This chapter examines the role of public policies as drivers of knowledge creation and innovation for sustainable growth. It puts the focus on the role of positive financial incentives as complement to policy stringency measures, providing a comprehensive set of available and new measures of public support. It provides insights into the size, nature and direction of support for energy and environmental resilience and transformative policy goals, as well as new analysis of the relationship between support and environmental innovation outcomes.
Measuring Science and Innovation for Sustainable Growth
Abstract
Public policies are key drivers of new knowledge and innovation for sustainable growth. This chapter focuses on measures of government financial support for research and development (R&D) and innovation-related activities that positively induce investment. While measures of environmental policy stringency, such as taxes or standards intended to achieve environmental goals, have increased steadily across OECD countries since 1990 – for several years – these have not been complemented by equivalent growth for government support for research and innovation directed to energy and environmental goals. Specifically:
For most of the time since the early 1980s and the oil crisis, government budgets for R&D on energy and the environment have struggled to increase in real terms and as a percentage of gross domestic product (GDP), mainly responding to energy affordability and security considerations. In 2023, the median OECD country dedicated less than 7% of its national R&D budget to energy or environmental objectives. This is even lower when accounting for the growing role of R&D tax incentive support.
Similarly, government institutions that perform R&D dedicate 10% of their total R&D expenditures to pursue energy and environmental objectives, on average across OECD countries.
Less than 5% of all government budgets for R&D across OECD countries are dedicated to supporting low-carbon energy technology development.
Support for R&D for nuclear energy has decreased significantly since the 1980s, with investment in other technologies like renewables and batteries struggling to compensate for the decline in nuclear energy investments.
However, government support for R&D on energy and the environment started increasing again in the mid-2010s and picked up significantly in 2023, according to multiple measures:
Government budget allocations for energy and environmental R&D in the OECD area grew by 29% in 2023, boosted by Japan’s Green Transformation initiative budget. Government spending on energy R&D and demonstration reached a global total of USD 50 billion (US dollars) in 2023.
Hydrogen and nuclear appear to be the main foci of low-emissions energy demonstration support, followed by ammonia and cement.
The International Energy Agency (IEA) estimates that overall government spending (not just R&D and demonstration) on clean energy investment support rose nearly 25% from 2021 to 2023, outpacing growth in fossil fuels in the same period.
Analysis of COVID-19 recovery and resilience fiscal packages from 2020 to 2022 shows that 51 OECD, EU and G20 countries allocated USD 1.29 trillion in spending to develop and deploy low-carbon technologies.
Analysis of R&D portfolios for governments and funding agencies across 19 OECD countries and EU programmes shows that energy and environmental goals:
Are embedded across a wide range of funding agencies and projects that also contribute to other societal goals, such as industry competitiveness and innovation and health.
Account for approximately 20% of project awards funding in 2023, representing nearly USD 200 billion.
Saw an appreciable boost in 2021, particularly for Sustainable Development Goal (SDG) 7 (Affordable and clean energy) and SDG 13 (Climate action). This, however, appears to have been relatively short-lived, with a return to pre-crisis levels in 2023. R&D budgets do not appear to be translating immediately into actual R&D projects.
When put in perspective, public support for R&D and demonstration appears dwarfed by support for downstream adoption of existing energy and environmental technology. Specifically:
Overall, only 6.5% of total low-carbon technology funding in COVID-19 recovery packages was channelled towards support for R&D and demonstration projects (about 2.4%), while the majority supported deployment and adoption of existing technologies and practices.
A pilot study confirms that support for business innovation with environmental objectives tends to be relatively more oriented towards downstream adoption of innovation activities, which fall outside the scope of systematic innovation measurement efforts.
Governments intervene in co-financing demonstrations and venture capital (VC) investments. The share of deals involving government VC support in energy, resources and sustainability in OECD countries between 2000 and 2022 was around 6%, putting the sector among those with the highest reliance on government VC involvement.
New OECD analysis shows that targeted innovation policies appear to influence the likelihood that businesses will develop and introduce technologies and practices that positively impact the environment.
For example, a 1 percentage-point increase in the proportion of enterprises receiving public funding for R&D or other innovation activities is, on average, associated with an approximately 2.1% increase in innovation-active firms introducing green innovations that reduce CO2 emissions among users of their products. In contrast, R&D tax incentives do not have such effects.
While public support does appear to influence business behaviour, the main reported motivations driving the adoption of environmental innovations are: 1) improving the enterprise's reputation; 2) current or expected market demand for environmental innovations; 3) environmental regulation; and 4) cost of energy, water or materials.
A broad set of policy measures aims to address environmental challenges such as climate change. As explained in Chapter 1, market failures within innovation systems hamper the production of new knowledge and its translation into practical applications that can result in environmental improvements and better use of natural resources. Current levels of private sector investments do not appear sufficient to meet the challenge (IEA, 2020[1]). Transformative change at the scale and depth required by climate change and other environmental challenges call for ambitious levels of science, technology and innovation (STI) investment over an extended period, covering all parts of the innovation chain, from exploratory fundamental research to the deployment and diffusion of tested technologies (OECD, 2024[2]). Some of the low-carbon technologies necessary to reach net-zero emissions already exist. However, their costs need to be reduced to become fully competitive with more polluting alternatives to enable their rapid deployment at scale (IPCC, 2022[3]). Other technologies are still in their infancy and need to be further developed.
The adoption of environmentally friendly technologies can also be a driver of future competitiveness for countries. Global supply chains are being reshaped, and countries that rely heavily on fossil fuels are at particular risk in today’s geostrategic context, highlighting the importance of establishing reliable supply chains relating to energy and environmental technologies. The current geopolitical context provides additional motivation for aligning environmental, economic and energy security objectives when it comes to policies relating to science and innovation investments. Evidence regarding the magnitude, nature and impact of public support can inform policies to seize available opportunities. All in all, the rationale for public policy is particularly strong due to the unique features of investments in STI that aim to help combat climate change and other environmental challenges.
Governments typically use spending, pricing, or regulatory measures to induce economic activities that are deemed socially desirable and undersupplied. The implementation of different types of environmental policies, such as taxes or restrictions on pollution, helps reduce incentives to engage environment-harming activities. As noted in Chapter 1, implementing the “polluter pays” principle though these policies in isolation may lead to undesirable consequences such as mounting costs, which also may be levelled at those who are most economically vulnerable. There is ample consensus that they need to be complemented by policies that encourage the desired transformations and support those most directly negatively impacted by them.
Government financial support for STI in the public and private sectors is a key part a portfolio of measures (OECD, 2024[4]). Without this support, including for fundamental research, progress on low-carbon innovation, as illustrated for instance by the recent advances in the cost reduction of renewable energy technologies, would not have been possible. Understanding the intended and actual directionality of public support can help in policy reform. Investing in internationally comparable and comprehensive data on support for research and innovation can also provide the international community with the evidence required to help foster a level playing field, where countries can compete and co-operate under a shared rules-based system.
This chapter puts the focus on available and new measures of public support along the spectrum of knowledge generation and deployment activities that characterise science and innovation systems. It provides insights into the support for energy and environmental resilience and transformative policy goals and provides new analysis of the relationship between support and environmental innovation outcomes. It concludes with a re-examination of the balance between stringency-oriented and support-oriented policies.
Measurement rationale
Government budget-setting processes set the priorities for countries’ use of public financial resources. How much does this process contribute directly to paying for R&D, and within that, how effectively does it aim to support energy and environmental objectives? Data on government budget allocations for R&D (GBARD) provide an overview of public policy resource allocation priorities for R&D contributing to the economy-wide generation of knowledge in areas relevant to a number of societal challenges, such as energy and the environment (Box 4.1).
Indicators of Government Budgets for R&D (GBARD) measure the funds that governments allocate through budgets to R&D to meet various socio-economic objectives (SEOs), using the classification in the nomenclature for the analysis and comparison of scientific programmes and budgets (NABS).
The environment SEO covers R&D aimed at improving pollution control, including identifying and analysing sources of pollution and their causes, and all pollutants, including their dispersal in the environment and the effects on humans, species and the biosphere. The development of monitoring facilities for measuring all kinds of pollution is included, as is R&D for eliminating and preventing all forms of pollution in all types of environments.
The energy SEO covers R&D aimed at improving the production, storage, transportation, distribution and rational use of all forms of energy. It also includes R&D on processes to increase the efficiency of energy production and distribution, as well as the study of energy conservation.
SEOs are implemented in GBARD statistics as mutually exclusive categories. Measured funds for the environment and energy exclude funding supporting R&D relevant to these objectives if another objective better represents the funder’s intention, e.g. defence, transportation or the general advancement of knowledge. The allocation of R&D budgets to SEOs is conducted at the level that most accurately reflects the funder’s primary purpose. This often matches the mandate(s) for the specific ministry, department or agency in charge of disbursing of funds. It is, therefore, conceptually a top-down measure of intended directionality. GBARD measures provide timely information on funding plans and commitments intended specifically to fund R&D and exclude extrabudgetary data sources of funds provided by other international organisations. OECD and Eurostat data cover the ensemble of individual EU Member States as direct funders of R&D through their budgets. However, they do not currently capture EU institutions’ own budgetary funding for R&D conducted within or outside the EU area.
Source: Authors, based on OECD (2015[5]), Frascati Manual 2015: Guidelines for Collecting and Reporting Data on Research and Experimental Development, http://dx.doi.org/10.1787/978926.
Indicators of government R&D budget support for energy and the environment
Across OECD countries, a growing proportion of government support for public and private R&D is being channelled through instruments and funding oversight arrangements that transfer responsibility for where to invest in R&D performers. R&D budgets for most defined policy objectives have remained flat and been outpaced by non-directed R&D support instruments. This has been particularly the case for energy and environment-oriented R&D budgets, which remained unchanged in real terms from the early 1980s until the mid-2000s, when funding towards these goals resumed growth, reflecting increases in oil prices (Figure 4.1). Mexico, Japan and France stand out as energy-oriented countries in their R&D budgets, while Colombia, Latvia, Finland and New Zealand are among the countries most oriented to environment preservation goals (Figure 4.2). In terms of absolute spending levels, Japan stood out in 2023 as the leading investor due to how its multi-year Green Innovation Fund was recorded in budgets.
Note: More detail on measurement in box 4.2. From 1991 onwards, trends in the OECD area are derived from underlying estimates for each OECD country (except Costa Rica). These are chain-linked with trends prior to 1991, which are estimated using a sub-sample of OECD countries for which GBARD data are available for the entire 1981–1991 period (namely Australia, Austria, Belgium, Canada, Denmark, Finland, France, Greece, Ireland, Italy, Norway, Spain, Sweden, United Kingdom and United States and West Germany).
Source: OECD calculations based on OECD (n.d.[6]), Main Science and Technology Indicators (MSTI) Database, www.oecd.org/sti/msti.htm (accessed March 2025).
As a percentage of each country’s total government budget for R&D and combined total (USD million values on top)
Note: Data refer to 2013 and 2023 except for Austria, Greece, New Zealand, Sweden (2011, 2021), Chile, Estonia, Germany, Luxembourg, Switzerland (2012, 2021), Colombia, Denmark, Croatia (2012, 2020), France (2019), Slovenia (2012, 2016).
Source: OECD (n.d.[7]), Research and Development Statistics Database, www.oecd.org/sti/rds (accessed February 2025).
