Tyler Cowen - Black Swans

Uh, the question is, what would it mean if there were more black swans than we had thought? And when one has discussions on risk and uncertainty, it's very frequently the case you hear talk of radical uncertainty or Knightian uncertainty. Or outcomes which we cannot enumerate or fully encompass in our minds. And sometimes this talk frustrates me a bit because it's not always well specified, or it can be a little vague, and I wonder, what does it mean? But these are concepts, they've stuck in the public discourse. There's even a best-selling book by Taleb, The Black Swan, written about this. So I thought maybe there's a way I could try to make this question more operational and break it down and see if we could get anywhere talking and thinking about black swans. So my nomination for the operational question is, "What would it mean if there were more black swans than we had thought?" I could have done, "What would it mean if there were fewer black swans than we had thought?" But I think in 2012, actually, going the other direction is more appropriate. So let me just walk through some options and try to disaggregate this problem a bit and, and just make it more specific. So if we decided after the financial crisis that there were more black swans than we had thought, one reaction we might have, the most instinctive reaction, would be to think that we as individuals should hold safer portfolios. And I don't want to rule out this reaction, but I think in some ways it may be inappropriate to the current crisis, because part of the current crisis is people deciding at black swan-like levels that they want to hold safer portfolios. It's by no means all of the crisis, maybe it's a symptom of the crisis, but it's clear that it's happening. So if for a 10-year US security you're earning in real terms -2%, uh, which has been the case lately, to then come away with a piece of advice, "Well, in a world like this you should hold more 10-year notes," well, possibly -2% real is the best you can do.. But it hardly sounds like useful advice. I think also that advice is mixing the private return and the social return. Even if an individual wanted to hold more U.S. Treasury securities or Swiss bonds, whatever, from a social point of view, probably that is suboptimal. So another way of posing the question, you know, what should we do if there were more black swans, or posing the answer, is to ask, what could we do that would increase the value of social hedges? 'Cause just buying a Treasury security probably doesn't increase the value of the social hedge. And here I have a few nominations for things we could do that would lower social risk. But actually, I find it surprising how impotent a lot of them are. One thing we could do is just have freer immigration. So when bad things happen, say to Ireland or to Spain, to some extent we can mitigate the costs by letting people leave and go somewhere else. But you may not know it or feel it living in London, but only 2 to 3% of the world's people live somewhere outside of their native country. So immigration is still remarkably small. Another thing you might think we should do to increase the value of the social hedge is to invest more in education. And this you hear all the time. But I think one thing interesting about the current crisis is how much we've seen that what are apparently commonsensical secondary measures they're not nearly as powerful or as useful as we might have thought. So if you look at the United States and you ask yourself, the absolute return to a college degree, how has it done today compared to the late 1990s? It was actually 10 to 15% higher in the late 1990s, and the gains we see are concentrated in post-secondary education. So we can't all quite get PhDs in economics,. But the notion that education is this useful social hedge, it's actually turned out, at least in the last 10 to 15 years, to be a lot less powerful than we might have thought, for all the talk promoting education. I also think a great deal— if systemic risk is now higher, does that mean we should have larger countries or smaller countries? And I think you could argue this either way. You could say, well, the smaller country is more useful because the smaller country is nimble.. So think back to the '80s, New Zealand, a fairly small country, they have a lot of problems, they reform their way out of those problems, it's become quite a healthy economy. But there's another argument that says the smaller economies don't handle risk as well, they can't save their own banking systems. You look at a place like Ireland or you look at a place like Dubai which required external assistance. And I think actually on that my intuition is that the case for larger countries becomes stronger if you live in a world where you suddenly discover it has more black swans than you think. Another thing one could do is have your fiscal and/or monetary authority create more safe assets. Our Fed has done this. We now pay interest on reserves. That's a new interest-bearing safe asset. It's a lot of value there. It's a kind of— a short-term Treasury bill with some important differences, but nonetheless it's a safe asset. The demand for safe assets goes up, the supply has gone up. That may have helped prevent a disaster, but in absolute terms it hasn't brought a stunning recovery. Maybe it was just the best we could have done. But in general, if you think in terms of black swans and, and you're fond of Taleb's ideas, I find it a very interesting exercise just to ask yourself and ask people, What do you think about current market prices? Because in my view, current market prices can't really make any sense. So here's a world where, say, a 10-year note from the United States is yielding -2% real. How do you interpret that? One way of interpreting that is to say, well, the stock market must really be risky if