Managing Uncertainty, Mitigating Risk

Tackling the Unknown in Financial Risk Assessment and Decision Making

Nick Firoozye, Fauziah Ariff

• Publisher: Palgrave Macmillan
• Year: 2016
• Print ISBN: 9781349565757, 9781137334534
• eBook ISBN: 9781137334541
• Topic: Business
• Subtopic: Corporate Finance
• Index: Business

Managing Uncertainty, Mitigating Risk proposes that financial risk management broaden its approach, maintaining quantification where possible, but incorporating uncertainty. The author shows that by using broad quantification techniques, and using reason as the guiding principle, practitioners can see a more holistic and complete picture.

1

Introduction

Background

Risk and uncertainty are ever present and many would argue that the world is even more uncertain now. However, it may not be that there is more uncertainty in the world today, but it may seem that way, since markets, geographies and political institutions are so intertwined and interconnected that the occurrence of seemingly insignificant one-offs could cause profound consequences the world over. Banks and financial institutions are now forced to tackle uncertainty with such immediacy and urgency due to the speed and magnitude of these impactful incidents. Moreover, the current state of heightened regulatory uncertainty is particularly relevant and, in many ways, responsible for the increased emphasis on compliance and the growth of corporate defensiveness. In a state of defensiveness, no player wants to introduce major innovations, whether they be in product areas or in risk practices, for fear of a regulatory backlash. Bottom line: financial institutions now have to operate in a more challenging and legally fraught environment than in the past.

Debates on risks and uncertainty have not receded and are still very much at the forefront, even some seven years after the crash of 2008. We did not view 2008 as just another economic cycle of boom and bust, but a year which forced us not only to accept uncertainty as part of the new environment in which we operate but also to incorporate uncertainty into our decision-making. The difference of the 2008 crash from other financial crises is that it affected almost all key financial markets, from sovereign debt, swaps and mortgage markets to credit and equity – and, of course, partly due to policy, the money market and global liquidity. It was truly an existential crisis for the financial world. Unlike the Asian economic crisis, or the Long Term Capital Management (LTCM) / Russian crisis, or even the US savings-and-loan collapse in the 1980s, where the impact of each crisis was largely localised to a specific region or primarily affected a specific sector, this time the crisis hit the largest and most developed economies, the US and Europe, in a wide range of sectors over a prolonged period. The banking and securities industry had created such a translucent web of global interconnectivity that no one anticipated that a few triggers could have the broad effect and deep repercussions on the global economy. Risks and uncertainty are not new in the economy, but the severity, volatility and magnitude of their impact is new.

It has been more than seven years since the fall of Lehman Brothers, but the banking sector has not yet returned to the normalcy of pre-crisis years, and it probably never will. We see how complex and jittery the economy is when seemingly minor unrelated events can still affect the global stock market. We can nowadays see one-off events causing huge unanticipated swings in the stock market, and we see how tried-and-tested economic policies have had to be ditched entirely as they could not produce the results they usually do. In turbulent times, markets are harder to predict, economies are harder to predict, and policies are harder to choose.

Although nothing much can be done about outcomes which were not foreseeable, there can no longer be justifications for being unprepared for reasonably likely, foreseeable events, especially if these rare events or situations cause such extreme consequences. For the most part, a range of potential outcomes can be determined and the events can be prepared for and measured and need not be disregarded when it comes to risk assessments.

In characterising future unknown events, by uncertainty we mean specifically Knightian uncertainty – anticipatable events which cannot be characterised by a single probability. This in contrast to risk, by which we mean Knightian risk. Knightian risk is, by definition measurable, and is the primary subject of today’s financial risk management, for which probability is a sufficient characterisation. Virtually all models and methods in use in financial risk management are based on the premise of probability being sufficient, save for stress testing and scenario analysis, which are generally added to the current framework in a somewhat ad hoc manner, without any theoretical justification.

There has been a recent popular focus on black swans or Donald Rumsfeld’s unknown unknowns, events of which we have absolutely no knowledge and which are completely unforeseen – in fact, unforeseeable. The focus of risk management should be on the incomplete knowledge we have, not just the knowledge we do not have, and whether this incomplete knowledge is something that is actionable. Under most circumstances, it is actionable. As with all states of incomplete knowledge, it is accompanied by uncertainty both within and beyond the realms of probability.

