Abstract:
China's patenting activity has surged over the past two decades, yet questions remain about the quality and sources of innovation. We develop a new method to measure the importance of a patent for innovation, based on the use of a Large Language Model to process patent text data and a new theory of the innovation process. We apply this method to study the evolution of patenting in China from 1985 until recently, and also classify patent ownership using a comprehensive business registry. Our method and data yield several novel facts about Chinese patenting. Among these are that the patents which are important for innovation have become less important on average; that knowledge within China has become more important than knowledge outside of China for directing innovation in China; and that knowledge produced by Chinese entities has been more important than knowledge produced by foreign entities in China. These findings have implications for China's growth trajectory and reflect both global trends in the decline of innovativeness and potential effects of domestic policy.