Should Geography Determine Auto Insurance Rates?

A large auto insurance company in California has decided to comply with Proposition 103 and base rates on people's driving records rather than where they live:

Major Shift in Auto Policies Rates would emphasize safety and mileage, not ZIP Codes, for drivers insured by Auto Club, by Marc Lifsher, LA Times: In a move that could presage lower auto insurance premiums for many of California's 23 million drivers, the state's fourth-largest provider has agreed to base its rates on how safely and how much its customers drive rather than primarily on where they live.

The plan, to be announced today by the Automobile Club of Southern California and California Insurance Commissioner John Garamendi, ... is a victory for consumer advocates, who say that rates based largely on ZIP Codes saddle city dwellers with higher premiums than suburban and rural drivers with similar records. It also is a coup for Garamendi, who has been working to force insurers to comply with requirements approved by voters with the passage of Proposition 103 in 1988.

In revising how it sets rates, the Los Angeles-based Auto Club is breaking ranks with the state's other major insurers. It expects that 88% of the 993,000 drivers it insures will see their bills drop 7% on average, or $134 annually. Most other Auto Club customers will see their premiums rise less than 5% when the new rate-setting formulas take effect...

Proposition 103, the landmark auto insurance initiative, required that rates primarily be tied to a person's driving record, number of years licensed and total miles driven each year — not the ZIP Code where a vehicle is registered. But changes in industry practice have been held up for the last 17 years. Insurers, which argued that where a driver lives is an essential factor in assessing risks and costs, persuaded former Insurance Commissioner Chuck Quackenbush to allow ZIP Codes to remain a priority in setting rates. Quackenbush's regulations and later court rulings kept the old rules largely in place.

In December, Garamendi announced that he would propose his own regulations aimed at fulfilling the mandate of Proposition 103. ... Garamendi's proposed regulations, which would greatly diminish the importance of ZIP Code rating, are under review at the California Office of Administrative Law. If cleared there, the rules would take effect after July 18. ...

Rates would go up for more than 60% of motorists living in rural and suburban communities and go down for people living in such large cities as Los Angeles, San Francisco and Sacramento, insurers contend.

Garamendi's regulations "ignore factors that greatly contribute to risk associated with certain regions. Population density and congestion are two such factors," said a recent statement from an industry-supported coalition, Californians to Stop Unfair Rate Increases. ...

If the market is competitive and rates are already set to price risk efficiently, and geography plays a significant role in the risk assessment, then complying with Proposition 103 would move away from the efficient outcome unless it corrects a market failure.

For that reason, I tried to think of some reason, some market failure, that would justify intervening in the car insurance market and forcing insurers to leave geography out of the rate equation even though my initial instinct is that this would reduce efficiency.

The part people find objectionable, I think, is that their rates are linked to factors outside of their immediate control. Where you live, not your individual skills as a driver determine your rates. In big cities, there is a congestion externality, and that extra congestion increases the chances of an accident. There are also other factors such as how often cars get stolen that play a role in pricing geographic locations as well. Again, assuming people take appropriate precautions, this is outside of their individual control.

But is the extra insurance cost outside of their control? Can't a person choose where to live? Don't we, in choosing to live in a crowded city instead of a rural area also choose higher insurance rates voluntarily?

On the other side, don't we, as a society, expect flexibility in the labor force? Don't we expect and want people to take jobs wherever they might be so that the economy is robust to shocks? If you are unemployed and receiving unemployment benefits and you are offered a job, you have to take it or you lose your benefits in many states. If so, do you really have choice about where to live, or is it forced upon you?

If the externality is avoidable, i.e. if teachers, factory workers, military personnel, accountants, and so on, and so on, have complete freedom of choice about where to live so that they can avoid the higher rates if they wish, then rates based on geography are equitable. But if most people must live near their jobs and do not have full choice about where those jobs are located, then perhaps some intervention to share the extra costs across geographic areas is in order.

I'm not sure if that gets there or not, but that's the best defense I could come up with. Maybe you can do better.