people are basically paying the government to hold their money for them. If the stock market is really risky, I should be buying a lot of shares now and I will make a fortune because the equity premium will kick in and save me. The risk of equity over the long term has always been an illusion. It's still an illusion today. Voilà, medium-term path to significant riches. That's one view. Another view is to say something like, my goodness, -2% real, stocks must really be risky and that risk has to mean something. So if I put all my money into stocks, I could just be completely wiped out. That risk is at some kind of unprecedented level. It's also possible. There's yet a third view, which I suppose is the one I hold, which is to say something like, the rate of return on Treasury securities and stocks, it had some well-defined relationship in the past, which we call the equity price premium, like 1% on T-bills, 7% on stocks.. But to get your 7%, you might have to wait 20 years, and you weigh off your risk aversion and consumption smoothing preferences, and so on. And there's some equilibrium based on partial market segmentation. I kind of always bought that story. I haven't given it up, but I'm not sure it applies to 2012. So it's quite possible that today there's been some disconnect between the real rate of return on Treasury securities and the expected risk premium on stocks, and we simply can't infer any relationship.. And the market, in a sense, is looking to discover or find out again what that relationship will be. And it will eventually settle on an opinion, and then there'll be some big snap and some price somewhere, or a bunch of prices, will move a lot very quickly. And we don't know what to infer from seeing -2% real, say, on 10-year notes. And I don't know if that's more or less scary than the view that the risk of stocks is real. But again, if you're tempted to think in terms of black swans, One way of making that question concrete is simply to ask yourself, what is it you think a person should be buying today? And I know that a lot of economists like to separate their research judgments from their individual portfolio decisions, but I don't see any reason why we consider this logically tenable. I mean, everyone here has some money, if only lunch money. You do something with it, and that's what you think. So I don't want to hear like, "This is what my theory says, this is what I do." I want to know what you think.. And you all have to take stances on black swans and— not asking about your portfolios, but look at what you've done and figure out what you think is pretty much what I'm saying here. There's another approach to answering the question, what should we do if there were more black swans than we thought? And this is the notion in the psychology and behavioral economics literature, sometimes called risk intelligence by psychologists. There are some interesting papers that look at professional bookies and compare them to gamblers, including gamblers who go bankrupt. When it comes to their risk intelligence, they're actually surprisingly similar in a lot of ways empirically. The big difference seems to be that the people who are professional bookies are far more dispassionate. The people who gamble, they use gambling as a way of getting some kind of temporary or maybe sometimes permanent high. So that's the big difference. Both bookies and habitual gamblers have pretty high risk intelligence, but it's how they manage their emotions. So a lot of behavioral economics, people compare the behavior to a model. Some of the talks before got at this. I never trust the models. So if people don't match the model, I don't know whether to reject the people or reject the model. I think an underrated approach, it's partly Jesse Shapiro's idea, but just is to look at the smarter people. So Jesse has a paper in the QJE. He says, "Look at people with higher IQs. See what they do." I'd modify that a bit. I'd say, "Look at people with higher gambling IQs, the professional bookies, the ones that have survived, of course. What do they do?" And they're more dispassionate. So maybe that's another answer to this question, "What should we do if there were more black swans?" is to be dispassionate. Now I've introduced this idea of black swan and I know I haven't quite defined it. It's a very amorphous, nebulous notion. It's like some kind of uncertainty or risk that you can't quantify or write down or, or somehow pin down. But in my next few points I'd just like to look at that notion a little more closely. I'm not sure that's the most useful way to think about it. Something you can't quantify or talk about or pin down. And sometimes I think it's actually the opposite. So if we look at the latest financial crisis, What we see is that in a bunch of ways it looks oddly familiar. In parts it looks a lot like mid to late 19th century business cycles before there's an effective lender or completely effective lender of last resort, where you have sudden financial and credit collapses based on speculative bubbles, and then there's a kind of recovery. I'm not saying they're exactly the same, but it could be that what we've done by having a shadow banking system and by having a lot of financial interconnectedness, has recreated some earlier era where the actual ability of a lender of last resort to solve all problems is now under question. And of course, one sees that with the ECB and the Eurozone. But if that's the case, if what really happened is that something from the past came back to haunt us, was it really a black swan? Here's a way to think about it. Maybe the black swan is the stuff we've been believing for the last 10, 15, 20 years, the idea of the Great Moderation. Maybe that's a kind of black swan. And maybe the mistake we make as humans is not ignoring the black swans, but we talk up our own black swans, like, "I'm gonna do