Statistical discrimination
Which of the following is OK?
* Because women drivers (as a group, on average) demonstrably have fewer accidents than men, charge women less for car insurance.
* Supposing that people living in Britain but with French nationality could be demonstrated (as a group, on average) to have fewer accidents than those of British nationality, to charge French drivers less for car insurance.
* Same thing, black drivers.
* Same thing, those who live in an area lived in by a lot of black people[*]
* Same thing, those who attend a Christian church service at least 26 times per year. (Of course on the basis of non-discrimination, before allowing this you'd demand that the insurance company also checked the profile for other religions, and perhaps for those who are in some sense convincingly atheist).
* Same thing, those who own a large home[**].
* Supposing that sex is a factor in life expectancy, incorporating it into the life expectancy used when selling an annuity.
* Supposing that women (as a group, on average) could be demonstrated to remain in their first job for fewer years than men, to offer lower initial salaries to female recruits.[***]
It seems to me that insurance is a field which is almost entirely about profiling people according to whatever demographic or other information you can possibly lay your hands on, in order to compute the most accurate possible odds on something bad happening to them. In other words, discrimination.
I'm pretty sure that insurers don't prefer to charge women less because they subscribe to some biological or psychological theory about gender disparities in aggressive or risk-taking behaviour, or because they have some quaint notion that women only have the car one day a week anyway to do the big shop, or because they think women are "better drivers". It's because a statistician with records of every claim ever made against the company, told them that female customers are cheaper to have on the books. [****]
Changing the price of insurance probably won't change what the statistician sees. I suppose it's possible that men have more accidents because they pay more for insurance, that that the disparity will vanish once the prices change. I can't quite see the mechanism by which that would occur.
Employment, on the other hand, is an area where we expect people to make decisions based on relevant factors directly displayed by the individual, not demographic averages. So we don't think much of employers who say things like, "well, women might quit to have kids, and so we don't really want women working here", and instead we wittily riposte, "if you weren't such an ass-hat to women, maybe your female employees wouldn't quit".
Is it definitely clever though to move insurance away from gender discrimination? Or looking at it another way, since insurers already do also take into account a lot of individual factors such as past driving record, is it clever to provide car insurance in a way slightly closer to the NHS model that you pay based on a "fair" contribution, and slightly less based on the statistical risk taken by the insurer, given all information available?
Maybe it is clever, especially when the factor under consideration is one that the individual cannot change. In which case what about annuities? If you have a genetic trait that will most likely kill you by 75, thus making you cheaper to provide an annuity to, should insurers be permitted to offer you a bigger annual payout for the same price, or is that unreasonable discrimination because you *might* live just as long as anyone else? Is a rare genetic trait ethically different from a very common genetic trait (sex)?
On the other hand, maybe it's dumb, and if prices are indeed equalized then insurers will suddenly want female customers more than they want male ones, because they'll be more profitable. But they won't be allowed to charge them less. So if you're a new driver, expect to see a lot of "free" crap coming with your car insurance, that insurers think women will probably want, and that men will pay for in return for the fact that they still cost more to insure. Because a court can't actually stop people crashing into stuff, even in the name of equality. And if you're an experienced driver, maybe expect more punishment for making claims, since insurers are being asked to discriminate more based on individual records and less on demographics.
Also, since I've just claimed that sex is genetic: does this ruling solve or create any problems for transgender drivers (or other insurees), that everyone should know about?
[*] In fact car insurance of course does discriminate by postcode. Postcode correlates somewhat with race, although I doubt that insurers take race directly into account when setting rates.
[**] This does tend to make your insurance cheaper, because typically it means you can park off-road, but is not the specific basis of the discrimination.
[***] A red herring really. Even if employers believe this (and I don't know how many do, just that it's the kind of thing I've heard in the past offered as a possible reason) this single factor on its own is not responsible for the whole mess of the gender pay gap. But let it stand for any hypothetical statistical factor meaning that, all else being equal, female employees do on average benefit your company slightly less than male employees for some reason which cannot be predicted with confidence for particular women and men at time of hiring.
[****] Or perhaps they're more price-discriminating, and all that's really happening is men are just as expensive but more profitable because they'll pay any old ridiculous fee. If the advice came from the marketing department rather than from the actuaries, then perhaps there's a very different argument to have, but I don't think it did.
