Why Does TGV Ridership Overperform Models?
I’ve found some TGV ridership data with which I can check the model I use for high-speed rail ridership projection. The model is trained on Japanese data and has flaws in Japan too, but I’ve wanted to see how well it ports to Europe, where I don’t have as complete a dataset of ridership between pairs of metropolitan areas. Thankfully, I have just found a short Omnil report from 2015 about intercity rail ridership at the Parisian terminals, from which we can extract some information. The TGV overperforms the model substantially; this can be explained with fudge factors, but those fudge factors only work if we assume that the aspects of the TGV that seasoned rail advocates hate don’t matter much.
The model
As a reminder, the model posits that the annual ridership in millions between two metropolitan areas with populations 
The model is very accurate for ridership between Tokyo and other cities on Honshu; it overpredicts inter-island ridership, but becomes correct if we replace the Japanese air/rail modal splits with European ones, where taking the train over five hours is more normal than in Japan. I would expect that in isolation, European ridership should overperform it, because fares here are much lower, about 0.10-0.11€/kilometer compared with about $0.23/kilometer on the Shinkansen. French ridership significantly overperforms, beyond what the fares alone can explain, as we will see.
We will need to modify the model as written above for the French case anyway. TGV ridership relies on direct through-service from Paris to every city in France, including many that are not on the network of dedicated high-speed lines (called LGVs); trains serve those by diverting from the LGVs to classical lines, on which they travel more slowly. Therefore, while we can apply the model as above for connections that entirely use LGVs, like Paris-Lyon or Paris-Marseille, we need to consider the slower speeds for connections that use classical lines. For those, we assume that trains average 220-225 km/h; this is the rough average speed of the express Shinkansen trains as well as that of the TGVs to Lyon and Marseille. Thus, the model, at travel time t, is,
The floor of 500 km, or in this case a trip time of 
TGV ridership
I have never been able to find city-to-city or station-to-station ridership figures in France. The Omnil report is no exception: it reports ridership at the Paris stations and breaks down where people are going by region of France in the geography of 2015, before the merger of some regions.
The total ridership at the Paris stations, including TGVs, low-speed intercity trains, and other regions’ regional trains (TERs) but not Paris-area regional trains (Transilien), is 443,000/day; of those, the TGVs comprise 239,000 and the slow trains 204,000. The four Parisian terminals with TGVs – Gare de Lyon, Gare du Nord, Gare Montparnasse, Gare de l’Est – have 92% of the TGV ridership in the region, while the other 8% are at suburban stations on bypasses around the city, like CDG Airport. Ridership is asymmetric: two-thirds of those 443,000 daily riders don’t live in Ile-de-France, which is what we should expect of a commuter-heavy ridership profile. Within Ile-de-France, 63% of passengers originate or are destined to Paris itself and another 21% for the Petite Couronne suburbs, showcasing destination centralization – Paris is only 17% of regional population and about 33% of regional employment, but 63% of those interregional and intercity trips go there and not to the suburbs.
There is also a breakdown of where passengers are connecting, by region of France or country. Picardie is increasingly an exurb of Paris, to the point that as France was debating the merger of regions in the early 2010s, one proposal was to detach its southernmost department, Oise, and attach it to Ile-de-France; 19% of the non-Francilien passengers originate there and 10% of Franciliens go there, for a ratio of nearly 4:1. More relevantly to high-speed rail, Rhône-Alpes is 9% of both non-Francilien and Francilien ridership, for a ratio of about 2:1, and a total of about 40,000/day, or around 13 million/year. PACA is 5% of non-Francilien and 7% of Francilien ridership, for a ratio of about 1.4:1 and a total of 25,000/day or around 8 million/year.
So we need to evaluate our model against an observed ridership of 13 million between Paris and Rhône-Alpes, and 8 million between Paris and PACA. Both sets of numbers involve multiple city pairs, with fairly long tails: France is a country of small metro areas, the median person living in a metro area of 330,000, whereas half of Japan lives in the metro areas of Tokyo (37 million), Osaka (18 million), and Nagoya (9 million).
French metro areas and the model
France recently changed its definition of metro areas. The old one, the aire urbaine, was similar in definition to the American metropolitan statistical area; the new one, the EU-wide functional area, generally spits out slightly larger numbers, though it still seems tighter than the Japanese definition. The functional area of Paris, comprising Ile-de-France, about half of Oise, and surrounding communes, has 13.2 million people. The new definition splits Nice and Cannes apart, which is good, since both have TGV service to France.
| Metro city | Population | Trip time | Prediction |
| Lyon | 2.29 | 1:58 | 4.586 |
| Grenoble | 0.72 | 3:01 | 0.999 |
| Saint-Etienne | 0.5 | 2:58 | 0.771 |
| Geneva (French part)* | 0.44 | 3:13 | 0.485 |
| Annecy | 0.3 | 3:45 | 0.321 |
| Chambéry | 0.26 | 2:52 | 0.49 |
| Valence | 0.26 | 2:12 | 0.805 |
| Bourg-en-Bresse | 0.14 | 1:50 | 0.49 |
| Marseille | 1.88 | 3:07 | 2.02 |
| Nice | 0.62 | 5:48 | 0.24 |
| Toulon | 0.58 | 4:02 | 0.47 |
| Cannes | 0.39 | 5:18 | 0.199 |
| Avignon | 0.34 | 2:40 | 0.701 |
The Rhône-Alpes metro regions combine to a predicted ridership of 8.95 million; actual ridership is higher by about 50%. The PACA metro regions combine to a prediction of 3.63 million; actual ridership is higher by a factor of maybe 2.2.
Note that the prediction is already based on some optimistic assumptions. The trip time is the best that can be sustained multiple times a day; the issue of frequency is ignored, so the effective trip time on connections from Paris to cities like Annecy with a train every three hours gets no malus, even though the Japanese city pairs that the model is trained on get multiple express trains per hour. This is relevant, because as we examine fudge factors below to rescue the model, we need to keep ignoring or at best minimizing the malus due to poor frequency and lack of trip spontaneity in the ticketing system.
Fudge factors explaining the overperformance
We need to explain why Rhône-Alpes overperforms by 50%, and PACA by more than 100%.
Fares
The average JR East Shinkansen fare revenue in 2020-1 was ¥23.8/passenger-km (source, PDF-p. 50), and has risen little in the last 10 years. The average TGV fare revenue in 2019 was 0.10€/passenger-km (source, pp. 16 and 20) and has likewise little changed in nominal terms. These differ by a factor of 1.6. The elasticity of high-speed ridership with respect to price varies widely by study; the Italian study linked above says -0.37, one Spanish study says -0.59, and Börjesson’s lit review says -0.59 for non-business trips and -0.72 for business trips. A value of -0.5 explains a factor of 1.27 overperformance by itself; a value of -0.6 explains a factor of 1.33.
In fact, Germany, charging similar average intercity rail fares to France, seems to overperform the Shinkansen model too. I have little data here, only line-wide Berlin-Hamburg and Berlin-Munich, both of which look like they overperform by about 20%. This can result from a 30% overperformance mitigated by the issue of lower speed: the modeled prediction is based on trip times, but when trips are shorter than about 2:15, the model stops seeing the impact of slowdowns – Berlin-Hamburg is 1:44 and Berlin-Leipzig is 1:13, where at Shinkansen or TGV speeds they’d be 1:17 and 0:45 respectively.
Metro area size
French metro area definitions, even with the new functional areas, are somewhat tighter than Japanese ones. The functional area of Berlin has 5 million people, but reckoned the Japanese way (1.5% of the age 15+ population commuting to the central city), practically all of Brandenburg would count, a population of 5.7 million in total. This is likely more significant in PACA, where the above-listed metro area are 80% of the total population, than in Rhône-Alpes, where they are 90%. It’s possible even Paris is a bit bigger than 13.2 million – but only a bit, since Ile-de-France and Oise together only have 13.1 million. This factor can scrounge some extra ridership, but probably no more than 10%, maybe a bit more in PACA.
Leisure travel
Provence is renowned for its tourism, which generates extra trips out of Paris beyond what we should expect from population alone. This should disproportionately affect Nice and Cannes; for what it’s worth, I’m seeing seven weekday trains from those cities and Toulon to Paris, I believe all skipping Marseille, and 14 trains from Marseille; if we take ridership as proportional to the offer, this does show some Riviera overperformance relative to Marseille, though not by much.
Of course, the majority of Paris-PACA ridership comprises Provençals, not Franciliens. But perhaps the 1.4:1 ratio of Provençals to Franciliens is atypically low, and the 2:1 ratio in Rhône-Alpes is more normal of capital-province relations; I have no Japanese numbers on this, and would overall expect to see similar asymmetries in both countries, given their similar level of economic capital-centricity. If 2:1 is typical, then the extra leisure ridership from the capital to make it 1.4:1 adds a total of 14%, which is far less than PACA’s overperformance relative to Rhône-Alpes.
