TL;DR: We got an unusually rich batch of labor market data this week, which looked neither great nor bad. The most probable current state of the labor market is “cooling slowly”, with no trend change in the past month.
In this note I’ll cover:
What’s Next
What to Make of Mass Layoff Announcements
An Update on the Shutdown
The Federal Reserve
Chicago Fed Unemployment Rate Nowcast
Claims for Unemployment Insurance
Morning Consult Unemployment Rate
Indeed Job Openings
Conference Board Survey
ADP Weekly Data
What I’m Reading
More Below Chart.
1. What’s Next
Normally, I’d be getting you psyched up for (or anxious about) the next round of JOLTS and jobs data due in the first week of November. But those won’t be published next week.
We’ll be getting more data on unemployment insurance claims, Indeed job openings, and ADP. Revelio will publish their October job growth estimate. And we’ll get an updated unemployment rate nowcast from the Chicago Fed.
However, with the shutdown probably getting closer to its end, I’m starting to think about the resumption of government data collection/publication and what we should expect the data to show.
2. What to Make of Mass Layoff Announcements
This week was dominated by breathless headlines about mass layoffs (two prominent examples: 48,000 at UPS, 14,000 at Amazon). How worried should we be?
I’d like to urge some perspective. The first thing to remember is that even during the past year, a period of lower-than-typical layoffs, we averaged 1.73 million layoffs a month. The Amazon and UPS announced layoffs, combined, are what the US experiences roughly every 26 hours.
Of course, that’s not the end of the story. Perhaps these announcements are representative of what smaller companies are doing, and despite recently low layoffs we’re about to experience a layoff wave. A lot of things are possible! But we’ve had multiple waves of mass layoff announcements going all the way back to late 2022 or early 2023, and aggregate layoffs have risen only modestly since then. I wrote about the information value of mass layoff announcements (spoiler alert: low) back in early March. I’m inclined to think the significance of this wave is overblown, too.
The third point I’d like to remind everyone: layoffs have not played a big role in labor market cooling over the past 3-3.5 years. A massive decline in hiring (which paused in mid-2024, but may have resumed in late summer) was a much bigger culprit. There are obviously lots of stories about young people struggling to find jobs out there, but not many about “mass hiring freezes”. But there probably should be - they are much more important than mass layoffs for explaining the current state of the labor market.
In “the boy who cried wolf”, the wolf eventually materializes. We’ve had surges in aggregate layoffs in the past and it will happen again at some point in our lifetimes. Maybe the wolf will show up as soon as late 2025. But it hasn’t come yet.
3. An Update on the Shutdown
In the past week, the anticipated shutdown length from Kalshi hasn’t budged: 44.8 days. That puts us at mid-November, only two weeks away. Please consider wide error bands around the market’s prediction: lots of things can happen and outcomes are uncertain.
I don’t think it’s coincidence that we’re (probably) approaching the end of the shutdown at the same time as headlines about real consequences increase: the government may cut off SNAP benefits, and airports have grounded flights periodically. It’s a natural negative feedback loop that has prevented durable economic damage from shutdowns in the past.
If you want to be a worrywart, I’d think of how that feedback loop would malfunction this time around.
4. The Federal Reserve
We had an FOMC rate decision, policy statement and Chairman Powell press conference this week. I’ll keep my comments in this section limited to the labor market. In the press conference, Powell said:
“…[A]t the July meeting we saw downward revisions in job creation, we saw a very different picture of the labor market, and suggested that there were higher downside risks to the labor market than we had thought. And that suggested that policy, which we had been holding at a -- I would say modestly, other people would say moderately, restrictive level, needed to move more in the direction, over time, of neutral. If the two goals are sort of equally at risk, then you ought to be at neutral, because one of them is calling for you to hike, and one of them is calling for you to cut. So if that got back into balance, then you’d want to be roughly at neutral. So in that sense it was a risk management, and I would say the same about today. Sort of the same logic. But as I mentioned, going forward is a different thing.”
Elsewhere, he said:
“Yeah, I mean in principle if you were to see data that suggested that the labor market is strengthening, or even that it’s stabilizing, that would certainly play into our decisions going forward. So, and we do have -- we get some data. The labor market is a place where we get, for example, we get the state level data on initial claims, which are sending sort of a signal of more of the same. We also get job openings, and we’ll get lots of survey data, we’ll get the Beige Book and things like that. So we’ll have a -- we’ll have a picture of what’s going on in the labor market. And the fact that we’re not seeing an uptick in claims, or a downtick really in openings, suggests that you’re seeing maybe continued very gradual cooling, but nothing more than that. So that does give you some comfort.”
The Fed’s two rate cuts this fall have been anchored around the worry that we were approaching an inflection point for the labor market. That inflection point has not obviously materialized. Our read on things is blurrier than usual, but I’d expect the data I’m writing about to look a lot worse in those circumstances. It’s worth reflecting on what happens if we remain on the same, “maybe continued very gradual cooling” trajectory through mid-December.
5. Chicago Fed Unemployment Rate Nowcast
In my opinion, this is one of the most significant indicators in the absence of government data - a proxy, based on other data (a lot of it covered in this note) of one of the most important labor market metrics we have.
The gist is that not much has happened in the past month. Over the past 2 months, the unemployment rate rose by 0.03 percentage points. Naively extrapolated over the next year, that would land us at 4.50% by next August. That’s pretty close to the Fed’s current projections for next year, which also incorporate further improvement in core inflation and 50 more basis points of cuts beyond this point.
6. Claims for Unemployment Insurance
We’re still extremely fortunate that the Department of Labor is collecting and publishing this data, albeit without the convenience of a weekly report. (They’re publishing a huge CSV of weekly state level data and I’m aggregating it.)
The main takeaway is that not much has changed beyond the federal government. Regular initial claims are roughly where they were 1-2 years ago (layoffs steady) and continuing claims are rising very slowly (consistent with Powell’s very gradual cooling in the labor market). We have seen an indisputable rise in labor market distress among current/former federal government workers (both initial and continuing claims), but its impact on overall labor market distress is very small.
Let’s start with regular initial claims. My estimate is they were around 219K in the week ended 10/25, right around where they were in late October 2023 and 2024.
Regular continuing claims in the week ended 10/18 were up a little relative to 2024 and 2023. I think this reflects what Chair Powell calls “very gradual cooling” in the labor market, which we’ve been seeing for a long time with no meaningful change.
The above metrics all exclude federal government workers. Their claims for unemployment insurance have spiked dramatically, to around 10K on initial claims (week ended 10/25) and 23K for continuing claims (week ended 10/18).1 My sense is initial claims by federal workers are peaking or have already peaked; continuing claims should continue to rise for a while (possibly until the shutdown ends).
While these spikes for federal workers are pretty stark, they are small in the grand scheme of things. Initial claims for federal workers are equivalent in size to about 4-5% of regular initial claims; continuing claims for federal workers are equivalent over 1% of regular continuing claims. I use regular continuing claims to nowcast unemployment due to permanent layoffs, but if we include the federal worker continuing claims the impact on the nowcast is barely visible:
One thing I’ve been interested in is how much leakage we have from various recent policies (the shutdown, RIFs) onto the private sector. And the answer is that there’s been a little, but it seems to be peaking (just like federal worker claims).
I wouldn’t characterize this as the last word on the subject. If the shutdown goes on much longer than financial markets expect, we’ll probably find good reason to return to this chart.
Finally, you’re used to the nowcast that doesn’t currently work: Google Trends data on unemployment insurance searches. This shows signs of having peaked, though the massive gap between search and action remains:
7. Morning Consult Unemployment Rate
During the shutdown, the wonderful people at Morning Consult have made their labor market data public. I love their data, because they try to track the unemployment rate - something few other folks stepping into the shutdown data desert have attempted.
The Morning Consult data tells us a similar story to claims: we’ve seen only a small impact, if any, from the shutdown. A pessimist will observe that the survey unemployment rate has stopped falling; an optimist will mention that, as of yet, we’re still marginally below year-ago levels.
8. Indeed Job Postings
At this point you’re probably getting bored - so be it. Indeed’s job postings data is among the highest quality labor market metrics we have. (Though long-time readers know I greatly prefer turnover to openings as a measure of labor market heat.)
Postings continue to gradually climb down at the same rate as before the shutdown. This is a bit of a Rorschach Test between mild optimism and mild pessimism. The mildly pessimistic take is “the labor market is very gradually cooling and the shutdown has not had much impact”. The mildly optimistic take is that the current rate of decline is comparable to what we saw when the labor market was fairly stable, between mid-2024 and mid-2025.
9. Conference Board Survey
The Conference Board’s monthly Consumer Confidence survey produces one of the most road-tested unemployment rate proxies we have: the labor market differential. This measures the delta between the share of respondents saying “jobs are plentiful” and those saying “jobs are hard to get”.
This indicator ticked up a tiny bit, though this isn’t particularly reassuring - it’s exhibited fairly significant residual seasonality, improving around year-end only to plunge in the spring and summer.
But I don’t know how much to worry about that ongoing decline. Traditionally folks have mapped this indicator against the unemployment rate or the quits rate, and it’s tracked pretty well. But over the past year or two, it’s become decoupled. I don’t know whether that means that the BLS data is too optimistic, or that the “vibecession” is starting to contaminate this august indicator.
For a long time I thought this indicator might be vibecession-proof as it asks something relatively concrete: the availability of jobs, vs. the abstraction of the state of the economy. But it’s possible that even though indicators like the unemployment rate are relatively stable (or deteriorating only a little), people are very pessimistic due to what they read in the news.2
Chart courtesy of Greg Daco.
10. ADP Weekly Data
A not-quite-secret is that the Fed gets weekly data from ADP. ADP also publishes this data with a significant lag. There was a recent kerfuffle with between the Fed and ADP because ADP was annoyed that Governor Waller cited that weekly data in a speech. Fortunately ADP has decided to resolve that kerfuffle by publishing the weekly data! 3
The gist is that ADP shows employment growth perking up a little relative to the summer (which was probably “peak pessimism”). The recent employment gains are not much above zero, but that might be enough to keep the unemployment rate steady given very low labor force growth.
11. Reading List
At this point my reading list (unchanged for a long time) is a recurring punchline, but:
“Seeing economic data through the fog of immigration estimates” by Jed Kolko (Link)
“Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data” by Lichtinger & Massoum (Link)
“How Retrainable are AI-Exposed Workers?” by Hyman, Lahey, Ni & Pilossoph (Link)
“Evaluating the Impact of AI on the Labor Market: Current State of Affairs” by Gimbel, Kinder, Kendall and Lee (Link)
OK, I have to run to Trick or Treating with my kids! There may be typos or bad stylistic choices here, if so please ping me and I’ll correct them.
DC was late submitting its numbers, and has higher than normal exposure to federal workers, so I had to impute their numbers. Possible it will get revised next week!
I know some folks will chime in with “hiring is really low, so jobs are indeed hard to get”. True. But the decline in hiring stopped in 2024 and I suspect any subsequent decline in the last few months has been small - we’ll see when JOLTS revives.
Though as far as I can tell, only in chart form. Hopefully the underlying data will be published in CSV form, too.
On the layoffs from UPS, I have a few friends who work for them. According to my friends , it’s more older employees taking packages so the company can hire new guys at a much lower rate of pay for five years until they fully vest. According to them, they are already starting to hire new workers. Make sense?