Trains are not Planes

Trains and planes are both scheduled modes of intercity travel running large vehicles. Virgin runs both kinds of services, and this leads some systems to treat trains as if they are planes. France and Spain are at the forefront of trying to imitate low-cost airlines, with separately branded trains for different classes of passengers and yield management systems for pricing; France is even sending the low-cost OuiGo brand to peripheral train stations rather than the traditional Parisian terminals. This has not worked well, and unfortunately the growing belief throughout Europe is that airline-style competition on tracks is an example of private-sector innovation to be nourished. I’d like to explain why this has failed, in the context of trains not being planes.

How do trains and planes differ?

All of the following features of trains and planes are relevant to service planning:

Stations are located in city center and are extremely inconvenient to moveAirports can be located in a wider variety of areas in the metro area, never in the center
Timetables can be accurate to the minuteTimetables are plus or minus an hour
Linear infrastructureAirport infrastructure
High upfront costs, low variable costsHigh upfront costs but also brutal variable costs in fuel
Door-to-door trip times in the 1.5-5 hour rangeDoor-to-door trip times starting around 3 hours counting security and other queues
In a pinch, passengers can standStanding is never safe
Interface with thousands of local train stationsAll interface with local transport is across a strict landside/airside divide
Travel along a line, so there’s seat turnover at intermediate stopsPoint-to-point travel – multi-city hops on one plane are rare because of takeoff and landing costs

Taken together, these features lead to differences in planning and pricing. Plane and train seats are perishable – once the vehicle leaves, an unsold seat is dead revenue and cannot be packaged for later. But trains have low enough variable costs that they do not need 100% seat occupancy to turn a profit – the increase in cost from running bigger trains is small enough that it is justified on other grounds. Conversely, trains can be precisely scheduled so as to provide timed connections, whereas planes cannot. This means the loci of innovation are different for these two technologies, and not always compatible.

What are the main innovations of LCCs?

European low-cost carriers reduce cost per seat-km to around 0.05€ (source: the Spinetta report). They do so using a variety of strategies:

  • Using peripheral, low-amenity airports located farther from the city, for lower landing fees (and often local subsidies).
  • Eliminating such on-board services as free meals.
  • Using crew for multiple purposes, as both boarding agents and air crew.
  • Flying for longer hours, including early in the morning and later at night, to increase equipment utilization, charging lower fares at undesirable times.
  • Running a single class of airplane (either all 737 or all 320) to simplify maintenance.

They additionally extract revenue from passengers through hidden fees only revealed at the last moment of purchase, aggressive marketing of on-board sales for ancillary revenue, and an opaque yield management system. But these are not cost cutting, just deceptive marketing – and the yield management system is in turn a legacy carrier response to the threat of competition from LCCs, which offer simpler one-way fares.

How are LCC innovations relevant to trains?

On many of the LCC vs. legacy carrier distinctions, daytime intercity trains have always been like LCCs. Trains sell meals at on-board cafes rather than providing complimentary food and drinks; high-speed rail carriers aim at fleet uniformity as much as practical, using scale to reduce unit maintenance costs; trains have high utilization rates using their low variable operating costs.

On others, it’s not even possible to implement the LCC feature on a railroad. SNCF is trying to make peripheral stations work on some OuiGo services, sending trains from Lyon and Marseille to Marne-la-Vallée and reserving Gare de Lyon for the premium-branded InOui trains. It doesn’t work: the introduction of OuiGo led to a fall in revenue but no increase in ridership, which on the eve of corona was barely higher than on the eve of the financial crisis despite the opening of three new lines. The extra access and egress times at Marne-la-Vallée and the inconvenience imposed by the extra transfer with long lines at the ticketing machines for passengers arriving in Paris are high enough compared with the base trip time so as to frustrate ridership. This is not the same as with air travel, whose origins are often fairly diffuse because people closer to city center can more easily take trains.

What innovations does intercity rail use?

Good intercity train operating paradigms, which exist in East Asia and Northern Europe but not France or Southern Europe, are based on treating trains as trains and not as planes (East Asia treats them more like subways, Northern Europe more like regional trains). This leads to the following innovations:

  • Integration of timetable and infrastructure planning, taking advantage of the fact that the infrastructure is built by the state and the operations are either by the state or by a company that is so tightly linked it might as well be the state (such as the Shinkansen operators). Northern European planning is based on repeating hourly or two-hourly clockface timetables.
  • Timed connections and overtakes, taking advantage of precise timetabling.
  • Very fast turnaround times, measured in minutes; Germany turns trains at terminal stations in 3-4 minutes when they go onward, such as from north of Frankfurt or Leipzig to south of them with a reversal of the train direction, and Japan turns trains at the end of the line in 12 minutes when it needs to.
  • Short dwell times at intermediate stops – Shinkansen trains have 1-minute dwell times when they’re not sitting still at a local station waiting to be overtaken by an express train.
  • A knot system in which trips are sped up so as to fit into neat slots with multiway timed connections at major stations – in Switzerland, trains arrive at Zurich, Basel, and Bern just before the hour every half hour and depart just after.
  • Fare systems that reinforce spontaneous trips, with relatively simple fares such that passengers don’t need to plan trips weeks in advance. East Asia does no yield management whatsoever; Germany does it but only mildly.

