Foreword

Financial markets are all about risk management. Banking and capital markets activities throw up all manner of risk exposures as a matter of course, and these need to be managed accordingly such that stakeholders are comfortable. “Market risk” traditionally referred to risks arising from a change in market factors, and when we say “risk” we mean risk to the profit and loss account or to revenues. These market factors might be interest rates, foreign currency rates, customer default rates, and so on. Managers of a financial institution should expect to have some idea of the extent of their risk to these dynamic factors at any one time, so that they can undertake management action to mitigate or minimize the risk exposure. This is Finance 101 and is as old as commerce and money itself.

Measuring market exposure has always been a combination of certain methods that might be called scientific and others that might be described as application of learned judgment. I have always been a fan of “modified duration” for interest rate risk and I still recommend it. Of course it has its flaws, which estimation method doesn’t? But when Value-at-Risk (VaR) was first presented to the world it appeared to promise to make the risk manager’s job easier, because it seemed to offer a more accurate estimate of risk exposure at any time. And the latter was all it ever was, or claimed to be: an estimation of risk exposure. A measure of risk, no better and no worse than the competence of the person who was making use of the calculated numbers.

Unfortunately, in some quarters VaR was viewed as being somehow a substitute for “risk management” itself. It didn’t help that the assumptions underpinning every single methodology for calculating VaR were never widely understood, at least not at the senior executive level, which made explaining losses that exceeded the VaR estimate even more difficult than usual. In 2012 JPMorgan announced losses of up to $9 billion in a portfolio of corporate credits that were managed by its London-based chief investment office. Depending on which media report one follows, the VaR number reported for the bank as a whole the day before the announcement was alleged to be between 1 percent and 10 percent of this value. Is there any point in going to the trouble of calculating this estimate if at any time it can be demonstrated to be so completely off the mark?

The short answer is yes and no. VaR is a tool, nothing more nor less, and like all tools must be used within its limitations. One could argue that a bank the size and complexity of JPMorgan is going to struggle to ever get a meaningful estimate of its true risk exposure under all marker conditions, but therein lies the value and the worthlessness of any statistical measure like VaR: it is reasonable for some, indeed most, of the time but when it does get out of kilter with market movements the difference could be huge (an order of magnitude of up to 100 times out, if some recent headlines are to be believed). It reminds one of the apocryphal story of the statistician who drowned in a lake that had an “average” depth of six inches.

The circle is complete of course. It was JPMorgan that gave the world VaR back in 1994 (one or two other banks, including CSFB, were applying a similar sort of methodology around the same time), and eighteen years later the bank saw for itself just how inaccurate it could be. Does that mean we should dispense with VaR and go back to what we had before, or look to devise some other method?

Again, yes and no. The key accompanying text for any statistical measurement, VaR most definitely included, has always been “use with care, and only within limitations.” That means, by all means, continue with your chosen VaR methodology for now, but perhaps be aware that an actual loss under conditions that the model is not picking up could well be many times beyond your VaR estimate. In other words, bring in your interest rate risk and credit risk exposure limits because the true picture is going to be in excess of what you think it is. That is true for whichever firm one is working at.

But that isn’t all. Knowing VaR’s limitations means also seeking to develop an understanding of what it doesn’t cover. And this is where Max Wong’s very worthwhile and interesting book comes in. In the Basel III era of “macroprudential regulation,” Mr Wong applies a similar logic for VaR and presents a new concept of Bubble VaR, which is countercyclical in approach and would be pertinent to a bank running complex exposures across multiple markets and asset classes. But I also rate highly the first half of the book, which gives an accessible description of the vanilla VaR concept and its variations before launching into its limitations and how Bubble VaR is a means of extending the concept’s usefulness. The content herein is technical and arcane by necessity, but remains firmly in the domain of “know your risk,” which is something every senior banker should be obsessed with.

This book is a fantastic addition to the finance literature, written by that rare beast in financial markets, management consulting, or academia: a person delivering something of practical value for the practitioner and that advances our understanding and appreciation of finance as a discipline. Finance 201, so to speak, for everyone with an interest in financial risk management.

Professor Moorad Choudhry

Department of Mathematical Sciences

Brunel University

16th December 2012