Labor's Share of Output Hits 53%, Lowest Since 1947

Labor's share of nonfarm business output fell to 53% in the third quarter, the lowest reading in records going back to 1947, according to the US government productivity report released on a Thursday last week. One day later, Friday's jobs report showed unemployment falling to 4.1% from 4.2%. The two headline numbers looked strong together. Read side by side, they describe an economy in which workers produce more and capture less of what they produce than at any point in nearly 80 years.

The number matters because it measures the split of everything businesses produce between the people who work and the capital that employs them. Labor share counts how much of output goes to workers as wages, salaries and benefits. Picture a company with $1 million in output. At a 53% labor share, $530,000 goes to employees. The other $470,000 goes to capital and to costs that are not labor. That residual share is now the largest in the postwar record relative to labor's.

What did the two reports actually say?

The Friday jobs report carried three strong figures:

• Joblessness fell to 4.1% from 4.2%.
• Household employment rose by roughly 463,000.
• The employment-population ratio rose to 59.3%.

The Thursday productivity report was the bigger story. It showed:

• Productivity, output per hour of work, rose 3.6%.
• Hourly compensation rose 2.9%.
• Unit labor costs — what businesses spend on labor for each unit produced — fell 0.7%.
• Output grew at a 3.7% annualized rate while hours worked barely moved.

Because pay grew more slowly than output per hour, businesses spent less on labor for every unit they produced. Workers got more productive. Each unit of that productivity returned less to them in compensation. The gap between the two growth rates, 3.6% against 2.9%, is the arithmetic behind the falling labor share.

Where did the extra money go?

The extra money shows up in profits. Corporate profits reached about $4.7 trillion at an annual rate in the second quarter, up 20.8% from a year earlier. Paychecks moved the other way. Average hourly earnings, adjusted for inflation, slipped from $11.38 in January to $11.26 in September, measured in constant dollars. Wages and salaries fell to 42% of gross domestic income in 2025, the lowest reading in the past decade.

The profit figure and the wage figure tell one consistent story across two different datasets. Output per hour accelerated. Compensation per hour did not keep pace. The difference accumulated in corporate margins, which expanded at a rate not seen in the postwar distributional record.

The composition of the gains compounds the distributional effect. When labor's share shrinks, more of the money goes to shareholders, and shareholders are concentrated at the top. The top 1% of households own 50.9% of corporate equities and mutual fund shares held outside 401(k)-style plans, according to the Federal Reserve. The bottom half of households owns 0.6%. A record-low labor share moves income toward the top of the wealth ladder, because the households receiving capital income are overwhelmingly the households that already hold the most wealth.

Is AI the reason?

The roots of the shift go back to the labor shortage of 2022, according to Tom Barkin, president of the Richmond Fed. Companies short on workers at that time put money into automation and new operating models, Barkin told Bloomberg's Odd Lots podcast in August, and they are collecting the return on that investment now.

AI's effect, in Barkin's account, operates mostly through hiring. Executives are deciding, in his words, to "see if you can't fill the job using AI first and then we'll hire later." That sequencing — automation first, headcount second — would show up in the data exactly as it did in the third quarter: output rising at 3.7% annualized while hours worked barely moved.

The claim has a credible skeptic. Economist Adam Posen, appearing on the same show, pushed back and linked slow hiring partly to the reshuffling of jobs after Covid. "You can't automatically say the lack of hiring is AI," Posen said. One quarter is too short to prove a permanent structural shift in the economy's labor-capital split.

The caveat is real, and the direction of the trend is nonetheless unambiguous across the data. Businesses are learning to grow output faster than they add workers. Whether the cause is AI-driven automation, post-pandemic job reshuffling, or the delayed return on 2022's automation investments, the measured result in the third quarter was the same: a 79-year low in labor's share.

Why should investors and policymakers care?

The risk for investors is that workers are also customers. Fatter margins hold up only as long as the people buying the products keep spending. Signs of strain may already be showing. The personal savings rate fell to 4.1% in the third quarter, down from 5.3% in the first. That decline suggests households are dipping into savings to maintain their spending.

The mechanism is straightforward and self-limiting. Companies suppress labor costs through automation and restrained hiring. Households, whose income is mostly labor income, sustain demand by drawing down savings rather than by earning more. Profits rise. The savings buffer that props up the profits shrinks.

Two signals matter over the next quarter or two. The first is whether unit labor costs keep falling in the revised third-quarter data and the first fourth-quarter productivity release. The second is whether the savings rate keeps dropping. If both trends continue, profits stay strong for now while consumer demand rests on thinning household savings.

What happens if both trends hold?

That mix — falling labor costs paired with falling savings — can carry corporate earnings into 2027, according to the analysis, but only while households keep spending what they used to save. The condition is not permanent. At some point savings stop falling, and demand must then be supported by income rather than by drawdown.

For the AI industry, the stakes run in both directions. Barkin's account implies that automation investment, including AI, is already measurable in the national accounts as a shift from labor compensation to capital income. Posen's caution implies that attributing the entire shift to AI overstates the technology's current reach. The third-quarter data cannot settle the argument on its own.

The revised productivity figures and the next savings-rate reading will carry more evidence. A repeat of 53% or lower on labor share, with unit labor costs still falling, would strengthen the structural case. A rebound in hours worked or in hourly compensation would strengthen the post-Covid reshuffling explanation. Either way, the 1947-2025 record now shows labor's slice of the economic pie at its thinnest point in the entire series.