Jobs Day: September 2024

Pretty strong shock if you ask me

Going to start of with something I noted briefly (and with a hand-drawn chart) on Bluesky earlier this week: the recent decline in JOLTS measures has been huge. Like great-recession-huge:

JOLTS Job Openings Rate Sans Seasonal Adjustment.

Let’s draw some of the equilibria and put it on a log scale (the “correct” view per the DIEM¹). The difference between the pre-pandemic equilibrium and the current estimate of the one we are approaching is comparable to the Great Recession. If you measure from the top of the stimulus peak in 2022, they’re almost exactly the same size.

JOLTS NSA Job Openings Rate on a Log Scale with Equilibria shown (pre-Great Recession, pre-Pandemic, and the estimate of the current equilibrium.

What does this mean? While the level of the job openings rate is higher than most of the pre-Pandemic equilibrium period — given the whole “vibes-cession” discussion — I can’t help but think people might not be attuned to the level. It is also true that we just left a period of relatively high inflation. That inflation is over and has been for awhile now:

CPI Inflation (month to month). Forecast uses the vintage March 2022 data (before the Fed started raising rates). Inflation has been consistent with the dynamic equilibrium rate ~2.5%/y since late 2022.

In fact, that inflation has been over since around the time JOLTS started to fall². And while people complained about inflation while it was happening, the vibes-cession zeitgeist felt later. Maybe it wasn’t?³

Regardless, there has been a huge shock to JOLTS measures. Here are quits and hires rates:

This hasn’t manifested as a major increase in the unemployment rate (though JOLTS declines have preceded an unemployment rise in the past):

Unemployment rate rising from just below 4% to just above 4% over 2023 and 2024.

While the shock to JOLTS has been huge, the shock to unemployment has been tiny (“the littlest recession”):

‘Core’ U3 consisting of headline unemployment minus temporary layoffs. The Sahm-rule-like shock detection threshold was triggered back in early 2023.

The magic return to the pre-stimulus dynamic equilibrium has been the most puzzling — is it an example of a restorative force? Regardless, it does seem like a good bet going forward. The current path seems likely to merge with the expected level after the pandemic in the counterfactual of no stimulus shock⁴. In contrast, JOLTS job openings has exceeded this heuristic — openings are well past the no-stimulus counterfactual, “over-correcting” by about 100%.

Initial claims continues to be (hypothetically) useless — riding at the noise floor since early 2022:

An apparent “noise floor” in the rate of initial claims (ICSA divided by CLF16OV in FRED-speak)

So those are my big takeaways from Jobs Day September 2024: JOLTS shock = huge, stimulus information = evaporating, and initial claims = useless.

---

Update +15 minutes

Forgot the wage growth model! Unlike other labor market measures, wage growth appears to be heading back to the pre-pandemic equilibrium:

Atlanta Fed Wage Growth Tracker (see here) with DIEM forecast.

1. Dynamic Information Equilibrium Model. It is based on an assumption that a process variable A is growing (or declining) ~ exp(a t) for some a on average.

2. Still kicking myself over the fact that I did have a chance to forecast the rise in inflation based on my “gravity waves” hypothesis that major changes in labor force participation (WWII demobilization, women entering the workforce, … post-pandemic return to work) are associated with spikes in inflation.

3. I don’t really know anymore as every year since 2016 has been ridiculous.

4. I call it the stimulus shock, but that’s mostly as a label. The shock appears as soon as the Biden stimulus does and lasts from early 2021 to the end of the year. It might not be stimulus. It might be vibes. Maybe everything is vibes.