Industrial policies have regained importance in recent years and are now a crucial element of science, technology and innovation policy portfolios. Adopting an industrial ecosystem perspective – namely going beyond sectoral boundaries to consider both upstream and downstream industries, as well as the diverse set of stakeholders involved – can help design more effective industrial policies. This chapter distils the insights from three recent studies, offering practical examples of how to define the boundaries, stakeholders and challenges of the automotive, renewable energy and energy-intensive industrial ecosystems. It highlights the value of adopting an ecosystem perspective and the importance of relying on robust evidence coming from diverse data sources. The chapter provides insights on policies that foster growth and support thriving, resilient industrial ecosystems and economies.
OECD Science, Technology and Innovation Outlook 2025
Abstract
Well-designed industrial policies – which include a variety of instruments such as research and development (R&D) grants, government venture capital, tax expenditures, loans and loan guarantees, etc. – can help address complex challenges like slowing productivity growth, the resilience of global value chains and the transition to a low-carbon economy.
Sectoral policies that are often taken to implement industrial policies are not necessarily well suited to address these challenges as they do not account for key actors located outside sectoral boundaries or for the interdependencies linking them.
More effective is an industrial ecosystem approach that identifies all relevant stakeholders associated with a given technology or product, including large and small firms, start-ups, technology providers, workers, trade partners, and investors.
The industrial ecosystem perspective can help policymakers design better targeted and more effective policies that account for interdependencies between upstream, core and downstream stakeholders.
Transitioning to an industrial ecosystem approach entails developing a robust data infrastructure that brings together granular data from multiple sources.
Innovation and industrial policies for industrial ecosystems can help address major challenges such as access to critical inputs, the lack of skills, and barriers to technology development and diffusion.
Ecosystem-based industrial policies represent an attractive middle ground between sectoral policies that are too narrow in scope and horizontal approaches that are not necessarily sufficient to address current challenges.
OECD countries are facing an increasing number of challenges – including slowing productivity growth, the transition to a low-carbon economy and the uncertainty caused by a tense geopolitical landscape. There are also major opportunities, including the emergence of artificial intelligence. Major crises, like the COVID‑19 pandemic and the Russian Federation’s war of aggression against Ukraine, have exacerbated these mounting challenges and shown the fragility of global value chains, contributing to the renewed interest of many economies in industrial policy.
In this context, designing, implementing and evaluating industrial strategies aimed at enhancing productivity, fostering value chain resilience, and accelerating the development and diffusion of novel technologies has become a policy priority. A growing body of research has shown that purposeful industrial strategies can contribute to achieving these goals (Criscuolo et al., 2022[1]; Lane, 2020[2]; Criscuolo et al., 2019[3]; Dechezleprêtre et al., 2023[4]), especially when industries are characterised by spillover effects, benefit from economies of scale or are impeded by co-ordination failures.
The challenges facing OECD economies are unprecedented in both scope and complexity. The twin transition toward digital and sustainable technologies, for example, involves stakeholders not only from a variety of different institutions, including the private sector, government, academia and civil society, but also from very different sectors of economic activity. Designing policies capable of addressing the needs of stakeholders as different as multinational firms active in the energy sector, small start-ups designing specialised predictive maintenance software and local consumers associations represents a complex challenge, one that requires robust empirical evidence and strong, adaptable collaboration among all the parties involved.
Sectoral strategies, despite being a staple of industrial policy (Criscuolo et al., 2022[5]), are ill-suited for facing such an array of challenges. First, by focusing their interventions on firms belonging to a single sector, they can miss critical upstream suppliers of inputs, without whom downstream firms, even if adequately supported, may fail to thrive. Similarly, lack of support for downstream firms may stifle upstream firms unable to find adequate demand for their outputs. Upstream and downstream sectors are critical in a context where value chains are characterised by severe bottlenecks and are subject to natural and geopolitical threats. Beyond networks of production, sectoral strategies can fail to properly take into account providers of technologies that lie outside the boundaries of a given sector, as well as valuable human capital. These are essential to consider in a scenario of slow and dispersed productivity growth (Berlingieri, Blanchenay and Criscuolo, 2017[6]; Berlingieri et al., 2020[7]), where innovations appear to be increasingly difficult to find (Bloom et al., 2020[8]).
The industrial ecosystem approach aims to overcome these shortcomings by explicitly accounting for the wealth of actors and relationships that underpins modern industrial production. Figure 6.1 portrays a schematic, simplified representation of an industrial ecosystem, which “encompasses all players operating in a value chain, from the smallest start-ups to the largest companies, from academia to research, service providers to suppliers” (European Commission, 2020, p. 16[9]). The figure includes three main blocks of sectors – core, upstream and downstream – each composed of a variety of actors, including large, established firms that are commonly associated with a given industry; smaller firms and start‑ups; and academic research centres and finance providers. Core sectors include the firms identifiable through (and typically targeted by) a sectoral approach: for example, firms belonging to the manufacture of motor vehicles are at the core of the automotive ecosystem. Upstream sectors supply inputs such as raw materials, intermediate goods, capital equipment and technologies. Downstream sectors use the outputs of core industries as inputs for further production, for further use by installers or other service providers, or for final demand. For instance, the steel industry (upstream) provides inputs to automotive manufacturers (core), while the car retailers (downstream) use outputs from car manufacturers. Note that the figure provides a simplified version of the many actors and relationships underpinning ecosystems: for example, it does not include the many feedback loops that link upstream, core and downstream sectors.
Source: OECD.
The industrial ecosystem concept – rooted in an analogy between economic and biological ecosystems (Moore, 1993[10]) – draws heavily on similar paradigms, such as national innovation systems (Nelson, 1993[11]; Lundvall, 1992[12]; Freeman, 1995[13]), regional innovation systems (Cooke, 2004[14]), local clusters (Porter, 1998[15]), sectoral systems of innovation (Malerba, 2002[16]) and entrepreneurial ecosystems (Stam, 2015[17]; Stam and van de Ven, 2021[18]). Innovation ecosystems, in particular, share many characteristics with industrial ecosystems, both of which trace their origins to the concept of the National Innovation System – a framework to which the OECD contributed substantially (OECD, 1999[19]), especially thanks to the work of Freeman in the 1990s. As the field progressed, the OECD also contributed to the above-mentioned related fields, especially on sectoral innovation systems (OECD, 2006[20]).
Both these concepts emphasise that firms (or entrepreneurs) cannot be regarded as operating in a vacuum but are rather interwoven in a complex web of relationships that includes several different actors. As these actors depend on each other for their survival and growth, “ecosystems” represents not only a crucial unit of analysis for business leaders (Jacobides, Cennamo and Gawer, 2018[21]; Iansiti and Levien, 2004[22]) but also the most suitable unit of analysis for policymakers interested in strengthening their economy’s resilience and promoting their growth (Box 6.1). However, key differences exist between the two types of ecosystems. Their objectives differ: innovation ecosystems focus on fostering collaboration in research, development and commercialisation of new technologies that address shared priorities whereas industrial ecosystems aim to increase the value added generated within a specific industry. Additionally, their boundaries are distinct. Innovation ecosystems are not constrained by industrial considerations and broadly encompass all actors contributing to innovation. In contrast, industrial ecosystems have a narrower sectoral scope but include actors who may not directly contribute to innovation yet play a crucial role in the ecosystem’s overall success.
Industrial policy has been a crucial component of economic strategies at least since the Industrial Revolution (Juhász and Steinwender, 2023[23]). Nevertheless, the use of active industrial policies had fallen out-of-favour at the end of the 20th century (Warwick, 2013[24]) due to concerns regarding governments’ ability to identify the most promising areas for investment, the risk of political capture of subsidies and budget constraints. Emerging trends and challenges like those described at the beginning of this chapter – productivity slowdown, the climate crisis, shocks to global supply chains and uncertainty caused by geopolitical tensions – have contributed to making industrial policies attractive again (Juhász, Lane and Rodrik, 2024[25]; Mazzucato, 2021[26]).
OECD countries have recently begun to design coherent industrial strategies again. For example, in 2025 the UK Government (2025[27]) recently published the UK’s Modern Industrial Strategy, and Italy (Ministry of Enterprises and Made in Italy, 2024[28]) has published its industrial strategy delineating its industrial priorities and how the Italian government aims to support them. Similarly, numerous large-scale policy initiatives, such as the “European Green Deal” (2019), the “Next Generation EU” Fund (2020), the “Korean New Deal” (2020), the “Inflation Reduction Act” (2022), the “EU New Industrial Strategy” (2020, updated in 2021), the “US CHIPS and Science Act” (2022), and the “EU Green Deal Industrial Plan” (2023), have shown governments’ commitment to a more active role in driving industrial development. The Draghi Report (Draghi, 2024[29]) also highlighted the growing importance of industrial policies, calling for better co-ordination across the European Union to ensure their effectiveness.
Ecosystems – industrial ones (Andreoni, 2018[30]), as in the focus of this chapter, but similarly innovation (Adner, 2006[31]) and entrepreneurial (Stam and van de Ven, 2021[18]) ones – represent natural targets for industrial policy interventions, as they avoid narrowness concerns that may otherwise hinder targeted industrial policies (Criscuolo et al., 2022[1]). As such, ecosystem thinking has already permeated policymaking initiatives. For example, the Dutch “Top Sector” approach has incorporated systemic considerations of upstream industries. Nevertheless, recent emergencies, such as the COVID-19 pandemic, have highlighted the need to further expand the scope of systemic analysis. For example, the European Commission adopted the industrial ecosystem concept in its 2020 industrial strategy, to subsequently bring it into even sharper focus in the 2021 amendment, where the Commission described industrial ecosystems as the lenses though which it analyses the EU Single Market Economy.
