Stress Testing: A Board-Level Issue

WHITE PAPER Stress Testing: A Board-Level Issue Stress Testing: A Board-Level Issue Table of Contents About the Authors.............................................................................................1 Abstract............................................................................................................2 Introduction......................................................................................................2 Linkage to VaR..................................................................................................3 Constructing Stress Tests ...............................................................................5 Liquidity Risk ...................................................................................................6 An Integrated View of Stress Testing...............................................................8 Risk Architecture..............................................................................................9 Information and Data Management...............................................................11 Summary.........................................................................................................13 About SAS Risk Management for Banking...................................................15 ® Appendix.........................................................................................................16 i Stress Testing: A Board-Level Issue About the Authors Content for this paper was provided by Dr. Robert M. Mark and Antonis Miniotis. Mark is the Founding Partner and Chief Executive Officer of Black Diamond Risk, which provides corporate governance, risk management consulting, risk software tools and transaction services. Mark is also the founding Executive Director of the Master of Financial Engineering Program at the UCLA Anderson School of Management. He serves on several boards as well as on Checkpoint’s investment committee. He is on the executive committee of the board of the Professional Risk Managers’ International Association. In 1998, he was awarded the Financial Risk Manager of the Year by the Global Association of Risk Professionals. Mark is an adjunct professor and the co-author of two books: Risk Management (McGraw-Hill), published in 2001; and The Essentials of Risk Management (McGraw Hill), published in 2006. He served as the Chief Risk Officer and Corporate Treasurer and was a member of the management committee at Canadian Imperial Bank of Commerce (CIBC). Miniotis is a Solutions Architect in the Americas Risk Practice at SAS who specializes in risk solutions related to regulations as well as internal market, credit and firmwide risk. He has worked for SAS since 1998, and has been involved with SAS Risk Solutions since 2001 in presales and product management roles. He holds a PRM certificate and a master’s degree in operations research. 1 Stress Testing: A Board-Level Issue Abstract The recent global financial crisis painfully revealed the need for better, more comprehensive stress testing in the financial industry. Stress testing focuses on analyzing risks associated with rare or extreme – but still relevant – events. Even if something that could go wrong never materializes, firms should still have a contingency plan for such events (i.e., a firm’s response to Murphy’s Law). A stress test is the foundation for this contingency plan, and serves to answer two types of questions: • How much could I lose under certain stress scenarios? • How could I lose more than X dollars? This white paper discusses the motivation for stress testing and describes the role stress testing plays in the management of risk. It also provides insight into the different types of stress tests firms commonly employ – such as sensitivity analysis and scenario analysis. Further, it illustrates the importance of integrating stress test measures with value at risk (VaR) measures for financial and operational risks. The paper also describes important risk architecture and data management considerations. Introduction Stress tests are particularly good for identifying and quantifying the impact of a loss of either trading or funding liquidity – which is a key feature of a crisis. Stress tests are also very good for determining in advance which actions can handle large market moves, such as those that take place in market crashes (e.g., the subprime credit crisis). Large market moves could include significant changes in implied volatility, correlation changes, etc. There are several types of stress test approaches.1 One type is sensitivity analysis, which tests the impact on the portfolio of changes in different factors (or parameters) one at a time (e.g., a parallel shift in interest rates by 200 basis points). Another approach is scenario analysis, which tests what happens if some realistic but rare scenarios materialize. Scenarios are basically multifactor stress tests (e.g., a dramatic rise in defaults, an asset bubble, high commodity prices, dramatic flight to quality, real-estate deflation, etc.). Historical scenarios are custom-designed scenarios that serve, among other things, to highlight a current point of failure that might arise due to a repeat of a recent historical event – for example, the dramatic decline in housing prices that took place in the 1 Crouhy, Michel, Dan Galai, and Robert Mark. The Essentials of Risk Management. New York: McGraw-Hill, 2006. 2 Stress Testing: A Board-Level Issue 2007 to 2009 financial crisis – or a more distant historical event, such as failure of the systemically important Long-Term Capital Management (LTCM) hedge fund. There are also top-down stress scenarios and bottom-up stress scenarios. A top- down scenario is applied at the income statement and balance sheet level, whereas a bottom-up scenario is typically applied at the transaction level. When developing stress tests, it’s important to establish stress test limits. The purpose of stress test limits is to determine if the size of potential losses related to extreme events for the current portfolio is larger than the organization is willing to tolerate. Stress test limits should attempt to model a participant’s behavior and action taken to stay below a stress test limit in a stress environment. In other words, management response to an extreme event should be a critical component of the stress framework. Stress test limits also serve to meet regulatory requirements, such as Basel II. Stress tests work well to highlight hidden weaknesses in a risk management system, such as hidden “hot spots” in a portfolio under extremely negative market conditions. An integrated, bankwide analysis of risk for portfolios under stress uncovers critical interdependencies. With SAS Risk Management for ® Banking, analysts can stress test an entire bankwide portfolio as well as individual business units for the amount of market, credit, operational and asset liability management (ALM)-related risk. For example, this covers interest rate risk in the trading book (such as the risk arising from trading a bond) and in the banking book (such as risk arising from managing the asset/liability structural gap). Linkage to VaR VaR models 2 provide useful information, but should not be considered indicative of the amount at risk in extreme markets. In other words, VaR calculations are unlikely to reveal much about extreme events. One reason is that a typical VaR holding period may be too short to reveal the full impact of unusual features that arise in stress situations. There are three main approaches to calculating VaR. Analytic VaR assumes that the analytics properly describe the risk (for example, it may assume that percentage price changes follow a multivariate normal, or that price changes follow a lognormal distribution). Monte Carlo simulations (MCS) depend on choice of price process (such as lognormal, mean reverting or jump diffusion), but don’t require a closed-form analytic solution. Historical simulations assume 2 Crouhy, Michel, Dan Galai and Robert Mark. Risk Management. New York: McGraw-Hill, 2002. 3 Stress Testing: A Board-Level Issue that past probabilities are prologue. In other words, future risk is determined by past events within a historical data window. For reliable results, stress test measures should be fully integrated with VaR measures. Many examples prove that VaR in isolation of stress scenarios is of limited use. These include markets with low historical volatility that may be subject to large discrete movements encompassing the intersection of multiple risks – such as market, liquidity and credit risk, shown in Figure 1. It also includes markets that experience considerable regime switches. Stress tests are valuable for capturing risk in trending markets over multiyear time periods, such as during a housing bubble. Stress tests are also helpful in cases where instruments with nonlinear price movements (that is, complex exotic derivative transactions) respond to key risk factors that are not easily captured by traditional VaR models.