Abstract: China′s new energy vehicle industry subsidy policy aimed at encouraging enterprises to develop global competitiveness through R&D and innovation, but it has merely accelerated the rapid expansion of industrial scale in the short term while the enthusiasm for R&D investment is seriously insufficient. The new energy vehicle industry is facing risks such as technological disruptive innovation expectations and market demand uncertainty. It is difficult to ensure the normal development of the industry with the "invisible hand" of the market alone. China′s new energy vehicle subsidy policy, while stimulating the rapid increase of the industrial scale, has also caused enterprises to invest in low-tech projects, and the industry faces a structural imbalance. Under the background of the verification of new energy vehicles′ fraudulent behavior, it is very important to evaluate the effects of subsidy policies from the perspective of policy transmission and provide theoretical basis for policy optimization. The effects of subsidy policies have been controversial. One view is that the positive incentive effect of the subsidy policy is obvious, which promotes investment. Frequently subsidized companies will increase R&D investment, thereby promoting corporate R&D behavior, and subsidized companies often have stronger innovation output. Another view is that although industrial policies such as subsidies have a stimulating effect on industrial development, this effect is achieved by expanding supply and stimulating demand, rather than R&D and innovation; Moderate subsidies will stimulate manufacturers′ R&D investment, but it will have a crowding effect when it exceeds a certain amount. At the same time, due to the existence of many transaction costs, the subsidy policy is difficult to effectively promote the technological upgrading of the new energy automobile industry, and to a certain extent, it even has a restraining effect. These studies focus on the effects of subsidies and provide enlightening empirical evidence for the selection and formulation of policy tools. However, few studies involve the penetration process and focus of policy tools, so it is difficult to determine the drawbacks of the current new energy vehicle industry policy from the mechanism. Recent studies have recognized that the effect of industrial policy is not only related to the type of policy tools, but also to the mechanism of the influence of policy tools on the behavior of innovation subjects. This study uses empirical research to compare whether scale or R&D innovation plays a significant intermediary role in the transfer of subsidy policies to corporate performance, and in-depth reveals the mechanism of subsidies for China′s new energy automobile industry at the enterprise level. Methodologically, this article uses the content analysis method and Bootstrap mediation effect model to combine text analysis with empirical evidence. First of all, an in-depth text analysis of the content of the policy was conducted to obtain the direction and mode of action of the policy. Furthermore, this study compares the mediation paths of enterprise′s scale and patent behavior in the relationship between subsidy policy and enterprise performance. The design of the text part introduces Rothwell and Zegveld′s three-level division of policy tools (demand side, environment side, and supply side), and refines it based on industry characteristics. At the same time, combining the key events and industrial characteristics in the development of the new energy vehicle industry, the four major links of the innovation value chain are introduced, namely research and development, production, sales and use. The logic structure of the analysis revolves around the transmission of policy tools on the new energy vehicle innovation value chain, and finally forms a two-dimensional policy analysis framework for the new energy vehicle industry. The empirical study of the intermediary relationship uses the Bootstrap method proposed by Preacher and Hayes. Bootstrap does not require a distribution hypothesis, which prevents the coefficient product test from violating the distribution hypothesis. And it does not rely on standard error, thus avoiding inconsistencies in the results of different standard error formulas. The simulation study found that Bootstrapping has higher statistical power than other methods, so it is currently the ideal test for mediation effects.The text analysis results show: There are policy disadvantages in the investment of science and technology in the policy tools and the R&D link in the innovation value chain. From a policy point of view, the content of the policy clause itself affects industrial scale more than product development. Based on this, we draw the conclusion that the policy effect is biased in the transmission process. In order to further verify this, we use the mediation model for further analysis. The empirical results show that public subsidies have significantly promoted the improvement of the enterprises′ financial performance, scale and patent output. Moreover, the size of the enterprise and the patenting behavior play significant mediating roles in the policy effect on the enterprise′s performance, and the enterprise scale exerts stronger effect than the patenting behavior. Therefore, new subsidy policies such as dual-credit policy should take into account factors such as institution, subsidy tools and industrial characteristics in a comprehensive way to avoid policy deviation in the "government-enterprise interface". The innovation of this article is reflected in: (1) theoretically, using the expansion of enterprises and R&D innovation behavior as an intermediary, and analyzing the phenomenon and mechanism of policy action point deviation through text content analysis, further deepening the transmission mechanism of policy effectiveness. (2) In terms of methods, the policy text analysis and empirical analysis are combined, and the content transmission method and Bootstrap mediation effect model are used to analyze the policy transmission path, and the research logic is more complete. (3) In terms of practice, this paper proposes that industrial policies need to be implemented in the area of corporate R&D promotion.

Key words: new energy vehicle, patent behavior, content analysis method, bootstrap method