Should Fares Be Higher During Peak Hours?

Metro in Washington DC is considering fare surcharges for using the busiest downtown stations during the peak commute period.  This is proposed as an alternative to a generalized 50-cent surcharge during the busiest part of the peak.  The transit advocates at Greater Greater Washington are on board.  The Transport Politic’s Yonah Freemark objects:

But more fundamentally, one should ask whether it makes sense for a transit system to charge extra for exactly the service it is supposed to provide best: journeys to and from the downtown core at peak hours. Should the District of Columbia push for years to increase the number of office jobs downtown if it decides to reverse the game later on and disincentivize the use of the region’s primary transit service to get there? Why penalize the people who are using the system in exactly the way that the system was designed to work?

There’s a pretty clear businessman’s answer to that rhetorical question, but it’s not something that most people want to hear:  Transit ridership is determined by choices people make from among their available options.  If transit is already much more attractive than the other available options (as is usually the case for commutes into a major central business district (CBD) where motorists would face huge congestion and parking costs) the cruel facts of supply and demand argue that the fare for transit service in those situations can probably rise without much loss of ridership.

Of course, since we’re talking about a government product, you won’t win the argument using such cold calculus.  Your argument has to be about fairness.  Here, the case for peak surcharges is that they help people notice, and take account of, the very high cost of providing additional transit capacity during the peak period.

Service added to a network to handle peak-only overloading is very expensive service to operate. For example, drivers usually must be paid four-hour minimums for shifts that are actually as little as two hours, and the peak determines the number of vehicles that must be owned and maintained.  Peak demand that flows in only one direction, as in the classic American single-centred city, also generates the huge inefficiency of moving all those vehicles, entirely or mostly empty, back in the reverse-peak direction.  Sometimes, peak ridership is so much higher than midday that the fare revenue makes up for all these inefficiencies, but not often.  In fact, my experience with American bus operators is that few of them have really counted the cost of their peak-only services.

To insist that peak service be priced equally to midday or weekend service is to argue that the riders of those off-peak services should subsidize the peak.

There’s also an argument for peak surcharging that connects with urbanist goals for transit.  People who have the option of travelling off-peak should be encouraged to do so, because  off-peak capacity is usually abundant, while peak capacity is naturally scarce and costly for the transit agency.  This strategy helps build off-peak patronage, which supports more all-day service, which leads directly to transit that is more relevant to the entire life of the city rather than just the commute.  All-day frequent service is the only type of service that can support transit-oriented development and thus change the shape of the city in more sustainable ways.  From an urbanist perspective, then, a gentle but persistent effort to shift demand away from the peak makes sense.

I have trouble even seeing a social justice argument against peak surcharges, as the average peak-period traveler usually has a higher income than an average midday traveler (because lower-paying jobs, mostly in service industries, are less likely to start and end on the usual peak). So insisting on flat fares all day could actually be seen as regressive.

In cases where (a) peak-only service adds hugely to system costs but (b) doesn’t generate the farebox return to cover that impact, I can see only one credible reason not to surcharge the peak (or discount the midday, which is the same thing.)  But it’s a good reason:  surcharges/discounts are an added layer of complexity in the fare system, and complexity in fares is a serious barrier to patronage, and especially to the informal, spontaneous use that lets people experience transit as essential community infrastructure. That’s the reason I usually use, but it only takes you so far in a network that’s groaning under peak pressure and desperately needs to encourage people to travel off-peak if they have the option.

My only comment about the specific Washington debate is that any surcharge should really be tied to the direction of demand where capacity is costly, namely into the city in the morning and out in the afternoon.  A generalised surcharge on all peak travel will hit some people who make circumferential or reverse-peak trips where capacity is abundant and the incremental cost of peak service is zero.  Thus, if the surcharge is tied to certain CBD stations, it should ideally apply only to station exits in the morning and entries in the afternoon.   Again, though, the cost in system complexity — “how long does it take to explain this?” — has to be counted, and that often counts for quite a bit.

This posted is expanded from my comment on the Transport Politic post on this subject.  See also the comment thread there.