Metro area coverage
The PACA cities have multiple stops. The population distribution in the Riviera is linear, and multiple cities with extensive leisure (like Saint-Tropez) are served by the TGV. Marseille likwise has a second stop at Aix-en-Provence, close by car to its northern suburbs to the point that I’ve heard it called Marseille-bis. If we split metro Marseille’s population 2:1 between Marseille and Aix, then the 0.8 exponent in the model produces a 14% increase in ridership. 14% and another 14% from leisure combine to 31%, which explains the majority of the PACA overperformance relative to Rhône-Alpes.
Competition with air in small cities
The TGV competes with cars and planes; domestic buses are almost a non-factor, and were entirely a non-factor in 2015 (they’re called Macron buses because it was Macron, as minister of economics in 2014-6, who passed the reform that allowed them). In Rhône-Alpes, competition is entirely with the car: Lyon is just close enough to Paris that air travel can’t compete; in PACA, competition is mostly with the plane, especially beyond Marseille.
The population distribution in both Rhône-Alpes and PACA may favor the train. The issue is that the secondary cities of Rhône-Alpes are around three hours from Paris, at which point the train is strongly favored but planes normally still exist, as in Marseille. However, those cities are scattered all over the region, and so there is no single airport that could serve them, except Lyon – and if the choice is to take the train for three hour or to drive an hour to Lyon-Saint-Exupéry, then the train can just demolish air competition.
In PACA, the same is true for the secondary cities. Nice has a strong airport with many flights to Paris, buoyed by the leisure market, but Toulon and Avignon don’t; on the eve of corona, Toulon-Hyères had 500,000 passengers a year, most not bound for Paris.
I believe this effect on air-rail competition is more significant in Rhône-Alpes than PACA. However, air competition is overall more significant in PACA than in Rhône-Alpes, and thus it likely effects a similar boost to TGV ridership in both regions, or perhaps is more significant in PACA, explaining the remainder of its overperformance.
Some conclusions
I don’t think the TGV’s overperformance of the model invalidates the model. Most of the overperformance in Rhône-Alpes can be explained by fares alone, and I think the rest can be explained by the modal split versus air being more favorable than in Japan given the small size of Annecy, Saint-Etienne, and so on. Most of the overperformance in PACA relative to Rhône-Alpes can then be explained by leisure travel and the good metro area coverage of the TGV thanks to Aix and the linear population distribution of the Riviera. However, these fudge factors have implications for rail planning in France, Europe, and beyond.
Connections to smaller cities
The modeled prediction is that Lyon and Marseille comprise little more than half the ridership to Paris from their respective regions. Moreover, the overperformance of TGV riderhip relative to the hinkanssen model likely comes disproportionately from smaller cities, due to their lack of good air connections. This underscores the importance of good service not just to million-plus metro areas but also to the tail of metro areas of half a million, give or take. Those metro areas are less important in rich Asia or the US, but are important throughout Europe.
This service to smaller cities can take the characteritic of TGV-style direct connections to Paris on classical lines. In Switzerland, the Netherlands, Austria, and increasingly Germany, service to smaller cities is provided through timed connections at carefully-chosen nodes; the Swiss network particularly excels at this. But the French system’s ridership is such that it not obviously inferior, and is unlikely to be inferior to the German system at all. Thus, a country like Poland or Britain can safely choose between the French and German system, or even mix them.
The issue of frequency
The low frequency of TGV services to smaller cities – trains run every two to three hours, often timed to just miss regional trains – should be visible as a serious malus to ridership. But it isn’t. Perhaps it exists and countermands the effect of lack of air competition to cities the size and distance class of Grenoble – but Grenoble is not Nice, and air competition there even under more favorable scenarios to planes would be second-order.
At the same time, there are markets where the TGV is visibly much weaker. The TGV’s modal split between provincial regions is not good. Because trains from Paris to Marseille don’t stop at Lyon, and trains from Paris to Lyon don’t continue onward to Marseille, the Lyon-Marseille city pair cannot piggyback on strong connections to the capital the way same-side pairs of provincial Japanese cities can. The dedicated Marseille-Lyon trains have an inexplicable six-hour gap, with frequent service on both sides of it, and the Toulon-Lyon trains are even worse. The modal split is evidently weaker – in 2009, nearly everyone drove betwen Lyon and Toulon (the 2023 number in the link are speculation for what if an LGV is built to Nice), even over a rail-friendly distance of about 390 km, averaging around 130-150 km/h.
So while the system that centers direct trains to Paris is not suspect, the lack of frequency on shorter connections between secondary cities is. This could be resolved with buying rolling stock that makes boarding and alighting faster, with two door pairs per car rather than just one; TGV connections not including Paris run local, and since the trains are not optimized for many stops, those connections have low average speed, which in turn discourages SNCF from providing more frequent local connections.
Liberalization
The EU is increasingly forcing national railways to allow on-rail competition. This is an idea imported from the UK, where John Major’s privatization of British Rail split up operations and infratructure, the latter eventually renationlized; in Japan, privatization broke up JNR into regional JR companies, each responsible for both infrastructure and operations as in the pre-nationalization era of rail, and in the US, the breakup of Conrail likewise restored the pre-nationalization status quo. SNCF resists the mandate for competition in increaingly spiteful ways: it makes up excuses why RENFE can’t operate on its network, and where it does operate, it won’t even let its crew use break rooms at French stations. Eurocrats, even more progressive ones, treat SNCF as public enemy #1.
And SNCF’s anti-competitive monopoly on domestic rail travel generates high rail ridership. Italy and Spain have both seen sharp increases in ridership from the competition mandate. But Madrid-Barcelona, offering worse frequency and a more broken market than the domstic TGVs (domestic TGVs are split just between lower-price OuiGo and higher-price InOui brands; Spanish high-speed trains have more classes of train on thinner markets), don’t perform nearly so well. Madrid-Barcelona riderhip in 2019 was 4.4 million; the modeled prediction is 4.1 million for this city pair alone, and 6 million including intermediate trip to Zaragoza. Riderhip ha risen since the introduction of competition in 2020, and media coverage has been laudatory, and at times depreating of France for failing to liberalize – but the 50% growth in ridership cited in most articles still leave the line barely overperforming the high-fare Shinkansen and strongly unerperforming the TGV.
European media should be less credulous of promises of private-sector efficieny and recognize that the TGV’s model of public-sector monopoly, with integration between infraatructure and service (even if this means shoving direct trains to Paris on trunk line rather than building a Swiss integrated timed transfer system), produces better outcomes than competition. Germany has the same model too and, relative to how slow its trains are, has good outcomes too; Switzerland, the undisputed leader of European rail ridership, resists privatization entirely. Private competition did not invent high-speed rail, and where it has been introduced it has so far failed to produce outcomes on a par with what the TGV has with entirely public operations.
Pedestrian Observations 
Italo + TrenItalia seems to produce quite high ridership on Turin-Milano-Roma-Napoli. I was quoted something like they currently run together 8 TPH throughout the day?
It isn’t quite an all-day standard hour timetable. In the busiest hours the services that do Bologna – Rome tends to be
– Milan – Rome fast – 4tph (2tph from each operator)
– Milan – Rome via Florence – 2tph (1tph from each operator)
– Venice – Rome via Florence (1tph from each operator)
So yes, up to 8tph.
Services from Milan often start from Turin; services to Rome often go on to Naples and Salerno.
Bologna – Florence; Florence – Rome; Rome – Naples – Salerno almost have metro frequencies but alas no metro-like ticking.
I suspect leisure travel accounts for a fair part of the difference.
According to OECD figures for 2019, France had 207 million domestic tourist trips, or around 3.1 per person; while Japan had 312 million domestic tourist trips, or just under 2.5 per person. France also draws thrice as many international visitors outright – 91 million versus 32 million – for obvious reasons like permeable land borders. Overall, that’s 4.4 tourists per capita against 2.7, a proportionally larger pool of leisure demand the TGV can tap on compared to the Shinkansen.
Distribution of traffic within the year could also be a factor. France presumably has summer and ski season peaks, but nothing like Japan’s Golden Week-Obon-New Year trifecta when everyone travels at once in the space of a few days. During those periods, the Shinkansen lines hit standing capacity and turn away many potential riders; conversely, I imagine TGV services being able to absorb a steadier stream throughout the year.
Some other musings on how leisure might be especially significant in explaining France-Japan ridership disparities:
– If we generalise from trends in aviation, travel for holidays or friend/family visitation is less sensitive to travel time (ameliorating the TGV’s frequency and speed weaknesses), and more sensitive to price (the main downside of the Shinkansen).
– International visitors, a segment that France decisively leads, might have a greater inclination than their domestic counterparts to pick fast trains: not wanting to bother with a rental car (fuel, tolls, parking, navigation, international driving permit, etc.), special tickets/deals, cramming multiple locales in one itinerary, and so on.
– The Riviera tourism area has fair TGV connectivity. Hokkaido and Okinawa, Japan’s third and fourth most-touristed prefectures, have token and zero Shinkansen service, which further shrinks the pool of leisure demand the network can actually vie for.
Another difference between France and Japan which you frequently mention is that Japan through-runs while France has separate nonstop trains to Lyon and Marseille.
Perhaps one lesson from this data is that through-running is bad for convenience and ridership? That is to say, the inconvenience of long intermediate stops outweighs the gains in frequency. Perhaps, since we are talking about relatively long trips, people are willing to schedule around infrequent trip times if it means a faster and more comfortable ride.
