2 Innovation capabilities as a guide for successful policy design

Introduction

Recent years have seen a resurgence of industrial policies worldwide. These policies have mostly been driven not by new insights into their efficacy but by governments responding to challenges such as climate change, supply chain disruptions and national security concerns. In part, recent industrial policies reveal governments’ expectations about which industrial activities are most likely to offer long-term benefits to the economy.

By employing a range of industrial policy instruments governments are also making (explicitly or implicitly) a wide range of scientific and technological choices. These choices shape the economic incentives for a stakeholder – whether individual or institutional – to facilitate the generation, acquisition and diffusion of new scientific, technological and production knowledge. As a result, industrial policies influence the innovation path taken by a region or country by choosing where to allocate human and financial resources through a range of public policy instruments.

As discussed in Chapter 1, economic thinking would endorse governments investing in those activities, individuals and institutions that facilitate the generation, acquisition and diffusion of new scientific, technological and production knowledge. Consequently, successful industrial policies should aim to develop new capabilities, nurture nascent ones and maintain any existing advantages over other countries.

But which are the correct scientific or technologically related capabilities to target with industrial policies? For instance, which fields of scientific research should government funding prioritize? Which promising embryonic technology should get government funding to achieve commercial viability? Answering such questions is not straightforward. It requires the conviction that supporting a nascent local industry today will generate critical input for other local industries at competitive prices in the future, or that it will generate substantial spillover benefits to the local economy.

Market mechanisms often provide signals that are too incomplete to inform such decisions. Evaluating the benefits and costs of such interventions is crucial in the evolving landscape of industrial policy. This is because the innovation that fuels progress, economic growth and competitiveness is a multidimensional force embodying various facets of human endeavor across nations, regions and industries. Among the many relevant dimensions of innovation are the people and institutions related to the production of science, technology and products.

The empirical approach undertaken in this chapter focuses on these three key dimensions of innovation: science, technology and products. At the macro level, advanced national economies typically perform in all three of these dimensions. Yet advanced economies may greatly differ in terms of the specialization, intensity and combination of and the subcategories within these three dimensions. Some economies excel in scientific research but struggle to translate scientific outcomes into technological advances, leading to an untapped potential. Others might exhibit exceptional ingenuity in one technological field, yet face challenges in transforming these advances into commercially viable products.

As will be discussed further in the next section, most economies show some production capabilities but only a far smaller group of economies are able to show scientific capabilities, and a still smaller group to show technological ones. Why is it that certain countries manage to create viable, competitive products with a role in global value chains but find it difficult to produce scientific discoveries and technological breakthroughs of their own?

This chapter focuses on innovation capabilities measured by the scientific, technological and production know-how – tacit or codifiable – existing in each country or region. Assessing capabilities in these three dimensions is crucial for evidence-based policymaking but is not straightforward. The chapter provides a novel empirical analysis of the current set of innovation capabilities in economies for international comparison. This relies on a body of economic literature focused on economic and technological relatedness and complexity that is applied to data on scientific publications, patent applications and international trade.⁽¹⁾For a summary of this literature see Box 2.2, and Balland, P.-A., T. Broekel, D. Diodato, E. Giuliani, R. Hausmann, N. O’Clery and D. Rigby (2022). The new paradigm of economic complexity. Research Policy, 51(3), 104450. DOI: https://doi.org/10.1016/j.respol.2021.104450.

The first section of this chapter defines innovation capabilities and discusses how they can be measured using data on scientific publications, international patents and international trade. The second section introduces the innovation complexity concept by exploring the qualitative differences between innovation capabilities, particularly with regards to diffusion. It also discusses how complexity can be a key factor in explaining economic growth. The third section introduces the concept of relatedness in order to shed light on how current capabilities can be leveraged to develop new ones. The last section concludes the chapter with takeaways, remarks and general policy implications.

Defining innovation capabilities

Innovation capabilities represent, in essence, the ability of a country to deliver competitive outputs in a certain field of the innovation process. In many cases, these outputs include the skills and knowledge embedded in tools, procedures or computer codes that can be easily shared or shipped around the world. However, often they are tacit, meaning that they are embedded in individuals but are not readily codifiable and hence not easily transferrable.⁽²⁾For a seminal discussion on tacit knowledge, see Polanyi, M. 1966. The Tacit Dimension. Chicago: University of Chicago Press, 4. The fact that they are not easily transferrable makes their understanding and measurement crucial for innovation policymaking.

This section focusses on innovation capabilities in terms of the scientific, technological and production know-how – tacit or codifiable – that each country has.⁽³⁾This chapter's definition and measurement of innovation capabilities follows closely the work of Pugliese, E., G. Cimini, A. Patelli, A. Zaccaria, L. Pietronero and A. Gabrielli (2019). Unfolding the innovation system for the development of countries: Coevolution of Science, Technology and Production. Scientific Reports, 9, 16440. DOI: https://doi.org/10.1038/s41598-019-52767-5.

The scientific dimension: Scientific capabilities include the research, discovery and generation of knowledge. This is achieved through a culture of exploration and experimentation. Scientists and researchers push the boundaries of the world’s scientific knowledge by discovering, perfecting and combining existing capabilities in each scientific field.

The technological dimension: This includes all the methods that transform existing scientific and technical knowledge into concrete processes and products. Engineers, applied scientists, developers and designers collaborate to bridge the gap between theory and practice. They translate what are usually abstract concepts into functional technological capabilities. These capabilities take the form of technical procedures and tangible tools that can be applied to different technological fields. For instance, engineers and developers relied heavily on the scientific fields of quantum mechanics and materials chemistry to develop semiconductor devices, lasers and optical technologies. Today, products and processes rely on these latter technologies to innovate and researchers use them to push further the frontiers of science.

The production dimension: This contains the whole spectrum of production capabilities; namely, all those capabilities needed to produce all the goods and services commercialized in an economy. To optimize the production of a given industry outcome, entrepreneurs combine production capabilities by hiring labor with specific skills, acquiring technologically advanced equipment and incorporating more sophisticated inputs. With varying degrees of sophistication, this happens at the scale of large corporations, as well as at the scale of small companies and start-ups. Ultimately, innovation is realized through efficient production methods, supply chain optimization and customer-centric offerings. Production stakeholders and their capabilities play a crucial role in ensuring that innovations reach end-users and drive economic value.

The innovation path

To a great extent, the path to successful and innovative products can be traced all the way back to some technological and scientific capabilities. Many of the most advanced innovations have originated from basic exploratory science. Scientific breakthroughs can open the door to ground-breaking capabilities, giving birth to new technological solutions that boost economic growth and, more importantly, assist in addressing societal challenges. The scientific and technological discoveries of penicillin and semiconductors, for instance, led to groundbreaking innovations. These innovations first boosted direct growth in the health and electronics industries, respectively, and later spread productivity growth throughout the economy.⁽⁴⁾For a discussion on the discovery and development of penicillin and semiconductors and their contribution to economic growth, see WIPO (2015). World Intellectual Property Report 2015: Breakthrough Innovation and Economic Growth. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=3995.

A relatively linear path from scientific discovery and technological development to industrial production is still noticeable in today’s medical innovations such as novel medicines and medical implants. Typically, a pharmaceutical product new to the market can be linked to a scientific finding of a molecule and the technologies developed subsequently to synthetize it at scale. The same applies to advanced medical implants – such as pacemakers and artificial organs – that resulted from the synergy of a scientific understanding of human biology and technological capabilities in materials engineering and miniaturized electronics.

However, mastering scientific capabilities does not necessarily lead to product and process innovation. This is for several reasons. First, scientists may lack the incentives to link with other actors because innovation is not their primary goal. Second, scientific capabilities can be very theoretical and not easily applicable when related to the most fundamental science. Third, the specific settings of scientific institutions – for example, organizational practices and culture – may differ considerably from those of private institutions leading to barriers in establishing science–industry linkages.

In addition, a country or a company does not need to master all the scientific and technological capabilities required to successfully develop new production capabilities. Indeed, skilled workers often acquire production capabilities by systematically using advanced equipment rather than through formal scientific or engineering training. This is what is known in the economic literature as learning by doing.⁽⁵⁾Arrow suggests that learning is the product of experience and hence hypothesizes that innovation (technical change) can be related to experience. He defines experience as “the very activity of production which gives rise to problems for which favorable responses are selected over time.” See Arrow, K.J. (1962). The economic implications of learning by doing. The Review of Economic Studies, 29(3), 155–173. DOI: https://doi.org/10.2307/2295952.

Similarly, not all technologies develop all the way to goods and services commercialized in the market. For instance, studies based on surveys of applicants find that between a third and a half of patents are never used commercially.⁽⁶⁾Using an extensive Pat-Val survey, Torrisi and colleagues and Giuri et al. found that between a third and a half of the patents surveyed were used only strategically or not at all. Other studies have found even lower results due to other regulations – such as medicine approval – preventing patented products from being commercialized. See Torrisi, S., A. Gambardella, P. Giuri, D. Harhoff, K. Hoisl and M. Mariani (2016). Used, blocking and sleeping patents: Empirical evidence from a large-scale inventor survey. Research Policy, 45(7), 1374–1385. DOI: https://doi.org/10.1016/j.respol.2016.03.021; and Giuri, P., M. Mariani, S. Brusoni, G. Crespi, D. Francoz, A. Gambardella, W. Garcia-Fontes, A. Geuna, R. Gonzales, D. Harhoff, K. Hoisl, C. Le Bas, A. Luzzi, L. Magazzini, L. Nesta, Ö. Nomaler, N. Palomeras, P. Patel, M. Romanelli and B. Verspagen (2007). Inventors and invention processes in Europe: Results from the PatVal–EU survey. Research Policy, 36(8), 1107–1127. DOI: https://doi.org/10.1016/j.respol.2007.07.008 Moreover, several technologies are created from other technological capabilities without requiring the related scientific capabilities.⁽⁷⁾Balland and Boschma find that many EU regions specialize in technologies without mastering the related scientific capabilities. See Balland, P. and R. Boschma (2022). Do scientific capabilities in specific domains matter for technological diversification in European regions? Research Policy, 51(10), 104594. DOI: https://doi.org/10.1016/j.respol.2022.104594. Technological advances can stem from creative combinations and applications of existing tools and concepts. For instance, 3D printing (i.e., additive manufacturing) is a technology that has evolved significantly in recent years. However, the basic principles have been known for decades. Innovations in 3D printing often involve the development of new materials and a refining of the printing process rather than any groundbreaking scientific advancements. The technology is widely used for rapid prototyping in various industries, allowing for the quick and cost-effective production of prototypes and customized products.

