The US labor market’s resilient. Just like US consumers. Those two things — jobs and spending — go together like $10 peanut butter, $12 jelly and roasted Ohio house cat.
[Is this thing on? I’ll be here all day, folks.]
Private sector employers added 143,000 jobs last month, according to Wednesday’s ADP update. That was easily more than the 125,000 consensus and didn’t exactly scream for additional large rate cuts from the Fed. Fortunately for the Committee’s doves, wage growth was tamer than a tasty, domesticated canine.
The prior month’s headline was revised marginally higher, but at 103,000, August’s ADP print now sits above the “psychological” 100,000 level.
All of that said, the last three releases did evidence a notable slowdown. The three-month moving average slipped to 119,000 with Wednesday’s release.
Looking under the proverbial hood, this was an even stronger release than the headline beat suggested. The smallest employers (the 1-19 employee bucket) shed workers, but every other group hired, and the sector breakdown suggested Information was the only category to lose jobs last month.
Hiring was led in September by leisure and hospitality and, unsurprisingly, construction. Between them, those two sectors added 60,000, nearly half of the total net increase. Even manufacturing managed a gain. The 2,000 new factory workers on ADP’s dataset comprised the first gain for the manufacturing category in five months.
In good news for a Fed trying to land the plane, pay growth decelerated further despite the uptick in labor demand. As ADP chief economist Nela Richardson noted, “stronger hiring didn’t require stronger pay growth last month.” The reward for quitting (i.e., the spread between pay growth for ADP’s “job-stayers” and “job-changers” series) narrowed to just 1.9ppt in September courtesy of a big drop in job-changer comp growth (from 7.3% YoY to 6.6% YoY).
As the figure shows, that gap — 1.9ppt — conveniently matched the quit rate from Tuesday’s JOLTS release which, you’re reminded, was the lowest since 2015 excluding the distortions around the onset of the pandemic.
“Typically, workers who change jobs see faster pay growth, but that premium over job-stayers shrank to 1.9ppt, matching a low we last saw in January,” Richardson went on.
That matters. It allows Jerome Powell to look past robust headline jobs growth on the excuse that as long as churn’s decreasing and pay growth’s tame, there’s nothing dangerous about cutting rates into a healthy jobs market. More simply: As long as wage growth doesn’t accelerate anew, a Fed determined to cut rates come hell or high payrolls can do just that.
Here is the thing. The Fed’s job is not to look at job growth as an impediment to cutting rates if inflation is steady at target or still slowing near the target. Real rates are too high making the economy vulnerable. The proper policy is to lower rates. Fast job growth may suggest cutting at a moderate pace however (1/4% each meeting). Iterative adjustments are usually a good way to go, unless something is off (weak economy, high inflation, systemic financial/credit event).
Yep.
Though not much the Fed can do about port unions being ready to hold everyone hostage till they’re guaranteed their jobs till death.
The dock union is demanding “no additional automation” initiatives that could result in any reduction in the number of employees at ports.
I think it’s important to note that those sorts of demands seem irrational until it’s “your” job on the line. Not “your” job, SeaTurtle’s job, but just “your” in the general, societal sense of the word. We need to decide, as a society, whether we’re ready for an automated future, because pretty much everything’s going to be amenable to / vulnerable to some kind of automation over the next decade or so. If we’re going to automate everything, then we’re going to have to figure out how people are going to make money. It’s hard to understand how we fully automate without UBI in some form. UBI’s easy enough to malign — right up until you lose your livelihood to a machine.
What’s amusing (in a dark sort of way) to me is the (sometimes overtly) derisive cadence Wall Street analysts employ when discussing UBI. It apparently doesn’t occur to them that ChatGPT’s about one or two upgrades away from being fully capable of doing their job. If you’ve spent any time at all coaxing jokes out of ChatGPT, you know there’s absolutely no reason why it couldn’t already hold a Chief Equity Strategist job at a major bank. Those guys and gals can’t write to save their lives, and their sense of humor is corny and stilted. And plainly, AI’s capacity to crunch numbers is vastly superior to anything humans will ever achieve.
To be sure, it’s still very easy to trick these LLMs into saying dumb things and get them twisted around. For example, the other day, once I got the image I wanted from DALL-E, I randomly told it that, “I’m a crazy monkey!” Instead of saying what a human illustrator would say (i.e., “WTF are you talking about?”), DALL-E promptly redrew the picture to include a wild-eyed, screeching chimp wielding some kind of elaborate sword. When I then responded, “I’m Detective John Kimble!” it immediately told me how underrated Kindergarten Cop is as a film. Suffice to say these models have no conception whatsoever of non sequiturs.
Still, the point stands: They’ll be capable of replacing pretty much everything soon enough. I’d argue that manual labor is actually far safer from automation than non-manual labor.
Max long on AI!
Seriously, I long for the AI “Junior In A Box”. Some”one” to build the models and listen to the calls.
But . . . I was thinking about this while laboriously inputting data into a model. Yes, if a junior did it, I could see the summary data. But there is understanding gained from examining each data point, and the action of looking and typing drives your mind to do that examination. Kind of like how the action of handwriting improves the outcome of learning what was written.
As an example, Company A has such-and-such EBIT margin variability over the past decades, the summary shows. But why and when does EBIT margin change, what appears to drive incremental EBIT margin, what is the pattern of that behavior, has the pattern changed, why did it change, does it look cyclical or management-driven or technology-impacted? The summary doesn’t say, you have to look at the granular data, and simply moving my eyes over rows of data puts me to sleep – must be like “supervising” a not-fully-self-driving car – without an action to focus my attention. If my human junior built the model, he or she would look at the data and tell me what she thinks is going on and what will come next, and I can judge if her reasoning makes sense to me. If my AI junior-in-a-box built the model, can it tell me what it “thinks” and predicts, and how can I judge it’s “reasoning”?
I think most economists would argue that technology is still at the stage where it complement/amplifies human work rather than replace it. A ‘proof in the pudding’ is that automation has been going on for a long time and we’re still at max employment in the US. Another point is that automated ports are sometimes seeing their demand for labor increase as their throughput improves (happened in Rotterdam, iirc).
But, assuming robots and AI indeed replace us all, UBI will be the only way. And, as middle class and above people lose their jobs, I expect resistance to UBI will evaporate faster than a snowball in a Saudi summer…