Meme Weeding: Regular Funding

American and British discourses on public transportation are too obsessed with getting regular funding. In the United States, it’s a commonplace among advocates that what the mode really needs is a regular source of funding protected from the usual political process. American thinktanks trying to talk about construction costs, such as Eno, even spoke with fascination about the bipartisan consensus in favor of expanding the Madrid Metro, in contrast with the American situation. In the United Kingdom, it’s specialized to some aspects and their construction costs, but the “feast or famine” line on electrification is particularly common. In truth, there’s quite a lot of funding regularity and political consensus in parts of the United States and United Kingdom that nonetheless cannot build anything, and conversely, some of the most prolific, lowest-cost builders in the world do so in an environment of funding uncertainty, especially when we consider the effect of the Great Recession.

Spain in the 21st century

The very low-cost metro construction program in Madrid under Manuel Melis Maynar in 1995-2003 is well-known to both academic researchers and advocates, and Melis’s own writeup on how this was achieved is a starting point for the entire discourse. This program, going back earlier in the 1990s and continuing right up until the 2008 financial crisis and the onset of the Great Recession, was marked by political competition between the left and the right over who could deliver more metro construction, and consistent commitment to building more infrastructure across Spain. At national scale the later part of this period is characterized by rapid, low-cost construction of high-speed rail building upon the success of the Madrid-Seville line (opened 1992). This was not a purely rail-based system – airports were expanded rapidly in the 2000s to accommodate the growing economy and growing international travel market.

And then all of this crashed after 2008. Some lines were still built if funded before – close to half of Spanish high-speed rail by funding is lines that opened post-2008, nearly all well underway when the recession set in. Other lines were canceled, some of which have not been restored, such as Madrid-Lisbon. Barcelona kept building L9 and L10, the most expensive in Spain, but Madrid’s construction program broadly wrapped up. I encourage readers to go to Alexander Rapp’s visualization and compare the rapid growth in 1995-2010 with the almost total lack of new lines in 2010-2025. If there’s one place that meets the definition of feast-or-famine, it must be Spain.

And downstream of the famine, construction costs are still very low. The high-speed rail built through the famine is more expensive, but only because it’s the Basque Y, majority in tunnel due to topography. The metro lines Madrid has built recently are still cheap: the Line 11 extension to Conde de Casal to open next year costs 84 million €/km, only mildly more than the Melis-era program in real terms, the difference largely attributable to the fully-underground character of the extension whereas the Melis-era program was only 73% underground. Feast-or-famine hasn’t dented Madrid’s ability to build cheaply, and so the plan as the economy is finally recovering from the crisis is to resume building rapidly, the entire 33.5 km Line 11 plan budgeted at the same low cost.

Italy and the reset button

Italy’s period of regular construction in the 1960s, 70s, and 80s was initially characterized by low costs for the first Milan lines from the 1950s and early 60s, but these grew quickly, and by 1970 they were if anything higher than in Britain and not much lower than Germany’s, due to the extensive corruption of the First Italian Republic. The 1990s process of mani pulite was such a transformative change that people periodize Italy since then as the Second Italian Republic, with rather different politics from the First, with far less corruption and far more left-right polarization.

In the interim, there was a famine in metro construction as rules were changed to require more transparency. The system that emerged from the famine had lower real costs than that of the 1980s with its corruption. A second famine occurred, to some extent, in the Great Recession, albeit nothing as drastic as in Spain; the Spanish economy was characterized by high growth with a housing bubble in the years leading up to the Great Recession, where Italy in the 1990s and 2000s up to 2008 already had the weakest economic growth in Europe, so in effect there weren’t such massive infrastructure projects to be canceled. The famine is over, and the Italian economy has exceeded its 2007 peak and returned to healthy growth, with further construction of metro lines. Costs broadly remain low – Turin’s Line 2 has a high projected cost by Italian standards, but the extensions in Milan don’t appear any more expensive than the ones built 15 years ago.

Nordic costs

Stockholm built its three-line metro from the 1950s to the 1990s, with most construction complete by 1978. That feast, driving up Stockholm’s modal split by the 2000s and 10s to Parisian levels despite not being a particularly large city, was followed by famine, with only two short extensions opened thereafter, in 1985 and 1994. Since then, only one urban rail tunnel has opened, Citybanan, opening in 2017. Nya Tunnelbanan has proceeded, with the 19 km extension to open in stages beginning next year.

Swedish costs are sharply up. The original T-bana was among the cheapest metro systems built in history, not much more expensive than lines built in the 1900s-10s in New York, Paris, and London. Relative to Sweden’s GDP per capita, it is, I believe, unsurpassed in its low costs. Madrid’s costs were higher than this. Citybanan is harder to compare, as a commuter rail tunnel with mined stations beneath older T-bana stations, on two levels in the case of Stockholm City/T-Centralen. But Nya Tunnelbanan is definitively more expensive, its costs still below the global average but no longer so Earth-shatteringly cheap.

And yet, this is not too different from the evolution of construction costs in Norway and Finland, where construction has been more gradual. We know why Stockholm’s costs have risen, a combination of mid-project regulatory changes that led to cost overruns and a general increase in real Nordic costs due to the adoption of more privatized project delivery and loss of state capacity. Copenhagen, which built its metro based on these more privatized ideas from the start since it only started building in the 1990s when those ideas were in vogue, has had these higher costs from the start (again, meaning “somewhat lower than 2020s global average” as opposed to “setting world records for GDP per capita divided by per-km metro costs”).

On the narrow matter of electrification, it’s notable also that the Danish program, a concentrated feast rather than a gradual multi-generational program as in Italy, is not particularly expensive. As it is recent and intensive, I if anything encourage English-speaking countries to study it and benchmark their programs to it, precisely because it gives us a snapshot of electrification of many different main lines. (This is also true of Israel.) In contrast, the more gradual rolling electrification programs of Italy and France are harder to benchmark to, because at this point the lines being wired are selective, for a combination of low traffic (the higher-traffic lines having been wired generations ago) but also ease of electrification (the more difficult lines being deferred due to failing a cost-benefit analysis).

Consensus in the United Kingdom

In Madrid, PP and PSOE both politically supported metro expansion and made that an election theme. The same can be said of a number of Continental European comparanda, such as Paris. However, London (on rail in general) and the UK (on High Speed 2 and electrification) are no different. HS2 and Crossrail were both matters of bipartisan consensus in the 2010 election, just as Grand Paris Express was a point of consensus in 2012. If anything, HS2 and Crossrail were thus less politicized than GPE, on which the Socialists and Republicans differed on which exact lines to build within the concept of inner-suburban Paris orbitals; a closer British comparison to GPE is the electrification program, which too was a point of agreement between Labour and the New Tories but then after the 2010 election the Coalition modified the program to prioritize secondary-city commuter lines over the Midland and Great Western Main Lines.

Moreover, physical construction in London, unlike British rail electrification, has shown remarkable consistency. The current period after the opening of Crossrail might be the first famine in a generation. However, the famine is not the cause of high costs but rather is caused by them: Crossrail 2 was being planned in the 2010s, but canceled due to outrageously high costs, and yet it is likely a scaled down version of it will go back to the planning board and be approved soon. An extension of the Bakerloo line to Lewisham is in the pipeline as well, at much higher per-km costs than those of the Northern line extension to Battersea and the Jubilee line extension, and the question of whether to build it at all is again downstream of the high costs, not upstream of them.