Measurement rationale
While the role of government in R&D is often perceived as that of a provider of funds, government institutions play a vital role as R&D performers, especially in conducting basic and applied research. While not exclusively, they are mostly funded by government budgets, which allows them to undertake long-term and high-risk R&D projects that might not be feasible for private entities. Depending on their set-up, their research agendas can align more with national priorities and strategic interests, such as public health, defence, energy security and environmental protection. This can help establish effective support relationships with industries critical for a country’s economy, not only undertaking relevant R&D but also providing a broader range of scientific and technical services.
Government institutions that engage in R&D are critical for supporting industries such as agriculture, fishing, and extractive activities that rely directly on natural resources. These industries and the communities whose livelihoods depend on them also rely on scientific information on the state of those resources and their main threats, such as pests. In addition, they drive technology development, for example, in measurement technologies, climate change-resistant plant varieties, etc. Furthermore, government R&D institutes drive R&D in many critical network-based industries, such as nuclear power and electricity networks.
The public sector comprises government-controlled institutions across different institutional sectors of the economy, such as the government, the business sector, and the higher education sector. Data on state-owned enterprises (SOEs) are reported as part of the business enterprise (BE) sector and business enterprise R&D (BERD) statistics.
Government expenditure on R&D (GOVERD) refers to the total spending by government institutions on R&D activities, including basic and applied research as well as experimental development. This expenditure typically covers research conducted by government agencies and public research institutions. There are challenges in determining the precise share of GOVERD spent on SEOs, as defined in the previous section, including environment and energy. The government sector presents fewer difficulties in reporting R&D performer-based expenditure statistics. Difficulties arise, particularly in separating R&D from other related activities, when these are undertaken simultaneously within an organisation. Government entities may, for instance, engage in activities such as general-purpose data collection for monitoring of natural or social systems or development of infrastructures, and to advance these goals, perform internal or external R&D.
The higher education (HE) sector comprises universities and other tertiary education institutions, independent of their sources of finance or legal status. In many countries, a significant part of R&D performed in the HE sector is conducted within institutions that are part of the general government sector, but exhibit considerable autonomy. R&D measurement standards treat the HE sector as a standalone reporting sector. Statistics on HE R&D (HERD) by socio-economic objective are rarely available because HERD is funded and organised differently, with universities, departments and academic staff making independent decisions on which topics to research, which limits the ability of reporting institutions to provide meaningful breakdowns by policy-oriented SEO purposes. In fact, a standalone SEO, General University Funds, exists to allocate institutional funds given to HE institutions for R&D which cannot be traced to a given objective.
Source: Authors, based on OECD (2015[5]), Frascati Manual 2015: Guidelines for Collecting and Reporting Data on Research and Experimental Development, http://dx.doi.org/10.1787/978926.
Measures of R&D for energy and the environment by government institutions
The total amounts and percentage of GOVERD across various countries dedicated to energy and environment SEOs reveals that while these sectors receive a notable share of public R&D funding, they generally represent a small to moderate portion of overall government R&D spending, close to 10% ( Figure 4.3). The distribution of spending varies significantly, with some countries prioritising environmental research, others focusing more on energy, and some maintaining a balanced approach. New Zealand has the highest share allocated to environmental R&D, whereas Japan, Germany and France invest significantly in both sectors. Compared with 2012, the combined share of environmental and energy R&D has remained relatively stable for most countries, though some shifts are evident.
As a percentage of total government R&D expenditure (GOVERD), selected economies for which data are available
Note: Data refer to 2012 and 2022 except for Austria, Greece, New Zealand, Sweden (2011, 2021), Chile, Estonia, Germany, Luxembourg, Switzerland (2012, 2021), Colombia, Denmark, Croatia (2012, 2020), France (2019), Slovenia (2012, 2016). Values shown on top represent the total GOVERD spent on environment and energy, expressed in USD millions PPP.
Source: OECD (n.d.[7]), Research and Development Statistics Database, www.oecd.org/sti/rds (accessed February 2025).
Measurement rationale
Public support for energy and environment-related R&D is highly heterogeneous. Support for energy R&D may be directed towards a wide array of technologies, from low-carbon to those relating to advanced coal, oil, and natural gas extraction, as well as carbon capture and storage (CCS) for fossil fuel use. Tracking support provides transparency on how much is spent on high-carbon innovation compared to low-carbon energy, helping policymakers evaluate whether funding aligns with their priorities. Herein lies policy interest in more granular estimates of public support for energy technology than offered by SEO-based R&D budget statistics presented earlier in this chapter.
Furthermore, public support for energy and environmental technologies is not limited to upstream R&D activity; it can also support demonstration and deployment activities well into the downstream. While the rationale for such forms of support may not be high in other technology areas, the unique features of energy and environmental markets may call for such types of support interventions, filling gaps left by private sector investment due to high risks and long timelines from ideation to commercialisation.
Measures of government support for low-carbon energy R&D
The IEA collect data on public support for R&D and demonstration (RD&D) by technology area (Box 4.3). These data show that support for low-carbon technologies, which account for most public support for energy RD&D, has only caught up recently with the peak levels of the early 1980s, despite a recent upsurge in investment in R&D on renewables and energy efficiency (Figure 4.4). The role of low-carbon energy technology within R&D budgets is also very heterogeneous across OECD countries and minimal in several of them (Figure 4.5).
Energy RD&D covers research, experimental development and demonstration to extract, convert, generate, transport, distribute, store, control, and rational use of all forms of energy. The subject scope of IEA’s RD&D data is broader than the Frascati Manual's energy data presented earlier. The former comprises all programmes that focus on sourcing, transporting, using energy and enhancing efficiency, regardless of their ultimate SEO. IEA data collections aim to include energy-related programmes that might be included under other SEOs.
In addition, the IEA concept of RD&D differs from the Frascati concept of R&D in that it also includes all kinds of “demonstration projects”. Demonstration activities test these innovations in real-world conditions, proving their feasibility, reliability and scalability. Many clean energy technologies – such as advanced solar photovoltaics, hydrogen production, and CCS – require large-scale demonstration before they can attract private investment and be widely adopted. Without adequate funding for demonstrations, promising innovations may remain stuck in laboratories or pilot stages, delaying their contribution to energy transitions. The scope of IEA RD&D data also differs in that it includes SOEs. Despite these differences, OECD energy R&D and IEA RD&D energy data coincide for several countries, reflecting difficulties in identifying budgetary support for energy demonstration without including deployment efforts and ensuring reporting by SOEs. While such challenges are being progressively overcome, the key insight provided by IEA RD&D stems from the possibility of tracking R&D investment in different energy technologies with varying levels of implications for decarbonisation objectives.
Source: Authors, based on OECD (2015[5]), Frascati Manual 2015: Guidelines for Collecting and Reporting Data on Research and Experimental Development, http://dx.doi.org/10.1787/978926 and IEA (2011[8]), Guide to Reporting Energy RD&D Budgets/Expenditures Statistics, https://www.iea.org/reports/iea-guide-to-reporting-energy-rdd-budget-expenditure-statistics.
USD millions PPP, IEA member countries
Note: Figures in USD millions in 2022 purchasing-power equivalent prices. Fossil fuel RD&D expenditures are depicted in red on the bottom scale so that the sum of other categories approximates support for low-carbon technology (categories like hydrogen may not be entirely low-carbon).
Source: OECD analysis, based on IEA (2025[9]), Energy Technology RD&D Budgets (database), https://www.iea.org/data-and-statistics/data-product/energy-technology-rd-and-d-budget-database-2.
As a percentage of total government budgets for R&D
Note: For Canada, Italy and United Kingdom, the recent shares are based on 2022 due to missing data for 2023.
Source: Authors, based on IEA (2025[9]), Energy Technology RD&D Budgets (database), https://www.iea.org/data-and-statistics/data-product/energy-technology-rd-and-d-budget-database-2 (2022 data) for low-carbon R&D budgets; and OECD (n.d.[6]), Main Science and Technology Indicators (MSTI) Database, www.oecd.org/sti/msti.htm for total government R&D budget (GBARD).
Government venture capital
Apart from grants, another popular means of public support for start-ups in recent years has been government venture capital (GovVC). This policy tool is often intended to complement private venture capital by funding innovation-driven firms that might not attract traditional VC investment. Governments have developed a variety of GovVC programmes that use public funds to invest in start-ups (Berger, Criscuolo and Dechezleprêtre, forthcoming[10]). The financial support can be channelled indirectly, such as though private VC firms, which themselves invest in selected start-ups, by providing tax credits or matching funds that augment private capital commitments. Governments can also take a more active role by directly owning and managing a VC fund (Berger, Dechezleprêtre and Fadic, 2024[11]).
Based on analysis and data infrastructure recently developed by the Productivity, Innovation and Entrepreneurship (PIE) division within the OECD STI Directorate (Box 4.4), governments tend to be somewhat more involved in sectors typically perceived as high-tech (see Figure 4.6). The share of deals involving GovVC in energy, resources and sustainability; healthcare; and manufacturing is around 6%. In about half of these deals, GovVC makes standalone investments without private VCs involved. In other sectors, GovVC accounts for about 4% of deals, and standalone investments from GovVC are rare.
Note: The sample covers VC deals in OECD countries in the period 2000-22. Mixed CV entails a combination of VC funding from government and private actors.
Source: Berger, Dechezleprêtre and Fadic (2024[11]), “What is the role of Government Venture Capital for innovation-driven entrepreneurship?”, https://doi.org/10.1787/6430069e-en, based on the OECD Startu-up Database.
The OECD project “From knowledge to innovation: Building the evidence base to inform innovative entrepreneurship and risk-finance policies” (KnowInn) identified a list of GovVC investors that provide public funding for innovation-driven entrepreneurial firms and other innovative firms. The original list was assembled using several sources, including data from proprietary vendors (Crunchbase, Dealroom, Pitchbook and Preqin), open-source data from the academic literature, and Invest Europe. To ensure the quality of the list, delegates of OECD countries to the Committee on Innovation, Industry and Entrepreneurship were invited to validate the results. This made it possible to validate 308 GovVC entities across 37 OECD countries.
The government VC data has been combined with the OECD Start-up Database, which contains firm- and deal-level data spanning 38 countries and covering all major industries. Firm-level data come from Crunchbase and Dealroom. The two commercial database providers complement each other in terms of coverage. Crunchbase has strong coverage in the North American market, while Dealroom focuses on the European market. Both databases include information on companies, VC deals, and exits (e.g. acquisition). They are assembled from a combination of methods, including through investor networks, online searches, community contributors, and the analysis of news and media feeds.
The GovVC database does not include funds where the government is an indirect investor. This restriction is a result of limited data availability, as information on the ownership structure of hybrid funds (between private and public entities) and other recipients of funds is not publicly available. A second limitation of the data is that more successful firms are more likely to be included, which may exclude firms with small angel investments or unsuccessful firms. For about one in three deals, information on investors or the amount invested is not disclosed. Furthermore, there is no direct mapping between the industries listed in Crunchbase and Dealroom and official economic activity classifications such as the International Standard Industrial Classification of All Economic Activities (ISIC) or its regional or national variations, including the Nomenclature of Economic Activities (NACE) or the North American Industry Classification System (NAICS), or. Instead, both databases provide tags that correspond partly to technologies and partly to activity areas in which firms operate.
