Doggie-bags from Overtaxed Restaurant Analogy

26-36In the last post, commenting briefly on how some San Franciscans were lamenting the deletion of the uncrowded and redundant bus line 26-Valencia, I compared the 26 to a restaurant ….

Yes, we all like to ride uncrowded buses.  I really enjoy dining in uncrowded restaurants too.  And yet the odd thing is, however much I patronize my favorite uncrowded restaurants, they seem to go out of
business sooner than the crowded ones do.  I wonder why that is.

As always, the danger of fast posting is that you end up with comments that are much smarter than the post.  Commenter Spyone, for example, explained some of the economics of the restaurant business.

In transit, the vehicle itself is the service provided to the customers, and (to a limit) additional customers do not increase cost. In a restaurant, a tiny fraction of the customer’s bill pays for his chair.  Due to this fact, a restaurant actually operates best if it is rarely full. The extra chairs represent emergency capacity (for special events and the like), while it is the staffing that is adjusted to meet
projected demand.

Which oddly leads me to realize on of the major problems with creating a frequent network: in most transit systems, staffing is per-vehicle rather than per-passenger, so it is more cost effective to serve more passengers by using bigger vehicles rather than more vehicles. If frequency is your goal, the opposite is what you want: more vehicles (possibly smaller vehicles if that represents any kind of savings) running more often.

In both restaurants and buses, what matters is not the crowdedness of the space but the ratio of customers to employees, because at least in the high-wage developed world, the employees are the dominant element of cost.  The attentive service good restaurants provide is the result of getting this ratio right.  In most transit in the high-wage developed world, the right ratio of passengers to drivers is “as high as possible up to the limits of civilized comfort,” which is admittedly an elastic and culturally determined limit.

The dominance of labor in transit’s cost explains why there’s so little money to be saved by running smaller buses in areas of smaller demand.  I’ve met many small-bus proponents who talk as though we could just chop all our big buses in half and have twice as many small ones, but alas, it doesn’t work that way.  Small vehicles may be marginally more fuel-efficient and offer a more intimate look-and-feel, but these are small factors compared to the constant cost of labor.

Cap’n Transit added:

I’ve seen plenty of uncrowded restaurants that have stuck around for years.  To amplify what Spyone wrote, one difference between restaurants and buses is that a popular restaurant is able to raise its prices to both reduce crowding and compensate for the revenue lost. If your demand curve runs the right way, you can often make more money serving less customers.


In fact, the express buses here in New York cost $5.50, more than double the regular $2.25 fare, and they rarely have anyone standing. It’s my understanding that they’re still subsidized, but less than they would be if they were $2.25. On one level it would be best for the agency if every route – or every itinerary, or even every trip – had a different fare based on the supply and demand. Of course, it would also be confusing and frustrating to the passengers, and probably wind up driving more business away.


Actually, I explored that idea in a thought experiment, here.  What if fares were inversely related to crowding?  We could endure crowded conditions for a bargain, or pay more for more space, just as we do on the airlines.  It’s actually possible to imagine, with the next generation of smartcards, a fare system where fares varied in real time according to how crowded the vehicle was.  Essentially, you’d be charging for transit the same way you and a few friends might pay for a ride you take together — by splitting the actual cost among you.  I’m not recommending this, but I do recommend thinking about it as a way to broaden your notion of what fares can do.  The Cap’n goes on:

Whenever I hear about a route being eliminated, though, I always worry that it’s a self-fulfilling prophecy, or at least a vicious cycle. Who wants to bother waiting around for twenty minutes if you can walk a block east and get a bus in five? It makes me wonder: what if they ran that route every five minutes for a month? If people knew that the route was going to be frequent and dependable, would they start really using it?


Often, a route that does something unique in the network may fail because it was never run frequently enough, but the 26-Valencia, which was only a block from parallel service on Mission St., was not one of those.   Yes, you might have saved the service by investing massively in it, but that wouldn’t have changed the fact that the Valencia and Mission markets are mostly overlapping, so you’d have ended up dividing the market between these two streets, yielding a much less efficient network.  The most efficient network gives you one line that goes the direction you want to go, and then runs the greatest possible frequency on it, including a mixture of local-stop and limited-stop service if the corridor is long, as Mission is.