Despite relative consistency in funding and politics, British projects do not have any consistent planning regime. Every project is delivered as a single purpose vehicle (SPV/SPDV; D stands for delivery). The Jubilee line extension was in theory designed in-house, but because it was clear the model was giving way to privatized project delivery, the in-house team was treated as an SPDV and its expertise withered away after the line opened. The same has happened with subsequent megaprojects, the lessons of Crossrail now existing only as a Learning Legacy report; in contrast, GPE, initially planned as an SPDV under British influence, is transitioning to a permanent bureaucracy for building metro and commuter rail tunnels across France.

Consensus in New York and California

The United States since 2010-11 has been characterized by almost total unwillingness by the Republican Party to assent to major public transportation and intercity rail projects. Chris Christie set the stage by canceling the ARC tunnel between New Jersey and New York without replacement; as the project has high benefits matching its outrageous costs, it was rebooted by Amtrak, at even higher costs. Subsequently, governors elected in the 2010 wave returned federal money for intercity rail, in Ohio, Wisconsin, and Florida.

And yet beneath this national picture, one finds remarkable consistency in funding in New York and California. In New York, the Metropolitan Transportation Authority has planned its capital spending in five-year cycles going back to reforms in the 1980s under Richard Ravitch to discourage maintenance deferral, with regular funding in these cycles and little partisan controversy thereover. Expansion projects used to be a matter of bipartisan deals, with Republicans like Governor George Pataki (r. 1995-2006) dealing with Democrats like Assembly Speaker Sheldon Silver (r. 1994-2015) to agree on funding for their constituencies, respectively the suburbs and the city. The more recent Republican Party has been less convivial, but also far less electable, and the state is by now run by a single party, with consistent funding and a regular cursus honorum for the political appointees overseeing the MTA. There’s no feast-famine cycle, nor is there any siege mentality over what if the Republicans will defund the MTA, because the state money for the MTA is still there and nothing electable in the state is canceling that.

California is largely the same. California Republicans do not have the same history of bipartisan moderation as New York Republicans, but the state moved on from them a generation ago already. In this century, the only Republican who was even mildly competitive in a statewide election, Arnold Schwarzenegger, was already in the 2003 election noted as atypically moderate, and after winning pursued a broad consensus in transportation and supported California High-Speed Rail. Below the statewide level, the form of mandatory referendums in the state has led to rather consistent funding with specific sales taxes deeded to transportation, to the point that the Los Angeles County MTA has capital funding projections through 2060 with a list of priorities to be funded in each period telegraphed decades in advance.

And neither New York nor California is capable of building. Consistency in funding has not led to the formation of a consistent civil service. Single party dominance, with the Republicans choosing to nominate sacrificial lambs like Larry Elder rather than to moderate enough to win the median voter in a wave election, means that there is no need for political appointments, as professionals know in advance what the state’s ideological direction is going to be, and yet agency upper echelons are stacked with politicals and expect subject matter experts to be silent when the politicals and their staff are in the room. Costs in California are not only high but meteorically rising, in San Francisco, in San Jose, and in Los Angeles. In New York, despite real savings from our advocacy for shrinking station size, the projected costs of IBX are unreal and those of the 125th Street extension are full of lard, the savings wasted on deeper stations and ever higher soft cost charges.

American activists keep talking about the need for consistent funding. But where funding in the United States is the most consistent, outcomes are consistently bad, just as they are in the parts of the United States that are one election away from cancellation. This is not just a bad inheritance – the MTA let its in-house expertise go in favor of a consultant-centric system, and MTA CEO Janno Lieber takes credit for having implemented the globally-failed design-build system locally, with essentially all discourse that isn’t coming from us at the Transit Costs Project backing further privatization and SPDVization of project delivery out of a false belief that London and the knowledge of global English-only consultants are worth imitating.

This way, the MTA, with a capital budget of $13 billion a year and no real fear of large cuts, delivers subway extensions at a rate of about one new underground station per 5 years, 9 stations equipped with elevator access per year, and sporadic commuter rail interventions that never actually improve speed or convenience. Consistent funding is not going to make it better, because its funding is already consistent and yet it is run by politicals who look down on subject matter experts and who have New York solipsism and react acerbically to being told they need to learn, and so it can’t build, only manage its own decline.

Quick Note on Subway Operating Costs

A London Underground benchmarking report using CoMET data from 2013 compared operating cost breakdowns and revenues between the Underground and its international peers. CoMET data is in principle anonymized, but it’s not hard to find which city is which, and in particular, whereas London, Paris, and Berlin all spent around $6/car-km on operations in 2013 costs, New York is the city labeled “Am” that spent $10/car-km.

More recently, I followed up on these costs by looking at 2020s data, finding little for Berlin and even less for Paris, but finding exact costs per car-km for New York and per train-km for London. London has seven cars per train, from which we can impute, in 2024 PPP prices, $6.2/car-km in London; New York’s exact costs are $11.58/car-km. BVG’s costs are bundled across modes, but the total costs for 2024 reported in the Lagebericht und Jahresabschluss 2025 are 1.6B€; if U-Bahn costs per car-km and bus costs per bus-km were as in New York and tram costs per tram-km were as in Boston, the two rail modes compared on a per train length basis (thus, a Berlin U-Bahn train of 100 meters is deemed to be six New York City Subway cars), the total would be exactly twice on a PPP basis. If the same comparison were made with an adjustment for bus speed (17.9 km/h here, 11.3 km/h in New York), make it 1.6B€ vs. 2.6B€. Either way, it’s very likely Berlin’s U-Bahn operating costs are in the $6-7/car-km range in 2024 prices.

All of this is remarkable, because prices between 2013 and 2024 rose, by a factor of 1.34 in the US. And yet, despite this inflation, London more or less kept its operating costs unchanged, New York had an increase of slightly less than half the inflation rate, and Berlin likely had a small increase like New York or even smaller.

Moreover, none of the three systems engaged in massive automation over this period, not even the incremental automation of Paris. Furthermore, New York’s subway costs in the short and medium runs tend to rise when ridership decreases and fall when it increases, as fixed costs are spread across more service; the number of employees per unit of service provided rose when ridership fell after WW2 and fell when it recovered. However, the period 2013-24 was not one of major service increases in any of the cities: all three opened new lines, but only short ones, and none of the three embarked on a scheme to massively increase service – London had some increases but New York if anything provides less service now than it did in the early 2010s. Thus, no short-term shock can explain the over-the-decade fall in real operating costs in all three cities.

This contrasts with buses, which are dominated by variable labor costs. In the United States, the cost of running a bus is the wage of the operator plus various overheads, of which the largest is the wage of the maintenance crew. In Europe, bus driver wages are lower, but buses are also cheaper to procure and more fuel-efficient, and overall the system is dominated by wages rising faster than inflation and by variable and not fixed costs.

Devolution Best Practices

British commentators on infrastructure have spoken about overcentralization in their country, often also related to other state capacity issues. This led to a false belief that the United Kingdom (or just England) is too centralized, and to a discourse on devolution that I think doesn’t really get what the success and failure cases are. Beyond very small states like Israel and Singapore, we see extensive devolution of the management certain functions, like health, education, and local public transport, but not others, like intercity infrastructure, social insurance, and most taxes. Germany and Canada both have thriving Land- or province-level politics through having large enough units that democratic elections can be meaningful, and through devolving many functions to the same level that those elections are in fact meaningful. The United States suffers from a severe local democratic deficit because of wrong scale and because each function is devolved to a different level so elections are lower-salience and turn on personalities.

The local democratic deficit

Any devolution has to be to a level high enough that it meaningfully represents how people live their lives. People do not live and work in neighborhoods but in cities and metropolitan areas. This is true regardless of municipal boundaries: a once independent small town that has become a suburb of a major city has the same status as a neighborhood, with only a minority of its employed population working within its jurisdictional limits, and as such is not a good geography to devolve functions to.

The reason for this is that to be meaningful, elections need to be contested in boundaries that people can reasonably discuss with their social circles and peers. Otherwise, people cannot be sufficiently informed. It’s bad enough that I’ve repeatedly seen Germans whose social circles are predominantly immigrant and English-speaking barely know the name of the mayor and not be able to name a single policy that the current Berlin government is responsible for. When there are different jurisdictional limits, this effect is magnified and hits people even if they are very aware of political trends.

Scale

Devolution requires the services devolved to be at the right scale for the level of government. The military, for example, relies on national and even supranational scale, to the point that there are flailing attempts in Europe to involve the EU more. Pensions and social transfers, heavily dependent on a locality’s tax base and age distribution, are almost always national, and if they’re devolved it’s with extensive equalization payments, as in Canada.

In contrast, health and education are both attractive targets for devolution.

Schools don’t really benefit from scale beyond that of the individual school or small group of schools. They rely so much on teacher autonomy and subsidiarity that the state can merely provide loose supervision, for example by setting high school leaving exams. States can still be centralized on this matter – France infamously used to have one class schedule for the entire country – but there are few benefits from this, so in practice this gets devolved, and in France primary, middle, and high schools are devolved respectively to communes, departments, and regions. Even on values matters, for example requirements for teaching sex education over religious objections, there is nothing a ministry of education inspector can do to coerce a community where there’s consensus among the teachers and administrators to the contrary. The American problems with the devolution of education are not about scale (for one, they affect large urban departments if anything more than small suburban ones), but about the above issue of the local democratic deficit and sub-metropolitan devolution.

Health care is similar. Nordic single-payer health care systems are not centralized the way the National Health Service in England is. Finland is divided into 21 health care regions (“Hyvinvointialueet,” or well-being services counties), with state funding and standards but local management; there’s emergency reciprocity in the sense that a Helsinki resident who gets hit by a car in Oulu will get treated in Oulu without getting charged more than if they were hit by a car in Helsinki, but the same Helsinki resident will need to find a primary care doctor and therapist in Helsinki and not in Oulu. This works, since the scale of health care doesn’t extend much beyond the hospital or network of doctors.

This, in turn, has led to a decentralist mentality that views the state as centering coercion and prefers to center care at a much local level. But pensions are care and rarely devolved. Conversely, police is coercion and almost always is, since good police work requires cultivating local ties and community trust and therefore the police is ideal for devolution even as far down as a municipality, and even national police forces are organized by region (for example, the Paris Police Prefecture).

I bring up the care-coercion dichotomy and New Left and NGO attitudes, because infrastructure is usually care-coded, but cannot be devolved beyond the level that it serves. Decisions on local and regional transport have to be done at the metropolitan level, which in Germany is that of the Verkehrsverbund and in Scandinavia is that of the county or region, the latter offering better democratic oversight than the purely intergovernmental Verkehrsverbünde. Intercity transport cannot be devolved at all: high-speed rail and tightly integrated medium-speed rail in the Swiss style both require tight top-down prioritization and lend themselves to national rail bureaucracies, even in generally devolved countries like Germany. Just as the EU flails at military integration, it flails at rail integration, with nobody proposing a coordinated rail system and only some activists, generally regarded as idealistic and reformist like Jon Worth, even discussing regulations like passenger rights and booking harmonization.

Co-devolution

There are functions comprising a substantial share of the public-sector budget and its decisionmaking that can be devolved, and can be devolved to wildly different levels. After all, between the school and the province there are multiple orders of magnitude of population. To reinforce democratic elections, it is best if as many different functions are devolved to the same level, to heighten the salience of its political issues. These may include all or most of the following:

  • Primary and secondary education
  • Tertiary education if the units are large enough
  • Health care management (with state funding to deal with differences in resources and age pyramids)
  • Road transport
  • Public transport
  • Police
  • State administration (for example, the population registry in Germany is devolved)
  • Zoning and development in general

Germany and Canada don’t devolve the same functions. For example, universal health care in Canada is administered by the provinces, whereas in Germany it’s regulated federally with a system of national Kassen and private insurers. However, they share a tendency to devolve multiple high-salience functions to the first-level federal units. Canadian provinces are thus stronger than American states compared with both the federal government and local governments.

In both countries, this interacts with the scale of these units, all large enough to be coherent units of how people live and not too local, to allow for competitive elections. In Germany, that some areas are more left-wing than others just means that the possible coalitions under the proportional system are different, and in truly left-wing places like Hamburg, SPD and the Greens compete for first place. In Canada, there are provincial parties separate from the federal ones, so that even Alberta has competitive elections where federally it is safely Conservative.

The United Kingdom has consistent, coherent regions to devolve to in Northern Ireland, Wales, and Scotland. But in England outside London, it’s never been sure what to do, and the result is that its devolution practices have been a mess of different tiers of authorities, whose boundaries shift with every generational government review. The Metropolitan Counties are popular as of late as subjects of devolution, since they are coherent metropolitan areas and have sufficient population for scale, with population levels between that of Northern Ireland and that of Wales. One can even speculate that counties, or regions comprising a few counties with strong historic ties like East Anglia, Yorkshire, and possibly a Greater Lancashire region including both Greater Manchester and Merseyside, could succeed if many functions are devolved to them at once. In all cases, that politicians who win in these regions can be considered strong candidates for the prime ministership, with Boris Johnson having so ascended from London and Andy Burnham mooted in the media as a replacement for Keir Starmer, underscores that this devolution can lead to the election of strong politicians rather than petty managers of local gossip.

The United States has consistent boundaries of its states, counties, and municipalities as well. But each function is devolved to a different level. State administration is done by states or counties, schools are done by districts that vary between the municipality and the county, policing is done by a combination of overlapping jurisdictions generally at the municipal level, roads are a mix of county and state programs. State and local parties do not exist, and in safe states, governors get ahead not through good management of the state but through partisan outrage bait and performance of resistance to federal authority that indicates either the governor or a member of the federal cabinet broke the law and ought to be in prison.

Infrastructure and devolution

Good infrastructure programs can happen at any level. Usually the same countries that have low infrastructure construction costs for local urban rail also have low costs for high-speed rail; the main exceptions are that Italy has high high-speed rail costs and low urban rail costs and France has low high-speed rail costs and medium urban rail ones, and neither of these patterns is about devolution. It is far more important to follow good procurement practices and keep in-house expertise in the public sector than to optimize which level makes decisions on infrastructure. Indeed, in highly unitary France, RATP (state-owned) will contract smaller cities to offer them public-sector consulting services for infrastructure projects of the type that are generational for them but routine for Paris.

It’s important to avoid devolving the basic functions of the state, which in this case include procurement standards, engineering standards, contract law, and homologation. On the last one, it is notable that vendors at InnoTrans have come to list the countries where their products have been homologated, because it’s not a single EU-wide process and this weakens the Common Market.

It’s equally important to understand that this conversation, especially in its British context, is not really about infrastructure. It’s about other functions. English centralization produces poor outcomes in construction; so do Canadian federalist devolution to the provinces and American hyperlocalism. The commonalities to the three countries lie elsewhere, chiefly in project delivery and engineering standards that ensure high costs and poor outcomes.

The RPA Publishes Bad Report Against Through-Running

The Regional Plan Association released a new report, called New York Penn Station: Constraints and Considerations for Meeting Future Demand. It argues against through-running on the grounds that it would reduce capacity, and asserts that any solution to station capacity after the Gateway tunnel, the Hudson Tunnel Project, opens in the late 2030s must include “station expansion or some other form of system expansion.” There’s something disappointing about this report, not because it’s wrong (although it is) but because it seems to just rehash past arguments without seriously addressing past criticisms of the $17 billion Penn expansion.

The lead author Rachel Weinberger and I talked about Penn Station capacity issues in 2024, after I wrote these two blog posts attacking Amtrak for its assumptions that imply additional tracks are required at Penn Station. Since then, the Effective Transit Alliance has done additional work, modeling the interfaces between the trains, the Penn Station platforms, and the concourses to arrive at feasible dwell times and capacities. Unfortunately, the RPA isn’t really engaging with any of that work, and retreats to just asserting that through-running would reduce capacity at Penn Station.

Capacity and dwell times

The most advanced technical work on the subject of capacity at Penn Station, the question being whether the bare Hudson Tunnel Project can perhaps with minor facelifts provide enough capacity to run 48 trains per hour between New Jersey and New York and the answer being yes, was done in the above-linked ETA report. We found the source of past claims by Amtrak that very long dwell times are necessary and deconstructed their assumptions, and modeled based on the current (post-Moynihan Station project) capacity of vertical circulation elements how long it would take passengers to clear each platform under rush hour assumptions.

The interface between the train and the platform itself is not the limiting factor; my two above linked blog posts from 2024 go over this and find a very short minimum dwell time, of 2 minutes or even less. The limiting factor is vertical circulation between the infamously narrow platforms and the concourses. Nonetheless, passengers can clear these in about 3 minutes. Notably, the narrowest of the platforms, platform 9, is compliant with NFPA 130 and its requirement that passengers be able to clear the platform in case of a fire in 4 minutes; the other platforms (except the wide platform 10) are not compliant, because they have two adjacent tracks and NFPA 130 assumes both tracks can be occupied, in which case the load doubles. Across all platforms, the one with the highest clearing time, platform 4, can clear a 1,620 passenger NJT bilevel in 4.83 minutes. Writing timetables to cycle trains between the platforms so that no platform comes close to having queuing between trains, is a routine exercise.

The report instead asserts that,

The time that the train dwells at the platform is 7 minutes in reasonable operating conditions: 5 minutes to unload passengers, 1 minute to check for stray passengers and then 1 minute for schedule recovery. There’s an additional 2.5 minutes to clear the interlockings and for train safety separations.

Accounting for all these aspects, under reasonable operating conditions, each track can accommodate 6 trains inbound per hour using drop-and-go service.

There is no citation for the model used, nor justification for either the 5 minute figure or for why an additional minute for schedule recovery is required. There is no explanation for why the 2.5 minutes to clear the interlockings matter to the capacity of each platform – once a train is past the platform, it’s gone and the capacity to measure is that of running track, not the platform. On outbound trains, it ominously says that,

The process of bringing an empty train into Penn Station, loading passengers and then departing takes 9 minutes in reasonable operating conditions during the PM peak period. This includes 7 minutes for passengers to hear the track announcement, descend to the platform and load into the train and 2 minutes for schedule recovery.

“Hear the track announcement” is an Americanism. In Germany, the track numbers are printed on one’s ticket, even on intercity trains, even at capacity-constrained stations with track shortages like Köln Hbf. The track for each fixed branch should be scheduled months in advance and known by regular passengers. Precisely because demand is asymmetric – toward Manhattan in the morning, away from Manhattan in the afternoon – the on and off peaks do not coincide at all, and encouraging passengers to get to the platform even before the train arrives would not overload the platform or the access and egress points.

The report completely missed the consequence of the asymmetry of demand when it finally asserted, based on modeling for which the report provides no details, that,

It would take an estimated 6 minutes for passengers to deboard, clear the platform and reach the concourse and another 4 minutes for passengers to descend to the platform and board the train during the AM peak. 2 additional minutes must be added to provide a buffer that ensures schedule adherence. The trains also require 2.5 minutes for clearing interlockings and safety separation, resulting in an hourly capacity of 4 trains per track and potentially 8 per platform.

A good rule of thumb here is that if Munich manages to slot 7 numbered branches, rising to 9 when one includes sub-branches, through one central trunk tunnel with 30 peak trains per hour, and comparable ridership to all three New York commuter rail systems combined, then nobody needs to add 2 minutes to the dwell time for schedule recovery, or 2.5 minutes for interlockings. The RPA is welcome to release its model for why it should take 10 minutes for passengers to board and alight; ours is open for inspection on GitHub and finds that the busier of the two can be done in 3 with a bit of buffer time and the less busy of the two is essentially free since so few passengers ride reverse-peak, and the train can leave even if some passengers that got off it are still on the platform on their way to the escalators.

The Parisian issue

The RPA report doesn’t talk about Paris. It doesn’t rebut the point that the RER is a good comparison case for New York commuter rail capacity, but it’s clearly lurking in the background. It does mention the RER as an ill comparison for the benefits of through-running (see section below), on the grounds that “trains follow each other along the same paths,” whereas in New York there is more and, measured by number of stations if not distance, earlier branching. But it doesn’t really address the point that if the RER can run 24-27 trains per hour per track in each direction, and the Munich S-Bahn can run 30, then so can through-running paths at Penn Station.

Reverse-engineering from what the report does say and from what the biggest points of contention have been when I talked to the RPA on this, they clearly think it matters that the European comparison cases have multiple city center stations to spread the load. Penn Station, in contrast, is the single central business district station, in a high-kurtosis city with far higher job density within walking distance of the station than can be found in any European city.

And yet, as I explained in my second original post from 2024, the effect coming from New York’s single city center station versus Paris’s multiplicity thereof largely cancels out the effect of much higher overall ridership on the RER than on New York commuter rail. Averaged over the peak four hours, the highest resolution I have for Paris, the sum of peak boardings and alightings per train is actually a bit higher on the RER E at Haussmann-Saint-Lazare, as of the 2010s when it was still a terminal, than it was at Penn Station in 2019 just before corona. The same sum at each of Gare du Nord and Gare de Lyon on the RER D is respectively 20% and 15% less than at Penn Station; on the RER B at Gare du Nord, it’s also 20% less. The Gare du Nord numbers are unlinked, so passengers interchanging across the platform between the RER B and D are counted in both trains’ ridership, but from the perspective of the train-to-platform interface, this is still a flow that the train and its doors and platforms must accommodate. If that’s doable in a 55 second dwell time, then trains at Penn Station can unload in 2 minutes at the peak and the rest is just a matter of counting platform-to-concourse vertical circulation elements and adding up their capacities as in the model described in the above section.

