Since its first edition in 2006, the OECD Compendium of Productivity Indicators has shed light on short- and long-term trends in labour and multifactor productivity across OECD countries and where possible accession countries over the period 1995-2023. It has also examined trends in key components of productivity, namely capital and labour inputs. Beyond economy-wide developments, the publication also offers granular insights at industry and firm level. In addition, it provides insights into productivity developments in 2024 and presents two new analyses on multi-factor productivity by industry and on cyclical adjustment of multi-factor productivity in manufacturing.
OECD Compendium of Productivity Indicators 2025
In 2023, labour productivity – measured as GDP per hour worked – for the total economy averaged across all OECD countries saw a modest increase of 0.6%. However, it fell by 0.9% in the euro area – the largest drop since 2009 – continuing a downward trajectory observed since 2021. In contrast, labour productivity increased by 1.6% in the United States, close to the pace seen in 2019 (Figure 1.1).
The subdued labour productivity growth in 2023 unfolded against the backdrop of a challenging macroeconomic environment. The global economy slowed markedly to 1.7% in 2023 from 3% in 2022 amid headwinds, such as tighter financial conditions and heightened geopolitical tensions. Despite modest economic growth, labour markets remained resilient. Hours worked grew by 1.1% on average across OECD countries, mostly through an increase in employment.
At the same time, globalisation appeared to have lost momentum, with global trade growth weakening notably (OECD, 2024[1]). Since trade and foreign direct investments play a vital role in facilitating knowledge diffusion and driving innovation, a weakening of global economic integration is likely to weigh on future productivity performance (Aiyar et al., 2023[2]).
In addition, real interest rates remained elevated in 2023 in most countries (OECD, 2024[1]). Restrictive financial conditions in most major economies, with high lending rates and tight credit standards, continued to dampen firms’ investment and economic activity. Elevated borrowing costs tend to raise operational expenses and create uncertainty around the interest rate outlook, making it harder for firms to commit to long-term investments (Duquerroy, Istrefi and Mouabbi, 2024[3]). This, in turn, can slow innovation and impede productivity growth, as constrained access to credit limits firms’ capacity to adopt new technologies and invest in efficiency-enhancing measures.
Labour productivity growth is estimated to have stagnated at around 0.4% in 2024 on average across OECD countries excluding Türkiye (Chapter 9). However, these estimates are subject to significant uncertainty. Labour productivity growth is expected to have improved modestly in OECD Asian countries, averaging 1.8% in 2024. In contrast, productivity growth would have remained stagnant in OECD South American countries and several European economies outside the euro area. Meanwhile, labour productivity growth is expected to have declined slightly in both the euro area and North America.
GDP per hour worked, Per cent
Note: Official data on labour productivity data across OECD countries are only available from 2011 onwards due to the unavailability of data on hours worked for Korea prior to that year, following a methodology revision by Statistics Korea in 2017. Data of this series for years before 2011 are based on estimates using past vintage data.
Source: OECD Productivity Database (2025).
The OECD average labour productivity growth masks significant variation across economies in 2023. It was positive in about half of the countries, and negative in the other half, with varying magnitudes. Strong labour productivity gains were seen in a number of non-EU countries, contributing to the positive growth in the OECD average. Costa Rica, Latvia and Romania were the top performers, whereas Ireland, Estonia, and Luxembourg experienced the steepest declines (Figure 1.2).
In 2023, economic activity in the euro area, and particularly Central and Eastern European countries, was notably weak, with GDP growth slowing to 0.4% from 3.5% in 2022, reflecting high energy prices and other headwinds (Chapter 2). On the contrary, growth remained robust in the United States and Japan experienced above-trend growth. Although the overall economy slowed down in the euro area, employment rates stayed robust, and total hours worked in this region rose by 1.6% in 2023. Labour hoarding, where firms retain employees despite weak demand, contributed to a decline in labour productivity in many countries (ECB, 2024[4]).
GDP per hour worked, Total economy, Per cent
Source: OECD Productivity Database (2025).
