In an AAIS Webinar, FLAMES: The Modern Fire Protection Evaluation Model, AAIS Vice President Phil LeGrone, and AAIS Senior Risk Strategy Lead Matt Hinds-Aldrich, PhD, outlined the new fire model, FLAMES, and the advantages it offers over traditional models for underwriters assessing and managing non-cat fire risk.
An Urgent Need
According to Dr. Hinds-Aldrich, non-cat fire claims account for more than $25 billion in losses every year, and are responsible for household and commercial damage across the country. Fire risk models have attempted to predict fires for decades, however the historical assumptions about fire risks used by these legacy systems fail to account for variances that can impact expected losses (such as differing capabilities and claims severity across communities) and instead focus narrowly on factors such as distance to fire hydrants and fire stations.
Collaborative Research
The problem was clear. To solve it, AAIS drew on input from people across the fire service industry, including insurers and Community Risk Reduction (CRR) and fire department experts. One message heard loud and clear was the problem arising from the lack of variation in claims when using legacy fire models. As a result, carriers were “banding classes together to get a lift in their models.” Regulators were concerned as premiums weren’t changing when the classes did.
CRR experts noted that the current formula used by legacy systems, which centered on number of apparatuses, number of firefighters, and number of training hours, was not a true indicator of “potential success.”
Data To The Rescue
Fortunately, according to Mr. LeGrone, “There was a lot of data out there that could be thrown at this problem to better quantify the risk aspects related to fire protection around the total fire risk.” The AAIS team drew on numerous streams of data – including claims data, community data, fire data, economic activity and additional data such as weather and crime information. (Some of it generously shared by AAIS Members and other industry stakeholders looking to support industry innovation.)
Seeing the Difference
While current methods of measuring fire risk reflect fire protection capabilities within the geographic boundaries of individual fire departments, they ignore variances in capabilities and claims severity across communities. FLAMES, on the other hand, is a country-wide, severity-based model that works at ZIP code-level granularity, tuned for each line of business and adaptable for state-specific requirements.
In the webinar, Mr. LeGrone and Dr. Hinds-Aldrich showed maps of fire loss from both AAIS’s former linear model, and the FLAMES model.
Both models show fires that occurred in the Tampa-St. Petersburg areas of Florida over an eight-year period. In the legacy model, data was input and dots assigned to fires based on the model’s three buckets of fire protection: protected, partially protected, and not protected. All based on distances.
The FLAMES Model met regulatory and state-specific requirements and with zip code-level variations in risk. The presentation showed how AAIS’s FLAMES model is a much more robust approach to modeling fire risk.
The FLAMES model was initially built to focus on homeowners and countrywide models were developed. The model can be adapted to fit state requirements. So far, FLAMES has been officially filed in the state of Florida and as of January 1, 2021, AAIS had built it into its Florida Homeowner’s by Peril Program or FL HOBP.
Roadmap Ahead
FLAMES is an ongoing initiative with no endpoint. The FLAMES team is excited to evaluate data and feedback to consistently refine the model, and will be continually enhancing the model. Currently, the AAIS FLAMES team is working to expand the model to include dwelling properties and commercial lines.
Want to learn more? Contact an AAIS Engagement Manager, download our FLAMES white paper.