It is important to emphasize that innovation capabilities do not float in a vacuum. They are embedded in those individuals and organizations that facilitate the generation, acquisition and diffusion of new scientific, technological and production knowledge. These innovation stakeholders include firms and academic institutions (such as universities and public research organizations). They also include public institutions without a primary scientific or technological mission such as government agencies, financial institutions and intellectual property (IP) offices. The collection of all these stakeholders in a country, region or industry defines a living “innovation ecosystem.”⁽⁸⁾For an introduction to the notion of innovation ecosystems and the strands of the economic and social sciences literature discussing it, see Chapter 1 of WIPO (2022a). World Intellectual Property Report 2022: The Direction of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/wipr/en/2022. The industrial and innovation policies explored in Chapter 1 influence the paths taken by innovation ecosystems and their capabilities by allocating human and financial resources through public policy instruments.

How can we measure innovation capabilities?

Scientific, technological and production capabilities have their own internal consistency, yet they are also interdependent in generating innovative ideas, technologies and products. The degree of sophistication and interconnectedness of these dimensions characterizes the innovation ecosystem of a given country, region or city.

How can these three capability dimensions be measured? Typically, economic literature estimates capabilities by using a different set of outputs for each dimension. Peer-reviewed scientific publications reflect advances in science, whether incremental or breakthrough discoveries, as they are a tangible, credible and easy-to-disseminate source of new scientific information. Patent applications capture the exclusivity requests for new technologies – either methods, products or both – that are novel and have an industrial application.⁽⁹⁾To be granted, a patent must be for an invention that is novel, have a sufficient inventive step and which is susceptible of industrial application. Like scientific publications, the patent application process requires public disclosure and therefore facilitates the dissemination of technical information.⁽¹⁰⁾For an example of using scientific articles and patents as proxies for scientific and technological capabilities, see WIPO (2019). World Intellectual Property Report 2019: The Geography of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4467. Lastly, exports are considered to be an indicator of a country’s ability to provide competitive goods and services, implying that there is an efficient mechanism behind their production.⁽¹¹⁾For an example of using scientific exports as proxies for production capabilities, see Hidalgo, C.A. and R. Hausmann (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences, 106(26), 10570–10575. Box 2.1 details the data used to measure capabilities in this chapter. 

Innovative outcomes are highly concentrated in just a few countries. Over the past 20 years, the top eight countries (five percent of the countries covered in this analysis) account for 50 percent of exports, 60 percent of scientific publications and 80 percent of international patenting. Technological and scientific outcomes are significantly more concentrated than exports. As shown by the three indicators in Figure 2.1, today the world’s scientific publications, international patent families and exports remain concentrated in large countries.

A few economies – namely China, France, Germany, Japan, the Republic of Korea and the United States – have been among the top countries in all three indicators for the last five years of available data. Not surprisingly, as Figure 2.1 shows, most innovation outcomes are concentrated in high-income economies. However, the size of any economy also matters. China and, to some extent, India are two notable exceptions to the high-income economies’ concentration thanks to their large size.

However, income and size are not the whole story. There are notable differences across economies in terms of scientific, technological and production shares. For instance, Germany has a greater concentration of scientific articles, patents and exports than its share of GDP would predict (Figure 2.1). Brazil’s shares of exports and scientific articles are above its GDP share but the share of international patents is not. Indonesia’s share of exports is above its GDP share, whereas the shares of scientific articles and international patents are substantially below. 

In addition to revealing notable differences across economies, the data on scientific articles, international patents and exports enables the exploration of innovation capabilities in much greater detail. These three indicators combined can shed light on more than 600 scientific, technological and product fields representing a wide range of innovation capabilities. We have grouped these product fields into 11 scientific, 14 technological and 15 production domains.

Figure 2.2 illustrates the world's output for these capabilities according to their relative size in the period from 2017 to 2020. Among the 11 larger scientific domains, the chemistry domain accounts for 22 percent of all scientific outputs, while the engineering, and the physics and math domains account for 16 percent and 14 percent, respectively. This same trend applies to the 169 scientific capabilities identified. The engineering field accounts for 5.8 percent of all scientific publications, followed closely by the fields of chemistry (5.3 percent) and physics (4.3 percent).

Among the 14 technological domains summarized in Figure 2.2, two stand out. The information and communication technologies (ICTs) and biopharma domains account for 18 and 16 percent of all international patents, respectively. Within the 172 detailed technological capabilities, the medical and veterinary pharmaceuticals field accrued 8.6 percent of all the international patents from 2017 to 2020, followed by the computing technologies (6.6 percent) and electric digital communication (6.3 percent) fields.

Among the 15 production domains, the machinery and transport equipment domain accounts for almost 30 percent of all exports, followed by manufactured goods and articles (21 percent) and chemicals (10 percent). Among the 285 production capabilities, the motor vehicles field accounted for 3.6 percent of all international trade, followed closely by crude fuel minerals (3.14 percent) and telecommunications equipment (3.11 percent). ⁽¹²⁾When the fields are grouped into 14 production domains, the machinery and transport equipment domain accounts for almost 30 percent of all exports, followed by the domains of manufactured goods and articles (21 percent) and chemicals (10 percent). 

Why do economies specialize?

It is notable that countries do not produce the same share of outcomes across scientific, technological and production fields. Why do economies specialize in certain capabilities? Specialization offers a multitude of benefits that contribute to an economy’s growth and efficiency. By focusing on their strengths, countries and regions are able to achieve a higher level of productivity and innovation. Specialization encourages the development of expertise, leading to improved production processes and higher-quality outputs. This, in turn, fosters healthy competition, drives technological advances and enhances productivity and ultimately boosts overall economic performance.

In contrast, as discussed in Chapter 1, countries that are too specialized can be vulnerable to externalities such as global supply chain shocks. An over reliance on a particular set of industries can make them less resilient to external shocks, international market volatility and value chain disruptions, among other things. Many industrial policies implicitly apply a rationale in support of generating capabilities in strategically important industries – that is, diversification – by redirecting surplus resources toward such strategic industries and away from the most production industries. The same logic is applied to scientific and technological capabilities.

However, diversification and specialization are not necessarily opposing concepts. An orchestra, for instance, requires a set of specialized musicians in order to perform. Each musician is trained to play their instrument. Together they form a diversified group capable of playing the most sophisticated works. One-man bands that play many instruments but specialize in none are at the opposite end of the spectrum. By not being specialized in the way an orchestra is they are only able to produce simpler works. Furthermore, when individuals and firms specialize the innovation ecosystems of which they are a part can gain new, combined capabilities. This increases the ecosystem’s diversity and the opportunities to combine different capabilities. Both make more sophisticated outputs possible.

The lion’s share of most innovation capabilities is enjoyed by only a few countries. Figure 2.3 shows how the presence of all 626 innovation capabilities varies between four selected countries, namely, France, Malaysia, Tunisia and Uganda. These four countries represent four different income groups.⁽¹³⁾A very similar chart can be achieved by selecting almost any other four countries from each income group. The vertical axis shows a country’s world share for each capability. Most of France’s contribution to each capability surpasses one percent of the world’s total outcome. In other words, for most capabilities, one or more out of 100 patents or scientific publications are produced by a French inventor or researcher. Similarly, a French company is responsible for at least 1 US dollar for every USD 100 exported for most products.

The same is true for the majority of high-income economies, which are together responsible for more than 65 percent of most outputs in each field. Indeed, any given capability has at least one country that has accumulated more than 10 percent of the total. In most cases, the countries in question are the United States (with a more than 10 percent participation rate in 70 percent of capabilities), Japan (22 percent of capabilities), China (53 percent), the Republic of Korea (11 percent) or Germany (19 percent).

These and other similar economies at the innovation frontier often accumulate a considerable diversity of capabilities, making them substantive players in the global innovation arena. For instance, Italy and Japan managed to build successful motorcycle hubs based on their advanced engine and mechanical capabilities (see Chapter 4). At the same time, these two motorcycle innovation powerhouses differed with regards to many other innovation capabilities, meaning that their products evolved along successful but different technological paths.

In such a skewed and concentrated landscape, other economies are hard to find. In Figure 2.3, the selected upper-middle income economy, Malaysia is seen to display several production capabilities above the one percent threshold, a handful of scientific capabilities and only one technological capability. Tunisia and Uganda have just a few production capabilities above one percent. Small high-income economies such as Singapore, the Republic of Ireland and New Zealand often also struggle to achieve capabilities above one percent, given the size of their economies.

This means that most non-large, non-high-income countries have limited absolute resources available to allocate to the production of outcomes in all capabilities. Often, such countries achieve an above average share of outcomes in a small set of capabilities. They therefore have to prioritize the distribution of resources in order to build specialization in those key capabilities in which their economy may have a natural or historical advantage. Choices such as these can be the result of long planned industrial policies, aimed at leveraging existing economic advantages from nature or history. They can also be aimed to completely change the existing capabilities.