All of these innovations require public planning and integration of timetable, equipment, and infrastructure. These are also the exact opposite of the creeping privatization of railways in Europe, born of a failed British ideological experiment and a French railway that was overtaken by airline executives bringing their own biases into the system. On a plane, my door-to-door time is so long that trips are never spontaneous, so there’s no need for a memorable takt or interchangeable itineraries; on a train, it’s the exact opposite.


  1. Matthew Hutton

    Sleeper trains at least can make journeys of up to 12-13 hours competitive with flying. And while sleeper beds aren’t as cheap as a low cost carrier they are definitely cheaper than business class and I’d guess roughly as comfortable.

    Secondly if SNCF stopped pretending England doesn’t exist in terms of its route planning I reckon it might see a significant uplift in TGV journeys. Perhaps 50%.

      • Matthew Hutton

        😀. It’s more that I think people going by train beyond Paris is probably not something people from Scotland/Wales are likely to do as its too far!

        As a couple of examples why didn’t the French sleeper trains to the south of France connect sensibly with a Eurostar that leaves London at 6pm or 6:30pm that people could catch after work? And why don’t the TGV trains through to Spain or Italy also connect sensibly with an early Eurostar out of London?

          • Alon Levy

            Macron should start by saying France will welcome Ukraine to both the EU and NATO as soon as possible. For a europhilic neoliberal, he lapses into the worst aspects of French nationalism far too often.

          • Borners

            Don’t worry, they always do this. French leaders are either annoying* or terrible. Macron is annoying. All is normal. This is lightyears better than the Balkans years dithering.

            Its also completely understandable that France/Italy/Germany are a little unnerved at the possibility of a victorious total war its not something they’ve experienced.

            *annoying to the English and usually other Anglo cultures also.

    • Phake Nick

      In Japan the rule of thumb for sleeper trains is that they would be competitive when it “depart after the last flight and arrive before the first flight”.

      But the problem with sleeper trains is that it is a niche market. Even between major Chinese cities, which is that largest market size you can get, the number of high speed night trains are still limited to single digit departure each days. Then there are also days when maintenance is needed. High speed rail track maintenance cannot be done when there are other trains traveling right next to the maintenance personnel. It is not worth to reschedule maintenance to daytime when there’d be more demands.

      As for sleeper train on conventional tracks, again demand is the main problem. Like again in Japan, even before Shinkansen get extended to its current form and when there were still more sleeper trains, most routes only get a few sleeper trains even when they are trips between major cities. In comparison airlines are flying tons of 747s between them, each of them probably carrying as much passengers as a single night train which have fewer places due to sleeper arrangement. All in all it can be concluded that night train is just a niche. And even for remaining night trains, the rail companies would still rather put those passengers onto buses than to men the stations and trains at night as well as causing extra restrictions on track maintenance need to be imposed when the night trains are in operation.

      • Matthew Hutton

        Unless you are prepared to get up very early in the morning in general it is pretty difficult to get a plane that gets you anywhere useful before noon, and getting a plane in the evening after work is possible but there’s a pretty narrow window of departure times (and therefore destinations).

        Plus planes are worse for the environment than trains so I’m pretty sure lots of people are prepared to accept a slightly slower journey in exchange for a greener journey.

        • Phake Nick

          Even if you pick a later flight, say 7:30 departure from Tokyo, it can already arrive Fukuoka at 9:30. If the night train arrive later than 9:30, then why don’t passengers sleep well at home in the previous night at one’s own bed, and wake up a little bit early to get the flight?

          As for environmental friendliness, surely a train is usually much more environmentally friendly than a flight, but if a train is empty then it would be burning all the fuel transporting nothing and thus wouldn’t be environmentally friendly.

          • Matthew Hutton

            Ok let’s say I get up at 6am, leave the house at 6:30am, get to the airport at 7:30am, Airport slowness and breakfast followed by an earliest flight departure of 9:30am, flight lands at 11am, more post arrival airport slowness and I’m in the city centre by maybe 12 noon?

          • Phake Nick

            Why would you spend two hours in airport before departure and then another hour after arrival, for a domestic flight?

          • Matthew Hutton

            To be fair in Europe most flights are international, but even for domestic flights there’s security checks etc that aren’t instant. Plus if you travel by road to the airport there can be traffic, and the bus from the car park to the terminal isn’t super quick, and if you go by public transport it can be delayed.

          • Matthew Hutton

            And the hour after includes the time to get from the airport to the city centre of the destination.

          • Tom M

            For Sydney/Melbourne, it’s a genuine three-hour CBD to CBD airport journey…maybe three and a half if you hit traffic on the Freeway at the Melbourne end. Two hours at the airport pre-departure is pretty generous. One hour even in the US is usually sufficient to get through security and pick up breakfast for an am departure. Numerous times I’ve done a 5:30 am leave the house in NYC, get on a 7 am flight out of LGA / JFK and be in Pittsburgh / Cleveland or similar city downtown by 9:30 am.