Source: Berger, Dechezleprêtre and Fadic (2024[11]), “What is the role of Government Venture Capital for innovation-driven entrepreneurship?”, https://doi.org/10.1787/6430069e-en.
Government support for low-carbon investment
Government support for demonstration projects
For certain energy technologies, constructing and operating large-scale demonstration projects is a crucial step in moving a new concept or design from pilot plant to commercial availability (IEA, 2025[12]). Such projects are often needed in multiple different technological, geographical, regulatory and market settings in order to build experience and inform regulations and standards. Demonstration projects are featured in the policy and strategy documents of major economies, and new funding mechanisms are being developed. Government information and other announcements are tracked in the regularly updated IEA Energy Demonstration Projects Database (Box 4.5).
The IEA Demonstration Projects Database seeks to map major demonstration projects of clean energy technologies globally. Each project provides information on location, sector and technology grouping, status, capacity, timing and funding, when available. Information about projects is based on submissions from six governments, which is complemented with the IEA’s own research. The selection of projects is based on the IEA’s definition of demonstration projects, which considers the TRL of the technology being tested, its significance in terms of technology development and the regional distribution of existing projects.
A “demonstration project”, according to common usage in the energy sector, is typically one of the first few examples of a new technology being introduced onto a given market at the size of a single full-scale commercial unit. It involves far more time, cost and risk than a prototype and significantly reduces investor risk for subsequent installations. Demonstration projects are usually loss-making investments when considered in isolation, with their combination of capital requirement and risk placing them squarely within what is often referred to as the “valley of death’’, a stage when technologies can fail to progress commercially even if they have high market potential.
Source: IEA (2024[13]), Clean Energy Demonstration Projects Database, https://www.iea.org/data-and-statistics/data-tools/clean-energy-demonstration-projects-database and IEA (2024[14]) Advancing Clean Energy Demonstration Projects, https://www.iea.org/reports/advancing-clean-energy-demonstration-projects.
As of April 2025, this database lists around 200 active projects that, if realised, would account for investments of nearly USD 61 billion over the 2022-35 period, including USD 32 billion of public funds. However, roughly 60% of all funding relates to projects not yet under construction, of which many have pending final investment decisions but are awaiting financial or policy decisions outside their control (Figure 4.7). The funding is highly skewed towards a small number of very large projects, as fewer than 15 (of 200) account for half of the USD 61 billion in total estimated investment. Energy demonstrations are concentrated in North America, Europe, the People’s Republic of China (hereafter “China”), and a few other advanced economies in the Asia-Pacific region, such as Australia and Japan. Of the selected active projects over 2022-35, nearly 60% of the total funding announced or allocated is for projects in North America (primarily in the United States), 25% in Europe and 10% in China.
In USD billions
Note: CO2 mgmt = CO2 management. Hydrogen projects with exclusive application in a specific end-use sector (e.g. use of hydrogen in steelmaking) are reported in the corresponding end-use sector and otherwise under “hydrogen”.
Source: IEA (2025[15]), The State of Energy Innovation, https://www.iea.org/reports/the-state-of-energy-innovation
The IEA’s The State of Energy Innovation (IEA, 2025[15]) concludes from these data that current trends indicate a greater emphasis on supply-side rather than demand-side technology. It points out that if the active demonstration projects listed in the database proceed as planned, funding would be concentrated in two supply-side sectors: hydrogen and hydrogen-based fuels and nuclear power generation. Around one-half of the total funding was for projects focusing on hydrogen and hydrogen-based fuel production, and another quarter on power generation, such as advanced nuclear designs and floating offshore wind. These trends suggest that there are still many untapped funding opportunities on the demand side in sectors such as aluminium, aviation and shipping. Heavy industry, where projects for low-emission cement and steel account for a significant share of all projects, is an exception. Beyond R&D and demonstration, the IEA estimates that overall government spending on clean energy investment support rose nearly 25% from 2021 to 2023, outpacing growth in fossil fuels in the same period (IEA, 2023[16])
Support for low-carbon technology in COVID-19 recovery packages
Fiscal spending policies adopted after the coronavirus (COVID-19) pandemic have been presented as a unique opportunity to “build back better” and stimulate the economy while accelerating the transition to a low-carbon economy. A recent OECD paper (Aulie et al., 2023[17]) analyses the impact of fiscal measures adopted as a response to or in the aftermath of the COVID-19 pandemic on the development and deployment of innovative low-carbon technologies – i.e. innovative technical solutions characterised by a low greenhouse gas emission intensity, compared to existing alternatives.
This work builds on the OECD Low-Carbon Technology Support (LCTS) database, a newly developed repository of funding measures announced in 2020-21. The database focuses exclusively on fiscal measures (usually adopted as part of COVID-19 recovery packages) that aim to support the development or diffusion of low-carbon technologies. To be included, a measure must imply the use of government spending and have been announced as a reaction to or in the aftermath of the COVID-19 pandemic. Only measures aiming to deliver long-term economic or environmental benefits are considered; short-term rescue measures are excluded. Measures also need to be additional with respect to baseline policies, i.e. completely new measures or significant budgetary expansions of pre-existing measures.
The LCTS database includes measures announced from February 2020 until December 2021 and updates to these measures made after December 2021, where relevant (e.g. the US Inflation Reduction Act and some updated EU Member States’ Recovery and Resilience Plans). There can be a lag between the investment announcement and when these become publicly accessible, which implies that some low-carbon technology measures are not included in the LCTS database. The LCTS database includes 1 166 measures totalling USD 1.29 trillion worth of announced spending (not necessarily actual disbursements). The data includes measures adopted by 51 economies, including members of the OECD, the European Union and the G20 (Group of Twenty), which account for 89% of global GDP and 79% of global annual CO₂-equivalent emissions. However, coverage for non-OECD and EU members is likely to be less comprehensive.
Each measure in the LCTS database was allocated to a sector of activity: energy, buildings, transport, industry, agriculture, carbon capture, utilisation and storage (CCUS), water/waste management and other activities. Each measure is also linked with a specific technology, such as renewable energy generation, electric vehicles or energy efficiency in buildings. Finally, the innovation stage of the targeted technology is categorised by reference to guidance elements and definitions in the OECD Frascati Manual (OECD, 2015[5]) (R&D and boundaries), the IEA framework for measuring public support for RD&D (IEA, 2011[18]),the Oslo Manual on measuring innovation (OECD/Eurostat, 2018[19]) as well as academic literature and the IEA Technology Roadmaps.
Source: Aulie et al. (2023[17]), “Did COVID-19 accelerate the green transition?: An international assessment of fiscal spending measures to support low-carbon technologies”, https://doi.org/10.1787/5b486c18-en.
The analysis shows that 51 OECD, EU and G20 countries allocated USD 1.29 trillion in spending to develop and deploy low-carbon technologies as part of recovery and resilience fiscal packages since 2020 (Figure 4.8). Around 40% of total low-carbon technology government support has been directed to the energy sector (generation, transmission and distribution), 33% to the transportation sector and 14% to the buildings sector. With a mere 4% of the total funding going to industry – which is responsible for 23% of global CO2 emissions – industry appears to be the “forgotten” sector of the low-carbon technology public spending measures adopted in the aftermath of COVID-19. This being said, enhanced deployment of renewable electricity may indirectly enable further electrification in the industry sector.
In USD billions
Note: For each sector, a proportion of spending can be dedicated to multiple technologies within the sector or to the sector in general, but not to specified technologies. Measures amounting to around USD 90 billion target more than one sector (“cross-sector”). These measures target multiple technologies across multiple sectors or do not specify the low-carbon technology, such as measures targeting “green/clean/low-emission technologies”, “green transition”, etc. The category “other sectors” includes agriculture (USD 6.7 billion), water and waste management (USD 3 billion) and education/skills (USD 2.5 billion). Measures that do not target any sector or technology in particular (USD 8.5 billion) are excluded.
Source: Aulie et al. (2023[17]), “Did COVID-19 accelerate the green transition?: An international assessment of fiscal spending measures to support low-carbon technologies”, https://doi.org/10.1787/5b486c18-en, based on OECD (2023[20]), Low-Carbon Technology Support Database, available at https://www.oecd.org/en/topics/sub-issues/quantifying-industrial-strategies.html, version May 2023.
Most of the low-carbon technology public funding in the recovery programmes has been used to scale up existing technologies, as opposed to targeting technologies at early stages of development. However, this focus on the deployment of mature technologies varies significantly across countries (Figure 4.9). Overall, roughly 6.5% of total green funding was channelled to support for R&D (around 4.2% or USD 53.6 billion) and demonstration projects (around 2.4% or USD 30.4 billion). An additional 7.8% (USD 100 billion) of low-carbon public spending is channelled at emerging technologies at the pre-adoption stage, based on their technology readiness level (TRL), not specifically mentioning R&D or demonstration. In total, therefore, around 14.3% of post-COVID funding targets pre-commercialisation phases, while 72.7% of funding is allocated to adoption and deployment phases (13% of funding cannot be linked with any innovation stage). Globally, emerging technologies in the R&D, demonstration and pre-adoption innovation stages received USD 184 billion. Country breakdowns are also available in Aulie et al. (2023[17]).
As a percentage of total low-carbon technology support spending
Note: Measures are categorised as “R&D” if they explicitly mention “R&D” and categorised as “Demonstration” if they explicitly mention “pilot (project)”, “prototypes”, “limited number of projects”, ‘trials”, “demonstration”, “test/testing”, or “laboratory for the development of ...”. Measures are categorised as “Pre-Adoption” based on the technology readiness level (TRL) or if spending relates to other emerging technologies, not specifically mentioning R&D or demonstration. The remaining funding in the LCTS database supports measures which target the adoption phase, several phases or for which the innovation stage cannot be specified.
Source: Aulie et al. (2023[17]), “Did COVID-19 accelerate the green transition?: An international assessment of fiscal spending measures to support low-carbon technologies”, https://doi.org/10.1787/5b486c18-en, based on OECD (2023[20]), Low-Carbon Technology Support Database, available at https://www.oecd.org/en/topics/sub-issues/quantifying-industrial-strategies.html, version May 2023.
Among technologies specifically targeted, hydrogen ranks first in terms of funding for both R&D and demonstration (Figure 4.10). For example, the US Infrastructure Investment and Jobs Act included a USD 8 billion measure to finance research demonstrating the commercialisation of clean hydrogen production and use in the transportation, utility, industrial, commercial and residential sectors. Japan dedicated USD 1.8 billion to research projects related to hydrogen use in steelmaking processes via the Green Innovation Fund. Whereas a large portion of funding at the demonstration stage supports CCUS, nuclear innovation and smart grid technology projects, funding for the R&D stage of the same technologies remains limited.
As a percentage of total low-carbon technology support spending
Note: The category “General low-carbon R&D funding” includes funding related to measures specifically supporting R&D or demonstration but that do not target any specific technology. This category includes broader low-carbon measures and sector-specific but not technology-specific measures.
Source: Aulie et al. (2023[17]), “Did COVID-19 accelerate the green transition?: An international assessment of fiscal spending measures to support low-carbon technologies”, https://doi.org/10.1787/5b486c18-en, based on OECD (2023[20]), Low-Carbon Technology Support Database, available at https://www.oecd.org/en/topics/sub-issues/quantifying-industrial-strategies.html, version May 2023.
Measurement rationale
Both the 2015 OECD Daejeon Declaration on Science, Technology and Innovation Policies for the Global and Digital Age (OECD, 2015[21]) and the 2024 OECD Declaration on Transformative Science, Technology, and Innovation Policies for a Sustainable and Inclusive Future (OECD, 2024[4]) stress the importance of STI in achieving sustainable development and accelerating progress towards the Sustainable Development Goals (SDGs). There is particularly high interest in measuring the contribution of government R&D funding to specific SDGs. SDG-relevance mapping frameworks allow governments, academic institutions and businesses to allocate resources to projects that contribute directly to sustainable development (Roy et al., 2018[22]; IPCC, 2021[23]; IPCC, 2022[3]; IEA, 2021[24]; Borchardt et al., 2023[25]). SDG-based metrics are used to inform sustainable and transformative policies (OECD, 2024[26]). These metrics can also enable governments to prioritise R&D funding for projects addressing global challenges like climate change. Mapping SDGs to R&D “project”-level information in the OECD Fundstat database provides a unique opportunity to get a granular and nuanced understanding of the contribution of STI to sustainable development (Box 4.7).
The methods applied for allocating project R&D funding closely resemble those described in Chapter 2 for scientific publication output. These apply the same machine-learning-based probabilistic classifier for documents describing research activities and findings, trained on survey data with researchers’ self-assessment of SDG relevance. Compared to alternatives, this method’s main advantage is that it is informed by actual assessments of relevance that make it possible to establish conceptual links between upstream basic research and goals, even in the absence of references to explicit terms contained in the project descriptions. The probabilistic approach allows a given project to be fractionally allocated to multiple objectives, resulting in additive estimates of projects and, most importantly, funding. While the SDG framework was not conceived as a complete classification framework for all potential policy goals, its coverage of societal goals of R&D is rather complete as it includes goals on industry, innovation and infrastructure, as well as education. Furthermore, the classifier also factors in the possibility of no relevance to any SDGs since that was a choice provided to survey respondents. As it turns out, a relatively small fraction of surveyed researchers found their research to be of no relevance to any SDG, contrary to unsupervised machine-learning methods guided by semantic similarity to language contained in SDG descriptions. For the analysis of specific environmental and energy goals, the detailed SDGs provide a clear connection mechanism.
In addition to scientific publication data (see chapter 2), this classifier has been applied to the OECD Fundstat database, which contains project or award-level R&D funding data for agencies from 19 OECD countries and several EU programmes implemented by the European Commission. The analysis is limited to 2015-23, a time period where the percentage coverage of R&D funding in the OECD Fundstat database relative to OECD R&D budget statistics is relatively stable, and to R&D projects with available funding information. Coverage of R&D funding is heterogeneous across countries, with a weighted average percentage of 58% in 2021, and this may result in an over- or under-estimation of funding towards specific goals. This can be particularly the case for France and Germany. For some countries and agencies, observations of analysis are more programme-like, while in others, awards may capture individual project components or contributions from different agencies to the same project.
Source: Authors, based on Aristodemou, Appelt and Galindo-Rueda (forthcoming[27]), “Assessing the directionality of R&D funding towards societal goals through new data sources and AI-assisted methods”.
Estimates of energy and environmental R&D funding based on the SDG classifier
Analysis of R&D portfolios for governments and funding agencies across 19 OECD countries and EU programmes shows that SDG-defined energy and environmental goals accounted for approximately 20% of funding in 2023, representing nearly USD 40 billion (Figure 4.11). Funding for SDG 7 (Affordable and clean energy) leads at 5%, closely followed by SDG 12 (Responsible consumption and production) (4.6%), SDG 13 (Climate action) (4.2%), SDG 15 (Life on land) (3%), SDG 14 (Life below water) (2%), and SDG 6 (Clean water and sanitation accounts) (1.2%). Canada, Norway and Lithuania are the countries in the Fundstat database with the highest focus on energy and environmental goals, at just over 30% of their entire funding portfolio.
As a percentage of total R&D funding provided by each country and EU-EC institutions within the Fundstat database
Note: This indicator, with international comparisons, is sensitive to potential coverage biases. Funding for R&D project awards is fractionally allocated according to estimated SDG relevance probabilities using the SDG classifier developed in Aristodemou, Appelt and Galindo-Rueda (forthcoming[27]).
Source: OECD analysis, based on the OECD Fundstat database (v. 2024), March 2025.
Looking at funding trends for broad categories of SDGs since 2015 (Figure 4.12), funding for environmental and energy SDGs has been on the rise, with no significant shifts in their relative importance. Funding for economic prosperity goals dominates the funding landscape, followed by health, which receives just under twice as much funding as energy and environment combined.Project-level funding data show in a particularly clear way the funding boost to almost all funding areas during the COVID-19 crisis. As a percentage of total funding covered in the Fundstat database (Figure 4.13), there is an appreciable increase in funding share dedicated to some energy and environmental SDGs in 2021, particularly those of SDG 7 (Affordable and clean energy) and SDG 13 (Climate action). This, however, appears to be short-lived, with a return to pre-crisis levels in 2023.
R&D funding awards for 19 OECD countries and EC-EU programmes in the OECD Fundstat database, in USD billions
Note: Funding for R&D project awards is fractionally allocated according to estimated SDG relance probabilities using the SDG classifier developed in Aristodemou, Appelt and Galindo-Rueda (forthcoming[27]). Environment = SDG 6: Clean water and sanitation, SDG 12: Responsible consumption and production, SDG 13: Climate action, SDG 14: Life below water, SDG 15: Life on land; Energy = SDG 7 (Affordable and clean energy) and Prosperity = SDG 8 (Decent work and economic growth), SDG 9 (Industry, innovation and infrastructure), SDG 10 (Reduced inequalities), SDG 11 (Sustainable cities and communities).
Source: OECD analysis, based on the OECD Fundstat database (v. 2024), March 2025.
As a percentage of total R&D funding within the Fundstat database
Notes: Funding for R&D project awards is fractionally allocated according to estimated SDG relevance probabilities using the SDG classifier developed in Aristodemou, Appelt and Galindo-Rueda (forthcoming[27]). See chapter notes for a list of the SDGs and SDG categories.
Source: OECD analysis, based on the OECD Fundstat database (v. 2024), March 2025.
Figure 4.14 presents estimates of countries’ relative specialisation in the energy SDG, on the one hand, and on the other hand, the combined environmental SDGs, based on estimated funding allocations to individual SDGs. The results show an overall positive correlation between both measures but with some exceptions. Japan is highly specialised in energy, but the opposite applies to environmental goals. There are starker specialisation differences for energy, with a much wider range between the least specialised (Austria and the United States) and the most (Lithuania and Germany), than for the environment.
Specialisation indices and country share in total Fundstat energy and environment R&D funding
Note: The use of this indicator for international comparisons is sensitive to potential coverage biases in the Fundstat database. The relative specialisation log-index is computed as (the logarithm of) the ratio between a country's funding share in a specific SDG category relative to the country’s total funding and the global funding share for that SDG category relative to total global funding. Values above 0 indicate above-average relative specialisation in funding for that SDG category compared to the global average, while values below 0 suggest below-average specialisation. Log specialisation indices are depicted as positions (x-axis for energy; y-axis for environment).
Source: OECD analysis, based on the OECD Fundstat database (v. 2024), March 2025.
Mapping the R&D funding landscape according to SDG relevance
Users of the SDG framework are often concerned about the extent to which different SDGs overlap, relate to and potentially contribute to each other. A probabilistic assignment of projects and the amount of funding they allocate to SDGs helps provide a basis for interpreting the association between different SDGs and different projects characterised by their unique relevance “footprint” to the entire set of SDGs.
The principal component analysis (PCA) visualisation technique presented in (Figure 4.15) shows a base two-dimensional map of SDGs on which it is possible to locate all projects in the Fundstat database. Instead of 18 (17 SDGs plus “none” option) dimensions, 2 dimensions capture 35.5% of the variation in the underlying SDG probability assignment data. The different SDG positions reflect their distinctiveness vis-à-vis each other, pulling away from the case of no SDG relevance at the centre of the figure. SDGs appear to cluster in four distinctive areas within each quadrant. The top right quadrant is predominantly occupied by SDGs centred on environmental protection, SDGs 12, 13, 14 and 15. The sole presence of SDG 9 (Innovation) in the top left quadrant hints at economic performance-oriented projects. The bottom left quadrant is solely occupied by SDG 3 (Health), while the bottom right quadrant is more heterogeneous, comprising mostly people-related SDGs, including education and poverty. Smaller clusters around environmental sustainability SDGs (13, 14, and 15) suggest thematic convergence, reflecting the interdisciplinary and interconnected character of environmental research. The position of SDG 7 (Affordable and clean energy), right in between industry performance and environmental goals, is indicative of connections with both and potentially the pursuit of a “double dividend” for projects that score high on this goal.
Principal component analysis (PCA) biplot of estimated SDG relevance probabilities
Note: The PCA is performed on the probabilities predicted by the SDG classifier developed in Aristodemou, Appelt and Galindo-Rueda (forthcoming[27]) on government-funded R&D projects in the OECD Fundstat database. The PCA identifies two main dimensions as drivers of overall data variance, depicted as the main axes, and clusters SDGs according to their commonalities and distinctiveness. Bubble size reflects the estimated share of total R&D funding for each SDG. The two principal components shown explain approximately 35.5% of the variance in the original dataset. The analysis is limited to R&D projects with available funding information.
Source: Aristodemou, Appelt and Galindo-Rueda (forthcoming[28]), “Assessing the directionality of R&D funding towards societal goals through new data sources and AI-assisted methods”.
Indicators of multiple SDG relevance of R&D funding
Another way to examine the interconnectedness between SDGs is to decompose funding projects that are predominantly related to one goal and examine the other goals to which they are relevant. Figure 4.16 illustrates the distribution of projects by primary or dominant SDG categories and, within those, presents the distribution of funding across all broad SDG categories, starting with the dominant one. The results indicate that the primary broad goal accounts for at most 70% of all funding going into an average project. In the case of funding for projects with a primary environmental goal, 18% of funding is deemed relevant for economic prosperity and 14% for health. In the case of energy, 14% is deemed relevant to environmental goals and 10% to economic prosperity.
Distribution of funding relevance within projects’ dominant SDG groups (horizontal), as % of dominant
Dominant SDG group (funding distribution proportional to width)
Note: The Marimekko chart shows how projects principally under a SDG group contribute other secondary SDG groups. The x-axis represents different dominant (primary) SDG groups. The y-axis shows the relative distribution of secondary SDG groups within each dominant (primary) group. The analysis is limited to R&D projects with available funding information. How to read: Within “Health” dominated R&D awards, 26 (100-74)% of funding is relevant towards other SDG groups (of which 9% is Planet).
Source: Aristodemou, L., et al. (forthcoming[28]), “Assessing the directionality of R&D funding towards societal goals through new data sources and AI-assisted methods”.
Distinctive topics of energy and environmental R&D funding
One of the main advantages of working with information at the project level, including detailed descriptions, is the ability to explore the topics covered within broadly classified categories, such as the SDGs. Figure 4.17 presents the results of an unsupervised topic modelling exercise for R&D projects with a dominant energy or environmental SDG goal, calibrated to result in a manageable number of topics (37 arise), labelled using ChatGPT-4.5, and displayed visually in a simplified fashion using the same PCA described earlier and with bubble sizes representing the amount of funding estimated to be applicable to each topic.