This chapter builds on the industrial ecosystem approach and draws on three sectoral case studies: the automotive, renewable energy and energy-intensive industries (EII) ecosystems. All three ecosystems are affected by (at least some) of the challenges delineated above: for example, the automotive ecosystem has experienced severe disruption in its value chain linked to chip shortages during the COVID-19 pandemic; furthermore, this ecosystem is characterised by increasing digitalisation and the need to reduce the carbon footprint of its end products. The renewable energy ecosystem is at the forefront of the transition toward a low-carbon economy, but at the same time is markedly dependent on several critical minerals that have the potential to turn into bottlenecks. EIIs are crucial upstream industries whose outputs are incorporated into a wide variety of downstream industries. They are, therefore, crucial for the competitiveness of multiple sectors but also face challenges related to their high and only slowly decreasing emissions intensity. Consequently, these three ecosystems constitute particularly insightful cases on the opportunities provided by the industrial ecosystem approach.
The remainder of this chapter focuses on showcasing how the industrial ecosystem approach, as operationalised by these three studies, can help policymakers address the challenges delineated at the beginning of the chapter. First, the industrial ecosystem approach can provide evidence on who the main actors in an ecosystem are, what role they play and their linkages. Second, the industrial ecosystem approach can shed light on the challenges, bottlenecks and dependencies that these actors face. Third, it can aid in designing more effective industrial strategies.
The defining characteristic of an industrial ecosystem is that it extends beyond traditional sectoral boundaries to include both upstream and downstream activities, as well as a broader range of stakeholders than just private companies. This requires establishing criteria to define non-sectoral boundaries and identify key actors within the ecosystem. This section outlines the approach used in the three industrial ecosystem studies, with a particular focus on the diverse data sets used to define ecosystem boundaries, including input-output, trade, innovation, and workforce data. It also highlights the variety of stakeholders included in each industrial ecosystem. The section emphasises how relying exclusively on sectoral classifications would overlook critical inputs essential for thriving ecosystems (including raw materials and innovations) and would also provide a distorted view of the leading actors and the significance of a given ecosystem, for instance in terms of workforce.
Identifying relevant ecosystem stakeholders using multiple data sources
There are various approaches to identifying the boundaries of an ecosystem and, within it, the most important firms, sectors and economies. Despite the importance of moving beyond an exclusively sectoral approach, industrial classifications such as the International Standard Industrial Classification (ISIC), the Statistical Classification of Economic Activities in the European Community, and the North American Industry Classification System, remain important building blocks for defining an ecosystem.
Identifying the core sectors composing an ecosystem can be relatively straightforward or complex, depending on the specific type of ecosystem. Table 6.1 presents the sectors at the heart of the automotive, renewable energy and EII ecosystems. On one end of the spectrum, the “Manufacture of motor vehicles, trailers and semi-trailers” (ISIC rev. 4 Division 29) is a natural starting point for defining the core component of the automotive ecosystem. For the EII ecosystem, defining core sectors already requires additional assumptions, as there is currently no internationally agreed-upon definition of what an EII is. The approach adopted in Dechezleprêtre et al. (2025[32]) is to consider as core parts of the EII ecosystem all the 2-digit ISIC manufacturing sectors whose energy input cost share over total inputs costs is above the median. These sectors include: “Coke and petroleum” (ISIC rev. 4 19), “Non-metallic minerals” (ISIC rev. 4 23), “Chemicals” (ISIC rev. 4 20), “Basic metals” (ISIC rev. 4 24), “Paper” (ISIC rev. 4 17), “Rubber and plastics” (ISIC rev. 4 22), and “Wood” (ISIC rev. 4 16). Finally, at the other end of the spectrum, the renewable energy ecosystem core is particularly difficult to define based only on specific sectors. No sector at the 4‑digit level of the ISIC classification corresponds to the production of electricity specifically from renewable sources, and a sector corresponding to the manufacture of solar cells, solar panels and photovoltaic inverters was only created as part of the ISIC Rev. 5. In addition, market leaders in electricity production often produce electricity from various sources, including fossil, nuclear and renewable ones, making clear‑cut sectoral distinctions especially complex. Firms manufacturing other capital goods necessary for the production of renewable energy (e.g. hydraulic turbines and wind turbines) can be found in the “Manufacture of engines and turbines sector” (ISIC rev. 4 2811).
A lesson learnt across the three studies is that increasing the level of granularity of the analysis can help to better delineate ecosystems’ boundaries, as within broadly defined sectors firms are highly heterogeneous. For example, there is considerable heterogeneity in energy intensity among 4-digit sectors within the same 2-digit sector, and there is also vast heterogeneity among firms within 4-digit sectors (De Lyon and Dechezleprêtre, Forthcoming[33]). Therefore, a definition of the energy-intensive ecosystem based on energy input cost share over total input costs defined at the 2-digit, 3-digit, 4-digit or firm level would give very different results. Unfortunately, granular data are often difficult to obtain and unevenly available across industries and geographies. Furthermore, there is typically a trade-off between the granularity of the data and their sectoral coverage, which often makes it unfeasible to conduct analysis at a highly granular level, despite the value of such analysis for policymaking.
The identification of the core components of an ecosystem needs to be integrated with the identification of all its other non-core parts, which nevertheless play an important role in the ecosystem. Input-output relationships, trade relationships, technological and workforce relationships can all contribute to the definition of an ecosystem beyond its core components. The following sub-sections examine each of these linkages and explain how they are operationalised in the three studies on ecosystems.
Notes: Only the hydraulic turbines and wind turbines subset of the “Manufacture of engines and turbines” is considered relevant. The “Manufacture of solar cells, solar panels and photovoltaic inverters” was only added to ISIC rev. 5.
Input-output data
Upstream and downstream relationships are crucial for the delineation of an ecosystem. Taking them into account allows identifying critical inputs that core sectors include in their production, and vital markets that core sectors rely upon for selling their outputs.
Input-output tables make it possible to analyse both upstream and downstream linkages. For example, the significance of upstream sectors in the automotive ecosystem can be determined by assessing how much of the value added embedded in the final demand for automotive products originates from other sectors. Conversely, the importance of downstream linkages can be measured by examining the portion of value added in the final demand of other sectors that can be attributed to the automotive industry. In other words, upstream linkages are reflected in the portion of the value of motor vehicle production generated by sectors other than automotive manufacturing itself (e.g. the chips controlling airbags or the metal of the car frame), while downstream linkages are represented by the portion of value added generated by the motor vehicles sector embodied in the production of other sectors (e.g. the transportation sector uses outputs from the automotive ecosystem). OECD Trade in Value Added (TiVA) data1F1 reveal that in the automotive sector, upstream linkages are especially important while downstream linkages are less so, i.e. value added produced in the motor vehicle sector does not contribute significantly to other final products. Notably, TiVA data have limitations: first, they provide only an aggregated view of value flows, which does not allow for granular product-level analysis; second, they do not encompass capital investment when it comes to intermediate transactions. Other data sources, like value-added tax data (Criscuolo et al., 2024[34]) can, therefore, be used to complement them.
Figure 6.2 shows the production network underpinning the EII ecosystem. It shows the breadth of both upstream and downstream linkages, highlighting how numerous sectors contribute to the value added by core EII industries, while many downstream sectors benefit from EII outputs. For example, the “Construction” sector is closely linked to “Non-metallic minerals”, as well as “Rubber and plastics”, “Wood”, “Chemicals”, and “Basic metals”. Similarly, the “Motor vehicle” and “Electrical equipment” downstream industries depend on various EII core sectors, pointing to EIIs’ overarching relevance as providers of critical inputs.
Notes: The edges are weighted by the magnitude of the intermediate input flows while the nodes are weighted by the value added of each sector. The “Paper” sector is aggregated with “Printing and reproduction of recorded media” (ISIC 18 – rev. 4) in the underlying data. The figure only shows the seven main upstream and downstream sectors for the whole ecosystem. “Wholesale and retail” is both an upstream and a downstream sector but is located among the upstream sectors given that the value of its inputs supplied to energy-intensive industries are higher than for the inputs it sourced from them. Manufacturing n.e.c. is not included as a downstream sector given its general definition, instead the next downstream sector is reported (“Electrical equipment”).
Sources: Dechezleprêtre et al. (2025, p. 23[32]), based on the OECD Inter-Country Input-Output (ICIO) database, 2023 edition.
Having identified both core and non-core components of an ecosystem, it is also possible to define which countries play the most relevant role within it. Figure 6.3 highlights the countries (and sectors) that contribute the highest share of value added embedded in the final demand for motor vehicles. The figure serves as a sharp reminder of the importance of accounting for the entire ecosystem, as the “Rest of the ecosystem” contributes substantially, and heterogeneously across countries, to the automotive value added. For example, although the European Union’s (EU) motor vehicle sector remains the largest contributor among motor vehicle sectors, the People’s Republic of China (hereafter “China”), when accounting for the entire ecosystem, has a higher share of value added. Similarly, the United States’ automotive core sector’s value added is lower than Germany’s, but the rest of the US ecosystem generates considerably more value added than the German one.
Notes: The graph can be interpreted as follows: 23% of global value added embodied in final demand for motor vehicles comes from the European Union. Among these, 9.5 percentage points come from the automotive sector, while 13.5 percentage points come from the rest of the ecosystem.
Sources: Adapted from Dechezleprêtre et al. (2023, p. 22[35]), based on OECD, Trade in Value-Added (TiVA) Database, https://www.oecd.org/en/topics/sub-issues/trade-in-value-added.html (accessed in February 2022).
Trade data
Trade data can provide additional insights beyond value-added data, particularly in ecosystems where sectoral boundaries are less clear-cut. In such cases, granular information on specific products or technologies becomes especially relevant. The renewable energy ecosystem is a clear example of this, as it does not correspond to a single sector but rather comprises specific capital goods, products and raw materials. Due to their granularity, trade data are particularly well-suited to capturing the actors and relationships within such ecosystems. However, it is important to note that trade data fail to capture production that does not cross any border, pointing to the complementarity between trade and value-added data.
