Knowledge sourcing by multi-locational firms in Chinese cities: Mining and integrating local knowledge stocks in private and state-owned firms

Abstract

A linked firm-patent dataset is generated for a sample of approximately 400 domestic, multi-locational firms operating in China over the period 2001-15. Analysis focuses on the character of technologies produced within the different R&D plants of these firms. We show that most multi-unit firms produce different kinds of knowledge within their R&D units and that the likelihood of technology differences increases with the number of R&D plants within each firm. Evidence of a geography of knowledge sourcing is presented that rests on the technological similarity between the patents produced within a firm's R&D plants and those produced in the cities where those plants are located. We also find that the average complexity of patents generated by firms is positively related to the number of R&D units that they operate. That these benefits emerge for multi-locational firms operating within a single country suggests that the constraints on knowledge diffusion operate at a relatively disaggregate spatial scale. Finally, our results indicate that collaboration between headquarters and non-headquarters R&D units within firms has a differential impact on the complexity of knowledge produced and that state versus private ownership of firms influences the nature and impact of intra-firm collaboration.