By the year 2030, most Americans may only head into the office a few days a week, lots of urban skyscrapers will have been converted into apartments, suburban land prices will have declined, and we’ll be “driving” — or at least sitting in autonomous cars — way less. Goodbye to traffic congestion.
That’s a bit of the possible future sketched out in a recent blog post – “What happened to traffic?” – by transportation expert David Levinson of the University of Minnesota and author of the Transportationist blog. I chatted with Prof. Levinson about his fascinating speculation for my Ricochet Money & Politics podcast. Here are some highlights from our conversation:
You wrote what I think is a remarkable blog post recently. It’s received quite a bit of attention. You titled it “What Happened to Traffic.” And your thesis, I think, is that traffic is declining and will continue to decline dramatically in the coming decades. And that decline is not only the result of some deeper trends, but will itself cause a pretty majoring restructuring of American society. Now to many if not most people, it probably seems like traffic is getting worse, so explain your thesis.
Well, it depends on where you are as to whether traffic’s declining, but national statistics have shown that per capita travel in vehicles is roughly where it was in the late 1990s. And vehicle miles traveled, the number of miles that cars are moving is roughly where it was in the early 2000s. And this is after a 90-year increase in the amount of automobile traffic, from, you know, the 1910s to the early 21st century.
So people have sort of this expectation that traffic will continue to increase because it has increased in the past for such a long period of time. And this is built into traffic forecasts. It’s built into the way people view the world. But beginning in the early 2000s, in particular after 9/11, with a number of societal changes, including things like increased gas prices, changing demographics, changing employment, the amount of travel that people were engaging in individually has leveled off and has declined on a per capita level. So this is sort of the first thing that we’ve observed.
Now, a lot of technologies have a lifecycle. They have an S curve associated with them. So they start off, they grow slowly even, there’s a period of very rapid growth. Then it levels off. And then something new happens and the S curve begins to decline. And so we sort of see that in a number of things that we no longer use as much we used to. U.S. mail volume increased for decades upon decades until the 1990s. And it started to level off in the 1990s with the rise of email and the Internet, and then, in the early 2000s has fallen off a cliff.
So is that going to happen with travel? And so this is the scenario that I’m painting. And so it’s a future scenario. I don’t want to say that I predicted that this would happen, but this is one thing that might happen that nobody is taking any account of right now.
Right. The blog post is set up as someone looking backwards from 2030 and outlining a timeline of events.
So the idea is that we have sort of three things [that might happen] as future scenarios. The first scenario is that traffic picks up and continues to grow into the future. I think this is becoming less and less plausible over time because why would people travel more unless they could travel faster. And until people can travel faster, they’ve allocated an hour, hour and a half a day for traveling already. There’s not a lot of growth in the amount of travel that people can do individually.
Now, of course, there’re still a few more people being added to the population, about 1 percent growth per year in the U.S., so you might see a little bit of increase associated with population growth. But on the other hand, we’ve got these offsetting factors which lead people to travel less.
And so the second scenario is that per capita travel is essentially flat. You might see a small growth in total amount of traffic with population growth.
So I want to paint a picture of a third scenario where traffic actually falls.
So why would that happen, and what is the impact if it does?
There are a few possible causes. One is perhaps we get more productive with what we do on the internet, how we work. We don’t need to work at certain places in order to get work done. There’s nothing magic about a five-day workweek. It’s very industrial age. And if we’re going to be working with information most of the time, we don’t need to be in a physical place to do that.
Now a number of people have what’s called the five-four schedule where they work five days a week one week and then four days a week the next week. This is already something that’s happening.
So if you’re an employer, you look at this and you say well, if people are going to be coming into the office to meet with other people, we need conference rooms, but not desks. Or we need fewer desks.
Well, space is money, right? I’m a firm: Why do I want to rent more office space then I really need if I can offload work to my employees’ homes in a sense? The employee’s will be happier in many cases because a couple of days a week, they’ll be able to work from home. And then, of course, they still check their email at home other days of the week. And then they come in three days a week. Well, I can reduce the amount of space I’m running by 40 percent. That’s a huge savings.
