This was a great panel, and if you care about economic data, you should definitely listen to the whole thing. But for those of you who would rather get some highlights, here's what I found most interesting: 🧵

Video of yesterday's "Future of the Fed" panel at the AEA meetings. David Romer, Janet Yellen, Loretta Mester, Anastasios Orphanides. Moderated by Anil Kashyap. videosolutions.mediasite.com/Mediasite/Pl...

BLS has an FAQ re:shutdown impacts to Nov. CPI released tomorrow. In short, Nov. CPI index values will be calculated by comparing Nov. prices to Oct. prices where available (from alternative/non-survey sources) and Sept. prices otherwise (most prices). www.bls.gov/cpi/

Adding sailors and mariners to this list. Our lives depend on good marine forecasts. This is so sad.

"Executives say they still believe generative AI will eventually transform their businesses, but they are reconsidering how quickly that will happen within their organizations." www.reuters.com/business/bus...

There was some speculation recently that the DRP wouldn’t be this big a drag on October’s NFP print but those estimates from OPM last summer turned out to be pretty accurate.

Confess I had not thought through the implications of no data collection in Oct. for Nov.'s jobs report until I read BLS's FAQ. 1)CPS is a panel. No collection in Oct means no new respondents initiated into the survey, and so many more need to be added in Nov. 2)above clearly has implications for

This was also my take-away from the Sept jobs print. If payroll prints are trending down only because of labor supply (immigration) the unemployment rate should be trending down or steady, not heading up. Unemployment up means demand is slackening more than supply

Fed's Williams: Official government data is the gold standard.

Probably the most constructive thing you’ll read today. GOLDMAN: “.. Shelter inflation is set to sustainably undershoot its pre-pandemic pace. .. PCE housing inflation will slow to .. 2.1% year-over-year in December 2026. .. another reason we expect near-target inflation next year.”

Staffing is down ~25% at BLS due to DOGE/DRP and the hiring freeze. But this impacts price programs (largely collected by BLS feds) more than jobs. CPS (household data in the jobs report) is collected by Census field staff. They are down staff too but have been prioritizing economic indicators.

A nice thread from Guy on JP's comments yesterday on NFP growth being overstated. To which I will add a caveat - the birth/death model is not the principal driver of the wedge between CES and QCEW these last two years. (Warning, statistical esoterica ahead) 🧵1/9

A nice piece on algorithmic pricing worth your time. A question I got frequently when I led BLS was why the agency went into stores and didn’t just collect prices online. My answer, “prices online are often different than in the store, so we do both” often surprised people.

Bulldogs have two modes: 1. snoring adorably, 2. barking at you until you open the treat jar

A thorough review of the issue. One missing part that economists don't seem to see is that since 2008 the problem isn't that NYC is expensive. It's that Ypsilanti is now also getting more expensive. Vibecession: Much More Than You Wanted To Know www.astralcodexten.com/p/vibecessio...

So great to see BLS back to work measuring and informing us about our large and dynamic economy. It’s a great institution full of dedicated civil servants deeply committed their mission. They have had a tough year, and everyone is covering for someone who left. I thank them all for their service.

BLS confirms that October jobs report will not be released, but that October payroll numbers will be published with the November report

When Canada tried this is was a data disaster for large swaths of the country

While everyone is still digesting how it feels to not see a Sept jobs report until late November, remember this feeling the next time folks complain about monthly revisions and suggest publishing data with a 2 or 3 month lag.

Allow me to translate. October data was not collected as field staff were all furloughed. In some cases, that data can be recovered from businesses. In other cases, the window to collect the data has closed (consumer prices, labor force status) and the data hole may be permanent.

All of BLS has been furloughed since October 1st. That means no one was collecting data in Oct. You can’t just walk into a Costco in mid-November and find out what the price of a good was in October. That’s why an October CPI isn’t possible.

The usual caveats apply when comparing ADP to BLS jobs reports. But add an additional one for October; this report was anticipated to be a big negative headline print - an estimated 150,000-200,000 federal workers left government at the end of September. www.reuters.com/world/us/us-...

It’s been just one month with little official economic data.

A very nice rundown of what private data sources can and cannot tell you about the state of the economy right now.

