The Most Important Number in the Jobs Report Isn’t -23,000
In this day and age why are we making key decisions off data that are wild guesses?
You've reached your free article limit
You've read 0 of 1 free Pro articles.
Normally, you can find some parts of the jobs report that “fight” against the headline. That somewhere in the details is a potentially different narrative. Maybe it’s because I’m lazy on a summer Friday, but it’s difficult to see what that is in the July jobs report.
But let’s start with the most important point.
Instead of worrying about what Fed Chair Warsh said or didn’t say, how often he might or might not say things, to whom he may or may not speak to, maybe we should worry about making decisions based on garbage data?
As you know, I’ve argued for years that it seems insane that in this day and age, where we have real-time data on almost every aspect of our lives, we are content to kind of stick a finger in the air, and take a wild stab at estimating jobs. I think the data source task force is the most important and potentially useful thing Warsh has created. The data task force is too limited in scope, if anything.
The establishment headline is -23,000. The estimate was +80,000. Revisions for past two months were -103,000! Maybe the estimate would not have been for 80,000, if the initial reports for the prior months reflected reality and didn’t need to be revised lower?
Not sure this is “good” news for the workers, but wage growth was anemic — even as the renewed fighting in the Middle East is pushing up the price of energy products.
The unemployment rate has dropped to 4.1% from 4.3% in the past two months. On the surface, maybe that is good. But the unemployment rate is based on the household survey, which had a loss of 87,000 in July, which is “better” than June’s loss of 507,000 jobs. The four-month total number in the household survey is more than 600,000 lost jobs! The unemployment rate is only lower because the participation rate has dropped to 61.4% from 61.8% in two months. And let’s not forget, this is occurring with record capex on data center/AI build.
OK, now my “favorite,” the birth/death model. I will admit I’m not sure how the seasonally adjusted birth/death model translates into a number of jobs in the establishment survey, but this “model” showed 235,000 jobs added by new businesses being formed. Maybe companies are being formed to take advantage of the AI/data center spend. It is in line with last July’s birth/death adjustment, so maybe it didn’t impact things. However, I always struggle when “plugs” or “models” seem to be bigger than the actual numbers.
Honestly, I have no idea if Friday’s numbers are the aberration or whether it was what was originally reported, but that is the point!
How are we making decisions based on data that seem to be a wild guess (apologies to wild guesses)? Sure, if there is no way to get better, more accurate, timely data, then we’d have to live with it. We had to live with carrier pigeons at one time. I find it difficult to believe that a nation that put astronauts on the moon, cannot figure out a better way to calculate data that are so crucial to decision making!
So, yes, Houston, we have a problem, but the bigger problem isn’t Friday’s numbers, it is that we don’t really know what numbers are correct or not!
On the other hand, my view that we don’t see a hike this year looks better now than it did a week ago — and that is with no “deal” in Iran, which any deal, will also help.
Have a great weekend and if you missed it, I had long interview on Bloomberg TV yesterday, addressing many pertinent topics (starts at the 1 hour 43 minute mark).
https://www.bloomberg.com/news/videos/2026-08-06/bloomberg-surveillance-8-6-2026-video
