Why American subways cost so much to build

Why the Tunnel Is the Easy Part and Oversized Stations, Overstaffing, and Procurement Make American Subways Scarce

Subway construction costs in the United States are among the highest in the world, and that single fact quietly decides how many people can live close to good jobs, schools, and each other. This page serves futurism in transport and abundance and post-scarcity in a practical form. Its angle is that American cities do not lack the money or the tunneling machines to build abundant rail; they lack the institutional habits that let Madrid, Stockholm, and Istanbul build many kilometers for the price of one. If that is right, rail scarcity is a choice that can be reversed, and the fix is learning rather than spending.

The gap in plain numbers

A 2019 presentation from the Transit Costs Project at New York University's Marron Institute compared recent subway extensions. New York opened two extensions in the 2010s, the 7 line extension and the Second Avenue Subway, for roughly 1.3 to 1.6 billion dollars per kilometer. Recent Paris Metro extensions ran about 250 million dollars per kilometer, and the most recent Madrid Metro extensions, on Lines 2 and 9, cost about 85 million dollars per kilometer.[1]

The same presentation set the totals side by side. New York planned to spend about 35 billion dollars between 2005 and 2030 on subway and commuter rail expansion and receive about 15 kilometers of new tunnel. Paris planned to spend about 40 billion euros over a similar period for 228 kilometers, 187 of them underground. Madrid, a much smaller city, spent about 10 billion euros from 1995 to 2015 on 234 kilometers, around 180 of them underground. The presentation's conclusion was direct: the limiting factor to subway expansion is cost.[1]

That is the scarcity story in one comparison. A city that pays ten times more per kilometer does not simply get a more expensive network; it gets a much smaller one, and most neighborhoods never get a station at all.

Where the money actually goes

The project's final report, published in 2023 after case studies of Boston, New York, Istanbul, Italy, and Stockholm, estimated that Phase 1 of the Second Avenue Subway cost 8 to 12 times more than a composite baseline built from its low-cost cases.[2] The surprising part is what drives the gap. It is mostly not the tunnel.

The report breaks the premium into factors that multiply together, so several premiums of 25 to 50 percent compound into an order of magnitude.[2]

  • Oversized stations, a factor of about 2.06. Stations made up 77 percent of the hard costs. The dig for the 96th Street station is 2.6 times longer than its 187 meter platform, and the other two station digs are about twice the platform length and use a more expensive deep mining method. In the comparison cities, station digs ran only 3 to 20 percent longer than the platforms.
  • Custom systems and finishes, a factor of about 1.35. Each of the three stations was designed separately, with two different escalator contractors and different numbers of exits, crossovers, and elevators. Systems and finishes came to roughly 46.5 parts against 53.5 parts of civil works, where Paris and Rome run close to 15 to 40.
  • Labor share, a factor of about 1.5. Cost estimators told the researchers that labor is 40 to 60 percent of hard costs in the American Northeast, against 19 to 30 percent in Turkey, Italy, and Sweden. The report attributes the gap to overstaffing of both field crews and office staff, including white collar overstaffing of 40 to 60 percent in Boston, and says at most a quarter of the difference comes from differences in pay.
  • Procurement, a factor of about 1.85. Secrecy and distrust between agencies and contractors, thin in-house capacity to manage contractors, weak competition, and attempts to push all risk onto private builders lead bidders to price in red tape, delay, and defensive design.
  • Soft costs, a factor of about 1.2. Third party design and project management added 21 percent to Second Avenue Subway hard costs, against a typical 7 to 8 percent in the comparison cases.

A follow-up post by one of the authors asked readers to quote the premium as a factor of 9 to 10, not as the falsely precise 9.34 that the multiplication produces.[3] The same post notes that the tunneling premium in New York is low compared with the station and systems premiums. In other words, the boring machine is the least of the problem.

Why this is an abundance question

The report argues that rapid transit is unusually valuable because it lets cities grow without choking on traffic. It cites research estimating that building more housing in high-demand cities such as New York and San Francisco could add about 1.4 percent of GDP when the transport network stays fixed, against a far larger estimate of 13.5 percent in work by Hsieh and Moretti, and it suggests the gain is closer to about 10 percent nationwide when transit can grow with the city.[2] Those are model estimates, yet the direction is clear. When rail costs fall, both housing and access to work become less scarce.

