Wednesday, May 30, 2012

The Metaphor of the Bridge

We based our last post on the observation that a good metaphor is worth many equations in problem solving. Today we come to one metaphor that I find particularly useful. But before we do, let us ask: what is the best way to conceptualize risk forecasting? To many it appears as no more no less than clairvoyance, peeking into the future with the help of incomprehensible Greek symbols, fighting "Against the gods" as Peter Bernstein put it. Alas, the prophets of market risk are not very good and the critics are only happy to rub it in asking the efficient-markets guru Mr. Bernstein "Against the gods? And who won?".
And therein lies one of the bigger problems of risk modeling. It is thought of as something that looks into the future and sees the possibilities, through some magic maths; it is not by accident that "Against the gods" is filled with stories of Blaise Pascal and Gerolmo Cardano and other combinatorial wizzes (and no mean gamblers or at least correspondents of gamblers; Blaise Pascal as the first risk manager and Chevalier De Mere as an inquisitive portfolio manager, so to speak) crunching the numbers. After this setup it is only too easy for Dr. Naseem Taleb to come along and point out that possibility set is infinite results are catastrophic and human mind is quite and quite finite, so that the pretense of a number crunching wizz is drowned in endless hyperspace. The Icarus flies too high and burns his wings again and again.
Let us put an end to that nonsense. Risk forecasting is not really about reaching into the future. It is about looking at the soundness of the structure that is of your portfolio or of an asset class or of a market as a whole. I would call it financial engineering had not that word been so discredited with the witch soup of CDO^3s and so on. Wait, aren't we, as risk managers, supposed to think of all of the possible events that could occur: wars, earthquakes, flash crashes and what the Fed will do next? Not as risk managers. Some time ago we have introduced a metaphor for stress testing: stress testing a portfolio is like crash testing a car ( We are not concerned with all of the possible (infinite number of) triggers that might cause a loss, but rather with the vulnerability of the car (portfolio). When a car is crash tested, testers do not specify whether it is a tree or a pole that is being hit or if the wall is painted green or black, it is only the vulnerability of the vehicle that concerns them. All events are classified into only a few classes: frontal impact, side impact etc.. So it is with portfolio risk. 
But what about the bridge, didn't we promise a bridge? Yes, we did, and we will deliver two of them. The first very useful bridge for the mind of a risk manager comes from a fertile mind of Jon Danielsson (RiXtrema Senior Advisor)  and you can hear more aboout it on The bridge is a structure that responds to those crossing it. So, people crossing a bridge and the bridge itself form a system which generates its own measurable endogenous risk and people reacting to the bridge causes the bridge to react (hint: a deleveraging cycle at work). 
But let us take this a bit further. We already established that risk forecasting is not about considering vast set of possibilities (a finite number however large is really no closer to infinity than a zero) and assigning each of them a probability. Risk management is about looking at the soundness of a structure. If an engineer examines a bridge, some classes of dangerous possibilities can be thought of (a.k.a. stress testing) and simulated. But what if we are just looking to estimate the riskiness of the bridge (er, portfolio)? Then we need to assess the soundness of the structure without specifying any shocks.

 For those of you familiar with the RiXtrema approach for early warning systems in asset allocation, you may recall that we use two categories for assessing the soundness of the financial system: risk mispricing and loss of conviction (see the New Paradigm of Risk Management video by me on Risk mispricing is the excessive exuberance in taking risks (like when a country experiences massive short term capital inflows you can take your pick in the Eurozone for examples). This first condition is like a bridge being overloaded, carrying more load than it has ever carried historically. However, risk mispricing can go on for a long long time (bubble is the most profitable stage of the cycle), so we need a second factor, which we call loss of conviction. This looks like little cracks appearing somewhere in the structure. In the absence of the preceding mispricing, cracks could still be attended to and mended. However, if risk was mispricing (the bridge overloaded), then look out. Whenever you hear someone talking of 'soft landing', it likely means we have reached the stage or are at least very near to the loss of conviction . George Soros (one of the pre-eminent risk managers, despite not carrying that title) imaginatively called it the 'twilight zone of the bubble', when people play the game, but you can see that they no longer believe it. So, risk is high when 
a. Bridge has been persistently overloaded (risk mispricing)
b. Bridge is showing some structural abnormalities (loss of conviction)
When that happens there is no longer a possibility of a 'soft landing'. This is why our models showed persistently high risk for Greece/Portugal through every bailout along the way; why they showed elevated risk for Eurozone (but not US) corporate debt and equity in the summer of 2011 and this is why they are showing high risk for Austrian bonds and Irish bonds (again). The precise timing is never known, but the pattern is unmistakable. It repeated for centuries. And the metaphor of the bridge is only one of the ways we can visualize that.

No comments: