The Bureau of Labor Statistics (BLS) has faced growing scrutiny in recent years as monthly revisions to nonfarm payrolls have been persistently negative, at times sizeable. Over the 36 months ending December 2025, 27 out of 35 revisions (77%) were downward, with an average revision of –35,000 jobs.
More importantly for investors, this pattern may not be random. If payroll estimates systematically overstate employment late in the cycle, what appears as labor market resilience in real time may instead reflect a lagging and overly optimistic signal. For analysts, portfolio managers, and policymakers, this creates a meaningful risk: payroll data may be least reliable precisely when it matters most for assessing recession risk, economic momentum, and the likely path of policy.
If benchmark revisions follow their typical late-cycle pattern, 2025 payroll growth could ultimately be revised materially lower, potentially into negative territory. If so, current estimates may be overstating labor market resilience and misinforming asset pricing. Further, the jobs numbers can make the economy look healthy right when it’s actually starting to deteriorate.
This raises a key question: do these revisions reflect flaws in the survey methodology, or do they reveal something more systematic about the labor cycle itself?
