By Nicole Goodkind
Friday's jobs report was a mess, and not just because of the numbers.
The Bureau of Labor Statistics reported a loss of 86,000 private sector jobs in February. ADP, which processes payroll data for more than 26 million workers, reported a gain of 63,000. Two data sets analyzing the same economy over the same month are pointing in opposite directions.
Factor in a nurses' strike, winter storms, and a population reset delayed by the 2025 government shutdown and only incorporated into Friday's release, and the picture becomes harder, not easier, to read.
This isn't a one-month problem. The BLS overcounted employment by 911,000 jobs for the 12-month period ending in March 2025. December's payroll figures, initially reported as a gain of 48,000, were revised this week to show a loss of 17,000. Survey response rates have been falling for years. The BLS recently noted that its Current Population Survey can no longer reliably detect a 0.2 percentage point change in the unemployment rate. A half-million swing in job openings gets labeled "little change" because the confidence bands are so wide the number is statistically meaningless.
The birth-death model, the statistical engine that estimates job creation from new and closing businesses, has become especially hard to calibrate "when the economy is undergoing big changes," Federal Reserve Chair Jerome Powell said at a press conference in September. In the past, he called it noisy. That is a polite word for it.
A new paper presented Friday at Chicago Booth's Clark Center U.S. Monetary Policy Forum says the Fed has more options. Philadelphia Fed President Anna Paulson, a voting Federal Open Market Committee member this year, and San Francisco Fed President Mary Daly both offered remarks on the findings.
Other Fed officials like Governors Michelle Bowman, Lisa Cook, and Christopher Waller, Chicago Fed President Austan Goolsbee and Cleveland Fed President Beth Hammack were in attendance, listening closely to see what came of the presentation ahead of their own policy deliberations later this month at the March FOMC meeting.
The paper's central argument is that private-sector data from companies like ADP, Vanguard, and JPMorgan Chase can signal where the labor market is headed in ways that official government statistics, under increasing strain, cannot.
The headline finding is that private data predicts the BLS first print better than the Bloomberg consensus of professional economists, can anticipate whether those figures will be revised up or down, and carries meaningful predictive power for employment growth up to six months out.
The researchers behind the paper ran a real-world test. In the spring of 2025, while BLS first prints showed employment holding up, ADP data was already signaling deceleration as early as May. By June, the gap between the official first print and what private data was forecasting the third print would show was about a full percentage point. The conclusion: had the FOMC incorporated those signals into its policy reaction function, a quarter point cut could have been justified earlier than it actually occurred. The eventual third print confirmed the private data was right.
The stakes are high: a rising unemployment number could prompt the Fed to lower rates to stimulate economic activity.
At Friday's conference, Daly acknowledged the Fed has used private data this way before, pointing to the early weeks of Covid when official statistics were so delayed and incomplete that the central bank had little choice but to look elsewhere. "When they need it, we use it," she said.
But Daly was careful about what the 2025 case study actually proves. Some FOMC members, she suggested, were already seeing the labor market weakness that private data was flagging. The problem wasn't the information, it was the committee. Inflation uncertainty and wide dispersion in economic growth forecasts meant that even members who read the private signals clearly couldn't build consensus to act on them.
She also noted that BLS revisions are actually getting smaller over time, a reminder that official statistics remain a global gold standard. Paulson's framing was more structural. Policymakers, she said, are like archaeologists trying to reconstruct a mosaic, choosing which tiles to weigh and how. The problem right now is that some of the oldest and most trusted tiles are cracking.
The paper's authors are careful to note that private data is a complement to official statistics, not a substitute. ADP's sample skews toward midsize and large businesses. Vanguard's 401(k) data, which tracks hiring and income trends for five million workers in real time, tilts toward employers large enough to offer retirement benefits. JPMorgan's data covers 20 million bank customers but isn't always easy to access publicly.
Plus, every private source carries a durability risk. The Chicago Fed has run a credit card data partnership for years to help forecast retail sales, but they have lost it twice for short periods because of provider issues.
The question hanging over all of this is what comes next. Kevin Warsh, Trump's nominee to replace Powell in May, says he is forward-looking rather than backward-looking, a pointed critique of the Powell era's data dependency. Whether an incoming Fed chair philosophically skeptical of reactive policymaking will be receptive to more data inputs remains an open question.
But when the government and the private sector cannot agree on whether the labor market added or lost tens of thousands of jobs, it is also an urgent one.
Write to Nicole Goodkind at nicole.goodkind@barrons.com
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
March 07, 2026 01:30 ET (06:30 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments