Benchmarking of Stress-testing Practices in relation to Treasury Activities

Annex – Good Practices for Stress Testing The following practices are adopted by AIs with more robust stress testing program. (a) Segregation of duties is maintained by confining the execution of stress-testing to units independent of the front office. (b) Perform comprehensive stress-testing covering both banking and trading books. This is a recommendation set out in Sections 2.2.4 and 2.2.5 of SPM IC-5 that stress-testing should cover both the trading and banking books with the adoption of an integrated approach and providing results on an institution wide basis. (c) Clearly document the major stress assumptions in the relevant stress-testing policy and / or reports. (d) Adopt a policy requiring management to review the relevance and appropriateness of stress scenarios and parameters at least once a year. (e) Perform stress-testing on movements in foreign exchange rates, parallel and non-parallel yield curve shifts and, where applicable, narrowing of Prime / HIBOR spread for designated management committee’s regular review. In addition, perform stress-test showing the impact on net interest income as a result of a series of adverse stepwise interest rate movements. (f) Perform stress-testing scenarios that simulate previous major crisis situations, such as the 1997 Asian Crisis, the 1998 Russian Crisis and the 2001 U.S. Crisis. In addition, certain AIs perform stress-testing on a hypothetical oil crisis and a hypothetical crisis of major international currencies. (g) Generally perform stress-testing at least once a month, and more frequent stress￾testings, such as on a daily or weekly basis, on more risk sensitive portfolios (e.g. more volatile or highly leveraged products) for risk management unit’s review. (h) Define risk tolerance for stress-testing results with reference to impact on capital adequacy ratio. Excess of pre-set risk tolerance limit requires escalation to senior management for decision on remedial actions. (i) Assess the probability for each stress scenario and present both the gross and expected losses accordingly to management for their information and decision. (j) Present stress-testing results for designated management committees’ review with trend analysis. Trend analysis with proper explanation on large variances improves management’s understanding of an AI’s risk exposures.