Fairness in Insurance Pricing
Behavioral pricing exploits what you’ll pay, not what you’ll cost. That may be fair in retail—but not in mandatory insurance.
When I wrote Pricing Insurance Risk (with John Major) and we appeared on a podcast with David Wright, I talked a lot about fairness in insurance pricing. One listener, an economist, pushed back: I never defined what I meant by fair, and surely, they said, all freely agreed prices in a competitive market are fair. That seems reasonable on the surface, but I don’t think it holds for insurance. This post discusses why.
Insurance isn’t like most products. It’s often priced based on private individual information and its purchase can be required by law. That makes fairness a complex and nuanced question.
Why insurance is different
First, price discrimination is built into the insurance product. The whole point of underwriting is to assess the buyer’s individual risk and reflect that in their price. We accept this—risk varies by person, so prices should too. Risk-sensitive pricing helps drive risk awareness and safer behaviors.
But insurers also have access to information most sellers don’t. Much of it is personal, and some of it feels off-limits. You can’t use race or sex to price a coat. Why should insurance be different?
Second, some insurance is mandatory. You can’t legally drive without liability cover. You can’t get a mortgage without homeowners’ insurance. That changes the ethics. It’s no longer a voluntary transaction between equals.
Three kinds of pricing
Here’s how I break it down:
- Dynamic pricing: changes based on time or supply. That’s fine. If you book a popular flight on a Friday night, you know what you’re getting into, and you’re the one getting home earlier.
- Price discrimination: reflects expected cost per buyer. In insurance, this is essential.
- Price optimization: charges more to those who’ll pay more—regardless of cost. That may be common in retail. But in compulsory insurance, it feels wrong.
The last one is the problem. You’re not pricing for risk. You’re exploiting behavior that you know only as a by-product of underwriting.
What makes a rating variable fair?
Some rating factors are clearly relevant. Take auto insurance. A clean driving record? Absolutely. Good credit? At first, it seemed unrelated. But research shows a link between financial and physical risk-taking [1], and now it’s widely accepted.
Other variables—like race, sex, or age—are generally banned. Some jurisdictions such as the EU require “unisex” pricing. Age is trickier, and often “years licensed” is used as a direct measure.
Then there are grey-zone variables. How often you shop around. How long you stay with one insurer. Your online behavior. These may predict claim costs—but also willingness to pay. And that’s the problem.
Fairness and the WSJ test
Here’s a simple test: imagine your pricing algorithm on the front page of the Wall Street Journal. Would you be proud of it? Would your customers accept it? To be clear, I’m talking about the variables used and their importance to the price, not the actual factors and relativities.
That’s how I think about fairness. Not a legal definition. A societal one. If it feels exploitative, it probably is.
What should regulation look like?
I’m not in favor of price regulation. Let the market set rates. But regulate the inputs, not the output. And regulate them through disclosure.
Insurers should be required to disclose:
- The variables they use in pricing,
- How much each variable contributes to variation in premiums (not coefficients, just explanatory power, similar to a California auto filing).
That gives away very little—competitively—but reveals a lot about fairness. And it would allow third parties such as aggregators or quoting sites to build transparency tools consumers can actually use.
Compare that to today’s approach: detailed, technical rate filings that are a dance of obfuscation between company and insurance department actuaries. Only competitors read them!
Final thought
Markets are powerful. But they don’t guarantee fairness—especially when the purchase isn’t optional and the seller knows everything about you. In those cases, we need guardrails. Not to fix prices, but to make sure we’re charging for risk—and not for risk-irrelevant behaviors.