High-Speed Rail Ridership Estimator Applet
Thijs Niks made a web applet for calculating high-speed rail network ridership estimates. This is based on the gravity model that I’ve used to construct estimates. The applet lets one add graph nodes representing metro areas and edges representing connections between them. It estimates ridership based on the model, construction costs based on a given choice of national construction costs, and overall profitability after interest. It can also automate the exact distances and populations, using estimates of population within a radius of 30 km from a point, and estimates of line length based on great circle length. The documentation can be found here and I encourage people to read it.
This is a very good way of visualizing certain things both about high-speed rail networks and the implications of a pure gravity model. For one, Metcalfe’s law is in full swing, to the point that adding to a network improves its finances through adding more city pairs than just the new edges. The German network overall is deemed to have insufficient financial rate of return due to the high costs of construction (and due to a limitation in the applet, which is that it assumes all links cost like high-speed rail, even upgraded classical lines like Berlin-Hamburg). But if the network is augmented with international connections to Austria, Czechia, Poland, Belgium, the Netherlands, France, and Switzerland, then it moves into the black.
To be clear, this is not a conclusion of the applet. Rather, the applet is a good visualization that this is a conclusion of the model. The model, with the following formula,
is open to critique. The minimum distance d can be empirically derived from ridership along a line with intermediate stops; I use 500 km, or around a trip time of 2:15. The constant c is different in different geographies, and I don’t always have a good explanation for it. The TGV has a much higher constant than the Shinkansen (by a factor of 1.5), which can be explained by its much lower fares (a factor of about 1.7). But Taiwan HSR has a much higher constant than either, with no such obvious explanation. This is perilous, because Taiwan is a much smaller country than the others for which I’ve tested the model (Japan, South Korea, France, Germany, Spain, Italy). There may be reason to believe that at large scale, c should be lower for higher-population geographies, like the entirety of Europe; the reason is that if c is truly independent of population size, then the model implies that the propensity to travel per individual is not constant, but rather is larger in larger geographies, with an exponent of 0.6. This could to some extent be resolved if we have robust Chinese data – but China has other special elements that make a straight comparison uncertain, namely much lower incomes (reducing travel) and much higher average speeds (increasing travel).
The other issue is that the value of c used in the applet is much higher than the one I use. I use 75,000 for Shinkansen and 112,500 for Europe, with the populations of the metro areas stated in millions of people, the distance given in kilometers, and the ridership given in millions of riders per year. The applet uses 200,000, because its definition of metro area is not taken from national lists but from a flat applet giving the population in a 30 km radius from a point, which reduces Paris from 13 million people to 10.3 million people; it also omits many secondary cities in France that get direct TGVs to the capital, most notably Saint-Etienne and Valence, collectively dropping 12% of the modeled Paris-PACA ridership and 37% of the Paris-Rhône-Alpes ridership. (Conversely, the same method overestimates the size of metro Lille.)
Potentially, if the definition of a metro area is the population within a fixed radius, then the 0.8 exponent may need to be replaced with 1, since the fixed radius already drops many of the suburbs of the largest cities. The reason the gravity model has an exponent of 0.8 and not 1 is that larger metro areas have diseconomies of scale, as the distance from the average residence to the train station grows. Empirically, splitting combined statistical areas in the US into smaller metro areas and metropolitan divisions fits an exponent of 0.8 rather than 1, as some of those divisions (for example, Long Island) don’t have intercity train stations and have a longer trip time; it is fortunate that training the same model on Tokyo-to-secondary city Shinkansen ridership results in the same 0.8 exponent. However, if the definition of the metropolitan area is atypically unfair to New York and other megacities then the exponent is likely better converted to the theoretically simpler 1.
Pedestrian Observations 
No discussion of the exponent 2 applied to distance? That value seems rather steep to me.
It looks empirically true on Japanese data (mostly Sanyo Shinkansen) and the snippets of within-country European data I have. It is also within the range of negative exponents of elasticity of ridership with respect to travel time.
This is really cool that this is a web app, and some data is better than none, but is Japan really the best country to base stuff on?
