As the Biden administration continues to promote its infrastructure plan, one problem stands to raise procurement costs and lead to disappointing outcomes: the Buy America provisions. The original Depression-era law (Buy American) requires federally-funded contracts to only use domestic materials; and a Reagan-era extension (Buy America), which targets public transportation and rolling stock specifically, requires domestic manufactures, banning the import of buses and trains that receive any federal capital grant.
The law is harmful. The jobs Buy America provisions create come at a cost of between $250,000 to over $1 million per job “saved,” and only last about five years. The equipment it provides is often shoddy, as well.
The implementation of Buy America rules since 2016 has been especially bad. The provisions of the law have an escape clause: If using domestic steel pushes the cost of a bid to 25 percent more than using foreign steel, imports are permitted. But waivers require the approval of the Federal Transit Administration, which stopped processing them in the last administration to align its practice with Trump’s stated policy of protectionism. The Biden administration is regrettably keeping the Trump-era practice. It would improve the state of American infrastructure in the future to fully repeal the law and legalize unrestricted imports, but via executive action it is possible, as an intermediate step, to grant waivers whenever imports are significantly cheaper or higher quality.
The historic costs of Buy America
Until the 1980s, Buy America laws kept foreign competition away from American transit systems. As long as the United States was at the technological frontier of bus and train manufacturing, this was not a major problem. To this day, the European and Japanese rail markets are separate since both sides are protectionist, and the negative impacts are visible but minor. However, in the postwar era, American mass transit went into a gradual process of decline. Railroads and transit agencies deferred maintenance and tried to stretch the lifespans of old trains, and the market for new passenger rail vehicles shrank. The only big buyer was the New York City Subway.
In the 1970s, New York was at the depth of its era of deferred maintenance, but did purchase new vehicles — in fact, it aimed to buy more advanced trains, with air conditioning and higher top speeds, hoping to attract people back from the car to the subway. The two models it ordered, the R44 made by the St. Louis Car Company and the R46 made by the Pullman Company, were both defective. The R44 had three times the failure rate of other train classes, and in 1974 the state comptroller recommended a lawsuit against St. Louis Car; the manufacturer went out of business the same year. The R46 had a host of problems including defective brakes, and in 1979 the subway had to bring the older models the R46 was meant to replace out of retirement while the R46 was being reserviced; in the same year, the subway sued Pullman for delivering a defective product, and won, contributing to the company’s exit from the business.
The United States entered the Reagan era with no domestic rail manufacturing industry to speak of. New York’s next order, the Japanese-made R62, was not federally funded due to Reagan budget cuts. It was technologically conservative to avoid the R44 and R46’s mishaps but was nonetheless considered a success. Subsequent federal funding would require vendors to open transplanted factories in the United States for further orders, but the U.S. remained a peripheral market to a global industry whose cores were Western Europe and Japan, and more recently South Korea and China.
While the United States remains a peripheral player in transit, innovations developed elsewhere take a long time to reach here. Buying the same modular products that work elsewhere is difficult because the vendors have to effectively reproduce the manufacturing capacity in the United States, whose market is small relative to the effort this requires. In particular, American subway car technology has largely stood still since the 1980s and early 1990s. For example, recent European subway trains have what are known in industry parlance as open gangways, which permit passengers to walk through from one end of the train to another. American trains do not have open gangways and are only beginning to experiment with the technology. In fact, in San Francisco, Bay Area Rapid Transit’s Fleet of the Future is similar to a modular product running in London, Singapore, Stockholm, and Toronto, but without open gangways.
Buy America, today
In Europe, single-deck electric rolling stock costs about $100,000 per meter of train length. This rule of thumb applies to light rail vehicles, metro trains, regional trains, and even intercity trains, except very high-speed trains, which cost nearly double. Bilevel trains cost around 50 percent more.
In the late 20th century and the very beginning of the 21st century, American single-deck electric trains cost about the same. Examples include the early-21st century New York orders, the Fleet of the Future, and the electric trains bought by the Long Island Railroad and Metro-North, both in the New York suburbs. The same few factories would churn out these trains as they always had. The transit construction cost database maintained by my research group at NYU’s Marron Institute excludes rolling stock because at the time we started looking at comparative costs, circa 2010, there was no American rolling stock cost premium.
But technology marches on. Sticking to 1980s technology in the 2010s and ‘20s has real costs, and eventually this turned into a procurement cost premium. Moreover, the major vendors are consolidating in response to both new market entrants in Europe and global competition with Chinese vendors, which means the base of established companies with American factories is becoming even less competitive. When a new firm has to establish its first factory just for a single order, costs are by nature higher.
