Those Jobs Numbers? Looks Like the Joke’s on Us.

Thousands and thousands of people staring at screens. Highly educated, well-trained professionals, staring at screens. Millions, even billions, maybe even trillions, of dollars, yen, euro, and other currencies poised to move around based on the data that comes out at 8:30 a.m. ET.

And what did we actually get? Another mishmash of data that are so different from expectations. Revisions were bigger than the number itself. Are we supposed to make something of this data?

Let’s pretend that the data are valid and go through the motions under the assumption that this is anything other than a wild guess, using methods and technologies that should have been updated years ago, if not decades ago.

Here are the numbers:

Only 29,000 jobs vs. expectations of 90,000 job. Ugh.

60,000 of downward revisions after last month’s 55,000. Ugh.

Only 46,000 in private payrolls, which also got revised down from the initial guestimate of 127,000 to 89,000. Ugh.

Somehow the 3-month moving average is now 51,000 this month and last. Hooray?

Hourly earnings dropped, good for wage pressure inflation, bad for anyone trying to make ends meet.

The unemployment rate rose 0.1%, but largely due to an increase in the labor participation rate of 0.2%. I kind of like that, assuming it is remotely accurate, which seems like a silly assumption to make, but one that we are all forced to make. The underemployment rate actually ticked down, which again, if real, is good. All of this is based on the household survey, which said we added 406,000 jobs in September. One is known to be more wildly inaccurate than the other hugely inaccurate number, but 406,000 sounds pretty darn good, if it was real, which it probably isn’t.

The birth/death model showed job losses of 190,000.That is not seasonally adjusted, and I don’t know how much that subtracted (or possibly added) to the non-farm payrolls headline number, but again, it seems “strange” that a “plug” (or calculation) is in the same order of magnitude of the number it is helping to “true up”.

Bonds and Absurdity

We will get the obligatory rally in bonds (and the numbers are “tepid” enough) that the bond rally should help equities along, but there is a better chance of me breaking par, than these numbers being truly useful in judging anything. We will use the numbers because “we have nothing better” but that increasing seems like a giant cop-out.

Seriously, if you were in a class and were told billions of dollars would be made or lost on a number that even the people looking at the number don’t believe, you’d shake your head in incredulity that this really was going on. And it’s not been for just for a month or two, but for years and years.

Can’t we use electronic paycheck data to get a pretty accurate number for all those who are getting paychecks? Wouldn’t that cover a big part of the economy. Wouldn’t knowing, with a high degree of exactness, how that part of the workforce (those who receive at least one check in a month) is evolving over time be useful? We could attempt to get the rest, but something that covered a vast swath of the economy, and was reasonably likely to be accurate and real-time, would seem a good goal.

Yes, I continue to hope to somehow land on a data task force committee, because this just seems more and more unbelievable all the time.

It is very good that Fed Chair Kevin Warsh in particular keeps telling us not to focus on any one month’s data, nor any one data series. That I’m in full agreement on, which is why I expect this initial reaction to fade.

You’d like to think ADP can capture some of what I’d like to see, and maybe last month’s 36k and this month’s 90k are more accurate? But who knows. But, also a reason not to fully trust today’s data.

Two pieces of data that I think are more difficult to “fake” or (more accurately, get absurdly wrong), are the Job Openings and Labor Turnover Survey, or JOLTS, “quit” rate, which remains low at 1.9%. People could be so happy with their current jobs they aren’t even thinking about another job (I know I am), but generically, I think it represents a “take this job and shove it” vibe, which increases when the perception is that equal or even better jobs are easy to find. The JOLTS “hires” rate came in at 3.3% this month. It has been stuck right around this level, which is a little lower than the 3.8% or so we typically saw in 2016 through 2019. That’s consistent with a no hire, no fire employment situation.

Bottom Line

We can examine data that the Fed may use to hike, pause or even cut, but the big drivers remain the same: The Iran war; the price of diesel; compute spend — uncluding the threat of cheap Chinese compute; and whatever is going on with European bond yields. The German two-year bond yield has declined from 3.31% on Monday to 2.99% as of Friday morning. The French 2-year has moved from 3.62% to 3.77% in that same time span. From 31 basis points to 78 basis points in a week is pretty ugly.

We also see some real weakness in credit spreads. I cannot remember the last time credit spreads seemed particularly interesting on the macro front, but they are again, with the credit default swap index moving from 50 to 60 in two weeks. Certainly not alarming but can’t help but look.

With the bond market shaky, and the excitement of this data like to fade as the morning wears on, we will all be back tp placing bets on the above issues, which seem like good topics for Monday’s report!

At the time of publication, Tchir had no position in any security mentioned.