The catastrophe risk modeling community is now much more responsive to the insurance and reinsurance business, although there’s clearly still challenges and opportunities to further refine the models, according to industry experts.
Speaking on a panel earlier today at the 2023 Bermuda Convergence event, moderated by Peter DiFiore, Managing Director, Neuberger Berman, experts from across the sector discussed the rise of catastrophe risk models and how these have developed to support the risk transfer industry.
“When I look at where we’ve come, the amount of global coverage, the number of perils, what we’re actually able to bring forward, and you think about what goes into the models, and you think about the development of hazard, the ability to quantify and understand the hazard, we’ve done a phenomenal job of being able to bring that forward,” said Dan Dick, Executive Director, Global Head of Property Analytics, Aon.
Dick went on to note that these models have not only been used for reinsurance and insurance-linked securities (ILS) transactions, but also for rate making and even loss mitigation.
“All of this stems from the fact that we’re bringing a common language and a common view of risk forward. Obviously, there’s still opportunities and challenges in front of us to further refine the models,” said Dick.
“I think what we’ve been able to do over the last 30 plus years is really coalesce an industry around risk and an understanding of risk and a common language of risk, and I think that is a phenomenal win unto itself,” he added.
The panel also included George Freimarck, Business Leader – Catastrophe Modeling, Applied Research Associates, who explained that the days of going to insurance conferences and there always being a session on cat models and were they right, which model is more correct than the others, has in many ways disappeared.
Freimarck feels that this is because “people are much more accepting of the models, because the results are much more in line with what’s happening with the policies, with the reinsurance structures.”
“The modelling community is much more responsive to, and the product is much better suited for, this business. So, echoing what Dan was saying, we’re much more in line with what the business is all about,” continued Freimarck.
The third speaker on the panel was Volkan Sevilgen, CTO & Co-Founder of Temblor, a catastrophe modeling company specializing in seismic hazard and risk assessment.
One of the things Sevilgen highlighted during the discussion was the difficulty, in earthquakes models specifically, of being able to test the models because large earthquakes are so infrequent.
“To test a California model, you need about 2000 years for these large earthquakes to appear, so you know that your model works. Nobody has 2000 years to test a model,” he said.
“We took a separate approach, and that approach is, California is about 1/2000th of the earth’s area. So, if you build a model that covers the entire globe, and it’s consistent with its methodology, you can test your models in few years very effectively.
“And the good part of it is you actually do not seem to lose resolution. So, you are naturally trading some kind of certainty by expanding the areas. And the challenging part would be then you are really working against some of the government models that have been established in the markets and legacy, and the price has been somewhat set, and then changing that past pricing has been slightly challenging for people to accept,” added Sevilgen.
