Is Mass Transit Green?

Brad Templeton:

That transit is a significantly greener way to get around than private car travel almost goes without saying in our thoughts and discussions. Disturbingly, this simply isn’t true. … City diesel buses and electric trolley buses are both mildly worse than the car in energy efficiency. Light rail systems are also slightly worse, on average, though it varies a lot from city to city. Commuter rail and subway (heavy rail) trains tend to be a bit better, but not a lot better.


Passenger Miles Per Gallon

What’s not in these numbers … energy to make and recycle cars and transit vehicles. … to build and maintain roads … and tracks … to extract, refine and ship fuel …

In spite of [these numbers], it is always the green move for any individual to take existing mass transit over their car. That’s because the transit is running anyway, so the incremental cost of carrying one more passenger is indeed less than just about any private vehicle.

This is a common way to analyze marginal costs, but I wonder.  When one rides mass transit one not only makes the train a bit heavier, one also makes it a bit more crowded, discouraging other passengers.  Worse, one makes all future transit planners estimate that a slightly higher fraction of the population is willing to ride mass transit, encouraging them to build more and large transit systems.  It seems to me that this last effect could bring the marginal cost of using mass transit back up to near its observed average cost, i.e., about the same than cars.

GD Star Rating
Tagged as: ,
Trackback URL:
  • nick

    It’s illegitimate to analyze mass transit capital projects based on marginal rather than total costs, and that’s the most common use this kind of study is put to. Worse, they are failing to account for the real options value of taking a car.

  • q

    please label your x axis

    • Tuttle

      I believe it is labeled, as MPG.

    • q

      yes — so i see it is.

      i wonder what “fuel to wheels” means for bicycles.

      • q

        sorry to involute, but brad templeton does answer this question.

  • mattmc

    Where do Maglev trains fit into this? Going 300MPH has it’s advantages.

  • chuck

    the x axis is miles per gallon per passanger for actual average passanger loads.

    Where’s new york city?

    it seems to me that a shorter version of this post would be ‘the best way to start up mass transit is for everyone to build it overnight and have everyone switch the next morning.’

    really though, perhaps the real lesson is don’t do mass transit for geen purposes without also having the economy properly price carbon emissions.

  • Nancy Lebovitz

    When one rides mass transit one not only makes the train a bit heavier, one also makes it a bit more crowded, discouraging other passengers.

    There’s a middle range of optimum attractiveness. If subway stations are too empty, people afraid to use them.

  • Marshall

    In spite of [these numbers], it is always the green move for any individual to take existing mass transit over their car.

    This argument can also be used to justify “flying off to Borneo” for a week.
    The planes are going to fly there anyway, so you’re not significantly adding to the amount of fuel used.


    • Robin Hanson

      Yes, that is a similar argument. People can more easily notice that the number of planes fueled responds to passenger demand; it is harder to notice that the number and size of transit systems similarly responds to passenger demand.

  • Will Pearson

    It looks like the MPG for the car is based on all car travel (long distance as well as commuting). It’d be interesting to see similiar figures for commuting car travel which is generally significantly less fuel efficient, due to the stop-start nature and lower speeds. City MPG vs highway MPG can vary a fair bit dependent on car.

  • Seinberg

    Even if it’s true that mass transit is approximately the same as cars in terms of raw fuel efficiency, it’s certainly far, far more efficient at actually transporting people than cars are. New York City would grind to a screeching halt if it didn’t have its subway system.

    What the measure also misses is that cities like New York don’t need to sprawl as much as a city like Los Angeles in order to grow, since its mass transit system can accomodate such high densities of people. With things closer together, it also encourages fewer highways (NYC/Manhattan is the only major US city not to have a highway running through it), more pedestrians, and more bicycle riders. It takes a shorter distance to get places. Any given block in Manhattan usually has at least one or two people riding a bike, and in some areas there are as many or more bikes as there are cars. In fact, I know very few people living in Manhattan who even own a car, thereby saving space, energy in the production of the car, etc. There are so many ancillary benefits of mass transit.

    • Chris

      Manhattan has no highway through the middle, but it has two highways on the east and west sides (FDR Drive and the Henry Hudson Parkway). It’s also only two miles wide…

      The other boroughs have plenty of highways.

      • Seinberg

        Yes, along the sides. Nothing through it, however, which is what I said. But true enough, re: the other boroughs.

    • adam

      As a metropolitan area Los Angeles is denser than New York. New York doesn’t just mean Manhatten island.