Figure 6.4 provides an example of the level of detail achievable when analysing trade data. It shows the main exporters of selected capital goods that are at the heart of four key renewable technologies: solar photovoltaic, wind power, solar thermal and hydropower. The analysis rests on the identification of a set of specific products, defined at the 6-digit level of the Harmonised System classification, for each of these technologies, rather than on a set of specific sectors. Therefore, while value-added data would not be able to identify key players active in the ecosystem, trade data can show export trends of capital goods across technologies. For example, China appears as a key player in the renewable energy ecosystem, being the most sizeable exporter in all technologies but solar thermal, where Mexico plays the leading role. Furthermore, trade data, similarly to value-added data, can be used to portray the flows of upstream goods. In the case of the renewable energy ecosystem, these correspond notably to critical raw materials, which will be a cornerstone of the trade dependency analysis presented in the next section.
Another advantage of trade data over value-added data is that they are more readily available and can therefore provide more up-to-date snapshots of the key countries active in an ecosystem than value-added data can. For a direct comparison, the EII ecosystem paper (Dechezleprêtre et al., 2025[32]) considers both value-added and trade measures, but the value-added measures are only available up to 2020 while the trade measures are available up to 2022.
Note: The figure shows the three largest exporters by technology. Data are reported in current USD billions. RoW = rest of the world.
Source: Dechezleprêtre et al. (2024, p. 15[36]), based on UN Comtrade database.
Innovation data
The importance of countries within an ecosystem is not limited to their contribution to value added and trade but also stems from their involvement in innovation activities. This is crucial to consider as innovation can provide a forward-looking perspective on the ecosystem: while value added and trade portray today’s leading actors, innovation measures provide a glimpse on who is developing the capabilities that can help become tomorrow’s leader.
There are several measures of innovative activities, including the number of active start-ups, investment in innovative firms, academic publications, among many others. Patents (OECD, 2009[37]), despite not being a perfect measure of innovation (Dziallas and Blind, 2019[38]; Acs and Audretsch, 1989[39]), include a wealth of information that can be especially helpful in describing innovation landscapes, and reveal information on the key actors of an industrial ecosystem (Supriya, 2023[40]).
The technological domain of patents can be identified based on patent classifications such as the Cooperative Patent Classification and the International Patent Classification. Dechezleprêtre et al. (2025[32]) use the correspondence table developed by Goldschlag, Lybbert and Zolas (2020[41]) to identify patents protecting technologies used in the EII ecosystem (such as innovations in steel, plastic or chemical products).2F2 It is then possible to recover the sector of activity in which companies developing these patents operate. The data reveal that inventions relevant for the EII ecosystem are not only developed by firms operating in core EII sectors, companies from other industries may also develop innovations relevant to EIIs. For example, a company operating in the machinery sector might invent and patent a technology for a machine used in wood cutting. Figure 6.5 portrays patents filed in EII technologies (on the y-axis) and the sector of the firms filing them (on the x-axis). Core EII sectors, highlighted in bold, appear to be among the leading innovators in these technologies, as can be inferred by the size of the dots in the figure. However, only 30% of the patents filed in EII technologies in the period 2018-2022 were filed by firms in core EII sectors, indicating that most EII patents were developed outside the boundaries of the core sectors. Among these non-core sectors, “Computers and electronics” play an especially important role, as the sector filed about 21% of EII-related inventions worldwide, being the leading source of innovation for “Non‑metallic minerals” technologies (with 36% of total patents). Firms belonging to the “Machinery” sector also contribute significantly to new EII technologies, especially in the fields of “Paper” (30% of patents related to “Paper” are from the “Machinery” sector) and “Wood” (23% are from the “Machinery” sector). Overall, this figure emphasises that limiting the analysis to EII core sectors would miss key innovation providers, supporting the importance of moving beyond a solely sectoral approach.
Notes: EII: energy-intensive industry. Data refer to IP5 patent families, by earliest filing date and applicant’s location. See Annex B of Dechezleprêtre et al. (2025[32]) for further details on the patent data coverage and methodology.
Source: Dechezleprêtre et al. (2025, p. 43[32]), based on OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, and ORBIS©, version 2022.1, Bureau van Dijk, October 2024.
This is further confirmed in Figure 6.6, which portrays countries’ revealed technological advantage (RTA) for EII technologies, calculated as a country’s share in EII patents compared to its share in total patents. As the range of possible values for the RTA goes from 0 to +∞, the RTA is shown in logs. In the figure, a log(RTA) greater than zero indicates that a country is specialised in a particular technology, compared to the world average. A log(RTA) below 0 indicates under-specialisation. The figure distinguishes between two different sources of innovation: core EII sectors (see previous sections for the complete list) and all other industries that generated patents in EII technologies. The figure clearly illustrates the importance of adopting an ecosystem approach, as the size of dark blue bars is often overshadowed by light blue ones, indicating that a country’s specialisation in EII technologies stems primarily from sectors outside the core components of the EII ecosystem. Brazil, Luxembourg and Poland are clear – but not unique – cases of this tendency. The figure also shows how drastically the ranking of countries would change if only core sectors were considered. Finland, whose specialisation in EII technologies is entirely driven by core EII sectors, leads the ranking in Figure 6.6, but the scenario is markedly different when considering the rest of the ecosystem. For example, Brazil, which does not rank high in terms of value added or employment among the countries most involved in EIIs, shows very strong technical specialisation in EII technologies, driven by non-core EII sectors (especially mining and quarrying). A sectoral approach would likely overlook these peculiarities of the EII innovation landscape, underscoring the value of the ecosystem approach for policymakers.
Notes: EII: energy-intensive industry. Data refer to IP5 patent families, by earliest filing date and applicant’s location. Only economies with a high matching rate to ORBIS© with more than 100 IP5 patent families in total for each set of industries are included. IP5 patent families are defined as sets of patent applications protecting the same invention filed in at least two intellectual property (IP) offices – with at least one application filed in one of the five largest IP offices worldwide (IP5): the European Patent Office, the Japan Patent Office, the Korean Intellectual Property Office, the State Intellectual Property Office of the People’s Republic of China, and the United States Patent and Trademark Office. Patents for inventions related to EIIs are delineated using the concordance developed by Goldschlag, Lybbert and Zolas (2020[41]) that maps codes of the co-operative patent classification to the industry classification (ISIC, rev. 4), using a probabilistic approach. Only economies with more than 1 000 IP5 patent families in total for each period are included.
Sources: OECD calculations based on OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, and ORBIS©, version 2022.1, Bureau van Dijk, March 2025.
In the automotive area, adopting an ecosystem approach helps identify the fastest growing technologies, which in turn can enable the identification of economies that are building the capabilities that should be key for future growth in automotive. From this perspective, the surge in patents related to autonomous vehicles – which more than doubled between 2012-2015 and 2016-2019 – stands out as a key trend likely to shape the ecosystem’s evolution in the coming years. Within the automotive ecosystem, countries differ in their technological specialisation. The United States, for instance, holds a strong RTA in the autonomous vehicle segment while other countries focus on mature, less innovation-intensive areas such as combustion technologies. This suggests a stronger potential for future leadership of the United States in the automotive ecosystem than would be suggested by considering their overall RTA across all automotive-related technologies.
Workforce data
Data on the workforce can also contribute to defining ecosystem boundaries, as a non-negligible portion of the occupations generated by a given industrial ecosystem lie outside the boundaries of its core sectors. Figure 6.7 portrays this for the specific case of the renewable energy industrial ecosystem in the United States (but this holds true for the other countries considered in the analysis).3F3 The figure shows that renewable energy vacancies are spread across various industries, regardless of the observed technology group. Most vacancies are concentrated in “Manufacturing” and “Electricity, gas, steam and air conditioning supply”, but employment is not limited to these core renewable sectors, including manufacturers of capital goods and adopters (such as firms in the electricity production sector). Many renewable energy vacancies are found in the “Professional, scientific and technical activities”, where renewable innovations are being developed, as well as in the “Construction”, “Finance and insurance”, and “Administrative and support service activities”. Overall, only approximately 39% of all vacancies are in what might be defined as core sectors, while approximately 61% are in other parts of the ecosystem.
This reinforces a key point from the previous section on innovation data, and a fundamental rationale behind the adoption of an ecosystem perspective: accounting only for the core sectors of an ecosystem overlooks a critical share of relevant stakeholders. This represents arguably the most crucial message for policymaking coming from the ecosystem perspective.
Note: The figure displays the distribution of renewable energy vacancies across sectors and technology groups in 2022 for the United States. Observations with missing sectors have been removed and not counted. The intensity of red colour corresponds to the share of vacancies, with all red-coloured cells collectively summing up to 100%. The intensity of blue corresponds to the sum of shares either across rows or across columns. Industry information is sourced through the North American Industry Classification System in the United States and aggregated into 1-digit sections for better comparability. Fractional counting is employed for cases involving multiple technologies. If a vacancy has been tagged through both a generic and a specific keyword, only the specific technology is considered. Technology groups are assigned to renewable energy vacancies based on the keywords used to identify them.
Source: Adapted from Dechezleprêtre et al. (2024, p. 44[36]), based on Lightcast data.
Accounting for the heterogeneity of industrial ecosystems stakeholders
It is also important to account for the heterogeneity of actors involved in industrial ecosystems. For instance, different types of firms – in terms of age, size or geographical location – have different roles in industrial ecosystems and tend to benefit from different sets of policies (Criscuolo et al., 2022[5]). Additionally, firms located upstream, at the core or downstream in a given ecosystem are also likely to play different roles, and the design of industrial strategies can benefit from taking these differences into consideration. Finally, ecosystems’ actors not only encompass firms but also public research organisations (PROs) and finance providers, among others, that need to collaborate to ensure the ecosystem’s success. This view is at the heart of mission-oriented policies: proponents of this approach acknowledge the need to consider and co-ordinate “systems” of stakeholders to address major challenges – like the transition toward a low-carbon economy (OECD, 2024[42]) – that are systemic in nature (Larrue, 2021[43]).
