Warsh's low-turnover line survives a July in which hires fell 278,000

The identifying problem in JOLTS is that two very different labor markets print the same low quits number. In one, workers stopped moving because the 2021-2022 churn already put them where they wanted to be, which is the version Fed Chair Warsh described at Jackson Hole and which Wolf Street reads as that rematching settling out ^[1][2]. In the other, they stopped moving because there is nowhere to go. Layoffs are the tell for that: they fell about 6.0% from a year earlier, with the three-month average sitting at the lower end of the prepandemic range ^[5][7]. Employers who are pulling back do not usually fire fewer people while they do it.

Then the slot ledger, which is where the mechanism lives: separations vacate slots, unfilled slots become openings, filled openings become hires ^[4][11]. Add July's three separation channels, 3.06 million quits plus 1.67 million layoffs and discharges plus 350,000 retirements and other separations, and you get roughly 5.08 million slots vacated ^[1] against 5.05 million hires, a gap of about 30,000 ^[2]. That is small enough for the rounding in three published figures to swallow it, so I would not build a payroll call on it, and the source is explicit that hires do not speak to payroll changes ^[11]. The check that does hold: 3.06 divided by 5.08 is 60.2%, which matches the 60% share BLS attributes to quits ^[4][9].

Harder to fit into the settled-rematching story is the hiring line. Hires fell about 5.2% in the month while quits fell about 4.9%, so in absolute terms the hiring decline was 1.77 times the size of the quits decline ^[4][5][6], and openings went the other way, with the three-month average about 1.8% above a year ago ^[8]. That leaves 1.44 openings for every hire ^[3], and because BLS counts an opening only where the work exists, the job could start inside 30 days and the employer is recruiting outside the firm, from a survey of 21,000 HR departments rather than scraped postings ^[8], with temp-agency and contractor roles excluded ^[9], the 7.27 million is not a stale-listings artifact. (Retirements and other separations, 7% of the total, are creeping up off a 25-year low set in 2025, which adds slots that no quit created ^[6].)

This is probably wrong in one direction, or rather, the more interesting version of the risk is that low quits and low layoffs can coexist for quarters precisely because nobody is doing anything, and the read gets settled only by whichever moves first. My view: the July print is consistent with turnover normalizing near the peak of the prepandemic years ^[12], which means soft quits on their own should not move a rate view, and the series that deserves the attention budget is layoffs. The rematching read fails if the layoffs three-month average climbs out of the lower end of its prepandemic range while quits stay near 3.1 million, or if hires keep sliding at roughly 5% a month while openings hold, because at that point the openings are decorative. Worth noting the survey's scope: it tracks turnover, not employment growth or unemployment, which came in the July jobs report on August 7 ^[13].