Erika McEntarfer, director of the U.S. Bureau of Labor Statistics, is attending a staff event on August 1 last year when she sees a strange email on her phone that Friday afternoon. Whether she wants to respond to a social media post by President Donald Trump, a journalist writes. “I immediately thought: hmm, let me go to my office and take a good look at this.”
Behind her desk she sees the message from the president. “I have instructed my Team to fire this Biden Political Appointment, IMMEDIATELY,” reads a long, angry post on Truth Social. Earlier that day, the Bureau of Labor Statistics released U.S. labor market figures for July. These were bad – and they go hand in hand with revised figures for June: a significant deterioration.
Trump accuses McEntarfer of manipulating the figures “for political purposes.” “She will be replaced by a much more competent and qualified person.”
McEntarfer, dryly: “At that point I already had many more emails.”
The resignation of McEntarfer (1973), a respected labor market economist with a long track record in the US federal government, shocked many Americans. Trump’s move was widely seen as political interference in a traditionally strong apolitical agency, comparable to the Dutch CBS, which is crucial for insight into the world’s largest economy.
Erika McEntarfer
The monthly labor market figures, plus the inflation rate, are the most important indicators of the American economy. The central bank, the Federal Reserve, relies heavily on it when making interest rate decisions. Stock market traders, pension funds, retail investors – they all have the dates on which the BLS publishes figures in their diaries.
Both Democrats and Republicans immediately condemned McEntarfer’s firing. NRC approached her for a conversation about the impact of her dismissal on herself and on the BLS – and also about her new research. She now works at the Californian Stanford Institute for Economic Policy Research, where she studies the effects of AI on the American labor market – a hyper-current topic. The conversation, which takes place by video, is one of the rare interviews she has given since August.
How did that Friday go?
“It was an ordinary one jobs day – until it wasn’t anymore. At eight o’clock in the morning we had updated the Minister of Labor about the figures. There were some unusually large revisions to previous figures, but these were explainable. It happens regularly because some of the required data only arrives later. I asked if they had any questions, they didn’t have any.”
Behind her computer later that afternoon, McEntarfer initially doesn’t believe the president’s message. “I thought he was mainly threatening, just like against [centralebankpresident] Jerome Powell. But I knew immediately: we had to prepare a media response. I started thinking: my head of communications was at the dentist, how could I reach her?
“It was only then that I realized that I also had an email from the Presidential Personnel Office. So I opened it and saw a resignation letter.”
You previously hinted that you didn’t believe it because the effect on the economy would be so obviously bad.
“In general, you don’t want to shake markets by damaging confidence in economic data. Good economic data, such as the US has, pays off. In lower interest rates, more investments, more security. I was so naive to think that the people who [Trump] from pulling the trigger and firing Powell would also tell him that doing so would damage markets’ confidence.”
After McEntarfer’s dismissal, Trump wanted to appoint economist EJ Antoni, affiliated with the very conservative think tank Heritage Institute, but ultimately decided against it. The BLS is now again run by an economist with experience in government.
Was it scary to be targeted by the president?
“Oh yes, absolutely. I was worried about my family’s safety and felt guilty towards them. Just to be on the safe side, we moved out of our house for a week and stayed elsewhere until the commotion died down a bit. I mean… it was real crazy…” McEntarfer sighs and seems to be having a hard time. “Let me think about how I explain this.”
There is a silence of 24 seconds. “It’s very difficult to describe what it’s like to be in a media storm. Journalists are very good at finding things about you that you don’t even know are there. I got a barrage of messages. It went on and on for months. It was a very difficult time.”
Even before her dismissal, her job had become more difficult, McEntarfer says. Shortly after Donald Trump took office, she was contacted by DOGE, the tough austerity agency led by Elon Musk.
McEntarfer was initially hopeful about that club, she says. “I was told that they would help us with AI.” That in itself was welcome. The BLS had ideas for this, such as using artificial intelligence to make it easier for companies to complete BLS surveys.
