 I'm Martin Wolff from the Financial Times, and it really is my immense pleasure to have this discussion with Al Rath, who won the Nobel Prize in 2012 for the work that we're going to be discussing. In particular, to discuss his recent book, Who Gets What and Why, the Hidden World of Matchmaking and Market Design, I have been reading the book and loved it, and I have the great advantage of being probably or possibly in the same position as many in the audience is that I know really nothing about this, so this is not an area of economics I've focused on at all. It's far too clever for me and vastly too clever for me, so I'm going to learn as much as I hope as you. So let me start off, if I may, in our discussion with a quote from your book, you will recognise it and it gives us a good place to start. Markets are human artefacts, not natural phenomena. Market design gives us a chance to maintain and improve some of humanity's most ancient essential inventions. So what is market design? It sounds very hubristic, so what does it mean? Market design is part of the engineering, part of economics. It grows out of game theory and it's concerned with the rules of market places, which are where markets happen. And sometimes when the rules work well, the markets work beautifully and sometimes they work less well and market design helps us fix markets when they're broken, or sometimes to create new markets where none exist. Well, give us an example of the sorts of conditions in which such conscious design is necessary. And you have in particular, let's run through the whole book, this notion of matching markets are very important. Labour markets are incredibly important. The marital market, the housing market, the market for interns and husband, give some examples of these and what special problems they create. Well, you mentioned matching markets. Let me start with commodity markets, which are the kind of markets most people think about in a commodity market, the price. Explain, because just for these points, you don't mean commodities in the sense of oil, coal, you mean something broader than that, don't you? I do, but oil and coal are not bad examples. One of the things I say in the book is that God made wheat, but the Chicago Board of Trade made number two hard red winter wheat. And number two hard red winter wheat is a commodity that you can trade by price. Another example you give is Ethiopian coffee, recent example of grading. That's a very nice example. And once you have a commodity market, you can deal anonymously with your counterparties. You don't need to know who they are, you don't need to inspect the product that's already been sufficiently defined. And so the job of a commodity market is just price discovery to find the prices at which supply will equal demand from moment to moment for all the commodities on the market. The crucial thing about a commodity market in this sense is that it's something standardized. Everybody knows what it is. It's the same for everybody. It's a commodity, and therefore you can have wide, anonymous markets in it. And you don't care who you're dealing with, so you can have what you want if you can afford it. But in many markets, you can't just choose what you want, even if you can afford it. And those are matching markets. Markets in which you can't just choose what you want, you also have to be chosen. And so labour markets are like that. You can't just choose to work for the financial times. You have to be hired. And neither can the financial times just decide that you should work for them. They have to compete with the Daily Mail. God knows who. Almost nobody as far as I know. So in labour markets, prices don't clear the market. You need other institutions. You need applications and interviews and maybe writing samples. So many markets are like that. Stanford University doesn't choose its entering class by raising the price of tuition until just enough students want to come. It's expensive to go to Stanford, but lots of people would like to come at that price. So there are all these other market institutions of application and admissions. And in some matching markets, not only don't prices do all the work, in some markets we don't allow them to do any of the work at all. So I've been involved in designing public school choice systems for some American cities. And in kidney exchange in most of the world, with the exception of the Islamic Republic of Iran, we don't allow monetary markets in kidneys. But there aren't enough organ donations. And so in the United States and spreading around the world, we now have something called kidney exchange in which if you wanted to give someone you loved a kidney, and you were healthy enough to give a kidney, but your kidney didn't match the person you loved, I might be in the same situation. And maybe I could give a kidney to your patient and you could give a kidney to my patient. And that's an exchange. That's the kind of thing economists are supposed to know about. So, before we get to some of the examples, just give us some examples of what the problems are, sort of generic problems of matching markets. You talk about thickness of the market, obviously the number of people who are transacting in it. You have this lovely phrase, exploding offers. So give us some example of what are the design, what are the problems, the special problems that these sorts of market circumstances create in various cases, some of which you've already indicated. Well, the first job of a marketplace, as you say, is to make the market thick, to bring enough people together so that they can consider lots of possible transactions and find the good ones. And not everyone likes a thick market. Everyone likes a thick market on the other side of the market, but not everyone likes to be in the market at the same time