Knowledge Characteristics of Industrial Production Networks and the Quality of Corporate Innovation: Evidence Based on Machine Learning and Patent Big Data

Innovation quality is difficult to evaluate and study externally, yet it is critically important for economic actors to achieve sustainable development. Based on the market microstructure of industrial production networks, this paper characterizes and investigates how the knowledge characteristics of production networks inform corporate innovation quality. First, decision-making about corporate innovation quality is used to develop a mathematical model that explores how knowledge characteristics of production networks shape innovation quality. Second, big data, advanced machine learning models, and statistical methods are used to measure the innovation quality of Chinese listed corporate groups based on tens of millions of patent-related data points. Finally, Chinese production network data are used to confirm that heterogeneous knowledge in production networks promotes innovation quality, while homogeneous knowledge inhibits that quality. This effect also generates spillover effects on innovations that do not directly involve production network collaboration. The theoretical framework can be applied to green innovation decisions; the main effects ultimately drive increases in corporate returns on capital and production efficiency. This study has implications for governments wanting to drive innovation through market-oriented mechanisms and for firms wanting to make scientific decisions.