Professional Curiosity Series · Week 46, 2026
A monthly jobs number is an estimate. Boards should decide in advance what it is allowed to change.
FalconBridge Partners · November 2026 · Own-account research, evidence date 9 November 2026
On 6 November the US Bureau of Labor Statistics reported that payrolls rose by 165,000 in October, against an average monthly gain of 48,000 over the previous year.
The agency's own 90% confidence interval on that single monthly change is plus or minus 122,000 jobs.
That does not make the number wrong. It makes it provisional, and it raises a question boards and lenders rarely ask out loud: how much should one release be allowed to move a hiring plan, a covenant conversation or a capital decision?
This piece is part of FalconBridge's Professional Curiosity series: questions we take up because they matter, research on our own account, and publish. It is not client work, and it draws on no client's confidential information.
We went to the primary sources to find out. We read the release as BLS publishes it, including its revision history and survey response rates. We read the research on how firms respond to provisional macroeconomic data, and how lenders build labour figures into credit models. We classified 68 references to labour data in a year of SEC annual reports and proxy statements, and compared eight regimes that already make decisions on noisy data.
165,000
Payroll gain, October 2026 (first estimate)
48,000
Average monthly gain, prior 12 months
±122,000
BLS 90% interval on one month's change
Source: U.S. Bureau of Labor Statistics, Employment Situation and Technical Note, as published November 2026.
The figures are the agency's, not ours. The long-run average revision between first and third estimate is about 56,000, as large as the current average monthly gain. Revisions can also run one way: in 2025 the first prints for May and June, 139,000 and 147,000, stood at 19,000 and 14,000 after the 1 August 2025 release. And the surveys behind the headline are answered by materially fewer respondents than in 2020, which the agency's own response-rate tables show.
What the evidence points to
Less over-reaction than failure to filter. Research on firms shows they respond to the error in early GDP estimates as if it were real, with later reversals in profitability. We found no study measuring the same behaviour in boardrooms for payroll data, so the risk is plausible rather than proven. But the filter is visibly missing where you would expect it. Of 48 sampled annual reports that cited the unemployment rate, none said which estimate the figure was or whether it would be revised.
Lenders already do this well
Their allowance processes use consensus forecast paths, probability-weighted scenarios and quarterly updates, and the Federal Reserve states plainly that its stress scenarios are not forecasts. The exposure sits elsewhere: in the board pack, and in the borrower whose covenant leaves little room if a strong print turns out weaker.
Two things surprised us
1 · Waiting is a risk too
Official announcements of US business-cycle turning points have come between 4 and 21 months after the turning point. A board that waits for certainty can act late on a deterioration that is already confirmed.
2 · Boards use different words
In a year of filings, “nonfarm payroll” appeared in one annual report. In proxy statements, “labor market” meant the market for talent and pay in every context we could classify.
Our view, for what it is worth
The useful discipline is not to react less but to filter first, and to agree the rule before the data arrive. Regimes that live with noisy data share a few habits: they act on averages rather than single points, require two independent signals, set thresholds in advance and make it harder to expand than to protect. Applied to October 2026, that reading says watch, not act: confirm on the next release, and make no expansionary commitment on one print alone. The thresholds we drafted are illustrative and have not been back-tested.
When the jobs number reaches our board, does anyone say which estimate it is and how precise it is?
Have we agreed what would make us act, before the next release rather than after it?
Is the bar for expanding on a strong number higher than the bar for protecting against a weak one?
The study behind this piece
Decide in advance what a jobs number may change
Monthly US labour data in board and lender decisions
The study sets out its sources and the gaps it could not close.
These are questions, not advice. The decision stays with the board. If the public record misses something you know, we would like to hear it.
Around this piece
USA · Week 46, 2026 · BLS: US payrolls rose 165,000 in October; unemployment rate changed little at 4.1%
More Professional Curiosity
A conversation about your decision
What needs to be understood
before your next decision?
Bring the proposition, question or direction you are working on. Together, we can define the support it requires.
Start a conversation