Government-driven knowledge networks as precursors to emerging sectors: a case of the hydrogen energy sector in Korea

1. Introduction

It is generally thought that developing new sectors has shown a positive relationship with economic growth (Amsden, 1989; World Bank, 1993; Stiglitz, 1996; Fagerberg, 2000). Saviotti and Pyka (2004) put forward qualitative changes such as interactions, competition and the entry of new actors to describe a country’s economic growth. Their contribution has attracted attention to emerging sectors, especially of researchers in developed countries that need to nurture new industrial sectors for sustaining economic growth and creating jobs. In addition, since climate change has become a global issue, a growing number of developed countries have become interested in the emerging sectors of sustainable energy.

The emergence of a new sector, however, involves a number of components, factors and complex processes. Therefore, a multi-faceted perspective is needed to understand the entire process, from the seeding to mature phases. Various actors aiming to overcome many barriers are involved. An emerging sector demands challenges, especially in terms of risks and uncertainties, which are related with not only technological, but also socio-economic aspects.

It is required for emerging sector participants to address these challenges. Participating firms need to monitor the rapidly changing environment and integrate their internal and external capabilities (Teece et al., 1997; Gulati et al., 2000; Hagedoorn et al., 2000). To mitigate risks and uncertainties, governments and public research organizations (PROs) can support appropriate R&D activities and promote network formation among actors for collaboration, while universities can help build the knowledge base and provide human resources to the innovation system (Freeman, 1991; Mowery and Sampat, 2005; Spencer et al., 2005).

All these activities, however, cannot be coordinated by a single actor in a short period of time; interactions lie at the core of activities in the emerging period. There are many motives for actors in their interactions within emerging sectors. For instance, actors can acquire technological capabilities and resources to gain new scientific knowledge and maximize economic feasibility (Mowery et al., 1996; Hagedoorn et al., 2000), and they can come to agreements through negotiations within the society (Callon, 1995). Heterogeneity in the composition of actors in terms of knowledge components, capabilities, and various other characteristics will manifest in the evolutionary processes of a variety of relationships that range from cooperative to competitive, market to non-market, and formal to informal (Van de Ven and Garud, 1989; Nelson, 1994; Dosi et al., 1997; Gulati et al., 2000; Marlerba, 2002, 2004, 2006).

Extant literature has contributed to our understanding of the evolution of industries. Utterback and Abernathy (1975) described an evolutionary process that involves the co-development of technology and institutions via repeated interactions among participants. Rosenkopf and Tushman (1998) and Orsenigo et al. (1998, 2001) tried to uncover the relationship between the evolution of knowledge and industrial networks. Van de Ven and Garud (1989) and Malerba (2002, 2004, 2006) took into account the patterns of transformation of the systems. Nelson (1994) suggested the concept of co-evolution—which represents interactive learning, changing environments, and selection mechanisms—to evolve together in organized directions. However, many challenges still lie ahead. In particular, the primitive form of the emerging sector has yet to be uncovered, and few attempts at the empirical studies of the formative stages have been made because of the difficulties in establishing explicit evidence that incorporates a multi-faceted perspective.

In this study, we examine R&D networks in which actors share visions and knowledge as the origination of emerging sectors and the role of government in building knowledge networks. We identify the building blocks of the sectoral systems of innovation (i.e. knowledge, actors and networks, and institutions) which form even prior to the commercialization process, so that we can regard knowledge networks as precursors to emerging sectors. The analysis of knowledge networks will show the necessary components of emerging sectors that are integral to knowledge networks and their evolution with respect to interactive learning and changing environments in organized directions that are set by key actors. In particular, we will show that knowledge networks can be driven to be formed by a government, and that a government plays a key role in the expansion and structuration of knowledge networks.

For an empirical case study, we consider the emerging hydrogen energy sector in Korea, as an increasing number of actors have been involved in this sector and are interacting with each other to form the knowledge network for this sector. The social network analysis method is used to visualize the network formation and its patterns of evolution. The methodology requires us to look back by investigating the interactive relations among actors as well as actors’ positions within the network.

Supported by the empirical findings, this article can provide added theoretical value for a narrower understanding of the building blocks of particularly emerging sectors than the existing literature on innovation systems, which has provided us with the identification of the building blocks (i.e., Malerba, 2004).

This article consists of five sections. In Section 2, we review prior literature on the formation of emerging sectors and the role of government. In Section 3.1, we introduce the current situation of the Korean hydrogen energy sector to describe the environment and conditions of transition. The study methodologies are described in Section 3.2. Section 4 presents the findings and interpretations from the perspective of social network analysis. In Section 5, we summarize our arguments and suggest further research topics.

2. Literature on the emergence of new sectors

2.1 From industry formation to the emergence of the sectoral systems of innovation

A number of studies have reported the dynamic patterns of entry and exit, turbulence, and innovative activities in industrial sectors. From the perspective of industry life-cycles, Utterback (1994), Klepper (1997), and Klepper and Graddy (1997) argued that industries evolve through a couple of stages over time. Industry life cycle theory presents the patterns and characteristics of the sectors in which the value chain has already been established, and the necessary components for emergence have been identified to a certain extent. Therefore, it would be valuable to explore the primitive forms—the precursors—of emerging sectors.

