When using public transport, we often need to transfer to a different route during of the journey. These transfers have costs – I can think of the following, but there may be more.

  1. If we're reading, working, etc. on board, we need to stop, losing flow.
  2. We spend time moving from one route to the next.
  3. We spend time waiting for the next departure.
  4. There are lots of passengers moving between routes. They take up space and cause queues when on- and off-boarding. This also increases the risk of accidents.

As far as I can understand, we could decrease the necessity of transfers by having multiple routes to the terminals (final stations).

I've reviewed the subway maps of London, New York and Oslo, finding that none of them do this to any significant degree – the terminals usually have just one route leading to them.

As an example, let's look at Oslo, the simplest of the three:

enter image description here

Imagine that you're travelling from Østerås (orange, left) to Ellingsrudåsen (orange, right). No transfers required – route 2 takes you all the way.

But if you're travelling from Østerås to Mortensrud (purple, bottom right), you would have to transfer from route 2 to route 3 at some point.

Let's imagine that we take some of the capacity from routes 2 and 3 to serve routes 6 and 7. Route 6 (gray) serves Østerås-Mortensrud, and route 7 (brown) serves Kolsås-Ellingsrudåsen:

enter image description here

With this design, some transfers could be avoided. If we added enough routes, many more transfers could be avoided. Granted, we would have to wait longer for a route that went exactly to the station we want to go, but we would always have the option of taking the first train and then changing like before.

(Note that I'm not suggesting adding new infrastructure, just using the existing one differently.)

So, I've been asking myself, why don't public transport authorities do this? These are the reasons I can think of:

  1. Simplicity for passengers: When only one route serves a particular station, you don't have to think about which route you'll board. Also, the maps are simpler with fewer routes.
  2. Simplicity for transport providers: More routes would mean more difficult (and therefore costly) planning and management.

From my (admittedly ignorant) perspective, the above reasons don't seem sufficient to outweigh the advantages of fewer transfers.

Have I overlooked some problems with adding more routes? Or am I underestimating the costs and/or overestimating the benefits of more routes?

Note that I have previously asked this question on Engineering Stack Exchange. I'm asking it again here because I didn't get satisfactory answers, and because I think it's more of an Economics question.

  • $\begingroup$ Without trying to digest the details above, I can tell you that in the US "public transit" is controlled by politicians whose interest conflict with those of the general public (especially the lower economic levels, where public transit is most important). $\endgroup$
    – Hot Licks
    Commented Mar 9, 2018 at 18:47
  • $\begingroup$ Shouldn't they also add some routes from Kolsas to Vestli? And from Frognerseteren to Mortensrud? And from Bergkrystallen around the loop to Mortensrud? And so on? Where do you stop? This map has 8 terminii(?) and so you would end up with 64 routes! Apparently they have decided that these combinations are the most useful (or perhaps the most practical) 5 routes. $\endgroup$ Commented Mar 14, 2018 at 5:40
  • $\begingroup$ @immibis I think the maximum number of routes would be 1 + 2 + 3 + 4 + 5 + 6 + 7 = 28. Also, I'm not sure if it would make sense (or be physically possible due to the directions of the lines) to go all the way to 28. For example, I don't think a route from Kolsås to Østerås would save travellers any time compared to taking a bus or something. $\endgroup$ Commented Mar 14, 2018 at 13:25
  • $\begingroup$ Right, I forgot to account for routes from a station to itself and for routes being bidirectional. I only see 3 routes here that are obviously close together (Kolsas-Osteras, Bergkrystallen-Mortensrud, and Frognerseteren-Sognsvann, pardon me for not having a Norwegian keyboard), so that still leaves 25... $\endgroup$ Commented Mar 15, 2018 at 6:13
  • $\begingroup$ @immibis Yeah, we could have quite a lot of routes, and that does lead to some complications. But it's still not obvious to me that the low number of routes we have today is superior to a higher number of routes, all things considered. $\endgroup$ Commented Mar 15, 2018 at 10:14

2 Answers 2


The people who do the planning do explicitly account for the value of time and convenience to transit users. At least, in all the systems I've worked on. (source: I have had jobs where I worked on exactly this, for London, for the Netherlands, and for international train travel)

Interchanges are modelled with penalties which cover:

  • the time taken to move from one platform to the other
  • the time to wait for the connecting train
  • and a penalty to reflect the inconvenience, which is calibrated through Stated Preference surveys.

