Agencies without Agency: How Dispersed Power Derails Transportation Infrastructure - American Affairs Journal

The United States has the unfortunate distinction of having some of the highest transportation infrastructure costs in the world. Urban rail costs in the United States are 50 percent higher than in Germany, double the costs of Norway, and nearly triple the costs of Sweden.¹ Phase One of New York’s Second Avenue Subway was built at the eye-watering cost of $2.5 billion per mile, eight to twelve times the cost of similar projects in European countries.² Phase Two is budgeted even higher, at $4 billion per mile.³

Outside of New York, an extension of Chicago’s Red Line is expected to cost more than $1 billion per mile, which transit researcher Alon Levy describes as “almost a world record for an elevated line.”⁴ California’s high-speed rail line, aimed at eventually connecting San Francisco and Los Angeles, is projected to cost over $200 million per mile, or over $126 billion altogether.⁵ This is roughly four times the cost of building high-speed rail in other countries (which average roughly $54 million per mile) and ten times the average costs of building high-speed rail in Spain.⁶

The exorbitant costs of rail construction in the United States mean that many such projects never get off the drawing board. In cities such as Austin, Atlanta, New York, and Philadelphia, enormous projected costs have resulted in rail projects being canceled or scaled back significantly.⁷ And it’s not only rail infrastructure. Road tunnels, highway bridges, and roadways are all more expensive to build in the United States than in major European countries.⁸ A study by economists Zach Liscow and Leah Brooks found that the inflation-adjusted cost of American highway construction tripled between the 1960s and the 1980s.⁹

Even something as simple as bus transportation is unusually difficult and expensive in America. In San Francisco, it took nearly twenty years and $300 million to build a two-mile bus lane on Van Ness Avenue.¹⁰ The average diesel bus in the United States costs around $500,000, more than twice the cost of diesel buses in European countries. Electric buses average closer to $1 million, three times the cost of the price of comparable buses in Asia.¹¹

Wherever we look, we see the same thing: American transit agencies struggle to rein in costs when building transportation infrastructure. In the United States, we pay far more for our transit systems than virtually anywhere else in the world.

Solving this problem means understanding what’s driving it. A confluence of factors makes U.S. transit agencies ineffective at controlling costs and delivering projects. First, over the last several decades, transit agency authority has been increasingly dispersed, checked, and otherwise restrained. Second, at the same time, agencies have been under pressure to outsource as many of their functions as possible, resulting in their technical and oversight capabilities being hollowed out. These two trends, along with the bad incentives to which they contribute, have caused transit costs to spiral out of control.

Democratic Dispersion

In 1919, a young Dwight Eisenhower, then a major in the U.S. Army, joined the military’s first cross-country caravan undertaken by car and truck. A convoy of 310 men (including a fifteen-piece brass band) drove from Washington, D.C. to San Francisco, three thousand miles away. Thanks to the poor state of the country’s roads, which ranged from “average to non-existent,” the trip took over two months, with the convoy averaging just five miles per hour.¹²

The trip, which the New York Times would later describe as taking “60 days and 6000 breakdowns,” impressed upon Eisenhower the need for an “adequate, all-weather road US road system.”¹³ This impression was strengthened in 1945, when Eisenhower, holding the post of Supreme Commander of the Allied Forces, traveled on Germany’s Autobahn after the country’s surrender in World War II. Germany’s rail transportation network had proven relatively easy to disrupt by way of Allied bombing raids (a single bomb could take a track out of service for days), but the country’s wide, robust highways proved harder to break: even a bombed out road could still be used by cars and trucks.¹⁴ A major U.S. highway system thus wouldn’t just make travel easier; it would be a valuable tool for defense. When Eisenhower was elected president in 1952, he decided that a national highway system would be a top priority, and in 1954, he asked members of his administration for a plan to get “50 billion dollars worth of self-liquidating highways under construction.”¹⁵

After several years of politicking and horse-trading, Eisenhower largely got his wish. In June of 1956, the Federal-Aid Highway Act was signed into law, which authorized $25 billion for a forty-thousand-mile National System of Interstate and Defense Highways, funded by an increase in the federal gas tax.¹⁶ A few months later, construction of the first segments began, and the first eight-mile stretch of interstate highway opened in November of 1956 near Topeka, Kansas.¹⁷ By 1964, nearly half of the new Interstate system was open or nearing completion.¹⁸

Initially, the new interstates were by and large met with enthusiasm. Newspapers printed headlines such as “Interstates a Boon to Iowa,” and property values near the highways rose.¹⁹ In his history of the Interstate, Tom Lewis notes that in 1956, “across the country, editorials complained not that the highways were being built, but that they were not being built fast enough.”²⁰

