Perspectives from Around the World

So Many Cars, So Few Roads: A problem for cities everywhere in the world

Gilles DURANTON
Professor, The Wharton School, University of Pennsylvania

As the world develops, it is becoming more urban. One person in two now lives in cities and nearly three billion new urban dwellers are expected by 2050. There is a simple reason for this big urban shift. As we become less dependent on farming, cities offer more opportunities to work, shop, find friends or mates, and enjoy leisure. In turn, this urban advantage depends on urban dwellers being able to seize these opportunities. For this, providing good accessibility to urban dwellers is fundamental. We sometimes call this challenge the urban transportation problem.

To understand why this is a "problem", consider drivers in an American city. Each kilometre they drive costs them about one minute and thirty seconds to two minutes of their time. Looking at the choices made by urban dwellers when they choose between different transport alternatives, transportation economists have concluded that drivers tend to value the time they spend in their cars at about half of their wage. Hence, each kilometre costs about 30 cents of their own time to American drivers. Then, we must add about 6 cents for fuel. Drivers must also add a few more cents for the depreciation of their car. Put together, a kilometre by car costs about 40 cents to urban drivers in America.

These costs paid by drivers are not the only costs of driving a car. Drivers also impose costs to others. Urban drivers in America impose about 3 cents per kilometre of accident costs, a couple of cents for local pollution, less than a cent for global warming, and, another 3 or 4 cents for congestion. In total, American cars cost more than 50 cents per kilometre to operate in cities but only 40 are paid by drivers. In other countries, the calculation will differ slightly. For instance, in Japan or in the UK, fuel is more expensive whereas in developing countries the value of time is lower. It remains nonetheless that, the world over, not all of the costs of driving are paid by drivers. And when something is too cheap, consumers will consume too much of it.

What are the solutions? The costs of accidents and pollution are (very) roughly proportional to fuel consumption and thus can be dealt with by the gas tax. Put differently, through the gas tax, governments can charge drivers for the pollution and accident costs that they impose on society. What about congestion? It is much more complicated because it depends on traffic conditions. Rural drivers or drivers who drive at 4 a.m. do not impose much congestion, if any. On the other hand, at peak hours the congestion induced by an extra kilometre driven can be much above the 3 cents per kilometre figure quoted above. Many estimates range between 30 cents to 60 cents per kilometre. Thinking about the extra driver that stands in the middle of a big intersection at 5 p.m. and blocks everyone else, Nobel Prize winner Bill Vickrey even made a plausible case that each kilometre driven by this last driver could costs hundreds of dollars to society.

Aggregate congestion in American cities can be estimated to cost the U.S. economy at least 100 billion dollars every year in lost time for car drivers. The total cost of congestion is perhaps twice that once we take into account the cost of wasted fuel and that of idle trucks. The costs of congestion may be even higher in relative terms in developing cities where driving conditions are even more chaotic and bound to get worse as more and more households buy cars.

Sadly, a good case can be made that a lot of what cities gain in extra productivity, they lose in congestion. Because of congestion and, in many developing countries, pollution, cities are not exploiting their full potential. What are the solutions?

Historically, urban planners and the political system in many countries have responded by increasing the supply of roads. America developed an extraordinary system of urban highways and many Asian cities have built elaborate systems of elevated roadways. But when it comes to congestion, building more roads is a losing proposition. Actually, traffic increases nearly exactly proportionately with the roadway. This is because more capacity makes taking the road more attractive and, as a result, any capacity gain is nearly exactly offset by rising traffic. There is considerable latent demand for driving, especially in the more central parts of cities at desirable hours. In the end, road building does not reduce congestion. That does not mean that building roads is always wrong. It increases capacity. If done well, one can also eliminate bottlenecks, but this is not a general cure to congestion.

More recently, urban planners have also been seduced by engineering solutions. Ramp-metering is one example. Preventing too many cars from entering a highway when it is already very busy can have positive effects. It is better to force cars to queue before entering the highway rather than clog it. But again, this type of engineering fix is ultimately self-defeating. Making a highway more fluid at peak hours only makes it more attractive and is bound to attract more drivers. Here again, solutions like ramp metering are useful because they allow more cars to go through but ultimately they cannot be expected to improve driving speed much. As driving gets easier, car drivers will want to drive more.