Alon cites Berlin-Munich – a through-running service with more egregious slow zones than anything in Japan – as another route that overperforms the model, which makes me think the other factors are more significant.
In France the absence of through running is inconvenient for people who are not travelling to/from Paris. And for SNCF those people do not matter. SNCF is all about moving people to to/from the Capital. The moment you try to use a train to get from to the system usually fails completely.
But Paris is so big that most HSR demand is to/from the capital. By the model, Paris/Marseille has higher demand than Lyon/Marseille even though Lyon is much closer.
Yes SNCF should improve its non-capital service, but they’re probably right that this won’t be highly profitable.
The non-Paris trips, and especially those that avoid the LGV Sud Est, only need to cover the marginal cost for running the train.
That’s the cost of the electricity to run the train, plus the driver and guard and the marginal maintenance costs for the track and train. If you have to buy extra trains to run the service then I guess you would also need to include the capital cost of perhaps a 4-8 car TGV train set.
You probably really don’t need a very high loading to cover those costs. Probably 50-100 passengers on the train is enough.
Certainly in Britain they run a pretty frequent evening service on a lot of lines, and aside from the last train which is often pretty busy in my experience the other evening trains typically are very lightly loaded.
The model already takes trip times into account, so the mostly nonstop services in France are already getting a bonus for speed.
Given even Inverness-Aberdeen has a two hourly train all day and pretty much every other major UK city pair has hourly service between them maybe the French just need to run an hourly express train from Lyon to Marseille?
Indeed. Two 2m metro areas sub 2-hours apart is a viable point-to-point market. At such journey times frequency, clockface scheduling and ticketing flexibility make a lot of difference to the overall generalised journey time. It’s these reasons and not the point-to-point service nature that explain the low ridership / mode share.
In fact most Lyon – Marseille services originate from non-Paris locations in the north like Lille and Strasbourg, so it’s not as though they are exclusively reliant on point-to-point demand.
Leaving Marseille St Charles are 12 daily TGV serving Lyon Part-Dieu, and 5 towards the LSE station east of Lyon. The only 6 hour gap is at night.
Wether competition improves service depends a lot on how good the incumbent is. In Italy competition forced Trenitalia to get its act together and do things like actually making it possible to book trains easily. RENFE is notoriously bad at actually selling tickets, and hopefully will be forced to improve there.
Shinkansen is considered as an extremely posh way for leisure travel or visiting your loved ones (except holiday homecoming). While Japan as a whole is no less wealthy than France, an average Japanese worker earns quite a bit less income than an average French. With less money (and with less vacation days) than the French, the Japanese don’t leisure travel as much as the French. The statistics posted by John D. above also support this.
Therefore, the main use of Shinkansen is for business trips. Indeed, according to a government survey[1], about 45% of all intercity rail trips are business trips. Combined with commuters, about half of all intercity rail trips is work-related.
[1]: The results are from https://www.mlit.go.jp/common/001193659.pdf and the introduction to this survey is found in https://www.mlit.go.jp/tetudo/tetudo_tk9_000006.html
If I would have to guess for an explanation, it would be a difference in disposable income which is lower in Japan, and thus that the price bites a bit more in Japan. In particular, Japan often has more realistic (and much cheaper) secondary travel options. Though this is probably by design, as many parts of the Japanese system are rather close to capacity (or at least would be if pricing was lower and allowed more commuting). Basically the Japanese are fine with pusing some ridership to cheaper options.
Not an expert here, just sending a hypothesis from Spain.
Could it be because of the TGV connections to the airports? An international traveller going to a secondary french metropolis can rely easily on the trains from Paris Charles de Gaulle or Lyon Saint Exupéry, while one landing in Madrid Barajas or Barcelona El Prat is more likely to take a connecting flight to go to, say, Alicante or Zaragoza. That’s also part of what allows french goverment to ban “too short” internal flights.
I cannot think many more reasons why the plane still keeps a 25% of market share of the trips between Barcelona and Madrid, when train is a bit shorter and much much more convenient: https://www.elnacional.cat/es/sociedad/tren-supera-avion-12-puntos-cuota-trayecto-barcelona-madrid_925899_102.html (BTW, according to this article, Madrid – Barcelona in rail already reached 5,6 million passengers in 2021)
I’m also very skeptical about “liberalizations” that turn public monopolies into private almost-monopolies. Having (possible) competition doesn’t stop the awful website of Renfe from being a recurrent joke in Spain: https://verne.elpais.com/verne/2020/01/30/articulo/1580392755_303687.html
But in this case, maybe liberalization can open the door to a different way to do business that does actually provide a benefit for the user? Specifically, I’m imagining Vueling or Air Europa operating trains and selling combined tickets plane+train through Madrid or Barcelona airports. That would require Adif to build a connection between the conventional line and the high speed line (like the ones that already exist on a few other places), but with Madrid being the 5th busiest airport in Europe, and Barcelona the 7th, probably it’s worth it?
(Rescued from spamfilter – I only saw this now, after replying to the shorter version of the comment below with fewer links, sorry.)
Renfe runs too few trains and they sell out days in advance. To absorb the remaining air demand they need to physically run more trains. Hopefully this is something Iryo and SNCF will address.
There might be a historical factors, the «artère impériale» was a prestigious line way before high-speed trains.
Another factor in favour of trains in France is strike atomicity: in case of strike, you are stuck at your departure or destination city. With the planes, you could get stuck at the airport…
Okay, but the Tokaido was prestigious even before there were any trains.
Fair point.
My experience in the academic world is that in France, even broke students would take the TGV, while broke Japanese students would find alternatives to the Shinkansen. In Japan, it is possible (if extremely tedious) to plan a route using the locals trains, and that route would be significantly cheaper. In France this never was an option, local trains are too sparse and too unreliable to make this a viable option. Gas prices are also higher in France compared to Japan, and the cars are heavier – many students in the university I did my postdoc in in Japan had keijidōshas.
Wait – my Japanese rail ridership numbers are for all JR trains, not just Shinkansen.
Then I don’t know. 😀
The cheap option in Japan isn’t the local trains, at least between the major cities — it’s the overnight buses, which can take you from Tokyo to most other regions in ~12 hours overnight when all rail services are shut down.
Depending on the day of travel and the time, how far ahead you reserve, whether or not Venus is in retrograde etc. the cheapest way to get from New York to Philadelphia can be Amtrak. The cheapest way is usually to make it all NJTransit – it’s possible to do it by rail – and the most expensive is to be almost as masochistic and use SEPTA to get between Trenton and Philadelphia. It depends on how frugal you are. If you want to be really frugal, use PATH between Manhattan and Newark and transfer to NJTransit there.
Al, TGV ridership predictions from Alon Levy. He doesn’t try to factor in the greater number of midsize cities in Germany vs France. Bob Campbell
*They
France actually has a longer tail of metro areas in the 200,000-500,000 region, whereas German metro areas tend to be a bit bigger, around 1,000,000. And also this isn’t relevant to the two German lines for which I have data – I did take into account everything on these lines.
Wouldn’t the ICE have more through passengers from neighbouring countries than the TGV does?
The French TGV network connected with Switzerland in 1981 in Geneva. These days, there are connections to Lausanne, Basel and Zürich and there were improvements to shorten travel times between the countries (ligne des Carpattes). Meanwhile the only ICE connection is to Zürich (via Basel), sometimes it goes up to Interlaken. So I suspect the TGV network drains more from Switzerland than the ICE one does.
At least the line from Munich is electrified now.
It’s true that France-Switzerland with the TGV Lyria is good.
Paris-Milan ironically is good for London etc with the 12:43 service but the first train of the day runs much too early in the morning. It’s also (understandably) slow through the alps.
Paris-Barcelona only has one train a day that’s timed for ongoing connections from Barcelona with the later train missing the last service of the day to Madrid. Additionally the first train is too early for London. Perpignan is well connected to Northern Europe with some good late arriving TGVs, but Perpignan-Barcelona has arguably worse rail service than Inverness-Thurso or Glasgow-Fort William with the first reasonable train being the afternoon TGV from Paris.
Paris-London and Paris-Brussels-Amsterdam are well served but expensive.
For Germany the timings in general connect well with the Eurostar in Brussels and have strong connections with Amsterdam as far as I can see. Less sure about the other connections from Germany to be fair.
Paris – Milan is weird because it does not follow the much straighter and shorter path via Vallorbe and the Simplon that the TEE took.
There are a few more TGV than ICE connections to Switzerland. The ICE always goes via Basel. The TGV goes mostly either via Basel or Geneva. Half the TGVs and ICEs end at the border station. The other half go on to a few Swiss cities in both cases (Lausanne, Zurich vs. Chur, Zurich, Interlaken). ICEs go to a number of German cities (Hamburg, Kiel, Berlin, Dortmund / Cologne, Frankfurt, Stuttgart) and Amsterdam, calling at many more on the way. The TGVs all go to Paris (except a summer train to Marseille once a day), with Dijon as the only significant French stop.
There are more than twice as many Germans in Switzerland than French but France gets a lot more tourism. Hard to guess which service gets more passengers.