          • Matthew Hutton

            60 minutes is the minimum check in time at Heathrow – I guess 90 minutes before would be OK if you got a taxi to the terminal. And I guess at smaller airport you could reduce that accordingly.

          • Matthew Hutton

            Right – so given Japanese public transport runs like clockwork you could get a train that arrives at the airport 20-25 minutes before your flight and that would be reasonable.

      • Borners

        Yeah, East Asian evidence is real. But there are caveats. Green politics/values are big deal and holidays in Europe are much longer than East Asia, so the niche is bigger. And the holiday time issue is real, the JR companies are creating hotel sleepers, basically train cruise tours. They are aimed at wealthy old people. Note this one’s main area isn’t Shinkansen/Airport friendly. Much of Europe isn’t so Airport-HSR dense as Japan is.

        Also a real effort at Europe wide sleepers needs to focus on how to get operating costs down (more standardized rolling stock?). Underrated reason for the the collapse of sleepers in Japan is that the most viable routes (Shikoku, Sanin, Hokkaido) weren’t electrified. Bespoke diesel used once every 24 hours? It still baffles me that they didn’t electrify the Hakodate mainline after opening the Seikan tunnel.

        • Phake Nick

          Holidays being much longer in Europe than Asia spread out people’s travel. As people won’t say they MUST spend all that three or five days of the holidays fully in the countryside or back into the ancestral home area. There can be more variation in departure and return time as well as destinations. Hence the load would spread across the network. And European cities are also more distributed across different part of their countries, unlike East Asian countries where half of the country’s economy are on a single side or in a single cluster in the country, causing majority of demand being travel from that particular part to all other directions.

          As for train cruise, they are nice money printer, and is a nice experience for tourists on-board, but their usefulness as a transportation tool is effectively zero. Somewhat similar to the position of Amtrak’s long distance trains, but even more experience-oriented. Hence they should be ignored when discussing transportation network and service. Like you won’t talk about tourist coaches when discussing bus service of a city.

          > Underrated reason for the the collapse of sleepers in Japan is that the most viable routes (Shikoku, Sanin, Hokkaido) weren’t electrified.

          Routes to Shikoku, San’in, and before that to Kyushu, are all electrified.
          Route to Hokkaido have the Hokkaido segment not electrified, but rest of the route they are pulled by electrified locomotives. Diesel locomotives used by the sleeper train when it enter Hokkaido is shared with other trains in the prefecture.
          And those regular sleeper trains also have less income than Shinkansen despite their cost, as they cannot charge the Shinkansen extra fare, according to the ancient fare system being put in place before JRs being split up.
          Also, why would one electrify a track for single digit number of sleeper trains in operation? Even around the time before Shinkansen opened there were only ~10-ish night trains departuring out of Tokyo every night to all the direction, and this number alone obviously isn’t worthwhile for the cost of installing and maintaining electrification equipment.

          >It still baffles me that they didn’t electrify the Hakodate mainline after opening the Seikan tunnel.

          Number of through passenger trains from Honshu to Sapporo was in the number of single digit. Until the opening of Shinkansen to Hakodate.
          Why would one electrify a track for single digit number of trains?
          Also, the conventional tracks are very likely to be abandoned after the opening of Shinkansen to Sapporo, as no one is willing to bare to operation expense of the old tracks, despite being vital logistical route for freight train. Current proposal is to build freight ships instead.

        • Sascha Claus

          Green politics/values are big deal and holidays in Europe are much longer than East Asia, so the niche is bigger.

          Most (Central) Europeans prefer to spend their holidays by holidaying, and not by travelling five days out and five days back with six days on the beach in between. If people would like to spend more time on the road, we’d see more people taking multi-day bus trips to the Mediterranean and less ravel by low-cost airlines.
          And green values … how much has LCC air travel shrunk and how much has (night) train travel grown since ‘Greta’? A lot of media coverage doesn’t mean a lot has changed—remember that media likes to cover the outliers.

          • Borners

            Oh Green-values travellers are swamped by LCC normies and Bavarian SUV campers. But they exist (see German Green party). Minorities can make chunky niches large enough perhaps to meet Rail minimum efficient scale requirements.

    • Alon Levy

      SNCF knows England exists – it’s that place where it charges twice the fare on Eurostar that it does on domestic trains but then the extra revenues go to Chunnel track access fees and the costs of maintaining security theater and non-Schengen passport control.

      Sleeper trains can sometimes be competitive with flying, but the operating costs are a lot higher. For one, the train can’t have very good utilization – the equipment is rarely useful for day trains. For two, there’s no passenger turnover – in theory the train can makes some 3 am stops, in practice those are roundly mocked (for example, on Amtrak).

      • Matthew Hutton

        Is it not feasible to charge enough to cover the costs? And can you not use the trains in the daytime as compartment trains?

