Lloyd’s Comments on Cat Models
“Catastrophe models are useful, but have their limits,” says a comment on the Lloyd’s Web site (www.lloyds.com). “That is the lesson learnt by the insurance industry following last year’s devastating hurricane season.”
Lloyd’s observed, as have many others in the industry, that the loss models of last year’s hurricanes “varied alarmingly from model to model.” A number of analysts have in fact pointed out that Katrina should have been, according to most models, a maximum $10 billion event, not $40 billion plus. Lloyd’s acknowledged that while its “market’s risk management framework stood up well to the rigorous hurricane season, and its Realistic Disaster Scenarios were useful, but as insurers throughout the industry discovered, no model is perfect.”
Paul Nunn, manager of Lloyd’s Loss Modeling Unit, analyzed the situation as follows: “Insurers in and outside of Lloyd’s had already begun to reassess how they used models. There was a realization that, while they were a valuable tool, they could not be used in isolation or as a substitute for the experience gained in underwriting and covering past events. While vendor hurricane models were quite well established for wind damage, they had limited capability for estimating the losses from associated perils such as floods, which caused as much damage in New Orleans as the windstorm itself.”
He added that offshore exposure models were still very new, and so did not yet have the chance to prove their reliability against actual events. “Hurricane models are effectively property models and will not account for the many other classes of business impacted such as marine or cancellation insurance,” he continued. “While the modeling firms have been quite quiet about potential holes in the models, it has become very apparent to insurers who have been signing the checks.”
Nunn reiterated that while the models do have real value, Lloyd’s “Realistic Disaster Scenarios” might well be in need of “stiffening, as the “evolving consensus that the risk environment for US hurricanes is increasing.”
Eddie McGlauchlin, director of Risk Consulting and Head of Marsh’s Modeling Unit, indicates that the market will have to undergo a cultural change if it is to use risk models in the way they’re intended to be utilized. “Models were never meant to be the definitive answer to the issue,” he explained. “They need to be used in conjunction with good sound judgement. They are not black boxes which hold all the information you need. This is not an exact science. The fact is that we are in the risk business and therefore our business is inherently risky. We simply need to take a structured view.”