The extant literature has also examined the formation of industrial sectors from the perspective of sectoral innovation systems. Pavitt (1984) came to recognize that different sectors have particular patterns of technological change that are influenced by the knowledge base, innovation processes, and interactions among heterogeneous agents. Thereafter, attempts were made to describe the technological development in various sectors using a framework that was integrated and consistent across several dimensions (Nelson, 1994; Edquist, 1997; Malerba and Orsenigo, 1997). As a result of these efforts, Malerba (2002, 2004) introduced the concept of the sectoral systems of innovation. A number of articles on the sectoral systems of innovation emphasized the multi-dimensional, integrated, and dynamic views of the innovation process. Malerba (2002, 2006) highlighted that emerging sectors did not originate in a vacuum, but from several existing sectors over which new clusters spanned. He gave examples of sectors such as Internet–software–telecoms and biotechnology–pharmaceuticals to depict the integration of knowledge and technology, new interrelations between actors, and the expansion of boundaries. In addition, Jacobsson and Bergek (2004) took into account sectoral transformation to explain the emergence of the renewable-energy sector.

Malerba (2002, 2004) also explained that the transformation of sectors results from the co-evolution of building blocks, through which it is possible to understand the mechanisms and dynamics of sectoral systems. He identified three key building blocks of the sectoral systems of innovation: knowledge and technologies, actors and networks, and institutions. It drew an important suggestion that the building blocks can be considered necessary components for emerging sectors. Although the value chain analysis may provide us with an evidence of building blocks in emerging sectors, it is not appropriate when a sector has not yet matured to a certain stage, such as the widespread commercialization of a dominant design.

The results from several case studies on new sectors provided some insights into the necessary components for emerging sectors. Russo (2003) emphasized institutional environments, and Giarratana (2004) stressed that a sound technological base, niche-based products, and patent systems are key factors in emerging sectors. Meanwhile, communities and organizations that bridge actors for sharing know-how and practices were cited as being a necessary component for new sectors (Mezias and Kuperman, 2000; Sapsed et al., 2007). The creation of knowledge through strategic alliances among firms was also mentioned as a driver of new sector formation (Murtha et al., 2001).

While valuable for documenting the necessary components for emerging sectors, the identification of a more integrated and comprehensive framework that incorporates these factors and their dynamics can yield fresh insights into the emergence of new sectors. As we will illustrate, one such framework could be the knowledge network.

While a sector is evolving, the network for that sector plays a central role. Gulati et al. (2000) ascertained that firm performance is heavily dependent on network structure, network membership, and the modality of inter-firm ties. Heterogeneity in terms of both observable variables such as size, age, and location and unobservable variables such as knowledge and capabilities is connected to the activities of industries. Inter-organizational linkages are common strategic means of acquiring resources and enhancing competitiveness. By connecting heterogeneous partners, actors can broaden their knowledge base and internalize external capabilities in a rapidly changing environment (Dosi et al., 1997; Hagedoorn et al., 2000; Powell et al., 2005). Actors with limited resources need to spread their risks and costs by linking with others. In particular, motives for risk reduction help increase profitability when big investments are involved (Mowery et al., 1996; Hagedoorn et al., 2000). Heterogeneous actors who have different technological backgrounds and various capabilities can create desirable solutions to address the difficulties in technology development, and this helps to overcome the technological challenges of emerging sectors (Mowery, 1988; Shan et al., 1994).

Different participants play various roles in the network. In history, large firms have often appeared as key actors who led the early stages of emerging sectors. The reason large firms first took the risks was probably due to the difficulties and uncertainties that require management in an emerging sector, which include the risk of failure, the demand for high internal capability, and the need for capitalizing the knowledge base for a new sector (Agarwal and Audretsch, 2001; Ganco and Agarwal, 2009). Large firms’ risky investments fuel the emergence of a new sector, and their R&D activities contribute to the deepening and broadening of the knowledge base. However, the key actor can differ by sectors and countries (i.e. national innovation systems). For example, in the case of the computer industry, it was universities that carried out government-funded basic research that made remarkable contributions during the early stage of the sector’s development (Malerba and Orsenigo, 1996). Entrepreneurial start-ups often appear as major actors when the sector is very young, especially in technology-intensive sectors (Agarwal and Audretsch, 2001). In some countries that have the significant tendency of government intervention in R&D, the role of PROs should be emphasized. The roles of PROs are worth noting because: first, PROs can perform either basic or intermediate R&D activities, which private organizations often lack. Second, PROs can play a bridging role in linking distanced actors to create new opportunities for collaborations, as well as in bridging knowledge flows (Sapsed et al., 2007; Cassi et al., 2008). In addition, PROs are under the relatively direct influence of government policies, so governments can affect the evolution of the network, aiming for the emergence of a new industrial sector by steering PROs in the network.

The expansion of relationships fosters the scale-up of experiments, and this provides momentum for transition by achieving consensus and social acceptance. As a result, a set of relations (i.e. a network) offers a negotiation space that generates a space, a period of time, and resources where innovative activities can take place (Law and Callon, 1992). The importance of networks in the emerging period was thoroughly examined empirically in a study by Barley et al. (1992). The results of their study showed us that the sampled firms had established over 2200 relationships to strengthen their competitiveness from 1973 to 1989 in the formative stages of the biotechnology sector.

A network is a result of not only supply–demand relationships, but also the knowledge and institutional factors that are embedded in the system. The means of gaining benefits from linked actors differ between the mass-production sector and the science-based sector. The rules and customs of society permit the linkage of some—but not all—actors (Leung, 1993; Kogut, 2000; Owen-Smith and Powell, 2004). A knowledge network that is formed prior to the value chain significantly embeds knowledge and institutional factors, since, as we explained previously, actors in emerging sectors focus on linking with heterogeneous actors to diversify their knowledge base and to build a consensus, not on transacting their products.

Still, there exists a literature gap between the investigation of the dynamics of industry evolution and the observation of the emergence of a new species. Although it is required to see time-lined data in the early stage of, or even earlier than, the emergence of an industry, the investigations so far are often limited to a-point-in-time snapshots for the status of the industries. To see the early figures, we need to identify what might exist before the formation of an industry, in which there was seldom any possibility to find business partnerships between firms. Instead, other forms of networks of actors, in which they link to each other and some actors play bridging roles between actors, could be observed. Knowledge networks may be a strong candidate to be observed prior to the emergence of industries. A growing number of researchers have become interested in the presence of knowledge networks in various industrial sectors (Okamura and Vonortas, 2009). In addition, government policies that favor networks will influence network formation and evolution (Breschi et al., 2009). For example, a contract research network formed by government R&D funding can help in the forming of science networks, that is knowledge networks (Wagner and Mohrman, 2009). This article will attempt to make a contribution to fill the literature gap by linking those knowledge networks to the emergence of industrial sectors, since knowledge networks have the potential to be transformed to networks in industrial sectors. We will discuss this transformation later in this article through our analysis and explanations on the changing roles of various actors and the evolutionary pattern of network structures.

2.2 The role of the government in emerging sectors

Governments can allocate resources to avoid the duplication of projects through research portfolio diversification, which enables researchers to choose different strategies. Also, funding agencies can manage project schedules to avoid inefficiencies (Dasgupta and David, 1994). Such methods provide for the relevance of public intervention to reduce risks and facilitate the flow of knowledge. They are of significant importance in the emerging period during which available resources are limited and trajectories towards a dominant design remain invisible.

One role of governments is to promote the formation of knowledge networks (Freeman, 1991). Governments would want to seed knowledge networks through STI policies. Government policies facilitate collaborative networking among firms, universities, and PROs that are involved in the formative stages (Polt, 2001). Governments can carry out their role by forming networks and placing key actors in appropriate positions. The structure of the network and similarities among actors in the network are important in the decision-making process, resource access, and entry points, which also influence the path to emerging sectors that can be determined by government actions (Law and Callon, 1992; Callon, 1993).

Technological programs can form knowledge networks among actors. Government programs contribute by guiding the technological trajectory using a flexible portfolio of programs to avoid inefficiencies associated with R&D activities such that it does not lock irreversibly (Callon et al., 1992; Callon, 1995). At the same time, a network of private and public entities represents an effective channel that provides information regarding the players and their development trends (Tijssen and Korevaar, 1997). Government R&D programs can work as this channel, since they help actors to be invited into the knowledge network, where they share information and knowledge with each other (Cassi et al., 2008). These programs help both the firms and the governments to allocate their resources appropriately and to decide the right time and place for entry during the emerging stage.

It should be noted that firms will have some obligations if they are engaged in government research programs that aim to promote collaborations among various actors. The trade-off of receiving support is sharing certain knowledge with their partners. Government programs are particularly effective in emerging stages, when markets are rarely formed and competition is not active. Firms within the network can gain numerous opportunities to reduce risks and uncertainties by sharing them with the government. Because of these positive effects, invitations to government programs are easily welcomed by private actors, despite certain obligatory requirements placed upon them (Tijssen and Korevaar 1997; Polt, 2001; Audretsch et al., 2002). Furthermore, in some countries, there exist strong connections between the government and private firms, which is significant especially in recently developed countries whose firms may have limited internal capabilities and whose government has a strong tendency for intervention.

Following this literature, we need to further examine the formation and evolution of a knowledge network regarding whether it becomes a precursor to an emerging sector, and the role of governments in forming and managing the knowledge network. To find evidence, an investigation of the characteristics of the evolutionary pattern of knowledge networks is needed.

3. Case analysis

3.1 The emergence of the hydrogen energy sector in Korea

The hydrogen energy sector has received increasing attention as a possible option for sustainable development. When many developed countries began to invest in hydrogen energy development, the emerging hydrogen sector became observable. Further, the Korean government, with its relatively strong authoritative views, has publicly funded R&D projects as a primary means of facilitating a variety of inter-organizational cooperation and has invited various organizations to engage in joint research (Kim, 1993, 1997). In this respect, the Korean case provides us with a good case to study a probable precursor to an emerging sector from a government-driven knowledge network by large R&D programs.

After the second oil crisis in 1979, the Korean government, concerned with energy security, enacted the Alternative Energy Technology Development Promotion Act in 1987 to reduce its heavy dependence on fossil fuel and to promote R&D activities on renewable energy, including hydrogen energy. The first government-funded R&D program on hydrogen energy was launched by the Ministry of Science and Technology (MOST) in 1989 (KISTI, 2003; OECD, 2006). A total of USD 63 million was invested in fuel-cell technologies between 1988 and 2002. Of the total investment, the ratio of public funding was 55 % (OECD, 2006), a relatively high proportion among OECD countries.

Eight major R&D programs were initiated by the government. The first program, “New and Renewable Energy Technology Development (Hydrogen division)”, was launched in 1989. It covered all of the sub-fields in hydrogen energy, from hydrogen generation to consumer products and vehicles. Therefore, a variety of actors were engaged and large investments were made in this program. While this program was proceeding, two G7 Programs related to alternative energy and vehicles underpinned the former. The programs in their early stages focused mostly on basic research. Thus, PROs and universities received a large portion of the allocations.

