Transcription
A hedge fund manager is forced to eat his own cooking. So they have usually 50% of their money in their fund, and when they lose money, they've lost any day more than 50 times what the next largest client has in their fund, in proportion of their net worth. It's the same thing of why now skin in the game is very important because, um, for example, helicopter pilots, there are two dimensions of skin in the game.
The first dimension is what I call the crux of randomness. It's the agency problem, and that's sort of known in economics, but not, but given that economics doesn't know about fat tails, there's a fat tail twitch to it. But helicopter pilots, for example, in Brazil, um, requested that helicopter maintenance people take random rides on helicopters. And sure enough, okay, the thing improved. Okay.
The but the dimension of skin in the game that I'm investigating now is completely different. It's evolutionary in, in the following sense. A lot of people engage in this. They're the fools of randomness. They have the crooks of randomness. You have the fools of randomness. The fool, a lot of people believe their own. You see the idea? Like if you look at economists, they believe their own stuff. It's not like they're gaming the system, and when a measure is wrong, you know, their objective function isn't, you know, that they're not penalized by their own mistakes. So, but in nature, you don't have that.
In nature, anyone who endangers others, all right? Okay. You don't have evolution unless those who endanger others are themselves at, at the same, bear the same risk. So, let's say, if, have you been on highways or highways here to get here? Okay, go on a highway. Uh, any per, you remember that pilot, that crazy guy, that pilot? Okay, anybody could do it on a highway. You don't need to be, be on a plane. You can kill 30 people. You can go wild and kill 30 people on a highway. No, you go against traffic and kill 30 people. Why aren't there that many of these? Can someone tell me?
Sorry. >> No, it's not because they have skin in the game. They could be crazy and not have skin in the game. I'm not at the crooks of randomness agency problem. I'm at the fools of randomness. Why? Because they're dead. Like this, the guy who killed the, and some people, he's dead. So you can't have that higher ratio in the population of these people because they end up killing themselves. You know, if they, when they kill others, they kill themselves as well. You see, they're filtered out of the system. You see? So, so it's like entrepreneurs who make mistakes, they're dead. You see, they're there. So this is pretty much, uh, what people don't realize as a filtering tool that you, you're because in an opacity, in a very opaque environment.
>> Sorry. >> Yeah. But okay, there are exceptions, but it helps that they're dying as well, and in the operation, you see, it does help. But I mean, of course, you have exceptions with ISIS, but typically traditionally, the ratio of people who are boss and the population has stayed lower, you know, than, than the level you need to blow up the planet because the nuts like Alexander, Napoleon, and these people would go to battle, would stay at the front line in battle. Hannibal, the nut, complete nut, if you look at it, there's people say make him a hero, he's a complete irrational fellow. All right. Hannibal was at the front line, okay? He was first in battle, and that was the, all these. So that was, that's traditionally what, what has happened. Two, I mean, suicide bombers kill themselves, so they're filtered out. They kill other people, but I don't think that we can talk about suicide bombers as a real danger to the system as a whole, given that the number of casualties coming from them is still very, very minute. Every, say, in America, 7,000 people die, and say, multiply by 20 to get the number of people who die on the planet, and count how many of these come from suicide bombers, and you realize that we're still talking about low risk, much bigger risk.
>> Cost to our society. >> Okay, I'm not, this is, okay. We're talking about skin in the game and as a filtering tool for people who endanger others. Okay, so the, the, I know we're not judging whether terrorism is dangerous or not. I believe it is dangerous, but there's another, I believe it's very, very dangerous psychologically and stuff like that. I'm talking about the mechanism of filtering people in a pool. You see, as a mechanism of filtering bad traits. I was a pit trader, and the mechanism of filtering there was a mechanism of filtering, and, and, and, and people would go bust with their own money. They wouldn't survive hiding risk, and effectively traders don't like to hide risk. What happened is they, they clawed.
