Anreize, Auktionen, KI – Märkte neu gedacht

00:00:09: – Sprecher 1 ZEW Wirklich Wirtschaft Background – ein Podcast des Leibniz Zentrums für europäische Wirtschaftsforschung in Mannheim

00:00:17: – Sprecher 3 Nicht jeder Markt regelt sich von selbst. Das kennen wir von Strommärkten oder den Auktionen von Mobilfunkfrequenzen. Komplexe Märkte zu gestalten und durch ein optimales Design schnell effiziente Lösungen zu finden, gehört zu den Hauptanliegen von Marktdesign. Wie wichtig hier eine systematische Herangehensweise ist und wie die Theorie sich am besten in die Praxis umsetzen lässt, darüber spreche ich in dieser Folge mit niemand geringerem als einem Nobelpreisträger. Herzlich willkommen bei Wirklich Wirtschaft, dem Podcast des Leibniz-Zentrums für europäische Wirtschaftsforschung in Mannheim. Mein Name ist Katrin Müller und mein Gesprächspartner hier im Studio ist heute Professor Paul Milgrom von der Stanford University. Unser Gespräch führen wir auf englisch.

00:01:03: Sprecher 3 Hello professor Milgrom, and welcome. I'm really delighted that you could join us in our brand new podcast studio. Thank you for taking the time to talk with us about market design and its challenges, and about your research. Of course, as you know, market design is one of the research fields here at CDW. And to be honest, sometimes it is really hard for us as communicators to explain to journalists or the general public what market design stands for.

00:01:31: Sprecher 3 So let's start right away with a basic question. What fascinates you about market design and how would you define it?

00:01:38: Sprecher 4 Well, like, you know, the market design emerged intellectually in the 1990s, partly because, more and more markets were being run online as web browsers came into being. And, we came to understand that the detailed rules of the market could have, pretty big impact on how they performed. I got involved in 1993 right at the very beginning, with the design of radio spectrum auctions and, that inspired my interest.

00:02:12: Sprecher 4 And I've had some wonderful fun since then with designing even more complicated markets.

00:02:18: Sprecher 3 You are now one of the leading scientists in this field, and you received the Nobel Prize in Economic Sciences in 2020. What has changed for you since then? Except that we know now, thanks to your coauthor and friend, Robert Wilson. Did you have a doorbell camera?

00:02:34: Sprecher 4 Yes. Well, I do have a door camera. Well, it certainly has, impacted the way people see me. The Nobel Prize is very famous. I remember my, grandson, came back from school the day after and said, even my teacher knows what that prize is. He was, so impressed. So, you know, it's a very famous prize.

00:02:59: Sprecher 4 And, people take me seriously. I don't have money. I don't have to prove things to people anymore. When I have ideas, people are more willing to listen. So those are nice things. I mean, those are, a pretty big impact, actually.

00:03:14: Sprecher 3 That sounds good. One of your great contributions in this field has been to bridge the gap between theoretical auction models and real world implementation, which insight from theory turned out to be crucial for the actual implementation of auctions. And why is the historical approach so important for you in that context?

00:03:33: Sprecher 4 Okay, well, those are two questions, right? As an economist who came to this as a theorist, studying market design and, contracts and other things, incentives were always important to me. And I think it was a surprise to some people, the amount of attention, we paid to the incentives that were created by market rules.

00:04:00: Sprecher 4 And in fact, in the work I'm doing right now, it still surprises people sometimes how much attention I paid to incentives. In, in the case that I talked about yesterday and I'll talk about again today, actually, I was talking about the history of the markets because, market design, what what people do at any time is responsive to what's going on in the market.

00:04:23: Sprecher 4 Often there where we're solving a problem that has existed. And as the technology changes and there's one problem solved, another problem emerges as the main one to be solved. And so we see a sequence of innovations often. And they're, they're linked to by history. You want something that will be easy for people. You want solutions that will be easy for people to adopt so that they don't resist your solutions.

00:04:50: Sprecher 4 And so that these things are implemented successfully. So so the history matters for that reason.

00:04:55: Sprecher 3 So let's talk more generally auctions and markets allocate scarce resources. And this has distributional consequences. Exactly w we work for example on auction markets for energy or mobile frequencies, as well as on matching markets for day care places or the distribution of free food. How does effective market design reconcile efficiency and fairness?

00:05:19: Sprecher 4 Yes. Yeah. So some of my students have been very active in that area. Period for Jack. Mohammad Akbar Pau and others and, and you know, it comes up also in, in, in markets that involve money, as you've described, but also in free markets and kidney exchange and, and so on. And, and, and new methods are being developed, to do that, to take and to take those things into account.

00:05:51: Sprecher 4 I guess I can talk about one example Mohammed was involved in, in, in the change in, kidney transactions in Iran. Iran is the only country in the world where where you can actually, pay someone to donate a kidney. And, what Mohammed had noticed is that that people who accepted payments were usually poor people, who were willing for money to give up, kidney and extra kidney and, and, he thought that competition among them resulted in low prices and that if you cared about, equity and redistribution, you might want higher prices for kidneys.

00:06:33: Sprecher 4 And he convinced the Iranian government to substantially increase the price. And, it didn't have much effect on the demand side, but on the supply side, it created, more income for people, but also created a need for rationing, because when you increase the price, the supply goes up. So there's some fancy market design that's involved in doing that.

00:06:54: Sprecher 4 But, it's an example of, of, taking into account, redistribution.

00:07:01: Sprecher 3 So on your opinion, should market designers explicitly build in social objectives beyond pure economic efficiency?

00:07:09: Sprecher 4 So when I'm working on market designs, usually what the objectives are, are, specified by the people who are hiring me. When I did the US incentive auction. Broadcast incentive auction, for example, you know, they had a time frame within which they wanted things done. They had, a certain public image that they wanted to create.

