10° 40' N, 61° 30' W

Monday, February 17, 2003

The Congestion Charge

I don't drive in London, but the biggest story locally, besides last Saturday's antiwar march, is the start of congestion charging in London. The £5 charge is extremely controversial, and it's failure could lead to London's mayor, Ken Livingstone, losing reelection next year. I don't have much to say on the politics of it, but I would like to shed some light on the economics of the charge.

Livingstone himself has attributed the idea for his scheme to Milton Friedman, who wrote a paper in 1951 titled "How to Plan and Pay For the Safe and Adequate Highways". London's scheme, as implemented roughly follows this, so posting about this would not be terribly insightful. The paper I have in mind, though, is one published in 1974 by the Harvard economist Martin Weitzman in the Review of Economic Studies titled "Prices vs. Quantities". Weitzman's work has been a seminal one for those researching common-pool resource problems, akin to the tragedy of the commons. Congestion can be thought of as one such problem.

Weitzman was looking for the best way for regulating pollution under uncertainty, with the choice between a quota- or a price-based instrument depending in large part on the degree of uncertainty that policymakers are willing to tolerate in the achivement of a particular outcome. The first type of instrument, a quota (or limit) arises when a central planner (I know some call the mayor "Red" Ken, but no pun is intended) sets a a cap on the amount of allowable pollution; in this context the planner sets a ceiling on the amount of vehicles that can enter the city centre, rationing out the permission to enter.

Sound farfetched? Well such a system is currently in use in Athens, and used to be in Paris, where authorities only allow licence plates with particular numbers into the city centre on certain days (usually the rationing alternating between odd and even numbers every day). A quota can also be set to very close to zero, with only limited traffic; Rome is implementing such a scheme. The trouble is, as the Economist writes, is that such regulation is blunt, and cannot be fine-tuned to demand as a price instrument--in other words, it assumres that regulators can determine an optimal limit to congestion. It also assumed that the net number of vehicles is fixed or growing very slowly; its effects would be reduces if there is a boom in car sales (as happened in paris, where people bought second cars with the necessary registration to beat the system. Such quotas also require enforcement, which has to be paid for; and regulation raises no revenue.

A second means of implementing a quota is though the creation of a market, as the US did for the reduction of sulphur dioxide emissions. How is this a quota? Well a market requires some degree of excludability--the ability to stop others from consuming the same good you consume. In a 1995 article, Paul Krugman explains how it would work:

Suppose we issue registered drivers in a metropolitan area a specified number of "rush hour points," which they are free to sell at their discretion, and require anyone who drives during rush hour to present (electronically) the appropriate number of points. Let us also create a market in these points. Some people will continue to drive every day; they will need to purchase extra points. Others will find new ways of getting to work and will sell their points for whatever the, ahem, traffic will bear. How market-oriented can you get? And such a scheme, like congestion fees, would make most people better off, either because they get extra money by selling their points or because the cost of buying points is more than offset by a quicker commute. (Italics added)


Krugman describes such a scheme and congestion charging as being basically equivalent, as they both involve " . . .creating a market incentive for people . . . to take into account the costs we impose on others by driving during rush hour." This market, like a quota, assumes that the planner can set an optimal ceiling on congestion--though it is adjustable, for as more information becomes available the planner could buy permits or issue new ones.

Krugman was right on that last point, but he is wrong about the equivalence of the two schemes. One problem is distributional--who do the initial specified number of permits get disributed to? Non-drivers among those can expecte to receive a windfall gain, while drivers among the excluded will suffer windfall losses. Such a market is also government subsidised, in terms of enforcement--like a quota, it raises no revenue for the state, and does not pay for its own operation.

Price measures include Livingstone's charge. Unlike both regulation or a market this does raise revenue, but as it is a £5 flat-rate, it too is a blunt instrument--it will not limit congestion in the event of unexpected surges in demand. While it cut congestion by 25% today (some of that due to the fact that it is half-term this week--school is out) the long-run effect on congestion remains uncertain. A more innovative way both of controlling congestion and raising revenue is the electronic road pricing in use in Singapore, where congestion is measured and the price of accessing the city centre adjusted accordingly every half-hour. This is a closer to a true price system, in that it passes on the "cost" of congestion; evidence from Singapore shows that it has cut congestion considerably.

Like most "markets", the congestion charge, as implemented, has "distortions"--vehicles that receive discounts or are exempted. Some of these create odd incentives--the 100-% discount for alternative fuel vehicles, for example, might reduce air pollution, but does not do anything about congestion, which is what the charge is ostensibly about; the discount for residents of the charging area, similarly, does not affect congestion.

The technology--a series of 700 cameras taking pictures of the licence plates as vehicles enter the zone--is untried, and could be the undoing of the scheme. Singapore's use of low-powered radio, checked by overhead gantries, is far superior, as is a GPS-based system. The trouble for the mayor is that such sytems would require a national government mandate that all vehicles have them installed, with severe penalites for those caught without them. Toll plazas, as in Norway, are out of the question in the city centre, though could work closer to the M25 beltway. Cameras, untested as they are, are entirely within the mayor's control, and this, in all probability, is why this option was chosen.

If the system works, the chance is that congestion charging will become a more common means of dealing with traffic congestion in major cities, as many economists have proposed; in the New Yorker last September, for example, John Seabrook looked in the possibility of congestion charging in Manhattan. Although it's technology will need to be replaced in the long run, and the design improved, let's hope that London's experiment does not fail.