Diy 13 / shutterstock / Paul Spella / The Atlantic

In the spring of 1944, as the Allies secretly prepared to invade France, a midwestern factory with the pleasant name of Willow Run produced one B-24 heavy bomber every hour. An astonishing feat, even now. Every day, 125 miles of copper wire and 8.6 million rivets went in one end of the factory’s mile-long assembly line; every day, 24 completed airplanes, each two stories tall and 110 feet wide, came out the other. Then these strange newborns taxied to the airfield next door, met a waiting crew, and took off for Europe. This is how 42,000 Americans—working full-time at a place that had been, a few years earlier, a small farm outside Detroit—furnished a violent war nearly 4,000 miles away. If the great in “Make America great again” refers to anything just and mighty, it means Willow Run.

Little wonder, then, that Donald Trump chose the old factory for the site of his first visit to Michigan as president. It was March 15, 2017, and Trump was in a jubilant mood. “I love the people of this state,” he told a crowd of autoworkers and their C-suite bosses. “You did me a big favor.” Months earlier, Michigan had, by the narrowest margin in its history, delivered Trump its 16 Electoral College votes. The state had not gone red in a presidential election in 28 years.

Trump had a good idea how he had flipped it. “During the campaign, I came to Michigan again and again, and I made this promise, that I am going to fight for your jobs,” he said. Now he would make good on that vow. He claimed to have persuaded the “Big Three” American automakers—Ford, General Motors, and Fiat Chrysler—to preserve or create a total of 3,600 auto jobs in the United States. But the country needed to think bigger. “We must embrace a new economic model,” he said. “Let’s call it the ‘American model.’”

With this declaration, Trump began one of the greatest fiascos of his presidency. His initiative would eventually entangle three federal agencies, four automakers, and 13 states. It would devour decades of billable time from legions of corporate lobbyists and lawyers. And it would stymie every person who hoped to buy a new car in the United States from 2021 to 2026—or, for that matter, every person who planned to use an American highway, or breathe American air.

That day at Willow Run, Trump was also about to make a promise that he couldn’t keep.

Trump’s “American model,” as he described it, established a new bargain between American companies and American workers. As part of this deal, the federal government would cut regulations and reduce burdens on U.S. companies. In exchange, those companies had to hire more American workers and buy American products. “That is how we will succeed and grow together—American workers and American industry side by side,” Trump said.

This vision, depending on the angle, looked like either a proudly nationalist way of doing business or an oddly lopsided agreement. If the companies got their rollbacks but then didn’t hire American workers, what could the government do about it? By cutting regulation first, it had already sacrificed its clout.

But this point went undiscussed, because Trump knew which rule he wanted to clip first: the Corporate Average Fuel Economy standards, or CAFE standards, which governed the gas mileage and tailpipe pollution of new cars, light trucks, and SUVs. “I’m sure you’ve all heard the big news,” he said. “We’re going to work on the CAFE standards so you can make cars in America again.”

The CAFE standards did not actually concern whether cars could be made in America. But since 2012, the rules had required that new vehicles must get about 5 percent more fuel-efficient every year, while simultaneously restricting the amount of heat-trapping carbon pollution they could emit. The rules had, in a word, worked: They had pushed the country’s fleet-wide fuel economy to an all-time high, prevented millions of tons of carbon from blanketing the atmosphere, and saved consumers hundreds of billions of dollars at the pump. They were also one of the most consequential climate-change protections in American history.

But aspects of the rules were inflexible, and automakers had grown to resent them. The chief executives of Ford, General Motors, Fiat Chrysler, and other automakers had implored Trump to roll back the standards. Now he was doing it—all, he made sure to note, in the name of saving auto jobs. “If the standards threatened auto jobs, then commonsense changes could have and should have been made,” he told the autoworkers at Willow Run.

