to black markets. And the things we were trying to ban become governed by criminals.
B
Right, right, right. So let's see if there's sort of a categorization of these. And so maybe you've already done it, because in the book, the way that they're structured is big categories, like sex is a category, but there's a way in which there's a different dimension along which they might be categorized, which is these are things that we would like to ban because we think they are bad for your health.
A
There are some things, but that's why we want to ban them.
B
Yes. And a little bit would be okay. And that's why it's hard, which is a little bit of a drug as a medicine, but a lot of a drug is an abuse.
A
Right. So we try to ban heroin because we think it's bad for you, but that's not why people tried to ban same sex marriage.
B
No. So there's a second reason for bans, which is it's not bad for you right now, but I think it's bad for your immortal soul. And potentially that's the reason.
A
Right.
B
Which is that there's something beyond the current case that might be bad.
A
We can think some things Might not be so bad by themselves, but they might put us on a slippery slope to a less sympathetic society than we would like to live in.
B
What's an example of that?
A
Well, it's okay for you to donate a kidney to someone you love and you can save their life, but you can't sell a kidney. That's against the law. You can't be paid for it.
B
Okay, so this is an example of. And this is part of the reason why some matching markets have money in them, and some matching markets don't. And sometimes the reason they don't is because if they did, it would be repugnant and that it would lead to poor people selling parts of their bodies to rich people. And we find that repugnant right now.
A
Of course, there's a terrible shortage of kidneys for transplants. So as a market designer, you can start to think about ways to increase the supply by being more generous to donors without opening it up to having rich people buy kidneys from poor people. You could, for instance, think about a market in which only the federal government could pay a donor for a kidney, and in which we allocated the kidneys the way we do the Seastoner kidneys now, based on considerations of health and need, so that it would never be the case that rich people had more access to kidneys than poor people.
B
Does the addition of money to a market generally make it more efficient?
A
Money is a great market design invention that makes lots of transactions possible that really wouldn't otherwise be. So I moved from Boston to the bay area in 2012, and in order to move, I sold a house in Boston, and I came to the Bay Area, and with the money that my house had sold for in Boston, plus some more money, I was able to buy a house in the Bay Area. If you couldn't use money, if we had to trade houses, I would have had to find someone in the Bay Area who wanted my Boston house and whose house I wanted. It would have been impossible to move.
B
And so the very fact that using money for a transaction might be the repugnant component, like for kidneys. It's very interesting because there's a theoretical harm that is there that is sort of like, well, that would be icky if poor people were selling their organs. But there's an actual harm of people not having kidneys when they need them.
A
Absolutely. And there would be an actual harm of people not moving to jobs across the country if they couldn't transact in houses. But let me give you an example. I mean, a familiar example where Adding money to the transaction makes an otherwise not illegal transaction illegal prostitution. What makes sex prostitution is cash. Right. We no longer have laws against adultery, let alone other kinds of extramarital sexual, but we do have laws against prostitution. And what makes a transaction a prostitution transaction is money.
B
And yet what makes a boxing transaction not assault is money.
A
Yep, that's right.
B
Weird.
A
Boxers agree to assault each other. Absolutely. So the world is complex. You can make exceptions in all sorts of directions. Incidentally, it is legal here in California to pay for sex acts in the movie industry. The courts have decided that when you pay actors to have sex on camera, you are paying for acting, not for sex. Paying for sex is illegal, but paying for acting is, of course, a great California industry.
B
And so repugnance can make markets less efficient in various ways, but also create lots of externalities that are repercussions of making something illegal. You mentioned black markets. There was a story of Rhode island that accidentally made prostitution indoors legal.
A
Yes.
B
And what were the effects?
A
So that's a complicated story. It's an interesting one. So there were two economists who studied it, Scott Cunningham and Manisha Shah, and they found that sexually transmitted diseases went down and assaults, sexual assaults, went down during the six years that there was legal indoor prostitution in Rhode Island. But you'll be happy to know that the Rhode island legislature stopped that loophole. And it's no longer legal.
