[cptskippy]

I think you're spot on.

I worked for an Auto Insurance Broker in a former life and they had a state come down on them pretty hard for discriminating against minorities. At the time it was pretty common to jack up the rate factors on particular minorities for a pretty legitimate business reason.

Instead of using race as a rate factor they switched over to a combination of zip code, income, and vehicle age which encompassed pretty much the same demographic but actually ended up improving their overall loss ratio.

I would hope risk analysis has evolved since those days but if not there's a huge opportunity for any aspiring data scientists out there who wish to become actuarials.