Stress Testing to Fill in Gaps Left by VAR
Vendors and financial institutions have historically concentrated on market risk and credit risk with a slow move to integrate the two, but industry sources are pointing to another aspect of risk to fill in the holes that may be left by those calculations. Speaking at the first annual Risk Management Convention and Exhibition sponsored by the Global Association of Risk Professionals (GARP), Philip Best, managing principal at the Capital Markets Company, emphasized the importance of stress testing on a daily basis. Best describes the objective of stress testing as a way to identify scenarios that could cause significant loss, outside of regular market events. Stress testing addresses a wide variety of market situations, which could affect a financial institution in extreme situations,” explains Best.
While vendor systems may offer generic stress testing capabilities, Best says, “There are relatively few risk systems out there which really support stress testing well.” Instead vendors have concentrated more on value at risk calculations and statistical models, which Best says do not properly assess risk in many scenarios. The two systems that currently offer “good” and “reasonable” stress testing capabilities, according to Best, are Algorithmics and MKI, respectively. Best says stress testing represents a “new challenge for risk systems vendors,” which they will be responding to more and more in the future.
Recent financial crises in Asia and Russia as well as the Mexican peso crisis have prompted firms to move beyond credit and market risk. “Before the peso crisis, the classic value at risk model was showing a very low level of risk,” explains Best. “After the crisis, the measure increased, showing a lot more risk, but it was too late and the banks had already suffered the loss. Stress tests needed to be in place to identify the fact that this could happen, the models are very bad at predicting extreme events.” Best defines two guidelines that are essential for firms to perform effective stress testing—one, it must be done on a daily basis and two, it must account for systematic stressing of major risk factors.
Deborah Williams, research director at Meridien Research agrees. “Simulations (such as VAR) are not meant to look at extreme events, they are meant to look at an average day.” She explains that while a VAR simulation at a 95% confidence interval is meant to show what will happen 95 days out of 100, the other five days could pose a significant threat. “If you know that five days out of the next 100 your valuations are not going to fall in the normal range calculated by VAR and you have no idea what they will look like, that’s asking for trouble. If you’re not doing stress testing, you’re leaving yourself open, it’s a critical part of all risk management,” Williams says.
But it’s up to the risk manager to decide what type of stress testing is necessary for their portfolios as institutions utilize different methodologies depending on what they consider extreme market conditions. “Every portfolio is going to look a little different and a good risk manager plays with a portfolio and runs lots of different kinds of stress tests to see what makes the portfolio move. They look for the sensitivities and stress those things on a continuous basis,” says Williams.
“Risk management is still very new and putting VAR in place was a big step forward and was pushed heavily by regulators,” adds Best. “But in a very timely way the financial markets reminded everyone that it’s now time to take the next step.” He describes the next step as more routine stress testing procedures which take into account a banks positions and apply large numbers of prices changes in succession and in different combinations. In other words, he estimates that firms need to perform thousands of stress tests on a daily basis. “I think over the next two years we’ll see a big move toward stress testing,” concludes Best.
Tweet This