The Worst Jobs Report Since The Pandemic
Chief Economist Andrew Flowers dissects a "bad" jobs report and whether it could signal a permanent shift for the US labor market.
Photo Credit: Aditya Vyas
The headline number for the February 2026 jobs report is bad, but after you dig into the details – welp, it isn’t much better. There was a shocking decline of 92,000 jobs last month. An important factor was a Kaiser Permanente strike, affecting over 30,000 healthcare workers in California and Hawaii. Plus, federal government jobs declined by 10,000 amid a partial government shutdown. But those factors alone don’t explain away that this is the worst jobs report since the beginning of the pandemic.
Looking at private employment excluding healthcare — so removing the two largest distortions in the data — still shows a decline of 58,000. When labor market growth depends on a single industry, we will inevitably end up with large swings like today.
Indeed, there was weakness across the board, with declines in construction, leisure and hospitality, manufacturing, and professional services, among others. Meager gains in financial activities (+10k) and in retail (+2k) were really the only notable positives.
Killing the Golden Goose
As we’ve covered in detail, healthcare has been the “golden goose” of the labor market – the powerhouse engine of employment growth that’s propped up the labor market. But that took a turn for the worse last month, particularly for Offices of Physicians – a sign that the Kaiser Permanente strike weighed down job gains.
The strike is one factor. But what about all the funding cuts from the federal government? There was $1 trillion in cuts to Medicaid to be phased in over the next decade. There were cuts to state-level Affordable Care Act subsidies, which haven’t been fully reinstated. Perhaps this is a turning point for the industry. It’s just one month of data, so it’s early to be catastrophizing — but it’s a concern.
Ominous signs: Downward revisions, tech hiring and construction slowdowns
There are other concerns, too. This report had many data points for the labor market “doomers” to latch onto. Downward revisions to the tune of 69,000 for January and December is a sign that the hoped-for reacceleration isn’t materializing; and, worse yet, they may be a harbinger of future negative revisions.
Then there was weakness in the Information sector, a rough proxy for tech hiring. Steady job losses there raise fears of A.I.-fueled displacement, whether rightly or wrongly. The combination of weak tech hiring and strong productivity figures from earlier this week will only inflame the A.I. debate.
And then there’s the slowdown in construction jobs, which had been a minor bright spot over the past year (far less than healthcare, but at least generally positive). Perhaps the best explanation for construction weakness is weather-related.
An Older Population
All that we’ve covered until now is focused on the so-called “establishment survey” of the jobs report, which informs the payroll tallies. But there’s other bad news on the household survey-side of the jobs report. That’s the data source the calculates the unemployment rate, which rose from 4.3% to nearly 4.5% (just barely rounded down to 4.4% in February).
Each year, the Census updates its population estimates. And this year was a doozy, as my former boss Jed Kolko explains. Basically, the population and thus the labor force got revised to be noticeably older, in part due to lower immigration. As a consequence, the overall labor force participation rate and employment-to-population ratio were revised lower. That means the total talent pool, whether employed or looking for jobs, is less engaged with work. The talent pool is shrinking.
What does this mean for recruiters?
It means there’s weak job growth and a lower supply of available talent, and both trends are likely slipping lower. Healthcare itself had a bad month, even after taking into account the Kaiser strike. Is that because of something more permanent shifting? We don’t know yet. Finally, it’s worth repeating a mantra we have at Recruitonomics: don’t over-index to one data point. That said, this was a bad one!