PART I The Accounting- Based Approach SECTION A The Balance Sheet Based Approach
CHAPTER 2 Introduction to the Balance Sheet Based Approach to Stress Testing CHRISTIAN SCHMIEDER LILIANA SCHUMACHER The balance sheet approach is the natural approach to stress testing banks and other fi nancial institutions. The reason is twofold: first, balance sheet information is publicly available for financial institutions in a standardized format, 1 which facilitates peer comparison and systemwide stress tests; second, the approach allows a bottom- up view on the vulnerabilities of institutions and hence the identification of important risk drivers. These two dimensions are key advantages of the approach compared with market- based models, which are based on markets perceptions of default probabilities and other risk drivers embedded in market prices (such as stock and bond prices and credit default swap spreads), which are not necessarily available for all banks, even in developed markets. The balance sheet approach to stress testing has, in part, been spurred by the Financial Sector Assessment Program (FSAP), conducted by the IMF and the World Bank since the late 1990s. This method, introduced by IMF staff in the earlier years of the FSAP, represents the first generation of (balance sheet based) stress tests and is documented by Čihák in Chapter 3. The contribution was not only to allow for stress tests covering the key risks faced by banks (credit, market, liquidity, and contagion risk) but to do so in a meaningful yet practical, transparent, and flexible way to account for differences in financial systems and their vulnerabilities. Work outside the IMF in this area has also been abundant in recent 1 This assumes that banks use the International Financial Reporting Standards rather than national Generally Accepted Accounting Principles. However, even in the latter case, it is usually relatively straightforward to run countrywide stress tests and, with expert judgment, to make the necessary cross- country comparisons. years, as documented in central banks Financial Stability Reports. By now, many central banks and/or regulatory agencies run stress tests on a regular basis, and some publish the results of the tests on their respective Web sites. At present, the key balance sheet frameworks used at the IMF are the Excel- based balance sheet approaches introduced in Chapter 3 and subsequent calibrations and enhancements. The choices depend on the complexity of the banking system being assessed, the data at hand, and the experience of the stress tester. More integrated stress testing frameworks (such as the ones used by the Bank of En gland and the Austrian Central Bank) are more complex and are perceived to be black boxes to some extent, although the recent prominent crisis stress tests conducted by the U.S. and Eu ro pe an authorities have leaned in the direction of the simpler balance sheet based approach. Despite its popularity and many advantages, the balance sheet approach also faces its own challenges. These include the ability to adequately capture the risks in a comprehensive, meaningful, timely, and preferably forward- looking manner; as well as systemic aspects, such as the risk- transmission process among banks, fi nancial institutions, and other market participants that can create or help propagate systemic risks (e.g., sovereigns). Extensions of the basic balance sheet described in Chapter 3 and the related approaches presented in this volume seek to tackle these two dimensions as further outlined in the following. DATA CONSIDERATIONS The quality of a balance sheet based stress test depends on the scope and quality of available data and the selection of an
14 Introduction to the Balance Sheet Based Approach to Stress Testing adequate technique. Timely and high- quality data are prerequisites for a meaningful balance sheet based stress test but will likely remain a major limitation for some time, for the following reasons: data are very often limited in scope (e.g., typically on new products), and this can be crucial (as the crisis has shown); data quality is also often limited (such as in the case of missing values and short time series). 2 The practical implication of this is that the information used to conduct balance sheet based stress tests ideally should be more granular than the pure fi nancial statements information (i.e., assets, liabilities, and items from the profi t and loss account). Additional information may be available in banks regulatory Pillar 3 reports and from publications by regulatory authorities. 3 Valuable data provided by authorities and banks may also be found on the Web pages of supervisory authorities (e.g., most notably the Eu ro pe an Banking Authority with the 2011 EU bank stress tests, when more than 3,000 bank variables were published). 