Was it rigged? / What prompted Trump’s ire was the BLS’s monthly Employment Situation report covering Current Employment Survey (CES) data for July (technically, for the middle of the month). The top-line number for job creation was a disappointing 73,000, about half of a typical month over the past two years. But that wasn’t necessarily the worst news.

As is common, the report included revised estimates for the previous two months’ data. Earlier BLS estimates put job creation at 144,000 for May and 147,000 for June, and Trump and his supporters had used those numbers to discredit claims that his tariffs and other economic policies were hurting the economy. But the revised numbers were an anemic 19,000 and 14,000, the largest revision since April 2020, in the depth of the COVID pandemic.

Several factors explain the changes. Some 30–40 percent of businesses—many of them small businesses but also state and local agencies—are a month or two late in returning their voluntary BLS survey. The CES response rate has fallen from 60 percent to 43 percent over the past decade, increasing the sampling error. Over the same period, the BLS budget has fallen by 12 percent (not counting the 8 percent additional cut proposed for this year). A further complication: Unexpected events do happen by chance; the Wall Street Journal noted, “The jobs estimate can also be off in either direction by 136,000 in any given month because of statistical chance.” Last but not least: Trump’s chaotic policies, his Depression-throwback tariffs, and his bossy and threatening ways toward businesses and groups he considers disloyal are sufficient to account for the lackluster employment situation and the slowdown of GDP growth.

Those explanations are far more likely than the idea that McEntarfer “rigged” the numbers. For one thing, she probably had no direct input on the report: The chief statistician is not normally directly involved in a regular publication, leaving it to the staffers tasked with that duty. Any direct political interference from senior agency management would presumably create a break in the methodology or workflow process, creating the possibility of detection in an audit. The BLS employs close to 2,000 people, and many of them are involved in the gathering and analysis of survey data. Organizing a successful conspiracy among so many people would be virtually impossible. As we learned in the 2020 election, many civil servants, including humble ones, show personal integrity and would resist corrupt interference. In addition, security protocols are in place to protect data, including siloed workflows.

Federal statistical agencies have a tradition of political independence. As a National Academies of Sciences handbook instructs, “Federal statistical agencies must be independent from political and other undue external influence in developing, producing, and disseminating statistics” (National Research Council 2021). This principle is probably part of the Western classical liberal tradition more than a feature of American exceptionalism as such. In many Western countries—think of Canada, the United Kingdom, the European Union, and its major member countries—it is unlikely that official statistics can be politically manipulated or that a chief of government or head of state would politically survive his firing a chief statistician. One likes to think that any such attempt would be followed by a cascade of protest resignations.

Possible consequences / The direct consequence of political interference such as the firing of a chief statistician because his—or, indeed, her—bureau is producing statistics that displease a political boss will be to undermine the perceived reliability of official statistics, not only for the targeted agency but also the 12 other federal statistical agencies.

The incentives of McEntarfer’s successor will be to avoid being disavowed and fired. Even without bogus numbers, the format of reports can be changed or spun to please the political boss. Otherwise, what was the point of firing McEntarfer? Perhaps somebody from Fox News could do the job better?

Trump’s own words about what he wants from government statistics were insightful. He declared that BLS statistics must be “fair and accurate.” “Accurate,” of course, but what is a “fair” jobs number? That favored workers get more statistical jobs? It’s worth remembering that, earlier this year, Trump disbanded the Federal Economic Statistics Advisory Committee, which provided expert guidance on maintaining the integrity, relevance, and methodological rigor of official economic data.

That’s why there’s concern about the long-term consequences of the firing. Individuals form expectations based on their experience and understanding, expectations that guide their interactions. Now, there is growing expectation that federal data are not reliable, and people will make decisions with that in mind. It is not certain that the next president can reverse that damage quickly—or even that he would want to do so instead of competing in a race to the bottom of massaged statistics. Reestablishing the independence and reliability of federal statistical agencies could take more time than it took to build their former reputation.

Bad examples / What is certain is that more political interference would make matters worse. A well-known example of political tampering with statistics was the falsification of budget deficit data in Greece in the 2000s, which contributed to the government’s debt crisis in the early 2010s. The crisis required the intervention of Eurostat, the European Union’s statistical agency, as well as bailouts from the International Monetary Fund (IMF) and European institutions over several years. In reward for his integrity, the chief statistician responsible for correcting the situation was prosecuted several times by the Greek government.

About the same time, Argentina’s president, Néstor Kirchner, removed a prominent government statistician for reporting high inflation numbers that complicated the path of his wife, Cristina Fernández de Kirchner, to succeed him. The loyalist he installed dutifully reported an inflation number Kirchner preferred, and Fernández de Kirchner cruised to victory. Soon afterward, inflation reached 25 percent according to private economists. It would take a decade for Argentina to begin generating reliable inflation reports. Inflation continued to rage, reaching 289 percent in 2024, before new president Javier Milei’s policies began taming it.