> At least the line from Munich is electrified now.
That line is an unmitigated disaster. 2/3 of the trains are late, often by 1/2h or more, making it slower and certainly less reliable than the old diesel service. Who could have known that a single-track line with a mix of long-distance and local trains is unmanageable, even when electrified…
A parsimonious model is usually better than a complex one when data is sparse, but perhaps one is reading a bit too much into this particular model construction. Sounds like there are a few, albeit isolated but dare I say represented examples that suggest perhaps this model may systematically underestimate European demand.
If posters’ accounts of fare levels, load factors and lower prevalence of leisure travel are credible, then observed Japanese ridership numbers probably represent constrained demand (capacity constrained as well as having excess demand ‘priced off’); and indeed if the model is trained on Tokyo data it’s possible the modelling error / bias may be larger for European sized flows and distances which are at the lower end of the Japanese data distribution.
In the absence of transparent data we can use other proxies to gauge potential strengths of markets. Looking at ‘best in class’ European examples, Swiss and Dutch inter-city networks and Italian HSR, annual seats transported would probably outnumber predicted annual ridership. Based on seats, it’d be interesting to see how Paris – Lyon compare with Milan – Rome when this model is applied.
On the liberalisation point – it’s worth noting Europe really has only two networks that can be considered truly excellent – the Swiss and the Dutch. Switzerland has a plethora of private railways. Only the Dutch can boast ‘the state know best’. The other state operated networks are various shades of conservative and antiquated, SNCF and Renfe included. Madrid – Barcelona is clearly artificially constrained – tickets sell out days in advance – a bit of disruption from Iryo is well overdue. It would be nice if national governments could all specify Passenger Service Obligation (PSO) contracts with nationally coordinated clock-face timetables, or state-owned operators offered that service in the first place, the disappointing truth is few governments / national operators currently have that competence. In the absence of that competence open access competition is the next best thing.
Your model has a floor of 500km or ~2h15m – probably wise when fitted to your specific Japanese dataset to avoid spurious results at the lower distance / travel time end. In a European context though a 2-hour journey becoming a 1-hour journey would clearly be expected to make a big difference. Incidentally if I apply your model to London – Birmingham and London – Manchester with the floor on, I get very close to existing (pre-Covid) ridership numbers. If I remove the floor and apply HS2 journey times I get 15m annual journeys on each flow, which sounds slightly ridiculous but actually plausible for the long term. HS2’s business case is pretty much predicated on long-term quadrupling of capacity (HS2 services not carrying inter-mediate demand which is half the existing market, and near doubling of train lengths from 218/265m to 400m in the long run, and maintaining the 3tph frequency).
A couple points.
1. Follow the link in the lede graf – I have a link to prefecture-to-prefecture travel volumes there. Taiwan massively overperforms the model, by a factor of about three; I think the issue is that the model assumes that if national population grows by a factor of x, ridership grows by a factor of x^1.6 instead of by a factor of x, and when that is corrected, Taiwan normalizes. (You can’t use that trick in Europe because people travel intra-Europe all the time.) South Korea looks like it overperforms Japan but not by as much; Seoul is enormous.
2. Post-liberalization, the Milan-Rome trip volume was 3.6 million in 2018 (source), just between those cities, without intermediate points like Bologna and Florence. The problem with comparing it with the model is that there’s wide variation in estimates of metro area sizes. Eurostat has Milan at 4.3 million and Rome at 3.7 million, so the prediction would be 1.5*(4.3*3.7)^0.8/9 = 1.5, but then I’ve seen estimates of metro Milan that go up to 7 million – Milan is like Boston and Frankfurt in that it has satellite metro areas and it matters whether you include them, and I suppose we should since they’d all be using Milano Centrale to get to Rome. If it’s 7 million for Milan and a high-end 4.2 million for Rome, the model thinks it’s 2.5 million and then Milan-Rome overperforms to the exact same extent as Paris-Lyon.
3. Switzerland has private operators but in the traditional way, with their own tracks, like BLS. It doesn’t have competition in the style of Italo vs. FS, where they compete on price and don’t cross-honor tickets; it instead has outside operators that can run through but still have to accept the Swiss tickets.
4. The 500 km floor is weird, I admit. But for what it’s worth, Taiwanese data by station (I don’t have O&D pairings) bears it out – Kaohsiung’s ridership is not much weaker than Taichung’s, even though it’s significantly farther from Taipei. As I said in the post, I suspect what is happening is that at an average speed of 200+ km/h, the effect of higher trip times on ridership countermands the effect of high-speed rail competing with cars better when trains take two hours and cars take five than when trains take 40 minutes and cars take an hour and a half. But then if the average speed is lower, as on Lyon-Toulon (in addition to bad frequency) or Florence-Bologna before the new line opened or probably Berlin-Leipzig and Berlin-Hamburg, then it does straight out reduce ridership even within the floor. Ideally I’d want to replace the floor with a flat term to be added to t (or to d) but I don’t have enough city-to-city data outside Japan for this, unfortunately.
5. I fully expect HS2 to overperform the model if the fares resemble domestic TGV or ICE fares rather than Eurostar fares and if there is no security theater (which I suspect is one of Spain’s problems). The 500 km floor in that case should imply similar flows from Birmingham and Manchester, which I don’t think is a stupid prediction – they’re both so close to London by Paris-Marseille standards that the access and egress times would be large compared with the in-vehicle travel time, especially out of Birmingham.
Is it possible that Taiwan’s numbers are a result of high motorcycle ownership relative to other developed countries? Perhaps, one who owns a car will tend to use it for long distance trips, but one who owns a motorcycle will prefer a train for long trips.
Maybe? But the modal split for the train is a majority of all trips (not just air/rail trips) on some city pairs in France and Japan – I believe it’s about 50% in Japan at 500 km range in general, so it should be more than this for Tokyo-Osaka.
Your 500km floor is hugely problematic because it doesn’t make sense a 300 km journey gets exactly the same ridership as a 500km one and because the actual driver on ridership is actually *time* not distance, because people evaluate their journeys by how long it will take (along with price and convenience-what time of day is the trip, how close to their destination, etc.) not how far it is. Since many of these networks have speed variability (Paris-Strasbourg is built and run at different speed standards that Paris-Lyon, as would future services like Chuo v. Tokaido, DC-NY v. DC-Pitt., anything across the Alps) putting a flat floor on distance means capping trips at different times. Many of the criticisms of you choices based on this model can be traced to this floor (i.e. Memphis gets HSR because of all the riders taking REALLY long trips to Chicago and Orlando that people normally fly, but Oklahoma City doesn’t because it is “too close” to Dallas so it’s ridership looks meh despite being a very convenient distance for HSR).
I can’t speak to details of how to develop different formulas, but I can suggest:
-Adding a Max as well as a min term so you are not capturing phantom long distance ridership that actually flys
-Using a floor/ceiling based on time not distance. This requires a bit more work to look up/compute the time for each city pair, but I believe you do this anyway (for networks intended to takt like your Germany plan you don’t even need to compute it, you define it by assigning city pair travel time to meet the takt).
-Using a multi part formula with different constants/exponents (short trips, optimum trips, long trips). Although most mode share percentage graphs I’ve seen tend to have little ridership beyond 5hr, a pretty linear relationship between 1-5 hr, and few gains below 1 hr, so perhaps a Min/Max is all that is needed.
-At a minimum set a different floor of perhaps 250-350km (a 350kph line with the express having a time made good of .7 top speed covers 250km in an hour, a 250kph line does 350km in two hours, somewhere in the 1-2 hr range is where HSR mode share stops rising).
The 500km number bandied about comes from 30 years ago when you could arrive at the airport 15-20 minutes before departure and still make your flight. Versus a three hour train trip. It’s closer to 750 and four, four and half hours today.
It’s probably not worth it to build 300-ish miles or 450 km of high speed tracks between Saint Louis and Memphis for the 350 people an hour, of demand, that Memphis could generate if it was on the Tokiado Shinkansen. It’s not. Memphis is sometime perhaps far far in the future because in the Northeast or Midwest in 300 miles you trip over metro areas of a million people or more every 100-ish. And the origin or destination is New York or Chicago, not Memphis.
The advantage of a longer train ride vs flying is that you can work on the train whereas you cannot really work while flying. Then there’s the environmental benefits – a sizeable chunk of my lifetime carbon emissions are from flying. Plus a whole chunk of the population doesn’t enjoy flying to a greater or lesser extent.
The problem with longer trips in general is that often the fares are extremely high and/or the journey has (multiple) crossings between stations in a big city and/or the schedule requires leaving very early in the morning.
There is also potentially a data issue that the data available is very limited as each operator appears to have no idea that people have bought tickets to other destinations – certainly Eurostar is like that.
I know it’s problematic, but that’s what the numbers out of Japan look like – Shinkansen ridership at 300 km and at 500 km looks the same, keeping city sizes constant. This looks like competition with cars canceling out the effect of longer trip times on the overall market.
And don’t forget that Paris-Nice actually has pretty healthy TGV ridership even well beyond five hours, even with a healthy air market buoyed by high leisure travel volumes (unlike Grenoble or Toulon or Saint-Etienne).
> 4. The 500 km floor is weird, I admit. But for what it’s worth, Taiwanese data by station (I don’t have O&D pairings) bears it out – Kaohsiung’s ridership is not much weaker than Taichung’s, even though it’s significantly farther from Taipei. As I said in the post, I suspect what is happening is that at an average speed of 200+ km/h, the effect of higher trip times on ridership countermands the effect of high-speed rail competing with cars better when trains take two hours and cars take five than when trains take 40 minutes and cars take an hour and a half.
Taichung HSR is outside of the city center and the medium speed train take a bit longer, but are cheaper and bring you right into the city center.
People prefer faster travel time, and more are willing to adjust their schedules to get that. I know that transit folks prefer consistent clockface schedules of local trains. However, faster schedules attract more riders from a greater distance to the station than a more frequent schedule would.
France (and Spain) have some of the shortest travel times, so taking a train there is a no-brainer when compared to Germany. If a commuter/regional train travel time is increased by 10 mins at 1tph, you can probably capture drivers from 5 miles further away than you could by going slower at 2tph.
Germany overperforms pre-liberalization Spain relative to trip times, and I think they’re on a par post-liberalization, with Germany taking the malus to average speed that isn’t visible to the model below 2 hours.
[citation needed]
PS RENFE operations are a fucking disaster. Random service gaps, infrequent, crazy ticketing … about as fun as US airlines, but without even the profit motive. Just pointless useless hostility. Even a tourist, one who admires tremendously overbuilt new infrastructure, can see that, right? Right? But right, indeed, some “people prefer” abuse, and come to so love Big Brother that they will defend him against all apostates. I mean, look at “adirondacker12800” (or don’t, please don’t.) Punish me more, it’s what I’ve grown to expect! You have no right to expect anything but the punishment that I endure! Stockholm Syndrome isn’t just pop psych.
Agreed.
People do fly a lot with Ryanair who usually fly from less convenient airports but is cheaper. In fact it’s the worlds most popular scheduled airline.
It’s no secret that they have a different doctrine.
Mainlines are flying transit, they provide an anywhere-to-anywhere network, with decent headways. They sell schedule flexibility. However, they have relatively limited ability to not sell this feature to people who can produce their own more cheaply, or rather, to be precise, they have limited ability to avoid incurring the costs of producing it even if in the end it gets thrown away in the ticketing system.
Ryanair and the others like it are a flying multitude-of-carpools. They deliberately buy schedule flexibility from the people who can produce it cheaply, so that they can produce transportation without this feature, allowing them to sell tickets cheaper. Obviously this makes them useless for people who want to buy schedule flexibility. Why pretend that this is a problem?
Of course, this situation is only tenable because the two sorts of airlines need to share very little infrastructure (basically, only ATC). There is a moderate surplus of existing airports suitable for low-costs, and there is no linear infrastructure in the air. On rail, having to share infrastructure makes such a situation much less possible; to a first approximation, the existence of a flexibility-producing service on a line makes it such that everything else has to pay the costs of being able to produce flexibility, even if they don’t. (At an international hub airport, landing fees are high even if you don’t sell connecting tickets.) If a line’s traffic is dominated by an hourly takt, then any “surplus” train needs to take a trainpath that recurs every hour in the timetable whether there is a train using it or not.
People prefer the shortest generalised travel time. In-vehicle time and wait penalty are both part of the generalised travel time. If a market sustains a half-hourly point-to-point service level, then the increase in in-vehicle time through calling at intermediate city centre stations (usually at least a 15-minute penalty relative to shooting past at 300km/h) outweighs the frequency gain beyond 2tph.
How is it a 15-minute penalty? The TGVs aren’t great trains for this, sure, but Velaro Novo specs get you a stop penalty of around 3 minutes plus the dwell time, and the dwell time can be held down to 2 minutes (at Frankfurt and such it’s 4 even with changing direction, and also most cities are smaller than Frankfurt).
Paris-Chambéry is 2h50 on the early morning Milan TGV that stays on the fast line the whole way and 3h12 on the Trenitalia service that goes via Lyon-Part Dieu.
That’s in excess to the stopping penalty as both trains have one stop because the TGV stops at Mâcon-Loché TGV instead.
In Britain as we’ve discussed before the time penalty for going via an intermediate city is 45-60 minutes so is even worse.
Yes, in the specific case of SNCF having built its infrastructure to bypass Lyon rather than stopping through it.
I guess in Tokyo they avoided bypass tracks – but the inner Tokyo sections of the northern Shinkansen line were very expensive.
Other than Tokyo in most other cases either bypass tracks exist or the legacy approaches are used for intermediate stops.
The real solution for the French is to just run a half hourly or hourly consistent all-day service between some of their intermediate city pairs as we would in Britain and to just call it a day.
Obviously in some cases an 8 car train might be excessive – and either you split the train into two and run two 4 car trains, or you run an 8 car train and live with it. The half hourly Didcot-Maidenhead-London Paddington slow services are an (IMO excessive) 8 car train, but that’s what they have so that’s what they run.
You need to take alignment into account too. Chances are you are going to have a 300km/h straight alignment through a major city, and using the legacy alignment through Lyon Part-Dieu is as good as you’ll ever get. Then there’s the typical European dwell to consider. A 15-minute penalty relative to going straight through St Exupery is probably an underestimate.
The typical European dwell is not 15 minutes or even close to it – again, Frankfurt is 4 and the limiting factor there is the direction reversal. A lot of this boils down to SNCF’s airline imitation and hostility to seat turnover.
I was looking at some of the UK rail industry costs in order to make a response to the consultation about shutting ticket offices. And basically by comparing 2019/20 to 2020/21 it looks like in reality basically all of the railways costs are fixed.
Crudely that makes things extremely simple. It means that in order to justify a peripheral service where there is clearly spare capacity such as Lyon-Marseille or Lyon-Montpellier you basically only need to cover the guard/driver and train maintenance costs – and if you have to buy additional trains to run the service you also need to consider the cost of the train.
So I’m not really sure how much all of this stuff costs. However with an average fare of €0.10/km I really think if you can get an average of maybe even 50 passengers per train that would be enough. Or maybe even less if you can just use your existing trains more efficiently.
I didn’t say a dwell would be 15 minutes. I said a through city centre stop with a stopping and alignment penalty would be around 15 minutes relative to bombing through via St Exupery at 300km/h.
More relevantly… Lille-Europe was built for through-service, so the path through it is pretty fast. Part-Dieu, yeah, it’s not a good through-station because it was never designed as such once SNCF built the LGV Rhône-Alpes, but both Lille and Strasbourg are fine for this (no idea about Bordeaux), and there are plans to turn Marseille into a viable through-station but I can’t tell how serioues they are since the LGV PACA is paused.
Lille and Strasbourg are both smaller cities which helps.
Additionally Lille got the opportunity to have a short direct high speed line to the three most important cities in Europe if it accepted the line going straight through the middle of it. And given how nice Lille city centre is it has done well out of that.
And Strasbourg presumably wants to maintain the ridiculous status quo where the EU parliament decamps there every 6 weeks – which means strong international connections are pretty important for it – as is it not pissing off the French or German establishments by being inflexible.
Lille is detour writ large. Whether from the perspective of London – Paris or London Brussels the current routing through Lille-Europe categorically does not represent the straightest route possible. Between Bois-Grenier and Camphin-en-Carembault the current D-shaped route is 36km, and at (my guestimate) an average of 220km/h non-stop takes 10 minutes. A direct cut-off would be 17km, at a (conservative) 270km/h would only take 4 minutes. This 6-minute difference is what I mean by the alignment penalty. Notice how Paris – Brussels conveniently bypasses Lille – it’s the only way to keep the line geometry sensible.
A through alignment is almost always more indirect and slower than a bypass alignment. The economic and political constraints are such that you have to respect existing street patterns and railway infrastructure orientations for building new stations, or in many instances the only way is to plug into the existing infrastructure. Where new stations are built, routes through Antwerp and Bologna are both slower (way below 300km/h and less direct than a theoretical bypass (an Antwerp bypass would be Duffel -Noorderkempen, and Bologna bypass would skirt around the southwest of the city with 7000m curve radii). The European norm, where there are no bypasses, is in fact simply plugging into the existing network with minimal or no upgrade. Just look at Liege, Aachen, Augsburg and Ingolstadt – in all cases no better than Lyon. Oh and let’s not forget Bad Hersfeld.
The problem with going through one place and stopping there is that every other place will demand you going through them and stopping there. So that 15-minute penalty (combined alignment penalty and stopping penalty) is never one off – you’ll have several of them. The only way to bring Berlin – Munich down to 3 hours (a market similar in unconstrained volume and distance to Rome – Milan, so should target a similar journey time) is to ruthlessly bypass a lot more places. Theoretically London – Amsterdam is a strong point-to-point market and the current journey time of 4 hours (even if all the Brexit nonsense is done away with) is on the wrong side of being unquestionably competitive against air. That journey time could be brought down if it didn’t have to suffer the alignment and stopping penalties at Lille, Brussels and Antwerp (though admittedly, realistically only Antwerp has a slim chance of getting a bypass).