The latest fad among urban planners is to attempt to get drivers to switch from their cars to public transportation. One reasonable way to do it is to provide nice and efficient public transit. This sounds attractive. But the same logic still applies. If one can convince a driver to take the bus instead of the car, this frees some space on the road and that space will be taken by someone else. Public transportation can have lots of virtues but reducing congestion is not one of them. The data show this.

So if the cure for congestion cannot be found on the supply side, it has to be found on the demand side. Some cities in the developing world such as Mexico City or Bogota impose quantitative restrictions based on a car's plate number. For instance, plates ending in 3, 6 or 9 cannot drive in the city on Mondays between 6 a.m. and 10 a.m. and between 4 p.m. and 7 p.m. This type of regulation is fairly easy to implement. This is the good side of those restrictions. But everything else about them is not very good. Mainly, there is a way around these restrictions: People keep several cars with different plate numbers. Not only does this make these restrictions ineffective, but also it can make the situation worse because having lots of old cars on the road brings lots of negatives. Old cars are unsafe and pollute a lot more than more recent cars. There is another reason why quantitative restrictions are not appealing. They allow some drivers to make yet another trip of little value to them like going to the supermarket a third time on the same day because they have a lucky number on their plate. At the same time, these regulations also prevent other drivers from doing something they really value like going to work.

If the solution cannot come from increasing supply or from quantitative restrictions to curb demand, it needs to come from pricing. The idea is simple: Make drivers pay for the congestion cost they impose on others. Just like we make polluting plants pay for the costs of their emissions, we need to make drivers pay for the delays they impose on others. This does not mean that we should expect completely fluid traffic at 8 a.m. Making roads so expensive so as to allow only a handful of drivers at full speed at peak hours would not be efficient. Instead the idea is to discourage unnecessary trips at peak hours and keep traffic moving. Since there is no way to judge which trip is really necessary and which trip is not, the only possibility is to rely on a price and the willingness to pay of drivers to take the road. Drivers will be willing to pay for important trips. They are less willing to pay for less important trips.

Although simple, this idea is far from popular. Many objections against it have been raised. The first is that this is just another tax. While governments are often short of revenue, the objective of charging for congestion should not be raising revenue, it should be about improving efficiency. Charging for congestion could be made revenue neutral by lowering other taxes. The second objection is that running a road network where drivers are charged according to where and when they drive is technologically difficult and costly. There is something to this criticism. Many of the benefits from the London congestion charge disappear in the enormous costs of administering this scheme. But poor implementation should not discredit a good idea and technology is changing fast. While plate recognition systems like in London are extremely costly to operate, transponders like those used to go through tolls in many countries are cheap and highly reliable. New global position system (GPS) based systems are being developed. They promise extreme accuracy at minimal costs. We also know that while real time congestion charging is still a challenge, simple "cordon" pricing where drivers are charged a fixed amount at certain times when they enter a given perimeter can yield tremendous improvements. Singapore -- the poster child for a congestion pricing -- operates a very sophisticated pricing scheme. Stockholm in Sweden operates a much simpler scheme. This simple scheme has nonetheless improved driving conditions very significantly. A third criticism against congestion pricing is that it threatens our privacy. This accusation carries much less weight today when most people let cell phone provider track them at all times through the signals of their cell phones. But even if we take privacy seriously, there is a simple answer to this concern. Cars can be equipped with anonymous boxes. Congestion charging does not require fingerprinting drivers and having to know who is where at all times. It only requires anonymous payments.

The last accusation against congestion charging is that this is somehow "unfair". Using the jargon of economists, congestion charging is often accused of being regressive and affects the poor more. While there is no doubt congestion charging has distributional implications, it is simply untrue that it taxes the poor more, be it only because the poor often do not have cars! It is true that among drivers, paying a fixed fee for better driving conditions benefits those drivers who value their time the most. Those who potentially lose the most are those who are forced to abandon their cars and use public transit instead. Here is the solution that some cities like London or Stockholm have adopted: Use a good share of the revenue from congestion charging to provide more and better public transportation. The poor who are using public transit even in absence of a congestion charge will see their lot improved. Those who are induced to leave their cars will actually find a much better alternative. Finally, those that remain in their car will experience much better driving conditions making it worthwhile for them to pay a charge for driving at peak hours. In theory everyone can be made better off. While we may not be able to achieve this ideal, a large majority of travellers can still gain in practice.

March 1, 2014

Article(s) by this author