A Strange Definition of "Externality"

I just came across a surprising paper from economists Aaron S. Edlin and Pinar Karaca-Mandic published in Journal of Political Economy in 2006 which claims to address the “accident externality” due to driving. Here’s the abstract:

We estimate auto accident externalities (more specifically insurance externalities) using panel data on state-average insurance premiums and loss costs. Externalities appear to be substantial in traffic-dense states: in California, for example, we find that the increase in traffic density from a typical additional driver increases total statewide in- surance costs of other drivers by $1,725–$3,239 per year, depending on the model. High–traffic density states have large economically and statistically significant externalities in all specifications we check. In contrast, the accident externality per driver in low-traffic states appears quite small. On balance, accident externalities are so large that a correcting Pigouvian tax could raise $66 billion annually in California alone, more than all existing California state taxes during our study period, and over $220 billion per year nationally.

So if I move to California and start driving, not only do I have to pay for my own insurance, but everybody else’s insurance premiums will go up by a fraction of a penny; in total everyone else will pay an extra couple of thousand dollars. Edlin and Karaca-Mandic call this an externality.

I don’t get it. This extra cost is only an externality if you can pick somebody out as the “extra driver”, but you can’t do that. Each driver who pays insurance is covering the cost of all the others drivers’ “externalities”, just as they are covering the cost of his.

Let’s rejuggle the books to eliminate the externalities: assume there are 20 million drivers in California and the accident externality I impose on each of them by driving there is one hundredth of a cent (for a total of $2000). I’ll take this entire cost on myself; my premium goes up $2000. But now I’m not going to cover other drivers’ accident externalities. My initial premium included one hundredth of a cent of cost due to each of those 20 million drivers, so we can eliminate that; my premium goes back down $2000 to its original level.

Entirely through economic sleight of hand, Edlin and Karaca-Mandic make a case for an extra $66 billion tax on California motorists. They not only avoid mentioning that the costs they are trying to account for are already taken up by insurance premiums (they actually calculate the costs by analyzing how premiums vary with traffic density), but also suggest that the tax be imposed on those premiums, at a rate of 200–400 percent! If the money were actually being spent to address the externality (the increased risk of accidents), then insurance premiums would fall accordingly and the state would take over the bulk of the insurance industry, with responsibility for covering all accidents caused by other cars on the road. In practice, however, the authors propose only that the state pocket the money as tax revenue. Insurance premiums would still have to take traffic density into account, so lower density could lead to lower premiums, but this would be nothing like enough to cover the tax increase and the cost of driving would be much much higher. (These higher costs are the only reason to assume any reduction in traffic density.)

If you want to discourage driving to decrease pollution, to prevent traffic deaths, or to alter the character of a community, I can understand that. Justifying taxes with economic obfuscation, however, is not cool.

(Brought to my attention by Andrew Gelman.)