Figure 2.3 (x-axis) shows how much countries have accumulated of each capability relative to world share.⁽¹⁴⁾This indicator is the relative comparative advantage (RCA), expressed by the formula:share of capability x in country y / share of capability x in world total. Those capabilities on which countries concentrate relatively more than the world average have a value greater than one. Not surprisingly, France has a substantial number of scientific, technological and production capabilities above the world proportion. In respect to these capabilities, France is specialized in both absolute and relative terms. Similarly, most of Malaysia’s absolute specialization production capabilities also appear with relative specialization. More importantly, Malaysia displays relative specialization in many scientific and technological capabilities also.

The above-mentioned small high-income economies appear quite often among those countries specializing in relative terms. This is in accord with anecdotal evidence. For example, Finland and Poland created new and successful video game industry capabilities based on their strong capabilities in relative terms in ICT technologies, computer science, imaging science, computer services and audiovisual services, among others (see Chapter 5).

Combining the two axes in Figure 2.3 captures the current state of the global innovation landscape, considering both absolute and relative capabilities of economies. Both absolute and relative specializations are essential if we are to understand how resources are allocated and what are the capabilities of a given economy. On the one hand, countries whose capabilities lie in the lower-left quadrant of the figure are not specialized, since they are not contributing enough to the field, yet, at the same time, they are not trying to do so either, relative to other countries. On the other hand, the remaining three quadrants show some level of specialization that allows a country’s innovation capability to be identified.

In the rest of this chapter, innovation capabilities are measured as the combination of absolute and relative specialization exemplified by the remaining three quadrants in Figure 2.3. The top-right quadrant shows those economies that are concentrating larger shares of innovation capabilities, while also doing relatively more than the world average. Only a small set of economies – for example, France and to some extent Malaysia – can have a large set of innovation capabilities in terms of both absolute and relative specialization. The upper-left quadrant shows those capabilities where large and advanced economies – for example, China, France, Germany, Japan and the United States – show an absolute but not relative specialization.

A country making a prominent contribution in a given field is considered as specialized in that field, even if in that country the share is relatively low. Most of the other economies specialize mainly in relative terms (see the lower-right quadrant in Figure 2.3). Like most economies in the lower middle-income and low-income groups, Tunisia and Uganda share the fact that the majority of their capabilities are in the lower-right quadrant. These two economies are not considered to be specialized in all these relative capabilities. For instance, Tunisia and Uganda have fewer than five yearly patents per capability. Therefore, despite being specialized in relative terms, these capabilities cannot be considered comparable to those of other countries specialized in the same field.⁽¹⁵⁾The final method considers a country to be specialized in a given innovation field if (1) its absolute specialization is within the top two times inverse Herfindahl–Hirschman Index (HHI); or, (2) its relative specialization (i.e., RCA) is greater than unity and its absolute specialization is above the bottom two times HHI. The method is inspired by Hausmann et al. (2024) and detailed in Moscatelli et al. (2024). Hausmann, R., M.A. Yildirim, C. Chacua, S. Gadgin Matha and H. Hartog (2024). Global Trends in Innovation Patterns: A Complexity Approach. WIPO Economic Research Working Paper No. 80. World Intellectual Property Organization; and Moscatelli, F., C. Chacua, S. Gadgin Matha, H. Hartog, E. Hernandez Rodriguez, J.D. Raffo and M.A. Yildirim (2024). Can we map innovation capabilities? WIPO Economic Research Working Paper No. 81. World Intellectual Property Organization.

Distribution of capabilities around the world

How widespread is each of these scientific, technological and production capabilities globally? On average, 21 out of 154 countries specialized in every innovation capability in 2017-2020. This represents a moderate increase of five percent since 2001–2004 when it was 20 countries.

However, this average number hides the fact that economies vary in their capabilities all the time, sometimes incorporating capabilities and at other times dropping them. Figure 2.4 shows the scientific, technological and production capabilities for eight selected countries over a 20-year period. During these two decades, China, India and the Republic of Korea saw a big increase in capability diversification. China in particular had a remarkable increase in technological capabilities due to its boom in patenting. China jumped from being specialized in only 16 percent of all technological capabilities during the 2001–2004 period to being specialized in 94 percent by the 2017–2020 period. China’s technological capability diversification appears to have been preceded by an earlier diversification in scientific capabilities, which was already at 73 percent during the 2001–2004 period and jumped to 100 percent by the 2013–2016 period.

The Republic of Korea also saw a large increase in both scientific and technological diversification, as both scientific and technological capabilities were around 40 percent during the 2001–2004 period and then jumped to 66 percent and 83 percent, respectively, during the 2017–2020 period. In a similar way India saw its scientific and technological capabilities rise from 42 percent and nine percent, respectively, during the 2001–2004 period up to 68 percent and 21 percent, respectively, during the 2017–2020 period.

During the same period, Germany and Japan saw a reduction in capability diversification, as the two economies experienced a drop in all three capability dimensions. While the United States dropped production capabilities, it remained stable in scientific and technological capabilities at around the maximum diversification level. Mexico saw little change in the amount of scientific, technological and production capabilities in which it specialized. While Mexico is still one of the most diversified economies in Latin America, it has lagged in all innovation capabilities in terms of diversification when compared to China, India and the Republic of Korea. At the other end of the spectrum, the United Republic of Tanzania shows very little specialization in any of the three dimensions, although it has shown some progression in the diversification of production capabilities.

There is also substantial variation across capabilities. Figure 2.5 summarizes how common capabilities are by showing the number of countries specialized in at least one innovation capability grouped by 40 scientific, technological and production domains. There is a general pattern of technological domains being less common among countries than are scientific domains, which in turn are less common than production domains. However, the pattern does not apply across all domains or fields. For instance, during the 2017–2020 period more than a third of economies specialized in at least one capability related to the domain of exporting natural lipids (36 percent of countries in the data) or the travel services domain (38 percent); publishing scientific articles in the medical science domain (34 percent); or applied for patents relating to the machine technologies domain (36 percent). All these domains are almost six times more ubiquitous than the audiovisual technological domain (6 percent).

In general, scientific domains show less variation in terms of ubiquity. The clinical medicine domain (59 percent) is the most ubiquitous, whereas the biochemical and biotechnological domain (24 percent) is rarer across countries. In the production and technological domains, the range is much wider. The less ubiquitous production domains are financial services (11 percent) and cultural and recreational services (three percent). The already mentioned audiovisual domain is the least ubiquitous technological domain, followed closely by the electronics (10 percent) and semiconductors and optics (10 percent) domains.

A quick inspection of Figure 2.5 also indicates that the rarity of capabilities – that is, a capability in which fewer countries are specialized – increases with the level of sophistication typically associated with the activities related to the field in question. For instance, in the production capabilities groups there is an increase in rarity as products and services come to require additional transformations and involve more technological equipment. The capability to export machinery and transport equipment, for instance, is rarer than miscellaneous manufactures, which in turn are rarer than most activities in the primary sector. That is not to say that the primary sector always lacks technology (see the discussion of the agricultural sector in Chapter 3).⁽¹⁶⁾For a thorough analysis of innovation in the mining sector, see Daly, A., D. Humphreys, J. Raffo and G. Valacchi (eds). (2022). Global Challenges for Innovation in Mining Industries. Cambridge University Press. DOI: https://doi.org/10.1017/9781108904209.

Furthermore, Figure 2.5 indicates a change in the rarity of capabilities over time. Overall, technological capabilities display a slight increase in the number of countries able to produce technologies during the last two decades. The two remaining dimensions display a more heterogeneous behavior. Most scientific domains have seen an increase in the number of countries specializing in at least one field within that domain. The three exceptions are the chemistry, engineering and technology scientific domains, which are much less ubiquitous today. Likewise, production has shown an increase in country ubiquity for the majority of its domains. The most notable exception is the domain containing capabilities in exporting cultural and recreational services, which has also seen a remarkable decrease in ubiquity.

Countries exhibit a heterogeneous distribution of capabilities. The diversification phenomenon becomes increasingly present as economies grow in terms of both population and income. Within the same income group, a country’s size explains the number of capabilities in which it is specialized. For example, both Colombia and Republic of North Macedonia belong to the upper-middle income group. The large difference in the size of their respective populations explains to a large extent why Colombia is specialized in almost four times as many capabilities as the less populated Republic of North Macedonia. Conversely, higher income economies tend to have a larger set of capabilities when ecosystems from different income groups are compared. For example, the less populated Australia can match the diversification of larger, more populous countries such as India.

Having a similar number of capabilities is only one part of the story. High-income countries also tend to specialize in a different set of capabilities compared to the rest of the world. Figure 2.6 maps countries based on the proximities of their skills sets and groups them into four clusters. Cluster 1 includes the majority of high-income economies no matter their size. As countries move further away from this cluster their average income decreases. Countries such as the Republic of Ireland and Serbia show a similar level of population and capabilities. However, they have a great difference in terms of income and, despite both countries being clustered within the same group (cluster 2), the Republic of Ireland appears much closer to cluster 1.

This would imply is that some capabilities are more valuable than others. Countries eager to gain new capabilities ought therefore to consider carefully where to aim. Diversifying without a strategy could mean relocating scarce resources into fields that might not prove to be as beneficial. Therefore, it is important to determine the value of capabilities to find where the most rewarding opportunities are likely to be found. The next section focuses on the qualitative differences across innovation capabilities, regardless of dimension or domain.

Are all innovation capabilities equally important?

