The Bureau of Labor Statistics (BLS) has come under increasing scrutiny lately as monthly revisions to nonfarm payrolls have been persistently negative, sometimes significant. In the 36 months ending December 2025, 27 of 35 revisions (77%) were downward, with a median revision of -35,000 jobs.
More importantly for investors, this pattern is probably not random. If wage estimates systematically overstate employment late within the cycle, what appears to be labor market resilience in real time could as an alternative reflect a lagged and overly optimistic signal. This poses a big risk for analysts, portfolio managers and policymakers: payroll data could also be least reliable on the time when it’s most significant for assessing recession risk, economic dynamics and certain policy direction.
If benchmark revisions follow their typical late-cycle pattern, job growth could ultimately be revised significantly downward in 2025, possibly into negative territory. If that is the case, current estimates may overstate the resilience of the labor market and misrepresent asset prices. Additionally, employment numbers could make the economy look healthy just when it actually starts to deteriorate.
This raises a key query: Do these revisions reflect deficiencies within the survey methodology or do they reveal something more systematic concerning the work cycle itself?
