Faculty & Research. Working Paper Series

Transcription

1 Faculty & Research Towards a Metanational Research and Development (R&D) Organization: Imagination, Reality or the Future? An Exploratory Study by Y. Helble L. Chong and A. De Meyer 2003/78/ABCM/TM Working Paper Series

2 INSEAD Working Paper Title: Towards a Metanational Research and Development (R&D) Organization: Imagination, Reality or the Future? An Exploratory Study by Yvonne Helble, MBA/MA/MIM, Doctoral Student at the University of St. Gallen in Switzerland and Visiting Research Associate at INSEAD. Prof. Dr. Li Choy Chong, Professor of International Management and Director of the Asia Research Center at the University of St. Gallen in Switzerland. Prof. Arnoud De Meyer, Deputy Dean of INSEAD, Dean of Administration, Akzo Nobel Fellow in Strategic Management and Professor of Technology Management and Asian Business. 1

3 Abstract This paper applies the concept of the metanational organization to the domain of international R&D and postulates the characteristics of a metanational R&D organization. While Doz, Santos and Williamson (2001) illustrated in detail the metanational organization in a general firm context, this concept has not yet been examined for R&D organizations. Metanational R&D organizations are characterized as R&D organizations which are capable of optimally leveraging the technological hierarchy and which use a large number of knowledge bases to their advantage. This means that metanational R&D organizations leverage different technological stages within their R&D organization based in many critical knowledge clusters. Based on an exploratory study of 71 interviews with 51 R&D subsidiaries of multinational firms in Singapore, using a standard interview guide, this paper examines different models of international R&D organizations in comparison to the metanational R&D organization. In metanational terms, Singapore is seen as periphery. Only few firms conduct research in Singapore, most R&D subsidiaries focus is on development. With Singapore s planned transition into a knowledge-based economy a shift towards more research is essential. This would also imply a change in the perception of the periphery and higher technological sophistication creating a new critical knowledge cluster, a necessary condition for R&D organizations to also tap into the periphery and thus to become metanational. Given this situation, Singapore is one of the better locations to identify metanational R&D organizations in the making. Our findings indicate that (1) the concept of the metanational organization applies to only a limited extent to R&D organizations, (2) that different international R&D organizations have different performance implications on a subsidiary level and (3) that the notion of the metanational R&D organization is certainly not mere imagination, even if it is reality for only 2

4 a few R&D organizations today. It could, however, possibly be a new organizational model of international R&D organizations in the future. Key words: international R&D, metanational organization, Singapore. 3

5 1. Introduction The field of international R&D management and as such of different international R&D models was a marginal subject for research up to the 1980s (Granstrand, 1999; Gerybadze and Reger, 1999; Ronstadt, 1977, 1978; Behrman and Fischer, 1980). Academic interest in this topic started in the early 1990s (De Meyer and Mizushima, 1989; De Meyer, 1993). Up to that point in time, international R&D management has only attracted peripheral and passing attention in the literature (Granstrand, Hakanson and Sjölander, 1993). One of the reasons for the relatively late development in theory and practice of international R&D management was a widely held assumption that corporate R&D activities needed to be centralized. It was argued that the protection of firm-specific technology was only possible by centralizing corporate R&D activities. Furthermore, it was believed that only through such centralization could economies of scale, and as such the necessary efficiency in corporate R&D, be reached Another important reason for centralized corporate R&D is historical: R&D facilities were traditionally established close to headquarters (Gerpott, 1990). However, with the internationalization of other firm s activities such as sales and production, the internationalization of corporate R&D followed suit, seemingly overthrowing the long-held belief in the centralization of corporate R&D (Gerybadze and Reger, 1999). A study by Kuemmerle (1999) of 32 multinational firms in 5 countries shows that R&D efforts abroad by these firms increased from 6,2% in 1965 to 25,8% in While European firms are at the forefront of this internationalization process, US firms and Japanese firms seem to have less internationalized their R&D activities compared to their European counterparts. European multinational firms operating in small countries have particular high shares of foreign corporate R&D. Firms based in Belgium and the Netherlands, for instance, perform more of their R&D activities outside the home country than inside it (Granstrand, Hakanson and Sjölander, 1993). Swiss firms spend more than 50% of their R&D expenditure abroad (Boutellier, Kloth and Bodmer, 1996). Foreign shares of R&D for firms in larger European countries such as Sweden and the UK vary between 23 to 42% (Granstrand, Hakanson, and Sjölander, 1993; Granstrand, O. 1999). These figures contrast with foreign R&D activities of 10%-12% by US firms (von Zedwitz, 1999). Japanese firms are late, but fast internationalizers regarding their R&D activities, increasing their foreign share of R&D from 1,58% in 1987 to 5% in 1991 (Granstrand, 1999). 4