Those “commonsense” changes are now complete. Late last month, amid the coronavirus pandemic, the Trump administration published its new version of the tailpipe-pollution rules. President Trump, in his only tweet on the occasion, noted that he was “helping U.S. auto workers by replacing the failed Obama Emissions Rule.”

The president may believe this. But he is not saving jobs with the rollback—he is destroying them. According to his own administration, the rollback will eliminate nearly 13,500 jobs a year. This is clearly stated and repeated throughout the nearly 2,000-page rollback, which is free to read online.

It first says that it will kill jobs on page 1,409, in a passage that announces the changes will “slightly reduce estimated U.S. auto sector labor hours.” Several hundred pages later, it clarifies the meaning of slightly: A table on page 1,721 shows that Trump’s plan will lead to 13,474 fewer “person-years” compared with the existing Obama-era standards. Person-year is a fancy word for job.

When the president says his rollback will save jobs, or help U.S. autoworkers, he is either lying or being lied to. His prized rollback will eliminate nearly four times as many auto-industry jobs as he once bragged about saving. When asked about this, the White House referred me to an earlier statement that said the rollback “will continue to improve United States air quality while also protecting the jobs of our great American autoworkers.”


Three years ago, the rollback merited a major policy address and a trip to Detroit. Now it is an afterthought. It was finalized on March 31 amid a public-health crisis and a nauseating recession, with only a presidential tweet and a five-sentence press release to show for itself. Finalized is a legal term that means a regulation has been approved by administration officials and will soon carry the force of law. But this particular rule was also finalized in the sense that the Hastings Cutoff finalized the unhappy Donners. By publishing this rule without fanfare, and without fixing its key flaws, the White House is as good as killing it.

What happened? Here, it is worth consulting the rollback itself, since the rule, in its voluminous magnanimity, will do much more than eliminate jobs. Trump’s rollback will make owning a car more expensive for the average American. It will intensify climate change and increase the number of Americans killed every year by toxic air pollution. And it will—in a surprising concession to the Green New Deal—outright discourage Americans from driving.

The previous sentences, like the job-killing stats, are facts printed within the rule itself. (They can be found, respectively, on pages 1,414; 723; 1,524; and 1,589.) Altogether, according to the Trump administration, the rollback will inflict as much as $22 billion of net damage on the American economy.

This is not how things are supposed to work. Since the Reagan administration, new federal regulations have had to pass some kind of cost-benefit analysis. Their value to the American economy, measured in a holistic but monetary sense, must exceed the cost of implementing them. It was once a major goal of Republican governance that every federal regulation should pass a cost-benefit sniff test. But this rollback has cut the cheese.

“Accepting the administration’s own numbers—and some of them are highly suspect, and most are just wrong, but accepting the numbers upfront—the rule is net costly,” Richard Revesz, the Lawrence King Professor of Law at NYU, told me. He is director of the university’s Institute for Policy Integrity. “This rule is actually causing deaths, even under their analysis.”

The Trump administration says the rollback’s effects are not all bad. “Striking the right balance on fuel economy standards ensure Americans have the freedom to select from a wide range of safe, clean, and fuel-efficient vehicles to find the one that best meets their needs,” Sean Rushton, a spokesperson for the Department of Transportation, said in an emailed statement. (The U.S. Environmental Protection Agency did not respond to a request for comment.)

Under federal law, the government can run its cost-benefit analysis under a few different discount rates, so named because they discount costs levied on people in the future. The administration prefers a 7 percent rate, because it claims that this figure more adequately reflects real-world market behavior. “At a 7 percent discount rate, the rule projects more than $16 billion in net benefits,” Rushton said. This is true, as it goes: That number is in the rollback. It is one of four cost-benefit analyses in the rollback. If you average them together, they come to negative $3 billion.