Fat tails is when a small number of observations cause the major effect on the properties. Okay. Something you learn in school. Uh, this is my second lecture in this room. I usually, uh, you know, I, I, contrary to what you think, I don't lecture a lot. So, uh, but when I was, uh, here, I criticized something they teach in school called the law of large numbers, which I told them automatically whenever you hear it used or you hear something called linear regression equated with, right? And let me explain why, because if you're sampling, you need a vast, it doesn't work in a real world. Small number of observations, deviations, people what would call them outliers determine all the properties. If you're sampling wealth in America or in the world. All right, you sample the first 100 million Indians. All right, not going to give you the average. All right, unless you hit on a few top people. And if you, you know, that Bill Gates, for example, is wealthier than the bottom billion or something. All right. So you get the idea of fat tail. So that's sort of what, what I mean by fat tail, and it has consequences all across.
There are domains like this one that are not fat-tailed, and that's the rule. The rule is when you have large deviations, okay, the, the, the average and the maximum are close to each other. Okay. And this one is when you have large deviations, one observation typically will represent the bulk of the maximum. Is what counts. Actually, you can get most of the properties from the maximum or the second or the third or the first 10 or first, whatever. But you can have a billion people, you want to get the total wealth of the planet, you get sample the one top 1%, you get more than 50%. Okay, don't tell the tax authorities, because in fact, numbers show about between 40 and 70%. All right. So within, so you get the idea. Okay. That's what I mean, fat tail. And there's a super fat tail. If you take violence, small number of conflicts represent the bulk of people who died in, in history, for example. Okay. And that's super fat tail. You have two classes of things. Uh, death by, uh, knives and death by nuclear weapons or death by violent, by big arms. Although sometimes on the contagion, death by knives can be like this. Okay. No.
Second point, why the world has become, is becoming more and more fat-tailed. Something called, I don't know if you've heard of it, called globalization. Okay. So you can imagine Google in the 18th century starting in a garage dorm and running the world. Okay. It's, it's not so small advantages win or take all effects come from that. And the danger with that is that in the ecology, for example, the diversity in nature is proportional to the, the size of the an island, for example, or a continent. A large continent will have many more species than an island, but will have a lot more, uh, but have a large continent will have fewer per square meter. So, and you lose that diversity. But and then, of course, now variables that we discovered 2007, uh, are becoming more and more fat-tailed. In other words, you have fewer crises, but when they happen, they're deeper. Unpredictability. That's probably, probably what's behind unpredictability that you have, uh, nothing happening, and then a big problem called the turkey problem. And the turkey problem, because a turkey is fed by a butcher for a thousand days, and every day confirms to the statistical department of the turkey that the butcher loves turkeys with increased statistical confidence until Thanksgiving minus one, two days, you have a big surprise for the turkey, right? And a huge revision of belief, and then you can see the statistical machine didn't work. That's pretty much what's happening on the fat tails. We need different machinery, or at least you can tell yourself that the mean is not visible using conventional methods. By sampling the mean, you don't get the mean.
There's a book by a guy called Steven Pinker, "A Drop of Violence." And when we looked at data, we realized two things. One, that he was wrong. He didn't know how to compute the data. And the second thing, if anything, there's a rise of violence, but more concentrated. So if you take history, it's, uh, fewer and deeper. That's pretty much fat tail, and the conventional test statistics don't work for that. Okay. Now, this is number one, and this pretty much explains why, uh, if you're, if you throw garbage after garbage of data, it's not going to work. But, but, but you can see extremes from data. So if you can use data, big data for extremes only, take or for targeted things. You ask a question, is he, does he know anyone who has a beard without a mustache? All right. It's very good for terrorism, but it's not good to predict socioeconomic events. Okay. Except then after the fact, I mean, it's like predicting catalysts. All right. You can't, you know, you're not predicting. And then when, when you have a, like a bridge collapsing, people, you know, don't analyze in engineering. You don't analyze, uh, the, the colors of the last truck that was on it that may have caused the collapse. You look at how fragile the bridge is. Okay.