Their airports are extremely well located compared to plenty of other countries and the Asians have also avoided a lot of criticism on their environmental practices which can be pretty weak.
Frankly Japan’s effort on CO2 emissions per capita is pretty poor with a drop of only 15% from their peak – and South Korea and Taiwan have barely seen their emissions drop at all. That’s compared to the US with a drop of around a third, or Britain or France who have seen a drop of around half.
Malaysia and the Philippines in particular are extremely weak on plastic pollution with the Malaysians causing more plastic pollution than the rest of the rich world combined by some margin.
Certainly some of my madder European trips by rail have cost a similar amount to flying including airport access which is nowhere close to true for Tokyo-Kyushu.
the Japanese had very low emissions compared with Americans to start with. Measuring declines only is unfair.
The d set at a minimum 500 km because car travel would predominates at these distances doesn’t make sense. It excludes most HSR travel. Average TGV trip is about 435 km. It is based on unsupported assumption. In a 2015 report, the lek consultancy had HSR market share at 64% on 405 km Tokyo -Osaka, 65% on 431 km Paris Lyon, 61% + 11%conventional rail on 441 km Paris-Strasbourg, and a few days ago we were discussing the 81% on Nanterre- Lyon 473 km. If the formula is only valid on longer trips, how does it applies on 349 km Taiwan HSR or in South Korea.
The formula pretends all distances less than 500 km are treated as if they were 500 km. This is borne in Shinkansen ridership – the ridership from Tokyo to Sendai (325 km away) or Nagoya (342 km), relative to city population, is the same as from Tokyo to Shin-Osaka (515 km away). Within France, relative to city sizes and distances, even setting Paris-Lyon to the minimum 500 km distance rather than the actual 431 km, Paris-PACA overperforms Paris-Rhône-Alpes, I think because of fares and tourism travel.
Thanks, now I understand why you have a 500 km basis. The question is whether we should measure distance as the crow flies as I did when I said that there were 405 km between Tokyo and Osaka, or along the tracks (515 km) as you did . You said that Paris-PACA TGV ridership outperformed your model because of (lower) fares and tourism which is reasonable. I suppose you used track distances which are considerably higher than air ones. Paris-Nice is 686 as the crow flies, 916 km by rail. SNCF fares are market based and its main competitors are airlines on this market. How does the model perform if you use air distances ?
There aren’t enough datapoints to try to figure out the exact difference between air and rail distance empirically. Theoretically, rail distance should be better, because if there is a large difference between the two then it should penalize the train.
I like the model, by C is an absolute black box. Yes, the French seem really obsessed with riding trains, but that requires an explanation within the model, not outside. Are French roads much crappier and/or expensive? Do they have really terrible intercity bus service (or not at all)?
C is probably not fixed, either – Spain has been seeing double digit HSR traffic for years now (it has doubled since pre-pandemic numbers for long distance HSR travel, 1.5x for short range HSR services). Part of the reason is that by opening the market now there are actual trains running, instead of overpriced Renfe vanity projects.
The explanation within the model is that French and German fares are much lower than Japanese ones. Standard elasticities of travel volumes with respect to price explain most to all of the difference.
This also helps explain Spain. Madrid-Barcelona ridership is now within the parameters predicted by TGV ridership where it used to be within those predicted by Shinkansen.
Regarding the question about the driving experience in France: This does make me wonder if tolled-via-annual-placard (e.g. Austria, Switzerland) vs tolled-by-trip-and-distance (e.g. France, Italy) vs no toll (e.g. Germany) makes a difference for driving demand.
Yes it does make a difference. We tend to forget what is already paid like car maintenance, insurance or annual permits when deciding on trip options. Highway tolls are usually almost as expensive as a TGV ticket and fuel is also costly.
The Autorité de Regulation des Transports follows both highways and train traffic. It turns out that long distance (> 100 km) train and tolled highway passenger-km figures (at 2.2 pax/car) are comparable in France. Numbers were equal pré-Covid. Train ridership dipped below highways during the pandemic but is now 10% higher because gasoline prices when up after the Ukraine war.