As an alternative measure of productivity, multifactor productivity (MFP) growth captures changes in economic performance beyond those explained by capital and labour. MFP growth either stagnated or even turned negative in most OECD countries in 2023 (Figure 1.3). The slowdown was particularly pronounced in Luxembourg and Austria, where MFP fell by around -2% in 2023. By contrast, MFP grew fast in Slovak Republic and Slovenia, with growth rates of 4.6% and 2%, respectively in the same period.
Total economy, Per cent
Source: OECD Productivity Database (2025).
The slowdown in MFP growth was the main factor behind the weak labour productivity performance. The dominance of MFP contribution over capital deepening, defined as capital services per hour worked, in explaining labour productivity growth has been evident in the past for long (OECD/APO, 2022[5]). While the contribution of capital deepening was negligible in some countries, it was negative in many others at the total economy level (Chapter 2).
This variation in the role of capital deepening across countries in 2023 partly reflects differences in investment patterns – an essential component in measuring capital services. Investment rates (i.e. investment over GDP) varied across OECD countries in this period. Around half of the OECD countries experienced an increase in investment rates, while the other half saw stagnation or even a decline. This variation likely reflects differences in how firms and households responded to heightened uncertainty, rising costs, and tighter credit conditions.
Investment rates also varied across asset types. Much of the increase was concentrated in construction-related assets, such as other buildings and structures, and in non-ICT capital goods, including other machinery, equipment, and weapons systems. By contrast, investment in ICT equipment and intellectual property products (IPP), key enablers of innovation and digital transformation, declined in most countries (Chapter 4).
Over time, capital deepening has shown a declining contribution to labour productivity growth (Chapter 6). This can be attributed in part to a slowdown in net investment since the global financial crisis. The overall weakness in investment is likely driven by subdued spending on non-digital tangible assets, coupled with rising depreciation (OECD, 2025[6]).
In most OECD countries, labour productivity growth in 2023 was primarily shaped by changes within industries. The cross-industry reallocation effect, reflecting shifts in total hours worked between industries, played a minimal role in the same year (Figure 1.4). In about half of the countries with available data, the within-industry contribution was negative or near zero, pulling down overall productivity growth (Chapter 3).
Knowledge-intensive industries, including information and communication and finance, recorded negative productivity growth in many countries in 2023. Manufacturing made the strongest positive contribution to overall labour productivity growth in several countries, including Slovak Republic and Denmark. Meanwhile, energy-related activities, comprising electricity, gas, steam and air conditioning supply, had a strong negative impact on economy-wide labour productivity growth in Greece and Croatia (Chapter 3).
While the overall impact of cross-industry reallocation was limited, 2023 saw a relative increase in the share of hours worked in low-productivity sectors, such as transportation, accommodation, and food services (see Chapter 3). This shift partly reflects the easing of labour shortages, which had persisted since the COVID crisis. In several OECD countries, quit rates slowed and job vacancies decreased in 2023 (OECD, 2024[1]).
Large firms in OECD countries consistently exhibit higher labour productivity on average compared to their smaller counterparts (Chapter 7). Indeed, large firms (those with more than 250 employees) tend to invest more heavily in information and communication technology (ICT) capital and digitalisation and account for a signification share of innovation (Garcia-Masia, Hsieh and Klenow, 2019[7]; Lopez-Garcia et al., 2024[8]) and trade more than smaller firms. In manufacturing, the productivity gap between large and smaller firms is on average more pronounced than in the business economy as a whole, reflecting returns to scale in capital-intensive activities.
Percentage points
Note: The static reallocation effect refers to the reallocations of hours worked between industries with different productivity levels. The dynamic reallocation effect accounts for the reallocations of hours worked between industries with different productivity growth rates. The sum of the two makes up the cross-industry reallocation effect. The within-industry effect reflects labour productivity developments that are not the result of reallocation of hours worked between industries. It is measured by the labour productivity growth in each industry weighted by the industry share in total value added (Chapter 5). For United States, these three effects do not fully sum to the labour productivity growth due to the use of Fisher volume indices for GVA. However, the residual is typically small, especially when averaged over several years (see the methodological note).
Source: Authors’ calculation based on OECD Productivity Database (2025).
