Let’s recap negative job revisions and also discuss a reader comment “This was not a poor nonfarm payroll report.” 
One of my readers commented “This was not a poor jobs report.”
I wonder what planet such true believers are on.
Job Report Stats
- Private Payrolls: Only 74,000 (and highly likely to be revised lower)
- Manufacturing Payrolls: -24,000
- Revisions: -86,000
- Involuntary Part-Time Work: +246,000
- Total Full-Time Work: -438,000
- Total Part-Time Work: +527,000
- Multiple Job Holders: +65,000
Of the 74,000 jobs added, 65,000 of them were people taking a second job!
To the above data, my dear reader responded “I would hardly characterize this as a very poor report. It was an OK report, certainly not indicative of the economy currently being in a recession.”
The comment displays a massive amount of economic cluelessness about jobs and lagging indicators.
Great Recession Job Details
- The most recent revision in December of 2007, one month before recession, the economy added 105,000 jobs.
- The most recent revision for January 2008, with the economy in recession, shows the economy added 1,000 jobs.
Dot-Com Bust Recession
- In March of 2001, the month preceding recession, jobs were +145,000 at the second revision,
- Eventually, the BLS revised that to +51,000.
August 1991 Recession
- In July of 1981, one month ahead of recession, jobs were a whopping +383,000! The first revision was to +351,000. The second revision was to +265,000,
- In August of 1981 (the recession had already started), the initial report was +61,000. The first revision was to +49,000. The second revision was to +21,000.
- In September of 1981 (recession was two months old), the initial report was -54,000 but the BLS revised that up to +47,000 on the first revision, then an amazing +132,000 on the second revision.
August 1991 Recession Revisions
To recap, three months into the 1991 recession (the December report for November) the BLS was reporting +132,000 payrolls.
Only later (I believe a benchmark revision over a year later), did the BLS start reporting negative numbers.
Even then, the most recent revision for the month ahead of the recession was still +111,000.
Well not to worry, QTPie tells me “This certainly not indicative of the economy currently being in a recession.”
Things Consistent With Recession
- Private Payrolls: Only 74,000 (and highly likely to be revised lower)
- Manufacturing Payrolls: -24,000
- Revisions: -86,000
- Involuntary Part-Time Work: +246,000
- Total Full-Time Work: -438,000
- Total Part-Time Work: +527,000
- Multiple Job Holders: +65,000
Hmm. Somehow that list looks familiar.
Things Also Consistent With Recession
- September 3: Construction Spending Growth Slows in May, Stops in June, Negative in July
- September 5: Fed Beige Book Shows Flat or Declining Economy in 9 of 12 Fed Districts
- September 6: Payroll Report: Manufacturing Sheds 24,000 Jobs, Government Adds 24,000, Big Negative Revisions
100% Consistent With Recession
For those who may have missed it (or ignored it on the basis of an OK jobs report) please consider a Key Recession Indicator Gives Stronger Recession Signal in August
A modified McKelvey recession indicator with no false positives or false negatives since 1953 suggests we are in recession now.
Not to worry, jobs are “OK” because the BLS is very believable. If you can’t trust the BLS monthly job reports, what can you trust?
Oh, there’s just one more small thing consistent with recession, Fed rate cuts. When is it that the Fed starts starts cutting rates?
Additional metrics consistent with recession:
(1) Household Survey series “Employed, Usually Work Full Time” (FRED series LNS12500000), year-over-year change, negative indicates recession. Series has been negative each month for past 7 months. No false positives in series (available since 1970).
(2) “Part Time For Economic Reasons” (FRED series LNS12032197) now shows spike atop a gradual uptrend – same as all prior recessions in series.
(3) Frequency of negative revisions to Establishment Survey data. Mish has covered this. I would emphasize that a major cause of negative revisions is when the advance estimates’ Birth-Death model (which is not real-time) misses turning points. Then more accurate later data prove the B-D to have been in error, and revisions are issued.
