More Sanity Checks on My High-Speed Rail Model
Writing my Eastern Europe high-speed rail post meant I needed to go back to my ridership projection model for high-speed trains. This model aims to analyze large networks like the entire Shinkansen or TGV and project traffic for extensions of these systems, for upgrade of partially-built networks like that of Germany, and for entirely new systems like any proposal for the United States. To compensate for the relative paucity of training data, the model has as few variables as possible – just four:
The populations of metro areas A and B are in millions, distance is in kilometers, and ridership is in millions for year. The four constants – 75,000, 0.8, 500, and 2 – are roughly motivated by Shinkansen data, with some European sanity-checks; the 0.8 exponent is also motivated by some practical attempts to subdivide metropolitan areas into pieces, some on the line and some off it. But is this right?
And as it turns out, the model nails trips from Tokyo to most metropolitan areas in Honshu. Defining Tokyo as the prefecture plus Kanagawa, Chiba, and Saitama Prefectures (#15-18 in the spreadsheet), the spreadsheet says there are 2,742,200 annual JR trips between there and Aomori Prefecture (#6); the spreadsheet predicts 2.59 million trips between Tokyo and the prefecture’s two cities, Aomori and Hachinohe.
Two years ago already, I made a spreadsheet with the model’s predictions for ridership between each pair of metro areas on the Shinkansen network. This can be compared with actual numbers of prefecture-to-prefecture trips by mode (except cars). For most long-distance trips, the numbers in the second sheet matter, comprising all Japan Railways trips; for interregional ones for which slow lines maybe an alternative, it is perhaps better to use the fourth sheet, which excludes season passes used by regular commuters. The numbers in the spreadsheet are directional, so cells should be added to the ones diagonally across.
The problem is that the model severely overpredicts inter-island trips. Tokyo-Fukuoka is 4.61 million per the model; the actual count per the spreadsheet (Fukuoka is #44) is 1,203,600. Tokyo-Hakodate (#4) is likewise 954,000 in the model and 378,200 in reality. Osaka-Fukuoka fares better but still notably underperforms: the model says 10.2 million, whereas the spreadsheet, defining Osaka as Osaka, Kyoto, and Hyogo Prefectures (#30-32), gives 7,162,900, even though the distance is similar to that of Tokyo-Osaka.
Another category that the model overpredicts is trips through Tokyo. Those are less convenient by Shinkansen as they require a transfer, and the transfers are not timed, though the frequency on the Tokaido Shinkansen is such that untimed transfers are not the end of the world. The model predicts Osaka-Sendai at 2.21 million and Nagoya-Sendai at 1.97 million; the spreadsheet says metro Osaka-Miyagi Prefecture (#8) is 194,500 trips, and from metro Nagoya (Aichi Prefecture, #27) it’s 234,300. It’s not even quite a matter of distance: Osaka-Sendai is 840 km, about the same as Tokyo-Hiroshima, which the model gets correctly. It’s just a through-Tokyo effect, like the inter-island effect.
Fortunately both categories can be corrected by taking plane trips into account, listed on the last sheet. Tokyo-Fukuoka by both air and rail is not 1,203,600 trips but 12,313,300; this is a five-hour trip, at which point in Europe the modal split is fairly even, for example on Paris-Nice (p. 3) or more generally in France (source, p. 33). If we include air trips, Osaka-Fukuoka is 8,199,900, Tokyo-Hakodate is 1,482,200, Osaka-Sendai is 1,624,000, Nagoya-Sendai is 445,100. If we take the combined air-rail market and apply more common European modal splits, there is no inter-island penalty, just a bit of a penalty for Osaka-Sendai and a big penalty for Nagoya-Sendai.
Finally, the model chokes on short-distance ridership, generally underpredicting it. It’s hard to exactly count between two relatively close provincial cities, since very short regional trips are included and not just intercity ones. But even excluding season passes, Miyagi-Aomori is 1,140,400 trips a year, where my model says 446,000.
