Donald Trump had many priorities upon returning to the Presidency, and one of the most pressing was to get Winston Churchill back into the Oval Office. A bust of Britain’s wartime Prime Minister, of whom Trump is a “big fan,” had kept him company throughout his first term, and for the past six months it has sat next to the office’s gilded fireplace, lurking in the background of Trump’s meetings with other world leaders. Trump admires Churchill’s glower, which he channelled for both his mugshot and his official Inauguration portrait, and he sees himself, much like Churchill, as the singular savior of a beleaguered nation. Perhaps it was a Churchill quotation—an apocryphal one, as it happens—that Trump had in mind last week when he fired Erika McEntarfer, the commissioner of the Bureau of Labor Statistics: “I only believe in statistics that I have doctored myself.”

McEntarfer was nominated as head of the B.L.S. in 2023, by Joe Biden, and confirmed by the Senate, in an 86–8 vote, with most Republicans, including now Vice-President J. D. Vance, joining Democrats in the majority. She has a background as a high-flying but uncontroversial government economist, with stints at the Census Bureau, the Treasury, and the Council of Economic Advisers. Much of her work has been focussed on the provision and analysis of labor-market data: the exact subject that got her fired.

On August 1st, the B.L.S. released its monthly jobs report, covering July. To insulate them from political interference, such reports are released on a strict schedule and are not available even to the B.L.S. commissioner (or to the President) until shortly before they are made public. As is commonplace, the July report also included revised figures for May and June: the B.L.S. relies, in part, on self-reporting from a sample of public and private employers, and there is typically a lag. This time, the revision—which knocked the estimated number of new jobs created over those months from two hundred and ninety-one thousand down to thirty-three thousand—was extraordinarily large, puncturing the bullish picture of the economy that had been building over the past few weeks, and bringing to a sour end a week of trade-deal-related good press for the Administration. Trump, offended by this, announced on social media that the numbers were “RIGGED” and that McEntarfer was out.

In fairness to Trump, it’s hard to imagine many Presidents being unruffled by such bad news. The jobs report is one of the country’s most important indicators of economic health, watched closely by investors and policymakers alike. It provides a deceivingly simple headline verdict on the government’s economic performance—an especially controversial topic at present, but always a major focus. In 1971, when favorable data from the B.L.S. weren’t played up in a manner that Richard Nixon wanted, he instituted what would become known as a “Jew count,” to extirpate supposedly subversive elements within the bureau. Four people, singled out for having “Jewish-sounding” surnames, were demoted or reassigned. Nixon’s interference led to new rules to protect the integrity of government statistics, among them a regulation that governs when and how reports are released. Presidents, of course, kept grumbling about unfavorable numbers, but their influence over them waned.

The Bureau of Labor Statistics has had to tighten its belt in recent years. Since 2010, its budget has fallen by twenty per cent in real terms, according to Bloomberg. Although further budget cuts under Trump have affected the agency, it fared relatively well under DOGE, perhaps because it’s considered so important, or perhaps because it sounds so boring. Still, Trump has been trying to undermine the bureau for years; he’s been criticizing it since long before he formally entered politics. In October, 2012, the jobs report showed that employment had fallen below eight per cent for the first time since Barack Obama became President. Trump—mostly known back then for his stint on “The Apprentice” and his promotion of the birther conspiracy—told CNBC that the number was “not correct,” suggesting that it had been manipulated to help Obama’s chances of reëlection. “After the election they will put in a correction,” Trump said. (In post-election reports, the September numbers remained unchanged.) Labor statistics formed the climax of Trump’s 2015 campaign announcement, during which he insisted, “Our real unemployment is anywhere from eighteen to twenty per cent. Don’t believe the 5.6. Don’t believe it.” Responding to a major downward revision to the previous year’s job-creation figures, released in August, 2024—a horrible moment for Kamala Harris’s fragile Presidential campaign—Trump accused the Biden Administration of “fraudulently manipulating job statistics to hide the true extent of the economic ruin that they’ve inflicted on America.” In the same speech, he said, “They wanted this to come out after November 5th when it wouldn’t have meant so much, but it came out a little early, so there’s a patriot in there someplace, right?” That patriot would have been McEntarfer.

If Trump’s critique over the years has been incoherent, it has at least been consistent: the metric of truth—whether it is the revisions to a jobs report that are false or the original numbers, for example—is what makes him look better, or makes his opponents look worse. As in other arenas, the facts shift depending on Trump’s personalist view of the world and on his rhetorical needs. Amid the muddle, it’s tempting to grab at ironies: a bad jobs report increases the pressure on the Federal Reserve to lower interest rates, as Trump has been demanding, but if this one is faulty, perhaps the pressure’s off? This kind of thinking, however, misses the point. Trump’s fluid relationship to facts doesn’t mean that they don’t matter to him: they matter a great deal, so long as they say what he wants, and, if they don’t, he’s happy to ask for new ones. Firing McEntarfer drives home a message that should have been obvious by now: Don’t trust the wonks, trust me.