        • Sascha Claus

          You can do a lot if things. Rail operators are there to make money; private profit-oriented ones (obviously), and state-owned ones as well. I doubt there’s any railway company whose executives have any personal, monetary reason to care about climate change instead of bottom line.

          And daytime compartment trains … when night trains arrive and have been cleaned, the morning rush hour is likely over. What could one do with compartment trains that can’t be done with local trains that have some hours to spare between morning and evening rush? (I know a very, very niche application that’s even more niche than night trains, and certainly not able to operate without public subsidy, let alone earn money to cross-subsidise night trains.)

          • Tiercelet

            Well don’t keep us in suspense, mate. What niche application do you have in mind?

      • Phake Nick

        What about OBB Nightjet?

        Granted, all those OBB Nightjet services are 1000km or less 1x daily connection or even less frequency than that, which speak about the limited passenger volume it attract, and 1000 km is clearly HSR range that would have worked much more wonderfully with HSR and with more passengers enjoying probably hourly departure between all those big cities served by OBB Nightjet, nevertheless the Nightjet as a service in itself seems to be able to work out the finance of night time service on conventional tracks.

        • Matthew Hutton

          To be fair Paris-Vienna is 11 hours by daytime train. So high speed rail isn’t really working for that journey.

          • Matthew Hutton

            There is a TGV or ICE the whole way from Paris to Munich which still takes ~6 hours to do ~800km.

          • Phake Nick

            That show it is pretty much not as high speed as desired. At average timetable speed of ~130km/h, even Japanese conventional narrow gauge commuter trains can run express service that are only a slight bit slower than this.

          • Matthew Hutton

            Fair! I’m impressed that Japanese narrow gauge can average close to 130km/h. That’s impressive.

            British trains with a top speed of 200km/h merely manage 165km/h as a maximum average speed.

            There’s clearly even more than I thought to learn from Japan..

          • Phake Nick

            I was using the “slight bit slower” a bit liberally, they have track max speed of ~120-130km/h while timetable average is ~100-110km/h, so the time lost is just about 20%. In this aspect the British trains aren’t losing.

            And this show the problem. Even with the Birtish 200km/h trains averaging at ~165km/h, Munich to Paris at ~800km should only take less than five hours, instead of ~6. And we are talking about those British trains that aren’t high speed trains.

          • Alon Levy

            Right now Paris-Munich takes 5:30. Karlsruhe-Stuttgart is 332 km in around 3 hours (and British trains would not average more than this on these tracks, which are a good deal curvier than the WCML and ECML), and Paris-Karlsruhe is 500 km in 2:30.

            And the average speeds of trains in Britain between major cities are not 165; if you gerrymander to pick two stops on a main line fulfilling the same role today that Crewe and Preston did in the Railway Mania, you can pretend high-speed trains are faster, too, a trick used by anti-rail hack Wendell Cox to negatively compare HSR proposals, using their major city-to-major city average speeds, with existing ones, using their nowhere-to-nowhere average speeds.

          • Matthew Hutton

            I agree I’ve been bold, but not that bold 😀. The fastest Edinburgh to London train does an average of ~160km/h over the whole journey, and the hourly express trains do an average of ~145km.

            Yeah clearly I missed the fastest 5h30 trains – probably because I was looking at Paris-Vienna and cutting it short.

          • Phake Nick

            >Right now Paris-Munich takes 5:30.
            >Karlsruhe-Stuttgart is 332 km in around 3 hours (and British trains would not average more than this on these tracks, which are a good deal curvier than the WCML and ECML),
            >and Paris-Karlsruhe is 500 km in 2:30.

            What about Stuttgart-Munich?

  2. wiesmann

    There are two points you post does not address which I feel are relevant differences between train and planes: border crossing and luggage.

    Border crossing have two implications, one is custom checks, the other is which operators are allowed to serve the route. Custom checks are generally absorbed within the security circus in airports, whereas for trains, sometimes they are done before boarding, sometimes on board. Which operators are allowed to run the route was one issue that privatisation was supposed to help, the Paris-Milano route is not served by both Italian and French operators. Before this you need either one country to accommodate the other (DB ICE trains run in Switzerland), or some shared venture (Lyria for TGVs running in Switzerland), or some common entity handling this (like the TEE of yore).

    Luggage does not matter for people doing day trips with their carry on, but there is a whole class of travel where they matter, this is also where airplane get additional revenues. This in turn impact the facilities, like security gates, the airport ones are not designed for people with luggage – something nobody seems to have explained to the French…

    • Phake Nick

      Most HSR proposals being thrown around the world nowadays doesn’t involve border crossing. The only threes that I can think of that would involve border crossing, are Vancouver-US Pacific Northwest, Montreal&Toronto-Detroit&Chicago, and the now-cancelled Singapore-Malaysia.
      Both the Vancouver route and the Singapore route involve only 1 major station across the border, and thus can copy the arrangement of Eurostar.
      US-Canada trains can also use TSA pre-clearance to smoothen cross border traffic like air travel.

