Technology Obsessions in the U.S. National Transit Database

The US National Transit Database (NTD) is out for 2008! (Yes, a one year lag counts as fast in this business.)  For those of you who aren’t compulsively drawn to spreadsheets, Cap’n Transit tries to get to the bottom of farebox recovery, the percentage of operating costs paid by fares.

Farebox Recovery Ratio Agencies
70-200% Lincoln Tunnel buses, inclined planes, Hudson River ferries, SEPTA [Philadelphia] trolleybuses
40-69% Big city rail, college town buses
30-39% Big city bus and light rail
0.1-29% Small and medium city bus and light rail, plus assorted boondoggles
0 Free services


Farebox recovery is not, of course, the only thing that matters.  The ratio combines three separable factors that have completely different significance, and that can be explored separately in the database:

  • Operating cost per hour of service, which is mostly about labor rates and management efficiency.
  • Riders per hour of service, which is about the efficiency with which the system attracts ridership.
  • Average fare revenue per rider, which is about fare levels, fare structure, and fare enforcement.

(The farebox return ratio (fare revenue / operating cost) is the product of the latter two factors, divided by the first.)

My own frustration with the National Transit Database is that it assumes that technology is the most important distinction between transit services.  This is not an ideological statement on the government’s part, more just an effect of the way transit agencies tend to keep data.   A big transit agency has separate operating divisions for rail, trolleybus, standard bus, etc, so it’s easy to report data in those terms.

But the “bus” category combines a whole bunch of totally unrelated things that happen to be done with the same versatile vehicle, including:

  • Frequent Networks.  Frequent, all day service on major lines, carrying passengers both ways all day.  Justified by ridership goals.
  • Peak Express.  One-way services running only during the rush hour, running full one direction and usually empty out-of-service the other.
  • Local Coverage services.  Infrequent services wandering around in low-density areas, carrying few people most of the time.  This is every agency’s lowest-ridership service.  It is never justified by ridership goals, but rather by social needs or expectations of equity, or what the Brits and Aussies call “social inclusion.”

In most networks I look at, at least 80% of the service falls clearly into one of these categories.  The largest gray area is between Frequent Networks and Local Coverage services.  Sometimes a line is a Frequent Network spine with Coverage branches, or sometimes a line is just on the edge where it might be grown in to a Frequent Network route or might be shrunk into a Coverage service.  In my experience, good network planning has the effect of increasing the separation between these two types of service, and thus increasing the clarity of the agency’s thought about the conflicting purposes that they represent.

So the regular “bus” category is a mixture of very unlike things.  On the other hand, the Philadelphia trolleybuses jump out in Cap’n Transit’s table because they are all Frequent Network services.  The infrastructure that trolleybuses require is cost-effective only if you use it intensively, so there are not many low-frequency trolleybus lines in the world.  The Philadelphia trolleybuses are an especially small group of routes very focused on high demand areas.  But the ridership outcomes don’t have anything to do with the fact that these are trolleybuses.  It’s just that trolleybuses have to run frequent services, and are thus a subset of Frequent Networks, which attract high patronage.  Most Frequent Network services are run by standard buses, but the standard bus can be used in so many other ways that the NTD category “bus” lumps them all together and thus obscures this crucial distinction.

In fact, the Cap’n’s table, and the NTD in general, are great for throwing up spurious statistics about the performance of different technologies.  Trolleybus manufacturers can say:  Look how productive trolleybuses are!  This is another great example of the statistician’s warning:  ‘Correlation is not causation!‘  Trolleybuses can only be used in high-ridership places, so they tend to end up with high ridership if you look at them in isolation.  But that doesn’t mean you need the trolleybus to get the ridership.  What matters is where the service runs, how frequently, how fast, and how reliably.

