Revisions to the jobs report issued by the Bureau of Labor Statistics (BLS) are at the center of a political firestorm after President Trump fired the agency’s head earlier this month.

The agency’s most recent report revised down employment numbers for May and June by a whopping 258,000 jobs, drawing accusations by the president and his allies that the numbers were manipulated for political purposes.

That’s not true, most economists say. BLS instead revises its numbers to account for more information from its nationwide surveys, and the agency remains the gold standard for macroeconomic data in the U.S.

Still, there are measures that the bureau could take, its supporters say, to modernize the collection of its survey data, particularly for its population survey one of two surveys used to compile the jobs report. A group of former BLS heads has asked Congress to fund the agency with at least $770 million for the upcoming fiscal year.

“The greatest way to restore confidence would be ensuring that they have the resources they need,” said Kyle Ross, a fellow at the left-leaning Center for American Progress.

Why the jobs report gets revised

Each month, the BLS surveys a sample of more than 120,000 employers by email and phone, aiming to collect data on wages, total employment and other characteristics. At the end of the month, it publishes an initial estimate of how many jobs the U.S. has added from the data it has.

The BLS also conducts a survey of households to track the employment status and take-home wages for the country at large.

In the next two months, the bureau issues updates to its estimates, incorporating additional responses to the surveys and adjustments for seasonal changes. 

While the August revisions surprised many economists, they weren’t the first time the BLS made large changes. During the pandemic, the agency had to make significant revisions to many of its estimates; in the summer of 2021, for example, it marked down its estimate for June to September job growth by 626,000 positions.

Several key BLS surveys have struggled with falling response rates over the past two decades. The Federal Reserve Bank of San Francisco estimates that response rates to the employment survey are around 45 percent, down to about 60 percent prior to the pandemic.

However, the limited responses do not appear to have impacted the size of the BLS’s revisions after 2022, the bank said in March.

Over more than 60 years of data collection, the agency’s initial job estimates have gradually become more accurate, according to analysis by Ernie Tedeschi, an economist at the Yale Budget Lab.

Concerns over other BLS metrics

Advocates say that while Trump’s claims of political bias are baseless, the agency could use extra funding to be able to modernize particularly on its Current Population Survey, which polls households instead of businesses on employment.

Friends of the BLS, an advocacy group that includes former commissioners William Beach and Erica Groshen, asked Congress in May to fund the agency with at least $770 million for the upcoming fiscal year. 

In a letter to appropriators, the group said that additional Congressional funding would allow the agency to go forward with long-planned updates to its data collection and methods. 

Among other modernization efforts, the agency is hoping to implement an online response model for its Current Population Survey.

Additional funding, Beach and Groshen said, would also help the BLS maintain detailed data for important statistics like the Consumer Price Index, which tracks price inflation. The agency relies in part on data collectors who fan out across the country to monitor prices of goods and services. 

“The field person will literally pick up a jar of, if I could say Pringles, and they’ll say, well last month, we had 36 Pringles in here, and it’s this month, it’s the same price, but we only have 32 Pringles in here,” Beach, who was Trump’s BLS pick during his first administration, told the Bloomberg podcast Odd Lots in April. “That means that the product has actually gone up in price.”

Last summer, in response to budget constraints, BLS mulled cutting the population survey’s sample size by 5,000 households.