Convergence: Risk modelling 'not fit for purpose'

Current risk modelling is not fit for purpose, according to a pioneering researcher on the links between climate change and extreme weather events. 

Massachusetts Institute of Technology professor Kerry Emanuel, speaking at ILS Bermuda’s Convergence Conference, said models that relied on historical data did not reflect changes brought about by climate change or increases in exposures. 

Emanuel said most hurricane risk models would need 500 years of historical data to get a robust record, but these did not exist. 

“Most risk assessments are based on historic hurricane statistics, but most hurricane records are too short to be fit for purpose,” he said. 

Even if there was a comprehensive historic record, it would be a poor guide to the present, given the effects of climate change on extreme weather events and changes in exposure, he said. When Hurricane Harvey hit Houston in 2017, the cost of flooding was three times higher in real terms than it would have been in 1970, he said.

“Frankly it frightens me that  that so much of what we do is based on statistics that are not fit for purpose,” he said. 

Emanuel said the answer to the problem lay in applying global models to risk modelling and using physics-based models for catastrophe models. 

“A revolution is happening but it needs to happen faster,” he said in regard to applying physics-based modelling, adding that the re/insurance industry could help to drive modelling to a physics-based foundation. 

Emanuel said earlier the US was seeing a demographic shift in which people were moving to regions that were more exposed to catastrophes and climate risk.  

He noted that the risk landscape was changing. He said in addition to tropical cyclones, the US was also experiencing severe convective storms, winter storms, flooding, wildfires, droughts and heatwaves.  

He said the potential intensity of hurricanes was increasing but said they were not necessarily more frequent. He noted as well that hailstorms were causing $1 billion of insured damage annually, primarily to crops, and the areas affected was moving eastward.  

Earlier, the Convergence conference heard about the latest research on severe convective storms (SCS), which have cost insurers billions of dollars in losses in recent years. 

Kelsey Malloy, a research scientist on severe convective storms at Columbia University, said modelling the likelihood of SCS was difficult. 

This was partly because past data was dependent on past reports, which were unreliable. She noted that hailstorm reports were concentrated on where there were roads and in more heavily populated areas. 

This means the increase in hailstorm reports could be because people were getting better at reporting them.  

She said instead of using data, it made more sense to model the risk based on the presence of the ingredients that make up a thunderstorm. These included the presence of updrafts, windshear. 

In that way, it is possible to model a probability distribution and map where risks are more likely to occur.    

She said this helped to model more accurately where and how frequently SCS were likely to occur,  although it was also helpful to work in influences on the storm, including the effect of El Nino or La Nina on severe convective storms. 

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