THE ROLE OF STRESS TESTING #
Both stress tests and VaR measures [including economic capital (EC) measures] attempt to transform scenarios into loss estimates. The loss estimates’ distributions provide the basis to compute VaR at a very high confidence level. Stress tests tend to look at far fewer scenarios compared to VaR measures. There could be up to three major differences in the definition of loss estimates between stress tests and VaR/EC measures:
- Stress tests usually define losses from an accounting perspective, while VaR/EC measures usually take a market view of losses (i.e., deducting the opportunity cost of equity capital).
- Historically, stress tests have looked at longer time horizons, whereas VaR/EC measures have looked at point-in-time losses only.
- Stress tests do not focus on probabilities, but instead focus on ordinal rankings such as “severe,” “very severe,” and “extremely severe.” In contrast, VaR/EC measures focus on cardinal probabilities when interpreting the results of using Monte Carlo simulation (i.e., complicated statistical models) or historical simulation (i.e., actual past results). For example, a 99% VaR loss is interpreted as a 1-in-100 event.
COMPLEMENTING STRESS TESTS WITH VAR MODELS #
VaR/EC models compute expected losses using the following formula:
expected loss = PD x LGD x EAD
where: PD = probability of default
LGD = loss given default
EAD = exposure at default
In applying such models, financial institutions usually use a Merton model to simulate defaults and credit quality. With this model, asset returns are simulated using a factor model and default would occur when the simulated asset value falls short of a threshold (related to the borrower’s leverage) at the end of one year.
STRESSED INPUTS & STRESSED VAR #
Stressed inputs have been used in analysing market risk and for both supervisory and internal purposes within financial institutions. The revised Basel market risk capital framework mandates the use of stressed inputs. The revised framework also mandates the use of thorough and wide-ranging stress tests for financial institutions that employ internal models to compute the necessary market risk capital amounts. The Basel III revisions mandate the use of stressed parameters to calculate the default risk capital charge for counterparty credit risk.
Stressed VaR can be used to analyse the potential losses for an investment portfolio. In addition, it can be used to compute the capital charge for credit valuation adjustments (CVAs). The CVA represents the expected value or price of counterparty credit risk. For the VaR of CVA, it is important to select the correct stress period so as to compute a stressed CVA VaR that is higher than an unstressed CVA VaR.
KEY ASPECTS OF STRESS TESTING GOVERNANCE #
ADVANTAGES & DISADVANTAGES:
A key advantage of using stressed risk metrics is that they are conservative. In other words, the nature of the calculations should allow more-than-sufficient capital to be set aside for unexpected losses for future stressed events.
On the other hand, a key disadvantage is that risk metrics are stressed and will not necessarily respond to current market conditions. Instead, they will be impacted mainly by portfolio assets.