专利保险与未来产业技术发展——来自专利文本语义识别的微观证据

Abstract

Against the backdrop of intensifying global technology competition, whether the patent insurance pilot policy can effectively hedge firms' risks in disruptive technological innovation, facilitate forward-looking deployment in future industries, and cultivate first-mover advantages has become an important question demanding urgent answers. Drawing on national tax survey data and the full sample of invention patent applications, this paper employs a pre-trained language model together with a self-constructed future industry technology lexicon to measure firms' future industrial technological development at the firm-year level. Exploiting the characteristic fact that local governments provide patent insurance premium subsidies exclusively to locally registered firms, this paper delineates the administrative boundaries of the 23 patent insurance pilot regions using ArcGIS and constructs a Difference-in-Discontinuity (RD-DID) design to identify the causal effects of the 2012 patent insurance pilot policy. The empirical results show that the patent insurance pilot policy significantly elevated the level of future industrial technological development among firms near the boundaries of pilot regions, exhibiting pronounced geographic discontinuity characteristics. Mechanism analysis reveals that the policy first improves the regional patent litigation risk environment as a necessary precondition, and subsequently operates through three firm-level channels: deepening R&D collaboration, facilitating knowledge flows, and alleviating financing constraints. Heterogeneity analysis shows that the policy effects are more pronounced among large-scale firms, regions with weaker intellectual property protection environments, patent-intensive industries, and high-technology manufacturing sectors. Further decomposition of the policy effects indicates that the promotional effects stem from genuine increases in firms' overall innovation output rather than the reallocation of R&D resources across different fields.