Amtrak Doubles Down on False Claims About Regional Rail History to Attack Through-Running

Amtrak just released its report a week and a half ago, saying that Penn Expansion, the project to condemn the Manhattan block south of Penn Station to add new tracks, is necessary for new capacity. I criticized the Regional Plan Association presentation made in August in advance of the report for its wanton ignorance of best practices, covering both the history of commuter rail through-running in Europe and the issue of dwell times at Penn Station. The report surprised me by making even more elementary mistakes on the reality of how through-running works here than the ones made in the RPA presentation. The question of dwell times is even more important, but the Effective Transit Alliance is about to release a report addressing it, with simulations made by other members; this post, in contrast, goes over what I saw in the report myself, which is large enough errors about how through-running works that of course the report sandbags that alternative, less out of malice and more out of not knowing how it works.

Note on Penn Expansion and through-running

In the regional discourse on Penn Station, it is usually held that the existing station definitely does not have the capacity to add 24 peak trains per hour from New Jersey once the Gateway tunnel opens, unless there is through-running; thus, at least one of through-running and Penn Expansion is required. This common belief is incorrect, and we will get into some dwell time simulations at ETA.

That said, the two options can still be held as alternatives to each other, even as what I think is likeliest given agency turf battles and the extreme cost of Penn Expansion (currently $16 billion) is that neither will happen. This is for the following reasons:

  • Through-running is good in and of itself, and any positive proposal for commuter rail improvements in the region should incorporate it where possible, even if no dedicated capital investment such as a Penn Station-Grand Central connection occurs. This includes the Northeast Corridor high-speed rail project, which aims to optimize everything to speed up intercity and commuter trains at minimal capital cost.
  • The institutional obstacles to through-running are mainly extreme incuriosity about rest-of-world practices, which are generations ahead of American ones in mainline rail; the same extreme incuriosity also leads to the belief that Penn Expansion is necessary.
  • While it is possible to turn 48 New Jersey Transit trains per hour within the current footprint of Penn Station with no loss of LIRR capacity, there are real constraints on turnaround times, and it is easier to institute through-running.

The errors in the history

The errors in the history are not new to me. My August post criticizing the RPA still stands. I was hoping that Amtrak and the consultants that prepared the report (WSP, FX) would not stick to the false claim that it took 46 years to build the Munich S-Bahn rather than seven, but they did. The purpose of this falsehood in the report is to make through-running look like a multigenerational effort, compared with the supposedly easier effort of digging up an entire Manhattan block for a project that can’t be completed until the mid-2030s at the earliest.

In truth, as the August post explains, the real difficulties with through-running in the comparison cases offered in the report, Paris and Munich, were with digging the tunnels. This was done fairly quickly, taking seven years in Munich and 16 in Paris; in Paris, the alignment, comprising 17 km of tunnel for the RER A and 2 for the initial section of the RER B, was not even finalized when construction began. The equivalent of these projects in New York is the Gateway tunnel itself, at far higher cost. The surface improvements required to make this work were completed simultaneously and inexpensively; most of the ones required for New York are already on the drawing board of New Jersey Transit, budgeted in the hundreds of millions rather than billions, and will be completed before the tunnel opens unless the federal government decides to defund the agency over several successive administrations.

The errors in present operations

The report lists, on printed-pp. 40-41, some characteristics of the through-running systems used in Paris, Munich, and London. Based on those characteristics, it concludes it is not possible to set up an equivalent system at Penn Station without adding tracks or rebuilding the entire track level with more platforms. Unfortunately for the reputation of the writers of the report, and fortunately for the taxpayers of New York and New Jersey, those characteristics include major mistakes. There’s little chance anyone in the loop understands the RER, any S-Bahn worth the name, or even Crossrail and Thameslink; some of the errors are obviously false to anyone who regularly commuted on any of these systems. Thus, they are incapable of adjusting the operations to the specifics of Penn Station and Gateway.

Timetabling

A key feature of S-Bahn systems is that the trains run on a schedule. Passengers riding on the central trunk do not look at the timetable, but passengers riding to a branch do. I memorized the 15-minute off-peak Takt on the RER B when I took it to IHES in late 2016, and the train was generally on time or only slightly delayed, never so delayed that it was early. Munich-area suburbanites memorize the 20-minute Takt on their S-Bahn branch line. Some Thameslink branches drop to half-hourly frequency, and passengers time themselves to the schedule while operators and dispatchers aim to make the schedule.

And yet, the report repeatedly claims that these systems run on headway management. The first claim, on p. 40, is ambiguous, but the second, on the table on p. 41, explicitly contrasts “headway-based” with “timetable-based” service and says that Crossrail, the RER, and the Munich S-Bahn are headway-based. In fact, none of them is.

This error is significant in two ways. First, timetable-based operations explain why S-Bahn systems are capable of what they do but not of what some metros do. The Munich S-Bahn peaks at 30 trains per hour, with one-of-a-kind signaling; major metros peak at 42 trains per hour with driverless operations, and some small operations with short trains (like Brescia) achieve even more. The difference is that commuter rail systems are not captive metro trains on which every train makes the same stops, with no differentiation among successive trains on the same line; metro lines that do branch, such as M7 and M13 in Paris, are still far less complex than even relatively simple and metro-like lines like the RER A and B. The main exception among world metros is the New York City Subway, which, due to its extensive interlining, must run as a scheduled railroad, benchmarking its on-time performance (OTP) to the schedule rather than to intervals between trains. In the 2000s and 10s, New York City Transit tried to transition away from end-station OTP and toward a metric that tried to approximate even intervals, called Wait Assessment (WA); a document leaked to Dan Rivoli and me went over how this was a failure, leading to even worse delays and train slowdowns, as managers would make the dispatchers hold trains if the trains behind them were delayed.

The second consequence of the error is that the report does not get how crucial timetable-infrastructure planning integration is on mainline rail. The Munich S-Bahn has outer branches that are single-track and some that share tracks with freight, regional, and intercity trains. The 30 tph trunk does no such thing and could not do such thing, but the branches do, because the trains run on a fixed timetable, and thus it is possible to have a mix of single and double track on some sporadic sections. The Zurich S-Bahn even runs trains every 15 minutes at rush hour on a short single-track section of the Right Bank of Lake Zurich Line. Recognizing what well-scheduled commuter trains can and can’t do influences infrastructure planning on the entire surface section, including rail-on-rail grade separations, extra tracks, yard expansions, and other projects that collectively make the difference between a rail network and crayon.

Separation between through- and terminating lines

Through-running systems vary in how much track sharing there is with the rest of the mainline rail network. As far as I can tell, there is always some; near-complete separation is provided on the RER A, but its Cergy branch also hosts Transilien trains running to Gare Saint-Lazare at rush hour, and the Berlin and Hamburg S-Bahn systems have very little track-sharing as well. Other systems have more extensive track sharing, including Thameslink, the RER C and D, and the Zurich S-Bahn; the RER E and the Munich S-Bahn are intermediate in level of separation between those two poles.

It is remarkable that, while the RER A, B, and E all feature new underground terminals for dedicated lines, the situation of the RER C and D is different. The RER C uses the preexisting Gare d’Austerlitz, and has taken over every commuter line in its network; the through-connection between Gare d’Orsay and Gare d’Invalides involved reconstructing the stations, but then everything was connected to it. The RER D uses prebuilt underground stations at Gare du Nord, Les Halles, and Gare de Lyon, but then takes over nearly all lines in the Gare de Lyon network, with the outermost station, Malesherbes, not even located in Ile-de-France. Thameslink uses through-infrastructure built in the 1860s and runs as far as Petersborough, 123 km from King’s Cross on the East Coast Main Line, and Brighton, the terminus of its line, 81 km from London Bridge.

And yet, the report’s authors seem convinced the only way to do through-running is with a handful of branches providing only local service, running to new platforms built separately from the intercity terminal; they’re even under the impression the RER D is like this, which it is not. There’s even a map on p. 45, suggesting a regional metro system running as far as Hicksville, Long Beach, Far Rockaway, JFK via the Rockaway Cutoff and Queenslink, Port Washington, Port Chester, Hackensack, Paterson, Summit, Plainfield, New Brunswick, and the Amboys. This is a severe misunderstanding of how such systems work: they do not arbitrarily slice lines this way into inner and outer zones, unless there is a large mismatch in demand, and then they often just cut the outer end to a shuttle with a forced transfer, as is the case for some branches in suburban Berlin connecting to S-Bahn outer ends. Among the above-mentioned outer ends, the only one where this exception holds is Summit, where the Gladstone Branch could be cut to a shuttle or to trains only running to Hoboken – but then trains on the main line to Morristown and Dover have no reason to be treated differently from trains to Summit.

Were the report’s authors more informed about just the specific lines they look at on p. 41, let alone the broader systems, they’d know that separation between inner and outer services is contingent on specifics of track infrastructure, including whether there are four-track lines with neat separation into terminating express trains and through locals. But even if the answer is yes, as at Gare de Lyon and Gare d’Austerlitz, infrastructure planners will attempt to shoehorn whatever they can into the system, just starting from the more important inner lines, which generate more all-day demand. There don’t even need to be terminating regional trains; the Austerlitz system doesn’t, and the Gare de Lyon and Gare de l’Est systems only do due to trunk capacity limitations. In that case, they’d recognize that there is no need to have two commuter rail systems, one through-running and one not. Penn Station’s infrastructure already lends itself to allowing through-running on anything entering via the existing North River Tunnels.

Branching

S-Bahn systems usually try to keep the branch-to-trunk ratio to a manageable number. Usually, more metro-like systems have fewer branches: Crossrail has two on each side, the RER A has two to the east and three to the west, the Berlin Stadtbahn has two to the west plus short-turns and five to the east, the Berlin North-South Tunnel has three on each side. The Munich S-Bahn has five to the east and nine to the west, and the combined RER B and D system has three to the north and five to the south, but the latter has more service patterns, including local and express trains on the branches. Zurich has so much interlining that it’s not useful to count branches, and better to count services: there are 21 S-numbered routes serving Hauptbahnhof, of which 13 run through one of the two tunnels, as do some intercity trains.

If there are too many branches, then they’re usually organized as sub-branches – for example, Munich has seven numbered routes through the central tunnel, of which two have two sub-branches each splitting far out. Zurich has fewer than 13 branches on each side, but rather there are several services using each line, with inconsistent through-pairing – for example, the three services going to the airport, S2, S24, and S16, respectively run through to two separate branches of the Left Bank Line and to the Right Bank Line.

The table on p. 41 gets the branch count mildly wrong, but the significant is less in what it gets wrong about Europe and more in what it gets wrong about New York. A post-Gateway service plan is one in which New Jersey has 12 branches, but some can be viewed as sub-branches (like Gladstone and the Morristown Line), and more to the point, there are going to be two trunk lines. The current plan at New Jersey Transit is to assign the Northeast Corridor and North Jersey Coast Lines to the North River Tunnels alongside Amtrak, which is technically two branches but realistically four or even five service patterns, and the Morris and Essex, Montclair-Boonton, and Raritan Valley Lines to Gateway, which is four branches but could even be pruned to three with M&E divided into two sub-branches. The Erie lines have no way of getting to Penn Station today; to get them there requires the construction of the Bergen Loop at Secaucus, with an estimated budget of $1.3 billion in 2020, comparable to the total cost of all yet-unfunded required surface improvements in New Jersey for non-Erie service combined.

If the study authors were more comfortably knowledgeable of European S-Bahn systems, they’d know that multi-line systems, while uncommon, do exist, and divide branches in a similar way. The multiline systems (Paris, Madrid, Berlin, Zurich, and London) all have some reverse-branching, in a similar manner to how New York is soon going to have the New Haven Line reverse-branch to Penn Station and Grand Central. The NJT plan is solid and stands to lead to a manageable branch-to-trunk ratio, even with every single line going to Penn Station via the existing tunnel running through.

The consequence of the errors

The lack of familiarity with through-running commuter rail is evident in how the report talks about this technology. It is intimately related to the fact that the way investment should be done is different from what American railroaders are used to. For one, there needs to be much tighter integration between infrastructure and scheduling. For two, the scheduling needs to be massively simplified, with fewer operating patterns per line – usually one, occasionally two, never 13 as on the New Haven Line today. The same ignorance that leads Amtrak and its consultants to assert that the S-Bahn runs on headway management rather than a fixed timetable also leads them not to even know how through-running commuter rail networks plan out their routes and services.

From my position of greater familiarity as both a regular user and a researcher, I can point out that the required investments to make through-running happen in New York are entirely in line with the cheap surface projects done in the comparison cases. New rolling stock is required, with the ability to run on the different voltages of the three networks – but multi-voltage commuter rolling stock is the norm wherever multiple legacy electrification systems coexist, including Paris, London, and Hamburg. Some extensions of electrification and high platform conversions are required – but these are not expensive, and the latter is already partly funded at reasonable unit costs. Some rail-on-rail grade separations are required – but those are already costed and very likely to be funded, potentially out of the Bipartisan Infrastructure Law.

Penn Station would be used as the universal station in this schema, without the separation into a surface terminal and a through- underground station seen in Munich and Paris. But then, Paris and Munich don’t even universally have this separation themselves; Ostbahnhof was reconstructed for the S-Bahn but is still a single station, and the same is true of the RER C. In a way, Penn Station already is the underground through-station, built generations before the modern S-Bahn concept, complementing and largely replacing surface terminals like Hoboken and Long Island City because those are not in Manhattan.

None of this is hard; the hard part is the Gateway tunnel and that’s already fully funded and under construction. But it does require understanding that the United States is so many decades behind best practices that none of what American railroaders think they know is at all relevant. It’s obligatory to understand how the systems that work, in Europe and rich Asia, do, because otherwise, it’s like expecting someone who has never learned to count beyond 10 to prove mathematical theorems. The people who wrote this report clearly don’t have this understanding, and don’t care to get it, which is why what they write is not worth the electrons that make up the PDF.

Taxes are not About Urbanism

I caused a ruckus on social media when I pointed out that a Venn diagram commonly posted on Strong Towns and by American urbanists in general about taxes, services, and urbanism, is completely false:

The truth is that the overwhelming majority of taxes and services are about things that don’t meaningfully depend on local density. The overwhelming majority of those services are not even locally provided. In the United States, in 2022, taxes split as 67.5% federal, 19.1% state, 13.4% local. In Germany, in theory the division is similar, but in practice 95% of taxes are levied federally, but a portion are distributed to the states by formula.

Most of what all of this goes to is welfare programs. La Protection Sociale en France et en Europe puts the proportion of GDP going to social protection at 32% in France, 29% in Germany, and 27% Union-wide, all as of 2022. This includes programs whose recipients loudly insist that they are not welfare, including old age pensions and health care, comprising 82% of the above total. Government spending ranges around 50% of GDP here; France ranks first in the OECD, at 58%, so that its share of government spending going to social protection, slightly more than half, is if anything a little less than the European average. In the United States, these programs (Social Security, Medicare, Medicaid, income security) comprise two-thirds of non-interest federal spending, and one third of state and local spending. A couple additional percent – more in the United States than in Europe with one notable exception – is the military plus veteran-specific welfare in the US. None of this has anything to do with how much urban sprawl there is, or how municipal finances are structured.

Local taxes in the United States mostly go to schools. Education expenses are a combination of teacher and administrator salaries and facilities spending; these are not about density. Entire comparative tomes have been written about education funding, none touching density, because it doesn’t matter; for example, the OECD’s Education at a Glance database goes into per-student spending, teacher salaries, administrator salaries, student-teacher ratios, the public-private school division, the vocational-academic track division, graduation rates, the earnings advantages of graduates at various levels, and correlations with civic participation. The most notable features of the United States are high educational segregation, and low teacher salaries compared with the average for university-educated workers. This is not an urbanist issue; American urbanist activists have been spending generations avoiding talking about school segregation, to the point that the activists who do talk about education from any angle are more or less disjoint from the ones who talk about recognizably urbanist issue. The Venn diagram above is a good example of this: instead of talking about curricular rigor or teachers or segregation or how schools are funded, urbanists prefer to pretend reducing urban sprawl is the solution (not mentioned: any of the problems internal to, say, New York City public schools).

Once we get to items that do depend on what urbanists care about, it’s at the bottom of the list of budgetary priorities. Transportation costs are a small share of GDP; in the US, an extreme case in road costs, they’re 2% of federal spending and 5.6% of subnational spending, which averages out to somewhat more than 3% of GDP. German federal investment in roads is around 0.25% of GDP, and federal investment in rail is similar but planned to be double that in 2026-27. I haven’t seen more recent numbers, but in 2018, public transport finances across Germany amounted to 14.248 billion € in costs and 7.363 billion € in fare revenue; the difference, about 7 billion €, was 0.2% of 2018 GDP, covering about 16% of work trips Germany-wide.

So, yes, reduced sprawl would reduce the required spending on roads, but it’s not going to be especially visible in tax bills, which are dominated by other things. The biggest drivers of the cost of private transportation are the costs of buying, maintaining, and fueling the car. Those drive American spending on transportation to 17% of household spending, of which 8.3% are not cars, and buying the car and fuel for it are two thirds of the rest (that is, 10.3% of household spending). This is high per capita and per GDP; it is also not what urbanists who try to calculate operating ratios for roads talk about – the problem with private transportation is not the public cost of the roads, but the private cost to households.

Then there’s policing. Strong Towns has a particular infamy on the subject – Chuck Marohn essentially canceled himself among anyone concerned with civil rights when, during the initial Black Lives Matter protests in 2014, his take was about the lack of placemaking in suburbs like Ferguson, requiring people to protest in the middle of roads. On social media, multiple people insisted to me that higher density permits higher-quality (?) or lower-cost policing; at least as far as police brutality goes, the pattern is that it happens 1-2 orders of magnitude more in the US than in the rest of the developed world, and that in the US, it happens to black people about three times more than to white people, and it happens less in the Northeast than elsewhere, even in blacker-than-average places like Buffalo or Philadelphia. Urban departments are probably somewhat more professional than rural sheriffs’ departments, but this probably isn’t really a matter of suburban sprawl development but jurisdiction size, and is also not at all what the Venn diagram is trying to say. “Big central cities that have dense cores and can annex suburbs end up having very high police spending but get more professional departments out of this” is nowhere in the Venn diagram. It is not implied in it. It is directly opposed to what Strong Towns says about municipal finances, which is about long-term budgeting for municipalities with given borders and not about the desirability of delocalizing decisionmaking.

It’s always tempting to try to turn one’s issue into an omnicause. High taxes and poor services? Sprawl. Police violence? Sprawl. Schools? Sprawl. High housing costs? Sprawl. Government corruption? Sprawl. It creates an illusion of seriousness, of solving systemic social problems, while really only droning about one peripheral issue. It creates an illusion of outside-the-box solutions, while really ignoring real tradeoffs, in this case between taxes and government services. And it creates an illusion of solidarity and alliance between different advocacy groups, while really giving activists for one cause permission structure to bring up their pet cause no matter how unimportant it is in the moment.

Mass Transit on Orbital Boulevards

Herbert in comments has been asking me about urban rail on ring roads; Nuremberg has such a road with an active debate about what to do with it. Ring roads are attractive targets for urban rail, since they tend to be wide commercial throughfares. The one in Nuremberg is especially attractive for a tramway, or possibly a medium-capacity metro if one can be built cheaply; this is an artifact of its circumference (18 km) and the city’s size, reminiscent of the Boulevards of the Marshals hosting Paris Tramway Line 3, and the Cologne Gürtel, most of whose length has a tramway as well. Significantly closer-in ring roads, often delineating the medieval or Early Modern walls, are too small for this.

The history of such rings tends to be that they were built based on the extent of the industrial city. Cologne’s was built in the 19th century to connect growing bedroom communities to one another, where they previously only extended along the radial boulevards connecting them to the historic center. The Boulevards of the Marshals delineated the inner end of the Thiers wall from the 1840s; the Périphérique motorway is where the outer end had been. The upshot is that the construction standards are rather modern – for one, the roads are wide. Another upshot is that those roads are often destinations in and of themselves, so that radial rail lines have stops at them; the Métro has stops at every intersection with the Boulevards of the Marshals, generally named after the nearby gate (for example, I lived near Porte de Vincennes, due east along Métro Line 1).

This contrasts with older rings, including one visible on the screenshot above. Those older rings come from premodern city walls, and may not always have enough width to make it easy to build two tram lanes in the center or to do cheap cut-and-cover without disturbing the residences and businesses too much. Even when they do, they’re so close to the center the time savings from a ring at that radius are moderate. Jarrett Walker has long pointed out that people don’t travel in circles, giving the example of the Vienna Ring Road, which has two U-Bahn lines on different sections of it but no continuous ring, as a 5.3 km circle is too small to have viable long relatively linear sections. In Paris, old boulevards closer in than the ring forming Métro Lines 2 and 6 generally have Métro stops but it’s inconsistent, and there’s no coherent circular route to be built.

The modal question – tram or metro – is complicated by special elements of orbital boulevards, which sometimes cancel out, and can work differently in different cities.

In favor of light rail, there’s the issue of speed. Normally, the advantage of subways over tramways is that they’re faster. However, on a circumferential route, the importance of speed is reduced, since people are likely to only travel a relatively short arc, connecting between different radials or from a radial to an off-radial destination. What are more important than speed on such a route are easy transfers and high frequency. Easy transfers could go either way: if the radial routes are underground then it may be possible to construct underground interchanges with short walking, but it isn’t guaranteed, and if there are any difficulties, it’s better to keep it on the surface to shorten the walk time. This has in general been an argument used by pro-tram, anti-subway advocates in Germany, but on routes that rely on multiple transfers, potentially three-legged trips, it is a stronger argument than on a radial line from a suburban housing project to city center.

Frequency is especially delicate. It can be high regardless of mode. Driverless metros can reach 90-second headways or even less, but those are achieved on very busy lines, which need that frequency for throughput more than anything, like Lines 1 and 14 in Paris with their 85-second peak headways. In practice, an orbital tram, especially one in a smaller city than Paris, needs to be prioritizing frequency in order to shorten the trip, not to provide very high throughput, which means that the vehicles could be made smaller than full-size metros, to support frequency in the 3-6 minute range. This could be done at-grade with light rail, or underground with very small-profile metros akin to those used in small Italian cities like Brescia, or even some larger ones like Turin.

In favor of metro, there is the cost issue. The same factors that make speed less important and frequency more important also make it easier to build a metro. If the road is wide enough, which I think the one in Nuremberg is, then cut-and-cover is more feasible, reducing costs. The low required capacity permits intermediate-capacity metros (again, as in Brescia or some smaller French cities), with stations of perhaps 40-50 meters, reducing their construction costs. Nuremberg in particular has had some very low U-Bahn construction costs, so its ability to build an orbital U-Bahn should not be discounted. That said, even at Nuremberg costs – around $100 million/km in 2023 PPPs for U3 extensions – the extra speed provided by such a line, say half an hour to do a full orbit compared with a little less than an hour on a tram, may not be worth it necessarily, whereas such a speedup on a line that passengers may ride for 10 km unlinked would be extremely beneficial.

InnoTrans is Souring Me on On-Rail Competition

I’m at InnoTrans this week, which means I get to both see a lot of new trainsets and talk to vendors for things I am interested in. Those are interesting conversations and much of the content will make it to our upcoming report on high-speed rail in the Northeast and to some ETA reports. But then, in broad stroke, the presentations about the trains here have deeply bothered me, because of how they interact with the issue of on-rail competition. The EU has an open access mandate, so that state-owned and private railways can compete by running trains on tracks throughout the Union, with separation of operations and infrastructure (awkwardly, state railways do both but there are EU regulations prohibiting favoritism, with uneven enforcement). As a result, I’m seeing a lot of pitches geared specifically for potential open access operators, all of which remind me why I’m so negative about the whole concept: it treats infrastructure as a fixed thing and denigrates the idea that it could ever be improved, while enthroning airline-style business analysis.

Proponents of the model cite higher ridership and lower fares due to the introduction of competition in Italy and Spain, but even then, it’s never really invented anything new, and only gotten some city pairs in those two countries to the service quality that integrated state-provided services have always had in France and Germany. In effect, the EU is mandating a dead-end system of managing trains and making a collective decision not to invest in what worked – namely, building and running high-speed lines.

For people unfamiliar with the argument, I wrote a year ago about TGV ridership and traffic modeling. The TGV overperforms a model trained on Shinkansen ridership, which can be explained based on lower fares, leisure travel to the Riviera, and underdeveloped air competition in small metro areas whose residents mostly drive to larger ones to fly. Relative to the same model, Italian and Spanish ridership underperformed the TGV before competition, and rose to roughly match the TGV or be slightly deficient after competition. So competition did lead to growth in ridership, as the competitors added service and lowered fares – but it only created what the French state did by itself. The German state seems to have French-like results: the trains here are much slower than in France, but relative to that, ridership seems to be in line with a TGV-trained model on the handful of city pairs for which I have any data.

This is causing quite a lot of the buzz in the intercity rail industry in Europe to center cross-border competition and new entrants. But this is, judging by the examples the proponents of competition look up to, not creating anything new. It’s not moving rail forward. It’s just filling in gaps that some state-owned railways – but not the largest two – have in their operations.

And worse, it’s making the long-term issues of intercity rail in Europe worse. There’s practically no cross-border high-speed rail construction in Europe, nor any serious push for making it happen. After a great deal of activism by Jon Worth (and others), the European Commission is announcing regulatory measures in its agenda, starting with passenger rights in case of delays. Physical construction is nowhere on the horizon, nor is there any serious advocacy for (say) a Paris-Frankfurt high-speed line or a Bordeaux-Basque Country one. This is a recent development: in the 2000s there was more optimism about high-speed rail, leading to plans like Perpignan-Figueres. But since then, the TEN-T corridor plans turned into low-speed lines and vaporware, and there’s no real interest in fixing that.