      • Seinberg

        Los Angeles has a far lower density in the metropolitan area. In fact, more than 3 times denser:

        New York City (including the boroughs) — 27,440 people per sq mile.
        Los Angeles — 8,205 people per sq mile

        Manhattan alone — 71,201 people per sq mile

      • Seinberg

        Sorry, that should say 1/3 as dense as New York 🙂

      • Tuttle


        Manhattan + the boroughs comprise only about 1/3 the population of the NY Metro Area.

        NY Metro’s density is 2,744/sqmi.
        LA Metro’s density is 2,665/sqmi.

    • Jess Riedel

      It’s not the New York doesn’t need to sprawl as much as LA because of some fantastic metro system. It’s that it’s cheaper to sprawl in LA because there is more space, so developers do so.

      • Seinberg

        It’s cheaper to build out on Long Island or in New Jersey, too! But there’s still less sprawl in New York.

  • chuck

    I think it is also worth pointing out that for a very long time, mass transit was not about green as much as affordable transportation and ‘livability’ issues, etc.

    When one rides mass transit one not only makes the train a bit heavier, one also makes it a bit more crowded, discouraging other passengers. Worse, one makes all future transit planners estimate that a slightly higher fraction of the population is willing to ride mass transit, encouraging them to build more and large transit systems. It seems to me that this last effect could bring the marginal cost of using mass transit back up to near its observed average cost, i.e., about the same than cars.

    If only there was some alternate universe where we could experiment with this and see what it would be like to have a well-integrated mass transit system in a high population density environment. For the sake of furthering discussion, I propose we call it “London Town” or “Tokyo Ville”.

    Are those systems subsidized? Are the subsidies out of proportion, per passenger, to how road systems are subsidized through road building and policing and land use?

    Seriously though, it sounds like you are contemplating how mass transit gets critical mass? I think it is like any other networking problem, where a two-node network is not very valuable but 1000 node network is highly useful, even if it is crowded at peak times. So you just carry the costs until the network builds out knowing that it’ll pay for itself in the long run.

    You should read Matthew Yglesias if you’d like an intelligent liberal viewpoint on this. One classic ‘conundrum’ is how you get efficient mass transit in cities/suburbs built around cars. It isn’t a snap, but you’ve got to change zoning to allow density and you’ve got to plan.

    There are 3 kinds of wealth – energy, knowledge, and networks.

    • snogglethorpe

      FWIW, rail transport in Tokyo (for the most part) is privately run by [a rather large number of] profit-making companies (even the previous public rail [JNR] has been privatized), and does not receive government subsidies.

      As I understand it, the government does sometimes subsisize particular improvements to accomplish goals of its own — for instance, paying a rail company to turn a surface line into a subway (presumably to free up the space for development).

      • Michael Turner

        There are in fact massive government subsidies for construction of rail in Toko — since 1962, about 70% of subway construction costs are covered by governments.

        Since much of the ongoing cost of any large infrastructure project is the cost of capital, if you don’t consider government subsidies for construction and possibly the provision of credit on relatively easy terms, you’re missing a big part of the financial picture. What’s the difference between a government subsidy and a loan at a rate of interest hardly above the inflation rate? You tell me. How about if you can remove capital costs from your balance sheet, report a “profit” as a result, and thus qualify for a subsidy proportional to that “profit? Well, then you use the subsidy payments to cover interest on the loan, right? Maybe even pay back some of the principal.

        Don’t get me wrong — I love the train system here. I think it’s the way to go. But mass transit has always required subsidies, and probably always will.

  • magfrump

    9 passengers on a bus? Most buses I ride have much more than that, often more than 20 passengers, which would presumably make them twice that efficient (give or take).
    Passenger numbers aren’t listed for San Jose Light Rail, Trolley buses, or heavy rail, so I’m wondering how you might expect to adjust those.

    Also consider the massive adjustments to cars for having 1.57 average passengers versus solo; I think the real lesson here is that carpooling is efficient at the same level of mass transit.

    • Mike

      The study claims to use average passenger loads. If it seems there are always more people on the bus when you ride it, this is a selection bias.

    • Tuttle

      “Also consider the massive adjustments to cars for having 1.57 average passengers versus solo; I think the real lesson here is that carpooling is efficient at the same level of mass transit.”

      If traveling with 1.57 people in car is roughly equivalent with mass transit, then would not carpooling with 3 or 4 people actually be far more efficient than mass transit?