UPDATE:  In response to questions in the comments, let me clarify that I am talking only about situations — likely to arise only in major cities — where the differential cost of peak service is substantially higher than the differential in ridership, such that midday riders are effectively cross-subsidising peak riders.  Any proposal for peak surcharges should rest on a thorough accounting of all the costs of peak-only service, including (a) minimum pay hours that exceed work hours, (b) deadhead time [time running empty to/from the garage or to return from a one-way peak run], and (c) all procurement, maintenance and storage costs for fleet used only during the peak. 

23 Responses to Should Fares Be Higher During Peak Hours?

  1. Jack Horner May 6, 2010 at 6:05 am #

    A peak surcharge on transit is conceptually similar to road congestion charges.
    Road congestion charges attempt to force motorists to meet the external cost of their motoring (external aka spillover cost: a cost which my actions impose on you without compensation. When I enter a congested road I not only suffer delay but also increase the delay suffered by everyone else, without compensating them).
    Road congestion charges improve economic efficiency by forcing more economically rational choices about where/when to drive in a way which reduces the considerable external costs of peak period motoring (whether you regard them as fair is another matter).
    In the transit case, the public subsidy of the trip is a cost external to the rider. The peak traveller causes an above average externality cost by using a service with below average cost recovery.
    Possible contrary argument: the externality cost of road congestion is greatest in peak hours, so the benefit of getting people onto transit at this time is greatest, so perhaps a lower financial cost recovery for transit at this time is tolerable for the sake of the benefit to remaining motorists of reduced congestion (ideally they would compensate the transit authority for this).
    But this is a second best argument: it would not be needed if there are appropriate road congestion charges.

  2. Alex B. May 6, 2010 at 6:51 am #

    Good points, Jarrett.
    I realize that you’re trying to speak in general terms here, but I’d want to point out that Washington, DC’s Metro already has a substantial difference between peak and non-peak fares. This isn’t a question of going from a flat fare to a rush hour surcharge, it’s a matter of making the difference between peak and non-peak even greater.
    Off-peak fares vary based on distance, you pay either $1.45, 1.95, or 2.45.
    Peak fares (M-F, opening-9:30 am, and from 3 pm-7 pm) vary based on distance between $1.75 and $4.60. They are not broken out into three discrete levels like the off-peak fares, your fare may increase by 5 or 10 cents for traveling one extra station down the line.

  3. Sean N. May 6, 2010 at 10:31 am #

    The argument to use peak fares to shift demand is an excellent one, IMHO. But here in Vancouver it’s effective because “peak” fares aren’t “peak” at all but are actually “weekday” fares, applying every weekday until 6:30PM. If Vancouver really wanted to spread demand out then they’d only apply peak fares during peak times, say 7-10AM and 3-6PM.

  4. anonymouse May 6, 2010 at 10:56 am #

    How does the peak fare interact with things like unlimited passes? If there’s only one kind of unlimited pass, and it includes peak fares, then what you accomplish is moving occasional riders to the non-peaks.

  5. Tom West May 6, 2010 at 12:00 pm #

    The other way of looking it is that you are offering adiscoutn to off-peak users. In your classic commuter-orientated system, stuffed peak horu buses provide enough revenue to off-set the lower income per revenue hour generally found on off-peak buses.
    I’ve also seen student bus passes avilable in two types – a limited hours version designed to be able to get them to and from school (generally not valid weekday evenings, weekends or holidays) and a unlimited, more expensvie version that is valid any time.
    In England, seniors get free bus travel – but only after the morning peak. On the British rail system, travelilng after 9am on a weekday allows you to use a cheaper ticket.

  6. Andrew May 6, 2010 at 12:28 pm #

    The trouble with peak fares is that it encourages people to drive, because peak fares increase the cost of public transit compared to the cost of driving. For peak fares to work, there also have to be peak road tolls, high parking fees or some other deterrent from driving to work.

  7. M1EK May 6, 2010 at 12:59 pm #

    I pretty much fall into this camp:
    “so your formula that off-peak passengers are subsidizing the peak is not correct.”
    It’s very difficult to come up with passenger numbers on a system that would support your contention that off-peak passengers are the subsidizERS rather than the subsidizEEs. Those empty buses or trains may cost slightly more in labor, but it can’t possibly be THAT much more to make up for the bus/train being full in rush hour (i.e. hundreds of fares versus a handful of fares would more than make up for the cost of weird shifts and empty vehicles the other direction – this also ignores the fact that in many cities, reverse-commute peak vehicles are not all that crowded but still carrying more people than the off-peak vehicles are).