this. This is new. Things have changed. Never before. A different world now." And you talk yourselves into all kinds of nonsense based on your immediate experience. And maybe the real error is people just don't study history enough.. And we spend too much time living in worlds full of imaginary black swans that maybe are false or don't really exist. I take this view pretty seriously. Also, if you look at what are potential candidates for future black swans, I'm not sure anyone delegated me the right to be the person who names these candidates, but I think a pretty simple list would include climate, as in Elizabeth's talk, war, financial crises, financial panics, And again, you might think we haven't had a lot of war lately, or we haven't had a climate crisis lately, but if you look at history, those are the things which are quite common. And we've talked ourselves into this new view that maybe they're not that important because we're wealthier, we're smarter, we have better institutions. But again, the stuff that's being called black swans, maybe very often we can think about it better as just being history. So one possible moral lesson is that if we think there are more black swans in the world, somewhat counterintuitively, maybe what we should do is study history more. You know, you look at the euro today, ask yourself a simple question: which is the black swan, the collapse of the euro or the euro? It's a tricky question. I think a lot of people are seeing collapse of the euro as the black swan, or possible collapse. There are currency unions in the past, but I don't think one really like the eurozone And I would argue that maybe like the shocking thing that I wouldn't have thought actually could have happened was the Eurozone itself. And a lot of people predicted it would have problems. Martin Feldstein was one of them, Milton Friedman, many others. And, you know, in the years leading up to the Eurozone, I actually never believed they would do it. I read all the newspaper articles and thought, yeah, yeah, yeah, you know, it's talk, but at the last minute someone will find a reason to back out. Of course, I was totally wrong about that.. But I think my mistake there raises the question of what's really the black swan? And we too much take that for granted. Another feature of black swan-like talk that I'd like to open up for investigation is this notion that black swans, the possibilities cannot easily be enumerated or written down or thought about, that they somehow are beyond categorization. That's a difficult claim to judge because if they were beyond categorization, it's not clear how you would show that. You could pull in your black swan from the other room there, walk in with your black swan and say, "Here, here's my black swan," but then you're identifying it. And in that sense, in that moment, maybe it ceased to be a black swan. Let's turn again to some literature from behavioral economics. I've spent a lot of time lately studying What can behavioral economics learn from studying chess games? This to me is a very interesting topic. It's interesting for at least two reasons. First, published chess games, they're all played by experts, grandmasters, world champions. So getting back to the very first session, the question, how much can we trust the experts on their behavioral biases? Well, look at what the experts do. But the really neat thing about using chess to study behavioral economics is in a chess game you really do have a great model of what's the right move, and that's what the computer says. So in so much of behavioral economics, it's what the person does and there's what the model says. Do you trust the person? Do you trust the model? A lot of people out there trust the model. Very often I trust the person. But you go to chess, I mean, I trust the computer. I don't trust the person. Play the person against the computer, computer wins almost every time. That's a no-brainer. You don't really have to argue that. So if you want to measure how people make mistakes, admittedly a select group of experts use computers to study chess games. And another thing about chess games is the moves in a chess game, there are no black swans. Everything in principle can be enumerated. It's a completely fixed closed system. It's not as if an eagle swoops down on the board and like the rook disappears and then all of a sudden, you know, a figure from Harry Potter performs some magic. The pieces are the pieces and the rules are the rules.. So there can't be black swans. But what we see happening in chess games is the kind of mistakes grandmasters make. They're exactly the same kinds of mistakes that are supposed to be black swan mistakes. So what grandmasters very often do, very often they calculate extremely well, as of course computers are calculating perfectly. But the grandmasters, looking ahead in their decision tree, they spot moves which intuitively look ugly or unlikely, and they dismiss them. They don't look at them closely. And the computer, in its stupidity/genius, doesn't know that those moves look intuitively ugly, and the computer doesn't dismiss them. And the way the computers beat the top players is that the computers don't dismiss the ugly moves. In chess, there's even a phrase. We call it a computer move, a move that looks ugly. It's like, "Oh, that's a computer move." The computer will recommend something. You say, "Ah, it's a computer move." You laugh, it can't be good. Computer crushes you, and so on. Now that doesn't prove that other black swan mistakes take that same form, but I find it indicative that even when you can enumerate everything, the real mistake people make, like they kind of see these possibilities. They're there in the decision trees of the grandmasters. They just dismiss them because it sounds counterintuitive. Like the years before the financial crisis, I thought about like, "Could