Among the things that irk me these days are these yuppie urban planning folk. It’s not to say they don’t have fair and reasoned points at times, but the more radical of them espouse the economic and environmental benefits of ‘abolishing’ the suburb and the automobile. I find such a position incorrect and worth of thorough refutation.
To clarify, this piece will look at the features, benefits, and drawbacks of relying on automobiles versus a comprehensive mass transit system. The purpose of a transportation system of course, is to transport people safely, conveniently, speedily, and affordably to and from their destinations, as well as goods and cargo. The United States, which is the country this piece will be relevant to, has taken the route of automobile dependency to a scale unusual among nations, though not dissimilar to the routes taken by other developed, anglophone countries such as Canada and Australia.
Matter of Service
Off the bat, we must recognize some of the basic advantages of private automobile ownership. For one, a private automobile provides on-demand and direct point-to-point travel. Cars can be driven and parked at home, work, the mall, etc., unlike busses and trains that must continue on their respective fixed routes. A bus makes regular stops to pick up and offload passengers but is able to reroute, whereas a train is incapable of departing from fixed tracks. A car, in contrast, can be driven precisely from where a commuter is located, whether at home or elsewhere, to precisely where they wish to go, on demand, on their own time, and via a relatively direct route.
With regard to the other facets of service and amenities of travel, the automobile has several major advantages, and one great drawback. The pleasantries of automobile travel include the customization of one’s vehicle, stereo sound, privacy, control over speed and route, as well as climate control, something that can be quite nice to have in various climates. The great drawback of the personally owned and operated automobile (but hopefully not in the near future) is the requirement that a (or in most cases, the) passenger be their own pilot.
This requirement can be advantageous for those who enjoy driving, but most people, especially those in more densely populated areas or traveling through rush hour, do not appreciate driving in stop-go traffic. And some people don’t particularly enjoy driving in the first place. That being said, there are on the whole some clear advantages to automobile transportation.
Another more contentious issue is the rate of travel, which, as I will demonstrate, works in favor of automobiles, not public transit. According to a study published in Scientific Reports titled ‘Disparities in travel times between car and transit,’ which looked at four cities (São Paulo, Sydney, Stockholm, and Amsterdam), the average trip via transit takes from 40 to 160 percent longer than a similar trip via automobile. Looking at Census data as organized by Mike Maciag, we find a similar picture in the United States, where transit commute times are on average roughly double the car commute times in the 25 metros with the greatest utilization of public transportation.
For example, in the New York-Newark-Jersey City metro area the mean commute via public transit was 51 minutes, compared to just 29 minutes via personal automobile. In Boston-Cambridge-Newton and Washington-Arlington-Alexandria, driving came ahead by 19 and 16 minutes respectively.
Now it’s worth mentioning that these sorts of comparisons are being made between work commute times, not all everyday travel. In the Phoenix Metro (my home), any time on the weekends or outside rush hour traffic (7-9AM and 5-7PM, give or take), you can traverse at an average rate of 70 or 75 mph on the freeway and 45 or 50 mph on the surface streets. This may be relatively weak anecdotal evidence, but is the case in cities like Phoenix and Las Vegas. Fast rates of travel are normal outside of rush hour in most American cities.
The only real argument against the automobile regarding rate of travel is the issue of traffic congestion, a phenomenon almost entirely limited to rush hour commutes. Regardless, even with rush hour traffic the automobile is the fastest mode of transportation in most American cities. Only in places like San Francisco and New York does the automobile begin to lose out, with rush hour travel via automobile typically at 10 mph versus 15 to 20 for inner-city rail alternatives.
When using Google Maps’ travel distance and speed tool, I found the average of high and low estimates for car travel speed in a sample of ten random automobile commutes in ten different major metropolitan areas during rush hour (8AM Monday morning) to be 37 mph, and when excluding auto commutes in Manhattan and D.C., the average speed of travel increases to 41 mph. In the cases of Manhattan and D.C. the average estimated rates of travel were 19.9 and 23.6 mph, slightly faster though mostly on par with the mean rate of travel via the New York subway, albeit without the time required to travel by foot to the station and then again from the disembarkment station to final destination.