So this is so far, um, my idea. And here, now let's talk about something, skin in the game and moral hazard. You've heard of a place called Wall Street, right? This is very generalized, actually, even outside Wall Street. Uh, if you make a, you know, banks, they like to make money steadily. All right? Under fat tails, you don't see, because I said, large lot, large numbers operate very slowly under fat tails. Very, very, very slowly. Okay? You need vastly more data to figure out if someone is really making money. Yet, corporation, everybody, they get a bonus, something called a B. If you make money for a year, you get a bonus at your end. No. So you make money year one, you get a bonus. Year two, you get another bonus. Year three, if you live in New York, then suddenly all your jokes become funny. All right? You make money three years in a row. You get more and more money under management. The bank gets bigger, they expand, then okay. And then you continue a lot of bees. And then comes a point when you have a turkey problem, right? When, hey, you know what? This Thanksgiving one is too. The equivalent for the bank, 1982, banks lost more money than the history of banking, and we didn't even have bonuses then at the time, right? More in the history of money center banking on, on one event, 2007, 2008, 4.7 trillion before, of course, the taxpayer came to rescue them. Okay, in an implicit way. And of course, here what people do is they write a letter saying, the odds of this event are so low that we can see it. And surely, there's a sentence, surely it's as much of a surprise to us as it was to you. That was in 1998, uh, when, uh, Long-Term Capital Management blew up, and you saw it again in 2007, 2008, with absolutely no linguistic evolution. Same, same sentence. All right.
Now, this I've generalized. I'm generalizing this to the following. Any situation in which you have the upside without the downside, you're invited to fool people with statistical properties. You have hidden risks that blow up rarely, and then using the Marco, which Marco is all that people will give you here, will tell you that it's very safe when, in fact, it's hidden risk, and you cannot make these claims based on, um, the structure of the portfolio. So, with time, those who survive, whether corporation, the minute corporation go to the market, they start development, is payoff visible profits and steady and hidden losses not born by them. I called that the Bob Rubin trade. $120 million collected from, uh, City Bank. And when City Bank happened, he said, "Well, it was so unexpected." He didn't return the 119 million he should return. All right. He kept, you know, and keep one for the drivers and stuff like that. But no, 120 million. Okay. So, I, I called that, I said, you know, John Gotti, for example, the mafia, they never made that money. Okay. Same thing happened recently after the crisis. I don't know if you heard JP Morgan, the whale. Well, same thing. All these people were getting $30 million of bonuses on something that really, uh, was hiding risk. And of course, they lost it all, and then they spun a story, and people kept their previous bonuses. No clawback. All right. Or minor clawback, if it ever happened once in Swiss bank. All right.
So, but this you can generalize to any profession. >> Ones that are more popular. >> Sorry. Someone said something. I know you, you have your microphone up, but feel free to interrupt if you're angry or something. You know what I'm saying? So, no. Okay. So, here we have a class of, uh, uh, thing, and I call it no skin in a game. If you have skin in the game at all times, you don't have that problem. Okay. This sort of explains this kind of setup explains modernity. Modernity was people getting benefits from their actions, and with the adverse effect not being affecting them because they're not visible. They're delayed, and they hit them later. Okay. And you can generalize to a lot of situations in which, say, you're a bureaucrat, you're going to do something that improves your, your year-end, uh, job assessment, but then you hide risk, and then, of course, when things blow up, you say, "Oh, it's unexpected." Okay. And, uh, so you have an invitation to have steady thing rather than volatility. And effectively, based on this principle, you're going to see when you look at countries and anything, anything that's very volatile, when things are very volatile, guess what? They're more stable. When things are steady, they're very stable. There's something people in finance are definitely, I mean, it's like a generalized fraud because if you look at a metric called Sharpe ratio, which is average return divided by standard deviation of return, the first thing is standard deviation of return doesn't work under fat tails or measuring, uh, risk. Uh, the ones with the highest, the funds from Lehman Brothers, not Le, the firm that went bust, Bear Stearns, not Lehman Brothers, Bear Stearns, the funds, the highest funds, the ones that went bust were from Bear Stearns. They had never lost money until they lost money. Pure turkey problem. Okay. All right.
