Mount Transit, Mount Auto, Mount Next

US Transportation Trends
Comment: The graph shows both linked and unlinked transit trips, as the way transit trips are counted has changed, and there is no continuous series of both over the entire period

In the US, we have seen a great struggle play out in the twentieth century between what David Jones calls Mass Motorization and Mass Transit . The conflict between the modes continues to this day, and has become a morality play in the culture wars. While they mostly serve different markets, they compete for users, and roadspace, and funding, and the hearts and minds of travelers. They are competing on old turf though, as the graph shows, both modes appear to be in decline, transit for decades, the decline of the auto-highway-system is just beginning.

To develop a metaphor Kevin Krizek and I used in Planning for Place and Plexus, the US spent from the late 1880s through the early 1920s climbing Mount Transit. Transit was the most important mode of travel (after walking) in large and medium-sized US cities. The rise of transit was enabled by the electric streetcar, itself a product of electricity, harnessed by Edison and others, and the modern railroad, developed beginning in 1825 with the Stephenson’s steam-powered Stockton and Darlington Railway. Transit peaked in the US in the 1920s, but for a spike during World War II when oil and rubber were rationed, crimping the US of the automobile. From the end of the War forward, it began a steady decline from which it has not really recovered. Despite the so-called resurgence of transit, and receiving about a quarter of federal surface transportation expenditures, transit trips per capita remain below 1990 levels.

The US spent almost the entire twentieth century climbing Mount Auto. From the 1920s onward, the automobile was the dominant mode of travel for Americans, accumulating more miles per capita than other modes. While the Great Depression slowed the auto’s growth, it did not result in decline. There was a brief downturn during World War II, and a couple of hiccups in the steady rise of mileage due to oil shocks in the  1970s and early 1980s (Yom Kippur War through the Iranian Revolution, before oil deregulation), early 1990s (Gulf War), and early 2000s (9/11). But the later 2000s and 2010s have seen a sharp downturn in auto use per capita. This drop is greater than the drop during World War II in absolute terms (though the War saw a drop of 23% off the pre-war peak, and the 2012 drop is 7% below 2005). It is complemented by an apparent downturn in total miles of paved roads.

In The Transportation Experience, William Garrison and I trace the policy, planning, and deployment of transportation technologies across time. Both the auto and transit follow the classic lifecycle model or S-curve of birth, growth, maturity, and decline. (One hesitates to say “death”, since so few technologies actually disappear, fixed route streetcars are still with us for instance. Kevin Kelly has found that no technologies actually vanish, though obviously they diminish in importance). The S-curve allows us to mathematically approximate the process of growth and decline of technologies. It is in many ways natural (if we start with 0 vehicle kilometers traveled by car per capita in 1900, surely the number has to go through 5000 VKT before it reaches 10000 VKT, and 10000 before 15000. One million people must own a car before two million can. Similarly, technologies don’t disappear overnight (although transit really came pretty close). These are long gradual processes, and the occur over many technologies which see growth and decline. Transportation is among the slowest of these technologies, as fixed infrastructure is expensive to build and long-lasting.

The growth curves reasonably fit the data for total system size or total system use for a number of technologies in retrospect. A collection of such curves, and descriptions of the development of the associated technologies can be found at the Transportation Deployment Casebook, which is the result of student projects for a few years in my Transportation Policy course.

The difficulty is to use such curves in prediction. There are some observations though, the left and right sides of the curve (from the inflection point, where the rate of growth changes from increasing to decreasing) are approximately the same amount of time. (So it takes about as long to go from 10 to 50% of the final market size as it does to go from 50 to 90% of final market size.)

A key issue is the determination of how large will the system get at its maximum? It depends on the system. For instance if we are modeling the number of US states that will adopt some policy, the maximum is 50 (unless the US adds states). If we are modeling the percentage of cars that will have some advanced technology, and we believe it will become universal, then we can say 100%. But if we are modeling a continuous number, rather than a share, it is harder. What is the maximum number of kilometers people will travel in a year? What is the maximum number of trips? We can make guesses; we can even make informed guesses, but we can never know for sure until after the fact. However, if the rate of growth has slowed (we are on the right half of the S-curve), we can make a much better guess than if growth is increasing at an increasing rate (the left half of the S-curve).

Is the decline in auto permanent, like what happened to fixed route transit services in the US (which is well below one-fifth of its previous importance), or just a brief digression from the steady march of increasing per capita vehicle travel that has been following the same drum almost continuously from 1910 to 2000? History will tell us for sure, but the evidence for “Peak Travel” has been mounting. This does not mean there will never be a year in which auto travel again rises. The economy and gas prices still fluctuate, and a boom year with low gas prices following a recession with high gas prices might very well temporarily bump traffic upward, but that is really short-term noise. In the absence of external events (technological shifts, demographic shifts, social shifts), the curve appears to have peaked. But over the longer term, a significant technological shift could profoundly change how people use the automobile. If there were only one possible significant technological or social shift, this might be predictable, but there are numerous technological and social shifts in play.

There are many reasons people are not driving more, but “saturation” satisfies Occam’s Razor. There is only so much time in the day. For a worker who spends at least 8 hours at the office and 8 hours asleep, how much time is reasonable to actually spend traveling as opposed to the other things that comprise life. Each additional minute traveling is one less minute doing something else. The literature on the travel time budget is rich, and while people do want some separation between their home and work lives, most people do not want to spend too much time (say more than 90 minutes per day) traveling on a regular basis. The travel speeds of current technologies limit distance.

