Rendering reliance on it for indications of employment trends useless.
By Wolf Richter for WOLF STREET.
The ADP National Employment Report data, released today by payroll processor ADP, was massively revised going back to 2010, now showing erratic and massive month-to-month job gains and losses going back to 2010 that are completely different than the data for the same period released a month ago, rendering reliance on the report for indications of employment trends useless.
The chart shows ADP’s new revised data of the month-to-month changes in employment (red columns), and the prior data released on January 7, 2026 (blue columns).
There is hardly any correlation between the red columns and the blue columns. And these erratic differences between new data and the prior data go all the way back to 2010, when ADP began this data series.
For example:
In 2025, the new version (red) shows job declines in March, April, and May, when the old version showed substantial job gains (blue).
Then for the second half of 2025, the new version (red) shows much bigger job gains of 345,000 for June through December, than the old version (131,000).
For 2024, the new version shows big job losses in February and March (red), while the old version showed moderate gains (blue).
And then again in September and October 2024, the new version showed job losses (red), when the old version showed massive job gains (blue).
For 2023, the new version shows huge job gains for May, June, and July, while the old version showed much smaller job gains.
These massive differences go back all the way back to 2010.
The entire data set was massively revised.
ADP’s entire data series going back to 2010 was heavily revised, and shifted down by about 2.5 million jobs across the entire period.
In the press release, ADP said about the revisions:
“The January 2026 report reflects a scheduled annual revision of the ADP National Employment Report. The data series has been reweighted to match the Quarterly Census of Employment and Wages (QCEW) benchmark data through March 2025.
“Beginning this month, in addition to the annual benchmark revision, the ADP National Employment Report also will reflect data from the most recent QCEW release.”
The QCEW is produced by the Bureau of Labor Statistics and is based on companies’ quarterly employment tax filings. The BLS adjusts its own nonfarm payrolls data to the QCEW on an annual basis.
ADP gets its data from the payrolls it processes. But many companies use other payroll processors, and many companies process their own payrolls, and ADP has no data on them, including some very large employers. ADP then extrapolates its limited and possibly skewed sample to the national population. It then periodically adjusts this data to the government’s data, the QCEW by the BLS.
The QCEW is the gold standard for US payroll data, but lags far behind because it’s based on quarterly tax filings with delayed filing deadlines; and it is not complete either.
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the mug to find out how:
Run two MACDs that are Fib related harmonics (pick your periods) and when they diverge, look for a turn. Works for Cumulative Tick Indicator and the Adv/Decl and Up/Dn Vol ratios as well.
Huh. I knew other companies were competing with ADP but I’m surprised they did this. Thanks for posting it.
I have never found the ADP Report to be useful, because it never seemed close to accurate, and totally ignored it. And now ADP itself has confirmed my view. The changes made may cause it to be (close to) accurate on a going-forward basis, but it will take years to see if that is the case. So, for the next few years, at least, I’ll continue to ignore it.
At this stage in my life when I see major revisions of the past (such as this one which seems outlandish without a detailed explanation) I ask myself “Are the revisions to increase factual recounting of the past or is this to rewrite history for the sake of the present?”
This is a good reminder of why investors and business people need a well-staffed, not-politically-influenced set of national statistics agencies.
The private sector data is guesswork at best.
This is what is what we’d be left to rely on when we can’t trust the BLS or Census or BEA. Frankly, at that point, I’d just diversify into various developing nations.
As they say: “Never trade on ADP”
Those August – November 2024 revisions are pretty striking… not that we had anything going on during that period 🤔