Photo by Oran

From time to time, the American Public Transportation Association (APTA) will publish transit savings reports that essentially compare the localized monthly cost of taking transit versus the monthly cost of driving.  It’s obviously a simplified comparison, but a fairly credible baseline to examine the cost differentials between the two modes.  The national savings, on average, amounts to $816 for the month of November, and $9,797 when calculated over the whole year.

The list of the top 20 cities puts Seattle at 4th among the company of San Francisco and Philadelphia.  According to APTA, a Seattleite saves $992 a month on average, just by switching to transit.  The methodology:

APTA calculates the average cost of taking public transit by determining the average monthly transit pass of local public transit agencies across the country.  This information is based on the annual APTA fare collection survey and is weighted based on ridership (unlinked passenger trips).  The assumption is that a person making a switch to public transportation would likely purchase an unlimited pass on the local transit agency, typically available on a monthly basis.

APTA then compares the average monthly transit fare to the average cost of driving.  The cost of driving is calculated using the 2011 AAA average cost of driving formula.  AAA cost of driving formula is based on variable costs and fixed costs.  The variable costs include the cost of gas, maintenance and tires.  The fixed costs include insurance, license registration, depreciation and finance charges.  The comparison also uses the average mileage of a mid-size auto at 23.4 miles per gallon and the price for self-serve regular unleaded gasoline as recorded by AAA on November 18, 2011 at $3.38 per gallon.  The analysis also assumes that a person will drive an average of 15,000 miles per year.  The savings assume a person in two-person household lives with one less car.

What the report doesn’t take into account is the employer-paid fare subsidy for transit users or the upfront vehicle purchase price parking subsidy for car users, both of which represent largely sunk costs that aren’t readily apparent each time a trip is made.  While I suspect that the differential would widen vary considerably* if those accommodations were made, that could only be assumed if mode use were mutually exclusive.  Since many transit users do own cars, it would be interesting to see how numbers would stack up if costs for both were factored together.

*Depending on locality, driving costs could vary with parking costs; the counterclaim, of course, is that in places where parking is free, those expenses are still passed onto the employee in another form.  Quantifying that, however, would be beyond the scope of such an analysis.

48 Replies to “APTA: Seattle the 4th Highest City in Transit Savings”

  1. As it happens I have calculated the incremental or “marginal” cost of transit as compared to driving. I haven’t recalculated based on the 2010 data but in 2009 a car owner would have to reduce their driving by an average of 27 miles a day in order for a transit pass to begin to make economic sense. This is because the marginal cost (gas plus vehicle wear) for a small sedan is only about 14 cents per mile as of 2010.

    What this means is that any policy tools available to reduce vehicle ownership will help shift the balance in favor of riding transit by taking vehicle costs and change them from sunk to marginal costs. I’m a big fan of the car sharing system that’s being experimented with in Paris and of course we already have Zip Car in the US.

  2. You say the fixed costs in the report include depreciation, which implies that the upfront vehicle purchase price is taken into account. I don’t know what else would depreciate in this model if not the price of the vehicle.

    And either I’m way out of average or the numbers are complete BS. Car insurance costs me a little less than $50/month. I rarely fuel up twice a month, but assuming I did, gas is generally around $40 a tank. My “expensive” car was $20K new and it’s 10 years old and going strong. (My cheap truck was $8K new and is going stronger.) With oil changes twice a year and the occasional repair my maintenance is minimal–say $600/year. Plates are cheap–in Bellevue, at least. Parking is free at work–in fact, parking is free almost everywhere on the east side.

    So I’m looking at $50 insurance + $80 gas (this is way high) + $160 depreciation (assuming my car is worth nothing today) + $50 maintenance. So in total I’m looking at less than $340/month to drive, and I rounded up aggressively. If you look at my wife’s case (she pays to park in Seattle) it’s less than $500/month.

    I look forward to seeing this report because it sounds like BS to me.

    I blame the 15000 mile/year estimate. Maybe that works for some cities, but a commuter from Bellevue to Seattle will travel 100 miles a week (10 each way * 10 trips a week) for 52 weeks meaning 5200 miles driven to commute. Seattle’s a small city by any measure so average mileage is low, but when I bing for “average mileage for year” the first page of results shows numbers from 10.5K (supposedly an EPA estimate) to 12K and a couple instances of 15K. Maybe APTA conveniently picked numbers on the high end of the range to make transit look more affordable?

