1、 Foreword
The report of the 19th CPC National Congress mentioned science and technology more than 10 times and emphasized innovation more than 50 times, pointing out that innovation is the first driving force to promote development. The government's policies to encourage enterprise innovation mainly include direct subsidies and indirect reductions. Direct subsidies (fiscal subsidies) directly stimulate enterprises' innovative behavior and reduce private costs in R & D investment; Indirect reduction (tax preference) indirectly encourages enterprises' innovative behavior and reduces the tax cost of enterprise income. However, the initial investment of enterprises in order to obtain policy support will also bring additional costs to R & D activities. Considering the early-stage investment of enterprises, it is necessary to choose which policy of financial subsidies and tax incentives can best promote enterprise innovation, analyze the extent to which these two policy tools can promote enterprise technological innovation and economic growth, and study whether one policy is more conducive to enterprise innovation than the other, and whether there is the "best combination" of the two policy tools. At present, China's fiscal and tax support for high-tech enterprises is large and the standards are relatively unified. Therefore, this paper selects high-tech enterprises as the research object, and provides empirical evidence for these problems by analyzing the relationship between financial subsidies, tax incentives, R & D investment and enterprise innovation performance.
2、 Theoretical analysis and research hypothesis
（1） Fiscal and tax policies: the difference between fiscal subsidies and tax preferences
Fiscal subsidies and tax incentives are policy tools for the state to stimulate enterprises to carry out innovative production. In terms of support methods, financial subsidies are direct cash assistance given by the government, and tax preferences are tax deductions deducted from tax obligations. In terms of support, both have reduced R & D costs, but financial subsidies are more deterministic. In terms of support timing, financial subsidies provide early-stage funds for R & D projects, and tax incentives provide post compensation for R & D projects. Zhang Honggang [1] believes that financial subsidies are direct economic subsidies and benefits given to products of a certain sector by policy makers who focus on a specific industry for a period of time. Tax preference is an indirect subsidy, which reduces the tax burden cost of enterprises through tax relief and other measures, and encourages enterprises to carry out innovative production.
（2） The relationship between financial subsidies, tax incentives and innovation performance
Due to the huge investment in the early stage of innovation, the private rate of return is usually far lower than the social contribution rate, so public subsidies are needed to encourage enterprises. Paunov[2] pointed out that manufacturing enterprises that received financial subsidies were less likely to stop innovative projects during the financial crisis. Howell [3] also found that financial subsidies have a significant impact on patent output when ranking energy R & D using the SBIR proposal. B? Ler et al. [4] studied the relationship between Norwegian tax policy and corporate profits and found that tax policy can stimulate the innovation performance of enterprises. In addition, Cappelen et al. [5] also believe that enterprises that obtain tax credits are prone to produce new production processes, which facilitate the development of new products and improve the innovation performance of enterprises. For the impact of both on innovation performance, foreman peck [6] found that small and medium-sized enterprises supported by the government in the UK grow faster, and the support of these two policies will increase innovation. Brub et al. [7] also found that Canadian enterprises that receive both tax credits and financial subsidies have created more new products than those that only use tax credits. Based on the research of scholars, this paper proposes hypothesis 1:
Hypothesis 1: fiscal subsidies and tax incentives have a significant impact on innovation performance.
（3） The relationship between financial subsidies, tax preferences and R & D investment
The impact of fiscal subsidies and tax incentives on innovation performance is an important topic discussed by scholars. Financial subsidies can increase the R & D investment of enterprises [8], but lee[9] believes that when the government increases financial subsidies, enterprises will choose to reduce their r&d investment and use the remaining funds to expand the inventory. Similar to financial subsidies, there are differences on the impact of tax incentives on R & D investment. Kobayashi[10] found that the tax credit will increase the R & D expenditure of small and medium-sized enterprises in Japan, and have a greater impact on enterprises with limited liquidity. However, cowling [11] does not believe that the tax credit will have a significant impact on the R & D expenditure of small and medium-sized enterprises in the UK and Spain. Some scholars also estimated and compared the impact of the two policies on R & D investment. Liu Guangqiang [12] believed that financial subsidies and tax incentives have different incentive effects in different industries and different enterprise environments. Haegeland et al. [13] found that the additional cost of tax credit is higher than that of financial subsidies when studying Norwegian enterprise data. According to the research of scholars, this paper puts forward hypothesis 2:
Hypothesis 2: financial subsidies and tax incentives have a significant impact on R & D investment.