So now I've set up. I've explained unpredictability. I've explained fat tails. I explained why how people tend to position themselves on delayed, uh, uh, blowups. If you have a job assessment, no morals, and no skin in the game, a hedge fund, incidentally, you don't have that problem. You know why? >> Why? >> Skin in the game. >> They're forced. Exactly. He said it. He said forced. A hedge fund manager is forced to eat his own cooking. So they have usually 50% of their money in their fund. And when they lose money, they've lost any day more than 50 times what the next largest client has in their fund, in proportion of their net worth. It's the same thing of why now skin in the game is very important because, um, for example, helicopter pilots, there are two dimensions of skin in the game. The first dimension is what I call the crooks of randomness. It's the agency problem. And that's sort of known in economics, but not, but given that economics doesn't know about fat tails, there's a fat tail twitch to it. But helicopter pilots, for example, in Brazil, um, requested that helicopter maintenance people take random rides on helicopters. And sure enough, okay, the thing improved. Okay. The but the dimension of skin in the game that I'm investigating now is completely different. It's evolutionary in, in the following sense. A lot of people engage in this. They're the fools of randomness. They have the crooks of randomness. You have the fools of randomness. The fool, a lot of people believe their own. You see the idea? Like if you look at economists, they believe their own stuff. It's not like they're gaming the system, and when a measure is wrong, you know, their their objective function isn't, you know, that they're not penalized by their own mistakes. So, but in nature, you don't have that. In nature, anyone who endangers others, all right? Okay. You don't have evolution unless those who endanger others are themselves at, at the same, bear the same risk. So, let's say, if, have you been on highways or highways here to get here? Okay, go on a highway. Uh, any per, you remember that pilot, that crazy guy, that pilot? Okay, anybody could do it on a highway. You don't need to be, be on a plane. You can kill 30 people. You can go wild and kill 30 people on a highway. No, you go against traffic and kill 30 people. Why aren't there that many of these? Can someone tell me?
Sorry. >> No, it's not because they have skin in the game. They could be crazy and not have skin in the game. I'm not at the crooks of randomness agency problem. I'm at the fools of randomness. Why? Because they're dead. Like this, the guy who killed the, and some people, he's dead. So you can't have that higher ratio in the population of these people because they end up killing themselves. You know, if they, when they kill others, they kill themselves as well. You see, they're filtered out of the system. You see? So, so it's like entrepreneurs who make mistakes, they're dead. You see, they're there. So this is pretty much, uh, what people don't realize as a filtering tool that you, you're because in an opacity, in a very opaque environment.
>> Sorry. >> Yeah. But okay, there are exceptions, but it helps that they're dying as well, and in the operation, you see, it does help. But I mean, of course, you have exceptions with ISIS, but typically traditionally, the ratio of people who are boss and the population has stayed lower, you know, than, than the level you need to blow up the planet because the nuts like Alexander, Napoleon, and these people would go to battle, would stay at the front line in battle. Hannibal, the nut, complete nut, if you look at it, there's people say make him a hero, he's a complete irrational fellow. All right. Hannibal was at the front line, okay? He was first in battle, and that was the, all these. So that was, that's traditionally what, what has happened. Two, I mean, suicide bombers kill themselves, so they're filtered out. They kill other people, but I don't think that we can talk about suicide bombers as a real danger to the system as a whole, given that the number of casualties coming from them is still very, very minute. Every, say, in America, 7,000 people die, and say, multiply by 20 to get the number of people who die on the planet, and count how many of these come from suicide bombers, and you realize that we're still talking about low risk, much bigger risk.
>> Cost to our society. >> Okay, I'm not, this is, okay. We're talking about skin in the game and as a filtering tool for people who endanger others. Okay, so the, the, I know we're not judging whether terrorism is dangerous or not. I believe it is dangerous, but there's another, I believe it's very, very dangerous psychologically and stuff like that. I'm talking about the mechanism of filtering people in a pool. You see, as a mechanism of filtering bad traits. I was a pit trader, and the mechanism of filtering there was a mechanism of filtering, and, and, and, and people would go bust with their own money. They wouldn't survive hiding risk, and effectively traders don't like to hide risk. What happened is they, they clawed.