An Introduction to Basic Credibility

In my talk, I will try to reinforce and expand on the ideas Gary Dean presented in his talk. I will start off my talk by using the following set of graphs taken from my "Student's Guide to Buhlmann Credibility and Bayesian Analysis" to illustrate some simple credibility ideas in terms of experience rating or individual risk rating. The goal of experience rating is to use an individual risk's experience were observed to be worse than average, we would predict his future experience would also be likely to be somewhat worse than average. Therefore, we would be likely to charge this insured somewhat more than average. As mentioned by Gary Dean, credibility quantifies how much worse or better an insured's future experience is expected to be based on a particular deviation from average observed in the past. These graphs should illustrate some of the ideas Gary Dean mentioned, such as why more weight is given to an individual's experience in certain situations. Also, those of you familiar with linear regression should see much that is familiar. (With the widespread use of personal computers, anyone can do a linear regression.)