Each year, the Bureau of Labor Statistics revises its monthly employment figures, which come from a survey of employers, based on more comprehensive data it obtains later from state filings. But former President Donald Trump called this year’s revision a “total lie,” baselessly claiming that “the Harris/Biden administration has been caught fraudulently manipulating job statistics.”
There’s no evidence for such claims, and the Trump campaign hasn’t provided any.
The BLS, nonpartisan number-crunchers who publish a wealth of statistics on employment, wages, inflation and more, announced on Aug. 21 a preliminary estimate that the number of jobs created over the 12 months ending in March would likely be adjusted downward — 818,000 lower than the original estimate of 2.9 million jobs. That’s an adjustment of -0.5% to the March level of employment, larger than the average revision over the last 10 years of plus or minus 0.1%, the BLS said. The final estimate will be issued in February 2025, when the January employment report is released. That’s when the final revisions have been issued each year dating back to 2004.
While the revision would be sizable, there have been other large revisions in the past. The annual revision for 2019, under Trump, was a reduction of 514,000 jobs, or -0.3% of the initial March 2019 employment estimate. The 2009 revision was a reduction of 902,000, or -0.7% of the original March 2009 estimate.
The former president and his campaign have repeatedly claimed that something nefarious is afoot with the latest preliminary estimate. “It really isn’t a revision, it’s a total lie, total lie. There’s never been any revision like this. They wanted it to come out after the election, but somehow it got leaked,” Trump claimed in North Carolina on Aug. 21.
The revision wasn’t “leaked.” The BLS had announced on July 5 that it would release the preliminary revision estimate on Aug. 21, around the time it had done so in past years.
“The Harris/Biden administration has been caught fraudulently manipulating job statistics to hide the true extent of the economic ruin that they’ve inflicted on America,” Trump continued. “The new data from the Bureau of Labor Statistics shows that the administration padded the numbers with an extra, listen to this one, 818,000 jobs that don’t exist.”
As it stands now, the U.S. has added 15.8 million jobs under President Joe Biden, and the revision — if it holds up — would drop that number to about 15 million. That’s not evidence of “economic ruin,” as Trump put it.
Two days later in Las Vegas, Trump claimed, “They reported fake jobs,” again saying the figures wouldn’t have come out but for “a patriot leaker.”
“There’s no evidence whatsoever of any manipulation or padding,” David Wilcox, a senior fellow at the Peterson Institute for International Economics and director of U.S. economic research at Bloomberg Economics, told us in a phone interview. “Last week’s announcement was completely formulaic in two respects. One was that it followed the pattern of many years observed of a preliminary estimate being announced, generally speaking, during the summer, prior to the final revision being implemented in, with the release of January data. So this is a pattern that has been many years in the making.”
Second, Wilcox noted that BLS had announced on July 5 that it would publish the preliminary estimate on Aug. 21.
How BLS Estimates Employment
The BLS uses what’s called the “establishment survey” to estimate the change in nationwide employment each month. It’s a survey of about 119,000 employers and covers about 30% of U.S. employment.
“It’s a well-constructed survey with weights that are carefully calculated to try to make the results as representative as possible of the types of employers in the United States by size and location,” Wilcox, a former director of the Federal Reserve Board’s statistics division, told us. He is also chair of the Federal Economic Statistics Advisory Committee, a group of independent experts who advise government statistical agencies, but he stressed that he was not speaking on behalf of the committee in our interview.
Like any survey, the establishment survey is subject to error. To get a more comprehensive estimate, the BLS uses state unemployment insurance tax filings that employers submit to determine what taxes they owe to unemployment benefit programs. That data isn’t as timely as the monthly surveys, but it covers about 97% of U.S. nonfarm employment.
Once a year, based on the March level of employment from the unemployment insurance filings, the BLS adjusts its monthly estimates in a process called “benchmarking.”
The 818,000 estimated adjustment announced on Aug. 21 “was a little larger than normal,” Wilcox said, but a downward revision was expected by many analysts, who were keenly interested in how much the figure would show the labor market had been softening.
Goldman Sachs analysts said they had expected a downward revision of 600,000 to 1 million jobs. However, they also noted that BLS’ estimated revision may be about 500,000 too large, because many immigrants who are working but in the country illegally are likely not included in the unemployment insurance system — but would have been included in the establishment survey figures — and because the unemployment insurance data has often been revised upward.
The BLS will finalize the estimate and make the final adjustment in February, as it has done for more than two decades. The final figure likely won’t be exactly 818,000, “but historically, the preliminary estimate has been a reasonably good guide to the final benchmark revision,” Wilcox said.
We asked Wilcox about the BLS’ nonpartisan nature, and he said, “There’s only one political appointee at the agency, and that’s the commissioner, the No. 1 person, and you can bet your bottom dollar that the protests from the rank and file, from the senior leadership, up and down the organization, would be deafening if there was any move on the part of the administration to try to influence the numbers that BLS produces.”
We asked the Trump campaign for support for his claims that the original BLS estimates were somehow fraudulent and his claim that the preliminary revision estimate “got leaked” before the election. We didn’t get a response.
Perhaps Trump’s “leak” comment is a reference to the BLS posting the estimate at 10:30 a.m. on Aug. 21, instead of 10 a.m., as it had announced it would do. Some eager analysts called BLS during the delay and were able to get the estimate, as the New York Times reported.
But that’s not evidence that the estimate wasn’t supposed to be released at all, or that “a patriot leaker” had forced the BLS’ hand. The agency’s summer release of the preliminary estimate tracks with the same procedure it has followed for years, including during Trump’s administration.
Editor’s note: FactCheck.org does not accept advertising. We rely on grants and individual donations from people like you. Please consider a donation. Credit card donations may be made through our “Donate” page. If you prefer to give by check, send to: FactCheck.org, Annenberg Public Policy Center, 202 S. 36th St., Philadelphia, PA 19104.