Ben Casselman

I missed fire elmo. (Well, “missed.”)

The report is published directly by BLS. Neither the commissioner nor anyone outside BLS (including the president, Fed chair, etc) see the number until they are final. Which means that if anything got changed, there are dozens of people who would know.

White House Pulls Pick to Lead Labor Data Agency

Except that isn’t true. The person Trump tried to install to run the BLS wasn’t confirmed, and just had to withdraw. The person running the agency in the interim is the longtime deputy commissioner who also served as acting head under Biden.

www.nytimes.com/2025/09/30/b...

“It’s another obstacle being thrown in their path,” said @ericagroshen.bsky.social, #CornellILR senior labor market advisor and former Bureau of Labor Statistics commissioner, about the potential impact of a federal government shutdown on the bureau’s efforts to hire new staff.

The "low hire, low fire" narrative remains in place: Layoffs have edged up, but they remain low, especially when we look at a rate rather than a total number.

The quits rate has basically been bumping along for the past year or so, below its 2019 level but no longer really falling.

The hiring rate (gross hiring, not the net job change we measure in the Friday jobs reports) has been below its long-run average for more than a year. It had seemed like it was leveling off, but might be falling again now, though hard to say definitively just yet.

But it IS getting harder to find a job. There is now less than one job opening per unemployed worker. Not a low rate by historical standards, but definitely weaker than just before the pandemic (and way weaker than at the peak of the reopening boom).

A few quick #JOLTS 📈:
Starting with: Job openings are way down from their peak, but they've fallen slowly if at all in recent months. No obvious sign that labor demand is falling off a cliff.
#NumbersDay

Job Openings and Labor Turnover Summary - 2025 M08 Results

Still no evidence of major cracks in the job market in August JOLTS data. Openings, hires, layoffs all close to flat. Revisions were modest. #NumbersDay
www.bls.gov/news.release...

So yes, private data can be useful, and can help us manage a short lapse in public statistics. But it can only get us so far.

(There are also other issues with private data, like short time series, inconsistent methodologies, etc.)

But even when the estimates are entirely based on private data, they are typically benchmarked to some public source. (This is how the ADP model works now.)
That won't matter much in the short run. But the longer we go without gov't data, the further these estimates will drift off their benchmarks.

In some cases, this means private sources literally incorporate government data into their estimates. (This is the way the old ADP model worked -- it was heavily based on lagged BLS data.)

Private data providers (at least the good ones) know this. And so they weight their data (or provide the necessary information to let others weight their data) to match the economy as a whole.
But this brings us to issue #2: Most private sources rely to some degree on government data.

Two main reasons for that:
First, private sources are hardly ever as comprehensive as government sources. This isn't about *size.* It's about *scope.* ADP has lots of salary data, but it skews toward larger companies. Zillow has lots of rent data, but it skews toward professionally managed units.

3. "Fine, we'll just rely on private data."
(This one is @marthagimbel.bsky.social bait)
There is lots of private data out there, and we'll be leaning on it heavily during a shutdown. But hardly anyone who has looked closely at these data thinks they can come close to replacing government sources.

There are long-run challenges facing our stats agencies, and attrition/staffing shortages under Trump have made them worse. But basically everyone I talk to who knows these agencies and knows their data, including many who are VERY critical of this administration, say they still trust the numbers.

I'm in regular touch with people inside BLS who I'm confident would tell me if they were being pressured to change numbers.
(Are YOU a staffer at BLS or another stats agency with something I should know? Reach out!)

EJ Antoni, Trump's controversial (to say the least) pick to run BLS has not been confirmed. So the agency is still being run by Bill Wiatrowski, a "BLS lifer" who also served as acting commissioner under Biden.

Is the Jobs Data Still Reliable? Yes, at Least for Now.

2. "The numbers were cooked anyway so who cares?"
I understand the skepticism after Trump's firing of the BLS commissioner back in August. But there is NO evidence that BLS (or any of the other major stats agencies) are cooking the books.
www.nytimes.com/2025/09/05/b...

One Casualty of a Shutdown: Government Data (Published 2023)

In 2023, the Biden administration said econ data would stop if the government shut down (the two sides eventually reached a deal to avoid a lapse in funding).
Here's my (very similar) story from back then:
www.nytimes.com/2023/09/28/b...

1. "Aha! Any excuse to hide the bad numbers!"
I won't pretend to know anyone's motivation. But important to note that econ data has basically always stopped during government shutdowns. A jobs report was delayed in 2013, and we skipped a GDP report entirely in 2018.

Three responses I'm hearing a lot of here:
1. "Aha! Any excuse to hide the bad numbers!"
2. "The numbers were cooked anyway so who cares?"
3. "Fine, we'll just rely on private data."
Let's address all three in a 🧵:

Is the Jobs Data Still Reliable? Yes, at Least for Now.

Fair question. I addressed that at more length in the story below, but the short answer is that BLS is still being run by the same career civil servants who were running it before (Trump's nominee hasn't been confirmed yet).
www.nytimes.com/2025/09/05/b...

Government Shutdown Could Delay Economic Data at a Critical Moment

Friday's jobs report will provide key evidence on whether slower hiring is turning into deeper weakness in the labor market.
If there is a jobs report on Friday at all.
My story on how a shutdown could leave us flying blind at a vital moment for the economy. #EconSky
www.nytimes.com/2025/09/30/b...

Except that most of the data that goes into the "nowcast" comes from government sources. I imagine it should do fine forecasting September CPI in a couple weeks, but it will struggle after that.

See follow-up post. This is the way it has worked in past shutdowns as well.

Maybe so, but no reason at this point to think the data would be false. BLS and other stats agencies still being run by career civil servants and there’s been no sign so far of any political interference in their work.

And also a lapse in data collection for the 2025 ACS.