Remember that feeling of bewilderment after your first few weeks in your first job after university? That wrenching realisation that, while the theories that you had laboured to understand may have been illuminating, they were too abstract to be applied to the real world? Reading Riccardo Rebonato’s intriguing book brings those memories flooding back. For while Rebonato well understands, approves of, and writes about quantita-tive probability and risk theory, his day job involves actually managing financial risk. Hence he appreciates the limits both of theory and of applying it to real world situations.

Rebonato had a lot that he wanted to get off his chest in this book, and certainly it tumbles out. The book is thereby not an easy read: form to an extent follows content. The serious reader has no alternative but to plough through it, line by line, chapter by chapter. Nevertheless, a persuasive thread of argument runs through it, along the following lines.

There is today a basic problem today with the management of financial risk: too much attention is paid to measuring it; yet too little to how to reach decisions based on it. The widely-used value-at-risk (VAR) frame-work in particular is employed too uncritically.

Moreover, quantification is often spuriously precise. The properties of the tails of the distributions with which the management of financial risk is perforce concerned tend to be inferred from distributional forms that derive their apparent persuasiveness from the tightness with which they fit not the tails, but rather the main body, of the curve. Risk managers in turn tend to infer rather more than they should from these (assumed) tail properties: in many cases, the most that can legitimately be said about such probabilities is that they are nonzero.

Hence the world needs a better mousetrap. Rebonato asserts that, ultimately, managing financial risk is about making decisions under uncertainty, and that better techniques are available. He favours a more subjective approach, involving probability, experimental psychology, and decision theory, as employed in a number of the physical and social sciences, notably ‘Probabilities-as-degree-of-belief’ and ‘Probabilities-as revealed-by-actions’. These tools are less precise, but that is a virtue: like Keynes, Rebonato would prefer to be approximately right than precisely wrong.

At least two important consequences follow from all this.

First, there is a disconnect between young recently-trained ‘quants’, who are adept at solving well-defined, technical problems (Rebonato dubs them ‘ignorant experts’), but may not understand the full dimensions of the issue, and their managers. They in turn understand the questions that need answering but do not understand their quants’ techniques, and so have to take on faith that the quantitative approach chosen will not only be technically correct, but also ‘makes sense’ for the problem at hand.

Second, it has been difficult for the regulators, looking in from the outside, to appreciate this degree of disconnect.

There is considerably more meat in this wise, practical, yet unpretentious book than can be summarised in a short review. And it is also contains a number of memorable one-liners: particularly apposite in the current environment is the observation that: “It is much easier to pass risk around than to make it disappear.”

John Llewellyn

Lehman Brothers