I & I Editorial BoardEditor’s note: After this editorial posted, the Bureau of Labor Statistics released a preliminary “benchmark revision” of its jobs data from April 2024 through March 2025, saying it had overcounted jobs created by more than 900,000 — an absolutely stunning failure.
When President Donald Trump tapped economist EJ Antoni to head the Bureau of Labor Statistics, the elites in Washington clucked their thick tongues. Antoni might have a doctorate in economics, but he’s not “widely recognized”! He’s tweeted some things that turned out wrong! Worse still, he seems to like Trump!
But on Friday, the BLS proved once again why it is in desperate need of an overhaul by an outsider such as Antoni.
When it released its monthly jobs report on Friday, the BLS said the economy had created 22,000 jobs in August – a weak number that generated countless headlines.
But who knows what the actual number is? These days, you could throw a dart at a wall while blindfolded and be as accurate as this agency.
In each monthly report, the BLS revises the previous two months’ numbers as additional data come in.
You’d think that these revisions would be rather small, given the massive size of the survey it conducts each month – more than 100,000 businesses and government agencies.
You’d be wrong.
The chart below shows the initial report, and the subsequent revisions, since the November elections. Does this look like the work of top professionals?
The BLS revised its job number for July up by 8%, and that was as close as it’s been for months.
While the BLS initially said that 147,000 jobs had been created in June, it whacked that back by 133,000 in its first revision, and by another 27,000 in its latest revision.
So, instead of a gain of 147,000 jobs, it turns out the economy lost 13,000 jobs in June. The BLS was off in its initial estimate by 108%!
For May, the BLS said the economy gained 139,000. Then, in its first revision, it increased that to 144,000. But in its second revision, it put the gain at a mere 19,000. How is that even possible?
So far this year, the BLS has claimed the economy created 483,000 jobs that later turned out not to exist.
This isn’t a new problem. As we noted in this space last month, the BLS has an absolutely abysmal track record when it comes to accurately reporting the number of jobs gained or lost in a given month (here’s the chart we ran).
But while the initial estimates make headlines, the subsequent – often massive – revisions get ignored.
During the latter part of the Biden administration, the BLS wildly exaggerated job growth month after month, which it then sharply downgraded in later reports. Many, including us, reasonably began to wonder if its mistakes were politically motivated. Every month, President Joe Biden would brag about strong job gains, only to have the BLS quietly erase most of them in its later revisions.
See: Just How Bad Is The BLS At Its Job? Our Findings Will Shock You
It’s still possible that politics is playing a role in how those allegedly pure and unbiased number crunchers at the BLS make their estimates.
But it’s looking more and more like plain old incompetence.
Except in the case of the BLS, it’s rising to the level of malfeasance because its initial jobs reports can have ripple effects throughout the economy.
The “widely respected” economists who’ve been running the BLS can’t or won’t fix this glaring – and worsening – problem. So why not give Antoni a shot? Things could hardly get worse.
— Written by the I&I Editorial Board
Unleash Prosperity
Is there not a narrative explanation by the BLS each reporting cycle regarding the cause of revisions?
You can usually tell when somebody is incompetent at his/her job when he/she always needs a bigger eraser.
It use to be that the eraser the BLS used was for one revision. Now it is for two revisions.
One can’t manage the economy using capricious data. Data is needed to see if you are doing a good or bad job.
When the data is unreliable (as it has significantly been beginning, I believe, with the employment numbers in the Biden Administration), those who read the numbers can’t know if the things they do to improve the economy is right or wrong.
I’m not sure if I&I is correct as far as the BLS data under the Biden Administration is concerned. The employment numbers always seemed to be fantastic in the first month they were reported. And they always were reported as being proof of the vitality in the economy.
Then when it was revised a 100K less employment in the economy wasn’t unusual. This went on repeatedly (and, of course, the revision wasn’t blared in the legacy media’s headlines).
The thing is: It always went one way, in the direction that favored the Biden Administration, even though it was always revised downward.
I believe there was so much incompetence and political shenanigans in the BLS reported data in the previous Administration that now the BLS can’t adjust fast enough to give the real data.
The BLS-data ship is veering and it will take time to adjust the inertia.
As I&I concludes, the new chief of BLS could hardly do a worse job. The BLS under the Biden Administration had turned into a left-wing Hydra monster. Perhaps now under a new BLS chief there will be pristine and actual data-in the current month-reported!
Hallelujah!
I owned a restaurant that was somehow included in the BLS survey. Every month I got a call from Washington D.C. and was asked how many full time and part time employees I had during the month. Since there is a lot of turnover of employees, this was hard to define and I was busy, so I just blurted out my estimate, often later realize I had given inaccurate information. I once tried to quit the survey, saying I didn’t think I was accurate enough and I was also disinterested. However the rep talked me out of it and said “You represent all small businesses in Oregon” If my participation actually represented this, it’s no wonder the data sucks.
A comment to an article on another website asked why the BLS could not just look at payroll tax withholding data reported to the IRS to estimate changes in job numbers. It wouldn’t be perfect, for example not catching jobs paid via 1099, but perhaps there are ways to get those numbers as well. I’ve no idea what the pros and cons to that are but it seems reasonable.