Trump has fired the chief of the Bureau of Labor Statistics, Erika McEntarfer. Trump's justification was that there were too large and, in his opinion, politically motivated revisions to the American employment figures ("Nonfarm payrolls"). I've taken a look at the historical revisions of the labor market figures. It should be noted that one of the reasons for these revisions is that when the first employment estimate is released, it's based on a smaller data foundation than when the final figure comes out. As more responses come in, the estimate becomes more accurate, but this also leads to revisions. It's essentially a survey of American businesses. BLS simply asks businesses how many employees they have in a given month, and based on this, BLS produces an estimate for total employment in the US. Total employment in the US is around 163 million people. Historically (since 1979), monthly revisions have averaged 57,000 people (in either direction). Out of 163 million, this is practically nothing - about 0.03%. There's also nothing to indicate that this has gotten worse over time - if anything, the opposite. Over the past 5 years, the average data revision has been 55,000 people per month. That's marginally less. And this despite the fact that in 2020 - during COVID - there were very large revisions in certain months. That said, revisions have been increasing again over the past 1-2 years, and it appears that simply fewer businesses are responding to BLS's survey than before - and it's taking longer to get responses in. But to say that the numbers are particularly biased is simply not correct. What one might consider is whether it's even meaningful to make so much of the variations seen from month to month. Is it even meaningful whether employment rises by 50,000 or 200,000 in a given month? Relative to employment of over 160 million people, that difference is minimal. If BLS, instead of publishing a figure showing the NUMBER OF PEOPLE CHANGE in employment, had published the percentage change in employment, there would be almost no story to tell.