18 Responses to Technology Obsessions in the U.S. National Transit Database

  1. Danny February 2, 2010 at 8:38 pm #

    I know this is asking for more confidential information, but I would like to see the FRRs of the individual routes of the companies that run the XBL. I know the XBL provides a distinct advantage that helps to raise fares and ridership, leading to higher FRRs…but I also know that those same bus companies run many routes that don’t use the XBL. Are those routes also profitable? If not, then why would a private company run them at all?
    As a lot of people here have found out about me, I’m a big fan of profitability in transit. I think that the only way transit can ever be truly sustainable is if you don’t have to rely on subsidies that may or may not be there tomorrow.
    I understand that most of time the goal of a transit service isn’t to make a profit…but shouldn’t we still be doing more to learn from the companies that do make profits providing transit? Isn’t a feeder line much more likely to stick around during budget cuts if its FRR is 80% as opposed to 10%? And even if you do have a few “equity” lines running 10% FRRs, wouldn’t it be nice to have a few frequent service lines running at 150% to be able to subsidize those lines?
    I know some people here might think I’m passionate about beating dead horses, but really I’m just passionate about transportation. I think profitability is the most crucial underlying issue in a transportation network. It pains me to see arguments about the pros and cons of different types of rail technology that never mention a thing about cost or sustainability…because it doesn’t matter how aesthetically pleasing, high capacity, smooth riding, or fast your transportation is if it is cut from service when you need it most.

  2. Rhywun February 2, 2010 at 9:28 pm #

    My understanding was that trolleybuses are more economical in the long run–witness their extensive use in the olde Warsaw Pact countries–and the only reason they aren’t employed more frequently is due to the higher startup costs (i.e. lack of foresight). One could just as easily argue that there aren’t many low-ridership trolleybuses in the world *because* of this bias.

  3. Aaron M. Renn February 2, 2010 at 9:28 pm #

    Bus and rail are also part of an integrated system. When there are frequent transfers between modes, it becomes a tricky revenue and cost accounting challenge.

  4. Cap'n Transit February 2, 2010 at 10:06 pm #

    Thanks for the link, Jarrett!
    Danny, if you look at the schedules of the XBL companies, you’ll find that some have very few runs outside the XBL.
    Take southern Rockland County, for example. The lines that take middle-class commuters through the XBL (and also into the George Washington Bridge terminal) are run by Red and Tan Lines (a.k.a. Rockland Coaches, now owned by Stagecoach) with a farebox ratio of 79%.
    Most of the Red and Tan commuters have a car or two at home, and drive everywhere except Manhattan. The commuters who don’t have cars (and tend to have lower incomes) rely on the local buses operated by Transport of Rockland, a public agency run by the county government with a farebox ratio of 18%.
    In short, there’s some cross-subsidy, but not as much as you might think.

  5. EngineerScotty February 2, 2010 at 10:12 pm #

    A big part of the trolleybus-vs-combustion choice is which power source costs more money. In many parts of the world, there are plenty of powerplants but fossil fuels are expensive; in much of the US (in particular outside the West Coast), electricity is expensive.
    A technical question on trolleybusses: Can the two-wire power systems needed for trolleybusses (which cannot conduct return current through the tires) co-exist with overhead electrical systems used for rail (which generally only provide the supply voltage)?

  6. Jarrett February 2, 2010 at 10:41 pm #

    Scotty. Yes the two can be found in the inside lane of Market Street in downtown San Francisco, where both streetcars and trolleybuses run. As I recall, they’re next to each other over the lane, but I’m sure a local will correct me.

  7. anonymouse February 2, 2010 at 11:37 pm #

    On Market Street, the streetcars use trolley poles and so just use the left wire, which is at +600 Volts. The right wire is bonded to ground along with the rails. I have no idea how the switches actually work, but it must end up working out somehow. On systems where the trains use pantographs, you can’t share the actual wires (you have to run them parallel), but you can certainly have crossings, for example on Muni Metro lines, and on the streetcar in Seattle, although the crossings are more complicated, somewhat less reliable, and somewhat more difficult to maintain than with trolleypole-only systems. I believe in Europe there is at least one place where streetcars (with pantographs) share a tunnel with trolleybuses, presumably with three wires in total, the streetcar wire being hung a bit lower and slightly offset to one side, with the pair of trolleybus wires a bit higher and offset to the other side. I’d assume that Seattle didn’t go with this option because their light rail uses a higher voltage (1500 VDC), which would raise safety concerns in case of a trolleybus dewirement.

  8. Winston February 2, 2010 at 11:49 pm #

    One note on the database. The Urbanized area definitions can lead to very odd results. For example, in the case of the SF Bay area is composed of 5 different main urbanized areas: (1)San Francisco-Oakland, which includes San Francisco, all the cities in San Mateo County except Half Moon Bay, and the swath of development from Richmond to Fremont (2)Concord, which includes Central Contra Costa County plus Dublin and Pleasanton but not Livermore, (3)Antioch, which includes everything Northeast of Willow Pass, (4) Livermore, (5) San Jose.
    On the other hand, all of the vast urban sprawl surrounding NYC is counted as one urbanized area. In many ways this makes sense since the urban pockets that make up the bay area are separated from each other by large tracts of undeveloped land while the NYC area is basically hundreds of miles of uninterrupted sprawl. However it can lead to counter intuitive results.