Comments

[M1EK]

Insurance is a classic market failure - regulations make the barriers to entry quite high; and other regulations to attempt to limit price have created perverse incentives to raise rates much higher on people who have gotten a couple of speeding tickets (for instance) than the risk would truly warrant. (An oligopy of heavily-regulated insurance companies is prevented from raising rates X% for everybody, so they need to make up the difference on the class of drivers on which they ARE allowed to raise rates).

[M1EK]

Forgot to mention the most important piece: any market for a good or service for which the government mandates the purchase can't be said to be remotely free or functional, and thus merits every bit of regulation it gets. If we required that everybody buy one hamburger per day, a lot of the competition between burger joints would probably disappear.

[crack]

The market failure is the fact that the zip code where you live is not inherently the zip code you drive in the most. The problem isn't necessarily using zip code, but what is used to determine a zip codes cost. Right now I don't know, but it seems to be based on accidents in the zip code. If a zip code has 10x accidents a day but only x accidents involve people from that zip code then it seems like they are bearing the brunt of bad driving from other zip codes. If its based on accidents by drivers in a zip code that just seems a little arbitrary, the problem drivers may do only a miniscule amount of driving in their zip code.

I think the type of insurance would matter a little too, liability should be based more on the person where as comprehensive would have a definite storage component. Again though, its possible the car spends a substantial amount of time outside the registered zip code.

[Andy]

Two things:
1) Against your argument -- salaries would presumably be higher in high-accident areas to compensate employees for the extra expense
2) For your argument -- is there a racial/discriminatory aspect to this? I think it is akin to redlining -- geography is a very good proxy for race these days.

[c roast]

How about Pay-As-You-Drive Insurance (PAYD)?

Premiums for a vehicle are directly related to how much is driven – and savings are related to how little driving takes place. Converting insurance into a variable cost gives the happy motoring public a new incentive to reduce mileage. Another alternative could be pay-at-the-pump-insurance (PAP). This nicely pounds the low mpg behemoths, makes sure that everyone is minimally insured (New Mexico, are you listening?) and the boost in the pump price reinforces good behavior (less driving).

In either alternative, the happy motorist purchases a minimum amount of insurance from the insurance company. In the PAYD alternative, the insurance company measures the VMT of the vehicle and re-prices accordingly. In the case of PAP, the state manages the cost collected at the pump in an insurance pool for the insurance companies.

Check out: http://www.vtpi.org/tdm/tdm79.htm

[vorpal]

Actually, this should ultimately lower the costs, because of lower overhead. Insurers no longer have to disect every zip code in the book to establish rates. since there is less information to base the rates on, they'll have to make other parts of the business more efficient to compete.

To the companies, the cutomers are a statistical set, but to the customers they are individuals and shouldn't be included in any demographic group for policy decisions.

If we didn't have laws against undue profiling (which this is) not one middle-class black person in America could get a mortgage loan. Under "free market" approach, a black male would have to pay outlandish amounts for car insurance... just for being black. In fact, the zip code factor is a wiley way of achieving the same thing as racial profiling.

Ultimately, basing the rate solely on driving record achieves all the market incentive necessary to promote safe driving...which is the goal of the state. This way, the total cost will be kept down because nobody will want to get into an accident, and the cost of dissecting demographic groups will be dispensed.

[vorpal]

In general, cooperative behavior is more efficient than competitive behaviour because people are pulling together. When possible, it is better to use cooperative behavior.

In the case above, the people of California are agreeing to get rid of some criteria. They are cooperating. In the future, they are competing to be safe, inexpensive drivers because they can't establish cooperation on driving extremely safely. Some people just don't seem to want to be safe drivers and some people want to drive more expensive cars.

This is a sensible blend of cooperative and competitive behaviour. Some call it "capitalist" and "socialist", because they are divisive. But once you get to brass tacks, capitalist and socialist just me competitive and cooperative behavior in economic systems.

I recommend that all economists read the above, because I think it will clarify a lot of fuzzy thinking they have in their head.

[billy]

Together with your post about the "New York Paradox", it makes an interesting juxtaposition. Will the low wage service employees in New York pay insurance set for a market of high paid executives? Or vice versa?

The assumption underlying ZIP code pricing is implicitly the same. Its a proxy for class. The insurance industy vehemently objects when they are asked to publish the pricing model and rates, under the pretension that it's a trade secret.