I’m getting around 2:30 with stops at Erfurt (but not Halle) and Nuremberg (but not Ingolstadt). It requires a 300 km/h Berlin-Halle alignment, which DB is ignoring even though it would be cheap for the same reason the current line is already 160 km/h, and then closing the gaps in Bavaria (which is in the long-range plan); once that happens, ICEs on the route wouldn’t be sharing tracks with anything other than a 300 km/h ICE between the city limits of Berlin and Munich, which would also allow DB to slash the 25% schedule padding to something more reasonable.
The 3-hour plan was already compromised to hell, by planners who think in terms of thresholds instead of the benefits of saving each incremental minute.
Yeah, Berlin
Yeah, Berlin – Halle needs a proper 300km/h line. Realistically a Berlin – Halle – Erfurt – Nuremberg – Munich stopping pattern ought to be acceptable and deliver a (sub-) 3-hour journey time. Those places are all small enough to not produce excessive churn (and with enough trains calling) and don’t have horrendous alignment penalties. The Ingolstadts and Bambergs should be on regional expresses that connect into ICEs.
It’s just terrible that people in the real world don’t conform to the formula.
Not an expert here, but could it be that TGV gets more ridership because it connects directly with the airport terminals in the Paris-Charles de Gaulle and Lyon – Saint-Exupéry airports? Therefore, for international trips, it can distribute the passengers to the rest of smaller cities without a connecting flight.
I’m also very skeptical about “liberalizations” that turn public monopolies into private almost-monopolies. Having competition doesn’t stop the awful website of Renfe from being a recurrent joke in Spain: https://verne.elpais.com/verne/2020/01/30/articulo/1580392755_303687.html
The Omnil link breaks down ridership by train station within Ile-de-France, and CDG is barely a rounding error.
Everybody thinks airports are really really popular places. They aren’t. They are looking at people in the check-in lines who will be the same people in the security lines who will also be the same people at the gate. Looking at the annual passenger count is deceptive because it counts the passenger who get off a plane at gate 1, walks across the concourse, to gate 2 and gets on plane. Who gets counted twice. And gets to see many many passengers go by if they have a long layover. And everybody thinks sending the train to the airport is a great idea. For other people.
…..railfans everywhere think spending billions and billions to tunnel across Philadelphia to get to PHL is a fabullousssss idea. So that Washintonians can use the train to go past BWI really fast or New Yorkers can go past EWR really fast.
Well, checking the Omnil link mentioned before (which I should have opened before), I get 237k TGV daily passengers in the Paris stations (taking the page 3 graph, with the total users and percentages per type).
Checking in wikipedia ( https://en.wikipedia.org/wiki/List_of_the_busiest_airports_in_Europe ) I get in Charles de Gaulle airport 57,474,033 / 365 = 157k daily plane passengers, and in Orly 29,187,269 / 365 = 80k, so TGV and plane ridership seem to be comparable figures (even with the asterisk of the twice counted passangers).
OFC when you compare that with a metro system or a commuter train, that’s a completely different story. Spain did a costly mistake letting conventional train rot while building the HSR.
Airport stations are popular when done RIGHT – meaning minimal transfers for flying passengers. The CDG station requires a few escalators to get to the platform, where you can catch a direct train to Bordeaux, Brussels, Strasbourg, Lyon, Marseille, Nice, etc…
The key is that the station is part of the airport and its amenities, rather than something like Newark Airport, where you lose a significant number of passengers due to the vicious cycle of poor amenities, poor service, and poor investment. Extending the AirTrain to Newark Penn Station would provide airport customers with a better experience through better amenities, better and more frequent service, access to PATH, and light rail. It would save millions in costs of maintaining an extra unnecessary station, and save 5 mins for trains not having to stop at the airport, among many other benefits.
Sadly doing a proper underground station under the airport terminals is not something America does well, preferring endless connections (See SFO, LAX, ORD, STL, etc..)
There’s no place to put Airtrain in downtown Newark. People who are not in Manhattan are allowed to use the airport too. Explain why Jamaica underwhelms.
JFK had 31 million passengers in 2019. Round that up to 36.5, that’s 100,000 a day. 50,000 arriving and 50,000 departing. Really busy day, around Thanksgiving, 100,000 arriving and 100,000 departing, you manage to get 25 percent to use mass transit, they aren’t all going to be using Penn Station and it works out to a train an hour. Airtrain is good enough. No, you can’t run a shorter train every 15 minutes because you’d be swapping a half full six car train for a 12 car train that is standing room only, during rush hour. The 100,000 a day doesn’t account for the people who are changing planes at JFK, who arrive at the airport on an airplane and depart it one one. They don’t care if the bus sucks, the train is almost as bad, where the taxi stand is, how expensive car rentals are or the quality of airport hotels. Airtrain is good enough.
If building the AirTrain station underground or above the platforms is not possible – as is typically done, then given the low density around the station, one could be built above one of the many parking lots.
There’s probably some good ROI on rebuilding the station to untangle the PATH track threading.
Multiple stations (EWR + Newark Downtown) kill ridership via infrequent service, a smaller number of destinations, transfer requirements, or limited amenities.
Politics aside, I don’t understand why you’d fight for the EWR station when extending AirTrain to Newark Downtown elevates the passenger experience in many ways.
My wife always insists on a taxi to JFK because dragging suitcases is bad enough on a subway, but then figuring out schedules and transferring is just too much. What’s the value of spending billions on transit that requires transfers when you immediately ignore most of your market?
It’s too bad you pick inconvenient hotels. Changing from the train at Newark Airport isn’t that much different than what you fantasize it could be at Penn Station Newark. Except that you’d be doing it over concourses you aren’t going to use.
It’s too bad the swooping tracks offend your aesthetics. Penn Station Newark works the way it is, very well and changing anything would screw it up. Eastbound PATH trains are a cross platform transfer from eastbound trains to Penn Station New York, on Track 1 or Track 2. Westbound the platform is above westbound suburban and Amtrak trains on Track 3, 4 and 5 and it’s a walk down a ramp or the stairs to them. Move the PATH tracks and that doesn’t work anymore.
The link at the top of the article says CDG has 8,700 daily riders, which is 2.7 million a year, maybe a bit more since I think this doesn’t peak. This is not a lot compared with the airport’s traffic.
@alon.
CDG Station TGV numbers for
* 2021 are 6.3 million
* 2019, they were 15 million
Source: https://ressources.data.sncf.com/explore/dataset/frequentation-gares/table/?q=87271494&disjunctive.nom_gare&disjunctive.code_postal&sort=nom_gare
The CDG Airport Passengers numbers for
* 2021 were 26.2 million
* 2019, they were 108.
(Source: Google search)
While we cannot confirm whether TGV passengers were also boarding a flight, the numbers are not as insignificant as you make them out to be.
@adiron – Ah, the swooping PATH tracks provide a more effortless transfer convenience based on how you describe that.
That’s not just the TGV station but also the RER one, so this includes O&D travel to the airport from Paris.
I’d be curious how Korean HSR ridership performs on this model. It’s a composite of both French network design and Japanese operating principles, and the external factors are different from either:
1. Through-running of almost every city, including Seoul (to Haengsin)
2. Lots of legacy rail operation, both in the through-running and in tails (though the latter often on newly-upgraded rail, and sometimes on new alignments, e.g. Pohang)
3. Many smaller-ish cities have airports, and though service at these tends to be international because the country is small, there are still flights from e.g. Ulsan/Gwangju; just the same, getting KTX-branded services is always politically popular
4. Seoul is more dominant than either Paris or Tokyo
5. The biggest HSR-connected leisure travel connection is also the second largest city (Busan)
6. Both frequency, stopping patterns, and fares are more Japanese, even though the network design has much more in common than the French
I don’t have ridership figures on hand, but the passenger-kms vs. the number of rail passengers from OECD data says that most trips are probably long-distance to Seoul.
KTX overperforms the Shinkansen model, but it depends on when and which line. Before SRT opened, KTX ridership slightly overperformed. Since then, it’s significantly overperformed, though not nearly as much as Taiwan (it’s around 1.5-2x, not 3x). Reasons may include,
1. Lower fares – I think at this point fares are still lower than in Japan even as incomes have mostly converged.
2. SRT providing better coverage within Seoul.
3. Smaller country size, as in Taiwan but to a smaller extent.
Seems odd that at least in reference to your HSR traditions post, the model doesn’t quite match. If Japanese style metro-like operations (organization) are most impactful when it comes to eventually guiding line design (concrete), and naturally a metro-like system should be the most materially/operatively efficient style of rail in reference to ridership (generating the most ridership/km rail), why does a model trained on the Japanese system suggest that other national HSR traditions induce more people to ride? Of course, there are exogenous factors, but it seems an unintuitive result.
Interestingly, if we take the ‘big picture’ of annual trips per capita:
– Shinkansen: 370,451,000 trips / 126.6 million = 292,620 trips per 100,000
– TGV: 110,000,000 trips / 67.4 million = 163,200 trips per 100,000
– ICE: 99,000,000 trips / 83.1 million = 119,130 trips per 100,000
European figures are approximate but the margins are clear enough.