      • Henry Miller

        Every once in a while someone throws out some form of US-Mexcio high speed line as well, and I suspect you could make the numbers work on paper to extend such a line farther into central America, but the lack of wealth means I doubt it would work in the real world even if you could make customs work out.

        • Eric2

          San Antonio-Monterrey is the only reasonable candidate (except the trivial San Diego-Tijuana). From Monterrey you could connect to central Mexico (Mexico City and maybe Veracruz). But the next significant city is Guatemala City which is much too far, poor, and mountainous for a HSR connection to the north.

      • adirondacker12800

        Montreal is closer to New York or Boston than Chicago is to Toronto. So is Philadelphia. Boston and New York are about the same distance from Toronto as Chicago is.

  3. Martin

    All of these points I think make a lot of sense, except the critique of yield management on prices.

    That is just actually pricing the product people want to buy (a trip at a specific point). E.g. prices of apples goes down when there apple are plenty, and go down when people want to eat them.

    You can argue for a planned economy approach (fixed prices), if supply is so vast, and the cost of fulling supply through extra capacity is so cheap, that you can get basically provide free capacity at all times at little cost (there are public services like that). Despite your claims, I do not see that as a particularly common situation for trains in the world at all. Rather the opposite, with massive capital demands for new tracks, trains, and larger stations in most contexts, and also that HSR actually has non-trivial running costs.

    Otherwise, I see absolutely no point why you should price a 05:00 train the same as the 08:30 one, or the summer train to the beach the same as the winter train. The systems I have seen that mostly use fixed pricing such as the Taiwanese and Chinese HSR, have massive issues of overbooked trains (and all kinds of gray markets started operating to meet this demand), combined with virtually empty trains at other points, and broadly seems terrible inefficient on these points (for the Taiwanese one I think fares are regulated by law).

    I think in Japan the lack of yield management basically does not collapse because there are nearly always completely parallel tracks to the Shinkansen tracks that basically fulfill the same function of regulating demand. They have also lost lots of traffic to (slower!) low-cost airlines, losing market share only on cost, a market that probably wouldn’t exist if there existed reasonable prices for Shinkansen tickets at unattractive time points. I think East Asian ticket prices are diffidently not best industry practice, and while for a tourist or spontaneous traveler uniform pricing feels convenient, not being able to find a ticket over the next 5 days, not matter how ridiculous the time (Taiwan and China) is definitely not.

    • Matthew Hutton

      I agree that cheap advance tickets are good. But I do think that having flexible tickets available with unlimited capacity at €0.20/km or so is probably a good idea.

    • Eric2

      Last time we discussed yield management, I think the most reasonable idea I saw was that yield management is OK if it is easy to switch your ticket to a different train if your schedule changes.

      • Matthew Hutton

        I think that would be very hard to achieve in practice. People don’t like even small amounts of friction.

    • Richard Gadsden

      The UK train system requires very high flexible fares because it has a policy of not requiring reservations for any train, so if you have a flexible ticket you can just board any train. The result is that very high prices are used to deter passengers from using the most popular trains and there is also heavy crowding on some trains – standing for 2+ hours is extremely unpopular, and is caused by not allowing reservation-only trains.

      Being able to make the 8:30 arrival a reservation-only train would allow the price of tickets for that to be lowered to something more reasonable. The train would sell out, of course, but people would then get the 8:10 or 8:50 arrival and so on. They could then sell discounted tickets for less popular trains.

      Obviously, this is only an approach for longer-distance services; no-one wants to have to book their commute in advance.

      The German system is similar in theory, but I don’t think it generates the sort of mass crowding that the ECML (in particular) is famous for. Not sure what they do to discourage people from using Flexpreis tickets and all getting the same train.

      • Matthew Hutton

        To be fair I think the main benefit of flexible fares isn’t for people who want to get into London early – it’s more for people who are going to a business meeting or to meet a friend or whatever and aren’t sure how long it will go on for. And it seems pretty unlikely that someone would go a long way by train and find their welcome ran out at 6:30am.

        On the other hand perhaps the railway could be more flexible and run more long distance trains at peak into London if that is still desirable. Perhaps Welwyn North could not get any trains in the peak direction at peak times as that would free up capacity as that is the pinch point – and then to mitigate the harm perhaps the peak fare for Welwyn North could be cheaper.

      • Phake Nick

        Why not just let passengers stand in the train then, if they want to reach destination at the desired time, and assuming that it is not possible to run extra trains?

        • Richard Gadsden

          That is what happens at the moment, but trains don’t have that much standing capacity.

          When you reach the stage where the train is full and people are standing for two hours or more, people get angry.

          The wcml uses really high prices to prevent this; the ecml ends up with people paying 2/3 of the wcml fare, still way too much to be stuck standing.

          • Phake Nick

            Then why not have a premium class of service? Tell people to pay for the premium if they want a seat in a premium class, else accept having to stand.
            Or else, run more services? I don’t think like 20 minutes headway is the best that can be provided

          • Richard Gadsden

            There are lots of things that I think they would like to do but aren’t by the regulator. First class is technically not guaranteed a seat either, but I’ve never seen it full.