The popular focus on black swans is understandable in light of the financial crisis. Yet, most risks are not black swans – for the most part, proximity brings knowledge, and black swans turn either into perfect storms or uncertain one-offs the closer we get to them. The risk of a eurozone breakup was an unknown unknown in 1970, long before the introduction of the euro. Well after the introduction of the euro, the breakup of the eurozone, however improbable, moved from the unknown unknown to the merely uncertain.

Similarly, the subprime crisis, in spite of the fact that it caught many by surprise, was anything but an unknown unknown. Many hedge funds and speculators and even some investment banks foresaw the possibility of trouble ahead, and many took advantage of it, illustrating that it was anything but a black swan (for instance, John Paulson;¹ Goldman Sachs to varying degrees, depending on the reporter;² Michael Lewis’ The Big Short³ also documents the fact that a large number of speculators benefited from the ongoing crisis). Lehman’s bankruptcy, as remote as it was, was not unfathomable, given that Lehman had such a close brush with the same outcome in the aftermath of the Russian and LTCM crises. Rare events, or sets of events forming rare situations or causal chains, are, by definition, highly improbable and may be to some extent outside the realm of normal statistics or probability. Yet, they are almost never entirely beyond our knowledge or consideration.⁴

Under normal circumstances, the normal rules apply, and probability and the current range of probabilistic quantitative models continue to work reasonably well. When these game-changing events are seemingly imminent, it is clear that the old models can no longer apply.

One could argue that the number of one-off possibilities are endless and it would be impossible to run scenarios for all the possible outcomes. We can nonetheless model a sufficiently large and complex set of uncertain outcomes so as to give some clarity to the possible event chains. Moreover, what is important in assessing investments is not to produce and assess an infinitude of outcomes but to narrow down and examine only the set of reasonable and realistic combinations which are representative enough to allow us to adequately prepare for them and to cover a range of similar situations.

The events leading up to the US Fiscal cliff in 2013 were not unknown unknowns; but instead the generic possibilities were truly foreseeable. Again, the circumstances could not be incorporated into current risk frameworks given the lack of directly relevant data. We argue for a less restrictive perspective on what is deemed to be fact and knowledge, and we challenge the thinking that subjective beliefs play no role in finance. Most industries welcome expert opinions and qualitative data in preparing for possible scenarios, and incorporate these into risk assessments.

Financial risk management is newer in some ways than risk management is in other sectors; however, with its access to massively big data, it has become so ‘data’-dependent as to possibly lose sight of a more objective goal of managing threats and reducing adverse outcomes.

Risk management in other fields gives valuable insights into dealing with the fundamental concept of uncertainty and how it should be incorporated into an expanded financial risk management process in a process-driven framework. The strength of this resulting framework is its adherence to a firm set of principles which are both technically rigorous and practical. In writing this book, we took the cue from existing applications of risk, planning tools and methodology in finance and other industries while also taking into account key characteristics of the economic and financial world. The resulting approach and methodology had to be nimble, mathematically disciplined and scientifically mature in order to overcome the limitations of overly restrictive mathematical frameworks and ensure that uncertainty would become a criterion in all risk management considerations.

Objectives

We wish to address the issue of uncertainty and assess possible outcomes or combination of outcomes, rare or otherwise, which cause undesirable consequences. In addition, we are interested in how uncertainty can be incorporated into the current risk assessment framework and risk management system without adversely affecting processes and the bottom-line of banks.

Our goal is threefold. First, to present a robust foundation for the inclusion of uncertainty in managing foreseeable game-changing events which financial institutions are faced with today. Second, to illustrate the necessity of using qualitative inputs to manage these uncertainties and the need for a more rounded approach to address them. Current probability-based models are neither sufficient nor appropriate for managing these types of inputs. Third, to offer a framework and an implementable, flexible and workable solution for the incorporation of uncertainty into current risk assessment structure of banks. It is also important that this solution is based on a mathematical foundation that would blend well with current mathematical frameworks and not deter risk managers from continuing their current mandates and operational goals.

Scope of the book

Our focus is on the one-offs as well as the game-changing events and their uncertain consequences in great detail. In fact, when writing this book, we toyed with the notion of calling it The Risk Management of One-Offs! This title did not seem appropriate, since we also review possible future events which are by themselves not rare but may lead to other extremely adverse consequences and outcomes.

These potential outcomes can come in various forms and each will have its own characteristics and remedy. They can essentially be classified as an event, or isolated outcome that could spark off huge effects; a risk situation; or a combination of improbable but manageable events occurring together, sequentially or simultaneously, each requiring its own remedy; a causal chain; or a series of triggering events, linked to one another, causing an adverse situation; or a perfect storm, a combination of improbable and seemingly independent events which together exacerbate a risk situation. All these will be discussed in later chapters.

The subject of our concern is how to reduce the unknowns, these external triggering events and risk scenarios; as much as possible. When uncertainty is managed, banks are more resilient and can proactively manage their portfolio positions to lower possible revenue shortfalls or, if the stars are aligned, maximise revenue generation. Fortunately, risk management is not about the accurate prediction of the future. The emphasis is on optimising our portfolio position, avoiding or mitigating potential risks. Although these factors which drive and trigger reactions are so dynamic and interconnected that they may appear almost impossible to predict, often, we are able to foresee a range of possible outcomes or situations, and even deduce distinct possible events to which we can assign rough likelihoods to augment the historical data. This can be achieved through inputs from experts who are able to link possible causal interdependencies for a given risk situation. After all, historical data alone, especially from periods of relative normalcy, cannot adequately guide us in the risk management of idiosyncratic and game-changing one-offs.

When probability adequately describes the world, linked events and states of the world can be modelled with Bayesian networks. However, when we move to the realm of uncertainty, the linkages can only be modelled through credal networks. This set of modelling techniques forms the basis for our risk and uncertainty assessment method, uncertain VaR or UVaR, a method that can be used and monitored as new states of the world transpire. Given a complex set of interdependencies, we can derive chains of outcomes based on their relative likelihood and further plan and prepare for representative scenarios or eventualities. While actual losses may be unavoidable, scripting our response to a risk situation can reduce the likelihood that these large market shocks are catastrophic for the firm. Although we may not be able to predict exactly the causes and sequence of triggering events, the post-event remedies can be quite similar.

This book sets out a framework and method specifically for uncovering these remedies and scripting responses to uncertain risk situations. It is written for decision makers, risk practitioners, quants and academics in finance who want to tackle uncertainty, potential one-offs and game-changing events. Our approach should allow for a broader number of inputs and a range of probabilities to ascertain a more inclusive risk profile, one which takes into account the possibilities of game-changing events and situations. Decision makers would be able to plan for these possible adverse events as there are new tools for the identification and direct measurement for these uncertainties. Uncertainty should be managed and dealt with strategically as a means of assessing the possible gains and opportunities, while mitigating and minimising exposure to possible losses.

Outline of the book

The book is organised into three parts. Part I, which consists of Chapters 2–5, reviews the landscape and delves into the subject based on academic thought and literature, the birth of modern finance and evolution of risk management. This topic may not be new for risk practitioners but it is necessary to begin the book by highlighting that modern risk management’s emphasis on probability and quantitative risk has sidelined the management of uncertainty. Chapter 3 discusses the rationale for managing uncertainty and how current regulatory and risk assessment tools are insufficient. It is very clear that the banking and securities sector is currently stifled by regulatory uncertainty and is functioning under a cloud of political anxiety. Regulatory requirements which adhere to probability-based tools and historic data alone will not be sufficient to tackle uncertainty, no matter how much the models are tweaked. Taking on uncertainty requires a different sort of mathematics, combining data with more qualitative inputs. Learnings from risk management in other sectors are also discussed in Chapter 3. Chapters 4 and 5 review the subject of probability, the mathematical underpinning of financial risk management, including a discussion of its broad-based applications in modern finance. In the process of describing these canonical models of modern finance, we also discuss recent reformulations...

Table of contents

• Cover
• Title Page
• Copyright
• Contents
• List of Boxes
• List of Figures
• List of Tables
• Preface
• Acknowledgements
• 1 Introduction
• Part I: Setting the Landscape
• Part II: Managing Uncertainty: The Essentials
• Part III: Framework, Methodology and Case Studies
• Appendix: Model Uncertainty
• Notes
• Index