sheilas' wheels within sheilas' wheels

oddities of insurance

Insurance is odd stuff. I was reminded this morning of a strange event in European motor insurance, which arose out of a decision in 2011 that it was illegal, under the Gender Directive, to discriminate in pricing between men and women.

This came as a bit of a shock to the industry; it was considered, at the time, to be a very well established fact that women drove more safely than men; they have fewer accidents at slower speeds. There was even an entire sub-brand of one of the UK’s largest motor companies dedicated to this fact (older readers might remember “for bonzer car insurance deals, girls get on to Sheila’s Wheels”).

You can imagine the state of the editorials at the time, about political correctness going mad and that. I seem to remember at least some people in the industry thought that it was completely unsustainable to ignore this basic actuarial fact, and that the EU would either have to change their policy or see premiums jacked up to unaffordable levels. (In fairness, they were basically correct about the same argument applied to life assurance, where it simply isn’t possible to pay men and women the same annuity rates).

A funny thing happened, though. A few years later, the gap between the premiums charged for car insurance to men and women had grown much wider. When forbidden from using gender as a cheap and cheerful risk selection tool, the insurers started putting more weight on things like miles travelled, claims, convictions and occupational categories[1]. And it turned out that the previous gender discrimination policy had been nothing like discriminatory enough; women were much safer drivers, and hadn’t previously been getting anything like enough credit for it.

Nobody really knows why this happened; it’s widely suspected that the insurers had just got into the habit of shading the gender gap downward because “it can’t possibly be that much”, or because they were worried about incentivising couples to lie about who was the main driver. Or simply because pricing was strategic, and as long as nobody rocked the boat too much they could get away with it.

The conclusion I take away is that once more, a simple bonus-malus system (the “no claims bonus”) is not easily defeated, and also that most of the time, big features of datasets are big features that are easy to recover from correlated characteristics. (This is what makes me a big sceptic about futurists talking about genetics in health insurance, telemetrics in motor or AI in everything – if something’s big enough to be relevant to pricing, it’s generally really big and easy to find).

Envoi: The thing which got me thinking about it was that someone on the social media highlighted a passage in a recent UK Supreme Court decision saying that:

“In the case of sex discrimination, it is impossible to see how an assessment of the differential risk known to be posed by, say, women and men drivers, could possibly be made by reference to actuarial or other reliable data sources that had also to take account of certificated sex based on a Gender Recognition Certificate. There is no rational basis for thinking that having a certificate could make a difference to the risk posed by drivers of different sexes.”

Lawyers, I have noticed in my career, often have excellent general knowledge, but it is sometimes very out of date. Once a fact gets into a lawyer’s head, it is very difficult to shift it.

[1] Occupational categories are where pricing models sometimes throw up some real funnies. I once spoke to an actuary who had worked for a motor insurer that basically refused to cover anyone working in the theatre profession. I made some weak joke about show people, and he said that it was actually driven by the tendency of backstage staff to give lifts to visiting musicians, and that “you only need to have a claim once for a Fiat 500 with a Stradivarius on the back seat and it puts you off the whole industry”.