Also if you drive maybe 10-20k kilometres your marginal driving cost is close to the fuel cost.
There’s a lot of people in that category.
Someplace where there is no congestion and parking is free. Many railroad destinations are in places where traffic is awful, involves tolls/fees and parking is difficult.
This is why I often say your transit system should encourage everybody to be a “unlimited rides monthly family pass.” Each of those is important as they are reducing friction of riding for somebody by making worry about the cost of an single trip.
Now the subject here is High speed rail. I’m less confident this is the right answer – high speed rail implies long trip times. Almost nobody is going to hours long trips on a regular basis, and a monthly pass wouldn’t help them. The exceptions are likely making business trips and so they expense the cost of the ticket and don’t care about discounts.
If your high speed rail is to get to a different city in the same CSA than the pass is useful – a lot of people are making that trip, and you want to encourage it. However this is balanced against those trips costing the operator a lot and I’m not clear how much is marginal and how much is a fixed price (land costs, regular inspections), and how much is marginal (cost of energy to run the train and cost of the wear on the train) . For the part is a fixed price the monthly pass makes sense, but if marginal is a significant cost of the trip you need to charge heavy users more or your lose a lot of money.
The people who want to go long distances will be pissed off that people who are getting off at the first stop are sitting in the seats.
For any trip it is inexcusable to not have room for everyone. For long time trips – say more than 10 minutes that space must include a seat and be near seats for everyone else you are traveling with. For shorter trips you can ask the non-disabled to stand (but disabled here is very wide: covering people who can walk just fine and for any other purpose would not qualify). Anytime a transit vehicle is more than 70% full you should be adding more service (preferred – in most cases frequent service brings more riders) or larger vehicles (for trains this implies longer platforms and thus can be significant cost, but a larger bus isn’t that much more). Either way routes busy enough to ask this question are generally also busy enough to pay for themselves from the farebox alone, so only a small subsidy is needed to provide more service.
If you really must run > 80% full (while building more service) then you need some form of assigned seating for groups. There are a lot of options here, try to find the on best for riders.
Remember we are not running airplanes. The economics of running a half-full bus or train are very different from airplanes. So be careful what you learn from airlines – they have things to teach transit, but they also have some different situations and so not everything they do applies.
The transportation company has expenses to pay for. Running empty seats around for the majority of the trip isn’t the way to do it.
The evidence points to break even being under 25% full in terms of covering the marginal costs.
So maximising frequency (and therefore overall ridership) is within reason the way to go.
The challenge is what to do when your major terminals or approaches are full as in Britain.
25% may be enough to break even at UK rail fare levels but is far from enough in France.
TER services average 30% occupancy ratio, and ticket sales do not cover track or operating expenses. Adding another train inevitably results in additional losses.
Unsubsidized Services Librement organisés or SLO are expected to pay more than marginal costs and most likely struggle to break even at 70% seat occupancy ratio. Most of the approaches to Paris are congested at peak hours. Adding tracks would be very expensive. As Adirondacker already explained, this scarcity issue is addressed by congestion charges.
The real reason these low passenger trains can run in Britain is that the variable track access charges are 25p a mile for a 2 car train with some rolling stock.
25p/mile is about 0.18 €/km is indeed very low. Average track access charges were €10.5/km for French TER services in 2010, and are most likely higher today. They are mostly covered by direct subsidies. SNCF réseau argues that the maintenance of the legacy network in a state of good repair would require another € billion annually.
The €10.5/km that the French are charging is definitely covering a significant chunk of the non-marginal costs though. Assuming the same efficiency as the British 98% or so of the track fees that are paid in France are covering non-marginal costs.
To be fair in Britain there are fixed costs that are negotiated with network rail to cover at least part of those as well quite reasonably.
But the way it is priced in Britain means that a mid-evening train with a handful of passengers on board makes sense to operate. Operating such services does help the railway overall.
It’s not a challenge at all. There are these things called reservations. Make one.
The electronic brains commonly called computers will gnaw at the numbers and figure out how to price things to encourage price sensitive customers to use the less popular options.
