Productivity Growth
US labour productivity grew at 2.8% per year from 1947 to 1973. From 2005 to 2024, it has averaged 1.4% — half the post-war rate. Every other advanced economy shows a similar pattern. Whether the slowdown is a measurement issue, a temporary pause, or a new normal is one of the most-debated questions in economics.
Key insights
The slowdown is widespread
OECD-wide labour productivity growth has roughly halved since the 1995–2005 IT-boom era. The US, UK, France, Germany, Italy, Japan, South Korea all show the same pattern. Aggregate growth in the OECD has fallen from ~2% per year to ~1%. The decline cannot be blamed on one country, one sector, or one shock.
Multiple candidate explanations
Possibilities include: (1) measurement — digital goods and free services are mis-counted, true productivity is higher than headline; (2) the easy gains from electrification, internal combustion and IT are exhausted; (3) declining business dynamism — fewer startups, less labour-market churn; (4) declining R&D productivity — more researchers required to produce each new idea (Bloom et al. 2020); (5) aging populations and the secular shift to services.
AI is the leading re-acceleration bet
Generative AI has produced the first plausible case for a productivity re-acceleration since the IT boom. Early studies on coding (GitHub Copilot), customer service, writing and basic legal tasks show 20–40% time savings on specific tasks. Aggregate effect remains uncertain — productivity gains require capital deepening, organisational change and complementary investment to show up in national statistics. Historical precedent (the dynamo, the computer) suggests 10–20 year lags.
US labour productivity growth — five-year averages
% per year, real output per hour, nonfarm business sector
Show data table
| % per year | |
|---|---|
| 1947–73 | 2.8 |
| 1973–95 | 1.5 |
| 1995–2005 | 2.9 |
| 2005–10 | 1.5 |
| 2010–15 | 0.9 |
| 2015–20 | 1.2 |
| 2020–24 | 1.4 |
Key Finding: Four eras: post-war boom (1947–73), slowdown (1973–95), IT revival (1995–2005), and the current sustained low (2005–today).
Productivity growth — selected countries (2010–2024)
Average annual % growth in labour productivity
Show data table
| % per year | |
|---|---|
| South Korea | 2 |
| USA | 1.3 |
| Germany | 0.8 |
| France | 0.7 |
| Japan | 0.6 |
| Canada | 0.5 |
| Italy | 0.3 |
| UK | 0.5 |
Key Finding: South Korea and the US lead; Italy, Canada and the UK have averaged about 0.5% or less.
Methodology & caveats
Labour productivity vs TFP
Labour productivity = output per hour worked. Total Factor Productivity = output per unit of combined labour and capital input. TFP is the residual that captures 'everything else' — technology, organisation, skill, market structure. Labour productivity can rise even if TFP is flat (more capital per worker); the productivity slowdown debate is largely about TFP.
Measurement issues
Digital goods (free search, free maps, social media) generate consumer surplus but little measured output. Services productivity is hard to measure (what's the 'output' of healthcare or legal services?). Hedonic adjustments for IT goods have grown harder as products integrate hardware and software. Brynjolfsson, Eggers and Gannamaneni (2018) estimated mis-measurement at ~0.5pp/year — meaningful but not the whole story.
Why it matters
Productivity growth is the single most important determinant of long-run living standards. A 1pp difference in annual productivity growth compounded over 50 years produces a 65% gap in income. The post-2005 slowdown, if permanent, means each generation's gain in living standards would be much smaller than the last one's.
Why the numbers vary
The US figure is output per hour in the nonfarm business sector, which excludes government and non-profits. OECD comparisons usually use GDP per hour worked for the whole economy, which is lower and harder to measure. Hours differ too: surveys and administrative data give different counts, and the pandemic shifted the mix of workers, so 2020-21 productivity jumped because low-wage jobs vanished, not because anyone worked better. Revisions are large, and a five-year average is a safer read than a single quarter.
What would change the picture
Productivity data would show AI's effect as a sustained rise above about 2% a year in the US, the IT-era pace, alongside faster total factor productivity. A single strong year proves little. In Europe the debate is different: the gap with the US since 2005 lies largely in technology-producing sectors, so energy prices and market size matter as much as AI adoption.
Cite this page
GlobeChart. “Productivity Growth.” Last updated 2026-10-31. https://globechart.com/economy/productivity-growth/
Please also credit the original publisher named under each chart: BLS Productivity and Costs; OECD Productivity Statistics. Spotted an error? Tell us.