Trump’s political project may represent a rejection of Washington technocracy, but it’s hard to run a government, or to accomplish your agenda, without stats. The federal bureaucracy is sustained by dozens of statistical agencies, units, and programs, all of which collect data in fields including crime, Social Security, animal diseases, housing, behavioral health, income, small businesses, large businesses, crop prices, transportation, and energy. There is a U.S. Chief Statistician, housed within the powerful Office of Management and Budget, who oversees this dispersed but furiously productive apparatus. The swamp makes its own food, in the form of paperwork.

During Trump’s first Administration, some of these programs were downsized, and some have now essentially ceased to exist—there’s not much for a chief data officer at the U.S. Agency for International Development to do nowadays, even if the post is mandated by Congress. But Trump has gone from being somewhat laissez-faire about data in his first term to being obsessed with data in his second term. Since January, he has moved aggressively to exert control over information produced, hosted, and published by the federal government. On his orders, agencies have buried climate data, rewritten history, redefined categories and definitions (most notably of gender), cut research, and undermined or eliminated independent auditors. Trump’s Commerce Secretary disbanded the Federal Economic Statistics Advisory Committee in March; in June, his Health and Human Services Secretary cleared out the committee that makes vaccine recommendations. Meanwhile, the Administration has been seeking unprecedented access to data, held by the states, on Medicaid and food-stamp beneficiaries, ostensibly as part of a push to eliminate fraud and waste, but actually for the purpose of immigration enforcement. As California’s attorney general, who is resisting the request, put it to the Times, “The ongoing pursuit of data and the ongoing pursuit of power are interlinked inextricably.”

Data are never perfect, and they’re also not neutral: what a government decides to record, and what it does with those records, are always a reflection of its priorities. (In the U.S., where political life has turned on race for centuries, the census has included racial identity, in a telling variety of categories, since 1790; in France, where one’s status as French theoretically supersedes racial identity, government collection of data on race is illegal.) Viewed from this angle, McEntarfer’s removal is simply the next step in Trump’s project of making the federal bureaucracy, and the information it produces, into a tool of his own authority. In a more open form of government, transparency and freedom of information might empower outside actors—Congress, businesses, voters—to make informed decisions and act as a counterweight to (or rally in support of) the executive, while safeguards for privacy provide a check against government overreach. Trump, recognizing that knowledge is power, is centralizing it.

The consequences may be hard to control. Other countries that have experimented with manipulating their economic data, or that have incentivized bureaucrats to do so, have faced difficulties. In Turkey, where government benefits are tied to inflation, the gap between the official rate (which topped out at eighty-five per cent in 2022) and the real rate (an estimate in June 2022 put it at a hundred and sixty per cent) pushed millions into poverty and helped fuel a property bubble. During China’s Great Leap Forward, village officials, under intense pressure to meet agricultural-production standards, routinely inflated their figures, further driving up targets and contributing to a famine that caused tens of millions of deaths.

After McEntarfer’s firing, the White House put out a press release criticizing her for “a lengthy history of inaccuracies and incompetence” that had “completely eroded public trust in the government agency charged with disseminating key data used by policymakers and businesses to make consequential decisions.” Whether this was spin, or downright revisionism, depends on your point of view. The B.L.S. came under scrutiny during the Biden Administration, too—after a flubbed jobs-report rollout last year, during which a handful of banks got early access to the data, an internal inquiry called out a number of human errors, and admonished the agency, in terms that only a bureaucrat could love, to “develop a culture of enterprise-wide collaboration, break down silos, and work across organizational lines to ensure success.” The budget cuts and staffing reductions of the past few months have led to concerns that the bureau was stretched too thin, and it’s conceivable that Trump might have shared those concerns, though McEntarfer’s summary firing suggests otherwise.

For many observers, it’s Trump who is causing the erosion of public trust. Two former B.L.S. commissioners, including William Beach, whom Trump appointed in his first term, signed on to a statement condemning McEntarfer’s removal and paying tribute to their former colleagues at the bureau. Michael Strain, an economist at the conservative American Enterprise Institute, wrote, “It is imperative that decisionmakers understand that government statistics are unbiased and of the highest quality. By casting doubt on that, the President is damaging the United States.” (Charles Murray, a co-author of “The Bell Curve,” one of the more infamous examples of the way cherry-picked statistics can lead to questionable conclusions, replied “Agreed.”)

McEntarfer has not commented publicly since her firing, but her boss, Lori Chavez-DeRemer, the Secretary of Labor, has. She wrote that she supports Trump’s decision to replace the commissioner so as to “ensure the American People can trust the important and influential data coming from BLS”—but it may be too late for that. When it comes to public data, the appearance of interference can be as damaging as actual meddling, and the removal of an independent and credible official makes it harder to tell the difference. Trust can be earned, but distrust can be taught. Trump’s most orthodox supporters learned not to trust the government long ago. Everyone else is now learning the same lesson. ♦