The Limits of Congestion Pricing

It may be the best solution to city traffic, but that doesn't mean it's a perfect one

Image
Reuters

In October we described the "fundamental law of road congestion," which explains that building more roads fails to decrease congestion, and in fact increases it, because doing so tempts more people into their cars. The authors of that paper believe the most effective response to city traffic — nay, the only one — is congestion pricing.

Of course that doesn't mean it's a perfect solution. In response to our October post, Felix Salmon at Reuters argued that congestion pricing is a perpetually unpopular approach to clearing city traffic. At first, drivers are upset they must pay more to use the same route they've always used. As congestion pricing clears roads, it inevitably induces some additional demand, and traffic increases, so that prices must be raised even more. And so the fight continues.

Still that fight seems worthwhile — provided that pricing truly eases congestion. At least one transportation scholar isn't so sure. In an upcoming issue of Urban Studies, Moshe Givoni of the University of Oxford offers a detailed examination of London's congestion pricing system and concludes that "questions can be raised with regard to its practical effectiveness."

London began its pricing program in 2003. At first the benefits were enormous: in response to the rising cost of driving into the city, congestion fell by roughly 30 percent. But those gains quickly reached a plateau, and since then congestion has started to near its pre-pricing levels, Givoni reports:

In London, congestion is measured as the average excess delay (minute/km), which is the delay to traffic compared with the free-flow speed — the average speed during night time. In 2002, the average excess delay on roads inside the CC [congestion charging] zone was 2.3 min/km — the base congestion level before CC was introduced. This had fallen to 1.6 min/km in 2003, the often-quoted 30 per cent reduction in congestion. The level of congestion remained the same in the following year but started to increase thereafter. In 2005 and 2006 it increased to 1.8 and 2.1 min/km respectively. Thus, congestion has almost returned to its pre-charging level in 2006.

After the pricing scheme went into effect bus ridership exploded in the city, but Givoni isn't convinced that pricing was the primary reason for this newfound popularity. Sure ridership went up after the program began, he writes, but service increased, fares decreased and quality improved. Maybe more people in London are riding the bus because they don't want to pay more to drive. Or maybe they're riding it because they've been given a better bus system.

Of course it was the congestion charges that funded the bus improvements, so some connection certainly exists. But Givoni's point is that it might have been even more cost-effective for London just to implement the bus improvements from the get-go. He also reports "no clear evidence" that the congestion pricing program decreased air pollution in the city, either directly or indirectly. All told, he concludes, the available evidence is enough to "cast doubts on the overall perceived success of CC in London and thus to promote a more cautious pursuit of CC in other cities."

Givoni isn't the only scholar urging caution. In a theoretical study scheduled for the February 2012 issue of Transportation Research Part A, a group of Norwegian researchers examines the secondary impacts of congestion pricing — in particular, its effect on labor markets. While congestion pricing does seem to provide a general welfare gain, it does so by creating disparities in unemployment or population:

We show that while welfare gains can be achieved through optimal charging, this may come at the price of decreased integration. This may manifest through either greater centralisation tendencies in population, or through unemployment disparities between regions.

The research group drew their conclusions by modeling two hypothetical regions linked by a major piece of infrastructure that suffers heavy congestion, in effect simulating suburbia. They created an employment shock by having companies from one region move to the other in response to the congestion, and observed what might happen to the populations of each region in the aftermath of this event.

When the regions were 50 kilometers (roughly 30 miles) apart, congestion pricing created a substantial and long-lasting unemployment disparity — i.e., people couldn't afford to drive into the other region for work. When they were 20 kilometers (12.5 miles) apart, congestion pricing resulted in a wide population disparity — i.e. people simply moved to the region where there was more work.

None of this is to say that congestion pricing isn't a good way for cities to handle traffic problems. The problems presented above all have their counterpoints. Congestion pricing may be unpopular, but so is traffic in general, and so are politicians who do nothing about the problem. London's congestion may be growing again, but more aggressive pricing increases might keep these levels low (and it's worth noting that Givoni ended his study in 2006, before London raised prices again). While pricing may create regional labor disparities, it still raises general welfare, and if combined with improvements to transit systems those disparities could shrink.

Still the larger point remains: traffic has become such a problem in major metro areas that even our best solutions might not feel like solutions at all.

Photo credit: Jason Reed/Reuters

About the Author