August 2, 2017; Richard Bookstaber: People Cannot be Controlled like Automatons thus Crises Repeats

The McAlvany Weekly Commentary
with David McAlvany and Kevin Orrick

David, our guest today wrote a book ten years ago that was just amazingly prophetic called A Demon of Our Own Design. Richard Bookstaber, of course, is who I’m talking about. This is a man who understood the inner workings of the way Wall Street supposedly hedged themselves and insured themselves from the downturn that he then was predicting.

David: Let me tell you a number of things that I really enjoy about Richard Bookstaber. One, he opens his first book with some of the most memorable lines you will ever read. Second, he knows the very best Korean barbecue restaurants in mid-town Manhattan. And we have had some very enjoyable conversations there. And lastly, the newest of his books is a real challenge. The End of Theory: Financial Crises, the Failure of Economics and the Sweep of Human Interaction.

In it he is asking a number of questions, one of which, a very important one, in fact, is what is the role of economic theory? What is our understanding of economic theory, and are theories just a partial explanation of the real world? Are they adequate? Do they do their jobs? When we are tied to any one theory, is that, in fact, what blinds us and makes us more vulnerable? In The End of Theory, he is making a statement, he is asking a question. He is looking at the intersection of finance and science, our understanding of the world at a much deeper level, and I can’t wait for the conversation today.

Kevin: David, just to point out, that book that he had written, A Demon of Our Own Design, was published one year before we had the financial crisis. And he called for it. Now he is also actively involved in trying to keep those types of crises from happening. But now he is applying with this End of Theory book – I love the book, by the way – a science called Complexity Science. What he is basically saying is, the theories that we thought were good economic theories break down because there is a dynamic changing environment, continually, non-linear, that has to be modeled if you’re going to try to understand what is going to happen next. I’m really looking forward to the conversation.

David: Richard Bookstaber spent 30 years on Wall Street, and his focus has been on risk management. When he finished his Ph.D. at MIT he went on to oversee risk management at Morgan Stanley and Solomon Brothers. He has experience in your major hedge funds, Moore Capital and Bridgewater, and actually spent six to seven years working with the U.S. Treasury and the U.S. government trying to fix what went wrong in 2008 and 2009 and come up with a better framework for dealing with crisis. So, at this point, working with the University of California and their 110 billion dollar pension fund, he has a broad array of experience – theoretical, academic, but also real life practical dealing with the trading functions of some of Wall Street’s biggest firms, and we have lots of questions for Richard Bookstaber today.

*     *     *

Richard, the last time you wrote a book it was out of an imminent sense of concern that a financial crisis was being cooked up in the kitchen of the world of structured finance and counter-party interconnectivity. Your time on Wall Street as a manager of risk had you looking at the danger signs and signals. That was in 2007. And you spoke up about it in your book, A Demon of Our Own Design.

Now ten years later I see you looking at the financial markets again, writing in The End of Theory, and you’re looking not at leverage as the primary risk in the market, but instead at liquidity. Have we learned enough, in your opinion, from the global financial crisis, to avoid the next crisis, or are you writing again with a sense of imminent danger?

Richard: I’m writing again with a sense of danger. I don’t know that it is as immediately in the future as I saw it in 2007 because frankly, we have looked at the aftermath of the crisis in 2008 and we haven’t learned the simple lesson that the message that we were using then aren’t sufficient in dealing with crises. So inevitably, crises are going to start, and if we don’t get the right tools in place, we won’t be able to control them any better in the future than we did in 2008.

David: Mathematics has become so central to the way we operate in the field of economics, and that really became popular shortly after the Industrial Revolution. You had explanations, you had theories, you have today that are steeped in equations. And that, as they call a neoclassical approach to economics where it is saturated with math. Looking back it has failed to anticipate every major economic crisis. And this is what you suggest, that economists have to give up the myth that the economy is a simple mechanical equilibrium system. Can you expand on that?

Richard: Yes. The problem is that we’re people. We’re human. We have experiences and those experiences change the way we interact. We interact, and the very fact that we interact changes the environment, and economics misses that because as you mentioned, it treats us as if we are part of a mechanical system. Economics patterned itself, and still patterns itself, after physics. And it wants to believe that it can look at us at automatons. But that’s not the way we are, and especially that’s not the way we are during a crisis. So I think that the economics is so rooted in the way that we look at the world, and economics itself insists on trying to use axiomatically based mathematical methods that just are not going to be suited to dealing with crises that are an inherently human event.

David: You said in your book that when we take human nature into account and when we amplify that by the effect of crisis, what do you find? We find manifold failures in economics, failures that require a different approach to understanding crisis. So you write about the end of theory – that is the title of your book – the end of economic theory as we know it. Where do we go if the old theories are, in essence, inadequate – perhaps not dead because they are still practiced by economists, but no longer really workable?