There has been a notable change in government programs since MOST commenced the “High Efficiency Hydrogen Production R&D Program” in 2000. Its name was changed to “Hydrogen Energy R&D Center”, and it has been successful in developing generic scientific knowledge. In turn, three programs funded by the Ministry of Commerce, Industry, and Energy (MOCIE) have been launched to commercialize this generic scientific knowledge.

Such strong support by the Korean government was partly influenced by the establishment of the Kyoto protocol (OECD, 2006; Hronszky et al., 2008), and then the announcement of US President Bush’s Hydrogen Fuel Initiative for huge funding for hydrogen energy and hydrogen automobiles in 2003. After these events, the Korean government also announced a big push for hydrogen energy. These actions by the government convinced and encouraged various stakeholders, including private firms, to become involved.

In addition, some local activities are worth noting. Although local governments have recently started to incubate local start-ups, local governments and players are likely to rely on the support of the central government due to Korea’s highly centralized characteristic in the public sector. For example, while Jeonbuk province supports its venture network in the Jeonbuk Science Park in the area of hydrogen energy, they also implement central government programs.

If government support stops, linkages might not be sustained. In fact, some actors who left the network to pursue independent R&D, such as Ssangyong Cement Ltd, have chosen not to pursue hydrogen R&D in the end. In addition, several electronics companies of the Samsung Business Group left to develop their own network and have shown some performance, but they came back into the network after 2005. These events may indicate that government support shows significant benefits, despite the required obligations involved.

The Korean government has invited heterogeneous actors with different backgrounds to become involved in the network. Large firms, in particular, have participated aggressively, along with PROs and universities. These large firms hail from traditional sectors, such as the automobile, electronics, and conventional energy sectors. As large firms dedicated considerable efforts in developing hydrogen energy technologies, small and medium-sized enterprises (SMEs) that specialize in supplying components have started to participate in the network as well. In addition, there is a growing sense of recognition, among stakeholders at least. Their shared vision of a Hydrogen Economy includes economic prospects, while the government is also concerned with energy security and sustainability (Hronszky et al., 2008).

3.2 Methodology and data

Social network analysis is an interdisciplinary methodology that focuses on the relationships among interactive actors. It is based on the belief that the inter-organizational relationships in a network are the most important explanatory variables for complex events. Social network analysis has attracted the attention of disciplinary fields, because it successfully addresses research questions and challenges that concern the behavior and performance of actors. In this context, a wide range of issues are studied through social network analysis (Wasserman and Faust, 1994).

Patents and publications are very traditional and accepted sources of data for examining the networks of innovators. Both the strengths and weaknesses of the uses of patent and publication data are widely recognized. Murray (2002), Smith (2005), Cantner and Graf (2006), and Klitkou et al. (2007) conducted research by using data on patents and publications. The study of Klitkou et al. (2007) is particularly relevant because the authors discovered strong connections between science and technology in the hydrogen energy sector, which is the subject of this article.

Data on collaborative agreements is frequently used to analyze networks. The use of such data is appropriate for representing relationships among actors. Moreover, the data are available from well-organized databases, such as Bioscan and MERIT-CATI. Data on collaborative agreements were used by Hagedoorn and Schakenraad (1992) to study information technology and by Owen-Smith and Powell (2004), Powell et al. (2005), and Roijakkers and Hagedoorn (2006) to investigate biotechnology.

Transaction data are also applied in research relating to STI policy. Bonaccorsi and Giuri (2001) used transaction data in the commercial aircraft industry to establish the evolutionary dynamics that are interconnected with the structure of the industry.

Lastly, government R&D programs can be an important source of data. Meyer (2005) made use of German government R&D program data to show linkages between scientific and technological knowledge in the nanotechnology sector. The use of government R&D programs as indicators of linkages among actors is not common, but can be valuable in some contexts, for example, in emerging sectors that require government intervention due to extremely high levels of risks and uncertainties.

In this article, eight large government R&D programs of Korea in the field of hydrogen energy research were examined and utilized for social network analysis. These programs were major hydrogen R&D programs that were implemented between 1989 and 2005. Each program included dozens of projects that focused on specific research topics. Table 1 presents the main features of the eight government-funded R&D programs. The data was obtained from the Korea Research and Development Integrated Management System, which is maintained by the Korea Institute of Science and Technology Evaluation and Planning (KISTEP). To compensate for missing entries, we supplemented the database of KISTEP with the final official reports of R&D projects that were published by project groups and relevant documentations from R&D agencies that manage the projects.

The structure of the programs is shown in Figure 1. R&D programs are funded by MOST and MOCIE, and both have similar structures. The basic research unit is a project. Some projects had been carried out independently, but of late, most projects have been carried out under the supervision of national R&D organizations. National R&D organizations arrange R&D consortia for achieving specific R&D targets. Most projects are tied to sub-groups to share knowledge and solve problems. Between government and national R&D organizations, there are intermediate national agencies that manage funds, handle administrative matters, and evaluate results at the final stages of the projects.

The ties in the network described in this article are categorized into three types as shown in Figure 1. The first connects a sub-group leader with project leaders. The second links a project leader and participant firms for a specific project. The third connects a project leader to universities and PROs for a specific project. Whether all the participants in a specific project are tied with each other or whether ties are formed only between the upper- and sub-level organizations depends on how the participants make contracts on research topics and share their knowledge in implementing research. It should be stated that we have not considered the relations between national R&D organizations and their own sub-groups.