But the scale of the interstate project, which required the government to acquire 1.5 million acres of land via purchases and eminent domain, meant that opposition inevitably appeared.²¹ Conservationists were upset at the loss of scenic beauty as highways were carved through the landscape; residents of rural areas were suspicious of what it would do to their way of life. Citizens “whose land and houses were being ravaged by interstate construction” began to mobilize opposition.²² This was exacerbated by the fact that interstate planning took place almost entirely based on what highway engineers believed would be most efficient; public reaction to the large-scale construction project was at best a distant concern, and more likely to be ignored altogether.²³ Nothing more clearly illustrates this mindset than a 1963 plan to carve an interstate route through the Bristol Mountains in California by detonating twenty-two nuclear bombs, instantly creating a channel 325 feet wide and eleven thousand feet long, while saving an estimated $8 million in construction costs. Only uncertainty as to how long it would take the radiation to dissipate stopped this plan from going forward.²⁴

Early on, increasing opposition to interstates was blunted by the fact that many of the early segments were built in lightly populated rural areas.²⁵ But as interstate construction began to encroach on major cities—by the late 1960s, over sixty thousand homes a year were being torn down for highway construction—citizen resistance grew more fervent.²⁶ In 1964, tens of thousands of protestors assembled in Golden Gate Park in San Francisco to protest against any new freeways through the city, and several planned highways were cancelled.²⁷ In 1969, the Riverfront Expressway, an elevated interstate segment that would have cut across the city of New Orleans, was cancelled following intense activist and local citizen opposition.²⁸ Similar opposition sprung up in cities across the country: from Los Angeles to Seattle to Miami to Boston.²⁹ By the 1970s, highway construction was widely unpopular.

And opposition to highway construction had teeth. The 1969 National Environmental Policy Act (NEPA) required that the environmental impacts of any major government action (including highway construction) be studied, and citizens soon began to use this as a tool to delay or block highway projects. In 1971, a suit was filed to prevent an interstate in Washington state from being completed on the grounds that they had not properly complied with NEPA.³⁰ Today, the Federal Highway Administration (FHWA) performs more NEPA-required environmental studies than any other federal agency except the Forest Service.³¹

NEPA was just one of many laws and court rulings that gave citizens the power to slow down or stop highway construction projects. In 1966, Congress passed the National Historic Preservation Act (NHPA), which required federal agencies to take into account the effect a project may have on historic places.³² The NHPA was successfully wielded by activists against New Orleans’ Riverfront Expressway. The French Quarter through which the highway would be built had been designated a national historic landmark in 1965, and a lawsuit filed against the project claimed that the effect on said landmark had not been properly considered as required by NHPA.³³ That same year, the 1966 Department of Transportation Act required highway projects avoid construction through public parks, recreation areas, or wildlife refuges unless there was “no feasible and prudent alternative.”³⁴ Citizens in Memphis, Tennessee, successfully sued to stop a highway being constructed through a public park in what became a landmark Supreme Court decision,  Citizens to Protect Overton Park v. Volpe.³⁵ The 1973 Endangered Species Act made it possible to shut down projects that might harm endangered species, and it was used to stop construction of the I-10 in Mississippi when a lawsuit was filed claiming that it would cut through the only remaining habitat of the Mississippi sandhill crane.³⁶ Construction only resumed after the Mississippi Department of Transportation agreed to purchase and protect 1,960 acres of land adjacent to the Interstate, which today is the Mississippi Sandhill Crane National Wildlife Refuge.³⁷ As the number of environmental laws continued to rise throughout the 1970s, so too did citizens’ ability to prevent new highway construction.

As a result, highway engineers increasingly found themselves forced to take citizen preference into account when designing highway projects, rather than simply designing whatever would be cheapest to construct.³⁸ The 1970 Federal-Aid Highway Act required the secretary of transportation to “create guidelines that would assure a full consideration of the social, economic, and environmental effects of a federally funded highway,” and it also required highway designers to consider the noise effects of highways on the surrounding area.³⁹ In 1973, the Department of Transportation hired consultants to determine how best to involve citizens in the planning process.⁴⁰ A Federal highway department history notes that “The highway agencies that had once considered themselves premier engineering organizations soon found their transportation departments staffed with ecologists, biologists, planners, and a wide range of other specialists who could help work through the thickets of environmental law.”⁴¹

In his book Why Nothing Works (2025), scholar Marc J. Dunkelman characterizes this change as part of a broader shift in U.S. governance. Dunkelman describes American governance as pulled between two contradictory impulses: the “Hamiltonian” desire to centralize government power and give it the ability to accomplish things, and the “Jeffersonian” desire to disperse and check government power in order to minimize its ability to harm its citizens. During the period from roughly the late 1930s (starting with the New Deal) to the 1960s, the Hamiltonian impulse was ascendant, as large, powerful bureaucracies, often operating with little to no oversight, wielded unprecedented power in shaping American life. Emblematic of this era was Robert Moses, New York City’s transit baron and head of the Triborough Bridge and Tunnel Authority, who figuratively and literally bulldozed through opposition to build the highways, bridges, and other infrastructure he thought the city needed.

But, in part as a backlash against the abuses of Moses and similar figures, since the 1960s, the Jeffersonian impulse has held the upper hand in U.S. politics. Government power has been dispersed, checked, and reined in. Power that might once have been concentrated in one agency (or, in the case of Moses, a single man) was spread more thinly, “pushed down and out to those who might well have objected.” As Dunkelman notes, “the once powerful bureaucrat’s burden under the new regime wasn’t to engineer a road, or a rail line, or whatever else—it was to navigate the requirements and obstacles, to create a record of each factor contributing to a decision, and to anticipate the lawsuits that were sure to be filed. . . .”⁴²

This dispersal of power and the ability of citizen opposition to slow down, stop, or reroute highway projects made it increasingly expensive to build highways. Economists Zach Liscow and Leah Brooks found that the inflation-adjusted cost per mile of interstate highways rose from around $8 million per mile in the 1960s to over $30 million per mile in the 1990s.⁴³ Increases in the prices of labor and materials and increases due to changes in the physical difficulty of construction (such as needing to build through wetlands or existing urban areas) explain little of the overall increase. Brooks and Liscow conclude that one of the main drivers of cost was the rise of “citizen voice,” a combination of “social movements, legislation, and judicial doctrine” that gave citizens ever more influence over the construction of interstate highways starting in the 1970s.⁴⁴

As an example, Brooks and Liscow describe the construction of a twenty-eight-mile segment of I-696 in North Detroit. The first portion of the highway was completed in 1964 at a cost of $13 million per mile in 2016 dollars. The second portion was completed in 1979 at a cost of $48 million per mile. And the final section was completed in 1989 at $86 million per mile, nearly seven times the cost of the 1964 segment. The final section, which would cut through a Jewish neighborhood in the suburb of Oak Park, was so expensive in part because of extensive citizen resistance: Michigan law required city approval of interstate routes that passed through them, and citizens were able to lobby their local governments to oppose the project. Only by extensively modifying the final leg of the segment could approval be obtained:

In the end, the final compromise in 1981 that allowed the highway to go forward required the state to 1) hire a rabbi to consult on the project, 2) depress the entirety of the middle leg, 3) build three 700ft long plazas above the depressed highway, 4) install noise walls along most of the route (they had already agreed to do this near the zoo, but expanded the reach), and 5) install a network of pedestrian paths.⁴⁵

This sort of citizen opposition and hamstringing of agency authority isn’t limited to interstates. In the 1960s, Washington, D.C., planned an extensive metro rail system, in part due to rising citizen opposition to a planned highway that would encircle the city: the final metro plan included nearly one hundred miles of rail transit and was projected to cost $2.5 billion when it was completed in 1980. To build and manage the system, a new agency, the Washington Metropolitan Area Transit Authority (wmata) was created.

But wmata lacked the authority to simply build the route it deemed best, and thus routinely needed to make expensive changes to the transit plan to secure the approval of other organizations and government agencies. The architect of the Capitol refused to allow a station to be built under the legislative building, forcing the route to be moved a few blocks away. The University of Maryland denied the construction of a station on its campus, requiring a complex redesign of the project and a new station to be built elsewhere. Tunnels for the metro needed to be routed around projected future tunnels being planned by the highway department. A planned station at Farragut Square, a small park near the White House, was rejected by the National Park Service because it would require a statue of David Farragut to be moved, forcing wmata to build two stations, Farragut North and Farragut West, a few hundred feet apart.

wmata was also beleaguered by citizen opposition and lawsuits. One citizen lawsuit forced wmata to hold public hearings for any construction decision that would affect private property; another forced it to redesign the system to be wheelchair accessible, adding $65 million to the cost of the project as elevator shafts were hacked through already placed concrete. A 1973 lawsuit forced the project to retroactively comply with NEPA, delaying the project as the various environmental reports were prepared and adding an estimated $120 million to its overall cost. At one point, engineers were forced to change the planned route to accommodate a single historic tree. In his history of Washington’s metro, Zachary Schrag notes that “every park, every historic structure, every sidewalk could become an obstacle,” and accommodating the demands of various interest groups steadily drove up costs “a few million dollars at a time”:

Burying the line near the stadium to meet neighbors’ demands added $18 million. Diving under a road, at the insistence of the Virginia Department of Highways, cost $8.5 million. . . . The changes may have benefited the region as a whole, but they destroyed the engineers’ projections. As Jim Caywood of De Leuw, Cather [an engineering consulting firm on the project] put it, “there’s no way in this world that you can build a mammoth public works project such as Metro within a reasonable budget with all the outside influences. They won’t let you do it.”⁴⁶

When the metro was finally completed in 2001, its cost had risen to $10 billion, rather than the $2.5 billion originally planned. It was called by some “the biggest boondoggle in the history of all mankind.”⁴⁷

We see these exact same issues at work today with California’s High-Speed Rail project, which is supposed to build a high-speed rail line connecting the cities of San Diego, Los Angeles, San Francisco, and Sacramento.⁴⁸ The project first began to take shape in 1996 when the state legislature passed the High-Speed Rail Act; this was followed in 2008 by the passage of Proposition 1A, which authorized issuing $9.5 billion in bonds to begin planning and construction. But from the very beginning, opposition from citizen groups, conservation organizations, and local governments impeded the project.

California also has its own stricter, state-level version of NEPA, the California Environmental Quality Act (CEQA), which has been wielded as a powerful cudgel against the project. A 2008 lawsuit filed by two California cities and four conservation groups argued that the required environmental impact report for one section of the rail line had been insufficiently thorough, forcing the California High-Speed Rail Authority (chsra) to rescind and rework it.⁴⁹ When the revised report was released in 2010, another lawsuit was filed arguing it was still deficient, forcing the authority to rescind and rework it a second time.⁵⁰ A 2015 lawsuit was filed to stop electrification of one existing section of rail, filed by citizens who objected to the visual impact of the overhead wires; this legal obstacle delayed the project for nineteen months until it was thrown out by a judge.⁵¹

Collectively, there have been dozens of CEQA lawsuits filed against the project, each one delaying the project and adding expense as plaintiffs have extracted various accommodations⁵²: $500,000 to a church in Kern County for noise mitigation renovations; millions of dollars and a change in route to Dignity Health, who owns a hospital along the route; an elevated section of track for the city of Shafter so the line would not interrupt traffic through the city, as well as grade separations for adjacent freight trains;⁵³ $6 million for legal fees and conservation easements for farmers in Madera and Merced counties; $10 million to Kings County to pay for the cost of moving a fire station; $1 million to the city of Corcoran “to make up for the aesthetic effects from the rail route.”⁵⁴ In Northern California alone, chsra has estimated it spent 2,400 hours fighting CEQA lawsuits.⁵⁵

Beyond CEQA, the chsra has been hindered by its limited authority, which has forced an arduous and expensive process of approval-seeking from various other organizations who have the necessary jurisdiction, and who may not be particularly motivated to be compliant. A 2025 report from the Inspector General of California for High-Speed Rail noted that the authority “lacks leverage to help ensure that third parties do not themselves needlessly delay negotiations.”⁵⁶ Securing approval to build a portion of the line through the city of Wasco added $26 million across thirty-seven different change orders, and five years of delay.⁵⁷ In unincorporated Madera County, securing construction permits required $30 million worth of change orders.⁵⁸

A 2025 report from transit nonprofit Circulate San Diego, entitled The Powerless Brokers: Why California Can’t Build Transit, noted that these issues of authority are endemic to the construction of transit infrastructure in California. Unlike the famous “Power Broker” Robert Moses, who simply forced his way through any opposition, California transit agencies must secure approval to build from numerous third parties: local governments, utilities, special districts, and so on, all of whom can extract their pound of flesh. The report notes that “While transit authorities have been tasked with building transit, to a significant degree they are powerless to do so.”⁵⁹

Transit agencies that build low-cost transit infrastructure, by contrast, are often immune from these sorts of power restrictions that hamper their effectiveness. Spain, for instance, famously builds some of the least expensive transit infrastructure in the world: the Madrid subway lines built between 2003 and 2007 cost on average just $114 million per mile in 2023 dollars, compared to the $2.5 billion per mile of the New York Second Avenue Subway.⁶⁰ Spanish transit construction is free from many of the authority issues that hinder U.S. projects. Spanish law allows the government to take required land via eminent domain before compensation is determined, and disputes around compensation can’t hold up the project. In the United States, litigation around eminent domain acquisitions can drive up costs and delay projects for years. For California High Speed Rail, property compensation for eminent domain acquisitions in the Central Valley area alone ultimately amounted to $4.5 billion, far more than the $332 million originally estimated.⁶¹

Spain’s railway authority, which builds intercity and commuter rail, is the Administrador de Infraestructuras Ferroviarias (ADIF); it is able to override local zoning, and any activity that falls within railway infrastructure zones is under the ADIF’s purview, eliminating the need to negotiate with overlapping authorities. Spanish environmental reviews are also less burdensome than in the United States; it is difficult to use the judiciary to slow down projects. Environmental reviews can’t be challenged on their own (only the entire project approval can be challenged), lawsuits do not suspend construction, and courts are unlikely to grant injunctions against approved infrastructure projects.

Beyond citizen opposition and dispersal of power, U.S. transit agencies find their hands bound by a variety of other restrictions that drive up infrastructure costs. The Davis-Bacon Act requires that federally funded transit projects pay contractors “prevailing wage rates,” which are often based on higher local union wage rates rather than true market rates.⁶² “Buy American” laws require federally funded transit projects to purchase often more expensive American-made components. One study noted that requiring American-made steel added $2 billion to the cost of U.S. highway projects between 2009 and 2011. Another study noted that American-made buses were about twice the price of buses manufactured in Japan and South Korea.⁶³ Transit agency agreements with labor unions often dictate unnecessarily large labor forces on transit construction projects: the union agreement for construction on the Second Avenue Subway required staffing levels up to four times that of subway construction elsewhere in the world.⁶⁴

Building code requirements are often unnecessarily strict: U.S. transit tunnels, for instance, require cross passages, i.e., tunnels that connect two parallel train tunnels, to be spaced much closer than is required elsewhere in the world, entailing more expensive tunneling.⁶⁵ Contracting rules often require agencies to select the “lowest responsible bidder,” rather than the bidder that is likely to perform the best. This can have the effect of driving up construction costs if the bidder then performs poorly and various change orders are required to complete the project. (Robert Moses, by contrast, simply picked whichever contractor he liked.⁶⁶)

The primary problem with the dispersal of agency power is that delivering a large infrastructure project effectively requires control: control over how the project is designed, of when and how the work gets done, of what should be included in the project at all. The more that control is eroded—the more citizens have the ability to stop or delay projects, the more that outside authorities must grant approval, the more that regulation limits an agency’s freedom of action—the less influence agencies have over the project, and the more difficult it becomes to deliver projects quickly, cheaply, and effectively.

The Great Hollowing Out

As shifts in the political climate have increasingly constrained and checked government agencies’ authority to build infrastructure projects, a parallel trend has hollowed out their capabilities from the inside. Transportation agencies have been stripped of their technical and managerial expertise, and their capabilities in managing the construction of large infrastructure projects have withered as a result.

In the case of the Interstate Highway system, the actual construction was performed by private contractors from the very beginning—but that construction was performed under the watchful eye of an army of government engineers and technical experts. Federal engineers at the Bureau of Public Roads (later renamed the Federal Highway Administration) determined how the highways should be built and created standards for things like pavement designs, concrete mix ratios, lane widths, and signage requirements.

Between 1956 and 1960, $22 million (roughly $250 million in 2026 dollars) was spent on experiments to determine the best design for highway pavement alone.⁶⁷ Design of individual sections of the Interstate was in large part done by state highway departments, many of whom handled “all phases of the work, beginning with planning, location and design, and ending with supervision of construction and maintenance.”⁶⁸ And once construction was underway, it was closely monitored by government engineers, who conducted rigorous “inspections-in-depth” and who were “frequently looked upon to help solve complicated design or construction problems.”⁶⁹

But highway agencies were gradually shorn of their ability to closely manage and monitor highway construction projects. At the federal level, by the 1970s, the FHWA found that it was “not . . . able to maintain the previous level of project-level reviews” and began delegating more responsibility for project review to the states.⁷⁰ Today, due to its “limited resources,” the FHWA utilizes “risk-based stewardship and oversight,” only reviewing those projects deemed to be high-risk ones.⁷¹ As its construction oversight role has shrunk, internal technical capabilities at the FHWA have withered: the agency’s own construction management guide notes that “engineers coming aboard in the last two decades have not had the same field experience and technical exposure that FHWA engineers once had.”⁷²

A similar decline has taken place at state highway departments. The FHWA notes that while state Departments of Transportation (DOTs) historically employed in-house design and construction administration staff, it has over time become “difficult for some agencies to maintain an appropriate staffing level.”⁷³ Michigan’s highway department staffing levels declined by 38 percent between 1974 and 1994, even as highway spending doubled. A 2003 study by the National Highway Cooperative Research Program noted that 80 percent of surveyed state DOTs had flat or declining staffing levels; it also found that some states had seen recent substantial reductions up to 25 percent in staff.⁷⁴ A 2008 Government Accountability Office (GAO) report similarly found that the majority of DOTs had the same or fewer technical staff members than five years prior.⁷⁵ Overall, state DOTs and highway departments lost around forty thousand staff between 1997 and 2020, a decline of around 20 percent, even as highway spending rose significantly.⁷⁶

This reduction in highway department staff has come hand in hand with increased subcontracting. Between 1990 and 2009, Texas went from 15 percent of engineering spending going to consultants to 49 percent.⁷⁷ New Jersey’s DOT went from spending $3.9 million on consultants in 1993 to $105 million in 2002.⁷⁸ By 2002, the states of Illinois, Indiana, Iowa, Louisiana, and Rhode Island were outsourcing virtually all of their preliminary engineering work.⁷⁹ In 2008, the GAO found that more than half of state DOTs were outsourcing 50 percent or more of their engineering work, and a later report found that by 2022 outsourcing had increased even further.⁸⁰

Outsourcing engineering work may reduce what DOTs spend on employee salaries, but it adds to the cost of building highways. Without sufficient in-house technical expertise, plans and concepts provided to outside consultants are often underspecified; this adds risk to the contractors (who must then raise their bids accordingly) and results in scope changes as the project proceeds, creating costly change orders. When economists Zach Liscow, Will Nober, and Caitlin Slattery surveyed state DOTs, they found “broad agreement that state DOTs have become more understaffed and that reliance on consultants drives up costs.”⁸¹ Analyzing state highway construction costs and DOT employment, Liscow, Nober and Slattery found that one standard deviation increase in consulting costs was associated with a 20 percent increase in highway costs per mile, that state DOTs with more employees had lower per-mile costs, and that states with average-quality employees had 30 percent higher costs per mile than states with high-quality employees.⁸²

We see these same phenomena in other areas of infrastructure construction. On California’s High-Speed Rail project, a 2011 report noted that chsra is “significantly understaffed and has struggled to oversee its contractors and subcontractors, who outnumber its employees by about 25 to one.”⁸³ A later 2018 report echoed these concerns, noting that contract managers tasked with ensuring the proper completion of construction work were highly reliant on outside consultants, “hindering [chsra’s] ability to control costs.”⁸⁴ This lack of oversight resulted in chsra paying invoices for work without confirming that it had been completed and approving change orders for millions of dollars more than was recommended; in some cases, change orders were approved for work that was determined to be included in the original contract.⁸⁵ As of this writing, chsra has approved more than $2.3 billion worth of change orders.⁸⁶

Similarly, the Transit Costs Project notes that one of the primary contributors to the high construction costs of U.S. rail projects is a lack of in-house capacity to perform design work, create effective plans, or manage outside consultants.⁸⁷ When the Massachusetts Bay Transportation Authority (MBTA) wanted to extend its Green Line subway (the largest capital project in the MBTA’s history), it had only six full-time employees to manage the project. New York’s Metropolitan Transit Authority (MTA) went from a capital projects management staff of 1,600 in 1990 to just 124 in 2011. The result of this lack of staffing? Poor project management and uncontrolled spending:

[O]ur cases show that consultant teams need a client who knows what it wants and is technically competent enough to direct the consultants rather than allowing them to design overly elaborate stations or propose additional studies that don’t advance the project. In our GLX case we saw that when the project scope ballooned and consultants studied project alternatives that were obviously unviable, such as constructing a tunnel under an existing right-of-way, there was no one at the agency to rein in the consultants and direct them effectively. In New York, we were told that consultants were seen as an endless resource to study every challenge  that emerged multiple times rather than an expensive, specialized unit hired to execute the MTA’s vision.⁸⁸

In 1966, the federal Office of Management and Budget (OMB) issued Circular A-76, which formalized a previously informal policy of the government not competing with private enterprise.⁸⁹ A-76, which has been updated several times since first published, states that the federal government “will not start or carry on any commercial activity to provide a service or product for its own use if such service or product can be procured from private enterprise.”⁹⁰ Thanks to the vigorous enforcement of this policy, over the next several decades the government progressively outsourced more and more functions: executive branch employment was essentially the same in the late 1960s as in the 2020s, even though the U.S. population increased by 70 percent over that period and the federal budget increased by more than 300 percent in real terms.⁹¹ This outsourcing has similarly hampered the government’s ability to deliver projects across a variety of government agencies, and there have been similar calls to bring back in-house technical expertise and reduce reliance on contractors in areas ranging from IT and software procurement⁹² to naval ship design to NASA’s space missions.⁹³

The less transit agencies have in-house technical and managerial capabilities, the less they are able to draw upon their knowledge and experience to plan and manage projects effectively. They become “bad customers” that are unable to plan cost-effective projects, ensure work is being performed correctly, rein in consultant spending, or properly manage contractors. And this has all resulted in expensive, poorly conceived projects being the norm rather than the exception.

Steak for Stakeholders

The dispersion of transit agencies’ control over projects and the gradual reduction of their technical capacities together contribute to a third factor that hampers our ability to build transit infrastructure effectively: bad incentives. Too often, the various stakeholders in a major transit construction project can extract benefits to themselves at the expense of the project as a whole. The less transit organizations have the authority or the willingness to prevent this, and the more the organizational ecosystem encourages it, the worse this problem gets.

Transit projects, such as highways or rail lines, are often large pieces of horizontal infrastructure. By their nature, these projects will intersect numerous parcels of land, governmental jurisdictions, and political boundaries, and thus require the approval of numerous stakeholders who are often incentivized to extract concessions in exchange for that approval. The several examples covered above are just a sampling of the problem that plagues transit construction. The Powerless Brokers report on California transit notes that the weakness of transit agency authority gives other agencies the incentive to repeat the pattern: “Sometimes third parties are fairly transparent about their motivations. Explaining the City of Shafter’s engagement with the California High-Speed Rail project, their city manager stated that the project provided an opportunity to extract ‘this very expensive infrastructure for the city, earlier than when we would’ve been able to accomplish it by ourselves.’”⁹⁴

Similarly, the route for California’s High-Speed Rail project was chosen in part to secure the approval of various political interests.⁹⁵ While it was originally conceived as connecting Los Angeles with San Francisco, during the planning process the route was changed to take a forty-one-mile detour through the city of Palmdale in northern Los Angeles County, which added an estimated $8 billion to the cost of the project. Mike Antonovich, a member of the Los Angeles County Board of Supervisors (whose jurisdiction includes Palmdale), advocated for the route change and was able to secure it in exchange for approving the lease extension of a large development project of a land developer who was also a member of the rail authority board.

More generally, the Transit Costs Project notes that in nearly every low-transit-cost country studied, project design was done by professionals “with little political input.”⁹⁶ The sort of political meddling common in the United States, on the other hand, was “associated with higher costs,” as elected officials are likely to demand expensive additional studies for contentious design decisions, overcommit to projects (making it harder to pull out and easier for third parties to extract concessions), and to view transit projects as job creators, leading to “overdesign and overbuilding.”⁹⁷

Incentive problems are also at work within the agencies and organizations tasked with designing transit infrastructure. If an agency hires outside engineering consultants, for instance, those consultants are seldom incentivized to minimize the overall costs of the project: since their fees will be a function of how much work they do, consultants benefit if a project expands in scope or cost. The Transit Costs Project notes that consultants need to be prevented from “design[ing] overly elaborate stations or propos[ing] additional studies that don’t advance the project.”⁹⁸ A manufacturer of tunnel-boring machines on the Second Avenue Subway project, for instance, noted that they got “stacks and stacks and stacks” of instructions from consultants, many of which made little sense and may have simply been them “trying to do something to justify how much they’re getting paid.”⁹⁹ In their analysis of state highway DOTs, Liscow and coauthors note that outside consultants are likely to have “misaligned incentives,” a concern that is echoed in various auditor reports for California’s high-speed rail.¹⁰⁰

We see similar incentive problems on the construction side. Brian Rosenthal’s 2017 New York Times article “The Most Expensive Mile of Subway Track on Earth” provides numerous examples of bad or misaligned incentives driving up costs in New York transit construction, most notably (once again) in the Second Avenue Subway project. Union agreements resulted in staffing levels four times as high as typically used in other countries, a phenomenon which, according to an internal report by engineering firm Arup, is rife in underground construction in New York:

There are “nippers” to watch material being moved around and “hog house tenders” to supervise the break room. Each crane must have an “oiler,” a relic of a time when they needed frequent lubrication. Standby electricians and plumbers are to be on hand at all times, as is at least one “master mechanic.” Generators and elevators must have their own operators, even though they are automatic. An extra person is required to be present for all concrete pumping, steam fitting, sheet metal work and other tasks.¹⁰¹

On the same project, the Sandhogs union (a union for underground construction workers) was able to secure a $450,000 payout for each tunnel-boring machine used (a technology more than fifty years old) as compensation for jobs lost due to “technological advancement.”¹⁰² This generous union compensation is partly a product of the fact that New York’s MTA does not actually negotiate with unions; instead, negotiation takes place between the union and the construction contractors, two groups who benefit when construction costs are as high as possible.

Another incentive problem driving up costs on the construction side comes from the shifting of risk: transit agencies often try to shift project responsibility to contractors tasked with implementing it, which results in them raising their bids in response. The Transit Costs Project notes that this risk-shifting and price-raising is endemic to transit construction in the United States and much of the English-speaking world; Liscow and coauthors note the same phenomena occur in highway construction.

Transit projects must also often be designed “defensively” to include various features, construction methods, or studies primarily to reduce the risk of litigation by unhappy residents or interest groups. Environmental reviews, which today take years and millions of dollars to complete for a major transit project, have been forced to become incredibly thorough because agencies try to make them “litigation-proof,” reducing the risk of someone filing a lawsuit by claiming some particular option wasn’t sufficiently considered.¹⁰³ Similarly, U.S. rail projects have almost completely eschewed “cut-and-cover” station construction, which requires tearing up the street, building the station, then covering it up because of its disruptive effect and the risk of inviting citizen lawsuits. Instead, they opt for more expensive but less disruptive construction methods that keep excavation entirely underground.¹⁰⁴

Bad incentives can also be the product of how projects are funded. Municipal buses in the United States, for instance, are often expensive because of excessive customization, rather than using standardized, off-the-shelf bus designs. This excessive customization is partly a product of how they are funded: the federal government will cover 80 percent or more of the cost of buses, giving agencies little incentive to minimize cost.¹⁰⁵ Likewise, the Transit Costs Project notes that projects are often chosen based on how to maximize federal contributions to a project, rather than the most sensible option.¹⁰⁶ Report coauthor Alon Levy notes that “other people’s money,” i.e., funding sources that don’t have influence over design decisions, reliably leads to high costs.¹⁰⁷

Residents, interest groups, consultants, and every other stakeholder will always be incentivized to follow their own best interests. The process for building a major infrastructure project needs to be designed to take this into account and shape it so these various interests align as much as possible with the overall goal of building a cost-effective project. But far too often in the United States, they don’t.

Is There a Way to Fix It?

At a high level, fixing the problem of high U.S. transit costs means addressing all the things that cause them: restoring agency authority, rebuilding technical and managerial capability, and changing the incentives around how transit projects are constructed. We can broadly place solutions for these problems into three categories.

First, we can change the rules and regulations that shape transit agency actions and project construction: Davis-Bacon rules that dictate project wages, Buy American rules that require using U.S.-made components, and overly strict building codes. It also includes changes to the various rules that have had the effect of dispersing transit agency power: NEPA, historic preservation laws, and other regulations that give citizens, interest groups, and other government agencies de facto control over transit projects.

Changing rules and regulations is theoretically simple, but it is rife with political difficulties in practice. Many policy regimes have constituencies that will resist modifying or unwinding them: unions for Davis-Bacon, environmental groups for NEPA, the National Fire Protection Association for code requirements. Changes are likely to be drawn out and contested. For instance, there have been numerous congressional efforts in recent years around “permitting reform,” i.e., changing the federal permitting process, including compliance with NEPA and similar laws, to make it easier to build infrastructure.

Both Democrats and Republicans are interested in easing the federal permitting process—the former to make it easier to build clean energy projects such as wind and solar, the latter to make it easier to build oil and gas projects—but neither side has proved particularly willing to compromise and come to an agreement. By 2024, nearly a dozen permitting reform bills had been submitted to Congress but only minor changes (the Fiscal Responsibility Act of 2023, which slightly reduced NEPA requirements, and the advance Act of 2024, which streamlined nuclear power permitting) have been enacted.¹⁰⁸ The most comprehensive permitting reform bill, the Energy Permitting Reform Act of 2024 (championed by Senators Joe Manchin and John Barrasso), cleared the Senate’s Committee on Energy and Natural Resources but died in the House, in part due to widespread opposition from environmental groups.¹⁰⁹

But it’s nevertheless clearly possible to change the rules around infrastructure construction to make the process easier and cheaper. The 2005 Energy Policy Act, for instance, changed rules around environmental compliance for oil and gas drilling: specifically, it allowed many drilling projects to be excluded from NEPA analyses via a “categorical exclusion,” which reduced the time and effort it took to permit such projects.¹¹⁰ With sufficient political will, reform here could be achieved.

Second, we can rebuild institutional capacity. This goal would encompass the restoration of in-house technical expertise in transit agencies so that they can properly manage, evaluate, and perhaps design projects themselves, improving procurement methods and agency decision-making, and reducing internal red tape and bureaucracy. It also means changing policies and habits that drive up costs, such as the tendency of agencies to offload project risk to outside contractors.

Like changing rules and regulations, this restoration is also difficult, though it is difficult in a different way. Bringing back institutional capacity does not necessarily require any major legal reform: the Transit Costs Project notes that none of its recommendations for improving U.S. passenger rail construction require large-scale legal changes. But changing the DNA of an institution and cultivating the knowledge and experience to manage projects effectively can take years or decades; meeting these objectives requires the sort of dedication that many believe U.S. transit agencies are unwilling to marshal.

Whether U.S. transit agencies can rise to the challenge of remolding themselves is an open question, but there’s reason to think it’s possible. For one, U.S. transit agencies often move swiftly and efficiently when the circumstances require it. Following the collapse of a highway overpass, for instance, these agencies are often able to rebuild them extremely quickly. When an overpass on I-85 in Atlanta collapsed following a fire in 2017, it was rebuilt in just forty-four days, thirty days ahead of schedule and under budget.¹¹¹ Similar rapid overpass reconstruction efforts took place on I-95 near Philadelphia in 2023 (restored in just twelve days) and on I-10 in Los Angeles following the Northridge Earthquake in 1994 (restored in three months, seventy-four days ahead of schedule).¹¹² The I-35W Saint Anthony Falls Bridge in Minneapolis, which collapsed in 2007, was rebuilt in just fourteen months, three months ahead of schedule and two years faster than the original 1967 bridge was completed.¹¹³ When circumstances allow it—when a project has nearly unanimous support, when permitting and procurement difficulties are swept away, when contractors are properly incentivized—U.S. transit agencies often prove themselves capable of going above and beyond expectations.

Finally, the third strategy to reduce the costs of building transportation infrastructure is to simplify what we build and how we build it. Agencies should eschew unnecessarily complex or custom designs in favor of standardization, use less expensive construction methods where appropriate (such as cut and cover transit stations), resist pressure to include expensive modifications to a project to please various constituencies, and base their designs on what’s proven to yield low-cost construction elsewhere.

Simplifying what we build and how we build it is in some ways the easiest—in a sense, it just requires designing better projects—but much of it can’t be done without deeper legal and institutional reforms. The mess of American transit construction was a product of multiple influences, working over a period of years; cleaning it up will likewise be a yearslong project.

This article originally appeared in American Affairs Volume X, Number 3 (Fall 2026): 25–47.

Notes