Is there really a problem with the number of jobs in the country?
Questions about the quality of employment data came into the spotlight this month when President Donald Trump fired the boss of the Bureau of Labor Statistics. totalling. Trump claims that the bureau will produce biased numbers against him without evidence.
One flash point for statistical dust was a 133,000 revisions to previously reported for seasonally adjusted non-farm employment in June. As a result, profits have been reduced from 147,000 to 14,000.
In an age of immediate analysis, thin job creation has changed the economic storyline.
The original data suggested that the new administration’s unorthodox business agenda is boosting employment. The revision raised fears that Trump’s new policies were being cooled down.
And the wild claims of the president’s massive inaccuracy are more than just political theatres. Sowing questions about the reliability of government data can undermine trust in businesses and consumers and damage the entire economy.
A small slice
I’m a numbers guy and I know that this job market math can really be sneaky.
With that warning being stated, I found my trusty spreadsheet watching a tracking of BLS’s own revisions. The historically nonpartisan Bureau of Statistics details its revision in 1979, so it is very open about its methodology.
Focus on June 2025. That revision wiped out 90% of previously reported profits. Only 13 months after 1979, the first revision grew.
And then we look back on the past 47 years. The first revised version of a typical month reflects the average swing of 41,000 employment changes up and down. That’s about a third of the average 125,000 jobs added per month since 1979.
So June was definitely an eye-catching person.
However, try a different lens to see these changes. Think of the modification as a slice of all US workers with employers and not farmers. By the way, self-employed people are listed in this tally.
The 133,000 downward revisions in June are not 0.1% of 160 million U.S. jobs. And the sharing is comparable to the average results dating back to 1979.
The revision is a statistical improvement comparable to a small slice of the entire job market. These adjustments only look great when looking at the one-month swing of employment.
Therefore, it can be argued that the first draft number is 99.9% correct.
This is one of many reasons why all political stripe economists have recently said the US is the gold standard for employment data. They all condemned the BLS Chief for firing.
Furthermore, let’s all acknowledge that we all have issues with economic data. Worrying about a month-long turn of any business variable doesn’t mean that you’re firmly giving anyone anything good, bad or indifferent to what’s going on.
Easy work
The revision is not an error entry. They are part of the work tracking process.
Think about that. Have you ever wondered how national employment numbers will be announced at the end of the month?
These figures are based on a survey of US workplaces that amounts to around 600,000 US dollars. One Wrinkle: Not all bosses respond quickly. Regardless of who is doing it or what topics are being surveyed, the response rate is low and the response rate is lower.
Please understand that the 1-Friday work report is that all of us chat only reflect what BLS collects in the first 12 days of the previous month. The initial survey typically covers about three-quarters of the employers in the survey.
Therefore, the first estimate of the BLS for employment in June, released on June 3, was simply rapid mathematics.
On August 1st, BLS gave us her second peep in June. The survey results have increased. In other words, it was a better snapshot of the employment picture. And the trend wasn’t pretty.
Incidentally, there are at least two revisions to the June number.
The first is listed in the upcoming monthly employment report scheduled for September 5th. The second revision since 1979 generates an average of 31,000 changes. And that’s 36% of the time.
Trump’s BLS critique was driven by the second revision issue in May, along with the first in June. It saw a downgrade of 125,000, the fifth-largest job count since 1979.
The next June revision is scheduled for February 2026, when BLS will implement its annual “benchmark.”
Would you like to see what other unemployment insurance-based statistics say about the second quarter job market, including data that sparks the May and June debate? The report will be published in December.
Essentially, tracking the job market – at least how the federal government has done it for decades involves making quick estimates to improve the data as new information flows.
Where is the beef?
Is Trump justified complaint that the job counter isn’t fair to him?
Take a look at all 53 months he has been in office through his two conditions.
Job’s tally included 26 downward one-month revisions. That’s 49% of the time. These revisions changed job counts on average 40,200 jobs. The final outcome of the revision was to add an average of 100 jobs to total employment.
Compare that track record with the first month revision of 557 months since 1979. 267 downward revision or 48%. The average swing was 41,000 and the average net addition was 1,800.
It’s difficult to see a lot of bias.
Jonathan Lansner is a business columnist for the Southern California News Group. He can be contacted at jlansner@scng.com
Original issue: August 6, 2025, 11:40am EDT