  9. dejv February 3, 2010 at 12:40 am #

    On systems where the trains use pantographs, you can’t share the actual wires (you have to run them parallel).

    You can. It’s not a good solution though because the wire for pole should be as straight as possible while the wire for pantograph must zig-zag and insulated electric system for trolleybuses is safer. So this is preferred solution to this problem today.

  10. Ted King February 3, 2010 at 1:42 am #

    Re : SFMuni’s ETB + PCCs + LRV
    I suspect the electrical engineers had a field day with the co-existence of three modes of electrical transit. There are at least two noteworthy stretches in the network :
    Church St. from Duboce to 17th St.; and
    Ocean Ave. from San Jose to just beyond Phelan (CCSF).
    I have seen on a couple of occasions a wire-service truck near Church and Market. Those brutes with a big cab and a lifting bed seem almost as big as an Abrams MBT (M1A1).
    Church St.
    Lines – 22-Fillmore (ETB), F-Market (PCCs), J/N (LRVs)
    Cross and LRV merge at Duboce;
    Cross at Market;
    Cross and ETB merge at 16th;
    LRV/PCC merge at 17th (allows the historic streetcars to run out Church / 30th / San Jose to the car barn).
    Ocean Ave.
    Lines – 49-Van Ness/Mission (ETB), F-Market, J/K/etc.
    Cross and LRV merge at San Jose Ave.;
    Cross and LRV merge at I-280 (yard entrance);
    Cross and ETB merge at Phelan terminus.
    FYI – PCCs stay on San Jose Ave. and turn in to the old yard just southeast of Geneva + San Jose Aves.
    P.S. There is a third epic tangle near the Ferry Building (Market + Embarcadero) but I haven’t taken the time to see how nasty it is. That area has a terminus for several ETB lines, the J-turn for the F-Market on to the Embarcadero, link tracks for the E-Embarcadero, and the LRVs surfacing just to the south.

  11. jfruh February 3, 2010 at 6:55 am #

    Out of curiosity, how are passengers with prepaid weekly or monthly passes, or even timed tickets, figured into farebox recovery data? Many systems now have electronic fareboxes where you need to tap or slide your card even if it’s an unlimited pass, but on others you just flash it to the operator. Is there some attempt to figure out what routes people with transit passes generally ride? Are transit pass funds just uniformly spread all over the system?

  12. Danny February 3, 2010 at 7:16 am #

    Wow, it looks like Ted King understands MUNI better than MUNI does!!
    Typically accounting on that level isn’t necessary. You count your total fare revenue, your total boardings, and the operating costs, and you have everything you need for a system-wide FRR. Only if you wanted to understand route-by-route FRRs, you would need to count individual boardings.

  13. Art Busman February 3, 2010 at 8:40 am #

    My frustration with the NTD is that they enjoy wasting tax money forcing agencies to perform ridiculous boarding surveys to get passenger trip mile data which hardly anybody uses compared with revenue hours, trips, cost per passenger, farebox recovery, etc. This is a perfect example of how government just loves to waste money for sake of wasting money. It’s our constituents and their businesses who send tax money to the federal government to fund programs like NTD and what do they do? They force US to spend our own money to give them near irrelevant data THEY don’t even use. It’s like the Justice Dept forcing every cop out there to start sampling mileage trips for transporting criminals in his car. Hmmm, 2.5 miles on average to transport criminals from the crime scene to jail. Soooo useful data!!!!!