      1. My point about depreciation was that Sherwin says the report doesn’t take upfront vehicle purchase price into account. My depreciation calculations were even more aggressive: I’m taking the total cost of the car and dividing it by the number of years we’ve driven it, which is 100% depreciation.

        The AAA link is interesting. It’s targeted at Seattle, and includes the 15K miles/year number that I believe is ridiculously high in this small city. It estimates the total cost of driving a large sedan at $7707/year, which is $642/month, or close to $200 cheaper than the APTA figures if you assume that riding transit is *absolutely free*.

        I understand that screwing with data to help your argument is standard practice. But this report makes it seem like they can’t do math. This reminds me of when Claudia Balducci told me that it’s cheaper to build a new track along 112th than it is to run the train along the existing tracks on 118th. I just kept thinking, how do you do that math???

      2. Missed the part about depreciation. Fixed now, thanks.

        The 15,000 miles/year is one of three AAA baselines. I won’t disagree that the average annual mileage is probably lower in Seattle but you can’t localize the findings to your own commute. Parking may be free in Bellevue but it isn’t in Seattle.

        And I don’t know why you had to bring up East Link, but I’ll assure you that the engineered cost estimates said that building along 112th was cheaper, not Claudia Balducci.

      3. AP, the reason we pay domain experts like engineers is because often times reality runs counter to the layman’s intuition.

        “How do you do that math?” It’s simple: you actually do the math, rather than make assumptions.

      4. I brought up EastLink because it’s another instance of numbers I don’t believe. I came out of the EastLink discussions feeling like Sound Transit screwed with the numbers so that they could get the result they wanted.

        My larger point here is that it doesn’t help anyone’s cause to push unbelievable numbers. I don’t care if you’ve hired engineers to do the math. If the numbers sound like BS, it weakens your message. Reality may often times run counter to intuition but people with agendas–such as APTA–often times flat out lie.

      5. I agree with AP about the numbers seeming way off. About the only way I can think of that car ownership will cost you $1000 more per month is if you buy a new Mercedes every three years and you drive it every day from your house in Snoqualmie to your job in downtown Seattle, park it in the most expensive downtown parking garage, have a job that somehow pays you enough to afford the new Mercedes every three years and yet doesn’t subsidize your parking, and you have insanely high insurance rates because you’re a 22 year old male who gets a speeding ticket every other month.

        I fully agree with their point that transit can save you money if you can find a way to ditch the car completely. It’s why my wife and I share a single car instead of paying for a second one just so I can get to work a bit faster. The cost argument is worth making, but they should use realistic numbers. People are going to completely dismiss this chart without a second thought because the numbers are so ridiculous. There’s no way we spend $12k/year plus the cost of a bus pass to keep our car running, and few people in Seattle can really say otherwise.

      6. AP, I could point at any project and call BS on the numbers but if I don’t have empirical evidence to back that up, it would be my credibility that is being weakened.

      7. AP: Capital cost is separate from depreciation, so you also need to account for the fact that your 20K is not invested elsewhere — that is, if you could have found a 5% a year return on 20K, you’d be making an extra 1K/year, meaning you’re essentially paying an extra $80/month to have the car. 5%/year is pretty aggressive in today’s economy, so you might choose to estimate something less — but it’s definitely more than 0.

        Other costs a model might take into account include amortized cost of speeding/parking tickets and amortized cost of insurance deductables (that is, the amount you pay in an accident * likelihood of an accident). Neither of those should be very much for most people, of course.

      8. 5%/year is pretty aggressive in today’s economy,

        You won’t pull down 5% on a CD or bond but most people have debt. If you pay down a credit card it may be worth 12% or more. Even on a mortgage most people are paying at least 5%. In fact, the majority of people buy new cars finance them and even if the dealer tells you you’re paying some absurdly low rate in fact the real rate was rolled into the price (i.e. they will always cut you a better deal for cash money).

      9. Sherwin, are you implying something here? I never thought you, in particular, would think I have any credibility on this blog. I don’t agree with everything said here. That alone is damning in some circles.