B
Right. It was almost a clerical error that made it legal. It was nice as experiments go, I mean, because it's not like there was a whole cultural movement around it that correlated with the reduction of assault.
A
That's what makes it a natural experiment. Yeah, but in the spirit of trying to treat these things fairly, let's take a moment to talk about paternalism, which is a little bit like repugnance. If you have children and they want to stay up late on a school night, there's something they want to do, and you as their parent, don't want them to do it because they'd be sleepy in the morning or whatever. And as a parent, if you don't exert some compulsory force, you might be neglecting your children. That is, they should go to bed on time so that they can go to school the next day. So for children, we don't mind paternalism. We might even expect it. Now, there's a whole field of behavioral economics that suggests that those of us adults, that is people who've had enough birthdays so that we're not legal children anymore, might not always be in full control of what's in our best interest, not just for lack of information, but if you like to finish whatever's on your plate, you might like to have smaller portions served to you than otherwise because you prefer to eat less than you will eat if it's available, things like that. So there was an effort to ban giant servings of soda, for instance. That failed in New York because people don't like paternalism directed at them. But you could see an argument, and that's part of the argument of making heroin illegal. We think that heroin is bad for you and it's addictive. So if you start, you might not be able to stop. And we would like to prevent you from starting. But of course that's a market where we'd like. There'd be no heroin, but we haven't succeeded. There's lots of heroin.
B
What's interesting is that the transaction enables the taking of heroin. But if there was no transaction and still the taking of heroin, we wouldn't like it. In other words, there would still be a degree of repugnance. And what's interesting at looking at this through the lens of transaction is, well, let's take a illegal immigrants. That's a transaction. People come into the country and there's people who hire them, and we punish the people who have come in illegally and we don't punish the people who are hiring them.
A
Although we might sometimes. I mean, we might very occasionally.
B
But it's an example of no, we don't want that transaction in part because we've regulated the labor market in some cases, and we don't want escapers from our other regulations. And I noticed this thing about repugnance and paternalism and to some extent it's who's regulating whose transaction. So the dynamics of old people legislating or regulating the young, rich people regulating the poor, or religious people optimizing for the afterlife and non believers optimized. So in general, it's often us legislating them.
A
That's right. Republicans, people want to do it and other people don't want them to do it. And sometimes we make it illegal to do it. And sometimes that works and sometimes it doesn't work. One of the things I ask in the book is why is it so easy to buy drugs but so hard to hire a hitman? We have similar laws against selling heroin and murder for hire. But murder for hire is a tiny thing in the United States. It doesn't make it into the national crime statistics, whereas drug overdoses and selling heroin is the big thing. And our prisons are Full of people with drug convictions. So similar laws, you know, if we catch you, we put you in prison for a long time. Similar laws have very different effect, partly because we're more tolerant of drug dealing than we are of murder.
B
I'm sort of coming to a summary conclusion that democracies might make different decisions
A
than economists, for sure, on many issues. Like carbon taxes.
B
Like on carbon taxes. Right. That there's a whole range of things, especially those things that we find repugnant. And it's hard to assess the second order effects. So I want to start talking about work itself. And one of the things I noticed reading about repugnant transactions is that there's something that I think it's named in the book, which is an anti repugnant transaction. It's a transaction that two people maybe don't want to do and a third party really thinks they should. And so it's not an inhibitor. And vaccines are an example of that, which is there's parties on both sides, but some of them feel that it should be done.
A
And vaccines have big positive externalities. It's to my interest that you not give measles to my kids.
