Modelling: Cat risk modellers admit shortcomings
There needs to be a “paradigm shift” in catastrophe risk models and the way they are used, according to a leading modeller.
At the same time an influential academic has warned practitioners against placing too much reliance on them. Although acknowledging they can be useful, he said there was still an over-riding need to use judgement when making risk decisions.
Modelling is not a precise science, said Peter Taylor, a former reinsurance underwriter and now research fellow at Oxford University. “Models don’t represent the risk. They’re valuable tools, but they’re too coarse,” he told the IUA (International Underwriting Association) Catastrophe Risk Management Seminar. “Judgement is back, thank goodness.”
To illustrate the point, Taylor said events of the severity of Hurricane Katrina and last year’s Japanese earthquake and tsunami were simply not captured in the models. Predicting extreme events is not a science but a series of differing views, he said; one way forward is to use more than one model. “Probably the use of multiple models is here,” he said.
He quoted from Hemant Shah, CEO of leading modelling firm Risk Management Solutions, who recently acknowledged shortcomings.
Cascading risks
“Cascading risks” are an increasingly important factor, and one that is hard to capture in models, the president of Weather Predict Consulting Inc told the event. Craig W. Tillman said that these phenomena, which cause knock-on effects well beyond the original event, are frequently underestimated in Business Interruption plans. This is not a new problem, he acknowledged. The 1906 San Francisco earthquake was one such catastrophe. However, the growing complexity of modern societies had increased the risk of cascading failures. Last year’s Japanese tsunami and Thai floods had illustrated these tendencies, highlighting weaknesses in supply chains and the failure of companies to understand them.
|
|
A PARADIGM SHIFT Modellers and users must change to enable more resilient risk management practices Hemant Shah, CEO, Risk Management Solutions Whilst stating that catastrophe risk modellers had served with distinction, Shah acknowledged shortcomings and the need for change. “We all became too complacent. While models can help optimise a book of business, an over-reliance on models can lead to fragile portfolios that are prone to surprises, whether from Mother Nature or from the models themselves. As a modeller, I have been slow to recognise these rigidities....A paradigm shift is required.” |
“Models don’t represent the risk. They’re valuable tools, but they’re too coarse,”
Peter Taylor
research fellow at Oxford University