To represent social networks, Ucinet software, developed by Analytic Technologies, was used. Spring-embedding layout, which is based on an algorithm of Kamada and Kawai (1989), was adopted to present the network maps, wherein nodes with more connections are located nearer to the centre, while less-connected nodes are placed on the periphery. The shape of a node indicates the type of the associated organization: circles denote firms, squares represent PROs, and triangles are universities. Besides the shape, the colour also distinguishes the node of a firm with regard to the size of the corresponding firm.¹ Gray circles stand for large firms, and dark circles represent SMEs. Non-R&D bodies, including the government and quasi-government agencies, were intentionally omitted because the government and quasi-government agencies are not directly involved in a knowledge network in which participants created and exchanged knowledge.

The concept of centrality is that a central actor who has the highest centrality is the most active and influential within the network. In this article, the degree centrality of a node is applied as an index that indicates the total number of ties of the node. It is not concerned with the number of projects with which an actor is involved. Valued ties and binary ties are used to measure degree centrality. The valued ties allow multiple counting with one adjacent node, but the binary ties do not allow it. Both types are utilized in different ways. The former is typically applied to absolute indexes to present the level of concentration, and the latter is applied to calculate the proportional indexes that need the theoretically maximum number in the network (Wasserman and Faust, 1994). In the case of node size, it varies with the degree centrality of the valued ties. In addition, the strength of ties stands for the simple count, without any weighting, of shared ties.

4. Findings and interpretations

Figure 2a–c show the government-driven knowledge network for those years. All organizations are abbreviated, and the full names of the organizations are listed in  Appendix A. Visualizations are useful for intuitively viewing changes in the network and for suggesting what to look for in the network before we analyze and interpret the network in detail. As can be seen in Figure 2, the scope of the government-driven knowledge network expanded over time with an increasing number of new actors and additional ties. Eventually, the knowledge network, in which the central actors visibly stood out, became quite complex.

4.1 Patterns of network development: structural changes over time

When actors who become interested in a specific technological domain get together, it would be beneficial to develop relationships through which they can achieve technological progress before the new sector actually emerges (Mowery, 1988; Shan et al., 1994; Hagedoorn et al., 2000). On the one hand, when non-profit organizations as well as private firms join an emerging sector, risks and uncertainties may decrease. On the other hand, technological opportunities may increase through the integration of diverse knowledge and sets of experience (Utterback, 1994; Gulati et al., 2000; Robinson et al., 2007). At the same time, actor roles vary with their positions, specialties and backgrounds, and their diversity is necessary for building a solid foundation of an emerging sector.

Figure 3 presents the rapidly increasing numbers of actors and ties in the knowledge network related to hydrogen energy in Korea.² A notable aspect that was revealed by the figure is that, once the numbers of actors and ties exceeded certain levels, then they dramatically increased, which is suggestive of a snowball or bandwagon effect. Large, newly launched programs that were initiated after 2003 led to a sharp increase in the number of both actors and ties. This bandwagon effect was a result of interactive learning between stakeholders, including the government, resulting in shared expectations of the technological feasibility and economic affordability of hydrogen energy.

Figure 4 displays the average degree centrality of valued ties by the types of organizations. On the whole, the average degree centrality increased for all organizations. While the numbers were not significantly different from each other in the early years, the differences grew over time. Eventually, the average degree centrality of PROs and large firms rose to 12, while the corresponding number for universities and SMEs remained lower at around 4. This implies that there is a difference in terms of the division of labor between central and supporting actors. PROs and large firms occupied central positions, and these organizations have played key roles. The importance of PROs and large firms has been growing, and the strongest ties have been created between them. On the other hand, SMEs and universities were in the periphery, carrying out rather smaller R&D activities than PROs and large firms under sub-contracts.

In the early stages, both indexes showed a low value, but they soon turned higher as shown in Figure 5. High centralization indicates the existence of outstandingly-central actors, but high density implies that actors may not critically rely on others for interactive activities. Central actors did not perform all of the coordination, because non-central actors had direct ties to a certain extent in the early stages. The number of actors in the network was small, and their research was focused on scientific pursuits. Thus, this resulted in the formation of a small scientific research community.

However, there was one significant drop in density and centralization around 1998. We could find a clue to the drop from changes in government research programs: the entry of the automobile industry as the “Next-generation Vehicle Technology Development” program. It should be noted that this was not the first R&D project in the field, and there had already been basic research in the field since 1992 as we showed in Table 1. Nonetheless, this large program accelerated the entry of actors from the automobile sector.

The structure of the network began its reorganization after this drop. This can be found in the density and centralization figures in the later stages. After the new actors took their places, high centralization reappeared in a couple of years; however, density did not. The gap between centralization and density had increased over time, which indicates that while a number of new actors had joined in connection with Hyundai Motor Company, a major Korean automobile manufacturer, new linkages had been concentrated on a few actors. As this shows, a dense network may not be conducive to the coordination of heterogeneous goals and schemes across actors (Katz and Tushman, 1979). The central actors should bridge actors and mediate the coordination of tasks. Knowledge and information were passed to central actors and disseminated effectively to others who sought them. This was a hub-and-spoke structure with low density and high centralization. Such a hierarchical structure resulted from placing a priority on pursuing efficiency in the allocation and coordination of resources in order to avoid the risks of the premature solidification of networks when the scope of the network expands from a small community to a larger one.