      10. “In fact, the majority of people buy new cars finance them and even if the dealer tells you you’re paying some absurdly low rate in fact the real rate was rolled into the price (i.e. they will always cut you a better deal for cash money).”


        The advantage of taking public transportation over owning a car is strongest if your alternative is financing or leasing a car.

        If you can buy the car outright, the advantage is much smaller.

        If you can buy the car outright and you always buy used (and are competent enough to get high-quality used cars rather than getting high-maintenance junk), the advantage is even smaller.

        However, horrifyingly, most people really are in the first position, where they can’t comfortably afford to buy a car outright at all. Which means that mass transportation saves an awful lot of money for most people.

        Those of us who can easily buy new cars for cash have a different reason for liking public transportation, namely that we don’t have to *drive*. (Or to put it another way, it’s cheaper than a chauffeur.)

      1. As it happens, ZipCar’s website has a page that compares cost of car ownership to ZipCar costs, and makes ZipCar look way cheaper for occasional drivers. But if your car is past the steep part of its depreciation curve their estimate of ownership cost is way too high. For me (I have a 12-year old car with no associated debt and I make a few short-ish car trips per week typically) ZipCar is basically a wash in terms of monthly costs. The bad parts would be having to reserve cars and keep to a schedule. The benefit would be nicer/newer/more flexible cars. I guess another benefit is that my total assets would become more liquid overall when I sold my car (a car loan can provide the same liquidity, which makes ZipCar’s inclusion of finance charges in car ownership costs sort of fair). Anyway, I didn’t think the benefits were worth making a considerable change in my life, but if an asteroid spontaneously destroyed my car tomorrow there’s a good chance I wouldn’t replace it.

  3. If I have time today, I am going to have some fun ridiculing the figures in this “study.” But, I will start with this.

    If a monthly Puget Pass for unlimited transit in our area costs about $100 (the cost varies), then a person would get unlimited transit use for $1,200 per year.

    The actual operating cost of transit in our area (not including any of the capital costs of transit, as the study includes the capital cost of traveling by auto) is around $0.80 per passsenger mile. So, a person using transit to travel 15,000 miles per year would be getting about $12,000 worth of trips, just in operating costs. But, they would be paying only $1,200 for those 15,000 miles.

    So, a person using transit for 15,000 miles of trips per year would be getting a tax subsidy of about $10,800 per year. And that is just on operating costs of the transit trips — not including any of the capital costs.

    In other words, the reason transit riders save a lot of money compared to motorists, is because transit trips are incredibly highly subsidized by taxpayers, while motorists pay all of the costs of their trips themselves.

    These enormous tax subsidies for transit are really hurting our tax base, as Metro and ST combined are spending over to $1.5 BILLION per year in operating and capital costs to provide just a small fraction of all trips taken in our area each year.

    These gigantic tax subsidies for transit are unsustainable. Transit subisidies are taking way too much revenue away from basics like education and law enforcement.

    1. Please read a basic Urban Economics textbook, and it will explain why a natural monopoly is only efficient if it is subsidized. The average cost curve always stays above the marginal cost curve, so if you charge riders the full cost, the good will be produced at less than the efficient level. It works the same for public utilities and every other natural monopoly. You can argue about subsidies all you want, but in the case of transit they make pure economic sense according to classical economic principles.

      1. This is one of the most curious comments I have read yet, and that is saying something.

        Public utilites are not subsidized, generally. Users pay for their electricity and water — they are not subsidized by sales taxes or MVET’s. If City Light needs more revenue, they raise electric rates — they don’t raise the sales tax. How much subsidy do you think Seattle City Light gets each year from sales taxes, or other taxes?

        Same with the U.S. postal service. They are not supposed to be subsidized — they are supposed to generate all their revenue from charges from customers for sending mail. Postal rates will be going up again very soon, because the Post Office is losing money.

        What is happening with Metro and Sound Transit is that they are exceptionally INefficient. Are you trying to claim that Metro and ST are “efficient” in any way? They are producing their “goods” at vastly higher than the “efficient” level. In what way are mostly empty trains and buses during off-peak hours “efficient”?