Most Americans understand that rightsizing environmental rules requires trade-offs. The government can make it cheaper for Americans to drive, or it can reduce pollution from cars, but it can’t usually do both. Navigating such trade-offs is messy and disagreeable; it is what makes environmental policy hard. But with this rule, the Trump administration has somehow chosen neither option: It both discourages driving and worsens air pollution; increases costs for consumers and eliminates auto jobs. It accomplishes no obvious public-policy goal. Asked to choose between Door No. 1 and Door No. 2, it walks into a wall.  


For tedious reasons, three government entities share custody of the tailpipe-pollution rules: the U.S. Environmental Protection Administration, the U.S. Department of Transportation, and the state of California.

This is a dicey situation even at the best of times. The regulators at the EPA and in California are generally thought—by outside lawyers, academics, and former officials from the Obama and Trump administrations—to be more expert. They have more legal freedom, and they are focused on limiting carbon dioxide and other climate-warming pollutants. The staff at the DOT, meanwhile, is focused on fuel economy, which it has been regulating since the 1970s.

But as I wrote earlier this year, in a feature for The Atlantic, it was the small team at the DOT that took control of the rollback. Within months of Trump’s Willow Run address, its career staff and political appointees boxed out the EPA team. They worked on the rule by themselves for months, and devised computer models that seemed to show that loosening the standards would save hundreds of lives and billions of dollars.

When the DOT presented its models, Trump officials gleefully seized on them—they revealed, after all, the righteousness of the administration’s cause. As I reported, when EPA career staff tried to point out the flaws in the DOT’s wishful math, their concerns were ignored.

Separately, the Trump administration blocked California from having any influence on the DOT’s process. Under the 1970 Clean Air Act, California can set stricter air-pollution rules for cars and trucks. Any other state can then opt in to its stringent rules. (Thirteen states and the District of Columbia have done so to date.) But the White House revoked California’s authority to set these stricter standards for climate-change pollution, freezing it out of the revision process.

So in the summer of 2018—with the DOT’s models in hand, and California sidelined—the White House lunged for the most extreme option available: It proposed freezing the standards entirely. Starting in 2020, it said, carmakers would not have to make their vehicles more efficient than they were the previous year.

This proposal had two problems. The first issue—and comparatively the minor—was that the automakers hated it. Yes, they had asked Trump for a rollback four days after he was elected; yes, their CEOs had joined the president onstage at Willow Run; and, yes, their lobbyists had spent the previous 18 months bellyaching about the Obama-era rules to anyone in earshot—but Trump’s proposal was untenably tacky. The automakers had expected something that obscured the extent of the rollback through the doling out of credits and accounting guidelines and adult discretion. They blanched that Trump would freeze the rules. It made them, as automakers, look like villains. And more important, it wasted their years of investment in fuel-saving technology. (The oil industry, however, was delighted. It had spent months covertly lobbying for the freeze, a New York Times investigation later revealed.)

The second, and more serious, problem was that the draft proposal was riddled with errors. They came in every size. At one point, the authors of the rollback forgot to divide a crucial number by four. At another, they made strange assumptions, including that Americans would drive 20 percent more if they had more efficient cars.

But their crowning achievement was a catastrophic confusion of supply and demand. The new DOT model assumed that as cars got more expensive, more people would own and drive them. This backward math led the authors to argue that the Obama-era rules—which increased the price of cars—expanded the size of the U.S. car fleet and, with it, increased the number of car accidents and fatalities. This confusion allowed the Trump administration to declare that its rollback was safer than the Obama policy. Officials even named the rollback SAFE, short for the Safer Affordable Fuel Efficient Vehicles rule.

The DOT programmers, for their part, had misunderstood how the auto market works: As cars become more expensive, in fact, fewer people drive them. By the end of 2018, academic economists and the Honda Motor Company had diagnosed the issue, identifying in SAFE hundreds of billions of dollars in claimed safety benefits that did not exist.

The administration spent the following year and a half trying to figure out where to go next. SAFE, it was clear, could not pass its cost-benefit analysis without the safety benefits. As recently as February, The New York Times reported that the rule might be significantly delayed, pushed to this summer or put off entirely. But White House officials evidently decided to put the rule out of its misery and publish it.