If the pricing model and rates are known, then this regulation would not have been needed.

[Bruce Wilder]

"If the market is competitive and rates are already set to price risk efficiently . . . "

These seem like two misplaced assumptions to me.

Insurance is a game of chance between insurer and insured, in which the insurer has to choose a strategy for assessing risk, based on particular, but limited and imperfect information, and, also, a related strategy for building a portfolio of policies.

The insurance companies are both guarding against adverse selection and moral hazard, and, at the same time, the insurance companies are, in effect, attempting to practice adverse selection and moral hazard, in reverse.

The State has intervened to require both that everyone buy insurance (which, notice, also requires that the State act to make sure insurance is offered to everyone, and that the State monopolize the decision to 'refuse insurance', i.e. revoke a drivers' license), that policies have standard terms, and that the factors used to determine rates be known and approved.

In effect, although auto insurance looks like a predominantly "individual" insurance market, it is, in effect, a patchwork of overlapping groups, with people assigned to groups, according to the allowed risk-assessment factors. Your rate is a function of your group, though one company's definition of your group might differ from another's.

The rate for your group is not a simple and competitively stable outcome of the pooling of risk within your group. First, your group is having accidents with people in other groups, so there is no possibility of a perfect pooling of risk. (The largest component cost of my own auto insurance is the fee for being insured against having an accident, which is the fault of an uninsured driver.)

Second, the insurance companies are playing a complex and dynamic game among themselves to manage their portfolio risk. They do not define groups in exactly the same way, and pursue other strategies to improve their ability to skim (or pack) their competitors' groups.

There is no reason to suppose that there are, or could be, an ideal set of rules. Probably, there are some benefits from periodically changing the rules of the game, and encouraging some of the insurance companies to try completely different risk assessment/portfolio management strategies. Such changes will break down settled patterns of oligopoly, signalling, etc.

If I could intervene to re-structure auto insurance to make it more "efficient", I would be more concerned with adminstrative efficiency, and less concerned with what's largely an illusion of "competitive market" efficiency. I would adopt "no fault" as a rule; that allows some big administrative efficiencies, and makes risk pooling more practical. I would make liability insurance universal, and collect the fees for it in a combination of a registration surcharge and a gas tax; the State could package and farm out the State's whole pool of policies in a series of auctions or negotiations with insurance companies. It would then be a straight-forward group insurance market, where competitive prices would be achievable. As a monopsonist buyer of insurance, the State would actually be in a more advantageous position to identify and pursue the public interest vis a vis the insurance companies, than they are as a regulator of what is, now, an obscurely complex game.

[johnchx]

I don't think I buy the labor market mobility argument. First, labor immobility due to bona fide costs of re-location (including higher living costs in other areas) isn't a market failure at all. It's an efficient response to price signals. If moving to take a job is a money-losing proposition, the prospective employee *shouldn't* move.

Second, it's not clear that equalizing insurance prices across zip codes actually enhances "total" willingness to relocate. While it would increase the willingness to move from formerly low-cost regions to formerly high-cost regions, it would also reduce the willingness to move from formerly high-cost regions to formerly low-cost regions. Without some empirical evidence, I don't see a clear reason to assume which of these effects (if either) would dominate.

[lonesome moderate]

Almost everything that one has to pay for varies in cost from one zip code to another, and some things (notably rent and utility bills) vary more than insurance does. But there seems to be a consensus in many states that, for this one product, the free market is not good enough. Somehow there's a moral imperative for insurance, and only for insurance, that people who live in low cost areas should subsidize people who live in high cost areas. I honestly don't understand this. What is it that makes insurance so different from other goods and services?

[Bruce Wilder]

"Somehow there's a moral imperative for insurance, and only for insurance, that people who live in low cost areas should subsidize people who live in high cost areas. I honestly don't understand this. What is it that makes insurance so different from other goods and services?"

Insurance carries inherent cross-subsidization with it: the people, who never make a claim are paying for the people, who do have losses and claims. That's the nature of the product.

It is different in a variety of ways, but it is not uncommon for price discrimination or other schemes to result in some kind of skew that could be construed as a subsidy of one group by another. People, who eat early in restaurants, get cheaper food; people, who make reservations early travel cheaper. The political complaint here arises from being unable to control your insurance rates.