（4） The relationship between R & D investment and innovation performance
In order to obtain core competitiveness, enterprises need to constantly carry out innovation. R & D innovation can enable enterprises to obtain new knowledge, new technology and new production capacity. On this basis, enterprises can produce new products and improve enterprise efficiency. Therefore, R & D is an important factor to promote technological progress, optimize production structure and maintain economic development. Li Lu et al. [14] and Qiu Yunjie et al. [15] found a significant positive relationship between R & D investment and innovation performance. Li Zhong et al. [16] found that R & D investment mainly improves the innovation performance level of enterprises by improving the R & D efficiency of enterprises. Based on this, this paper proposes hypothesis 3:
Hypothesis 3: R & D investment has a significant positive effect on innovation performance.
（5） Intermediary effect of R & D investment
Financial subsidies and tax incentives have a direct causal relationship with R & D investment. Whether they promote or hinder R & D investment, R & D investment always has a significant positive effect on innovation performance. Theoretically, enterprises that enjoy financial subsidies and tax incentives will pay more attention to innovation activities, and enterprises will correspondingly increase R & D expenditure, which will affect innovation performance. Although financial subsidies and tax incentives are compensatory measures for the government to the innovation behavior of enterprises, their incentive effect on innovation activities may be indirect, and the specific impact needs to be achieved through the intermediary variable R & D investment. Qiu Dong et al. [17] subdivided R & D investment into government investment and enterprise investment, and studied the relationship between them and innovation performance respectively. The results show that enterprise investment plays a complete intermediary role between government investment and innovation performance. Based on this, this paper proposes Hypothesis 4: Hypothesis 4: R & D investment has an intermediary effect between financial subsidies, tax incentives and innovation performance.
3、 Data and methods
（1） Data source
The government's support for high-tech enterprises is large and the standards are relatively unified. At the same time, high-tech enterprises also pay more attention to innovation activities, so this paper takes the gem high-tech enterprises as the research sample. Excluding st enterprises and enterprises that did not continuously disclose data from 2014 to 2016, a total of 255 listed companies were obtained. The financial data comes from cninfo. Com, and the patent application data is obtained from the data of the national patent office.
（2） Variable selection
1. Explained variable
The measurement methods of innovation performance: first, the number of patent applications, authorizations and effective numbers of enterprises; Second, the sales revenue of new products and the sales profit margin of new products. Because the information related to the sales revenue and profits of new products has not been clearly disclosed, so the relevant data cannot be obtained. Therefore, this paper selects the patent output of the enterprise in the current year as the explanatory variable, and uses the ratio of patent output to assets per million to measure the innovation ability of the enterprise.
2. Explanatory variables
(1) Financial subsidies
Financial subsidy is the government's financial support for enterprise innovation and the most direct cash subsidy for enterprise innovation investment. This paper uses the ratio of government subsidy enjoyed by enterprises in the current year to total assets at the end of the period to measure.
(2) Tax preference
The goal of the government to implement tax incentives is to stimulate the innovation performance of enterprises, but there is uncertainty in the compensation of tax incentives for innovation results, so it is necessary to verify the implementation effect of tax incentives. At present, preferential tax policies mainly include accelerated depreciation, additional deduction and a preferential tax rate of 15%. These preferential policies will eventually be reflected in the calculation of income tax expenses. In order to enhance the comparability of tax incentives among enterprises, this paper selects the ratio of income tax expenses to EBIT as the measurement index of tax incentives.
3. Intermediary variables
The intensity of R & D investment reflects the relative degree of R & D investment of different enterprises in enterprise expenditure, which is the performance of R & D investment after excluding the influence of enterprise size and so on. This paper mainly measures the investment intensity of R & D funds, which is expressed by the ratio of R & D expenses to operating income.
4. Control variables
Due to the risk aversion of shareholders and the high investment and high risk of innovation activities, excessive concentration of equity may inhibit the innovation activities of enterprises; If R & D capital investment is a necessary condition for enterprise innovation, then human capital is the key factor of innovation and plays an important role in innovation activities; Large scale enterprises have strong financial resources, advanced technology and equipment, and the investment in innovation activities accounts for a relatively small proportion; The longer the enterprise is established, the more comprehensive the innovation knowledge it has. However, due to the lack of basic knowledge, the enterprise with a shorter establishment can only improve the innovation output through a large amount of investment; The asset liability ratio reflects the capital structure of an enterprise. Too much debt is easy to fall into financial difficulties, leading to the inability of enterprises to innovate. Too little debt means that they do not make full use of financial leverage, which will reduce the profitability of enterprises and restrict the development of enterprises. Therefore, this paper selects equity concentration, human capital factors, enterprise size, enterprise age and capital structure as control variables. The definition of variables is shown in Table 1.