Of course in a city like New York the rate of travel via automobile is faster than it might be otherwise due to half or more of daily commutes being made via foot and public transportation, reducing the level of congestion on the roads. Regardless, most cities are not as densely populated as New York City. The fact that a place as dense and expensive to live in as Manhattan manages to operate a transit system only roughly as fast as commute by automobile is remarkable, especially when we begin to consider the economics.
In Washington, D.C., the Metrorail is quite a bit faster (average speed at 33 mph) and thus a more attractive option, but like any innercity rail system, you will find yourself walking a mile or two to the station to board the train and then another mile or two from the point of disembarkment to your final destination, an hour of walking at a brisk pace.
Again, this whole comparison, centered around weekday work commutes, ignores the utter lack of congestion and thus extremely fast automobile travel American cities experience on weekends and outside rush hour. Transit simply cannot match the flexibility and speed of automobile transit absent a large level of congestion, but congestion is not a permanent nor even the majority state of the roads.
Matter of Cost
Perhaps more important than any other issue is the matter of economics, or more specifically cost. The relatively speedy Washington Metrorail I mentioned was built in 1980 at a remarkable expense, as much as $50 million per mile (in 2019 dollars), and extended for nearly $170 million per mile in 1990. Altogether about $16 billion (in 2019 dollars) were spent on building the D.C. Metrorail from 1970 to 2000, an average of roughly $155 million per mile.
I can’t find data for passenger miles traveled but if I were to extrapolate the roughly 174 million riders in 2018 over the 44 years the D.C. Metrorail has operated, and take a guesstimate mean commute length per passenger of around 5 miles, the capital cost alone per passenger mile is generously calculated at around $0.42. Even if a more generous 250 million passengers each year with an average trip distance of 10 miles (for reference, D.C. stretches 8.5 miles from Silver Spring to the Potomac River), the capital cost still floats at $0.15 per passenger mile.
On top of the capital cost (which does not include any capital spending since 1990), in 2018 the Metrorail had approximately $0.14 in operating expenses per passenger mile under the same assumptions. Not a perfect analysis by any means, but any underestimate of miles traveled per passenger should be outweighed by the purposeful overestimate of passenger trips (there were not 174 million trips the year it opened in 1976). When we stack all of this up, we can generously assume a cost of no less than $0.30 per passenger mile, and likely closer to $0.40 or $0.50.
The D.C. Metrorail, of course, is just one single example of innercity rail in the United States, and perhaps not a particularly good one, but as a matter of fact federal, state, and local government together spent $1.5 trillion in 2019 dollars on public transit from 1960 to 2019, and $1.13 on public transit per passenger mile in 2018. With passenger miles traveled via public transit since 1960 amounting to 2.4 trillion, the cost of government subsidization alone, without factoring in transit fee and pass revenues has amounted to more than $0.60 per passenger mile in total over that period, twice my bottom line for the D.C. Metrorail, and closer to what I expect it cost in reality.
I will grant that public transit can be economical but in limited circumstances. For example, the New York City subway system, especially in Manhattan, appears to operate at a very competitive cost per passenger mile compared to cars, and provides generally advantageous commute times during rush hour compared to driving. In Manhattan, which is only about two miles across, an individual is never more than about half a mile from a subway station.
Manhattan does save on the construction and maintenance of streets. I estimate Manhattan’s street capital to be roughly $1.5 to $2.0 billion, or around $1,000 per capita. The Phoenix metropolitan, area according to my estimate, has at least $55-60 billion in road capital, around $11,000-12,000 per capita (roughly 1,200 lane miles at $2 million per), albeit built gradually over time, closer to a cost of $200 per person per year including annual maintenance and repair costs. For New York, of course, the cost is more like $20 per person each year. So in terms of road capital it certainly helps to have density, but there’s another much greater expense that is exacerbated by density.