Principal component analysis (PCA) biplot of estimated topic relevance probabilities
Note: R&D funding is fractionally allocated to topics based on standardised probabilities predicted from the topic model using the same method as reported in Aristodemou et al. (2023[29]), with the addition that the topic labelling is performed using OpenAI’s ChatGPT-4.5. The analysis is limited to R&D projects with available funding information. The topic model identifies 37 topics. The PCA identifies two main dimensions as drivers of overall data variance, depicted as the main axes. Bubble size reflects the estimated share of total R&D funding for each topic. The two principal components shown explain approximately 27.5% of the variance in the original dataset.
Source: Aristodemou, L., et al. (forthcoming[28]), “Assessing the directionality of R&D funding towards societal goals through new data sources and AI-assisted methods”
The results are largely exploratory at this stage and require expert analysis to improve the parametrisation of the topic modelling, the precision of labelling and the interpretation of the results. They provide some initial insights on the themes underpinning funding bodies’ support to environmental and energy goals over an extended period. It is revealing to see clustering into topics such as solar and battery technologies, energy-efficient renovations, electric vehicles, smart charging infrastructure, neuromorphic computing, superconductors, and biotechnology, potentially capturing the scientific complexity, ecosystem health, marine sciences, biodiversity conservation, and integrated climate policy assessments, emphasising ecological sustainability, climate policy, and social inclusion, sustainable transport, smart grids, infrastructure technologies, green finance, and economic resilience. Several of these topics can be connected with elements of environmental taxonomies, such as the EU green taxonomy (see Box 4.8) or the IEA clean technology guide, which provides additional information on the level of maturity of technologies.
The Research Council of Norway (RCN), together with six other funding agencies for public support to business development in Norway, was instructed by Norway’s government to report on how the funded projects contribute with a significant positive effect on six climate and environmental categories based on the EU taxonomy, and whether the projects do significant harm to any of the categories.
2024 was the first year for this reporting, and it was designed by the ministries as a pilot project that will be evaluated in 2025. The different agencies solve the task in different ways depending on the number of projects and type of funding instruments. However, the reporting itself must be unified and comparable to that of a common statistics bank.
RCN has developed a machine-learning algorithm to classify more than 6 500 projects. From the corpus of application texts, 559 projects were selected for human annotation by two to three case officers each. These data were divided into training data (80%), validation data (10%), and test data (10%). In approximately 20% of the projects, there was a strong disagreement between two case officers on at least one of the categories. This was partly solved by introducing a third person to annotate these, but it illustrates that the classification is not simple, and in many cases, there is no obvious correct answer.
Since the purpose of the model is to report statistics on an aggregate level, the model was tuned to balance false negatives and false positives. The model was run on the total set of projects in mid-January 2025 and reported to the statistics bank. It will be re-run monthly with newly funded projects added.
Source: Case study provided by Research Council Norway.
Measurement rationale
Businesses play a major role in creating and using new knowledge to introduce new products and processes that transform economies and societies. Governments’ innovation policies, especially in market economies, seek, among other things, to set a framework for business innovation and provide a range of effective incentives to encourage greater alignment between business innovation efforts and greater societal good. Therefore, the use of positive financial incentives for firm innovation is one of the major defining aspects of innovation policy. As businesses account for nearly 70% of total R&D across OECD countries, it is apparent that advances towards energy and the sustainability transition require the active participation of this sector. Public support can help mitigate, in part, the disincentives for private investment in R&D and other innovation activities that are particularly marked in the case of environment-related technologies (OECD, 2025[30]). Measuring public support for innovation entails more than just capturing direct support for R&D; it also requires taking into account other support instruments. Governments may use a combination of innovation policy tools to encourage science and innovation by businesses for sustainable growth. This includes financial support for R&D and innovation through direct measures of government support (e.g. grant funding, government procurement) and indirect support through the tax system.
New insights from the mapping of business innovation support at the programme level
The results from recent OECD business innovation support mapping pilots (OECD, 2023[31]) provide some novel insights into the directionality of business innovation support across three dimensions: 1) the directionality of business innovation support by SEO; 2) the distribution of government support for business innovation with green and industry-related SEO by policy instrument; and 3) the distribution of government support for business innovation with green and industry-related SEO by STI activity supported (Box 4.9).
There is scope for achieving a more comprehensive understanding of the landscape of government support for business R&D and its impact. Recent OECD work proposed (OECD, 2023[32]) and tested (OECD, 2023[31]) a framework for measuring how governments direct resources in support of innovation, building from the analysis of individual instruments and programmes to full support portfolios. Viewing directionality as a multifaceted concept, the framework addresses various dimensions of directionality (e.g. policy objectives of business innovation support, the type of support mechanism used to provide support and the type of STI activities supported) while also providing quantification guidance in gross and net subsidy terms to allow international comparability:
Systematic measurement and documentation are still needed, particularly for indirect support instruments and those focusing on innovation activities beyond R&D. Since internationally compiled statistics on R&D budgets do not distinguish how these are allocated by objectives and beneficiaries, the OECD has been conducting an exploratory project to map government financial support for innovation and develop a measurement framework. This work has revealed the need to factor in financial support mechanisms that may facilitate a return on investment, procurement of innovative solutions and the provision of infrastructure, goods and services, often mediated through R&D specialist organisations with a mandate to serve business needs.
Source: OECD (2023[32]), “OECD framework for mapping and quantifying government support for business innovation”, https://doi.org/10.1787/f30547e7-en; OECD (2023[31]).
The five OECD countries covered in the OECD pilot study – Australia, Canada, France, Norway, and the Netherlands – make extensive use of tax incentives to support innovation. As Figure 4 shows, the importance of tax incentives is also reflected in the distribution of business innovation support by SEO. Given the predominant role of tax incentives and other programmes designed as horizontal, business initiative-driven instruments without thematic or sectoral constraints, most business innovation support is by default oriented towards the SEO “industrial production and technology” since no other objective better reflects the government’s intention. The breakdowns of measured support by SEO also reveal some specific national priorities. When it comes to supporting business innovation directed to specific objectives, the areas of “environment and energy” (combined into one category) stand out. With over one-quarter of total business innovation support, Norway displays the largest orientation towards “environment and energy”, followed by the Netherlands and Canada1.
Amount of support within each category as % of total government support identified in the SUPRINNO pilot study
Note: Figures for France, Norway, and the Netherlands refer to calendar year 2021, while those for Australia and Canada refer to fiscal year 2021-22. The figures reported, especially those for Australia and the Netherlands, are likely to understate the amount of business innovation support the government provides at the subnational level. Government support for business innovation is classified by socio-economic objective based on the Nomenclature for the Analysis and Comparisons of Scientific Programmes and Budgets (NABS). In the case of programmes with multiple objectives, an equal-weight fractional approach has been applied to the funding.
Source: OECD (2023), OECD MABIS-SUPRINNO project.
When focusing on business innovation support with a green or industry-related SEOs and comparing the distribution of this support by policy instrument (Figure 4.19), differences in the role of direct and indirect support mechanisms are noticeable. While business innovation support with an industry-related SEO is provided primarily through the tax system in the five OECD countries considered, grant funding represents the primary vehicle for business innovation support with a green policy objective.
Figure 4.20 provides additional insights into the differences in the type of STI activities that are supported through business innovation support, with a green vis-à-vis industry-related SEO. While business innovation support with an industry-related SEO predominantly entails support for R&D, business innovation support with a green policy objective entails support for R&D and more downstream innovation activities, either alone or in combination with R&D.
Amount of support within each category as % of total government support identified with a green/industry SEO
Note: “Prov. of infrastr., goods and services” = Provision of infrastructure, goods and services. Figures for France, Norway, and the Netherlands refer to calendar year 2021, while those for Australia and Canada refer to fiscal year 2021-22. The figures reported, especially those for Australia and the Netherlands are likely to understate the amount of business innovation support provided by governments at the subnational level. Government support for business innovation is classified by socio-economic objective (SEO) based on the Nomenclature for the Analysis and Comparisons of Scientific Programmes and Budgets (NABS). Government support with a “green” (“industry”) SEO comprises support in the NABS categories “environment” and “energy” (“industrial production and technology”). In the case of programmes with multiple objectives, an equal-weight fractional approach has been applied to the funding. The category “Multiple instruments” encompasses policies for which instruments are not separately identifiable.
Source: OECD (2023), OECD MABIS-SUPRINNO project.
Amount of support within each category as % of total government support identified with a green/industry SEO
Note: Figures for France, Norway, and the Netherlands refer to calendar year 2021, while those for Australia and Canada refer to fiscal year 2021-22. The figures reported, especially those for Australia and the Netherlands, are likely to understate the amount of business innovation support governments provide at the subnational level. Government support for business innovation is classified by socio-economic objective (SEO) based on the Nomenclature for the Analysis and Comparisons of Scientific Programmes and Budgets (NABS). Government support with a “green” (“industry”) SEO comprises support in the NABS categories “environment” and “energy” (“industrial production and technology”). In the case of programmes with multiple objectives, an equal-weight fractional approach has been applied to the funding. The category ‘Multiple instruments’ encompasses policies for which instruments are not separately identifiable.
Source: OECD (2023), OECD MABIS-SUPRINNO project.
Measurement rationale
Policymakers worldwide are not only interested in characterising public support for environmental innovation and technology. They also expect indicators to inform questions such as whether financial support measures, alongside other non-financial measures, deliver value for money and bear fruit in terms of spurring R&D efforts, patenting and innovation by business, including innovations with environmental benefits. However, as noted in Box 4.10, while necessary components of the evidence base, indicators have intrinsic limitations as sources of empirical evidence on policy impacts.
STI indicators provide useful information on the performance of a system, including its strengths, weaknesses and the nature of policies. They can also help track changes over time. However, this is insufficient: decision makers also need to know how conditions in one part of the system influence other parts and how the system works to create outcomes of interest, including the effects of policy interventions. The study of science and innovation for sustainable growth can benefit from the different elements that help put such types of evidence in place.
Associations between the components of an innovation system can be identified through descriptive and exploratory analysis of available indicators. Statistical analysis provides tools for exploring relationships of dependence between different indicators – for example, innovation outputs and inputs – conditional on other characteristics, such as firm size, age and industry of main economic activity. However, a statistical association between two variables (for instance, an input to innovation and an environmental performance output) does not necessarily imply causation. Additional evidence and reasonable assumptions include a plausible time gap between an input and output, replication in several studies, and the ability to control for relevant confounding variables. Unless these conditions are met (which is rare in exploratory analyses), a study should not assume causality. Research on policy interventions must address self-selection and plausible counterfactuals: what would have happened in the absence of a given policy intervention?
Source: Authors, based on OECD/Eurostat (2018[19]), Oslo Manual 2018: Guidelines for Collecting, Reporting and Using Data on Innovation, 4th Edition, https://doi.org/10.1787/9789264304604-en.
The relationship between support and innovation with environmental benefits across size groups, industries and countries
While there is a growing literature examining the causal impacts between policy interventions, including innovation policy tools2 such as R&D tax incentives (e.g. Dechezleprêtre et al. (2023[33]), Pless (2019[34]) and Pless and Srivastav (2022[35])), and outcomes such as business R&D, patenting, innovation introduction and economic performance, there is scope for enhancing the evidence on the impact of government support for business R&D and innovation on green transformation outcomes in a multi-country setting.
