Hal Varian says those of you who live in high density areas don't pay nearly enough for your auto insurance:
Beyond Insurance: Weighing the Benefits of Driving vs. the Total Costs of Driving, by Hal R. Varian, Economic Scene, NY Times: ...[G]enerally, the more people on the road, the more accidents will occur, at least up to the point of total gridlock. Even adding an obsessively safe driver to the road could increase total accident costs, since reckless drivers ... would now have one more car to run into. ...
If the number of accidents increases as the number of cars per mile of road increases, then adding a new car to the road will increase the expected accident costs for existing drivers. The new driver has to pay for the expected accident costs she may incur (via her insurance premium) but not for the increase in accident costs that her presence creates for other drivers on the road.
Economists say that the new driver imposes an “external cost” on the other drivers. Since drivers only face their own cost of accidents in their insurance premiums, they don’t face the full cost of their driving, which should include the costs they impose on others.
How big is this external cost? This is the question investigated by Aaron S. Edlin ... at the University of California, Berkeley, and Pinar Karaca-Mandic ... at the RAND Corporation, in a recent paper, “The Accident Cost From Driving,” published in the Journal of Political Economy.
The critical factor affecting accident costs turns out to be traffic density. In states where there are few cars per mile of road, the external accident costs are much less than in states with more congested roads. The authors estimate that an additional driver increases statewide insurance costs in California by $1,725 to $3,239, depending on the method used for the estimate. The comparable number for North Dakota is about $10.
In an economist’s ideal world, the benefits of driving should be compared with the total costs of driving, which include both the costs borne by the driver and the external costs that the driver imposes on others. In economics jargon, the external cost should be “internalized.”
Mr. Edlin and Ms. Karaca-Mandic estimate that internalizing those external costs nationwide would require a substantial tax on driving, on the order of $220 billion a year. In California alone, such a tax would raise around $66 billion. By comparison, the state income and sales tax revenues for California in 2004 were about $72 billion.
On the other hand, drivers in states like North Dakota with low traffic density would have negligible driving tax rates since the presence of an additional car in such places has almost no effect on the accident rates.
Internalizing the accident cost of driving would require a tax that varies with both traffic density and the amount of driving a person does. The obvious candidate would be a local gasoline tax.
But a gas tax is not ideal since good and bad drivers would be charged the same rate per gallon. One could also consider a tax on insurance premiums, which generally take driving records into account. Of course, a combination of taxes would be possible as well.
Unfortunately, the political support for such measures is virtually nil. ... People have made decisions on where to live and how far they are willing to commute based on the current costs of operating vehicles, and a dramatic change in the cost ... would be extremely unpopular.
Still, one might consider taking some modest steps toward making drivers face the full costs of their decisions. A moderate increase in the gasoline tax whose proceeds were plowed back into mass transit might be politically palatable. Driving during commuting hours in New Jersey and California isn’t very pleasant, and many drivers may prefer fast and convenient public transit.
Many commentators have called for a higher gasoline tax to discourage the use of oil. Some are motivated by environmental concerns and some by international politics. The proposal has gained little traction since the payoffs for reducing gasoline consumption seem distant and uncertain. Accident costs of driving give yet another reason to try to reduce our oil dependence. But unlike the other reasons, the payoff for driving less should have an immediate and concrete effect: fewer accidents.
Here's the argument about the externality from the introduction to the paper:
Consider two vehicles that crash as one drives through a red light and the other a green light. Assume that the accident would not occur if either driver took the subway instead of driving: hence, strictly speaking, both cause the accident in full, even though only one is negligent. The average accident cost of the two people’s driving is the damages to two vehicles (2D) divided by ... two ... (i.e., D per driven vehicle). But the marginal cost exceeds this. In fact, the marginal cost of driving either vehicle is the damage to two vehicles (2D per driven vehicle)—fully twice the average cost. Surprisingly, this observation holds just as much for the nonnegligent driver as for the negligent one. Drivers pay the average cost of accidents (on average, anyway), not the marginal cost, so this example suggests that there is a substantial accident externality to driving...