Start-ups are a category of firms commonly under policymakers’ lenses, due to their contribution to net job creation (Criscuolo, Gal and Menon, 2014[44]) and the deployment of novel technologies (Audretsch et al., 2020[45]). Despite their crucial role in driving economic growth, start-ups often face greater challenges than established firms due to their lack of legitimacy (Lounsbury and Glynn, 2001[46]) and limited financial history (Da Rin, Hellmann and Puri, 2011[47]). As a result, they benefit more from certain policies, particularly those that ease capital constraints and facilitate market entry and exit (Criscuolo et al., 2022[1]). One example of start-ups’ importance for innovation activities comes from the renewable energy ecosystem: in this ecosystem, incumbent firms (aged 20 years or more) file most inventions related to renewable energy. However, as Figure 6.8 indicates, younger firms (less than five years old) are not only filing a considerable share of patents in renewable energy technologies, but also a disproportionately high number relative to their contribution to overall patenting: when considering all other technologies (shown by the grey bar), young firms are responsible for around 12% of patents globally, but this share reaches 37% in geothermal technologies, 34% in marine energy, over 25% in waste and hydropower, and around 20% in biomass/biofuels and solar technologies.
Notes: LDES: long-duration energy storage. Data refer to IP5 patent families in renewable energy technologies, by earliest filing date. To be included in the sample, the patent family must be filed in at least two patent offices, one of which is among the IP5 offices (US Patent and Trademark Office, the European Patent Office, the Japan Patent Office, the Korean Intellectual Property Office, and the National Intellectual Property Administration in China) “Other technologies” refers to all technologies not related to renewable energy.
Sources: OECD calculations based on OECD, STI Micro-data Lab: Intellectual Property Database, OECD Start-ups Database and ORBIS©, version 2022.1, Bureau van Dijk, March 2025.
There are other important factors to consider when looking at the heterogeneity of firms involved in industrial ecosystems. One such element is their position – upstream, at the core or downstream – within the ecosystem. This is true not only in terms of inputs and outputs of production, but also in terms of innovative activities (Adner, 2006[31]). In this respect, the above-mentioned wealth of information included in patents can again play an important role, as patents include detailed data on both the prior-art cited by a patent (backward citations) and on other patents that would later mention the focal patent as part of their prior-art (forward citations).
Figure 6.9 portrays this analysis for the EII ecosystem. The central portion of the figure includes the different types of EII technologies while the left-hand side of the figure shows backward citations – from the EII ecosystem to its knowledge base – and the right-hand side shows forward citations – from other industries to the EII ecosystem. Focusing on the left-hand side, the figure shows that while some core components of the ecosystem, like “Chemicals”, mainly rely on inventions from within the sector, others, like “Wood”, do so to a much lesser extent. Furthermore, most EII inventions appear to benefit from patents filed in the “Chemicals” and “Rubber, plastics and minerals” sectors. Finally, there are sectors outside the core components of the ecosystem that play an important role as a knowledge base for the ecosystems: for example, “Machinery” and “Computers and electronics”.
The right-hand side of the figure shows that the EII core-sectors have positive spillovers on other industries: for instance, 14% of forward citations of “Non-metallic minerals” inventions are made by patents in “Computer and electronics”. Taking downstream adopters of innovation into account is essential for fostering industrial ecosystems, as research has shown that market demand serves as a “pull” for innovative activities (Mowery and Rosenberg, 1979[48]).
Notes: EII: energy-intensive industries. Data refer to patent applications filed under the Patent Cooperation Treaty, by earliest filing date and location of the applicants. Patents for inventions related to EIIs are delineated using the concordance developed by Goldschlag, Lybbert and Zolas (2020[41]) that maps codes of the co-operative patent classification (CPC) to the industry classification (ISIC, rev. 4), using a probabilistic approach. The technological scope is complemented with patents related to the decarbonisation of industries, including: selected climate change mitigation technologies in the production or processing of goods and patents for capture, storage, sequestration or disposal of greenhouse gases, identified using CPC codes. Cited technologies refer to the top 3 technologies listed in backward citations made in EII patents filed between 2018 and 2022. Citing technologies refer to the top 3 technologies listed in forward citations made to EII patents by patents filed between 2018 and 2022. Forward citation linkages exclude self-citations.
Sources: Dechezleprêtre et al. (2025, p. 39[32]), based on OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats (accessed in October 2024).
Finance providers, like banks, venture capital funds and angel investors, play a significant role, too, especially with respect to the above-mentioned funding challenges start-ups face. PROs – public research centres and other government institutions – and universities are other groups of stakeholders that should be considered when designing industrial policies targeting industrial ecosystems. Figure 6.10 focuses on their role within the renewable energy ecosystem. It shows that while innovations in this ecosystem mainly occur within private firms (with only 8% of patents originating from PROs and universities), the share of renewable energy patents owned by PROs and universities is relatively higher than that observed for all patents (6%). Additionally, there are specific technologies for which PROs and universities play a particularly important role, in particular long-duration energy storage (20%), marine energy (18%) and geothermal (17%). The technologies where PROs and universities tend to have a more relevant role are less mature technologies – like long-duration energy storage and marine – while in mature technologies like wind, industry takes a more outsized role in patenting activities, in line with existing evidence on academic innovation’s closeness to the technological frontier (Roche, Conti and Rothaermel, 2020[49]).
Notes: LDES: long-duration energy storage. Data refer to IP5 patent families. Patents filed by public research organisations and universities only include patents for which the type of applicant is identified. To be included in the sample, the patent family must be filed in at least two patent offices, one of which is among the IP5 offices (US Patent and Trademark Office, the European Patent Office, the Japan Patent Office, the Korean Intellectual Property Office, and the National Intellectual Property Administration in China) “Other technologies” refers to all technologies not related to renewable energy.
Source: Dechezleprêtre et al. (2024, p. 31[36]), based on OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats (accessed in January 2024).
Overall, the co-existence of these heterogenous stakeholders within the often-fuzzy boundaries of industrial ecosystems stresses the importance of designing policies capable not only of considering this multitude of actors, but also of fostering collaboration among them.
Delineating the boundaries of industrial ecosystems and identifying the ecosystems’ stakeholders are essential building blocks to identify possible challenges, bottlenecks and dependencies within ecosystems. Indeed, these often stem from co-ordination frictions among ecosystems’ actors (e.g. a lack of funding for start‑ups), “choke points” around specific technologies or resources (e.g. niobium for wind technologies in the renewable energy ecosystem), or critical but ignored relationships with upstream or downstream firms. This section examines three types of challenges that affect ecosystems: trade, technology and skills.
Supply chain dependencies and bottlenecks
Supply chain bottlenecks can be particularly detrimental for industrial ecosystems due to the interdependency that characterises them, as the failure of a single link (e.g. sudden export restrictions on a critical mineral or the shutdown of a crucial factory) may cascade down to the rest of the ecosystem. For example, the automotive ecosystem’s struggles during and following the COVID-19 pandemic can be linked to a failure of upstream suppliers to provide the chips needed for car manufacturing (Haramboure et al., 2023[50]). The resulting delays negatively affected not only core automotive sectors, but also other upstream sectors with high reliance on the automotive sector, as well as final customers. Monitoring and mitigating these supply chain risks is a key objective for policymakers interested in increasing the resilience of their economies (Bonnet and Ciani, 2023[51]).
The renewable energy ecosystem is also exposed to supply chain shocks. Figure 6.11 focuses on trade dependencies for capital goods employed in the production of renewable energy.
Notes: Three conditions need to be met to consider that country i is dependent for its supply of good k. First, gross imports of good k by country i need to be overall highly concentrated (Herfindahl-Hirschman Index >0.410). Second, the largest non-OECD partner needs to account for a high share (>10%) of country i’s imports of good k. Finally, other OECD Member countries’ import share for good k in country i needs to be less than 20%. Additionally, to avoid deeming as dependent a country with a high domestic production, two additional conditions were included. The ratio of exports to imports of the good k in country i must be lower than the 90th percentile of the distribution and the imports from country i of good k need to amount to USD 1 million or more. Conditions are verified over a three-year average, as shown in the figure specifically for the periods 2012-2014 and 2019-2021.
Source: Dechezleprêtre et al. (2024, p. 17[36]), based on UN Comtrade database.
Figure 6.11 shows that, among all products, the imports of metallic magnets and of turbines of a power exceeding 10 000 kilowatts are particularly exposed to potential shocks, as several countries are dependent on a limited number of exporters for a large share of their imports (of the 26 identified dependencies, 23 are linked to trade flows with China). This implies that, should this limited number of exporters restrict exports of metallic magnets and turbines, countries relying on them would likely face supply chain disruptions. The figure also reveals that trade dependencies have increased for most products across the two periods considered.
Identifying skills’ bottlenecks
The industrial ecosystem approach can also provide valuable insights into another critical challenge: skills bottlenecks. A shortage of relevant human capital, for instance, has been identified as a significant obstacle to the ambitious semiconductor-oriented initiatives recently launched by the European Union, Japan, Korea and the United States (see Box 6.2).
Online vacancy data can provide useful insights on the most required occupations within a given ecosystem. However, disentangling the occupations facing high demand but also high supply from those with demand-supply mismatches is not a trivial task. Figure 6.12 compares the occupational distribution of vacancies in EIIs (on the y-axis) and the distribution across all manufacturing vacancies (on the x-axis) for a sample of countries, with the denominator representing the number of online job postings in either manufacturing (x-axis) or energy-intensive industries (y-axis). While this does not allow the identification of precise mismatches, it pinpoints occupations that are especially salient to EIIs.
Notes: The figure features a scatterplot illustrating distributions. Each point represents a specific occupation in one of seven European countries analysed in 2022, namely Austria, Czechia, France, Germany, Italy, Spain and Switzerland, while the black line is a 45° line. Dots above this line correspond to over-demanded occupations in a country. Between 0.04% (Switzerland) and 4.0% (Spain and Czechia) of vacancies in Lightcast do not have any occupational information, with the specific percentage varying by country. Vacancies in the manufacturing sector were identified at the 1- and 2-digit level, which may include some vacancies in energy-intensive industries (EIIs) for which sectoral information was only available at the 1-digit level. Coloured dots represent the most over-demanded occupations in the EII ecosystem.
Source: Dechezleprêtre et al. (2025, p. 50[32]), based on Lightcast data.