        [ad hom]

      2. Hmm…I suspect that the capital costs for utilities are mostly subsidized. Was the entire cost of buying the Cedar River watershed and building the pipes passed on through higher rates? Maybe there’s another reason, I’m not sure. For transit, the average cost per person to operate is higher than the marginal cost of serving one extra person, so in that case it is efficient to subsidize it. The Postal Service is a good example of my point, actually. They should be subsidized–the fact that they are not is why they are going out of business.

        You are using a different definition of efficient than the one economists use. Sure, any transit agency could probably operate in a way that gets the same service for lower cost, but that’s not what economic efficiency means. Economic efficiency means the marginal cost equals the marginal benefit–that produces the greatest societal benefit for the societal cost.

      3. Capital costs for utilities are not subsized. Utilities sell bonds for large projects, and those bonds are then paid off with revenue from ratepayers — not with tax subsidies. Read this story on the $1.75 billion Brightwater sewage treatment plant. King County sold bonds to finance the construction, and the bonds will be paid off over the next 30 years or so with revenue from ratepayers on their sewage bills. Brightwater is not being paid for with tax subsidies:

        You have no idea what efficiency means. Most economists would argue that a capitalist economy is far more efficient than a socialist economy. In general “efficiency” means doing the most you can with the least effort. A full bus is “efficient”. A mostly-empty bus is NOT “efficient.”

        In a capitalist system, you don’t have many empty buses, because they lose money due to NOT being efficient. In a socialist (tax subsidized) system you can have a lot of empty buses, because a government beurocracy decides how many buses to operate and where they should go — not a market-based business entity.

      4. Norman, all public utilities are subsidized. The eminent domain powers are the most blatantly obvious part of it, while the tax exemptions for municipal bonds are another obvious part. But yes, many are minimally subsidized….

        …and most of these low-subsidy privately-owned public utilities *DO* underprovide service, massively. Look at rural electrification (FDR had to spend government money to get it done), or broadband Internet / fiber to the home (which isn’t getting done, right now).

      5. Of course, while these private natural monopolies UNDER provide services, suggesting that they should be government-run instead…. the counterargument is that direct government-run natural monopolies have a tendency to OVER provide services, as the state highway departments to with roads. :-P

      6. Public utilities are not TAX-subsidized. Sales taxes or MVET revenues are not given to Seattle City Light to help pay their expenses, as they are to ST and Metro.

        I’d like to know your definition of the “right” level of service. I would argue that it is level that people are willing to pay for when price is determined in a free market and there are no subsidies. By that measure, transit in our area is obscenely over-supplied.

      7. Stop lying, Norman. I just told you that tax exemptions for municipal bonds are a subsidy — they are, of course, a TAX subsidy. The availability of municipal bonds for utility work is *a tax subsidy*.

        Now will you please stop lying about “no tax subsidy”?

  4. Here’s the link to the AAA data. And here’s a link to Orca card options. If you’re going to approach the mobility of a car you’re going to need the full meal deal $171 pass. And since a car can haul around you and a friend you’ll need a second pass, let’s say at the $3.50 level so you can at least cross county lines on ST and the lake on Metro for an additional $126. And since you’re not going to be able to get a bus everywhere all the time throw in $60 for taxi fares. That’s $357/mo to approach a comparable level of service. AAA puts a mid-size sedan at 57.3 cents/mile or $716/mo. Toss in $150 for monthly parking downtown and you’re at $866. Total savings $509/month, $6,110 a year.

    1. I strongly disagree that most people would need to buy the $4.75 pass if they got rid of their car. The passes are priced so that the break-even point is at 18 round trip rides in a month. Buy a pass that covers the most expensive trip that you would do at least 18 times a month and have some e-purse cash to pay for any more expensive trips that you take once in a while.

      Even with a car, I don’t leave the city of Seattle most days. Thus a pass that covers Metro’s one-zone peak fare of $2.50 would probably be the right one for me. That pass costs $90. Once in a while I would take the light rail to the airport (which costs $2.75) or ride a bus to the Eastside during peak hours (which costs $3.00). That doesn’t mean I should spend the extra $9/month for a $2.75 pass or $18/month for a $3.00 pass. Only if I make that trip 18 times a month would that be worthwhile.

