A couple of days ago, Ezra Klein wrote a post on "pay as you drive" insurance that I wanted to respond to, but I, er, forgot. Well, no time like the present!
The question is how you move towards such a system. Currently, low-mileage drivers subsidize high-mileage drivers. Progressive, for one, is rolling out a pay-per-mile scheme, and it's a pretty good bet that only low-mileage drivers will sign up. This might make the project unprofitable. Or it might spur to throw their lot in with low-mileage drivers, raising rates on the high-mileage drivers or off-loading them onto other insurers. This in turn might force the other insurers to move to pay-as-you-drive schemes. It's essentially the same risk shifting that happens in the individual health insurance market, where insurers price their product to advantage healthy enrollees and keep trying to drive out sick people. The difference is that we actually want to discourage driving, or at least make people pay for it, while we don't want to keep folks from getting necessary health care. This is that rare devious insurance cost shifting scheme that I actually like!
This, of course, highlights the reason that adverse selection is even nominally a problem in health insurance markets: the government works very, very hard to prevent firms from pricing to risk.
The main problem I see in the movement to pay-as-you-drive insurance is that so much of the benefit is a positive externality. Progressive gets moderately better profits with the ability to price discriminate more effectively, but I assume that if this experiment actually works, those profits will be rapidly competed away. Meanwhile, most of the benefits go to people who get them whether or not they have pay-as-you-drive insurance: people driving on less congested roads, breathing less smoggy air, enjoying the beachfront property that is not covered by rising sea levels. I'm happy to think that we might move to a new, better equilibrium, but the cynic in me has doubts.
Number of miles driven doesn't necessarily correspond to chance of getting in an accident. Driving in empty rural areas has less risk then driving in congested urban areas.
In addition to what Dave noted, I see other serious problems with this scheme.
1. The young and inexperienced sometimes drive many miles and sometimes don't, but have a very high accident rate involving expensive payouts.
2. The elderly often do not drive many miles, but create many minor accidents.
3. Long-distance drivers who spend most of their time on the freeway probably account for fewer accidents than high-mileage drivers who burn their time over numerous short trips.
Insurance rates are normally structured to reflect the risk to the insurer, as anyone who is or has been a young-adult male can testify, especially if they ever owned a vehicle classified as a sports car or a coupe. The insurers have correspondingly gotten very good at understanding what factors are more likely to invoke an expensive or frequent number of payouts, and price discriminate accordingly.
Insuring by the mile, however, requires some sort of effective monitoring program and it isn't clear to me that the mileage has a strong relationship to the insurer's risk. Many companies already offer a rate-reduced "pleasure/recreation" classification for low-mileage vehicles incorporated into an umbrella policy that already has a fully-covered commuter vehicle, although AFAIK these aren't usually monitored unless the insurer has reason to suspect abuse.
Don't most accidents happen within 5 miles of home? I believe it has something to do with people being more familiar with the area and therefore more careless. It sounds more like long distance drivers compensate short distance drivers to me.
Number of miles driven doesn't necessarily correspond to chance of getting in an accident. Driving in empty rural areas has less risk then driving in congested urban areas.
Sure, but I'm pretty certain that auto insurance companies in most states consider the location of the driver and charge less in empty rural areas with lower congestion and accident rates. There may be some states where this is prohibited by law-- California was considering such a change a few years back, but I didn't hear what came of it.
I also know that New Jersey had (and may still have) caps on rates in cities (and thus had the rural areas pay more) under their old extremely regulated insurance system.
aMouseforallSeasons--
Again, merely because an insurance company is considering looking at miles driven doesn't mean at all that they'll stop looking at location, or anything. And if you look at the original post or article, you'll see that Progressive is actually talking about installing monitors in the cars; you're right, many insurers do offer lower rates to those who drive but rarely, but the discounts are generally pretty low precisely because of the difficulty of verification.
Don't most accidents happen within 5 miles of home? I believe it has something to do with people being more familiar with the area and therefore more careless.
It has more to do, as I understand it, with the vast majority of driving being within 5 miles of home. Most accidents are close to home, but that doesn't mean that the accident rate is higher.
What I like is Mr. Klein's statement of the fact that "Currently, low-mileage drivers subsidize high-mileage drivers" as though this is a new discovery and only just being addressed by Progressive. State Farm was in part founded on the belief that rural farmers who drove their personal cars rarely (on weekends to go in to town, etc.) were lower risk and should pay less in insurance costs than non-rural drivers...in the 1920's. http://en.wikipedia.org/wiki/State_Farm_Insurance
Actually, rural driving is more than twice as dangerous per mile driven compared to urban driving, per DOT stats. Faster speeds, fewer street lights, animal risks, etc.
