President Donald Trump claimed without evidence that the massive revisions to the latest jobs report constituted a “scam,” accusing one of the top overseers of government statistics of cooking the books in a vendetta against his presidency.
“In my opinion, today’s Jobs Numbers were RIGGED in order to make the Republicans, and ME, look bad,” Trump said in a message on Truth Social Friday.
As a result, he fired Dr. Erika McEntarfer, the commissioner of the Bureau of Labor Statistics.
Trump, in a message posted to Truth Social Sunday, incorrectly claimed McEntarfer “had the biggest miscalculations in over 50 years.” In fact, the revisions the BLS made to recent jobs reports were neither historic nor evidence of corruption.
Established in 1884, the Bureau of Labor Statistics (BLS) is an independently operated body within the US Department of Labor. The Labor Secretary, a member of the president’s cabinet, has oversight of the BLS, but it is run by a Senate-appointed commissioner.
The BLS collects data on a number of key economic concerns and produces critical reports on a regular basis. Those include data on prices, inflation, productivity, spending, pay, workplace injuries, employment and unemployment.
More than 2,000 people work for the BLS, including a number of professional economists and survey takers who regularly contact businesses and employees. The economists analyze the data and produce reports for the public and the government.
The BLS collects jobs data in two separate surveys. The first is done in part with old-fashioned door knocking. Survey takers go home to home throughout the country asking people for their employment status and their demographic information.
A second survey, known as the Current Employment Statistics (CES) survey, is collected from thousands of businesses and government agencies using a variety of methods: via telephone, internet surveys, and – for large corporations – through an automated data transfer. By law, participation in the CES survey is voluntary, but state laws in New Mexico, Oregon and South Carolina require businesses to produce data for the survey. A law in Puerto Rico also requires the territory’s businesses to submit data to the BLS.
CES survey respondents submit monthly employment, hours, and earnings data for all paid workers to the BLS from their payroll records. The data is collected for the pay period that includes the 12th of the month.
BLS economists first edit the data to detect processing and reporting errors. If mistakes are detected, BLS employees contact the business for clarification.
Then, BLS staff prepare the data for a monthly report by estimating America’s employment, hours worked and earnings. To extrapolate the data for the entire country, BLS economists add in some educated guesswork, based on seasonal hiring trends. The BLS also smooths out the data with calculations known as seasonal adjustments to avoid huge spikes and dips in data each month.
The BLS also protects the raw data and its estimates by processing them through statistical tests, so individuals can’t gain access to any particular employer’s data.
Every month, typically on the first Friday, the BLS produces its Employment Situation Summary, known colloquially as the monthly US jobs report.
The report is generated from the two surveys: The household survey provides demographic data and the unemployment rate. The business survey provides data on pay, hours worked, and the number of jobs the US economy added or subtracted.
In addition to that particular month’s jobs information, the report also revises up or down the previous two months’ jobs totals.
The BLS considers its initial jobs numbers to be preliminary when they’re first published, because some respondents fail to report their payroll data by the BLS’ deadline. Low survey responses can make the report more challenging to estimate. But the BLS continues to collect the payroll data as it’s reported, and it revises the data accordingly.
The data are also revised because of the seasonal adjustments. If the more complete data comes in well above or below the preliminary data, revisions can be exacerbated by the BLS’ seasonal adjustments, which sometimes need to be recalculated.
The data is revised several times: in each of the two months following the initial report, and then a preliminary annual revision in August and a final annual revision in February.
The July report, issued Friday, included revisions for May and June that were historically large, but they were not unprecedented.
May’s jobs total was revised lower to 19,000, down from an initial estimate of 139,000 – a total revision of 120,000 jobs. For the June jobs total, the BLS on Friday said the US economy added just 14,000 jobs, down from a preliminary estimate of 147,000 – a revision of 133,000 jobs.
The BLS tracks each month’s revisions dating back to 1979, but the BLS introduced a new probability-based sample design for revisions in 2003. Between 1979 and 2003, the average monthly revision was 61,000 jobs. Since 2003, the average monthly revision is only a slightly more accurate 51,000 jobs.
But you don’t need to look so far back to find larger revisions than the ones the BLS reported over May and June. In 2020 and 2021, jobs numbers were all over the map because of the pandemic: Revisions in four months during those years were bigger than the revisions from last month’s report – including the largest-ever 679,000-job revision in March 2020, attributed to particularly poor survey responses during a nationwide lockdown.
There have been bigger revisions outside of the pandemic, too, including a 143,00-job revision in January 2009.
Trump has complained about a preliminary annual revision that was issued in August 2024, that showed the US economy had added 818,000 fewer jobs over the past year than previously reported. Trump in a Truth Social post on Friday incorrectly called that revision a “record”: A 902,000-job revision in 2009 was larger. And the final 2024 revision, issued in February, showed that the 2024 data was overestimated by 589,000 jobs. The BLS said the difference between the initial and final annual revisions was due to information received in US tax returns.
Trump also correctly noted on Friday the the BLS revised lower initial jobs totals in August and September 2024 by a combined 112,000 positions before last year’s presidential election. But that revision was not out of the ordinary – several revisions were larger earlier in the year and in previous years. And October’s jobs numbers, reported just days before the election, constituted the worst month for jobs since the pandemic.
Many businesses and government organizations rely on the BLS data for their decision-making about investment, pay and hiring decisions. The Federal Reserve, in particular, relies on the BLS data to help guide its monetary policy and rate-setting.
Federal Reserve Chair Jerome Powell last week said the entire economy relies on strong data.
“Good data helps not just the Fed, it helps the government, but also helps the private sector,” Powell said at a press conference last week. “It’s very hard to accurately capture in real time the output of a $20-plus trillion economy, and the United States has been a leader in that for 100 years, and we really need to continue that, in my view.”
In June, Powell told Congress he was concerned about the “trajectory” of weaker data.
In addition to the Fed and the private sector, the BLS notes its survey data is used by the National Bureau of Economic Research to determine whether or not the economy is in a recession. It’s used by the Conference Board to produce economic indexes that help companies better understand the current business cycle. And it helps the BLS produce several other reports about jobs and the economy throughout the year.
Alternatives to the BLS data have proven limited. Payroll processing company ADP produces a monthly private payrolls report that fails to capture government hiring and is notoriously out of sync with the BLS report. Its inconsistencies have led economists to largely ignore the report.
Other surveys, such as a layoffs report from outsourcing and placement services firm Challenger, Gray and Christmas, is sometimes instructive but far less robust than the BLS report.
The BLS report has its challenges too. For example, the household survey that includes the monthly unemployment rate and demographic data is considered rather volatile because of its smaller sample size and declining response rates. But the business and government employment survey is widely considered by economists to be the gold standard.
“BLS is the finest statistical agency in the entire world, it’s numbers are trusted all over the world,” former Commissioner for the Bureau of Labor Statistics William Beach told Kasie Hunt on CNN’s “State of the Union” Sunday.
“I do believe, though, that the president’s attack on the commissioner and on the bureau is undermining that infrastructure, could undermine that trust over the long term,” he added.