Had not seen this reported elsewhere, that ADP has stopped sharing micro data with the Fed since the government shutdown. Article says Waller 'disclosed' this arrangement in an August speech, angering ADP, but the Fed has been transparent about this project for years (see, Cajner, et al., 2018).

Just pointing out the incoherence of ‘we must replace BLS leadership because of falling response rates’ and ‘we aren’t going to let you stem falling response rates by following up when they don’t respond’. (Census collects data for BLS too.)

This is a good summary.

The big story here is of an agency of dedicated statisticians and public servants working tirelessly to improve economic data in a climate of budgetary cuts to data collection. They have been very innovative in meeting that mandate. But we should also just fund our economic infrastructure.

The preliminary benchmark revision of -911K amounts to -0.6% to March 2025 payroll employment. Combined with 2-month revisions, recent total revisions are big but hardly unprecedented, & smoothed over the business cycle the payroll survey has gotten more accurate over time.

Your daily reminder that US economic data is of extremely high quality even when compared to other advanced nations and the envy of the world

The first wave of the survey underpinning nonfarm payrolls has seen falling response rates, but in contrast to accusations the BLS has "failed to take corrective action" several targeted steps have kept 2nd & 3rd wave response rates at or above 90%--Gold standard data from dedicated public servants

APOLLO: “.. Census data is starting to show a slowdown in AI adoption for large companies. “.. The US Census Bureau conducts a biweekly survey of 1.2 million firms, and one question is whether a business has used AI tools .. to help produce goods or services in the past two weeks.”

The larger-than-usual downward revision last month was in large part driven by a negative skew in the job growth distribution among late reporting firms. That’s unusual, but it’s happened before when the pace of job growth slows rapidly. This print is more evidence that was the case

A key challenge for BLS is adequate funding for data collection and modernization. During my tenure I tried to identify the causes. One person told me, ‘your problem is that while BLS has no enemies you also have no friends’ meaning no powerful interest groups advocating for economic data.

"Payroll employment revisions indicate a slowing labor market—not a political agenda" A great resource for journalists & others who want to understand recent events with BLS. @upjohninstitute.bsky.social Pres Mike Horrigan helped lead BLS for decades. #EconSky #️⃣#️⃣ www.upjohn.org/news/payroll...

It has been the honor of my life to serve as Commissioner of BLS alongside the many dedicated civil servants tasked with measuring a vast and dynamic economy. It is vital and important work and I thank them for their service to this nation.

BLS explaining its CPI data collection reductions "Roughly 15 percent of the sample in the other 72 areas also was suspended from collection, on average." But it reassured this kind of reduction will have limited effect on CPI www.bls.gov/cpi/notices/...

That was my theory going into the piece, but it turned out to be not only false but seemingly the opposite: the biggest employment gains young male grads have made in the past year have been in software and engineering jobs

I’ve talked to many young people in recent years who are interested in an econ PhD but don’t want a career in academia. Glad to see NABE stepping in here to provide more information on nonacademic careers for economists. (Only half of recent US econ PhDs work in academia!)

Attn users of CPS microdata: the April 2025 has been re-posted on the Census site to correct some errors in the weights and geography variables. Links in next tweet.

Can’t beat this commute to the office (walk to Woodley Park metro via the Zoo)

Even the guide dogs are posing among the blossoms on the mall today

CPS data users take note:

A nice defense of the utility of vacancy data for policymakers, and how vacancies might rise and fall with recruiting intensity

Wisdom of the crowds: first flight from DC to ASSA meetings I’ve ever been on that wasn’t packed with other economists. Seemed like a bad sign and sure enough it was.

Home in the frozen north for the weekend helping muck out these guys stables

Lots of folks assume more N is always better but not if your sampling design or methods are flawed. If your sample is biased, a larger N will make your estimates more precise, but still wrong. Statisticians refer to this as the ‘big data paradox’. A good intro here:

Last jobs day at Postal Square, last post-briefing coffee at Union Station, packing up to join the first wave moving to our new digs in Maryland next week.

Crowdsourcing some ideas for remarks for upcoming events - what do my fellow economics data watchers want to know more about?

The US Census Bureau's Business Trends and Outlook Survey shows that AI adoption by US firms has risen to 6% the highest on record. (Data goes back to late 2023.) Additionally, the share of firms *planning* to use AI has risen north of 9%.