Instead, the interest is in letting the private sector lead. State-built high-speed railways – more or less the only high-speed railways – are not in fashion. The private sector is not going to step in and build its own (despite the sad Hyperloop capsule on display at InnoTrans), but instead look for underserved city pairs to come into, competing with state railways. It’s a story of business analysts using techniques brought in from the airline industry rather than one of infrastructure builders.

And it’s exactly those airline-imitating business analysts who are why RENFE, FS, and Eurostar underprovided service to begin with. The airline world lives off of segmenting the market; there are periodic attempts at all-business class airlines, and low-cost carriers entering and exiting the market frequently. It does not build its own infrastructure, not think in terms of things that could work if the infrastructure were a little bit better. A railway that thinks in the same terms might still build, but will not build in coordination with what it runs. It will do the exact opposite of what Switzerland has done with its tight integration of infrastructure and operations planning; therefore, it will get results inferior to those of Switzerland or even France and Germany.

The trains on display at InnoTrans announce proudly that they are homologated for cross-border travel, listing the countries they can operate in. The main high-speed rail vendors here – Siemens, Hitachi, Talgo, Alstom – all talk about this, explicitly; Alstom had a presentation about the Avelia Horizon, awkwardly given in an American accent while talking about how the double-decker cars with 905 mm seat pitch (Shinkansen: 1 meter) minimize track access charges per seat.

In contrast, I have not seen anything about building new lines. I have not seen booths by firms talking about their work building LGVs or NBSes. I have not seen anything by ADIF selling its expertise in low-cost construction; there are some private engineering consultants with booths, but I haven’t seen ADIF, and the French state section of the conference didn’t at all center French construction techniques. The states that have figured out how to build high-speed rail efficiently seem uninterested in doing more with it than just completing their capital-to-provinces networks; even Germany is barely building. Naturally, they’re also uninterested in pitching their construction, even though they do do some public-sector consulting (SNCF does it routinely for smaller French cities). It’s as if the construction market is so small they’re not even going to bother.

Every other booth at the conference talks about innovation with so many synonyms that they swamp what the firm actually does. But beneath the buzzwords, what I’m seeing, at least as far as physical infrastructure goes, is the exact opposite of innovation. I’m seeing filling in small gaps caused by last generation’s bad airline imitation with a different kind of airline imitation, and nothing that moves intercity rail forward.

Quick Note on Capital City Income Premiums

There’s a report by Sam Bowman, Samuel Hughes, and Ben Southwood, called Foundations, about flagging British growth, blaming among other things high construction costs for infrastructure and low housing production. The reaction on Bluesky seems uniformly negative, for reasons that I don’t think are fair (much of it boils to distaste for YIMBYism). I don’t want to address the construction cost parts of it for now, since it’s always more complicated and we do want to write a full case on London for the Transit Costs Project soon, but I do want to say something about the point about YIMBYism: dominant capitals and other rich cities (such as Munich or New York) have notable wage premiums over the rest of the country, but this seems to be the case in NIMBY environments more than in YIMBY ones. In fact, in South Korea and Japan, the premium seems rather low: the dominant capital attracts more domestic migration and becomes larger, but is not much richer than the rest of the country.

The data

In South Korea and Japan, what I have is Wikipedia’s lists of GDP per capita by province or prefecture. The capital city’s entire metro area is used throughout, comprising Seoul, Incheon, and Gyeonggi in Korea, and Tokyo, Kanagawa, Chiba, and Saitama in Japan. In South Korea, the capital region includes 52.5% of GDP and 50.3% of population, for a GDP per capita premium of 4.4% over the country writ large, and (since half the country is metro Seoul) 9.2% over the rest of the country. In Japan, on OECD numbers, the capital region, labeled as Southern Kanto, has a 14.5% premium over the entire country, rising to around 19% over the rest of the country.

In contrast, in France, Ile-de-France’s premium from the same OECD data is 63% over France, rising to about 90% over provincial France. In the UK, London’s premium is 71% over the entire country and 92% over the entire country excluding itself; if we throw in South East England into the mix, the combined region has a premium of 38% over the entire country, and 62% over the entire country excluding itself.

Now, GDP is not the best measure for this. It’s sensitive to commute volumes and the locations of corporate headquarters, for one. That said, British, French, Korean and Japanese firms all seem to prefer locating firms in their capitals: Tokyo and Seoul are in the top five in Fortune 500 headquarters (together with New York, Shanghai, and Beijing), and London and Paris are tied for sixth, with one company short of #5. Moreover, the metro area definitions are fairly loose – there’s still some long-range commuting from Ibaraki to Tokyo or from Oise to Paris, but the latter is too small a volume to materially change the conclusion regarding the GDP per capita premium. Per capita income would be better, but I can only find it for Europe and the United States (look for per capita net earnings for the comparable statistic to Eurostat’s primary balance), not East Asia; with per capita income, the Ile-de-France premium shrinks to 45%, while that of Upper Bavaria over Germany is 39%, not much lower, certainly nothing like the East Asian cases.

Inequality

Among the five countries discussed above – Japan, Korea, the UK, Germany, France – the level of place-independent inequality does not follow the same picture at all. The LIS has numbers for disposable income, and Japan and Korea both turn out slightly more unequal than the other three. Of course, the statistics in the above section are not about disposable income, so it’s better to look at market income inequality; there, Korea is indeed far more equal than the others, having faced so much capital destruction in the wars that it lacks the entrenched capital income of the others – but Japan has almost the same market income inequality as the three European examples (which, in turn, are nearly even with the US – the difference with the US is almost entirely redistribution).

So it does not follow, at least not at first pass, that YIMBYism reduces overall inequality. It can be argued that it does and Japan and South Korea have other mechanisms that increase market income inequality, such as weaker sectoral collective bargaining than in France and Germany; then again, the Japanese salaryman system keeps managers’ wages lower than in the US and UK and so should if anything produce lower market income inequality (which it does, but only by about three Gini points). But fundamentally, this should be viewed as an inequality-neutral policy.

Discussion

What aggressive construction of housing in and around the capital does appear to do is permit poor people to move to or stay in the capital. European (and American) NIMBYism creates spatial stratification: the middle class in the capital, the working class far away unless it is necessary for it to serve the middle class locally. Japanese and Korean YIMBYism eliminate this stratification: the working class keeps moving to (poor parts of) the capital region.

What it does, at macro level, is increase efficiency. It’s not obvious to see this, since neither Japan nor Korea is a particularly high-productivity economy; then again, the salaryman system, reminiscent of the US before the 1980s, has long been recognized as a drag on innovation, so YIMBYism in effect countermands to some extent the problems produced by a dead-end corporate culture. It also reduces interregional inequality, but this needs to be seen less as more opportunity for Northern England as a region and more as the working class of Northern England as people moving to become the working class of London and getting some higher wages while also producing higher value for the middle class so that inequality doesn’t change.

The MTA Capital Plan Falsifies Subway History

The 2025-29 capital plan is out, and it is not good. There’s an outline of an ETA report to be released soon going over issues like accessibility, rolling stock costs, and the new faregates. But for now, I’d like to just focus on a high-level issue and how it relates to the subway’s history: State of Good Repair. The capital plan has a summary history of past capital plans on PDF-page 8, and it calls the 1990s and 2000s an era of underinvestment and deferred maintenance, the exact opposite of reality. It treats 2017 as a keystone year for system renewal, which it was not; it was, however, the year current MTA chair Janno Lieber was hired as the head of MTA Construction and Development (formerly Capital Construction). In effect, the plan falsifies the history of the system in order to treat the current leadership as saviors, in service of a plan to spend more money than in 2020-24 while having less to show for it, washing it all with the nebulous promise of State of Good Repair.

The history of State of Good Repair

Traditionally, capital investment is conceived as going to expansion. In New York in the first two thirds of the 20th century, this meant new subway and elevated lines, new connections between subway lines, station upgrades to lengthen the platforms, and new transfers between stations that had previously belonged to different operators. Maintenance was treated as an ongoing expense.

The finances of the subway after WW1 were shaky, and from the Depression onward, it never made money again. Of the two private operators, one, the IRT, was in bankruptcy protection during the Depression, while the public operator, the IND, was debt-ridden due to exceptionally high construction costs for that era and overbuilding. This made it attractive to defer maintenance, on the subway as on mainline rail everywhere in the United States. In 1951, bond money was designated for Second Avenue Subway, but then the money was raided for other priorities, including smaller extensions but also capital renewal, such as replacing the almost 50-year-old IRT rolling stock fleet.

