No Recession in Sight
Here's a Tweet that caught my eye today.
This post on Change in Nonfarm Payrolls and Average Payrolls Change in U.S. Economic Cycle represents the entirety of the analysis.
These two lines are the entirety of the article (emphasis mine).
Nonfarm payrolls have averaged 184k the past 12 months, more than the cycle average of 170k. This does not happen in front of recessions.
Image: LPL Financial LLC
Where's the Proof? Heck, where is the supporting evidence?
How about some recession bars?
How about any semblance of analysis to back up the claim?
It's not even clear what the author means by "cycle average".
All Employees vs Employment Level
Leading and Lagging Indicators
All Employees is from the establishment survey.
The Employment Level is from the household survey.
OK November bounced. Why?
The GM strike ended and there were additional seasonal factors as discussed in Jobs Surge in Strike-Ending and Seasonal Adjustment Rebound.
I dove further into the details with Trump Tweets "Manufacturing Blowout": What's the Real Story?
I presented six charts. Here the key one.
Manufacturing Average Weekly Hours Year-Over-Year Detail
The GM strike ended on October 25. And Thanksgiving was November 28, the last possible date, further skewing BLS job sampling.
Yet, there was no improvement in manufacturing hours of actual production workers!
2015 vs Today
The dip in 2015 was energy related. This dip isn't.
This manufacturing dip is related to a global slowdown in trade enhanced by Trump tariffs.
What About Hiring Trends?
I am glad you asked.
Please consider Hiring Trends in Recessions by Size of Firms
Slump in ISM and Freight Traffic
Recessions are difficult to predict. I certainly provide ample proof.
But at least people ought to discuss the actual data instead of making baseless claims.
I originally stated "Employment is a very lagging indicator that peaks just ahead of recessions."
Based on my charts as shown, a reader accurately corrected that assessment.
The charts show employment sometimes leads, sometimes lags, and sometimes is coincident. On average it is coincident.
But the charts reflect revisions that are in fact lagging. So the correct interpretation counting is likely coincident to lagging.
Mike "Mish" Shedlock