When we focused on each of the two major programs to better understand the relationships amongst actors, we observed differentiated patterns in the expansion and structuration of the networks. The 21st Century Frontier R&D Program by MOST focused more explicitly on basic science, while the Hydrogen and Fuel Cell RD&D Program by MOCIE focused on commercialization and industrialization, which emphasises the importance of demonstrations and deployments. Although it is not appropriate to divide the hydrogen sector into two parts according to the singular shape of the network, two network visualizations were intentionally shown in Figure 6 to clearly demonstrate the difference in actors’ roles in the two programs. In the MOST network, PROs and universities are seen in central locations, while in the other, the MOCIE network, large firms are. Shown in Table 2, the participants’ share of universities is larger as 42% and 17% of PROs in the MOST network, than those of 32% and 11%, respectively, in the MOCIE network, with an inverted proportion of firms. It is again clarified by the centrality analysis, since the MOST network has the highest centrality of PROs as 8.2, and an identically high figure of centrality for large firms in the MOCIE network. The higher efficiency index of 0.65 for the MOST network than that of MOCIE’s 0.58 denotes the MOST network has a form of network that is more open, which can be beneficial in seeking and combining new knowledge.

There are 27 shared actors between the two networks out of a total of 116 actors. Those 27 actors that participate in both programs have an average efficiency index of 0.65, which is much higher than that of non-shared actors at 0.49. This explains why the shared actors such as the Korea Institute of Energy Research and Hyundai Motor Company link not only to two programs but also to other actors by playing bridging roles.

Structuration affected the integrative process in the network as well. Figure 7 illustrates that the ratio of isolated actors to the total number of actors significantly decreased from 80% to 0% with the progressive incorporation of these actors in the knowledge network. This indicates that the importance of central actors grew over time, while peripheral actors often tended to make relations with them, either through direct ties or via third parties, since the resources and information that could be gained by tied actors are critically beneficial. As a result, the knowledge network became more tightly integrated.

The government did not require actors to establish partnerships for every project; instead, the government organized them into R&D programs. With this rather subtle intervention, the formation of a huge, single network cluster with 129 participants is remarkable.

4.2 Sector emergence process

The establishment of relationships is a way of enabling actors to integrate the knowledge and complementarities that they do not possess (Nelson, 1995; Edquist, 1997; Robinson et al. 2007). A new knowledge-base underlying an emerging sector can be created through a process in which heterogeneous actors bring different types of knowledge and competencies, whereupon they are subsequently integrated (Carroll et al., 1996; Malerba and Orsenigo, 2000; Malerba, 2002, 2004). It implies that the knowledge base of an emerging sector results from inter-sectoral interactions and evolutionary processes. Most ideas and designs that shed light on the technological barriers that arise in emerging sectors are derived from current relevant knowledge domains. Hence, the dynamics of the knowledge base and their boundaries are incremental and compatible with existing sectors, rather than punctuated and disruptive.

This process can be examined in analyzing critical actors and their roles and changes. Figures 2a–c and 3 show the diversity in the types of actors, which is clearly evident in the Korean case. In addition, Table 3 shows a list of major participating firms that are sorted by the category of sector and other central actors with degree centrality for 1989–2005 in which the first entry of each sector is shown.

In the early stage, actors from the public R&D sector and universities played key roles in forming the knowledge base. While PROs developed intermediate technologies by themselves, they also played roles as network hubs, where knowledge in the network flows through. PROs also organize sub-contracted projects under the main programs. Though the ratio of universities decreased over time as shown in Figure 3, a small number of universities (e.g. Seoul National University) have been key players since 1989. They have disseminated scientific knowledge that they have generated and obtained from the international scientific community.

A number of traditional sectors became associated with the hydrogen energy field. One such sector was the conventional energy sector—oil, gas, and electricity—represented by firms such as the Korea Electric Power Company and GS-Caltex. Another was the mechanical sector, represented by firms such as Hyosung and Samsung Engineering. Relatively recently, the importance of the automobile sector has been increasing, with participating firms such as Hyundai Motor Company. The range of sectors involved has expanded with time. As Patel and Pavitt (1997) proposed, the profile of technological competencies should not be restricted to one field within one specific sector. Rather, several relevant knowledge domains should be integrated to underpin the formation of a new sector. This indicates that the knowledge base of an emerging sector does not form instantaneously, but gradually over time.

The observed hydrogen and fuel-cell technology-specific entrants, most of which were entrepreneurial start-ups, appear in 2001. GS Fuel Cell was the first specialized firm, diversified from GS Caltex, while other start-ups were not spin-offs of existing firms. In other words, companies that target emerging sectors appear, which can be seen as a phenomenon that further reinforces the knowledge base for the emerging sector (Carroll et al., 1996). The involvement of technology-specific diversifying entrants and start-ups indicates the level of maturity of the knowledge base in the hydrogen energy field, and also indicates the beginning of a new sector formation.

With the generation and accumulation of sectoral knowledge, opportunities become eventually more accessible to firms that undertook mainly commercial activities (McKelvey, 1997). Thus, the research focus gradually shifted from basic to applied research, and this change was observable in Figure 3. Universities accounted for 60% of actors in the early years, but they fell to only 33% in 2005. Instead, universities started to strengthen their role of training. For example, Yonsei University, Chonbuk National University, and others established hydrogen-related departments and courses in their schools. In contrast, the ratio of commercial, for-profit firms, including large firms and SMEs, grew from 30% in 1989 to 53% in 2005. This pattern is also in line with the gap between average centrality in Figure 4.