Richard: First of all, I want to make it clear that economics might work well in some setting. If you’re trying to figure out what the demand for dresses will be as a function of people’s income, that may work. But when you talk about a crisis, you have some difficulties. Let me just go through four characteristics that I think really pinpoint the problems of economics in dealing with crises. And it might be that they affect economics more broadly, but really, a crisis is the refiner’s fire. That’s where the issues are going to be most manifest.

So the first thing is that in a crisis it doesn’t look like the past. A crisis has dynamics that really don’t occur day to day. Yet economics assumes the future is going to look like the past. It assumes what in mathematics is called ergodicity. And so if you start with the idea that I can look at the past, I can develop models and formulas that can be estimated based on the past, then you’re missing the essence of the crisis, because a crisis isn’t just a bad draw from an urn. It’s a different urn. It’s different than the past. So that’s one problem.

Another problem is that the world simply can’t be modeled through equations. We can’t draw out equations, for example, and have that determine where we will be over time. We can’t follow life. We can only live life. And if you want to know where the economy will be over time, you can’t solve now for where it will be, you have to follow the path along as you go there, either in real life, or using computer simulation. So those are really two essential characteristics that you have to have in place to deal with crises.

David: In essence, the future is unknown.

Richard: Yes, the future is unknown, and certainly the future of a crisis is unknown as you look at the non-crisis past. One of the problems also is, we create and we invent. And what we invent, and what we create, we can’t, almost by definition, anticipate right now. So you have what is called in some circles, radical uncertainty. It is not as if you have a roulette wheel spinning around and you don’t quite know what number it will hit. You don’t even know the possibilities – what numbers are on that roulette wheel.

And yet for economics we assume that we know every possible fate of the world that can occur, and the probability that it can occur. Again, that may work in some settings but it certainly doesn’t work over the span of a crisis. There is a book that was written some years ago called This Time It’s Different. It was written by economists who argued that there are essential characteristics of every crisis. And that may be true at some level, but really, what creates the crisis can help propagate changes from one crisis to another because we have different strategies, we have different sorts of financial instruments. We are steeled based on the past crises, which almost assures that the future crisis isn’t going to look like the past.

David: Then there is a complexity of human interaction where individual choices combine and it’s not something that deductive mathematics can really get at.

Richard: Yes, that’s right. What does human interaction lead to? It leads to complexity, and sometimes unaccountable complexity. Look at what can happen when there is a stampede after a concert, or even congestion with traffic. Nobody is trying to create the stampede, or create congestion. Each person is doing what they think makes sense as they drive along the highway, and then somehow the aggregate actions of everybody can lead to a period of congestion, or everybody trying to leave the stadium, because then they create a crushing stampede.

Economics doesn’t model or allow for that, because economics assumes we are all optimizing, we don’t really affect each other in a major way, and it is structured mathematically to have an equilibrium world, so things do go off the tracks. It is like there is this mathematically based rubber band that pulls things back on the track. But of course, in real life that is not the way things go. And there are actually different levels of complexity. So start with a non-complex system – a purely mechanical system. Theory can work there and physics can work there.

Now add a little bit of stochastic vibration. It still can work because the vibrations that are imposed usually are really constant and well behaved. Now, allow for dynamical interactions and you start to get to the point that things can run off the tracks, and standard theories can’t work. And now go one step further to what happens in the financial markets with what George Soros called reflexivity. When I do something, it changes your experiences and your view of the markets, which then changes what you do. And then when you do what you do, I look at it and change my expectations. Now you’re beyond the realm of what a theory can solve. You have to, as I said, kind of follow truth along the path to see where it is going to go. You can’t start with a bunch of equations, plug in numbers, and think you’re going to get to the answer.

David: I guess part of what we’re talking about is a critique of the assumptions that go into an approach to economics, particularly in the context of crisis. As you say, it works in certain circumstances, but in the context of crisis, there is that refiner’s fire of economic theories where, frankly, in terms of a stress test, what we find with modern economic theories is they get the fail, not the pass. What you are suggesting is something called an agent based model which you prefer to the older economic theories. Talk to us about simulating circumstances and attempting to map complex and adaptive systems using an agent based model.

Richard: Here is the point. We are all agents. We all do what we want to do. We all have different rules, or heuristics, on how we operate. Economic systems were all totally rational, optimizing people. And so they kind of put us all together as one, what they call a representative agent. And as soon as you do that, you don’t have the possibility of interaction. So each of these modelings start with a recognition that there are a lot of different people, all of whom are doing different things, and you have to start it at micro level.

The best example, I think, to get sense of agent based modeling is to look at traffic flows. In fact, agent based modeling is used a lot in modeling traffic. When you look at traffic on a highway, you have a lot of different agents. Every car, every driver, is a different agent. Each driver has a different sort of heuristic, or the way that they go about driving. Some speed, some are tailgaters, some tend to go slowly, and some go slowly in the fast lane.