For a household, this shows up as a problem in living rather than an engineering chart. A neighborhood without a station means long bus rides, a car payment many families cannot carry, or rent paid for location instead of space. A city that could build ten stations for the cost of one would change those daily constraints for far more people.

The money at stake is large. The report notes that the 2021 Bipartisan Infrastructure Law set aside between 91 and 108 billion dollars for non-mainline public transit with usual appropriations included.[2] At New York unit costs that buys a few dozen kilometers of subway. At Madrid unit costs it would buy many hundreds.

What low-cost builders do differently

The report's recommendations read less like a construction manual and more like advice on how institutions learn.[2]

  • Keep a permanent, well-staffed engineering team inside the public agency, funded from regular budgets, so that the agency can design, supervise, and remember what it learned instead of renting that knowledge from consultants on each project.
  • Standardize station designs and systems so lessons from one station carry over to the next. Standardized does not have to mean bleak. The report points out that Stockholm built its T-bana with standardized stations and distinctive art for about 3.6 billion dollars in 2022 terms, around one fifth the per kilometer cost of the Washington Metro built in the same era.
  • Build stations as shallow and as short as the chosen method allows. Shallow stations are less expensive and also faster for riders, who spend less time on escalators.
  • Publish maps, drawings, itemized costs, and contracts in readable form without forcing the public through records requests.
  • Learn from peer agencies abroad, including sending junior and mid-level planners to conferences rather than only senior managers.

The report also warns that a family of reforms popular in English-speaking countries, built around shifting risk onto large private contractors, has tended to raise costs wherever it spread, naming England, Singapore, Hong Kong, and Canada.[2]

Honest limits

The comparisons rest on contract documents, budgets, and hundreds of interviews, not a controlled experiment, and the authors themselves describe the factor breakdown as less precise than its charts suggest. Labor premiums in their own data range from 30 to 70 percent around a central value of 50. Cities also differ in geology, existing utilities, train length, and accessibility rules, so a factor of ten is a broad estimate, not a law. The low-cost cases are not frozen either; the Stockholm chapter itself examines why costs there are now rising. Finally, the labor finding concerns staffing levels and agency turf conflicts, and the report says pay explains little of the gap, so it offers no case for cutting wages.

  • Explain why a station dig 2.6 times longer than its platform can double the cost of a whole line, and suggest who inside an agency should have the authority to stop that kind of overbuilding.
  • Compare the totals for New York, Paris, and Madrid and estimate how many additional stations New York could have built at Madrid unit costs.
  • Discuss whether public agencies that keep their own engineers act like a commons of knowledge, and what happens to that commons when each project is handed to outside consultants.
  • Describe how standardized station designs could still give each neighborhood a sense of identity, using the Stockholm art stations as a starting point.
  • Weigh the claim that pushing risk onto private contractors raises costs, and design a contract approach that shares risk more evenly between an agency and a builder.
  • Imagine your own city cut rail construction costs by a factor of five, and outline how housing, commuting time, and car ownership might change over twenty years.
  • Investigate whether the same habits behind high subway costs, such as bespoke design and thin in-house expertise, also appear in hospitals, schools, or housing projects near you.

See also

References

  1. ↑ 1.0 1.1 Alon Levy, "What is the Cost of Building a Subway?", presentation, NYU Marron Institute, November 19, 2019. https://marroninstitute.nyu.edu/uploads/files/costspresentation2.pdf
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 Eric Goldwyn, Alon Levy, Elif Ensari, and Marco Chitti, Transit Costs Project: Understanding Transit Infrastructure Costs in American Cities, NYU Marron Institute, final report updated February 11, 2023. https://transitcosts.com/wp-content/uploads/TCP_Final_Report.pdf
  3. ↑ Alon Levy, "Our Construction Costs Reports are Out!", Pedestrian Observations, February 6, 2023. https://pedestrianobservations.com/2023/02/06/our-construction-costs-reports-are-out/