In the piece, you offer the scenario that maybe by 2025 you’ll see the kind of the four-three schedule, so maybe you’ll take every other Monday and Friday off, I guess.
And how soon this happens, I mean, this is just speculation, as to whether it’s 2025 or not, but the trend I think is in place and the logic is place for this. And so the question is when are people going to do this. You know, a lot of companies are nervous about being first with having different kinds of human resource structures or different kinds of social trends in place because it’s a risk and they don’t want to look silly. And people are much more comfortable doing something if they can see how it’s worked in other places. But somebody’s going to go first.
And so I’d dub this the flipped office. We have what we call the flipped classroom in a lot of places, where the professor is not lecturing so much, but the professor is helping students solve problems and they record the lecturing. The students can watch the lecture at leisure on their computer. And when they do this, they can stop at a place they didn’t understand. They can review it. And it’s – you know – in many cases, it’s productive.
As a professor myself, I’m doing this a little bit, but I’m not fully doing this yet. But I would imagine that in 10 years, I’ll have all my lectures recorded and doing more and more problems in the classroom.
And this changing of the work week has huge implications for real estate, both in the suburbs and the cities?
And so if people are going to work three days a week, we need 40 percent less office space. So the good office space will stay offices. The marginal office space will have to get converted to something. I mean, they’ll even be torn down or be changed over to some other land use.
And so I invented the skyscraper crash of 2021. Real estate goes through these periodic spasms of overexpansion and then it crashes. And we’ve seen this a lot. We saw this with the savings and loan bubble in the early 1990s. We saw this in the early 2000s. We saw this with the recession in the 2008, 2009. So real estate gets frothy. And my estimate is that 2021 is about the time for it to get frothy again.
So there’ll be more land available [in cities], more structure available to live in, residential real estate. And we might see shifts. In a lot of countries, the wealthy people live in the core cities and the poor people live in the suburbs. In the U.S., it hasn’t generally been that way. I mean, obviously, wealthy people live in a few cities like in New York or Chicago, but often the core cities have been populated by much poorer people. But if suburban real estate becomes a lot cheaper and it turns out that it was overbuilt because of this change in demand pattern, we might see more lower-income folks moving into the suburbs. And just as we once had like flight from the cities, back in the 1960s, we might see a reversal of that and immigrants moving into the suburbs in much greater numbers.
So shorter workweeks and greater urbanization, both of which equals we’re just not driving as much, right?
If you’re in the city, you don’t need to drive as much just because the distances are shorter. And the public transit is better. And the more people who use public transit, the better public transit gets. And so this is one of the nice positive feedback systems that we have. If it’s working in a good direction, the more people who use public transit, the more frequent the buses will be. And the more frequent the buses are, the less you have to wait for public transit. And the more attractive public transit becomes.
Now, of course, this worked in reverse in the mid part of the 20th century is fewer used public transit because cars were now affordable. In the 1950s, public transit service was cut back and as public transit service gets cut back, more and more people want to use cars. And that makes public transit gets worse and worse. But there’s no reason that system can’t work in the opposite direction. I mean, it worked in the opposite direction in the 1880s and through 1920s. And in core cities, if we’re increasing the population in the cities, we should see better public transit service.
So far this is basically an extrapolation of existing technology and demographic trends. How do driverless cars fit into all of this?
And so driverless cars are interesting, right? So we’ll probably be able to drive by, I mean, sort of at the high-end, driverless cars within 10 years. Not everyone will right away. But you can sort of imagine in 20 years all new cars will be driverless. You’ll still have older, traditional cars on the road because it takes a while to change over the fleet.