So what you need to do if you want to understand what the flow of traffic might be like, let’s say that you are wondering whether you want to go an off ramp at one place versus another. You can, in the simulation, pepper the roadway with a bunch of cars that have these different characteristics, and now in the simulation let the clock move forward, second by second, or ten seconds by ten seconds, each car moves along, and suddenly, one of the cars based on the way it behaves, changes from one lane to another. When it changes, the car behind it, based on what that car happens to do, puts on the brakes because it doesn’t like to tailgate.

Now, the environment of all those cars around there has changed, and so each of them reacts based on what has changed. And that is happening on all these different parts of the road, and the end result can be that you discover that when one car does this, and another car does that, you end up getting congestion. Well, the same thing is what should be done in financial markets. You have agents there, and when you are talking about large-scale crises, you don’t have to look at the individual investors.

You can say we have banks – we have City Bank, we have Morgan Stanley, we have Goldman-Sachs. We have assets managers – Fidelity, and actually the portfolio managers within Fidelity. And we have hedge funds – Citadel or Bridgewater – those are the agents, and we can identify them. We don’t have to do what an economics model does where it says, “Assume N banks, G asset managers and M markets.” We know the markets and we know the banks. So that is really the beginning structure of how an agent based model would look at the financial system. And you’re not trying to boil it down to a representative agent. You’re not trying to find a set of formulas that do it. You’re trying to create the reality of the market in the simulation and then see how it might move forward over time.

David: So, in that context, the world cannot be solved, as you like to say, it has to be lived. Is this approach valued – the agent based model. Is it valued for providing better understanding and a more appropriate policy response to crisis? Or is there the possibility of getting out in front of a crisis and having mapped all possible responses, and then actually influence outcomes?

Richard: That is where I would like it to go. Right now, to answer the first part of your question, it is a new paradigm that in some respects, and in some areas of life a crisis would replace standard neoclassical economics. So, it tends to not be pushed by people whose livelihood and reputation rests with neoclassical economics. That includes academics in the U.S., it includes the Federal Reserve, and so on. There is interest in it in Europe, interestingly, because they are not as entrenched. So it is slowly making its way into the policy area. And I left industry for a period of six years, actually, after the crisis. I went to work in Washington to help reformulate the system, and I developed an agent based model to help asses vulnerability.

And it remains to be seen how far that goes, and what the regulators decide to do. But I actually believe that agent based modeling can be used to try to understand an get ahead of a crisis – not to predict a crisis, necessarily. It’s like in the military. You may be able to understand where the vulnerabilities are in your defense, but you may not know exactly where the enemy may try to come in. So we can understand the vulnerabilities and anticipate – kind of have a war plan, a battle plan, in place so that when something happens we can say, “Oh, I understand which agents, which institutions in the market are under pressure. I understand if they’re having to sell that’s driving prices down.”

I’m seeing prices go down 5%, 10%, 15%, 20%. And because I understand the dynamics, I, that is, a big pension fund or sovereign wealth fund or whatever, I am willing to go in and buy. Because if I buy 25% down and things recover, I’ve made a lot a money, but the very fact that I’m buying helps to dampen the crisis because the reason prices are dropping more and more is some people are being forced to sell. And there doesn’t seem to be anybody on the other side.

So my argument is not to try to get this model in the hands of regulators that tend to be really slow on the draw, although if they have it, that’s great. It is to get it in the hands of people who have capital at the ready, to make them more willing to go into the market during times of market dislocation. And when they do that, they can help, maybe not stop the crisis, but dampen it.

David: You say that a financial crisis is an emergent phenomenon that has broken through the containment vehicles, and that what is locally stable can be globally unstable. And this sounds like what you just described – understand the interactions between players and some of the cause and effect, perhaps mapping out what responses would be under pressure. Talk to us a little bit about emergent phenomena, and again, that idea of global instability.

Richard: Think of what goes on at each institution. I have worked in a lot of different areas. I have worked in banks and brokered dealers. I was in charge of risk management at Solomon and Morgan Stanley. I’ve worked at some of the bigger hedge funds – Moore Capital and Bridgewater. I have worked in government. Now I’m working at a large pension fund. Everybody in these organizations has a serious concern about risk management. They all are trying to be prudent in what they do, just like every car on the roadway is trying to be prudent in how it drives. It is not trying to create congestion.

But the problem is, if all you do is look at your institution and what you’re doing, and don’t understand how what you do might change the environment and change how other people act, you’re missing the essence of what creates the crisis. And you know, this agent based methodology is used to try to understand the nature of stampedes and congestion. So it’s not being applied in an unusual way when we try to do it for crises. But that’s the point, that everybody can be locally prudent. Everybody within their organization can do what they think makes sense, but the end result, on a global basis, can be globally, let’s say, imprudent, or globally unstable. And that is the essence of when you get a crisis, when, as you said, you’re kind of breaking through the containment vehicle that is embedded in standard institution-specific risk management.

