TNR BLOGS

July 03, 2009 | 7:55 PM
July 03, 2009 | 7:37 PM
July 03, 2009 | 7:12 PM

March 09, 2009 | 5:19 PM
March 09, 2009 | 5:16 PM
January 07, 2009 | 12:20 PM

July 01, 2009 | 10:33 PM
June 30, 2009 | 8:42 AM
June 29, 2009 | 9:09 AM

July 26, 2008 | 2:24 PM
July 23, 2008 | 1:55 PM
July 17, 2008 | 3:56 PM

July 03, 2009 | 10:13 PM
July 02, 2009 | 12:57 PM
July 01, 2009 | 7:02 PM
COLUMNISTS
TODAY'S STORIES
04.01.2009
The End of All-You-Can-Drive Insurance

Suppose that, instead of paying for gas each time they filled up, drivers paid a fixed sum every six months to cover their fuel costs. This sum would vary based on the driver's age, type of vehicle, and location, but wouldn't change with the number of miles driven. It sounds absurd, of course, but it's exactly what happens in the auto-insurance market. Accident risk—like fuel use—increases in roughly linear fashion with each mile a person drives, but insurance costs the same (or almost the same, since some insurers offer small discounts to drivers with self-reported low mileage) whether you drive 500 miles each year or 50,000.

It's a huge market failure, one that leads people to drive much more than would be economically optimal. The way to fix it is to make drivers pay for insurance on a per-mile basis. According to Brookings researchers Jason Bordoff and Pascal Noel (who describe their work in a Resources for the Future web commentary), getting all drivers to switch to pay-as-you-drive insurance would cause an 8 percent reduction in the total number of miles driven in the United States, roughly the same reduction that would result from a $1-per-gallon increase in the gas tax. Better yet, because a few high-mileage drivers account for a disproportionate number of miles driven, some two-thirds of drivers would actually see their insurance rates go down.

Bordoff and Noel have some common-sense proposals for encouraging pay-as-you-go insurance—most notably a tax credit to cover the initial cost to insurance companies of installing remote mileage-monitoring devices in customer's cars or setting up a network of odometer-inspection stations. But what they don't point out (at least, not until you read pretty far into their longish discussion paper on the subject) is that once a critical mass of drivers switched to pay-as-you-go, it would create a cycle of snowballing economic incentives that would pretty quickly lead most people to switch, even if they had to cover the initial cost of installing a mileage monitor in their cars.

Why? The first people to switch to pay-as-you-drive insurance would be the ones who drove the least. Because this would effectively remove the lowest-risk drivers from the pool of people buying lump-sum insurance, the prices for lump-sum insurance policies would have to go up. This would cause slightly higher-mileage drivers to switch to pay-as-you-drive policies, which would cause the price of lump-sum policies to rise even further, and so on. In the end, only the highest-mileage drivers would stick with lump-sum policies, and they would pay a price that reflected the true cost of providing them with insurance.

This process, better known as adverse selection, is what causes individually-purchased health insurance to cost a lot more than a comparable employer-provided plan. But the auto insurance market is one place where it could have a positive effect—and an impressively large positive effect at that. After all, the Lieberman-Warner cap and trade bill would have raised gas prices by only 25 cents per gallon, causing a much smaller than 8 percent decrease in driving. This means that in the short run, at least, changing the way that auto insurance is priced could have a much bigger impact on people's driving habits than getting a climate bill passed.

--Rob Inglis

Posted: Sunday, January 04, 2009 4:34 PM with 25 comment(s)

Comments

You must be logged-in to comment.

Not a subscriber? Click here to get a digital or print and digital subscription to The New Republic!

kme said:

The links in this piece are broken. Interesting topic, though.

January 4, 2009 2:03 PM

Brad Plumer said:

Should be fixed, now

January 4, 2009 3:08 PM

benjamin81 said:

This truly is the proverbial better mousetrap. I hope this gets implemented as fast as possible.

January 4, 2009 6:14 PM

srodar said:

An interesting concept but one that fails to account or give consideration to residents of rural areas who, out of necessity, must drive more miles. People who live in the Southwest and Rocky Mountain regions have no public transportation alternative and have to drive longer distances for schools, groceries and other basic needs. Even though more miles are driven in rural areas the accident risks are lower due to less traffic. So basing premiums on mileage may not be the best criteria for setting rates.

