Tuesday, March 3, 2009

Road congestion and diminishing returns

Carol Coletta from CEOs for Cities (and our keynote speaker in the last annual conference) links a fascinating study on road congestion and car usage in the US. It turns out that traffic congestion in America's cities declined 30% in 2008, a truly amazing figure in a single year. The reason behind the suddenly unclogged, agreeable roads was a 3% decline in total miles traveled by drivers during the year.

How come such a small drop of driving produced such a gigantic drop in congestion? Basically, road and highway capacity has a tipping point: a certain stretch of asphalt carries more and more cars up to a point where the average speed drops bellow a certain threshold (40 mph in highways) and traffic slows down dramatically, creating a traffic jam. A road can take only a certain amount of cars before it becomes useless, and sometimes a tiny drop in the number of drivers can make a drive much more pleasant by moving the the volume under the level where capacity is maxed out.

As Carol says, we have two ways then to lower congestion. One is expensive, and favored by DOTs everywhere: add more lanes, cut bottlenecks, and basically throw concrete at the problem. For a massive amount of money we will have a road that can take more cars before being clogged, but that will end up getting congested sooner or later, as drivers have free access to it anyway. The second option is fairly cheap: raise the price for road use, aka congestion pricing.

Last year we saw something similar to congestion pricing in the form of very high gas prices. The added cost moved a very small percentage of drivers to carpool or look for alternatives, pushing congestion down in a dramatic fashion. A more rational, less indiscriminate way of doing the same is coming up with a system (let's call it "not-a-toll") that makes driving more expensive when roads are busier (higher "demand", thus higher prices) and much cheaper when they are not full of conmuters. That is, congestion pricing.

The basic idea to have in mind is that driving in Connecticut right now is not free. When I get on my car and drive to work, I am spending time. When I-91 is not cooperating, I am spending a lot of time that could be used in other more profitable endeavours (sleeping). The big problem with me sitting in I-91 in traffic is that the price of my conmute is not only being paid by me in lack of rest, but by all the other drivers as well stuck in traffic. If only 3% of us (wild estimate) decided to work from home or take the train (if it was there), the other 97% will enjoy a road with 30% less congestion, being able to add a good 20 minutes sleep every day.

As ConnDOT is not able to have direct control of our work hours, what they can do is make this pain I am inflicting other explicit: if I am going to be lazy and not walk 10 minutes to the station for the convenience of using my car and annoying everyone else on the road, I will have to pay X dollars. Some conmuters will be lazy and eat the cost, but if a number big enough cheapens out and teleconmutes, works 10 am to 6 pm to avoid the charge or takes the train, driving will be much faster for very little cost. If well done, transportation costs on aggregate can be much lower, as the cost of congestion is reduced dramatically.

Is this an endorsement of tolls? Not necessarily; depends on many factors. Congestion is not distributed evenly, after all; and the lack of alternatives could just turn this into a tax. What it means, however, is that congestion charges could make a lot of sense in many cases.


  1. Please note that building a public transit system that only takes 5-10% people off the road can have massive benefits on those who are still driving, by the same logic. Just food for DOT thought.

  2. Actually, let the Onion talk about that better: