No, as I said, Sheila's Wheels, (as part of Esure), is a direct insurer, who do not offer their products through brokers, which essentially, is all a price comparison website is, an automated broker. They seem to do this, primarily, so that you *can't* work out that Joe Q. Nobody Insurers who don't have an epic, epic marketing budget has a more appropriate policy for you, personally.
So, Sheila's Wheels, who have *always* sold policies to men, and have often been cheapest for them anyway, but as they don't appear on comparison websites, they work on men assuming that they can't get insurance from them, when in fact, they absolutely can. I imagine what will happen now is, their prices will go up, somewhat, but the gender bias of their customer base won't change much.
The whole thing with the Man/Woman, 18-25 vs 25-60 question is not that they're spectacularly accurate indicators of risk, it's that they're spectacularly easy to measure. For instance, one major contributory reason why women have fewer accidents than men is that on average, summed over the whole population, a woman is less likely to be driving home on busy roads, just as darkness is falling, because on average, a woman is less likely to have a full time, 9-5 job.
Now, an insurer cannot with any hope of a reasonable, truthful answer ask you "over the last 12 months, how many miles did you drive during twilight, in rush hour?" Everyone's estimate of that will be different, and everyone will lie anyway, if they know there's no way the insurer could possibly check. They can, however, prove what gender you are, and so they figure "hey, we can't get causation, we'll have to settle for correlation."
So, if you stop the insurance companies making these spectacularly broad rule of thumb judgements, they'll have to do something else instead. One thing that's been touted is a GPS system that actually bothers to measure when and where (and to a limited extent, how) you drive; Your behaviour, rather than your gender, determines how much of a risk you are. Accept the system in year one of your insurance contract, in exchange for a discount, then your behaviour in that year will drive the cost of the renewal.
Well, asking whether or not you drive to work is one of the things you can quote on, and one of the things they can issue endorsements on. Two of the standard responses for main usage are "Social, Domestic and Pleasure" and "Social, Domestic and Pleasure (excluding commuting.)" The latter is obviously cheaper.
I seem to remember some kind of fashion for "no driving after 7pm" policies not so long ago, primarily to create a cheaper policy for the 18-25s. I think they were something like fully comp policies which revert to being Third Party, Fire and Theft after 7pm; still technically legal to drive, but much lower cover.
I suppose that this change will see a range of similar excesses, endorsements and so forth come into existence.
I think I'd be concerned that third party insurance shouldn't become too set-about with caveats, and ideally no kind of insurance should be so complex that the driver ever has difficulty working out whether they're fully-comp covered or not at any given moment.
You don't want people forced to choose between abandoning their vehicle or breaking the law, because they're stuck in a traffic jam at nearly midnight on 30th April, and their insurance is only good when there's an R in the month.
It wasn't clear to me that "direct insurers" never by any means appear on price comparison sites. Presumably only brokers can access quotes databases anyway, and most (all?) comparison websites in any case only list insurers that pay them commissions.
"hey, we can't get causation, we'll have to settle for correlation."
Of course, and driving in twilight isn't pure causation either. Actuaries work with conditional probability and conditional expected values. They don't try to get to know their customers well enough on an individual basis to predict whether they're going to fuck up on a particular evening and fail to spot a cyclist, and as you say they can't draw conclusions from things that are easy to lie about, or easy to change your mind about.
If there is a correlation, then it will affect the conditional probabilities regardless of the cause. Insurers probably *could* reasonably ask whether you typically drive to work or not, much as they ask where you typically park.
Why only a limited extent? Why not wire the thing up to all the car's controls? Then the only limit is what they can interpret from anything the driver does to the car. For that matter, maybe they can get information from what you're listening to on the radio.
I just mean limited in "I can tell you drove at these speeds, I can tell you brake this hard, I can tell you drive this far at a stretch, and this often." I can't as readily make subjective judgements like "you cut that guy up", or make a distinction between "You broke well there to avoid an obstacle" and "you left it a bit late braking for that junction".
In the end, you're right, if you can add a sensor for it, you can probably score someone's driving on it in one way or another.
So, Sheila's Wheels, who have *always* sold policies to men, and have often been cheapest for them anyway, but as they don't appear on comparison websites, they work on men assuming that they can't get insurance from them, when in fact, they absolutely can. I imagine what will happen now is, their prices will go up, somewhat, but the gender bias of their customer base won't change much.
The whole thing with the Man/Woman, 18-25 vs 25-60 question is not that they're spectacularly accurate indicators of risk, it's that they're spectacularly easy to measure. For instance, one major contributory reason why women have fewer accidents than men is that on average, summed over the whole population, a woman is less likely to be driving home on busy roads, just as darkness is falling, because on average, a woman is less likely to have a full time, 9-5 job.
Now, an insurer cannot with any hope of a reasonable, truthful answer ask you "over the last 12 months, how many miles did you drive during twilight, in rush hour?" Everyone's estimate of that will be different, and everyone will lie anyway, if they know there's no way the insurer could possibly check. They can, however, prove what gender you are, and so they figure "hey, we can't get causation, we'll have to settle for correlation."
So, if you stop the insurance companies making these spectacularly broad rule of thumb judgements, they'll have to do something else instead. One thing that's been touted is a GPS system that actually bothers to measure when and where (and to a limited extent, how) you drive; Your behaviour, rather than your gender, determines how much of a risk you are. Accept the system in year one of your insurance contract, in exchange for a discount, then your behaviour in that year will drive the cost of the renewal.
Reply
(The comment has been removed)
I seem to remember some kind of fashion for "no driving after 7pm" policies not so long ago, primarily to create a cheaper policy for the 18-25s. I think they were something like fully comp policies which revert to being Third Party, Fire and Theft after 7pm; still technically legal to drive, but much lower cover.
I suppose that this change will see a range of similar excesses, endorsements and so forth come into existence.
Reply
You don't want people forced to choose between abandoning their vehicle or breaking the law, because they're stuck in a traffic jam at nearly midnight on 30th April, and their insurance is only good when there's an R in the month.
Reply
"hey, we can't get causation, we'll have to settle for correlation."
Of course, and driving in twilight isn't pure causation either. Actuaries work with conditional probability and conditional expected values. They don't try to get to know their customers well enough on an individual basis to predict whether they're going to fuck up on a particular evening and fail to spot a cyclist, and as you say they can't draw conclusions from things that are easy to lie about, or easy to change your mind about.
If there is a correlation, then it will affect the conditional probabilities regardless of the cause. Insurers probably *could* reasonably ask whether you typically drive to work or not, much as they ask where you typically park.
Reply
Why only a limited extent? Why not wire the thing up to all the car's controls? Then the only limit is what they can interpret from anything the driver does to the car. For that matter, maybe they can get information from what you're listening to on the radio.
Reply
In the end, you're right, if you can add a sensor for it, you can probably score someone's driving on it in one way or another.
Reply
Leave a comment