The dicier situation is that of Tokyo-Shizuoka (#26). My model predicts 14.1 million annual trips between Tokyo and the combination of Shizuoka and Hamamatsu, the two largest cities between Tokyo and Nagoya. The spreadsheet gives the number at 37,264,700, or 26,458,300 excluding passes. Even that likely includes some legacy rail trips: there are 9,382,000 non-pass trips between Shizuoka and Kanagawa Prefectures and 15,646,300 between Shizuoka and Tokyo itself, where from Nagoya and Osaka the ratio is more like 1:4; Shin-Yokohama is located in a less central place than Yokohama Station, and I suspect most of the 9.4 million figure rides the Tokaido Main Line.
To complicate things further, while my model underpredicts trips of around 150-200 km, it doesn’t do so to Tokyo-Sendai (10.5 million predicted, 10,924,000 actual), Tokyo-Nagoya (32.4 million predicted, 28,391,100 actual), or Tokyo-Niigata (11.1 million counting Nagaoka and Sanjo, 10,436,600 actual), and it actually overpredicts Nagoya-Osaka (19.2 million; the actual is 13,026,800). Thus, I am still reluctant to change the 500 km minimum in the denominator below which distance is deemed to no longer matter, even if I include a fudge factor for Nagoya’s repeated underperformance.
The one snag that may be worth addressing is the scale issue inherent in the 0.8 exponent. The exponent seems broadly correct from Japanese data. Moreover, when I break metro areas like New York into smaller constituents and compute their individual trip times to other places to apply the formula to each, the exponent seems correct: large metropolitan areas are so big that many of those smaller pieces are far from the train (for example, Long Island) and have additional trip time to consider, reducing overall ridership.
However, the 0.8 exponent also means that as the area of study grows, the model expects per capita trips to rise. Doubling the population of every metro area increases overall ridership by a factor of about three, which may be correct at small scale (it probably means those regions get stronger connecting lines and better train frequency), but should fail at large scale. Evidently, Taiwan overperforms the model by a factor of about three. Just shrinking the exponent to 0.5 wouldn’t work – it would lead to massive underprediction of Tokyo-Osaka ridership and overprediction of Tokyo-Sendai, Tokyo-Hiroshima, and Nagoya-anywhere. There may need to be a fudge factor for systemwide population.
Best I can say is that you should trust the model in the range of system sizes between Spain and Japan, so around 45 to 120 million. It may be possible to go above 120 million in geographically larger regions, like the Eastern United States (around 190 million) or even the entirety of the EU from Warsaw westward (around 370 million), if the large geographic extent means that in practice the network has its own diseconomies of scale from the long end-to-end distance. In other words, the pure 0.8 exponent is obviously false if we double the population of each metro area – but if we instead double the size of the area of study by including more cities in it, then the average distance grows and then the exponent of 2 in the denominator countermands the effect.
Pedestrian Observations 
what can be a long distance high speed international rail that has Istanbul as a terminus? Bucharest-Budapest or Sofia-Belgrade too long to make sense?
I think they’re too long and the topography is too painful.
The problem is not just the length, it’s that Sofia, Belgrade, etc aren’t huge cities, international links tend to underperform especially when there is a hard border involved, and the countries in question don’t have much reason to build domestic HSR, because their cities are not huge and distances are not long. But for example Serbia is building a 200 km/h line north to the Hungarian border and presumably Hungary will continue it to Budapest. It would be reasonable to extent this south as well, to Sofia. So that is the kind of network is more likely for the Balkans than full HSR.
large metropolitan areas are so big that many of those smaller pieces are far from the train (for example, Long Island)
It’s not if the Hikari and Kodama are stopping in Brooklyn, Jamaica, Farmingdale and Yaphank. Since it’s all the same loading gauge etc. as a the commuter system, they can stop in New Rochelle and Stamford too. Bound Brook, West Trenton and Jenkintown, someday.
Yeah, you can slice this as granularly as you’d like; I poked around metropolitan divisions of the base MSA and subsidiary MSAs in the same CSA, some served (New Haven, Trenton) and some not (Kingston).
No reason why some century in the future the Buffalo-DC Kodama doesn’t stop in Kingston. Like Empire Service trains do today. Though there are consultants, lobbyist etc. who might more interested in Buffalo-Harrisburg. Saratoga Springs-Harrisburg, soak up some of the peak hour demand? Though that would more likely be a morning and evening Hikari that doesn’t stop in Kingston. Kingston-New England would probably involve changing trains in Albany. All “free” because the tracks would be there for other reasons. So is sending a Kodama or Hikari through Stamford instead of Farmingdale….. Farmingdale is closer to Penn Station than White Plains Airport. And the tunnel would be cheaper to build because it could be cut and cover.