Autocratic rulers typically assert control over national statistics agencies, with dire consequences. In January 2022, Turkey’s Recep Tayyip Erdoğan fired the head of the country’s statistical agency after a report showed that the annual increase in the consumer price index had reached 36 percent the month before. According to IMF data, the increase peaked at 72 percent in 2022 and had only diminished to 59 percent in 2024. We can reach back further to the many stories of Stalinist Russia and Maoist China, where blatantly false economic numbers were produced either at the rulers’ command or to avoid their wrath.

>Usefulness of official statistics? / Most people rely on the accuracy of federal statistics, whether they know it or not. For example, the Consumer Price Index—also produced by the BLS—determines the indexation of federal tax brackets and Social Security benefits. Official statistics influence business and investor confidence and behavior. The information of all market participants shapes such things as the interest on government bonds and mortgage rates. A loss of trust in government data would add a risk premium to government debt and could potentially affect trillions of dollars in assets.

Non-manipulated government statistics also provide useful information for increasing our knowledge of social and economic processes. These statistics are based on known methodology, explained in detail in handbooks available on the internet, and openly debated. They provide long and consistent time series.

Official statistics are certainly not perfect, including those of the BLS, which is why public and scholarly discussions are necessary. Statistical bureaus remain government bureaus, after all, with their own incentives and biases. Politicians are not better.

Private substitutes could at least partly replace official statistics. Some already exist, such as those of ADP, a payroll company that produces employment estimates in cooperation with Stanford Digital Economy Lab. “During the 17-day government shutdown in October 2013,” notes the Wall Street Journal, “private-equity firm Carlyle Group began publishing proprietary estimates for retail sales, inflation and economic growth” (Saeedy et al., 2025).

But private data don’t typically have the scope, details, continuity, or methodological transparency of government data. Some must be purchased. They often rely, at least in part, on official statistics. Perhaps, if faith in official statistics declines, more academic research groups and private think tanks will produce government-like data—even if MIT’s effort a couple decades ago to do this for inflation, the Billion Prices Project, fizzled out partly because its online data-gathering was limited by the absence of many service prices.

Hopes and dangers / Looking on the bright side, there may be a benefit from an erosion in the credibility of government statistics: It may deprive fuel for proposals of government intervention. The continuous flow of statistical information on politically defined “social problems” generates a growing demand for government solutions. Many statistics were conceived precisely for that purpose or are used that way, such as inequality and poverty statistics (Lemieux 2023). John Cowperthwaite, the laissez-faire colonial administrator who was a key contributor to Hong Kong’s booming economic growth, banned the production of macroeconomic statistics for that very reason (The Economist 2017).

The dark side is at least as easy to predict. In a blog post earlier this year, I wrote about Syldavia, a fictitious French country (Lemieux 2025). (The name is borrowed from The Adventures of Tintin.) I wondered, how would the ruler of Syldavia respond to inflation and economic stagnation in lieu of the Golden Age he had promised? The first step, I suggested (but only the first step), would be to “try to hide the inconvenient numbers by stealthily ordering the deep state’s statistical agencies to cook the books.” No doubt, if the chief statistician were honest, a firing would be needed.

A false belief that everything is rigged outside the reign of morally pure politicians risks becoming a self-fulfilling prophecy, but with the rigging done by impure politicians and demagogues. Friedrich Hayek’s The Road to Serfdom seems more prophetic than ever with its warning of the rule of the worst and “the end of truth” (Lemieux 2021). But America has not yet gone totally Syldavian, and there is still hope.

Readings

• Kiernan, Paul, 2025, “What Happens When Politicians Meddle with Economic Data: Argentina’s Example,” Wall Street Journal, August 10.
• Lemieux, Pierre, 2021, “Where Are We on the Road to Serfdom?” Regulation 44(3): 50–52.
• Lemieux, Pierre, 2023, “Is Economic Inequality Bad, Large, and Increasing?” Regulation 46(4): 53–57.
• Lemieux, Pierre, 2025, “Price Increases and Possible Consequences,” EconLog, February 14.
• National Research Council et al., 2021, Principles and Practices for a Federal Statistical Agency, 7th ed., National Academies Press.
• Saeedy, Alexander, et al., 2025, “Trump’s BLS Firing Tests Wall Street’s Reliance on Government Data,” Wall Street Journal, August 4.
• The Economist, 2017, “Meet the Invisible Hand Behind Hong Kong’s Rise,” October 5.