6 Large multinational firms are the drivers behind this internationalization process. Declining communication costs facilitating the exchange between R&D locations, rising R&D costs due to shortened product life cycles, global competition and an increasing complexity of technologies are the main reasons for this development (Brockhoff, 1998; Gassmann and von Zedtwitz, 1998). The increasing share of R&D activities conducted outside the home country raises numerous research questions, for instance on the R&D internationalization process and on international R&D management, most notably on different international R&D models. This paper is dedicated to the latter research question, namely to further develop the field of international R&D organization. More specifically, this paper attempts to examine the concept of the metanational organization as described by Doz, Santos and Williamson (2001) in an R&D domain. Furthermore, we examine whether different international R&D organizations result in differential R&D performance. This paper proceeds as follows. First, the literature review is presented. Based on the review of traditional models of R&D organizations, as given in the literature, and based on the concept of the metanational organization, this paper develops a model for a metanational R&D organization in a second section. In a third part, the sample, data sources, and analysis techniques are explained and results presented. A final section concludes. 2. Literature Review Since the concept of the metanational organization is new, its application to the R&D context has not been discussed so far. Linked only in a general context, previous literature examined the determinants for R&D internationalization (De Meyer 1993; Florida, 1997; Kuemmerle, 1999; Le Bas and Sierra, 2002; Patel and Vega, 1999; Ronstadt, 1978). The different determinants indicate why R&D organizations conduct R&D abroad. 5

7 Patel and Vega (1999) and Le Bas and Sierra (2002), for instance, examine patterns of technological activities outside their home countries. Both authors indicate that adapting products and processes to suit foreign markets and to provide technical support are still major factors for R&D internationalization, which means that R&D subsidiaries exist to extend abroad the firm-specific advantage of the parent firm (Meyer-Krahmer and Reger, 1998; Kuemmerle, 1999). This implies that R&D knowledge created in the home base is transferred and applied to the host country s market or in other words the home base R&D knowledge is exploited by applying it to the host market. In contrast to these market driven determinants, a second stream of literature emphasizes technology driven determinants for R&D internationalization. Florida (1997) analyzes foreign direct investment into R&D in the US by foreign-affiliated R&D laboratories, and as such the R&D internationalization into the US by European multinationals. Kuemmerle (1999) focuses on two industrial sectors, namely electronics and pharmaceutical, and examines laboratory sites by 32 multinational firms domiciled in five countries. Both authors agree that the major determinant for R&D internationalization is more technology driven than market driven; that is multinational firms seek to harness external new scientific and technological capabilities abroad. The knowledge of the home base is therefore augmented by knowledge in the host country. This technology driven determinant is in contrast to the market driven determinant as identified in the studies by Patel and Vega (1999) and by Le Bas and Sierra (2002), as discussed above. Overall, the literature suggests that firms in many industries decreasingly undertake overseas R&D activities as a mere support function for existing sales or production operations (Bartlett and Ghoshal, 1991; Belderbos and Iwasa, 1999). The competitiveness of firms is becoming increasingly dependent on their ability to establish a presence at an increasing number of locations in order to access new scientific capabilities (Edler, Meyer-Krahmer and Reger 2002, Kuemmerle, 1996 and Shan and Song, 1997). This development is particularly 6

8 important in the context of increasing R&D costs, fiercer foreign and domestic competition, intensifying product differentiation, and market globalization (Mutinelli and Piscitello, 1998). The following figure summarizes the main determinants of R&D internationalization. Figure 1: Major determinants of R&D internationalization Technological Advantage of Home Country Strong Home-Base Exploiting Home-Base Augmenting Weak Market Seeking Technology Seeking Weak Strong Technological Advantage of Host Country Source: Kuemmerle, 1996; LeBas and Sierra, Analysis of the literature on the determinants of R&D internationalization (see figure 1) shows most notably the distinction between home and host country base. In effect, the literature is characterized by a dichotomous nature. Depending on the home country s strengths or weaknesses either the home base is exploited or augmented. A home country s strengths or weaknesses refer to the technological advantage or disadvantage of the home base. 7

9 It is argued here that the determinant home base exploiting is logically also a market seeking determinant, since the home base is exploited by adapting products and processes to the new market and hence the market is sought after. This market-seeking determinant applies even if the home country is not strong in a particular technical field. Existing products of the parent firm are adapted to a local market even though these products may not be of a high level of technological sophistication. The same holds true for the technology-seeking determinant, which at the same time is also a home base augmenting determinant. The home base is augmented by seeking new technologies in the host country, meaning investment in cross-border R&D takes place in order to acquire knowledge from overseas locations with a specific scientific or technological competence. The review of the literature shows that the determinants of R&D internationalization have been the subject of a variety of studies. While the literature suggests that the determinants of R&D internationalization are more technology than market driven, the emphasis on home versus host country is still evident. What is missing in the literature, however, is a more holistic view on the determinants for R&D internationalization. Such a holistic view can be found in the idea of an international network of R&D laboratories (De Meyer, 1993). Such an international network R&D organization is considered crucial for the creation and diffusion of both internal as well as external know-how (De Meyer, 1993). This network idea is further developed in the concept of the metanational organization, which is explained in the next section. 3. Conceptual Framework 3.1 The Metanational Organization According to Doz et al. (1997) and Doz, Santos and Williamson (2001), MNCs (multinational companies) face two major challenges today in maintaining their competitiveness. First, transferring knowledge created in the home base to the host country is 8

10 insufficient for creating sustained competitive advantage. In today s competitive environment, the international exploitation of strategic advantages based on the home base is no longer a valuable strategy. This is the case because new critical locations of technology and sophisticated customer demand emerge in non-traditional geographical areas. For example, initially manufacturing sites were established in low cost locations. Over time, such manufacturing sites can, however, develop their own abilities to create new technologies. Semiconductor manufacturing plants in South East Asia, for instance, have made innovations in the technologies for the packaging of semiconductors (Doz et al., 1997). The challenge lies therefore in overcoming the dichotomy of home versus host country base (see figure 1). This is called the location challenge. The second challenge comes from the fact that MNCs face increasingly more and more complex knowledge. Knowledge originates from more dispersed parts of the world; products and customers require more complex knowledge combinations (Doz et al., 1997). Firms so far have been able to manage simple knowledge, but are required to manage increasingly complex knowledge. This challenge is particularly important in an R&D context where knowledge is of a highly complex nature, especially in research. It is assumed that the higher the degree of complexity of knowledge, the higher the resulting sustainable competitive advantage (Doz et al., 1997). According to the metanational framework, a metanational advantage is necessary when these challenges become relevant for MNCs. Three distinct capabilities are required to build such a metanational advantage, namely sensing, mobilizing ever more complex knowledge and putting it into operations. In other words, these MNCs need to create sensing nodes to detect critical knowledge, mobilize this specific knowledge and then connect and apply this knowledge (Doz, Santos, and Williamson, 2001). This in turn implies that MNCs need to augment their capabilities to leverage highly complex and dispersed knowledge to create competitive advantage (for more details see Doz, Santos and Williamson, 2001). 9

11 The term meta stands for beyond. According to Doz, Santos and Williamson (2001), this term has been chosen because metanational firms do not draw their competitive advantage from their home country or from various host countries, a dichotomy which is overcome in the metanational context. Metanational organizations view the world in its entirety with pockets of different, specialist knowledge. In order to examine how the metanational framework applies to R&D, the next section is devoted to the more traditional analysis of international R&D organizations. This will enable us to evaluate how these models cope with the two challenges we have mentioned. Afterwards, we will elaborate on what the metanational R&D organization could be. 3.2 Traditional International Models of R&D Organizations A study by Kuemmerle (1999) of 32 multinational firms in 5 countries, for instance, shows that R&D efforts abroad by these firms increased from 6,2% in 1965 to 25,8% in As a result of this development and a growing literature on international R&D management, different models of international R&D management have been identified and are depicted and illustrated below: 10

12 Figure 2: Different International R&D Organizational Models Decentralized R&D Polycentric, decentralized R&D Integrated R&D Network Distribution of internal competences and knowledge bases Centralized R&D Ethnocentric, centralized R&D R&D Hubmodel Geocentric, centralized R&D Competition Cooperation Degree of Cooperation between R&D Locations Source: Gassmann (1997); Gassmann and von Zedtwitz (1999) In the ethnocentric, the most traditional model, all corporate R&D activities are centrally located at headquarters. As a result, no transnational R&D processes take place. Such a centralized approach is based on the assumption that corporate R&D must be centralized and implies that only such R&D activities allow economies of scale and synergies to be realized and unintended technology transfer to be avoided. On the other hand, given the primacy of the home base in this model and hence the transfer of home base knowledge to host countries, this organizational form may lead to inadequate local market adaptations. 11

13 Further competitive disadvantages occur through the organizational rigidity of such a concept and the omission of foreign technological inputs since knowledge is created in the home base only. Obviously, the described challenges, namely the location and knowledge complexity challenges, cannot be addressed in this model. As opposed to the first concept, the second model is international in nature and adopts a geocentric, centralized approach to R&D. The physical R&D location is located in the home based, but international R&D knowledge is acquired through intense collaboration with local manufacturing, suppliers and lead customers. As a result of this, this international R&D structure is sensitive to local markets and can benefit from foreign technological trends. The international awareness of the R&D personnel is fostered as well, since researchers are exposed to international markets. This model might be inadequate if an R&D organization becomes increasingly international. The centralized character might hinder a critical presence in diverse R&D locations. Overall, this R&D model is more suited to address the identified challenges of location and knowledge complexity in comparison to the ethnocentric model. The primacy of the home base is reduced and critical knowledge is seized through various international collaborations. The third model is the polycentric approach, where various R&D locations in different foreign markets prevail without centralized control. These R&D locations have mostly been established as a result of local distribution facilities and manufacturing plants. Consequently, the decrease in the primacy of the home base assures a high degree of local sensitivity and the utilization of local resources. Due to the strong decentralized character of this R&D organization and its lack of coordination mechanisms, parallel R&D efforts might result in duplication and thus in major costs for a MNC. Knowledge from dispersed locations is seized, but might not be integrated and leveraged. Opportunities for innovation through combination are foregone in this model. 12

14 The fourth model, the Hub-Model, is basically an in-between model, which has on the one hand a central R&D location like the ethnocentric model, but has on the other hand dispersed R&D locations. The central R&D location is leading in most technological fields. On the dimension of the degree of cooperation, it is situated between cooperation and competition. Due to coordination of R&D efforts, a high degree of efficiency is achieved and a suboptimal resource allocation avoided. The resulting high coordination costs of this model might be due to the lack of effective knowledge management. The last model is that of the integrated network. Within this concept, the central R&D location loses its overall dominance. All R&D locations are equally important, play a strategic role and interact through multiple and complex coordination mechanisms. Each R&D location focuses on one specific product, component or technology area. Consequently, the home versus host country dichotomy loses its significance. Each R&D location focuses on one specific technological field. The obvious benefits of this model are the realization of specialization, learning and synergy effects, and the utilization of local competencies. On the other hand, however, high coordination costs can occur because of the complexity of the model. Furthermore, the decision making process is made more difficult due to the equal status of the respective international R&D locations. Given these characteristics, it is obvious that this model is the most advanced model with regard to tackling the challenges of location and knowledge complexity. While this division into five basic international R&D models proves an overview of international R&D models, it is not an all-encompassing classification. Not all R&D activities of international companies can be clearly attributed to one of these models. Furthermore, the differentiation between the different models is not always completely clear-cut; the hubmodel in particular as an in-between model does not seem to be fully defined. However, this classification of international R&D models constitutes a contribution to visualizing different international R&D concepts and serves in enhancing our understanding of international R&D 13

15 activities. Furthermore, this classification shows where current R&D organizations are lacking in capabilities in terms of the identified challenges of location and knowledge complexity. This paper attempts to expand this framework by proposing a metanational R&D organization, which assumedly is capable of addressing the identified challenges. 3.3 The Metanational R&D Organization How the two identified challenges can be overcome under this model is analyzed in the following. Location challenge: Large number of knowledge bases In accordance with the metanational framework, the international R&D organization is viewed in its entirety. Thus, overall R&D allocation is optimized and integrated accordingly. R&D organizations adopt a more comprehensive strategy, allocating their R&D activities anywhere in the world where specific knowledge is available. This knowledge does not only refer to internal, but also to external knowledge bases in contrast to the traditional R&D models, where only internal knowledge bases are considered (see figure 2). Internal knowledge bases refer to knowledge bases within the R&D organization, whereas external knowledge bases refer to knowledge bases residing outside of the R&D organizational boundaries, namely in the external research environment. The following example attempts to illustrate this comprehensive approach: While Bangalore (in India) has specific knowledge in software development, Ang Mo Kio (in Singapore) has specific expertise in the development of electronic-related products and Shanghai (in China) has a strong manufacturing base. Hence, an R&D organization would conduct its software development activities in Bangalore, its electronic development activities in Ang Mo Kio, and an R&D unit with manufacturing support in Shanghai and thus take advantage of the technological hierarchy within R&D (see figure 4). Following such a metanational strategy, the R&D organization is no longer concerned with exploiting or 14

16 augmenting their knowledge base form a home country perspective. The corporate R&D organization is more concerned with tapping into specialized knowledge centers anywhere in the world and integrating this technological expertise to the fullest for the R&D organization, resulting in a highly specialist R&D organization. Hence, the R&D site abroad is not determined from a home base perspective, but is determined by a holistic perspective (where it is best to allocate these R&D activities based on the global landscape of the R&D organization). The international R&D organization acts as a global scanner picking up and exploiting new technology wherever it evolves (Zander, 1998). Hence, previous literature is extended by including this determinant, which will be referred to in the following as nondichotomous (ND) determinant of R&D and which is depicted below. Figure 3: Determinants of R&D internationalization Technological Advantage of Home Country Strong Home-Base Exploiting Home-Base Augmenting Weak Market Seeking Non-dichotomous determinant (globally optimizing) Technology Seeking Weak Strong Technological Advantage of Host Country Source: Authors extending Kuemmerle (1996) and Le Bas and Sierra (2002) 15

17 This non-dichotomous determinant is presumably the starting point of a metanational R&D organization for tapping into different knowledge bases world-wide. As noted above, a metanational organization needs the capability to sense, mobilize and integrate knowledge bases world-wide, regardless of home or host country base. It is assumed here that sensing different knowledge bases world-wide would require MNCs to increase their degree of R&D internationalization. Otherwise, if R&D activities are restricted to a low number of knowledge bases, it will be difficult to sense dispersed knowledge. This difficulty applies to a great extent to R&D since R&D is mostly focused on the triad nations (namely US, Europe and to some extent Japan). The periphery has so far been neglected, especially because it is considered that due to lower value added activities there, R&D organizations do not need to consider the periphery. But examples prove that through constant technological upgrading and developments in their science and technology policy (also see section 4), the periphery can contribute to the whole R&D organization and should not be neglected. This negative perception of the periphery also renders the mobilizing of different knowledge bases difficult. The integration of knowledge refers again mostly to the integration of knowledge residing in the triad nations. A metanational R&D organization, however, would take advantage of a large number of knowledge bases (internal and external knowledge bases) including the periphery (see dimension of knowledge bases on the x-axis in figure 5). Knowledge Complexity: Leveraging of the technological hierarchy The second dimension of the proposed metanational R&D organization (see figure 5) refers to leveraging the technological hierarchy, which refers to all technological levels depicted in the technology ladder in figure 4, this can be R&D knowledge with regard to market support, manufacturing support, development or research. As stated in the above mentioned example, this R&D knowledge can reside in development expertise in South East Asia, in R&D knowledge related to manufacturing in East Asia and research in Europe. It is 16

18 essential to sense this knowledge within the technological hierarchy (see figure 4). The next step is to mobilize this knowledge. It is important that during this step the periphery is not ignored and that all relevant R&D knowledge regardless of its technological level is considered, especially because different R&D functions require different technological capabilities (see figure 4). To integrate this knowledge is the third component in the metanational advantage and requires the capability to not only integrate the different knowledge bases, but also to integrate operation, improvement and generation capabilities in a fruitful way to create competitive advantage, since different technological stages require different management skills. This second dimension is shown in figure 4 below and depicted on the y-axis in figure 5: This knowledge complexity challenge is especially important in an R&D domain where knowledge is of a highly complex nature, especially in research Figure 4: Leveraging of Technological Hierarchy in the Metanational R&D Organization R&D headquarters Sense and mobilize different knowledge bases in the technological hierarchy R&D subsidiary in host country Technological Hierarchy: Pure Science Unit (S) Basic Research Unit (R2) Applied Research Unit (R1) Exploratory Development Unit (D2) Advanced Development Unit (D1) Manufacturing Support Unit (M2) Market Support Unit (M1) Other R&D subsidiaries Integration Capabilities Generation Capabilities Improvement Capabilities Operation Capabilities Source: Authors extension of Medcof, 1997; Costa and de Queiroz, 2002; Amsden and Tschang,

19 After describing the different dimensions which are part of the metanational R&D organization, the proposed framework for the metanational R&D organization is visualized below. Furthermore, the positioning of the metanational R&D organization is compared to the traditional international R&D models. Figure 5: Proposed Model for the Metanational R&D Organization high Geocentric R&D organization Metanational R&D organization Integrated R&D network Leveraging of Technological Hierarchy R&D Hub Model low Ethnocentric R&D organization Polycentric R&D organization few Number of Knowledge Bases many Source: Authors As can be seen from figure 5 and based on the discussion of the traditional international R&D models, the metanational R&D organization is an extension of the integrated R&D network. Within the model of the integrated R&D network, the home base loses its overall dominance, all R&D locations are equally important, play a strategic role and interact through multiple and complex coordination mechanisms. As can be seen, the three 18

20 capabilities for a metanational advantage, namely sensing, mobilizing and integrating different knowledge bases, are partly apparent in this international R&D model. Hence, an integrated R&D network would have to reinforce these capabilities to reach the state of being a metanational R&D organization. The same holds true, but to a lesser degree for the R&D Hub Model, an in-between model as described above. The home base is more important than in the integrated R&D network. The central R&D location is leading in most technological fields. On the dimension of the degree of cooperation (see figure 2), it is situated between cooperation and competition. The importance of the home base should be increased and knowledge management skills increased. The ethnocentric R&D organization is furthest away from the metanational R&D organization, the primacy of the home base being critical. As opposed to this concept, the geocentric, centralized approach of R&D is more international in nature. The same applies to the polycentric R&D organization. Both the geocentric as well as the polycentric R&D organization are strong in one dimension of the metanational R&D organization (namely leveraging of the technological hierarchy for the geocentric model, and a large number of knowledge bases for the polycentric model), but lack another dimension. The geocentric R&D model can leverage the technological hierarchy because international R&D activities are coordinated. In contrast, the polycentric R&D organization can tap into different knowledge clusters, but is highly decentralized. Finally, the metanational R&D organization acts as global scanner picking up and integrating technologies in all relevant and emerging critical knowledge clusters. Thus, this proposed type of R&D organization takes advantage of a large number of knowledge bases and at the same time is highly effective in leveraging the technological hierarchy. The following table gives an overview of the differences of the traditional R&D models versus the metanational organization. 19