(4) Establishment Survey, “All Employees, Total Private” (FRED series USPRIV), percent change from year ago. Series signals imminent or ongoing recession whenever reported change drops to 1.5% or less. This series goes back over 70 years and has been right 12 of 13 times. No misses, but there was one “false alarm” in 1952 about a year before the actual recession of 1953-1954.
All of these are again signaling recession now.
Thanks
I will look into that as a second indicator
Yes Mister Mish, its an “A Multiple Jobs (i.e., Walmart shopping cart attendant + Door Dasher + Amazon MTurker) Economy”
And the media like NY Times is touting how healthcare (Medicaid, Medicare, Veteran Affairs Health Admin) is the one sector that is defying the slowdown in hiring
Too bad they don’t also include in this news reporting that Biden’s Medicaid budgets are at least $100 billion more than the Pentagon and national security budgets ! ! !
Unlike Birdbrain Newsom or Walz Nut, Ron DeSantis has been tapping the brakes on Medicaid spending as far as making sure it is not providing coverage for households that at least make 3 times the poverty level and still qualify for subsidy with the Silver Plan on the Affordable Care Act exchange.
and that does not include the illegals cost in various states. That bill has yet to have been paid.
I estimate New York to be asking for $20 billion post election, mid 2025.( what did Biden give last time? I think 12Billion
Q2 GDP revised up to 3% just days ago.
Q3 GDP tracking about 2% with only three weeks left in the quarter.
The inertia of inflation & the economy are certainly slowing, but $2T in deficit spending & 12.5M extra mouths to feed are very buoyant for the economy.
How can anyone be surprised that the NBER hasn’t declared a recession . . . YET.
“Higher” education assigns degrees by government DEI quotas. They sell acceptance letters, and they sell grades. Corporations cannot depend upon the face value of a diploma. Human resources departments compensate by offering short term trial employment called internships. The managers assess competence and hire the capable. Months ago I was told that internships are hard to get in the hot computer programming arena. Times have been tough for a good while. Here’s the good news. Alyssa Heinerscheid left a plum job for someone who did their homework.
What you write makes sense to me, Mish, that more possible negative revisions may be on the way for these payroll numbers. But what your analysis really highlights – at least for me – is that it is possible – if not likely according to your historical numbers – that monthly payroll numbers will show positive gains even at the beginning of a recession. So shouldn’t we try to add to the recession-call analysis what most likely actually led the NBER to call recessions in the past, and so the future? Real GDP changes
For 1981 recession rGDP percentage change (from previous quarter) – according to FRED data:
Q2 1981: -2.9%
For 1991 recession:
Q3 1990: 0.3%
Q4 1990: -3.6%
For 2001 recession:
Q1 2001: -1.3%
For 2008 recession:
Q1 2008: -1.7%
Most recent (for a recession?):
Q4 2023: 3.4%
Q1 2024: 1.4%
Q2 2024: 3.0%
Q3 2024 (1 month in): 2.1% estimate (according to GDPNow forecast)
So the NBER can use whatever multitude of data it chooses to call a recession. But if it chooses April or May of 2024 for the start, it would be the first time in 40+ years that its recession call did not start or closely correlate with a negative quarterly real GDP growth rate
Sorry, Mish, but you’re missing the point I was making in those comments. I agreed with you that it is possible for there to be decent job growth immediately before, and for a very short while into a recession. The point I was making was that that decent job growth into a recession lasts a very short time period, while you have been saying we’re in a recession for a while now. This post does not address that point. Instead, it reinforces a different point on which we do not disagree.
Revisions matter…
Lots of “astute” comments here except no one is talking about how many more people left the labor force and enrolled in social security.
The latest social security snapshot.
https://www.ssa.gov/policy/docs/quickfacts/stat_snapshot/2024-07.html
Total recipients: 72,554m
Over 65 added from last month: +141,000
Total cost: $121.4 billion per month or $1.4 trillion per year. Those numbers will only grow bigger and faster.
I think this has a large part in why the recession just doesn’t come, for all the job losses, there are simply more openings as more and more people retire.
Will one overtake the other? If so when? That’s the trillion dollar question.
I suspect there is a large component of Fed believers–the ‘soft landing’ approach might be self-fulfilling–creating an extended period of low to zero real growth. In fact, it was one of the possible scenarios in 2007-2009. The result–the Fed had to stimulate until 2015 or so–the Wall Street bailout.
Meanwhile, the switch from full-time to part-time jobs is indicative of recession.
I like it that you are finally looking backwards to other recessions. A little more info would be helpful, but it’s a welcome start.
Yeah, watchdog reporter Burnie Thompson says context is everything and background always matters
An interesting insight on forecasting validity: https://dailyreckoning.com/when-the-models-fail/
Extrapolation of past trends and implied relationships into the future is necessarily fraught with uncertainty, and a set of assumptions that model users might not understand, or appreciate.For example, the reaction time of manufacturers to changes in consumer spending.
Perhaps the biggest problem with models is when we assume causality, and make decisions accordingly. The classic example, man-caused global warming–which is not to say it doesn’t exist, but causality in endemic given its starting point. ‘You find what you are looking for.’
Not too far different; casual correlations, when data sets align and there is no structural relationship. Equally, building models where time leads and lags play a crucial role.
Regarding employment and recession… commonsense should be sufficient to expect a decrease in full-time employees and an increase in part-time employees. Seeing it reflected in the data does not PROVE a recession is underway; but sure as Shinola, someone is polishing given the revisions. Politically driven, or simple model inadequacy?
At major economic turns, few realize the turn is occurring. Clues are all around us in real time. We don’t need to wait for delayed information from the government. Look at how many older people have gray hair roots showing because they stopped coloring their hair. People wear clothes that have stains or holes in them. Look for smaller crowds at coffee shops. Look for lighter commuter traffic and cars with broken mufflers. Look for fewer orders at work. Fewer people will landscape their yards, or more people may let the grass grow taller between cuttings.
Yes, good signs to look for. But many of those things happen after the turn, and take time to show up. They should all become much more visible in months ahead than they are right now.
At the turn, you’re still up at the “top”, like driving a car over a hill. The main difference isn’t how high up you are, it’s just that you’re now heading down and need to use the brake pedal, instead of heading up with your foot on the gas.
“At major economic turns, few realize the turn is occurring.”
Correct – Look at Fed Chair Ben Bernanke’s denial three months into a recession!
The number 1 sign: 18-wheeler traffic on the major interstates. You ain’t shippin’ ’cause no one is buyin.’
When Bush the Elder was running for reelection against Clinton in 1992, Clinton ran on it being a lousy economy. 4th Qtr GDP growth ended up being >4%. 2024 seems like 1992 in reverse. 4th Qtr will eventually prove to be worse than “expected”. I think Mish doesn’t believe in conspiracies (much), but I would guess than 80% or more of BLS staff are democrats, and I doubt they can refrain from letting their desired outcome influence their reports. So, they slant things to help the democrat.
Bush’s administration did preside over a weak economy; that hadn’t fully resolved by the 1992 election. Anyway, Clinton didn’t win entirely because of the economy, there was also Ross Perot splitting the conservative vote.
Very true regarding Clinton-Bush. It was a turning point in media bias. Most of what Clinton achieved with the economy was the result of Newt Gingrich, yet few people know. Whitewater should’ve excluded him from office. Again, yet few people understood its significance–maybe it was too complicated for the masses. Clinton’s womanizing was also well known, including the coverup of what happened at Oxford.
As for the bias in Federal government employment, I’d suggest your 80% is low.
Footnote: Even now, Wikipedia does not mention Newt Gingrich in the Bill Clinton article. IMHO, Clinton signed Gingrich’s legislation to avoid impeachment, now referred to as ‘secret pact.’.
Most of the economic boom had nothing to do with anyone in power at any level (congress/senate/president).
It was 100% driven by the rise of the internet and computing efficiency increasing worker production. In 92 when Clinton won, pretty much the only people with PC’s on their desks were programmers and probably half of those had no connection to any other PC even in their own office! By 2000 when he left it was almost impossible to find any office worker anywhere without a PC on their desk along with full internet access.
I am sure those in the trenches are not doing anything. Someone would rat.
I had a discussion on this yesterday who believes the manipulation is in the fundamental model, set incorrectly (namely the birth/death model).
That’s more plausible in that it involves less collusion.
But the B/D model was wildly low coming out of covid.
My conclusion remains: Unlikely to be purposely biased, especially when QCEW will point that out!
We’ve seen (up-close) the impact of DEI in the US Secret Service. We KNOW the US Military is the same. We KNOW the DOJ is openly biased. Why would the rest of the US government be immune?
“We’re just super incompetent” is the cover story for the Trump assassination plot. Limited hang out.
“Someone would rat”. 1) I doubt it. People who work at BLS want to keep their job. 2) who would they rat to? The media who would ignore it or the government who would also ignore it? The entire government media complex is on the side of one party.
Exactly, keep your head down and nose pointed to the keyboard and crank out those deliverables, audit reports, etc
Don’t even give the perception that you could be a “potential rat”.
And if you want to score points, have a photo of the Obama family on your desk in your cubicle. And make sure you are seen at the monthly American Federation of Guvmint Employees (AFGE) meetings.
Mister Mish obviously has never worked in the federal bureaucracy, LMAO.
Maybe you have never worked in the federal bureaucracy, Mister Mish.
Not sure how you would fare working at least 5 years (to be vested in the Federal Employee Retirement System) such as your mental sanity in the Beltway cubicle farms at BLS or BEA, or the Census Bureau.
Yeah go file a complaint with the Inspector General and then wonder why they put you in the complaint hamster wheel while it seems like all of a sudden you are suffering subtle PPP (prohibited personnel practices).
All of a sudden you realize you are now a subject of an investigation for mishandling sensitive data, or something else.
Then go to the Washington Post or NY Times and see what they say about your complaint about how you believe they are misreporting economic statistics to help the Democrat Party during an election yar.
Failure to fix an obviously wrong model that is catastrophically misleading precisely when it is most needed … is exactly the sort of “not our fault” error that a bureaucracy will generate on its own.
The manipulation isn’t overt, it’s not in the B/D model, but in the failure to implement a much better system. Given that trillions of dollars are at stake in any recession, and the technology exists to track business startups and closures in near-real-time, there is absolutely no excuse for the U.S. not to have a better system.
P.S. Ditto for tracking employment – should be done via IRS tax withholding data, not antiquated surveys.
Exactly as the IRS has the data right there as far as federal tax withholdings
they could also go to Social Security and ask about Social Security tax receipts
and as far as gig jobs, they could survey or query the major gig employers like Door Dash, Amazon MTurk, etc.
“Lies, damned lies, and BLS statistics.”
https://www.barrons.com/articles/labor-markets-jobs-unemployment-economic-growth-productivity-ad86e019?st=5evb5hsluctdr9p&reflink=mobilewebshare_permalink
An article in Barrons indicates there are 1.2 job openings for each worker. This is the same in about 30 countries and will be along term issue. I live in the econically sensitive metro detroit area and no indication of a slowdown. There are layoff in the auto industry but subdued.
Early days yet. The early responders will be survivors. The risk, of course, being the recession ‘evaporates’ with more stimulus, and auto sales take off.
However, the thing about Detroit: autos drive the economy, with a multiplier to total employment (likely around 8:1)
Quite possible, things happen slowly and then suddenly.
What we don’t know though is how many of those job openings are:
1) Generic openings that more or less always remain posted but are rarely ever filled
2) Posted positions paying far too little to attract any serious candidates
Irregardless this ratio is much higher than It has been historically.
It could be that more companies are posting phantom or ruse job openings just to give perception that they are growing and also to conduct market research as far as the quality of the job candidates (compared to their current employee slaveforce).
This practice may not have been as prevalent in the past when it was harder to post job openings (i.e., no widespread use of internet).