Self-Driving Cars Can’t Cure Traffic, but Economics Can

http://www.nytimes.com/2017/03/08/upshot/self-driving-cars-cant-cure-traffic-but-economics-can.html

Version 0 of 1.

It’s easy to get giddy about self-driving cars. Older people and preteens will become more independent and mobile. The scourge of drunken driving will disappear. People will be able to safely play video games while on the freeway to work.

But there is one problem autonomous driving is unlikely to solve: the columns of rush-hour gridlock that clog city streets and freeways. If decades of urban planning and economic research are any guide, the solution is unlikely to come from technology but from something similar to Uber’s surge pricing: charging people more to use driverless cars at rush hour.

Not that technology companies aren’t trying to find other solutions to congestion. Traffic is one of the few problems that fabulously wealthy people can’t buy their way out of. This helps explain why Elon Musk, the founder of Tesla and SpaceX, wants to bore subterranean freeways under Los Angeles and build a hyperloop train half the length of California. Or why Larry Page, Google’s co-founder, is interested in flying cars.

This is in addition to other, less revolutionary efforts, from companies like Sidewalk Labs, which is owned by Google’s parent company and which aims to ease congestion by helping cities make better use of data. And of course there is the self-driving car, which, in addition to making roads safer, is supposed to help manage freeways by smoothing human flaws — like a tendency to engage in antsy braking and sudden lane changes — that make traffic worse.

These various technologies share a common theme. One way or another, they promise to expand the nation’s roads — literally, in the case of Mr. Musk’s tunnels, figuratively, in the case of flying cars, and efficiently, in the case of self-driving ones. While it is possible that one or all of these technologies will increase road capacity to the point at which no amount of traffic will fill them, history gives us reasons to be skeptical.

Decades’ worth of studies show that whenever cities add roads, new drivers simply fill them up. This isn’t because of new development or population growth — although that’s part of the story — but because of a vicious cycle in which new roads bring new demand that no amount of further roads can satisfy.

This has been studied at rush hour, studied on individual freeway projects and studied with large data sets that encompass nearly every road in the United States. With remarkable consistency, the research finds the same thing: Whenever a road is built or an older road is widened, more people decide to drive more. Build more or widen further, and even more people decide to drive. Repeat to infinity.

Economists call this latent demand, which is a fancy way of saying there are always more people who want to drive somewhere than there is space for them to do it. So far anyway, nothing cities have done to increase capacity has ever sped things up.

The extent of this failure was chronicled in a 2011 paper called “The Fundamental Law of Road Congestion,” by the economists Gilles Duranton, from the Wharton School of the University of Pennsylvania, and Matthew Turner, from Brown University.

The two went beyond road building to show that increases in public transit and changes in land use — basically, building apartments next to office buildings so that more people can walk or bike to work — also fail to cut traffic (or do so only a little).

This doesn’t mean public transit and land planning are bad ideas, or that widening freeways is a bad idea. When roads are bigger, more people can get around. More people see family; more packages are delivered; more babies are lulled to sleep. It just means that none of those measures have done much to reduce commute times, and self-driving cars seem unlikely to either.

That’s where charging people during busy times comes in. “Maybe autonomous cars will be different from other capacity expansions,” Mr. Turner said. “But of the things we have observed so far, the only thing that really drives down travel times is pricing.”

This is because the average person prefers the privacy and convenience of riding in a car. Only when the drive is far enough or the traffic is bad enough — or a taxi costs enough — will more people choose to bike, car-pool, hop on a train or postpone a trip. This same pattern shows up in all kinds of places. You can see it in car-centric cities like Los Angeles and Houston, where yearslong, multibillion dollar lane-widening projects did little to speed commutes.

But you can also see it in New York — by far the most transit-friendly city in the United States. Ride-hailing services like Lyft and Uber have led to increased auto traffic, according to a February study from Schaller Consulting in Brooklyn, while subway trips dipped slightly in 2016 because of a fall in weekend usage.

Bruce Schaller, principal of Schaller Consulting, said if the growth in ride-hailing services continued, it would inevitably push the city toward some sort of congestion pricing system, an idea New York has floated and rejected.

“There will be so many cars on the streets, and freight deliveries and buses and everyone else will be so slowed down, that people will get fed up and demand a solution,” he said. “And we have a solution in pricing.”

Things like matinee movies, red-eye airfares and happy hour drinks have accustomed Americans to the idea of variable pricing depending on time. But charging more for roads is toxic, at least in the United States.

Mayor Michael R. Bloomberg met abject failure with his attempt at congestion pricing in New York. Yet London, Singapore and Stockholm have all put in such systems effectively.

In the United States, the most common objection is that road pricing is regressive: Rich people get to drive alone while the masses huddle on a bus. Also, people just don’t like paying for things that they are used to having free.

Economists are hoping that may change. Several states, including California, Texas and Minnesota, have added high-occupancy toll lanes with different pricing during rush hours.

“This idea of congestion pricing is not completely dismissed the way it once was,” said Clifford Winston, an economist at the Brookings Institution.

Mr. Winston said the eventual introduction of self-driving cars would probably lessen consumer opposition to paying more to use roads during peak periods. Ride-hailing apps have taught consumers to accept surge pricing, and people are generally less resistant to paying for something new. The result would be something like variably priced lanes dedicated to fleets of robot vehicles.

If that happens, one of the hidden benefits of this revolutionary new technology will be that it got people to accept an idea that economists started talking about at least a century ago. And you get home a half-hour earlier.