If your break even occupancy is too low that implies you are charging too much and you could potentially get more passengers by charging less. If you are a private transit company this is a profit optimization exercise, but much transit is not private for-profit and they have other goals. Every person riding transit isn’t driving a much more dangerous car, nor are they flying with the high emissions (car emissions are probably worse than flying though a full car might be better). As such more transit should be asking how can I maximize my ridership while covering my costs, not how to maximize profits. What too low is I leave as an exercise – I’m not sure how to analyze it, between peak vs off-peak, better service leading to more riders, and higher costs discouraging riders this a hard problem.
@adirondacker12800 Reservations are a bad option. Sometimes they are the option you are forced into, but they are still a bad option. Reservation implies you can’t go when you want to go, and in turn a car that will go when you feel like it starts to look better. Most of your trip won’t be in traffic, particularly if we are talking about high speed rail which mostly crosses rural areas, and you are in control of when feel like going. If you must you must, but you shouldn’t have to.
Your implication is bogus. No matter what is your break even occupancy ratio you could get more riders by charging less. Rail travel is not a positional good people buy precisely because it is very expensive. In other words, elasticity to ticket price is negative. The question is how negative? Is elasticity between 0 and -1, meaning a price cut will bring more passengers but not enough to compensate for the discount? If on the other hand, price elasticity is below -1, say -1.3, a 10% price cut will bring 13% more passengers. You can be sure that the major players make every effort to find out where they are in that respect and keep the data under wrap. Elasticity varies with distance (there are trips where rail has more pricing power), time (peak vs through), patrons (business vs personal travel). Mandatory reservations and yield management are the tools allowing you to best allocate your limited resources. People who are more cost sensitive will often travel outside of peak hours. If you want to travel at peak times you either have to pay more or to reserve the cheap seats in the first hours SNCF put them on the market. If you need maximum flexibility, major carriers may offer, at a price, an elite status to their premium customers. Sure, reservations have a cost, but this cost has been steadily decreasing with technological advances. You may complain about them to your railroads. airlines, dentist and doctors but I don’t think they are going away.
It’s nice that there is infinite capacity on your planet. Here on Earth there isn’t.
Here, on Earth, there will be more demand than there is capacity. You have to allocate it someway. Adults understand that you might have to pay more for things when demand is high. And that when demand is high you might want to make reservations early. Because there won’t be any seats if you show up a few minutes before departure expecting to get one.
Most of the time capacity is effective infinite because capacity exceeds demand, or you can make capacity exceed demand just by adding more trains (cheap but 2 year lead time). Where there isn’t enough track space you can just build more tracks – it will take 5-10 years, but tracks to capacity mean there is more than enough demand to pay for the costs. (except perhaps in the US where we can barely build and then only for outrageous prices).
The economics of trains are not the economics or an airplane. Mandatory reservations make sense for airplanes because the economics mean you are better off leaving it on the ground if it isn’t near full. A train with just a few passengers by contrast pays for the marginal costs of running that train. Capital costs need not be paid for by all trains equally, all you need is the additional wear and tear be paid for and the total capital costs be paid for across all trains. Even here sometimes the comfort of knowing they could take a late train if they run late even though they never actually take it can be worth running an empty train.
That doesn’t matter though as trains are not only about pure economics. They are nearly all publicly run and thus subject to other public concerns like reducing traffic congestion on the highways, reducing highway deaths, reducing pollution, service for the poor, jobs, tourism, and other such things. Many of those things are helped by running more service and so nearly all trains get only a minority of their income from fares, and governments are thus willing to subsides to various amounts. (several of the above concerns also cause airplanes to get subsidies)
I disagree in most cases. It is a tool, but it normally isn’t the best tool. The best tool is running more service. Only where that isn’t possible (either short term situations or while building something) should you use it. Yield management means you are intentionally turning customers away from what they want to. Depending on your elasticity and you can often make more money doing this, but only at the expense of people not making trips they want – your rail line gets optimized for economics but the whole system of the city gets less economic activity done and so you are worse off overall because you over optimized for one thing when you owners want the whole system.