      • magfrump

        Traveling with 1.57 people is on the low side, so carpooling with 2 people would be efficient on the same level, which is what I meant. Regardless my point was that eliminating solo drivers is a good basis for efficiency.

        I realize that my observations will deal with selection bias somewhat, but that steers the argument towards making more efficient bus routes which will have higher numbers of passengers (say, by eliminating those routes which regularly have fewer than 3 passengers).
        Also, as has been mentioned many people who ride public transit don’t have cars, which means they aren’t building cars or parking, which can make them more efficient by other measures.

      • Tuttle


        If you intend to increase “efficiency” by getting rid of solo car drivers, wouldn’t you also want to get rid of any bus routes which average under 7 passengers or so?

        Perhaps everyone should just sit at home.

    • q

      it’s probably selection bias.

      however, the above chart indicates that there is a clear case for density and mass transit on transportation energy efficiency grounds.

    • Nick Tarleton

      As a bus passenger, you should expect to observe a higher-than-average occupancy, since you’re more likely to be on a more popular bus.

  • q

    clearly the more miles one goes the more energy you use.

    therefore it seems a reasonable strategy to give people economic substitutes within smaller radii. then they can have utility that is as high as it is now but travel less.

    but my premise is that people travel as much as they do for a good reason, ie to go to work, shop, school, recreation, etc. hardly any of that travel is in and of itself desirable.

    in my book, that means communities which are either higher density or which have more diverse commerce.

    i’d be willing to bet that few people actually _enjoy_ most of the time they spend in their car. some do, and some enjoy some of it — probably a lot like smokers and smoking — but i’d be willing to wager that people spend far more time in the car (or on transit) than they would wish to.

  • Andrew

    In general it’s best to be skeptical when someone pipes up and says “I’ve looked at some data in a field I know little about, and guess what? All the people who’ve spent their lives studying this field are wrong!”

    Case in point. Templeton takes his data on fuel efficiency per passenger-mile from a USDOT table, which includes a very clear disclaimer that the information is based on national averages, NOT merely averages of car use in areas where transit is available as an option (i.e. mostly cities and a few suburbs).

    In cities, motor vehicles are less fuel-efficient than on highways, by a variable amount but usually by about 30% more MPG than city fuel efficiency. However, the data which Templeton simply compares transit (intercity buses and rail are listed separately) with the national average for automobiles. In reality, mileage depends on the layout of the city, the locations of residential and work areas, etc. In short, congestion matters. The average for a car’s mileage is about 11,000 BTU/pm in New York.

    There is another factor to consider. Mass transit affords an opportunity to switch a motor vehicle fleet to energy-efficient vehicles much more quickly and seamlessly than the process of individuals making private, uncoordinated choices to switch to energy-efficient vehicles. Thus, when a city with 100,000 daily bus commuters switches to an all-hybrid bus fleet that is 50% more fuel-efficient than the existing buses, it is as though 100,000 cars in Montana (or wherever) were swapped by their owners for 50% more efficient models. So in NYC to take an example where I have data, the BTU/pm for the average private car is actually about four times greater than the same figure for buses.

    Rather than interpret the data himself, Templeton should really have consulted with actual transit planners to clarify issues on which he was uncertain, instead of concluding that his counter-intuitive results constitute bold “myth-busting”.

    • Seinberg

      This stands to reason. I find it very telling that nobody has bothered to respond to your message. The subway is surely similarly more efficient with the density of people who can ride it.

  • Tom Swann

    Is the MPG label right?

    If so, it’s clearly not the right unit.

    Should be MPG per Passenger.

    • Tom Swann

      Silly me, passenger miles per gallon.

  • AP

    MPG alone is a poor measurement of how “Green” a particular method of transport is. The graph above doesn’t take into account the different types of fuel/energy each method uses, e.g. 100 miles in a Tesla charged by solar panels is obviously orders of magnitude better than 100 miles in a Jet ejecting C02 into the toposphere, or wherever it is they fly, which, I gather, is more harmful by far than doing so at ground level.

    Most inner-city rail systems are electric powered, which, depending on your sources of electricity, could produce more or less carbon NET than traveling the same distance by petrol.

  • Unnamed

    Could we estimate the marginal energy cost of an additional mass transit passenger by looking at the effect of population density (since density adds passengers)? If places with higher population density tend to have more energy efficient mass transit systems, then that suggests that the marginal passenger uses less energy than the average passenger (taking all of the system-wide effects into account).

