For decades, catastrophe (cat) models have been the backbone of underwriting, risk pricing, and reinsurance placements. But as weather extremes intensify and losses defy historical norms, insurers are questioning whether traditional models are still enough.
For Kim Roberts (pictured), head of North America peril at Guy Carpenter, catastrophe models are still indispensable, but they can no longer stand alone.
“Absolutely, there’s still a place for cat models,” Roberts told Insurance Business America. “They’ve been around for decades and hold tremendous institutional knowledge. But we now have a greater understanding of the uncertainty around their results. The future might not look like the past, and that’s what these models rely on.”
Roberts said the property insurance market’s abrupt recalibration in 2023 forced carriers and reinsurers to reassess how they quantify and price risk.
“We had a couple of years with a rather shocking rise in frequency and severity that caused a hard reset,” she said. “It was, ‘Okay, we need to re-evaluate what we’re doing; do the models we have actually reflect reality?’”
That reset led to tighter capacity and higher pricing across several peril lines. But as losses eased in recent years, Roberts said the market has begun to normalize, with reinsurers regaining appetite and innovative tools supplementing traditional cat modelling.
“Now, reinsurers are more willing to deploy capacity,” she said. “There are still hot-button perils and regions that cause concern, but we’re making headway.”
The shift has also given rise to what Guy Carpenter calls a more comprehensive “view of risk”: an approach that integrates catastrophe modelling with other data-driven strategies and underwriting insights.
In North America, severe convective storm (SCS) has been one of the hardest perils to model accurately. “The models we have don’t always align with real-world experience,” Roberts said. “But insurers are becoming more strategic about managing that risk, and reinsurers are responding.”
She noted that simple but disciplined aggregation management remains key. “You can’t have too much concentration risk, but you also need to marry that concentration with hazard data, not just cat model outputs.”
Guy Carpenter’s GCAT Severe Thunderstorm Risk Index provides one such alternative view, giving insurers a fresh lens on SCS exposure outside traditional model parameters. But Roberts emphasized that success goes deeper than accumulation control.
“Insurers need to know their data: roof age, roof condition, materials, siding, deductible structures... these details matter,” Roberts said, noting that strategies like excluding cosmetic hail losses or adjusting deductibles can “meaningfully reduce exposure.”
Wildfire is another peril where modeling challenges persist. Despite recent advancements from Verisk and Moody’s, Roberts noted existing models still fall short of capturing one of the most critical dynamics: how wildfires behave once they enter populated areas.
“Most models stop at the wildland interface,” she said. “They’re built to account for vegetation fuels, not structures themselves as fuel. But in California and other regions, we’ve seen how fire moves from wildlands into urban areas. That’s where the next modeling frontier lies.”
While the 2025 California wildfires did not result in large insured losses, they underscored the urgent need to model “fire in the built environment.” The good news, Roberts said, is that insurers are increasingly factoring in structure-level resilience, from defensible space to roofing materials, and even forming partnerships with wildfire mitigation companies that help policyholders prepare and protect their properties.
As the 1/1 reinsurance renewal season approaches, Roberts sees a more balanced and orderly market emerging. Capacity is returning, and reinsurers appear increasingly open to underwriting severe convective storm and wildfire risks, provided insurers can demonstrate a robust, data-driven strategy.
Still, she cautioned, the days of relying solely on a cat model output are over.
“Data is table stakes now,” Roberts said. “You can’t just show up to a reinsurance meeting with your cat modelling results. You need to bring your full story: how you understand your risk, how you’re managing it, and what you’re doing to build resilience.”
Ultimately, catastrophe models remain the cornerstone of modern underwriting; however, they are no longer the sole foundation. “These models are tremendously valuable,” Roberts concluded. “But combining them with new tools, new data, and a deeper understanding of risk… that’s how insurers are going to thrive in a changing world.”