B
Right? That's exactly right. And work falls into both cases. So one case is sex work, repugnant. We've got that. One person we had on the show was Elizabeth Anderson from the University of Michigan, and she talks about the history of the work ethic. We had a long discussion about Ireland and the descendants of Bentham were the ones who made policy decisions about Ireland. And what they saw was they saw the Irish poor who could live on a quarter acre of potatoes plus a cow. And they were actually the most well nourished Europeans of their time with a quarter acre and a cow that didn't look like work, that looked like leisure. And so leisure in general was associated with vice. And so first of all, the poor don't want to work. They want leisure. If you give them leisure, they'll reproduce. And there was all these things about it. And so that was a part of the argument for workhouses, which is we're going to force you to sell your property because you're starving to death. And we're going to put you into workhouses where you will be forced to consume this virtuous good work. And so is that something that you see, which is, let's call it virtuous transactions or anti repugnant transactions?
A
So there's something I call protected transactions, which is close to what you're describing. So marriage is a protected transaction when there was a lot of political opposition to same sex marriage. The laws that were passed saying that the federal government wouldn't recognize same sex marriages was called the Defense of Marriage Act. It was defending marriage against same sex marriage because marriage is worth defending. And of course, a lot of things having to do with sexual, we think you shouldn't engage in unless you're married to each other, things like that. I mean, there's a long tradition of repugnancies around sexual transactions that intersect with being married or not being married. And I think part of the reason for that, and I write about this part of the reason those things have changed so much in contemporary society is birth control technology and fertility, like ivf, assisted reproductive technology. So there was a time when sex between a man and a woman often led to pregnancy and pregnancy often led to birth. And a lot of social norms were designed around making sure that babies were born into families so that they'd be protected. And incidentally, for a long time it was not legal to sell contraception in many places. But a lot of these things in the US go to the Supreme Court. So there was Griswold versus Connecticut. Was the contraception case in the 1960s, not ancient history made it legal everywhere to sell contraceptives in pharmacies. So it's legal now to have contraceptives, it's legal to have in vitro fertilization. And so all of a sudden it's possible to have sex without pregnancy, without a high risk of pregnancy. And it's possible to have a baby without having sex. So unmarried people can have sex without presenting a great hazard to society about babies without families. And same sex couples can have marriages and have babies. You have to add surrogacy into the picture to give a same sex male couple babies. So all of a sudden, you know, our ideas about who can have sex with whom and who can form a family with whom have changed pretty radically in the last century.
B
And there's a sociological hangover or muscle memory in some cases about situations that were a problem, you know, a hundred years ago. So we named them repugnant. And there's a quote that's right.
A
So Clarence Thomas, in his concurring opinion for overturning Roe v. Wade, which said that women have the right to choose whether to have an abortion or not whether to end a pregnancy, in his concurring opinion that that wasn't a right of Americans, but rather could be legislated by each state, he went on to say, now that we've overturned that mistake, we should also overturn the mistakes the Court has made about contraception, about consensual sex between adults and about same sex marriage.
B
Yeah.
A
So Justice Thomas would like to revisit re litigate a bunch of controversial, repugnant transactions that were resolved in a permissive direction, in a liberal direction. He would like to make them less permissive again. Which just goes to show that these hard fought battles are never fully over.
B
No, they're not. Well, there's more to talk about in terms of work and repugnance. An example is welfare and how a transaction that was the government giving people aid when they're out of work, the desire to force people into work, the belief that taking away financial aid would force people into work, which is a natural good. So there was something definitely there. But I want to talk about. A lot of your work has been the market of job matching, essentially, like with the medical residencies. A lot of the work that I do is about how work is allocated inside companies. So, you know, you have a team, the team receives projects and how those projects get allocated. It's messy on a lot of levels. One of the levels is it's not very thick. You don't get a lot of projects at a time and people don't come free all the time. So sometimes it's completely linear. It's first come, first in, first out. But sometimes you have two people who are free at the same time. Two projects come in and then you can allocate them to huge benefit, by the way, huge. Two people have completely different interests. I mean, one of the reasons I asked about the variation that needs to be on one side of a matching market is that our research shows that there's nearly unlimited variation in terms of what people want from a matching like that. And if you can just be aware of it, you can make that match much, much better. And sometimes there are opportunities for people to trade tasks, which is, you know, we're both working on these two projects, we're just going to swap tasks. But compared to the beautiful like math problem of kidney exchanges, this is a bucket of noodles. And so are there principles that can be brought into that kind of mess? Hey everybody. On June 16th, I'll be speaking at one of my very favorite venues. It's the Future Talent Summit in Stockholm, Sweden. To get Tickets, go to futuretalentsummit.org that's all one word and enter my speaker promo code elevenfold, which is eleven fold to get a big discount. If you're in the area, I would love to meet you there. That's futuretalentsummit.org promo code elevenfold.