> “[the model] actually overpredicts Nagoya-Osaka (19.2 million; the actual is 13,026,800)”
Worth noting that Kintetsu has some market share on that route with its Hinotori and Urban Liner services – looking at page seven of the MLIT spreadsheet for non-pass travel on non-JR private railways (民鉄), that’s another 1,230,000 trips. Highway buses are also a competitor given the short distance – from page nine of the spreadsheet, they represent a further 1,149,600 trips.
I suspect the rest of the gap is down to Nagoya being more car-oriented by Japanese standards (Aichi of course being Toyota’s home turf), which could mean greater willingness to make the relatively short drive down the Tomei or Shin-Tomei Expressways.
> “If we include air trips, Osaka-Fukuoka is 8,199,900, Tokyo-Hakodate is 1,482,200, Osaka-Sendai is 1,624,000, Nagoya-Sendai is 445,100. If we take the combined air-rail market and apply more common European modal splits, there is no inter-island penalty, just a bit of a penalty for Osaka-Sendai and a big penalty for Nagoya-Sendai.”
I think another instructive comparison is Osaka-Kagoshima: it’s virtually identical to Osaka-Sendai in overland distance (~870 km), travel time (4.25-4.5 hours by train), and availability of air service, and down on population (Kagoshima Prefecture is only two-thirds as populous as Miyagi, and Sendai is a much larger city than Kagoshima-shi), yet generates slightly more air-rail trips (~1,774,000).
I wonder if we’re seeing the relative intensity of social and economic ties between regions. Osaka-Fukuoka, Osaka-Kagoshima, and Tokyo-Hakodate stick within ‘West Japan’ and ‘East Japan’, respectively, whereas the penalised Osaka-Sendai and Nagoya-Sendai both traverse that imagined divide.
Supposedly there are significant cultural differences (1, 2) between West and East Japan, not just a different electricity standard.
Tokyo-Fukuoka has a strong travel market, it’s just almost entirely by air.
Even Tokyo-Fukuoka seems to underperform relative to Tokyo-Sapporo. The spreadsheet shows 12,350,000 trips from Greater Tokyo to Fukuoka Prefecture across all modes, compared to 11,079,000 between Greater Tokyo and Greater Sapporo (Central Hokkaido, 道央). Both are inter-island trips of comparable distance (~850 km flying); Central Hokkaido has only two-thirds the population of Fukuoka-ken, yet manages to come close in trip count.
Most obvious explanation is Hokkaido is a popular domestic tourist destination
Which again demonstrates how socio-cultural and economic confounders can easily skew the ridership prediction model
Again as I mentioned in NE US HSR post, I think it is crossing big cities that matter instead of specific operating arrangement
Like (by prefecture):
Hiroshima:
-Shizuoka 1.5M (Model) 140K (Actual)
-Aichi 5.3M (Model) 532K (Actual)
-Gifu 0.66M (Model) 43K (Actual)
Okayama:
-Shizuoka 1.1M (Model) 85K (Actual)
-Aichi 3.0M (Model) 340K (Actual)
-Gifu 0.37M (Model) 27K (Actual)
Yamaguchi (including Shimonoseki):
-Shizuoka 0.41M (Model) 36K (Actual)
-Aichi 1.6M (Model) 127K (Actual)
-Gifu 0.22M (Model) 12K (Actual)
Fukushima
-Iwate 0.29M (Model) 81K (Actual)
-Aomori 0.26M (Model) 50K (Actual)
Tochigi
-Iwate 0.39M (Model) 54K (Actual)
-Aomori 0.31M (Model) 32K (Actual)
Gunma
-Toyama 0.45M (Model) 30K (Actual)
-Ishikawa 0.31M (Model) 54K (Actual)
(You should double all the actuals, don’t forget that the actual spreadsheet is directional.)
Yeah, the way Hiroshima-Kansai slightly overperforms the model (6.7 million vs. 5.8 predicted) while Hiroshima-Nagoya grossly underperforms (1.05 million vs. 3.2 million predicted) is weird – it’s way more than any plausible Nagoya underperformance fudge factor; note that Tokyo-Hiroshima is on-point even without adding air trips to the mix.
