Auto rating by territory

[jerrytuttle]

Hi. Please remind me what the thinking is about this:

Suppose I am principally garaged in suburban territory A, I have an accident with someone who is principally garaged in suburban territory B, and we have an auto accident in crowded city territory C where we are both declared equally at fault.

So my claims experience goes into the ratemaking bucket for territory A, the other driver’s claims experience goes into the ratemaking bucket for territory B, but shouldn’t the claims experience really go into the ratemaking bucket for territory C? Why don’t we do this? Do we code territory of occurrence?

By the way, I once saw an enormous claim under Michigan no-fault, by an out-of-state driver who had the accident in Michigan. The carrier paid the Michigan claim, but the carrier was unable to get reimbursement from the Michigan Catastrophic Claims Association Fund because the carrier was not licensed in Michigan. (Yes, I know the law has changed on mandatory buying unlimited no-fault.)

[magillaG]

It seems like if policies were rated by taking a weighted average if the roads a person drove on, then the loss should be allocated to city C, or even the specific spot in city C. We could imagine using telematics to do this in a particular kind of world (maybe an ideal one from an actuary’s point of view).

However, if rating is done by garage, then this doesn’t work anymore. Essentially we are hoping that people with garages close to each other drive over the same roads, roughly, and that this serves as a proxy for the rating that more exactly captures risk. In that case, the losses need to be allocated to each garage.