as their competitors. So, in many labour markets, we see exploding offers, which are early short duration offers. So, if you're trying to hire, so private equity makes exploding offers to young investment bankers, for instance, they give them an offer and say, don't shop this around, you have to let us know today whether they'll accept it. So, you've got a young and talented person, or they hope is a young and talented person, and they say, we've got this offer and you've got essentially 30 minutes to decide, and that, of course, removes the chance of going to all the other people and finding they can make a better offer. Exactly. Exercising a bit of monopoly power, effectively. That's right. A little bit of temporal monopoly power. You have to decide before knowing what other offers might come in if you were to wait. And many labour markets suffer from exploding offers. And in some of them, we've been able to organize centralized clearing houses that remove that problem. So, many American doctors, for instance, almost all American doctors get their first jobs through a labour clearing house called the National Resident Matching Program, which solved this problem of exploding offers. It made a thick market. It also solved another problem that marketplaces have to deal with, which is congestion. Once you have a thick market, everyone on the marketplace at the same time, there might be too many transactions to consider in the time available. And that's a serious problem. Think about Airbnb, which is a marketplace that's recently been designed and turned into a giant hotel company. They had an enormous congestion problem to solve, which you can understand by thinking about how Hilton hotels would work if you could only inquire about one room at a time. That's sort of the story with Airbnb. The majority of their hosts only offer one room or one apartment at a time. So if you had to reserve a hotel in Davos by calling the hotel and asking about a particular room, you'd say, how about room 412? And they'd say, I'm sorry, room 412 is booked. And then they'd hang up and you'd have to call back and say, how about room 414? They'd say, oh, you want room 414, of course. That would be a difficult problem, but that's the problem that Airbnb had to solve. So they're very much a mobile communication kind of company because they have to have a website that allows you to see what rooms are still available. And to relieve congestion, they do things like, if I'm offering a room and you try to reserve it, my room will become invisible to other people even before I've accepted your reservation because what they don't want is that when I finally get around to accepting your reservation, I should have to reject five people who have waited all day to find out whether they could get the room. Another problem that you mentioned is the incentive in certain markets, you mentioned this in labour market conditions, to give offers to people earlier and earlier, and in which neither side can have close to the information needed to make a sensible decision. Could you explain why that incentive arises and what problems that creates? It's a fascinating example. So markets unravel in ways that makes the market thin and also deprives them of information and it comes from a kind of unproductive competition. So right now, if you were a graduate of a fancy American law school, the job that you might like to have is to be the clerk of an appellate judge and you graduate from law school after three years, but you would have gotten the offer to clerk for an appellate judge, an exploding offer, possibly in the summer after your first year of law school. So you'd have gotten it two years in advance. And this has happened to house officers in Britain, it happened to doctors in America. And it happens because judges who are a little late find that the clerks they're most interested in have already taken offers elsewhere. And in the United States, the doctors have solved this problem, but the lawyers have not. So markets can persist in inefficient states for quite some time. They don't automatically correct themselves. That just confirms one's general view that lawyers don't understand rules. The lawyers understand the rules very well. It's fascinating to study their markets because many of them earn their livings by following the letter of the law while evading its spirit. They don't understand the spirit of the rules, perhaps a better way. So how do you...actually, before we go to the solutions to these various sorts of problems which are obviously profound, talk a bit about, because I think it's fascinating and very, very important, immense, as it turned out, social implications, about the transplant problem, as a problem. You mentioned something. What are you trying to do and what are the various design problems you have to tackle to make a kidney exchange work? OK. So transplantation is the treatment of choice for a number of diseases involving organs, and I've been involved with kidney exchange. The reason there can be kidney exchanges is because you have two kidneys. We can't do a heart exchange. Not yet. There aren't living donors of hearts, but there are living donors of kidneys. But there aren't enough organs for transplantation. This is a problem around the world. One feature of this problem is we almost universally make it illegal to pay an organ donor, so there's a queue of more than 100,000 people waiting for kidneys in the United States this morning. But we only get about 11,000 deceased donor kidneys each year, so people wait for a long time. It's costly to wait. They die while waiting in the thousands. So there's a real shortage. And one approach that economists have taken but without success so far is to change the law about whether you can compensate kidney donors. The fact that we make