It is an article of faith among urban & transportation planners that auto travel is very expensive and heavily subsidized to boot. They use this claim over and over to justify increased subsidies to transit.
Fortunately, we have very good information on the actual cost of driving, the actual costs of transit, and the actual subsidies to both. We also have reasonable information about the social costs of driving and can speculate a bit about the social costs of transit.
What does it cost to drive per passenger mile?
- According to the Bureau of Economic Analysis, Americans spent $988.2 billion on driving in 2005 (see line 69). This includes capital costs, operating expenses, and taxes, but not any highway subsidies.
- According to the Federal Highway Administration, Americans drove 2.99 trillion vehicle miles in 2005. Counting only passenger vehicles, and not heavy trucks, and it was 2.75 trillion vehicle miles.
- According to the US DOT, the average auto carries 1.63 people, so 2.75 trillion vehicle miles equal 4.48 trillion passenger miles.
Divide 4.48 trillion into 988.2 billion and the cost of driving averages 22.0 cents per passenger mile.
The National Transit Database provides costs and passenger miles for transit. As I explained in a previous post, the data are not terribly easy to read, as operating costs, capital costs, fare, and passenger miles are all in separate files. So I’ve gathered them all into one spreadsheet for you.
According to cells Q1371 through S1378 of this spreadsheet, average capital and operating costs per passenger mile are:
| Mode | Capital | Operating |
| Bus | 0.14 | 0.78 |
| Trolley bus | 0.48 | 1.13 |
| Light rail | 1.46 | 0.57 |
| Heavy rail | 0.24 | 0.36 |
| Commuter rail | 0.26 | 0.39 |
| Demand response | 0.25 | 3.09 |
| Other | 0.31 | 0.45 |
| Total/Average | 0.25 | 0.59 |
It is appropriate to point out that the capital costs in any given year are not attributable to the passenger miles carried in that year but should be spread out over roughly 30 years worth of passenger miles. On the other hand, in the long run agencies will need to sustain a continuous investment in capital costs, so the costs for some modes such as buses are probably pretty close to the long-run costs per passenger mile.
In the above table, the costs for trolley buses and light rail are a little high, reflecting heavy investments currently being made in those two modes, but the rest are probably about right. Since light rail and trolley buses are a relatively small portion of total transit travel, the overall average of 25 cents a passenger mile is also roughly correct.
So transit costs 84 cents per passenger mile, nearly four times the cost of driving. Transit fares average 20.4 cents per passenger mile (interesting how the transit industry keeps fares close to the actual cost of driving), so the subsidies are nearly 64 cents per mile.
How much are the subsidies to driving? That answer is also available from the Federal Highway Administration. In 2005, federal, state, and (mostly) local governments spent $39.2 billion in non-highway user fees on roads. Offsetting this, federal, local, and (mostly) state governments diverted $21.3 billion from highway user fees into non-highway programs. So the net subsidy was $17.9 billion. (By comparison, transit received $30 billion in subsidies in 2005.)
The highway subsidy works out to a subsidy of 0.38 cents per passenger mile. Added to the consumer cost and the total cost of driving is 22.4 cents per passenger mile — still a lot less than transit costs.
What about the social costs of driving? Back in the 1990s, UC economist Mark Delucchi did a massive series of studies of these social costs, which you can download from the University of California Transportation Center. He summarized his work in an article in Access magazine, which is also published by the University of California Transportation Center.
Many auto opponents have attributed extremely high social costs to autos and roads. After reviewing the literature, Delucchi concluded that most of these claims “rely on outdated, superficial, nongeneralizable, or otherwise inappropriate studies.”
Delucchi himself estimated (in the Access article) that social costs averaged about 6.9 cents per passenger mile. Two cents of this was health problems due to air pollution, and he based his air pollution calculations on Los Angeles in 1990. Cars are cleaner today and just about everywhere is cleaner than Los Angeles, so the cost of air pollution should be quite a bit lower. He also attributed a significant chunk of costs to congestion and accidents, but some would say that those costs are paid mainly by the users, not society in general.
Transit also has social costs. Buses produce far more ozone-producing air pollution per passenger mile than cars. The coal-fired power plants needed to supply electric rail transit with energy also pollute. Total social costs might be less than for cars, but they are still more than zero.
Using Delucchi’s 6.9 cents, the total cost of driving is 29.3 cents and the subsidies are 7.3 cents per passenger mile. If the social costs of transit are 2 cents per passenger mile, the total costs of transit are 86 cents and the subsidies are 66 cents per passenger mile. So transit costs are nearly three time auto costs, while transit subsidies are more than eight times auto costs.
This doesn’t mean transit is evil. What this should suggest instead is that the transit industry needs to make itself more efficient. This could be done by introducing competition into the industry or, at the least, contracting out transit to private operators who spend about 60 percent as much, per passenger mile, as public agencies. It goes without saying that an efficient transit industry would not sink billions into rail capital improvements when buses can do everything rail can do at a far lower cost.
