科技金融政策、资金网络与企业创新绩效——基于潜在狄利克雷分布模型

Science and technology finance policy, capital network and enterprise innovation performance：Subject recognition based on LDA

摘要:

政策是科技金融的灵魂，多种类型的科技金融政策可以全方位提升企业创新绩效，借助资金网络的效应使得绩效进一步释放。本文基于2011-2020年国家层面科技金融政策文件，运用LDA算法进行聚类分析，将科技金融政策划分为宏观导向类、资金支持类、平台建设类和专利保护类，深入挖掘各类政策对企业创新绩效的提升效果及资金网络在其中发挥的调节效应。研究发现，四类科技金融政策对企业创新绩效均具有显著提升作用，资金网络中心度增强了宏观导向类、资金支持类和专利保护类三类科技金融政策对企业创新绩效的提升作用。进而从政策、网络与企业三方出发，提出对策建议。

Abstract:

Policy is the soul of science and technology finance. Various types of sci-tech finance policies can improve enterprise innovation performance in an all-round way, and further release the performance with the help of capital network effect. Based on the national sci-tech financial policy documents from 2011 to 2020, this paper uses Latent Dirichlet Allocation for cluster analysis, divides the science and technology financial policy into macro orientation, capital support, platform construction and patent protection, and deeply excavates the promotion effect of various policies on enterprise innovation performance and the regulatory effect of financial network. The results show that the four types of sci-tech financial policies can significantly improve the innovation performance of enterprises, and the capital network centrality enhances the promotion effect of three types of science and technology financial policies on the innovation performance of enterprises: macro orientation, capital support and patent protection. Then from the policy, network and enterprise three aspects, put forward countermeasures and suggestions.