This is how the internet works. Someone comes up with a meme, and it takes on a life of its own. I’m going to have some fun with it.
By Wolf Richter for WOLF STREET.
Back in mid-December, a hullaballoo erupted on the internet and the social media. The headlines and YouTube videos and Tweets were everywhere, and the pivot-mongers latched onto it and bandied it about with utmost energy. It was just what they’d been waiting for. Here are some of the headlines:
- NY Post: “Biden administration overstated Q2 job growth by 1 million: Philadelphia Fed”
- Zero Hedge: “Here Comes The Job Shock: Philadelphia Fed Admits US Jobs ‘Overstated’ By At Least 1.1 Million.”
- CNBC: “Philadelphia Fed suggests BLS overstated job growth in second-quarter by a million jobs”
- Reuters: “Missing jobs mystery puts Fed on back foot.”
- Mish Talk: “The Philadelphia Fed Just Revised Jobs Lower by 1.2 Million for Q2. Hello ‘Jobs are too strong for recession’ advocates. We have major revisions and more are likely coming.”
This “BLS” is of course the Bureau of Labor Statistics that releases the monthly employment report, of which we got the latest installment on February 3 – along with said revisions.
The whole thing arose on December 12 when the Philadelphia Fed’s new experimental fancy-schmancy algo predicted that the BLS had overreported jobs growth by 1.1 million and would therefore make a downward revision of 1.1 million jobs on February 3.
Oh boy! On February 3, the BLS made instead an upward revision of 813,000 jobs to nonfarm employment for the period through December 2022 – which blew the Philadelphia Fed’s experimental fancy-schmancy algo out of the water.
On top of this upward revision of 813,000 jobs of the past data through December, the BLS reported nonfarm payrolls jumped by 517,000 jobs in January (all data here is seasonally adjusted). These employment numbers make sense because the labor market has been historically tight, triggering all kinds of secondary effects, including the biggest wage gains in four decades and a spike in the employment of people with disabilities.
But the news sites and financial blogs and YouTubers all had made a lot of hay while they could back in December. I got links to their stuff in my inbox for days, and the WOLF STREET comments were bombarded with it, as it usually happens when something as titillating as this shows up in the headlines.
At first, I politely put some perspective to these comments, but it kept going for two weeks, and I finally took out my old double-barrel goose gun and shot these comments down, as you can see. Here are some samples in chronological order, from polite to double-barrel goose gun (as you can see, if you don’t read our comments, you’re missing out!):
Commenter, Dec 16: “Data will clearly show that the Fed has been hiking (way too late as always) right into the teeth of a recession, which began already in Q1 of 2022. The Philly feds jobs data already shows this. Versus the amply stupid establishment survey. Off by more than 1 million jobs for Q1. Q2 will be worse. Q3 will show we have been clueless. But we won’t see this for months in the revised data.”
To which I replied: “These benchmark revisions take place every year. The BLS adjusts its figures every year with these benchmark revisions, sometimes up sometimes down. The Philadelphia Fed attempts to estimate those revisions on a quarterly basis. They’re an estimate of future benchmark revisions. I posted the Philly Fed text in a comment under the prior article. Read it…”
Commenter, Dec 17: “Steven VanMetre just released on YouTube showing the BLS overstating employment figures in 2022 by at least 1,000,000? The Fed relies on the BLS and believes we have a strong economy when the opposite is true….”
To which I replied: “What you’re referring to in terms of the BLS: The BLS does an annual benchmark revision, and it does it once a year every year. The Philadelphia Fed, using the same data that the BLS uses for its benchmark revisions, tries to estimate on a quarterly basis what that annual revision from the BLS might look like. And that is now suddenly gospel? The internet is full of clickbait garbage about that.”
Commenter, Dec 29: “Any comment on the Philadelphia Fed pointing out that the BLS got the Q2 payroll count wrong by *1 million* – essentially erasing *all* jobs allegedly added during that period.”
To which I replied: “I’ve commented on it a gazillion times already. And I shot it down. You just didn’t read it. Go find my comments. I’m tired of wasting my time on it. This has already been debunked, including by the BLS itself. This was an experimental new algo that the Philadelphia Fed came up with to predict quarterly what the annual BLS adjustment will be. And it was BS. But the pivot crowd sure jumped on it and made a fool of itself once again.”
So here is what happened:
On December 12, the Philadelphia Fed released a report about its new experimental algo that uses the same data that the BLS uses for its benchmark revisions. The algo attempts to estimate what the annual benchmark revisions by the BLS will be, when the BLS reports them on February 3, 2023.
The Philadelphia Fed masterpiece said that only 10,500 jobs were created in the March through June period, instead of the 1.12 million jobs reported by the BLS, and that the BLS had therefore overstated the job creation by 1.11 million jobs during this period, and that the BLS’s annual benchmark revisions would therefore be a negative 1.11 million.
This algo has not been tested, and there was no way to say how accurate its prediction of the BLS benchmark revisions would be. But for anyone looking seriously at the labor market, the assertion by the algo that only 10,500 jobs were created in the March through June period was astounding. Someone must have smoked something very special when they designed the algo.
On Friday, February 3, the BLS released the actual benchmark revisions to nonfarm employment of the Establishment Survey in its monthly jobs report. And it went into the opposite direction
For the period through December 2022, the BLS revised nonfarm employment upward by 813,000 jobs. This is on top of the jobs created per the unrevised data. On a revised basis, the total number of jobs created in 2022 rose to 4.81 million.
For anyone watching this historically tight labor market – and all the side effects of it – the upward revision wasn’t much of a surprise.
The revisions go back to 2021. This chart shows the revised data for nonfarm payroll jobs (red) released on Friday, and the unrevised old data (green), released a month ago. Both are through December:
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the beer and iced-tea mug to find out how:
Would you like to be notified via email when WOLF STREET publishes a new article? Sign up here.