A modeling expert said insurers who evaluate their exposure to hurricanes must look beyond catastrophe models and use other underwriting processes in order to truly assess their catastrophe risk.

Karen Clark, founder of the first catastrophe modeling firm and currently president and chief executive officer of Karen Clark & Company in Boston, reiterated statements she made earlier this year that catastrophe models are great tools but should not be overemphasized.

Ms. Clark said the industry has gone from all underwriting and no models 20 years ago, to virtually the opposite today, and she stressed that neither one of those extremes is an optimal approach to insuring against catastrophes.

Models, she said, are great tools, and modelers do the best job they can, but she added that the problem is models are based on a lot of scientific judgment rather than hard scientific facts. Therefore, companies need to understand the limitations of models when using them to manage risks, Ms. Clark said.

She said that in conversations with insurers, she recommends three categories beyond strict model results when it comes to protecting against severe catastrophe losses.

In independently assessing catastrophe risks, Ms. Clark said that companies need to have an idea of what their risks are before looking at model results. They could do this, she said, by looking at historical events, and determining what losses would look like today and how much exposure they have in those areas.

Companies need to independently assess their potential losses so that they can put model results in context, she advised.

Ms. Clark also suggested that insurers develop full model transparency and then test the credibility of the modeling results.

She said there are many reasons why a model result may not adequately reflect risk on a particular book of business. “Companies can't just take the results as fact. They have to fully vet them. They have to dissect them, test them, see what the sensitivities are,” she noted.

Finally, Ms. Clark said that insurers need to thoroughly test the completeness and accuracy of the exposure data that goes into the models. Insurers cannot get credible information out of models if they do not put the correct data in, she said.

Ms. Clark's recommendations came in advance of the 70th anniversary of the “Long Island Express.” The Long Island Express was a Category 3 hurricane that struck the Northeast on Sept. 21, 1938–the last major hurricane to hit the area.

The chance of another such storm hitting the area in a given year, Ms. Clark said, is about 1 percent, going by the prediction that storms of that intensity hit the region every 100 years. She said most estimates range from about 75-to-125 years.

Speaking to preparedness today versus 1938, Ms. Clark said the population today would be alerted sooner because of better warning systems and tracking of storms.

But she warned that Northeast hurricanes tend to be “very fast moving” compared to hurricanes in other regions, and she also noted that the track of a storm could change quickly. The Northeast, she said, could get less of a warning compared to other areas.

Ms. Clark cited one estimate of $300 million in total economic damage from the Long Island Express in 1938. The same exact storm today would cause around $50 billion in total economic damages, and about $15-30 billion in insured damages, she said.

Losses would be even higher if a similar size storm tracks farther west, where stronger winds on the right side of the hurricane would occur over more populated areas, said Ms. Clark

Special Reports /

Modeling Expert Tells How Insurers Can Vet Models

Related Stories

Recommended For You