David: Borrowing from some scientific ideas that go back better than 300 years, you talk about the three body problem as an illustration of how modern economic theory falls short in the context of crisis. As you just said, you have an institution that things about themselves and maybe one other entity, but not beyond that. Maybe you can review for us how the interactions of multiple factors make the assumptions of stability within the system hard to swallow. It’s not practical, unreasonable.

Richard: People might say, “Okay, fine, you’re talking about interactions and stampedes and areas where standard mathematical methods can’t deal with the problem. But how often does that happen? And my answer is, that type of problem exists almost everywhere where humans are involved. And to get that point across, I go into what you are talking about, this three body problem. This problem is one where let’s say you have, in the model, three planets. They are just out in space all by themselves and they are rotating around each other. This is a very simple problem. There are just three planets. Their interactions are very well defined based on simply the velocity relative to each other, how they are moving in the three dimensions of space, and gravitational force, which is very well known. Yet, generally speaking, if I show you right now where those three planets are, what their masses are, and the direction each one is going, and I say to you, “Give me a formula that will tell me where those planets will be, say, three months from now,” you can’t do it. You can do it for some specific cases. We can do it, for example, for our solar system, but broadly speaking, you can’t do it. The only way you can figure that out is, in simulation, follow those planets second by second, see how these interactions lead them to go, and sometime they will just fly apart. More often than not, they just fly apart. But when do they do it, and how do they do it? You can only know by following along that path. So my argument is, if something that simple, where there is no uncertainty – it’s all mechanical – can’t be figured out using mathematical methods, why do we think that they can be useful when you have many institutions, a lot of uncertainty, and the various heuristics and rules that people are using? Between traffic, as an example, and the three body problem as a motivation, I think you kind of get to the point where, in my mind, the need for something like agent based modeling is pretty apparent.

David: The focus in your first book was on complex, interconnected systems. Certainly leverage played a significant role in the last crisis, and there are a variety of ways for leverage to proliferate. I know bank balance sheet leverage has come down since the global financial crisis. What are your impressions of derivative market expansion, corporate balance sheet leverage increasing, margin debt exceeding all-time highs in nominal terms, and as a percentage of market capitalization. Is leverage no longer the factor it was during the last global financial crisis?

Richard: I think there has been a lot of focus on leverage, and I don’t think it is as dominant a factor now as it was then. I think we are sorting of fighting last year’s war. Not that we should ignore leverage, but our focus is very much on it. There is another side to this scissors. Leverage is important, but another thing that is important is liquidity, the capacity of the market to absorb any flood of selling that occurs. So leverage creates the need under some shock to have to sell into the market. If you lower leverage, the amounts of selling will reduce. But what happens to prices is determined by the level of liquidity in the market?

So you can have less selling, but have prices still drop a lot if nobody is on the other side who is willing to buy without a big concession in price. And my feeling is that we need to start focusing more on the potential illiquidity of the market. And people are starting to do that, to some extent, but it is a very difficult problem to deal with. But I would say that we need to know three things if we want to deal with crises, and actually, in each of these models these three things are paramount.

We need to know leverage, because we need to know at what point people may be forced to sell, at what point might they be selling to get margin call. We need to know liquidity so that we understand how much prices might be driven down by that selling. And we need to know where the pockets of concentration and crowding are in the markets so that there could be a lot of people trying to run out the door at the same time. And if we just focused on leverage we’re only focusing on one part of the problem.

David: You’re at kind of an interesting intersection. As you said, you spent six years working on reformulating the system. Working in D.C. you were able to have a direct impact on the Dodd-Frank legislation and influence the Volcker rule. But here we are in a world where those pieces of legislation actually add some pressure to the liquidity issue you just described. How do we look at liquidity in a world Dodd-Frank and the Volcker rule, where market makers are actually discouraged from playing the role of liquidity provider to the market?

Richard: Yes, this is a very important point. I support the Volcker rule because you don’t want the market-makers, the banks, to also be trading against their client, which is essentially what the Volcker rule prevents from occurring, but the problem is, a lot of the incentive for broker-dealers to provide liquidity, to be market-makers and take on traditions was because they made a lot of money by trading against their client. So when you take away one thing, they no longer have the incentive to do the other. So there is sort of a collateral damage, or unintended consequence to the Volcker rule, which is that it has reduced the incentive, the profit motive, for banks and broker-dealers, to support a market.

And that is the reason now I think liquidity has become more of an issue. I don’t know quite what the solution is for that because I would not want to repeal the Volcker rule and go back to where we were. It might be, over time, that other institutions take up the slack, and take on the role of being market makers and supporting the market, and actually, in the extreme event it could be that the institutions I was talking about a while ago, the big sovereign wealth funds and pension funds that have a lot of cash on the ready, can provide that function when you really need extreme levels of liquidity.