Pantheon, meanwhile, was curious about why a bus route engages such feelings of attachment, even among people who admit they didn’t ride it much:



Following the links in the Streetsblog article, one comes across a page of tributes to the 26 which read like obituaries for a dearly departed loved one. Here’s a taste:

“I’ve lived a half-block from Valencia for well more than six years now, yet I can count the number of times I’ve ridden the 26-Valencia on 1.5 hands. It’s almost always a foggy ride, not due to the weather, but more to how much liquor I’ve imbibed. Or sometimes, it was simply the amount of warm pizza in me, and with the wind-chill factor factored in, and the randomness of a 26-Valencia magically showing up to cart my friends and loved ones on down the avenue to the safety and warmth of our homes …”.

I think this illustrates the deep emotional attachment people develop towards bus routes. Once established, they become part of the urban fabric. They are almost like a kind of mobile architecture. And if taken away, you miss them the way you would miss a grand old building that was destroyed to make way for a parking garage.

One of the joys of life in a big city is that anything, even a bus line, can be a gateway to associations and memories,  like Proust’s madeleine.  Pantheon goes on:

Here’s hoping the residents of Valencia can recover from their loss, perhaps heal their sorrows with some liquor and warm pizza, and enjoy the short walk home.

I’m sure they’ll be fine, because the walk from Mission to Valencia is very short indeed.

11 Responses to Doggie-bags from Overtaxed Restaurant Analogy

  1. Cap'n Transit December 10, 2009 at 5:42 am #

    It’s not just the dominance of labor costs; it’s also the price cap that people put on a system that is seen as a form of welfare. As I wrote last summer, many people who can afford to drive do so because they can be sure the vehicle won’t be crowded. Transit for all is still important, but offering an uncrowded alternative is one way to attract the famous “choice riders.”
    Ride the #7 train at rush hour here in New York and you will see a number eastbound passengers who take the train two stops west, from Grand Central to the end of the line at Times Square, so they can get a seat. If people will spend that extra five minutes, imagine how many would spend an extra two dollars.

  2. anonymouse December 10, 2009 at 8:16 am #

    “you’d have ended up dividing the market between these two streets, yielding a much less efficient network.”
    I think to some extent people realize that this will divide the market, but they think that this is a GOOD thing. In their view, Mission is already too full of buses, which means that the service becomes crowded and unreliable. They’d prefer the less crowded, more reliable service on the 26 if it were actually competitive on time. They might have a point: the 14 alone runs an articulated bus every 5 minutes, and there’s also the 14L and 49, which also run artics. Maybe if they had proper transit priority and real POP reliability would be less of a problem on Mission.

  3. EngineerScotty December 10, 2009 at 10:19 am #

    Don’t transit vehicles in one European city (Paris, IIRC) offer a “first class” cabin and a “coach” cabin, with no difference in amenities whatsoever between the two? The only difference is that fewer people are willing to pay the higher price to ride in the “first class” cabin, so it is generally less crowded.
    At any rate, this is one good reason to prefer rail over busses for a high-density corridor (especially one where the entire corridor is strong). The passenger/driver ratio can get far higher with a train than it can ever get with a bus.

  4. marco December 10, 2009 at 10:41 am #

    The attachment of San Franciscans to bus route 26 reminded me a similar story about Paris line 38:

  5. rhywun December 10, 2009 at 10:54 am #

    Depending on the length of the trip, a higher fare is absolutely worth it to some people if the choice is between standing and sitting. Which is why I gave up on the Express Bus from my neighborhood in Brooklyn to Manhattan: it was still standing room only!

  6. Cap'n Transit December 10, 2009 at 12:54 pm #

    Paris eliminated first-class metro cars in the 1980s, and first-class commuter rail cars in the 1990s, I believe. I don’t know the reasons, but it would suggest that the logistics of such a system are hard to manage.
    Also, the 38 bus in Paris parallels the route of the popular #4 metro line.

  7. Alon Levy December 10, 2009 at 6:39 pm #

    Cap’n: could it be capacity issues? The RER A is the busiest rail line in the Western world, and is at capacity even with double-decked 200-meter cars. At some point, it no longer makes sense to reduce capacity in order to run first-class cars.

  8. Cap'n Transit December 10, 2009 at 7:17 pm #

    Yes, that could very well be, Alon.