I know that insurance companies already have credits and surcharges for variables like mileage, type of car, age of driver, geographical location, distance from work,and other risk factors that are know to affect claims. Might be best to leave this to those who study risks more carefully and consider more than just mileage.

January 5, 2009 11:02 AM

reganad said:

I don't think this makes sense.  If the number of miles driven did, indeed, increase the risk, you can bet that that would be a factor in underwriting insurance policies already.  It seems to me that since the insurance companies do not take miles driven into account, or do not put much weight on the annual mileage when determining car insurance rates, then there is not a good correlation between annual mileage and claim rates.

This being the case, I don't see why insurance companies would make the change.  They can't significantly lower the price on a policy for a car driven, say, 3,000 miles per year if its risk profile isn't significantly lower.  

I can see why environmentally, it seems like a good idea to have these types of policies.  But practically speaking, or speaking from a business point of view, it seems like a gimmick.  The gas tax would be a better plan--then governments would have more money for infrastructure improvements, and the highest utilizers would still be paying the most.

January 5, 2009 12:57 PM

satyendra said:

" Even though more miles are driven in rural areas the accident risks are lower due to less traffic." Srodar, really? I was under the impression that the accident rate, i. e., the number of incidents per driver or per mile, was much higher in rural areas.  Yes, the sheer number of accidents must be lower.

January 5, 2009 1:00 PM

dhauck said:

srodar beat me to it, but that's what I was going to say, as well - not all miles driven are the same.  I would be hard-pressed to believe that a 5 mi drive in Wauseon, OH is as accident-prone as a 1 mi drive in Manhattan.  And as was pointed out, rural persons are forced to drive more miles by nature of their situation.  So companies would have to keep the currently used information, AND would now have to cross-reference it all with miles driven.  And drivers would have to stop in at a checkpoint two or four times a year to have their odometer read - I'm sure that would be just as quick and easy as the E-Check stations we Ohio drivers currently put up with.  

If insurance companies thought they could save themselves money by doing this, you'd better believe they would already be doing it.  The fact that they are not seems to indicate that they consider the other factors - age, region, history - to be far more significant than mileage.  Pressuring them to add this new criterion to their calculations, just so you can stick it to high-mileage drivers, seems a poor use of government authority.  Stick with the carbon tax.

January 5, 2009 1:04 PM

Brad Plumer said:

srodar and dhauck-- Jason Bordoff of Brookings gave a few reasons why insurance companies don't already do this in this paper (starting on page 16):

www.brookings.edu/.../07_payd_bordoffnoel.pdf

In some cases, yes, the monitoring costs for many firms are higher than the private benefits, although the *social* benefits appear to outweigh the costs (in which case it might make sense to offer tax credits for firms to implement this system). In other cases, state regulations stand in the way. Also, a bunch of insurance executives told Bordoff they're reluctant to switch to this system because they're afraid of infringing on the patents held by Progressive, a firm that *does* offer pay-as-you-drive insurance.

January 5, 2009 1:55 PM

dshenkle said:

Why not both?  Get an 8% mileage reduction with a carbon tax and also collect insurance premiums at the pump, with an additional tax to fund all liability insurance costs.  Presto, no uninsured motorists.

January 5, 2009 2:05 PM

srodar said:

Having lived in Manhattan, suburban NYC, Maryland and now New Mexico, I can say the accident risk of driving in densely urban areas is far greater than driving in rural areas. I know auto insurance rates are higher in urban areas than rural areas and there must be some underwriting basis for differences. My main point above is that there are lots of risk factors that determine rates other than mileage and mileage is already considered somewhat by insurers who factor in distance from work into premium rates.

Trying to balance out all the different factors and coming up with a way of tracking miles driven may cost more than any savings that might be achieved.

January 5, 2009 2:09 PM

satyendra said:

Srodar, understood that a 5-mile drive in NM is much less risky than in Manhattan - for one individual.  However, accident rates per person-mile must be much lower in Manhattan, right? I guess I can still see how insurance rates are lower in NM as you're insuring the person and not the person-mile.

