Predictive Modeling, Predictive ?

We are in the result stages of how predictive modeling use by insurance companies, is panning out. The results are looking pretty good for insurance companies. Many who use it are reporting underwriting results that are 40% better than prior to its use. This is a significant number. What this translates into is more profit. However on the street level it is a hard thing to explain to customers. Many times we get premiums that are surcharged because the predictive model picks up a negative criteria based on the insureds financial history. Many times this is not "predictive" in the eyes of us agents who know the clients. For example, I had a client who was going through a nasty divorce which his credit score was negatively impacted. His business was running like a clock and his loss history was fantastic. However his "predictive model" developed rates that were surcharged because the computer picked up his poor credit score and modeled his business as a higher risk. This was definitely not the case, but the computer metrics can't figure out that his divorce was casuing his credit score to fall. This is were predictive modeling is flawed and that is when it comes down to really knowing the customers. As long as the insurance carriers will allow agents input when "modeling" insureds, then it can work. The system can work better when the insurance companies allow agents to put a human touch into the predictive model. Until next time be careful out there and know you risks. K

Popular posts from this blog

Delaware Legislature Preparing To Revisit Dram Shop Liability

Get employees back to work quickly to keep work comp costs down

Important to notify insurance company of potential liability claims