Catastrophe models are coming under fire following Hurricane Katrina, with some fearing that many initial insured loss estimates generated by the models will be inaccurate.

Stephen Way, chairman and chief executive officer of HCC Insurance Holdings, told Reactions that some insurers invest too much faith in computed estimates, which often turn out to be too low.

Mr. Way fears some insurers will struggle to cope with their losses and that this will have a negative effect on the reputations of all insurers and reinsurers. ?I would not sleep at night if I used just the models,? he said. ?We don?t rely on them or trust them but I think other companies do ? and they are in for much bigger losses than they expect.?

HCC uses catastrophe model figures as a small facet of risk assessment, he said. ?We do risk assessment manually. It?s boring, hard, tedious and reliable.? HCC has not yet released an estimate of how much Hurricane Katrina will cost it.

Dane Douetil, chief executive of Brit Insurance Holdings, agrees that loss estimates are too low. In a press conference on September 6 he said that Katrina could cost $50 billion in insured losses, due to the flooding in New Orleans that followed the hurricane. This figure is higher than most of the risk modelling firms? highest estimates.

Risk Management Solutions on Friday raised its prediction of insured losses to between $40 billion and $60 billion with flood damage in New Orleans contributing between $15 billion and $25 billion of this. Its peers? estimates are much lower.

Joe Annotti, vice-president of public affairs at the Property Casualty Insurers Association of America, says that loss predictions are often too generous. ?Sometimes they come out with these estimates that are so broad and often they turn out to be too high. It?s almost alarmist,? he said. And he believes catastrophe models are useful. ?I think they are very important tools for the risk assessment community. I wish I had a better alternative. But how else can you base your prices??

@EDITRULE: