Toll Roads and Externalities

A recent paper by Aaron Edlin and Pinar Karaca-Mandic has focused my attention on the potential of toll roads. The basic question is whether,

...driving entail[s] substantial accident externalities that tort law does not internalize? ...If so, this implies that a one percent increase in aggregate driving increases aggregate accident costs by more than one percent.

This may seem obvious. Any error in tort judgments would reduce deterrence enough to make it suboptimal. But, argue Edlin and Karaca-Mandic, its not so simple;

The reverse, however, could hold. The riskiness of driving could decrease as aggregate driving increases, because such increases could worsen congestion and if people are forced to drive at lower speeds, accidents could become less severe or less frequent. As a consequence, a one percent increase in driving could increase aggregate accident costs by less than one percent, and could even decrease those costs.

Edlin and Karaca-Mandic find that

...traffic density increases accident costs substantially whether measured by insurance rates or insurer costs. ...a typical extra driver raises others’ insurance rates (by increasing traffic density) by the most in high traffic density states. In California, a very high-traffic state, we estimate that a typical additional driver increases the total insurance premiums that others pay by roughly $2231 ±$549 each year.

What is the best way to internalize the externality? Gas taxes are, as Edlin and Karaca-Mandic point out, politically untenable. They propose

...requir[ing] insurance companies to quote premiums by the mile instead of per car per year? This simple change could reduce driving substantially by moving a fixed cost to the margin without raising the overall cost of driving.

To some extent insurance companies already do this. Nonetheless I'm not sure that this solves the problem. A friend of mine lives in Riverside, CA and commutes to LA at 3 am. He would get hit by the Edlin premium but is in fact reducing the externality. Even in LA he is really only a risk to himself at 3 am. Another option is toll roads. The problem is that currently most toll roads do not congestion price or differentiate by vehicle size (beyond trucks); a factor White [2002], for example, finds significantly affects accident costs. The transaction costs of internalizing this externality via toll roads may be too high. But technology, according to the Economist, is changing. Toll roads can now congestion price and change higher fees to SUV. The Economist notes that a Swiss toll system which charges for the distance driven

".. seems to be an unmitigated success. To general surprise, it was up and running on time. And it achieved its main objective: reducing truck traffic across Switzerland, which increased by 7% during the late 1990s. In the year after the system's debut in 2001, the number of trucks on Swiss roads fell by 5%. What is more, transport companies now try much harder to ensure that their trucks do not cross the Alps empty. Financially, things appear to work, too. Operating costs amount to only 6% of revenues, estimated at €575m last year.

Addendum: Tyler and Alex debated this issue you can follow the debate here.