（3） Model design
In order to test the hypothesis, this paper designs the following model:
The independent variables of model (1) are fiscal subsidies and tax preferences, and the dependent variable is innovation performance, which is used to test hypothesis 1; The independent variables of model (2) are fiscal subsidies and tax preferences, and the dependent variable is R & D investment, which is used to test hypothesis 2; The independent variable of model (3) is R & D investment, and the dependent variable is innovation performance, which is used to test hypothesis 3; The independent variable of model (4) is financial subsidies, tax incentives and R & D investment, and the dependent variable is innovation performance. The direct effect of R & D investment on innovation performance is tested through D3, and then the intermediary effect is tested through A1, A2, B1, B2 and D1, D2 to determine whether R & D investment is a complete intermediary effect or a partial intermediary effect between financial subsidies, tax incentives and innovation performance.
4、 Empirical research
（1） Descriptive statistics and correlation analysis
Descriptive statistics is to analyze the distribution trend of data and find out the possible extreme values in the data. In Table 2, there is no case that the standard deviation is much larger than the mean value, so there is no extreme value problem in the data. The standard deviation of innovation performance is 0.0193, and the average value is 0.0138, indicating that the innovation performance difference of enterprises is small, but the overall level is low. There is no significant difference in financial subsidies and R & D investment, but the difference between the maximum and minimum tax preferences is large, indicating that the contribution rate of tax preferences to the income of high-tech enterprises is quite different. The standard deviation and full range values of human capital factors, enterprise size and enterprise age are small, indicating that the development of enterprises in these three aspects is relatively balanced. The average asset liability ratio is 30.5510, which does not reach the appropriate level of 40% - 60%, which shows that high-tech enterprises can raise R & D funds by borrowing.
There may be multicollinearity between the variables in the same regression equation, but the variables are difficult to exist completely independently, and the correlation within a certain range will not seriously affect the results. It can be seen from table 3 that the Pearson correlation coefficients of the explained variables, explanatory variables and intermediary variables are basically around 0.1, and only the Pearson correlation coefficient of financial subsidies reaches 0.27, but it does not exceed 0.3, which is in the weak correlation ranks, meeting the basic requirements of linear research. The coefficient of some control variables is high, but it does not exceed 0.6 as a whole. The Pearson correlation coefficient is moderately correlated between 0.5 and 0.8, indicating that there may be multiple collinearity between variables. Therefore, this paper will carry out D.W. collinearity test on variables.
（2） Result analysis
1. Direct effect analysis
It can be seen from table 4 that the D.W. value is between 1.863 and 2.064, indicating that there is no serious error autocorrelation. The test results of model (1) show that financial subsidies have a significant positive effect on innovation performance at the level of 1%, while tax incentives have a significant negative effect on innovation performance at the level of 5%, which verifies hypothesis 1 of this paper. The regression coefficient of financial subsidies in model (2) a1=0.187 (p<0.01), indicating that financial subsidies have a significant positive effect on R & D investment; The regression coefficient of tax preference a2=-0.084 (p<0.01), that is, there is a significant negative relationship between tax preference and R & D investment. Hypothesis 2 is tested. The results of model (3) show that R & D investment has a positive incentive effect on innovation performance at the level of 1%, and hypothesis 3 is verified.
2. Intermediary effect analysis
In model (4), R & D investment has a significant positive impact on innovation performance at the level of 5%. According to Baron and Kenny's intermediary research model, the regression coefficient of R & D investment in model (3) (c3=0.125, p<0.05) meets the basic conditions for exploring the intermediary effect. In model (2), the regression coefficients of the explanatory variables fiscal subsidies, tax incentives and the intermediary variable R & D investment are significant at the 1% level, and in model (4), the regression coefficients of the intermediary variable R & D investment and the innovation performance of the explained variable are also significantly correlated (d3=0.096, p<0.05), so there is an intermediary effect. At the same time, the regression coefficient d2=-0.059 (p>0.05) between tax incentives and innovation performance in model (4), so R & D investment has a complete intermediary effect between tax incentives and innovation performance; The regression coefficient between financial subsidies and innovation performance d1=0.134 (p<0.01). According to the step-by-step test method of intermediary effect, if the direct effect of the explanatory variable on the explained variable is still significant, the intermediary variable is a partial intermediary effect between the explanatory variable and the explained variable, so the intermediary variable R & D investment is a partial intermediary effect between financial subsidies and enterprise innovation performance. See Figure 1 for the specific intermediary effect. In Figure 1, financial subsidies and tax incentives mainly affect innovation performance through R & D investment. Although there is a significant correlation between financial subsidies, tax incentives and innovation performance, the impact coefficient of the two on innovation performance is not very high, and the impact coefficient of R & D investment on innovation performance is also very small, so it is also necessary to analyze the impact effects between variables. When analyzing the robustness test, this paper will also further analyze the conversion efficiency of the intermediary effect.