            And the lines are at headway capacity due to too many routes sharing one set ot tracks into London.

          • Matthew Hutton

            I mean strictly you could shut Welwyn north and gain a few extra slots of capacity. And certainly you could shut Welwyn north in the peak direction.

    • Henry Miller

      I assume we are talking only about inter-city trains here, and not some form of intracity rail. Intarcity rail should have most people on an unlimited rides month pass, and as such there is no ability to price people to a cheaper train.

      Either way, it is in general best add more capacity when your trains get full. The cost of trains is largely in the track, so a two half full trains isn’t much more costly than one full one. If possible run more trains, as frequency is very nice for users and will increase ridership at bit in many cases (the need to add more passing sections or miss connections at a knot may make this undesirable). Adding more cars is also possible, but may require longer platforms.

      Still, yield management may sometimes be required to get people off that busy 5:00 train. However since 100% full trains isn’t really needed for train operation you don’t need to be as confusing about train pricing.

    • Phake Nick

      > I see absolutely no point why you should price a 05:00 train the same as the 08:30 one, or the summer train to the beach the same as the winter train.

      Simple. Don’t run the beach train in winter and don’t run the train at 5:30. Or only provide discount on them but not overcharging other trains above standard fare.

      > The systems I have seen that mostly use fixed pricing such as the Taiwanese and Chinese HSR, have massive issues of overbooked trains (and all kinds of gray markets started operating to meet this demand), combined with virtually empty trains at other points, and broadly seems terrible inefficient on these points

      The demand is there, no matter what pricing method you adopt, you cannot prevent people from trying to travel in the holidays. And you cannot build enough track to meet the demand for the few-time-in-a-year travel season. Even buses and air flights in China which some of them do have flexible pricing still face this problem, as the entire country have holiday season at the same time.

      > I think in Japan the lack of yield management basically does not collapse because there are nearly always completely parallel tracks to the Shinkansen tracks that basically fulfill the same function of regulating demand. They have also lost lots of traffic to (slower!) low-cost airlines, losing market share only on cost, a market that probably wouldn’t exist if there existed reasonable prices for Shinkansen tickets at unattractive time points.

      Those parallel tracks and LCCs are only offering minimal extra capacities compares to what the Shinkansen offer. They have much lower frequency and also can carry fewer people per trains than what a Shinkansen train can carry.
      But it is an advantage of LCCs, that they can move aircraft from international flights to domestic flights during domestic travel seasons. It is desirable to simply let LCCs absorb this extra demand. Something China/Taiwan cannot do because their domestic holidays are also when people travel the most with LCCs, unlike Japanese LCCs which have sizable market from inbound travelers who typically have different travel seasons compares to Japan domestic holiday times.

      Also, Shinkansen do have early bird tickets for trains in inconvenient time, and they do pull trains from operation in low demand days, labeling those trains as “temporary”.

      > I think East Asian ticket prices are diffidently not best industry practice, and while for a tourist or spontaneous traveler uniform pricing feels convenient, not being able to find a ticket over the next 5 days, not matter how ridiculous the time (Taiwan and China) is definitely not.

      As I mentioned above the problem is entire country having holiday and travelling in the same direction together. No matter rail, ship, planes, or even highways, they are all packed to the max. Nothing you can do in a single mode to solve this problem.

      • Alon Levy

        (European LCCs can’t easily move planes between domestic and international flights, because practically all flying here is international and intra-European; when Norwegian provided intercontinental service, it was with 787s, not the 737s of European service.)

        • Phake Nick

          But most of Europe are now in same border control and custom union. Intra-Eurpopean flights still need international customs or passport control?

          • Herbert

            You need to establish your identity to get on a intra-Schengen flight and Schengen is not the same as the customs union (so customs checks but no immigration check when going to Switzerland).

            But for the operation of an LCC it makes little difference whether a flight is Schengen or not. They’ll just have the pax prove that they have/don’t need a visa and that’s it…

    • Alon Levy

      SNCF yield management isn’t just charging more at busy times than at less busy times; DB does the same, and I think Korail has a system of discounted fares for this as well. It’s much more pervasive, consisting of,

      • Fare buckets based on how far in advance one buys.
      • Fare buckets for refundable and nonrefundable tickets.
      • Different fares based on where one buys tickets, including an overpriced website marketed to American tourists.
      • Non-interchangeable tickets in case of train cancellation: when my Eurostar was delayed 1.5 hours, I could not change the ticket to a London-Lille departure with a connection to a domestic TGV back home, because TGV and Eurostar tickets are not interchangeable for each other even though Eurostar is majority owned by SNCF. (In contrast, at the Franco-German borders tickets are interchangeable between SNCF and DB.)
      • Multiple brands with non-interchangeable tickets: OuiGo for the ordinary Macron voter (Le Pen voters drive), InOui for the person who also likes En Marche’s Parisian candidates.

