P1.T4. Valuatin & Risk Mdels Principles fr Sund Stress Testing Practices and Supervisin Binic Turtle FRM Study Ntes By David Harper, CFA FRM CIPM www.binicturtle.cm
Principles fr Sund Stress Testing Practices and Supervisin, May 2009 DESCRIBE THE RATIONALE FOR THE USE OF STRESS TESTING AS A RISK MANAGEMENT TOOL.... 3 DESCRIBE WEAKNESSES IDENTIFIED AND RECOMMENDATIONS FOR IMPROVEMENT IN: THE USE OF STRESS TESTING AND INTEGRATION IN RISK GOVERNANCE... 4 STRESS TESTING METHODOLOGIES AND SCENARIOS... 5 2
Principles fr Sund Stress Testing Practices and Supervisin, May 2009 Describe the ratinale fr the use f stress testing as a risk management tl. Describe weaknesses identified and recmmendatins fr imprvement in: The use f stress testing and integratin in risk gvernance Stress testing methdlgies Stress testing scenaris Stress testing handling f specific risks and prducts. Describe stress testing principles fr banks regarding the use f stress testing and integratin in risk gvernance, stress testing methdlgy and scenari selectin, and principles fr supervisrs. Describe the ratinale fr the use f stress testing as a risk management tl. Stress testing refers t bth the mechanics f applying specific individual tests and t the wider envirnment within which the tests are develped, evaluated and used within the decisinmaking prcess. A stress test is cmmnly described as the evaluatin f a bank s financial psitin under a severe but plausible scenari t assist in decisin making within the bank. Stress testing is an imprtant risk management tl that is used by banks as part f their internal risk management and is prmted by supervisry banks thrugh Basel II capital adequacy framewrk. Stress testing alerts bank management t adverse unexpected utcmes related t a variety f risks and prvides an indicatin f hw much capital might be needed t absrb lsses shuld large shcks ccur. Stress testing has gained prminence and credibility within banks as a tl that cmplements risk management and capital planning. Stress testing supplements ther risk management appraches and measures. It plays an imprtant rle in: Prviding frward-lking assessments f risk Overcming limitatins f mdels and histrical data Supprting internal and external cmmunicatin Feeding int capital and liquidity planning prcedures Infrming the setting f a banks risk tlerance Facilitating the develpment f risk mitigatin r cntingency plans acrss a range f stressed cnditins. 3
Stress testing is mre imprtant during gd ecnmic times Stress testing is especially imprtant after lng perids f benign ecnmic and financial cnditins, when fading memry f negative cnditins can lead t cmplacency and the underpricing f risk. It is als a key risk management tl during perids f expansin, when innvatin leads t new prducts that grw rapidly and fr which limited r n lss data is available. Describe weaknesses identified and recmmendatins fr imprvement in: The use f stress testing and integratin in risk gvernance Bard and senir management invlvement is critical in ensuring the apprpriate use f stress testing in banks risk gvernance and capital planning. This includes setting stress testing bjectives, defining scenaris, discussing the results f stress tests, assessing ptential actins and decisin making. At banks that were highly expsed t the financial crisis and fared cmparatively well, senir management as a whle tk an active interest in the develpment and peratin f stress testing, with the results f stress tests serving as an input int strategic decisin making which benefited the bank. Stress testing practices at mst banks, hwever, did nt fster internal debate nr challenge prir assumptins such as the cst, risk and speed with which new capital culd be raised r that psitins culd be hedged r sld. The financial crisis has als revealed weaknesses in rganisatinal aspects f stress testing prgrams. Prir t the crisis, stress testing at sme banks was perfrmed mainly as an islated exercise by the risk functin with little interactin with business areas. This meant that, amngst ther things, business areas ften believed that the analysis was nt credible. At sme banks, the stress testing prgram was a mechanical exercise. While there is rm fr rutinely perated stress tests within a cmprehensive stress testing prgram (e.g. fr backgrund mnitring), they d nt prvide a cmplete picture because mechanical appraches can neither fully take accunt f changing business cnditins nr incrprate qualitative judgments frm acrss the different areas f a bank. In many banks, stress tests were carried ut by separate units fcusing n particular business lines r risk types. This led t rganizatinal barriers when aiming t integrate quantitative and qualitative stress testing results acrss a bank. Prir t the crisis, many banks did nt have an verarching stress testing prgram in place but ran separate stress tests fr particular risks r prtflis with limited firm-level integratin. Risk-specific stress testing was usually cnducted within business lines. While stress testing fr market and interest rate risk had been practiced fr several years, stress testing fr credit risk in the banking bk has nly emerged mre recently. Other types f stress tests are still in their infancy. 4
