The September employment report, published on October 2, showed the American economy added 29,000 jobs. Economists had forecast 84,000. The unemployment rate rose to 4.2%, leaving 7.1 million people counted as unemployed, and revisions took 60,000 jobs out of the two previous months, turning July from a gain of 21,000 into a loss of 10,000.
Two weeks earlier, the September purchasing managers survey reported hiring accelerating at the fastest pace since June 2022 and business activity at a five year high. Both of these describe the same month. Only one of them is a count.
Soft data and hard data measure different things
A purchasing managers survey asks whether conditions improved. The payroll report counts jobs on employer records. When they diverge, the usual resolution is not that one is wrong but that they are answering different questions: a firm can report that hiring conditions improved while adding nobody, and a diffusion index registers the direction of change rather than its size.
There is a second explanation worth holding, which is that sentiment surveys lead and payroll counts lag, so a survey can be right about a turn that has not yet shown up in the data. That argument has been made in both directions this year and it cannot be settled from one month.
The revisions are the part that deserves the most weight. A single monthly print carries a wide margin of error and routinely gets rewritten. The pattern of the last several months has been consistent downward revision, and a trend of revisions in one direction is more informative than any individual headline, because it says the initial estimates are systematically missing something.
What the composition is telling you
Whitmore's Watchlist:
RHI (Robert Half Inc.): A staffing business, which is about as direct a read on hiring demand as a listed company provides.
XRT (SPDR S&P Retail ETF): Consumer spending, which follows employment income with a short lag.
IEI (iShares 3-7 Year Treasury Bond ETF): Intermediate Treasuries, where a softening labour market shows up before the long end notices.
Health care added 17,000 jobs, construction 11,000 and manufacturing 9,000. Financial activities lost 7,000, and that sector has now shed 129,000 positions since May 2025.
The concentration matters. Health care hiring is driven by demographics and public funding rather than by the business cycle, so a report in which health care is most of the gain is a weaker report than the headline suggests. Strip it out and the private economy added very little.
Average hourly earnings rose 5 cents to $37.81, up 3.0% over the year. That is the first time in a long while that wage growth has been running below the headline inflation rate, which is around 3.4% on the Federal Reserve's preferred measure. Real wages going backwards is a different consumer story from the one the record equity market implies.
What it does to the October meeting
The committee raised rates on September 16 and sixteen of nineteen participants projected at least one more increase before the end of the year. That projection assumed unemployment at 4.1%. It is now 4.2%.
A central bank with a dual mandate reads this report as the first real tension between its two obligations this cycle. Inflation at 3.4% headline and 3.0% core argues for tightening. Payrolls at 29,000 with negative revisions argues for waiting. John Williams had already said there was no need for urgency before this report landed, and futures pricing for an October increase has been sitting close to a coin flip since.
What the report does not do is settle anything. One soft month after a strong survey is not a trend, and the committee has two more employment reports and two inflation prints before December.
For a portfolio the useful observation is that the two sides of the curve are now reacting to different reports. The front end moves on employment because employment is what gives the committee permission to pause. The long end has spent the last fortnight moving the other way entirely, with the ten year above 5.3% and the thirty year above 5.7%, on supply rather than on data.
Whitmore's Take: A 29,000 print with 60,000 of downward revisions is a labour market losing momentum, not one that has broken. Worth watching the revisions rather than the headline, because that is where the trend has been showing up first.
Written by Daniel Whitmore
Millionaire Insiders