In previous post, the limitations of VaR were mentioned. To address the limitations, stress-testing is used to complement VaR. Stress-testing measures potential losses expected under extreme market scenarios. VaR and Stress limits set are known as risk appetite lmits.
Different types of Stress Tests:
1. Predefined Px/Vol metric. A grid that displays simulated PL under various spot px vs volatilities scenarios. Eg. 25% vol vs 10% px. This indirectly controls greeks like vega, gamma etc by limiting simulated PL to certain threshold levels.
2. Historical. This is to assume a repeat of past crisis scenarios.
3. Hypothethical. Forward-looking based scenarios. Assumptions about future stress event based on current economic outlook (GDP, Unemployment..). Usually done in collaboration with the economics/research unit of the bank.
4. Reverse stress-testing*. Based on understanding of risk profile of business, a worst case set of metrics is set up and then deduce a scenario that could trigger the event to occur.
In coming up with the shock quantums used for stress-testing, historical data is often used to callibrate the scenarios and data will have to cover sufficient crisis periods to ensure scenarios will be stressed enough.
* Illustration of how a reverse stress test is applied:
1) Historical data retrieved.
2) Compute holding period returns (eg. 3-day)
3) Average of max 3-day gains/losses. This forms shock quantums to be used.
4) Directional stress scenarios can vary each time depending on how risk profile of biz changes.
5) Deduce the event that can trigger the scenario. Or this step can be skipped as one just have to assume that regardless of what happened, this is what the portfolio will suffer if market were to go limbo. This basically exposes the vulnerability of the business.
Banks usually use a combination of the above to support their risk management measures.
No comments:
Post a Comment