Posted On: January 2, 2007, 6:31 pm CST

Posted By: get real

Ok Ray you and Einstein seem to think certain things can be used as predictors and others can't. Which are they? It seems clear that credit is not. More because you don't understand the science behind predictive models than anything else. Zip seems ok, age seems ok. You say that credit isn't 100% but then you dispute when LLCJ and I say what about young drivers what about those with violations but no accidents. That is all about predicting losses but it isn't 100%. So how is it different? No rating variable is 100%

You say that driving record is much more reliable but that only shows how little you understand about rating and segmentation. In fact I would give up CLUE reports and MVRs long before I gave up credit reports. @80% of the market is clean but over 90% have a credit report. It is much more predictive and reliable.

What about model year on liability? Your argument would say it doesn't matter the year of the car for liability right? But it does! Model year is a predictor of future losses even for liability.

There are hundreds of competitors in this market place more than for any other product I can think of. Here is the bottom line bare bones facts. If someone can sell insurance for less they will. So why would companies choose to over price customers if someone else would just come in and write them for less and make money?

The answer is that the companies that don't use credit are losing their shirts from adverse selection. I have seen it over and over again.