How Car Insurance Companies Explain Racism

I used to work for a very small auto insurance company. Specifically, I worked on the underwriting software the company used to rate its customers. The rating each customer gets then determines the premiums they paid for insurance. And there were plenty of things that affected each customer’s rating. The big ones were age, gender, marital status, driving history, and the zip code of the customer’s primary residence.  

After each customer gets rated, the highest “rated” (read: highest liability) driver gets assigned to the most expensive car available on the policy. So that means if there’s a BMW 5 series and a Honda Civic on the policy and there’s a 16 year old male (statistically the most risky) and a 44 year old female (the least risky), the teenager gets assigned to the Bimmer while his mom gets the Civic, regardless of who drives what in real life. Or if you have two vehicles but just one driver on the policy, that driver gets assigned to the more expensive one as if he only drove that vehicle.

The reason has to do with minimizing liability. It would take a lot more time and effort to verify that each driver on the policy is driving the car they say they would drive. And that would cost the insurance company more money. So instead of that, they just assume the worst case scenario and then charge accordingly. All of this happens with the explicit approval of each state’s insurance commissioner, who reviews and signs off on the underwriting methods for each company, as well as any changes in base premiums.

This is also an inadvertent explanation for how racism works. Because insurance rating, like racism, is simply a calculation on how to treat different people according to a set of traits shared among groups of people. Racism deals with treating people differently on the basis of their skin color. Car insurance underwriting considers more traits than racism, but it’s still a form of discrimination by proxy.

If you think that it makes sense to group drivers into cohorts based on age, gender, marital status, zip code, driving history, etc, you’re using the same principle that racists use. Because you’re judging an individual driver based on shared characteristics of a “similar” cohort rather than the individual himself. It gives a slightly more negative connotation to the saw “birds of a feather flock together”, doesn’t it? 

We’ve all heard the stereotypes. Asians are good at math and bad at social interaction. Blacks are athletic and loud. Hispanics are ignorant and hardworking. Whites are entitled and naive. And nobody reading this would say we deserve to discriminate people based on their skin color. Imagine the public furor if FICO boosted the credit scores of white people and penalized black people. But we tacitly accept that it is proper to discriminate based on age, gender, and other observable characteristics.  

There are plenty of male teenagers who are very responsible drivers. Unfortunately they comprise a very small minority of male teenagers at large. The insurance company has no way of knowing whether one male teenager is more responsible than his peers at large unless it spent a large amount of time and money doing so. The end result is that the responsible male teenager gets lumped in with the others and therefore pays much higher insurance premiums than he deserves in an ideal insurance rating system.

Ideals are hard to implement in reality, though. Reality is sloppy, lazy, arrogant, and extremely judgmental because it’s created by people who are mostly sloppy, lazy, arrogant, and extremely judgmental. The fact of the matter is that everybody has mental shortcuts and uses them on a daily basis. When it comes to mundane things like calculating car insurance premiums, nobody has a problem with it. 

But when it comes to highly charged topics like racism, those mental shortcuts have to be hidden considering that the penalty for honest thoughts on racial prejudice is pretty stiff. And that’s a shame because sunlight really is the best disinfectant. If you’re not allowed to talk about racial prejudice (except in the most politically correct of ways, which is inherently intellectually dishonest), then it gets harder to bridge reality to the ideal.  

I’ll end this with two thoughts:

1. The insurance company has an excuse for being prejudicial. It would be expensive and illegal for the insurance company to hire people to check on its customers’ lives to calculate the ideal insurance premium (the true expected value of the insurance policy plus operation costs plus a 3-5% profit).

2. People and institutions (which is really just another word for people) will continue to use discriminatory shortcuts/heuristics so long as they work, and well past the point where they don’t. Changing their attitude requires constantly proving them wrong. Unfair, but life’s unfair.