We have established that an economy’s innovation capabilities are related in part to both its degree of development and size, and partly to the specialization choices that an innovation ecosystem makes to further improve its functioning.⁽¹⁷⁾See WIPO (2022a). World Intellectual Property Report 2022: The Direction of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/wipr/en/2022. But what about the qualities of these capabilities?

Assessing the worth of capabilities and their potential impact on a country’s ability to innovate involves considering several factors. They include market demand, profitability, entry barriers, scalability, risk and uncertainty. Of course, compiling detailed data measuring all of these factors internationally is not easy.

Economists solve this issue partially by asking “who does what?” and “what is done by how many?”. A first step is to assume that ubiquitous capabilities are easy to adopt, and that rare ones are harder. However, this is not always the case. Some capabilities can be rare just because the incentives to develop them are low. Likewise, there may be widespread capabilities whose rewards are so high that countries are motivated to develop them, even at a high cost.

Hence, a second step is to look at how diversified are those countries that have these capabilities. As mentioned above, a broad set of capabilities allows innovation ecosystems to create increasingly sophisticated outputs. Therefore, if a rare technology appears exclusively in diversified countries, then it is the result of this process; and for it to be developed, it must be leveraged with other capabilities. Conversely, if this same technology were to appear in non-diverse countries, it would mean that countries do not need extra know-how in order to develop it, making the process simpler.

The complexity concept

Combining the diversity of countries and the rarity of their capabilities is formalized as the complexity concept (see Chapter 1). Hidalgo and Hausmann designed the complexity indicator with the aim of measuring the level of know-how embedded in a given place.⁽¹⁸⁾The economic development literature has used Hidalgo and Hausmann’s economic complexity indicator (ECI) extensively. The ECI allows a reduction in the dimensionality of the problem of understanding why economies grow. Previous efforts had aggregated data on firms, households, government and customs in order to build national accounts indicators, such as gross domestic product (GDP), investments, consumption and trade indicators. Aggregation loses much of that information by collapsing the different entries. By adopting a network analysis approach, the complexity methodology preserves more information by transforming the data into an indicator that still captures what it is that countries do. For a discussion, see Hidalgo, C.A. and R. Hausmann (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences, 106(26), 10570–10575; and Hidalgo, C.A., B. Klinger, A.L. Barabási and R. Hausmann (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. For a summary of the literature, including similar metrics such as the Fitness concept, see Balland, P.-A., T. Broekel, D. Diodato, E. Giuliani, R. Hausmann, N. O’Clery and D. Rigby (2022). The new paradigm of economic complexity. Research Policy, 51(3), 104450. DOI: https://doi.org/10.1016/j.respol.2021.104450. Computing the complexity indicator involves an iterative process considering (a) how each country specializes in each capability and (b) the number of countries specializing in each capability. In other words, complex capabilities are those that are rare and only diversified innovation ecosystems are able to make use of them. Conversely, complex innovation ecosystems are those that specialize in capabilities that are rare and in which only other diversified innovation ecosystems are specialized.

Figure 2.7 illustrates the first step in computing the complexity indicator for all 626 innovation capabilities. This establishes a reciprocal relationship between the capabilities mastered by a country and the number of countries that master a capability. The countries panel (Figure 2.7a) plots the inverse relationship between how many capabilities a country is specialized in (diversity) against the average number of countries also specializing in this same set of capabilities (ubiquity). There is a clear downward trend shown.

As countries become more diversified in general their capabilities become less common across other countries. For instance, Afghanistan is specialized in just two capabilities – fruit and nuts, and spices – which are very common, with on average about a quarter of countries specializing in them. Conversely, Germany specializes in more than 500 capabilities, and on average less than an eighth of other countries specialize in any one of them. Not surprisingly, virtually all high-income economies are to be found at the bottom right of the figure displaying both more diverse and also rarer capacities.

The capabilities panel (Figure 2.7b) plots the same relationship from the perspective of capabilities: how many countries specialize in each capability against the average number of specialized capabilities in those countries. This panel also shows an inverse relationship between the commonness of a given capability (ubiquity) and how diversified those countries that specialize in the same capability are. For instance, 59 countries (38 percent) specialize in the scientific capability tropical medicine but these same countries on average specialize in under a quarter of innovation fields. Conversely, a handful of countries specialize in the technological capabilities of audiovisual information storage and printing machines but on average these same countries specialize in 80 percent of all capabilities.

As a result, technological innovations in audiovisual information storage and printing machines are more complex innovation capabilities than the scientific production of tropical medicine. However, the inverse relationship is not straightforward This is because the rarity of a capability is not the only criterion to determine its complexity. Some capabilities may be rare because they are not attractive enough to acquire or because they are located in specific places in the world. For example, a handful of countries specialize in the production capabilities ores and concentrates of uranium and thorium and rubber, although on average those same countries specialize in less than 10 percent of capabilities. Consequently, uranium, thorium and rubber are less complex than physics, particles and fields.

An iterative process following the same principles resolves these exceptions and produces an innovation complexity indicator that ranks the complexity of all innovation capabilities.⁽¹⁹⁾See Box 2.2 for a summary of the main metrics relating to the complexity economic literature applied to innovation complexity. For more details on the complexity indicator and algorithm, see Hausmann, R., M.A. Yildirim, C. Chacua, S. Gadgin Matha and H. Hartog (2024). Global Trends in Innovation Patterns: A Complexity Approach. WIPO Economic Research Working Paper No. 80. World Intellectual Property Organization; and Moscatelli, F., C. Chacua, S. Gadgin Matha, H. Hartog, E. Hernandez Rodriguez, J.D. Raffo and M.A. Yildirim (2024). Can we map innovation capabilities? WIPO Economic Research Working Paper No. 81. World Intellectual Property Organization. Complex capabilities are those that everybody wants but few know how to develop. Figure 2.8 ranks innovation capabilities according to their complexity level. Overall, production capabilities require a lower amount of know-how to be developed, followed by scientific and technological capabilities.

The complexity spectrum

Primary sector capabilities such as the mining or agrifood industries seem easier to adopt for those countries with lower levels of accumulated know-how and diversity. Goods and services, however, appear to include a wider spectrum of capabilities in terms of complexity. In the manufactured goods domain, for instance, the manufacture of lead goods is one of the 10 percent least complex, whereas the manufacture of machine tools is in the top 20 percent most complex. For services, manufacturing services come within the lower 10 percent of the complexity spectrum, whereas exports related to IP services is one of the top five percent of fields.

Scientific capabilities are usually the intermediate dimension in terms of complexity. These capabilities likewise show up at either end of the spectrum, much like manufacturing and services, but tend to require more know-how. Earth sciences, for instance, is a mid-tier domain with 15 fields that lie between the 20 percent and 70 percent range of complexities. At the lower end are the paleontology and marine biology fields, while at the higher end are the environmental and geological engineering fields.

In general terms, scientific fields get incrementally more complex as they start to depend on more sophisticated equipment and machinery. For example, despite being a sophisticated subject, theoretical physics has a relatively low dependence on laboratory infrastructure. Technically speaking, theoretical physicists can contribute to advances in their field armed only with a pen and paper. In contrast, in order for scientists to contribute in applied physics they must conduct experiments in labs and rely on machines maintained by engineers, and so on. An indication of such complexity can be found in the average team size of scientific subjects. For instance, scientific publications in theoretical physics have almost half the average number of authors per paper compared to those in applied physics.⁽²⁰⁾Wuchty and colleagues find that theoretical physics has an average of 2.33 authors per paper against 4.08 authors per paper in applied physics. See Wuchty, S., B. F. Jones and B. Uzzi (2007). The Increasing Dominance of Teams in Production of Knowledge. Science, 316(5827), 1036–1039. https://doi.org/10.1126/science.1136099.

Technological capabilities are by far the most complex set of capabilities. All are among the 60 percent most complex fields. This across-the-board pattern indicates that the transformation of ideas – especially scientific ones – into patentable technologies is of itself a rare capability. As has been seen in Figure 2.1, concentration is much higher for patent outcomes than for scientific ones.⁽²¹⁾For a discussion of the evolution of scientific and technological concentration – both at country and regional level – in further detail, see WIPO (2019). World Intellectual Property Report 2019: The Geography of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4467. In addition, capabilities related to audiovisual, electronics and semiconductors are among the most complex ones. This pattern is in accord with the recent boom in innovation in ICTs and the even more recent boom in digital technologies. Only a few very advanced countries have managed to systematically generate technologies in these fields. These capabilities appear exclusively in diversified countries that have managed to collect know-how across different dimensions.

Figure 2.9 develops Figure 2.8 by contrasting two economies – the Republic of Korea and Egypt – with different innovation capabilities. In the first panel of Figure 2.9, the Republic of Korea shows a wide distribution of capabilities that covers most of the domains, including the most complex ones. For instance, and not surprisingly, the Republic of Korea is specialized in all fields related to semiconductors, ICTs and audiovisual technologies. The market evolution of world-leading Korean companies in these technologies – such as Samsung, LG Electronics, SK hynix and LG Display – is a concrete indication of the Republic of Korea’s innovation capabilities in these fields. 

In turn, the second panel of Figure 2.9 shows Egypt’s innovation capabilities, which are mostly a subset of those displayed for the Republic of Korea. They are concentrated in domains where complexity is lower. In the case of production capabilities, Egypt is specialized in agrifood, mineral fuels, and to a certain extent manufacturing and chemicals capabilities. In the case of scientific capabilities, Egypt is specialized mainly in capabilities related to chemistry, applied and fundamental biology, and engineering. Nonetheless, Egypt shows no particular specialization in any one technological capability. 

Why should an innovation ecosystem care about innovation complexity?

In the same way that capabilities can be ranked by complexity, an innovation complexity indicator can be produced in order to rank the innovation complexity of a given country (see Box 2.2). Conceptually, the country innovation complexity indicator ranks countries according to the level of sophistication of their innovation capabilities. High-complexity countries are specialized – in absolute or relative terms – in the most complex innovation capabilities. Figure 2.9 shows an example of the underlying difference in the complexity of countries – between the Republic of Korea and Egypt – a country’s complexity being the average complexity of their capabilities.

Economic literature has found a strong relationship to exist between complexity and economic performance.⁽²²⁾See Balland, P.-A., T. Broekel, D. Diodato, E. Giuliani, R. Hausmann, N. O’Clery and D. Rigby (2022). The new paradigm of economic complexity. Research Policy, 51(3), 104450. DOI: https://doi.org/10.1016/j.respol.2021.104450. First, not only are developed countries more diversified they are also more complex. Vibrant innovation ecosystems can generate elaborate and unique technologies that lead to the creation of complex products. Second, studies find that economies attaining technologically complex production structures typically see higher economic performance. Countries with greater complexity are also more likely to have future economic growth.⁽²³⁾Hidalgo and Hausmann find that the measures of complexity correlate with a country's level of income and that the relation is predictive of future growth. They suggest that development efforts should focus on generating the conditions that would allow complexity to emerge and thus generate sustained growth and prosperity. See Hidalgo, C.A. and R. Hausmann (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences, 106(26), 10570–10575. Furthermore, these more complex economies are more likely to be resilient, by observing longer-run patterns of economic performance.⁽²⁴⁾Balland and Rigby find that in US cities technological complexity correlates strongly with longer-run patterns of economic performance. See Balland, P.A. and D. Rigby (2017). The geography of complex knowledge. Economic Geography, 93(1), 1–23. Moreover, the reward for higher complexity goes beyond economic growth. Higher complexity is found to correlate with less inequality, lower greenhouse gas emissions and more economic development.⁽²⁵⁾Hidalgo and colleagues find that economic complexity correlates with higher economic growth, less inequality, less greenhouse gas emissions and more economic development. See Hidalgo et al. (2022).Still, some caution is needed when interpreting these results, as in most cases economic research has found a strong correlation without a strong empirical setting to test causation, see Kogler, D.F., E. Evenhuis, E. Giuliani, R. Martin, E. Uyarra and R. Boschma (2023). Re-imagining evolutionary economic geography. Cambridge Journal of Regions, Economy and Society, 16(3), 373–390.

As a result, the complexity literature views economic development as a structural transformation process. Countries grow by transforming their production structure from one dominated by low-tech, ubiquitous activities (primary products) to a more advanced structure with rarer outputs that are more reliant on human capital (manufacturing and services).⁽²⁶⁾See Hausmann et al. (2024). Figure 2.10 shows how countries with a higher level of complexity typically have a higher GDP per capita. Additionally, countries with a higher complexity measure have a strongly predictable pattern of economic growth. Countries that have high complexity relative to their income level (below the trend line in Figure 2.10) grow faster than those that underperform in terms of complexity. It is in this underperforming group that the majority of Latin America and the Caribbean countries appear.

These results are in accord with recent economic research indicating that the economic and technological complexity of a country serves as a measure for intangible assets. This allows us to quantify the hidden growth potential of that same economy. A 10 percent increase in complexity is associated with a 0.45 percent increase in GDP per capita.⁽²⁷⁾Mewes and Broekel find that technological capabilities (and their complexity) are a strong predictor of economic growth for European NUTS 2 regions from 2000 to 2014. See Mewes, L. and T. Broekel (2022). Technological complexity and economic growth of regions, Research Policy, 51(8), 104156. DOI: https://doi.org/10.1016/j.respol.2020.104156. This is particularly true for those countries whose complexity is higher than their expected GDP per capita. As a result, country dynamics have very different patterns. Those economies with a lower complexity measure have a more turbulent economic growth path.⁽²⁸⁾Using a metric equivalent to complexity (fitness metric), Cristelli and colleagues find that the predictive power of complexity (fitness) to explain economic growth depends on the level of the former. Economies with a lower complexity (fitness) will have a "chaotic" path to growth, whereas those with a higher complexity will have a "laminar” (i.e., more predictive) path to growth. See Cristelli M, A. Tacchella and L. Pietronero (2015). The heterogeneous dynamics of economic complexity. PLoS ONE, 10(2), e0117174. Available at: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0117174. 

Figure 2.11 builds on the results of Figure 2.4 by keeping only the top 50 more complex fields for each of the three dimensions – namely, scientific, technological and production – for the same eight selected countries over a 20-year period. From Figure 2.11, it is clear that the impressive economic performance of China correlates with a big increase in that country’s complex capability diversification. China has gained complex capabilities in all three dimensions, its jump in the technological dimension being the most impressive. In two decades, China went from being specialized in only seven out of the top 50 complex technological capabilities to 47 out of the top 50.

The same country’s jump in complex scientific capabilities is also noteworthy. China’s specialization in the top 50 complex scientific capabilities went from 26 out of 50 during the 2001–2004 period to 50 out of 50 by the 2013–2016 period. While Germany and Japan have seen a reduction in all capability diversification (Figure 2.4), Figure 2.11 shows that Germany has actually maintained its high degree of diversification in the top complex scientific capabilities, while Japan has done the same for the top complex technological capabilities. The Republic of Korea also shows a pattern of high diversification into complex capabilities. These results indicate that successful innovation ecosystems may drop less complex innovation capabilities but rarely drop the more complex ones.

In contrast, India shows progress in diversifying in the top complex scientific capabilities but less progress in the production and technological ones. Similarly, Mexico seems to have low diversification in all the top complex capabilities, particularly technological ones, while the United Repubic of Tanzania has yet to specialize in a single top complex capability.

Leveraging capabilities to catch up

As was seen in the case of China, the Republic of Korea and to some extent India, these rankings are not fixed. As countries gain and lose capabilities their complexity levels change. China’s new capabilities included technological know-how in ICT and transportation and scientific capabilities in medical science and clinical medicine. The addition of these complex capabilities mean that China’s is now 18 positions higher in the complexity ranking than 20 years ago.

Such changes occur across all countries although often in a less dramatic fashion. Overall, countries have increased their diversity during the past 20 years. This rise is mainly driven by countries in East and Southeast Asia and to a lesser extent those in southern Europe and South America. Other regions have experienced a reduction in diversity. North America, eastern and western Europe all saw a reduction in the number of capabilities during the same period. This trend may partly explain why some Western countries – such as the United States and European Union (EU) countries – are adopting industrial and innovation policies designed to recover some of the capabilities they have lost.⁽²⁹⁾As discussed in Chapter 1, recent years have seen a revival of industrial policies in response to a variety of new challenges faced by governments. These include a need to reduce carbon emissions to mitigate climate change (e.g., the European Green Deal and the US Inflation Reduction Act), avoid shortages of strategic goods due to global supply chain shocks (e.g., during the COVID-19 pandemic) or to support those high-tech industries critical for national security (e.g., semiconductors). 

Several middle-income countries such as the Republic of Korea China and more recently India have consistently increased their level of know-how overall by adding more complex know-how to their capabilities. As the result, the Republic of Korea has succeeded in becoming a high-income economy.⁽³⁰⁾Soh and colleagues document how the Republic of Korea is today a highly industrialized global leader in innovation and technology. Their report focuses on the Republic’s transition from a middle-income to a high-income economy. They indicate that the Republic of Korea has succeeded by focusing on building global capabilities in innovation and technology among others that they consider to be the foundations of long-term growth. See Soh, H.S., Y. Koh and A.Aridi (eds) (2023). Innovative Korea: Leveraging Innovation and Technology for Development. Washington, DC: World Bank. Available at: http://hdl.handle.net/10986/40234. While continuing to be a upper-middle income economy, China’s impressive growth during the past two decades has left it on the verge of obtaining high-income status. More recently, India’s continuous growth has put it on track to becoming a upper-middle income economy. Diversifying to more complex capabilities has helped, and continues to help, the economies in question move closer to the level of sophistication of the high-income economies.

Lower middle- and low-income countries alike are showing a decline in complexity levels, however. Rather than adding complexity, both groups have become “trapped” into focusing incrementally on less valuable capabilities thereby jeopardizing their ability to grow and exacerbating income inequalities around the world.⁽³¹⁾Pinheiro and colleagues find that EU regions and countries at an intermediate level of income can be “trapped” in low complexity activities because they the lack relevant capabilities required to move into more complex ones. See Pinheiro, F.L., D. Hartmann, R. Boschma and C.A. Hidalgo (2021). The time and frequency of unrelated diversification. Research Policy, 51(8), 104323; and Pinheiro, F.L., P.A. Balland, R. Boschma and D. Hartmann (2022). The dark side of the geography of innovation: relatedness, complexity and regional inequality in Europe. Regional Studies, 1–16. DOI: 10.1080/00343404.2022.2106362. 

How can countries choose which capabilities to pursue? Over the years there have been several unsuccessful efforts made by policymakers to “recreate Silicon Valley” in their respective regions and states.⁽³²⁾For a discussion of the geography of innovation and the unsuccessful attempts to create “cathedrals in the desert”, see Chapter 1 of WIPO (2019). World Intellectual Property Report 2019: The Geography of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4467; and Crescenzi, R., S. Iammarino, C. Ioramashvili, A. Rodríguez-Pose and M. Storper (2019). The Geography of Innovation: Local Hotspots and Global Innovation Networks. WIPO Economic Research Working Paper No. 57. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4471. The famous Californian hotspot was a small rural community at the beginning of the 20th century and is now recognized internationally as a major hub for technology and innovation, making it one of the places in the world with the most diverse and complex know-how. This success story resonates strongly in policymaking. However, other economies may be unable to replicate the multiple factors that made it possible.

