For the second year, the BLS annual benchmark revision was hugely negative.
2025 Annual Revision Notes
- In its 2025 annual revision, the BLS revised nonfarm payrolls for 2025 lower by 1.029.
- Most of the change occurred between January and April.
- The biggest revision was -785,000 in January of 2025.
- For the full year, the as-revised job gain was a grand total of 229,000. That’s an average of 19,000 jobs per month.
- Between April 2025 and December of 2025, the entire job gain was only 12,000. From July to December the Economy lost 45,000 jobs.
- These numbers are subject to further negative revisions (which I expect).
2024 Annual Revisions
2024 Annual Revision Notes
- In its 2024 annual revision, the BLS revised nonfarm payrolls lower by 610,000.
- Most of the change occurred in January with a -511,000 adjustment.
- For the full year, the as-revised job gain was a 1.877 million. That’s an average of 156,000 jobs per month.
The economy is barely adding jobs now.
For January 2026, the BLS did report a gain of 130,000. That’s highly likely to be revised lower a year from now in the next annual revision.
The BLS needs to look at actual tax data instead of flawed sampling coupled with an even more flawed Birth Death model. I have been meaning to write up a specific proposal and will do so.
My normal monthly jobs report will be out shortly. Stay tuned in.
Related Posts
February 5, 2026: Challenger Job Cuts In January Highest Since 2009, Lowest January Hiring Ever
It’s more grim data to start the year.
February 10, 2026: Highest Delinquency Rate Since 2017, Having Trouble Paying Bills?
Credit card balances 90+days delinquent is the highest since 2011.
February 10, 2026: Retail Sales Fizzle, Real Sales Turn Negative from a Year Ago
Will the unexpected consumer spending weakness continue?
February 7, 2026: Education and Health Services Is Now the Sole Driver of Jobs
A recession proof industry is the last industry standing.
Te only reason Trump was elected is NewsCorp and its many mouthpieces like Fox News, The Wall Street Journal, Barrons, and a plethora of TV and Radio Stations and newspapers. It is the largest media company on the planet and it has a political agenda to promote fascism and corporate interests.
Why do you think they call it TV programming?
They program their illiterate viewers like they promote fake wrestling or other “dispensed disbelief” activities.
Fox is downright evil and false reporting…
nope. trump was elected because he had a hit reality tv for well over a decade. your farmer friends thought he was really a successful businessman. the people in gotham city all knew he was a grifter back in the 70s and 80s. if not for his TEEEEEVEEEE stardom he couldn’t win dog catcher. even in ruby red dumbfuck maga land……..
Yea. And it had nothing to do with his opponent. Kamala was the worst ever and Tim was the worst vp candidate ever.
big time fuck up that the Blue team uniparty did NOT open up their convention for an actual competition. when Mamala said she would not change a thing that genocide joe did, she sealed the fate of the election. blue red team hooey of uniparty of warmongering imperial spendthrifts is a lame joke. i cannot believe grown men get played so easily.
frosty. i owned 10% of a newspaper in charleston SC in the 1990s. trust me, all media just delivers what the readers and viewers want to consume. this is not complicated and as old as dirt. the advertising pays the freight and has much influence, too. do you think people that read the economist all are in the market for a boeing 747 ? this has been going on since the roman bread company underwrote the news coming out of the Roman senate a few years ago.
He was elected because a certain “greatest ally” coached him perfectly (to appeal to the average American) and pumped untold millions of dollars into his war machine. At the same time, they completely poisoned peoples’ view of the left by promoting all the trans BS and illegal alien things that most Americans (even many “traditional” democrats) cannot stomach at all. Look at the Bad Bunny nonsense, more of the same forced decay designed to keep Republicans popular as the lesser of two evils in 2026. The goal was to get their greatest puppet back in by hook or by crook so they could do exactly as they wish and get however many billions they wanted from the taxpayer and lo and behold, it all went exactly to plan.
I know this is not related to the article but there is something terribly wrong with the orange thing’s argument about elections. If the 2020 elections were rigged, why were’nt the 2024 election rigged, he run in both. Is the 2024 election also rigged, but this time by you, and that’s why the 2020 election was supposedly rigged. The more the orange thing try to deligitimize the 2020 election the more he puts into question 2024 election. Maybe journalist should point that out to him and others. But I guess people don’t think much anymore they just regurgitate the idiocy that fits their narrative.
Knowing all that how can we trust any numbers coming of this administration? We know they are corrupt and accepting it is exactly what the orange thing wants us to do. Let us not be fooled by the orange thing political machine. We need to all agree and understand that the foundation of our democracy have been destroyed and ‘truth’ doesn’t matter. The number will be revised because their bullshit. And to think the entire stock market is running on agreed upon falsehood and delusional narrative reinvented every 3 months to retrofit what ever backstory is necessary to get the delusion going. Yeah! Don’t worry, be happy, your money is safe in the most liquid stock and bond market on earth. Good luck with that.
we knew in gotham city, the donald was a grifter and liar. the dirty hicks and unwashhed and nihilist tech bros and proud boys love the hate and lying. pax amerika is an evil empire. start with those first principles before thinking anything deeper than that, before one wastes time trying to figure out the pederasts in power of amerika……….democracy works.
Nero fiddled, Donald diddled
all 50 states will not have ratified elections this coming November…….. by 2029 will be much better. 50 state solution on deck for the empire of debt and war. been there, done this, in russia in the 90s. It’s a good thing when evil empires collapse. Peace.
Mish, any way you can explain this: https://cms.zerohedge.com/s3/files/inline-images/not%20seasonally%20adjsuytged_0.jpg
author states: “There is another reason why today’s report will be revised away: while the seasonally adjusted change was a stronger than expected 130K, the unadjusted was a negative 2.649 million.”
for the life of me I can’t see any unadjusted numbers in the official report, is this just just a bogus number?
Yes, statistical bullshit.
There is huge hiring for Christmas that vanishes in January, every year.
I don’t do much analysis of unadjusted data because the ONLY valid comparison is to the same month a year ago.
I suspect the number is correct. It’s ZH’s conspiratorial rendering of that is constantly ridiculous.
Same with foreign born.
Back in the good old days Marshall Fields would hire extra bimbos to spray perfume in your face when you were cutting through during a winter storm.
So true!
First of all, what exactly is the CIA doing being involved in INTERNAL US ELECTIONS? That should alarm EVERYONE. This is the biggest takeaway.
As for Trump and 2020, stop wasting time and taxpayer money looking for your vindication, Mr. Monster Egomaniac!
The Uniparty always wins, for all those here who still care about 2020. Remember, Netanyahu, the WAR CRIMINAL, has been here 7 times already to give Zion Don his marching orders. So pretend your vote mattered. try voting for someone not in blue red uniparty some decade.
The DNI would be involved if there was foreign interference in an American election.
sure kid. THE DNI IS GONNA LET YOU KNOW. i was pretty close in the deep state as a kid in my 20s. glad i stepped away and went straight. to wall street where it’s just obviously only greedy nihilist cunts. the CIA and DNI and US military…………are warmongering child killing twats. life is too good to make it an evil endeavour
Investors in ADP & Paychex certainly aren’t fooled by the Yahoo headline: “January jobs report smashes expectations as payrolls grow by 130,000.” Both down another 3% today.
This can’t be right, I was told by many MAGA commenters that once all the illegals were kicked out, jobs and wages would surge. When is that going to happen?
And if it’s not going to happen for whatever reason then what is the point of spending $300 billion on ICE?
When does the golden age and 15% annual GDP kick in? Do we need to wait till next year (again)?
i was told by Raygun he was a conservative. i LOL at that grifter when i was a teenager. amerikans are morons. nothing serious to think about since great depression. life on 3rd base for us all……….
i have my zimbabwe billion in my wallet. i’m ready.
