When the BLS Data Flow Resumes

A Quick Preview

TL;DR: Once the shutdown ends and the data flow resumes, we’re still going to have an outdated understanding of the US labor market until at least early December.

This piece is split into:

1. The Big Picture

2. The Current Population Survey (CPS)

3. The Current Employment Statistics (CES)

4. JOLTS

5. Claims for Unemployment Insurance

6. The Quarterly Census of Employment and Wages (QCEW)

7. On “Permanent Damage”

More below.

1. The Big Picture

The US government shutdown is on the verge of ending, and that means that US government agencies will resume collecting and publishing labor market (and other) data. What should we expect?

Conceptually, it’s worth separating out “data collection” and “data publication”. In some cases, collection was significantly disrupted, and there will be a permanent hole in the historical time series. In other cases, collection was either ongoing or can be “made up” after the fact; it’s a matter of compiling, processing and publishing that data, but we will have a complete historical record.

There will be some delay in data publication. We might get the September data relatively quickly, but that’s old news at this point and of relatively minor interest. Due to the disruption of October data collection, the first batch of timely labor market news will be the November BLS jobs report¹, due in December. So while the shutdown is over, we’ll probably still be relying on “alternate” labor market data for a few more weeks.

Surprises are possible, but most of the labor market data we saw during the shutdown was consistent with only a mild cooling. And that’s what I expect to see in that “fresh” November data.

Let’s go through the key data sources one by one:

2. The Current Population Survey (CPS)

“The household survey” in the monthly jobs report is my personal favorite and in my opinion the most valuable source of labor market data. It was not collected during October at all. I’m guessing that the November data will be collected, though with a bit of a crunch: the November CPS reference period is the week of November 2nd through 8th² and collection would have already started a few days go. My bet is that they can and will go through with November CPS data collection.

As far as publication, the never-released September CPS data will come out as part of the overdue September jobs report relatively soon - the shutdown happened just a day before they would have come out, so preparation must have been close to complete. Based on comments from the White House Press Secretary, we will not get an October jobs report, but instead will get November CPS data as part of the November jobs report sometime in December. That report was originally scheduled for Friday December 5th, but there could be a small or moderate delay.

A final thought: the absence of October data for eternity is a genuine misfortune, but it’s not the end of the world. Unless you’re a scholar, we’ll have November data points reasonably soon. That data will suffer from the lack of contextualization that we typically have (“don’t look at a single jobs report in isolation”), a problem that will be significantly ameliorated in early January with the December data.

A second related gripe I’ve heard is that “we’ll never know what happened in the October labor market”. I don’t think that is true. We have an OK (albeit imperfect, blurry) idea based on alternate data, and any durable impacts of October’s events (for example, elevated unemployment to the degree it occurs) will persist into November.

3. The Current Employment Statistics (CES)

The “establishment survey” in the monthly report is where nonfarm payroll employment (including employment by industry), average weekly hours and average hourly earnings comes from. It gets a lot of attention from labor market watchers, though in my opinion its usefulness is lower than usual at present.

That said, this series (unlike the CPS) will probably avoid an October lacuna. The CES has an extended reporting period to allow for slow-reporting establishments to provide data. (That’s where the revisions everyone hates come from!) Some folks have interpreted Press Secretary Leavitt’s comments as implying these will not be released, but I’m reasonably confident they will be published - as part of the November jobs report.³

I think it will be worth looking at response rates for the August, September and October data - these might be lower than normal, reducing the reliability of the estimates. But again, that is mostly a problem for historians and nerds like me - we’ll know the level of November employment by sometime in December, and have OK substitutes courtesy of several data providers. By early January, we’ll have 2 job reports in hand and the significance of the publication disruption will recede into memory.

4. JOLTS

August JOLTS data on turnover and openings was the last major release that came out of BLS before the shutdown. Like the CES, it’s a survey that has built-in mechanisms for delayed responses. So my guess is that we’ll have September data (collection would have started in October) and I am very sure we’ll have October data (collection would have probably started a week or so ago). I’m not sure about publication - they may decide to roll the September data into an October report (with release in December) or even (less likely) roll the September and October data into a November report.

5. Claims for Unemployment Insurance

Data on claims for unemployment insurance has been a godsend during the shutdown. Unlike the BLS data, the claims data are collected by individual states (as well as DC, Puerto Rico and the Virgin Islands), and the Department of Labor kept discreetly compiling them during the shutdown without publishing the weekly claims report. That allowed intrepid labor economists like yours truly to keep tabs on the data, the way we would in normal times.

The weekly reports should resume relatively quickly; at some point relatively soon the DOL will also backfill the missing official data (that will probably correspond closely to the data me and other economists compiled).

6. The Quarterly Census of Employment and Wages (QCEW)

The QCEW is a more accurate measure of employment and provides source data for the annual benchmark revision. (We already have a preliminary estimate as of September; the final revision, which will probably be less negative, is due in early February.) The Q2 2025 QCEW data was initially scheduled for December 3rd, but my guess is that publication will be delayed. Nevertheless, I’m pretty sure it will be published - the data has already been collected (by individual states). I do wonder whether the BLS fill feel stretched rushing to complete publication in time for the final benchmark revision (due in early February 2026).⁴

7. On Permanent Damage

The White House Press Secretary’s comments on “permanent damage” have raised questions about whether she is talking about something more serious than just the non-collection of CPS and CPI data. Here’s what she said, verbatim, according to Acyn:

The Democrats may have permanently damaged the federal statistical system with October C.P.I. And jobs reports likely never being released and all of that economic data released will be permanently impaired.

I know there are expansive interpretations out there, these seem to be narrow (politically-charged) comments about the October jobs and CPI reports, as well as the data contained within. Ominous interpretations about future data are not warranted, besides the obvious point that RIFs and underfunding are permanently damaging statistical data collection.

1. I suppose it’s possible that October JOLTS could come out slightly before then.

2. Earlier than a typical month, but fairly normal for November and December. The CPS often moves these reference periods earlier because they’re worried about response rates being disrupted by the late-in-the-month holidays.

3. The other option would be to publish “half a monthly jobs report” including October data for the CES but not the CPS.

4. Strictly speaking, we already have a full year of QCEW data as of September. But the Q1 data typically gets revised with the Q2 release, which gets incorporated into the final benchmark.