Tuesday, 1 September 2015

The magic of statistical analysis in insurance (or the weather needs to be more predictable)

My colleague and I were all set to go to West Palm Beach this week for the @LOMA #Lomatech conference. We were looking forward to participating in the inaugural event and were both speakers.

Then the weather intruded. Our friends at LOMA had a tough decision to make on Friday as Hurricane Erika was coming through the Caribbean. The storm had already done considerable damage and caused many deaths. Since the projected path covered the entire state of Florida, there was considerable concern. Once the governor of Florida declared a state of emergency for every county in Florida, LOMA made the only decision they could and postponed the event.

Why do I blog about this today? Because it is Tuesday, the day Erika should be decimating Florida, and I am sitting just outside Tampa enjoying a beautiful sunny day. Not a rain shower in sight.

My point? This is just one example of many of the challenges faced in the insurance industry. Our industry relies on an immense amount of data. We use historical data and averages and every kind of statistical analysis we can to ensure that our customers and the insurance companies are in a win-win situation (as much as they can be in the situation where they need their insurance).

This is a classic example of how all that analysis is wonderful, on average and over time, but can be completely inaccurate in a single event. Your mortality may suggest you will live well into your eighties, but you, as an individual, may not. My homeowners insurance, living in Florida, includes a significant cost associated with the storm that may never come. The county in which I live, Pinellas, which includes Clearwater and St. Petersburg, was last hit by a hurricane in 1921. Yet Tampa is number 4 on the National Oceanic and Atmospheric Administration (NOAA) list of cities most vulnerable for a hurricane. In fact, they state Tampa has an 11% chance for a hurricane every year.

That is the problem with probabilities. An 11% annual chance, but no hurricane for 94 years.

I find the entire process fascinating and continue to marvel at how well our industry does and how wonderfully they protect their customers. At heart, though, I still consider much of it to be magic.



from Celent Insurance Blog http://ift.tt/1LI3BKb
via IFTTT

No comments: