I am embarking on a long-term project to investigate why US construction costs are high using case studies, so everything I’m going to say so far is tentative. In particular, one of my favorite theories for most of this decade seems to be false based on the addition of just two or three new data points. That said, having spent the last nine years looking at topline costs and a few itemized breakdowns does let me reach some initial conclusions, ones that I believe are robust to new data. The context is that some mainstream American pundits are asking why, and I realized that I’ve written more posts criticizing incorrect explanations than posts focusing on more plausible reasons.
1. Engineering part 1: station construction methods
The most important itemized fact concerning American construction costs is that New York’s premium over Paris is overwhelmingly about stations. I have itemized data for a single line in New York (Second Avenue Subway Phase 1) and a single line in Paris (Metro Line 1 extension), from which I have the following costs:
Tunneling: about $150 million per km vs. $90 million, a factor of 1.7
Stations: about $750 million per station vs. $110 million, a factor of 6.5
Systems: about $110 million per km vs. $35 million, a factor of 3.2
Overheads and design: 27% of total cost vs. 15%, which works out to a factor of about 11 per km or a factor of 7 per station
These costs have some reinforcement with other projects in both cities. When New York built the 7 extension, there were calls for an intermediate stop in addition to the single stop built, and at the time the city definitively canceled the extra station, its cost was given as $800 million. Moreover, in Paris, another extension for which I have per-station cost data, that of Metro Line 12, costs €175 million for 2 stations and no tunnels, about $110 million per station, including overheads; the same is true of two more stations not on M12 given in a French report about the costs of Grand Paris Express (PDF-p. 10).
The difference concerns construction methods. In Paris, as well as Athens, Madrid, Mexico City, Caracas, Santiago, Copenhagen, Budapest, and I imagine other cities for which I can’t find this information, metro stations are built cut-and-cover. While the tunnels between stations are bored, at higher cost than opening up the entire street, the stations themselves are dug top-down. This allows transporting construction materials from the top of the dig, right where they are needed, as well as easier access by the workers and removal of dirt and rock. There is extensive street disruption, for about 18 months in the case of Paris, but the merchants and residents get a subway station at the end of the works.
In contrast, in New York, to prevent street disruption, Second Avenue Subway did not use any cut-and-cover. The tunnels between stations were bored, as in nearly all other cities in the world that build subways, and the stations were mined from within the bore, with just small vertical shafts for access. The result was a disaster: the costs exploded, as can be seen in the above comparison, and instead of 18 months of station box-size disruption, there were 5 years of city block-size disruption, narrowing sidewalks to just 2 meters (7′ to be exact).
In London, the Crossrail project was forced to mine stations as well, as it passes underneath and around many older Underground lines. Only one station could be built cut-and-cover, Canary Wharf, built underwater at very deep level. These stations have comparable construction costs to those of Second Avenue Subway. One way around this problem is to build large-diameter bores, as in Barcelona on Line 9/10, which used a bore so big it could fit two tracks with platforms. However, L9/10 has high costs by Spanish standards, and moreover the vertical access to the stations is exclusively by elevator, with lower capacity than escalators and stairs. A technique for slant bores for escalators exists in St. Petersburg, but I do not know its cost.
2. Engineering part 2: mezzanines
The other big problem with American metro construction methods is the oversized stations. This problem also occurs in Canada, where Toronto uses cut-and-cover stations like most of the world and yet has very high costs, as these cut-and-cover stations are palatial. But I do want to caution that this is a smaller problem than station mining, especially in New York. The total amount of excavation in Paris is barely lower than in New York.
But whatever the dig size issue is, one problem persists: American subway stations have mezzanines, usually full-length. This problem goes back to the 1930s. According to a historical review published in JRTR, costs in New York per kilometer rose to $140 million in the 1930s; in the 1910s and 20s costs were only $45 million per kilometer but there was extensive elevated construction, so per underground kilometer they were perhaps $80 million. This contrasts with $30-35 million per km on lines built in London and Paris from the 1900s to the 1930s.
A big cost driver in the 1930s was the higher construction standards. The subway built wider curves, even wider than those used in London and Paris. There were underground flying junctions allowing a complex system of branching on local and express trains to serve many different origin-destination pairs. And stations had full-length mezzanines.
The mezzanines have since turned into an American standard, featuring on all subsequent subways that I know of. BART has them under Market Street. Boston has them at some of the newer stations, alongside high ceilings at parts of stations the mezzanines don’t reach.
Outside the US, cities with such large station digs have high costs as well Toronto has had palatial construction at some of its newer stations, such as Vaughan Metro Center, leading to high costs even with cut-and-cover stations: while the Vaughan extension cost only C$320 million per kilometer, further projects in Toronto are slated to cost far more, including the single-stop Scarborough subway for C$520 million per km (only 18% less than Second Avenue Subway adjusted for station spacing) and the Downtown Relief Line at C$800 million per km.
Moreover, my recollection of riding the MRT in Singapore, another high-cost country, is that its stations are palatial as well, more so than recent American ones, let alone French ones. Singapore has high construction costs: the under-construction Thomson Line is to cost S$600 million per km according to information from 2012, and since then there has been a schedule slip, though I can’t find more recent cost estimates, and I do know of rail infrastructure projects with schedule overruns that stay within budget. Individual stations in Singapore are fairly expensive, with the central one (Orchard) approaching American costs at S$500 million, and in a speech full of excuses for construction costs, Singaporean transport minister Khaw Boon Wan mentioned that the new line has more exits per station, signaling larger station footprints.
3. Management part 1: procurement
The best industry practice, outlined by Madrid Metro’s Manuel Melis Maynar, is to award contracts by a combination of cost, construction speed, and a technical score judged by an in-house oversight team. Moreover, in Madrid there is separation between design and construction, in order to permit construction teams to make small changes as they go along without being wedded to their own plans. With this system, Melis built a wave of metros for an underground construction cost of, in today’s terms, $80 million per kilometer (almost all but not 100% underground), including rolling stock, which I have attempted to exclude from other lines whenever possible.
The American practice is to award contracts by cost alone. This leads to one of two problems, depending on the coast.
In California, the problem is, in two words, Tutor-Perini. This contractor underbids and then does shoddy work requiring change orders, litigated to the maximum. Ron Tutor’s dishonesty is well-known and goes back decades: in 1992 Los Angeles’s then-mayor Tom Bradley called him the change order king. And yet, he keeps getting contracts, all of which have large cost overruns, going over the amount the state or city would have paid had it awarded the contract to the second lowest bidder. In San Francisco, cost overrun battles involving Tutor-Perini led to a 40% cost overrun. This process repeated for high-speed rail: Tutor submitted lowest but technically worst bid, got the contract as price was weighted too high, and then demanded expensive changes. It speaks to California’s poor oversight of contractors that Tutor remains a contractor in good standing and has not been prosecuted for fraud.
In New York, this is not a problem, as the state makes sure to avoid shoddy work through overexacting specs, down to specifying the materials to be used. Unfortunately, this kind of micromanagement reduces flexibility, increasing construction costs in two ways. First, the direct effect raises the hard costs of construction, by about 15-25% plus overheads and contingency according to many contractors interviewed for Brian Rosenthal’s New York Times article on the subject. And second, since many contractors are turned off by the red tape, there is less competition – the 7 extension had just a single bidder – and thus contractors can demand an extra profit on top.
Some American cities try to get around this problem by using design-build contracts. However, these merely move the locus of micromanagement from the public to private sector. Madrid eschews them and prefers using public oversight to macromanage contractors.
While this may well by the single most important institutional factor in New York, it is not universal in the United States. In Boston, a manager at the MBTA, Jaime Garmendia, reassured me that the agency would “would cease to do business with that contractor in a heartbeat” if anyone acted like Tutor.
4. Management part 2: conflict resolution
In Madrid, Melis Maynar insisted on itemizing construction contracts. Thus, every contract would have a pre-agreed cost per extra item if changes were needed. Since changes are inevitable, this provides fast conflict resolution without expensive courtroom battles and without too much risk on the contractor.
I know of one additional example of itemization: in a paper studying electricity generation contracts in India, Nicholas Ryan compares cases in which there was a pre-agreed system for price escalation in case of changes in input prices and cases in which there were one-off negotiations whenever the situation suddenly changed. Pre-agreed escalation based on input prices leads to lower costs, first because there is less risk to the contractor, second because the negotiation happens in a situation in which if the contractor walks away the state can find another without incurring too much of a sunk cost, and third because the process attracts more honest contractors than Tutor.
In the United States, itemizing does not happen. Contracts are by lump sum, and every time a change is needed, there is a new negotiation, which involves lawyers and potentially courtroom litigation. Robert Kagan calls this tradition adversarial legalism, and contrasts it with European bureaucratic legalism, in which regulators and judges have more power than individual lawyers. Kagan gives an example of litigation about the Oakland Harbor dredging project. Tellingly, a civil rights-centered critique of the concept, arguing that adversarial legalism produces more liberal outcomes for minorities and the disabled (in the context of special education) – but when it comes to transit, the United States lags in wheelchair accessibility.
This is not intended as a broad attack on American legalism, although I do think such legalism also leads to worse infrastructure decisions in general. This is a specific attack on the tradition of using lawsuits to resolve conflicts between contractors and the state, rather than agreeing on itemized costs in advance, a technique that is legal in the US and that international firms, which have successfully bid on many American projects at American costs, are already familiar with.
5. Management part 3: project management
Some problems are not about procurement or the law, but purely about managerial competence. In Boston, consensus concerning the Green Line Extension seems to be that its high costs are the result of poor project management. The Green Line Extension’s costs were at one point estimated at $3 billion for 6.4 km of light rail in preexisting mainline rail rights-of-way; it’s so expensive that it was misclassified as a subway in one Spanish analysis, which still found it was a premium over European subways.
The current estimate is down to $2.3 billion, of which $1.1 billion was wasted in the initial project, and only the remaining half is actual construction costs of the restarted project. Several Boston-area insiders, including the aforementioned Jaime Garmendia, explain that the MBTA had no prior experience in managing a large project, and did not hire an experienced manager for it, leading to a pileup of errors. When it finally hired a new manager and a new team and restarted the project, costs fell, but not before a billion dollars were wasted.
The remaining cost of the extension, $190 million per km, is still very high for a light rail line. However, in conjunction with the other problems detailed here, this is not so surprising.
6. Management part 4: agency turf battles
There is little cooperation between different public transit providers in the US in the same region. Usually, the effect is only on operations. Whereas in Germany, Sweden, and Switzerland the fare within a metro area depends on the start and end point and perhaps on whether one rides in first or second class but not on whether one uses a bus, a tram, a subway, or a commuter rail line, in the United States fares are mode-dependent and transfers between separate agencies are not free. Nor do American agencies coordinate schedules between different modes of transit even within the same agency: the MBTA is forbidden to coordinate suburban bus and commuter rail schedules.
While this by itself does not impact construction costs, it can lead to overbuilding when construction for one agency impinges on another agency’s turf. This problem is particularly acute when mainline rail is involved, as there is an institutional tradition of treating it as a separate fief from the rest of public transit: “commuter rail is commuter rail, it’s not public transit,” said MBTA then-general manager Frank DePaola in 2016. Extensive turf battles may also occur between different commuter rail operators run as separate units, for example in New York. The same tradition occurs in Canada, where Toronto regional rail modernization plans came from an overarching planning agency, which had to force the commuter rail engineers and managers to go along.
I covered turf battles in a post from the end of 2017. In short, two distinct problems may occur. First, there may be visible overbuilding: for example, plans for California High-Speed Rail included a gratuitous tunnel in Millbrae, near the airport, in order to avoid reducing BART’s territory even though BART has three tracks at a station where it needs only one or at most two; overall, area advocate Clem Tillier found $2.7 billion in high-speed rail cost savings between San Francisco and just south of San Jose. The same problem afflicts plans for extra regional rail capacity in New York: the commuter railroads do not want to share turfs, forcing the construction of additional station tracks in Midtown Manhattan at great cost.
The second problem is that without coordination of capital planning and operations, schedules for construction may be constrained. I believe that this contributes to the high cost of Boston’s Green Line Extension, which is high by American light rail standards. Without agreement on construction windows, right-of-way modifications such as moving bridge foundations to make room for extra tracks become difficult.
7. Institutions part 1: political lading with irrelevant priorities
There is a kind of overbuilding that comes not from American engineering practices that became accepted wisdom in the 1930s, but from active interference by politicians. I caution that I do not know of any case in which this has seriously impacted tunneling costs, the topic I feel more qualified to compare across the world. However, this has been a problem for other public transportation and livable streets projects, especially on the surface.