00:07:33: Sprecher 4 They had simplicity objectives that they stated. And then, you know, politically, it was really very interesting. The, the Democrats and the United States style, which is their, you know, more left party, was interested in clearing a lot of radio spectrum for use in, wireless data. And the Republicans are more right wing party. Was interested in the amount of revenue it would raise to help, lower the public deficit.

00:08:03: Sprecher 4 And so, you know, and I was hired by the government, and I had to take as given what the, what the objections would be, but they can be quite complicated. The objectives, they, they're not always just as simple as efficiency.

00:08:19: Sprecher 3 Let's move on from the manmade markets to the digital driven world. Given the growing importance of AI in science, but also in the daily use, how can I influence market design and what are the consequences.

00:08:33: Sprecher 4 You might have asked the reverse question to how can market design help? I am, that's one of the problems I'm working on right now, actually.

00:08:40: Sprecher 3 Next question.

00:08:41: Sprecher 4 Your next question. Okay. Well, let's let's take them in order then the, how can I help market design? Well, I've noticed that, there are proposals to, to let AI assist bidders. I have some skepticism toward that. I tend to prefer simplicity. The we used, machine learning in the incentive auction as part of the development of the algorithms that were used by the auction.

00:09:11: Sprecher 4 We we, had an objective of making things simple for the bidders, but, but putting all the complexity under the hood, you don't need to know how your car works to know when you step on the accelerator what it does. And we had that, view toward, the reallocation and the moving around of radio spectrum.

00:09:32: Sprecher 4 And, and machine learning techniques were certainly a big part of doing that.

00:09:38: Sprecher 3 So from the other hand, what can market design do for AI?

00:09:43: Sprecher 4 Right now, my company auction, Imax has partnered with a security exchange called one Kronos. And we are working on developing, markets, financial markets for compute. It's very, very expensive to build data centers. And the, raising money to do that involves hedging the risks that are involved. And currently there's no way to hedge risks.

00:10:12: Sprecher 4 It's difficult to borrow from a bank, to invest in, processors and GPUs, when, nobody's quite sure what they're going to be worth a year from now, because we don't know what the next generation will be doing. But if you could hedge that risk, that would open up more financing and and a more rapid build out of the data centers.

00:10:39: Sprecher 4 We're going to need to, to take advantage of AI. So that's one of the projects I'm working.

00:10:44: Sprecher 3 On right now. That sounds interesting. So I guess the historical point of view doesn't matter there, because we don't have much history with the AI. But how do you do this?

00:10:53: Sprecher 4 Well, yes or no? Right. The, you know, when when you develop financial markets, you want them to work together with the existing markets to be familiar to the people who are investors in markets so that that doesn't look strange. So you take a look at, at other markets that have benchmarks, S&P 500 markets or, you know, that day where, where benchmarks are used and, and use that as a model to make the markets that we create look familiar so that people can use them easily.

00:11:29: Sprecher 3 So from the global markets to the German market, let me ask a question closely tied to the German context, what role can research institutions like CDW play in linking theoretical work with policymaking and practice, for instance, auction design and evaluation, data analytics, advisory work, particularly in the European context? Do you think that the European context and its regulation are more challenging, that in other regions.

00:11:59: Sprecher 4 There are places, you know, there are times here when I, when I think that lawyers run too much stuff and don't understand the, technical challenges involved. There are times, of course, when the when the technicians get too focused and, I mean, there's some balance to be reached. And I think the balance is different in the, in the US and, and, and in the European context, there, with respect to many things, there are the, scholars who are doing good work have, had things to teach to, regulators and to policymakers.

00:12:40: Sprecher 4 And one hopes that, that you, you know, that they do a good job at that. They're good at communicating. I think institutes like this emphasize the interface and, allow, research to be put into practical form where it might actually be used. So you mentioned electricity markets or you mentioned, organization of child care.

00:13:05: Sprecher 4 You know, those are nice applications of, of ideas that are relatively modern that didn't exist, you know, 20 years ago. So I think, the value of that is clear. And, the, the value going forward will depend on what research gets done going forward and, and our ability to communicate it to people who want to use it.

00:13:26: Sprecher 3 Before we close, one of our main tasks as the research institute is the development of young talents. Which open research questions would you recommend for young researchers in market design to explore other some hidden gems that you can suggest?

00:13:41: Sprecher 4 So, you know, one of the things I'm proudest of in my own career is my development of young researchers. I have, students who have done extremely well in the last, especially in the last decade, getting good jobs and developing new ideas. And the best of the young researchers are mostly coming up with their own questions.

00:14:01: Sprecher 4 They're not questions that I hand them. I understand that's different from the way some people develop researchers, but the, too much research, too much talent is wasted working on old problems that somebody tells you are important. They are problems that if they were that important, that person would have solved them themselves. If they could do either too hard or, really second tier problems.

00:14:28: Sprecher 4 The, the best young researchers that I work with are identifying and finding their own problems. And I keep challenging them when they come back with stuff that's not important. Said you think that's important enough? Is that something you really want to work on and send them back to find? Because there's certainly plenty of problems in the world that that need fixing, you know, from the climate to aging populations to immigration, you name it, there are lots of problems that need fixing.

00:14:58: Sprecher 4 And, a good young scholar ought to be able to find something where he can, do so. He or she can do some research and, and make it work better and do something valuable.

00:15:10: Sprecher 3 So thank you again for your time, Professor Milgram. It's been a privilege to have you here. I hope our listeners will be inspired to dive deeper into the market designs fears. Thank you for your insights. It has been really, really inspiring.

00:15:24: Sprecher 4 Okay. Thank you very much for having me.