Newsletter Sign-up

By MATTHEW BRODSKY, senior editor/Web editor of Risk & Insurance®

In early August, the two biggest catastrophe modeling firms competed to see who could get a press release out the quickest about the latest versions of their earthquake models. Both Boston-based AIR Worldwide Corp. and Newark, Calif.-based Risk Management Solutions Inc. have incorporated the latest scientific consensus into their catastrophe modeling software, which allows reinsurers, insurers and insurance buyers to gauge the probability and potential severity of certain disasters.

The third major U.S. modeler, Oakland, Calif.-based EQECAT Inc., told Risk & Insurance® it plans to incorporate the latest appropriate science, too, with the release of its model update in March 2010, though much of the cutting-edge science is already in its current version.

Without getting into the nitty-gritty of this science, we can take away from this news that loss estimates put out by the newest models will be different than those from past models.

RMS estimates that the "average" insurer across all lines of business could see 10 percent to 25 percent reduction in U.S. earthquake loss estimates. For commercial portfolios in California, the drop could be 5 percent to 15 percent; for residential portfolios, as much as 25 percent to 35 percent.

Peter Yanev, board member of Global Risk Miyamoto and co-founder of EQE, had his number-crunchers at Global Risk estimate the difference. For a 250-year event in the eastern and central United States, insurers could see reductions of 10 percent to 15 percent; for the West as a whole, a 10 percent reduction.

But for the dreaded Cascadia Subduction zone, which stretches from Alaska to Northern California, Global Risk's analysis shows an increase in loss estimates of 5 percent to 10 percent for 250-year events, up to 5 percent increase for a 500-year monster.

Still, for those lower numbers in California, theoretically speaking, quake insurance buyers should see their rates lowered by next year's renewals. Should. Ravi Singhvi, assistant vice president of risk modeling at wholesale broker Napco LLC, said risk managers shouldn't assume they'll get a break on rates. It'll depend on each risk manager's portfolio and concentration of risks in earthquake country.

Or as Yanev put it, "The numbers are not significantly different to make a whole bunch of noise about it."

SCIENCE SHIFTS

For insurers and reinsurers, these model upgrades could be something to at least grumble about. They reinforce the notion that catastrophe modeling software is far from perfect, said Karen Clark, president and CEO of a Boston-based eponymous consulting company and founder of AIR. Sure, they are the best tools out there for gauging potential losses from certain perils, but they are full of assumptions (different for each brand) that are based on incomplete science. Look at that latest USGS report, Clark said, and you'll find the list of recommended improvements that scientists plan to tackle in the next report.

It wouldn't surprise Yanev, who wrote the book (literally) on earthquake risk management ("Peace of Mind in Earthquake Country"), if new data came out in a few years that sent the modeling numbers back up. He added that he is pleased that the latest thinking of the scientific community has been incorporated into these insurance tools.

"That does not mean that we're anywhere near to having this thing right," he said.

Clark cannot reiterate enough how these earthquake model upgrades are only further indication that catastrophe risk management must involve other steps and tools than modeling.

The modelers themselves appear to recognize this. In discussing how the new RMS model allows users to better understand underlying uncertainty, RMS Chief Products Officer Paul VanderMarck said in a press release, "While science develops and models become increasingly sophisticated, estimating losses from catastrophic events remains an inherently uncertain endeavor."

August 20, 2009

Copyright 2009© LRP Publications