The benefits of through-running

Through-running works ideally when there are multiple city center stations, allowing the trains to function as urban rail as well as distribute passengers across multiple destinations. This is well-known to any group pushing through-running in New York, which is why so many such groups advocated for a tunnel connecting Penn Station with Grand Central, the so-called Alternative G in the ARC era in the 2000s, and why Tri-State’s proposal showcases a trunk line from Newark Penn Station to Sunnyside, and why ReThink heavily markets Secaucus and Sunnyside as secondary business centers. Here’s what ETA produced in 2023:

The upshot of this is that even with the dominance of Manhattan, any reasonable through-running system cobbled from existing and under construction infrastructure would unlock commutes from east of the Hudson to Downtown Newark, and from west of the Hudson to Long Island City (the station labeled Queens Junction is essentially Sunnyside Junction, walkable to a large fraction of Long Island City jobs), Flushing, and Jamaica.

It’s important context to understand why the RPA graphics denigrating through-running as limited in use don’t work:

In truth, even setting aside a Penn-Grand Central tunnel (which is doable at MTA construction costs at a lower cost than Penn expansion), the destination in New Jersey is likely to be not just on the North Jersey Coast Line as depicted, but on any of the Northeast Corridor, North Jersey Coast, and Raritan Valley Lines. At present, the first two carry 14 trains per hour at the peak, and are decently likely to be one’s pair east of the Hudson. Similarly, one’s destination east of Penn Station is unlikely to be Long Island as commonly defined to exclude Brooklyn and Queens but rather to be on the trunk in Long Island City or on multiple branches within the city or perhaps on the New Haven Line.

The upshot is a hefty share of the total through-Manhattan market would in fact be served by a through-running system, usually with one-seat rides, or if not then with transfers at Sunnyside or Secaucus rather than New York. Through-running is not about those mythical trips from Oyster Bay to the Jersey Shore that Adirondacker in the comments section mocks, just as it isn’t about trips from Aulnay to IHES on the RER B or from the Marne-la-Vallée branch of the RER A to Saint-Germain-en-Laye. It’s about the overlapping near-center markets, individually small compared with commutes to the central business district and yet collectively significant.

How Much Community Outreach Does the Urban Institute Think Italy and Turkey Do?

The Urban Institute just published a brief about community outreach for public transportation construction projects. The authors are Yonah Freemark, Gabe Samuels, and Christina Plerhoples Stacy; I don’t know the latter two but I know Yonah well and respect him and his expertise and global curiosity. So I honestly don’t understand why the brief concludes that the US’s problem is “inefficient community engagement” and the solution is to do more community engagement early in the process. Worse, the brief for some reason cites Italy and Turkey, via our transit costs reports, as places that do earlier, more effective community engagement, rather than as places with rather top-down decisionmaking and limited citizen voice. It’s sad, because the brief does go over the problems of the American process but can’t bring itself to the right conclusion, namely that community engagement should be curtailed; its only response to the problems of engagement are to suggest earlier engagement rather than less of it.

The brief is, well, brief. I recommend people read it in full. It quickly goes over the usual critiques of the American community engagement process: it is skewed toward higher-income residents, who are likelier to own a car; public meetings attract people with the leisure to attend during business hours; it leads to defensive design such as the reluctance to engage in surface disruption when building subway stations. All of these are real problems. But the brief tries to rescue the idea that public engagement should inform decisions, by criticizing the “decide, announce, defend” mentality of infrastructure project managers and by demanding early engagement.

Then the solutions proposed are a mixed bag. Some are good, and are lumped in with community outreach even when they’re not about any such thing. The invocation of our Italy and Turkey cases is about the public itemization of costs for infrastructure contracts, but this is not about any outreach but rather about contracting transparency for anti-corruption purposes, and the database is not easily legible to the general public. The same is true of in-house expertise, of simplifying project homologation, and of limiting contingencies; unfortunately, the brief frames the latter in the language of “create contingency budget limits” rather than stating the real problem, which is that federal regulations in the last decade began requiring much higher contingency than is normal, 40% rather than the 20% used in Turkey, and these are mirrored in the UK as an ersatz attempt to deal with cost overruns without addressing the underlying cost problems.

In contrast, the invocation of early and representative community engagement is awkward on the same list. The list tries to coast on the reputation of Italy and Turkey among parts of American transit advocacy for low-cost construction in order to justify more involved outreach laws. But neither Italy nor Turkey is a state with strong civil society empowerment in community engagement. Italian infrastructure construction heavily involves different administrative bureaucracies, for protection of labor, the environment, and historical monuments, but not the community. The community does not have the expertise to judge whether some construction technique is a risk to a Renaissance cathedral. Among our low-cost cases, Sweden has higher citizen voice empowerment and even permits some litigation, but far less than the United States or, as of late, the United Kingdom; it too relies extensively on bureaucratic legalism, and the outreach there tends to be done to large collective groups, for example umbrella unions on matters of labor, or a feminist bureaucracy on matters of gender equality.

What makes me bitter about this entire concept is that the good proposals the brief is trying to wed to outreach are about empowering bureaucrats, not civil society. Civil society does not build infrastructure. The administrative state does. Civil society extracts money and betterments from the administrative state whenever it is empowered to do so, because budgets that are third-order for the state are massive wins for petty actors. But 10 third-order items are a second-order item and 10 second-order items are a first-order one, and the costs of veto points and community actors mount. The solution isn’t to involve any such groups early; it is to not involve them at all.

Learning from Many Places is Better than Learning from One

An article from last week about a cost saving push in Seattle made me think again about how learning from lower-cost examples works. As the costs of the majority-above-ground Ballard extension are careening toward the $2 billion/km mark, the agency and civil society are looking for cost savings. Scott Kubly, the former head of the Seattle Department of Transportation, is proposing to reduce the costs of stations by shortening the platforms, citing Copenhagen’s combination of driverless operations and very short trains as a cost saver. Kubly says,

They built it at about a quarter cost. How they did that was shorter, more frequent trains, which leads to smaller stations, which leads to less excavation, which leads to faster delivery and a better passenger experience.

This, to be clear, is at best a second-order saving. The issue is that Copenhagen’s short trains and driverless operations are more or less unique to it, but its medium-low costs aren’t. An honest program of learning from Scandinavia ought to learn from all of the Nordic capitals at once, as much as possible, and focusing on what’s common to them and not on the differences. Stockholm in particular has lower construction costs than Copenhagen these days and has longer trains than Seattle: Stockholm T-bana trains are 140 m long (and Citybanan, with 200 m trains, cost $400 million/km in 2025 prices, less than a quarter as much as Ballard), Link trains are designed up to 116, Copenhagen Metro trains only 39. Nya Tunnelbanan manages lower construction costs than Copenhagen these days, so learning from low-ish Nordic costs should not center the combination of driverless trains and short platforms.

