US Wages Accelerate in a Tight Job Market

• US job gains continue at a robust pace…

• ….while wage growth accelerated

• The outcome reinforces Federal Reserve guidance

• Are multiple job holders distorting the gains?

• US Nonfarm payrolls, m/m 000s // UR (%), November, SA:
• Actual: 263 / 3.7
• Scotia: 205 / 3.6
• Consensus: 200 / 3.7
• Prior: 284 / 3.7 (revised from 261 / 3.7)

US job growth continued at a fairly robust pace last month as layoffs likely held back the pace of gains that would have been even more powerful otherwise. Wage growth accelerated.

The beat to consensus and my estimate was within the 90% confidence interval around nonfarm estimates (+/- 120k) but that never fusses markets. They just looked at the beat relative to the fear factor that’s waiting for the other shoe to drop and sold everything. The S&P is down by just ½% on the day but the two-year Treasury yield is up 13bps and the 10 year is up 4bps in a bear flattener. Terminal rate pricing edged up by about 12bps and has reversed about half of the decline following Chair Powell’s speech. The dollar initially strengthened but has given back most of that now.

There are probably no direct implications for the Federal Reserve’s script into the meeting the week after next. They’ve already largely pre-determined that the fed funds rate will go up by 50bps and the expected terminal rate will go “somewhat higher” which sounds like a consensus view around about 5%. One data point won’t change this for the Fed especially given concern toward forward-looking downside risks into 2023.

DETAILS EMPHASIZE WAGE GAINS

Wage growth accelerated. It was up by 0.6% m/m seasonally adjusted which translates into 6.8% at an annualized rate. That’s a resumption of the quickest wage gain since late last year (chart 1). It’s just one month of acceleration so far but the baseline gains leading into that month were already fairly elevated. The prior month’s gain was also revised up to about 0.5% m/m SA which translates into 5.7% m/m SAAR.

There was high breadth to the gain (chart 2). Services employment was up by 184k with goods up 37k, construction up 20k and manufacturing was up 14k. Within services, education/health up 82k, leisure/hospitality up 88l, government up 42k all state/local. Retail down 30k which goes against the strong holiday season angle (3 consecutive declines)

The unemployment rate held steady at 3.7% (chart 3). It is derived from the companion household survey that showed a bigger decline in the labour force (-186k) than the decline in employment (138k) from that survey. The participation rate slipped (chart 4). The household survey has a higher margin of error around it than payrolls but it also captures off-payroll small business employers.

Hours worked slipped by 0.1% m/m SA. They are tracking a gain of just over 1% q/q at an annualized rate (chart 5). Since GDP is an identity defined as hours times labour productivity this is supportive of Q4 GDP growth.

MULTIPLE JOB HOLDERS NOT THE CULPRIT

It’s worth taking a closer look at an issue that has drawn criticism of the payroll numbers this year. Do they overstate job growth in the US economy because of double counting?

On the one hand it’s true that nonfarm counts jobs, not workers (so double/triple etc counting risk) versus the household survey that tracks bodies and is therefore not susceptible to double counting jobs. The other household survey difference is that it also captures employers without formal payrolls better (ie: small biz).

The evidence points to how the household survey of employment fell by 138k in November with a 186k drop in the size of the labour force. So on the surface that might suggest that the discrepancy between the two job measures is due to double counting.

But we need to look at the household survey’s details to see if double counting was really an issue.

The household survey’s measure of multiple job holders has been mostly moving sideways since about mid-year. It has also mean-reverted toward pre-pandemic norms after getting crushed in the depths of the pandemic (chart 6). One cannot dismiss job gains because of unusual gains by multiple job holders and therefore treat payrolls as fake data. A job's a job in my books so if we're just mean-reverting then that's a.o.k. by me. Americans are doing what they have to in reverting to prior ways of paying the bills.

Further, I like looking at the household survey’s measure of folks who are working part-time for economic reasons which was flat (+25k) in November after a 183k drop in October which doesn't lean toward more people having to take on multiple part-time jobs and inflate the payrolls data for economic reasons. Part-time voluntary jobs (not for econ reasons) were down 77k in Nov and up only 41k. What has happened is that American part-time jobs for non-economic reasons have gone back up toward pre-pandemic levels but part-time for economic reasons has gone back down toward pre-pandemic levels (chart 7).

Another problem is that the household survey is wildly volatile and with much wider confidence bands than nonfarm. It was up by about 650k in August and Sept and is now down by 466k in Oct/Nov. The trend is volatile but still tentatively up.