  8. M1EK May 6, 2010 at 1:24 pm #

    Also a point I forgot to make but just made over at Yonah’s blog is that the individual incentives cannot be ignored; the off-peak passenger is more likely to be transit-dependent and paying a more discounted fare than is the peak passenger; you may, in fact, invoke the transit version of Laffer’s Curve if you’re not careful here (raise fares enough that your most lucrative customers go drive the car they doubtlessly have parked in their driveway).

  9. Alon Levy May 6, 2010 at 1:30 pm #

    Anonymouse: some cities, for example Berlin, have cheaper passes that are valid only after 9 or 10 am.
    M1EK: the principle about high costs reducing demand is called a demand curve, and has nothing to do with Laffer. For-profit companies compute demand curves all the time to figure out how to price their product. The difference with transit is that as a public service with positive externalities, it’s priced below the profit-maximizing level.

  10. Agustin May 6, 2010 at 5:18 pm #

    A quote from the blog post:

    “Transit ridership is determined by choices people make from among their available options. If transit is already much more attractive than the other available options (as is usually the case for commutes into a major CBD where motorists would face huge congestion and parking costs) the cruel facts of supply and demand argue that the fare for transit service in those situations can probably rise without much loss of ridership.”

    Does the argument stand up if there are not huge congestion and parking costs for driving? Or would people give up the bus in favour of the car at peak hours?

  11. Colin May 6, 2010 at 6:44 pm #

    The idea of charging extra for peak trips only on inward trips in the morning and outward trips in the afternoon happens to be great for people like myself who live in the centre and often work in the suburbs.
    To what extent would it further those incentives; making the city a more attractive place to live, and the train-station-based suburbs a more attractive place to locate your business. And if so, those “train suburbs” would attract more residents, and more activity/development in general.
    It would make the rail system a more attractive thing to base you life around, in every direction.

  12. Alon Levy May 6, 2010 at 9:03 pm #

    Off-topic: I want to commend you for putting a non-rhetorical question in your post title. More often than not, titles like “Should we do X?” or “is Y happening?” are just FUD or sotto voce.

  13. voony May 6, 2010 at 9:13 pm #

    Jack horner is right:
    “A peak surcharge on transit is conceptually similar to road congestion charges”
    bus fleet are sized according the peak surcharge not the average usage, so to accommodate the peak, you need to buy buses which gonna seat at the barn most of the time…
    Peak surcharge is “yield management” or drive the demand toward the offer you have (rather try to match your offer to the demand)…
    and yes people working after the other (janitorial services…) are actually double penalized:
    -they work in time slot where public transit offer is usually poor, but still pay the full price of it…
    I also believe peak pricing is not only a good management practice but accomplish also sort of social justice by providing lower fare for low income.
    the idea of a reduced pass valid only after 9am or so is very great…

  14. Jarrett at May 6, 2010 at 11:15 pm #

    I’ve added one paragraph of UPDATE reflecting key comments and questions in preceding comments.

  15. M1EK May 7, 2010 at 8:17 am #

    Alon, yes, it’s a demand curve, but we’re also talking about an aggregate amount of revenue here, in which it’s probably more useful to use an example that shows a similar change in aggregate revenues.

  16. BruceMcF May 7, 2010 at 8:40 am #

    First, of course they should be higher, but also, of course it should not be labelled as a peak hour surcharge.
    Rather, in the context of a general fare increase (all too common nowadays), it should be labelled as an off-peak discount.
    What Yonah has done is get caught up in a semantic confusion. Of course mass transit in particular does relatively better at catering to peak hour congestion in terms of both direct and indirect (square feet required per rail passenger versus square feet required per bus or car passenger) … but that is a relative not an absolute comparison.
    “Doing better” means that full cost increases for mass transit and full cost increases by a much higher amount for cars during peak hour.
    The off-peak discount in Sydney is that a round trip ticket bought in off-peak hours cost 120% of the one way ticket. For a low marginal cost of operation system with very high up-front capital costs for expansion, that is a quite reasonable peak/offpeak fare structure.
    Also note that this kind of fare structure for the mass transit backbone of an integrated transport system is also good for the complementary bus system, with transfer bus passengers being spread out more evenly between the center of the peak and the shoulders by those with more flexible schedules targeting the start of the off-peak discount period.

  17. david vartanoff May 7, 2010 at 9:44 am #

    I may be the lone voice here, but applying “yield management” or any of its cousins to public transit is just wrong on its face. First, the surcharges mostly cheat the lowest wage riders who cannot choose when to appear at work, and are least likely to default back to car use. Secondly, in this instance (and many others nation wide) the transit agency is being denied previously agreed funding from one or more revenue streams. WMATA is in the same bind as Caltrain and many others; ridership is high but the funding, often based on volatile taxes, has shrunk leaving them unable to serve the market they have created. Pushing riders back to cars is not a path to reducing GHGs or energy conservation. In WMATA’s case, the saddest part is that after three decades, they haved become so successful that poor original design and inadequate fleet are crippling their throughput.

  18. Alon Levy May 7, 2010 at 11:30 am #

    David, low-income workers often work in shifts other than 9-5.

  19. Sean T Hedgpeth May 7, 2010 at 11:47 am #

    This is a debate that needs to be heard. I think that a surcharge on peak service is the wrong way to phrase to a public hit with massive fare hikes and service cuts. Instead, midday service fares should be ‘frozen’ when other fares go up. The one way bias could also be fixed by providing a few stops of service in reverse commute, possibly qualifying for JARC funds.
    That being said, I do not think that express peak service should cost more, its cheaper to run than local. I think one of the reasons peak service is so expensive is that the buses get stuck in traffic, while the express bus can pass by. More bus only lanes and am/pm towaway zones need to be implemented.

  20. Jarrett at May 7, 2010 at 12:22 pm #

    The following comment was received from Ari,
    My dad tells me about “Dime Time” which Boston ran in the 1970s. The T dropped the fare—which was 25¢—to just a dime between 10 a.m. and 1 p.m. He said his office at Volpe in Cambridge would frequently decamp to the North End for pizza, but it never sounded like they worked too hard there. This was, of course, before most commuters started using unlimited use monthly passes for which the marginal cost of a midday trip is zero. I think BART and WMATA are the only major systems which charge solely per-trip.
    Oh, boy, there’s actually some stuff on the Interwebs about Dime Time. I smell a post coming. (Long and short of it: it didn’t cause many lapsed riders to come back to the system but did get a lot of existing riders to shift their trips.)

  21. Alon Levy May 7, 2010 at 8:08 pm #

    Singapore charges solely per-trip. Shanghai does, too, but if you have a multi-ride ticket, you get a 10% discount after having taken a certain number of trips in one calendar month. The MTR offers a day pass but prices it for tourists only. And none of those three systems has separate peak and off-peak fares.
    But Shanghai might find it useful to congestion-price Line 1. Or just build the relief lines to connect to the rest of the system better.

  22. B.T.Carolus May 7, 2010 at 8:30 pm #

    The other way to frame a differential price-scheme is to say that off-peak fares are lower than peak fares. LA’s Metro essentially does this for their senior/disabled pricing. The base fare for seniors is listed as $.55 (compared to $1.25 for regular riders) but this is technically the peak fare, only charged between 5-9 am and 3-7 pm. The rest of the day (16 hours) senior passengers are only charged $.25. In practice this means that senior passengers often delay their trips until after 9 am.

  23. Chris Stefan May 7, 2010 at 8:33 pm #

    King County Metro started charging higher peak (5AM to 9AM, 3PM to 7PM) in the 90’s rather than impose a general fare increase. They’ve kept the peak/off-peak distinction since then with the additional modification of removing the zone charges off-peak. Since the surcharges haven’t changed since they were imposed the relative difference between peak and off peak fares is much less than it used to be when the peak charge was first imposed.
    Note that the peak fares apply to all King County Metro routes during “peak hour”. King County Metro also doesn’t impose surcharges for commuter express routes.
    This doesn’t seem to have hurt ridership at all as the transit mode share, particularly for CBD commutes is relatively high.