فعالیت های تحقیق و توسعه جهانی شرکت های چند ملیتی ژاپنی در آمریکا: روش مثلث بندی

We examined 79 Japanese MNCs’ R&D subsidiaries in the US from the knowledge-based view. We found: (1) subsidiaries’ R&D strategies generally encouraged knowledge flows; (2) subsidiaries’ R&D alliances promoted knowledge flows; (3) R&D subsidiaries with process-oriented incentives promoted vertical knowledge flows; (3) autonomous R&D subsidiaries promoted knowledge flows from the local environments to the subsidiary; (4) R&D subsidiaries with a high level of knowledge flows accumulated a high level of knowledge; and (5) R&D subsidiaries with a high level of accumulated knowledge achieved high overall performance. Our interviews with 30 R&D subsidiaries and 10 parent companies supplement these findings.

Traditionally R&D activities in multi-national corporations (MNCs) were centralized and concentrated in a home country, mainly because of supply-side reasons such as possible scale economies (Vernon, 1966 and De Meyer, 1993) or higher appropriability of R&D efforts (Teece, 1987 and Grandstrand et al., 1993). However, MNCs have been rapidly globalizing their R&D activities, especially for the last two decades. A National Science Foundation study shows that US companies’ investment in overseas R&D has increased three times faster than company funded domestic R&D over the last 10 years—10.1% versus 3.4%. Overseas R&D now accounts for 12% of the US industry's domestic R&D expenditures (NSF, 2004). The largest US (and European) R&D spenders are even more concentrated in foreign locations (>33%) (Roberts, 2001). JETRO (2003b) also reports that the number of overseas R&D subsidiaries by Japanese MNCs has been increasing approximately 10% annually for the past decade.3 “Demand-side” factors seem to be replacing the supply-side factors as influences. Namely, in order to better serve diversified foreign demands, R&D activities must be decentralized (Grandstrand et al., 1993 and Iwata, 1994). Kurokawa et al. (1999), Roberts (2001), and Hirschey and Caves (1981) report a positive relationship between foreign R&D activities and export shares on the foreign market. Although these supply and demand-side considerations shed light on recent MNCs’ overall tendencies, such factors do not suggest how to efficiently organize global R&D activities. In an effort to fill such a gap in our understanding, this study investigates global knowledge flows in Japanese MNCs in the US, from the knowledge-based view (KBV) of the firm ( Grant, 1996 and Kogut and Zander, 1993). This paper attempts to answer the following five research questions: (1) why knowledge flows are important in managing global R&D subsidiaries?; (2) what factors determine knowledge flows among R&D subsidiaries, headquarters (HQ), and other subsidiaries?; (3) what factors determine knowledge accumulations of R&D subsidiaries?; (4) what factors determine performance of R&D subsidiaries?; (5) how should R&D subsidiaries manage knowledge flows for higher performance? In order to answer these questions, this paper has the following four major sections. First, the paper discusses global R&D activities by surveying the relevant literature. Second, it constructs hypotheses from KBV of MNCs. Third, it tests these hypotheses by using our survey data from 79 Japanese R&D subsidiaries in the US, supplemented by 30 interviews with these subsidiaries and 10 interviews with their parent companies. Finally, it concludes with managerial implications and observations for further research.

