Don't expect the world's top reinsurers to apply today's latest earthquake forecasting research to how they work their books of business just yet.
Dr. Alex Allmann, lead modeling expert for Munich Re, said he would need to see further testing before he got comfortable with anything he might find in a seismology journal. Allman's remarks came during a January symposium put on by modeler Risk Management Solutions Inc., in New York City's marble-encrusted Hudson Theatre. He was responding to a question regarding what level of uncertainty reinsurers are willing to live with.
"We really value the scientific consensus," said Dr. Mariagiovanna Guaterri, vice president of asset-backed/insurance-linked securities for Swiss Re Capital Management & Advisory, adding that objective feedback on any new forecasting method was needed. For instance, she said that most catastrophe bonds now use time-dependent probabilities.
For well-studied faults, time-dependence is a way to increase the chances of a temblor based on where the fault is in its quake cycle. For instance, if you know a certain fault ruptures every 100 years, the probability would be greater for an event 75 years since the last quake, compared with only 25 years.
Allmann said Munich Re began using time-dependent probabilities in 1997, though he isn't completely sold on this technique.
"I am still not convinced," he said, "that we should place 100 percent confidence in these time-dependent issues."
But he also cited additional instances when his company has absorbed current science, such as when it sent out underwriting alerts in 2004, 2005 and 2007 for facultative reinsurance underwriters with interests in such hotspots as Sumatra.
"It's not that they can't write business," he said about those hotspots. It's just a matter of watching accumulations there.
The scientists at the RMS event, many of whom were there as presenters of their latest forecasting research (such as how one quake can make other quakes in the same area more probable), admit much is still unknown about quake frequencies.
One major question mark, according to Dr. Geoffrey King of the National Center for Scientific Research in France, is the influence played by subseismology: the discovery that the upper layer of the mantle and the lowest layer of the crust move at different rates.
Another mystery surrounds GPS technology, which is used to measure fault movement down to the millimeter. But GPS data doesn't jibe with the energy released from observed faults when quakes occur, as Tom Jordan of USC's Southern California Earthquake Center pointed out.
Still, such an open discussion about the limits of quake forecasting is a good thing among an insurance crowd. Gordon Woo, catastrophe risk consultant for RMS, touched upon this when he said that seismologists historically have not fully disclosed the uncertainty in their work.
And again it's not to say that the work of these seismologists hasn't benefited insurers and modelers despite this uncertainty.
"There is an ongoing review of state-of-the-art forecasting techniques," said RMS' Arnaud Mignan. RMS will continue to monitor this science, examine its impact on losses and deliver the information to its clients, he said.
March 1, 2008
Copyright 2008© LRP Publications