France devoted 64.1 billion euros to research and development in 2024, mobilizing 529,000 people in full-time equivalents, of which 70% are researchers. Yet this sum represents exactly the same share of its national wealth as it did several years ago: 2.18% of GDP. Increasing spending without changing its relative trajectory is running in place while believing you are moving forward.

The essentials

  • Domestic research spending reaches 64.1 billion euros in 2024, up 1.8% in real terms, but effort remains stable at 2.18% of GDP (Ministry of Higher Education, November 2026).
  • Germany devotes 3.1% of its GDP to R&D, South Korea 4.9%: France has not closed this gap in ten years.
  • The increase in euros partially follows economic growth.
  • Diane Coyle argues that without metrics measuring the actual diffusion of innovations into the economy, spending is piloting blindly.
  • The challenge for 2035 is one of transfer: a significant portion of these 64 billion actually modifies the productive capacity of French companies.

64 billion that follow GDP without exceeding it

The 1.8% increase in real terms is genuine. It corresponds to thousands of additional positions, research contracts signed, equipment purchased. An economy in growth mechanically generates more spending in almost every sector. The absolute sum matters less than what it represents in national effort.

At 2.18% of GDP in 2024, France is stagnating. The projection for 2025, 2.19%, confirms this immobility. The Barcelona European Council of March 2002, as part of the Lisbon Strategy, aimed to approach 3% of GDP in R&D by 2010. France is at 2.18% in 2024, though it has surpassed 2.2% of GDP several times since 2010.

This plateau follows simple accounting logic: if R&D spending grows at the same rate as GDP, the ratio remains constant. This is the mechanism being observed. The 64.1 billion euros corresponds to research effort stable at 2.18% of GDP in 2024.

The 529,000 full-time equivalents mobilized constitute a considerable human pool. France trains researchers, finances laboratories, maintains a dense scientific network. But a pool does not automatically transform into competitive advantage if the mechanisms for transfer to the economy remain fragile.

The gap with Germany and South Korea reveals something structural

Germany at 3.1% of GDP and South Korea at 4.9% are not anomalies. These are economies that made an explicit political choice: fund R&D at a level exceeding ordinary growth, over several decades, and build institutions so that this investment flows through industry.

The Franco-German gap is one percentage point of GDP. At France’s 2024 GDP level, reaching 3% of GDP from 2.18% would represent approximately 24 billion euros in additional domestic expenditure on R&D (DIRD). This gap has widened over a generation, and no recent trend reverses it.

The South Korean comparison is even more instructive. South Korea devoted less than 2% of its GDP to R&D in the early 1990s. It deliberately raised this ratio decade after decade, by mobilizing large industrial groups, the chaebols, as vehicles for massive investment. The result is visible: the Korean economy has built a robotic and technological density among the highest in the world, at the cost of real social tensions, but to the benefit of transformed industrial capacity.

France does not start from the same historical point or the same industrial model. But the comparison shows that an R&D ratio can be deliberately piloted over the long term. It is not simply the passive reflection of economic structure.

Euros do not measure the diffusion of innovations

British economist Diane Coyle offers useful insight in The Measure of Progress. She argues that standard economic indicators—GDP, productivity, aggregate spending—are structurally blind to the quality of transformations. An economy can spend more on R&D and produce more patents without these activities actually modifying the productive capacity of its companies. These indicators measure input, not output.

Applied to French R&D, the diagnosis is precise. The 64.1 billion euro spending is a measure of activity: it counts functioning laboratories, working researchers, circulating credits. It does not measure diffusion: how many innovations from public research have been adopted by industrial SMEs, how many technology transfers have modified production processes, how many patents have generated lasting economic value.

This metrological limitation has direct political consequences. If policymakers steer research policy solely using the spending-to-GDP ratio, they know if the budget is growing. They do not know if it is transforming. It is somewhat like measuring the effectiveness of an education system only by the number of class hours taught, without ever examining what students retain.

Coyle’s critique calls for complementary indicators. Adoption rate of innovations in mid-sized companies, share of exports based on high-technology-intensive products, average time between academic discovery and first industrial application: these metrics exist partially. They do not yet structure French public debate on research.

The Research Tax Credit finances activity, not necessarily transfer

The structure of the CIR (Crédit Impôt Recherche) creates another barely visible effect: it rewards declared spending, not results obtained. The CIR does not distinguish eligible spending based on the existence of an industrial outlet, subject to its basis and rate rules. This design does not distinguish between activity and transformation. It incentivizes declaring research spending. This declarative logic overlaps with the measurement blind spot identified by Coyle: the main fiscal tool is not conditioned on the diffusion of R&D results.

The CIR is a major lever for supporting private R&D whose official objective is to increase research spending, even if its economic effectiveness remains debated.

France has the most generous private R&D fiscal support scheme in the OECD. The Research Tax Credit (CIR) represents approximately 7 billion euros in annual tax expenditure, a tax reduction granted to companies that declare research spending. The initial objective: to encourage private companies to invest more in R&D.

The assessment is mixed. The CIR has indeed increased declared private R&D spending. But several evaluations, including those conducted by France Stratégie, raise a persistent question: a significant portion of the scheme benefits very large companies that would have invested anyway, and less to SMEs and mid-sized companies that need it most to cross a technological threshold.

Philippe Aghion, whose work on Schumpeterian growth illuminates innovation dynamics, defended a position different from Coyle’s on one specific point. For him, R&D spending is itself an engine of productive transformation provided that the competitive environment pushes companies to genuinely innovate. Without this competitive pressure, fiscal incentives can finance research without creative destruction or productivity gains.

The two readings complement each other. Coyle emphasizes the absence of metrics on diffusion. Aghion indicates that diffusion requires a market that rewards innovation. Large companies benefit most from the CIR and the OECD recommends better targeting support to increase its additionality.

French public buyers illustrate this same transfer blockage: despite existing schemes, public procurement remains largely indifferent to innovative offerings, depriving young technology companies of a structuring outlet.

What works and what remains to be built

French public research produces world-class scientific results. The National Center for Scientific Research remains one of the most-cited research organizations in the world. Major schools and universities feed a pool of talent in mathematics, physics and engineering sciences recognized internationally.

Several recent initiatives attempt to narrow the gap between laboratory and market. France 2030 has directed billions toward priority sectors—hydrogen, semiconductors, health—with public-private co-financing logic. The Carnot institutes, created to facilitate partnerships between public laboratories and companies, have expanded their scope. A few regional clusters, around Grenoble for microelectronics, around Toulouse for aeronautics, show that dense research-industry integration is possible on French soil.

These partial successes have a common feature: they rest on strong institutional proximity between researchers and companies, built over several decades. They remain geographically concentrated and sectorially limited. The question of open diffusion among researchers, already tense at European scale, further heightens the risk that research gains remain captured by already best-equipped actors.

The challenge by 2035 is less about reaching 3% of GDP, a desirable but insufficient goal on its own, than about knowing whether invested euros actually modify productive structure. According to the OECD, France has fewer mid-sized companies than Germany, which notably limits integration into global value chains, and this structural difference weighs on the productivity gap.

Steering what you truly measure

This governance limitation is not unique to France. Most advanced economies steer their research policy with input indicators because they are easier to collect and compare. The difficulty lies in the nature of transfer itself: it operates over long timeframes, involves heterogeneous actors and produces diffuse effects in productive structure. Building reliable metrics for diffusion requires statistical infrastructure that few countries have yet deployed nationally. Acknowledging this difficulty does not exempt one from working on it.

It simply means that the measurement challenge is itself a distinct public policy problem, not a technical detail resolvable at the margins of the budget envelope debate.

Stability at 2.18% of GDP is not a catastrophe. France retains solid scientific capacity, significant public funding and pockets of industrial excellence. But this stability reveals a governance limit: research policy relies significantly on input indicators rather than transformation indicators.

Increasing the spending-to-GDP ratio should be a target. But if this increase is not accompanied by metrics measuring actual diffusion—industrial adoption, technology transfers, productivity of benefiting sectors—it will remain a dashboard objective, not a transformation lever.

The practical decision to make is known: build and publish annually transfer indicators alongside spending figures. Not to replace R&D accounting, but to make visible what it produces. Only at that condition do 64 billion become a steerable starting point, and not a reassuring number.


Sources

  1. Ministry of Higher Education, Research and Space, Expenditure on research and experimental development in France, detailed results for 2024 and estimates for 2025, SIES, November 2026, https://www.enseignementsup-recherche.gouv.fr/fr/depenses-de-recherche-et-developpement-experimental-en-france-resultats-detailles-pour-2024-et-104571
  2. Diane Coyle, The Measure of Progress: Counting What Really Matters, Princeton University Press, https://press.princeton.edu/books/hardcover/9780691253336/the-measure-of-progress
  3. OECD, Economic Surveys: France 2026, Organisation for Economic Co-operation and Development
  4. France Stratégie, evaluations of the Research Tax Credit (reports published since 2019)
  5. Philippe Aghion et al., work on Schumpeterian growth and innovation policy