More broadly, this proposal by Kubly (and by the generally good Robert Cruickshank, formerly of California HSR Blog) is convincing me that the real strength of the Transit Costs Project is that we’ve done deep dives into more than one success case. We technically have three low-cost cases: Stockholm, Istanbul, and a selection of Italian cities. But in Stockholm, too, the case looked at trends common to the Nordic countries, namely, the unfortunate tendency in all four to privatize project planning to international design-build consortia. Our conclusion also uses some medium-cost Parisian examples to check itself as well as a few German and Spanish specifics, and we’ve increasingly looked at some Chinese examples, though well short of a full case like our main three. We’re blinder than we’d like to be in East Asia, especially low-cost Korea, but our coverage in Europe is fairly good and we have at least some idea of what’s going on in Latin America. Our ideology is always that it’s most important to look at the commonalities of the places that work when distilling the best practices, rather than on the differences.

This matters, because the average Continental European (or Chinese) just doesn’t think about the United States when doing transportation engineering. A Swede, asked about the features of their transportation program, will focus not on what makes the difference between Scandinavia and the United States, but on the differences between Sweden and the other Nordic countries. The literature I saw when I wrote the Sweden case was replete with intra-Nordic comparisons, on every conceivable measure. Most of the literature came from the country I was researching, but some came from the other three Nordic countries. All of them are like this. Ask a Swedish planner about what makes Nya Tunnelbana work and they’ll cite pertinent features like the drill-and-blast construction method and why it’s superior to tunnel-boring machines, where Copenhagen in fact uses TBMs just fine; ask a Danish one and they’ll talk about the driverless operations and extremely high frequencies.

And this generalizes. Italy has used very short stations and driverless operations to keep down the costs in very small cities like Brescia, and somewhat longer but still short stations in Milan, but Rome Metro Line B1 uses 110 m long trains and was built for $300 million/km in 2025 prices. While Italians talk about this as a cost saver, Manuel Melis Maynar wrote positively of Madrid’s rejection of driverless operations as an example of cost saving through technological conservatism. I don’t doubt that in both cases they were right under the circumstances (Melis having written of extensions built 25 years ago), but just as the medium-low costs of the Nordic countries have to be properly attributed to shared features, the even lower costs in Southern Europe have to be properly attributed.

Far more important than looking at a Copenhagen-specific feature is answering all of the following questions, in increasing order of abstraction:

  • How large are the underground stations compared with the train length? How standardized are their designs?
  • How standardized are the RFPs for station finishes, signals, and electrical and mechanical systems?
  • Is there a single point of homologation for the system, or can a single suburban fire department demand construction in excess of code because the fire department head objects (as happened in Bellevue across Lake Washington)?
  • Does contracting follow good practices (i.e. the traditional system as used in Scandinavia until about 15 years ago), or is everything done with layers of consultants managing other consultants with opaque design-build contracts? Are contracts itemized or fixed-price? Is there technical scoring to ensure contractors race to the top and not the bottom?
  • How large is the in-house planning and supervision team? Does it have the capacity to manage a project the size of ST3, or even a single subway line like the combination of Ballard and West Seattle?
  • Are priorities decided by professionals or by political appointees? Does funding follow the Acquis or possibly another similar system governed by rule of law? Is a cost-benefit analysis mandatory, and do project designers expect that a poor cost-benefit analysis will lead to project cancellation?

Each of these questions, by itself, represents a cost saving larger than anything that could come from shrinking the stations based on driverless operations, a tradeoff that reduces station dig costs but increases systems costs and works in many places but is never going to turn American costs into Southern European or Nordic costs. In contrast, relying on one simple trick like Copenhagen’s requires deep understanding of why each of Copenhagen and Stockholm does what it does, which understanding is unlikely to exist in anyone who doesn’t know that these cities have comparable costs, both a fraction of the United States’.

Why High Speed 2 and Other European Lines Make Fewer Stops than the Shinkansen

At a meeting last week, Borners was asking me why High Speed 2 is designed not to make any stops between London and Birmingham. This distance, about 150 km between Old Oak Common and Birmingham Interchange, would have around 4.5 interstations on the Shinkansen, but in the UK, it runs nonstop. More generally, in Europe similarly long stretches without stations are observed, but not in Japan. This post goes over why; it is not due to poor design on the part of either side, but rather a response to the respective geographies of the countries in question.

Speed classes on the rail network

Modern intercity rail networks have multiple speed classes, comprising not just separation between high-speed trains and trains on the classical lines but also local and express trains on the same line, often the same track. Here, it is useful to go over the difference between Japan and Europe in general and Britain in particular.

On the Tokaido Shinkansen, the fastest trains, stopping only in Tokyo and its immediate suburbs, then Nagoya, then Kyoto, and then Shin-Osaka are called Nozomi and average 210 km/h end-to-end; trains making additional stops are called Hikari and average 178 km/h; trains making all Shinkansen stops are called Kodama and average 132 km/h. Below that class are limited express trains on the narrow-gauge network, all much slower. The original Kodama, inaugurated in 1958 just before the Shinkansen began construction, did the Tokyo-Osaka trip at an average speed of 80 km/h. The fastest trains on classical lines in Japan average around 100-110 km/h, on lines without Shinkansen, while lines parallel to Shinkansen, such as the Tokaido Main Line, are slower as they prioritize regional traffic at an average speed of 60 km/h or so.

Britain, in contrast with Japan, has rather fast trains on its classical network, as do other countries that chose to invest in upgrading their existing network rather than in building high-speed rail on it. The intercity trains today connect London with the major cities of the Midlands and North at average speeds of about 130 km/h, depending on city and line. The West Coast and East Coast Main Lines are both straight, built in the 1830s and 40s when it wasn’t clear trains could even round significant curves without derailing, in one of Europe’s flattest geographies, the exact opposite of Japan with its mountainous terrain and narrow-gauge lines. Both lines were four-tracked in the 19th century, more or less allowing fast intercity trains to run alongside slower regional lines without interference. In effect, trains offering Kodama speeds are available in Britain today.

Germany is in a similar situation to Britain. The topography here is hillier and the lines built slightly later, after engineers figured out that sharp curves were fine at the speeds of mid-19th century steam trains, but it’s possible to squeeze decent speed out of some of them, especially in flatter northern Germany. Berlin-Hamburg, exceptionally, averages around 160 km/h entirely on classical track, with timetabled overtakes between intercity and regional trains and extensive schedule padding to allow for recovery from cascading delays. Other lines average 120-130 km/h, for example to Leipzig and Cologne, so that Kodama speeds are already available, again, and the focus when high-speed rail is built is on the Hikari/Nozomi speed range.

Finally, France, like Japan and unlike Britain and Germany, chose to invest in high-speed rail more than to upgrade its classical lines. But it, too, already had high speeds on its classical lines: from Paris to Marseille and Nice, trains averaged 120 km/h before the TGV opened. Short-range intermediate cities like Dijon already had fairly good service available, so serving them was less important than maximizing the speed of longer-range connections like Paris-Lyon and Paris-Marseille.