      SJ for example has variable fares and not much of a takt, but I could get to Växjö with a decently timed connection to a regional train at Alvesta, and there isn’t (or at least wasn’t in 2014-6) anything as drunk as the OuiGo vs. InOui distinction. A Francized SJ would be running cheaper trains to Södertälje instead of Stockholm C and not even print connecting SL tickets together with them, so passengers would have to queue at Södertälje to buy commuter rail tickets, like how you can’t buy combined Arlanda Express + T-bana tickets.

    • Max Wyss

      With all due respect, but yield management is poison for rail travel. In fact, it is a sign of lousy service (which SNCF proofs any time).

      It is one of the core advantage of rail travel to have flexibility. Yield management defeats this advantage.

      The German model is a compromise, as it has just fixed discounted prices, but does not set the price for every individual seat.

      Unfortunately, SNCF got infested by phased-out airline managers.

      • Herbert

        There are advantages to steering flexible customers onto your empty trains.

        If all trains cost the same all the time, you’ll have even more demand on the high-demand trains and even less on the low-demand trains, because there is no incentive to use the low-demand trains.

        And we all know capacity is most expensive at the peak. So why not price that in?

        • Henry Miller

          > So why not price that in

          Are you a company running the train, or society/city?

          A company trying to get the most profit needs to price that in to make best use of infrastructure already built. Even then they need to carefully consider if building more infrastructure will bring in enough new customers as to be worth it (even though now the trains are never busy enough to need congestion pricing)

          As society and a city you are concerned about car traffic on expensive roads. You are concerned about the environmental effects of cars. You are concerned about the space cars take. You are concerned about safety. You want your people to be in better physical shape. You have many other concerns that I can’t think of. Getting people out of cars and onto transit very much helps with a lot of your other goals. Discouraging riding transit at anytime is not useful to your overall goals. You really want everyone on an unlimited rides family pass as then they will think about using transit for all trips even if they have a car.

    • Joseph

      Taiwan HSR does offer discounts for less desirable times, and judging from a quick fare search they are available only for early purchases. The fact that it’s crowded at peak times has far more to do with the extremely linear, one-way nature of intercity travel in Taiwan. And in my experience tickets are rarely hard to purchase outside the most obvious peak times- if you need to go south on a Friday, just go before the end of the work day (or very, very late at night).
      The TRA does not have discounts, though it arguably should.

  4. fbfree

    I can think of one other opperational difference of note: planes have higher crewing requirements than trains. Thus, the likelihood of crewing problems, and thus needing to know which plane to cancel based on reserved passenger demand, is higher than for trains.

    • Borners

      Shssh. You can’t say that trains are efficient in terms of labour costs! Think of what you’d do to airline shareholders, unions and Anglosphere ideologues of either stripe.

    • Max Wyss

      I would not say so… OuiGo requires almost a dozen staff at an intermediate station, and blocks a whole platform for half an hour, if not more. But we know why SNCF introduced OuiGo… (and, yes, it has nothing to do with providing lower fares)…

  5. plaws0

    Lol. I was watching trains at Springfield (Mass) Union Station on the tube of ewes yesterday. Amtrak 449 arrived from Boston on it’s way to Albany to meet 49 where they are combined to Chicago. I pointed out in chat that with a full three coaches, that the train hardly needed 8000 HP (2 Genesis units) and the assemble Amtrak apologists claimed that it was needed because of the grades. Did I mention 8000 HP for 3 coaches? I suggested that it would be better for all if that unit had been left at ALB eastbound and picked up again WB. I have no idea if this was routine or a special occasion (ferry move of OOS unit for service? I would have hoped BOS was more capable but … Amtrak).

    ANYWAY … 7 minute dwell time. Did I mention 3 coaches? I use coach in the generic sense since there was a not-a-bagliner that could have been a sleeper – but that means even *fewer* possible passengers.

    None of that is new, of course. Still sad, though, especially since Springfield should be a busy hub. As it is, it’s served by two railroads even if trains from both services relax there for hours at a time instead of being out earning revenue.

  6. adirondacker12800

    LCCs, which offer simpler one-way fares.

    Fare for London to Berlin, on Ryanair, for dates in June are 19.99, 22.99 or 24.99 or 25.99. If I want to go tomorrow and come back on Sunday it’s 137.99, 158.99, 185.99 or 260.91. I didn’t check anything else.

    • Alon Levy

      Yes, so that’s the advance vs. walk-up fares; contrast this with trans-Atlantic flights, where even for approximately the same dates, you pay less for a roundtrip if you stay over a Saturday night.

      • adirondacker12800

        Business travelers want to be home on the weekend. Leisure travelers don’t.

  7. Mikel

    high-speed rail carriers aim at fleet uniformity as much as practical, using scale to reduce unit maintenance costs

    *laughs in Spanish* Renfe has always been pretty drunk about this; they currently have 10 different classes of high-speed trains, and now they’re buying not one but five new classes of Talgos, totalling 49 trainsets*. I’m not making this up.

    For what it’s worth, the liberalization –not privatization– seems to have been moderately succesful so far. In the Madrid-Barcelona corridor, rail ridership is already 20% over pre-pandemic levels while air mode share has crashed. Renfe’s new fare structure is completely insane but I think we’ll have to wait a few years until Ouigo and Iryo have started operating in other major corridors and the market has stabilized (along with energy prices). Hopefully we’ll get a major shakeup and depolitization of Renfe’s leadership along the way.