A
There are, and some of them resemble the ideas of stability, say, that keep people in a marketplace. That is, if you persist in assigning people jobs they don't like, they often can quit and go to another company. So one question has to do with keeping your workers happy. But of course, firms also have opinions about who should be on which teams, because the teams are important. And your manager might not want you to switch jobs with someone else who he didn't choose for his team, but he chose you. So there are lots of places to think about that. And of course, one of the benefits of Silicon Valley for industries and for people is that it's thick with firms. So if you're a software engineer and you are unhappy with the tasks you're being given at your current company, you very likely can find a different company. So that's true, though, even without that flexibility. I was on a National Academy of Science panel that studied the labor market of the Air Force. So the Air Force has a lot of compulsory power, particularly for people who have just signed up and have a military obligation, right? So if you've just graduated from the Air Force Academy, they can tell you what your next job is, and you have basically no choice but to take it. But at some point in your career, you have exhausted your remaining military obligation, and they ask you to make a new commitment. But that comes with a new job. And typically the idea is they say, here's this attractive job. Would you agree to serve four more years in the Air Force, and we'll give you this attractive job next year? But they have a matching problem, because if you're, for instance, a pilot of cargo planes, you're very employable outside of the Air Force flying big planes. But on the other hand, as a firm, the Air Force has some of its own opinions about where people should go. And in particular, they think they should deploy to foreign places like Iraq when it's their turn. So it turns out a lot of people who no longer have military obligation resign from the service rather than go to Iraq if they don't feel like going to Iraq. And it depends what you like to do. If you like to fly attack helicopters at night, close to the ground, very fast, there aren't so many places you can do that. But if what you do is fly a cargo plane, there are lots of places you can do that. And you could work for FedEx, for instance. So the Air Force has to think about, is it a matter of principle that when it's your turn to deploy. You deploy. Or could we preserve this highly trained pilot service in the Air Force for a few more years if we were more flexible when they didn't want to deploy? And that's a tough problem for the Air Force.
B
It's interesting. It opens up the door a little bit to coercive labor and, and so coercive labor in the prisons. But also this idea of you've signed a contract with the military and beyond that contract, what you work on is not up to you. I had never really thought about that before. It's kind of the opposite of a market.
A
It's not up to you for specified periods of time. But if while they've got you, while you have a military obligation remaining, if they give you jobs you don't like, then as soon as your military obligation is over, you'll get a different job. So if they're going to train you to be a pilot, they ask you to commit to increase your obligation because of their training costs and efforts, but they don't get them all back in your remaining time. If you're an experienced pilot, they would like you to stay on. Incidentally, there are a couple of ways to lose pilots. One of them is to say, now it's your turn to deploy to Iraq. Another is to say, congratulations, you've got promoted. You're going to now deploy to a desk and not fly every day. And it turns out there are some pilots who would like to be running the Air Force, but there are other pilots who like to be sitting in an airplane flying it around. And some of them leave when they're told, you're now too senior to fly. You have to sit at your desk and make decisions.