21 Table 1: Comparison of traditional R&D models versus the metanational R&D organization Dimensions of Conceptual Framework Traditional R&D models Metanational R&D organization Location Challenge: Number of Knowledge Bases Determinants of R&D internationalization Dichotomous in nature Comprehensive in nature Extension of R&D Only in the triad Triad nations and internationalization nations periphery Degree of specialization Low to medium High Knowledge Complexity Challenge: Leveraging of Techno-logical Hierarchy Knowledge development Critical R&D personnel at headquarters or at key R&D Development of critical R&D personnel also in the periphery subsidiaries Locus of critical knowledge At headquarters Anywhere in the (innovation) and/or at key R&D subsidiaries R&D organization Source of knowledge Mostly internal Internal and external Source: Authors As illustrated in the table, the metanational R&D organization is based on a comprehensive strategy dropping the home bases versus host country perspective, is present also in the periphery and is highly specialized (the different stages in the technological hierarchy are optimized). In contrast, the traditional international R&D models emphasize the home versus host base dichotomy (to a lower or higher degree depending on the R&D model). 20

22 Characteristically, the traditional R&D models are also restricted to the triad nations and are characterized by a lower degree of specialization. The metanational R&D organization also differs from the traditional models with regard to leveraging of the technological hierarchy. Critical knowledge development takes place in the whole R&D organization, the locus of knowledge is anywhere in the R&D organization and the source of knowledge is internal and external in nature. In contrast, traditional R&D models develop critical knowledge mostly at headquarters and at key R&D subsidiaries. As a result, the locus of knowledge is residing at headquarters and at key R&D subsidiaries and its source is mostly internal. This distinction between international R&D organizations and the metanational R&D organization applies to all international R&D organizations, regardless of their degree of similarity to the metanational R&D organization. After proposing this framework for the metanational R&D organization, we shall empirically analyze to what extent existing R&D organizations present in the periphery have evolved into a metanational organization. Before we do this, the context of the periphery in terms of the metanational R&D organization is discussed, in our case Singapore. We will analyze why Singapore is an adequate location to examine the concept of the metanational R&D organization. 4 Singapore s Science and Technology Policy The following outline of Singapore s science and technology policy serves as a basis to understand the periphery context in terms of the metanational framework. Singapore s industrial policy has been based on a technology leverage strategy from foreign MNEs since the beginning of Singapore s existence in These foreign MNEs based their manufacturing base in Singapore due to favorable incentive schemes by the Singapore government and the availability of low cost labor. This technology leverage 21

23 strategy was mostly based on manufacturing (Matthews, 1999). Emerging economies, such as China and Malaysia, have increasingly gained importance in manufacturing due to their lower manufacturing costs compared to those in Singapore. China in particular has attracted a large number of foreign producers. Singapore s recent economic policy thus aims at the growth of economic activities beyond manufacturing. This in turn implies that a strong emphasis is placed on fostering higher valued added activities, e.g. R&D activities. Singapore s recent science and technology policy is therefore based on building up R&D activities as they are conducted in the triad nations (US, Europe, and Japan; the internationalization of R&D is still confined to these triad nations, see Edler, Meyer-Krahmer and Reger, 2002). Since Singapore has a relatively short industrial history with a focus on manufacturing, it faces the challenges of a smaller R&D scale compared to the triad nations and a weaker technical capability. While R&D activities are to some extent conducted in the electronics industry, Singapore s most recent science and technology policy attempts to build up R&D activities in the biomedical sciences field also. Therefore, the most recent efforts of Singapore s science policy focus on the biomedical sciences. The R&D capabilities in this sector comprise the pharmaceuticals, medical devices, biotechnology, and healthcare services sector. As part of this strategy, the Singapore government has decided to build the biomedical sciences industry as the economy's fourth pillar, the other three pillars being electronics, chemicals, and engineering (Wess 2002). The two main government institutions, the Economic Development Board (EDB) and the Agency for Science, Technology and Research (A*Star), are in charge of Singapore s economic policy in all of these four fields. This illustration of Singapore s science and technology policy shows that Singapore, as the context of the periphery, is well suited to study the emergence of a metanational model because it has a long history of FDI by multinational firms and has also evolved from a 22

