How have financial models failed us in the past? And are they leading us towards yet another financial crisis? These are two major questions James Rickards, best-selling author of Currency Wars will address at next month’s Global Investment Conference in San Diego (April 6 to 8). His keynote discussion will look at imbalances building up in the global financial system. However, in advance of the conference, we asked James a few questions about the problems with financial models central bankers and others use and how other techniques might provide a better predictive capacity To find out more about James’s presentation and the Global Investment Conference, please click here.
What’s the problem with today’s financial models?
I’m a critic of the models central banks are using. Those are models of so-called modern financial theory which really developed beginning in the 1950s and so they got a lot of acceleration in the 1970s and became the dominant way of thinking about risk and capital markets. They have fancy names – dynamic stochastic general equilibrium is one. They also involve a set of assumptions. The assumptions are, number one, efficient markets, namely that markets instantaneously and continuously incorporate new information, price accordingly and so there’s no way to “beat the market.” Second assumption is that risk is normally distributed, meaning, if you look at the time series of prices and you have up days and down days and gains and losses and you put them in a degree distribution, that degree distribution will look like a bell curve. And the third thing is that as people are rational wealth maximizers: they do things that will maximize their wealth and they avoid doing things that would diminish their wealth. So take these three things: efficient markets in terms of incorporating information, normally distributed risk and rational behaviour. All three of them are empirically wrong.
If the current models are wrong, where do you go?
So then the question is well, what ways of thinking about risk and capital markets do work and do correspond to what we actually observe in the way people actually behave? What I have discovered is that there are three very powerful tools and I use them in my analysis.
The first one is behavioural psychology. It has given us enormous insights in recent decades that people do very irrational things, or at least not rational as economists define it. It might be rational for an ice-age Neanderthal, but we’re not that far evolved from the ice age and the Neanderthal even though we think we are. The second one is complexity theory. It’s new stuff and has close cousins in chaos theory, network science, and graph theory. And then the third — it has a couple of names, just to make it confusing – but it’s called inverse probability. It’s also called casual inference and it’s probably better known under the heading of Bayes’ Theorem. Bayes’ Theorem is something you use when you don’t have enough information. If you’ve got lots and lots and lots of information, you can use normal statistical methods, regressions and correlations and covariance matrices to tease out your inferences or your forecasts.
So my toolkit is behavioural psychology, complexity theory and inverse probability. They’re all different but they complement each other nicely as they fill in each other’s blanks. If you’re using Bayes’ Theorem to solve a problem, you can begin to populate the numerator of the right side of the equation using insights from behavioural psychology because again it gives you ideas about how and why people are behaving the way they do and then when you have a hypothesis that something might be going wrong – that would be a Bayes’ Theorem-type prior – you might flip over to complexity theory and say, oh by the way the whole system is at risk so if this one thing goes wrong then it’s going to take down the whole house of cards.
You say that, in part thanks to the dominance of the old models, we’re going to get another financial crisis.
How does J.P. Morgan account for risk in their derivatives books? Well they use a model called Value at Risk, which embeds the assumptions I was talking about. They’ve got well over $1 trillion in notional value derivatives. So you say, “Oh my goodness, that seems like a lot of risk.” But the bank will say, “Don’t worry because it’s long-short, long-short long-short, it all nets out,” and Value at Risk is taking the net exposure and then adjusting that on the assumption that extreme events happen like in the normal distribution, which is a false assumption, but that’s what they do. And then they say, to a 98% certainty, we will not lose more than x dollars every two years. But complexity theory would tell us that we should think that the risk is in the gross, not the net and that actually, when you look at power curve, risk is an exponential function of scale.
Now to prove that, just look at AIG. AIG, on a net basis, was fine until the panic and then when AIG was bailed out in October 2008, nobody cared about the net exposure, nobody said it was all good because of the net, they cared about the gross exposure. If you were a counterparty of AIG and AIG was on the verge of bankruptcy, who cares what the net exposure on AIG’s books was? You were looking at a gross exposure vis a vis them. If they went under, you would lose money on all the gross value on all the trades you had with them. Everyone’s going down this path of thinking about net exposure and Value at Risk when in fact they’re lighting matches in the dynamite shed.
Would a derivatives clearinghouse help?
There are two problems. Number one, only some of the most plain-vanilla swaps have been moved onto derivatives exchanges, but it’s not nearly enough. In other words, the more exotic the derivative, the less likely it is to be on an exchange and the dealers want it that way because they make very good spreads on these trades. So is it a good thing? Yes. Is it sufficient? No. Number two, there’s an extent to which, even when you do that, all you do is replace counterparty risk with clearinghouse risk. What happens if individual members of the clearinghouse themselves fail? Now we get into layers of too big to fail.
The real way to do it is to break up the banks and ban most of the derivatives. Of course, once you ban it, you have no risk at all. By breaking up the big banks, it’s not as if any of the smaller banks can’t fail – they can, but it doesn’t matter. In other words, bank failures will always be part of the landscape, but you want the bank itself to be small enough so that when it fails it doesn’t take the system down with it..
To learn more about the Global Investment Conference, please visit the conferences section of the CIR website. If you are interested in attending this event, please email Alison Webb to be considered, as limited space available.