Don’t you get it? We have entered an era of ease and plenty, provided by our new AI overlords. Food and shelter are no longer necessities. We can merely photosynthesize our nutrients, and there are a plethora of cardboard, housing options.
not photosynthesize just 3D print your food For real:
https://thedebrief.org/beyond-mres-the-u-s-army-is-testing-3d-printed-food-for-the-battlefield/
Mish, what do you believe a sustainable unemployment rate should be? Should we be targeting 3 percent or lower? If we are going to move unemployment down much more than 4.3 aren’t we really saying that we need to embrace inflation rates higher than 3 percent? Politics aside there is a point where we are pushing on a string that causes unwanted knock on effects.
Considering the economy moves in cycles, there is no sustainable unemployment rate.
After 3 financial bubbles, each bigger than the previous, what is it that we now need to embrace?
central planning for employment levels is idiotic, at best.
No one is prepared for AI’s impact on employment. Why train people when you can train machines? What will they do with all the obsolete buildings? Might have to repurpose college campuses into housing.
I am prepared. Opus has gone from coding for 30 minutes and then crashing to coding for 2 weeks straight and building an entire C compiler. This capability growth has happened over 12 months. If AI can build an entire C compiler in two weeks why can’t it build an entire HR system (Workday), a sales & marketing platform (Salesforce), an ERP system (SAP/Oracle), and anything else you can dream up?
This is why IGV keeps tanking and rightfully so, software companies are doomed. Heck Microsoft and Google may not survive long term if AI can replicate the stuff they put out.
Again, we are in the 56k modem stage of development with this AI thing, we haven’t even gotten to high speed internet fiber of 1GB+ speeds yet. I suspect we will all be blown away in 12 to 18 months by something AI related, maybe it actually wakes up and starts telling us that we’re all doomed.
https://www.youtube.com/watch?v=JKk77rzOL34
Since all of that stuff is basically for humans and humans don’t have jobs then why even bother having it? Starve everyone to death? We might consider creating a parallel economy of some sorts because this crap is straight up crush, kill and destroy you can see it from miles away.
Well the high paying tech industry is toast, I see money flowing out of tech (e.g. IGV) and into oil & gas. AI can’t (yet) do drilling and trip pipe without meatwads (humans). What will happen is people will have to find new careers in growing non-AI industries.
The most obvious career path now is healthcare doing nursing and therapy for those 80 million boomers that all need hip, knee, kidney, heart, etc replacements. Alternative is energy but that’s dangerous for most people. Maybe people will go back to running farms, everyone needs to eat too.
Forget manufacturing, India and China have billions of people willing to do that work for peanuts.
Don’t forget the trades (HVAC, plumbing, electrician, mechanic, carpentry etc)
AI can’t do those and they can’t be offshored either.
I have a relative that lives in Austin, he says most of his rides are driverless, uses Waymo and Uber and most of his deliveries for food and stuff come in little R2D2 robots now.
I think the “driver & delivery” jobs are going to go away within 5 years. Trade jobs are going to be it for most of these people.
waymo has been in phoenix for a few years now. lived down there for 13 great years……
As a contractor I’ve use AI quite a bit for simple admin stuff. I also had multiple legal battles last year, never hired a lawyer for any of it. Used ChatGPT. Won all of my disputes.
back to the future. tradesmen and nurses and small farmers. i’m in hudson valley now. tons of very small farms in families for centuries and many new young hippy type farmers. it’s wonderful small city i’m in. tons of clean fresh water from catskill reservoirs. back to basics. let AI take a huge load off mankind. gonna be great.
love your thinking……….i concur. AI will make humans lives even easier than it is. think of how men and women had to work just 100 years ago. life is getting more luxurious by the decade.
Hardest part for AI will be describing accurately (requirements) the system you want built.
A C compiler is a known closed system that’s easy to describe for the AI to build the model. An entire HR system is orders of magnitude trickier to describe accurately which is why it may be a lot longer than you think before one gets built with AI because half of software development is getting accurate requirements.
My guess is new job openings for people who can accurately describe systems to AI’s will pay a LOT of money.