  14. Tom Rubin February 3, 2010 at 9:54 am #

    Farebox recovery ratio is perhaps the most misunderstood and misused statistic in transit.
    First, it ignores capital costs entirely, both the original costs to get a system up and running AND the very expensive capital renewal and replacement costs.
    Second, it is a percentage, not a dollar value, which can be very misleading. For example, the LA MTA bus system has a relatively low average fare, which leads to FFR a lot lower than other big city systems. However, because its costs are also relatively low, and its utilization so high, the taxpayer subsidy per passenger, and per passenger mile, are generally among the best in the nation. Yes, the low fares, and hense the relative low FFR, make this a good deal for the riders, but the taxpayers are getting a GREAT deal.
    Electric (trolley) buses have been dying out for decades. They are most useful in a city with extreme hills, like SF – an electric motor generates maximum torque at zero rpm, which is just what you need in getting up Sacramento with a full load. However the poles restrict speed; it is very difficult to exceed 35 mph without running a major risk of losing the poles. Also, there are no real cost advantages to trolley bus – when Muni says that trolley buses are cheap to operate because of Hetch Hetchy power, they are ignoring what that power would sell for. Also, before you string those uninsulated power lines at 600-750v, you might want to have a long talk with the local fire department that will have to fight fires on those streets. This is not a total dealbreaker, but, when you lose a bus with an engine, you lose one bus; when you have problems with a trolley bus line, you can lose the entire route — and leave the buses stranded.

  15. Ted King February 3, 2010 at 12:42 pm #

    Actually, an ETB network can be built with power zones and turnarounds so that a line break will have minimal impact in most cases. The nightmare scenario for SFMuni would be to have a line break at like Van Ness + Market, Tenth + Market, or Eighth + Market. The other bad spots are in those stretches in my earlier post (3 Feb.’10 20:42).
    As a child I rode the old Mack diesels up Sacto. from Chinatown during rush hour. A granny walking could beat those sardine cans. And that hill, Nob Hill, is one of the steepest in San Francisco. On some of the south side blocks there are steps instead of a regular sidewalk.
    P.S. Danny, I’ve used SFMuni for roughly fifty years. My knowledge is not encyclopedic but certain sections (Downtown, Richmond, Sunset, Mission, DC-Line) are well known to me. SF is a great town for a rail-fan.

  16. Jarrett February 3, 2010 at 1:26 pm #

    @jfruh. Most agencies do track their pass usage. Sometimes when you flash a pass at a driver you’ll see him push a button, which is counting your pass transaction. The data is helpful in getting pass pricing right, and also in evaluating the uptake of passes to determine if they’re being marketed well.
    @Art Busman. I couldn’t disagree with you more about the value of the data collection. Most of the data in the NTD is stuff good agencies want to know about themselves. For example, to do a network redesign I really need to know ridership by stop, and by time of day, so that I can see the usage on any segment of a route that I might be considering changing. Automation of data collection is slowly happening, reducing the marginal cost of each survey and offering management much, much more information about exactly how and where and when their service is being used. Agencies do care about their riders, but until recently did not have the tools to really measure the rider experience or determine how to best allocate resources.

  17. anonymouse February 4, 2010 at 9:10 am #

    Tom Rubin: there’s definitely a reliability tradeoff with diesel vs. electric traction, but it generally seems to be the case that while wire damage is more catastrophic to an electric system than simple vehicle failure in a diesel system, the electric system as a whole ends up being much more reliable than the diesel vehicles. There’s also the tradeoff of maintenance: wires do cost money to maintain, but often the maintenance of electric vehicles plus wires is less than the maintenance of diesel vehicles. This kind of makes sense: a diesel engine is a horribly complicated thing with thousands of little moving parts being moved back and forth by explosions two thousand times per second.
    On another note, modern ETBs have backup batteries, and so can run off wire for a few blocks. On Muni, this is standard procedure when there’s an obstruction that makes use of the wire impossible. And, depending on how well the wire is done, 35 mph is certainly not the upper limit of speed. The unofficial speed record for trolleybuses is somewhere around 65, and I suspect that with some care, 45 or even 50 mph would easily be possible in regular service.

  18. Ted King February 4, 2010 at 11:29 am #

    Another downside to a diesel bus is the transmission (aka gearbox). In a hilly city like S.F. that sub-system gets beaten to death. And SFMuni has had gear set problems in the past due to a flat set (e.g. suitable for Davis, CA) being installed instead of the usual hill set (the AMC buses). It took years for the laughter over that gaffe to die down.
    I’m glad that SFMuni is multi-modal. It makes them less vulnerable to oil price spikes and the city a little less noisy and sooty.
    P.S. I lived next to SFO for a few years and noticed a layer of soot from the jet engines all over the place. The noise from late night flights (e.g. a 747 Foxtrot to Japan at 0100) was bearable.