Figure 6.12 shows that some occupations are considerably more requested in the EII ecosystem than manufacturing in general. While there are country-specific differences, over-demand is most evident for labourers in mining, construction, manufacturing and transport (3 percentage points higher on average than manufacturing), closely followed by stationary plant operators (2 percentage points), and science and engineering associate professionals (1 percentage point). The metal and machinery trades stand out as a notable exception, as they are in over-demand in Germany, France, Austria and Switzerland (in descending order) but under-demanded in Czechia, Italy and especially Spain.
The semiconductor ecosystem is incredibly complex due to the high degree of specialisation in the various stages – design; wafer production; fabrication; and assembly, testing and packaging (Haramboure et al., 2023[50]; OECD, 2024[52]) – of its value chain. Furthermore, the semiconductor ecosystem is characterised by the same challenges described in this chapter: significant bottlenecks at specific technology nodes (e.g. extreme ultraviolet lithography) and a strong dependency on critical raw materials.
Initiatives such as the CHIPS and Science Act (United States) and the EU Chips Act (European Union) have acknowledged not only the importance of the semiconductor industry as a provider of key inputs for downstream industries (Haramboure et al., 2023[50]) but also the challenges this industry has to face. Consequently, these policies have provided generous incentives to companies operating in the semiconductor ecosystem, especially for the construction of new fabrication plants. However, the success of these initiatives depends on firms’ ability to access the necessary human capital, which can be in short supply, as suggested by anecdotal evidence on the construction of new manufacturing facilities (New York Times, 2024[53]) and empirical data on supply gaps (McKinsey & Company, 2024[54]; 2023[55]). This issue is not limited to the semiconductor ecosystem: the battery manufacturer Northvolt, a cornerstone of the European strategy for battery-making, experienced significant disruptions in its operations due to a lack of expertise with foreign equipment (Tagliapietra and Trasi, 2024[56]). Filling similar human capital gaps requires a co-ordinated effort from both private and public stakeholders: for example, recent OECD work (OECD, 2024[57]) on the Philippines semiconductor industry highlighted the potential for stronger co-ordination among education stakeholders to provide more targeted training.
In addition to education, the availability of skills is also influenced by migration policies, including visa regulations and the phenomenon of brain drain. The Philippines serves again as an interesting case, as the country has experienced significant emigration, which may hinder its ambitions in the semiconductor ecosystem. To address this, the government launched the Balik Scientist Program to incentivise scientists to return to the Philippines and help strengthen the local semiconductor ecosystem (OECD, 2024[57]).
Sources: Haramboure et al. (2023[50]); New York Times (2024[53]); McKinsey & Company (2023[55]; 2024[54]); Tagliapietra and Trasi (2024[56]); OECD (2024[57]).
Technological interdependencies and adoption struggles
The industrial ecosystem approach, with its holistic perspective, is particularly well-suited to address the challenge of developing effective policies for fostering novel technologies, which rely on a complex web of collaboration between diverse actors. Start-ups are key engines of creative destruction and productivity-enhancing reallocation of market shares toward more innovative and productive firms (Decker et al., 2017[58]). Universities and PROs are also crucial in building a strong scientific knowledge base on which marketable innovations can be generated and can play a disproportionately significant role in the early stages of some emerging technologies.
Figure 6.13 highlights research links between start-ups and universities/PROs in the automotive ecosystem. It shows that, consistently across automotive-related technologies, young firms significantly cite more academic patents, a result which resonates with the literature emphasising the value of linkages between start-ups and academia (Thursby and Thursby, 2011[59]; Perkmann et al., 2013[60]), and hinting at the role of start-ups in further developing early-stage innovations originating in academia. This highlights the importance of creating opportunities for these two ecosystem actors to collaborate, for instance though technology transfer offices and university-affiliated incubators and accelerators.
Notes: Patent families are only assigned to one of the four technologies. If a patent family can be linked to multiple technologies, it is categorised based on the following order of priority: hydrogen, autonomous, electric and combustion. To be included in the sample, the patent family must be filed in at least two patent offices, one of which is among the IP5 offices (US Patent and Trademark Office, the European Patent Office, the Japan Patent Office, the Korean Intellectual Property Office, and the National Intellectual Property Administration in China). A patent is labelled as citing an academic patent if at least one application in the patent family cited a patent filed by an academic institution.
Source: Dechezleprêtre et al. (2023, p. 42[35]), based on data from OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats (accessed in June 2022) and ORBIS©.
Taking an ecosystem perspective can also help better understand what parts of a given ecosystem have made the most progress in dealing with the key technologies of the twin transition. For example, given its relatively low energy efficiency, the energy-intensive ecosystem would particularly benefit from more rapid adoption of digital and low-carbon technologies. Indeed, digital technologies can have a positive effect on EIIs by providing efficient tools for monitoring, controlling and automating production and processes (Calvino, Dechezleprêtre and Haerle, 2025[61]). Between 2018 and 2022, the EII ecosystem developed digital technologies at a higher pace than other non-ICT manufacturing (7.4% of EII-related technologies on average also featured a digital component versus 6% in the rest of the manufacturing sector) but there is considerable heterogeneity across the sectors composing the ecosystem, for example between “Non‑metallic minerals” technologies (16%) and “Coke and petroleum” (less than 1%), and countries (for instance, EU27 countries are lagging behind in this context).
The solutions to challenges faced by ecosystems are also interdependent
The various challenges discussed in this section are inherently linked and addressing one can have positive spillovers on the others. Figure 6.14 provides one example of this link. It shows patents filed to develop substitutes, and improve recycling, for raw materials employed in the renewable energy ecosystem. As such, these innovative efforts can play an important role in alleviating supply chain dependencies on specific products and improve supply chain resilience. The figure shows that most of these patents are focused on silicon, at least for the 2017-2021 period. Nickel and niobium are also targeted, but considerably less so than silicon. All three raw materials are characterised by low substitutability – hence the importance of finding alternatives to foster ecosystems’ resilience – but nickel and niobium also face high export concentration, making them even more delicate for supply chain monitoring.
Notes: Data refer to IP5 patent families by filing dates. Patents for recycling or substitute for raw materials are identified using the search strategy described in Annex A B.2 and Table A B.3 in Dechezleprêtre et al. (2024[36]).
Sources: Dechezleprêtre et al. (2024, p. 36[36]), based on OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats (accessed in January 2024).
Figure 6.15 presents another example, portraying the relationship between technological specialisation and trade comparative advantage, focusing on a subset of renewable energy technologies, namely wind, hydro and solar power. The figure shows a positive relationship between countries’ RTA, lagged by ten years, and relative comparative advantage, in particular in wind and hydro power, but also for solar power. This suggests that developing stronger innovation capabilities goes hand in hand with strengthening export performance, and while the directionality and causality of this relationship is not yet well-established in the literature, the fact that RTA values in Figure 6.15 precede relative comparative advantage values by ten years suggests that productivity gains associated to innovation (Cassiman, Golovko and Martínez-Ros, 2010[62]) and upgrading to higher value-added activities within technology-specific value chains (Caliari et al., 2023[63]) likely contribute to explaining this phenomenon.
Notes: This regression is pooling all technologies together and controlling for technology and country fixed effects. Also, this regression controls for relative comparative advantage in the base year (2012). A placebo test was performed by regressing revealed technological advantage in 2017-2021 on relative comparative advantage in 2012. This correlation was still positive but with a much lower coefficient (0.09 vs 0.9) and t-values (1.79 vs 4.17) than in the original specification.
Source: Dechezleprêtre et al. (2024, p. 30[36]), based on OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats and UN Comtrade database.
This section, focused on the policy implications of what has been described so far, shows there is room for improvement when it comes to adopting an ecosystem perspective in policymaking. While the chapter so far has highlighted the importance of non-core sectors, the interdependence among heterogenous actors and the multifaceted nature of industrial ecosystems, examples from the ecosystem studies show that existing policies typically overlook these elements.
Existing industrial policies are yet to consider ecosystem interdependencies
Despite significant conceptual advancements in industrial and innovation ecosystems and increasing awareness about their importance, the adoption of ecosystem thinking is still not the norm in industrial policymaking. One clear example comes from the energy-intensive sector, which is crucial for economies as its outputs are integrated in a variety of downstream industries. In addition, the EII ecosystem has relatively low R&D intensity and is only slowly increasing energy efficiency. Existing policies appear to provide only partial responses to these challenges: for example, policy support targeting the EII ecosystem, in the countries sampled in the QuIS project appear to be overwhelmingly focused on tax exemptions on energy costs (Figure 6.16).4F4 As unbalanced as it may appear, the picture provided in Figure 6.16 is likely to be an underestimation of this type of support, as some economies have horizontal tax reductions on energy costs that are not covered in the QuIS project, since they alter the baseline taxation of the entire economy, including both households and businesses.
As a percentage of GDP
Notes: Data are available for 11 OECD countries. Instruments targeted to the ecosystem are defined as those that either target a specific sector of the ecosystem or specific objectives related to the ecosystem such as subsidising energy efficiency or energy input costs in manufacturing.
Source: Dechezleprêtre et al. (2025, p. 55[32]), based on the OECD QuIS database.
Tax exemptions on energy costs, coupled with the large extensions granted to EIIs under the EU Emissions Trading System and other carbon-pricing mechanisms – with a share of free allowances over total emissions of 82% (vs. 21% in the rest of manufacturing), on average, in the period 2005-2020 – respond to crucial concerns linked to promoting the competitiveness of sectors like EIIs, that are located upstream in many value chains and are also highly exposed to international trade. Nevertheless, a more balanced policy portfolio could be beneficial, as there is established evidence pointing to free allowances and energy tax exemptions both being contradictory to abatement incentives (Dechezleprêtre, Nachtigall and Venmans, 2018[64]; Flues and van Dender, 2017[65]). In other words, the incentives currently provided may not provide a strong incentive to innovate in green technologies, even if their logic answers to ecosystem concerns related to downstream industries. While this may simply call for a more balanced approach, it also points to the challenging trade-offs that policymakers need to face.
