BLS and Jobs Report Major Revision

[welderwannabe]

As many of you know, the BLS not only releases a headline jobs number every month, but they also revise the previous 2 months. They also then do a catch up every September to look at the Previous April-March. I previously posted some calendar year based analysis of up how often they revise up versus down, but since these major bulk revisions are done April->March, I wanted to show some data I have pulled together from the reports based on that schedule from this period. Again I have a strong belief based on going back to the pandemic that the BLS always seems to err on making the initial reports look better.

April 2024->March 2025

Aggregate number of 'headline' jobs reported: 2.18M jobs

Number of times they revised up: 8 times
Number of times they revised down: 16 times
Note: They revise each month twice, hence why there are 24 revisions.

Total number of jobs after all monthly revisions: 1.88M

Bulk job revisions released in September 2025 covering the above period: -911K

Total number of jobs after all revisions, including bulk revision: 969K

This means that the headline jobs numbers for the above period overestimated job creation by 125% and we dont find this out for 18 months past the start of the period, making this report almost totally useless. Please also note that the previous year was in similar shape with a huge revision that came out this time last year.

[ochotona]

They also backdate the start of recessions by sometimes more than a year! The stock market always knows... BLS does not.

[boglerdude]

Another problem with central planning, only a handful of people to buy off and change the narrative of the entire country

I suppose there are private sector consultants doing surveys but they dont make the data free.