The accumulation of sectoral knowledge provides novel insights in dealing with problems, which can, in turn, be considered technological opportunities. Along with increasing actors, innovation performance has been strengthened in the field of hydrogen energy in Korea. The number of Science Citation Index (SCI) publications from Korean researchers on hydrogen generation, storage, and fuel cells rapidly grew from only six in 1991 to 235 publications in 2005. The number of US patent applications from Korean organizations related to fuel-cell technologies increased from 8 in 2001 to 14 in 2002, 23 in 2003, 48 in 2004, and 120 in 2005, while the total number of fuel-cell patent applications in the US patent office increased from 630 in 2001 to 894 in 2002, 1161 in 2003, 1324 in 2004, and 1376 in 2005 (source: USPTO). The increase ratio of the number of applications from Korea was significantly larger, which demonstrates the rapid catching-up of the Korean hydrogen energy sector.

The legitimating process within existing institutions that Aldrich and Fiol (1994) and Nelson (1994) termed “gaining legitimacy” is important when speaking about the sectoral institutions in an emerging sector. Legitimation proceeds, and common notions and recognition in emerging sectors spread out among actors (Aldrich and Fiol, 1994). The founding of cooperative alliances, trade associations, scientific societies, and other network bodies creates critical mass, stimulates actors in setting high expectations, and accelerates the general public’s acceptance of the emerging technologies. In this context, the expansion and structuration of networks is a part of a process in which sectoral institutions are legitimated, and sometimes the bandwagon effect can be understood as an indicator of cognition across stakeholders with shared vision. An examination of national institutions can help to understand the legitimation process, since national institutions may shape sectoral institutions in a country (Malerba, 2002, 2004). Briefly, Korea’s national institutions are often demonstrated by the strong authorities and the concentrated capabilities in large firms (Amsden and Hikino, 1994; Kim, 1997, 1998). This can provide an explanation for the observed pattern of the entry of actors in the Korean case. PROs, universities, and large firms were the major actors at the outset; however the number of SMEs rose only after 2002. This delayed entry of SMEs may be because the capabilities of SMEs are relatively insufficient in comparison with those of large firms. Thus, SMEs started participating after the risks and uncertainties had been somewhat reduced.

In summary, we have identified the building blocks of the hydrogen energy sector by visualizing its knowledge network between 1989 and 2005. Knowledge networks could be regarded as precursors to emerging sectors before the market and the value chain are formed, and we could derive insights into emerging sectors by examining the following patterns of the evolution of knowledge networks (Nelson, 1994; Dosi et al., 1997; Malerba, 2006). First, a knowledge network evolves in a centralizing manner. The importance of key actors, such as PROs and large firms, grows over time. These actors are located at the centre of the network and develop strong relationships with other actors. Second, the pattern of evolution is integrative. That implies that actors who are isolated in their activities in the early years subsequently connect with other actors, thereby expanding the knowledge network. Third, the formation of the knowledge base is neither disruptive nor discontinuous. The sectoral knowledge base of an emerging sector is based not only on a new scientific field, but also on existing ones. In the course of this evolution, PROs link and mediate among heterogeneous actors, develop intermediate technologies, and, being located at the centre, serve as a hub for knowledge flow and reservoir.

5. Discussion

5.1 Building blocks of an emerging sector: Knowledge network as a precursor

In analyzing the case of the hydrogen energy sector in Korea, this study aims to show how a new sector emerges. It is widely known that all contemporary developments of emerging technologies cannot be led by a single organization, but can be interpreted as the result of the collective and coordinated activities of a variety of organizations. The hydrogen energy case study also shows that interactions are at the core of the emergence of new sectors.

This study shows that all of the principal building blocks were initiated and identified in the knowledge network, even before the appearance of the value chain. The network began to form with the entry of heterogeneous actors such as PROs, universities, and firms. They triggered the establishment and expansion of the network in the hydrogen energy field. In addition, other government agencies and financial institutions constituted another type of network as mentioned previously through the Science Park, for example.

We have also identified institutions in the analysis. We have shown that legitimation and recognition among stakeholders resulted from the expansion of the knowledge network. However, sectoral institutions in the early stages were not limited to legitimation regarding the hydrogen energy. There were institutional components at different levels from legislation to consensus: the Alternative Energy Technology Development Promotion Act in 1987, the Hydrogen Economy Master Plan in 2005, and the vision and norms that are shared in Korean government programs.

Korea’s hydrogen R&D network showed a single, large cluster of actors, although there have been two major R&D programs with building blocks which are subtly different from each other. The overall singularity of the network reflects the established knowledge base of the hydrogen energy sector in Korea, and that the actors share the basis of the knowledge base together. In terms of sectoral institutions, Korea’s national innovation system provides the hydrogen sector with institutional components of the sectoral system of innovation, such as legislation, funding systems, PROs, government policies, and cultural aspects. The actors and their networks are represented by a clear visualization of the network with various actors, evolving throughout 17 years of the time-line.