Looking to develop high-complexity technologies where there are no solid foundations is like building a palace on an iceberg. Not only it will be hard to build but its inhospitable environment will make it hard to maintain and access. With no visitors nearby and nobody to fix it the structure will surely be abandoned and crumble at some point.

Related and unrelated capabilities

Knowledge gets incrementally diversified as it expands. Schools, for instance, start by teaching fundamental concepts such as mathematics and language in the early years of education and then later introduce physics, chemistry, literature and foreign languages. Some capabilities are building blocks or platforms to develop new ones.

In this sense, innovation capabilities can be considered to be like a network connecting similar forms of knowledge. Figure 2.12 maps one such a network based on how often scientific, technological and production capabilities come to be found together in the same place. In this representation, the more complex capabilities appear in the lower right corner, as they are those that only advanced innovation ecosystems have developed these or adjacent capabilities. Most of the capabilities related to audiovisual, electronics and semiconductor technologies lie in that zone. In contrast, capabilities that require less accumulated know-how will appear more isolated, usually on the outskirts of the network on the upper left of the figure. This is the case for the production fields of many raw materials (iron and copper ores, cork, oils), food and live animals (cocoa, tea, rice), and some basic manufactured goods (such as those using tin or pearls and precious stones). Most of the intermediate complexity capabilities are at the center of the network.

As discussed in the previous section, capabilities differ in their level of complexity and which subset of these capabilities a country specializes in depends on many factors. In general, high-income countries have a rare set of skills that help them produce increasingly sophisticated outputs. These complex capabilities in turn boost productivity and wealth. Figure 2.13 reproduces the capability spaces in Figure 2.12 for Australia (top left), the Plurinational State of Bolivia (top right) and China (bottom-left and bottom-right quadrants). In all four, the non-grayed capabilities represent each country’s current innovation capability specialization. Like any country with a more advanced level of innovation capabilities, Australia has innovation capabilities that are more centrally placed. In contrast, countries with a lower complexity – such as Bolivia – display a lower number of capabilities and these are located almost exclusively at the border of the network.

This comparison raises two important questions: (1) How is it that Australia reached its current level of innovation capabilities? (2) How can Bolivia catch-up?

Some countries have been consistently adding innovation capabilities during the past two decades and have benefited from an increase in complexity. But how is it that countries gain new capabilities? Figure 2.13 shows the change in China’s capabilities over the past two decades. During this time China has gained complex technological capabilities in the ICT domain, particularly in speech or audio coding or decoding, electronic circuitry, electric elements for telecommunications, and computing methods and technologies. More importantly, by 2020 China had gained most of the complex capabilities it was lacking in back the early 2000s.

China’s experience suggests that a country’s current capabilities can indicate where to go next. This is known as the principle of relatedness.⁽³³⁾The principle of relatedness establishes that greater past relatedness predicts current specialization. In other words, the capability (product or technological) space conditions the future specialization and complexity of countries or regions. See Hidalgo, C.A., B. Klinger, A.L. Barabási and R. Hausmann (2007). The product space conditions the development of nations. Science, 317(5837), 482–487; Boschma, R., P.A. Balland and D.F. Kogler (2015). Relatedness and technological change in cities: the rise and fall of technological knowledge in US metropolitan areas from 1981 to 2010. Industrial and corporate change, 24(1), 223–250; Boschma, R. (2017). Relatedness as driver of regional diversification: A research agenda. Regional Studies, 51(3), 351–364; and Hidalgo, C.A., P.A. Balland, R. Boschma, M. Delgado, M. Feldman, K. Frenken and S. Zhu (2018, July). The principle of relatedness. In International Conference on Complex Systems. Cham: Springer, 451–457. In particular, Boschma (2017) provides a comprehensive analysis. Economies tend to diversify incrementally, moving into activities that have similar skills to those they currently possess.⁽³⁴⁾Yet, regions or countries do sometimes diversify into unrelated fields. See Pinheiro, F.L., D. Hartmann, R. Boschma and C.A. Hidalgo (2021). The time and frequency of unrelated diversification. Research Policy, 51(8), 104323; and Pinheiro, F.L., P.A. Balland, R. Boschma and D. Hartmann (2022). The dark side of the geography of innovation: relatedness, complexity and regional inequality in Europe. Regional Studies, 1–16. DOI: 10.1080/00343404.2022.2106362.

This behavior can be found in other contexts, too. For example, in the labor market people move between jobs that require a similar set of capabilities.⁽³⁵⁾See the work by Frank Neffke on this topic, Neffke, F., M. Henning and R. Boschma (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265; Neffke, F. and M. Henning (2013). Skill relatedness and firm diversification. Strategic Management Journal, 34, 297–316. DOI: https://doi.org/10.1002/smj.2014; and Neffke, F., M. Hartog, R. Boschma and M. Henning (2018). Agents of structural change: The role of firms and entrepreneurs in regional diversification. Economic Geography, 94(1), 23–48. However, new positions will often require additional skills that can be learned on the job. This new set of capabilities then allows workers to move into positions that require a different set of skills from the first, performing tasks that were not present during their initial training.

Any structural transformation is a path-dependent process. However, there is room for agency. Due to related diversification, economies with similar capabilities can specialize in different areas. Finland and Sweden, for instance, have a relative abundance of forests in their territory. Such easy access to a particular resource has resulted in these two countries developing forestry capabilities during the late 19th and early 20th centuries. For both Finland and Sweden forestry is a vital part of the country's economy and related capabilities. In a simplified illustration, over time, Sweden’s move into related skills can be summarized as forestry capabilities being leveraged successively into wood treatments, furniture, designs, logistics and eventually telecommunications. Finland’s move into related skills can be summarized as forestry capabilities being leveraged successively into woodcutting and pulping equipment, heavy machinery, electronics and later telecommunications and video games. Nowadays, these paths can be illustrated by the business model successes of IKEA and Ericsson, for Sweden, and UPM, Nokia and Rovio, for Finland (see Figure 2.14 for a diagram on related and unrelated development paths).

Different capability paths can originate from a pre-existing portfolio of capabilitiesFigure 2.14 Comparison of two capability strategies: maximum relatedness vs minimum timeSource: Modified from Hausmann et al. (2024) and Hidalgo (2022).

This example of a diverging evolution illustrates how relatedness can take two different paths. Sweden’s case is characterized by a maximum relatedness path (see the bottom line of the schematic diagram in Figure 2.14). The choice to stay in the realm of related skills has allowed it to build on its existing knowledge base and resources. It minimizes the risk associated with venturing into entirely new domains while capitalizing on established expertise.

However, Finland opted for a different path (represented by the top line in Figure 2.14). This required the country to develop skills and technologies that were less related to those it had at that time. This decision allowed Finland to quickly carve out a niche in a different sector, diverging from the path Sweden was taking. In contrast to Sweden, Finland did not limit itself to the immediate relatedness of the forestry sector. Instead, it pursued opportunities that provided a quicker transition to a new field. This approach is often associated with a “minimum time” strategy, as it aims to reduce the time required for economic transformation.

Both strategies can lead to successful economic development but they showcase different approaches to structural transformation based on relatedness and time considerations. Minimum time strategies rely on making a strategic leap into a less related, more complex capability. This move can open new opportunities if successful.⁽³⁶⁾See Boschma, R. (2022). Designing Smart Specialization Policy: Relatedness, unrelatedness, or what? In Anderson, M., C. Karlsson and S. Wixe (eds), Handbook of Spatial Diversity and Business Economics. Oxford: Oxford University Press, forthcoming. 

Countries that managed to make such a move have consistently shown the ability to skip certain stages of traditional economic development and move directly to more advanced or modern stages. However, the question is can this be considered to be innovation leapfrogging? The well documented development boom undergone by several East Asian economies driven by the information technology industry shows how policies can play a big role in accelerating the process of acquiring less related capabilities.⁽³⁷⁾See a detailed discussion in WIPO (2022a). World Intellectual Property Report 2022: The Direction of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/wipr/en/2022. Nonetheless, the related capabilities were eventually built, as illustrated in relation to the Republic of Korea and China (Figure 2.13), respectively. 

Leapfrogging is therefore unlikely because the requisite new capabilities would have to be acquired first. As a result of related diversification economies tend to specialize in different areas. Any attempt to follow these kinds of strategies must be aware of timing. The window of opportunity to “leapfrog” is often narrow. Targeting an unrelated activity too late may miss the opportunity to fully benefit from such a risky move. Targeting it too early may lead to failure and wasted resources.

This is because the principle of relatedness also works in the opposite direction.⁽³⁸⁾Neffke et al. (2011) was the first paper to show that entries are related to existing capabilities within regions, whereas exits are unrelated to such capabilities. See Neffke, F., M. Henning and R. Boschma (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Countries often lose those capabilities that are isolated from their related skills. Countries that are related to a certain field are more likely not only to enter this new field but also to maintain the related capabilities they already possess. Indeed, innovation ecosystems exit certain capabilities – especially complex ones – if they do not maintain the related capabilities already in their basket.⁽³⁹⁾Hausmann and colleagues formalize this in "density regressions," where density measures the extent to which there is activity present within a country that is related to a certain activity. A country’s existing technological portfolio is strongly predictive of not just the entry of new technologies but also their exit and growth. Entry and growth are more likely when related technologies are present, whereas technologies with few related technologies around them are more likely to disappear. See Hausmann, R., M.A. Yildirim, C. Chacua, S. Gadgin Matha and H. Hartog (2024). Global Trends in Innovation Patterns: A Complexity Approach. WIPO Economic Research Working Paper No. 80. World Intellectual Property Organization. 