All that said, miles driven actually is a pretty good first-order proxy for accident probability. One can come up with all sorts of confounding factors, but the reality is that all else being equal, if you drive twice as far, you'll get in just about exactly twice as many accidents.
(Different Dave writing, BTW)
What is needed is a black box that collects data on driving style, places driven, driver condition, etc. via a combination GPS, camera+sound, accelerometers, and vehicle data (speed, acceleration, braking, no. of Gs pulled in maneuvers, etc.).
This information can be stored in the black box, and then analyzed by a built in program that assigns a score to the driver for riskiness and price the insurance accordingly.
The question is, are we ready for the same type of controls on commercial airline pilots yet?
Don't most accidents happen within 5 miles of home?
Probably, but IIRC most *driving* happens within 5 miles of home.
Yeah, from what I understand it's a statistical artifact; you spend much more time within five miles of your house than anywhere else.
You know what happened when the ethnic found out that most accidents happened within 5 miles of home?
He moved.
I've never understood why you insure the vehicle(s), instead of the driver. The vehicle doesn't drive itself.
"I've never understood why you insure the vehicle(s), instead of the driver. The vehicle doesn't drive itself."
I think the value of the vehicle being driven might be a factor in insurance.
"The main problem I see in the movement to pay-as-you-drive insurance is that so much of the benefit is a positive externality."
I may be a little naive here, only being an economics undergrad, but isn't that the good thing about free markets? Competition allows you small windows of economic profit, enough to incentiveize a more efficient method, then your competitors adopt the more efficient method and force your prices down. Firms get normal profits, and we get positive externalities.
Yeah, I'm with Peter. I utterly fail to understand what Megan is talking about there.
No, if we're talking about what happens "in" the market, "we" get cheaper insurance. That's good for drivers, but it's not a positive externality, it's just competition working normally, as you say.
What she's saying is that the benefit of competition - lower insurance prices - will be pretty small. The main result will be less driving, which isn't "the good thing about free markets", since neither of the participants in the market (driver and insurer) want it. It's good for the people outside the transaction, who want less gas burned or less traffic, so it's a sort of side effect rather than primary goal.
My question is: you've got a large positive externality, and a possibly small-to-nonexistant expropriable benefit for the companies; they may not do it without a regulatory push.
When I was 16 it cost 300 bucks a month to fully insure me on a car that was basically rolling scrap metal.
Now it costs 50... for 500000 in coverage w/ a 1 million umbrella policy.
I drove then. I drive now. Jamming some people's insurance up by 50 bucks a year or whatever to compensate for mileage is really supposed to severely cut down on mileage? I'm not seeing the incentive.
If $4 a gas won't do it, marginal increases in insurance costs sure won't.
Re: Number of miles driven doesn't necessarily correspond to chance of getting in an accident. Driving in empty rural areas has less risk then driving in congested urban areas.
A proper pricing matrix would take this (and other factors, like the driver's safety record) into account. I doubt any insurance company would make miles driven the one and only rate factor.
Re: This, of course, highlights the reason that adverse selection is even nominally a problem in health insurance markets:
Health and driving are rather different beasts. Yoyr driving behavior is at least nominally under your control. You can choose how much you drive. You have only a very limited input to your health. Yes, uou can quit smoking and east better, but you hav no choice over your genetic inheritance and no one can change the biggest factor of all in health risk, their age.
Every auto insurer I've ever got a quote from asks for the mileage of my daily commute and the zip code where I park my car at night, so they are trying to factor in both the mileage and the traffic density in my location in calculating the rates. Of course, more accurate information such as actual mileage and exactly where you drive would be better, but would also invite more fraud. It's easier to turn back an odometer or to stick a GPS tracker into a friend's car instead of yours, than to fake a home address and employer's address in ways that would pass a records check.
Re: It's easier to turn back an odometer or to stick a GPS tracker into a friend's car instead of yours, than to fake a home address and employer's address in ways that would pass a records check.
You might not be able to fake the addresse, but you could easily fake claiming to work at home or ride the train several days a week when you don't.
"This is that rare devious insurance cost shifting scheme that I actually like!"
?? Govt forbid, everything isn't socialized.
"This is that rare devious insurance cost shifting scheme that I actually like!"
?? Govt forbid, everything isn't socialized.
>>Actually, rural driving is more than twice as dangerous per mile driven compared to urban driving, per DOT stats. Faster speeds, fewer street lights, animal risks, etc.