Preface
This is a story of the illusion of risk measurement. Financial risk management is in a state of confusion. The 2008 credit crisis has wreaked havoc on the Basel pillars of supervision by highlighting all the cracks in the current regulatory framework that had allowed the credit crisis to fester, and ultimately leading to the greatest crisis since the Great Depression. Policy responses were swift—UK’s Financial Services Authority (FSA) published the Turner Review, which calls for a revamp of many aspects of banking regulation, and the Bank of International Settlements (BIS) speedily passed a Revision to its Basel II, while the Obama administration called for a reregulation of the financial industry reversing the Greenspan legacy of deregulation. These initiatives eventually evolved into the Basel III framework and Dodd-Frank Act respectively.
The value-at-risk risk measure, VaR, a central ideology for risk management, was found to be wholly inadequate during the crisis. Critically, this riskometer is used as the basis for regulatory capital—the safety buffer money set aside by banks to protect against financial calamities. The foundation of risk measurement is now questionable.
The first half of this book develops the VaR riskometer with emphasis on its traditionally known weaknesses, and talks about current advances in risk research. The underlying theme throughout the book is that VaR is a faulty device during turbulent times, and by its mathematical sophistication it misled risk controllers into an illusion of safety. The author traces the fundamental flaw of VaR to its statistical assumptions—of normality, i.i.d., and stationarity—the Gang of Three.
These primitive assumptions are very pervasive in the frequentist statistics philosophy where probability is viewed as an objective notion and can be measured by sampling. A different school of thought, the Bayesian school, argues for subjective probability and has developed an entire mathematical framework to incorporate the observer’s opinion into the measurement (but this is subject matter for another publication). We argue that the frequentist’s strict mathematical sense often acts as a blinder that restricts the way we view and model the real world. In particular, two “newly” uncovered market phenomena—extremistan and procyclicality—cannot be engaged using the frequentist mindset. There were already a few other well-known market anomalies that tripped the VaR riskometer during the 2008 crisis. All these will be detailed later.
In Part Four of the book, the author proposes a new risk metric called bubble VaR (buVaR), which does not invoke any of the said assumptions. BuVaR is not really a precise measurement of risk; in fact, it presumes that extreme loss events are unknowable (extremistan) and moves on to the more pressing problem—how do we build an effective buffer for regulatory capital that is countercyclical, and that safeguards against extreme events.
This book is an appeal (as is this preface) to the reader to consider a new paradigm of viewing risk—that one need not measure risk (with precision) to protect against it. By being obsessively focused on measuring risk, the risk controller may be fooled by the many pitfalls of statistics and randomness. This could lead to a false sense of security and control over events that are highly unpredictable. It is ultimately a call for good judgment and pragmatism.
Since this book was first published in 2011, the financial industry has experienced a sea change in Basel regulation and new risk modeling requirements under the Basel III capital framework. There are also exciting developments in the modeling of risk at the research frontier. This revised edition is an update to include some of these topics, even though the primary objective remains to encourage an alternate paradigm of looking at market risk.
This book is intended to reach out to the top management of banks (CEOs and CROs), to regulators, to policy makers, and to risk practitioners—not all of whom may be as quantitatively inclined as the specialized risk professional. But they are the very influencers of the coming financial reregulation drama. We are living in epic times, and ideas help shape the world for better (or for worse). It is hoped that the ideas in this book can open up new and constructive research into countercyclical measures of risk.
With this target audience in mind, this book is written in plain English with as few Greek letters as possible; the focus is on concepts (and illustrations) rather than mathematics. Because it is narrowly focused on the topic, it can be self-contained. No prior knowledge of risk management is required; preuniversity-level algebra and some basic financial product knowledge are assumed.
In order to internalize the idea of risk, this book takes the reader through the developmental path of VaR starting from its mathematical foundation to its advanced forms. In this journey, fault lines and weaknesses of this methodology are uncovered and discussed. This will set the stage for the new approach, buVaR.
Chapter 2 goes into the foundational mathematics of VaR with emphasis on intuition and concepts rather than mathematical rigor.
Chapter 3 introduces the basic building blocks used in VaR. The conventional VaR systems are then formalized in Chapter 4. At the end of the chapter, readers will be able to calculate VaR on a simple spreadsheet and experiment with the various nuances of VaR.
Chapter 5 discusses some advanced VaR models developed in academia in the last decade. They are interesting and promising, and are selected to give the reader a flavor of current risk research.
Chapter 6 deals with the tools used by banks for VaR reporting. It also contains a prelude to the Basel Rules used to compute minimum capital.
Chapter 7 explores the phenomenology of risks. In particular, it details the inherent weaknesses of VaR and the dangers of extreme risks not captured by VaR.
Chapter 8 covers the statistical tests used to measure the goodness of a VaR model.
Chapter 9 discusses the weaknesses of VaR, which are not of a theoretical nature. These are practical problems commonly encountered in VaR implementation.
Since this book deals primarily with market risk, Chapter 10 is a minor digression devoted to other (nonmarket) risk classes. A broad understanding is necessary for the reader to appreciate the academic quest (and the industry’s ambition) for a unified risk framework where all risks are modeled under one umbrella.
Chapter 11 gives a brief history of the Basel capital framework. It then proceeds to summarize the key regulatory reforms (Basel III) that were introduced from 2009 to 2010.
Chapter 12 discusses developments in measuring and detecting systemic risks. These are recent research initiatives by regulators who are concerned about global crisis contagion. Network models are introduced with as little math as possible. The aim is to give the reader a foretaste of this important direction of development.
The final part of this book, Part Four—spanning five chapters in total—introduces various topics of bubble-VaR. Chapter 13 lays the conceptual framework for buVaR, formalized for market risk.
Chapter 14 shows that with a slight modification, the buVaR idea can be expanded to cover credit risks, including default risk.
Chapter 15 contains the results of various empirical tests of the effectiveness of buVaR.
Chapter 16 is a concluding chapter that covers miscellaneous topics for buVaR. In particular, it summarizes how buVaR is able to meet the ideals proposed by the Turner Review.
Lastly, Chapter 17 lists suggestions for future research. It is a wish list for buVaR which is beyond the scope of this volume.
Throughout this book, ideas are also formulated in the syntax of Excel functions so that the reader can easily implement examples in a spreadsheet. Exercises with important case studies and examples are included as Excel spreadsheets at the end of each chapter and can be downloaded from the companion website: www.wiley.com/go/bubblevalueatrisk.
Excel is an excellent learning platform for the risk apprentice. Monte Carlo simulations are used frequently to illustrate and experiment with key ideas, and, where unavoidable, VBA functions are used. The codes are written with pedagogy (not efficiency) in mind.