it illegal to pay for kidneys makes selling a kidney what I call a repugnant transaction. I devoted a chapter in the book to the wide variety of these and the consequences I've had. But in the meantime to try to get more transplants what my colleagues and I have helped our surgical colleagues to do is to institute a number of thick marketplaces for kidney exchange, which I began to describe to you. And one reason why there are as few living donor transplants as there are, there are about 6,000 a year in the United States, is that sometimes you're incompatible with the patient you would like to give a kidney to. And kidney exchange removes that constraint. It allows exchange of kidneys and the simplest exchange is between two pairs. But kidney exchange is a very congested market because the same law that forbids you from paying me for my kidney forbids us from making any contract about a kidney at all. We can't give valuable consideration for a kidney. So we can't write a contract that says we'll give my partner and I, we'll give a kidney to your patient today and you will give a kidney to my partner tomorrow. That's not an enforceable contract. So when we do pairwise exchange we do them simultaneously. That means we need four simultaneous operating rooms and four surgical teams. So that makes the market very congested. It's hard to consider larger transactions. And one of the big innovations has been non-simultaneous chains begun by non-directed donors. And we've had chains that have 60 people in the picture say 30 transplants or more. The biggest numbers are bigger now. And chains of this sort cover about half of the transplants we do this way in the United States. So kidney exchange has become a standard form of transplantation in the United States. And it's starting to spread. There's kidney exchange in Britain. It's a little bit here in Switzerland. It violates German law. In Germany I couldn't give you a kidney. Only a member of your family could give you a kidney. That's partly out of an abundant concern that maybe you had bought my kidney. Yes, I suppose the history of experimentation might possibly influence this, but I don't know. In the solution, we've sort of got to the solution on this. We can come back on other things. There's a key role played as a sort of triggering of the altruistic donor, as I understand it. Is that correct? Well, in a non-simultaneous chain. In the non-simultaneous chain. Any way you can get these huge changes. And that begins with an altruistic donor. And that way we don't have to close the circle. And that way if a link breaks and the chain ends, the cost is not as great because each patient donor pair receives a kidney before they give one. And that motivates them to, and also it's work for them and they can continue. And one of the interesting behavioural economics things about kidney exchanges is that people are much nicer than economists often give them credit for. As long as they're not economists, isn't that the general lesson from others? There is some evidence that if you study economics you become a very nasty person. I'm not persuaded by that evidence. So as a result, these chains are very successful in the US. There are very few broken links. Which is, I think, very, very interesting. Tell us what this, there's no, of course, official exchange solution to this. Tell us how this way of thinking about things should influence the way we look at marriage and marital partners. Well, just how widely it goes. Of course it's obviously a matching market. Well, so marriage markets have to be thick. And as the age of marriage has advanced in much of the West, traditional marriage markets have gotten less thick. So when you used to marry your high school sweetheart, that was the time when the market was thick and the reason you should... Because everybody was available. Everyone was available and later, hardly anyone was available, the market would be thin in the future so you should marry in high school. As we had more and more participation of both men and women in tertiary education going to college, that created a thick market for a longer time. So you no longer needed to marry your high school sweetheart. Maybe you could marry in college. And the internet is solving the matching problem. Well, as more and more women participate in the labor force, they find themselves sometimes in a thin matching market. And I think the internet attempts to build thick matching markets even when the people who you are meeting from every day are just your boring colleagues. And who you don't want to marry because you're no longer in a thick market like being in college. So I think that's partly why we're seeing increasing internet presence in the marriage market. You have a very interesting example which might well interest people here particularly of what you might call costly competition in the case of financial transactions. The market for speed, which we've heard a lot about. The super fast computer driven matching trades trading and the immense resources invested in order to reduce the trading speed by micro-millig seconds. And you make the argument and I'd like you to explain that this actually can reduce liquidity in the market. Maybe that's what we're seeing right now in all this turmoil. I don't know. So what is the problem and how might I think it's colleagues or put forward an extraordinarily ingenious solution. I'm just interested if you talk about this because it's something we're thinking about now. So in my book I talk about the work of my student Eric Boudish who's now a professor at Chicago and Peter Crampton and a colleague of theirs. And they look at high frequency trading and they observe that because of the investment of billions of dollars in faster and faster communication channels, it's now possible to get messages about prices back and forth between the New York Stock Exchange and the Chicago Mercantile Exchange in under nine milliseconds. And it takes you hundreds of milliseconds to blink your eyes so this is very fast. And because it's very fast it allows the traders who use these fast communication tools to sometimes trade on information in one market based on in the other market before the information has arrived for slower traders. This apparently reduces liquidity because liquidity providers give a bid and an ask and what they now have to worry about is that their stale bids and asks will be executed when they're behind the news if they're slower than the fastest traders. So they have to give wider spreads and offer less volume at each spread. And the solution that Boudish and his colleagues suggest is that we could run markets as call markets use them to clear supply and demand once a second. And for many of the commodities traded on the New York Stock Exchange and the Chicago Mercantile Exchange once a second allows a very, very thick market. Thousands of transactions are offered in a whole second. But when you look at the millisecond level the market is very thin and that's part of the issue. So this would involve rule changes that could happen at the exchange level but could also happen at the regulatory level. So what we will be trying to do is the aim will be to slow markets deliberately in order to ensure there are more participants in the market. That's what makes them thicker because at the microsecond they're just too thin. That's right. The markets are very thin and there's a hazard in putting an offer out to the market if other traders are faster than you and might only take your offer when it becomes stated. Now if I think about your general problem, one way to solve these sorts of problems is to turn things into commodities. You discuss that but in some cases that's very, very difficult. People are not commodities. Standard labour market theory in the old days used to treat them as if they were. So people aren't commodities. Time is precious. Time dimension. So just go through the sort that you mentioned, the resident problem. Just discuss a little bit more detail now. You've discussed a little of the problem. Selyd it out and how you solved it. Well, there's a clearing house for medical residents. It actually has existed for a long time and what it does is it... You even discussed some of the mistakes that were made. So that's very interesting. So go through that a little bit more. OK. Well, so let me talk about that in two ways. First there's the clearing house itself. So today that's a computerized clearing house and what medical students do is they go around on interviews and then the students and the jobs, the hospital residency programs each fill out rank order lists that are submitted to this centralized computerized algorithm which produces a suggested match that everyone is then asked to sign the necessary contract. So the first thing before I tell you what the computer does is that it's a computerized algorithm because it does a good job. It gives people an incentive to wait and participate in the big marketplace instead of trying to go early and do better. And if you make me an early exploding offer the fact that I know that there's this attractive market still waiting to happen allows me to decline your early offer and wait for a better one. Now then because it's computerized it has a chance of dealing with congestion even though everyone is on the market at the same time computers are very fast so offers can be made and rejected. Now none of this would be of any use if what went on in the computer were not helpful to the people in the marketplace. So it's important to have good algorithms that make it safe for people to put down their true preferences because you can't give people what they want if you don't know what they want. And many algorithms that we've encountered in centralized clearing houses don't make it safe to put down your true preferences. This has been a particular problem in school choice where many American cities adopted procedures in which if you didn't get your first choice you would do very badly. So when they asked you to rank your schools you had to think not just what schools you liked but which school you could get if you claimed it was your first choice because if you didn't get your first choice you'd come to harm. So you want to avoid that kind of problem and it turns out to be possible to do so. There's an algorithm called the Deferred Acceptance Algorithm that was developed by David Gale and Lloyd Shapley in the 1960s for which Lloyd Shapley shared the Nobel Prize that he and I won. David Gale was no longer living. And one of the things that I proved when I first started to study this was that this algorithm makes it safe for you to state your true preferences because even if you don't get your first choice you won't have any less chance of getting your second choice than if you had said it was your first choice. So it has that property throughout and that makes it safe for you to state your true preferences. The way I got involved... The key point just at the end of it you've discussed this term if I remember which comes out of this that it works because everybody ends up in the best possible position. They could be given the preferences of everybody else. So it's a stable equilibrium. What that means is when the match is finally proposed if you are a residency program and I am a young doctor that means it will never happen that I would rather have come to your program and you would rather have hired me than someone who you were matched to. So we can guarantee that that doesn't happen. That way if I come to you early and say can I have a job you are disinclined to say yes because you know that if