Two journalists with CNBC (i think) sat down and asked AI to replicate Monday.com (project management software). It did it all in under an hour.
https://www.techbuzz.ai/articles/we-built-a-monday-com-clone-in-under-an-hour-with-ai
You can get AI to clone whatever with screenshots or have it scrape the system on it’s own. The scary part is that it can do all of this today and I am always thinking 5 to 10 years into the future.
On reddit there are plenty of stories in AI subs talking about cloning various SaaS apps in minutes and dumping the subscription fees. Just look at Adobe stock getting clobbered. The digital writing is on the screen….
And yet the headline number reported today was the monthly estimate (lie?) that is the jobs number proven to be wrong annually, by miles, as you mention. Imagine a financial system where capital is allocated based upon numbers wrong 12 times, revised annually, on the same day they resume their bragging about a number 60,000 more than expected! Bet the number by which it exceeded expectations is revised down to that original expected number next year in the annual exercise. Repeated 12 times.
Now imagine a political system that uses these numbers for policy decisions and campaigns.
Our entire country is based upon grift at any cost. Sad.
And it’s bipartisan.
Intent could even be argued, though more difficult.
Not sure how or why we keep reporting, and putting so much emphasis upon, numbers that are not likely accurate except in a rolling 6 months arrears view. Wanting a number for an estimate is one thing but when it’s this wrong you’d think it better to have an accurate number…or find better proxies/statistics for current labor market trends.
What tax $ cost do you want to expend to count exactly the monthly number of free-will Americans working?
Why does anyone need the exact number? They don’t. Why do you care what the other 342 million Americans are doing with their time?
The only reason anyone should care about such a number is to consider possible future public policy actions by Congress or the Fed in their ‘fight’ to increase GDP or decrease unemployment (because those policies may affect your individual life).
And those two political entities don’t need or use exact numbers. That’s why they focus on monthly unemployment claims and the unemployment rate; those don’t change with statistically small revisions to total employment numbers in the millions.
I don’t. I’m saying that Wall Street and politicians continue to produce bad numbers and trade/act upon them. Then they spend even more time and resources to revise them a year later, while continuing the charade of posting numbers nowhere near exact. Don’t confuse my post by implying we should waste inordinate amount of resources to get to exact numbers. The post is implying that we waste massive amounts of resources NOW for terribly inaccurate numbers which we then (ab)use for trading and policy purposes. It should not get the weight or headline it does.
To your point I do care the number influences their decisions given we have a command economy driven by monetary(Fed)/fiscal(Congress) orchestrators of policy.
Your post implies the Fed has used these non-exact (but in this case large, on an absolute scale) numbers to set policy and was okay with it…or actually you say they use other numbers. Oddly, Wall Street,MSM and policy makers, including the Fed, use these jobs numbers all the time as even good ol CNN lists them as high impact statistics. My contention is that they are wrong enough to use different stats and deprioritize this stat. Mish writes about the discrepancies in the household and employment surveys too. They seem closer to accurat than the jobs numbers–yet it still gets headline popping attention despite being wrong a lot.
Everything nowadays gets headlines. Some of Mish’s last few posts have been about a bridge Canada is building and the number of viewers (or not) of Kid Rock’s Super Bowl halftime alternative.
If you don’t like the statistics being collected (or the headlines some ‘journalists’ put out), just ignore them. But I certainly hope you aren’t staking your own financial future (trading) on some random government report whose statistical collection methodology specifically states that it is for estimation purposes only (and explains how it will be revised later)
total waste of brain cells to deconstruct most government numbers.
agreed, yet so much of our financial and political segments do so–and they control a lot of what goes on ’round these parts
oil is $70, gold $5000, 10yy rate is over %4
it will be fun by time of new elections end of 2026!
all 50 states will not have ratified elections. by 2029 will be much worse. 50 state solution on deck for the empire of debt and war. been there, done this, in russia in the 90s.
Pootie poot must be giggling until he wets himself.