Similarly, there are a variety of complementary hypotheses as to why people are driving shorter distances, some of the important ones include:

  • Price of fuel – higher energy costs diminish travel
  • Size of the workforce – fewer people working leads to fewer work trips (due to both unemployment and labor force participation)
  • Telework – people working at home for the day leads to fewer work trips (but more non work trips)
  • Online shopping – buying over the internet at home decreases shop trips
  • Virtual connectivity – connecting with friends at home can substitute for visiting

Obviously different demographic sectors do these things in different amounts. Just as your grandparents may still receive a physical issue of the newspaper while you read online, young people are more likely to be early adopters than their parents and grand-parents. And the habits formed while young may very well persist over time.

These last three reasons for traveling less by auto (and overall) are due to information and communications technologies substituting for travel. But these are all non-transportation reasons.

There are active transportation modes, like walking and biking, which work well for short trips, and certainly have niches they can grow into if land development intensifies and people reorganize their lives to enable them (for instance, I am one of the 7% of Minneapolitans who walk to work, the numbers are much lower outside core cities, and nationally (3%)).

There are a slew of “new mobility options” which use information technologies to allow travel without owning an automobile, but are not yet visible in the transportation statistics. These include peer-to-peer taxi and ridesharing services, dynamic real-time rental cars, and the like. While these are useful in their niches, they are not cost-effective to be the main transportation mode for the vast majority of the population with the given technology. But they are supplements when the main mode doesn’t solve the job to be done.

Technologies allow people to do more of the same, and they allow people to do new things. It is easier to predict more of the same than new things.

I believe the more important technological changes over the next few decades are those associated with autonomous vehicles. Cars that drive themselves change how people use them. First in “the more of the same” category, we might see more travel.

Generally as the cost of travel declines, travel increases. Since fully driverless cars make it easier to drive (by reducing the cognitive burden on the driver) the initial effect would be that people would travel farther, to places they are less familiar with, and move to places farther from their place of work, to get more real estate for the dollar. Today’s commuter rail passengers travel farther (and longer) than auto users, and autonomous vehicles, where the passenger can do something else while traveling are more like commuter rails than are today’s cars

Such cars also can deposit drivers in front of buildings and park themselves, reducing the amount of time that drivers spend parking and accessing and egressing their cars, which would naturally lead to longer distances.

Third, such vehicles are likely to be safe at higher speeds, since humans won’t be driving, which will also lead to longer distances in the same travel time.

Further such vehicles expand mobility for those who are now restricted (the young, the disabled, and so on).

However such cars also make the so-called new mobility options much more useful in cities. Instead of owning a car, renting on demand becomes much more viable. The right-sized car can in principle be summoned at any time. And if a driver is paying by the minute when the car is used, instead of paying for a car loan or lease by the month (whether or not she uses it), the incentive structure the driver faces changes. Travel will be less frequent and more thoughtful. The daily pattern of transit for routine trips and a “nextcar” for special trips becomes feasible. The lack of nextcar options now pushes people to owning vehicles, and once they own a vehicle, they are going to use it. This lifestyle model works in cities, where transit can be a mainstay transportation mode, and nextcars are conveniently located.

It works less well in the suburbs, exurbs, and rural areas, where the baseline transportation mode cannot be as expensive on a per-trip basis as the nextcar’s rental model requires, but the density is not high enough to support fixed route transit on most corridors.

Obtaining better capital utilization out of our surface transportation fleet (like the airlines have achieved with planes that are in motion as much as possible) through nextcar like vehicles will reduce the lifespan of cars by using fewer vehicles more intensively, and wearing them out sooner. Thus nextcars will on average be newer than today’s fleet. As technology continues to advance with greater rapidity, this becomes increasingly important. The difference between a 2030 and 2020 model likely will be far greater than the difference between a 1970 and 1960 model car.

These are gradual processes. The rapid change in information technology can inform us of the direction of changes in transportation, but the pace cannot be replicated. The technologies are different. Building roads or rails have socio-spatial implications that laying fiber optic cables or constructing cell phone towers do not. The lifespan of a car (15-20 years) far exceeds that of a smart phone (about 3 years), so the technology people possess lags far behind the technology that is possible.

The mountain analogy implies society cannot climb to the peak of the next technology in the same market niche (for instance, serving daily transportation needs) until it climbs down the first. One can imagine a technological helicopter or zip line, or leaping off the peak (abandoning existing function technology, rather than just depreciating it over time) to accelerate transformation, but such sudden changes are rarely wise and even less politically acceptable, with entrenched interests having accumulated power desirous of maintaining (or expanding) the status quo.

If the Future of Transportation does not involve more information technology and more automation, I will be both disappointed and surprised. But the exact shape of what comes next is hard to say. In the 1980s, we had a vision of a future of telecommunications and information that was something like what the internet came to be, all the world’s information at your fingertips. But few foresaw that it would be supported by online advertising. The idea that a collaboratively built online encyclopedia would displace Britannica, and be of the world’s biggest websites, or that an online bookstore (a bookstore!) would be on a trajectory to become the world’s largest retailer were all unpredicted and unpredictable. So it is with transportation in 2014.

2 thoughts on “Mount Transit, Mount Auto, Mount Next”

Comments are closed.