January 5, 2009 2:58 PM

virginiagal said:

Just to think about - has it occurred to any of you that some of us commuting a long distance are not in a position to move closer to work right now b/c of the housing crash, and are not eager to change jobs in a bad economy?

This is NOT the time to make expensive changes that require that people either have extra cash on hand, be able to move easily, or be able to change jobs easily.

When you insure a person, who presumably drives about the same number of miles  each renewal period, many years of accident free experience are probably a good predictor of many more years of accident free driving.  The distance is already factored in the driving history.

If you want people to move closer to work (and trust me, when the economy stabilizes, that's very high on my list of priorities - I hate my long commute), I think stable housing and job markets, and higher gas prices (and they'll go up as soon as the economy stabilizes) will take care of it.  

Too many "incentives" too fast can really hurt ordinary people badly.  I care about the environment and global warming - a lot - but being able to live indoors and eat reasonably regularly really do have to take precedence right now.

January 5, 2009 3:18 PM

hrlngrv said:

Interesting point about Progressive's patent. If that is an issue, would this scheme require exercising eminent domain to acquire this patent from Progressive?

Next, auto insurance rates and policy language is currently regulated at the state level thanks to the McCarran-Ferguson act. Very costly to change the system state-by-state. Can of worms to partially repeal McCarran-Ferguson (it may be past time to do so, but there are a lot of insurance departments in the states that will fight this).

Those who do drive the most usually do so because they need to for work. To the extent their employers pay for their auto insurance, maybe no big deal, but to the extent employers require employees to buy basic auto insurance on their own, you're going to have the classic vocal minority complaining about legislation while other drivers who may see modest premium reductions at best don't care enough to lobby for the change.

We could cut down substantially on even more unnecessary energy use/pollution by slapping much higher fees on power boats and private aircraft. The Grace Commission made this recommendation back in the early 1980s. Didn't happen.

Remove personal interest from politics, and this sort of thing might fly. Good luck.

January 5, 2009 3:21 PM

thanbo said:

Rates are high in dense urban areas (particularly port cities) because the risk of theft is much higher.  The car-owner in Wauseon OH is probably not worried that someone will steal his car for a joyride, or to drive it to a chop-shop and sell it for parts to buy drugs, or if an expensive car, drive it to a ship which will smuggle it out of the country to some high-paying foreign buyer.

Brooklyn insurance rates are among the highest in the country largely because of theft and malicious damage.  Our last car was

a) broken into 2-3 times to steal the computer card which controls the engine, an $800 part which is easy to steal and resell for drugs;

b) stolen for a joyride, recovered 3 weeks later.

Our previous car lost its rear window when some moron lobbed an empty glass bottle of Kool-Ade at it.

Our insurance was about $1700-1800 a year in Brooklyn for those cars, while when the same cars had been registered in NJ and PA (we got both from my late father-in-law, at various times - yes, this was her father's Oldsmobile) my in-laws paid less than $500 for them, even though they were driven far more, and thus had higher risk of moving accidents.

Homeowners' insurance is also higher because of theft.  I could have gotten a rider for my wife's engagement ring, for 3% of the value of the ring per year.  That means that they expect you to have your ring stolen within 17 years.

January 5, 2009 3:24 PM

PHILDTM2 said:

As someone who commutes between Sacramento CA and San Jose (about 125 mi. each way) twice a week I would have no objection to paying insurance rates that in some way relected the number of miles driven.  However, if we're going that way, we should insist on accurate actuarial data being applied to determine those differential rates.  I am quite sure the following is NOT true:  Driver A and B are demographically identical, drive similar routes (of whatever kind) and both have excellent driving records -  Driver A drives 100 miles per week and Driver B drives 300, therefore on average will Driver B have three times the likely incidence of accidents compared to A?  I doubt it: driving is like anything else - the more you do of it the better you get.  Without any actual data of my own handy to back me up on this, I would suspect that Driver B is only about 20% more likely than Driver A to have an accident over the same time frame, though three times as many miles were driven.  The basic point is we need to insist on accurate actuarials being applied.  It would be shoddy and unfair to slap Driver B with triple the insurance premium of Driver A.