there be something wrong here?". And, you know, frankly, I got this call completely wrong. My, like, stupid human chess move was to say, like, US banks, banks elsewhere, senior management has its own money on the line. They can't possibly have made mistakes that stupid. That was, like, my mental countermove. Turns out it was wrong. But it was highly intuitive, right? They're decisive actors. They did have their money on the line. Reputation matters too. Like, how could this have happened?. And I got it wrong. I was like one of these human chess grandmasters— actually not even a grandmaster, but whatever level you're going to put me at. And I saw it. But I didn't see it in another way. So what I've been finding interesting, this work I've been doing on chess, is unrooting the sources of actual human error. And it's very often that possibilities which look theoretically or intuitively ugly are actually right. And the mistakes we're making are of that for it. What we should do actually is study more history. That would be a way of dealing with the fact that maybe the world has more black swans. Two or three other options I'd like to go through. One remedy I hear for the world having more black swans is this notion that what we need to do is create a lot more markets. Bob Shiller pushes this line, Robin Hanson, people behind prediction markets. And I think you could argue, maybe they don't view their own work this way, but the people who do microcredit and microinsurance, they're focused on other problems. But you can think of microinsurance as a way of creating new markets to protect against things that might go wrong. So the view here is, if you have enough markets, like A, people are more protected. And if you have enough markets, you get a lot more information. So when black swans come, At some point, they pop up in the market prices. So like on Intrade last night, the chance of Obamacare being struck down, it went up from 63% to 73% in the course of a few hours. I don't know what it means, nothing for certain, but one notices such things, and it's a kind of indicator that maybe something is on the way. So this view, like multiply markets, Basically, I agree with it. I don't see any reason not to do it. But I'm not sure how optimistic I am about it as a way to combat black swans. One thing I find very interesting is that in the corporate world, which of course is competitive and, you know, mostly private sector, prediction markets really aren't very popular. Hardly any company out there is interested in using prediction markets. You could imagine a company like having alternative projects put forward on the table and having— letting all the employees bet on them. And then depending which projects pay off, you compensate employees and you generate or mobilize knowledge inside the firm. But no one really ever does that. There are a few well-publicized cases of that happening, but when you look at them more closely, they're not usually like real prediction markets. They're more like the company picnic where they organize a relay race to get people working in teams and to know each other better. —So I find it, you know, one of the most interesting research questions. Why don't private companies use more prediction markets? I don't know the answer. Maybe to some extent, like, management is way more about morale than we think and way less about getting decisions right than we think. And that all you can do is get a few things right and you kind of have to hope you know them to begin with. And then you're just trying to sell your product full blast. And you spend all your marginal energies trying to keep, like, all the crazy people you've hired on the same page.. And you do that with morale and having them bet against each other, maybe it destroys that, whatever, I don't know. It's also interesting to see in developing markets to think about micro-insurance. What I find striking is how rare micro-insurance is, not how common it is. So there's an awful lot of micro-credit, and it also is passing a private sector test. Micro-insurance, much less so. And a lot of the micro-insurance you see is for death. Death is something relatively easy to insure, as we all know as economists. Adverse selection, well, there's when you die, but we're all going to die, so that market can do okay. When you look at the structure that insurance takes in a lot of poorer countries, typically what people just insure is their income. They don't insure against events very much. So, say, like, you live in rural Mexico, and your kids might be sick, you might be sick, you need money for the doctor. The way you buy health insurance is to ignore health insurance altogether and establish ongoing credit relationships with neighbors or moneylenders or a microfinance company or someone, whomever. And then when you have medical expenses which were unforeseen, you borrow more money and you pay it back by working harder. So your insurance is your kind of unused labor capacity at the margin. That's what you use to insure yourself. And I think we're just seeing that the ways in which people insure themselves against bad outcomes, they're really quite rich. And their desire to insure themselves against particular outcomes by betting in a market specifying that outcome or not, it's just more limited than we thought. And you go to Intrade, you look at the markets they have, or the markets that are out there, so many of them are like fun. Like, what are the best functioning prediction markets? Probably for sports, which is great, it's fine, but it's fun. —like celebrities, celebrity death pools, and so on. So when it's fun, people do it. When it's not fun, there's way less of it than we think. So I'm all for trying to, you know, mobilize more of these markets. But one of the things I've learned over the last 10, 15 years is just I'm surprised how little interest there