Creditflux Newsletter

The models used by most major financial institutions to calculate default risk generally understate the probability of default during periods of financial stress, according to a new study, A Quantile Monte Carlo Approach to Measuring Extreme Credit Risk.

So-called structural models, which use the relationship between historical equity returns and firm asset values to measure default probability, are employed by banks to measure distance to default, from which probability of default can be calculated.

However, if instead of using historical returns, the calculations are run using returns during periods of market stress (in this case the worst 5% of asset value fluctuations) a completely different picture emerges.

Under the normal calculation the distance to default for the financial industry from 2007 to 2009 was 2.09, which is equal to a probability of default of around 2%. Under the modified calculation, during a period of high volatility and high leverage, the distance to default was 0.89, giving a probability of default of 21%.

"At the period of the worst asset value fluctuations of the global financial crisis, default risk was substantially higher than indicated by the traditional measure," says Professor David Allen, one of the paper's authors at Edith Cowan University in western Australia. "Traditional measures may not identify extreme risk, and capital and provisioning decisions based on these measures may leave banks short during a downturn."

 A copy of the paper may be found here:  http://ssrn.com/abstract=1948311