4 The most detailed information on banks is available to supervisors (e.g., in the credit register and other supervisory data) and to the banks themselves, but these sources of information are confidential and not available to the public. Several chapters in this book address the data issues: Ong, Maino, and Duma (Chapter 4) demonstrate how solvency stress tests could be conducted in countries with weak data. Separately, Schmieder, Puhr, and Hasan (Chapter 5) provide examples of how to integrate information from other sources into balance sheet data, in order to make the assessment more meaningful and forward- looking. Their framework also accounts for data limitations and allows for risksensitive solvency and liquidity stress tests even in cases where only limited data are available. Ong and Čihák (Chapter 6) apply a traditional liquidity test to Iceland and document the limitations of solvency stress tests in identifying the buildup of risks in the lead- up to the crisis. This contribution remains one of the very few studies validating stress tests, an area to be explored further in order to improve stress testing framework on the one hand and 2 See Hardy and Schmieder (2013). 3 Pillar 3 of the Basel framework seeks to facilitate market discipline. Banks Pillar 3 reports are particularly useful for balance sheet based stress tests. They contain information on banks capital adequacy and comprehensive information on specific risk types namely, (counterparty) credit risk (including structured credit), market risk, operational risk, and liquidity risk and their relevance, some of which is standardized as foreseen by Basel II regulation. A cornerstone of this reporting is to provide information to market participants and thus reduce information asymmetries. An illustrative example is the Pillar 3 information provided by Deutsche Bank; the bank has published Pillar 3 reports since 2008 (http://www.db.com /ir /en /content /reports _2010.htm). Further information can be found in Basel Committee on Banking Supervision (2006, pp. 226 42) and the pertinent EU regulation (DIRECTIVE 2006 /48/EC). It is anticipated that Pillar 3 disclosure requirements will be amended (beyond disclosure on remuneration). U.S. banks do not (yet) publish Pillar 3 reports. 4 See http://stress- test.eba.europa.eu/ for further information. also to be reminded of their limitations on the other. Separately, the work by Schmieder and others (Chapter 7) represents the balance sheet based liquidity risk compendium to their solvency stress test in Chapter 5 and allows stress tests to be carried out even with limited information on liquidity. THE CRISIS AND REGULATORY REFORMS In recent years, various contributions have been added to the domain of balance sheet based approaches. The aims include (1) accounting for the lessons learned from the crisis (e.g., that the quality of capital matters); (2) making use of recent risk management techniques; and (3) making use of more granular information once it becomes available. A catalyst in this context was the introduction of Basel II, which lifted bankwide stress tests (beyond market risk) formally into the regulatory framework. The other driver was, naturally, the financial crisis, whereby the role of stress tests moved beyond what- if computations (which more often than not omitted policy actions) to that of crisis management tools used with the aim of regaining the confidence of markets by removing asymmetric information on the value of banks assets and liabilities between bank managers and market participants, as in the case of the U.S. and EU stress tests. These issues are explicitly targeted in the frameworks presented by Schmieder and his respective coauthors in Chapters 5 and 7: Chapter 5 introduces IMF staff s next- generation balance sheet based framework for assessing solvency risks, which is centered on Basel II/III developments. The framework allows the user to run multiyear tests for hundreds of banks and is more risk sensitive than previous frameworks (by adjusting risk- weighted assets for increasing portfolio risk) and more comprehensive. Chapter 7 presents the equivalent of the next- generation balance sheet based stress tests for liquidity risk. As in the solvency case, the purpose of the framework is to allow for more comprehensive, more risk- sensitive tests while keeping it transparent and relatively simple. In addition to traditional bank- run- type liquidity tests, the framework enables the assessment of maturity mismatches, including full- fledged cash flow tests used by banks and regulators. Likewise, the framework allows an analysis of the links between solvency and liquidity risks and the computation of proxies for the Basel III liquidity ratios, in all cases provided that data are available. PORTFOLIO RISKS AND SYSTEMIC ASPECTS Given their (static) nature, traditional balance sheet approaches need to be complemented with satellite models.
Christian Schmieder and Liliana Schumacher 15 Satellite models explore statistical regularities between macroeconomic variables and the quality of bank assets or income accounts. In this way, balance sheet items, including capital, can be projected under stress scenarios. An alternative approach is taken by portfolio models that use valuation techniques to project balance sheet items under stress scenarios (defined by the properties of the multivariate distribution of price changes). The focus of these approaches is the simulation of the stressed macroeconomic environment as a set of correlated changes in prices that are used to reprice all balance and offbalance-sheet transactions. Because these models target valuation, which is the core source of most changes in bank capital, they are good at integrating in a straightforward manner different kinds of risks, such as market and credit risks. In Chapter 8, Barnhill and Souto extend the basic portfolio approach presented in Barnhill, Papapanagiotou, and Schumacher (2002) to a set of institutions operating in the same fi nancial environment. For illustration, they apply the approach to a set of Brazilian banks to assess systemic risks coming from different channels of interconnection, such as a common exposure to changes in prices and claims in the interbank market. Using the same portfolio approach and following Aikman and others (2009), Barnhill and Schumacher (Chapter 9) integrate systemic solvency and liquidity risks, including their simultaneous feedback, by simulating the market response to deteriorating bank asset quality (funding liquidity) as well as the impact of asset sales (to address liquidity demands) on asset values (market liquidity). The framework is applied to a set of stylized U.S. banks using publicly available information. Separately, Avezani and others illustrate an alternative portfolio approach using an actuarial model in Chapter 10. This application of the CreditRisk+ model is based on the framework developed by Credit Suisse Financial Products. Prior to the crisis, the stress tests developed by IMF staff to date had largely focused on shocks to and within a par ticu lar fi nancial system. However, the global fi nancial crisis underscored the importance of anticipating cross- border shocks in an increasingly interconnected world. Among those who tried to address this gap, Cerutti and others (Chapter 11) com- bine satellite modeling with balance sheet based analysis to estimate the impact of potential ring- fencing on international banks (see also IMF staff work in the areas of network analysis and extreme value theory later in the book). CONCLUSION Notwithstanding the recent welcome developments in the balance sheet approach to stress testing, challenges remain. Frameworks need to be made gradually (even) more comprehensive and risk sensitive (e.g., by accounting for the evolution of business and/or risk- management techniques), while keeping them tractable. The modeling of macro- financial linkages and interconnectedness needs to be extended to capture all channels of risk transmission as well as the links between solvency, liquidity, and contagion risks. Most important, data need to be made more available to ensure reliability and to allow for up- to- date, preferably forward- looking, analysis. Finally, stress tests need to integrate the universe of financial institutions operating in a common financial environment. An example of the extension of stress testing techniques to nonbanks is provided by Impavido in Chapter 12, representing a nascent area of work at the IMF. REFERENCES Aikman, David, Piergiorgio Alessandri, Bruno Eklund, Prasanna Gai, Sujit Kapadia, Elizabeth Martin, Nada Mora, Gabriel Sterne, and Matthew Willison, 2009, Funding Liquidity Risk in a Quantitative Model of Systemic Stability, Working Paper No. 372 (London: Bank of En gland). Available via the Internet: http:// www.bankofengland.co.uk /publications /Pages /working papers /2009 /wp372.aspx Barnhill, Theodore, Panagiotis Papapanagiotou, and Liliana Schumacher, 2002, Mea sur ing Integrated Market and Credit Risk in Bank Portfolios: An Application to a Set of Hypothetical Banks Operating in South Africa, Financial Markets, Institutions and Instruments, Vol. 11, No. 5, pp. 401 43. Basel Committee on Banking Supervision, 2006, International Convergence of Capital Mea sure ment and Capital Standards: A Revised Framework Comprehensive Version (Basel, June). Available via the Internet: http://www.bis.org /publ /bcbs128.htm Hardy, Daniel C., and Christian Schmieder, 2013, Rules of Thumb for Bank Solvency Stress Testing, IMF Working Paper 13/232 (Washington: International Monetary Fund). Available via the Internet: http://www.imf.org /external /pubs /cat /longres.aspx?sk =41047.0