    • Tuttle

      Isn’t it a given that a marginal passenger requires less energy than the average passenger – for all transport types?

      If it’s true that the marginal passenger makes a bus more efficient, then it’s especially true that the marginal passenger makes a car more efficient, is it not?

      After all, with 1.6 passengers in the average car, the marginal passenger will have a much larger impact on efficiency that on a bus carrying 9 people. Even if the car has to drive a bit out of its’ way to pick up and drop off the marginal passenger, it should still benefit the car more than the bus on net.

  • Roland

    Conclusion: Mass transit only works if it is USED. Just having buses and trains driving around half-empty doesn’t work.

    The advantage of mass transit only plays out if it is really heavily used. In this case you will have the additional advantage of a drastic reduction of traffic congestion due to the reduced number of cars

    To achieve it cities need to implement mechanisms to reduce the number of cars, like charging hefty prices for driving in the inner city.

  • Brad Templeton

    I am aware — and cite several times — that these are national averages. The car figure in the DOE’s energy fact book is based on the EPA’s 22 mpg combined average, and that does include highway and city combined. And the transit systems are also averaged over cities and suburbs. Many cars do better than 22mpg city mileage and modern hybrids like the Prius will do more than twice as good as that in the city. Electric cars much better.

    But understand that, as I write but perhaps people didn’t read, most of the adjustments that can be considered, such as difference from national averages, well-to-wheels vs. fuel-to-wheels and so on at most make modest adjustments to the numbers.

    I believe, as I said at the start of the article, that most people work from the presumption that (as it is in Asia) transit is much better than private transportation. In the USA this does not seem to be the case. In some cities it is modestly better, perhaps up to twice as good, and in other cities it is actually a worse performer.

    If transit is only modestly greener, it’s not going to be a winning choice, not with private transportation’s overwhelming advantages in convenience and comfort in most cities. (There are cities where parking issues and congestion at rush hour make transit win in these categories, but these are the exception, not the rule.)

    The numbers seem to indicate that the actual winner would be small efficient vehicles, like scooters and city cars, particularly electric ones, beating both the cars and transit in every way, except in capacity at rush hour. Greener, cheaper and more convenient than both. Slightly less comfortable than luxury cars of course, and until robocars come, with the burden of requiring attention to drive.

    Read the follow up article linked in the blog post for notes on that. A city bus, for example, has 3,000lbs of bus per passenger on average, and still has 500lbs of bus per passenger when fully loaded. And it keeps starting and stopping. It’s not hard for electric scooters to do much better than that.

    • Andrew

      Intellectual curiosity into subjects which are not of one’s own field of study is highly admirable. But amateur interpretations of data must be coupled with expert guidance.

      I have seen the models used by engineers, and in New York City at least, the bus fleet has a fuel efficiency per passenger-mile which is two to three times higher than the average private car in the city. These models do indeed account for factors such as weight per axle. Doing otherwise would be foolish.

      There are real people (I am lucky to know some) who study transportation engineering for years and are paid real money to make the most accurate estimates possible given our current incomplete understanding. They rely on hundreds of pages of data and research and on computer models that took years to develop, not simply on one or two tables of data covering all rural, urban, and suburban areas in one of the world’s most populous nations.

      Before dilettantes dismiss the work of professionals with phrases like “does not seem to be the case” and “appears to be a myth”, a sense of intellectual humility and cognitive flexibility are called for.

      • Brad Templeton

        I welcome your cites for the figures you quote. The data I was using specified MTA motorbus figures at 1.95 million passenger miles and 5.4 billion BTUs for a BTU/pm of 2790 which is better than the average car nationwide, and I don’t have a problem with the suggestion that this is twice as good as the average NYC private car, though 3 times seems a bit high.

        Nobody doubts that NYC is the most transit-using city in the USA, as well as one of the most congested. And nowhere in my article do I state otherwise. I say several times the figures are national averages. Rather than just telling me I’m an amateur getting it wrong, it would be preferable if you correct we with your own cited facts — and that you correct what I actually wrote, rather than what you imagine I wrote.

        Other factors that may lead to misreading of the numbers: The figure for cars is for cars. It does not include trucks and SUV, which have their own bar on the chart. While I have not seen NYC specific numbers for cars, they might also be biased by incorrectly including cab drivers in car occupancy.

        Give me solid cited facts and I will correct my articles gladly. Speak in ad hominem and you will not receive further response. Show me how an MTA bus is more efficient, on average, in New York, than a Prius, especially one with 2 people in it.

  • Phil Goetz

    I know I should be asking Brad Templeton – but are these figures comparing city-driving mass transit to highway-driving cars?

    Also, I’d like to see a passenger-miles-per-death bar graph, too. I would guess mass transit would win, but don’t know.

    • magfrump

      Seconded on the passenger-miles-per-death front.

    • Brad Templeton

      Yes, I believe it wins on the deaths, especially national averages. Traffic deaths are more common on high speed roads than low speed ones, in addition.

      The figures compare all transit to all cars. Most cars get better mileage on the highway than on city streets, but hybrid cars like the Prius, and electric cars, get better mileage on city streets. Transit figures include both city street transit and longer range transit, such as commuter rail which competes with highway. Commuter rail tends to be more efficient than intracity transit, because it has higher loads, makes fewer stops, and often doesn’t run much at all outside of rush hour.

      However, don’t forget the core point. We usually work from the assumption that transit is much more efficient than private cars, and the assertion that transit is a bit more efficient than cars on city streets and less efficient than cars on the highway does not alter the conclusion that this is a poor assumption.

      If you prefer to ask a more specific question, “which is greener, a Prius or a city bus” many would be surprised to learn it is the Prius, by a good margin. (Leaving out the debate on the energy cost of making both of them.)

  • Seth

    “Worse, one makes all future transit planners estimate that a slightly higher fraction of the population is willing to ride mass transit, encouraging them to build more and large transit systems.”

    As opposed to adding more lanes on highways and arterial streets?

    On another note, one thing that bugs me about mass transit studies is that they don’t provide all-in cost comparisons. I’d like to know the breakdown of costs per person mile of everything. For example, road construction/maintenance vs. rail construction/maintenance.

    • Robin Hanson

      Yes a cost of driving cars is inducing more roads to be built.

  • Pingback: Tuesday morning links - Maggie's Farm

  • Bill

    Not a very dynamic analysis.

    For example, every person taken off the road with mass transit increases the mpg of the remaining persons on the road. Persons on the road experience more congestion and accidents with more persons on the road; taking persons off the road result in fewer fatalities.

    When you start viewing transportation as a system, it is hard to tweek one variable and not affect another component. You might want to model SYSTEM costs and SYSTEM mpg under various mixes of public transportation and car travel.

  • Phil Goetz

    If you’re talking about being green, you should ask about pollutants, not just about MPG. A motorcycle gets great gas mileage, yet typically pollute more than cars, because larger engines burn cleaner. Rail systems using electric power generation probably generate a lot less pollution per passenger mile.

    • Brad Templeton

      This depends a lot on the local grid. The U.S. grid is 50% coal, 20% natural gas, 20% nuclear and 10% hydro. In some places it is much heavier on the coal, in others more hydro.

      Coal is one of the dirtiest sources of power. It does CO2 but a whole lot more, and then has coal ash residue when done. Natural gas is CO2 and not too much else.

      As for hydro and nukes — well, no direct air pollution, but many people feel hydro has destroyed valleys, ecosystems and species. And people are of course mixed on how clean nuclear power is — some think it is the riskiest and dirtiest power, others think it is the cleanest.

      In terms of air pollution, coal is worst, diesel 2nd, gasoline 3rd, natural gas 4th, nuke and hydro much lower. It’s an open question if an electric train that gets half its power from coal and 20% from natural gas is cleaner than a diesel bus or gasoline car. By CO2, a bit cleaner but by other pollutants, not so clear.

  • michael vassar

    It is worth at least pointing out the MUCH lower net energy consumption per capita in mass-transit heavy cities like NYC as a suspiciously relevant datum in this discussion.

  • Ryan Carey

    If public transport runs closer to capacity, it’ll be pretty damn efficient.

    just because there are some plausible mechanisms of action for sitting on a train to encourage more trains to run does not mean that the marginal energy use of sitting on a train is the same as the average energy use of sitting on a train.

    Making the train fractionally heavier is trvial. Crowding is not so important, presumably its effect could be markedly reduced by sitting in the corner. As for the future transit planners, if they enlarge transit systems, so that there are more trains, this will not only reduce crowdnedness but also reduce waiting times for trains, encouraging trains to be fuller.

  • Pingback: Alternative Energy