Karen Clark is co-founder and CEO of Boston-based Karen Clark & Co., a catastrophe modeling company established in 2007 to help insurers better manage catastrophe risk. In 1987, Ms. Clark founded the first catastrophe modeling company, Applied Insurance Research, which became AIR Worldwide when it was acquired in 2002 by Insurance Services Office Inc. Ms. Clark recently spoke with Business Insurance Deputy Editor Claire Wilkinson about Hurricane Ida, the lessons learned, and how catastrophe models are helping insurers manage climate change. Edited excerpts follow.

Q: How did the modeling for Hurricane Ida work out?

A: Every hurricane is unique. Ida is a case in point. There was Ida part one and Ida part two. Part one was when it made landfall in Louisiana as a Category 4 storm. It was a typical Cat 4, with wind and storm surge damage. Our industry estimate for Ida part one was $18 billion, which seems to be holding well. We’re confident that the KCC model captured accurately Ida part one. But then Ida dissipated. The National Hurricane Center was not tracking it; it was not even a tropical depression. It was the remnants of Ida that interacted with a cold front over the Northeast that caused Ida to drop moisture over the mid-Atlantic and Northeast. Ida part two was a flooding event. There was lots of flooding in areas that are not even marked as high hazard areas. Privately insured flood damage is likely to be auto and non-commercial property, but only a small fraction of commercial properties have flood cover. Our industry database that we use to estimate losses is our estimate of how many properties — homes, commercial structures, autos — there are in individual zip codes. For flood you also must estimate how much of that is insured by the private market. That is still highly uncertain. We added another $3.3 billion for the private insured flood damage, but that is still a question mark, and we don’t know how it will pan out.

Q: What are the lessons learned?

A: Ida demonstrated what we expect from climate change in two areas. One it was another Category 4 storm. In 2020 we had Hurricane Laura, which was a Cat 4; in 2018 we had Hurricane Michael, which was Cat 5. So, in the last four years we’ve had a Cat 5 and two Cat 4s, which is a very high frequency. Normally you would only see a Cat 4 once every five or six years. It has been a lot more frequent, and that is exactly what scientists expect under climate change. Scientists also believe there will be more inland flooding from these tropical cyclones under climate change. The lesson here is that a hurricane model needs three perils. It used to be two: wind and storm surge. Now you need wind, storm surge and inland flood. The latest KCC U.S. hurricane model, version 3.0, does include all three perils.

Q: How are catastrophe models helping companies manage climate change?

A: We’ve already had a significant amount of climate change. From 1900 to 2020 there has already been a 1.1-degree Celsius temperature increase. The new Intergovernmental Panel on Climate Change report, a consensus work of scientists around the world, projects multiple scenarios. The best-case scenario by 2050 is that the temperature will increase another .4 degrees; expected middle of the road is .9 degrees, and the worst-case is a 1.3 degrees Celsius increase. If we look out to 2050 and we take the expected case, we have already seen more temperature increase to-date than we are going to see from now to 2050. What that means is the models that insurers are using today need to incorporate a significant amount of climate change and the impact. The KCC reference models already include climate change that has happened to-date for hurricane, wildfire and flood and for severe convective storm and winter storm. We are also offering additional projected views of risk based on those three scenarios for 2025, 2030 and out to 2050. As we update our reference models, we will be incorporating climate change as it occurs. 

Q: Is it challenging to provide an accurate estimate of climate-driven loss?

A: It is challenging. Even if we had no change in the climate, these models are based on many scientific assumptions, and that is why if you look at KCC model losses versus other modeled losses the numbers will differ. It’s not because the science is different but because the scientific assumptions are different. KCC has been addressing climate change from the beginning in its models. As other modelers start incorporating climate change it’s very important to be explicit and transparent on the assumptions. 

Q: Aren’t new climate models simply giving insurers the ability to increase rates?

A: It is a gradual trend, and for KCC models climate change is already built in there if you want to look one or two years ahead for your pricing. I don’t think anybody is going to build rates today based on 2050. With KCC models that should be a few low-single-digit percent changes to go out one or two years ahead and not a radical change to the number. If any model comes out with a radical change to the numbers for the next one or two years either they have faulty assumptions, or their reference model has not included climate change, because the temperature change that we’re seeing year on year is very small.

Q: What about wildfire risk?

A: The older models have not been giving a credible view of wildfire risk. As the climate changes and more people move into the wildland-urban interface it’s very important that the assumptions in the wildfire models are transparent and explicit, so all the users know exactly what is driving the loss estimates. It can’t be a black box.

Q: What’s next on the modeling front? 

A: A lot of the focus over the next couple years will be on more frequent model updates and ensuring that climate change is being built into the models appropriately. KCC is also expanding globally. We cover about 50 countries now, but we are working on more global coverage, particularly for our flood, severe convective storm and wildfire models. Other modelers are working on cyber. At KCC we feel that natural catastrophes still need a lot of attention because the models in the past have not been as accurate as insurers need, so we will be focusing on the bread-and-butter perils — hurricanes, earthquakes, wildfires.