This is the second and last post covering a recent interview with a catastrophe risk assessment expert. The interview was especially interesting in light of recent natural disasters and efforts by Florida lawmakers to protect insurance companies -- at the expense of policyholders -- from the risks these models have predicted.
The inventor of the first hurricane catastrophe model in the early 1980s, this expert admits that she trusts these models less and less. The models are seldom updated, but the results fluctuate wildly from year to year, she says.
She suggests that the models be part of an arsenal of risk assessment tools. Companies rely on them now without looking at more credible sources of information. It is as if insurers were trying to diagnose a problem with your car using only a tire gauge. They are not getting the whole picture.
This other information needs to be based on real data -- historical data for a specific geographic region. They need to look at the losses from Hurricane Andrew and figure out what it would cost today if an Andrew hit. Future scenarios should be built on what's happened in the past, too.
Insurers need underwriters.
Underwriters know the accounts they work on. They have information about that town, that kind of business, and that particular business that a model can't have. Underwriters know the subtleties; models are more big picture.
Catastrophe models, she says, "can get you in the ballpark." Insurers need underwriting to get a more focused picture of individual risk. These companies have relied too long on one tool while excluding all others.
A combination of chain saw and surgeon's knife, an approach that includes the best aspects of a lot of different tools would be the best of all possible risk assessment worlds.
Source: Insurance Journal, "P/C industry depends too much on catastrophe models, says pioneer Clark," Andrew G. Simpson, 04/14/2011