In the 1970s, the city’s poor finances meant it couldn’t subsidize operations and maintenance as much as before, and the maintenance deferral led to a systemwide collapse. NYCSubway.org goes over the various elements of it: chunks of equipment and material were falling onto the street from the elevated lines, and onto the tracks from the retaining walls of open cuts; trains had flat wheels and no lubrication, leading to such squeal that the noise was worse than that of Concorde; train doors and lights malfunctioned; derailments and fires were common. By 1981, the mean distance between failures (MDBF) dropped to its lowest ever, 6,640 miles (10,690 km). One third of the system was under emergency 10 mph speed restrictions, and a quarter of the rolling stock had to be kept in reserve to substitute for equipment failures and could not be run in maximum revenue service. The new trains bought for the system, the R44 and R46, used new technology, for example higher top speed for the use on long express sections, but were defective to the point that the lawsuits against the vendors, St. Louis and Pullman respectively, bankrupted them. The origin of the conservatism of rolling stock orders and the pattern that all American rolling stock manufacture is done at transplant factories owned by European, Japanese, or formerly Canadian firms, are both the result of this history.

The State of Good Repair program as we know it dates to the 1980s, when the MTA, starting with the leadership of Richard Ravitch, began to prioritize maintenance and renewal over expansion. This meant five-year capital programs, to reduce the incentives to defer maintenance in a single year, and a lot of openly crying poverty, where leaders both before and after Ravitch would prefer to extoll the system and downplay its shortcomings. There was large spending on capital as a result, but no Second Avenue Subway. Instead, money went to renewal. Rolling stock was more conservative; the R62 was also imported from Japan, since Reagan cut federal aid to mass transit and so the MTA was free from Buy America’s strictures (in contrast, today states prefer to preemptively obey even when they’re not sure they will get federal funding, and even demand in-state plants). Its mean distance between failures was far higher than that of all other rolling stock, and this greater reliability continued into the R62A, R68, and R68A orders; the systemwide mean distance between failures kept climbing throughout the 1980s, 1990s, and 2000s, to a peak of around 180,000 miles, or around 280,000 km, in 2005 and again in 2010-11. The slow restrictions that characterized the system in the 1970s were lifted, and rolling stock availability for maximum service rose.

The construction of Second Avenue Subway beginning in 2007 was not viewed as a rebuke to the SOGR program, but rather as the legacy of its success. Leaders like Lee Sander spoke of growth and new lines, setting the stage for what is now known as IBX and was known in then as Triboro RX. The political discourse in the United States at the time was one of transit revival, due to the then-new decoupling of car driving and oil use from economic growth, and the high fuel prices; this was also around the time American discourse discovered European and Japanese high-speed rail, setting the stage for Proposition 1A approving the construction of California High-Speed Rail in 2008.

The present of State of Good Repair

I’ve repeatedly criticized SOGR as a scheme allowing agency heads to demand money with nothing to show for it. The behavior of current MTA leadership is one such example; it is not the only one – Amtrak did the same under Joe Boardman in the 2000s. But it needs to be made clear that the SOGR program of the 1980s and 90s was an unmitigated success. There was visible improvement in the system due to better maintenance of fixed plant and more prudent capital investments, such as the trainsets bought in the era from the R62 to the R160. The present problems of the SOGR concept come essentially because its success in the 1980s and 90s led agencies to talk about it as the next hot thing, even while going in a rather different direction.

In the 2010s, the subway started facing new problems – but these were not problems of undermaintenance. The MDBF crept down to a little less than 120,000 miles at the bottom, in 2017, and was 125,000 miles in 2023. The oldest trains are the worst, but much of the problem comes from other issues than slow replacement of fixed plant. For example, the ongoing slowdowns on the subway – even in the 2000s it was slower than before the 1970s collapse, and speeds are noticeably lower when I visit than when I lived in the city in 2006-11 – come not from insufficient maintenance, but from tighter flagging rules, which are designed to protect workers on adjacent track, but in fact have coincided with more worker injuries than in 1999, with a particular deterioration in worker safety in the 2010s. Andy Byford’s Save Safe Seconds campaign was the right response to the slowdowns, and helped stop the bleeding.

And yet, the idea of SOGR persists, even though the problem it purported to solve has been solved. The worst offender is Amtrak: in the Obama stimulus, it asked for $10 billion for SOGR on the Northeast Corridor, promising trivial reductions in travel times; Amtrak’s chair at the time, Joe Boardman, was the very one who deferred maintenance in order to make Amtrak look more profitable on paper in the service of the Bush administration’s goal of eventual privatization, replacing David Gunn, who was fired because he refused to do so.

In effect, SOGR is now a byword for “investments that aren’t sexy.” Some of those investments are still solid, like those done in the 1990s. Others are wastes of money; their lack of sexiness makes them ideal for managers who rate themselves by the input of how much money their agencies get rather than by outputs like ridership or service quality, since the lack of visible output disempowers civil society and good government watchdogs.

MTA Construction and Development head Jamie Torres-Springer essentially uses this definition in his defense of the capital plan, saying “We looked very closely at a couple of asset types that haven’t been focused on in the past. And to some people, they’re not the most exciting assets. They’re the ones that ensure that we can provide service. It’s structures and power and station components.”

The MTA’s capital plan is likewise denigrating the agency’s own past, saying, of the era in which MDBF rose by a factor of about nine in the span of 15 years, “Investment lagged again in the 1990s and early 2000s” and “After years of progress in the 1980s, investment fell off, culminating in a ‘Summer of Hell’ in 2017. That year, New York’s subway had one of the worst on-time performance of any major rapid transit systems in the world, with only 65% of weekday trains reaching their destinations on-time.”

The problems of train delays are not about investment or about maintenance. Rather, the train delays were about overly ambitious schedules, compounding with the problems of excessive interlining. Of course, interlining had always been present in the system, but the combination of new trains with better braking and signal timers installed based on the performance of older trains meant that the schedules could not be met without slowdowns; managers, in turn, changed how they measured punctuality from on-time performance to wait assessment, the latter more appropriate for subway lines with high frequency (like New York) and little complexity (unlike New York). MTA President Ronnie Hakim, coming from a legal rather than technical background, also denigrated the idea of speed, viewing it not as an essential feature of public transit but as a source of legal liability.

Non-sexy investment can target this; Byford alleviated some of the slowdowns with Save Safe Seconds. In the future, deinterlining the system, starting from DeKalb Avenue’s scrambling of the B, D, N, and Q, where trains lose two minutes due to schedule padding entirely to protect from cascading delays, is necessary. But this is not SOGR – in fact zero dollars are required in capital spending to deinterline DeKalb. Nor is it invisible – this is a visible change on the subway map, which passengers and good government watchdogs can judge for themselves, trading off fewer one-seat rides for higher speed and reliability.

But neither Lieber nor Torres-Springer seems interested in inexpensive fixes. No: both rate themselves by how much money they get rather than by whether it does any good, hence the denigration of the era in which SOGR was a success. As political appointees, they also have no loyalty to the system and its permanent staff, or even to well-regarded leaders (Byford, again) who do not come from the same political milieu. They fail because they exist to allow incompetent governors like Cuomo and Hochul to control a system they have no business running.

Public Transportation and Gig Workers

An argument about public transportation fares on Bluesky two weeks ago led to the issue of gig workers, and how public transportation can serve their needs. Those are, for the purposes of this post, workers who do service jobs on demand, without fixed hours or a fixed place of work; these include delivery and cleaning workers. App-hailed drivers fall into this category too, but own cars and are by definition driving. When using public transit – and such workers rarely get paid enough to afford a car – they face long, unreliable travel times, usually by bus; their work travel is completely different from that of workers with consistent places of work, which requires special attention that I have not, so far, seen from transit agencies, even ones that do aim at service-sector shift workers.

The primary issue is one of work centralization. Public transit is the most successful when destinations are centralized; it scales up very efficiently because of the importance of frequency, whereas cars are the opposite, scaling up poorly and scaling down well because of the problems with traffic congestion. I went over this previously talking about Los Angeles, and then other American cities plus Paris. High concentration of jobs, more so than residential density (which Los Angeles has in droves), predicts transit usage, at metro area scale.

Job concentration is also fairly classed. In New York, as of 2015, the share of $40,000+/year workers who worked in the Manhattan core was 57%; for under-$40,000/year workers, it was 37%. It is not an enormous difference, but it makes enough of a difference that it makes it more convenient for the middle class to take transit, since it gets to where they want to go. In metro New York, the average income of transit commuters is the same as that of solo drivers; in secondary American transit cities like Chicago, transit commuters actually make more, since transit is so specialized to city center commutes.

Worse, those 37% of under-$40,000/year workers who work in the Manhattan core are ones with regular low-paying jobs in city center, rather than ones doing gig work. The difference is that gig workers work where the middle class lives, rather than where the middle class works (for example, food service workers at office buildings) or where it consumes (for example, mall retail workers). They still generally take transit or bike where that’s available (for example, in Berlin), because they don’t earn enough to afford cars, but their commutes are the ones that public transit is the worst at. They can’t even control where they work and move accordingly, because they by definition do gigs. In theory, it’s possible for apps to match workers to jobs within the right region or along the right line; in practice, the situation today is that the apps can send a worker from Bytom to Gliwice today and a worker from Gliwice to Bytom tomorrow, based on vagaries of regional supply and demand, and the Polish immigrant who complained to me about this with the names of those two specific cities wishes there were a way to match it better, but at least currently, there isn’t.