January 5, 2009 3:43 PM

satyendra said:

Phil, I had the same reaction as you, but according to Inglis's post "Accident risk—like fuel use—increases in roughly linear fashion with each mile a person drives"

January 5, 2009 3:59 PM

Andrew Davis said:

A lot of people in rural areas drive long distances to go shopping at the next town's WalMart, to attend cultural events, and to attend sporting events (especially youth sports).  I live in Kansas, and people drive long distances for soccer tournaments.  These are optional miles.  Sports leagues could be reorganized, and people could shop instead at their local merchant if it would save them money.

January 5, 2009 4:31 PM

hrlngrv said:

In re urban vs rural and the current state-by-state regulation of personal auto insurance, my money would be on states like NJ and CT adopting this, states like KS and MT not.

But let me take this in a different direction. People, especially children, watch too much TV. It's likely body fat percentage is positively correlated with hours of TV viewing. Also, TV requires electricity, which adds to carbon emissions. By the same reasoning as the cited article, structural changes that produce economic incentives to reduce TV viewing would be in society's best interests. Consider what it would take to make people pay for TV by the hour both in terms of the politics necessary to put such a program into place (not just TV viewers whining about it, also TV/cable content providers and advertisers) as well as the additional cost of fitting all TVs with usage meters.

January 5, 2009 5:36 PM

hrlngrv said:

I loved this line from the web site: 'Every household income group making less than $52,500 (in 2001) would save money on average.' Either every such household saves money or on average they save money (which could mean either than more save than don't, but could also mean those that the reduction for those that save would be larger than the increase for those that don't save). Given statistics' usual sematics, 'every' such household doesn't save 'on average'.

January 5, 2009 6:50 PM

virginiagal said:

satyendra and phil, I don't see anything in the source material that claims a LINEAR relationship, such that a driver that drives twice as far has twice the risk.  In fact, that is rejected by the paper itself.

It does say that risk increases with each mile driven.   That is intuitively obvious.  However, it does NOT say that the relationship BETWEEN DRIVERS is linear.  Those are two different things.  

In fact, the paper explicitly says the relationship between drivers is NOT linear.  

Quote from paper at www.brookings.edu/.../07_payd_bordoffnoel.pdf

"Notably, the relationship between VMT and accidents is not proportional. In aggregate, motorists that drive more tend to have fewer accidents per mile. One who drives thirty thousand miles a year will be involved in fewer than three collisions per million miles driven, whereas one who drives fewer than eight thousand miles a year will be involved on average in more than seven collisions per million miles driven (Figure 2)."

Also it makes a relatively fact-free assumption -

"While a comparison across different vehicles does not yield a proportional relationship, it is important to bear in mind that the risk of any individual vehicle being involved in an accident necessarily declines roughly proportionally with a reduction in VMT. Thus, while a high-mileage motorist may be only twice as likely to have an accident as a low-mileage motorist who drives one-fourth as much, if the high-mileage driver reduces her VMT by 10 percent she will also reduce her risk of getting into an accident by about 10 percent."

Even that assumption is based on indirect data - as the paper notes, "Reliable data on individual VMT variations and accidents to confirm this proportional relationship for individual drivers are lacking because insurance companies generally do not make accident data publicly available."  

In other words, they're guessing.  

The paper is taking a decline of overall miles driven and a proportionate decline in accidents and using it, without data supporting the applicability, as a proxy for a relationship in SUBSETS of the group - which IMHO is a big leap of faith.

The data they have shows nothing about relative proportionality of the data or behavior of those subsets.  The data they use to assume (and no other word fits) that such reductions take place is, and I quote,  "For example, a recession in 1981–82 caused a 10 percent reduction in VMT and a 12 percent reduction in insurance claims in British Columbia (Litman 2005), which is a more-than-proportional decline."

Pretty thin evidence for proportionality, either.

A linear relationship across different drivers is counterintuitive, and, if my quick read is correct, is explicitly contradicted by the paper itself.   A linear relationship in miles drive is unproven, and while they think it's obvious, I don't really see it and they certainly have not provided data to support it.

Finally, please keep in mind, again, that heaping extra expenses on those of us who cannot afford to move right now, who have to drive a long way to get to a job we hope we can keep, who do not have the choice of equivalent jobs nearby, and who are afraid to leave steady jobs in a terrible economy, is really pretty lousy.  