David: We have a new mega-buyer in the market, which is your index funds, where investors are going from owning individual stocks to prioritizing an exposure to a sector via an exchange-traded fund. So you have the mega-buyer on the one hand. Do you really have a mega market-maker on the other side should you see liquidations out of those ETFs? Because on one hand you have blind purchases being made, just one sector, not a particular company. There is really no discrimination in terms of a particular company within that sector. But if you have a blind purchase, you also have a blind liquidation. How does the market absorb the kind of liquidations that we might see with the transformation of that new mega-buyer in the market – those ETF or index funds?

Richard: The answer is, they probably won’t do it very well. When you have market-makers that don’t have a franchise reason to make market, which the broker-dealers used to have, because if they didn’t make the market, people would stop trading with them after the crisis left and they would lose their franchise. But if you have anonymous groups and they decide, “Boy, this is getting pretty dicey,” and they pull back, that is the end of the liquidity. And we saw that, to some extent, with the flash crash where some stocks dropped to one penny. I wouldn’t say that would occur again, but that is sort of the most dramatic example.

And it is good to point out the ETFs, because certain types of ETFs could be a focal point for this. If you have an ETF on the S&P 500, it is not so much of a problem because the S&P 500 itself is extremely liquid and trades very quickly, just like the ETF does. If you have an ETF, say, on high-yield bonds, there is a sleight of hand going on because it appears with the ETF that you can buy and sell intra-day very easily, that you’re in something that is very liquid. But high-yield bonds are not really liquid. Some of them almost trade by appointment. And it doesn’t bother us day-to-day because when we’re doing a little bit of trading back and forth with ETFs there are people on the other side.

But if things really go off the rails, as I was saying, the actual cash market, the actual high-yield bond market under the ETFs, will not be able to trade with the speed that typically retail investors expect, given what they tend to be able to do in normal times with the ETF. And the same is true with the sort of ETFs you are talking about that are based on factors based on specific industries and so on. So we haven’t had a real testing of the ETF market. But I think there is potential for dislocation, the fact that the ETFs make it appear as if there is a lot of liquidity and you can trade very quickly. But the reality of the underlying market is not the same. It can lead to big problems in the case of a major market dislocation.

David: I kind of want to segue to something a little bit different, but still borrowing from the ETFs. If we take another look at the current market context, how do you assess risk in an age of limited price discovery? You mentioned high-yield bonds. Current yield on European high-yield debt is about 2.8%. There is the central bank footprint in the market today which limits price discovery. How do you assess risk in that environment?

Richard: I can tell you what risk managers do to try to assess risk, and I’m a risk manager and this is what I do. At least, this is part of what I do. You look at the behavior of markets and assets over the past and you measure the risk that they have had. You notice that this asset, stock tends to move up or down a half percent or a percent in a typical day. Worst case is it might move up or down 3-5%, and so you say, “Okay, that’s the risk of the equity market.” We do the same for bonds. We do the same for other markets and assets.

And then you say, “I’m going to assume that the future is drawn from the same distribution as the past. The future will look like the past. And if that is the case, here is what my risk will be. So, you’re using mark-to-market of various prices of instruments. You’re looking at them historically, and you are just carrying that forward. Now, of course, this gets right back to the point that I began with. That probably works well 98% of the time.

Unfortunately, the 2% of the time is what really matters, the times that the future doesn’t look like the past, the times that we have an emerging crisis. And that is where you need to start using new methods, and at this point, I would have to say, I’m the only one who is really starting to employ, for on the ground risk management, these tools. But my hope and expectation is that over the next years it becomes more of a standard.

David: So, following on my last question relating to a central bank interaction within the context of the markets, you say that if you want to understand crisis you have to develop a system that can create a crisis. And there are several points, then, which you consider to be critical. If I’m recalling, I think it’s page 105 of your book, when you discuss context being an important factor you mention that we’re humans, and we need to allow for heterogeneity. And this is what comes to mind for me when I think about heterogeneity. We have central bankers who are playing the role of central planners through studying prices in the market, capping yields, and don’t we lose heterogeneity?

Richard: Well, it’s really what people do based on how the market is being managed and constructed. So think of the central bank as the one who designs the roadway and designs when to put traffic cones up and block exit ramps. They can deal with things at that macro level, but then what we do in our day-to-day action dictates how the market finally responds to that. And because of that, the central banks can only do so much. It is interesting that you have so much of a mechanism in place with, say, the Federal Reserve. They have more economists – they could create three universities based on the number of economists they have. But ultimately they can only do one thing, which is affect interest rates.

So they can only really do one thing – determine, say, the number of lanes on the roadway, and after that it is the interactions of the people in the market that determine how the traffic flow will behave. So in my book, and in my model, I really don’t emphasize the role of the central banks, partly because you can take what they do as the given, and then have the market operate on that, and because they tend to be very slow in how they respond. In 2008 they performed a very important role. They were the liquidity providers of last resort. They were the ones who bought all these instruments that were being flooded into the market. But it took them quite a while to get to that point.

And my focus – and I have to really kind of emphasize this – my focus is not only on periods of crisis, it is also on crises that really are seen within the financial system. So it tends to be crises that emerge pretty quickly before, say, a central bank can respond. If the crisis becomes so severe that it breaks through to affect the real economy, now you’re in a different world, and now the central banks are essential, and you have other kinds of random agents, or at least another agent, in the game, which makes it more difficult to model, and I’m not really there yet. I’m not focused on the point that the real economy gets involved.

David: Over and over again you write of the importance of human experience, and how economics in the modern form assumes a more simplified and repetitive process, which essentially ignores the humanity embedded in the choices we make. If you would, discuss being messy, being stupid, being unreasonable, being ambiguous as human beings, because this is the stuff of literature more than of science, is it not.

Richard: We know that we do things that don’t look reasonable or optimal currently. That’s part of human nature. And sometimes we will say, “Well, that was just stupid.” And sometimes we are just stupid, and we have to account for what George Soros and his theory of reflexivity calls fallibility. We sometimes just don’t get it right, and when we look back we say, “I can’t believe that I did that.” And so, in my book I used examples. I draw from examples of literature, because if you want to understand how people actually act, you can find that more in literature than from what economists are observing. They really are not observers of the human state.

But a lot of what we do that may seem, to an economist, to not be optimal behavior, actually does make sense. And the reason is that we all know that the world as we see it right now is not the world that may occur in a few years, or even in a few months, because we could end up with, so to speak, congestion of a different type because new instruments will be invented, or new methods will be invented. Just imagine, when I was young, the concept of the Internet was simply in nobody’s mind, and now it’s part of the world. So we might behave in ways that don’t seem optimal in the world as we see it today because we kind of have the sense that things can change in ways that we can’t totally anticipate. So we aren’t so fine-tuned to where we are today.

I used an example in my book A Demon of Our Own Design. I also referred to it again in my current book The End of Theory, of a cockroach. A cockroach doesn’t smell or hear or see. All it does in determining an escape from a predator is, if it feels a gust of wind on the little hairs on his leg. Now, that is not a very well-designed insect. And in any one environment, there are probably other insects that are more finely-tuned to that environment. If there is a jungle, there is probably some insect that has just the right mandibles to crush a really abundant seed.

But when that jungle turns into a desert and the desert turns into a city, that finely tuned optimal insect will be gone. But the cockroach, because it has a coarse behavior that works in a lot of different settings, will still exist. And we kind of have embedded in us that same characteristic. So we may not look super smart and optimal in any one setting or world, but we behave in a way that is going to be good enough to get us by as things change in unexpected ways.

David: I like the way that you discuss intelligence on that point, where there are benefits of selective ignorance. And for us, one of the marks of intelligence is choosing what to focus on and what to ignore. There are actually a lot of things that we can just completely be oblivious to, and that is a strength, not a weakness.

Richard: Yes, I give examples, actually one from literature, where people have perfect memory, they can remember everything, something that we might think an optimally designed human would have. One of these is by an Argentinian novelist that I cite a number of times in my book, Borges, or some people pronounce it “Borhays.” He has a number of novels sort of pose limitations on humans, some of which we may not even realize are there.

And he has, in one of his short stories, a man who falls from a horse and afterward, for some reason – and he, by the way, fell from a horse and had a head injury – this person falls from a horse and has a head injury, but his head injury leads him to have perfect memory. He can remember what the clouds looked like on a day four years ago. He can remember every crack in every ceiling he has seen.

But the problem is, it actually limits him, because he can’t have creativity. He can’t know what to ignore and what to focus on. And so, for example, if you even tell him a short story, he can’t really follow it because each word you say invokes other images that he has had from years ago. So while it would appear that our need to be selective in what we remember and what we focus on, and the fact that we seem to ignore information is actually a benefit. And if we did the opposite, we would be paralyzed.

David: But there is this assumption in the age of big data that there are many things that we can know and piece together and maybe even create an algorithm to manage emergent phenomena. Is that thinking that is consistent with agent-based modeling, or does that get us too close to controlling outcomes instead of influencing outcomes?

Richard: There are areas where I think artificial intelligence and its use of extensive data can help with complexity. But keep in mind, it’s using algorithmic methods, it is using simulations. It’s not trying to set an equation to solve, it’s trying to see how things have been done many, many times, and get rules based on that. I don’t know quite how far it can go. Agent-based modeling can be looked as an artificial intelligence model. I think it misses some of the essence of an agent-based model to do that. For one thing, it has a very clear structure and identifies the agents in the system, whereas a typical artificial intelligence approach is basically a black box. It uses neural networks and other methods where you throw in enough data and it kind of tries to figure it out. But I think that being armed with a lot of data, and being armed with simulation methods, can get you further along the path. I don’t if it can work with crises. In fact, I don’t think it can work with crises because, as I said, each crisis is different. You don’t really have big data when you deal with crises. It’s not like you have 10,000 different crises from the past, all of which are from the same world that you have right now. But there are cases where I think artificial intelligence can deal with the essence of a world that has dynamical complexity, and that can have the issues that arise from it, what we call emergence – different sorts of congestion and so on.