When a city announces a new public transit initiative, it comes with the expectation of an infusion of money. Usually this money comes from outside sources, such as higher-level governments, but even when it is purely local, individual stakeholders may treat it as money coming from other parts of the city. In this environment, there is an incentive to demand extra scope in order to spend other people’s money on related but unnecessary priorities. The most common example of this is the demand for street reconstruction to be bundled with light rail and even bus rapid transit.
The advocacy organization Light Rail Now claims that bundling street reconstruction has raised some American light rail costs. Moreover, I know examples of this happening for BRT. The Albuquerque project ART, which I covered in the context of electric buses, is one such example: it cost $135 million for 25 km, of which about 13 km were reconstructed to have wider sidewalks, trees, and street lighting. Moreover, in Tampa, the highway department insists that the transit agency find money for repaving roads with concrete if it wishes to run buses more frequently.
This is not just an American problem: the Nice tramway, which at €64 million per km for the first line is France’s costliest, spent 30% of its budget not on the tramway itself but on drainage, rebuilding a public plaza, and other related but unnecessary amenities.
Commuter rail exhibits this problem in droves. Either local suburbs or agencies that are captive to them insist on building large transit centers with plentiful parking, retail that is not necessary if trains arrive on time, and a sense of place. Spartan stations, equipped only with level boarding, shelter, and a convenient spot for connecting buses to drop people off on the street or at a bus bay, cost a few million dollars apiece in Boston and Philadelphia. In contrast, veritable palaces cost many tens of millions: the four stations of Penn Station Access, in the low-car-ownership Bronx, are projected to cost a total of $188 million per the 2015-9 capital plan (PDF-p. 225); in West Haven, an infill station cost $105 million including land acquisition.
8. Institutions part 2: political incentives
Politicians in the United States do not have an incentives to control costs. On the contrary, if anyone complains, their incentives are to accommodate even if costs rise as a result. While the American legal system favors the state over the individual in property takings, for example in contrast with the Japanese system, the political system favors NIMBYs and really anyone who complains. Infrastructure construction takes a long time and the politician who gets credit for it is rarely the one who started it, whereas complaints happen early. This can lead to many of the above-named problems, especially overbuilding, such as tunneling where elevated segments would be fine or letting agency turf battles and irrelevant demands dictate project scope.
Politicians have the ability to remove obstructive officials, as Governor Andrew Cuomo did when LIRR head Helena Williams opposed Penn Station Access on agency turf grounds. But they rarely have the will to do so. Coordination and good government are not their top priorities. American politicians who are ambitious enough to embark on big infrastructure projects govern their respective states and cities like comets, passing by quickly while expecting to move on to a bigger position within a few years. They can build better institutions if they want, but don’t care to.
This goes beyond individual high-profile politicians. In planning for the NEC Future project, a planner who spoke to me on condition of anonymity said that there was an unspoken assumption that there must not be impact to the richest suburbs in Fairfield County, Connecticut; such impact can be reduced, but not eliminated, and to forestall political controversy with very rich suburbs the process left that segment for later, never mind that it is the slowest portion of the Northeast Corridor today outside major city areas.
This problem can be mitigated by raising the political cost of poor infrastructure construction decisions. One way to do so is using referendums. In Switzerland, all major infrastructure construction must be approved by referendum. Thus, if cost overruns occur, the state must return to the people and explain itself in asking for more money. In contrast, California High-Speed Rail went to ballot on $9 billion (plus $950 million for connecting transit) out of a budget that at the time was estimated at $42 billion in year-of-construction dollars. The state did not need to identify funding sources for the remaining $33 billion, and thus there was no incentive to control costs, as it was not possible to complete the project for the budget on hand no matter what.
9. Institutions part 3: global incuriosity
The eight above factors all explain why American infrastructure costs are higher than in the rest of the world, and also explain high costs in some other countries, especially Canada. However, one question remains: how come Americans aren’t doing anything about it? The answer, I believe, has to do with American incuriosity.
Incuriosity is not merely ignorance. Ignorance is a universal trait, people just differ in what they are ignorant about. But Americans are unique in not caring to learn from other countries even when those countries do things better. American liberals spent the second Bush administration talking about how health care worked better in most other developed countries, but displayed no interest in how they could implement universal health care so that the US could have what everyone else had, even when some of these countries, namely France and Israel, had only enacted reforms recently and had a population of mostly privately-insured workers. In contrast, they reinvented the wheel domestically, coming up with the basic details of Obamacare relying on the work on domestic thinktanks alone. The same indifference to global best practices occurs in education, housing policy, and other matters even among wonks who believe the US to be behind.