In the part of control variables, human capital, as an intangible element of innovation investment, has the same impact on innovation performance as capital investment of tangible elements, which is a significant positive impact. Ownership concentration has a significant positive relationship with innovation performance, but a significant negative relationship with R & D investment. In the long run, shareholders hope that the enterprise will have a high innovation performance, but they hold a negative attitude towards the high investment in R & D costs. The relationship between enterprise size and innovation performance, R & D investment and equity concentration is similar. Large-scale enterprises have strong financial resources, and the proportion of R & D investment is relatively small, but the innovation benefits brought by technological improvement are considerable. The negative correlation of enterprise age is that the older the enterprise is, the more mature the innovative production of the enterprise is. The primary standard for mature enterprises to carry out innovative production is to reduce the innovation cost, which directly leads to the reduction of R & D investment; Insufficient R & D investment will directly weaken the follow-up innovation ability of enterprises, and insufficient innovation ability will also be detrimental to the improvement of enterprise innovation performance. Therefore, enterprise age has a significant negative impact on enterprise innovation performance. Due to the high risk of R & D and innovation of enterprises, creditors will worry about the solvency of enterprises, so creditors will reduce or even not invest in high-tech enterprises, so the capital structure has a negative impact on innovation performance.
3. Robustness test
In order to test the robustness of the conclusion, the bootstrap method of Mplus software is introduced to verify the relationship between financial subsidies, tax incentives, R & D investment and innovation performance. The control variables are not tested, and the results are shown in Table 5. After adding R & D investment, the direct effect of financial subsidies on innovation performance is 0.009, and the two-sided p value is greater than 0.05. The direct effect of tax incentives on innovation performance is -0.009, and the two-sided p value is also greater than 0.05, which shows that the impact of both on innovation performance is not significant. Therefore, it is determined that R & D investment plays a complete intermediary effect between financial subsidies, tax incentives and innovation performance. The mediation effect detected by bootstrap method is stronger than that detected by stepwise test method, but the overall results are consistent with the previous results. This may be due to the absence of control variables in the robustness test, which enhances the intermediary effect of R & D investment between financial subsidies and innovation performance. However, the total effect of financial subsidies on enterprise innovation performance is 0.232, of which the intermediary effect accounts for only 3.9%, indicating that the transformation ability of the intermediary effect to the total effect is not very strong. Since fiscal subsidies have a positive effect on innovation performance as a whole, the government needs to pay attention to the conversion efficiency between fiscal subsidies and innovation performance by increasing R & D investment. The overall impact effect of tax incentives is -0.073, and the intermediary effect accounts for 12.3% of the total effect. The conversion efficiency is relatively high. However, in view of the negative impact of tax incentives on Enterprise R & D investment, the government needs to consider appropriate adjustments to the tax incentives.
5、 Conclusion and Prospect
By constructing an intermediary effect model, this paper analyzes the relationship between financial subsidies, tax incentives, R & D investment and innovation performance. The main research conclusions are as follows: (1) financial subsidies have a significant positive effect on innovation performance, and have an inducing effect on innovation performance, while tax incentives have a significant negative effect on innovation performance, and have a crowding out effect on innovation performance; (2) The relationship between financial subsidies, tax incentives and R & D investment is similar to the relationship between them and innovation performance, but R & D investment has a significant positive impact on innovation performance; (3) When using the stepwise test method to test the intermediary effect, R & D investment plays a partial intermediary role between financial subsidies and innovation performance, and a complete intermediary role between tax incentives and innovation performance; (4) When using bootstrap method to test the intermediary effect of R & D investment, it is found that R & D investment is a complete intermediary effect between financial subsidies and innovation performance, tax incentives and innovation performance, and it is still a complete intermediary effect between tax incentives and innovation performance.
Targeted financial subsidies for innovation can help enterprises carry out innovative production quickly. Due to the volatility of enterprise profits, the incentive effect of tax incentives on enterprise innovation is uncertain, so tax incentives have a negative impact on innovation performance. In the process of practice, the government should give priority to fiscal subsidies, supplemented by tax incentives, expand the inducing effect of fiscal subsidies on innovation performance, reduce the crowding out effect of tax incentives on innovation performance, and better guide enterprises to carry out innovative production. In addition, enterprises should make good use of the financial subsidies given by the government to try to offset the crowding out effect brought by tax preferences, so as to achieve better innovation benefits. R & D investment plays an intermediary role between financial subsidies, tax incentives and innovation performance, which reflects the mechanism of financial subsidies and tax incentives on innovation performance. This paper provides a new perspective for the study of the relationship between the three, and provides a new entry point for improving the utilization of fiscal and tax policies, so as to better play the guiding role of fiscal and tax policies.
The limitation of this paper is to use the number of enterprise patent applications as the proxy variable of innovation performance, but the number of patents can not represent all the innovation output of enterprises, which may cause the lack of innovation performance information. In addition, the research object only covers the gem high-tech enterprises, and the follow-up research needs to take the national high-tech enterprises as a large sample.