“Instead, all the conversations were about: we want to eliminate this and this and replace it with AI. It didn’t matter which function.” She laughs. “I ultimately concluded that they just wanted to cut jobs and that it would be my job to replace them with AI.”
DOGE grew into a kind of “daily internal DDoS attack,” says McEntarfer. She had to fight time and time again against plans by the service, which accused her of not cooperating. “The last six months I was just trying to make sure the BLS wouldn’t shut down. It took me two months after my discharge to catch up on my sleep.”
The BLS ultimately lost 15 to 20 percent of its workforce, McEntarfer says. “That has had an impact. We have not been able to release certain inflation data.” The same goes for other government agencies, McEntarfer says, such as the U.S. Census Bureau, which conducts population surveys. “Agencies that were hit by DOGE are now trying to hire people again, but that has not been immediately arranged.”
We’re starting to run out of steam, and that was already before DOGE
Is understanding of the American economy in jeopardy?
“There is a lot of fear of political interference in various agencies. I think we need to be careful, we don’t know what the government is planning. But at the moment, the civil servants who work there still say that they experience no political interference, and I have no reason to doubt that. I also know that some people have postponed their retirement so that they can resign in protest as soon as interference occurs.”
McEntarfer actually sees the biggest risk from a different angle. “The budget of the BLS is 18 percent lower than fifteen years ago. When I was director, we had to cancel certain surveys because there was no budget for them. The number of respondents is also decreasing, which is a global trend. We are starting to run out of money, and that was already before DOGE.”
McEntarfer starts about the United Kingdom. “The national statistics office has run into problems there. Three years ago, their labor market survey completely collapsed. The Bank of England is now trying to make monetary policy with very little information about the labor market. That is quite a challenge. If something like that were to happen in the US, the consequences would be much greater – simply because the American economy is much more important to the global economy.”
During the government shutdown at the end of 2025, the BLS’s work stopped, but there was still commercial data on, for example, the labor market to provide a picture. Can’t they partly take over the function of the BLS?
“Commercial data can certainly play a role. Sometimes the quality is even better, because they receive better data from respondents. For example, the BLS already uses data on gasoline prices from a commercial provider.” But there are always risks involved. “When I was still at the BLS, private equity bought the provider of egg price data and dramatically increased the prices of the data set. We had to come up with a solution at the last minute.”
She knows that the BLS for gas prices still sends a small group of people out to gas stations – despite the commercial data it collects. “The agency still wants to train people who can do this, if the need ever arises.”
There are now many studies into AI in relation to the current weak labor market, and they overwhelmingly indicate that there is no connection
A barbershop in Austin, Texas. The US labor market is weak, which some attribute to AI. However, this is not yet the case in many professional groups, such as hairdressers, McEntarfer notes
Photo Brandon Bell/AFP
McEntarfer is now studying the effects of AI on the American labor market at Stanford, a topic that is receiving a lot of attention in the US. The US labor market is weak, and young people seem to have a harder time finding work after graduation, which some attribute to AI. Several companies have also linked significant layoffs to AI.
McEntarfer, who specializes in shifts in the labor market, is skeptical about this. “There is now a considerable amount of research into AI in relation to the current weak labor market, and they overwhelmingly indicate that there is no connection.” For example, unemployment is increasing in professions where AI does not have much effect at all, says McEntarfer, such as the hospitality industry.
According to her, research also shows that only one in five companies really incorporates AI into their business model. “People do use it to write emails. But the question is more: do you use AI in your HR department, at your help desk?” Often that is not the case, and in a way that makes sense, says McEntarfer. “Many companies have concerns about data security and liability. In healthcare, for example, you really need to keep a person involved in all processes.”
That is true now, but do you expect AI to have more impact in the longer term?
“My profession is really struggling with that question. Is AI a normal technology, for which we can look to the past to see how companies have dealt with innovation? Or is it an abnormal technology, for which the past offers us no tools? I don’t know, but I keep a pretty close eye on the data. And so far I see no signals that we are on an abnormal path. But it may take a very long time before we know for sure.”