Unfortunately, rather than consider these options, many planning advocates just call anyone who proposes them “anti–transit.” That’s like calling anyone who proposes to improve our health care system “anti-medicine.”
Yes, we must remember this research:
And these too:
DS
the team reported Thursday in the online version of the medical journal Lancet.
JK: Got any studies that have been out long enough to have been studied for flaws? Or do you just have this latest, “untested” study?
(At least you didn’t bring up that tired old, unproven, accusation of asthema being caused by cars.)
Of course busses pollute more than if all bus riders were in SUVs. http://www.ucsusa.org/clean_vehicles/big_rig_cleanup/rolling-smokestacks-cleaning-up-americas-trucks-and-buses.html
And electically powered light rail puts out uranium, thorium and mercury from the coal power plants. see http://www.ornl.gov/ORNLReview/rev26-34/text/colmain.html
The only solution is to forbid travel outside of our medeval style village walls. (overpopulation will stop being a problem after the whacked out enviros ban anti-biotics and electricity)
Thanks
JK
Good observations, Randall, but you left out an important cost and two of my favorite modes. The cost of the riders’ time is more than the fares they pay on transit, unless they are just riding to have a warm place to sit or sleep in the winter.
If one were to randomly select real trips in a metropolitan area such as Portland, most of the time I could ride my bicycle from origin to destination quicker than a transit rider who has to walk part way, ride buse(s) and/ or train(s) and walk again. My average speed is less than 13 mph, but I don’t have to wait for scheduled arrivals. The benefits of the exercise to my health and intelligence would partly pay for my time.
If instead of my bike I rode my motorcycle I would get there quicker than 99.5% of transit riders and 60% of car drivers. and have lower overall costs (including accidents and cost of my time) than if I drove a car or rode transit. (My estimates with estimated errors of 5 to 20 %).
I know you have to work with the data as reported, but if you don’t have data for a cost or a benefit doesn’t mean that you should ignore it completely. The demand-response category would look better if the cost of riders’ time were included. HOT lane variable pricing studies give us good data on the value of some commuters’ time.
Got any studies that have been out long enough to have been studied for flaws? Or do you just have this latest, “untested†study? (At least you didn’t bring up that tired old, unproven, accusation of asthema being caused by cars.)
1. Sure. Click the link for the release date. You can then go to The Lancet to learn that it is a peer-review-reviewed journal. You can then also visit a library to view any letters written to the journal about the paper. See also below, and other papers by the lead author, and other such simple research.
2. (unproven accusation) The numerous published empirical studies show internal combustion emissions exacerbate asthma symptoms. There are many many dozens of such papers.
DS
(unproven accusation) The numerous published empirical studies show internal combustion emissions exacerbate asthma symptoms.
JK: Then how come as pollution has been going down, asthma has been increasing?
Thanks
JK
Then how come as pollution has been going down, asthma has been increasing
There are multiple causative factors for asthma exacerbations. Kids sitting inside in our increasingly less porous houses helps increase exacerbations a lot.
But we know auto emissions have an effect. We also know they are not wholly responsible.
DS
So wouldn’t it be cheaper to mandate air exchange in houses than to try to alter how we get around?
Everything’s a tradeoff johng. Energy-efficient homes give us opportunity to collect our pet dander. Nonetheless,
I’m not a building guy, but at my last town my office was next to the building inspector, and I’d hear his conversations all day. There’s – IIRC – code in the Int’l Bldg Code for whole-house fans but AIUI its up to the states to decide to adopt.
But. It’s not solely internal air, it’s external air too, and age, and overall health, and time of year and allergens that affect asthma exacerbations (which costs us $10Bns a year). That link above has a wide range of findings on the differences, but I don’t think there is consensus yet on what causes what fraction.
Wrt altering how we get around, certainly we as a society can do better at internal combustion emissions. For example, I can’t go back to Sacramento in the summer any more, as I no longer am used to the air quality, because the LL O3 in the air is too much for my lungs, and I’m a bike rider and very fit [I used to ride home from work after dark near the end before I moved, as the AQ was so bad]. So healthy people are affected too.
At any rate, full transit in dense areas is more efficient energy-wise [with wide variation] if we’re talking environmental health as a goal (and I’m an environmental health guy, so I am anyway).
And certainly if we start including externalities in market pricing signals, we’ll have to alter how we get around (this, actually, is my area and bubbles up in my other work. I can say that many forces don’t want to include full pricing in market signals.).
DS
Dan At any rate, full transit in dense areas is more efficient [http://fairhaven.typepad.com/photos/uncategorized/passengerenergyuse.jpg] energy-wise
JK: I don’t see the density connection on that chart. All I see is that transit uses more energy than cars. Please explain how it supports you claim.
Dan [with wide variation (http://www1.eere.energy.gov/vehiclesandfuels/images/facts/fotw221a.gif)] if we’re talking environmental health as a goal (and I’m an environmental health guy, so I am anyway).
