Why Nonfarm Payrolls Was A Near Disaster

• Decent headline, lousy details

• Fudged seasonal adjustments avoided a payroll drop

• There was very poor breadth as most of the job market has ground to a halt

• Highest continuing claims point to rising unemployment

• Markets reacted dovishly to the details and claims data. So should FOMC officials.

• Nonfarm payrolls, m/m 000s / UR %, SA, September:
• Actual: 119 / 4.4
• Scotia: -20 / 4.2
• Consensus: 51 / 4.3
• Prior: -4 / 4.3 (revised from 22 / 4.3)

Putting aside the grievance that I was wrong about the headline this time given 119k payroll jobs were added in September (but on details made money for clients given the market reaction), I’m still calling this report a near disaster. It should be good enough for the FOMC to cut in December and I’ll explain why. Pairing the report to continuing claims data strengthens this argument on the path to the October and November readings. I now have more confidence in my -90k call for October payrolls.

Markets responded by pushing Treasury yields lower with 2s through 10s down by about 3–4bps. The dollar weakened in the aftermath. Pricing for the December FOMC meeting moved up by 3bps to about 9bps of a cut priced at the moment. My reading of vote-weighted FOMC members is such that I think December pricing is probably too low.

First, the seasonally unadjusted pace of hiring was close to my expectations. I had figured about 350k m/m and it came in at 317k. Seasonally unadjusted hiring is usually positive in September (think school year etc) but this September registered the fourth weakest month on record (chart 1).

Second, what was offsetting, however, was the strongest seasonal adjustment factor ever applied on record for a month of September (chart 2). With no explanation. These factors are fudged and subject to distortions imo.

Third, to give a sense of how much this matters, see chart 3. At almost any other typical SA factor for the month of September, payrolls would have shrunk. Perhaps by a lot!

Fourth, there was very low breadth to the gain. Leisure/hospitality was up by 47k and the education sector added 57K. There was literally nothing material across all other sectors certainly in the private sector (chart 4).

Fifth, one minor exception is that government added 22k entirely at the state and local levels as the federal government shed only 3k. Funny, I thought DOGE packages were expiring last month and that while more would drop off in October, there would be a pickup in the number of lost jobs in the federal government from the prior month’s 14k as they secured other jobs before packages ran out. Either they didn’t or be careful here.

Sixth, some of the job gain would be a perverse reaction to ICE raids. A growing number of jobs in the health care sector are accompanied by Visas and green card sponsorship which may reflect hotter demand as a way of avoiding expulsion than what was already becoming apparent before the raids (here). The share of Indeed job postings that include visas or green card sponsorships has been rising since 2021. Outside of health, job postings are tumbling and we’re weeks away from when about $35 billion of funding in the Affordable Care Act expires at year-end, driving insurance premiums sharply higher for an estimated 22 millions Americans and with it perhaps less demand for healthcare workers that may have been over-hired.

Seventh, the trend is soft. Only 193k jobs have been added in the five months since April.

Eighth, revisions removed 33k jobs over the prior two months. August turned negative.

OTHER DETAILS ADDED TO THE SOFT TONE

The unemployment rate edged up to 4.4% which has nothing to do with nonfarm as it’s derived from the household survey. The wonky household survey registered +251k jobs but the labour force expanded by 470k. The confidence bands on the household survey are multiples higher than nonfarm's.

There was little wage growth in September (0.2% m/m SA, 3% m/m SAAR). Chart 5.

Hours worked were flat (+0.07% m/m SA). Hours worked have ground to a halt in Q3 (-0.08% q/q SAAR) which means that solid Q3 GDP tracking is entirely driven by productivity (chart 6). Hours times wages get folks paid and they both came in softly. A flattening of hours worked while productivity accelerates is a telltale first-response cyclical warning sign for employment going forward.

There was a much higher collection rate this time given they had more time to collect the data (chart 7). That means lower revision risk to September when the next numbers come around. 

Weekly continuing claims shouldn't surprise anyone given advance tracking using the sum of states after applying seasonal adjustments, but they are at the highest levels since late 2021. Fired workers are not getting reabsorbed. Charts 8, 9.