Kevin's Weblog

I’ve also left the most complex of the state measures for last: Measure 42. (Or, at least, it seems like the most complex to me. Your opinion may vary.) This measure would ban the use of credit information in the pricing of any insurance product.

There’s a lot of history in the rhetoric around this measure. The author, Bill Sizemore, apparently has quite a history of pushing conservative ballot measures. He also recently got fined for some type of hanky-panky regarding signature gathering. Now, if signature gathering here works anything like it did in CA, I’m not going to hold that against him too much, because it just means that he got caught. The result is that much of the opposition to this measure has focused on demonizing these two characters, rather than debating the substance of the issue. There is an interesting question, though, of why this guy is pushing a measure that seems to be about consumer protection. Moreover, the only other person who submitted an argument in favor is Loren Parks, a billionaire living in Nevada with a history of big contributions to conservative causes. I had a conspiracy theory a couple days ago, but I forget what it was, and that’s probably for the best.

The financial backing for the opposition is coming almost entirely from big insurance companies, unsurprisingly. The arguments against are almost all variations on “your insurance bill will probably go up”, which is true since about 60% of people get a discount based on their credit history, while only about 30% pay more. Of course, less information for the insurance companies means more risk and less revenue for them; which is really their motivation.

Overall, this is a complex issue which isn’t really being fairly debated. The only reasoned discussion I have seen is from a curious actuary who decided to address it as an academic issue. (He’s not an Oregonian, so has no vested interest.) I basically agree with his analysis. I have no particular difficulty believing a correlation exists. Honestly, I think it’s more plausible than a conspiracy by the insurance companies to screw minorities and the poor. I also think the arguments about inaccurate credit reports are a bit of a strawman. Everyone can, and should be, getting free copies of their credit reports each year. Moreover, inaccuracies go both ways; I’ve noticed errors that look bad, which I make a point to correct, but I’ve also noticed omissions of some things which I’m not rushing to get added onto my report.

There is a large issue at hand here, though, which is what the future of insurance is in a world with more and more “personal” information available to companies. Financial histories, DNA analysis, automobile “black boxes”, video surveillance in public spaces; these are all examples which insurance companies would love to use to differentiate between customers. At some point, the whole concept of group insurance gets a little shaky. I don’t have a good argument where to draw the line here, but this may be an area where the concept of states as experimental testing grounds has merit.

Oregon has been conducting one such experiment for the past few years. The current laws allow credit information to be used for setting the initial premium on a new policy; but it may not be used to cancel, re-rate, or fail to renew an existing policy. As a Bayesian, I like this approach: use credit and other information to establish a prior, then going forward use the actual loss history for each period to refine the probability estimate. I’d like to keep this system in place for a while longer to see how it works in practice, so I’m voting against changing the system again so soon.