is in markets for a lot of things, even when the internet seems to give us the technology to organize them at very low cost. But still, he asked the question, "What should we do if there are more black swans than we think?" Organize more markets to me is still a perfectly good answer, even if it's not the end-all and be-all of the problem. Finally, we might choose to address things people are doing which are creating more systemic risk. And I would cite the comments that Randy made at the end, or toward the end of the last session, where he talked about tail risk. I think one of the big developments in financial markets and elsewhere is that individuals and institutions are taking on a lot more tail risk. For one thing, it's very hard to monitor. You can hide it in, you know, the trading book if need be. It's very hard for the trading book of a bank to be totally transparent, even to the bank's own shareholders, because the bank's trading, right? You can't just let someone buy one share of the company and see everything that's in the trading book. But I think there's a deeper reason why it's hard to stop tail risk, or two deeper reasons. One of them is the Arrow-Honderbrew model, which says you could pass 10 laws saying, no-tail risk investments, but there's always a way to recreate those positions just using very simple securities with dynamic trading. So I despair a bit as to whether we can really limit tail risk since there are ways of recreating extreme options positions using just like a stock and a T-bill. We all know this from Black-Scholes. The other problem with regulating tail risk is I think we're in a funny world —where penalties on wealthier people simply are not and cannot be very high. So let's say you're someone in a bank or even a governmental agency, you know, the mortgage agencies, wherever. It doesn't have to be private sector. And you have some tail risk strategy that will probably get you good returns for 3 or 4 years, but almost certainly blow up over 10 or 20 years or 30 years or whatever. If you're in that position, not only do you most likely have millions of dollars in a bank somewhere and a house in the Hamptons, but the value of your human capital is really going to be quite high. And the value of that human capital is very hard to take away. So if you take these risks, and even in the very worst-case scenario, your company, your bank, whatever, loses everything, uh, you're just not that much worse off. —so what the theorists in this room would call the "boil-in-oil equilibrium"—we either don't do, or we're unwilling to do, or we shouldn't do, or it's immoral, or unconstitutional. But in principle, you could imagine contracts where people who take tail risk suffer penalties far worse than losing their jobs. Like, imagine doing away with limited liability, and that might limit tail risk. But there's a bunch of reasons why that's hard to do. So the wealthier we get, The more it seems you can take on tail risk with relative impunity. And again, the worst people can do to you is send you back to square one. You still have all your human capital. You have what's in your bank account. Just be careful not to do it in an illegal way. And it seems we're getting possibly more tail risk. And you could make the argument, like, somehow markets figured this out. Like, before Black-Scholes, I don't think markets could have figured this out. Like we just didn't know enough. So it probably wasn't a problem. A year after Black-Scholes, 2 years after, people basically still hadn't figured it out, one would think. It's hard to know, but over time people are figuring out perhaps how to game financial markets. And that makes black swan problems potentially more severe. And we really don't know how to stop them. We don't know how much they're doing it. But I think we at least need to consider the possibility— that strategic use of Black-Scholes, combined with the fact that severe extreme penalties we're simply not willing or able to apply to people who do nothing worse than lose some money, mean we're in this situation where we're getting a bunch of risks that are too big to deal with, or institutions that are too big to fail, or too big to save. Or there are a lot of different ramifications of this, but it could be that in general we're on a path to have more systemic risk. Anyway, to sum up, I know I've covered a lot of distinct points, but my general points are really this. Like the notion of a black swan, I find it very useful in stimulating thought. But for me, it's often not specific or pinned down enough. I much prefer to operationalize the question. Like if you believe in black swans, or a lot of them, or more of them, what exactly does that mean we should do? I've considered some answers with varying degrees of usefulness.. But I think what those answers show is, like black swan problems, they're not really one big thing. You address them usefully by breaking them down, by disaggregating them. Black swan problems, they're often quite different than how they seem at first. For instance, maybe the real lesson is just we study more history and stop thinking so much about black swans. The problem is we don't always know what are the black swans and what not. And maybe we just ought to be in some ways more conservative thinkers or more dispassionate thinkers. But in any case, you look at market prices today, that -2% real return on the Treasury notes, and I think A, we ought to be scared, and B, we ought to all take that as some really pretty deep opportunity to reexamine what we think and ask like, what in the world is going on right now? Is it a coherent set of market prices? What is it telling us? What in our theories do we need to revise? And that I think is a process we are all just beginning to undertake. Anyway, those are my formal comments, and I'll leave it at there. Thank you all for listening.