The folly of measuring transportation costs per passenger-mile

When comparing costs of various modes of transit, units measured “per passenger-mile” are very common. It makes sense intuitively – people take trips of varying length, and longer trips are more expensive than shorter trips, so the desire to standardize and compare makes us want to simply divide the trips by their length and call it even. Both supporters and opponents of light rail use per passenger-mile costs and subsidies to justify their positions, the government keeps tabs on them, and Randal O’Toole at the Cato Institute has even used carbon emissions per passenger-mile to claim that cars are more environmentally friendly than rail.

The problem with measure at the rate of distance traveled is that the purpose of transit is not to travel long distances – these are not pleasure travelers trying to get as far from home as possible, but rather commuters trying to get to wherever their jobs and schools are located. But the distance to this “somewhere” is not a variable to be held constant – it actually varies with population and job density, which is highly correlated with mode of transit. Places with train lines generally have and allow for denser development and thus less distance between your house and your workplace or school – the difference in average commute distance between urban and exurban areas could be as much as an order of magnitude.

Measuring costs in terms of total costs per person and not per passenger-mile makes intuitive sense when you think about it in terms of personal finances. If you lived in Brooklyn and your office moved from Manhattan to somewhere out in suburban Long Island and your transit expenses rose from $89/month for a Metrocard to $300/month for the cost of the car plus insurance, gas, and upkeep, it would be hard to trick yourself into thinking you’re saving money. Your cost of commuting per-mile might have fallen, but you have to travel more miles to do the same task. This would obviously apply to your carbon footprint as well – emitting two pounds of CO2 to travel four miles by car is still more polluting than emitting one pound to travel one mile by bus.

Walking is also possible in high- to medium-density, transit-accessible areas, which reduces some numbers in the denominator of the cost per passenger-mile equation to 0, further heightening the disparity between total cost and cost per mile. And while walking is rarely an option for high-income earners commuting to work in cities, it is often an option for both quick shopping trips and low-wage local employment. The short car rides to the store to buy some milk and short car commutes to the local fast food restaurant to work that take place in the suburbs are within easy walking distance for most urbanites, the use of wheeled transportation being eliminated entirely from some trips while still retaining the same basic utility.

Of course commuting by car and by transit are not perfect substitutes. Cars can be more comfortable, walking gives you more exercise, and riding a train or bus allows you to devote more of your attention to other things. Some people have scenic drives to work, but some people enjoy the vibrancy and people-watching of walking to and then taking the train to work. People enjoy these activities in different proportions, but financial costs are a huge component of most people’s decision-making calculus, and if individuals aren’t going to be making decisions with their own money, then the government and commentators should at least be accounting for these subsidies and environmental costs properly.

  • Michael Lewyn

    I blogged about this issue awhile back at

  • Alon Levy

    “Randal O’Toole at the Cato Institute has even used carbon emissions per passenger-mile to claim that cars are more environmentally friendly than rail.”

    Alas, O’Toole is blatantly wrong, as a quick check of heavy rail’s per-passenger-mile emissions in the US will tell you. According to the FTA, average subway/el emissions in the US are 0.16 pounds per passenger-mile, which is emissions-equivalent to 82 passenger-mpg of gas. Drive a Prius on an average (probably rural) road at average (probably non-commute) occupancy and it’s still less emissions-efficient than rail in congested urban areas.

  • Stephen Smith

    I believe O’Toole averaged in bus emissions, too, with the idea that rail-based transit needs feeder buses to be effective.

  • Plucked

    I feel the need to start with pointing out a significant problems in your example – while it may cost the individual $89/month, the cost is substantially higher when you include all the subsidies for transit. Therefore, if you do not compare actual costs, it would be just as logical to provide massive subsidies paying for 80% of the cost of owning and using a car. If you do that, the car is cheaper. Then do you really want to get into the ludicrous of comparing different commutes? I can drive a car 10 miles to work, but what if my office moves from DC to New York? I need to take the Acela to work every morning – that is so much more expensive than owning a car, and thus mass transit is absurdly worse.

    Per-passenger-mile is the rational metric for comparison. Take what people actually travel and compare it based on how people get around – it avoids absurdly meaningless comparisons like those you cited.

    How do you propose accounting for the costs of transport in a way that is truly comparable? How do you divide the cost of mass transit if not by passenger mile? If a bus spends 90% of its route with 3 people on board, and the other 10% with an additional 30 people on board, do you propose dividing the cost of the bus by 33?). Take a Hummer and cram as many college students as you can into it and drive 10 feet – voila, now I get to divide the total cost and mileage of the solitary cross country road trip by the 17 students we fit in there.

    Per-passenger-mile is actually fairly favorable to transit, as trips are longer than they would be by car. Instead of driving a car directly to work, I could take a bus to the train station, take the train to another station, hop on another bus, and I’ve just traveled nearly twice as many miles (of course, I also happen to think the extra 50 minutes it adds to the commute to be significant, but convenience is hard to measure appropriately in mass populations, and thus is left out).

    If you can demonstrate how to measure the travel of hundreds of thousands of people in a more effective way than per-passenger-mile, you would shock the world and truly make a name for yourself. Go ahead – fabulous fame and fortune await.

  • Al

    If your office moved from DC to New York, I don’t think you would take the Acela– I think you would either move, or get a new job. However, if you did commute, are you seriously suggesting that you would drive 200 miles each way, and that this would be preferable to taking the train (which, I might add, is one of Amtrak’s few profitable routes)?

    New York City’s subsidy is around 33% for the subway. ( Yet you want an 80% subsidy for your car, on top of the road subsidies, and you call that logical?

    Per passenger mile is a ridiculous metric, unless the people are making the same trip in both cases– but they’re not. Consider: one person lives in a dense neighborhood, walks or rides a bike to do his shopping, and takes the bus five miles to work, which is subsidized by a dollar. Another lives in a distant suburb on a cul-de-sac and drives 25 miles to work, then drives another 25 miles for shopping and entertainment. This is subsidized by two dollars. By your standard, this is an improvement, since the subsidy’s gone from 20 cents to two cents a mile, and yet the more people switch to it, the more subsidies are required!

  • Pingback: Is “per passenger km” the right metric for comparing modes? « The Melbourne Urbanist()

  • Kurt Johnson

    Government doesn’t need to, “be accounting for these subsidies and environmental costs properly.” Government needs to stop all subsidies for any specific type of transportation. Mostly, we should have a “users pay” system.

  • Pingback: The folly of measuring transportation costs per passenger-mile | SMART()

  • Pingback: electronic music()