Mother Nature Temporarily Slammed US Payrolls, But a Distorted SA Factor Prevented a Negative Print 

• Payrolls ground to a halt

• Mother Nature is to blame as hurricanes walloped the US

• Two reasons why future reports are likely to sharply rebound…

• …and hence why the Fed will look through this

• If not for wonky seasonal adjustments…

• …payrolls would have posted a sharp decline…

• …which requires looking at smoothed data

• Markets overreacted at first

• US nonfarm payrolls, m/m 000s // UR %, SA, October:
• Actual: 12 / 4.1
• Scotia: 95 / 4.2
• Consensus: 100 / 4.1 (consensus payrolls range –10k to +180k)
• Prior: 223 / 3.9 (revised from 254 / 4.0)
• Two-month revision: -112k

US payrolls ground to a halt with a +12k print while the more volatile household survey posted a decline of -368k. Bear in mind that the 90% confidence intervals are +/-130k for payrolls and +/-600k for the very noisy household measure. The payroll number would have even weaker yet had it not been for a wonky seasonal adjustment factor. I’ll explain why all of this should be ignored and probably will be ignored by the FOMC next week.

FIRST, WHAT HAPPENED

First the data, and then I’ll turn to the massive caveats.

The weak estimates cited above were accompanied by a negative revision of -112k to the prior two months including –72k to September (now 223k) and –40k for August (now 78k).

The unemployment rate was unchanged at 4.1% only because it is derived from the noisy household survey that recorded the -368k job loss along with labour force shrinkage of -220k. The two effects cancelled each other out in the UR calculation.

Chart 1 shows the breakdown of the change in payrolls by sector. Goods sector employment fell by 37k and services employment was flat at +9k. Government added 40k. There was high breadth to the softness in payrolls with only education/health/social services and government posting material gains.

Hours worked were flat at 0%m/m SA. After only 0.56% q/q SAAR growth in hours worked in Q3 that meant productivity accounted for the solid 2.8% q/q SAAR GDP print, hours are once again tracking poorly in Q4 with just a 0.7% q/q SAAR gain baked in so far (chart 2). The same thing applies in that Q4 GDP growth will require productivity gains unless hours really pick up. The chart might also indicate that US workers have hit the wall in terms of willingness to work more hours. 

Wage growth was strong at 0.4% m/m SA. The 3-month moving average is about 0.4% after a trio of solid gains that have been pushing the year-over-year rate up from a low of 3.6% in July to 4% now. Chart 3 shows the pressures at the margin in that m/m SAAR wage growth picked up to 4.5%. As chart 4 shows, the real wage compression narrative remains incorrect; average US hourly wages adjusted for inflation have not been durably falling over the pandemic era and have in fact risen on a cumulative net basis to date.

WHY THE FOMC IS LIKELY TO IGNORE—AND YOU SHOULD TOO!

The FOMC is very likely to ignore the outcome for the following reasons.

1. It’s obviously a hurricane effect notwithstanding the BLS’s usual assertion that “it is not possible to quantify the net effect” of the hurricanes which they have said after prior negative hits to payrolls from hurricanes. Never go high on a payrolls estimate after major hurricanes and so the upper end of consensus around the 150–180k mark has the most explaining to do here.

2. Charts 5–6 show we’ll very likely get upward revisions in the next couple of months for two reasons. One is that more data will be collected if the pattern following past hurricanes is any indication, and on that count the BLS noted today that the response rate “was well below average.” The other is that there will be a resumption of more normal business effects that will lift payrolls going forward as per the activity after past hurricanes (charts 7–8). I wouldn’t be the least bit surprised to see the next report showing an upward revision to October and a strong November gain.

3. The seasonal adjustment factors remain just plain wonky. Chart 9 shows that October 2024 was the highest SA factor on record compared to like months of October in history. What that means is that payrolls, if anything, were overstated on first pass compared to a scenario in which the SAs were not subject to a recency bias in their estimation. Chart 10 shows by how much they would have fallen by looking at what would have happened to seasonally adjusted payrolls in October at varying SA factors applied to the same seasonally unadjusted estimate. Payrolls would have been significantly negative at any other SA factor in past month of October. 

The issue of SA factor distortions in payrolls has been around for many years, long pre-dating the pandemic, and there have been multiple pieces written about it over prior periods. And yet it's much worse than ever now given the recency bias to re-estimating SA factors post-pandemic. It’s like statisticians saying they know the data they are producing is garbage, but we refuse to adapt so do as you wish with the numbers.

And so what we need to do is smooth US payroll changes over the full year in this environment to take account of the likelihood that distorted SAs overstate job growth at the start of the year now, understate it in the summer, and then begin overstating it again. Upon doing so, the ytd average payroll gain is 170k this year versus 254k over the same period last year. Since last year didn’t have a notable hurricane effect, omitting this would probably boost ytd payrolls closer to 180k. That’s still a slowing pace of trend job creation, but probably close to estimates of the break-even rate as tighter immigration policy begins to slow labour force expansion once again—and with darker scenarios potentially ahead.

MARKETS ARE MORE SENSIBLY ALIGNED WITH THE FOMC’S DOT PLOT BY YEAR-END

The initial reaction to the numbers saw the US 2-year yield rally by about 13–14bps, but this is being reined in to about half that effect now. Undershooting on bad data isn’t unusual, but markets are likely reconsidering how to look at the figures in more balanced ways.

The only way I can justify the front-end rally after wonky payrolls data is that markets had gone too far shaving the odds of back-to-back 25bps cuts in Nov and Dec. Being more in line with that now with around 45bps of cumulative easing by year-end is a normalization effect to the dots that they are very likely to deliver barring something really unusual. Otherwise, if Ts are rallying because they really think the US job market is going off a cliff then it's the one-born-every-minute crowd driving that trade imo.