You are living in dreamland where capacity is not an issue because you just need to add more trains and in the rare instances where more tracks are needed “there is more than enough demand to cover the costs (except perhaps in the US)”.
I have news for you. HS2 is mainly built for capacity reasons, and the farebox is not covering 1% of the cost. It will serve Birmingham through a new station because New Street is full. It will start in OOC because neither the farebox nor the government are currently capable of financing the expansion and connection to Euston.
The Spanish government is financing the expansion of Madrid Chamartin and Atocha stations on the HSR side and connections allowing more trains to be diverted from Atocha to Chamartin because Atocha is full. A second HSR station is being built in Barcelona because Sants is at capacity. The fare box income isn’t going to pay for any of that, especially because it is currently not even covering operating costs. Your panacea, adding more trains without consideration of solvable demand, created the issue.
SNCF is spending over a billion € modernizing the signalisation over LGV Sud-Est to increase capacity from 13 to 16 trains per hour and fixing a bottleneck over its connection to Massy-TGV. The construction of the Bordeaux-Toulouse LGV is starting with the addition of 12 km of third track on the access to Bordeaux station (€ 900 million) and quadrupling over 19 km in Toulouse suburbs (€ 1 billion).
JR Central is building the Chuo maglev to address capacity issues on the Tokaido Shinkansen. It was planning to rely on the latter’s farebox to pay for the expense but is now receiving 3 trillion ¥ from the government.
There are many hours in the week and many km of HSR where capacity is not an issue but the gestion of scarcity is still key. Adding capacity in urban areas is expensive but that is often where it is needed. There is a lot of unused capacity on the AVE network to Galicia, Asturias or Basque region but 75% of the trips start or end in Chamartin. One just can’t ignore the problem. It is a make or break issue and equilibrium is needed. The scarcity of seats at peak hours may be an inconvenience for travelers but is the lifeblood of the system. The expensive seats that sell last are the ones paying the bill. One cannot always count on the taxpayer, especially in Trump’s America.
Here on Planet Earth people don’t want to travel in the dead of night. There needs to be a way to allocate capacity when it’s not the dead of night.
@dralaindumas, there are a lot of examples of weak pricing in the rail industry. Tokyo to Hakata is too expensive so people fly, or the French have no concept of an off-peak flexible ticket on the TGV which is the most car-like ticket, or the British system is absurdly over complicated, or the Germans don’t have peak fares.
Again, I don’t know how things work on your planet. Here on Earth, generally, the longer the distance the more a trip will cost. Hakata is really far from Tokyo. People, here on Earth, take into consideration the amount of time trips over long distances take. So do the airlines. And the bus companies. I’m sure some people drive.
Just for fun I asked Google “Paris Lyon train”. WIly wily electronic brain decided that I wanted to check fares. It offered fares between $30 and $133 for Tuesday of next week.
Yes it is. They seem to enjoy it. Partly has something to do with reaching capacity certain times of the day and charging a lot if you want to travel during them. Clever Google, for London-Manchester, is offering lowest fare of $29 with a change in Crewe and a high of $261.For Tuesday a weekl from now.
I asked the Google for “Berlin hanover train” Again it offered me fares for next Tueday. Lowest price is $11 and the highest I saw was $57.
Those dastardly dual mode locomotives Amtrak will be using means they are studying extending service to Ronkonkoma. I suspect something along the lines of Virginia services. The northbound terminating trains can instead continue thrrrrrrrrrrrrrrrrrroooooooooooough to Jamaica, Hicksville and Ronkokoma. Adults will be able to check the trip planner and find that the train to Newark and beyond departs at 8:17, 12:48 or 3:27 ( or something like that ) instead of just wandering down to the station and expecting a surprise.
People in Suffolk are jumping up and down in and exclaiming it might happen by 2028. No reason why it couldn’t happen tomorrow. The tracks, platforms etc. are there. Other people are pointing out it still hasn’t been studied thoroughly. I agree that 2028 is optimistic because the government is in control of Real Americans(tm) who think trains are a communist plot to sap and impurify their precious bodily fluids.
Very cool!