Cross-country analysis (Figure 4.21) of the relationship between different forms of public support for environmental innovation helps illuminate how different policy instruments work and what are their likely impacts. New OECD analysis, drawing on data from the 2020 Community Innovation Survey aggregated at the level of groups of firms within a given size group, specific industry, and each of the 16 different European countries, explores how the share of innovation-active firms introducing different types of green innovations varies in relation to differences in the share of innovation-active firms that received public funding – direct support for R&D or other innovation activities from local and/or national government and/or the European Union – and those that received tax support for R&D or other innovation activities. Receiving one form of support does not exclude the other (in contrast to the results presented in Figure 4.22 further below, where more detailed information on the sole and combined use of direct and tax support measures is available).
As Figure 4.21 shows, the share of innovation-active firms introducing different types of green innovations is, on average, not significantly linked to the share of innovation-active firms that received tax support. Direct public funding, by contrast, appears to incentivise the introduction of innovation with environmental benefits. The share of firms benefiting from direct public funding is positively and significantly correlated with the share of enterprises introducing green innovations for six out of ten types considered. The introduction of CO2 emissions mitigation and pollution-reducing technologies, both within firms and in end use, displays the strongest association with public funding: a 1 percentage-point increase in the proportion of enterprises receiving public funding for R&D or other innovation activities is, on average, associated with an approximately 2.1% increase in the percentage of innovation-active firms introducing green innovations that reduce CO2 emissions on the side of the user.
Estimated percentage change in the share of innovation-active firms introducing environmental innovations by type of green innovation and type of public support for R&D and other innovation activities
Note: Innovation-active enterprises are enterprises engaged at some time during the observation period 2018 to 2020 in one or more activities to develop or implement new or improved products or business processes. Public funding stands for direct support from local and/or national government and/or the European Union. The analysis covers Austria, Bulgaria, Cyprus, Czechia, France, Croatia, Hungary, Italy, Lithuania, Luxembourg, Malta, Poland, Portugal, Romania, Slovenia and the Slovak Republic. Horizontal lines mark the 95% confidence interval that covers the “true” elasticity with a probability of 95%. All regressions control for country (reference: Austria), industry (reference: manufacturing) and size class (reference: firms with 0-49 employees) fixed effects.
Source: OECD analysis based on micro-aggregated data from Eurostat (2020[36]), Community Innovation Survey 2020 (CIS2020), https://ec.europa.eu/eurostat/cache/metadata/en/inn_cis12_esms.htm (accessed March 2025).
New firm-level analysis of government support for R&D and innovation, and environmental innovation
The OECD microBeRD project was started in 2016 to shed light on the distribution and structure of business R&D (BERD) and the heterogeneity in the use and impact of government support across different types of firms and countries. microBeRD facilitates a co-ordinated statistical multi-country analysis, applying a distributed approach to the analysis of business R&D, tax relief and other relevant micro-data, based on close collaboration with national experts with access to confidential micro-data. A total of 23 OECD countries has contributed to the microdata-based analysis ever since, generating new insights into the impact of direct and tax support on R&D, innovation and economic outcomes (OECD, 2023[37]; 2020[38]), as well as trends in business R&D performance and funding (Appelt et al., 2022[39]).
Recent microBeRD project findings suggest that R&D tax incentives and direct funding are, on average, equally effective in raising total business R&D, with small R&D performers showing particularly strong responsiveness to tax incentives. However, R&D tax incentives boost investment towards incremental development more effectively than more transformational, higher spillover-potential knowledge. The effect of R&D tax incentives is effectively concentrated on “experimental development”, the D of R&D, while, unlike for directed support, the impact on research is virtually nil. Building on this work a new project phase set out to study the effect of direct and tax support on green transformation outcomes. So far, 14 OECD countries (Austria, Belgium, Chile, Czechia, Finland, France, Germany, Israel, Italy, Japan, Norway, Portugal, the Slovak Republic, and Sweden) have participated in the pilot analysis launched in September 2024, contributing micro-data on “green” transformation outcomes from the national business R&D survey, the business innovation survey and patent micro-data, where available.
The preliminary micro-databased statistics compiled in the green pilot provide some initial insights into the relationship between government support for R&D and innovation – tax and direct support – and outcomes related to environmental sustainability, such as the introduction of innovations with environmental benefits, energy and environment-related R&D, as well as grey, high-carbon and “green” (both low-carbon and broader environment-related) patent filings by business.
Figure 4.22 examines the link between government business R&D support measures and green innovation at the micro level, showing how the percentage of innovative enterprises that introduced a green innovation over the 2020-22 period varies across firms of different sizes and the type of public financial support for R&D that they received. In seven out of the ten OECD countries for which preliminary results by firm size are available, the percentage of innovative enterprises introducing green innovations is higher for larger-sized firms. The percentage of innovative enterprises introducing green innovations over the observed period tends to be lowest for firms that received no government support for R&D (ranging from 16% for medium-sized companies in Japan to 59% for large companies in Czechia), independent of the size class of firms considered. In most instances (16 out of the 27 cases considered), similar rates of environmental innovation can be observed for innovative firms that received only tax support and for those that received no financial support for R&D. By contrast, the percentage of innovative firms introducing green innovation is notably higher for firms that received only direct support for R&D (e.g. R&D grants, procurement) or both direct and tax support. In all of the 31 cases (24 out of the 28 cases) considered, the rate of environment-related innovation is higher for firms that received direct support only (including direct and tax support) compared to those that received no financial support for business R&D. This initial finding points to the importance of direct funding in steering the “green” direction of innovation.
Impact on energy and environmental R&D
The analysis also examines the relationship between government support for R&D and the environmental orientation of business R&D investment (Figure 4.23), measured by the share of energy (Panel A) and environment (Panel B) related R&D in total business R&D investment in 2021 (or closest available year). The results for energy and environment-related R&D largely mirror the results for environmental innovations (Figure 4.22). The share of both energy and environment-related R&D in intramural R&D is, on average, lowest among firms that received no support (energy: ranging from 2% in Israel and Japan to 19% in Sweden; environment: ranging from 2% in Japan to 12% in Norway), while there is no discernable increase in the share of energy or environmental R&D among R&D tax relief recipients. Direct funding, by contrast, tends to be associated with a significantly higher orientation of business R&D investment in energy or environmentally driven R&D. In the case of enterprises that receive direct funding only, the share of energy (environment) related R&D in total intramural R&D by business varies from 2% in Israel to 28% in Sweden (5% in Japan to 18% in Norway).
As a percentage of the total number of innovative enterprises within each size class
Note: This chart displays the percentage of innovative enterprises introducing an environmental innovation (i.e. innovation with environmental benefits that contributed significantly to environmental protection) over the 2020-22 period, drawing on 2022 Community Innovation Survey (CIS) micro-data and administrative R&D tax relief micro-data, where available.3
Source: OECD (2025[40]), OECD microBeRD project, https://oe.cd/microberd (accessed March 2025).
Impact of public support on patenting in environment-related technologies
Figure 4.24 shows that direct funding is associated with a higher rate of environment-related patenting than the “no support” or “tax support only” scenarios. This holds across the three OECD countries considered, regardless of whether a narrow or broad definition of “green” patents is applied – i.e. patents related to low-carbon technologies only or to a broader set of environment-related technologies (see Chapter 2, Box 2.8). Among enterprises that received direct support only, the share of low-carbon patents amounts to 20% in France, 9% in Japan and 13% in Norway; when using the broader definition, these shares rise to 24% in France, 12% in Japan, and 21% in Norway (where the analysis focuses on national patent applications only). In the “no support” scenario, the share of low-carbon patents is 5% in France, 8% in Japan, and 8% in Norway; under the broader definition, the corresponding shares are 7%, 9%, and 15%, respectively. Results are mixed in the case of high-carbon and grey patents, which relate to technologies improving the efficiency of high-carbon technologies. While direct support appears to be also associated with higher rates of grey and high-carbon patenting by business R&D performers in the case of France and Norway (relative to the “no support” scenario), no such differences are observable in the case of Japan. The higher rates of grey and high-carbon patenting for direct funding recipients in France and Norway may at least in part reflect the effect of common co-founding factors (e.g. firm size) correlated with the receipt of public support and patenting activity more broadly.
As a percentage of total intramural R&D expenditure (unweighted mean), 2021 or closest available year
Note: This chart displays the percentage of energy (Panel A) and environment (Panel B) related R&D in total intramural business R&D expenditures (unweighted mean) by type of public R&D support received by enterprises: 1) no financial support; 2) only direct support from local and/or national government; 3) only tax support; 4) both direct and tax support.4
Source: OECD (2025[40]), OECD microBeRD project, https://oe.cd/microberd (accessed March 2025).
As a percentage of the total number of patent applications filed by business R&D performing firms (patent family level), 2020 or closest available year
Note: This figure displays the shares of “grey”, high-carbon, low-carbon and environment-related patent applications (patent families) in the total number of patent applications (patent families) filed by business R&D performing firms in 2020 (2019 for Norway) by type of public R&D support received by enterprises: 1) no financial support; 2) only direct support from local and/or national government; 3) only tax support; 4) both direct and tax support. Grey patents capture technologies contributing to improvements in the efficiency of high-carbon technologies; high-carbon patents are related to technologies contributing to the exploration for and extraction of fossil fuels, as well as their transformation and the delivery of fossil fuel products to users; low-carbon patents are those related to climate change mitigation and adaptation; and environment-related patents have wider environmental scope.5
Source: OECD (2025[40]), OECD microBeRD project, https://oe.cd/microberd (accessed March 2025).
Climate change-related factors and other drivers of environmental innovation
Figure 4.25 examines which policy and other factors influence firms' decisions to introduce environmental innovations, showing the percentage of innovative enterprises that assigned high importance to nine different factors, ranging from environmental regulation and taxes to high energy costs. Despite significant differences across countries, the interim results point to four main drivers of green innovations: 1) improving the enterprise's reputation (from 22% in France to 47% in Portugal); 2) current or expected market demand for environmental innovations (from 16% in Austria to 46% in Norway); 3) existing environmental regulation (from 18% in Finland to 36% in the Slovak Republic); and 4) high cost of energy, water or materials (from 14% in Sweden to 41% in Portugal). Government grants, subsidies or other financial incentives for environmental innovations were rated by a comparatively smaller percentage of innovative enterprises (ranging from 8% in Sweden to 22% in Norway) as an important driver.
Percentage of innovative enterprises that rated factors as highly important for environmental innovation decisions
Note: This chart displays the percentage of innovative enterprises introducing an innovation with environmental benefits that contributed significantly to environmental benefits between 2018 and 2020, which assigned high importance to specific factors that influenced their decision to introduce green innovations. It draws on 2020 Community Innovation Survey (CIS) micro-data, with the exception of Austria, where 2014 CIS data are presented instead.
Source: OECD (2025[40]), OECD microBeRD project, https://oe.cd/microberd (accessed March 2025).
Further analysis looks at whether innovation performance is related to the business perceptions of the importance of climate-related factors, namely: 1) climate change-related cost or input price increases; 2) increasing customer demand for climate change mitigating products; 3) climate-related government policies or measures; and 4) impacts of extreme weather conditions. While these do not seem to explain overall innovation levels (Figure 4.26), they seem to help explain the incidence of environmental innovation among innovative companies (Figure 4.27).