The urban geography of Japan, were it in Europe, would thus be serviceable on classical lines. In contrast, the lower speed of classical trains in Japan means that exurban centers like Odawara, Mishima, Utsunomiya, and Takasaki greatly benefit from having local Shinkansen trains available. England is denser than Japan; there are places at similar range from London on or near the main lines, such as Milton Keynes and Northampton, but they already have fairly fast trains to the capital, and will have even faster trains when High Speed 2 opens even though they don’t get stations, because the removal of the intercity trains from the West Coast Main Line will allow for schedule rewrites reducing timetable padding and allowing for faster trip times between London and such intermediate points, which today are lower priorities than higher average speeds to Birmingham and points north.

City size and prioritization

On the one hand, as outlined above, there is less need for Kodama-speed service to intermediate cities in Europe than there is in Japan. On the other hand, regardless of need, such service must take lower priority in Europe, due to differences in urban geography.

In Japan, the Tokaido Shinkansen prioritizes Tokyo-Nagoya-Kyoto-Osaka traffic, as those cities outshine all others. Kodama trains could do the trip significantly faster than they do today, but are held at local stations to let Nozomi pass, and most trains are Nozomi rather than Hikari or Kodama. However, north of Tokyo, the situation flips. There is extensive commuter traffic, seen in much shorter average trips than on the Tokaido Shinkansen. The northern exurbs of Tokyo furnish extensive traffic, while the cities beyond commute range are too small to drive traffic all by themselves.

In the Tohoku region, by far the largest major metropolitan area is Sendai, population 2.3 million. The only other major metropolitan area served by the Tohoku or Joetsu Shinkansen, Niigata, has 1.4 million. The other cities don’t qualify as MMAs; the largest, Toyama on the Hokuriku Shinkansen, has an urban employment area of 1 million, while the others, such as Fukushima, Morioka, and Nagano, have around 500,000 each. Utsunomiya and the Takasaki-Maebashi region have 1.3 and 1.1 million respectively, enough that their needs should drive service planning as much as those of the cities to their north.

Britain is the exact opposite. Its metro area listings are somewhat outdated – I can only find 2001 data, compared with 2015 for Japan in the above paragraph – but we can compute based on metropolitan counties, designed to approximate metropolitan areas for the major secondary cities of England. Birmingham’s West Midlands and Greater Manchester are 3 million each, Leeds’ West Yorkshire is 2.4 million, Liverpool’s Merseyside is 1.5 million; at longer range, counting cities to be served on long extensions on classical lines at lower speed, Newcastle’s Tyne and Wear is 1.2 million, Edinburgh is 900,000, and Glasgow is depending on definitions between 1 and 1.8 million. In contrast with those, there is nothing that populous justifying its own station between London and Birmingham. Milton Keynes is too small, and an exurban station at the intersection with the under construction East West Rail between Oxford and Cambridge would not provide much added benefit over the existing direct express trains from Oxford and Cambridge to London.

The British situation generalizes. In Germany, not only does every line have quite a number of midsize cities at its end and beyond it, but also no city is so large that it sprouts subsidiary metro areas the size of Utsunomiya and Maebashi. When intermediate stops are built on high-speed lines, such as Montabaur on the Cologne-Frankfurt line, they are weak, and serve mostly for political purposes to defray NIMBYism; two such political stations are likewise included in the Hanover-Hamburg line. Even then, with stations that raise construction costs and compromise the alignment and timetable for no good technical reason, the stop spacing on these lines is wider than on the Shinkansen, which speaks to the difference in geography between Europe and Japan.

EU, Germany to Accelerate Rail Investment in Response to Iran War

As the American war on Iran led Iran to close the Straits of Hormuz, taking 20% of global oil deliveries offline, the German government and the EU have both promised that they are going to implement short- and long-term measures to reduce oil consumption and the national security risks it involves. These include investment in electric vehicles and German and European infrastructure packages on public transportation to reduce the extent of driving. Sources close to Chancellor Friedrich Merz added that even if the war is resolved soon and oil deliveries resume, the long-term package will reduce Russia’s oil and gas export revenues and improve European security.

In Germany, deficit spending will be used; members of the SPD left assure the coalition that there will be a two-thirds majority for it as Die Linke is opposed to the Iran war and supportive of both the green transition and visible decoupling from the American-led world order, and the Greens have long been supportive of such investments. The package is said to total 100 billion € in capital construction for urban, regional, and intercity rail, and negotiations are ongoing over the split, with the coalition insisting that nearly all money be spent on U- and S-Bahn extensions including future park-and-rides with electric vehicle charging stations and on high-speed rail lines and negotiators for the Greens demanding more money for regional trains and trams. The Greens’ working group on transportation insists on prioritizing regional trains, since the point of the program is to provide alternatives to the car where the current quality is too low, rather than to make already strong intercity lines stronger.

At the EU level, a package will be used to construct high-speed rail on the core cross-border lines in Western and Central Europe; farther east, financing for member state-led projects will be made available with member states choosing their own priorities, which can even be about issues other than transportation. The highest-priority lines are said to be Utrecht-Rhine-Ruhr, a connection from the LGV Est to Saarbrücken and Frankfurt via Metz and to Karlsruhe and Stuttgart via Strasbourg, completion of high-speed rail across the gaps in Belgium, an acceleration of construction of the connecting high-speed lines on the German and Austrian side of the Brenner Base Tunnel, an acceleration of the remaining gap from Perpignan to the rest of the TGV network, a new line from Bordeaux and Dax to Irún and the Spanish network, and a connection from Berlin to Poznań and the under-construction Polish network toward Warsaw. The German government assures the EU Commission that its own package will include a Berlin-Dresden connection to link with the base tunnel to be built to Czechia, where the Czech network will connect from it to Prague and the rest of the country.

The total cost of all of these lines at the EU level is said to be 50 billion €, but the package is expected to be much larger to include electric vehicle infrastructure and grants to the cohesion countries. Hungary may be included with especial subsidies in the event Péter Magyar wins the election this month, in order to provide him with more economic legitimacy, in which case a program connecting Budapest with the major secondary cities of Hungary as well as Bratislava and Vienna will be announced, a total of 600 km of Hungarian construction at what is estimated to cost 20 billion €.

Quick Note: Commuter Rail and Urban Bus Hubs

I’d like to introduce the concept of bus hubs with commuter rail connections, but in an urban rather than suburban context. This comes from the project we’re working on to about Queens buses on the assumption that LIRR is modernized, but it should generalize. The suburban bus context is a town center that buses converge on from many directions; the urban one is embedded in a much larger street network that must have multiple stops, and therefore the hub must be planned around both bus and rail service at once, rather than in a rail-first hierarchy.

