Elise Payzan-Le Nestour is an associate professor in the school of banking and finance at UNSW Business School. She is researching the way people perceive and react to financial risks, using a combination of theoretical and experimental methods. Payzan-Le Nestour spoke with Julian Lorkin for BusinessThink.
An edited transcript of the conversation follows.
BusinessThink: What is tail risk?
Payzan-Le Nestour: Payoffs that feature tail risk are payoffs that are not normal in the very precise sense that the chance to get an extreme outcome – with so-called black swans – is much higher, much bigger than with a normal distribution.
The question in our study was a very simple question: can people learn about tail risk? That is, can they learn, when the given distribution is not normally distributed, what the feature of a tail risk is.
BusinessThink: And how about black swans?
Payzan-Le Nestour: Black swans are extreme events. Tail risk refers to payoff distributions that have a probability of a black swan that is much bigger than when the payoffs are normally distributed.
BusinessThink: You've studied tail risk and these unexpected events. What have you found?
Payzan-Le Nestour: The premise of the study – my guess – was that probably people are not going to be very good at learning about black swans because in our natural environments there is no tail risk. Our ancestors didn't have to cope with tail risk, so our brain did not evolve to cope with tail risk. My guess was that probably people are pretty bad at learning about black swans.
And what I found is kind of the opposite. I found that under certain conditions people are actually quite good at learning whether and when a given payoff distribution is fat tailed – that is, it presents a high risk of black swans compared with when it's normally distributed.
'What I wanted to study was the capability, or absence of perhaps the ability, of people to implement statistical learning'
– ELISE PAYZAN-LE NESTOUR
BusinessThink: That's reassuring. And you've tested it with academic research and by getting a bowman to shoot arrows against a wall?
Payzan-Le Nestour: I designed an experiment that relied on a game, the Bowman game, which is exactly like a video game. And the reason for that is twofold. The first is that it is very important to use a non-financial task, which is a paradox, because to answer a financial question I designed a completely non-financial task. The reason for that is very simple. It was very important to control for financial illiteracy.
What I wanted to study was the capability, or absence of perhaps the ability, of people to implement statistical learning. By learning about payoffs that are fat-tailed, this was normal distribution which is a statistical learning program. And I wanted to isolate this particular statistical task.
Imagine, for example, if the majority of my subjects – that's participants – screw up in the task. Imagine that I design a financial task, then it will be unclear whether the [the participants screwed up] because of financial illiteracy, or because they didn't understand basic concepts, or perhaps because they couldn't learn, they couldn't implement the statistical learning task that we wanted to isolate in the study. That would be unclear.
It's very important to rule out these kind of confounds and so if you use a completely non-financial game like Bowman, in a sense that does the trick.
BusinessThink: And how does this actually test the rare events of black swans and tail risks, by having that bowman shooting against a wall?
Payzan-Le Nestour: The bowman shoots on a target on a wall. Basically, you have two kinds of bowmen. The master bowman kind is a bowman with room. Almost all the shots are near a target. That's the first kind. And the second kind is the novice or apprentice bowman. And with this kind, the shots can be basically anywhere in the wall; very unexpected.
Just to map that into the financial setting, then the master bowman is a metaphor for the normally distributed assets. And with the novice bowman, the apprentice bowman, the shots can be basically everywhere, they're all very unpredictable – so meaning a black swan. It's like much more often than they should; that's the metaphor for the assets, that feature, for the assets with which black swans are not that rare. Black swans are rare events. But when the tail risk is high, what it is saying is that the rare events, the rare disasters, are not that rare.
The apprentice bowman, that's to capture tail risk. The master is for normally distributed assets. And the idea's simply to see where people can learn when a given bowman is a master versus an apprentice. Obviously I designed the game in such a way that people have incentives to learn about that, because if they do learn about that they will earn a lot of money. And if they fail, that is we can tell from seeing the shots because we see a number of shots, and from that if they are good, intuitive statisticians they are supposed to infer whether the bowman is master versus apprentice.
So, people are good intuitive statisticians, they can do it and the game is designed in such a way that if they can do it, they're going to earn a lot of money. But if they are not good intuitive statisticians, then they cannot really tell from seeing 20 shots on the wall, they cannot tell whether the bowman is a master or apprentice. If they fail, their earnings in the game are going to be very bad.
And basically what I found is that the majority of the participants did play very well. They could tell after a few shots whether the bowman was an apprentice versus master, basically.
'And basically what I found is that the majority of the participants did play very well'
– ELISE PAYZAN-LE NESTOUR
BusinessThink: So, many people are actually fairly good at working out which is a good bowman. What does that teach us about people's ability to work out probability and risk?
Payzan-Le Nestour: We need to be a little bit cautious because concluding that people – finance practitioners – should learn in a sophisticated manner since the students participants did it, would be a little bit wrongheaded.
But we have to be careful. In the field of finance, things are different because what I call the stochastic structure is not well understood by many finance practitioners. If you do a test and you ask them, "What does tail risk mean?" I'm pretty sure that probably some of them won't be able to explain intuitively what it means in terms of the odds of getting a black swan, these kinds of things. Which is very problematic because then it largely explains what I would expect, but in the field the reasons are definitely not like what I found in the lab.
The reason that I would expect that finance practitioners, some of them, would not be very good at learning about tail risk is because they are not alerted; they're not warned about the different possibilities in terms of asset distributions. Just as they know what tail risk is about, the odds of a black swan if a given asset category features tail risk, what does that mean concretely?
So long as they are on their guard. If you don't tell them anything, they will probably screw up. And again, think about what I told you at the beginning – this is not surprising at all because the human brain did not evolve to cope with tail risk. It is not surprising at all that by default we don't understand very well what it is about. We are just a little bit naive about it. But that's completely normal. That's exactly what we would expect from an evolutionary point of view.
BusinessThink: So effectively we're saying that if people can see a random but rather bad event coming along, most of the time they're just going to cross their fingers and hope that it won't be happening.
Payzan-Le Nestour: Yes. Let's be a bit more specific on this. And I think again it's an important message because that's something to seize. I would emphasise because for regulators I think here we have big incentives to be not ideological, and I explain to people again the different possibilities and what tail risk is about. Because if we do that, I bet, based on what I found, that people are going to be able to protect themselves against tail risk because we will learn about it.
But if we don't do that, if we don't tell them anything then the outcome will be completely opposite to that. And by the way, this is what I think we have observed in the recent past – people being completely vulnerable to the occurrence of black swans, and when if a black swan eventually occurs it's a disaster. And in my view, that's partly because people just were not aware of black swan risk.
And so again, I think that here the key idea is that it's very important to educate finance practitioners about tail risk so that they can then by themselves learn about when a given asset category features tail risk and be on their guard.