A second example of policies that appear not to consider the entire ecosystem comes from the renewable energy ecosystem. In this case, support for R&D is increasing across countries, especially for green hydrogen and smart grids, particularly in Germany and Japan. However, historically, most of the support in OECD Member countries targeting this ecosystem has focused on deployment policies rather than direct R&D funding (Figure 6.17). This imbalance seems particularly pronounced in European economies, suggesting a potential oversight in fostering innovation within an ecosystem that has experienced a decline in patents across key technologies since 2010. In an ecosystem heavily dependent on critical raw materials and with capital goods providers concentrated in a few countries, prioritising support for deployment over support for innovation may reinforce bottlenecks rather than resolve them. This is especially the case given that countries like China – the main player in terms of capital goods and raw materials for the ecosystem – offer substantial direct subsidies to domestic companies (OECD, 2025[66]), thereby giving them an advantage over foreign competitors. Furthermore, Figure 6.7 already highlighted how the electricity sector, while being a relevant employer, accounts only for a limited share of renewable energy vacancies. More widely encompassing policies could benefit the numerous other sectors that account for renewable energy occupations.
Notes:. The public RDD value for the United States is based on the 2023 edition of the IEA Energy Technology RD&D Budgets database, as more recent versions no longer include the previously reported estimates. Funding for European countries includes EU-wide public RDD packages funded by the European Commission. These funds are allocated to each country based on its share in EU GDP in 2021. Funding is expressed in nominal terms (2021 prices) for deployment, and in constant 2023 prices and exchange rates for public RDD.
Sources: Dechezleprêtre et al. (2024, p. 56[36]), based on IEA Energy Technology RD&D Budgets database (accessed in December 2025); IISD (2024[67]).
Policies designed to support the automotive ecosystem appear to have moved toward a stronger integration of ecosystem thinking, which is crucial in a sector strongly affected by the twin transition toward green and digital technologies. In this area, the inclusion of support for upstream segments, such as the semiconductor sector, appears crucial to increase resilience, as recent shocks linked to the COVID-19 pandemic have shown. Moving in the direction of “CASE” (Connected, Automated, Shared and Electric) vehicles requires policies that support providers of digital technologies – an area where ICT companies play a key role.
Another upstream sector that has received direct government support is the battery industry: the establishment of the European Battery Alliance and its related Important Projects of Common European Interest funding suggest that ecosystem thinking is becoming increasingly embedded in policymaking. Russia’s war of aggression against Ukraine has put pressure on energy prices, further highlighting the importance of energy independence based on domestic renewable electricity production. From this perspective, policy initiatives like the Inflation Reduction Act, which include tax credits linked to manufacturing and investment to support electric vehicle and renewable electricity producers, can contribute to the shift toward electric transportation.
How can the industrial ecosystem approach support the identification of relevant policies?
The key policy lesson offered by the three ecosystem studies is that policies aiming to foster a given ecosystem should not target its core individual components but rather the variety of actors that compose them and the relationships characterising them. In addition, such policies should consider whether support needs to be tailored to different actors, including academia, investors, and small and medium-sized enterprises and start-ups, so that all stakeholders – regardless of size – can contribute to and benefit from the growth of the ecosystem. Policies that focus solely on core components can be short-sighted, as they may overlook factors such as upstream-driven innovation, demand in downstream markets, supply chain bottlenecks and other critical ecosystem dynamics. Similarly, “mission-oriented” policies emphasise the need for co-ordination across different policies (and across the actors overseeing them) to address overarching “grand” challenges like those posed, for example, by climate change (Mazzucato, 2018[68]; Larrue, 2021[43]; OECD, 2023[69]). This is a challenging task that requires capable public agencies, effective data infrastructure, alignment with national priorities and broad public consensus. Innovation, skills and competition-oriented policies should all be considered, as they all contribute towards creating more vibrant, resilient ecosystems.
Policies promoting innovation are crucial across the studied ecosystems as they contribute to building the capabilities needed for future success. There are various instruments available to policymakers seeking to support innovation, including R&D tax credits, grants for R&D expenditures, support to university-based research, etc. The literature (Bloom, Van Reenen and Williams, 2019[70]) so far has highlighted that these instruments, particularly tax credits and to a less clear extent R&D subsidies (Criscuolo et al., 2022[1]), are effective at increasing R&D spending and generating innovations (Appelt et al., 2016[71]; Dechezleprêtre et al., 2023[4]; OECD, 2023[72]; 2023[73]). Grounding the design of these policies in an industrial ecosystem approach can be beneficial; for example, the patent analysis discussed in this chapter identifies sectors and technologies upon which core parts of an ecosystem are dependent, and where support to investment in R&D might be necessary. For example, R&D support to automotive firms might be less effective if crucial innovations are happening outside the core portion of the ecosystem, in sectors like computers and electronics. Policymakers can rebalance R&D incentives to also target specific upstream sectors and technologies (in this example, digital ones) and exploit complementary instruments to facilitate the market-entry of start-ups active in these technologies. Figure 6.5 provided strong evidence on this point, as it clearly showed that a considerable portion of innovations related to EII technologies does not originate from the core components of the ecosystem. Another area where an ecosystem approach can benefit policymakers is in addressing supply chain bottlenecks: the identification of hard-to-substitute raw materials (as discussed above for the renewable energy ecosystem) is a precondition to design specific R&D incentives targeting substitute materials or recycling techniques.
Notably, strong complementarities exist across policies. For example, providing tax credits to R&D may not lead to innovative outcomes when required skills are lacking, as their absence tends to hinder the adoption of innovative technologies (Calvino, Criscuolo and Verlhac, 2020[74]). On the contrary, it may lead to inflationary pressures on R&D salaries (Criscuolo et al., 2022[1]) with negative consequences on other business functions. However, support to human capital is one of the most challenging areas for policymakers, as it can take years to bear fruit and typically relies on the collaboration between academia, technical and vocational education and training providers, and private firms. Adopting an industrial ecosystem approach can be beneficial for policymakers also in this context: in this respect, Figure 6.7 showed how job vacancies related to renewable energy technologies originate from a variety of disparate sectors, including but not limited to manufacturing and electricity, but also professional and scientific activities and construction, finance and insurance. A solely sectoral approach would, therefore, miss a critical mass of jobs related to the renewable energy ecosystem. Furthermore, it can highlight occupations that are demanded in both the core and non-core sectors of an ecosystem, rather than just in specific segments. This is helpful in mitigating potential “poaching” concerns which may arise if a policy lowers labour costs in one segment, leading to shortages elsewhere in the ecosystem, resulting in no net benefit overall. For example, in the EII ecosystem, digital skills are currently highly required across industries: investing in training for these skills specifically could lead to double dividends. To avoid talent bottlenecks, industrial ecosystem strategies should incorporate upskilling, reskilling and talent retention initiatives, ensuring that human capital development keeps pace with technological and sustainability ambitions.
Other areas for complementarity regard venture funding and competition policies. The previous sections have highlighted that start-ups are crucial economic engines, generating and diffusing innovations (especially for emerging technologies), with a positive effect on employment and on productivity via reallocation effects (Decker et al., 2017[58]; Calvino, Criscuolo and Menon, 2018[75]). Start-ups face tougher conditions than more established firms due to their lack of resources and financial track-record in the market. Obtaining access to venture funding is then crucial for them (Hall and Lerner, 2010[76]). However, scholarly reviews of public interventions in this area have found mixed results, with some studies pointing to crowding out issues and underperformance of government-backed venture capital versus private ones (Howell, 2024[77]), other studies finding comparable performance between the two (Berger, Dechezleprêtre and Fadic, 2024[78]) and yet others finding that a mix of private and government venture capital outperforms other modalities (Brander, Du and Hellmann, 2015[79]).
Similarly, competition policy has an important role to play in ensuring that established incumbents do not stifle innovation via aggressive acquisition strategies (Cunningham, Ederer and Ma, 2018[80]), especially in an economic context characterised by declining business dynamism (Calvino, Criscuolo and Verlhac, 2020[74]) in light of new evidence regarding the declining innovation rates of acquired companies (Berger et al., 2025[81]). Notably, competition is commonly regarded as having an inverted-U shape effect on innovation (Aghion et al., 2005[82]): positive up to a certain threshold (Levine et al., 2020[83]), negative afterwards (Kang, 2019[84]), as excessive competition may decrease firms’ resources available for investment. Grasping the role that various actors play in an ecosystem can help highlight where interventions might be needed: for example, the decreasing contribution to innovation by small and medium-sized enterprises in the renewable energy ecosystem might be linked to the increasing M&A activity experienced in this sector.
Overall, the complexity and interdependencies that exist among these policies – and potentially many others in areas such as public procurement, programmes for entrepreneurship, labour market policies, etc. – point to the need for co-ordination among policymakers in charge of different policies, which, even when centrally designed, are often administered and monitored by different ministries and agencies (OECD, 2024[42]) – as well as ecosystem stakeholders, grounded in robust empirical evidence.
Industrial ecosystems can be a crucial unit of analysis for policymakers seeking to design industrial policies fostering the growth and resilience of their economies. The three case studies used in this chapter (automotive, renewable energy and energy-intensive industries) have described concrete approaches to delineating industrial ecosystems (both their boundaries and stakeholders) and highlighted some of the key challenges affecting them.
Key messages include: first, sectoral definitions are relevant, but increasing the granularity of the analysis allows accounting for key components (e.g. upstream suppliers and downstream customers) and relationships (e.g. innovation linkages) that would be omitted in a standard sectoral approach. Second, ecosystems include a variety of heterogenous actors: their mapping, and the identification of the specific roles they play, and their respective importance, rests upon the use of a wide array of data (e.g. value added, trade, patent, online vacancies, etc.). Third, many of the challenges described, be it in the trade, human capital or innovation, originate in a well-defined part of the ecosystem but tend to spillover to other areas. Fourth, acting upon one challenge can be beneficial for others, too. For example, innovations generated upstream in the value chain will cascade to downstream sectors, with productivity gains going beyond sectoral boundaries, and positively affecting production and trade as well.