The upshot is that gig worker travel is, more or less, a subcase of isotropic, everywhere-to-everywhere systems, with no distinguished nodes. This has all of the following implications:

  • Travel by rail alone is infeasible – last-mile bus connections are unavoidable, as are uncommon transfers, with three- and at times four-legged trips.
  • The bus network has to have the usual features of a modernized, redesigned network, with high all-day frequency and regular transfers – suburb-to-city-center buses alone don’t cut it when the work is rarely in city center, and a focus on rush hour service is useless for workers who mostly travel outside peak hours. This also includes reforms that improve buses in general, regardless of the route taken: proof of payment, bus lanes, stop consolidation, bus shelter, signal priority at intersections.
  • For the most part, the buses that take gig workers to work are the same that could take residents of those neighborhoods to work, in the opposite direction. However, in areas with weaker transit than Berlin or New York, much of the middle class drives, making buses within usually lower-density middle-class areas infeasible. In contrast, those buses are still likely to be used by gig workers doing service work in the homes of those drivers.

The last point, in particular, means that one of the more brutal features of bus redesigns – cutting coverage service in order to focus on the more useful routes – can be counterproductive. This is, again, not relevant to large enough cities that their middle classes mostly don’t live in coverage route territory (even Queens doesn’t need this tradeoff, let alone Brooklyn). But in New Haven, for example, Sandy Johnston long pointed out that some of the bus routes just don’t really work, no matter what, because the areas they serve are too low-density, so the only way forward is to prune them.

This more brutal treatment can still be understandable at times. If the route is being straightened rather than eliminated, as we discussed for Sioux City years later, then it provides all workers with faster service – the meanders if anything are to big job centers that are a few hundred meters off the arterial, and gig workers are less likely to be using those meanders than regular service workers. Moreover, if the part being pruned is genuinely low-density, then it may well also have low density of destinations for gig workers. However, if the part being pruned has moderate density, and is just considered low density because the residents are rich enough they never take the bus, then it’s likely to be useful for gig workers, and should when possible be retained, likely with no extra peak service, only base service.

Evidently, routes like that are sometimes understood to have this class of rider, though perhaps not in this language. This is most visible in suburban NIMBYism against buses: a number of middle-class American suburbs oppose the introduction of bus service that may be useful for regular riders, for fear that poor people might use it to get to their areas; in Massachusetts, those suburbs are fine with buses making one stop in the periphery of their town, triggering a paratransit mandate under the state’s interpretation of the Americans with Disabilities Act (in most states it’s within 0.75 miles, but in Massachusetts it’s town-wide), but oppose any further penetration by regular transit.

To be clear, most of the things that would disproportionately benefit gig workers also benefit the network writ large: faster buses with off-board fare collection and (in denser urban areas) bus lanes would make a great difference, and so would shifting service away from the peak. But the network design principles at granular enough a level to discuss pruning marginal routes really do differ, and it’s important to get this right and, at the very least, avoid empowering aggrieved rich people who hire maids and then do local activism to make it harder for their maids to get to their houses.

To be clear on another point, none of these reforms would make traveling to clean a randomly-selected apartment in a residential neighborhood pleasant. But they could, through smoother bus travel time and transfers, replace a 1.5-hour commute with a 1-1.25-hour one, which would make a significant, if not life-altering, improvement in the comfort levels as well as productivity (and thus pay) of gig workers. It mirrors many other egalitarian social interventions, in producing a moderate level of income and quality of life compression, rather than a change in the rank ordering by income.

Quick Note: Flushing’s Growth and Through-Running

One of the dirty secrets of my (and ETA’s) New York commuter rail through-running proposal is that it barely connects Long Island to New Jersey. The later lines with the longer greenfield tunnels do, but the base proposal only through-runs the Port Washington Branch to New Jersey, and with some work it can also through-run some branches to the Hudson Line via Penn Station.

Credit: Kara Fischer, ETA; Flushing is not depicted on the map and is on the Port Washington Branch

It’s long been a criticism of the plan in comments and on social media that it doesn’t do anything to connect Newark with Jamaica. I’d like to address this briefly, since changes in work geography over the last decade have made the Port Washington connection more valuable relative to the Jamaica connection.

Job counts

For the main secondary centers that are or could be on this system, here are the job counts within 1 km of the station, in the business cycle peak years of 2007 and 2019:

Station2007 jobs2019 jobs
Newark Penn Station57,94444,171
Sunnyside – Queens Boulevard40,09263,096
Flushing17,02642,961
Jamaica11,88020,130
Stamford25,18925,141

Source: OnTheMap

Jamaica and Flushing both grew rapidly in the 2007-19 business cycle, but Flushing both started bigger and grew faster, to the point of approaching the job count near Newark Penn Station.

Long Island City has seen booming development, as the only near-center neighborhood in New York with significant construction rates; the number of residents has grown even faster, from 4,502 to 12,183 employed residents over the same period, but with a jobs-to-employed-residents ratio higher than 5, it is a business district first. Plans for an infill station at Queens Boulevard are on the MTA’s wishlist in the 20 Year Needs Assessment, at typically extreme MTA costs; this is separate from Sunnyside Junction, somewhat to the east, which has less development but could be a cross-platform transfer with East Side Access-bound trains.

Non-work trips

Flushing is a booming ethnic center for Chinese-New Yorkers. Jobs there serve the community wherever its members live, and so do non-work destinations, including cultural centers and well-regarded Chinese restaurants. This generates not only work trips, but also consumption trips. Without fast transit to Flushing, it’s a special occasion to go there for food, especially if one does not live on the subway; with fast transit, Flushing restaurants are capable of outcompeting more local alternatives for people arriving from inner New Jersey, and people from suburbs farther out may choose to take a more frequent LIRR than to drive.

Jamaica is not a regional center of much. There is one big trip generator there, other than the growing job center: JFK, via the AirTrain. Airport connections are valuable, but also overrated. The unlinked (likely total) ridership on the AirTrain in the first three months of 2024 was 1.924 million, or 21,143/day (not weekday), slightly higher than in 2019. This is not a high modal split, but airport arrivals are disproportionately going to Manhattan already, and the frequency between Penn Station and Jamaica is high enough that through-running and other modernization elements would only mildly increase this figure.

I can’t quite compare the two figures, since leisure trips, especially routine ones like going out to restaurants, are hard to measure. But Jamaica’s airport trips coming from better commuter rail are just not going to be significant in volume by the standards of the work trips of Long Island City or Flushing.

Through-running schemas

The reason I’ve advocated for through-running from New Jersey to the Port Washington Branch and no other LIRR line is operational. There is only enough capacity for at most 12 trains per hour, because the trains have to share tracks with Penn Station Access local trains to Stamford and with intercity trains. Connecting to an LIRR branch serving Jamaica would create complex branching, with the same line in Queens reverse-branching to different destinations, reducing reliability. It was hard enough to timetable the reverse-branched New Haven Line in our Northeast Corridor project. The Port Washington Branch, running completely separately from the rest of the system, sharing tracks only on the approach to and within Penn Station, is an ideal candidate.

It is a happy coincidence that the through-running schema for the LIRR that is easiest to implement also happens to serve the larger Queens business center between the two traditional ones. It would also be a great opportunity to build infill in Long Island City, which has emerged in the last few decades to be a much larger center. Another happy coincidence is that, while New Haven Line timetabling has been difficult, there is room in the schedule for two infill stations in Queens without upsetting the delicate track sharing between Penn Station Access local commuter trains and intercity trains within the East River Tunnels to Penn Station. Anything involving mainline rail through legacy cities is necessarily going to have to rely on tricks, waivers, and happy coincidences like this to cobble together a good system out of a region that had no reason to be built in 1900-30 around the commuter rail technology of the 1970s-2020s.

Costing Northeast Corridor High-Speed Rail

As our high-speed rail project draws to a close, we need to not just write down what is needed for running the trains but also how much it costs. This post should be viewed as a work in progress, and it will not surprise me if I’m missing things that will make it to the report later this year.

The rule for this post is that costs only matter going forward, not backward. If it’s already committed, it’s not part of the budget; in particular, the $6 billion Frederick Douglass Tunnel, already fully funded and in the design and engineering phase, is not part of the budget. In addition, only infrastructure is costed, not rolling stock (new rolling stock may well have negative cost relative to current plans, through buying standard EMUs and not esoteric trainsets like Massachusetts’ battery train idea or nonstandard LIRR/Metro-North-style EMUs).

Bypasses

All bypasses can be seen on this map, but not all bypasses are part of the plan – in particular, nothing between Stamford and New Haven seems worth it for now.

The main bypass we’re proposing, between New Haven and Kingston, is 120 km in relatively easy terrain, including two constrained river bridges (Quinnipiac and Thames; the Connecticut is easier), but no tunnels. The cost should be in line with non-tunneled high-speed lines in Europe, which in 2024 dollars would be around $5 billion.

The secondary bypass, around Port Chester and Greenwich, is 7 km of complex els crossing I-95 multiple times, and should be costed at the upper end of els, which is high hundreds of millions. Call it $1 billion together with a new bridge across the Mianus. The current projected cost for the Cos Cob Bridge replacement is higher, but it should be easier to rebuild the bridge a bit upstream to straighten the approach curves than to do it in situ; with a short section of 4% grades on each side, it should be possible to clear I-95 west of the river and keep the Riverside station east of the river while also having around 23 meters of clearance below the bridge. (4% grades are routine for EMUs; freight trains are so long that they can ascend these grades just fine, since what matters is the grade averaged over the length of the train.)

Frankford Junction is about 2 km of complex urban el, including a rail-on-rail grade separation; the per km cost is likely high, in the very low three-figure millions, but it’s 2 km and so $300 million should cover it.

The other bypasses are very short and in easy environments, for example easing the curve at Kingston (also discussed here), with costs dominated by the track connections rather than the physical construction of 1-2 km of at-grade track outside urban areas. Call this entire portion $6.5 billion total.

Grade separations

The starting point is that NJ Transit thinks that Hunter Flyover should be $300 million in 2022 prices (source, PDF-p. 151). This is as close as can be to a nonnegotiable element of the program.

At the other end of the New York metro area, there’s Shell Interlocking/CP 216, which must be grade-separated as well, and is even more nonnegotiable. I have not seen recent cost figures; it should be comparable to Hunter or somewhat more expensive given the right-of-way constraints. A $500 million placeholder is probably right.

Further north, the junction with the New Canaan Branch is flat and needs to be grade-separated, at a cost likely similar to Hunter, in a similarly built-up area. The Danbury and Waterbury Branches have flat junctions too, but traffic is low enough that they may be kept so (especially Waterbury), but if not, Danbury seems comparable in difficulty to Hunter and New Canaan.

In Philadelphia, the Chestnut Hill West Line (former R8) has a flat junction with the Northeast Corridor, and there are a variety of proposals for what to do with it; for decades, an advocate wish was the Swampoodle Connection, to have it transition to a closely parallel line letting it enter the city via the Reading side rather than the former Pennsylvania Railroad side that it’s on. It’s largely dropped off the wishlist, and instead a grade separation could be done for a cost comparable to that of Hunter, or maybe less (potentially much less) if it’s possible to abuse the line’s low ridership and close proximity to the Chestnut Hill East Line to have shutdowns to speed up the work.

On the other side of Philadelphia, the junction between the intercity and commuter rail approaches to 30th Street is flat as well, which also incorporates the branch to Media/Elwyn (former R3); this should be grade-separated as well.

In Boston, there are two flat junctions on the Providence Line. Canton Junction separates it from the Stoughton Line, and looks routine to either grade-separate (it’s a low-density area) or, potentially, even turned into a shuttle with timed connections to the Providence Line if absolutely necessary, given the demand mismatch between the two branches. The Franklin Line, farther north, has a similar flat junction around Readville, technically within Boston but in an area with plenty of space, but can be sent over to the Fairmount Line if there are difficulties, and may even preferentially go to Fairmount regardless (the main argument against it is service to Back Bay). The answer to “how much should this cost?” is “no more than around $150 million each or else it’s better not to do it at all.”

In total, these should be around $1.8 billion, with New Canaan and Canton but not Danbury or Readville.

Note that rail-on-rail grade separations for bypasses are already priced in, especially New Haven-Kingston, which is of comparable length to European high-speed lines that have been built, with grade-separated connections to legacy lines.

Portal Bridge

The Hudson Tunnel Project within the Gateway Program is funded, but some tie-ins are not. Most (such as Penn Expansion) are useless, but one is essential: a second Portal Bridge, to ensure four tracks of capacity from New York to Newark. The current favored alternative is a lift bridge, budgeted at $800 million; it is a movable and not fixed bridge, but it is not a causeway and has some clearance below, and would only need to open when a sludge barge comes from upriver, which can be scheduled overnight.

High platforms

Everything that touches the Northeast Corridor needs high platforms at all stations. The definition of “touches the Northeast Corridor” is complicated; for example, in New Jersey, there are 68 low-platform stations on the lines that go through Newark Penn or Newark Broad Street, of which 26 are funded for high-platform conversions for around $23 million each ($683 million/30 stations; the other four are on the Erie lines), but of the 68, only 10 are on the lines that would be using the North River Tunnels after the Hudson Tunnel Project opens (see map in ETA’s report). Even taking all 42 as required, it’s around $1 billion at NJ Transit costs, with nearly all benefits accruing to commuter lines.

In Massachusetts, the definition is easier – everything on the Providence and Stoughton Lines needs to be raised; the TransitMatters report explains that there are eight stations, plus two potential infills, with the eight costing around $200 million in 2020 prices, which should be closer to $250 million in 2024 prices. If Franklin Line work is also desired then it should be another $200 million, split across more stations but with shorter platforms. Note that the second phase of South Coast Rail, if it is built, would extend the Stoughton Line, but as the stations are all new construction, they will already have high platforms.

In Pennsylvania, nearly total separation of intercity traffic from SEPTA is possible from the get-go – the only track sharing is peripheral, in and around Wilmington, at low frequency on SEPTA. If the entire Wilmington/Newark Line is to be upgraded, it’s a total of 12 stations, all in four-track territory; SEPTA’s construction costs for high platforms are lower than those of the MBTA and NJ Transit, but much of its construction has been single-platform stations with shorter trains, and my guess is that those 12 stations are around $200 million total. The seven inaccessible stations on the Trenton Line, which, to be clear, does not need to share tracks with intercity trains at all, should be another $100-150 million (it’s a busier line, so, longer trains, and North Philadelphia is more complex).

In Maryland, two stations on the Penn Line are inaccessible, West Baltimore and Martin State Airport. West Baltimore is being upgraded as part of the Douglass Tunnel program, while Martin State Airport has a separate program, which appears funded.

In total, all of this is around $1.8 billion, with the benefits going to commuters at such rate that state matches would be expected; in Massachusetts at least, there are talks about doing it as part of the Regional Rail program, but no firm commitment.

Electrification

The variable-tension catenary south of New York, as users of the Northeast Corridor were reminded two months ago, is substandard. It’s long been a wish to replace it with constant-tension catenary, to both improve reliability and permit unrestricted speeds, up from today’s 135 mph (217 km/h).

Unfortunately, precisely because it’s a longstanding Amtrak project, the project definitions have been written in a way that is not compatible with any cost-effective construction. For example, Amtrak is under the impression that the catenary poles have to be redone because higher speeds require denser pole spacing; in fact, catenary systems sold routinely by European vendors allow high speeds at spacing that exists already on the legacy Northeast Corridor system.

This makes costing this more difficult; Amtrak’s official figures are of little relevance to a project that has even cursory levels of interest in adopting European practices. With the poles and substations already usable, the wire tensioning should cost less than installing new wires; around half of the cost of new-build electrification is the substations and transformers and the other half is the wires, so take the cost of new-build systems outside the US and Canada, cut in half, and then double back to take into account that it’s a four-track corridor. This is around $3 million/km, so around $1 billion corridor-wide.

Commuter rail lines that touch the Northeast Corridor need to be wired as well, and then it’s a matter of which ones count as touching, as with the high platform item. This includes 25 km of the North Jersey Coast Line, 72 km of the Raritan Valley Line, 31 km of the Morristown Line, 30 km of the Montclair-Boonton Line, 38 km of the Danbury Branch, a few hundred meters of Providence Line siding tracks, 6 km of the Stoughton Line, 34 km of the Franklin Line, and 15 km of the Fairmount Line. Much of the unwired territory is single-track, so lower per-km costs can be expected, on the order of $600 million total.

Together, this is about $1.6 billion.

Total

The sum of all of the above lines is $12.5 billion. It’s possible to go lower than this: the high platform and electrification costs are partly modernizing commuter rail that may not quite use the Northeast Corridor, and the Greenwich bypass may be dropped at the cost of 80 seconds (more, if Cos Cob Bridge speed limits have to be lower than what right-of-way geometry allows). A numerological $10 billion limit can still be met this way.

The Northeast Corridor Should not Host Diesel Through-Service

The vast majority of traffic on the Northeast Corridor comprises captive trains, only running internally to the corridor. However, a noticeable minority of trains run south of Washington, swapping the electric locomotive for a diesel locomotive. Those trains have a certain logic to them today – through-service is valuable, and north of Washington they more or less substitute for Northeast Regional service. But drawbacks to reliability remain, and if the corridor modernizes its operations, they will need to be removed; several elements of modern operations are not compatible with keeping either the Virginia trains or the long-distance ones on the corridor. Instead, these trains should be cut to Washington, with transfers to much faster, more frequent Northeast Corridor trains. Potentially, some Virginia lines could be electrified and then through-service could be offered, if they can fill the same size of train that future Northeast Corridor service could.