In order to fit 1.6 million people on a 23-square-mile-island (my city of Tempe has fewer than 200,000 people in an area of 22 square miles), you must build vertically, which is quite important because cost per square foot varies greatly by the number of stories built, with a mid- or high-rise building costing an additional $50 to $150 per square foot in building costs, not factoring in the cost of land. I figure it wouldn’t be fair to price in land because the land in Manhattan is only more expensive because of the demand, not because it is intrinsically more costly to acquire or build on.
In any case, for a decent sized family home of roughly 2,500 square feet, this is an additional cost of $125,000 to $375,000, or roughly $500 to $1,500 per month extra in mortgage expense. For an individual or young couple satisfied with 1,000 or fewer square feet, the additional cost is still great, at least $75-150 per month. The costs of dense, vertically built residential and commercial space overwhelmingly exceed the cost of having a Phoenix-style road system.
With transit requiring such density to be in-and-of-itself economical or speed-of-travel competitive, we must tack on this cost. For a family of two adults with two children the added housing cost of density comes out to appromxately $0.25 to $0.50 per mile traveled assuming 4,000 miles per person each year, and not including the roughly 50 percent additional cost of financing for a home over 30 years. A young couple or pair of roommates in a cramped apartment can tack on at least ten or twenty cents to each mile they travel as well.
Unless Americans will be content affording two-thirds or half as much floor space in their homes and shops, the costs of Manhattanification to our largely suburban cities in order to make public transit cost-competitive will be intolerable, and even afterwards half of all households would still own cars as they do in New York today. We’d still have hundreds of miles of car-friendly street and avenue in each major city, alongside endless freeways.
Now all of this could be written off if the cost of automobile transportation were so much greater, but in fact it is so much less. Even with a combined road capital and operating cost of about $300 per driver per year in Phoenix, at the average 14,500 miles driven by Arizonans each year, that comes out to just a couple cents per passenger mile assuming only one passenger per trip (which means the actual per passenger mile cost is actually even lower).
With an average car lifespan around 125,000 miles, the capital cost of the vehicles themselves should be on the order of $0.22 per passenger mile for the typical new vehicle and $0.02 for road and parking infrastructure. It’s of course possible to buy a nice, 5-year old used car for a half or third of the price of a new one. Next we can include fuel, which comes out to about $0.10 per mile conservatively (before-tax at $2 per gallon).
Maintenance, repairs, and insurance together average about $1,800 a year, or some $0.13 per passenger mile. To be comprehensive we can include roughly $80 billion in annual medical costs and productivity losses from automobile accidents, about $0.02 per passenger mile traveled, and further $0.05 per mile due to emissions assuming a $100-per-ton social cost of carbon equivalent emissions. Finally the mean automobile occupancy is 1.5 persons, yielding a holistic cost estimate of approximately $0.54 per mile and $0.36 per passenger mile traveled via automobile.
Even with the most generous estimate for cost of public transit per mile traveled, forgetting the implicit costs of inconvenience, time, and comfort, the automobile is still either cost competitive or cheaper altogether. On page 10 of the Department of Transportation’s 2015 National Transit Summary and Trends (link), the only form of transit that cost less per passenger mile was vanpool, which is just putting more people in an automobile. The rest of the options varied from slightly to greatly more expensive, before any sort of adjustment for inflation.
The final argument that could be made in favor of public transit is some kind of obscure economic benefit from density that outweighs the heightened costs of vertical construction and public transit. However, when we factor out the excess contribution of the financial sector (30% versus 7.5% for the nation) to New York’s GDP, their GDP per hour worked is roughly in line with relatively suburban states like Illinois, California, Washington, and Massachusetts. The seeming advantage is really gutted by the productivity of highly suburban Silicon Valley and Hollywood, or even the only moderately urban San Francisco.