Which runs up against the model saying Shinkansen ridership ought to be higher for the given population of each city-pair. Perhaps this means Shinkansen use is more distributed across the country rather than on specific city-pairs, which is plausibly linked to metro-style service?
I don’t have European passenger-km/route-km figures off-hand, but I suspect they’re also lower than the Shinkansen’s (99.332 billion / 2765 km = 35.9 million per km), which is likely another byproduct of intensive metro-like corridors.
Britain had 120 million “long distance” train passengers in the last financial year, and 140 million pre-Covid.
Pretty surprising we are beating the French with our high fares and worse service – https://dataportal.orr.gov.uk/media/2207/passenger-rail-usage-jan-mar-2023.pdf
DB Fernverkehr has a lot more trips than that. Many are not on ICEs, but the distinction between ICs and ICEs here is not sharp; currently, every ICE trip between major cities at longer distance than Cologne-Frankfurt can expect to spend most of its time on a classical line at a speed no higher than 200 km/h (230 for Berlin-Hamburg, uniquely).
Based on trips and passenger kilometers, the average Shinkansen trip is just under 270km.
TGV with a bit under 62 billion passenger kilometers over 110 million trips has an average trip length of over 560km, or over double that of Shinkansen.
Obviously Tokyo-Osaka and Tokyo-Nagoya are massively popular city pairs, but both of those are significantly longer trips than the average Shinkansen trip, and a lot of the ridership of Shinkansen from shorter trips enabled by the more metro like operations. While it lags on longer trips due to the higher fares.
Yeah, I think it’s mostly the fares – nearly all of the Paris-Lyon overperformance boils down to that, and the highest-end estimates I’ve seen in the literature for the fare elasticity even reduce the entirety of the overperformance to that.
No comments on your Conrail (ConRail on Day One in 1976)? Huh. The Consolidated Rail Corp was created to inherit the carcass of the failed 1968 Pennsylvania RR/NY Central RR merger that gave us Penn Central. In 1969, the still extant Interstate Commerce Commission required Penn Central to take on the bankrupt NY, New Haven, & Hartford. The whole thing lasted another 10 minutes and declared bankruptcy in 1970. Amtrak was attempt #1 to bail them out by taking over passenger service, largely from PC but also any other RR what wanted to “join” Amtrak. Didn’t work and in 1973 the court overseeing PC’s bankruptcy came to the conclusion that there was no way the company could ever be solvent. That lead to the feds creating Conrail to sweep up the remains of the PC and a bunch of other RRs (Erie-Lackawanna [a 1960 merger], Lehigh Valley, The Reading Co, Lehigh & Hudson River Ry, Central RR of NJ, & Pennsylvania-Reading Seashore Lines [joint venture of PRR & Reading]). Most of those other roads were built to haul coal to NY and Philadelphia.
As part of the Conrail legislation, the Northeast Corridor was dumped on Amtrak. That made sense, but recall that up until then, Amtrak owned *no* track and though RRs had “donated” their equipment when they “joined” Amtrak, Amtrak contracted out operation of most trains back to their original railroads (but not on the NEC?? – not sure). Amtrak didn’t get Conrail’s commute services and Conrail hung on to those until states were forced to take them over in 1983.
The feds sold off Conrail in 1986 and on paper it was a big success. What was left of Conrail was successful but there sure were a lot fewer places served and a lot fewer route miles. Successful enough that 13 years later, a bidding war lead to the company being split between NS (Norfolk Southern, not the other one) and CSX, with the former getting, roughly, what was left of the ex-PRR network and the latter getting most of what was left of the ex-NYC.
So … not exactly “pre-nationalization” – Amtrak, still a GSE, owns or otherwise controls the NEC (save for CDOT’s portion) as well as the line to Albany (save for the MTA’s section), and the line to Harrisburg. But the sections they don’t own or control are state-owned, just not the Federal government. Prior to Conrail’s creation in 1976, those lines were all privately held and after 1983, the non-Amtrak-owned sections were under public ownership either.
The whole thing wasn’t perfect but had it not gone that way, there is a good chance that the northeast US wouldn’t have retained *any* freight service. The northeast was being deindustrialized by corporations and the intermodal traffic had yet to take off (once it did, it became Conrail’s bread and butter) so it worked out OK.
But really, not its pre-nationalization state.
We were still burning coal for electricity in the 70s. Allowing the railroads to close down wasn’t an option.
Just to expand on the liberalisation point. Let’s bear in mind that traditionally, vertically integrated national operators tended to be both conservative and protectionist, and this has always hampered international traffic – a combination of ‘I’m not interested’ and ‘don’t you come on my patch’. Liberalisation was really the only effective tool at tackling that, by mandating separation of track and trains, and non-discriminative open access, supported by moving towards common technical standards and inter-operability.
In the old days international services would be part of bi-lateral agreements with everything quid-pro-quo – you are in effect tied down by the weaker partner – ‘I don’t want you to run extra services because that wouldn’t be fare, but I don’t want to run extra services myself because it’ll be too difficult’. Liberalisation allows anyone to unilaterally set up operations as long as the physical capacity is there.
In an ideal world national governments would design their PSOs in a way that defines the national ‘takt’ (whether its Swiss-style timed connection based or Dutch-style frequency based), so that A) you don’t leave obvious service gaps so Open Access are less likely to come in and B) spare capacity is well defined so your takt is never compromised. Let’s face it, your Spains and Frances of the world are not going to get to that sort of competence in a million years.
Put it this way, suppose Madrid – Barcelona would be an all-day 4tph market under an enlightened PSO world (with universal ticket acceptance, non-compulsory reservation and all the other trimmings). In the old vertically integrated monopoly world you’d get 1tph with the occasional second. In the Open Access world you’d get non coordinated 2tph with sometimes 3rd or even 4tph with no cross-ticket acceptance. All I’m saying is this – don’t wait for Spain to become Switzerland. Take the 2tph world.
A lot of this seems to be EU having had success with competition in passenger airways projected it onto railways. Even though basic theory-of-a-firm tells you that different industries are different. Tony Yates a former BOE economist calls this “economism” i.e. it actually ignores fairly basic economic theory.
I’d disagree on putting down the competition aspect too much. Competing services on the same railroad is madness, and only in Osaka-Nagoya do we have real intercity competition between JR West and Kintetsu (Japan has a lot more competing urban regional rail services e.g. the 4 Kyoto-Osaka lines).
I’d argue that other modes of transport mean there is a role competition, air, inter-city buses, cars etc. And that most of the best services in the world have some degree of price/revenue/passenger number discipline to avoid the soft-budget problem (NE Asia, SBB, SNCF). I don’t believe in profit, but I do believe in cash flow.
The issue is that other modes get hidden or explicit subsidies, which make clean price based competition iffy, my soundbite on this is that unless you have privatised parking Japanese style you shouldn’t privatise rail even on the Japanese vertical integration model.
Also British privatisation is better described as pseudo or cosplay privatisation. The Government decides the fares, the Union contracts, the services, the capital expenditure, the subsidies and effectively the timetable. I’d argue SBB is more “private” than Southern or Avanti. British rail operators with the exception of TFL look more like Amtrack than Gilded Age rail barons. Which makes sense given the wider British state commitment to socialism for cars and huge soft-budget problems.
Governments are not all competent. It took Italo coming onto the scene to force Trenitalia to up its game. Without the on-track competition Milan-Rome or Bologna-Florence would never have been up to 4tph. It’s all very well saying Trenitalia should have operated / the Italian government should have specified through a PSO 4tph service levels in the first place but they were never interested in doing so.
UK’s BR or DfT were never interested in anything beyond Newcastle and Edinburgh on the ECML. It took Open Access operators to establish there is a market from places like Sunderland and Hull and an overall service level of the ECML beyond 5tph.
My ideal scenario would be governments letting PSOs under a national takt structure, but in reality very few countries have the grown-up politics required to make that happen.
Then you won’t have high-speed rail. The issue with the privatized model is that it can get ridership comparable to what an all-public system has on the strongest city pairs like Madrid-Barcelona or Milan-Rome, but then the thinner markets don’t get this much service; Italy still has barely half France’s high-speed rail passenger-km, and I think around two-thirds its high-speed rail passengers. On pre-corona numbers, Italy somehow has fewer intercity p-km than it did in the early 2000s (link 1, link 2); in France, this is unthinkable.
The sad thing is that in Italian transport, the parts that are done well – the construction of local and regional lines – are the ones where there is no privatization, no self-contracting for PSOs, and no “governments are not competent” ideology. It’s done through bureaucratic legalism, with government agencies doing what they’re specialized at, and none of the surplus extraction seen in the UK or Germany whenever middle-class retirees think they can run the government better than the actual civil servants.
Certainly in the UK one of the issues with the civil service is that the highest grade of “doer” appears to basically get my starting salary adjusted for inflation plus the additional transport cost to get into London. And I worked for a medium sized company so my starting salary wasn’t that generous.
Given that, and the inability for the civil service to be flexible with the grading, you’re going to struggle to get good people from a STEM background to apply for civil service roles.
