Blame hurricanes if you like, but the impact is debatable. Only Hurricane Milton was in the reference period.
Hurricane Milton and Helene
The BLS Employment Report offers this Hurricane Explanation for the grim report.
Hurricane Helene made landfall on Florida’s Gulf Coast on September 26, 2024, and then tracked north into several other states. This was before the October reference periods for both the household and establishment surveys. Hurricane Milton struck Florida on October 9, 2024, during the reference periods for both surveys. Prior to the storm’s landfall, there were large-scale evacuations of Florida residents.
In October, the household survey was conducted largely according to standard procedures, and response rates were within normal ranges. The initial establishment survey collection rate for October was well below average. However, collection rates were similar in storm-affected areas and unaffected areas.
A larger influence on the October collection rate for establishment data was the timing and length of the collection period. This period, which can range from 10 to 16 days, lasted 10 days in October and was completed several days before the end of the month. No changes were made to either the establishment or household survey estimation procedures for the October data.
It is likely that payroll employment estimates in some industries were affected by the hurricanes; however, it is not possible to quantify the net effect on the over-the-month change in national employment, hours, or earnings estimates because the establishment survey is not designed to isolate effects from extreme weather events.
There was no discernible effect on the national unemployment rate from the household survey.
Jobs vs Employment
From September 2020 through early 2022, nonfarm payroll job gains and full time employment changes tracked together.
Starting around March of 2022, a divergence between employment and jobs became very noticeable, and I have been discussing the divergence since then.
Job Stats vs One Year Ago
- Nonfarm Payrolls: +2,173,000
- Employment: +216,000
- Full Time Employment: -1,000,600
Job Stats vs Two Years Ago
- Nonfarm Payrolls: +5,108,000
- Employment: +2,767,000
- Full Time Employment: +1,282,000
In the last year, nonfarm payrolls are up by 2.4 million while employment is only up by 216,000 and full-time employment is down by over one million.
Job Report Details
- Nonfarm Payroll: +12,000 to 159,105,000 – Establishment Survey
- Civilian Non-institutional Population: +209,000 to 269,289,000
- Civilian Labor Force: -220,000 to 168,479,000 – Household Survey
- Participation Rate: -0.1 to 62.6% – Household Survey
- Employment: -368,000 to 161,496,000 – Household Survey
- Unemployment: +150,000 to 6,984,000- Household Survey
- Baseline Unemployment Rate: +0.0 to 4.1% – Household Survey
- Not in Labor Force: 428,000 to 100,809 – Household Survey
- U-6 unemployment: +0.0 to 7.7% – Household Survey
Nonfarm Payroll Change by Sector
Key Points
- Government and Health Services are related to the surge of illegal immigrants and the need to address them.
- Professional and Business Services Employment in temporary help services declined by 49,000 in October. Since reaching a peak in March 2022, temporary help services has lost 577,000 jobs.
- Manufacturing Employment in manufacturing decreased by 46,000 in October, reflecting a decline of 44,000 in transportation equipment manufacturing that was largely due to strike activity.
Monthly Change in Nonfarm Payrolls
Monthly Revisions
- The change in total nonfarm payroll employment for August was revised down by 81,000, from +159,000 to +78,000.
- The change for September was revised down by 31,000, from +254,000 to +223,000.
- With these revisions, employment in August and September combined is 112,000 lower than previously reported.
Nearly every month there are negative revisions. These are huge revisions and hurricanes did not have an impact.
Part-Time Jobs
- Involuntary Part-Time Work: -67,000 to 4,557,000
- Voluntary Part-Time Work: -214,000 to 22,352,000
- Total Full-Time Work: -164,000 to 133,496,000
- Total Part-Time Work: -227,000 to 27,934,000
- Multiple Job Holders: -354,000 to 8,305,000
The above numbers never total correctly due to the way the BLS makes seasonal adjustments. I list them as reported.
It’s important to not that multiple job holders add to nonfarm payrolls but not the number of employed.
Hours and Wages
This data is frequently revised.
- Average weekly hours of all private employees was flat at 34.3 hours.
- Average weekly hours of all private service-providing employees was flat at 33.2 hours.
- Average weekly hours of manufacturers fell 0.1 hour to 39.9 hours.
An overall decline or rise of a tenth of an hour does not sound line much, but with employment over 160 million, it’s more significant than it appears at first glance.
