CatNet®: Redefining wildfire underwriting

Wildfires pose an escalating threat, consistently surpassing historical records. Since the year 2000 they have collectively consumed seven million acres annually – twice the yearly acreage in the 1990s.

This trend shows little sign of abating. In 2023, wildfires in Hawaii resulted in the biggest insured loss from any natural catastrophe event ever seen in the state, occurring in a region that was previously considered to be one of low catastrophe risk. At more than 1.2 million acres the Smokehouse Creek fire was the largest ever in Texas. Both highlight the urgent need for proactive measures to mitigate these risks.

Losses are only expected to grow in the coming years as global temperatures continue to rise and the pace of development in wildfire-prone areas shows no sign of slowing.

At the same time, the challenges facing insurers in assessing and managing these risks continue to mount.

This has become a predominant theme in California, which has more homes at risk for wildfire and nine of the 10 costliest wildfires in history. Several major insurers have chosen to non-renew a sizeable percentage of policyholders while others have paused writing in the state entirely due to increasing wildfire risks and soaring construction costs.

Industry concern over the decreasing number of insurers willing to offer coverage in wildfire-prone areas is rising, as homeowners and businesses are left with fewer options that are more expensive with more restrictive terms. So much so, that regulatory reforms are being considered in California that incorporate forward-looking catastrophe models and reinsurance costs into ratemaking and improve the rate review process, with the intention of better aligning pricing with risk.

Unfortunately, traditional risk evaluation methods have proven inadequate in the face of evolving wildfire threats. They rely on outdated assumptions and incomplete data sets that fail to recognize the full extent and complex interactions between drivers of wildfire risk and changes in extreme fire behavior.

Traditionally, models have been stressed by new fire behavior. Fires causing their own weather, crossing the Sierra crest, burning the same landscape in consecutive years, and entering urban landscapes are all recent examples of phenomena that were often not accounted for in modeling until well after they had caused major insured losses. At the same time, tools that can accurately and transparently forecast future wildfire risks have been hard to come by, largely because of challenges around "interannual" variability.

Interannual variability refers to fluctuations in the occurrence, intensity, and extent of wildfires from one year to the next. From weather patterns, natural cycles and the health and density of vegetation to human activities such as land management practices and urbanization, the risk and prevalence of wildfires is influenced by a range of evolving factors. In other words, something's changed in the past year and therefore the future is likely to be different.

Developing models that can assess this interannual variability and provide forward-looking insights into fire behavior have so far proven incredibly difficult – until now. It's become obvious that wildfire underwriting warrants an AI-first approach – a system that can handle a large amount of constantly-changing data about terrain, weather, and vegetation in a scalable fashion that reflects their complex interactions.

Using AI and comprehensive data harvesting, our collaboration with Bellwether enhances underwriting capabilities for insurers with an appetite for wildfire risk. Bellwether’s technology harnesses 600 layers of geodata from sources such as Google, including detailed information on canopy, vegetation, precipitation and wind speed. These robust datasets are then used to train machine learning models to accurately forecast wildfire risk in a specific local area.

This capability is integrated into CatNet®, Swiss Re's online natural hazard atlas, to help insurance companies offering property coverage in wildfire prone regions accurately assess their risk.

Supported by 20 years of historical data, this innovative approach yields physics-backed results and fresh insights on fire dynamics, unveiling neighborhood-specific risk factors and illustrating how past burns could influence future events. Consequently, these tools can forecast wildfire risk with precise accuracy for up to five years into the future.

Today, more than ever before, insurers require robust risk models based on high-quality data, rigorous analytics and advanced computing power to effectively anticipate and manage wildfires risks, mitigate losses and protect lives.

Through this partnership with Bellwether, CatNet® will be able to deliver precise risk assessments that can be used to identify mitigation strategies and lower the propensity for loss.

The ability to significantly improve predictability in a peril where risk has traditionally varied significantly year over year is unprecedented and sorely needed, providing underwriters with the confidence to make intelligent decisions, from accepting or declining submissions to setting pricing and terms.

At Swiss Re, we’re focused on continually innovating so that insurers may eliminate reliance on inadequate risk evaluation methods as they work to manage and mitigate evolving threats.

With smarter underwriting capabilities, portfolio performance can improve, and insurers can move from risk avoidance to risk management - reducing loss ratios and offering coverage to the insureds that need it, for the right price.