you can't get me you might be able to get me but you might be able to get people who you like even better than me. So there's no point in jumping the gun and trying to transact early. And that's how it solves the unraveling problem as well. So we've had a very good, I think to me, very fascinating discussion of the ideas and issues raised by the book. You now have six minutes in which to raise any questions. It would be a very brief question. I'll perhaps take two or three and say who you are and just please one sentence. And I'll put three together. Hi, I'm Praveen Nair from New York University and I could just finish reading your book. So I have a quick question. What if everybody prefers the same seller? So everybody prefers the same organization is the first question. The second one was could you help us allocate people to sessions at Davos? You know the interesting thing is we were discussing that very question because they're making such a mess of it and I thought he has a job. So that's interesting, exactly the same question. Okay, we'll come to that in a second. Anybody else dares to ask a question? Well, apparently not. Speechless Davos crowd, unheard of. You have flawed them. So what happens if everybody prefers the same seller? So if everyone prefers the same employer then at every stable outcome that employer gets its choice of employees. If all journalists would like to work for the financial times then at every stable matching the financial times gets its first choice. Because if they didn't then they would be a journalist and an employer who wished to work together who weren't working together. So it's very good to be in the position of being everyone else's first choice. But that's not a flaw in the market. And the other people get second choice. Yes, some people have to work for their second choice. Third and fourth and so forth. You said that's not acceptable to them. You are proposing voluntary unemployment. Well, one possibility, a crucial point here is you create the rules. People might start, of course this wouldn't happen in America, but you could imagine trying to find ways around the system. Corruption, calling on your father's best friend college who happens to be and so forth. How do you enforce the stability of a rule system? Which of course applies to the stability of any market system. That's a great question because of course the mathematical tools we use depend on models that look at a piece of the action that we hope is the most important part. Monitoring the market looks at changing, looking at the way the technology of the market is changing that might make that market obsolete. So when we do school choice one of the things we always try to check on is we say where did you assign children to school? And now let's look at who is in the classes come September. Are they where they are assigned or are they are different children showing up in the classes? Often that's how we help schools diagnose that there's some problem with their school choice process, the children aren't where they were sent. Is this because, it could be because the algorithm doesn't work properly in a particular case, but it could be because people are gaining the system? Often it's because people are gaining the system. People have large strategy sets so one of the things to be aware of is that when people are mutually trying to make agreements with each other they can often find ways to do so. That's why stability is such a compelling idea. It may be that I would like to work for you, but if you are not interested in hiring me it's much harder for me to gain the system and end up working for you than if we are both interested that I work for you. Then we can often find a way to make that happen outside perhaps of the rules of the system. And we see many systems where that happens where most of the action is outside of the official system. So let's give in the last minute or two that was the immense advantage of free consulting from one of the greatest market design theorists in the world. How should they be allocating people to the rooms that they have here? Well, market design is a game that everyone can play before we began. You suggested that perhaps they should have everyone who signs up for a session, give their credit card as well and if they don't show up for the session. And they don't cancel. Pay a charge. One thing that obviously that they could do is advertise more quickly empty seats. If that's the problem. That's a problem that comes up in school choice. One reason we like to have what we call unified enrollment systems is so that children don't have assigned seats at two schools simultaneously and you have to wait till the first week of the school year to find out where they're going to come. That causes a big problem in a number of cities and that's why we like to see district schools and charter schools in the same enrollment system. Yes, that makes a lot of sense. I think we can go much further. But I think the point up first of all, it's an immensely interesting enjoyable book. I think the non commodity markets and things that don't have the characteristics of commodities are immensely pervasive as soon as you look at it through these eyes. And the ways around that problem and that have been developed are I think absolutely fascinating. So I strongly recommend reading the book for those of you who are not experts in this area. And that might apply to a few more here. So thank you very, very much. I think that's been a most enjoyable and interesting discussion and you've done some amazing things. I haven't gone into the question of the markets we don't allow things. The markets we don't allow. Republic transactions of which there are so many, but perhaps next time we can discuss the drugs market. Thank you very much. Thank you very much.