24 manufacturing base into a base where high value added activities, e.g. R&D activities, are carried out. Equally importantly, Singapore is still at the periphery, i.e. it is not part of the triad nations in terms of R&D, and has little tradition in carrying out R&D activities. 5 Methodology 5.1 Sample and Data Sources An exploratory study was conducted before starting the fieldwork in Singapore. This exploratory phase led to the development of a questionnaire and was also used for pre-testing the survey instruments and as a crosscheck against questionnaire responses, thereby improving internal validity and the interpretation of findings. As the basis for in-depth interviews with the respondents served this questionnaire. Besides this questionnaire, which allows inter-firm comparability, the researcher gathered firm-specific data and data on management issues the respondent (senior R&D managers, R&D directors and/or managing directors) considered highly relevant for managing the respective R&D subsidiary. This approach allows gaining standardized data by addressing a specific set of questions for each firm, but the researcher can also investigate important firm-specific issues, which enable the researcher to better understand the firm-specific context. Furthermore, due to the direct interaction between researcher and research subject, the researcher can immediately respond to the information given, can ask additional questions, can clarify doubts, and can gather supplemental information through observation (Emory and Cooper, 1991). This research method also allows the researcher to constantly improve the in-depth interview (e.g. questions can be more clearly phrased, interaction between respondent and researcher can be optimized). The answers obtained from the R&D managers and director at subsidiary level were complemented by corporate archival research (annual reports, corporate webpages) 23

25 documenting their type of R&D organization. This archival research also served to identify the number of different knowledge bases of the corporate R&D organizations. The unit of analysis is the firm specific R&D organization (R&D site or R&D department). As unit of analysis, the R&D subsidiary was chosen because it is relatively easy to identify, its size can be determined and an R&D subsidiary involves a more long-term commitment than for instance a sole research agreement (Kuemmerle, 1996). The unit of analysis is R&D subsidiaries of MNCs. This study examines multinational firms, since they are particularly active in establishing R&D sites abroad (von Zedtwitz and Gassmann, 2002). Overall, the researcher conducted 71 interviews with 51 R&D subsidiaries of MNCs, two research institutions (Singapore Institute of Manufacturing Technology and Institute of Bioengineering) and the two main government bodies (EDB and A*Star) in this crosssectional study. A specific set of R&D subsidiaries was revisited several times to gain more in-depth insights into the type of R&D organization. Both the managing directors as well as the R&D managers for the different R&D departments were interviewed. Out of the 51 R&D subsidiaries, 20 are American, 19 European, 10 Japanese and 2 R&D subsidiaries with other parent firm nationalities. Overall, 15 R&D subsidiaries belong to the biomedical sector, 13 to the electronics sector, 6 to the chemical sector, 9 to the information technology and communication sector, 4 to the engineering sector and 4 to other sectors (food and aviation sector). The industry classification is based on the National Survey of R&D in Singapore 2001, which is conducted annually by A*Star. According to this National Survey of R&D in Singapore 2001, there are 206 private firms (either wholly foreign owned or have less than 30% local ownership) conducting R&D in Singapore in 2001 in the above-mentioned industries. The sample of 51 subsidiaries therefore translates into a response rate of 25%. 24

26 5. 2 Operationalization of Major Variables Leveraging of technological hierarchy This paper attempts to classify the R&D subsidiaries under investigation according to the conceptual framework. This is done by analyzing the illustrated dimensions, namely leveraging of the technological hierarchy and number of knowledge bases. The leveraging of the technological hierarchy is measured according to the R&D subsidiaries relationships with other R&D subsidiaries and R&D headquarters. In the R&D organization three types of relationships (tie modality) have been distinguished (Vereecke, Van Dierconck and De Meyer, 2002). First, relationships of human resources refer to the human resources flow between the different R&D sites and with headquarters. More specifically, the human resources flow refers to the development of critical R&D personnel in the internal R&D organization. It shows if critical R&D personnel is also developed in the periphery. In the interviews, the R&D managers were asked about the composition of their R&D personnel (number of personnel at technician level versus research scientist/engineer level). Technician level refers to R&D personnel holding a bachelor s degree, whereas research scientist/engineer level refers to R&D personnel holding a master s degree or a PhD. Analyzing the composition of the R&D personnel gave an indication to what extent critical R&D personnel was developed in the periphery. This approach is similar to the reasoning of Deeds, DeCarolis and Coombs (2000) who state in another context that the number of R&D personnel as percentage of the management team can be an important performance indicator. Here it is not seen as a performance indicator, but as an indicator for critical human resources development at subsidiary level. Furthermore, the R&D managers were asked to what extent their R&D subsidiary has influence on the acquisition and development of human resources and to what extent they conduct training for their R&D personnel. Overall, these items attempt to reflect to what extent development of critical R&D personnel occurs in R&D subsidiaries in the periphery. 25