In the end, the Trump administration solved the issue of SAFE not passing its cost-benefit analysis by having SAFE not pass its cost-benefit analysis. As mentioned above, the final rule includes four estimates of SAFE’s net benefits for the American economy—and they average to about negative $3 billion. According to the Trump administration, the rule is a drain on the economy.

But this may again overstate its value, because the final version of the rule may actually include a new error. Since the rollback is coming out now, it applies to cars made in model year 2021 through model year 2026. Its cost-benefit analysis, then, should primarily calculate how the policy will affect cars made in those model years. But the final Trump rule appears to include benefits from cars made in model years 2018, 2019, and 2020. The White House seems to retroactively apply its rollback to cars made in the past, then accredit the benefits to the future.

“It’s irrational,” Sylwia Bialek, an economist at the Institute for Policy Integrity at the NYU School of Law, told me. “The past has already happened. The agencies are behaving as if they were able to create a counterfactual past.” A rudimentary analysis suggests that this error creates $6 billion in benefits that don’t actually exist, but Bialek said the costs must be fully reanalyzed before anyone will know for sure.

The DOT disputes that the error exists, saying a similar method was used in the Obama-era standard. “In every [CAFE] rulemaking, including in 2012, we started with a baseline fleet in some model year prior to when the standards start (because we want it to be real vehicles that exist), and then we ‘bring that fleet up’ to the first model year being regulated by applying technology to each vehicle in the most cost effective way practicable,” an agency spokesperson said in an emailed statement. “We account for the costs and benefits of doing that, just like we did under the 2012 rule.”

The Trump administration has made a few other changes from the draft version. The biggest is that, instead of freezing the rules at their 2020 level, the rollback now institutes a slight annual increase of fuel economy. Automakers must now improve by about 1.5 percent a year. That is better than a freeze, but it may not mean much in the real world, because the industry already exceeds or matches that increase without any regulation. “A 1.5 percent annual improvement is really nothing. The industry, historically, has improved by anywhere from 1 to 2 percent,” Margo Oge told me. She was the chief vehicle regulator at the EPA from 1994 to 2012, and now sits on Volkswagen’s sustainability board.

Oge said that automakers may also not need to actually hit the 1.5 percent target, since the rules come with plenty of new credits and waivers.

“Our standards only set minimum average levels of performance, so those who prefer greater fuel efficiency will continue to be able to select from a wide range of highly fuel efficient models, and manufacturers will continue to meet that demand,” Rushton, the DOT spokesperson, said.

The automakers, for their part, have responded to the new rules with stupendous paralysis. The Alliance for Automotive Innovation, the industry’s trade group, has taken no position on the rollback. It published a statement last week that is notable chiefly for its literary accomplishment—somehow, each individual sentence has the flop sweats: “The auto industry has consistently called for year-over-year fuel economy and greenhouse gas improvements that also recognize that the standards originally developed almost a decade ago are no longer appropriate in light of shifting market conditions and consumer preferences.”

A few weeks after the Willow Run speech, Mitch Bainwol, then the president of the alliance, stood next to then–EPA Administrator Scott Pruitt and stressed that the imminent loosening of the tailpipe rules was not a rollback. “‘Adjustments’ are not rollbacks,” Bainwol wrote in an op-ed soon after. This intriguing claim was hampered by Pruitt’s insistence on referring to the rollback, repeatedly, both in person and in a now-deleted tweet, as a rollback. Automakers had failed to realize early on that the Trump administration was not playing the same game as them.

The alliance did not respond to a request for comment.

Meanwhile, the White House has only made things worse for automakers, especially American automakers. Take General Motors. Three years ago, Americans got a $7,500 tax rebate when they bought a GM-made electric car. Today, there is no tax rebate. Three years ago, GM’s assembly line stretched across North America. Today, the company has survived a fight over NAFTA but is girding for new fights over trade agreements with the United Kingdom and China. Three years ago, GM’s chief executive, Mary Barra, could brush off the occasional half-hearted haranguings of Barack Obama. Today, President Trump blames GM for killing Americans by not voluntarily building ventilators to help coronavirus patients.