David: Three more questions for you. One is coming back to the topic you lingered on in your first book. Complexity is not actually an accidental attribute of the financial system. As you argue, complexity is manufactured by Wall Street to gain a competitive advantage. How would you advise avoiding a similar outcome to 2008 and 2009 where gains were essentially privatized from the risks that were being taken through complexity created, but the losses ended up being socialized?

Richard: To the example of 2008, one example was that things would create increasingly complex derivatives that, essentially, the user couldn’t figure out, and that only they would be able to intelligently trade, so they could create a monopoly for that market where there is informational asymmetry, where they had the better sense of the instruments than the market did. And there are many other cases of that. This actually gets to another level of complexity. I talked a while ago about how you could march along from a purely mechanical system to a mechanical system that has noise in it, to a system that has the sort of dynamical complexity that occurs because of interactions, to the reflexive system that Soros discussed where those interactions change the environment, and change people’s expectations. But there is one level even beyond that, what I call strategic complexity, and that is what you are getting out here, that people can actually add complexity to the market that wouldn’t normally be there because it gives them an advantage because they are changing the environment in ways that could not be anticipated, but that they have an advantage with.

The best example for this is what happens in warfare. The whole point of a lot of theory of warfare is to confuse the adversary, to change the environment in ways that you can control and understand, but the adversary can’t. There is a very famous war strategist, John Boyd, who was a great fighter pilot in 1940s, and he essentially could beat anybody in a dogfight because he understood a principle of how you can move your aircraft in a way that changes his expectations and changes his environment and adds complexity which he can’t react to as quickly as you can because you induce the complexity in the first place.

In the financial system, we need to structure in a way that tries to minimize that. One way that we have done that is by forcing more standard noncomplex derivative instruments. Another way we can do it, which you mentioned is, make sure there is transparency, so nobody has an informational advantage by knowing prices or knowing the types of instruments where other people don’t.

Another way is to make sure people can’t change the market more quickly than other people can react. High-yield trading really is an avenue for allowing that sort of asymmetry to occur. Nobody should need to trade in microseconds, or milliseconds. Information just doesn’t move that fast, but it gives them an informational advantage because it can add a complexity to the market that other people can’t react to as quickly. So we can sort of scan the marketplace, find places with new strategies that we haven’t even anticipated yet might lead to this type of strategic complexity and if regulation is well formed it should try to minimize that from happening.

David: Does regulations then get involved? You saw Dodd-Frank with part of that Volcker rule. You were part of that, as well. But when you have that aspect of informational asymmetry, do regulators get involved? And at what level? Do they say, “Listen, with the advent of quantum computing we don’t allow that because it creates that informational asymmetry.” Or do you do something more basic, like break up the banks to where you’re basically restructuring the entire landscape?

Richard: Yes, that’s right. I think one thing a regulator can do is, as each new attempt occurs, try to create regulations that stop it from happening. Of course, regulators are pretty slow in acting and by the time they act somebody has thought of a way to game around it. A better way to do it, and another way to do it in a way that you have discussed is, reduce the debt of institutions that really can create those asymmetries. That, and many other things, lead to proposals to break up the big banks. So I wouldn’t say that this is the one motivation for doing it.

The issue with banks, by the way, is not just that they’re big, it’s that they span so many parts of the financial system. Banks help to fund the system. They also trade and create the markets within the system. And they are the body that dictates the collateral of the system. And there are really three waves – I think of three layers – for the financial system. There is the collateral that supports the funding, and the funding supports a lot of the trading, because many people invest based on leverage. So it is as if you have a building with three stories – the collateral story, the funding story, and the third story which is the one that we tend to see which is where the markets and the trading occurs.

Because the banks are in all three of those and act in all three, it is as if we have a shaft that goes all the way from the first floor to the third floor. So if there is a fire on any one floor, through the banks it can spread very quickly to the other floors. So I think the way to break up the banks isn’t just to create a whole bunch of smaller banks. It is to close off that shaft. Have some banks that can really focus on funding, other banks that really focus on market making. You know, you can have some things that are retail, some that are commercial, some that are much more into trading. But when you have banks that really could do all of those functions, a problem in one area can propagate much more easily and much more quickly to another area.

David: So, in the end, the agent-based approach is there to dampen the effects of crisis. Where does it need to be adopted to have the greatest positive impact on the human consequences of crisis? Are we talking policy circles, academic institutions, corporations, financial institutions, themselves? Who needs to adopt it to improve or mute the negative impact, the human consequence of crisis?

