The Employment Numbers Are being Fudged, So Says Jerome Powell

[progree]

22. Thank you. You're right, he did not. And a technical note from the BLS on the size of sampling error

Last edited Thu Dec 11, 2025, 04:50 AM - Edit history (1)

From the OP's link -- the Wall Street Journal article --
https://www.msn.com/en-us/money/markets/fed-chair-jerome-powell-says-us-may-be-drastically-overstating-jobs-numbers/ar-AA1S7cFs

Powell said that Fed staffers believe that federal data could be overestimating job creation by up to 60,000 jobs a month.

Given that figures published so far show that the economy has added about 40,000 jobs a month since April, the real number could be something more like a loss of 20,000 jobs a month, Powell said. We think there’s an overstatement in these numbers,” Powell said. “It’s a complicated, unusual, and difficult situation, where the labor market is also under pressure, where job creation may actually be negative,” Powell said.

. . . Powell’s concern involves a quandary that the Labor Department faces when measuring hiring: how to judge the number of jobs added or destroyed when new businesses are created or close down. Those jobs can’t be surveyed directly because it’s difficult for the government to reach out to brand-new companies or companies no longer in business.

Labor’s data arm, the Bureau of Labor Statistics, must use a statistical model to make a guess. In the past few years, that technique, called the birth-death model—referring to the births and deaths of businesses—has overstated job creation by hundreds of thousands of jobs a year, forcing significant downward revisions later. (("the past few years" -- so this isn't just a Krasnov era thing -progree. Highlighting by progree))

A falling number of timely responses to the labor surveys has increased the scale of a different set of monthly job-stats revisions, required after some companies hand in their payroll numbers late. A yearslong budget crunch and staffing shortages have also weighed on the agency’s capabilities. And, most recently, the extended government shutdown that ended in November set the agency’s work back by more than a month.

This doesn't sound like the numbers are "being fudged" .

============================================

Then there is this from the BLS - Employment Situation Technical Note (it's been this way for more than a decade -- it's not a Krasnov-era revision. A number of years ago it used to be +/- 120,000 instead of +/- 136,000. Somebody with time on their hands could find out at archive.org when it changed)

http://www.bls.gov/news.release/empsit.tn.htm

Based on what it says, there is a 90% probability that the Establishment Survey's non-farm employment increase is within +/- 136,000 of the stated number. And a 10% chance that it is off by more than 136,000.

Note, this is just the sampling error. There are other errors besides sampling error.

Correspondingly, again based on sampling error alone, there is a 50% chance that it is within +/- 55,800, and a 50% chance that it is outside that. So, for example for a reported job gain of 100,000, there is a 50% chance that it is between 44,200 and 155,800, and a 50% chance that it is outside that range based on sampling error alone. Note there are errors other than sampling error that add to the uncertainty

====================================

The latest available (thru September) nonfarm payroll jobs averaged only 39k/month over the last 5 months (May thru September),

'Non-farm payrolls: https://data.bls.gov/timeseries/CES0500000001?output_view=net_1mth

So for an average month of 39k/month for this period, the 50% sampling error bounds are 39k +/- 55.8k = minus 16.8k to +94.8k. And yet again this is due to sampling error alone. Other errors add to this uncertainty range.

=================================================

Note Powell said the numbers "could be... up to" 60,000 lower and "may actually be negative"

This is not the same thing as asserting that Powell's numbers ARE 60,000 lower and ARE negative.