To make matters worse, many countries are unable to take such a risk. This is either because they have tried before, failed and no longer have the will to try again; or because a lack of resources restrains them from making such a risky leap.⁽⁴⁰⁾Balland and colleagues showed that these kinds of policies have no certainty of success, making them very risky despite the potential high reward. See Balland, P.A., R. Boschma, J. Crespo and D.L. Rigby (2019). Smart specialization policy in the European Union: Relatedness, knowledge complexity and regional diversification. Regional Studies, 53(9), 1252–1268. As resources become scarcer there is usually a prevalence of conservative investments where the probability of success is much higher.

Looking for opportunities using relatedness and complexity metrics

Not every innovation direction is equally groundbreaking. Economists consider the concepts of innovation relatedness and complexity to be helpful policy tools in guiding the selection of priorities.⁽⁴¹⁾See Hidalgo, C.A. (2022). The policy implications of economic complexity. arXiv preprint arXiv:2205.02164. Available at: https://arxiv.org/ftp/arxiv/papers/2205/2205.02164.pdf; Rigby, D.L., C. Roesler, D. Kogler, R. Boschma and P.A. Balland (2022). Do EU regions benefit from Smart Specialisation principles? Regional Studies, 1–16; Deegan, J., T. Broekel and R.D. Fitjar (2021). Searching through the haystack: The relatedness and complexity of priorities in smart specialization strategies. Economic Geography, 97(5), 497–520; and Balland, P.A., R. Boschma, J. Crespo and D.L. Rigby (2019). Smart specialization policy in the European Union: Relatedness, knowledge complexity and regional diversification. Regional Studies, 53(9), 1252–1268. While the choices an economy could pursue are numerous, not all are equally related to pre-existing local capabilities. For example, given its ICT capabilities, a region such as Silicon Valley is more likely to innovate further in ICTs than in airplane technologies. The Toulouse region in France would likely be the opposite, as it is more related to airplane technologies than ICTs.

As discussed above, capabilities connect to each other according to their proximity and complementarity, which defines their relatedness. For example, autonomous vehicle technology requires technologies and expertise drawn from both the automotive and ICT industries. Consequently, automotive and AI technologies are both closely related to autonomous vehicles technology and, through it, they themselves are connected. In general, the more related, unique and sophisticated capabilities an innovation ecosystem has the more complex the technologies it will develop in the near future.⁽⁴²⁾See Hidalgo, C.A., B. Klinger, A.L. Barabási and R. Hausmann (2007). The product space conditions the development of nations. Science, 317(5837), 482–487; and Boschma, R., P.A. Balland and D.F. Kogler (2015). Relatedness and technological change in cities: the rise and fall of technological knowledge in US metropolitan areas from 1981 to 2010. Industrial and Corporate Change, 24(1), 223–250. 

The innovation capabilities of countries, regions and companies condition their ability to generate new outcomes. Countries and regions tend to specialize in technologies and products that are closely related to their past capabilities.⁽⁴³⁾Hidalgo, C.A., P.A. Balland, R. Boschma, M. Delgado, M. Feldman, K. Frenken and S. Zhu (2018, July). The principle of relatedness. In International Conference on Complex Systems. Cham: Springer, 451–457; Balland, P.A., R. Boschma, J. Crespo and D.L. Rigby (2019). Smart specialization policy in the European Union: Relatedness, knowledge complexity and regional diversification. Regional Studies, 53(9), 1252–1268; and Deegan, J., T. Broekel and R.D. Fitjar (2021). Searching through the haystack: The relatedness and complexity of priorities in smart specialization strategies. Economic Geography, 97(5), 497–520. For instance, Silicon Valley’s capabilities are more related to ICTs, whereas Boston’s relate to health technologies and Munich’s relate to automotive technologies. Similarly, countries and regions can only specialize in a higher complexity technology, once they have attained a higher relatedness to that technology.⁽⁴⁴⁾See Balland, P.A., R. Boschma, J. Crespo and D.L. Rigby (2019). Smart specialization policy in the European Union: Relatedness, knowledge complexity and regional diversification. Regional Studies, 53(9), 1252–1268. For example, EU regions have been found more likely to specialize in a complex product if it was more related to their recent specialization.⁽⁴⁵⁾See Deegan, J., T. Broekel and R.D. Fitjar (2021). Searching through the haystack: The relatedness and complexity of priorities in smart specialization strategies. Economic Geography, 97(5), 497–520. In other words, the current relatedness of countries and regions influences their future specialization, especially for complex capabilities. This makes it hard for regions to leap to complex technologies without having first built the underlying capabilities. Therefore only a few regions and countries are able to attain more complex products and technologies. ⁽⁴⁶⁾Balland and Rigby find that only a few US metropolitan areas are able to produce the most complex technologies. See Balland, P.A. and D. Rigby (2017). The geography of complex knowledge. Economic Geography, 93(1), 1–23. 

Smart specialization

In many respects identifying the relatedness and complexity of the top countries and regions of the world – such as Silicon Valley, Boston or Munich – is relatively straightforward. These regions already have high functioning innovation ecosystems that lead the way in transforming ideas into science and technologies that nurture the complex products of today and those of the future.

However, an understanding of potential specialization and diversification strategies based on relatedness and complexity tools can be extremely important for the design of innovation policies for middle-income economies and less developed regions. This process has been paired with the concept of smart specialization, as mentioned in Chapter 1.⁽⁴⁷⁾Foray, D., P.A. David and B. Hall (2009). Smart specialisation: The concept. In Knowledge for Growth: Prospects for Science, Technology and Innovation, EUR 24047 EN. European Commission. Available at: https://ec.europa.eu/invest-in-research/pdf/download_en/selected_papers_en.pdf; and Foray, D. (2014). Smart Specialisation: Opportunities and Challenges for Regional Innovation Policy, Regional Studies. London: Routledge. Smart specialization is an industrial and innovation framework that aims to illustrate how public policies, framework conditions and especially R&D and innovation investment policies can influence the economic, scientific and technological specialization of a region and consequently its productivity, competitiveness and economic growth path.⁽⁴⁸⁾See OECD (2013). Innovation-driven Growth in Regions: The Role of Smart Specialisation. Organisation for Economic Co-operation and Development. Available at: www.oecd.org/sti/inno/smartspecialisation.htm. 

Companies or regions differ in their production capabilities. Hence the direction they should follow will vary accordingly. Innovation economists therefore advocate for countries and regions to pursue a smart specialization strategy. Such a strategy aims to encourage investments that complement the existing local production or technological assets, so as to create future local capability and competitive advantage.⁽⁴⁹⁾See WIPO (2019). World Intellectual Property Report 2019: The Geography of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4467. Given the importance of priority selection in smart specialization strategies and regional innovation policy more broadly, scholars assert the need to develop better tools for informing a given region’s priority choices.⁽⁵⁰⁾See Deegan, J., T. Broekel and R.D. Fitjar (2021). Searching through the haystack: The relatedness and complexity of priorities in smart specialization strategies. Economic Geography, 97(5), 497–520; and Marrocu, E., R. Paci, D. Rigby and S. Usai (2023). Evaluating the implementation of Smart Specialisation policy. Regional Studies, 57(1), 112–128. DOI: 10.1080/00343404.2022.2047915. In other words, how can policymakers prioritize technologies or industries when designing innovation and industrial policies that build on the local innovation ecosystem?

Some regions are becoming increasingly able to produce scientific research at the international level but fail to transform this research into patented technologies.⁽⁵¹⁾See WIPO (2019). World Intellectual Property Report 2019: The Geography of Innovation. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4467. Conversely, some regions may develop technologies without possessing the related scientific capabilities, as shown in Balland, P. and R. Boschma (2022). Do scientific capabilities in specific domains matter for technological diversification in European regions? Research Policy, 51(10), 104594. DOI: https://doi.org/10.1016/j.respol.2022.104594. Despite not being able to contribute scientific outputs, yet other regions contribute to international trade but fail to transform that production capacity into the technological learning that leads to innovation. Such regions can benefit greatly from guidance as to where to focus their limited resources in order to remove the innovation roadblocks between science, technology and production. This guidance could also inform what role the IP system can play in assisting innovation policies.

Economists are increasingly suggesting that the complexity and relatedness framework is a useful toolbox for informing innovation policymaking, notably in support of smart specialization policies.⁽⁵²⁾Balland and colleagues define their smart specialization policy framework as four quadrants summarizing the cost–benefit trade-off of prioritizing specialization in a given technology instead of another one. In this approach, an attractive smart specialization policy will prioritize those potential technologies with a high relatedness and a high complexity (low risk and high reward) and oppose the “dead-end” policy scenario of prioritizing low relatedness and low complexity (high risk and low reward). Additionally, they describe a risky but potentially high-benefit strategy of developing new technologies from scratch (low relatedness but high complexity). Last, they point to a “slow-road” policy, which is where there is a relatively low-risk but also a low reward (high relatedness but low complexity). See Balland, P.A., R. Boschma, J. Crespo and D.L. Rigby (2019). Smart specialization policy in the European Union: Relatedness, knowledge complexity and regional diversification. Regional Studies, 53(9), 1252–1268. The report mimics this approach in Figure 2.15. By combining these metrics policymakers are able to understand which capabilities countries or regions possess and how rewarding they are in terms of complexity. Additionally, policymakers can explore which of the not-yet-developed capabilities can be more easily attained, given pre-existing capabilities.

Fulfilled opportunities and untapped potential

Figure 2.15 plots the complexity of all innovation capabilities against their relatedness density in Singapore over two periods. The top-left quadrant shows the capabilities for which Singapore was specialized in 2001–2004, while the top right shows the same for 2017–2020. The change from 2004 to 2020 indicates that Singapore successfully developed more complex capabilities. In 2001–2004, Singapore was mostly specialized in capabilities with a lower complexity (the bottom-right quadrant). By 2020, Singapore had managed to become specialized in capabilities with a higher complexity (the top-right quadrant).