If the two major R&D programs are intentionally segregated, some delicate differences arise in the building blocks between the two networks. First, the composition of the actors is slightly different in the two programs. A significant network of PROs and universities was found in the MOST network, while a larger share of firms was observed in the MOCIE network. By analyzing the network structures in centrality and efficiency, it was possible to identify which organizations played the bridging roles. In both programs, PROs and large firms were the bridging organizations. However, the bridging role of large firms is more significant in the MOCIE program. The MOST program network was based more on basic science than the MOCIE, which was based more on applied science and engineering. The cause of this difference can be explained by the different institutions, the third building block. The MOCIE program has focused on the RD&D of hydrogen energy for relatively short-term commercialization and industrial sector formation, while the MOST program aimed for long-term capability building and scientific research. It remains for future investigations that a different sectoral system of innovation and national system of innovation will provide us with different findings to enrich our understanding of the sector emergence over sectors and countries, since the building blocks can be varied quite largely from each other. However, it is still convincing that the sectoral system of innovation approach and the building blocks analysis will provide us with an appropriate framework in analyzing emerging sectors.

As a knowledge network is shaped, the evolutionary patterns of the network in the emerging sector become clear. The knowledge network grows due to the integrative activities of heterogeneous actors. Traditional sectors become involved in the network not disruptively, but with incremental changes. In this analysis, the hydrogen energy sector could be considered a science-based sector characterized by a variety of technology sources and the involvement of many large-sized innovative firms per Pavitt’s taxonomy (1984). Knowledge networks will evolve by external forces that are coupled with knowledge and institutional factors (Orsenigo et al., 1998, 2001; Pammolli and Riccaboni, 2002; Koka et al., 2006). In the end, knowledge links are likely to evolve to the value chain. For example, when POSCO Power started commercial electricity production with fuel cells in 2007, it established a business relationship with former R&D connections such as Korea Electric Power Co. and POSCO Machinery and Engineering.

In short, the knowledge network has all the three major building blocks of a sectoral system of innovation: actors and networks, knowledge bases, and sectoral institutions, together with the identifiable characteristics of each building block. The evolutionary pattern of the knowledge network shows significant similarities to that of industrial sectors. In addition, knowledge networks have high potential of transforming into value chains and business relationships. Hence, it is possible to regard knowledge networks as precursors to emerging sectors. This idea can be one of the major contributing aspects of this article to the existing literature. Our perspective that focuses on knowledge networks is effective in identifying the emergence of new sectors, showing how to find the building blocks of emerging sectors with more articulated forms. In the earlier stage of a sector emergence, the actors and their network can be found as a constitution of the actors of various existing sectors and the R&D network, respectively. The role of government and public research bodies need to be highlighted in this stage. In addition, the evolution of an R&D network also articulates the formation and development of a sectoral knowledge base, since the R&D activities should be based on relating scientific and technological knowledge. The sectoral institutions are less visualisable; however, we can assume there should be, for example, ways of collaboration among the actors, specified policies and legislations, the management of RD&D programs, and business relationships—some of which we mentioned in this article.

5.2 The role of government as a network organizer in the emergence of a sector

With respect to the evolutionary process, even novel technologies occasionally encounter failure. Given such an uncertain environment where even state-of-the-art technologies can fail, the role of governments and STI policies in the early stages can be critical in mitigating risks and uncertainties, which provides us with the added value of understanding the role of government in the emergence of new sectors, since not much SSI literature has shed light on this.

According to Spencer et al. (2005), the modes of government intervention can be varied across countries, such as “social corporatist,” “state corporatist,” “state nation,” and “liberal pluralist.” There exist a number of policy options in rather direct or indirect terms, for example, taxing, subsidising, giving tax benefits, and procurements. Especially for emerging sectoral systems of innovation, R&D funding and catalyzing the network formation can be complemented to those, which are proposed by the findings of this article. Korea mainly utilizes diffusion-oriented policies that are aimed at coordinating links between the private and public sectors as a group of the “state corporatism” countries that are characterized by highly centralized bureaucratic governments and social groups that have priority over individuals. This proposal is confirmed by this study, given that Korean government R&D programs drove the knowledge network of the hydrogen energy sector. We have recognized that heterogeneous actors coordinate the network and organize the negotiation process in which they have opportunities to change their interests and preferences, as well as the directions of technological development via government programs as Callon (1995) explained. This has been concerned with two points.

First, government programs help avoid the risk of knowledge network solidification. Actors can manage the trajectories of development and make flexible networks through the appropriate allocation of funding resources. In particular, PROs help steer the process from the top of the hierarchical network. Second, the government can maximize the effects of their policies by managing the timing of public intervention. The Korean government provided a hydrogen energy sector with initiatives to address energy security and climate change issues when other countries began raising those issues. Governments can give signals for the timing of the involvement by existing sectors, by organizing new programs in a planned manner. In addition, governments can manage the programs flexibly over time in accordance with the changing environment.

It has been reported that the Korean government tends to concentrate available resources on those emerging sectors that it deems to hold promise for future economic growth (Kim, 1997, 1998). The government-driven knowledge network in the hydrogen energy sector provides us with a good example in which an evidence for this traditional argument is still valid. This finding may be interesting not only to developing countries that benchmark Korea’s economic development, but also to developed countries that want to nurture emerging sectors, especially in the field of sustainable energy.

Although, historically, precursors to an emerging sector were not necessarily based on the formation of a knowledge network, this research demonstrated the pre-birth stage of a new sector in the field of the latest technologies, providing policy implications at least for the current situation. There remains a need for further research to generalize this argument, and such generalization requires examinations of cases involving different sectors and countries.

Acknowledgements

We thank Ben Martin, Ed Steinmueller and Malcolm Eames for their encouragement to very early version presentations, Franco Malerba for his attention to the initial idea of this paper during 2007 Globelics Academy. The authors are grateful for the comments received from two anonymous referees, which greatly contributed to improve this paper.

References

Appendix A