### Accidental Externalities: Reply to Comments

Hal Varian responds to some of the comments on his NY Times article "Beyond Insurance: Weighing the Benefits of Driving vs. the Total Costs of Driving," posted here:

Some replies:

External costs are costs on uninvolved third parties. The costs people in a crowd impose on each other are not external, because each member of the crowd "pays" the price whether in time lost in traffic jams or standing lines, or accident insurance...

It is true that the members of the crowd pay the average cost of congestion, but they don't pay the marginal cost (which is higher). This is the classic problem of the commons.

------------------------------

Check out pay-as-you-drive-insurance:

http://www.serconline.org/payd/talking.htmlYes, this is the right idea.

----------------------------

Where is the detailed statistical analysis that supports this claim?

It obviously isn't in the 850 word Times column. But if you take the time to read the paper at http://works.bepress.com/aaron_edlin/21 you will find detailed statistical analysis.

-----------------------------

And the call for a car accident tax is nonsensical. Car insurance premiums are supposed to be calculated to cover the real cost of accidents.

The premiums cover the average cost of accidents (obviously) but the marginal cost (i.e., the cost imposed on other drivers by an additional driver) is much higher than the average cost, at least in congested areas.

-----------------------------

The number of accidents likely do increase with the number of cars but only to the point where increased congestion lowers speeds and that is far sooner than gridlock.

The evidence presented in the paper says otherwise.

------------------------------

The arrogance on display is simply breathtaking, though. After 100 years of auto insurance, these economists assume that they have found a flaw in the system, which mere mortals, supposedly, have never noticed.

The original point was first made by Nobel laureate William Vickrey in 1968. So it only took 60 years to notice the problem, not 100. 8-)

-------------------------------

I sure do not see what the basis for the assertion, "drivers only face their own cost of accidents in their insurance premiums" is. The whole point of liability insurance is to pay for the costs to the other guy, of my hitting him.

The argument is independent of whether insurance is liability or no-fault, or whether or not there is insurance at all. Here is a more detailed example.

Suppose that Adam values driving at $150 and Eve at $70. Adam is currently driving and Eve is contemplating doing so. If only Adam drives there are no accident costs; if both drive the expected accident costs are $50 per car.

With these numbers, Eve rationally decides to drive since her value exceeds her costs: 70 > 50. Net social value from both driving is: (150 + 70) - 100 = 120. If Eve faced the full cost of her decision (the total accident costs) she would rationally decide not to drive yielding net social value of $150 (Adam's value of driving).

Now suppose that Adam and Eve both value driving at $150 but each has to face the $100 *marginal* cost of their decision. Net social value of both driving would be $300 - $100 = $200 which is higher than the net value of keeping one or both of them off the road. [It's true that there is an extra $100 of driving tax collected in this solution, but that can be used to reduce other taxes --- it is not a social cost.]

Suppose that each values driving at $70. The optimal solution here is for one to drive and the other not, yielding a net value of $70. This can be supported by charging 0 for the first driver on the road and $100 for the 2nd driver (i.e., the $50 marginal cost they face + the $50 marginal cost the would impose on the other driver.)

So the Vickrey tax gives the right solution when one should drive and the other not, or when both should drive. It is simply marginal cost pricing, slightly disguised.

Posted by Mark Thoma on Friday, November 17, 2006 at 04:24 PM
Permalink
TrackBack (0)
Comments (46)

You can follow this conversation by subscribing to the comment feed for this post.