You wrote about STOPS before and I was curious about it so I went to the FTA website and downloaded it. I’m sure I’m missing some plugins or data (only about a quarter of the way through the documentation) but I’m curious how your and other gravity models compare to STOPS results for the same model inputs? From what I see so far, STOPS inputs are very complex and gravity model inputs aren’t (by comparison) but I’m curious how the two may correlate?
And with respect to STOPS are you happy with the way its modeling was validated? Not a lot of data points unfortunately.
I’m going to have a really fun time drawing crayon HSR maps with this, that’s for sure.
Another question I have is at what network speeds (i.e. type of rail transit, say, low-speed intercity local, regional/commuter, medium-speed intercity) this type of model could be valid? Is a gravity model inherently a strong approximation of human behavior for high speed rail, but not for, say, a commuter line?
STOPS is not designed as an intercity estimator. It’s a local transit ridership estimator, and requires an initial ridership estimate to compare changes to; it cannot usefully estimate ridership without the initial number, which means that every New York project we analyze requires an assumption on systemwide ridership based on return to office trends.
Cool, thank you for replying. The reason I asked is I’m curious if there are publicly available models which estimate intercity ridership.
Very interesting appearance on the Volts podcast, by the way.
It’s the one other people come up with. With minor variations. It’s a starting point, a sanity check. Which is why New England is never going to need 12,000 seats an hour in either direction. There aren’t enough people.
It’s distance assuming the speed is the same. It’s time. Other people derive charts where the travel time from the origin to the automobile is not accounted for as is the travel time from the automobile to the destination. And come up with the same answers. Cars are good for short trips, airplanes are good for long trips and high speed rail is good for intermediate trips. In other news water is wet and fire is hot.
In ancient times when you could arrive at the airport moments before a flight and still make it onto the plane the travel time rail had to compete with was three hours. It’s four or four and half today.
…. it’s time, even for conventional-ish rail. New York to D.C. has high market share because it’s quicker than flying. New York to Boston which is the same distance give or take a few, is slower. Or Springfield Mass. is as far from New York as Albany is. Give or take a few. New York State has spent money improving the corridor and today it’s faster than driving. And gets a lot of riders. Connecticut and Massachusetts neglected things for decades. Springfield to New York is slower than the bus. It’s not as slow as it was a few years ago but it’s slower than a bus.
And it’s still just a starting point/sanity check. Low cost carrier, I forget which, pulled out of the Providence-Philadelphia air market. Which gave the legacy carrier the freedom to charge monopoly prices. Amtrak ridership boomed. So it’s more complex than just time or distance.
…. one of the minor reasons for the High Speed Ground Transportation Act of 1965 was that people didn’t like driving in cities and especially in Manhattan. When I want to go to Manhattan or Washington D.C. I check trains. When I want to go to Boston I drive, even though traffic and parking can be almost as bad as Manhattan. The – as in the once a day – Late For Sure Limited, is slooooooooow. Population and distance is a starting point. People, being unpredictable, will then do things like the French versus the Japanese.
Another starting point would be to recognize that , contrary to what you and Amtrak say, rail doesn’t have a high market share on the NEC. Between NYC and Washington DC = 364 km, Amtrak has a 15% market share (7 Acela+ 8% pulled coaches), buses have 12%, private cars 71%, airlines 2%. A high market share on a similar distance can be found between Taipei and Kaoshiung (349 km): HSR 42%, conventional rail 14%, buses 17%, private cars 27%, air 0%. Between Philadelphia and Providence/Boston I.e around 500 km, rail maker share is 6%. New bridges and tunnels between Manhattan and Newark, and a new generation of Acela trains are a good thing but aren’t going to make a big difference. Amtrak should stop pretending that they are almost there in terms of HSR. They should acknowledge that these numbers are embarrassing and take a good look at Alon’s and Marron Institute’s proposals.
I was considering the behavior of actual people today not how Americans would behave if they were parachuted into Taiwan. Or Japan or France. Or what they are going to do in 2040 when Amtrak manages to make things as fast as Metroliners were half a century ago. The distance between New York and Boston or New York and D.C. are almost exactly the same. It takes a lot of time to go through the squiggles along the coast of Connecticut. Since the distance is the same the difference is the time. Bypassing the squiggles someday will change the distance. But not by much. It will decrease the time.
