Growth Without Hiring

The economy grew at a 3.4% pace last quarter while adding, by the first count, about 50,000 jobs a month. Half the growth was stockpiling and an accounting quirk, much of the rest went to imported computers, and the economy now needs far fewer new jobs than it used to.

The reading

The first estimate of third-quarter GDP, published October 29, put growth at a 3.4% annual rate, the fastest in a year and up from 2.2% in the spring. Over the same three months employers added 152,000 jobs in all, about 51,000 a month: a loss of 10,000 in July, a gain of 133,000 in August and 29,000 in September. In 2015–19 the economy added nearly four times that many in a typical month. The unemployment rate barely moved, from 4.1% to 4.2%. The October jobs report, published November 6, revised August up to 186,000 and September to 60,000, so the quarter now comes to 236,000, about 79,000 a month; October added 165,000, and the unemployment rate was 4.1%.

Strong output and weak hiring don't usually travel together for long. When they seem to, there are three places to look: what the output actually was, how many workers the economy needs, and how much each one produces. All three are part of the answer this time.

New articles and charts arrive in one short email on Monday mornings. Get the weekly digest

Where the growth came from

GDP is built from pieces, and the Bureau of Economic Analysis reports how many percentage points of the 3.4 each one supplied.

Read together, the last two lines are the heart of it. Demand from American households and businesses was strong: their combined spending, from groceries to new buildings, grew at a 4.9% rate. But a large share of the equipment was built abroad. An imported server counts as investment when it is installed and is subtracted again as an import, and either way it was assembled by workers in another country. BEA's equipment figure is itself built mostly from import records, which is one reason it will be revised.

The trade balance, and goods alone. The deficit widened through the summer as imports of computers and capital equipment surged; it is the minus 3.3 points in the GDP report. Full chart and context →

How many workers the economy needs

Slow hiring has not pushed unemployment up because the number of people looking for work is barely growing. The Congressional Budget Office estimates net immigration at 410,000 people in 2025, against the 2 million it had projected a year earlier, and 570,000 in 2026. With the workforce growing that slowly, the breakeven, the number of new jobs a month needed to hold the unemployment rate steady, falls with it. A Federal Reserve Board study put it at about 155,000 a month in 2023–24, about 85,000 in 2025, and possibly under 10,000 in 2026. A St. Louis Fed estimate for 2026 runs from 15,000 to 87,000, depending on how many people arrive. On those numbers, 51,000 a month is roughly enough.

The people who are here are mostly working. The share of 25-to-54-year-olds with a job was 80.7% in September, higher than in more than 90% of months since 1948. What has changed is movement. Employers hired 3.3% of their workforce in August, lower than in about three months out of four since 2000, and laid off 1.0%, close to the record low. The frozen job market has the precedents for that combination.

The count itself is uncertain. The Bureau of Labor Statistics' preliminary annual check against tax records suggests payrolls in March 2026 were about 79,000 lower than now published, a correction it will make in February. The other monthly survey, of households rather than employers, showed nearly 900,000 more people working over the summer quarter, after showing a loss of almost 600,000 in the spring; it is far noisier, but it is a reminder that "about 50,000 a month" is an estimate with a wide margin.

Jobs added each month and the three-month average. With immigration down, the pace needed to hold unemployment steady is far lower than it was in the 2010s. Full chart and context →

How much each worker produces

Hours rose faster than headcount: total hours worked in the private sector were up about 1.3% at an annual rate in the third quarter, mostly because the average workweek lengthened. The Bureau of Labor Statistics measures productivity for businesses only, with its own output and hours figures, and its hours, which count the self-employed too, barely rose. The oil-reserve entry is government spending, so it falls outside that measure. The stockpiling doesn't: goods made and put on a shelf count as business output in the quarter they are made, and subtract from it when the shelf is drawn down.

Its first estimate for the third quarter, published November 5, put the rise in output per hour at 3.6% at an annual rate, as business output grew 3.7% and hours 0.1%, and at 2.1% from a year earlier. The spring was revised up to 2.1%, from 1.4%. Unit labor costs, what businesses pay in wages and benefits for each unit of output, fell at a 0.7% rate, and were 2.4% higher than a year earlier. Since the end of 2019, output per hour has grown 2.2% a year, against 1.5% in the 2007–19 business cycle. The same report put the share of business income paid to workers at 53.0%, the lowest since the series began in 1947. That is a little above the figure first reported for the spring only because the annual revision of the national accounts raised the recent figures; on the revised numbers, the share fell again in the summer.

GDP growth by quarter, labor productivity from a year earlier, and each quarter's first estimate. Spring 2026 was first reported at 1.5% and is now 2.2%. Full chart and context →

What history says

Output has outrun hiring before. After the 1990–91 recession, sustained job growth lagged the recovery in GDP by about four quarters; after 2001, by about eight. Those were "jobless recoveries," and researchers at the time blamed them mostly on weak growth, which is not today's problem. The closer parallel is the late 1990s, when investment in computers and the internet lifted productivity growth to about 3% a year from 1995 to 2004, double its pace of the previous two decades. What is new in 2026 is the third ingredient: a workforce that has nearly stopped growing.

What people argue about

An investment boom, or a statistical quarter. One side: the data-center buildout is real and large, and if output per hour is rising, it is arriving on the delayed schedule that steam, electricity and the personal computer followed, which what technology does to the numbers traces. The other: half the headline was inventories and an accounting entry, much of the boom is imported, and estimates of how much AI-related investment adds to growth disagree mainly over whether imports are subtracted; a Federal Reserve Board note put the net contribution anywhere from about 0.1 to 1 point a quarter in 2025–26. The first estimate also moves: BEA's advance figures have been revised by 1.3 points on average, up or down, by the time the data settle.

Good news, or a warning. One side: growth that comes from capital and productivity rather than more hours is the best kind, because it can lift pay without lifting prices. The other: real hourly pay has been falling since April and workers' share of business income is at a record low, so growth that doesn't need workers may not reach them either. Hourly pay has the numbers for both readings.

What the Fed should make of it. One side: 3.4% growth with 3.7% inflation is an economy running hot, which is why two of the twelve votes on October 28 were for another rate increase. The other: without inventories the economy grew 1.7%, inflation is mostly oil, and hiring is barely moving, so raising rates would cool a job market that is already still. Both are reading the same quarter.

See also

Sources

Corrections

The weekly digest

One email on Monday mornings: the week's new articles and charts, the numbers that moved most, and what comes out in the week ahead. No ads, and no tracking of opens or clicks.

Past issuesHow addresses are kept