The ecosystem approach has room for improvement. For example, many analyses still rely on value-added data that are currently only available at the 2-digit sectorial level. Analysing ecosystems based on more granular data could provide deeper insights for policymaking. Second, the network of customers and suppliers could be extended to more distant relationships to identify hidden bottlenecks and strengthen an ecosystem’s resilience. Third, the analyses presented in this chapter focus on firms, industrial sectors and countries. However, the literature in the ecosystem field has highlighted the role of geographically bounded clusters and districts, which is not accounted for in this work.
A promising area for further ecosystem research is the exploitation of firm-level value-added tax data (Criscuolo et al., 2024[34]) to describe firm-to-firm interactions. More granular ecosystems can also be analysed. One example is the ongoing work on the semiconductor ecosystem, based on a taxonomy distinguishing different types of chips and production facilities (OECD, 2024[52]). As these approaches become more widespread, and the level of precision and granularity increases, policymakers will be increasingly able to make targeted interventions based on clear-cut evidence and capable of not only targeting specific pain-points, but also confidently addressing the interdependency existing within ecosystems.
References
[39] Acs, Z. and D. Audretsch (1989), “Patents as a measure of innovative activity”, KYKLOS, Vol. 42/2, pp. 171-180, https://doi.org/10.1111/j.1467-6435.1989.tb00186.x.
[31] Adner, R. (2006), “Match your innovation strategy to your innovation ecosystem”, Harvard Business Review, Vol. 84/4, pp. 98-107; 148.
[82] Aghion, P. et al. (2005), “Competition and innovation: An inverted-U relationship”, The Quarterly Journal of Economics, Vol. 120/2, pp. 701-728, https://doi.org/10.1093/qje/120.2.701.
[30] Andreoni, A. (2018), “The architecture and dynamics of industrial ecosystems: Diversification and innovative industrial renewal in Emilia Romagna”, Cambridge Journal of Economics, Vol. 42/6, pp. 1613-1642, https://doi.org/10.1093/cje/bey037.
[71] Appelt, S. et al. (2016), “R&D tax incentives: Evidence on design, incidence and impacts”, OECD Science, Technology and Industry Policy Papers, No. 32, OECD Publishing, Paris, https://doi.org/10.1787/5jlr8fldqk7j-en.
[45] Audretsch, D. et al. (2020), “Innovative start-ups and policy initiatives”, Research Policy, Vol. 49/10, p. 104027, https://doi.org/10.1016/j.respol.2020.104027.
[81] Berger, M. et al. (2025), “Acquisitions and their effect on start-up innovation: Stifling or scaling?”, OECD Science, Technology and Industry Working Papers, No. 2025/10, OECD Publishing, Paris, https://doi.org/10.1787/b4efd3ab-en.
[78] 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.
[6] Berlingieri, G., P. Blanchenay and C. Criscuolo (2017), “The great divergence(s)”, OECD Science, Technology and Industry Policy Papers, No. 39, OECD Publishing, Paris, https://doi.org/10.1787/953f3853-en.
[7] Berlingieri, G. et al. (2020), “Laggard firms, technology diffusion and its structural and policy determinants”, OECD Science, Technology and Industry Policy Papers, No. 86, OECD Publishing, Paris, https://doi.org/10.1787/281bd7a9-en.
[8] Bloom, N. et al. (2020), “Are ideas getting harder to find?”, American Economic Review, Vol. 110/4, pp. 1104-1144, https://doi.org/10.1257/aer.20180338.
[70] 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.
[51] Bonnet, P. and A. Ciani (2023), Applying the SCAN methodology to the Semiconductor Supply Chain, European Commission, https://publications.jrc.ec.europa.eu/repository/handle/JRC133736 (accessed on 18 February 2025).
[79] Brander, J., Q. Du and T. Hellmann (2015), “The effects of government-sponsored venture capital: International evidence”, Review of Finance, Vol. 19/2, pp. 571-618, https://doi.org/10.1093/rof/rfu009.
[63] Caliari, T. et al. (2023), “Global value chains and sectoral innovation systems: An analysis of the aerospace industry”, Structural Change and Economic Dynamics, Vol. 65, pp. 36-48, https://doi.org/10.1016/j.strueco.2023.02.004.
[75] Calvino, F., C. Criscuolo and C. Menon (2018), “A cross-country analysis of start-up employment dynamics”, Industrial and Corporate Change, Vol. 27/4, pp. 677-698, https://doi.org/10.1093/icc/dty006.
[74] Calvino, F., C. Criscuolo and R. Verlhac (2020), “Declining business dynamism: Structural and policy determinants”, OECD Science, Technology and Industry Policy Papers, No. 94, OECD Publishing, Paris, https://doi.org/10.1787/77b92072-en.
[61] Calvino, F., A. Dechezleprêtre and D. Haerle (2025), “How can the digital transformation affect the net-zero transition? A conceptual framework for the “twin” transition”, OECD Science, Technology and Industry Working Papers, No. 2025/17, OECD Publishing, Paris, https://doi.org/10.1787/4ee71f6d-en.
[62] Cassiman, B., E. Golovko and E. Martínez-Ros (2010), “Innovation, exports and productivity”, International Journal of Industrial Organization, Vol. 28/4, pp. 372-376, https://doi.org/10.1016/j.ijindorg.2010.03.005.
[14] Cooke, P. (2004), Regional Innovation Systems: An Evolutionary Approach, Routledge, London.
[34] Criscuolo, C. et al. (2024), “Estonia’s firm-level production network: Lessons for industrial policy”, OECD Science, Technology and Industry Working Papers, No. 2024/13, OECD Publishing, Paris, https://doi.org/10.1787/e0f18b9c-en.
[44] Criscuolo, C., P. Gal and C. Menon (2014), “The dynamics of employment growth: New evidence from 18 countries”, OECD Science, Technology and Industry Policy Papers, No. 14, OECD Publishing, Paris, https://doi.org/10.1787/5jz417hj6hg6-en.
[5] Criscuolo, C. et al. (2022), “An industrial policy framework for OECD countries: Old debates, new perspectives”, OECD Science, Technology and Industry Policy Papers, No. 127, OECD Publishing, Paris, https://doi.org/10.1787/0002217c-en.
[1] Criscuolo, C. et al. (2022), “Are industrial policy instruments effective?: A review of the evidence in OECD countries”, OECD Science, Technology and Industry Policy Papers, No. 128, OECD Publishing, Paris, https://doi.org/10.1787/57b3dae2-en.
[3] Criscuolo, C. et al. (2019), “Some causal effects of an industrial policy”, American Economic Review, Vol. 109/1, pp. 48-85, https://doi.org/10.1257/aer.20160034.
[80] Cunningham, C., F. Ederer and S. Ma (2018), “Killer acquisitions”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.3241707.
[47] Da Rin, M., T. Hellmann and M. Puri (2011), “A Survey of Venture Capital Research”, https://doi.org/10.2139/ssrn.1942821.
[33] De Lyon, J. and A. Dechezleprêtre (Forthcoming), Firm Heterogeneity in Carbon Productivity Evidence from Representative Cross-Country Micro Data.
[32] Dechezleprêtre, A. et al. (2025), “A comprehensive overview of the Energy Intensive Industries ecosystem”, OECD Science, Technology and Industry Working Papers, No. 2025/09, OECD Publishing, Paris, https://doi.org/10.1787/bbf7d4c2-en.
[36] Dechezleprêtre, A. et al. (2024), “A comprehensive overview of the renewable energy industrial ecosystem”, OECD Science, Technology and Industry Working Papers, No. 2024/11, OECD Publishing, Paris, https://doi.org/10.1787/94dce592-en.
[35] Dechezleprêtre, A. et al. (2023), “How the green and digital transitions are reshaping the automotive ecosystem”, OECD Science, Technology and Industry Policy Papers, No. 144, OECD Publishing, Paris, https://doi.org/10.1787/f1874cab-en.
[4] 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.
[64] Dechezleprêtre, A., D. Nachtigall and F. Venmans (2018), “The joint impact of the European Union Emissions Trading System on carbon emissions and economic performance”, OECD Economics Department Working Papers, No. 1515, OECD Publishing, Paris, https://doi.org/10.1787/4819b016-en.
[58] Decker, R. et al. (2017), “Declining dynamism, allocative efficiency, and the productivity slowdown”, American Economic Review, Vol. 107/5, pp. 322-326, https://doi.org/10.1257/aer.p20171020.
[29] Draghi, M. (2024), The Future of European Competitiveness: Part B – In-depth Analysis and Recommendations, Publications Office of the European Union, Luxembourg, https://commission.europa.eu/document/download/ec1409c1-d4b4-4882-8bdd-3519f86bbb92_en?filename=The%20future%20of%20European%20competitiveness_%20In-depth%20analysis%20and%20recommendations_0.pdf.
[38] Dziallas, M. and K. Blind (2019), “Innovation indicators throughout the innovation process: An extensive literature analysis”, Technovation, Vol. 80-81, pp. 3-29, https://doi.org/10.1016/j.technovation.2018.05.005.
[9] European Commission (2020), A New Industrial Strategy for Europe, European Commission, Brussels, https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:52020DC0102&from=EN (accessed on 11 February 2025).
[65] Flues, F. and K. van Dender (2017), “The impact of energy taxes on the affordability of domestic energy”, OECD Taxation Working Papers, No. 30, OECD Publishing, Paris, https://doi.org/10.1787/08705547-en.
[13] Freeman, C. (1995), “The “National System of Innovation” in historical perspective”, Cambridge Journal of Economics, Vol. 19/1, pp. 5-24, https://doi.org/10.1093/oxfordjournals.cje.a035309.
[41] Goldschlag, N., T. Lybbert and N. Zolas (2020), “Tracking the technological composition of industries with algorithmic patent concordances”, Economics of Innovation and New Technology, Vol. 29/6, pp. 582-602, https://doi.org/10.1080/10438599.2019.1648014.
[76] Hall, B. and J. Lerner (2010), “The financing of R&D and innovation”, in Handbook of the Economics of Innovation, pp. 609-639, https://doi.org/10.1016/S0169-7218(10)01014-2.
