What an awful jobs report! And don't skip over the revisions

What an awful jobs report.

And I’m not even talking about the contraction of -23,000 jobs in July.

I’m talking about the revisions.

The Bureau of Labor Statistics estimates that the economy shed 23,000 jobs last month.

Economists were expecting 83,000 new jobs in July.

Oops.

Interestingly, despite job losses, the unemployment rate clicked down to 4.1 percent. That’s because the labor force participation rate fell to 61.4 percent, the lowest level since the pandemic.

But in my opinion, the big story is the revisions.

They were so extreme, even the mainstream media reported on them. Typically, we never hear about the revisions (and they happen constantly).

Remember how we got those "strong" job reports in May and June?

Well, May’s job numbers were revised down by 66,000, from +129,000 to +63,000. The June job data was revised down by 37,000, from +57,000 to +20,000. 

When you add it all up, the total number of jobs “created” in May and June was 103,000 lower than initially reported.

Downward revisions are par for the course

These downward revisions should come as no surprise. It happens with virtually every BLS report.

In January, the bureau made its end-of-the-year adjustment to the “birth-death model” it uses to determine job growth. That erased nearly half a million jobs from the economy. To be precise, the BLS wiped out 403,000 jobs with its model revision (At the same time, it revised December’s report down from 50,000 to 48,000 jobs).

With that revision, the U.S. economy only generated an average of 15,000 jobs per month in 2025. You probably don’t have that impression if you just saw the headlines as the BLS announced its employment data each month.

If it sounds like the agency is just making stuff up, well…

In fact, downward revisions appear to be standard operating procedure for the BLS. The agency erased nearly 1 million (911,000) jobs that it initially claimed were created between March 2024 and June 2025.  

In 2023, job numbers were revised down in 10 of the 12 months.

To be fair, compiling employment data is no simple task. Revisions should be expected. But why do the updates almost always remove jobs from the economy? One would think you’d see upward revisions nearly as often as downward, right?
Nope.

Between 2003 and 2024, the final annual BLS numbers were lower than the initial report 14 times compared to seven upward revisions. 

It's notable that markets typically only react to the initial numbers. You almost never see markets tank because the BLS erased a bunch of jobs from the economy with a few clicks of its calculator. The revisions happen quietly in the back alleys. Nobody pays any attention to them. That creates the illusion that the labor market is much stronger than it is.

It goes something like this:

This month, the government reports good news. Everybody celebrates. Markets move. The following month, the government quietly revises everything downward and reports that the good news was really bad news.
And nobody pays attention.

This month is a notable exception to that rule. We’re seeing the revisions emphasized. I guess you can’t hide under the cover of darkness forever.

This wouldn’t matter nearly as much if central bankers and government officials didn’t lean on this data to make decisions. But they do. And if the data is this unreliable, what does that tell you about the decisions based on this data?

Let’s be honest - when you look at the history, one’s got to wonder why anybody takes these numbers at face value.