      1. And if you go skiing say 20 times at Snoqualmie and make an early season trip to the Bake and a couple of blue bird days at Crystal… how much does that eat into the savings? It’s really only valid to compare a specific use like commuting. If you can be a one car family instead of a two car family then the savings are huge. If it’s just the cost of transit vs the marginal extra cost miles on the vehicle (~30 cents a mile) then it all comes down to time and convenience. NOT having to deal with traffic and parking in DT Seattle is the reason I will take transit if at all possible. But if we’re going to the opera with the grammas were not going to bus it (more money and way more time).

      2. I’m not sure what this comment has to do with my reply. All I was saying was that buying the most expensive ORCA pass is foolish from a financial perspective unless you’re actually doing the $4.75 Seattle->Tacoma Sounder trip on a regular basis.

        I agree that having a car is more convenient for some things (like going skiing), while transit is more convenient for others (like going downtown).

      3. I think the point is that it’s pretty ridiculous to make a blanket statement that riding transit saves you $1000 a month because you only come up with that number by comparing it to the cost of something completely different (driving 15,000 miles a year). It’s true most people don’t fully realize how much it cost to drive their car. It’s also true most people would pay a lot more before they’d give it all up. And if transit can reduce the car count by one then it’s a huge winner.

    2. Yeah – assuming the most expensive pass is the same as assuming 15,000 miles per year. Both are probably invalid for a whole bunch of users.

      1. Zipcar, like transit is very limited. If you live in on Capital Hill or the U-District it’s great. If you live in White Center or Hunts Point not so much (but I think car sharing and ride sharing will become a lot more prevalent). Taking a bus to get to Zipcar and then doing the same thing in reverse it isn’t so zippy. And hauling the baked yams and beer on the bus to grandmothers house sort of sucks.

      2. Not necessarily. If you’re staying at your destination for several hours or more, Zipcar costs around $80 because you have to pay for it the whole day, even while it’s parked. With a cab, on the other hand, you only have to pay for the miles you actually need the cab for. If you can take the bus one way, the return trip would have to be more than 25 miles for a one-way cab trip to cost as much as Zipcar. Plus, a bus-out-taxi-back travel plan leaves the option open of hitching a ride with someone for the trip back, which, if successful, means you don’t have to actually pay for the cab. Zipcar, on the other hand, you’re committed to paying for the whole day no matter what.

        I’ve noticed a general phenomenon that given the choice of accomplishing a particular trip through either a rental car or a taxi, most people are willing to spend far more money on a rental car. People usually don’t bother to calculate how many miles a trip is rationally compare costs. They simply go taxi=ripoff, period, the end. This is especially true for people traveling. Lots of people will spend $150 on a rental car just to get to and from an airport to avoid a $30-$40 cab fare each way.

      3. I think is true in Seattle. Perhaps it’s an issue with our cab laws, perhaps it’s the expectation that in Seattle you’re going to need/want the rental car anyway. Ultimately the power of operating your own transit system (aka the car) wins out.

    3. Most people are not normally carrying even two people in their car. For making a fair comparison, you don’t need a second pass.

      (Yes, if you normally go everywhere with two people in your car — for instance, if you and your spouse work in the same or adjacent workplaces — the advantage of public transportation is much reduced. This is exceedingly rare in the US.)

      1. It depends where you’re going. For non-commute trips, the carpool mode share (i.e. people in cars with at least one non-driving passenger) is over 50%. So one very realistic option for many people is to commute to work (thus saving money from tolls, gas, and downtown parking), but to use their car for all other trips (for which there are 2+ people to split the cost).

        That doesn’t save you anything on fixed costs like depreciation and taxes, but it sure does help for the variable costs.

  5. Seattle being the 4th highest city in transit savings means that even if fares increase dramatically it’s still one hell of a good deal! Increasing fares by a factor of four would make transit self supporting and it would still be hundreds of dollars less per month in out of pocket expenses vs driving. Service levels could be increased exponentially increasing transit share from 20% to 80% saving hordes of working poor from being slaves to the automobile. In fact, increased fares would pretty much spell the end of drive alone since buses would go everywhere at half the cost of driving.

