Creative Use of Data + Experience + Curiosity = Moving Past the Status Quo to Unimagined Solutions for identifying fire losses.
Many carriers are currently challenged by significant increases in the frequency and severity of fire losses in their personal and commercial books of business. MCA’s client was having similar fire-loss challenges with their Homeowners book of business but was struggling to find any insights within a $20M line of business.
To determine the root cause of the trends, the client had been using traditional methods to review results. As has been done for decades, these included looking at historical loss ratio analyses based on univariate cuts of rating variables—construction type, territory, credit, etc. In addition, the client had looked at individual underwriting files and completed large loss reviews within the claims organization to determine the drivers of the deterioration in results. There were no discernable trends developed.
The MCA Way — A New Approach
Starting with a similar approach, MCA began with five years of policy-level data and matched that information up with historical-loss information. To supplement the internal premium and loss data, MCA then partnered with HazardHub to easily append their 900+ variables at the address level. This approach provided a rich data set from which to apply advanced analytics to determine additional correlations.
Results — Revealing the Root Cause
The analysis revealed that the additional 3rd party variable that was most correlated with a poor loss ratio was a D or F from HazardHub’s “Property Fire Score.” The relative loss ratio on this small portion of the book was 2 to 3x worse than the remainder of the portfolio.
In addition to clearly identifying the worst-performing risks, the beauty of using this predictive modeling score was that it was highly intuitive. Underwriters have always been looking at the three key variables utilized—distance to the firehouse, the water source, and the makeup of the fire department—but not necessarily consistently. When underwriter experience and the model align, the normal challenges around implementation typically disappear quickly. Proactively identifying and addressing the few outliers on renewal and limiting any new writings significantly improved the overall performance of the portfolio.
One challenge that most companies in the P&C Mutual marketplace have historically faced is the amount of data that is available to them due to the size of their books. With the explosion in 3rd party data that is now available, this has become less and less of a constraint. The challenge we now see in many of these instances typically comes down to a few key themes—ability to match addresses in rural areas, the cost, and the speed and transparency of results returned.
Working with HazardHub was entirely different.
- Match rate – Based on the client’s addresses, the match rate was 99.75%, which allowed internal data to be supplemented with HazardHub’s 900+ address-level attributes.
- Cost – Most 3rd party vendors have minimum thresholds for these data projects. While $20,000 may be reasonable for large insureds to do POC work, the economics don’t work for smaller companies. HazardHub’s cost structure for R&D efforts was far more economical.
- Speed – MCA was able to build an API with HazardHub to append the 3rd party data overnight. Traditionally, this is a process that takes months.
- Transparency – With predictive model scores being all the rage, these “black box” scores can be intimidating to smaller companies. HazardHub took an entirely different approach…within a day, the Founder, John Siegman, and Chief Science Officer, Brady Foust, were on a Zoom call with us to explain the variables utilized and their approach to building the scoring models.
As we continue to work with many different clients in the Mutual marketplace, we are incredibly excited about the capabilities that HazardHub now provides to companies of all sizes.
Solutions for a Changing Landscape
The industry is changing, and traditional solutions may leave you falling behind. MCA can assist your Mutual company find hidden opportunities in your data so you perform steadily for the long term. Contact us at www.mutualcapitalanalytics.com so we can enable you to serve your customers and communities better and improve your business performance.