  9. Pantheon December 10, 2009 at 10:53 pm #

    The other aspect of the restaurant analogy that may be relevant to bus service is that of turnover. I was rather appalled when I first came to America at the way restaurants operate here, but it is because their entire business model is based on turnover. By contrast, a more service-oriented business model would necessarily mean higher prices, even if the restaurant were just as crowded.
    Take two restaurants – let’s call them Texas Steakhouse and Cafe France. Each has 24 tables. Both of them are totally full all night. But the difference is TS operates according to the American model and turns their tables over twice that night. CF is more laid-back, so it is actually the same people sitting at those tables all night. TS has actually served three times as many customers as CF, even though both looked “full” all night. This allows TS to sell food at lower prices, and CF needs to make sure their customers drank a lot of high-margin alcohol to turn a profit.
    Taking this analogy over to bus lines, the most economically efficient busline would not only be crowded, but would have lots of turnover throughout the line. It would achieve this by serving a high-capacity corridor with multiple start points and destinations along the line, and customer traffic going in both directions. This is what creates the potential for a high frequency service. By contrast, Express buses often need to be more pricey even if they are crowded, because they are more like Cafe France – the same customers for the whole route, and perhaps just as importantly no customer demand for the bus turning around and going right back the other way.
    This leads me to think that land use and culture are important factors in the economic viability of transit systems. One thing I have noticed about American cities is that people don’t actually live downtown. Thus American downtowns board up at 6 p.m., even in cities as large as Seattle. On the other hand, cities like Vancouver, Toronto, or Montreal will often be buzzing late into the night (European cities even more so). The simple fact of having large numbers of people living downtown creates a much more efficient transit network, because the demand goes both ways all day and night. And more efficiency means improved service and greater frequency. By contrast, a typical American city has a lot of empty buses running during peak times.
    This is another example of how land use patterns in America are an obstacle for transit networks. The problem is that a) people don’t live downtown and b) the cities are not very dense so there is little to no turnover along the line. A bus going from downtown out to bedroom communities X, Y, and Z will serve only one group of passengers for the whole line, and essentially no passengers on the return trip. It’s not like there are people who will get on at X and go to Z. So even if you get on downtown and the bus is crowded, it’s a bit of an illusion. Much like walking by Cafe France.

  10. Jarrett at December 11, 2009 at 10:17 pm #

    The following comment is by Peter Parker at Melbourne on Transit.

    Pantheon, I really like your analogy in favour of turnover or ‘hotseating’.
    Not only should bus routes have strong destinations along the way, but they need a strong destinations at both ends, to permit good utilisation for ‘reverse commuting’.
    Hence it’s a good plan to extend city – suburban buses at the next university or shopping centre or railway station further out instead of starting them from residential streets.
    If the bus is a suburban route that runs parallel to widely spaced stations on a railway line, the number of good connections to the train system may double if the customer can tolerate some backtracking. There’s a couple of cases on Perth’s Clarkson line of buses (463 & 464 IIRC) that offer a connection with every evening train at alternating stations even if the bus runs hourly to the train’s 30 minute headway. Very smart!
    But if there’s no logical termination point (maybe due to natural or street layout blockages eg freeways) and you’re stuck to a unidirectional commuter pattern only, then it might be more efficient to turn half the buses back halfway along the route so they can do another peak direction trip. This benefits the inner suburbs (by giving higher frequency) but at the cost of the outer areas. Where frequency is high it might be possible to arrange limited stops to give the outer suburban areas some speed but the gains might not be that much and you’re reducing inner area frequency. With trains you might be able to run a local/express two tier service on the longer lines, though track capacity is reduced if different stopping patterns share the one track.
    I’ve always thought that the number of unique passengers per bus or driver operating hour is a better measure than passenger pick ups per route kilometre.
    Using the time dimension reveals the big efficiency gains from things like reducing dead running and bus priority. And it quantifies the slow decline in efficiency in modes that must battle increasing volumes of mixed traffic.
    Even changing a traffic light sequence to include bus priority at a major shopping centre entrance has huge gains. Eg if two minutes is saved from 300 bus movements per day, that’s 600 bus minutes that could be used to boost frequency.

  11. Nathanael December 22, 2009 at 7:22 pm #

    ‘In most transit in the high-wage developed world, the right ratio of passengers to drivers is “as high as possible up to the limits of civilized comfort,” which is admittedly an elastic and culturally determined limit.’
    Of course, when you’ve got enough people, the way to maximize that ratio is to use trains. People really don’t mind having one driver in a giant 10-car train.