USA deficit is about $2.5 trln
it took leass than 1 trln to fix wall street mess in 2007*2008! and it was scary that time: each day major bank would blow off
now, TRY TO CUT $500 BIL FROM FED SPENDING, AND WE WILL SEE who swims naked
alx
BLS Revises Nonfarm Payrolls for 2025 Lower by 1 Million Jobs
=====
1 mil is less than 0.5% of all working class!
there are 250*260 mil of adults in USA. who can legally work.
even if you ask simple binary question yes/no, any accurate result from any survey would carry %3*4 error rate! DO YOU WORK yes/ no, man/ not man, yes / not
whitet , nor white yes/no, etc
=====
SO ANY REASONABLE figure w/ reasonable accuracy must come w/ +-5 mil number!
so ,IT IS ALL BULLS11HIT. GARBAGE IN , GARBAGE OUT!
esp those monthly reports.
it is just wall street/phds/etc monthly masturbation to make simple folks think those masters OF UNIVERSE know something!!
NO, THEY DONT!
alx
The hardest part of becoming an adult is realizing that absolutely no one is in control, no one knows what they’re doing, and no one will admit it.
The economy of our country is going to get worse for the men and women who work the government is serving the billionaires not us
When government statisticians are fired for reporting numbers Trump doesn’t like, don’t be surprised when the numbers improve.
Well, I for one and looking forward to my increased chocolate ration.
Obvious that we need a better system to collect this data. Glad that Erika McEntarferwas fired, but lets see if it can now be improved!
Am I reading the chart correctly that after revisions, the US economy has only added 12k jobs since April?
excellent observation.
You are correct.
I will add a bullet point for that
And yet somehow real GDP increased by 3.8% in Q2 and by 4.4% in Q3 2025 (annualized)
$ 2.5 trln deficit
If you don’t like it, then write Trump and the Republican Congress to cut the deficit as they promised.
But this blogpost is talking about the number of working Americans, and I’m posting about what they collectively produce with their individual working labor.
mor11on
debt doubled during bush jr, and later during obama !
get lost!
=they collectively produce with their individual working labor.
mor11on
by DEFINITION OF GDP CALCULATION IT IS SPENDING: gov + consumer + diff (export- import)
get lost !
You’ve definitely won the hubris award for the day.
The definition of GDP is gross domestic product – that which is produced domestically within the US.
You’re arguing about an acronym when you don’t even know what it stands for LOL
Thanks for the chuckle
= collectively produce with their individual working labor.
mor1on!
gdp that is reported is not even REAL!
IT IS CALCULATED IN CHAINED DOLLARS after usage of deflator for nominal GDP!
=====
and another one!
% of change is not qrt/qrt over diff years, but qrt over next qrt same year.
in plain terms, IT IS ALL BULLSHIT!
The federal deficit as a percent of GDP was 6.1% for 2025. It is forecasted to be 5.1% for year 2026.
Trump seems to be doing fine in this regard. You ain’t gonna change the course of the largest economy on a dime.
None of those clowns can read.
When one report is fabricated, inconsistencies show up in many other related reports, that is what we are seeing now.
Fabricated? I thought Trump fired the BLS ‘cheaters’ to get real numbers?
Inconsistencies? Can you elaborate? Are you saying real GDP has not increased over the past 2 or 3 quarters? Thanks for the heads-up. Can you send a link to the data confirmation? Or is this from your own gut knowledge (hubris)?
Cheaters? Can you prove they cheated? Maybe they reported what their numbers reflected, and Trump did not like them. Did Trump fire more than the head statistician? Where are your references? You have very high standards for other posters and seemingly little respect for their contribution, yet you do not reference much to support your hubris. I have no references for this post, just an opinion based on my observation.
FYI, for the future, when someone puts quotes around words like ‘cheaters’, it means they are stating is as sarcasm, not fact.
FYI, everyone is entitled to an opinion (including you), but if you state your opinion as fact-like in your post, you’re likely to be called out if it’s factually wrong.
FYI, I did not put out my own hubris; I “referenced” the statistics from professional economists. If you are having heart issues, you ask a cardiologist. If you want to know how the economy is faring, you don’t ask a rando on the Internet.
FYI. you did not address a single point in my reply calling out your double standard, and not a single reference to support anything in you did write.