There are no "high-cost areas" or "low-cost areas" independent of the scheme for creating groups and risk pools. There are high-risk and low-risk drivers, and any scheme will lump a lot of low-risk drivers with some high-risk drivers. Any scheme will do that, because that's what insurance does. Some people are not too happy with schemes that rely heavily on zip codes to do this lumping, and other people will be unhappy with whatever the new schemes are.

[Paul Gowder]

M1EK's comments are clearly correct. What meaning does the law of demand have in a market with massive barriers to entry for a product that is de facto (since it's theoretically possible not to drive -- just like it's theoretically possible to survive on mashed potatoes and vitamin pills 3 meals a day) compulsory?

[dWj]

I don't drive. When I moved to New York, I decided to live near a subway station rather than live far away enough to pay to insure a car. (Some of my coworkers make the opposite choice.)

I would imagine that it's much easier for car insurance companies to keep information on which zip codes drivers in accidents are from than in which zip codes the accidents took place. I'm sure they aren't trying to use the latter as a proxy for the former.

In any case, it's fine for this company to change its policy on its own; if zip codes don't matter, and differential pricing is incorrect, they'll make a lot of money attracting urban customers. If location does in fact entail a bona fide expense in terms of higher risks, they'll see their new, more urban customer mix drive up their losses. I'm happy to see a company try this experiment, though I'm equally happy not to be a shareholder in that company. In the long run, this will hopefully increase the ability of the market to enable the differences between real costs in living in different locations to accrue to people who are deciding where to live.

[johnchx]

Having thought about this a bit, and having listened to how Garamendi is framing the issue, I have a conjecture about the politics underlying this decision: it looks like another instance of individuals over-estimating their own capabilities and their personal control of outcomes.

Those who follow behavioral economics know that individuals routinely over-estimate their own knowledge and capabilities. Drivers are no exception: to hear people talk, you'd think that everyone you meet is an excellent driver, surrounded by idiots and maniacs. (Definition: anyone driving slower than me is an idiot; anyone driving faster than me is a maniac.)

So, when Garamendi says "good drivers need a break," listeners hear "good drivers LIKE YOU need a break."

Individuals also over-estimate their ability to influence outcomes (with their universally above-average skills and knowledge, not to mention dashing good looks). So they find it not merely counter-intuitive, but literally unimaginable that a risk model that includes (a) their driving records and (b) their home zip code could have significantly more predictive power than a risk model that includes (a) alone.

And yet...insurance companies (who, can we stipulate, know *vastly* more about the factors that correlate with risks of auto insurance payouts than bloggers do) seem to be under the impression that exactly this is the case: that a model that includes *both* of these factors has significantly better predictive power than a model that is based on individual experience alone.

But to admit that, even to ourselves, would be to admit that our environment and generic lifestyle habits have more influence on the probability that we will be in an automobile accident than does our individual driving skill.

[iknowall]

Search for in all major search engines simultaneously on the site http://www.iknowall.com.
Simultaneous search on Google, Yahoo and MSN Live Search.

Try http://www.iknowall.com

[Deepak(car insurance help)]

If your major interest is information related to Auto Insurance Car Quotes in Only Seconds or any other such as Auto Insurance Company Ratings, Street Rod Car Insurance, California Car Insurance Online Quote, Motorcycle Insurance Quotes, Free Car Insurance Quotes Ontario or Car Insurance Companies, this article can prove useful.

[mark]

Thanks for the providing good information - this is really helpful for car insurance.I found another good one that helped me find an incredible rate http://www.carinsuranceassistant.com/index.php?id=New+Car

they give you the best rates from lots of local providers

[Fairy]

Thanks for the providing good information - this is really helpful for car accident insurance.I found another good one that helped me find an incredible rate they give you the best rates from lots of local providers Car Accident Insurance

[alina]

Thanks for the providing good information - this is really helpful for car accident insurance.I found another good one that helped me find an incredible rate they give you the best rates from lots of local providers
Car Accident Insurance

[Rocky Rock]

Thanks for the article.
Rocky
Great Auto Insurance