Indeed, recent American rolling stock costs have been escalating. New York’s next train model, the R211, will finally replace the R44s at a cost of $3.7 billion, or about $130,000 per linear meter. Bilevel cars for Caltrain and New Jersey Transit cost about 35 percent and 60 percent more than their European counterparts respectively, the New Jersey Transit order also burdened by compliance with obsolete regulations. A Los Angeles light rail vehicle order from the last decade went up to $3.8 million per car or $140,000 per meter of train length (the average cost in Europe is under $100,000 per meter), while a San Francisco one went up to $170,000 / meter. In Boston, an order with little premium, built at a new factory in Springfield, Massachusetts, was given to Chinese manufacturer CRRC. But European and Japanese vendors had bid $135,000/meter, leading to worries about Chinese dumping and a congressional investigation.
If Buy America wasn’t bad enough, U.S. states also apply informal pressure on vendors to locate plants within their territory. Thus, instead of one national market protected behind trade barriers, the United States has several markets, one for each state with a large city that buys passenger trains. CRRC in fact established two separate American plants, one in Massachusetts and one in Illinois, to curry favor. The Los Angeles order was given on condition that the trains be built in Los Angeles’ own exurbs in Antelope Valley.
Other than the New Jersey Transit commuter trains, the trains detailed above aren’t technologically different from their European or East Asian counterparts, just more conservative in their lack of open gangways. And yet, they cost 30-70 percent more. In Europe, too, such cost premiums occasionally crop up when countries engage in informal protectionism. France is notorious for this, insisting on bespoke designs and informal guarantees of a made-in-France imprimatur for Paris commuter rail stock. It is incumbent on the United States to imitate Europe’s successes and not its failures.
Labor and industrial protection
Protectionism is often justified on grounds of labor and environmental standards. The American Prospect recently ran an article that approvingly mentioned the premium-cost Los Angeles light rail order as an example of good union jobs. The union threatened vendor Kinki Sharyo that it would launch environmental nuisance lawsuits unless the company agreed that the plant be unionized. In reality, the benefits of this style of protectionism go largely to managers and sitework consultants; blue-collar labor only gets scraps.
According to the Prospect, the light rail order, valued at $890 million, created 250 jobs paying $20 an hour, up from $11 for non-union manufacturing labor in the area. As a cost of about $1 million per job, the jobs themselves only pay $40,000 a year and only last a few years.
The Springfield plant in Boston has similarly lopsided capital-to-labor cost ratio. The order was $1.3 billion, including $843 million for the trains, with the rest going to related purchases not made at the plant, such as new signals. The plant has a total of 200 jobs. This was, to be clear, an order with little cost premium, but there is suspicion of dumping. The higher bids of first-world competitors would have amounted to a premium of about $1.2 million per job created.
Nor does rolling stock protectionism help jumpstart domestic industry. This is not infant industry protection: In 35 years of transplant factories, the United States has not been able to develop any domestic expertise in manufacturing trains, as the rest of the world keeps buying from European and East Asian vendors at either local transplants or those vendors’ home plants. Left-wing parties in Europe have a term for imports in this circumstance: parallel imports. According to this view, protectionism is justified on labor or ecological grounds, but imports from countries with strong unions and environmental regulations trigger neither justification. On the contrary, imports in such cases should be encouraged, because they can discipline domestic industry into charging lower prices, at the expense not of wages (since parallel imports are from similar-wage countries) but of monopoly profits.
Repeal Buy America
Buy America was intended to support American industry, but has had the opposite effect. Its damage may not be so evident wherever the United States already enjoys a global technological lead, but in the train market, where the United States is at the developed world’s rear, the problems are out in the open.
Worse, none of the benefits it was supposed to bring are real. It does not help create jobs: At a cost of $1 million per job that pays about $40,000 and lasts five years, it is at the bottom of the list of ways to improve working-class living standards. On the contrary, inefficient infrastructure procurement harms the very working-class people that rely on public transit for transportation.
Foreign manufacturers with factories transplanted in the U.S. have not led to any exportable product, as American technology in the area falls further and further behind the European and Japanese frontier. Nor has it helped create a unified national market. States use Buy America rules as a template to pressure firms into opening in-state plants, segmenting the nation into multiple state-based markets.
This is a harmful law that can and should be repealed. Too few jobs are created for it to be a true priority for the labor movement; $1 million for a single temporary job could instead buy a lot of college degrees, health plans, and permanent public-sector hires. Nor are factory sitework consultants an interest group worth appeasing the way the Chamber of Commerce writ large may be. All a repeal costs, politically, is an admission that a derivative principle was wrong; an admission that is inherent in any open legislative or executive change. Errors happen, and sometimes persist for generations. In hindsight, we can see that Buy America was among those errors, and this can and should be fixed.
Alon Levy is a Fellow with the NYU Marron Institute and founder of the urban transit blog pedestrianobservations.com