Workers have seen few benefits from these Trump-era changes. GM employed 5,000 fewer workers in the United States last year than it did in 2016. It now employs more people in China than it does unionized employees in the United States

On the day Trump was inaugurated, GM was worth about $55 billion. Two months ago, on the eve of the coronavirus crash, its valuation was already off by $5 billion. Today, it is worth $34 billion. And this dismal performance isn’t unique to GM. The S&P 500 autos index, which tracks the industry as a whole, lost 22 percent of its value from Trump’s inauguration to February 2020. It has lost roughly another third of its value since then.


If the final rule is so bad, then why did the Trump administration publish it? The answer no longer has much to do with the fuel-economy standards, as my reporting has shown. Rather, the White House hopes to permanently strip California of its ability to regulate greenhouse gases from vehicles. Revoking the state’s authority is a longtime goal of the oil industry, and it remains a goal of some automakers, even as they hem and haw about the tailpipe-pollution rollback itself. Stripping the state of its waiver is, from all signs, a major second-term goal for Trump. But legally, the White House cannot revoke California’s waiver without also rolling back the federal standards.

It has proved menacing in its attempt to revoke the rule. Last summer, four companies—Ford, Honda, Volkswagen, and BMW—reached a compromise with California on the fuel-economy rules. The compromise looked much like what it originally seemed Trump would give them: a less strict version of the Obama-era rules, with new credits and flexibility to lighten the burden. In reply, the Department of Justice slapped the four companies with an antitrust investigation. While the DOJ dropped the probe in February, it effectively silenced the companies’ criticism: No corporation, after all, is eager to discuss a matter under federal investigation.

Now this rollback will go to the federal courts. The standard for whether a judge can block or amend a federal rule is laid out in the Administrative Procedure Act. It asks whether a certain rule is “arbitrary and capricious.” Often, fights about this standard revolve around difficult questions of whether an agency’s authority has been stretched too far, or whether the agency calculated every cost and benefit that it needed to. This is likely not the ground on which this rollback will be fought, because, according to the rule itself, it is a drag on society.

“The Supreme Court has not yet held that cost-benefit analysis is required or that you need a positive cost-benefit result for the rules to be valid” in most instances, but a 2015 ruling by Justice Antonin Scalia “certainly suggests you have to at least think about it and talk about it,” Jonathan Adler, a law professor at Case Western Reserve University, told me. Adler is also a frequent conservative legal commentator.

In the past, courts have understood when an agency says it wants its cost-benefit analysis to be viewed compassionately, or in a favorable light, he said. But judges more likely to do so have generally been on the political left, Adler noted. “There is an irony in the fact that it’s generally been folks on the right who have argued for cost-benefit as a constraint on agency policy-making and agency action, yet this administration has pursued rule changes that it’s not clear can satisfy that criterion, or that can’t satisfy it under reasonable assumptions,” he said.

Revesz, the NYU law professor, said the SAFE rule failed that test. “This is a rule that, under any fair valuation, does more harm than good,” he said. “I would say that under any appropriate application of the standards of the Administrative Procedure Act, this rule should be set aside.”

And there it will join its peers. Presidential administrations have historically won about 70 percent of cases under the Administrative Procedure Act. According to a tracker maintained by the Institute for Policy Integrity, the Trump administration has won only 10 percent of its cases.

So it may come to seem fitting, poignant even, that the rule was first announced at Willow Run, for it will have another similarity with the products made there. As World War II drew to a close, the government was still owed planes under its contract with Willow Run, but it did not require them at the front. The factory still made the bombers. They took off from the airfield next door and flew directly to a scrappage yard.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.