At present, China’s S&T innovation is ushering in a period of major opportunities, and AI and big data technology are deeply integrated with other fields, so the breakthrough of major S&T innovation requires a deep combination of government and innovation subjects. However, S&T innovation policies of various departments in China are relatively scattered, and fragmented investment of S&T resources and distributed introduction of policy measures are relatively common, which is difficult to form an innovation force. Therefore, research on the effectiveness of comprehensive policy governance can provide certain enlightenment for policy formulation, and is also of great significance for enriching innovation theories.

This paper expounds the mechanism of policy collaboration affecting innovation activities, uses the text mining method to measure regional policy collaboration based on S&T policy documents of 31 provinces in the past 20 years, and conducts an empirical test on this mechanism. It is found that the collaboration of policy subjects and policy measures can promote innovation activities, mainly through enhancing the incentive effect of market vitality on innovation and reducing the inhibitory effect of rent-seeking behavior on innovation. S&T policy collaboration can also significantly improve the conversion efficiency of innovation investment. However, in regions with severe policy homogeneity, the promotion effect of policy collaboration on innovation activities will be weakened. Therefore, the government needs to improve the efficient cooperation between central and local policy subjects and strengthen the coordinated use of policy measures, so as to effectively promote the development of S&T innovation; and local governments need to make policies based on local actual development needs.

The marginal contributions of this paper are as follows: First, it extracts policy collaboration indicators from S&T policy documents using the text element extraction method, and considers the cumulative effect of policies decreasing over time in the measurement, making the measurement more reasonable. Second, the study on the mechanism of policy collaboration affecting innovation activities extends the depth of research on the impact of policies on S&T innovation. Third, it has certain reference value for the government to formulate macro S&T policies.