Former US President Donald Trump fired then-Bureau of Labor Statistics (BLS) Commissioner Erika McEntarfer last week after the agency revised down recent job numbers by more than 250,000.

The president's decision created a crisis at the leading US statistical agency, but it wasn't the initial spark for the BLS's difficulties.

Tasked with providing data that informs decision-making in US organizations and businesses, the BLS has been under pressure for several years due to budget constraints and low survey response rates. In recent months, they have also faced severe staff shortages due to the hiring freeze Trump imposed in January.

Former US President Donald Trump (left) and former Commissioner Erika McEntarfer of the Bureau of Labor Statistics (BLS). Photo: US Department of Labor

Former US President Donald Trump (left) and former Commissioner Erika McEntarfer of the Bureau of Labor Statistics (BLS). Photo: US Department of Labor

Trump claimed the BLS's regular data revisions were "deliberately manipulating the numbers for political reasons" to make his administration "look bad". He offered no specific evidence of BLS wrongdoing but said the scale of the jobs revisions was excessive.

Economists defended McEntarfer, stating that the agency's regular revisions ensure data completeness, consistency, accuracy, and reliability. It's a standard procedure dating back decades, stemming from the inherent tension between speed and accuracy.

In its latest report, the BLS revised down May and June job numbers by 125,000 and 133,000, respectively. The total decrease of 258,000 jobs over these two months was the most substantial change recorded since the Covid-19 pandemic in 2020.

However, BLS revisions occur monthly, and large changes are not unprecedented. Many analysts anticipated the June figure revision, given the unusual increase in education jobs when most schools were about to close for summer break.

Since 1979, the average monthly change in BLS jobs data has been 57,000. Revisions tend to be larger during periods of economic turbulence. Apart from the latest figures and the 2020 Covid-19 period, the BLS has revised monthly job figures down by more than 100,000 eight times since 2000, mostly around the 2008 financial crisis.

For instance, the BLS revised down January 2009 job numbers by 143,000 under the Barack Obama administration. The total number of jobs created in 2009 was revised down by 902,000 compared to the initial estimate, the largest annual revision in US history.

The number of jobs created in 2024 under President Joe Biden was also revised down by 598,000, sparking political debate at the time.

The latest revisions highlight a significant challenge the BLS has faced for years: quickly producing accurate economic data while increasing numbers of households and businesses submit late or no survey responses.

The BLS surveys approximately 120,000 employers monthly via phone or online to track employment numbers. This is one of the world's largest and most complex monthly surveys. By the end of the month, the BLS uses this data to provide an initial estimate of job numbers. They update this estimate over the next two months, incorporating late responses and seasonal adjustments.

A decade ago, about 20% of businesses didn't respond promptly to the BLS survey. That figure has doubled in recent months. If late respondents provide different information than the initial group, as happened last summer, the BLS is likely to make large revisions. For example, data from 1/8 showed approximately 109,000 fewer public school teaching jobs in May and June than previously estimated.

According to Omair Sharif, founder of Inflation Insights, this discrepancy largely resulted from late school responses to the BLS survey.

Jonathan Pingle, an economist at UBS, explained, "The declining response rate to the initial release survey makes for larger revisions. They get a significantly increased amount of supplemental information after the first deadline".

During his second-term election campaign in 2024, Trump expressed anger over another BLS revision process, a biannual benchmark revision that reconciles monthly data with a comprehensive federal dataset. This dataset theoretically includes nearly every employer nationwide, allowing the BLS to directly observe business starts and closures, a blind spot in its monthly data.

When initially released in 8/2024, this revision showed the economy had created 818,000 fewer jobs between 3/2023 and 3/2024 than initially estimated. Trump claimed this indicated earlier job numbers were inflated to make the economy appear stronger. The final revision, released in 3/2025, showed the initial figures overestimated job creation by 598,000, not 818,000.

Response rates to BLS household surveys have also declined. A decade ago, the BLS received responses from nearly 90% of households in its monthly survey. In recent months, that figure has dropped to approximately 70%.

Experts say the declining response rate has not yet significantly impacted reporting accuracy. Economists say the revisions, while potentially confusing, demonstrate the BLS's commitment to consistency and reliability, updating reports with the latest data regardless of the consequences.

David Wilcox, former head of the statistical advisory committee for a different government agency disbanded by the Trump administration, described compiling economic statistics as akin to "measuring the clothing of someone running at full speed".

Former BLS officials say budget issues have hindered investments in agency modernization. Congressional funding for the BLS has decreased by 22% since 2010.

In 2021, the BLS received $655 million, which increased to $678 million in 2024. However, while the overall US federal budget has grown significantly over the past few decades, the BLS's share has effectively decreased. In 2024, the BLS budget represented only about 0.01% of the total federal budget.

The Friends of the Bureau of Labor Statistics, an organization advocating for increased BLS resources, notes that the agency needs a larger budget to pursue long-term solutions for improving response rates and speed.

Trump's federal hiring freeze also prevented the BLS from replacing departing staff, particularly impacting inflation report data collection, which often requires hundreds of employees to visit businesses nationwide to check prices.

Staff shortages forced the BLS to halt price data collection in Lincoln, Nebraska; Provo, Utah; and Buffalo, New York. Last week, the BLS reported an average 15% shortfall in price data in other areas. The next BLS inflation report is due on 12/8.

Labor market economist Guy Berger commented, "I don't believe the BLS can't do better. But to say that they can easily do better is also not entirely based in reality".

Katharine Abraham, who led the BLS from 1993 to 2001, emphasized that data updates and revisions are part of the process. She wasn't surprised by the large May and June revisions, citing increasing difficulties in gathering responses, lack of investment in new methods, and a slowing economy, partly due to new tariffs.

"Things always get harder when you're at a point where the landscape can change quickly. What makes the challenge even greater is the shortage of staff and the agency not having enough resources to invest in tracking down survey respondents like they used to," she concluded.

Vu Hoang (According to WSJ, BBC, AP)