If Italy generates half of France’s high-speed rail passenger km on a third of France’s infrastructure km then it’s doing pretty well. Trenitalia wasn’t competent before the arrival of Italo – this isn’t an ideology, it’s a fact. Italo forced Trenitalia to offer a better service at lower cost. Italy’s regional trains are let as PSOs by regional governments – sometimes they are OK, but they are not a patch on Switzerland or Netherlands in terms of service level. Lombardy has slow lines and infrequent trains (certainly very thin off-peak timetables). Regional services either side of Verona don’t connect because they belong to different regions. Places like Mantua and Cremona have shockingly bad service levels and again timetables that don’t connect.
The important thing is this. Liberalisation does not stop competent authorities from specifying high-quality PSOs (Netherlands, Austria, Belgium), but where state-owned incumbents do not perform, it allows the market to do something better. You can’t protect existing state monopolies and simply to expect to flick a magical switch to change politics overnight.
And actually often even the wealthy do good stuff with their activism. Isn’t it good that the upper middle class retirees of Henley on Thames checked the water quality around the regatta and used their contacts to get it in the media when the water was found not to be clean?
And frankly isn’t it rather embarrassing that the water regulator doesn’t do this stuff every year to hold the water companies to account?
https://www.theguardian.com/environment/2023/jul/04/henley-regatta-organisers-complain-of-sewage-pollution-from-thames-water
And frankly the semi-private British railway service is a heck of a lot stronger on most of the “thinner” routes than the French service is.
Manchester-Birmingham and vice versa have two trains per hour on a clock face schedule all day from 6am to 6:30 or 7pm and with at least hourly service until 10pm.
In comparison if you look at Lyon-Marseille which is mostly hourly, apart from a few random 90 minute gaps, and with some trains that you can’t share tickets with the standard TGV services.
Manchester is about the same size as Lyon, and Birmingham is about twice the size of Marseille. So these aren’t equally thin.
Ok fair.
However I still think the region to region rail service is typically much stronger in Britain than it is in France.
And I think you will struggle to come up with a region to region destination pair where the French service is stronger than the British service.
Birmingham – Manchester is proposed to have 2tph point-to-point services post HS2, and HS2’s modelling predicts fairly modest loadings for 200m trains. With the best will in the world, Lyon – Marseille will be a weaker market as the distance is greater and one of the settlements smaller. Offering a consistent ‘same minutes past the hour’ timetable and plugging the many 2-hourly gaps, and perhaps a half-hourly peak timetable, is probably the best it can achieve (and needs to achieve).
Avignon-Nimes-Montpellier is a route that strikes me as one that would get a half-hourly (conventional) service if transplanted to Britain.
To be fair you could run 4 or even 2 car TGV services perfectly sensibly if there isn’t demand for more carriages.
The Scottish and Welsh run 4 car intercity 125 trains on some routes.
Many – perhaps even most – trains in Britain are only two cars long.
When it comes high-speed rail operation I don’t think the economics of non-standard train lengths (i.e. sub 200m) stacks up. The high-speed rolling stock market is a lot smaller so the costs of developing and maintaining sub fleets and micro fleets are disproportionately higher than with regional operations.
Selfishly using this blog as a personal bookmark as much as anything else – I’ve found a study looking at air vs rail competition in Europe.
Click to access 2020_03_Air2Rail_Koios_strategy_rev.pdf
Presumably you’ll have come across it already – would it help calibrating your model for Europe?
Can’t say all the numbers ‘look’ right – there are definitely gaps in the data.
What happened with the good old Occam Razor?
1) Every country over performs the model
2) The model is based on Tokyo
3) Tokyo is capacity constrained
4) Tokyo is getting a new HSR line
PS: Why other islands underperform?
Because they compete for resources: Everything north of Japan is designed to fit the Omiya-Tokyo timetable, everything south ultimately depends on the Tokaido Shinkansen timetable. Not only because Shin-Osaka is a JR Central station, but because the depot is east of the station (only the local Tsubame doesn’t goes to Osaka).
Okay, but the LGV Sud-Est is capacity-constrained as well.
But even on that line the capacity constrained bit has no stops and there’s a lot lot more space at Gare De Lyon for arriving/departing trains.
The interconnexion is obviously more complex but I’m sure it could be handled.
If the Tokaido Shinkansen can run 16tph at peak with stopping services you could certainly run at least that on the LGV Sud Est if you really wanted.
LGV Sud-Est is capacity constrained in a political way: “Let’s consider options”. If I recall correctly they still keep a 1,5 hour break for maintenance at 13:00, single deck Reseau’s still operate, etc.
The Shinkansen is constrained in a physical way: “We ran out of ways to fit more people in”. During Golden Week they allow people to go standing in the deck between cars, even in cars with reserved only seating.
From a business point of view there is a difference from the point you sell all seats and the point you need more capacity. In the former you have found equilibrium (assuming the fares are market based), in the latter you’re running below equilibrium (it’s worth to price out potential customers).
The Shinkansen isn’t capacity-constrained at 13:00 either. Both Japan and France have rush hour peaks.
The Shinkansen runs 16tph and the French top out at 12tph right? So the French could run 1/3 more trains if they wanted.
Besides Tokyo only has two lines into it, north and south, the French have south east, east, north east and west/south west. All of the others other than south east definitely have more spare capacity. I mean north east has a peak service of what 8? 2tph London, 2tph Brussels, 2tph Lille Flanders, 1tph interconexxion, 1tph Lille stopping?
I think the current peak on the Shinkansen is 14-15 tph. The TGV runs a bit less, yeah – I think it’s a matter of worse signaling and worse reliability due to too much branching.
And yes, the LGVs other than Sud-Est have far less traffic. Nord is the one that’s criminally underused, due to poor cross-border service; there’s more stuff to the north of Paris even ignoring London than to the southeast, but Thalys charges too much and the lines north and east of Brussels are too slow.
“I think the current peak on the Shinkansen is 14-15 tph”
They do manage to eke out a full 16 tph if needed. Tomorrow (July 17), there are 11 Nozomi, 2 Hikari, and 3 Kodama services departing Tokyo Station between 17:00 and 17:57.
Signalling upgrade to ETCS from TVM on LGV Sud Est is supposed to bring capacity up from 13 to 16 trains per hour.
https://www.railjournal.com/news/work-underway-on-lgv-sud-est-improvements/
Still, SNCF is not operating at infrastructure capability. While in principle there are (say) 13 unique hourly paths, almost none of them operate anywhere near every hour. I think a same-minutes-past-each-hour service pattern could look something like this (assuming a future scenario of Rhin/Rhone, Lyon-Turin, Marseille-Nice and Montpelier-Perpignan all complete) :
– 2tph Paris – Lyon point-to-point
– 2tph Paris – Marseille and Nice
– 1tph Paris – Montpelier St Roch via Nimes
– 1tph Paris – Barcelona
– 1tph Paris – Milan
– 1tph Paris – Geneva
– 1tph Paris – Basel
This is only 9tph. If Barcelona, Milan and Basel all become 2tph then that’s just 12tph. That still leaves up to 4tph for services from LGV Interconnexion Est (in reality I don’t expect more than 2tph to be taken up, one from London and one from Brussels).
While it is true the French have more branches they have far fewer stops. LGV Sud Est as a whole only has 2 stops both of which really only need one train per hour.
Certainly out of Paris at peak I find it difficult to understand why they can’t run a train in every theoretical slot with a missed gap every 30 minutes. Yes maybe you’d hit reliability a bit, but you could have a 3-5 minute longer stop out of the contended section to recover time.
And arriving into Paris I guess there’s some challenge, but I suspect the peak isn’t as hard, and I really don’t see delays of less than 5-10 minutes on a long distance service being a dealbreaker. Certainly with the number of platforms at French terminus stations there is no excuse for delays of that order to compound.
@ Weifeng, you’ve missed Dijon/Lausanne which probably gets hourly service at peak, plus I believe there’s also the valley leading up to Bourg-Saint-Maurice which gets service.
There are also TGV trains to the valley for Chamonix, but to be fair they could be extensions of the Geneva TGV Lyria very easily.
And I do basically agree that it seems extremely difficult to see how they are really usefully filling 11-12tph even at peak.
On a parallel path railway it doesn’t really matter if trains are 5 minutes late.
There were two routes I missed out – Annecy and Grenoble. I think they can share a path – either 2-hourly to each or split/join at St Exupery.
Does look Dijon has a second tph at peak times.
This then does take us to 14tph out of Paris and 2tph via Marne-la-Vallee, hitting the post-ETCS 16tph cap.
So I’ve tried to do a count of PM peak departures from Paris-Lyon. I think I’ve captured all the trains, and the result is slightly surprising even by French standards.
There are 5 departures in the 1600 hour, 10 in the 1700 hour, and 9 in the 1800 hour. There’s roughly 1tph from Marne-la-Vallee. So the 1700 hour is just one path shy of hitting the current capacity limit. Trains are roughly every 4 minutes which I think is the capability of the current TVM signalling.
Most surprisingly there are no Lyon trains in the 1600 hour. Also due to current long journey times the evening peak is essentially too late to run Milan and Barcelona trains, but you’d expect that to change once all the necessary high-speed infrastructure on the Milan and Barcelona routes is delivered.
(I’ve only looked at one day so could well have missed some services that don’t operate everyday)