B
Well, I think a lot of what I see as problems in work in general are the fact that it's often designed as market mechanics. When it should be engineered, it should be an engineered thing. And that's an example where that might happen because you forget that it's a market, which is, I'm going to treat you badly because you know what? You're in the military now and toughen up and you need to do this and, oh, this is a market, I forgot this is a market.
A
But on the other hand, they have this feeling that it's your turn to deploy. And we make people leave if they don't deploy because we have this idea of fairness. There are other people who are going to have to deploy, but now it's your turn. So that might not be a good policy. They do lose pilots. They lose Pilots, they lose other specialty services that are highly desirable in the general labor market. Another funny thing about the Air Force and armed services as labor markets is they hire almost everyone when they're 18 years old. They have very little lateral hiring. There are reserves, there's National Guard, there's some flexibility. But except for lawyers and doctors, there are very few people who enter the armed services in their profession. So of course, that's a problem for now. Cyber warriors, you'd like to be hiring Google engineers and bring them in as colonels the way we bring in transplant surgeons, but they can't easily do that.
B
On the list of repugnant transactions, I have a relative who has very poor impulse control and she is absolutely preyed upon by payday loans, cigarette vendors, alcohol vendors, bars, gambling opportunities. And it really looks like a pack of wolves. And so I have to say I find them incredibly repugnant.
A
They are. And of course, we had prohibition in the United States. We banned the sale of most alcohol by constitutional Amendment in the 1920s, and then we repealed it with another constitutional amendment in the early 30s. And it isn't that we solved the problems of alcohol. It was that prohibition failed to stop alcohol and instead gave rise to organized crime. So now we have legal markets for alcohol, but there's still alcoholism, there's still drunk driving, there's Alcoholics Anonymous, you know, there's addiction. The problems didn't go away. It's just that prohibition wasn't the solution that we hoped it would be. And I think that's sort of the story we are at with heroin. In other words, we need to experiment, we need to find other ways to deal with drugs because we're not winning the war on drugs, we're losing the war on drugs. But whenever we try to surrender, it doesn't accept our surrender. So we can't just legalize them because, you know, fentanyl is so deadly and it makes cities unlivable if there are open air drug markets everywhere. Many of the markets in my book I offer suggestions, but that's one where I think we still have to gather evidence and experiment to find better ways to deal with.
B
Well, and my parents, we lived in Tennessee when it was dry. And one of the things they said was that the people who lobbied for the continuation of Prohibition were the bootleggers and the Baptists.
A
The Baptists and the bootleggers.
B
The Baptists and the bootleggers. And it's very interesting because my friends who were in the weed business in California are all out of the business. Why? Because Corporations came in and just knocked them out. That's right.
A
Your friends who are in the weed business, if they were growers, they used to talk about how many plants they had. But farmers talk about what's the yield per acre.
B
Right. They've been completely out competed, and in part because they were not in advantageous locations.
A
No, they're in national forests sometimes.
B
Yeah, yeah, exactly. So in general, what are some of the recommendations you make for navigating these very difficult waters?