JK: Nice chart (fotw221a.gif). I don’t see how it supports you claim – it shows that the transit in highest density region in the country uses more energy than a new small car (2600 BTU for LA transit vs 2,488 for a new KIA) see: http://www.DebunkingPortland.com/Transit/BusVsCarTEDB.htm for details of the KIA. It appears that only four cities in your graph beat the KIA for energy usage and they are not Meccas of high density. Perhaps we should avoid high density to save energy.
Dan And certainly if we start including externalities in market pricing signals, we’ll have to alter how we get around
JK: Yeah, shut down mass transit and encourage small cars. Especially diesel buses and their horrific pollution and light rail and its putting mercury, uranium and thorium in the air.
Dan I can say that many forces don’t want to include full pricing in market signals.
JK: Probably starting with the transit agencies. In Portland, I suspect that Trimet would not want to raise its rates 400% to recover its operational costs. Probably have to increase 600-1000% to also recoup construction costs for the toy trains and road costs for the roads that the buses tear up.
Then there is all that nasty military stuff required to get all the excess fuel that buses use compared to small private cars.
BTW, did you know that in Europe, transit market share dropped around 20% in the last 20 years while the private car’s share of passenger-kilos went up TO 78.8%. Rail, bus & streetcar combined only amount to 16% these days. Time to quit trying to copy Europe! (Actually this is just another example of the planning “profession’s†tendency to lie to get people to behave in certain ways.) See: http://www.DebunkingPortland.com/Transit/EuroTranistShareLoss.htm
Thanks
JK
Dan, I forgot to mention that a cleaner version of your first refrence can be found at:
http://www.DebunkingPortland.com/Transit/VanPool1401.htm
I don’t see the density connection on that chart. All I see is that transit uses more energy than cars. Please explain how it supports you claim.
Maybe you are only focusing on buses. Focus on rail and vanpool.
DS.
Dan: Maybe you are only focusing on buses. Focus on rail and vanpool.
JK: OK, so where is the density connection, on that chart, for rail and van pool.
Thanks
JK
I used to live in Sacramento. The problem with air quality there is pesticide and herbicide residue, as it is throughout the Central Valley. As for internal combustion “improvements”, you surely can’t be referring to cars. You probably didn’t notice that CARB (California Air Resources Board) has put out an RFC for better measuring equipment because they want to reduce emissions even further. That’s because, once warm, a modern car’s emissions can’t be measured with current equipment. That’s right, pretty much all auto emissions is from cold start. Given that we measure in parts per billion, that’s saying something. I think it says CARB is due for elimination, they’re just saving their jobs by promulgating emissions requirements that are utterly pointless in terms of measurable environmental improvement. But then, when has any government agency voluntarily shut itself down once their job was done?
I used to live in Sacramento. The problem with air quality there is pesticide and herbicide residue, as it is throughout the Central Valley
No.
That’s only a tiny fraction of the sources [ http://tinyurl.com/2t7nu9 ].
That’s because, once warm, a modern car’s emissions can’t be measured with current equipment.
No.
Surprisingly, that’s incorrect too (full explanation here). BTW, cars sitting in the sun in CA emit significant VOCs from the fuel system, hence parking lot shade regs…
DS
The link to BEA study on the cost of driving is broken. It goes to the FHA instead. Anyone wanting to see the referenced table can go here, and find table 2.5.5
Also, the BTS tracks the cost of passenger miles more directly than the method you used; they track the average cost of owning and operating an automobile. They reckon the cost at $.52 per vehicle mile. Assuming your people/vehicle number applies, the cost is more like $.32 ppm for a personal transportation.
The vehicle occupancy rate you cite is for 2001. On average, vehicle occupancy rates declined about 14.2% between 1977 and 2001; extending that trend from 2001 to 2005 would suggest a vehicle occupancy rate of something more like 1.59 passengers per vehicle. This shifts your numbers very slightly, bringing the cost up to $.33 ppm
Your assumption that capital investments will be ongoing at the initial rate strikes me as misguided. Most cities around the country are heavily engaged in capital improvements to their mass transit right now, but there’s no reason to assume that cost will be ongoing. To get an idea of what the cost might be in the future, it’s useful to look at a city that already has a multi-modal system in place, like San Francisco, where capital expenditures are equal to about 25% of operating, versus the current national figure you’re using, where capital costs are equal to about 41% of operating.
Judah,
Thanks for correcting the link.
The BTS numbers for the average cost of owning an auto are hypothetical. I prefer to use actual numbers, which are the BEA numbers. Planners often use hypothetical numbers and fail to check them against reality.
It is true that vehicle occupancies declined between 1977 and 2001, but you can’t project a continued decline after 2001. All of the decline took place between 1977 and 1996. Between 1996 and 2001 there was almost no change (actually, a slight increase). Vehicle occupancies closely track household sizes (occupancy = household size minus one). Census data show that household sized declined between 1970 and 1990, but did not decline much between 1990 and 2000. So I feel comfortable using 1.6 per vehicle.