Now, if we have driverless cars, one of the really interesting things is that car sharing becomes a lot more feasible. So in a few cities now there are car sharing programs that car, Zipcar, Car2Go, things like that where if you – instead of owning a car, you can get one on demand and you pay, you know, $0.38 a mile with Car2Go. And it’s convenient and it saves you the cost of ownership because you might not use your car all the time.
The Car2Go is a very specific model. It’s a Smart car for two. So it’s a pretty tiny car, which is good for certain kinds of things. If I’m just taking myself and a backpack from place A to place B, this works really well, or buying some groceries. It’s not good for hauling stuff from the lumberyard. But a lot of people buy trucks because they imagine that they’re going to be hauling stuff around. And some people do that all the time if you’re on a farm. But a lot of people buy trucks and they don’t haul stuff around that often. Or they buy an SUV because sometimes they’re carrying a lot of people.
Imagine instead of buying the largest car that you might use, you buy the car that you’re going to use every day and you rent the car that you might use some of the time. And it’s really easy because there’s car sharing programs. And the car sharing programs are really convenient because the driverless cars drive to you. So you call up a car on your mobile phone. And in 10 minutes, the right sized car is delivered to you and you pay $0.38 a mile or whatever it costs. And you can have that car for that trip. And even if it costs you $10 or $20 for that trip, that’s a lot less than the $10,000 extra you’re paying over the life of the car for, you know, more expensive fuel and for a car that you don’t use as much, and all those kinds of things.
Skeptics of driverless cars seem more concerned about insurance and liability than the technology. Do you have any doubt that this is going to happen over the next 10 to 20 years?
Insurance should very much be in favor of this. I think what needs to happen is that the automakers need to be comfortable that their liability is not going to be so great. So inevitably a driverless car will still get into a crash because they’re dealing in a mixed environment. Somebody runs out in front of the street, and so on. Who’s going to be held liable for that? The driver wasn’t driving because it was a driverless car. And so what is the jury going to say? The car should have detected the, you know, the little kid who ran out in front of it. Well, yeah, but if a little kid just darted out, it might not have been enough time if the car was driving at speed limit. Who gets held liable for this? We don’t have any case law for this and we don’t have any liability laws determining this.
But I think that’s very much in everybody’s incentive, aside from maybe trial attorneys, to set up a legal framework where insurance pays for it and the automakers only pay if there’s real negligence and then only. This enables a much safer environment. The automakers would certainly be in favor of that. The insurance companies would certainly be in favor of something like that. And you got to imagine that eventually we’ll get our act together.
Now, the United States might not be the first to go, right? If you want to test driverless cars, the best place to test driverless cars is on a small island where you just replace everybody’s car with driverless cars and you make sure that you worked out all the bugs before you start deploying it widely in a large market.
So I don’t know why we haven’t – you know – just made an offer to one of the Virgin Islands and say, we’ll buy your 10,000 cars and put out driverless cars and see how it goes for a couple of years and make sure we’ve sort of worked out all the problems with software and hardware?
And I would think by the early 2020s, we’ll start to be able to buy them on our high-end Mercedes and BMWs and Lexus over time. But it’ll take a little while before the costs come down.
So how much should we spend on upgrading US infrastructure? Might it be the case that we’ll need nicer roads and bridges so these driverless cars can go, you know, 90 miles an hour on I-95? Do we need to spend a lot more money right now?
My sense is we don’t need to be building much new infrastructure, not to say that there’s no places where you wouldn’t build any, but it shouldn’t be federal priority certainly.
But what should be a federal priority is that if the interstate system is an important national thing to have, we want to make sure that we keep it. And that’s where we should be focusing our resources is maintaining the interstate highway system, making improvements to make it safer, make it work a little bit better. Make sure that the pavement is, you know, in good condition. Making sure the bridges are all in good condition. And that’s where the resources should be spent.
And what would be the role of better “pricing” traffic via tolls and congestion fees?