A
Well, so I think that when we create a black market run by criminals, that's seldom the right solution that we wanted when we tried to ban a market. So when that's the case, I'm largely in favor of legal regulated markets where we can exert more control. Not that that's easy. Not that that's easy. And again, alcohol is a good example. Tobacco is a good example. Right. Children aren't supposed to be able to buy tobacco, but there are more tobacco deaths than there are opioid overdose deaths. Tobacco is a very dangerous drug, although we're way down on the number of smokers. So there are big public health benefits to limiting tobacco, but banning it might very well just produce black markets of the heroin sort because it's addictive. So I think case by case, you have to think about what to do. You know, an interesting case is prostitution, because the workers in prostitution are of two kinds. There are voluntary sex workers who, maybe because they're poor, find that the best job they can have. But then there are trafficked women and children, people who are compelled and who aren't getting the benefit of their work. And what we'd like is to eliminate trafficking, but not necessarily to criminalize voluntary sex workers. Now, when I say that's what we'd like, that's not what everyone would like. I mean, what we do is we criminalize all sex work, both of trafficked workers and of voluntary workers. But one of the proposals I support in my book is a proposal put forward by my colleague Petra Pearson here at Stanford that she calls a hybrid Dutch Swedish market. Because both in the Netherlands and in Sweden, they've done experimentation with different ways to think about prostitution. And one of the things they do in the Netherlands is they license prostitutes. So you have to have a license, and that means you have to at some point have interacted with a social worker and things like that so they can start to think about who's trafficked and who's not trafficked. And one of the things they've done in the Netherlands that neither of these have Worked great. One of the things they've done in the Netherlands is they've decriminalized selling sex, but criminalized buying it. But that turns out not to work well. And the reason is sex, particularly street walking, is one of the most dangerous professions in the world. And therefore sex workers have to make lightning decisions, talking to customers and deciding whether they're crazy or not. And if it's a crime to solicit a prostitute, the customers no longer want to engage in a lot of banter. So if you're a streetwalker, you have to make an instant decision, do I get into this car or not? And that increases the danger. You can't talk.
B
Well, you've also selected for a population that's willing to break the law.
A
Absolutely. Absolutely. So what Petra's or Professor Pearson's proposal is, is why don't we license legal sex workers and make it a crime to engage with unlicensed sex workers?
B
Ah.
A
And that way we try to cut down on the trafficking which we want. And we aren't helping the legal sex workers by making them criminals. We could help them in ways that made their job less dangerous and safer and had health regulations. I think John Oliver may have had a piece in which he said, there's no Hallmark card in which you thank the officer who arrested you. So, you know, it's not doing a favor to voluntary sex workers to criminalize them. Right. It makes them criminals. It makes them vulnerable to crime themselves because they can't easily go to the police for help because they're criminals themselves.
B
We're going to hop to something that is less interesting, probably. But there's two methodological questions. One is that you advocate for laboratory studies.
A
Well, not laboratory. The States can be the laboratory of democracy.
B
Okay, Okay. I thought I read something where you're saying, when we're designing markets, we start off with a laboratory study. And I was like, what the heck does that look like? But then what you say is you roll it out in the world and you tinker with it. And it's because it's not until it comes into contact with the world that you can know a how it's going to work itself, but also what the second order effects are and what the
A
big strategy sets are, what people will do that you didn't anticipate.
B
And one of the challenges of legislation in general is that it is not very good at tinkering. It's big hammers. It's like it makes laws and it's hard to revise them. But what I still think is interesting about the tinkering is that in the design world, when it comes to product design, because we can't know how people are going to react to it, which is that's the metaphor we use for work inside companies is it's a product, it's not a metaphor. We actually believe that it's a real thing. Is that in the product design world it's 90% tinkering. Right. Which is that incrementalism is how you get to the right solution.
A
Right.
B
And that incrementalism is a couple of different things. It's discovery about what the customer wants. The customer is discovering what they want.
A
So you're right. It's hard to experiment with legislation. That's why I like, for instance in drugs, I like the fact that cities experiment and states experiment and countries experiment.
B
Got it. Yeah. That makes a lot of sense. You talked about some of the emerging disputes and one of them had to do with polyamory and polygamy. And that's something that you said it was, has been legalized in parts of Utah.
A
So they've decriminalized polygamy in Utah partly because of the same thing we talked about. You're not helping people by making them criminals. What they worried about is that people were being abused in polygamous marriages, women. But that they hesitated to go to the police because to be in a polygamous marriage was to be a criminal and they might take away your children. And so now they've made it more like a traffic offense. So it's not legal, but it's not a crime. And if you're being abused, you can go and get protection for you and your children.
B
West Hollywood just passed a law and that law essentially extends non discrimination protections to polyamorous. I want to say couples, but yeah, they're not couples.
A
They're good.
B
Yeah, exactly. And so we're starting to see that show up as extensions of existing law. So there's a few questions I ask at the end of every show and I haven't prepared you for them at all. I rarely do, actually. And so my question for you, Al, is what do you hire your work to do for you?
A
Well, I'm a professor of economics, so I teach economics and I get to think about economies and I get paid for it. So there's a part of the discussion of repugnance that says when you pay people, you're coercing them. But I don't feel coerced to be an economist. I feel privileged and I can earn my living by studying controversial markets, for instance, or by designing markets and thinking about market design. So I find it fascinating and fulfilling.
B
Yeah, puzzle solving. Big interesting problems. Often with friends, I would imagine.
A
Almost always with friends. Sometimes with people who aren't friends but who become friends. Market design is a very outward facing part of economics. So I frequently interact with people who aren't economists but who have some marketplace that they're interested in that's having difficulties.
B
Very interesting. Yeah, I'm sure for some people that would be a cost. That outward facing piece would be a cost.
A
Well, but Stanford doesn't make me do that. Stanford gives me the freedom to do that.
B
What does your work cost you?
A
I've been called an organ trafficker for my work in kidney exchange. Trying to make kidney transplants more available to people who might not be well insured, for instance. So that's a cost if you let it bother you. I've developed a thick skin and I've recently been at international transplant conferences where we're trying to make transplants more available to more people, including poorer people.
B
Right. I've been wanting to ask this question for a while. Is there any cost or downside to winning the Nobel Prize?
A
You get a lot more invitations than you used to, a lot more email. You get photographed a lot more. But on the other hand, my jokes are now funnier.
B
That was funny. Yeah. You know, it's a weird thing. I would think it would change reality around you. I was just listening to somebody. Oh, was he from Stanford? Where was he from? He was talking about signaling and how there are involuntary signals. My involuntary signal is I'm a tall man and so I walk around in the world and I can be threatening to some people. And so I might wear tie dye because nobody in tie dye is threatening. And so that's the voluntary signal and the involuntary signal. But it's weird. It does create a signal in the world that is out of your control.
A
Well, you can't affect your height very easily, so I can't tell because of the positioning of our screens. But I'm also a little taller than average. But you and I both are gray bearded. Now, you could change that if you wanted to. You could dye your beard black or blue or pink, which would change the signal in each case.
B
Well, thank you very much for joining me on the show today. It was fantastic.
A
Thanks for having me.
B
Really interesting and it gives me hope for these messy internal markets. So thank you. Thanks for joining me for another episode of Work for Humans. If you enjoyed this episode, please give us a five star rating. Wherever you listen to podcasts and share the show with one person you think would get value from it. Believe it or not, this really helps us grow the show and reach more people who want to build the kind of work that people really want. As always, thank you to my producer Jason names at 9th Path Audio for his insights into content and his high standard for quality. Final note, the opinions shared here are my own and not the views of Google or Cisco Systems. Thanks again for listening. See you next time.
This episode explores the intersection of markets, morality, and the design of systems that shape how society matches people to opportunities. Dart Lindsley interviews Alvin Roth, renowned for his work in matching markets (e.g., medical residencies, kidney exchanges), focusing on concepts from Roth's latest book Moral Economics. The conversation dives into "repugnant" or controversial transactions—those exchanges that societies permit, restrict, or outlaw based on complex social values—and how these decisions reflect deep assumptions about fairness, paternalism, and the social good.
The episode balances technical rigor and accessibility, using vivid real-world analogies and a conversational style. Roth maintains a measured, neutral perspective on controversial issues, emphasizing the descriptive and analytical, not opinion or advocacy.
This conversation reveals that markets are deeply entangled with human values and social norms. "Moral Economics" is about understanding—not just critiquing—which markets we construct, allow, or ban, and why. Roth's pragmatic and experimental approach to market design offers crucial lessons for both policymakers and organizations navigating complex, pluralistic environments where 'what is right' is contested and evolving.