Autonomous Cars and Induced Demand

27 Feb 2016

Last week I argued that autonomous cars can unleash a positive feedback loop transforming our cities for the better, by rendering parking an anachronism. In response, several commenters pointed to a post on Jarrett Walker’s blog that contemplates an alternative possible future. One where — land-use benefits notwithstanding — autonomy causes a dramatic increase in the total number of miles driven, and, by extension, terrible congestion. Walker, paraphrasing Mark Hallenbeck, imagines as follows:

A suburban father rides his driverless car to work, maybe dropping his daughter off at school. But rather than park the car downtown, he simply tells it to drive back home to his house in the suburbs. During the day, it runs some other errands for his family. At 3 pm, it goes to the school to bring his daughter home or chauffeur her to after-school activities. Then it’s time for it to drive back into the city to pick up Dad from work. But then, on a lark, Dad decides to go shopping at a downtown department store after work, so he tells his car to just circle the block for an hour while he shops, before finally hailing it to go home.

This is “induced demand” on steroids, and it’s a real risk. This is why the notion of fast-moving, tightly-bunched, coordinated fleets of cars is nothing more than a mirage. There’s no question that autonomy can increase the carrying capacity of all of our roads. But once the cars can drive themselves, there will be a tsunami of new traffic that will overwhelm the new capacity, and we’ll be right back where we started. Or worse.

There is, however, reason to be cautiously optimistic, if we can get two ingredients right.

The first is sharing. If most vehicles are shared, and private ownership is rare, we can dispense with half the trips in the nightmare scenario described above, including the horror of a private car driving around, empty, while its owner shops. The meteoric rise of Uber and Lyft and their ilk is encouraging on this dimension - an early sign that we’re getting used to the idea of “someone else’s car.”

The same could be said for Getaround and Turo (née RelayRides). These are particularly interesting, as they blur the boundaries between private and shared ownership. Even if suburban dad does own the car in question, isn’t it possible that, if it were easy, he would allow other passengers to pay to be driven around in it when he’s not using it? Maybe not everyone would. But when the money is on the table, many people will. Whether it’s strong sharing (pay-per-use access to fleets), or weak sharing (the compensated use of privately owned vehicles by persons other than their owners) doesn’t much matter. The economic logic of autonomous vehicles converges towards sharing.

There remains, of course, the fact that cars are tightly bound up with ideals of liberty and autonomy. And in their physical embodiment, we treat them as extensions of our homes, and in a sense, even of our own bodies. But I’m optimistic here, too. There’s some evidence for a generational shift in attitudes about driving. Which is not surprising, since the dream of automotive freedom has curdled into a triple nightmare of carnage, congestion, and environmental destruction.

Let’s say we manage to create a world where shared vehicles are the norm, and exclusively private use dwindles. We’re not out of the jam yet — we still have the risk of a massive increase in usage due to greater convenience and road capacity.

Fortunately, sharing itself creates a hedge against induced demand, fundamentally changing the economic psychology of driving. For most people who possess cars today, the real cost of each mile they drive is almost completely invisible. I’m not even talking about the external cost in congestion, pollution, etc. (more on that next time). Just the actual direct cost to the driver.) AAA estimates the all-in cost per mile at a stunning 60 cents per mile. For an average driver, that’s on the order of $10,000 per year - $833 each month - $27/day. And yet, each time you drive your 25 miles to work, does it feel like you’re spending $15? Not even close. We fill the gas tank twice a month. We only pay the insurance once a quarter. Maintenance is unpredictable. And we bought the car years ago, so a few thousand dollars of annual depreciation is the most invisible cost of all.

What if we could see the meter ticking, and paid for it as we went? It would completely change our behavior, in all kinds of ways, large and small. I’m sure of it. It would put a damper on the number of miles we drive, for a start. But it would also influence where we choose to live. It might even open our eyes to the tremendous savings that come from carpooling…and if you extrapolate from there and squint, you can begin to see the outline of a new model for public transit.

Of course — and this cuts against my optimism — in a world of full autonomy, each mile might also actually be cheaper, thanks to lower insurance costs. The vehicles will also be lighter and more fuel efficient, both because they’re sized for the right number of people, and because, if they never crash, they won’t need to be so over-engineered to survive crashes. But even here, the risks are mitigated. Because there’s no way to capture the economic benefits of right-sizing without sharing. The cheapest option for a family of four will obviously not be to own seven separate vehicles (four one-persons, two two-persons, and one four-person). To reap the cost savings that come from vehicles being the right size, they almost have to be shared.

Will autonomous cars lead to a massive rise in total vehicle-miles traveled? Will they uncork the mother of all induced-demand explosions? It’s a close call, but maybe not. Because autonomy can generate a virtuous cycle in land use patterns favoring proximity, density, walk- and bike-friendliness. And autonomy tips the scales towards sharing, which dampens the induced demand effect by making the direct costs of driving clearer and easier to see.

So I’m optimistic. But only cautiously, because to me, the forces of good and ill here are pretty evenly balanced. Fortunately, there’s a more radical step we can take that can, in one stroke, solve traffic for good. Stay tuned.