    1. Major problem with that line of reasoning. You raise fares, fewer people will ride the bus. Because an empty bus costs just as much to operate as a full bus, the operating costs will remain the same even if fewer people ride, and if enough fewer people rider, revenue will actually decline. You can reduce costs by cutting bus service to match the remaining ridership, but then the bus becomes less convenient, so fewer people ride. And you have to raise fares yet again to get fares to pay the operating costs. And this continues until you have zero or almost zero bus service left.

      In general, the more people use a transit system, the less the cost per user, so setting fares so that more people will use it is the best we can do.

      What I think we really need for transit to be able to function without such large taxpayer subsidies is technological improvements to allow buses and trains to be driven by a computer, eliminating the labor cost of hiring drivers. Besides reducing costs, the dramatically reduced marginal costs of operating a bus you already have one more hour each day will mean drastically improved frequency and span of all routes. This means more riders, which means more fare revenue and higher fare recovery rates.

      1. The key question is this. We know that if you increase frequency, you will increase ridership. We also know that if you increase fares, you will decrease ridership. So is it possible to raise both fares and frequency to a level where you have much more service than today, but for the same cost to taxpayers?

        In all likelihood, the answer is yes (or nearly so) for peak-express/commuter service, dubious for frequent arterial service, and no for geographic/social-equity service. The first almost entirely serves choice riders, while the last almost entirely serves “captive” riders, and arterial service is somewhere in between.

        This matches up with what you see in the developing world, where transit tends to be provided by private enterprise. Minibuses and jitneys serve major commute corridors at 1-2 minute headways. Service is so high on these corridors because it’s profitable, and people use it. Conversely, the unprofitable corridors don’t have service.

        One of the cruel ironies of Metro is that arterial service subsidizes commuter service! It should be the other way around — charge high fares to the people who are willing to pay (like Bernie said), and use the money to provide better service for people who can’t.

      2. We did have a chance to make Link driverless, but it crashed on MLK and SODO and sharing the DSTT with buses.

      3. While current technology may not permit a driverless transit vehicle that has any potential interactions with oncoming traffic, that is something that is likely to improve in the future. Already, there have been prototypes of cars capable of being driven, in heavy traffic, by an automated computer.

        In the future, technology will only get better and I do believe that the time will come, sooner than we think, when a computer-driven motor vehicle can be just as safe as a manual-driver motor vehicle.

        When this happens, we will start to see a gradual trend of general vehicle automation. Trucking companies will be keenly interested in using automated trucks to avoid labor costs and provide faster service by eliminating mandatory rest time required for human drivers.

        As individuals see automation in their personal cars, there will be a trend of a car becoming more like an extension of a home, especially with luxury cars.

        And, eventually, the shear costs savings will finally motivate transit agencies to embrace the technology, although I suspect issues such as fare collection, fear of crime, fear of lawsuits, and vociferous opposition from the bus driver’s union, will make city buses the last large fleet of motor vehicles to become automated. But, rest assured, whether it will happen in our lifetime, I don’t know, but I’m confident that unless we have a nuclear war or something, some day, it will eventually happen.

      4. Eric: There’s no question that computer-driven trains are safer than human-driven ones. And even with current technology, computer-driven cars are probably pretty close to being as safe (or safer) as human drivers. But humans are biased to think that we’re smarter and more capable than we really are.

        In all likelihood, computer drivers will need to be 1-2 orders of magnitude safer than humans before anyone is willing to trust them — and by that, I mean the companies that would produce them and the governments that would license them, not just consumers.

        And even after automated vehicles become street-legal, it will be a while before the capital costs are low enough to supplant the operating cost. And then, for a while, bus operators will start by having the operator there to open and close the doors, in the same way that many rail lines use ATO today. (That is, there’s no real operational savings.)

        My point isn’t that it won’t happen… just that the parameters are slightly different than what you laid out. For example, I could easily imagine a system where automated trucks (that weren’t carrying hazardous cargo) were allowed on highways, and there were designated exit zones where human drivers could board and take them to their final destination. Or, conversely, I could imagine a system where automated cars were allowed, but only for speeds of 20 mph or less, and a human driver would need to take over for higher speeds.

      5. Good points, Aleks. I’ve made those points before. It’s actually kind of depressing, but this bias strong humans have of feeling safer when a *human* is killing them than when a *computer* is killing them — this bias seems to be strong and unavoidable, and we just have to deal with it.

Comments are closed.