One of the predictions I have is that we eventually move over to tolling. Now, one of the reasons is if vehicle miles travel declines, the revenue source that we’re going to be using – that we’ve been using since the 1920s and are continuing to use to fund a lot of our road system — is the gas tax. Well, we’ll be consuming less gas because – gasoline because miles traveled goes down and we’ll be consuming less gasoline because fuel economy is improving and because we’re moving towards an electric freight.
We have to switch from the gas tax to something else to fund infrastructure. Now, what that thing is, whether it’s congestion charges or whether it’s using general revenue is a political decision. We’d be much better off if we had congestion charges and mileage charges. And we’re moving in that direction, it seems. But that’s got to be implemented and people are really nervous about the government tracking where they’re going.
Now, of course, the government in some ways is already tracking where you’re going because your cell phone is one you all times and the government’s tracking that. So I don’t really think we have much in the way of privacy anymore. But people are still nervous about this stuff and until they swallow the pill, a slightly bitter pill, that you don’t have privacy when you’re on the roads anymore. I mean, you have a license plate on your car and we’ve got cameras everywhere.
Once you swallow the pill that you don’t have privacy anymore, then, the main objection, the privacy objection to moving towards a mileage-based system, a lot of that disappears.
And that would replace the gas tax and it could replace a lot of property tax revenue. So most local governments fund roads via property taxes. And there’s no reason it should be based on property taxes, except that historically it’s been based on property taxes and it’s hard for local governments to implement gas taxes because if one of them implements a higher gas tax, then people can easily go to the next county over, the next town over and buy gasoline somewhere else. Whereas states, you’re less likely to leave state to buy gasoline.
So gas taxes have traditionally been at the state level and property taxes have traditionally been used at the local level to fund roads, but there’s no reason if you have a mileage-based user fee and you have a uniform rate you couldn’t use that to pay for highway financing. And then you can differentiate the price. So if you’re driving when it’s less congested, you don’t pay as much. And if you’re driving in areas with less traffic in general, you don’t pay as much. Would be a couple of things that you could do to try to soften the impact of this and to use the financing mechanism to help manage roads better.
Now, whether we’re going to go to a full-fledged highly dynamic congestion pricing system, that’s, I think, a few steps away, but it would be more efficient if we did.
So what’s the counter case, that traffic and congestion are only going to get worse?
So population is increasing about 1 percent a year. If we opened up the doors to immigration and [population was] increasing at 2 or 3 percent a year, that would have a significant effect on transportation. That would offset any changes in individual behavior at this point. I mean, the amount of travel that you and I are doing might go down a little bit, but if you start adding 3 percent to the population every year, we’d have to be – everybody would have to drive 3 percent less a year to offset that. And that’s a lot to ask from 200 million people.
So that’s something that could happen. You know, people could find that with driverless cars they can go a lot farther and it doesn’t bother them as much. So instead of being willing to drive 60 or 90 minutes a day, maybe they’d be willing to drive 90 to two hours a day or two hours to three hours a day.
I don’t think that’s terribly likely just because there’re so many things that people would rather do than be in a car. But if you’re in a car and it’s much more comfortable than it is today, I could see that sort of at the edges, people would be willing to travel a little bit farther. So I’d live farther away from my workplace.
So the question is, if you’re living farther away from your workplace, are you still going to work five days a week? You might be willing to do that if the car is more comfortable because you can, instead of paying attention to driving, you can pay attention to the Internet or you can watch movies or you can do work while you’re in motion.
But I’d still rather be in one place than in a car, I think. I’d rather be in my house than in a car. So how much better a place can I get that’s three hours away compared to one that’s two hours away? There’s a limit. You only have 24 hours in a day. You’re sleeping eight hours. You’re working presumably about eight hours. So the maximum that you could be driving is four hours in each direction and that you would have no time to do anything else. And so that’s sort of the upper limit. And if we’re driving one hour now, there’s certainly some play in that number, but I think there’s limits and practical limits to that.
I think it’d be more plausible that things are basically flat than that they went up.
Follow AEIdeas on Twitter at @AEIdeas.