It has been fascinating watching the progression of the forecasted path for Hurricane Joaquin — what a perfect example this is of the importance of a modern data and analytics platform!
The big news is that the hurricane is not expected to make landfall on the East Coast of the U.S., but the new forecast depends as much on analytics and big data as it does on actual changes in the storm’s path. The spotlight is now on the European Center for Medium-Range Weather Forecasts (the European model) vs. the Global Forecast System (GFS) run by the National Weather Service. The New York Times has a great article discussing the reasons for the changing forecast and, crucially, the differences between the two models.
This is an excellent lesson for insurers to see the power of modern data and analytics in action and what happens to models when they are not using the advanced capabilities available today. Fortunately, investment in analytics continues to rise, as detailed in SMA’s recent report, Maturing Technologies in Insurance. Almost three in four insurers are increasing their investment in analytics over the next three years. 48% of P&C insurers, in fact, are planning to increase their analytics investments by more than 10% annually during that time.
In recent conversation with key CAT modelers, we have learned that they are working to use their weather data and insights at a more granular level than ever before in coming releases. The advance of these CAT model tools creates opportunities for insurers in search of better predictive capabilities on weather events. An upgrade to the GFS model has been planned by the end of the year, taking advantage of soon-to-be-available computing capacity. Once it is up and running, it will be interesting to see how the upgraded GFS model compares with the current European model, especially when applied to future CAT events.
Insurers can take the continuing story of Hurricane Joaquin as a wake-up call — not only is analytics a critical area for investment, but the quality of the information and the computing capacity available have a major impact on how useful predictive modeling can be.