Corrected with both way traffic (Might different from actual figure a bit since I simply doublt it and the data both ways isn’t exactly equal in actuality):
Hiroshima:
-Shizuoka 1.5M (Model) 0.28M (Actual)
-Aichi 5.3M (Model) 1.1M (Actual)
-Gifu 0.66M (Model) 0.08M (Actual)
Okayama:
-Shizuoka 1.1M (Model) 0.17M (Actual)
-Aichi 3.0M (Model) 0.7M (Actual)
-Gifu 0.37M (Model) 0.05M (Actual)
Yamaguchi (including Shimonoseki):
-Shizuoka 0.41M (Model) 0.07M (Actual)
-Aichi 1.6M (Model) 0.25M (Actual)
-Gifu 0.22M (Model) 0.02M (Actual)
Fukushima
-Iwate 0.29M (Model) 0.16M (Actual)
-Aomori 0.26M (Model) 0.10M (Actual)
Tochigi
-Iwate 0.39M (Model) 0.10M (Actual)
-Aomori 0.31M (Model) 0.06M (Actual)
Gunma
-Toyama 0.45M (Model) 0.06M (Actual)
-Ishikawa 0.31M (Model) 0.11M (Actual)
I’m writing this on the Shinkansen from Hakodate to Tokyo ^^
For me some factors that might cause the model to overpredict Tokyo-Hakodate and Tokyo-Fukuoka.
In both cities the airport is well located and while the Shinkansen station at Fukuoka is central, the one at Hakodate is way out with a badly timed connection of 10 to 20 minutes waiting. Also the tunnel and approaches of the Seikan Tunnel are not running at high-speed. This makes the whole journey time around 4.5 hours.
Shinkansen prices are high, this means that tickets to either city are around 24k yen (around 160 Euro). Flight prices for the coming month start at 7k yen and only reach price parity for the coming 3 days. This compares to Paris-Nice, where one can get cheaper tickets in advance making it more price competitive.
Also there is only one train per hour, that has only reserved seats, making arriving at the station without pre-planning less feasible.
Another interesting point that came up for me on the topic of seat reservation. The train was fully booked for the Hakodate-Tokyo trip, because there was not one seat that was free all the way. So splitting the reservation at getting two different seats was necessary. This is why reservation only trains work well for SNCF, as you say they operate an airline at FL0. But for real train companies running trains with multiple stops, where people get on and off, this leads to wasted seats. The hybrid model of DB and others might be not too bad after all. (At the Shinkansen you have cars with only reserved seats and cars with only unreserved seats, at DB you have it mixed together, but a sign at the seat shows, from where to where it is reserved)
Off-topic, but what I found interesting, Japan feels like the perfect cars and trains urbanism that you’ve described. Cars and trains are both expensive, but work in general well. I found the extent of elevated expressways in medium sized cities surprising. Bus (and rail lines further out) frequencies are often laughable here and there is little timed connections. I saw P+R at quite a couple of stations, but often only surface lots. Road bypasses get built while train lines stay slow. What is saving some transit is that even in newly built suburbia lot sizes are tiny leading to some density.
It’s probably not due to airport. Hiroshima to Kumamoto and Hiroshima to Kagoshima still get overestimated 5 times over exact traffic, despite all three prefecture didn’t have convenient airport and also no direct flights between them.
exact trafficactual trafficHokkaido Shinkansen is almost exclusively tourist traffic, so frequency of 1tph is basically all that is needed now, at least until the extension to Sapporo is completed in 2030. Likewise the all reserved seating, as there is very little walk-up and ride traffic (ie.business passengers) north of Sendai. My experience having been a passenger several times on trains Shin-Hakodate Hokuto-Tokyo is the most people board at Morioka and even moreso Sendai.
That seems to be a result of the service being provided instead of cause.
As others may have mentioned, Tohoku (esp. the northern part) and southern Hokkaido just don’t have the population or density of industry/economic activity to generate significant traffic. They are backwaters historically, culturally, and in a psychological spatial sense (i.e. distance from the national center i.e. Tokyo) compared to the Tokaido, Keihanshin, Sanyo, and even Hokuriku regions/corridors.
Japan is very much into building more and more roads. Part of it that the Highway tolls create a very reliable revenue stream for the Highway system, while the LDP’s road and construction lobbies generate strong support for overbuilding. Japan likes building elevated roads and railways because they need fewer land takings than trenches etc while being cheap than tunnels. And concrete lots of concrete (East Asian growth machines love steel reinforced concrete). Also within local government the bureaucratic heft of the roads in transit departments is pretty high and is constantly widening roads nationwide.
That’s why Japan’s best transit cities are ones with Private Rail companies which provide rail a local heavy weight (i.e. Tokyo/Keihanshin/Hiroshima). JNR neglected and still neglects regional city services (e.g. no Sendai s-bahn). So the cities tended towards car oriented sprawl. But I would argue the basics of Japanese TOD are pretty much everywhere, Outer Suburbn Tokyo looks pretty similar to the out suburbia of Nagoya, Shizuoka, Sendai or Kita-Kyushu. The difference is Tokyo (and Keihanshin) did their railways better. And if you look at those cities modal splits, rail has generally help up or increased while its buses and active travel that have lost out to cars.
Buses are in a bad way in Japan outside the highway and night buses. Usually people just blame the declining subsidies but its more complicated than that. By the early 1980’s wages were rising such that they made buses less competitive and more expensive to operate and subsidise. Many of the leading bus operators were and are the urban private railway companies, and its telling most of them have downgraded their bus operations while increasing their rail operations. As Alon has always pointed, rail has superior economies of scale and handles increasing labour costs better.
People do rightly make negative comparisons with Seoul’s system of co-ordinated competition and revenue pooling. And certainly I think its a model that Japan’s urban prefecture should look at (looking at you Aichi). But I would highlight that Seoul is moving more and more towards heavy rail. Furthermore Korea has the worst cycling mode share in East Asia. I suspect its superior bus performance is linked to that.
Toyama proposed compact city model with infill station (“LRT-ka”) but given up due to costing too much to buy the trains for it?
The price of not intergrating/rationalising Chitetsu’s with the legacy JNR network for a S-bahn and channeling development along the rail lines. Too much of Toyama especially in the southeast is sprawl not served by an electric rail system within easy walking distance. Also some bad positioning of industrial zones.
The compact city city idea is still there, they just completed two new stations for the Hokuriku mainline for the Northeastern suburbs which again should have been built 50 years ago. The problem is you can’t force people to abandon the homes they built to live somewhere to preserve your infrastructure in the face of demographic decline. Downzoing those areas isn’t pointless but its not enough. That being said I do think Toyama is a model, its a much nicer looking city than others of its weight class. Major public facilities are reasonably consolidated in the centre, which is great for day tripping museum nuts like me.
A national government less in hoc to the construction industry would more aggressive push sprawl remediation policies in regional Japan, convert this land into either enviromental protection zones (esp along rivers) or returnit to agriculture. That would also mean creating aggressive tax subsidies for shrinking city centres. Unlike now where only parking gets that (n/b Japanese privatised parking works best when land isn’t dirt cheap, otherwise its a way of disposing of land cheaply given land comes with taxes).
And give people European length holidays so you’d get French style 2nd home culture.
I’ve always wondered if buses in other Korean cities are as good as Seoul is. Busan only just now is opening a BRT, and I have no idea how the physical infrastructure in other cities is, outside of the fact that bus terminals seem to be significant centers in, e.g., Gwangju, Cheongju.
I get the impression that, had Korea developed at the same time as Japan, it’d have much more similar rail infrastructure and urban fabric.
Korea has some clear Japanese inheritance. Inner cities have a lot of districts have Japanese-style dense low rise. Subways do through-running with surface rail etc.
But there are some real differences, Seoul didn’t have a pre-WW2 electric railway boom like Tokyo or Osaka. Korean urban development was a lot more car/bus centric as a result. The suburbs are pretty dense, but large roads tend to carve them up, so they aren’t as bike or pedestrian friendly as Japanese equivalents. Korea’s superior bus organisation and building capacity is out of necessity.
Population is only an estimate of how much “business” or “action” happens in a city. If your model overpredicts every kind of trip from Nagoya, that means Nagoya has less action than its population suggests. Indeed, it has become a running joke that Nagoya is the most boring city in Japan, the city that people want to visit the least.
This is probably also a problem for Niigata or Toyama.
Yeah because I’m sure Windermere, Penrith or tourist places in south west England get more traffic than the model predicts
That’s not a coincidence. Part of the reason Nagoya is boring is because Aichi screwed-up its heavy rail co-ordination prevented the city centre of Nagoya from developing more (compare Sakai to Umeda in Osaka). Which is why in the long run, I think Osaka will probably pull ahead of Nagoya again. Particularly if Japan’s car industry takes a hit.
Again? When can Nagoya ever be said as ahead of Osaka.
Not to mention Osaka can actually pull from the entire Kansai, Chugoku, Shikoku, and even Hokuriku area, while Nagoya can only pull from Mie, Gifu, Shizuoka.
IIRC traffic ratio of demand going from Hokuriku to Kansai vs Nagoya are 2:1, which’s already better than population would suggest for Nagoya, but still isn’t sufficient for Hokuriku to be pull toward more associating with Nagoya than Kansai, then there’s this Hokuriku Shinkansen that allow Tokyo simply pull those people into it
Nagoya/Aichi is richer than Osaka has been for decades.
https://en.wikipedia.org/wiki/List_of_Japanese_prefectures_by_GDP_per_capita
N/B These figures underrate suburban prefectures whose residents earn incomes in the central prefectures. Since 75% of the metro area of Nagoya is in Aichi compared to 50% of Keihanshin proper this actually underrates how much wealthier Nagoya is.
And population growth has Aichi ahead of Osaka/Keihanshin too.
Don’t get me wrong Osaka has more culture, history and food etc. But its got nothing like Toyota Keiretsu anymore. Osaka got too invested in small-scale artisanal manufacturing and its larger companies in relative sunset industries. Additionally Osaka was badly run for a long time which meant its local government has had a long running debt/pension overhang issue. You can see some of that in how the Osaka metro underperforms compared to Tokyo Metro because its “Monroe” doctrine prevented them from investing as much as they should have in through-running (Tanimachi line good god) while investing in duds (Imazatosuji good god). And not enough investment in universities where Osaka underperforms compared to places like Kyoto or Sendai let alone Tokyo.
I use Osaka to push people on “levelling-up” in the UK. To call Osaka the “Manchester of the Orient” today is an insult, yet it is still in relative decline. And not because they didn’t get enough money or autonomy (although they should have got more of both).
That said I think Aichi’s prosperity is fragile, dependent on single industry and a single company. So 2050 I expect Osaka to catch-up to Aichi but continue to fall behind Tokyo.
Are Chiba and Saitama really that poor, or is it all outcommuting?
@Alon, yes it is outcommuting, those two (and national-worst Nara) are “bed town” prefectures adjacent to the primary economic centers of their respective regions. Here it is presented on a map:
https://todo-ran.com/t/kiji/10709#:~:text=%E7%9C%8C%E5%86%85%E7%B7%8F%E7%94%9F%E7%94%A3%E3%81%AE%E5%85%A8%E9%83%BD,423.47%E4%B8%87%E5%86%86)%E3%81%AE%E9%A0%86%E3%80%82
Yes its outcommuting. Underlying prosperity of bed-town prefectures can be seen in their place in relative population performance
https://en.wikipedia.org/wiki/List_of_Japanese_prefectures_by_population
And real estate values too.
https://tochidai.info/
N/b the figures on this website are pretty raw (i.e. lots of non-urban land drags down the prefectural average), but one thing that does stand out is that Kansai is more expensive than it should be. That’s partly because its more land constrained than Kanto or Aichi (the largest and 2nd largest Honshu plains), but its also because Kyotoites are NIMBY tightwads and Osaka screwed up investment in its poorest Southeastern suburbs to waste it on dockland boondoggles. (this is why I rant about the Tanimachi line so much).
As a counterpoint, workers in Osaka still have a higher monthly salary than those in Aichi. Here is a spreadsheet breakdown by prefecture.
https://www.e-stat.go.jp/stat-search/files?page=1&layout=datalist&cycle=0&toukei=00450091&tstat=000001011429&tclass1=000001202310&tclass2=000001202312&tclass3=000001202331&tclass4val=0&stat_infid=000040029286
Higher GDP per capita is common in places with specialized industries, for example oil and gas industry in particular. However places that have these high value industries doesn’t mean they have higher centrality than places that are larger but have lower per capita outout.
@Alon Nara&Chiba&Saitama leak 12% of their population to other prefectures in daytime
None of this contradicts what I said. I would point to population change is as the real tell about relative prosperity of an urban area. And politics, Nagoya doesn’t have Osaka’s decline discourse. Nagoya’s mayor runs the “tax cut party” not “lets have a whatever the hell Ishin* means party”. And as I said, I think when/if Toyota’s machine enters some sort of decline, Osaka and Kansai will improve their relative position. And Osaka’s transportation policies with Naniwasuji and Monorail expansion compare very well with do-nothing Aichi. And even Kyoto is having to rezone to create more space given the ongoing tourism boom.
*N/b for non-Weeb readers Ishin in Japanese can be translated as restoration/renovation/revolution e.g. Meiji Ishin. So anybody advocating for it wants big change.
Niigata-Tokyo is right on-target, for what it’s worth.
I Japan bills just got passed, which from first look, looks like that marginal lines will be replace by buses (or alternatives, whatever that may be)
Actually I think they need more investment, local trains and trams with conductors, no IC card support, non-electrified, or just slow could become more attractive.
https://jen.jiji.com/jc/i?g=eco&k=2023042100610
They were using urban and intercity transit revenue to subsidize rural transit, but since coronavirus pandemic hit smashing revenue from urban and intercity network, there are no longer money to maintain those rural services. The train companies wanted to cut lines with passenger density below 2000 people a day but government’s new rule make this around 1000
Local lines is describing a lot here. The reality is that there isn’t a good case for lines like Kitakami line or Yamada line, very few people live along them or on either end of them. The missed/remaining opportunities in Japanese rural and regional rail are in the cities or along the mainlines. The latter is happening a bit with new stations along the mainlines in Aomori, Toyama etc.
People don’t always realise how low the population density is in Tohoku, especially one you move beyond Morioka-Sendai-Utsunomiya.
Tohoku is not really the worst, not even in Honshu. While Tohoku is of course sparsely populated, still there are large cities like Sendai or Morioka. So when all else fail, the municipal government still has money to intervene, even in very bad cases like Yamada line or Tadami line.
San’in region is even worse, being the backwaters of backwaters, inaka of inaka. The closing of Sanko line there has shocked the rail community; nobody expected that such a large amount of tracks (108km) was written off in one stroke, especially considering it happened in Honshu, not in some Hokkaido no man’s land. And the government of Shimane could do nothing to save it.
Though, it is arguable whether it is useful to keep these local lines afloat. For some lines, especially those in Hokkaido, even a bus replacement will get nearly zero ridership.
Sanin is pretty much everything wrong with Japanese regional rail. Too much wasted on lines to nowhere in the interior. The real ball game to save have a meaningful network, was a Shinkansen line to Yonago and turning the Sanin mainline into a freight bypass. Those decisions had to be made in the 1970’s given the demographic profile. That tells you just how badly the LDP ran JNR back then.
Problem is Japan’s local line were basically planned between 1890s-1920s, for an era that didn’t have cars. Various events slowed their construction that even until 1980s some of them were still going to be constructed, out of the principle of connecting different parts of the countries directly. This didn’t consider the rapidly changed need in rural area since 1950s with motorization dramatically reduced need for railroad, and development of higher speed travel mode including expressway and Shinkansen also make any possible need for such direct connection to be replaced by either direct bus or connection via bigger hubs that despite detour can still offer better travel time due to generational difference in technology involved.
While it is true that this trend was already set in the stone as of 1970s, that’s the time when both Japan’s expressway and Shinkansen were still forming their core network, thus it’s impossible for them to have resource to build Shinkansen into marginal area like San’in, and it’s also impossible in 1980s when JNR’s finance worsened massively and 1990s where they’re only restarting Shinkansen construction and extend toward Tohoku and Niigata and Nagano.
Hence it’s completely unrealistic to expect Japanese stakeholders at the time to be capable of making such decision in 1970s.
Alon’s model will normally overpredict ridership if applied to legacy rail intercity trains. But two lines stand out.
Osaka (19.3M)
– Fukui (0.7M): 2.4M model, 2.26M actual
– Kanagawa (0.7M): 2.4M model, 2.65M actual
Fukuoka (5.5M)
– Nagasaki (0.8M): 1.0M model, 2.6M actual
There’s no wonder why these two lines are being upgraded into Shinkansen.
What is the projection/expectation for Miami to Orlando?
In the future: Orlando to Tampa?
When I get a chance to translate the japanese spreadsheet, I’m going to try to fit a reduced population model (pop_A pop_B / (pop_A + pop_B)) to the data. This would help reduce the unphysical effects when splitting a region into smaller populations. For regions with comparable populations, it’ll give a result tending as (pop_A pop_B)^{~0.5}, which isn’t to dissimilar to your less than unity exponent of 0.8.
It’s a busy week, so I might only get to this on the weekend.
The order of the columns and rows in the spreadsheet is like this:
1. Northern Hokkaido
2. Eastern Hokkaido
3. Central Hokkaido
4. Southern Hokkaido
5. Hokkaido Prefecture
6. Aomori Prefecture
…
(Standard ordering used: https://en.wikipedia.org/wiki/Prefectures_of_Japan#By_Japanese_ISO )
…
50. Kagoshima Prefecture
51. Okinawa Prefecture
***
The order of the sheets is:
a) All trips
b) JR, all trips
c) JR, trips using commuter pass
d) JR, trips not using commuter pass
e) Private rail, all trips
f) Private rail, trips using commuter pass
g) Private rail, trips not using commuter pass
h) Car transit (sum of the next three categories)
i) Scheduled bus
j) Charter bus
k) Cars for hire (including taxis)
l) Scheduled ships (excluding car ferries)
m) Scheduled flights
Thanks.
Problem is since conventional rail line in Japan are slow, and expressway still isn’t that widespread in some area, it can take some places more than 2 hours or even more to travel to their nearest high speed railway station no matter by cars or trains, including for example Northern Kyoto, Northern Hyogo, Takayama City, Central/Southern Nagano, Ryomo area, Inland Fukushima, Coastal Fukushima/Iwate, Western coast of Yamagata/Aomori, and so on. Which would significantly change their demand on the high speed rail line.
Alon trained his data on the urban areas. I’ll do the same. I just want to check whether a model with better assymtotic behaviour is feasible.
*their
Certainly some places in Japan will be more than 2 hours away to any Shinkansen station, like the top of Mount Fuji or the Takamagahara Onsen. The problem though: nobody is living there.
Most of your examples are wrong. For example, Toyooka City in Northern Hyogo is certainly reachable in under two hours, starting from Himeji station. By car, take the E95 “Bantan Road”. By train, Limited Express “Hamakaze”. For Takayama City, take Ltd Exp “Hida” train to Toyama, which only takes 1.5 hours.
Tenryu River Valley in southern Nagano is pierced by Chuo Expressway and can reach Nagoya easily. The other valley in southern Nagano is serviced by “Shinano” Ltd Exp. Of course, some of the more remote villages are more remote, but very few people live in those villages.
Similarly for Fukushima, Kitakata city and Aizu-Wakamatsu City is quite close to Koriyama by car. More remote villages (like Tadami) of course exist but nobody lives there.
Western Yamagata (say, Sakata city) can reach central Yamagata pretty fast, where they can take the Yamagata Shinkansen.
***
The worst thing about your post is that there really exist cities in Japan where people need more than 2 hours to reach the nearest Shinkansen station, but you didn’t list any of them.
Kochi, Matsuyama, and Sapporo of course, but there also exist Honshu examples: Yonago city, Matsue city, Izumo city, or Shingu city (more than 200km by road from either Osaka or Nagoya!).
I am specifically talking about their distance to main Shinkansen stations of their prefecture. Of course places in Kiso and Ina valley can more conveniently access stations at Tokaido Shinkansen, but the point is this would make the calculation doesn’t work quite right when one try to assume in a formula that the entire Nagano is represented by Nagano station.
Very true, I agree.
The data for Tokyo-Nagano is especially problematic because of Karuizawa (tourist destination with disproportionate ridership) and Matsumoto (one-seat ride by Azusa).
Is that [max(d,500)]² ?
Yes.
Fantastic seeing a congenial colloquy here. Rare on the internet. Rare where most people work. Wonderful.