People are driving further, in many cases, b/c they can't sell their old houses, they lost their old job, they have to commute further to the new job, which may pay a whole lot less, and they can't afford to move until the economy gets better.  When the economy does get better, we have options.  Right now, you're taking people who are holding on teeth and toenails and giving them major expenses that they cannot absorb.    Hurting vulnerable people is lousy social policy.

January 5, 2009 7:07 PM

satyendra said:

Thanks for sifting through the Brookings report, Virginia. I didn't have the time and so took at face value Inglis's interpretation.

When I lived in Chicago I had a yucky commute to the N. W. suburbs through I-94.  Thereafter I resolved to only look for jobs that I could access via public trans.  I got one good job in a great downtown location that unfortunately moved to a suburb that had me crawling through O'Hare twice a day.  Even then I decided it was best for my career to stay.

January 5, 2009 9:59 PM

dhauck said:

Brad -

I read the sections you pointed out, but I'm still not convinced.  Of the three explanations given as to why insurance companies have not picked PAYD up on their own, only the second explanation, state regulation, seems to back Inglis' argument.  The first, excessive cost, is hardly a refutation of my own post.  In discussing the costs of PAYD, Bordoff is forced to resort to "social value" and "externalities".  Now, I've been on this blog a long time, and I understand both these concepts, but Inglis (and you, I think, in your response) seems to indicate the system would be self-sustaining in the long run, with just a one time tax credit to cover the initial costs of setting up the monitoring system.  I think Bordoff's heavy reliance on "social benefit" in a discussion of private business decisions would indicate he does not agree.  

This leads us to Bordoff's third explanation, Progressive's patent on the PAYD concept.  Inglis devotes a whole paragraph to describing the snowball effect of allowing low-mileage drivers to switch to PAYD, causing an upward shift in risk (and associated cost) of the remaining pool of drivers.  This would cause the next-lowest mileage drivers to want to switch, and so forth.  Except, apparently, PAYD is already offered - has there been a mass migration of low-mileage drivers to Progressive?  If not, why not?  If so, where is the snowball?  The answer to either question, I think, is in Bordoff's first explanation: the cost of a PAYD system is significant, its monetary benefits minimal.  So unless we lay the very heavy hand of government on the shoulder of the insurance industry, there will be no "snowball".  In that case, it's no longer a snowball, just a snow mound, artificially built up because a few people wanted it there.

Should we do it anyway?  As I said, I've been on this blog a while, and I understand, and in cases even agree with, the idea of social value trumping monetary value.  And I have no problem with using taxes to promote social policy -  I think a general carbon tax is a good example of using taxes to capture the externalities of heavy fossil fuel use, including long distance driving.  Certainly this proposal would reduce fuel consumption.  But would it do so as effectively as a carbon or fuel tax?  I think not.  Furthermore, PAYD unfairly targets those (like me) whose circumstances require long-distance driving, but who (also like me) have tried to mitigate that by driving small, high-mileage cars, and working odd hours to avoid heavy traffic.  So no, I don't consider leaning on the insurance industry, for the sole purpose of causing an indirect effect on fuel consumption, to be an effective use of targeted tax policy.  Stick with the carbon/fuel tax.

January 6, 2009 9:34 AM

satyendra said:

Also, some people go with insurers based on their ratings / reputation for paying claims promptly, not just absolute lowest cost.  I don't know much about Progress or their ratings, actually, I thought they were some sort of insurance broker that referred you to the lowest cost plan for you.  I didn't think they insured anyone directly.

January 6, 2009 10:25 AM

dhauck said:

satyendra - Actually, Progressive is a fantastic insurance company, and I don't just say that because their corporate headquarters is located about 10 mi north of  where I'm sitting and Cleveland needs all the employers it can keep.  They have a great customer service record, but more tellingly, they also have a great corporate culture (according to friends who work there) with a high pay scale, childcare for single parents, and an emphasis on public service.  I'd use 'em myself, except for... oh, yeah... they cost more.

January 6, 2009 12:31 PM

Environment and Energy said:

Speaking of congestion charges, here's a related idea making the rounds. As I mentioned last week

January 8, 2009 5:19 PM