The distance between Tokyo and Osaka didn’t change much when the Shinkansen opened. What changed was the time it took to get between the two. Or Paris and Lyon. …it’s time not distance.
Alon want to be frugal and make the trip between Boston and D.C. four hours. Because the distance isn’t going to change much… it’s time. Four hours isn’t very good time. Other people want to spend some more money and make it closer to three hours. … it’s time, not distance.
Yes, time matters. A weakness of Alon’s formula is that it assumes a 2h15 trip duration for a 500 km trip, a reasonable assumption if HSR tracks are available all the way but a big if. There won’t be any such HSR in America or the UK in my lifetime.
Alon’s formula, correctly in my opinion, doesn’t distinguish Americans, Europeans or Asians. People respond similarly to incentives.
Current US incentives do not favor passenger rail. Gasoline is lightly taxed by world standards, Interstate highways are mostly toll-free and, as you stated, Amtrak trains are not faster than 50 years ago. Laws and procedures make HSR construction difficult and slow. The projects completion is so remote that Amtrak and transit agencies emphasize the number of good jobs created rather than the results.
The benefits in terms of time-saving are determinant when French LGV projects are evaluated. Time is given a certain value, which is adjusted by taking into account recent experience. I don’t have the latest values but it was 43.3 Euros for 1 hour for business travelers and between 21.8 and 22.7 Euros for personal travel in 2010. Time savings are divided by two for modal shift passengers because it is assumed that their current choice is faster than what SNCF can presently offer. In the end, the Euro value of saved time is shared equally between SNCF, which will increase ticket prices accordingly, and passengers.
They admit it doesn’t account for every situation in all cases. It comes up with the same chart everyone else comes up with, cars are good for short trips, planes are good for long trips and high speed rail is good for intermediate trips. Time and distance are loosely coupled and people make decisions based on time. Because they have no frigging clue how distant their destination is. Until they look to see how many “miles” they are going to get for the trip.
Improved speed would be great and significantly faster trains would further increase demand, but I’m not sure they are meeting the current demand.
Does Amtrak put out the data on what percent of seats are currently going unoccupied between DC and New York? If there’s not many empty seats and the current tunnels under the Hudson are at capacity during peak hours then getting more market share is about increasing capacity not improving speed.
Amtrak can not measure how many people took some other mode, or stayed home, because they are sold out. When I checked a few moments ago, for travel tomorrow, between New York and Washington, first class on one Acela is sold out and business class on a few Regional is. The new Acelas are longer and there will be more of them.
They’ve needed more capacity since they opened the connection between the former Delaware Lackwanna and Western and Penn Station New York. And almost everybody abandoned the bus. Nearly 30 years ago. I seem to remember, when Access to the Region’s Core was canceled Amtrak proposed having Gateway open by 2025. Access to the Region’s Core should have been open by now.
The work between New York and Newark is all work that needs to be done anyway. It should save a few minutes. Assuming funding isn’t rescinded. A few minutes here, a few minutes there, pretty soon it will be fast as it was decades ago. Sometime… after we are all dead… they might get it down to average speeds the rest of the world has today.
Amtrak publishes monthly reports for the Northeast Corridor as a whole. The March one shows a 67.5% seat occupancy ratio on Acela and 63.9% on Regional trains.
This shows that Amtrak offer matches well with demand. Their trains are well patronized without being full. However, Amtrak does not cater to the price sensitive travelers. NYPenn to Washington DC prices next Tuesday go from $55 to $701. This is twice as much as what Adirondacker found on semi high speed European trips like Manchester-London and Berlin-Hanover. Average March revenue was 0.544 $/passenger-km on Acela, 0.284 $/pax-km on Regional, very high by world standards.
The budget conscious traveler can find bus tickets for $35 with a much larger choice of departure times than with trains. Rome2rio put the driving cost as between $42 and 90 for the 226 miles.
