One of the frequent criticisms I hear about the use of territorial rating and credit reports in pricing and underwriting auto insurance is that such measures are not “fair”. Critics say that auto insurance rates should be primarily affected by whether one is a “good driver” as reflected by their official driving records.
I have two problems with such criticism.
First, so many of the traits that really define whether one is a “good” or a “bad” driver are not caught on MVR reports. Many drivers are on their best behavior when within sight of a local constable or a state trooper. Their true behavior is rarely seen by ticket-writing officials, or if it is, it is relatively infrequently (or at least inconsistently) ticketed.
Second…even when a ticket is written, there are antics that can be engaged in to avoid a record from flowing through the system. See, for example, this discussion at the Actuarial Outpost:
Around here it’s pretty common to go to the DA and “plea bargain”, which means you agree to pay double the fine and they knock it down to a non-moving violation which keeps it from affecting your insurance.
And the critics complain about credit data quality….
1 response so far ↓
1 AnElk // 1 Mar 2007 at 6:47 pm
How does your view of using credit scores in risk classification mesh with the AAA’s Risk Classification Statement of Principles (http://actuarialstandardsboard.org/pdf/appendices/risk.pdf)? For example,
Is it based on clearly relevant data?
Does it respect personal privavcy?
Do risks tend to naturally identify with their classification?
Does it bear a reasonable relationship to the hazard insured against?
At what level can the insured control their risk classification?
Is it susceptible to convenient and reliable measurement? (”critics complain about credit data quality…”)
Does it stay constant over the period that the classification is assigned? (Companies don’t rerun it on every renewal)
Is the cost of utilizing credit reasonable in relationship to the benefit received?