Work is needed on SCS models, says Aon

Reinsurers may be counting their exposures on severe convective storms (SCS) quite blindly, finding the final tallies well outside what strained cat models might have told them to expect. A reckoning between real-world loss experience and the models’ estimates is clearly in order.

So says Katie Carter from Aon’s View of Risk Advisory where she is head for the Americas. 

“For SCS, there has been a challenge across the industry to have a model that aligns with experience,” Carter told APCIA Today. “The models have historically underperformed,” she said of a 20-year history of rising claims severity.

The reasons for the miss between experience and the models mirror the reasons why the billion-dollar storm count misses the point: changing demographics and housing development patterns outweigh any evolution in weather patterns. 

“Some perils, tornado being one example, may be over-represented in some models.”

“A tornado that occurred 20 years ago may have hit an undeveloped area; that same tornado today hits an area that has grown significantly,” Carter said. She said such changes can account for up to 80 percent of the difference. “That is the primary reason we see the miss in the models.” 

Being granular means getting peril-specific to cover the gap between experience and modelled output. Some perils, tornado being one example, may be over-represented in some models.

“When some of the models were first developed it was right after events in 2011, including the big Tuscaloosa tornado. Those cat models had a lot of the tail risk in tornado,” Carter said. Compare today: a recent midwestern derecho makes a great example, bringing hurricane force winds to communities never built to endure such stresses.

“The right kind of cat models need to reflect all the sub-perils in the tail risk,” Carter said. “It is not always a tornado that drives it.”

Benchmark and tweak

Reinsurers should benchmark their claims experience against the models and tweak them accordingly, Aon argues. That can be aggregate claims analysis to identify macro shortcomings, and location-level exposure data which Aon runs through its own Impact Forecasting for the deep dive. It does this to find out “not only what we are missing, but why we are missing it”.

The 2023 reinsurance market reset including the higher retentions that it ushered in may have protected reinsurers in 2023 and 2024 from many of the frequency losses, but it has not undermined the financial benefit that can come from benchmarking model experience, Carter said.

“For this peril, it has become a volatility issue around managing earnings potential,” Carter said. “It’s still very much a concern for our reinsurance clients.”  

Reinsurers are looking at ways to manage the volatility. Insights from model benchmarking are feeding actions such as implementing actual cash value of roof policies, nuances in how to implement deductibles in different hazard areas, additions of more hazard-dependant deductibles or percentage deductibles and more, “varying it based on the likelihood of the hazard”. 

For more news from the American Property Casualty Insurance Association (APCIA) click here.

Did you get value from this story?  Sign up to our free daily newsletters and get stories like this sent straight to your inbox.