🔍 Read the full analysis: Capital, Labour, And The Question Of AI’s Rewards on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
The U.S. Bureau of Labor Statistics reported that labor received 53.0% of value produced in the nonfarm business sector in the third quarter of 2026, the lowest share in its series, which began in 1947. The figures describe how income is divided between workers and capital owners; they do not establish that AI caused the decline or show how much of it is due to any single factor.
The U.S. Bureau of Labor Statistics reported that workers received 53.0% of value produced in the nonfarm business sector in the third quarter of 2026, the lowest share since the series began in 1947. The figure sharpens a long-running question about how economic gains are divided between labor and capital, as debate continues over whether technologies such as artificial intelligence could affect workers’ share.
The BLS published the quarterly productivity and costs report on November 5, 2026. Its measure covers the nonfarm business sector, which accounts for about three quarters of the U.S. economy, and compares compensation for work with gross value added. Compensation includes wages, salaries and benefits such as employer contributions to pensions and health coverage.
The reported share was 53.0% in Q3 2026. The source says the revised second-quarter figure was about 53.4%. The share stood at about 66% at the end of 1960 and around 63% in 2000, according to the historical figures cited in the source. The movement is a change in the division of measured income, not a direct measure of individual workers’ pay or household living standards.
The measure assigns the remainder to capital-related income, including profits, interest, rents and depreciation. Because self-employed people do not receive wages in the same way as employees, the BLS estimates the labor portion of their income using average employee pay per hour in their sector. That treatment is one reason the measure depends on statistical definitions and estimation.
Infographic 01 · The number to remember
53 cents of every dollar
How much of the value created in US businesses goes to the people who work there, 1947 to 2026.
US labour share · nonfarm business · % of value added (annual averages unless noted)
BLS publishes the series as an index (2017 = 100) and the latest level (53.0% in Q3 2026). The levels shown here are derived by scaling the index to that published level; they include an estimate of the labour income of the self-employed and use gross value added, so the capital side includes depreciation. Bars start at zero.
Sources: BLS, Productivity and Costs, Q3 2026 (5 Nov 2026); BLS labor share series PRS85006173 (via FRED).
Powered by Thorsten Meyer AI
The labor share provides one way to track whether the value produced by businesses is flowing more toward worker compensation or capital income. A decline can matter at an economy-wide level because even a percentage point represents a large sum when applied to the output of the nonfarm business sector. It does not, by itself, establish that every worker is earning less, that profits rose by the same amount, or that households experienced a particular change in income.
The result also offers a reference point for discussion of AI and automation. If technology enables firms to produce more with fewer workers or shifts bargaining power, it could affect how gains are distributed. But the third-quarter figure alone cannot identify AI’s role, distinguish it from other economic forces, or show whether workers in particular industries are affected more than others.
Popular right now
Dyson V9 Motorbar Cordless Vacuum
Roborock Q7 M5+ Robot Vacuum and Mop
Dyson V15 Detect Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
A Long-Running Distribution Debate
The division of income among workers, landowners and owners of capital has been an economic question for centuries. In 1817, economist David Ricardo described output as divided into rent, profit and wages, and called identifying the laws governing that distribution the principal problem of political economy.
For much of the postwar period, the U.S. labor share appeared comparatively stable. The source notes that John Maynard Keynes described the stability in 1939 as one of the best-established facts in economic statistics, while economist Nicholas Kaldor later included stable income shares among the patterns growth theories needed to explain. The U.S. series began in 1947, and the cited figures show the share near two-thirds for decades before declining after 2000.
The global trend is similar in direction but uses a different data series. The International Labour Organization estimated that labor’s share of global income fell 1.6 percentage points from 2004 to 2024, including 0.6 points after 2019. It said the decline represented $2.4 trillion in labor income in 2024, measured in constant purchasing-power dollars. Its 2026 trends report placed the global share at 52.6% in 2025, below the 2019 estimate of 53.0%. Those international figures should not be treated as directly interchangeable with the BLS’s U.S. measure.
““To determine the laws which regulate this distribution, is the principal problem in Political Economy.””
— David Ricardo, in 1817
What the Data Cannot Establish
The BLS figure establishes the reported share for the sector and quarter, but it does not explain why the share fell. The source does not provide a breakdown attributing the latest movement to AI, automation, changes in prices, industry composition, bargaining power or other causes. The ILO has reported evidence compatible with automation-oriented technological progress contributing to falling labor income shares in data from 2003 to 2019, but that finding does not establish that AI caused the U.S. Q3 2026 result.
The headline figure is also not a complete account of workers’ economic circumstances. It does not show changes in real wages, employment, hours worked or the distribution of pay among workers. The BLS estimate includes an imputation for self-employed workers, and the source identifies the second-quarter figure as revised. More detail would be needed to assess how revisions or measurement choices affect comparisons over time.
Upcoming Data and Further Evidence
The next step is to compare the Q3 result with subsequent BLS productivity and costs releases, which can show whether the latest reading persists or is revised. Future reports may also provide a clearer picture of how compensation, output and productivity are moving together.
To evaluate whether AI is changing the division of economic gains, analysts would need evidence connecting adoption in specific industries to productivity, employment, wages and profits over time. The current figures set out the distributional trend that motivates the question; they do not settle the cause or predict how AI’s gains will be shared.
Key Questions
What does a 53.0% labor share mean?
It means compensation for labor was measured as 53.0% of gross value added in the U.S. nonfarm business sector in Q3 2026. The measure includes wages, salaries and benefits, and estimates the labor portion of self-employed income.
Is this the lowest labor share ever recorded?
It is the lowest in the BLS series cited by the source, which begins in 1947. That claim applies to this measure and its recorded period, not necessarily to every estimate of labor’s share across all historical data.
Does the report show that AI caused the decline?
No. The reported figure does not identify AI as a cause. The source cites ILO research finding evidence compatible with automation-oriented technological progress contributing to declines in global labor income shares in 2003–2019, but that is not proof that AI caused the U.S. result in Q3 2026.
How does the U.S. figure compare with the global estimate?
The source puts the U.S. nonfarm business labor share at 53.0% in Q3 2026 and the ILO’s global labor income share at 52.6% in 2025. The figures cover different populations, periods and statistical methods, so they are not directly comparable.
Source: ThorstenMeyerAI.com