[50] Haramboure, A. et al. (2023), “Vulnerabilities in the semiconductor supply chain”, OECD Science, Technology and Industry Working Papers, No. 2023/05, OECD Publishing, Paris, https://doi.org/10.1787/6bed616f-en.
[77] Howell, S. (2024), “Government intervention in innovation”, Annual Review of Financial Economics, Vol. 16/1, pp. 367-390, https://doi.org/10.1146/annurev-financial-082123-105722.
[22] Iansiti, M. and R. Levien (2004), The Keystone Advantage: What the New Dynamics of Business Ecosystems Mean for Strategy, Innovation, and Sustainability, Harvard Business School Press, Boston.
[67] IISD (2024), Public Financial Support for Renewable Power Generation and Integration in the G20 Countries, International Institute for Sustainable Development, Winnipeg, Manitoba, Canada, https://www.iisd.org/system/files/2024-09/renewable-energy-support-g20.pdf.
[21] Jacobides, M., C. Cennamo and A. Gawer (2018), “Towards a theory of ecosystems”, Strategic Management Journal, Vol. 39/8, pp. 2255-2276, https://doi.org/10.1002/smj.2904.
[25] Juhász, R., N. Lane and D. Rodrik (2024), “The new economics of industrial policy”, Annual Review of Economics, Vol. 16/1, pp. 213-242, https://doi.org/10.1146/annurev-economics-081023-024638.
[23] Juhász, R. and C. Steinwender (2023), Industrial Policy and the Great Divergence, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w31736.
[84] Kang, H. (2019), “How does competition affect innovation? Evidence from U.S. antitrust cases”, Academy of Management Proceedings, Vol. 2019/1, p. 13918, https://doi.org/10.5465/AMBPP.2019.13918abstract.
[2] Lane, N. (2020), “The new empirics of industrial policy”, Journal of Industry, Competition and Trade, Vol. 20/2, pp. 209-234, https://doi.org/10.1007/s10842-019-00323-2.
[43] Larrue, P. (2021), “The design and implementation of mission-oriented innovation policies: A new systemic policy approach to address societal challenges”, OECD Science, Technology and Industry Policy Papers, No. 100, OECD Publishing, Paris, https://doi.org/10.1787/3f6c76a4-en.
[83] Levine, R. et al. (2020), Competition Laws and Corporate Innovation, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w27253.
[46] Lounsbury, M. and M. Glynn (2001), “Cultural entrepreneurship: Stories, legitimacy, and the acquisition of resources”, Strategic Management Journal, Vol. 22/6-7, pp. 545-564, https://doi.org/10.1002/smj.188.
[12] Lundvall, B. (1992), National Systems of Innovation: Toward a Theory of Innovation and Interactive Learning, Pinter Publishers, London.
[16] Malerba, F. (2002), “Sectoral systems of innovation and production”, Research Policy, Vol. 31/2, pp. 247-264, https://doi.org/10.1016/S0048-7333(01)00139-1.
[26] Mazzucato, M. (2021), Mission Economy: A Moonshot Guide to Changing Capitalism, The Penguin Press.
[68] Mazzucato, M. (2018), “Mission-oriented innovation policies: Challenges and opportunities”, Industrial and Corporate Change, Vol. 27/5, pp. 803-815, https://doi.org/10.1093/icc/dty034.
[54] McKinsey & Company (2024), “New tactics for new talent: Closing US semiconductor labor gaps”, https://www.mckinsey.com/industries/semiconductors/our-insights/new-tactics-for-new-talent-closing-us-semiconductor-labor-gaps?stcr=0C3963C0B794438ABAB92F3D7325622E&cid=other-eml-alt-mip-mck&hlkid=3cae96605f12454ba41540279f892979&hctky=11184079&hdpid=9a82e441-52c8-401b-9e84-6a48299ac0df (accessed on 20 June 2024).
[55] McKinsey & Company (2023), “Reimagining labor to close the expanding US semiconductor talent gap”, https://www.mckinsey.com/industries/semiconductors/our-insights/reimagining-labor-to-close-the-expanding-us-semiconductor-talent-gap.
[28] Ministry of Enterprises and Made in Italy (2024), Made in Italy 2030: Libro Verde Sulla Politica Industriale, Ministry of Enterprises and Made in Italy, Rome, https://www.mimit.gov.it/images/stories/documenti/allegati/Libro_verde_finale_2_2.pdf.
[10] Moore, J. (1993), “Predators and prey: A new ecology of competition”, Harvard Business Review, Vol. 71/3, pp. 75-86.
[48] Mowery, D. and N. Rosenberg (1979), “The influence of market demand upon innovation: A critical review of some recent empirical studies”, Research Policy, Vol. 8/2, pp. 102-153, https://doi.org/10.1016/0048-7333(79)90019-2.
[11] Nelson, R. (1993), National Innovation Systems, Oxford University Press, New York/Oxford.
[53] New York Times (2024), “What works in Taiwan doesn’t always in Arizona, a chipmaking giant learns”, New York Times, https://www.nytimes.com/2024/08/08/business/tsmc-phoenix-arizona-semiconductor.html (accessed on 21 February 2025).
[66] OECD (2025), “Government support in the solar and wind value chains”, OECD Trade Policy Papers, No. 288, OECD Publishing, Paris, https://doi.org/10.1787/d82881fd-en.
[52] OECD (2024), Chips, Nodes and Wafers: A Taxonomy for Semiconductor Data Collection, OECD Publishing, Paris, https://doi.org/10.1787/f68de895-en.
[42] OECD (2024), Mission-Oriented Innovation Policies for Net Zero: How Can Countries Implement Missions to Achieve Climate Targets?, OECD Publishing, Paris, https://doi.org/10.1787/5efdbc5c-en.
[57] OECD (2024), Promoting the Growth of the Semiconductor Ecosystem in the Philippines, OECD Publishing, Paris, https://doi.org/10.1787/01497fea-en.
[73] OECD (2023), “Income-based tax relief for R&D and innovation: An integrated view”, OECD Science, Technology and Industry Policy Papers, No. 161, OECD Publishing, Paris, https://doi.org/10.1787/2db44e8b-en.
[69] OECD (2023), OECD Science, Technology and Innovation Outlook 2023: Enabling Transitions in Times of Disruption, OECD Publishing, Paris, https://doi.org/10.1787/0b55736e-en.
[72] 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.
[37] OECD (2009), OECD Patent Statistics Manual, OECD Publishing, Paris, https://doi.org/10.1787/9789264056442-en.
[20] OECD (2006), Innovation in Pharmaceutical Biotechnology: Comparing National Innovation Systems at the Sectoral Level, OECD Publishing, Paris, https://doi.org/10.1787/9789264014046-en.
[19] OECD (1999), Managing National Innovation Systems, OECD Publishing, Paris, https://doi.org/10.1787/9789264189416-en.
[60] Perkmann, M. et al. (2013), “Academic engagement and commercialisation: A review of the literature on university-industry relations”, Research Policy, Vol. 42/2, pp. 423-442, https://doi.org/10.1016/j.respol.2012.09.007.
[15] Porter, M. (1998), “Clusters and the new economics of competition”, Harvard Business Review, Vol. 76/6, pp. 77-90, https://hbr.org/1998/11/clusters-and-the-new-economics-of-competition.
[49] Roche, M., A. Conti and F. Rothaermel (2020), “Different founders, different venture outcomes: A comparative analysis of academic and non-academic startups”, Research Policy, Vol. 49/10, p. 104062, https://doi.org/10.1016/j.respol.2020.104062.
[17] Stam, E. (2015), “Entrepreneurial ecosystems and regional policy: A sympathetic critique”, European Planning Studies, Vol. 23/9, pp. 1759-1769, https://doi.org/10.1080/09654313.2015.1061484.
[18] Stam, E. and A. van de Ven (2021), “Entrepreneurial ecosystem elements”, Small Business Economics, Vol. 56/2, pp. 809-832, https://doi.org/10.1007/s11187-019-00270-6.
[40] Supriya, L. (2023), “How patent information can help spot technology gaps and catalyze economic development”, WIPO Magazine, https://www.wipo.int/en/web/wipo-magazine/articles/how-patent-information-can-help-spot-technology-gaps-and-catalyze-economic-development-56565 (accessed on 6 March 2024).
[56] Tagliapietra, S. and C. Trasi (2024), “Northvolt’s struggles: A cautionary tale for the EU Clean Industrial Deal”, Bruegel’s, https://www.bruegel.org/analysis/northvolts-struggles-cautionary-tale-eu-clean-industrial-deal (accessed on 26 February 2025).
[59] Thursby, J. and M. Thursby (2011), “University-industry linkages in nanotechnology and biotechnology: Evidence on collaborative patterns for new methods of inventing”, The Journal of Technology Transfer, Vol. 36/6, pp. 605-623, https://doi.org/10.1007/s10961-011-9213-z.
[27] UK Government (2025), The UK’s Modern Industrial Strategy, UK Government, https://assets.publishing.service.gov.uk/media/68595e56db8e139f95652dc6/industrial_strategy_policy_paper.pdf.
[24] Warwick, K. (2013), “Beyond industrial policy: Emerging issues and new trends”, OECD Science, Technology and Industry Policy Papers, No. 2, OECD Publishing, Paris, https://doi.org/10.1787/5k4869clw0xp-en.
← 1. TiVA data describe the value added generated by each sector-country in the production of goods and services that are consumed worldwide.
← 2. More specifically, patents protecting technologies related to EIIs are identified through the concordance between technology classes assigned to patent documents, specifically the Cooperative Patent Classification (CPC) codes, and industries, as developed by Goldschlag, Lybbert and Zolas (2020[41]). The technological scope includes also patents related to the decarbonisation of industries, identified using CPC codes. See Annex B of Dechezleprêtre et al. (2025[32]) for further details on the coverage and methodologies used to build patent-based indicators.
← 3. The insights presented here are based on data from Lightcast, an employment analytics and labour market information firm that collects data from firms’ websites and online job boards.
← 4. Data from QuIS include 11 countries as to the 2022 vintage: Canada, Denmark, France, Germany, Ireland, Israel, Italy, the Netherlands, Slovenia, Sweden and the United Kingdom.