Data Insights and Fire Claims
California wildfires contributed heavily to making 2017 an expensive year for catastrophe claims, burning more than 10 million acres and resulting in a record number of fire claims, accounting for 30% of all fire loss costs. While 2018’s numbers have yet to be finalized, the recent Camp and Woolsey fires indicate that the trend is only worsening.
Tracking the trends
Fire peril is increasing sharply in both severity and proportion of catastrophe claims. For example, according to the LexisNexis Home Trends Report, nationwide claim severity increased by nearly 20%, and by more than 70% in California from 2016 to 2017. Additionally, more than 70% of fire claims in the state were due to catastrophe claims. Loss cost in 2017 increased nearly 22 points compared to 2016 levels. Seventy-seven percent of homeowners’ compensation for fire losses between 2012 and 2015 was for claims exceeding $100,000.
These patterns suggest that property and casualty carriers need to rethink their fire-specific products with more effective geographic and fire-data-driven approaches. That’s because unlike other insurance segments, current practices for scoring fire risk have not kept up with the information age and rely too heavily on expert opinions or simple business rules.
Data-driven products have the ability to go beyond the capabilities of expert underwriters to obtain an optimal solution with consistent empirical results. This type of scoring approach could also achieve consistent results without the expense of a large field staff tasked with inspecting fire stations and fire hydrants.
More accurate pricing can be achieved with new data sources, such as geographic factors, which take into consideration issues like the proximity to underbrush and wildland fuels, the slope of the terrain and other factors. Unlike other types of loss for homeowners, fire losses rely on the community’s fire response capabilities. To quantify insurance risk, carriers must be able to account for how fast a fire department can respond.
One way is to assess true driving time: mapping routes between fire stations and addresses, and considering types of roads and realistic conditions, such as the ability to slow down through intersections, construction and traffic patterns. This quantifies a realistic, best-case response time for a station. The data should also include factors that affect response effectiveness, such as volunteer fire stations or fire departments spending most of their time responding to non-fire medical events.
Another factor to assess involves regional loss statistics, because fire departments and areas that performed poorly in the past tend to perform poorly in the future. Regional loss statistics can be used to create a class of geographic predictors of fire loss that are directly related to how well insurance losses are mitigated.
How to get there: Fire response score
But utilizing all these new data assets isn’t enough on its own; carriers will also need to acquire enough loss data to precisely model the peril of fire. Some data vendors have introduced fire response score models for greater pricing accuracy. An effective predictive model can provide carriers with a simple 1–10 score, with base fire loss cost relativities ranging from 40% better than average to 80% worse than average.
For credible results with a 1–10 score, tens of thousands of fire claims are needed (based on a sample of 189,000 claims). Scores built on credible data can identify a significant number of policies as being unprofitable for fire risk. At the same time, scoring also allows carriers to identify properties located in lower risk areas and to price the risk appropriately for those owners.
Utilizing data-driven products provides multiple benefits over the current practices for scoring fire risk. They allow for a more accurate assignment of premium and loss because they do not rely on data tied closely to tax revenue and property values. Rather, by leveraging data that is predictive of loss — such as true driving time, distance to high-risk vegetation, actual response time and regional loss statistics — carriers can more accurately assess risk and price policies profitably.
As the risks increase and change, so too must the tools used to evaluate them. Data-driven products are another tool in the carriers’ toolbox to help them make more informed underwriting decisions and offer better rates for their risk exposures. Companies choosing not to take advantage of this new option could find themselves at risk of poor decision-making and increased risk for their portfolio.