How did Singapore do this?⁽⁵³⁾See WIPO (2022b)Global Innovation Hotspots: Singapore’s Innovation and Entrepreneurship Ecosystem. Geneva: World Intellectual Property Organization. Available at: www.wipo.int/publications/en/details.jsp?id=4623. The process is at least partly explained by the bottom-left quadrant of Figure 2.15, which shows the opportunities that Singapore had in 2001–2004. By 2004, despite being not yet specialized, Singapore had a set of highly related capabilities (opportunities), the majority of which were low in complexity. Singapore focused on the uppermost opportunities and by 2020 it had managed to transform that high relatedness potential into concrete complex specialization. As a result, with its new set of capabilities, the bottom-left quadrant shows a handful of new opportunities, most of which are now in the high complexity spectrum. This current scenario is beneficial for the country, as it can continue to improve its complexity level and benefit from the rewards. 

National and regional innovation policies can also exploit the relatedness between capabilities of different dimensions. Indeed, countries or regions are specialized in very different areas when it comes to trade, patents and scientific publications. How do these areas relate to one another? Can scientific capabilities for example translate into economic or technological capabilities?

Capabilities might not be directly related and may not co-evolve together, although the indirect effect of scientific capabilities on the absorptive capacity of countries, regions and companies has been documented in the economic literature. Studies have shown that patenting activity across countries correlates with scientific publications but not every scientific publication necessarily leads to patenting.⁽⁵⁴⁾See Pugliese, E., G. Cimini, A. Patelli, A. Zaccaria, L. Pietronero and A. Gabrielli (2019). Unfolding the innovation system for the development of countries: Coevolution of Science, Technology and Production. Scientific Reports, 9, 16440. DOI: https://doi.org/10.1038/s41598-019-52767-5. Similarly, other studies find that regional scientific capabilities in given scientific fields predict the development of related new technologies in the corresponding technological fields in the same regions. Recent studies find that countries are more likely to diversify in technologies that are related to existing scientific capabilities.⁽⁵⁵⁾These studies have explored the extent to which technology classes are related to scientific publications by analyzing citations of patent families in various technology classes to scientific fields, finding that countries are more likely to diversify in technological classes that are often citing specific scientific fields. See Furman, J. L., M. E. Porter, and S. Stern (2002). The determinants of national innovative capacity. Research Policy, 31(6), 899–933. DOI: https://doi.org/10.1016/S0048-7333(01)00152-4; Balland, P. and R. Boschma (2022). Do scientific capabilities in specific domains matter for technological diversification in European regions? Research Policy, 51(10), 104594. DOI: https://doi.org/10.1016/j.respol.2022.104594; Shin, H., D. F. Kogler and K. Kim (2023). The relevance of scientific knowledge externalities for technological change and resulting inventions across European metropolitan areas. Review of Regional Research, 1–17. DOI:10.1007/s10037-023-00190-9; and Hausmann, R., M.A. Yildirim, C. Chacua, S. Gadgin Matha and H. Hartog (2024). Global Trends in Innovation Patterns: A Complexity Approach. WIPO Economic Research Working Paper No. 80. World Intellectual Property Organization. A similar rationale follows the link between trade capabilities and the probability of entering new technological fields.⁽⁵⁶⁾Hausmann et al. (2024) explores this in a similar way as Shin et al. (2023) do for scientific fields. See Hausmann, R., M.A. Yildirim, C. Chacua, S. Gadgin Matha and H. Hartog (2024). Global Trends in Innovation Patterns: A Complexity Approach. WIPO Economic Research Working Paper No. 80. World Intellectual Property Organization; and Shin, H., D. F. Kogler and K. Kim (2023). The relevance of scientific knowledge externalities for technological change and resulting inventions across European metropolitan areas. Review of Regional Research, 1–17. DOI:10.1007/s10037-023-00190-9.. 

These connections can shed light on the untapped innovative potential of countries. The interplay between the three dimensions in the innovation frontier can help countries identify latent capabilities.⁽⁵⁷⁾In practical terms, this report defines as frontier innovation ecosystems those in cluster 1 of Figure 2.6. Figure 2.16 contrasts the untapped technological potential of a medium-sized high-income economy (Canada) with the untapped technological potential of a middle-income economy (Colombia). Figure 2.16 uses the proximity connections between scientific and technological capabilities in the economies from cluster 1 (in Figure 2.6) to estimate the number of patents that could be expected to be seen in a country based on its scientific publications if it were the average country in cluster 1. It refers to potential output in a capability given the current outcome on related capabilities.

Both countries have domains where, based on their scientific outputs, there is untapped technological potential. For Canada (Figure 2.16a) there is room for improvement in three of the most complex domains – audiovisual, electronics, and semiconductors and optics. The average economy in cluster 1 would produce more patents if it had the same scientific outputs as Canada. For example, given its scientific production, Canada produces half as many patents in audiovisual technologies and two-thirds as many in chemical technologies compared to the average cluster 1 economy. In contrast, with the same scientific output, Canada produces 16 percent more patents in civil engineering technologies than the average cluster 1 economy in Figure 2.6.

This insight can be powerful when it comes to identifying missing links between the stakeholders in an innovation ecosystem. By looking into how these dimensions interact in a well-functioning ecosystem policymakers can prioritize between domains and zoom into the relations between academic institutions, industry and the IP system, so as to identify the particular constraints that are stopping the economy from reaching its full potential. For less diversified economies such as Colombia technological capabilities are less present at the international scale, and its observed patents are far from reaching their potential. Indeed, Colombia’s transformation of scientific publications into international patents is in all fields less than 50 percent of that of the average cluster 1 economy. This is particularly relevant for biopharma and ICTs where Colombia produces a considerable related scientific output but realizes no more than 18 percent and six percent, respectively, of the technological transformation potential.

Conclusion: the key to successful development

This chapter has explored the empirical literature on innovation capabilities and presented new evidence based on data drawn from scientific publications, international patents and exports. In doing so, it has explored the potential relevance of using measures of innovation capabilities to inform the design of innovation and industrial policy.

First, the chapter has studied the need for a multidimensional measurement and an analysis of innovation capabilities. Categorizing innovation capabilities according to whether they are scientific, technological or production capabilities – measured by scientific publications, patents and trade data – seems to be a useful approach to mapping the different innovation ecosystems that exist around the world.

Second, to further understand the implications of a country’s specialization in certain innovation capabilities it is crucial to comprehend the quality of those capabilities. The complexity metrics illustrated add deeper insights that go far beyond how ubiquitous or rare any particular scientific, technological or production field might be. The empirical evidence shows that the development of more complex scientific, technological and production capabilities correlates with economic growth. Furthermore, the chapter has identified differences in the level of complexity of capabilities between dimensions in general, highlighting the fact that the ability to produce technological innovations is, overall, the most complex and rewarding of the three dimensions analyzed.

Third, the chapter documents the capabilities that exist in each innovation ecosystem that are predictors of new capabilities. The dynamics of innovation capabilities, relatedness and complexity present a framework for understanding the progression toward the economic and technological development of an innovation ecosystem – either local, regional or national. These interconnected concepts and metrics can help policymakers to adopt a strategic approach that encompasses the co-evolution of different domains and their interdependencies. By doing so, economies can address binding constraints, stimulate positive externalities and promote a resilient innovation ecosystem.

Lastly, the chapter has documented the importance of innovation diversification for countries and the relationship between science, technology and production. The ever-changing landscape of capabilities and their relatedness underscores the need for strategic diversification. This evolution is not a one-size-fits-all process. Instead, it allows countries to choose from diverse paths based on their unique circumstances. Some may opt for a strategy that builds progressively on existing skills, while others may aim to accelerate the transition to a new field by targeting less related domains, known as leapfrogging. The choice of strategy should be well-timed and align with a country's specific goals and resource availability. The timing of a venture into unrelated activities is of vital importance. Pursuing such a venture either too early or too late can result in missed opportunities and a waste of resources. Policymakers need to be able to recognize a narrow window of opportunity when it opens and have the related capabilities in place.

There are also some limitations to note. While very important, the scientific, technological and production dimensions are not the only dimensions related to innovation capabilities. For instance, non-patentable technologies and non-tradable goods also make a contribution to the innovation capabilities of an ecosystem. Yet, these two dimensions are poorly measured by scientific publications, patents and trade data. Moreover, country level analysis of these two dimensions might be too aggregated, confounding regional capabilities of countries with large territories. For example, it cannot be assumed that the aggregated capabilities of Silicon Valley and New York City apply to each other, and that they apply even less to many areas in the center of the United States. Lastly, some caution is needed when interpreting the results of the complexity and economic growth correlation. In most cases, economic research has found a strong correlation without a strong empirical setting to test causation. Moreover, there is still a limited conceptualization and understanding of the mechanisms through which these relationships are working, which limits the potential empirical tests.

Some of these limitations can be at least partially addressed by an analysis that is more qualitative and focused. With this in mind, the next chapters explore innovation capabilities and related concepts, such as relatedness and complexity, as they apply to three case studies: agricultural technologies (or “AgTech”) (Chapter 3), motorcycles (with a focus on e-bikes) (Chapter 4) and video games (Chapter 5). While most of the general findings of this chapter are there to be seen in these case studies, they take a much deeper and intuitive dive into innovation capabilities.

In sum, managing innovation capabilities and their relatedness is pivotal for those countries seeking long-term growth and competitiveness in an ever-evolving global economic landscape. By embracing the principles of complexity and smart specialization, comprehending related and unrelated capabilities, and making well-informed strategic decisions, countries can position themselves for success and sustainability in economic and technological development.