Here's an example from the Business Case Development Manual used for evaluating costs and benefits for transport schemes in Greater London, page E-10, showing a penalty just for the interchange itself of 3.5 minutes of travel time. That's in addition to valuing the waiting and walking time at the interchange:

enter image description here

(disclosure: I developed some of the methods, and calibrated some of the numbers, in that manual)

So, why aren't the lines designed for switching between tracks?

Well, sometimes they are: London's Circle, District and Hammersmith & City lines work very much on this basis: similar rolling stock on the same tracks, connecting different termini. London's Northern line works similarly, with two northern termini, two southern termini, and two branches in the middle, with trains switching between them.

At peak times, there is much less switching on the branches on the Northern Line. And the Circle line was recently redesigned so that it is no longer a circle.

The reasons for those two things, are the reasons why it isn't done more often.

The switching reduces the vehicle-capacity of the network. So it can move fewer passengers per hour. Which adds delays to everyone, at rush hour.

Trains are kept a minimum distance apart, in the interests of safety. Where two branches join, as they do at switch-points, that means that trains on both branches have to be kept that distance apart from trains after the switch-point. And the switch-point becomes a bottleneck, reducing the capacity of the two branches: the switch itself has a lower capacity than the sum of the capacity of the two branches.

So, during the busiest periods, there is less switching, and that minimises the overall system aggregated generalised cost. And that generalised cost includes the bullet points that I mentioned at the top of this answer. And when new lines are designed, the value of switching routes to different termini usually adds too much of a penalty to network capacity, to make it worth while: overall system utility is typically maximised with dedicated routes.

And if you don't want to do switching of vehicles on tracks, then you have to build additional grade-separated tracks to do the switching instead, and that's really really expensive, and isn't justified by the reduced generalised cost that arises from fewer passenger interchanges.

  • $\begingroup$ Thanks for answering. I don't quite understand how "[t]he switching reduces the vehicle-capacity of the network. So it can move fewer passengers per hour. Which adds delays to everyone, at rush hour." May I ask you to elaborate? $\endgroup$ Commented Mar 11, 2018 at 10:05
  • $\begingroup$ I've added a paragraph after the bit you've quoted. Does that help? $\endgroup$
    – 410 gone
    Commented Mar 11, 2018 at 11:00
  • $\begingroup$ Maybe. Does it depend on how many lines there are in the "common" section of the network? (In my original example, that's Majorstuen-Hellerud – see en.wikipedia.org/wiki/Common_Tunnel.) If all trains that go in one direction run on one single track, I don't understand how more routes would cause the switch-point to become a bottleneck. But if there are several parallell lines, I think I see how switching would reduce the capacity. $\endgroup$ Commented Mar 11, 2018 at 17:23
  • $\begingroup$ If there are parallel lines, then there's duplicate infrastructure, which is expensive. And it requires bigger stations, which is expensive. And if you want trains to weave, to serve different terminus-terminus pairs, then you either have switch points or build fly-overs ~(grade-separated crossings) $\endgroup$
    – 410 gone
    Commented Mar 11, 2018 at 17:50
  • $\begingroup$ @EnergyNumbers But there are parallel lines already, without the switching. Adding the switches wouldn't mean you'd need to build new parallel lines to maintain the capacity, because they are already there. $\endgroup$ Commented Mar 14, 2018 at 5:41

The people who do the planning explicitly account for 1) their cost of planning, 2) the cost of wages for transit operators, 3) the cost of infrastructure and 4) the cost of equipment (buses, etc.) ... etc. However, they might not account for the value of time or convenience to transit users.

The value of time of transit users may be represented in some manner through the political process, but in the economic modelling it is easier to ignore this than to try to motivate a specific number (or series of numbers weighted across types of transit users) for use in modelling costs and benefits.

One way to correct for this is to take surveys to try to find out how people value differences in commute times, although this may face legitimate disagreement about whether maximizing (for example) social welfare or production should be the primary objective in making use of such an estimation.

  • 1
    $\begingroup$ When you don't know, don't guess. It's ok just to leave a question unanswered for a bit longer. Adding guesses just adds clutter. $\endgroup$
    – 410 gone
    Commented Mar 10, 2018 at 21:53
  • $\begingroup$ Some thing are accounted for because they are easy to count. Some things are not accounted for. That's not a guess. $\endgroup$
    – nathanwww
    Commented Mar 10, 2018 at 22:39
  • $\begingroup$ It is not surprising that London UK and the Netherlands do so. This is not the status quo at present. $\endgroup$
    – nathanwww
    Commented Mar 11, 2018 at 4:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.