Thus there would appear not to be an inherent economic advantage in the New York City level of urban density. This is further but less meaningfully demonstrated by the large overall advantage of per capita and per worker GDP in the majority-suburban United States over the more urban Western Europe and East Asia, even on a city-by-city basis.
Matter of Culture
American car and youth culture is a relatively minor issue but I figured it was important enough that I’d touch on it. It is part of the traditional suburban middleclass coming-of age-ritual for Americans to learn to drive and get a car. Families take their SUVs and pickups on road trips and for camping. The automobile gives teens and young adults a way to get around like no other, including in and out of town, as well as a way to socialize with friends and family in ways not possible on public transit.
It’s true that American youth are increasingly putting off attaining a drivers license and learning to drive, but I would expect this to be more a function of the decreasing wealth and affluence of the middle class due to soaring income inequality since the 1960s and 70s and relative stagnation of the 2010s. Per capita domestic auto production has stagnated in the United States, in part due to the increasing lifespan of cars, but also perhaps due to a recently yuppier, more urban, millennial culture. Again it could come down to the lack of wealth and financial security common among today’s young adults.
Either way, the automobile is an important symbol of American freedom, affluence, mobility, and individuality. (For what it’s worth, of course). Not everyone appreciates the automobile. I’m certainly not a big car guy, but I do appreciate driving and having an car.
Matter of Externalities
By far the strongest arguments against the automobile are found in the realms of environmental protection, sustainability, and public safety. There is no question gasoline fueled automobiles are contributing to climate change and particulate pollution. Even with a switch to fully electric cars there would still be a substantial amount of particulate pollution from tires wear and disintegration. Automobiles are relatively dangerous and pollutive, and require miles and miles of asphalt and cement road, highway, and parking.
However, the bulk of these externalities can be dramatically reduced or even eliminated, such as tailpipe emissions via electrification. Tire emissions, for example (which are often criminally overstated), can be reduced through new tire designs, more frequent pressure checks, and new materials. Sound pollution, which isn’t unique to automobiles, can be greatly reduced with sound barriers around freeways, as well as vehicle electrification.
Mass adoption of new safety features such as automatic braking and lane assistance have proven to dramatically reduce auto accident and fatality rates. Indirect emissions from electric vehicles being powered by nonclean energy (again not unique to automobiles) could be solved with clean energy or by offsetting emissions. In short we already wield the technologies necessary to substantively combat the negative externalities created by automobile use.
Perhaps it’s worth mentioning the lesser externalities that do not apply to personally driven automobiles but to public transportation, such as the ease which a disease may be spread. This is not usually very relevant but in 2020 we were made to deal with the global spread of a deadly and highly contagious virus. Public transit certainly did not help quell the spread. Apparently it’s also been argued that public transit has negative social welfare effect in 24 out of 25 American cities (link).
The Futility of Mass Transit
Even if we grant that comprehensive public transit is economically offered in densely populated cities such as London and New York, the reality is that regardless there will be hefty car ownership. In London and San Francisco the rate of car ownership per household is 52 and 70 percent. In ultra-urban areas like Manhattan and Tokyo, the rate is still as high as 22 and 45 percent, respectively.
In Brooklyn, brownstones lining the streets are matched by an equal number of automobiles parallel-parked out front. In Osaka there are double decker and suspended freeways flowing through the city and inbetween high rise buildings and skyscrapers. In the District of Columbia the bulk of residences either have a car parked outfront or in a backspace garage.
Most cities do not feature a such high levels of density, nor the associated car traffic issues. More than a hundred, perhaps two hundred million Americans live in relatively spacious and quiet single family homes, found in thousands of metropolitan suburbs.