Hourly Earnings
This data is also frequently revised. Here are the numbers as reported this month.
Average Hourly Earnings of All Nonfarm Workers rose $0.13 to $35.46. A year ago the average wage was $34.10. That’s a gain of 4.0%.
Average hourly earnings of Production and Nonsupervisory Workers rose $0.12 to $30.48. A year ago the average wage was $29.29. That’s a gain of 4.1%.
Year-over-year wages are keeping up with year-over-year inflation after underperforming for many months.
Unemployment Rate
The unemployment rate hit a 50-year low in January and April of 2023 at 3.4 percent. It’s now 4.1 percent.
“The unemployment rate has bottomed this cycle and will generally head higher.”
I first made that comment many months ago. If there was any doubt, it’s now erased.
Alternative Measures of Unemployment
Table A-15 is where one can find a better approximation of what the unemployment rate really is.
- The official unemployment rate is 4.1 percent.
- U-6 is much higher at 7.7 percent.
Both numbers would be way higher still, were it not for millions dropping out of the labor force over the past few years.
Some of those dropping out of the labor force retired because they wanted to retire. Some dropped out over Covid fears and never returned. Still others took advantage of a strong stock market and retired early.
The rest is disability fraud, forced retirement (need for Social Security income), and discouraged workers.
Birth Death Model
Starting January 2014, I dropped the Birth/Death Model charts from this report.
The birth-death model pertains to the birth and death of corporations not individuals except by implication.
For those who follow the numbers, I retain this caution: Do not subtract the reported Birth-Death number from the reported headline number. That approach is statistically invalid.
Birth-Death Methodology Explained
I gave a detailed explanation of the model and why the hype is wrong in my June 8, 2024 post How Much Did the BLS Birth-Death Adjustment Pad the May Jobs Report?
I repeat, do not subtract the birth-death number from the headline number. That’s flawed.
However, it is now clear that the BLS is too optimistic about the number of jobs they believe are being created by the net creation of new businesses.
A Breakdown, by Sector, of the Negative 818,000 BLS Job Revisions
On August 22, 2024 I gave A Breakdown, by Sector, of the Negative 818,000 BLS Job Revisions
Those negative revisions are a direct result of the BLS Birth-Death model gone haywire.
Household Survey vs. Payroll Survey
- The payroll survey (sometimes called the establishment survey) is the headline jobs number. It is based on employer reporting.
- The household survey is a phone survey conducted by the BLS. It measures employment, unemployment and other factors.
If you work one hour, you are employed. If you don’t have a job and fail to look for one, you are not considered unemployed, rather, you drop out of the labor force.
Looking for job openings on Jooble or Monster or in the want ads does not count as “looking for a job”. You need an actual interview or send out a resume.
These distortions artificially lower the unemployment rate, artificially boost full-time employment, and artificially increase the payroll jobs report every month.
Final Thoughts
This was a dismal report.
Some of it can be blamed on the Boeing strike and the hurricanes. The negative 81,000 revision to August can’t be blamed on hurricanes.
The household survey which sets the unemployment rate did not seem to be impacted by the hurricanes.
I remain highly skeptical of the glowing report last month (some of which has been revised away), and BLS monthly reports in general.
The reports smack of oversampling large employers and undersampling small employers where jobs have been trending lower.
Full time employment is down by more than a million from a year ago. Don’t pin that on hurricanes and do expect more negative revisions.
Addendum
Employment fell by 368,000.
The unemployment rate was steady only because 220,000 people dropped out of the labor force.
To three decimal places I calculate the unemployment rate at 4.145 up from 4.051 up .094.
This is best viewed as an increase of 0.1 percentage points instead of unchanged.
The unemployment rate is actually 4.3% if you include the decline in employment in the number of unemployed. The drop out of laid off workers from the labor force has led to an underestimate of the unemployment rate.
Now you have two votes, thanks to Kate.
Pure opinion and a hunch, if October non farm payroll was this bad and both revisions wiped out 112K jobs, the dirty laundry will be aired pre inaugural for Harris or after Trumps is sworn in so either administration gets a “clean” slate to implement their respective jobs programs and economic growth programs.
Thanks for the detailed UE rate calculation! Up 0.1% is correct.
The Household Survey data once again support your the “emerging recession” thesis.
Of those 12 job categories, probably the best indicator of an impending recession is construction. It was positive by 8K and would have come closer to meeting the 12-month average without Milton’s interference.
With Q3 GDP of 2.8 and core PCE inflation accelerating for the 3rd month in a row, it’s not likely that a recession will arrive before 2025.
Of all the data, the single greatest leading indicator for recessions is the total employment level from the Household Survey, backed by other Household Survey data.
The establishment survey has limited predictive value, empirically proven.
The construction employment survey data in particular are not consistent with other data on the construction industry.
GDP is also not a leading indicator for recessions, in fact the initial GDP numbers were still coming out “positive” some 6-9 months into the 2007-2009 recession. Only later was the recession recognized in GDP revisions.
That is the initial reading jobs report which is always wrong compared to the final report:
https://media.scored.co/post/1L8GEdkrKpdh.jpeg
The initial market boost scam
It looks like the bond market sees a good possibility of a 25 bps cut when the Fed meets next week. IMHO they are beginning to lose control of everything on the curve except short-term rates. Bond vigilantes coming back to life after a multi-decade slumber? Could get ugly fast.
5 more months I’ll be joining the ‘unemployed’ after 30 years of federal service. My job is not going to be replaced. Technically I’m ‘voluntarily working part time’ since I’m only dong about an hour of work a day as I wind down.
ADPs model has October 2024 private payrolls surging and up at +233000.
I have not had a chance to review that yet but a quick check shows the expected.
Small employment only +4,000.
An excuse for the fed to cut another 0.25% next week.
How convenient.
Yeah, +12,000 is not a good monthly number; let’s hope (for all potential employees out there) it was due a lot to Boeing and the hurricane and returns to a bigger positive soon
But Mish, I have read a lot of other commentary out there about the difference between the results of the establishment vs. household surveys over the past two years, you have referenced now and before. Obviously either both are wrong or one is massively. And what I’ve read elsewhere says the establishment is probably more right because that polls businesses actually paying people. And the household survey asks a sample of households what they are doing. And this last one is more likely to undercount due to an immigrant surge for obvious reasons.
You reference that a lot of the actual hiring is due to needs to accommodate the immigrant surge. But you don’t (to my immediate remembrance) discuss much where these immigrants actually work themselves. If a lot or most of these immigrants are now working somewhere, wouldn’t the establishment survey show that likelihood better than the household survey?
I feel like this is an important issue to research and discuss, like the (now non-existent) difference between GDP and GDI
Undocumented immigrants would not be working at establishment-type jobs in general; they are more likely working for themselves or off the books for a small business which doesn’t meticulously follow the tax code.
Now that I think some more of this, I could argue the other way. It would be possible to get an establishment job with a fake social security number.
Probably a LOT harder than you think to get a fake SS number.
For people with ‘connections’ or people with computers/internet/black site knowledge it might not be too hard. But for a bunch of illegals crossing with only the clothes on their back and speaking only Spanish I suspect it’s VERY hard to do.
Hence they are working off the books.
Lots of businesses don’t care. Take a SS#: 123-45-6789; hire the person as a contractor. They sign the W9 under penalty of perjury.
I do the hiring; no withholding on job-shoppers; I send a 1099 to keep my books straight and take my writeoff.
That’s why I asked Mish; I don’t think any of us have a definitive answer. But the sources I’ve seen think these immigrants are being employed by all kinds of ‘establishments’: hotels, large construction companies, corporate farms, delivery services, meatpacking plants, etc. and not just mom-n-pop cash places. These establishments write payroll checks (without E-Verify) and it’s up to the worker to bear the brunt of cashing them, and getting SS benefits back (or not)
Hector Mexicano isn’t writing any code.
Si!
I’m on strike at Boeing, about a third have gotten full and part-time employment during the strike, another about 50% have savings to go 3 or more months, a small number live at home and are fine, and the last 10-15 are struggling on the outside or have crossed the line (2-3% estimate), so the Boeing strike effect on employment is fairly small in reality.
80 years of data says that the Household Survey correctly spots the emergence of recessions and the Establishment Survey does not.
That’s helpful to know. Can you forward the studies that prove this correlation/causation? I’m always up for being educated
And did those previous 80 years of data have such a distinct divergence between the results of the Establishment and Household surveys that Mish has pointed out for the last 2 years?
This is so ridiculously easy, you don’t even need to read studies, you can graph the data and see for yourself.
Go to http://FRED.StLouisFed.org
Type in “Employment Level” in the search bar.
Click the first link.
You now see the total employment level from the Household Survey going back to 1948.
Use the slider to vary the time window in the graph.
Note how the employment level levels off just before every recession, and then dips.
Note how it never does that at any other time.
Note that it has already leveled off, as Mish has pointed out repeatedly.
As a bonus, click on “Edit Graph”, change units to “Percent Change from Year Ago”, and see how the “leveling off and dipping” show up even more clearly.
Now, go back to the Search Bar, and type in Total Nonfarm. This gets you the Establishment Survey headline number. Note how the Total Nonfarm employlment data also levels off and then dips, but does so AFTER the recessions have started. (This is even more obvious if you do the “Percent Change from Year Ago” step.)
Conclusion: The Household Survey “Employment Level” data is a leading indicator of recession with no false positives and no false negatives. The Establishment Survey “Total Nonfarm Employment” data is a lagging recession indicator – it eventually confirms the Household Survey, but with a delay; you don’t want to use this to anticipate recessions because you’ll be late most of the time.
You don’t need someone else to tell you how to interpret this data. You don’t need to parse the debates of 1000 biased economists. You just need to take a few moments to look at the data yourself, see the obvious pattern, and recognize that This Time Is Never Different.
And yes, Mish’s graphing the same data (and more – the FRED database contains dozens of treasures like this one, if you explore).
forgot to mention: the Establishment Survey in FRED has been revised, you are seeing the eventual final values. The real-time values are even less helpful.
The Household Survey data doesn’t get revised AFAIK, because it’s methodology doesn’t require gathering more data..
I appreciate the long and involved response; I really do. And I use FRED data a lot myself
But you’re talking about eyeballing data to predict recessions. That employment level data in FRED you’re looking at is currently around 161 million people. And that is being estimated from the Household Survey, which surveys only about 60,000 households a month. That’s a lot of extrapolation.
The BLS itself (that actually conducts the two surveys) says the Household Survey is to be used primarily for the labor force status and unemployment rate because its sampling error is much larger than the Establishment survey targeting monthly about 120,000 businesses and agencies and 630,000 different worksites. The Establishment survey is used by the actual surveyors to estimate better employment, hours and earnings by businesses, not individual people answering the survey (although it has sampling error as well)
https://www.bls.gov/web/empsit/ces_cps_trends.htm
So yes, I ask about who is creating the data (and why) and who is analyzing it because I think that’s more important – and telling – than a simple eyeballed line. But everyone has their own opinion and that’s what makes a market.
BTW, this same website link – from the surveyor themselves – shows the difference for the last 30 years in the two survey results. The adjusted household survey results match almost exactly the payroll survey results, despite your FRED data claim. EXCEPT for recently (and in the late 90s) when the household survey was more negative than the payroll survey. That’s why I asked my question about the impacts of the immigration surge. The data IS different for the last 1-2 years, despite your Never Different claim.
The recession may be here already. Or maybe the household numbers are off (instead of the payroll numbers) just like GDI was revised upwards to match GDP recently. None of us will know for sure until later.
Thanks again for your response
Wisdom Seeker led you to the water, but whether you drink or not is up to you.
Mainstream economists routinely fail to anticipate recessions. Logically therefore, if you are interested in being able to predict recessions, you cannot rely on mainstream thinking.
As I see it, what you referred to as “eyeballing data” was someone who doesn’t know you explaining in non-mathematical language the same thing as a typical quantitative recession criterion:
“Stalls out and then dips”, if you think about it, shows up in year-over-year data as a line that falls to zero (no change in 1 year = stalling out) and then drops below (negative change year-over-year = dip).
So his criteria, applied to the graph of year-over-year percent changes in total employment from the household survey, translates to “whenever the year over year change drops to zero for any length of time, recession follows”.
I empirically back-tested this in 5 minutes by checking the graph using FRED and looking at each recession in turn.
Ah well, perhaps a hallmark of a major bubble is that so few people are willing to look skeptically at it. A bubble can only run to extremes when everyone chooses shelter themselves within the manic herd, rather than step away and ask “what would the end look like?”.
Thanks for your comment, too.
But I conduct quantitative analyses for work. So I stick with the characterization of the ‘eyeball’ technique described above. Both of you reference year-over-year changes. Of course, if we wait a year after the fact and can ‘see’ – years later – that when that YOY jobs number goes negative, we’re in a recession (that’s a primary reason the NBER calls one). That is not super helpful for predicting future recessions.
Of course, we can all see the economy is slowing down. Hell, the Fed raised interest rates a lot for two years for the explicit purpose of slowing the economy. But it is not “ridiculously easy” to know (for certain) if we are in a recession. I think that does a (probably unintended) disservice to Booth, Hunt, Mish and others that predict recessions and then end up being wrong at times – like in the past two years.
Maybe we should all clarify why we’re trying to ‘call’ a recession in the first place. Is it because we want to be more right than others on our dating it, or so we can move our asset portfolios accordingly. If it’s as easy as ‘eyeballing’ a graph, or looking at one-year-old data, do you think we’re currently in a recession? Have you moved most of your holdings to cash since corporate profits (and likely stock prices) will correspondingly drop if the recession is confirmed and serious? I’m glad I did not do that over the past two years with all the recession calls. But I’m thinking about it now (thus my questions), but primarily because short-term interest rates are still so high I’m more comfortable with that guaranteed return.
You may do quant analysis for a living, but so do I, and I’ve gotta say you are being willfully stupid here.
The year-over-year change in the Employment Level from the Househokd Survey goes negative in the realtime data BEFORE the recession.
That’s why it’s a leading indicator!
You can see it Right Now. You just have to look.
No need to wait a year!
P.S. The Establishment Survey data does not do that, there’s a few months of delay before declining employment hits the establishments. That’s why it’s a lagging indicator. Employment trends are led by smaller businesses not captured in the Establishment Survey’s initial reports, in large part because the birth-death model isn’t correct real-time.
The establishment survey is proven bullsheet and the QCEW reports prove it.
QCEW is hardest of hard data a 95 percent sample. Revisions -818,000.
BEDS a 90% subset has q1 net employment +493,000 for 2024 Q1 – Released yesterday. I had not reported this yet.
BLS Nonfarm payrolls are supposedly up 915,000.
Uh…. No
The cycle effect is at work.
Professional Business Services was not only down big last month but was revised down for September from +17K to down -9K! What a mess.
Goldilocks. No one needs to work. MMT and AI. And for the rest of us, word salad for breakfast, lunch and dinner.
What’s heavier than lead? Because when this thing goes down, its going to be worse than a lead balloon.
Its a disaster indeed.
Rates up anyway as our deficit and inflation problems are never ending
Gold is nearly two times heavier than lead, but will rise to surface ‘when this thing goes down.’ Goldilocks? Better believe it!
How the U.S. military lost a $250 million war game in minutesHow the U.S. military lost a $250 million war game in minutes – The Washington Post (archive.ph)
The lesson in the Millennium Challenge, 2002, missed by most, is not the susceptibility of the US to low-tech warfare. It is the ‘inability to think outside the box.’
The approach goes back to Sun Tzu, An advantage is obtained by doing what the enemy least expects. In the MC 2002, the enemy used commercial ships to launch missiles, etc. The US Navy was surprised and ‘lost’ most of a carrier task force, including the carrier.
We need to face facts: the US Military was short on critical and creative thinking. The situation is likely far worse at present, given the focus on DEI. The US Military is in dire need of leaders like Gen. Paul Van Riper, the enemy leader in MC 2002. Instead, it has Austin and friends.
Private sector in tatters. Maybe we can all get government jobs….So much damage done these past 4 years. Will take decades to fix. A strong bench for the Republicans though. Thinking Vance gets two terms after Trump followed by Ivanka for two and then Barron should be ready to take over.
The Republican presidential candidate’s views do not reflect any unifying principle other than self-interest.
You find Kamala’s rhetoric unifying? Have you found the last 4 years unifying?
hardly LMAO
“Hatred is the most accessible and comprehensive of all the unifying agents. Mass movements can rise and spread without belief in a god, but never without a belief in a devil.”
― Eric Hoffer, The True Believer: Thoughts on the Nature of Mass Movements
Kamala’s rhetoric very unifying at moving undecided voters to Trump.
I fear you are greatly overrating the average voter. I doubt they know the meaning of ‘rhetoric’ no matter how often they watch The View.
Kamala could be a cow pie with googley eyes glued on it, and still be a better president than trump.
The trouble with being brainwashed by the mass media is you fail to see it. This is true of both sides, btw.
Self-interest is infinitely better than cackling and mindless drivel.
Horrifyingly bad numbers. Terrible revisions.
Cry me a river on Wednesday.