Divergent trends in labour productivity growth are usually shaped by a combination of factors, including short-term cyclical factors, i.e. differences in the position of individual economies within the economic cycle, and structural aspects. In many euro area countries, both labour productivity and MFP growth tend to be highly procyclical, being low in times of low output growth and high when the economy rebounds. In contrast, countries such as the United States and Australia exhibit less procyclical behaviour, as labour is more flexibly adjusted to production (Dossche, Gazzani and Lewis, 2023[9]).
Yet, evidence suggests that cyclical factors had limited effects on productivity developments across OECD countries in 2023, at least in manufacturing (Chapter 7). Instead, variation in productivity performance was primarily shaped by structural factors, including differences in the business environment and regulatory framework.
Looking forward, productivity growth will be shaped by both ongoing challenges and transformative opportunities. On the one hand, geopolitical tensions and economic uncertainty appear likely to reduce investment as firms delay or scale back investments, leading to slower technological innovation and adoption and decreased overall efficiency in production processes.
On the other hand, new advances in Artificial Intelligence (AI), particularly Generative AI, could offer promising avenues for boosting innovation and driving future productivity growth. Yet, the potential productivity gains from AI come with notable challenges. Widespread adoption of AI will require new skillsets, placing pressure on both workers and employers to adapt. Unresolved issues around intellectual property and data ownership further complicate the regulatory landscape and may hinder innovation and diffusion (OECD, 2023[10]). Beyond economics, AI deployment also presents broader social, environmental and national security challenges (OECD, 2022[11]; OECD, 2024[12]). The energy-intensive nature of AI computing can lead to increased greenhouse gas emissions, potentially offsetting some of the productivity gains (OECD, 2023[13]). Reaping the benefit of AI will require complementary investments and institutional adaptation and productivity gains may only become visible over time (see Box 1.1).
Although Artificial Intelligence (AI) has been studied since the 1950s, only in recent years has it begun to transform economies and societies at scale. This shift has been driven by major advances in computing power, machine learning techniques, and the availability of large datasets. The emergence of generative AI, particularly following the release of tools like ChatGPT and Gemini in early 2022, has further accelerated the integration of AI across many areas of work and daily life.
There is no universally-agreed definition of AI
Different definitions of AI have been adopted across communities, commonly addressing AI as technologies that perform tasks by mimicking human-like senses, learning, and actions. For instance, OECD adopted a shared definition of AI, which is used in policymaking and legislative work (OECD, 2024[12]). According to this definition, AI refers to machine-based systems that, for explicit or implicit objectives, process input data and generate outputs, such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. These systems require intangible inputs like software, data, and human expertise with substantial computing capacity and complementary technologies. The statistical definition of AI, introduced in the System of National Account 2025, is different and defines AI as “a computer program operating a system capable of recognition, reasoning, communication, and prediction, simulating human recognition, reasoning, and communication”.
Quantifying the impact of macro-economic AI in the long run is challenging
Measuring the impact of AI on productivity presents significant challenges, particularly at the macroeconomic level. Firstly, AI amplifies traditional measurement challenges such as capturing intangible and public sector production in productivity measures. Without progress in these areas, assessing the effects of AI-related policies will remain difficult. Secondly, while micro-level studies or sector-specific analyses point to some evidence of productivity gains, those gains are harder to identify at the macroeconomic level. The complexity is compounded by the intangible nature of AI investments, such as data, knowledge, skills, or computing power and capacity, which are poorly captured in the current national accounts. Additionally, much like previous technological transformations, such as the rise of the internet, AI’s broader economic benefits often depend on complementary investments and institutional adaptation and may only become visible over time. This delayed impact, coupled with the diversity in how AI systems are designed and deployed, adds further complexity to measurement. Finally, current research also focuses mainly on advanced economies, neglecting international spillovers.
Early estimates of AI’s impact on productivity
Estimates of AI’s contribution to aggregate productivity growth vary, depending on the modelling framework and underlying assumptions. AI is estimated to raise labour productivity growth by between 0.5 and 3.5 percentage points per year over a ten-year horizon (Figure 1.5). Based on a micro-macro framework which combines micro-level performance gains estimates with exposure of activities to AI and likely future adoption rates, recent OECD modelling suggests that AI’s contribution to productivity growth in the G7 will range from 0.2 to 1.3 percentage points, with considerable diversity across countries (Filippucci et al., 2025, forthcoming[14]).