As a result, there was insufficient ability t identify crrelated tail expsures and risk cncentratins acrss the bank. Stress testing framewrks were usually nt flexible enugh t respnd quickly as the crisis evlved (e.g. inability t aggregate expsures quickly, apply new scenaris r mdify mdels). Further investments in IT infrastructure may be necessary t enhance the availability and granularity f risk infrmatin that will enable timely analysis and assessment f the impact f new stress scenaris designed t address a rapidly changing envirnment. Fr example, investing in liquidity risk management infrmatin systems wuld enhance a bank s ability t autmate end-f-day infrmatin, btain mre granularity as t unencumbered assets and frecast balance sheet needs f business units. Stress testing methdlgies and scenaris Stress tests cver a range f methdlgies and their cmplexity can vary, ranging frm simple sensitivity tests t cmplex stress tests. These methdlgies aim t assess the impact f a severe macrecnmic stress event n measures like earnings and ecnmic capital. Stress tests may be perfrmed at varying degrees f aggregatin, frm the level f an individual instrument up t the institutinal level. Stress tests are perfrmed fr different risk types including market, credit, peratinal and liquidity risk. Ntwithstanding this wide range f methdlgies, the crisis has highlighted several methdlgical weaknesses. At the mst fundamental level, weaknesses in infrastructure limited the ability f banks t identify and aggregate expsures acrss the bank. This weakness limits the effectiveness f risk management tls including stress testing. Mst stress testing mdels used histrical statistical relatinships t assess risk. They assumed that risk is driven by a knwn and cnstant statistical prcess. The crisis has revealed a cuple f serius flaws with relying slely n such an apprach: First, given a lng perid f stability preceding the financial crisis, backward-lking histrical infrmatin indicated benign cnditins s neither these mdels culd pick up the pssibility f severe shcks, nr the buildup f vulnerabilities within the system. Histrical statistical relatinships, such as crrelatins, prved t be unreliable nce actual events started t unfld. Secnd, the financial crisis has again shwn that in stressed cnditins, risk characteristics can change rapidly as reactins by market participants within the system can induce feedback effects and lead t system-wide interactins. These effects can dramatically amplify initial shcks as recent events have illustrated. 5
Extreme reactins (by definitin) ccur rarely and may carry little weight in mdels that rely n histrical data. It als means that they are hard t mdel quantitatively. The management f mst banks did nt sufficiently questin these limitatins f mre traditinal risk management mdels used t derive stress testing utcmes nr did they sufficiently take accunt f qualitative expert judgment t develp innvative ad-hc stress scenaris. Therefre, banks generally underestimated the strng inter-linkages between, fr example, the lack f market liquidity and funding liquidity pressures. The reliance n histrical relatinships and ignring reactins within the system implied that firms underestimated the interactin between risks and the firm-wide impact f severe stress scenaris. Mst banks did nt perfrm stress tests that tk a cmprehensive firm-wide perspective acrss risks and different bks. Even if they did, the stress tests were insufficient in identifying and aggregating risks. As a result, banks did nt have a cmprehensive view acrss credit, market and liquidity risks f their varius businesses. An apprpriately cnducted firm-wide stress test wuld have beneficially drawn tgether experts frm acrss the rganisatin. Fr example, the expertise f retail lenders, wh in sme cases were reducing expsure t US subprime mrtgages, shuld have cunteracted the verly ptimistic utlk f traders in securities backed by the same subprime lans. 6