Abstract:    At present, China′s economy is at a critical stage of transformation, and manufacturing enterprises are gradually shifting from arbitrage type to innovation. The success or failure of this transformation will directly affect whether China′s economy can successfully cross the middle-income trap. However, in recent years, with the high profit of financial investment and the decline of manufacturing profits, Chinese manufacturing enterprises have continuously broken through the main business and tried to make financial investment, displaying that the manufacturing industry has shown a trend of deepening financialization. Driven by the profit-seeking of capital, a large number of manufacturing industry capital began to flood into the lucrative financial and real estate industries. The result is that the allocation of financial assets in the balance sheet has become a very common phenomenon, which has caused the entity enterprises to deviate more and more from its main business. As mentioned above, the phenomenon that the manufacturing enterprises tend to be hollowed out is also called "entity enterprises financialization". When financial capital occupies a dominant position, industrial capital will lose its dominant power in the value chain, and then it is very likely to fall into the vicious circle of falling profits and corporate recession. Most of macro-level research results also provide negative opinions on financialization, while there are different conclusions (promote or inhibit) on the impact of financialization in micro-level research. Therefore, this paper takes the A-share listed companies in China′s Shanghai and Shenzhen Stock Exchanges as the research objects, and uses the nonparametric quantile-regression model to quantitatively study the nonlinear heterogeneous impact of the financialization of manufacturing enterprises on R&D investment. 
    Theoretically, this paper refers to the concept of nonlinear heterogeneity and combines the theoretical conduction mechanism analysis, describing the nonlinear effects and the heterogeneous effects of the degree of financialization on the R&D investment of manufacturing enterprises, thus revealing in depth the differences in impact patterns caused by financialization of the enterprises at different levels of innovation capability (corresponding to the different quantiles). 
   And the results are as follows: (1) The impact of financialization on corporate R&D investment has "inverted U-shaped" nonlinear characteristics, and there is a moderate level of financialization to promote corporate R&D investment maximization. (2) There are significantly heterogeneous effects of financialization in different quantiles, indicating that the degree of financialization has different impact models on corporate R&D investment. With the increase of the quantiles, the impact models of financialization on enterprises R&D investment change from the simple inhibition effect to the "inverted U-type" relationship which means that financialization will promote corporate R&D investment firstly, and then inhibit corporate R&D investment. In addition, the sensitivity of the R&D investment to the financialization is reduced firstly and then increased. (3) Distinguishing research subjects according to the system of ownership, this paper finds that state-owned enterprises have smaller financing constraints and are less sensitive to financialization, while non-state-owned enterprises have the opposite direction, that is, non-state-owned enterprises have greater financing constraints and are more sensitive to financialization. Through research in different industries, the result shows that financialization has a stronger promotion effect on the R&D investment of high-tech enterprises which represent the "inverted U-shaped" feature at all three quantiles compared to other industries. (4) Non-state-owned enterprises and high-tech enterprises are more sensitive to financialization in high quantiles, and the level of optimal financialization is increasing with the increase of quantiles. For companies located at low quintiles, R&D investment is more likely to be restrained by financialization, suggesting that companies lacking innovative inputs are more motivated to use financial assets for arbitrage. The research in this paper shows that Chinese enterprises should not blindly pursue the degree of financialization. In contrast, enterprises should choose the degree of financialization that is compatible with their own innovation capabilities. (5) This paper also calculates the degree of financialization corresponding to the apex of "inverted U-shaped", in order to provide decision-making basis for enterprises which aim to increase R&D investment by selecting a reasonable level of financialization. 
   Based on the results of this paper, we propose the following suggestions: (1) Chinese government should strengthen financial supervision and guidance to improve the financial system including measures like formulate relevant policies to actively create a good financial ecological environment, curb financial speculation, encourage resources to flow to R&D and production, and give full play to the "reservoir" function of financial assets. At the same time, it is necessary to strengthen the tracking and monitoring of the direction and field of currency flow in order to guard against the phenomenon that enterprises put focus on arbitrage activities. In the process of implementing financialization, enterprises must choose moderate degree according to their own innovation ability and innovation level. Conversely, companies should not pursue financialization blindly in case it will cause more serious inhibitory effects; (2) It is the duty of Chinese government to guide enterprises to correctly view financial behavior, rationally allocate financial assets according to the company′s own business status. Also, the government should help enterprises avoid short-sighted behavior of operators and focus on the long-term development of enterprises and the long-term interests of stakeholders; (3) Increase the strategically structural adjustment of state-owned enterprises as soon as possible, continuously improve internal governance, and formulate scientific and reasonable salary incentive system with the purpose of absorbing innovative talents, and then promote the innovative spirit of state-owned enterprises. (4) For one thing, reduce the market entry threshold and reduce financing constraints for private enterprises with entrepreneurial spirit and strong sense of innovation. For another, cultivate enterprises with high-tech content and considerable market potential, restrict the expansion of financial assets of companies with low technology content and weak innovation awareness. Meanwhile, enterprises with high energy consumption and low technology should be eliminated gradually. (5) Accelerate the adjustment of industrial structure, increase support for high-tech industries, and improve the innovation capability and production efficiency of entity enterprises. Reduce the gap between the profit of the entity and the return on financial assets, thus guiding the company to refocus on the main business.

Key words: financialization, R&D investment, nonlinear and heterogenous effect, nonparametric quantile panel