Richard: The most natural place, the place that people would think of first is with the central banks and regulators. The problem there, as I have mentioned, is first of all, it would be slow to adopt it because they very much have bought into a neoclassical approach. And even if they did, they would be very slow to act.

The second place people would think is in the institutions that they know – the banks, maybe the hedge funds. The problem is, we would like them to be the solution to the problem, but they probably almost always will be the problem, because they are the ones who have leverage. They are the ones who dictate the amounts of liquidity in the market. So, you’ve sort of crossed off the list the usual suspects. You’ve crossed off the list the institutions we tend to think about. What’s left? It is what are called the asset owners – pension funds and what I’ve talked about, sovereign wealth funds, which essentially are very big pension funds that are run by the countries, themselves. The reason I focus on them is that, first of all, some of them are very, very big. A lot of them are in the hundreds of billions of dollars of assets.

The other reason is that they don’t lever, and because they’re not levered they can’t be forced out of a market through margin calls.

And a third reason is that they have a very long-term horizon. The problem with a hedge fund is not only that it is leveraged, or the problem with a bank is not only that it is leveraged, it is that they have to account to their investors either every quarter or every month. So they have to take action, and they can’t say, “Hey, we invested, we put money into the market during this crisis. It’s down another 20%, but we have confidence that ultimately, in another two months or five months, things will revert.” They can’t do that, whereas a pension fund or sovereign wealth fund can do it. They may not have the will to do it, and that’s another issue.

But they do have a long horizon, much like, actually, the Federal Reserve, in this case. They can hit on a problem longer than the problem exists. If it is a liquidity and leverage-induced problem. So that is really where I would be focusing. And you know the thing that is interesting is, I got my Ph.D. in Economics. No place in my course work, and probably no place in the course work of 90% of people who are getting Ph.D.s, is the determined asset-owner or pension fund even mentioned.

David: Well, we’ve lingered on the theoretical for most of our conversation today. I wonder at a more existential level what worries you today? When you look at the financial landscape and you know that there are issues endemic in China. Maybe it’s the shadow banking concerns. You know that here in the U.S. market we have been in a period of very low volatility. There are a variety of factors that you could say, “Gosh, this just isn’t healthy.” What are the things that you as an individual investor look at and say, “This is not good?”

Richard: Just to make clear, I don’t really look at things as an individual investor. My focus is really on the institutions that I’m working for. I’m actually a terrible individual investor because I just don’t focus there (laughs). But the two issues that I am concerned about are, number one, the shadow banking in China. There is a lot of overhang of leverage. There are a lot of credit issues that aren’t fully understood and aren’t easily managed because they are not within the center of the large banks, that can lead to credit issues which can then spread.

Now, where does it spread? If it is really severe it can spread to our economy because we sell a lot to China. We get raw parts and products from China. The most immediate effect would be in the rest of Asia in the emerging markets; you could think of as Asia. And we also sell a lot there. And so they have problems and have less ability to consume, or they see greater risk and they don’t want to hold their money in risky assets. That will affect us, although I think the effects for us will be pretty muted compared to what would be going on in China, itself, and in Asia. But in any case, I’m not the only one – this is a pretty big topic right now, to look at China.

The other area where there maybe is not that much focus, although I think among a number of professional investors where is starting to be focus, is on the fact that we are in a very low volatility environment right now. What that means is, you look at, say, the equity markets, the stock markets. Prices just don’t bounce around as much as they used to. And that makes us think that there is a lot of stability, and because of that stability we are willing to borrow more money and lever more, we start to institute strategies that depend on lower volatility. If basically, things are moving around half as much, you’re willing to borrow twice as much. And banks are willing to lend you twice as much if they think this is the way the world will be.

But volatility tends to be mean-reverting. That is, there tends to be kind of a level of volatility that, long-term, we tend to get back to. So, there almost certainly is going to be a time, and it might be in the next month, it might be a year or two from now, where volatility increases. And when it increases the lenders, the banks, will say, “You know, now risk is high, we’re not willing to give you as much funding.” People who say, we told our investors we would keep our volatility to a certain level, now we have to sell assets because the assets are getting more risky.

And in both of those cases, more assets are being sold into the market, and that can depress prices, and that, in turn, can increase volatility even more. So I think looking at volatility is an area, because it has a lot of secondary effects that really can trigger a lot of these dynamics that I allude to when I talk about agent-based modeling. And there are strategies, and there are instruments that are really tailor-made with the assumption of a low volatility market.

David: Richard, I look forward to our next conversation. I hope it is, as it was many years ago over a Korean Barbecue somewhere in mid-town Manhattan. I really enjoyed our conversation today, and look forward to keeping in touch on these issues. These are so vital to understanding the direction of where we go from here, how we manage crisis, how we understand crisis, and the decisions that we need to make in light of the realities that exist. So thanks for joining us.

Richard: Thank you.