This is not merely a problem in public policy. In the private sector, the same problem doomed the American auto industry. American automakers have refused to adopt the practices of Japanese and German competitors even after the latter produced small cars better suited for post-1973 oil prices. They instead dug in, demanded and got government protection, and have been in effect wards of the American federal government for about 40 years.
American business culture does not care much for imitation, not does American society give high prestige to people who perfect something that someone else invented. The industry that teaches how to adopt best practices, consulting, has poor reputation in American culture. Instead, Americans venerate founders and innovators, an approach that works in industries where the US is in the global frontier, like tech or retail, but not in ones where it lags, like cars and the entire public sector. To avoid learning from others, Americans end up believing in myths about what is and isn’t possible: they insist they are so much richer than Europe that they have nothing to learn from across the Pond, and hang all their hopes on any flim-flam artist who comes from within American business culture who insists there is no real need for public transit or any of the other things Europe and high-income Asia do better.
In transit, we see it in politicians and agency officials who say things that are so funny they are sad, or perhaps so sad they are funny. Richard Mlynarik tells me of an official at either Caltrain or the California High-Speed Rail Authority, I forget which, who did not know Germany had commuter trains. Another Caltrain official, confronted with the fact that in Japan trains turn faster than Caltrain thought possible, responded “Asians don’t value life the way we do” – never mind that Japan’s passenger rail safety per passenger-km is about 1.5 orders of magnitude better than the US’s. In stonewalling about its safety regulations, since positively reformed, an FRA official insisted American trucks are heavier than European ones, where in fact the opposite is the case. Boston’s sandbagged North-South Rail Link process included a best practices section but insisted on only including North American examples, since European ones would make America look bad. To advocate for transit among Americans is to constantly hear things are not possible that in fact happen in various parts of Europe on a daily basis.
Canada is not much better than the US. Americans’ world is flat, with its corners in Boston, Seattle, San Diego, and Miami. Canadians’ world includes the United States and Canada, making it flat with the northern ends of the quadrilateral stretched a few hundred kilometers to the north. A study of a long-overdue extension of Vancouver’s Millennium Line to UBC has four case studies for best practices, all from within North America. This is despite the fact that in the developed world the system most similar to Vancouver’s SkyTrain in technology and age is the Copenhagen Metro, whose construction costs are one half as high as those of Vancouver despite cost and schedule overruns.
Meiji Japan sent students to the West to assimilate Western knowledge and catch up, avoiding the humiliations inflicted upon China in the same era and instead becoming a great power itself. The historian Danny Orbach, who wrote his dissertation on the historical arc leading from the Meiji Restoration to Japan’s World War Two atrocities, argues that Japan was able to modernize because it understood early that it was not at the center of the world, whereas China and the Ottoman Empire did not and thus only realized they were technologically inferior to the West too late, at the signing of the unequal treaties or at dismemberment. The United States at best thinks it’s the center of the world and at worst thinks it’s the only thing in the world, and this has to change.
Can this be reformed?
The answer is absolutely. There are no examples of good transit under construction in the United States, but there are many partial successes. The California State Rail Plan is moving toward coordinated planning, and Massachusetts has some inklings of reform as well. Boston’s ability to restart the Green Line Extension is to be commended, and the large gap in cost between the original project and the current one should encourage other American transit agencies to hire good project managers with a track record and pay them competitively; paying high six figures to a manager or even more can easily justify itself in ten-figure savings.
The legal problems can be reformed as well without turning the United States into something it is not. Politicians would have to be more courageous in telling constituents no, but so many of them have no chance of losing reelection that they can afford to piss off a small proportion of the population. Contracts could include itemized costs to control change orders. California already awards contracts based on a mix of cost and a technical score, it just needs to adjust the weights and figure out how to avoid doing business with Ron Tutor, and if possible prosecute him.
However, all of this depends on solving the last of the above nine problems. Americans have to understand that they are behind and need to imitate. They can try to innovate but only carefully, from a deep understanding of why things are the way they are in such global transit innovation centers as Spain, South Korean, Japan, Switzerland, and Sweden. They have to let go of the mythology of the American entrepreneur who does not listen to the experts. They can solve the problem of high construction costs if they want, but they need to first recognize that it exists, and that internal politics and business culture are part of the problem rather than the solution.