Two BLS reports, one monthly, the other quarterly highlights a 3.4 million overstatement in the monthly jobs reports. 
Understanding the Chart
- Nonfarm Payrolls are are from the Establishment Survey (CES). The sample survey was 666,000 individual worksites in 2023.
- QCEW (Quarterly Census of Employment and Wages) is a count of Unemployment Insurance (UI) administrative records submitted by 11.9 million establishments.
- QCEW is quarterly, within five months after the end of each quarter. The CES survey is monthly. For the comparison, I used end-of-quarter totals for CES.
- SA means seasonally adjusted. NSA means not seasonally adjusted.
The above comparison details are from the BLS publication Quarterly Census of Employment and Wages Concepts.
QCEW is a survey of 11.9 million establishments vs 666,000 for CES.
QCEW shows strong seasonal tendencies. Thus, it’s a mistake to compare seasonally adjusted CES (light blue line in the above chart) to QCEW.
It’s clear the monthly nonfarm payrolls are overstated every month. It looks random if you make the mistake of comparing SA data to NSA data. Here’s the correct application.
Nonfarm Payrolls NSA Minus QCEW
QCEW data is not in a friendly format to download. I manually plugged in the above quarterly numbers from a QCEW download into a better formatted CES download from the St. Louis Fed.
In normal times, neither heading into or out of recessions, the differences seem to vary randomly in a tight range.
At economic turns, the variance is much wider with a caveat that I only went back to 2019.
Understanding the Discrepancies
If the difference was nearly constant, say ~2.5 million, it would not matter. But if sharply increasing or decreasing numbers happen near recessions, the discrepancies do matter.
Comparing NSA numbers, Nonfarm payrolls are 3.42 million higher than QCEW numbers. That is the largest difference since 3.46 million at the height of the Covid pandemic.
The only other 3+ million difference was in the third quarter of 2020.
One more comparison will put things into proper perspective.
2022 Q4-2023 Q4 CES vs QCEW
- CES: 155,211,000 to 158,269,000 (+3.06 million)
- QCEW: 152,525,000 to 154,848,000 (+2.32 million)
CES reports 32 percent more job gains in 2023 vs QCEW!
Jobs Report on Friday
The monthly jobs report is Friday. The Bloomberg consensus estimate is 188,000 jobs.
I’ll take the under.
Regardless, whatever number the BLS reports is highly likely to be revised lower.
What’s Weakening?
Weakening Hard Data
May 24: Another Massive Revision, This Time Durable Goods, What’s Going On
May 23: New Home Sales Sink 4.7 Percent on Top of Huge Negative Revisions
May 22: Discretionary Spending Tumbles at Target, Shares Drop 10 Percent
May 22: Existing-Home Sales Decline 1.9 Percent, Sales Mostly Stagnant for 17 Months
April 15: Elon Musk Fires 10 Percent of Tesla Workforce, Prepares for “Next Phase of Growth”
A Second-Quarter Recession This Year Looks Increasingly Likely
Yesterday, I commented A Second-Quarter Recession This Year Looks Increasingly Likely
As I watch the evolution of consumer spending, housing starts, new home sales, and GDPNow trends, it appears the economy has peaked.
This is also why I expect a rate cut in July.
For discussion, please see I expect a Fed Rate Cut in July Despite Market View of 18.5 Percent Chance
It irritates me when people make stupid objections that I explained in my post.
Hint: Lawrence Bird
What about this statement do you not understand?
“If the difference was nearly constant, say ~2.5 million, it would not matter. But if sharply increasing or decreasing numbers happen near recessions, the discrepancies do matter.”
Leaving aside your unfortunate adjective (“stupid”), before I would see this as a recession indicator I’d need a longer look-back to prior recessions. I put an asterisk on a lot of covid-period data because that recession was unique. The economic data really aren’t designed for a black swan like that. They did chronicle it, of course, but I don’t give the data from that period much weight in trend analysis.
You may well turn out to be correct that the divergence you cite precedes a recession, but you did not go back far enough to persuade me of this particular data’s predictive value. I do think the economy is weakening, and have been saying so lately, but other data are much more valuable to me in developing that view.
Further, you offer no reason why the wider recent discrepancy that you cite would signal a recession. You have offered a (purported) correlation without a stated cause or even a relationship. Mind you, I LOVE numbers. I swim in a sea of numbers. But those who love numbers know their limitations, one being that a mere comparison between them is only a starting point.
Your post was interesting, but quite far from definitive or even especially valuable. Not stupid at all, but a swing and a miss.
How much credence are we expected to put into a random number generator operated by government bureaucrats too stupid to get real jobs?
Instead of obsessing over BLS minutia, just be honest…. the official statistics are gathered by morons and signed off on by people way past senile.
You’re talking to the wrong audience. You need to explain this to the folks who manage trillions in investments all over the world and make decisions based on these stats.
NVDA ==> wall street bs. Next week DIA might rise, before possibly dropping 45/55 pts.
The government needs to stop reporting on economic data monthly. Instead, they should computer a 3 month moving average, which I contend would be more reflective of what is really going on in the economy.
SpaceX’s Starship just landed softly in the ocean off Northwest Australia. It’s a milestone.
Has a spaceship not landed before on Earth ?
Only other SpaceX rockets. The Starship is much bigger and can put 150 tons into orbit.This one landed under its own power.
Data! Data! Data! Job posting from Indeed & St. Louis Fed.
https://fred.stlouisfed.org/series/IHLIDXUS
I’d like to see some diversification of these types of reports that source data from more than just the BLS. Whether the BLS is flawed or politicized can be mitigated to an extent by blending different data views. One source of data is bound to be wrong and serves as a single point of failure in any analysis.
Multi-point data analysis is what I do for my investment portfolio and things are slowing a bit but I don’t think it will hold but I’ll let you know after the May social security snapshot pops out.
The report comes from BLS, which does not interpret the broader meaning. They should use their own data and no one else’s. It is up to the users to interpret its meaning, and in doing that they should incorporate other data.
U3 is calculated from the household survey, correct? And the number of jobs in the monthly report come from the establishment survey, making the monthlies a blend of household and establishment?
“What exactly is slowing?”
The number of restaurants in California. Southern California based Rubio’s is closing 48 restaurants.
Prices are too high here. I use to eat here when lunch was closer to $8. Now it is nearly double that. And they don’t allow you to bring outside drinks to manage their inflation. Most of these restaurants would be better off filing for bankruptcy and letting commercial real estate go belly up. Then commercial real estate prices would truly crash and anyone wanting it could structure a business accordingly. The banks continue to sit on a bunch of bad assets and it is costing the economy. The banks would like everyone to blame the cost of doing business when it comes to employees but they never point the finger at themselves.
Yeah. I like how, “inflation is up 20% since 2020”, yet somehow all the prices have doubled.
Can probably be traced back to the rent they’re paying.
Um, idiot, the inflation rate is a blend of prices, weighted in a market basket of goods and services. Are you really this stupid?
Mish, is it possible(per ZH article today) Employment Authorization Document (EAD) is the answer?
I did a similar post a while back
It has to do with “foreign born” which I believe ZH stretched to illegal.
https://mishtalk.com/economics/us-born-employment-is-down-by-over-1-million-since-2020/
thank you, I will check it out, surprised i missed it.
Politics=Lying.
yep, career bureaucrats are cooking the books to help the Biden campaign
Hey, a guy’s got to get elected. What’s a few million here or there every third quarter of a presidential election cycle. Give a guy a break.
I like to use the word, “Break” as in the context of a JAW Fracture.
ZH has a good article on just this topic. Gov’t stats are the worst made sausage imaginable. And the input is weak from their sources, no one really wants to answer the surveys.
ZH Title not quite accurate IMO
https://mishtalk.com/economics/us-born-employment-is-down-by-over-1-million-since-2020/
Why don’t you believe the BLS? You are foolish enough to believe Chuck Schumer.
This is a nothing burger. The fact that the difference is stable for the five years shown (and probably longer) means it is irrelevant, it is the equivalent of picking a zero point for measuring potential energy.
You should also know better than to do charts showing an absolute difference of an item that is trending, you should be showing %s
If you breakdown the jobs numbers they’re overstated bigly, and then if you dig even further most of them are government jobs or menial ones that Biden’s illegals are doing. This economy is plain garbage due to high inflationary money printing.
They are not stable if you bother
CES reports 32 percent more job gains in 2023 vs QCEW!
What about this statement do you not understand?
“If the difference was nearly constant, say ~2.5 million, it would not matter. But if sharply increasing or decreasing numbers happen near recessions, the discrepancies do matter.”
Can you read?!
Job numbers are goosed by illegals and those who take care of illegals. It’s our new bestest economy ever