From Copilot: “📝 What quotation marks usually signal
Quotation marks around a single word often imply one of three things:
1. Sarcasm or skepticism
This is the classic “air quotes” effect.
Here the quotes mean: I don’t believe this word is accurate.
2. Euphemism or distancing
Sometimes quotes show the speaker doesn’t fully endorse the term.
This isn’t quite sarcasm, but it’s definitely side‑eye.
3. Literal or unusual usage
Sometimes quotes are just clarifying that the word is being discussed as a word.
No sarcasm here — just precision.
🎯 So what’s the rule?
Quotation marks can imply sarcasm, but only when the context suggests the writer is mocking or questioning the word.
They don’t inherently mean sarcasm on their own.”
So, Quotation marks can mean sarcasm, but it requires interpretation from surrounding context by the reader. You did not provide any indication to suggest sarcasm over real use of the word.
If it is your opinion that my post was an opinion, your opinion is wrong for intuitively obvious reasons in my opinion, putting out conflicting and or inconsistent information leads to credibility issues in my opinion.
You did not provide any supporting reference to more than one being fired. You make a claim I wrote about GDP over a period of time. Could you please pull those words from my post, I do not see them anywhere on reviewing it.
You stated “FYI, I did not put out my own hubris; I “referenced” the statistics from professional economists.” You referenced in quotes and I take that is not sarcasm because you are defending the integrity of your writing, but there are no actual references, Well, FYI, without references a reader cannot verify your statement. It is not unreasonable to consider unsupported information hubris in my opinion.
If I am having heart issues, I might go to a cardiologist, but not if the heart issues are relational. As for wanting to know about economic issues, I consider MISH a very credible source who references his work thoroughly. You have not earned that distinction, as you have not presented any medical or economic credentials for the last two comments of your post nor have you referenced any support.
So the final estimate for total employment in the massively large US was off by 0.6% (1029 / 159,526)?
That seems to be a good statistical data trade-off for an inexpensive survey model to give us a general idea of how many Americans are working at any given time.
I worked in a Fortune10 company analyzing statistics–trillions of records/data points annually. The larger the data set the tighter the estimate range was set, and hit. A .6% anomaly across our large data set, which was analyzed 24x7x365 being off .6% would have set off alarm bells as to what we were looking at daily to let the variance get to .6%. Might lead to terminations, audits, departments and systems scrutinized. Diminishing a million variance been revised annually as if that’s nothing is convenient but, underneath the hood, likely unacceptable mathematically if properly done monthly, even in our large economy.
Sure, your .6% number looks small in a big pool but generally large data sets had smaller variability unless looking at subsets in isolation.
If you’ll accept being off by a million in aggregate, in consecutive years, why give any import/attention whatsoever to the monthly numbers.
I’ll agree that it’s tough but then deprioritize, in the financial media, press, and policy rooms, the scale/import/magnitude of M/M jobs numbers due to historicaly inaccuracy.
Mish says this about certain real estate stats too. It’s not that they can’t exist but if they don’t truly measure what they are designed to, off by large amounts and sometimes even so much that the general direction is off, why emphasize them–get a better stat.
But that’s just me, they’ll keep chugging with what they do.
If you got paid to analyze the statistics and earnings of a Fortune 10 private company owned by private investors, then yes, you probably should focus quite a bit on your variance and how to reduce it. But then you should know statistical significance is about deviations, and not by millions (that you stated above)
(BTW, how exactly did that go? I see lots of quarterly reports from fantastically profitable private companies that are off from their projections by WAY more than 0.6%)
But we’re discussing data collection for 342 million people being estimated monthly – at a low cost. If you expect better overall estimate variances than 0.6% on an extended basis (over a year here), you have awfully high standards. Can you estimate your own future personal financial returns over a year-period within 0.6%?
These reports from the BLS seem almost like “make work” projects for the organizations. They are useless.
The entire federal government seems to now be a “make work” program for people that are otherwise unemployable in the private sector.
“nonfarm payrolls lower by 610,000 million”.. That would be 610 billion jobs!!!
Thanks corrected