JOLTS Recap (May)
Turnover Is Showing Signs of Stabilization
TL;DR: The “Great Stay” continued in May, though quits have stabilized and hires may be getting there too. Relative to a year ago, turnover is down substantially but net employment growth has not changed much.
The rest of this post covers:
The Big Picture
Beveridge Curve Thoughts
1. The Big Picture
There’s renewed anxiety in LaborWonkLand about whether the US labor market is getting closer to the recession tipping point. Most of the JOLTS data doesn’t validate that anxiety; the pace of cooling is slowing, not accelerating. (Though of course things could change.)
Between the spring of 2022 and the spring of 2023, the JOLTS data sharply deteriorated. Hiring and quits fell substantially. Even layoffs increased, from “insanely low” to “very low”. But over the past 6-12 months, the trajectory has improved.
Quits have been flat since early 2024. This is a “revealed preference” by employed people on the state of the labor market, and it’s no longer getting worse.
Hiring is still coming down, but at a slower pace.
And layoffs have stopped going up.
This could all change pretty quickly. Renewed declines in quits, a reacceleration of hiring declines or another step up in layoffs would suggest we’re getting closer to a downturn. But the dynamics look less worrisome in this data than they did a year ago, and what’s happening right now looks like we’re getting closer to a steady state.
Of course it’s a blah steady state. Hiring is at levels we saw in 2014/2015, when the unemployment rate was around 5.5%. This is not a great time to be looking for a job. But it is a pretty good time to have a job, with unusually low layoff risk across most of the economy.
2. Beveridge Curve Thoughts
Yesterday I emailed someone with the phrase “Beveridge Curge”, which was probably a typo but is also a funny accidental portmanteau of “Scurge” and “Curve”.
Scurge or not, the Beveridge Curve is the empirical relationship between job openings and unemployment. In “normal” times (orange or dark blue dots), the Beveridge curve is a diagonal almost-line - when openings are high, unemployment is low, and vice versa.
The post-pandemic period experienced a massive surge in labor demand that showed up more forcefully in job openings than in unemployment, leading the Curve to shift sharply outward (grey line). And by early 2022 - the peak of the Great Resignation - job openings were insanely high.
The surge in inflation during that period convinced many (not all) economists that policy needed to bring labor demand back down. Some of those economists, looking at the grey dots, thought this would entail a big surge in unemployment. Others, including Federal Reserve Governor Christopher Waller, were more optimistic: they thought you could bring down openings, and labor demand, down with a minimal impact on unemployment.
Waller and his allies were right. For about a year, from the spring of 2022 through the spring of 2023, openings plunged with no impact on unemployment (the yellow dots). And for a year after that, from the spring of 2023 to the spring of 2024, the impact on unemployment was very small (the light blue dots).
But now we’re back in the same region as the top left of the 2010s Beveridge Curve (orange dots), and the question is what comes next. Is our current segment of the Beveridge Curve still near-vertical, or does it have a more normal diagonal shape? Does further squeezing of labor demand lead to large increases in unemployment?
The truth is we don’t know. The shape of the Beveridge Curve is only clear in retrospect, and the intercept has shifted over time (compare the dark blue dots to the orange ones). It’s reasonable to assume that further declines in openings will have a higher unemployment cost than the ones that already happened, but we don’t know how big that delta will be.
And even if the Beveridge Curve is more downward sloping than it was (unclear), it’s important to remember that the perceived need to squeeze labor demand is much milder than it was in 2022. The Fed has already told us they don’t need to see labor market indicators cool much further to become convinced the labor market is “in balance”. Even on the diagonal Beveridge Curve segments, a very small decline in openings will have only a small associated increase in unemployment.
That means the risk is less “are we on the diagonal portion of the Beveridge Curve”, and more “can the Fed appropriately calibrate policy so as not to overshoot with labor market cooling”. It’s easy to say “we only want a tiny bit more labor market cooling”, much harder to actually deliver on that promise.