Firms that assigned a high importance rating to climate factors have, on average, higher rates of environmental innovation, as follows: 1) 24% for climate change-related cost or input price increases (6% for general innovation); 2) 32% for increasing customer demand (19% for general innovation); 3) 31% for climate-related government policies or measures (10% for general innovation); and 4) 22% for impacts of extreme weather conditions (6% for general innovation).
A complementary cross-country analysis of innovation survey data points in the same direction as the micro-level analysis (Figure 4.28). With the exception of government environmental policies or measures, climate-related factors do not seem to significantly influence innovation but appear to be correlated with different types of environmental innovations. The introduction of technologies reducing total CO2 emissions of end users, for instance, is positively associated with increasing customer demand. At the same time, government environmental policies or measures seem to be the key driver for the introduction of measures reducing CO2 emissions from production processes.
As a percentage of the total number of innovative enterprises
Note: This figure displays the percentage of enterprises introducing an innovation (innovative enterprises) over the 2018-20 period, distinguishing between enterprises that attached high importance to climate-related factors - cost increases, customer demand, government policies, weather conditions - for their business and those that assigned a medium or low importance to such factors or classified them as not relevant.
Source: OECD (2025[40]), OECD microBeRD project, https://oe.cd/microberd (accessed March 2025).
As a percentage of the total number of innovative enterprises
Note: This figure displays the percentage of innovative enterprises introducing an innovation with environmental benefits that contributed significantly to environmental protection ("green innovation") over the 2018-20 period, distinguishing between innovative enterprises that attached high importance to climate-related factors - cost increases, customer demand, government policies, weather conditions - for their business and those that assigned a medium or low importance to such factors or classified them as not relevant. The micro-aggregated statistics presented draw on 2020 Community Innovation Survey (CIS) micro-data.
Source: OECD (2025[40]), OECD microBeRD project, https://oe.cd/microberd (accessed March 2025).
Percentage change in the share of enterprises introducing any innovation and share of innovation-active firms introducing green innovations, by type of climate change-related factor
Note: Regressions at the industry level, including industry, country and firm size fixed effects. The analysis covers Austria, Bulgaria, Croatia, Cyprus, Czechia, France, Hungary, Italy, Lithuania, Luxembourg, Malta, Poland, Portugal, Romania, Slovenia and the Slovak Republic.6 Horizontal lines mark the 95% confidence interval that covers the “true” elasticity with a probability of 95%.
Source: OECD analysis based on micro-aggregated data from Eurostat (2020[36]), Community Innovation Survey 2020 (CIS2020), https://ec.europa.eu/eurostat/cache/metadata/en/inn_cis12_esms.htm (accessed March 2025).
Measurement rationale
Policies beyond those explicitly targeted at incentivising or supporting STI can have a positive effect on environment-related technology development and deployment. There is extensive academic literature documenting, for instance, the impact of emissions-trading schemes (Calel and Dechezleprêtre, 2016[41]), carbon taxes (Aghion et al., 2016[42]) or fuel efficiency standards (Rozendaal and Vollenbergh, 2024[43]) on patented innovations in low-carbon technologies. While policies targeting all sectors of the economy can impact innovation, both positively and negatively (Aghion, Bergeaud and Van Reenen, 2023[44]), environmental policies are likely to play a particularly important role as they shape the markets by incentivising the development and uptake of environment-related innovations. As countries implement stricter environmental policies, many of which overlap, there is an increasing need to summarise the incentives they collectively create to induce change in the economy.
Perspectives from the OECD environmental stringency indicators
Comparing the stringency of environmental policies across countries is not trivial because the mix of policy instruments can vary widely. Some countries, for example, rely relatively more on pricing instruments, such as carbon taxes, while others favour the use of non-market instruments, such as emission limits or standards. The revised OECD Environmental Policy Stringency (EPS) index consists of three equally-weighted sub-indices, which respectively group market-based (e.g. taxes, permits and certificates), non-market-based (e.g. performance standards) and technology support policies (Kruse et al., 2022[45]) (Box 4.12). Technology support policies are further divided into upstream (R&D support) and downstream (feed-in tariffs, auctions) technology support measures. The motivation for creating a separate technology support sub-index is that subsidies for R&D and feed-in tariffs operate differently from market- and non-market-based policies. While the market- and non-market-based components primarily target the negative externalities of emissions, the technology support component also targets positive externalities from R&D, which may lead to suboptimally low investment in the absence of public policy.
The EPS ranges from zero (no policy) to six (most stringent). The stringency of environmental policies is measured in different units. A carbon price is, for example, measured in US dollars per tonne of carbon dioxide (CO2) emissions, while an emissions limit for nitrogen oxides (Nox) is measured in milligrams of pollutants per cubic metre. To aggregate several policy types into a composite index of policy stringency, their stringency needs to be measured on a common scale. To this end, a data-driven approach is taken so that for each policy instrument, the raw data are ordered from the least to the most stringent observation across the 1990-2020 period. The lowest score of zero is assigned to observations with no policy in place. The remaining scores are assigned using the distribution of observations that have the policy in place. The highest score of six is assigned to observations with values above the 90th percentile of observations that have the respective policy implemented. To assign the remaining scores, the difference between the 90th and the 10th percentile is divided into five equal bins that define the thresholds.
The technology support sub-index entails policies that support innovation in clean technologies and their adoption, including:
Public R&D expenditure: The indicator represents the amount spent by the government for R&D on low-carbon energy technologies relative to the size of the country's nominal GDP. It includes renewable energy sources, energy efficiency, CCS, nuclear, hydrogen and fuel cells, other power and storage technologies, as well as other cross-cutting technologies and research as defined by OECD/IEA. It is calculated by dividing a country’s public R&D expenditure by its nominal GDP. The value is multiplied by 1 000 for readability.
Renewable energy support for solar and wind: This indicator represents the level of the price support for solar and wind energy technologies from feed-in tariffs (FIT) and renewable energy auctions, relative to the global levelised cost of electricity (LCOE). Some countries have replaced FITs with auctions. To capture this shift in policy making, the indicator represents the average awarded price from a wind or solar auction for country-year observations that replaced FITs with auction designs. The level of the price support is divided by the global LCOE to account for the decline in the costs of renewable energy production over the past decades. The value is the ratio of the price support (in USD/kWh) to the LCOE (in USD/kWh)
The main limitation of the OECD Environmental Policy Stringency index comes from the set of policies falling outside its coverage: the EPS focuses on policies aimed at curbing greenhouse gas emissions and local air pollution, and within this group of policies, it does not capture regulations across all sectors of the economy. For example, policies that regulate emissions from agricultural production are not included. In countries where agricultural production accounts for a relatively large share of total carbon emissions, the EPS may capture a relatively smaller share of the overall environmental policy mix.
Source: Kruse et al. (2022[45]), “Measuring environmental policy stringency in OECD countries: An update of the OECD composite EPS indicator”, https://doi.org/10.1787/90ab82e8-en.
Across OECD countries, environmental policy stringency based on the market-based and non-market sub-indices has increased steadily since 1990 (Figure 4.29). However, the technology support sub-index, which is based on government R&D expenditures in clean technologies, as well as solar and wind FIT, declined between 2012 and 2019.
Note: OECD average displayed. For each policy instrument, the raw data are ordered from the least to the most stringent observation across the 1990-2020 period. The lowest score of zero is assigned to observations with no policy in place. The remaining scores are assigned using the distribution of observations that have the policy in place. More details on the methodology used for constructing the indices are in Box 4.12.
Source: Kruse et al. (2022[45]), “Measuring environmental policy stringency in OECD countries: An update of the OECD composite EPS indicator”, https://doi.org/10.1787/90ab82e8-en.
The leading countries in terms of the technology support sub-index are Luxembourg, Switzerland and France, while Finland and Japan also show strong performance. On the other hand, Chile, Iceland, the Russian Federation and South Africa had no relevant technology support policy in place in 2020. France, China, Belgium and Indonesia have experienced significant growth in relevant technology support between 2010 and 2020. While many countries for which the EPS index was estimated have experienced a decline in technology support, the two other sub-indices focused on market-based and non-market-based policies exhibit a more uniform increase in stringency across all assessed countries. Chile, China and the United States are the countries that have seen the largest increases in stringency based on these two sub-indices (Figure 4.30).
Note: For each policy instrument, the raw data are ordered from the least to the most stringent observation across the 1990-2020 period. The lowest score of zero is assigned to observations with no policy in place. The remaining scores are assigned using the distribution of observations that have the policy in place. More details on the methodology used for constructing the indices are in Box 4.12.
Source: Kruse et al. (2022[45]), “Measuring environmental policy stringency in OECD countries: An update of the OECD composite EPS indicator”, https://doi.org/10.1787/90ab82e8-en.
References
[44] Aghion, P., A. Bergeaud and J. Van Reenen (2023), “The impact of regulation on innovation”, American Economic Review, Vol. 113/11, pp. 2894-2936, https://doi.org/10.1257/aer.20210107.
[42] Aghion, P. et al. (2016), “Carbon taxes, path dependency, and directed technical change: Evidence from the auto industry”, Journal of Political Economy, Vol. 124/1, pp. 1-51, https://doi.org/10.1086/684581.
[39] Appelt, S. et al. (2022), “Micro-data based insights on trends in business R&D performance and funding: Findings from the OECD microBeRD+ project”, OECD Science, Technology and Industry Working Papers, No. 2022/04, OECD Publishing, Paris, https://doi.org/10.1787/4805d3f5-en.
[27] Aristodemou, L., S. Appelt and F. Galindo-Rueda (forthcoming), “Assessing the direction of R&D funding towards the Sustainable Development Goals (SDGs) with a new classifier and the OECD Fundstat infrastructure”, OECD Science, Technology and Industry Working Papers.
[28] Aristodemou, L. et al. (forthcoming), Assessing the directionality of R&D funding towards societal goals through new data sources and AI-assisted methods.
[29] Aristodemou, L. et al. (2023), “Measuring governments’ R&D funding response to COVID-19: An application of the OECD Fundstat infrastructure to the analysis of R&D directionality”, OECD Science, Technology and Industry Working Papers, No. 2023/06, OECD Publishing, Paris, https://doi.org/10.1787/4889f5f2-en.
[17] Aulie, F. et al. (2023), “Did COVID-19 accelerate the green transition?: An international assessment of fiscal spending measures to support low-carbon technologies”, OECD Science, Technology and Industry Policy Papers, No. 151, OECD Publishing, Paris, https://doi.org/10.1787/5b486c18-en.
[47] Becker, B. (2014), “Public R&D policies and private R&D investment: A survey of the empirical evidence”, Journal of Economic Surveys, Vol. 29/5, pp. 917-942, https://doi.org/10.1111/joes.12074.
[10] Berger, M., C. Criscuolo and A. Dechezleprêtre (forthcoming), The Real Entrepreneurial State? A Taxonomy of Government Venture Capital.
[11] Berger, M., A. Dechezleprêtre and M. Fadic (2024), “What is the role of Government Venture Capital for innovation-driven entrepreneurship?”, OECD Science, Technology and Industry Working Papers, No. 2024/10, OECD Publishing, Paris, https://doi.org/10.1787/6430069e-en.
[48] Bloom, N., J. Van Reenen and H. Williams (2019), “A toolkit of policies to promote innovation”, Journal of Economic Perspectives, Vol. 33/3, pp. 163-184, https://doi.org/10.1257/jep.33.3.163.
[25] Borchardt, S. et al. (2023), Mapping EU Policies with the 2030 Agenda and SDGs: Fostering Policy Coherence through Text-based SDG Mapping, Publications Office of the European Union, https://doi.org/10.2760/110687.
[41] Calel, R. and A. Dechezleprêtre (2016), “Environmental policy and directed technological change: Evidence from the European carbon market”, Review of Economics and Statistics, Vol. 98/1, pp. 173-191, https://doi.org/10.1162/rest_a_00470.
[33] Dechezleprêtre, A. et al. (2023), “Do tax incentives increase firm innovation? An RD design for R&D, patents, and spillovers”, American Economic Journal: Economic Policy, Vol. 15/4, pp. 486-521, https://doi.org/10.1257/pol.20200739.
[36] Eurostat (2020), “Community Innovation Survey 2020 (CIS2020)”, Statistical Office of the European Union, https://ec.europa.eu/eurostat/cache/metadata/en/inn_cis12_esms.htm.
[46] Hall, B. (2019), Tax Policy for Innovation, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w25773.
[9] IEA (2025), Energy Technology RD&D Budgets (database), https://www.iea.org/data-and-statistics/data-product/energy-technology-rd-and-d-budget-database-2.
[15] IEA (2025), The State of Energy Innovation, IEA, Paris, https://www.iea.org/reports/the-state-of-energy-innovation.
[12] IEA (2025), The State of Energy Innovation, https://iea.blob.core.windows.net/assets/6ff289bd-5399-4625-9003-5218c2b9deab/Thestateofenergyinnovation.pdf.
[14] IEA (2024), Advancing Clean Energy Demonstration Projects, https://www.iea.org/reports/advancing-clean-energy-demonstration-projects.
[13] IEA (2024), Clean Energy Demonstration Projects Database, https://www.iea.org/data-and-statistics/data-tools/clean-energy-demonstration-projects-database.
[16] IEA (2023), Government Energy Spending Tracker, https://www.iea.org/reports/government-energy-spending-tracker-2.
[24] IEA (2021), Net Zero by 2050, IEA, Paris, https://www.iea.org/reports/net-zero-by-2050.
[1] IEA (2020), Clean Energy Innovation, https://www.iea.org/reports/clean-energy-innovation.
[8] IEA (2011), IEA Guide to Reporting Energy RD&D Budget/ Expenditure Statistics, IEA, Paris, https://www.iea.org/reports/iea-guide-to-reporting-energy-rdd-budget-expenditure-statistics.
[18] IEA (2011), IEA Guide to Reporting Energy RD&D Budget/ Expenditure Statistics, https://www.iea.org/reports/iea-guide-to-reporting-energy-rdd-budget-expenditure-statistics.
[3] IPCC (2022), Climate Change 2022 – Impacts, Adaptation and Vulnerability, Cambridge University Press, https://doi.org/10.1017/9781009325844.
[23] IPCC (2021), Climate Change 2021 – The Physical Science Basis, Cambridge University Press, https://doi.org/10.1017/9781009157896.
[45] Kruse, T. et al. (2022), “Measuring environmental policy stringency in OECD countries: An update of the OECD composite EPS indicator”, OECD Economics Department Working Papers, No. 1703, OECD Publishing, Paris, https://doi.org/10.1787/90ab82e8-en.
[40] OECD (2025), “Business R&D structure and dynamics and the impact of public support for business R&D”, OECD microBeRD project, https://oe.cd/microberd.
[30] OECD (2025), CSTP Ministerial Issues note: How to direct research and innovation funding and finance to address the climate challenge?, https://cdn-assets.inwink.com/693d4b04-1de5-4c06-808d-2f0c6b423782/6aa8eaf8-0fe5-471a-914f-f0ef810635ab?sv=2018-03-28&sr=b&sig=yryj3srzMRsviZhwPnpbcw%2FeEyUEe1Z4tpasPsvunJ4%3D&se=9999-12-31T23%3A59%3A59Z&sp=r&rscd=inline%3B%20filename%3D%22MIN24-Key%2520I.
[49] OECD (2025), Declaration on Transformative Science, Technology and Innovation Policies for a Sustainable and.
[4] OECD (2024), Declaration on Transformative Science, Technology and Innovation Policies for a Sustainable and Inclusive Future, OECD Legal Instrument, OECD/LEGA/0501, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0501.
[26] OECD (2024), “OECD agenda for transformative science, technology and innovation policies”, OECD Science, Technology and Industry Policy Papers, No. 164, OECD Publishing, Paris, https://doi.org/10.1787/ba2aaf7b-en.
[2] OECD (2024), “OECD Agenda for Transformative Science, Technology and Innovation Policies”, OECD Science, Technology and Industry Policy Papers, No. 164, OECD Publishing, Paris, https://doi.org/10.1787/ba2aaf7b-en.
[31] OECD (2023), Government Support for Business Innovation: Results from Measurement Pilots in Five OECD Countries, OECD unclassified document, https://one.oecd.org/document/DSTI/STP/NESTI(2023)3/FINAL/en/pdf.
[20] OECD (2023), Low-Carbon Technology Support Database, https://www.oecd.org/en/topics/sub-issues/quantifying-industrial-strategies.html.
[32] OECD (2023), “OECD framework for mapping and quantifying government support for business innovation”, OECD Science, Technology and Industry Policy Papers, No. 160, OECD Publishing, Paris, https://doi.org/10.1787/f30547e7-en.
[37] OECD (2023), “The Impact of R&D tax incentives: Results from the OECD microBeRD+ project”, OECD Science, Technology and Industry Policy Papers, No. 159, OECD Publishing, Paris, https://doi.org/10.1787/1937ac6b-en.
[38] OECD (2020), “The effects of R&D tax incentives and their role in the innovation policy mix: Findings from the OECD microBeRD project, 2016-19”, OECD Science, Technology and Industry Policy Papers, No. 92, OECD Publishing, Paris, https://doi.org/10.1787/65234003-en.
[21] OECD (2015), Daejeon Declaration on Science, Technology and Innovation Policies for the Global and Digital Age, OECD Legal Instruments, OECD/LEGAL/0416, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0416.
[5] OECD (2015), Frascati Manual 2015: Guidelines for Collecting and Reporting Data on Research and Experimental Development, The Measurement of Scientific, Technological and Innovation Activities, OECD Publishing, Paris, https://doi.org/10.1787/978926.
[6] OECD (n.d.), Main Science and Technology Indicators (MSTI) Database, http://www.oecd.org/sti/msti.htm.
[7] OECD (n.d.), Research and Development Statistics Database, http://www.oecd.org/sti/rds.
[19] OECD/Eurostat (2018), Oslo Manual 2018: Guidelines for Collecting, Reporting and Using Data on Innovation, 4th Edition, The Measurement of Scientific, Technological and Innovation Activities, OECD Publishing, Paris/Eurostat, Luxembourg, https://doi.org/10.1787/9789264304604-en.
[34] Pless, J. (2019), “Are ’complementary policies’ substitutes? Evidence from R&D subsidies in the UK”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.3379256.
[35] Pless, J. and S. Srivastav (2022), Unintended consequences of tech-neutrality: Evidence from environmental and innovation policy interactions, Working paper.
[22] Roy, J. et al. (2018), “Sustainable Development, Poverty Eradication and Reducing Inequalities”, in Masson-Delmotte, V. (ed.), Global Warming of 1.5°C. An IPCC Special Report on the Impacts of Global Warming of 1.5°C Above Pre-industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change, Cambridge University Press, https://doi.org/10.1017/9781009157940.007.
[43] Rozendaal, R. and H. Vollenbergh (2024), Standards for Greenhouse Gas Emissions and Fuel Economy Induce Innovation in Clean Car Technologies.
← 1. The focus on the central government and incomplete coverage of subnational business innovation support policies may potentially influence the distribution of business innovation support by SEO. Within the Canadian context, it is important to note that domains such as energy, environment, education (including universities), health, agriculture and transportation fall under the purview of provincial authorities. Conversely, the federal government assumes primary responsibilities in areas like defence, space and telecommunications.
← 2. Becker (2014[47]) and Bloom, Van Reenen and Williams (2019[48]) take a broad look at the various policies for boosting business innovation. For an excellent survey of the literature on tax policy for innovation, including income-based R&D tax incentives (“patent boxes”), see Hall (2019[46]).
← 3. In the case of Austria, France, Portugal and Sweden, figures refer to 2018‑20 instead of 2020‑22. Small firms are defined as firms with 10-49 employees, medium-sized firms as those with 50-249 employees, and large firms as those with 250 or more employees. Green innovation rates are reported for innovative enterprises by type of public R&D support received: 1) no financial support; 2) only direct support from local and/or national government; 3) only tax support; 4) both direct and tax support. Information from national innovation surveys was used to identify enterprises that received direct support. R&D tax relief micro-data, where available (Belgium, Czechia, France, Italy, Norway, Portugal and Sweden), were used to identify R&D tax relief recipients.
* In the case of Austria, Japan and the Slovak Republic, information from national innovation surveys was used to identify R&D tax relief recipients.
** In the case of Finland, the analysis focuses on direct government funding, as broader tax deductions for R&D salaries and purchased R&D services were only introduced in Finland in 2023.
*** In the case of France, a breakdown by firm size is not available, and figures refer to the total population of innovative enterprises, including micro firms (less than 10 employees).
← 4. In the case of Israel, Norway, Portugal and Sweden, figures refer to 2022 instead of 2021, while those for Italy refer to 2020. In the case of Austria, Germany, Israel and Japan, the analysis focuses on direct government funding. Israel did not offer tax incentives for R&D expenditures over the 2000-24 period, and Germany only introduced R&D tax incentives in 2020, the amount of this support being still very low in scale in 2021, the first year of operation.
* In the case of Austria and Japan, the analysis abstracts from R&D tax support as relevant R&D tax relief micro-data are not available.
** In the case of Israel, Norway and Sweden, the figures reflect the maximum value of the shares observed for subcategories of energy (environment) related R&D and thus provide only a lower bound for the total share of energy (environment) related R&D. This approach was adopted due to the partial overlap of R&D related subcategories, preventing the direct aggregation of values to an aggregate variable on energy (environment) related R&D.
← 5. For additional details on the definition and measurement of patents by sustainability category, see Chapter 2 (Box 2.8). At the time of reporting, relevant micro-aggregated statistics are available for France, Japan, and Norway.
* In the case of Japan, the analysis abstracts from R&D tax support as relevant R&D tax relief micro-data are not available.
** In the case of Norway, the analysis focuses exclusively on national patent applications.
← 6. The 11 dependent variables correspond to: 1) the share of innovation-active firms, i.e. enterprises engaged at some time during the observation period of 2018-20 in one or more activities to develop or implement new or improved products or business processes (“Any type of innovation”); and 2) the share of innovation-active firms introducing ten different types of green innovations over the 2018-20 period (see figure legend for the list of ten “green innovations”). The four independent variables correspond to the share of innovation-active firms that assigned a high importance rating to a specific climate-related factor: 1) increasing customer demand for green products; 2) increasing costs or input prices of non-green products; 3) impact of extreme weather conditions; and 4) government environmental policies and measures - for their business.