6.1. Contributions Our theoretical contribution to the study of KBV of MNCs is to substantiate the knowledge-based view of MNCs by our three-step model. Namely, first, we identified factors affecting knowledge flows. Then, we examined the extent to which knowledge flows lead to knowledge accumulations. Finally, we examined the extent to which knowledge accumulations lead to subsidiaries’ overall performance. Our data generally finds that high levels of knowledge flows lead to a high level of knowledge accumulations, and a high level of knowledge accumulations lead to high performance. However, we also find that a high level of knowledge flows occasionally does not lead to high levels of knowledge accumulations and performance, supporting our assumption that knowledge flows are cost-involving activities. For example, an R&D subsidiary might be able to achieve high performance without frequent knowledge flows from HQ, because of its highly experienced top management who have already accumulated enough knowledge for running the subsidiary. Our main methodological contribution to the study of MNCs from KBV is our use of the triangular approach. By combining survey methods with both structured and unstructured interviews, our study sheds light on the “black box” of R&D subsidiaries of Japanese MNCs. Other methodological contributions are to measure: (1) subsidiaries’ R&D strategies, (2) both technology- and market-related knowledge flows, (3) both vertical and horizontal knowledge flows, (4) both knowledge in-flows and out-flows, and (4) the effect of knowledge flows on knowledge accumulations and overall performance of subsidiaries. By such methods, we find that R&D strategies generally determine levels of knowledge flows, especially market-related and horizontal knowledge flows. We also find that R&D strategies tend to cancel the summed effects of the technology-related knowledge in-flows and out-flows between HQ and subsidiaries. Our study has determined that market-related and vertical knowledge flows tend to contribute more to the overall performance than do technology-related and horizontal knowledge flows. Furthermore, organizational factors appear to be more important for promoting knowledge flows than strategic factors supporting our view that trustful and democratic environments are needed for generating new meanings and innovations within MNCs. Finally, the previous studies on KBV of MNCs found that autonomous R&D subsidiaries promote knowledge flows (Gupta and Govindarajan, 2000) and that network-linked (i.e., HQ or other subsidiaries-linked) incentives stimulate knowledge flows (Minbaeva et al., 2003 and Foss and Pedersen, 2003). Thus, our finding that process-oriented incentives promote knowledge flows can also be unique.62 6.2. Managerial implications This study has the following implications for managers in charge of global R&D activities. First, we found that the defined control variables—e.g., communications medium, co-location, M&A, and experience—generally do not affect knowledge flows, knowledge accumulations, or subsidiaries’ overall performance. Thus, our data analyses suggest that knowledge flows are likely to be determined by managerial factors (i.e., strategic factors and organizational factors) more than contextual factors (i.e., control variables). Such findings suggest: (1) purposes and missions of R&D strategies should be clearly defined before launching global R&D sites, (2) process-oriented incentive practices rather than outcome-oriented incentives should be emphasized, and (3) subsidiaries should be allowed an appropriate level of autonomy for higher knowledge flow and performance. Second, knowledge flows among HQs, subsidiaries, other subsidiaries and local environments should generally be encouraged, because the amount of knowledge flows generally have positive associations with knowledge accumulations. Especially, market-related and sub–local knowledge flows should be emphasized, because these flows indicate positive associations with both knowledge accumulations and subsidiaries’ overall performance. However, managers should be cautious about the disadvantages (and costs) of knowledge flows. We found in our survey data that there are optimal levels of knowledge flows for knowledge accumulations. We also found in our interview data that too much knowledge flow from HQ to a subsidiary tends to lower subsidiaries’ performance. Third, R&D alliance activities generally should be encouraged at overseas R&D subsidiaries. We found that such alliance activities tend to promote knowledge flows, especially technical knowledge flows from local environments to subsidiaries. We also found that such alliance activities tend to lead to qualitative knowledge accumulations. Finally, the amount of specific knowledge accumulations—e.g., number of papers, patents, copyrights, and new products/services—should not be overemphasized. Such knowledge accumulations are not related to subsidiaries’ overall performance, and vary across different industries. 6.3. Limitations and directions for future research The findings in this paper must be interpreted with care, because of the following six limitations. First, although we indicated that Japanese R&D subsidiaries in the US represent the largest population, the largest population does not necessarily represent the average population. Thus, the external validity of this study could be severely limited. Second, Eisenhardt and Santos (2002) argue that knowledge transfer within an organization depends upon the characteristics of that knowledge, the sender, the recipient, and their mutual relationships. In this paper, however, we did not measure characteristics of knowledge, in terms of tacitness (Subramaniam and Venkatraman, 2001), “observability” or “system embeddedness” (Birkinshaw et al., 2002). Third, although we discussed the importance of “interpretative capacity”—i.e., how to interpret and process knowledge flows, and how to share knowledge with others—we did not measure such a capacity. Based on our interviews, we have a strong impression that successful knowledge-creating MNCs tend to have special and unique interpretative capacities (or biases). For example, one interviewed MNC, Honda, emphasizes product managers’ creative and intuitive concepts for new cars, occasionally disregarding marketing data, such as marketing surveys from potential customers. On the other hand, another interviewed MNC, Toyota, seems to emphasize marketing data rather than product managers’ new ideas. Fourth, since every MNC is a network, all intra-MNC knowledge transfers take place in the context of the network (Ghoshal and Bartlett, 1990). As contrasted with “dyadic” or “systemic” approaches to the examination of network-related phenomena (Gupta and Govindarajan, 2000), we focused mainly on the “nodal” level in order to simplify the phenomena. However, such simplification has disadvantages. Fifth, we did not examine knowledge flows within MNCs from a historical or long-term perspective. Zander and Kogut (1995) report that the median time to transfer knowledge was five years in their sample. However, our study examined neither the long-term effects of knowledge flows on subsidiary performance, nor the long-term effects of the subsidiary's performance on knowledge flows (i.e., the reverse effect). For example, our interviews with the parent companies reveal that managers at R&D subsidiaries tend to overestimate their performance. Finally, we did not investigate the joint (i.e., interaction) effects among defined strategic and organizational factors, mainly because of the limited number of cases in our sample. There could be some better “configurations” among these factors which lead to higher knowledge flows, higher knowledge accumulations, or higher performance. Accordingly, the opportunity for further research into the KBV of MNCs seems to be great. There are several promising directions for future research. First, investigation of MNC's interpretative capacities must be worthwhile future research. Such investigations should focus on human resource management (HRM)-related and R&D management-related issues, such as training and development activities for R&D personnel and expatriates in HQs, as well as R&D management practices in HQs and subsidiaries. A second line of productive inquiry would be to examine further alliance activities at subsidiaries. We investigated R&D alliance activities at a subsidiary's level by a simple method—i.e., asking managers about their attitudes toward alliances. More thorough investigation should be done on subsidiaries’ alliance activities, for example, by historical tracing of such activities and associating with subsidiaries’ long-term knowledge accumulation, as well as their contributions to HQs. Finally, although Japanese R&D facilities in the US represent a significant portion of global R&D subsidiaries, further research on, for example, German or Korean R&D facilities in the US or Europe, as well as Japanese R&D subsidiaries in Asia or Europe, will also broaden our understanding of this subject. Without such research, it is uncertain whether our findings are specific to the US R&D subsidiaries of Japanese MNCs in the current study or if they represent MNCs in general.