Risk modellers duke it out

“Puerto Rico’s insurance regulator has placed the Maria losses at $7bn for the industry,” stated Ron Diaz, executive vice president and international department head at Everest Re. This is a far cry from any of the estimates from RMS, AIR Worldwide, Karen Clark & Co or PCS, although the former was closest to that amount with a bottom range loss estimate of $13bn. Peter Sousounis, vice president of research at AIR Worldwide, explained the reasoning behind AIR’s initial estimate of $72bn to $34bn from Hurricane Maria in Puerto Rico. “There were some complications involving business interruption. These pharmaceutical companies and other industries were not just dead in the water in terms of business interruption. They were having generators shipped to them to try to recover and resume their business. Complicating insurance policies, a lot of these industries were self-insured, which translates to a higher deductible,” he said. “After we learned these things we felt that our second estimate of $25bn to $43bn is in line with other industry estimates, including Swiss Re which estimates the loss from Maria to be $32bn,” he added. According to Sousounis, AIR is also working to fill in gaps in its own industry exposure database following last year’s catastrophes. Glen Daraskevich, senior vice president at Karen Clark & Company, explained that speed and accuracy were not mutually exclusive concepts when it comes to modelling. “Some take the position that accuracy and speed are different things: ‘like we can give you estimates very quickly but they can’t be accurate’. At KCC that is not our philosophy. We don’t have a choice. We have to deliver both in real time. Time should be a factor due to the real life implications of these estimates,” he said. According to Pete Dailey, another implication of Maria is the focus on using multiple sources to collect data, and not relying on sources such as the National Hurricane Center. “I’m a little confused as to what brought about the change in your estimates. I think what you said, was that you don’t need ground observation of wind when you can use these environmental parameters, but when you used the environmental parameters that led to estimates that didn’t make a lot of sense,” he said in response to Sousounis’ explanation of how AIR came it its initial estimate, and why it revised it some weeks later. “What we’ve learned from events like Maria is that depending solely on environmental data and forecasts from the National Hurricane center and reconnaissance estimates prior to landfall, you’re going to tend to overestimate the losses, because the National Hurricane Center’s charter really is to warn the public, and so they tend to be conservative in their estimates.” All three firms stood by their most recent estimate for the event in Puerto Rico as losses continue to develop and the situation continues to evolve.

Blurred image of Reactions article content
Blurred image of Reactions article related content