    *that would be 10x S-106 AVE, 5x S-106 AVLO, 10x S-106 AVE RD, 5x S-106 AVLO RD, and 19x S-107 AVE RD.

    • Alon Levy

      Yes, and the introduction of the Italo raised ridership on the city pairs it competed on, which are the more peripheral ones poorly served by direct Frecce – things like Turin-Naples. But that’s in context of Trenitalia not really doing timed connections; Italy is Germanizing its regional rail operations with success, but its intercity rail is German in construction and French in operations, the opposite of good practice.

      • Mikel

        What does “French in operations” mean in this context? Did they copy SNCF’s weird “ground level airline” operations? 😦

        Renfe doesn’t do timed connections either (though they are sloowly warming up to the concept, mainly in Galicia), but it seems like Italo’s business plan was different from what’s happening in Spain. Here Ouigo and Iryo are going after the routes that are currently best served by Renfe, because they are the most profitable ones: Madrid-Barcelona, Madrid-Valencia/Alicante and Madrid-Sevilla/Málaga. The few non-Madrid that are somewhat competitive with Flying (Barcelona/Valencia to Sevilla/Málaga) don’t seem to be in their immediate plans. What opponents of liberalization fear (justifiably, IMO) is that Renfe will respond to dwindling profit margins on the largest corridors not by streamlining their intercity operations nor exploiting their monopoly of mainline regional rail through timed connections to their for-profit trains, but simply by jacking up prices on captive users of secondary corridors such as Madrid-Galicia, Madrid-Asturias and Madrid-Granada.

        Re: trains having lower variable costs, I’d love to hear your thoughts on track access charges — perhaps you could write a post about that? My impression is that in a monopolistic situation they’re just a fudge factor to make the infrastructure administrator nominally profitable and the service operator nominally subsidized, or vice versa. Whereas in a competitive market serious tradeoffs arise, e.g. is it okay to indirectly subsidize operators (both passenger&freight) through low track access fares if that takes market share away from less environmentally-friendly modes? For example, Madrid-Barcelona operators whine about how track access fares being nearly €9,000 per train (so at least €17-25 per ticket) hurts their ability to compete with planes on price, and that’s after Adif decided to offer a discount to encourage higher frequency service. In contrast, airports (both private- and state-owned) are usully run explicitly for profit and therefore try to extract the highest fees they can get away with.

    • Herbert

      Just like Lufthansa ordered a bit of every plane ever for the last fifty years or so…

  8. Tom M

    Is vehicle capacity additions a relevant difference? For an airline, in the short term, you can shuffle planes around to respond to demand changes. Mid-term, you can purchase or wet/dry lease aircraft pretty quickly from the second-hand market, and longer term buy new. For a train operator, if you have sufficient fleet size and track restrictions (loading gauge etc.) aren’t an issue, I guess you can shuffle trains around, but mid or long term, you’re only option is to buy new?? No second-hand market to pick up a short-term lease for a quick boost in capacity.

    • Alon Levy

      If the capacity you need is for busy times, then on an airline you’re either lugging empty seats or keeping those extra rush planes on the ground most of the day.

      • Phake Nick

        Airlines can and do lease planes from other part of the world during busy times, as people have holidays in different time in different parts of the world, like Islamic countries wouldn’t have much Christmas traffic while European countries wouldn’t have much Ramadan traffic.

  9. Patrick Jensen

    A couple of airline innovations I think are long overdue in railway operations are interlining and codesharing. It’s baffling how difficult it is to book train tickets across systems in this day and age.

  10. Mark N.

    You didn’t mention it explicitly as a key difference between rail and air transportation, only implicitly alluding to it with your last point, but a key difference of rail — and generally a great advantage over air — is its ability to serve those intermediate stops. This aspect, perhaps more than any other, convinces me of the great potential HSR has on a widespread scale in the United States. We even have a term for those (many) parts of the country inadequately served by air transit: “flyover country.” An HSR network could be an effective way to reconnect such areas, both economically and, perhaps more importantly, mentally, into the fabric of the nation.

    • Herbert

      The problem with that is that the party which dominates flyover country is aggressively anti rail on the national level

      • Mark N.

        Yeah, that’s the paradox that sadly creates barriers to precisely those policies which would benefit those areas the most.

  11. Herbert

    To be fair “a bunch of dead end stations spread haphazardly across the city” is bad from both an operational and a passenger comfort standpoint.

    Possibly a “Paris Hauptbahnhof” can never be built and a “London Hauptbahnhof” would be an even taller order, but having a category of train that serves Disneyland and no other Paris station would make sense if Disneyland had the added capacity (which it doesn’t)

    • Matthew Hutton

      London is pretty close to that with Euston, St Pancras and kings cross all within walking distance for most of the long distance trains to the north.

Leave a Reply to Richard Gadsden Cancel reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.