27 Second, we examined the innovation configuration to investigate in how far the locus of innovation is at the R&D subsidiary, at headquarters and/or at other R&D sites, namely whether R&D sources are at the subsidiary R&D site or whether core technologies are transferred for further development to the respective R&D subsidiary. This will give an indication of the primacy of the home base. In the interviews, respondents were asked to indicate to what extent they are able to participate actively in the global R&D program, to what extent they are recipients of core technologies from the home base and to what extent they can initiate own R&D projects. Moreover, the respondents indicated to what extent they conduct R&D activities in a field where headquarters or other R&D sites have no expertise and to what extent the innovation locus in their R&D organization is equally balanced. Overall, these items attempt to reflect the innovation configuration. Third, the degree of freedom the R&D subsidiary enjoys is examined. Respondents were asked to indicate to what extent their R&D site has to follow rules and regulations by headquarters, to what extent their R&D site can engage freely in external research collaborations and to what extent their R&D site can interact freely with other R&D sites. Overall, these three measures attempt to give an indication on leveraging the technological hierarchy and follow table Number of knowledge bases The number of knowledge bases is determined partly by the interviews and partly by corporate archival research. The answers obtained from R&D managers and directors at subsidiary level were complemented by corporate archival research (annual reports, corporate web pages) documenting their type of R&D organization and the number of different knowledge bases of the corporate R&D organizations. In general, the interview partners indicated the number of knowledge bases. Data on the number of knowledge bases was also found through corporate archival research. 26

28 The number of knowledge bases is considered small when the knowledge base is basically located at headquarters. The number is medium when the knowledge base is located at headquarters and a few key R&D subsidiaries. An R&D organization with a large number of knowledge bases has R&D locations in all critical knowledge clusters, also in nontraditional R&D locations. The following table summarizes the classification scheme for the various types of R&D organizations: Table 2: Classification Scheme for Different International R&D Organizations Type of R&D Leveraging of Technological Hierarchy organization Critical human Innovation Degree of Number of resources Locus also at freedom Knowledge Bases development at R&D subsidiary (autonomy) of R&D R&D subsidiary subsidiary Ethnocentric No No Low Small Geocentric No Yes 2 Low Small R&D Hub Yes Yes Medium Medium Model Integrated Yes Yes High Medium R&D network Polycentric Maybe Maybe High Medium Metanational Yes Yes High Large R&D organization Source: Authors 2 Even though the geocentric R&D organization does not have international R&D subsidiaries, it does engage in international R&D collaborations. Therefore, the innovation locus is not only at headquarters. 27

29 An R&D organization would be classified as an ethnocentric R&D organization if the number of knowledge bases is small, in this case only at headquarters. Logically, the leveraging of the technological hierarchy is low. Critical human resources development does not take place at the R&D subsidiary. The innovation locus is at the home base and hence not at the R&D subsidiary. Due to the primacy of the home base, the degree of freedom of the R&D subsidiary is low. The geocentric R&D organization is also present at headquarters only. But in this model, different international R&D collaborations are entered into. Therefore, the innovation locus is not entirely at headquarters. Thus, leveraging of technological hierarchy is higher than under the ethnocentric model. The R&D Hub model, the integrated R&D network as well as the polycentric R&D organization are all present in the triad nations. They differ with regard to their ability to leverage the technological hierarchy. By contrast, a metanational R&D organization is present in a large number of knowledge clusters and is very good at leveraging the technological hierarchy R&D Performance In order to analyze the performance implications of different international R&D organizations, different R&D performance measures are discussed in the following: From an organizational perspective a measure of R&D output refers to criticality, substitutability, interaction, and immediacy (Brockhoff, 1998). In this context, criticality reflects the degree to which the overall success of the organization depends on the work of the individual laboratory. The term criticality shows the possibility of substitution of work done in an R&D unit. The higher this variable, the lower the positive respective output of this R&D unit. Intense interaction of one R&D unit with other R&D units might increase the innovation output of the respective R&D unit. Finally, immediacy refers to the time lag between the 28

30 stoppage of work in a laboratory and the cessation of work in the whole organization, that is, to what extent a stoppage by the R&D unit constitutes a block for the rest of the organization. These concepts are relatively hard to measure due to their abstraction and are more useful in theoretical discussions than in an empirical study. Other R&D measures include its efficiency (to do things right) and its efficacy (to do the right things). Efficiency refers to the use of R&D resources in a favorable relation between costs and usefulness, whereby efficacy refers to R&D projects which contribute to the overall firm goals. In other words, low efficacy means a high share of money being wasted for R&D. The major disadvantage of these types of measurements is that these measures are exposed to considerable subjective evaluation by the R&D manager. This might lead to biased results. An alternative measure of R&D output is the number of drugs in development or in the pipeline. In general, one distinguishes between the number of products at stage I, II and III and the number of products approved for sale (Coombs and Deeds, 2000). The strength of a product pipeline assures future cash flows and increases the likelihood of firm survival. Hence, this measure of R&D output indicates the future potential value of a firm. Obviously, a product pipeline does not necessarily translate one for one into innovative products. Moderator variables, such as the firm s ability to effectively commercialize these products, can play a significant role. One possible solution to this problem would be to consider only the products in stage III of the pipeline since they have the highest probability of being translated into innovations. This measure, however, applies solely to certain industries such as the pharmaceutical industry. A further R&D performance measure refers to intellectual human capital, which can be measured as the number of Ph.D.s (master s degrees in sciences) as percentage of the management team (Deeds, DeCarolis and Coombs, 2000). This performance measure is 29