We have had negative revisions 11 straight months, but look closer.
Job Revisions – Six-Month and 12-Month Rolling Sum
The BLS generally tends to revise jobs higher, except near recessions.
The Great Recession, followed by the Covid recession stand out to the downside.
The housing bubble prelude to the Great Recession and the rebound from the Covid recession stand out to the upside.
The moral of this story is the BLS is very poor at economic turns.
BLS Nonfarm Payroll Revisions
Recent BLS Nonfarm Payroll Revisions
- Revisions negative for 11 consecutive months. December is uncertain.
- Revisions negative 29 out of the last 35 months dating to January 2023.
- A rolling sum starting July of 2024 briefly sent the 6-month rolling sum positive.
Two Repeated Stories, Both Wrong
These revisions are not political, nor are they a sign of incompetence, as frequently charged. Rather they are a sign of poor, antiquated data collection methods.
I’ve shared this story before, but it’s worth repeating.
In conversations with the BLS, I asked why they use the Birth-Death model (a major source of these errors), why they don’t they weed out duplicate Social Security numbers, and why they don’t use Treasury data.
The response was “We would like to but we don’t have access to the data.” I asked why not, and heard “We are not allowed to look at Social Security number or access Treasury data for security reasons.”
Bear in mind the simple solution is to encrypt the sensitive data and use sort-merge methods to produce reports, but we are dealing with the Government (and antiquated COBOL systems nobody seems to know how to fix).
Instead, the BLS concocted a Birth-Death model that is very wrong at turns. Those model errors were compounded by Covid, declining response rates, and a huge wave of immigration.
Benchmark Revisions
In addition to first, second, and third revisions to the monthly data, the BLS also has annual revisions.
The last set of revisions to Household employment data was a doozy. And it was very positive.
The benchmark revisions show above were to the Household Survey employment data. It represents undercounting of immigration, major sampling errors, etc.
The BLS revised the employment level from 161,586,000 to 163,831,000 in one fell swoop invalidating all historical comparisons unless one uses what the BLS labels “experimental data”.
This is ridiculous because the published key data is guaranteed nonsense. There was no magic increase of 2.25 million jobs in January of 2025.
Yet, if you believe Trump, that is what happened, moreover with all of it in US-born employment. The numbers are preposterous, but they keep making the rounds.
All posts on foreign-born employment suffer this flaw. There are no back adjustments.
More Benchmark Revisions Coming Up
Next month, more benchmark revisions are on coming up. Expect them to be negative.
On September 9, 2025 I commented New QCEW Data Indicates More Big Negative Revisions Coming to Job Reports
The discrepancy between jobs reports and quarterly data widens again.
QCEW data is highly accurate, but very lagging. It’s why I expect negative benchmark revisions next month.
Understanding the Enormous BLS Job Report Errors
For a full explanation of the above chart, please see my December 18, 2025 post Understanding the Enormous BLS Job Report Errors, What Really Happened?
Four things: Immigration, Birth-Death Model, Response Rates, Sampling.
Incredibly Bad Standard Numbers
Many people are reporting unbelievable year-over-year numbers, then attributing the improvement to Trump.
For example, on the Benchmark adjusted chart, Employment in November 2025 was 163,741,000 and November of 2024 was 161,661,000. That’s supposedly a gain of 2.08 million.
But that entire gain occurred in January of 2025 when the BLS threw the whole adjustment into a single month.
Foreign-Born Employment Nonsense
The numbers in my charts are seasonally adjusted. Foreign born employment is not adjusted, compounding comparison errors.
And we have no BLS revised data for foreign born employment. So, all such foreign and US-born comparisons with BLS data remain garbage.
A second major problem with foreign-born employment is the BLS makes no distinction between US citizens who were foreign born and genuine foreign workers.
How Fast is the US Shedding Foreign-Worker Jobs?
On December 11, 2025 I asked How Fast is the US Shedding Foreign-Worker Jobs?
Think carefully.
Foreign-born includes US citizens.
There is one more big issue. Since there is widespread disbelief in BLS numbers, why are we supposed to believe this set of BLS numbers?
Are Illegal immigrants suddenly responding to BLS household surveys?
Related Posts
December 27, 2025: Only a Third of CEOs Plan to Hire Workers in 2026
66% will fire workers or play wait and see with AI.
January 7, 2026: ADP: White Collar Jobs Clobbered in 6 of the Last 7 Months
Professional and Business Services jobs are down five months and seven of last eight.
January 9, 2026: Nonfarm Payrolls Rise by 50,000 with 76,000 in Negative Revisions
2025 closed out with a thud. Here are the details.
Assuming there are no more negative revisions (and what is the likelihood of that), the year-over-year employment gain was just 610,000.
I expect the next benchmark revision to wipe that out. Thud!
The above post explains in much more detail the “experimental” benchmark revision series.
January 9, 2026: AI Is Killing Select White Collar Jobs. What’s Hot and What’s Not?
Private education and health accounted for over 100 percent of job gains in 2025.
BLS and ADP are in agreement over white collar jobs.
When the only job strength is in recession-proof health care, generally the economy is already in recession.
I don’t understand why Social Security data was given to DOGE but not to the BLS. I would think that it would be the other way around.