Fortunately, this tradeoff still leaves the South with better service than it gets today. The forced transfer considerably speeds up the trip for New York-bound passengers, by more than the average transfer penalty even for passengers with heavy luggage. Nonetheless, a tradition of direct through-service from New York to diesel territory in the South will need to end.

Future Northeast Corridor service

Upgraded service, for example in our ongoing project at Marron for how to blend intercity and commuter rail on the corridor, should have all of the following features:

  • High speed: our current timetables have New York-Washington trains taking 1:53, at a top speed of 320 km/h; a blanket speed restriction to 217 km/h, the upper limit of the catenary today, would only slow down the trains to 2:04, the rest of the difference in speed from current trip times coming from reliability improvements, higher curve speed, higher acceleration, and minor curve fixes.
  • All-EMU configuration: nearly the entire world passenger rail market is electric multiple units rather than separated locomotives and coaches, EMUs outperforming locomotive-hauled cars in every aspect, and the exceptions are the less modern intercity and regional lines.
  • Single-class service: trains may have first- and second-class seats, but the trains should not be differentiated by speed – Spain has trains of different speeds on its high-speed line and charges more for the faster ones, and the resulting hit to frequency and interchangeability of tickets explains why Europe’s longest high-speed rail network has a fraction of the per capita ridership of France, Germany, or Japan, which (largely) lack this misfeature.
  • Affordable fares: the average fare should be in line with French and German norms, around $0.15 per p-km.

A train that does New York-Washington for a bit less than two hours and charges a bit more than $50 one-way on average – the current average is $106 on the Regional and $192 on the Acela – can comfortably expect ridership to quadruple, based on my two usual references on elasticity of high-speed rail ridership with respect to travel time and fares (Börjesson says -1.12 and -0.61 respectively, Cascetta-Coppola say -2 and -0.37 respectively). This forces running more frequency and longer trains. Frequency is a welcome addition provided there is capacity on the tracks for it; fortunately, there is capacity for an intercity train every 10 minutes in a post-Gateway timetable. The trains in question should be as long as possible, with platform lengthening where necessary to support 16-car trains, to maximize capacity.

The incompatibility of diesel trains

Very few diesel trains run on the Northeast Corridor today, and none run by Amtrak. The through-trains to the South run with engine changes: all Amtrak service today is run with locomotives, and at Washington, the trains change between diesel and electric locomotives. Nonetheless, even electric locomotive service as it is conceived today is incompatible with Northeast Corridor modernization, and future changes would still not make it compatible.

First, to the point on capacity: there is no way to run 16-car trains into the South. There isn’t enough demand for such trains. The Silver Star today runs nine coaches and the Silver Meteor runs 10, and on both trains, three coaches don’t sell seats but are used for baggage, lounge, and dining; the Palmetto runs six coaches, the Crescent seven, and the Cardinal five, each including two non-seat selling coaches. Speeding up the Northeast Corridor by an hour and a half can lead to ridership explosion internally to the corridor, but not on trains that take 30 hours today.

And second, there is no reasonable rolling stock for this, even if there were demand for a train with 16 cars or close to it. Locomotive-hauled trains would necessarily run slow, compromising not just top speed but also acceleration and, owing to the current equipment’s problems, reliability. The example train we’re using in our calculations, the Velaro Novo, has a power-to-weight ratio of 20 kW/t and an initial acceleration rate of 0.65 m/s^2. A pair of ACS-64 locomotives dragging 14 Airo coaches gets 14 kW/t but cannot accelerate faster than around 0.25 m/s^2 at any speed. The unpadded trip time for high-speed rail making one stop per state is 1:46; the unpadded trip time with the additional acceleration time of this example train and with a 217 km/h (135 mph) speed restriction is 2:09. If the timetable buffer time is still 7% then the trip time is 2:18, which means the train would be overtaken by about two faster trains, and this in turn would slow the trains further as more schedule contingency would be required for this more complex system. If the ACS-64’s problems or any interface with the freight-run Southern network forces more padding, then make it three overtakes.

The TGV used to couple a diesel locomotive in front of a high-speed trainset to reach Les Sables d’Olonne, before the branch to it was electrified. This option would eliminate the speed difference on the Northeast Corridor, but would also mean that expensive 16-car high-speed trainsets would be spending an entire day going to Florida at low speed and another going back, without being able to make back the cost through intensive operations measured in train-km per day.

Exceptions and the tail wagging the dog

The basic reason for prioritizing the Northeast Corridor’s internal traffic over through-traffic is the large mismatch in travel volumes. In fiscal 2023, the Northeast Corridor got 12,122,466 riders. The Virginia services got 1,300,776, the Carolinian 315,781, and the long-distance trains 1,308,211. A 4:1 ratio should tilt planners toward prioritizing the core over the long-distance trains.

Note that I have not, up until now, talked about Keystone service and trains to Springfield. This is because Keystone trains can run through to the corridor just fine. None of the reasons why the long-distance trains cannot do so applies: the Keystone corridor is electrified all the way to Harrisburg, and New York-Philadelphia is a significant enough portion of it that boosting speeds in the core (and acceleration everywhere) would lead to sufficient ridership increases to justify 16-car trains. Springfield service is currently unelectrified, and Amtrak generally runs shuttles with timed connections because of the mismatch in demand; it should be electrified, and through-service instituted.

On Keystone and the New Haven-Springfield line, the mismatch in capacity actually works in favor of through-service. The New York-Philadelphia section has the most demand, so having one third of the trains branch off to Harrisburg rather than continuing to Washington is a good way of assigning capacity. New Haven is not Philadelphia, but has so much commuter demand to New York that giving New York-New Haven an intercity train every 10 minutes is not so stupid; in contrast, unless a lot more is built, I suspect that 16-car trains running every 10 minutes between New York and Boston would end up emptier than most planners would prefer. Years ago, before I started looking at the track charts and the possible schedules systematically, I even used the greater demand of New York-Philadelphia to argue in favor of diverting some trains not just at New Haven to Springfield, but at Penn Station, to Jamaica and Long Island; as it is, my primary argument against sending intercity trains to the LIRR is timetabling complexity.

With Keystone and Springfield added back in, the traffic on the Northeast Corridor rises from 12,122,466 to 13,680,273. The ridership of the trains to the South that are to be cut from the corridor is 2,924,768, or 21% of the internal ridership; the tail should not wag the dog.

Is this even bad for the trains to be cut?

No. As mentioned above, the trip times would get a lot faster, it’s just that turning a 30-hour trip into a 28-hour one does not lead to a large ridership boom.

The extra transfer is annoying, but should be compared with the time cost of both running a slower train to New York and changing the engine at Washington Union Station. As explained above, the slower train would take a minimum of 2:09 between New York and Washington, stopping once per state. The scheduled time would be at least 2:18, and likely more, maybe 2:28 with 15% buffer time. The engine change takes about 20 minutes judging by southbound schedules on Virginia service trains; the wait time at Union Station is much longer northbound, because the train has to have more schedule contingency on the less reliable freight-owned section to make its slot on the more precisely timetabled Northeast Corridor. The most charitable interpretation, ignoring the extra required schedule padding, is that making passengers change trains in Washington would save them 45 minutes minus the wait time for the next train (at most 15 minutes).

The transfer penalty is extensively studied in the modal choice literature. For example, studying intercity trips in the Netherlands, de Keizer-Kouwenhoven-Hofker find that the penalty is 23 minutes, which already incorporates an imputed additional waiting time of 15 minutes. This penalty rises by seven minutes if the transfer is not cross-platform; a cross-platform transfer at Union Station would require the through-tracks to be upgraded with high platforms, as they are currently low-platform. It rises by a further seven minutes for passengers with heavy luggage. Even with all of those penalties, 23+7+7 = 37 < 45. And 45 is in a way a best-case scenario; there is a lot of padding involved in making a long-distance or even Virginia train make a specific slot on the corridor, as opposed to guaranteeing passengers a seat on the next available train, and this adds on the order of half an hour, counting both Alexandria-Washington and on-corridor buffer times.

The upshot is that while trains cannot run through from Virginia to a modernized Northeast Corridor, little is lost in making passengers transfer at Union Station. The transfer penalty is real but limited, even with luggage, and the speed gain from letting such passengers transfer to a faster train is noticeable, if small compared with the total length of a night train trip. It would break tradition, but offer a modest improvement in the quality of rail service on the long-distance trains using the corridor and the Virginia trains, in conjunction with the much larger improvement in the quality of internal Northeast Corridor service.