Percentage points
Note: The solid bar refers to the lower bound while the striped bar corresponds to the upper bound of the estimate. In cases where modelling predictions primarily focus on TFP, labour productivity is obtained using simple assumptions about the aggregate capital multiplier (Acemoglu, 2024[15]; Aghion and Bunel, 2024[16]; Bergeaud, 2024[17]). The estimates refer to the countries shown in brackets.
Source: (Filippucci, Gal and Schief, 2024[18]) (Figure A.10.)
OECD National Accounts Statistics (database), https://doi.org/10.1787/na-data-en.
OECD Labour Market Statistics (database), https://doi.org/10.1787/data-00046-en.
OECD Productivity Statistics (database), http://data-explorer.oecd.org/s/1v8.
OECD Productivity Dashboard, www.oecd.org/en/data/dashboards/oecd-dashboard-of-productivity-indicators.html.
References
[15] Acemoglu, D. (2024), “The Simple Macroeconomics of Artificial Intelligence”, Economic Policy, https://economics.mit.edu/sites/default/files/2024-04/The%20Simple%20Macroeconomics%20of%20AI.pdf.
[16] Aghion, P. and S. Bunel (2024), “AI and Growth: Where Do We Stand?”, Policy note.
[2] Aiyar, S. et al. (2023), Geoeconomic Fragmentation and the Future of Multilateralism, International Monetary Fund.
[17] Bergeaud, A. (2024), “The Past, Present and Future of European Productivity”, Paper prepared for the ECB Forum on Central Banking “Monetary policy in an era of transformation”.
[9] Dossche, M., A. Gazzani and V. Lewis (2023), “Labor adjustment and productivity in the OECD”, Review of Economic Dynamics, https://doi.org/10.1016/j.red.2021.11.006.
[3] Duquerroy, A., K. Istrefi and S. Mouabbi (2024), “Interest rate uncertainty and firm decisions”, Banque de France, https://www.banque-france.fr/system/files/2024-01/WP940_0.pdf.
[4] ECB (2024), Economic, financial and monetary developments, Economic Bulletin, https://www.ecb.europa.eu/press/economic-bulletin/html/eb202404.en.html.
[14] Filippucci, F. et al. (2025, forthcoming), “Macroeconomic productivity gains from Artificial Intelligence in G7 economies”, OECD Working Papers.
[18] Filippucci, F., P. Gal and M. Schief (2024), “Miracle or Myth? Assessing the macroeconomic productivity gains from Artificial Intelligence”, OECD Artificial Intelligence Papers, No. 29, OECD Publishing, Paris, https://doi.org/10.1787/b524a072-en.
[7] Garcia-Masia, D., C. Hsieh and P. Klenow (2019), “How Destructive is Innovation”, Econometrica, Vol. 87, https://doi.org/10.3982/ECTA14930.
[8] Lopez-Garcia, P. et al. (2024), ECB Economic Bulletin, Issue 2/2024, ECB.
[6] OECD (2025), OECD Economic Outlook, Volume 2025 Issue 1: Tackling Uncertainty, Reviving Growth, OECD Publishing, https://doi.org/10.1787/83363382-en.
[12] OECD (2024), “Explanatory memorandum on the updated OECD definition of an AI system”, OECD Artificial Intelligence Papers, No. 8, https://doi.org/10.1787/623da898-en.
[1] OECD (2024), OECD Economic Outlook, OECD Publishing, https://doi.org/10.1787/69a0c310-en.
[10] OECD (2023), “Intellectual Property and Artificial Intelligence: A Literature Review”, OECD Science, Technology and Industry Working Papers, https://doi.org/10.1787/794b33ef-en.
[13] OECD (2023), “The Climate and Energy Implications of Artificial Intelligence”, OECD Science, Technology and Industry Policy Papers, https://doi.org/10.1787/73cbede6-en.
[11] OECD (2022), “The OECD Going Digital Measurement Roadmap”, OECD Digital Economy Papers, Vol. 328, https://doi.org/10.1787/bd10100f-en.
[5] OECD/APO (2022), Identifying the Main Drivers of Productivity Growth: A Literature Review, OECD Publishing, Paris, https://doi.org/10.1787/00435b80-en.