Why Do Multinational Corporations Locate Their Advanced R&D Centres in Beijing?

Abstract

Theories of the globalisation of innovation assume that multinational corporations (MNCs) distribute their innovation activities hierarchically, with advanced technology being confined to the advanced industrialised countries, while more routine low-end innovation is decentralised in a few developing countries. The emergence of about 40 research and development (R&D) centres in Beijing, China, many of which engage in basic and advanced applied research, challenges the above assumption. This article argues that the cheap and abundant highly skilled labour of the latecomer countries is an essential factor in attracting global R&D activities but that this factor is far from being a sufficient condition for the presence there of advanced R&D activities. Through its analysis of the historical transformation of local institutions and of their co-development with MNCs, this paper identifies four major knowledge assets that explains why Beijing could attract advanced R&D activities. First, Beijing has developed a strong entrepreneurial culture that creates highly motivated engineers who are eager to learn new knowledge from abroad. Second, the experienced Chinese returnees provide a critical bridging role between Western R&D management knowledge and local engineer culture. Third, the lack of inter-firm trust and networks makes the entrance of MNCs into a ‘loose’ cluster much easier. Fourth, the large and dynamic Chinese market that desires high-tech products with low prices shortens the product life cycle, forcing MNCs to upgrade their R&D facilities in China. The findings show that the co-development of local institutions with the MNC R&D centres can create locational windows of opportunity for advanced R&D activities to be carried out in unconventional sites outside the Triad countries. This article concludes with the discussion on how Dunning's Ownership, Location and Internalisation (OLI) framework and Mathew's Linkage, Leverage and Learning (LLL) framework might be useful in explaining this new phenomenon.