When traffic congestion gets especially nasty, the first thing planners think to do is expand road capacity. More lanes should ease the pressure, right? Except, that doesn’t work. As Eric Jaffe points out over at Atlantic Cities, traffic tends to expand to fill capacity. He cites a new paper in the American Economic Review that finds that traffic “increases proportionately to roadway lane kilometers for interstate highways and probably slightly less rapidly for other types of roads.”

(Scott Olson/Getty)

A second option for planners, of course, is to expand public transportation. If there are more buses and subways, that should free up space on the roads, right? No again. Here are the authors of the AER paper, Gilles Duranton and Matthew Turner: “We find no evidence that the provision of public transportation affects [vehicle miles traveled. We conclude that increased provision of roads or public transit is unlikely to relieve congestion.”

As it turns out, this is a common finding. A World Bank study by Antonio Bento found that better bus service has essentially zero effect on total vehicle travel. (Boosting rail service can decrease car travel, but only modestly.) A University of California-Davis study by David Heres Del Valle also found no link between the quality of transit options and driving, although it did find that certain land-use policies could cut down on car travel somewhat.

This shouldn’t be so shocking. Road space during rush hour, after all, is a valuable commodity. When a highly valuable commodity is offered for below-market prices (or for free), we should expect shortages and long queues. In this case, the shortages are traffic jams. That doesn’t mean expanding roads is entirely useless. After all, if it means more people can travel on the roads at the times they’d prefer, that’s a tangible benefit. But bigger highways or better public transit shouldn’t be sold as a means of eliminating gridlock. (And, for those keeping score, that means this analysis in my earlier post on the congestion benefits of transit is wrong.)

So is there any way to reduce traffic congestion? Jaffe argues that the only viable option is congestion pricing — charging people a fee to use the roads during peak hours. Stockholm experimented with this program in 2007, Jaffe notes, and the results have been dramatic: “Transit ridership is up, traffic is down some 18 percent, and in some cases rush-hour delays have been cut in half.”

On the other hand, Felix Salmon adds some additional complications. Not only is congestion pricing unpopular — it’s human nature to prefer hidden costs (traffic delays) to explicit costs (fees) — but, over time, the fee starts to lose its effectiveness. Drivers adjust to the rush-hour charge and start clogging the lanes again. The only way to really liberate the roads from congestion is to do what Singapore does and constantly adjust prices until you get the desired amount of traffic. And sometimes that means cranking the fee way up — to painful levels (especially for low-income commuters). Economists love this idea because it’s a genuine market solution, bringing supply and demand into balance. But that doesn’t mean it’s politically painless.