The importance of two-way planning here is that within an urban grid, there usually aren’t obvious town centers, unless they are built. The topic of this discussion is not city center but outer areas where the commuter trains provide walk-on service, or would if they ran at modern (i.e. non-American) frequencies and charged the same fares as the buses and subway. For example, take Eastern Queens in and east of Jamaica:

Jamaica is an obvious node, and nearly all LIRR trains already stop there and the rest should, but farther east, it’s a question of how to set up the LIRR to interact better with the city, including its buses. A lot of questions and tradeoffs have to be addressed all at once:

  • Stop spacing, trading off station access time with trip times. The LIRR on the Main Line has four tracks through Floral Park so any infill station would be local-only, but that still matters for riders from points east like Floral Park and Hempstead. On the Atlantic Branch and the Montauk Line there are only two tracks, and extensive suburban ridership that would be slowed down by additional stops.
  • Intersections with main arterial streets that make for natural bus connection points. The Queens Village LIRR station is at the intersection with Springfield Boulevard, but a little to its west Francis Lewis intersects the Main Line without an LIRR stop, while Hollis is not so well integrated with the street network (it’s near Farmers Boulevard, which more or less parallels the Monrauk Line). On the Montauk Line and Atlantic Branch, strong east-west bus connection points are useful; St. Albans is at the intersection with Linden, but the Atlantic Branch has no such stop at the intersection with Linden.
  • Stops that permit buses to avoid congested nodes, in this case Jamaica; farther north, off-map, the same is true of Flushing. One of the goals of bus redesigns replacing traditional radial networks with grid is to get buses out of the areas where they are the slowest; this was explicit in the replacement of the radial network with the Nova Xarxa grid, increasing average speed even without any other interventions like bus lanes and stop consolidation.
  • Some bus routes duplicate rail routes and could be removed or curtailed, if enough infill stops are built: for example, Merrick and Guy R. Brewer Boulevards are very closely parallel to the Montauk Line and Atlantic Branch respectively.

Without doing more detailed work yet, I suspect that the speed-access tradeoff encourages more infill stops on the Main Line and Atlantic Branches and fewer or even none on the Montauk Line, since the Atlantic Branch only continues to fairly close-in suburbs whereas Montauk carries the Babylon Branch on it, and that the buses can then be moved to a grid to connect with these stations. But the point is that this is not a decision that can be made rail-first, unlike timed connections in secondary cities and suburban town centers. The top-down hierarchy that for example Marco Chitti brings up when setting up a Takt system breaks down when one does coordinated bus-rail planning within the city.

Why IBX Shouldn’t Connect to LaGuardia

Benjamin Schneider has an interestingly wrong proposal for how to extend the Interborough Express, currently designed to run between Southern Brooklyn and Jackson Heights, toward LaGuardia Airport. I know he cares a lot about urbanist issues and public transportation, so I’d like to explain what’s in this proposal, how it errs, and how it is similar to other problematic proposals, for example by the Regional Plan Association in the Third and Fourth Regional Plans in how it tries to make one centerpiece do too many things. We considered a similar plan for inclusion in A Better Billion and chose not to, and instead connect to the airport via the Manhattan-facing Astoria Line carrying the N and W trains.

What’s the proposal?

The idea is to extend IBX to the airport, in the following way:

The plan adds an infill station for a transfer to the LIRR and extends the line in a slightly roundabout way to connect to an infill Northeast Corridor station before veering to the airport. This distinguishes it from early ideas that didn’t make it to A Better Billion, namely a Y from IBX to both the airport and Harlem. The point of this is not just to connect IBX with the airport but also create a hub by connecting IBX to more things, in this case a transfer station designed to connect people from the entirety of the New Haven Line to LaGuardia.

Why doesn’t this work?

The general answer is that subway lines should be radial or circumferential and not mixed, and this is a mix – IBX is circumferential, connecting stations at a fairly consistent distance from Manhattan, and the extension to the Northeast Corridor (or even Harlem, well north of Midtown) would maintain this character, but a tail veering to the airport would suddenly be radial. Such lines always underperform, because they fail at both the function of a radial line, namely connecting outlying areas to city center, and those of a circumferential, namely connecting lines to one another better and providing near-center neighborhoods with additional service orthogonal to the radial direction.

The more specific answer is that we know where passenger demand to LaGuardia is, and it’s nowhere on IBX or for that matter on the New Haven Line. Airport passenger demand is extremely Manhattan-centric, and within Manhattan it centers on Midtown and the Upper East Side:

At the proposed IBX-NEC transfer point, just about every passenger from the airport would transfer to the commuter trains. The required infrastructure to build this might as well be used on a commuter rail branch, going to East Side Access as it is more central for air travelers than Penn Station based on the above map. IBX is more or less useless. Or, better yet, the Astoria Line can be extended as we propose, along an easier alignment that can be done largely above ground.

Now, what about airport workers? Those are usually mentioned, almost always as an afterthought, in various justifications for lines; I heard transit advocates use that line to argue for Andrew Cuomo’s backward air train idea back when he was still governor. Those are still poorly served by an IBX extension. On a map of airport employee residential density, it looks almost good:

The highest-density zip code on the map above is 11372, whose southwest corner is Jackson Heights. But what’s unclear from the picture above is just how circuitous a swing from Jackson Heights to almost Astoria to LaGuardia is. The straight line distance from Jackson Heights to the nearest potential transfer station location to Terminal B is almost twice that of the direct straight line distance from Jackson Heights to Terminal B. The street grid isn’t straight but neither would an IBX extension be, needing to keep going northwest before turning 120 degrees to the east to get to the terminals. For most people in these neighborhoods, IBX would not provide a trip time improvement over buses.

But more conceptually than this, rail improvements aiming to serve airport workers are generally a bad idea, because airport workers never cluster in one residential place on which a line can be built. The paired density maps are at different scales, and the ratio between the densest and least dense colors is much higher for the air traveler density map than for the airport employee one. OnTheMap gives, as of 2023, 11,000-12,000 airport workers, depending on whether one counts hotel workers across the Grand Central Parkway from the airport in the total. Out of 11,666 on a more generous count, only 3,200 even live in Queens and only 1,182 live in Brooklyn. The blob of seven high-density zip codes of worker origin plus the medium-density one between them (11377, just west of Jackson Heights) only furnishes 1,000 airport workers from all eight zip codes combined.

How does this relate to previous proposals?

There’s a tendency in New York planning, at all levels of officialness from the RPA down, to take one big project that’s politically agreed on and hang everything on it. The Third Regional Plan tried to tie in everything to Second Avenue Subway, to the point of bloating it to a four-track line (by the 1990s all planning was for a two-track line). Even commuter rail, in this case a LIRR Atlantic Branch connection to Lower Manhattan, was shoehorned into it, with through-service onto the subway. The Fourth Plan did the same with its Triboro proposal running through to Metro-North in the Bronx and with commuter rail through-tunnels trying to work around Gateway.

The result of such schemes is proposals that try to have a single line do work it cannot possibly do and would be compromised to the point of unusability if it were forced. In Los Angeles, for example, this is leading to a squiggle of a subway extension of the K (Crenshaw light rail) Line through West Hollywood, 6 km longer than it needs to be. In New York, this is leading to taking IBX, as pure a circumferential as one can be, and lading it with tunnels to destinations for which it doesn’t make sense.

It’s important to resist this temptation. If rail service to LaGuardia is desired, it should use the subway line that already points in that direction, whose alignment allows for an elevated extension, just marginal enough to the residential parts of Astoria to avoid NIMBYs, just close enough to still serve the neighborhood well. The overall planning complexity of two good lines is less than that of one bad line – the cost doesn’t magically increase just because the rail link from Astoria to LaGuardia is categorized as “N/W extension” rather than “IBX extension,” and it’s easier to supervise more, smaller projects if they’re parallelizable.