I rarely cite Zerohedge but I found “Tyler Durden’s” post there about the strange things going on in the labor market pretty interesting. Here’s the meat of it. There are two major metrics for employment: the establishment survey which tallies payroll head counts and the household survey which reckons individual employment situations. Normally, they’re pretty closely aligned but starting in March of this year something “snapped”. Now they’re diverging sharply.
Whether this divergence was due to wrong seasonal adjustments (a remnant of the overreaction taken by the Dept of Labor following the covid crunch to normalize for a new normal labor market), due to erroneous Birth-Death assumptions (here too, the Dept of Labor was assuming early cycle new business creation which clearly is wrong with the economy late cycle and millions of businesses shutting down, ignoring the open PPP fraud that took place in early/mid-2000s as everyone “opened up” businesses to get free money from the government), due to the Establishment Survey inability to tell the difference between full, part and multiple-jobs – as a reminder we first showed that since March, the US had lost 400K full-time jobs offset by far lower paying part-time jobs as well as double-counted multiple jobholders…
… due to the record high rate of estimation – recall the 49% Establishment survey response rate was much lower than the 70-75% rate typical in November, meaning the Dept of Labor was literally making numbers up to “complete” the survey…
… or some other reason, perhaps including the Biden admin tapping certain Bureau of Labor Statistics officials on the shoulder and advising them to show strong numbers if they want to keep their… well… jobs, we did not know, but we did know that according to the Household Survey, just 12,000 jobs were created since March, while according to the Establishment Survey – which moves markets and sets Fed policy – the increase in jobs over the same period was 2.692 million!
The divergence is now at an all-time high. Most policymakers including the Federal Reserve rely on the establishment survey. If the household survey is correct, that means that a lot of policy is being predicated based on bad assumptions.
Drum pointed this out quite a while ago.
Steve
Two other data points.
More complete data from the Philly Fed used to revise the Establishment survey data suggests instead of 1.2 million jobs created in Q2 of this year, there 10K jobs created, i.e. the job market already stalled during the summer.
Also, real time data of tax withholding remitted to the Federal Treasury has shown a consistent weakening YOY comparisons since the Spring; from 11% yoy growth in April to 0% in the last several weeks. The only that makes sense in an environment where workers are getting large nominal raises due to inflation is the number of jobs is stagnant or decreasing.
When there are several pieces of data from different sources that correlate with each other one ought to pay attention.
On a final sidenote, I think the “establishment survey” needs serious reform. There is absolutely no reason why it needs to be a survey, it is technical possible to have employers report SSN numbers when remitting withholding taxes, and have a definitive real time picture of the labor market.
It’s against present law.
Technically possible as in the technology exists to do it.
This is fixable if Congress is interested in it. Today, by law employers keep records of how much they withhold for each employee throughout the year, and then report that to the government once a year (W2). Its just illegal to report it more then once a year…. (?)
Yours truly pointed this out on this very blog months ago. And Zerohedge commented on it months ago as well.
Apparently it only counts if Drum says it.