This book brings together a number of very accessible and well written articles which together represent a comprehensive explanation of the ideas and methods which under-pin stress testing in many modern financial institutions. However, an actuary reading this book – or even scanning the contents page – will no doubt feel the same sense of frustration as me. As is all too common, “financial institution” and “bank” seem to be synonymous to the editor of this text, and as such the book almost exclusively explores methods for stress testing retail and wholesale credit portfolios. Unfortunately, Stress Testing for Financial Institutions is therefore unlikely to be essential reading for an actuary working in the traditional fields of Insurance or Pensions.
I say unfortunate, as many of the methods explored so succinctly will be familiar to the readers of this journal, as will many of the problems they are used to address – dependency structures, lack of relevant data or evolving regulatory requirements, for example. These familiar subjects are investigated from different angles to those taken in actuarial literature; as a result there is a lot of food for thought here and much we might learn from our banking equivalents. Indeed, as insurers move more and more in to the realms of making commercial loans on their own behalf, we may have more of a need for these alternative views.
The book is broken down into five sections:
1. Stress Testing Frameworks
2. Stress Testing for Corporate Credit Risk
3. Stress Testing for Retail Credit Risk
4. Stress Testing for Economic Capital
5. Stress Testing for Regulatory Capital
Each section contains 3 to 5 self-contained articles, written by experts in their fields. The book is written for readers familiar with the subjects, but contains enough narrative that an un-initiated reader with a reasonable understanding of financial maths could follow the arguments being made. However, the level of assumed knowledge does differ from chapter to chapter and it is a little frustrating to find some pivotal concepts explained only in the second or third chapter in which they arise.
Section 1, being the most generic, does carry some useful titbits for the reader who is not concerned with the specifics of credit risk. The discussion of how appropriate different risk measures are for various purposes is as clear as any I have seen elsewhere, whilst the cautionary notes on stress testing show that the book is a well balanced and honest appraisal of the state of things as they are, as much as a treatise on how things should be. For example, the idea that stress testing has a tendency to “formalise what might have been done by expert judgement anyway” and to give it an often undeserved veneer of certainty, is a problem that I have encountered in the work place, and is a message that bears repeating.
Section 4 offers the only chapter that is likely to be directly relevant to actuaries, on “Risk aggregation, dependence structure and diversification benefit”. This offers a very good introduction to a subject that many people are grappling with as insurance companies work towards Solvency 2 compliant Internal Models and as ST9 is gaining popularity. The chapter starts with simple definitions of different types of correlation measure and works up to fitting some of the more complicated families of copula. Diagrams are used to good effect to give a clear explanation of how these infamous mathematical structures work, and in showing how best to approach their use. The final part of this chapter explores the problems with fitting such models, and compares them to alternate possible methods.
The fact that this book touches on the practical difficulties of the methods it describes as well as their theoretical structures is certainly valuable, particularly to the more casual reader. The style employed is well structured and accessible, bringing clarity to complex ideas. At times, it is surprisingly entertaining – for example, the authors of one article discuss the problems associated with modelling extreme “tail” events by recalling a risk manager who met this problem by saying “I'm interested in the quantile where I get fired, but not beyond”.
This flippant quote largely sums up the reason that many readers of this journal will not go on to read this book; it contains little information that is likely to stop them getting fired, or help them in their day jobs in any other way. However, for those working in banks managing credit risk, this would be a good addition to the personal reference library. It would also be an excellent purchase for those looking for an articulate technical introduction to credit risk modelling and stress testing.