* Because women drivers (as a group, on average) demonstrably have fewer accidents than men, charge women less for car insurance.
* Supposing that people living in Britain but with French nationality could be demonstrated (as a group, on average) to have fewer accidents than those of British nationality, to charge French drivers less for car insurance.
* Same thing, black drivers.
* Same thing, those who live in an area lived in by a lot of black people[*]
* Same thing, those who attend a Christian church service at least 26 times per year. (Of course on the basis of non-discrimination, before allowing this you'd demand that the insurance company also checked the profile for other religions, and perhaps for those who are in some sense convincingly atheist).
* Same thing, those who own a large home[**].
* Supposing that sex is a factor in life expectancy, incorporating it into the life expectancy used when selling an annuity.
* Supposing that women (as a group, on average) could be demonstrated to remain in their first job for fewer years than men, to offer lower initial salaries to female recruits.[***]
It seems to me that insurance is a field which is almost entirely about profiling people according to whatever demographic or other information you can possibly lay your hands on, in order to compute the most accurate possible odds on something bad happening to them. In other words, discrimination.
I'm pretty sure that insurers don't prefer to charge women less because they subscribe to some biological or psychological theory about gender disparities in aggressive or risk-taking behaviour, or because they have some quaint notion that women only have the car one day a week anyway to do the big shop, or because they think women are "better drivers". It's because a statistician with records of every claim ever made against the company, told them that female customers are cheaper to have on the books. [****]
Changing the price of insurance probably won't change what the statistician sees. I suppose it's possible that men have more accidents because they pay more for insurance, that that the disparity will vanish once the prices change. I can't quite see the mechanism by which that would occur.
Employment, on the other hand, is an area where we expect people to make decisions based on relevant factors directly displayed by the individual, not demographic averages. So we don't think much of employers who say things like, "well, women might quit to have kids, and so we don't really want women working here", and instead we wittily riposte, "if you weren't such an ass-hat to women, maybe your female employees wouldn't quit".
Is it definitely clever though to move insurance away from gender discrimination? Or looking at it another way, since insurers already do also take into account a lot of individual factors such as past driving record, is it clever to provide car insurance in a way slightly closer to the NHS model that you pay based on a "fair" contribution, and slightly less based on the statistical risk taken by the insurer, given all information available?
Maybe it is clever, especially when the factor under consideration is one that the individual cannot change. In which case what about annuities? If you have a genetic trait that will most likely kill you by 75, thus making you cheaper to provide an annuity to, should insurers be permitted to offer you a bigger annual payout for the same price, or is that unreasonable discrimination because you *might* live just as long as anyone else? Is a rare genetic trait ethically different from a very common genetic trait (sex)?
On the other hand, maybe it's dumb, and if prices are indeed equalized then insurers will suddenly want female customers more than they want male ones, because they'll be more profitable. But they won't be allowed to charge them less. So if you're a new driver, expect to see a lot of "free" crap coming with your car insurance, that insurers think women will probably want, and that men will pay for in return for the fact that they still cost more to insure. Because a court can't actually stop people crashing into stuff, even in the name of equality. And if you're an experienced driver, maybe expect more punishment for making claims, since insurers are being asked to discriminate more based on individual records and less on demographics.
Also, since I've just claimed that sex is genetic: does this ruling solve or create any problems for transgender drivers (or other insurees), that everyone should know about?
[*] In fact car insurance of course does discriminate by postcode. Postcode correlates somewhat with race, although I doubt that insurers take race directly into account when setting rates.
[**] This does tend to make your insurance cheaper, because typically it means you can park off-road, but is not the specific basis of the discrimination.
[***] A red herring really. Even if employers believe this (and I don't know how many do, just that it's the kind of thing I've heard in the past offered as a possible reason) this single factor on its own is not responsible for the whole mess of the gender pay gap. But let it stand for any hypothetical statistical factor meaning that, all else being equal, female employees do on average benefit your company slightly less than male employees for some reason which cannot be predicted with confidence for particular women and men at time of hiring.
[****] Or perhaps they're more price-discriminating, and all that's really happening is men are just as expensive but more profitable because they'll pay any old ridiculous fee. If the advice came from the marketing department rather than from the actuaries, then perhaps there's a very different argument to have, but I don't think it did.