Looking Towards Autonomous EVs
Easily the most exciting and disruptive development in transportation technology is the electric self-driving car. Many of the costs associated with automobile operation are looking to fall by double digit percentages, in some cases more than half or two-thirds. What I’d like to do here is summarize the potential for autonomous electric vehicles to not just compete with public transit, but bury it.
Due chiefly to mechanical simplicity, electric vehicles today are capable of stretching their lifespans as far as 500,000 or perhaps soon 1,000,000 miles. For the sake of argument, given a mean lifespan of 250,000 miles and an average production cost including self-driving equipment at roughly $30,000 over the next decade, the capital cost of the typical electric automobiles could come out to some $0.12 cents per mile.
Electric automobiles are managing 100 to 150 MPGe, driving the cost of ‘fuel’ per mile to closer to $0.02. For example, the Tesla Model 3 offers around 250 miles of range for 50 kWh, only $0.025 per mile with the mean U.S. residential electric rate. Next we can stack on the $0.02 per mile road and parking infrastructure cost.
Maintenance and repair costs for electric vehicles have been averaging around three-fourths that of internal combustion engine vehicles, a total of around $900 a year, or some $0.05 per mile. While autonomous vehicles today have a higher accident rate, these accidents are on average much less serious, and 94 percent of automobile accidents are the result of human error. Assuming a matured self-driving technology some years from now, we could for the sake of argument guestimate conservatively an overall reduction in medical and liability insurance costs of about 75 percent (link), giving us a cost of around $0.03 or $0.04 per mile.
In the end, with a charitably increased mean occupancy of around 1.75 persons per vehicle due to increased ride sharing and autonomous taxis, we find that self-driving electrics could bring the national average automobile cost of transportation down to $0.25 per vehicle mile, and just $0.14 per passenger mile. This is a more than three-fifths reduction in cost compared to the entirely human-driven transportation economy of internal combustion engine vehicles, and a remarkable leap in efficiency.
Dedicated autonomous ride-sharing vehicles with a greater mean occupancy and greater utilization could be significantly cheaper, perhaps closer to $0.05 per passenger mile. It’s also worth pointing out that in conjunction with this reduction in cost, the electric self-driving automobile eliminates or greatly reduces the greenhouse gas pollution and accident externalities while offering improved service (autopilot instead of driving).
Another enhancement in service could come from extraordinary increases in traffic flow due to AI and autonomous fleet traffic management. According to one University of Cambridge study published in ScienceDaily (link), driverless cars operating together can increase the flow of traffic by ‘at least’ 35 percent, enough to eliminate congestion altogether at times in a city. According to AI researchers at UC Berkley (link), replacing just one out of fourteen human driven cars with a self-driving car doubled the average car speed. The team also found that AI traffic management in a Manhattan-style grid increased flow by a humble 7 percent.
I believe it can be said that not only is the automobile a superior mode of transportation all-around, it is becoming increasingly superior due to near-future innovations that will enhance every facet of automobile transportation, in some cases very substantially. Public transit by-and-large ceased to be the most convenient and economical means of transportation some fifty or sixty years ago, and never reached widespread adoption in the United States.
Proponents of transit may make the ‘gotcha’ claim that autonomous taxis are akin to transit, and in some sense they are right, but fundamentally an electric self-driving taxi will be a relatively low-occupancy automobile utilizing the road system as opposed to fixed tracks, stops, bus lanes, and bus stops. Further, it is a safe bet than many if not most Americans will continue to own and operate their own personal automobiles decades from now, even as ride-sharing and autonomous technology matures.
In my estimation, it is a great mistake to think public transit is a solution to any transportation problem in the United States, or remotely congruent with the existing city and economic capital arrangement, let alone the suburban American way of life. Americans will continue to seek space and flexibility, the freedom and individuality granted by the personally-owned and-operated automobile. I have no quarrel with the urbanites if they choose to live urban and car-less, it doesn’t phase me. But I will not allow pathological urbanism and its ideological assault on the majority-American way of life to go unanswered.