Standard Life Investments quantitative investment manager Jens Kroeske said there are inherent dangers in standard scenario analysis methodologies.
He believes investment managers should instead adopt a ‘multi-regime entropy approach’ that goes beyond the “normality assumptions and linear co-dependencies found in traditional models”.
Mr Kroeske said standard risk systems generally use the Black-Litterman approach for forward-looking analysis which is “almost guaranteed to leave the fallible user with a false sense of cognitive ease”.
“The Black-Litterman approach assumes a linear model of co-dependence that is independent of the size of the shock applied,” said Mr Kroeske.
“This is contrary to intuition: under historical conditions, the equity returns are positively correlated to the spread component of the credit returns and negatively correlated to the duration component of the credit returns.”
The overall effect is very weak correlation of equity and credit, he said.
“However, in extreme scenarios, such as a ‘credit collapse’, the spread component of the credit return will dominate and hence credit returns will be positively correlated with equity returns,” said Mr Kroeske.
A multi-regime entropy pooling approach, however, enables the degree or co-dependence to be a non-linear function of the severity of the shock.
“This is because it does not rely on a simple linear relationship between the shock and the inferred return,” he said.
“Relationships between assets are always viewed in the context of the shocks, and as such this enables us to focus on those parts of simulated outcomes that are relevant for the scenario at hand.”
Mr Kroeske said the multi-regime entropy pooling approach enables the creation of “much richer scenarios and more complex views”.
“The process by which we then incorporate expert views into the market model leads to a unique solution and is tremendously scalable,” he said.
Mr Kroeske said the model means investment managers are “better able to capture the pernicious nature of tail risks” and ensure “portfolio performance is robust to a comprehensive range of possible states of the world”.