What prevents Amtrak from offering cheaper fares? The new and larger Acela trains and replacements for the 50 year old Amfleet will help if the answer is lack of rolling stock. The Gateway and Baltimore tunnel projects will be the answer if the issue is lack of slots on the congested NEC. The problem could also be that Amtrak is simply a low productivity, high cost, carrier.
It shows that Amtrak has pulled off a marketing coup. They are able to charge premium prices for slightly better services. So people can brag “I took Acela”.
If the crack of dawn train is half empty that doesn’t change that the cocktail hour trains sell out. In the very short term they could run longer trains. The combination of longer trains and more of them is supposedly 40 percent more capacity on Acela. The long term solution is more tracks under the Hudson and eventually tracks over Long Island Sound. Though more tracks in New Jersey might be needed sooner.
Staff gets paid by the hour. The rent on the seat is by the hour too. That whole time and distance thing rears it’s ugly head again.
There are funner ways to satisfy masochistic urges. The bus or an automobile can get stuck in traffic.
I’m not in the mood to ferret out the tolls between Times Square and Capital Hill. The reimbursement rate for 2025 is 70 cents a mile. $45 is a very bizarre route and cooking the books hard. $90 doesn’t require cooking the books as hard.
@adirondacker12800, the Acela is much more reliable. Pretty much certain to be under 10 minutes late.
It has nothing to do with on time performance. If the train ( or bus or plane or ferry or aerial tramway that delivers you to a funicular ) is slow the cost per unit of distance whether you want to measure that in miles or kilometers or furlongs of pico-parsecs is going to be higher. Because it takes more minutes. Or fractions of a fortnight.
Thanks, Alon, for highlighting Rail Mapper. I had a lot of fun making it and can’t wait to see which networks other people will come up with!
The current version does not allow for segment specific cost estimates, but you can export any network to a fully functional Excel spreadsheet and it’s easy to adjust specific numbers there. Here is the Construction Cost matrix for Germany, for example, as spreadsheet.
I calibrated my default constant against the Northeast Corridor, trying to hit the same 80 million annual passengers. But it’s probably too high and I’ll likely revise it downwards.
One of the tricky things to consider when building out these networks is the question which stations to include. I’m working on a new version of the French network using the following rules for adding stations:
At least 100k residents within a 30km radius
The 30km radius does not (significantly) overlap with other stations
Has at least 1 direct TGV InOui connection to Paris per day
Artificially increases the length of regular rail segments by 2.5 times to account for the travel time elasticity of (150 kmh / 300 kmh)^-1.3
This would increase the covered population from 26 to 39 million and reduce the calibration constant from 200k to 105k (to still match the 130 million domestic passengers per year).
I’m also in the process of training the model on the Japanese Shinkansen network, which seems to land on a constant of 85,000.
If we assume the fare per kilometer is $0.25 for the Shinkansen and $0.15 for the TGV, the French constant should be ($0.15/$0.25)^-0.5 = 1.29 times higher than the Japanese one if we account for the price elasticity.
85,000 * 1.29 = 109,735
Which is reasonably close to 105k.
The 30km radius unfortunately crosses the border with Belgium, increasing the population estimate even though it’s unlikely many Belgian residents will use the TGV.
I’ll continue to make improvements based on the feedback.
Feel free to leave comments on my Google Doc!
CAHSR has recently released a 2025 Supplemental Project Update (dated August 22). If it cuts back to doing a Gilroy – Palmdale Phase One (Scenario 3 in the report), it estimates ~15 million annual riders (at ~$90 per ticket, generating ~$1.3 billion in passenger revenue). What is the ridership estimate coming from here and other places?
https://hsr.ca.gov/about/project-update-reports/2025-project-update-report/
The CAHSR update is available in various alphabets and languages including Hmong, but not in the universally understood language of travel times. We have 100 pages of obfuscation, technical jargon and self-praise, predictions of ridership, ticket sales and ancillary revenues like naming rights, dubious data on other projects (9 miles of tunnel on LGV Sud-Est, Madrid-Barcelona AVE costing $75 million/mile) but not a word on the different scenarios trip times.