Innovation and High-Quality Development of Enterprises—Also on the Effect of Innovation Driving the Transformation of China’s Economic Development Model

Abstract

Taking the A-share listed companies in Shanghai and Shenzhen from 2013 to 2020 as the research object, this paper discusses the impact of innovation on high-quality development at the enterprise level and evaluates the effectiveness of the current transformation of innovation driving China’s economic development mode. It is found that innovation inhibits the improvement of total factor productivity of enterprises overall and hinders the high-quality development of enterprises. This result also shows that China’s innovation driven economic development mode transformation has not achieved actual effect. The reason is that China is still in the dilemma that low-quality innovation occupies the dominant position of enterprise innovation. The inhibitory effect of low-quality innovation masks the promoting effect of high-quality innovation, which makes the net effect of innovation on enterprise productivity being inhibitory, and this inhibition is characterized by a loss of efficiency, which is caused by a mismatch of resources. Heterogeneity analysis shows that the inhibition effect of innovation is stronger in eastern enterprises, non-state-owned enterprises and the early implementation of innovation driven strategy. It is also found that the promotion effect of high-quality innovation on executive compensation is stronger than that of low-quality innovation, which indicates that carrying out low-quality innovation is an irrational and short-sighted behavior. The study provides important insights for China to optimize and improve innovation policies and to promote the shift of the overall innovation structure to high-quality innovation.

1. Introduction

The report of China’s 19th CPC National Congress made a clear judgment that their economy has shifted from a high-speed growth to a high-quality development stage, and it put forward the urgent requirement of improving total factor productivity [1], indicating that promoting high-quality development in the new stage has become the main melody of China’s economic development. In the speech on Comprehensively Strengthening the Protection of Intellectual Property Rights, President Xi proposed that “Innovation is the driving force for high-quality development” [2]. Hence promoting high-quality development by innovation will be an important core of China’s modernization construction in the new era. Since the publication of the planning outline of the national medium and long term science and technology development, China’s innovation has made great progress under the guidance of capital incentives of governments at all levels. By the end of 2020, the number of invention patents in China (excluding Hong Kong, Macao and Taiwan) has reached 2.213 million pieces, with an average of 15.8 pieces per 10,000 people [3]. The 2021 Global Innovation Index Report released by the World Intellectual Property Organization (WIPO) shows that China’s innovation index has risen to 12th place, ranking first in the world’s middle-income economy, achieving nine consecutive liters [4]. Although China’s scientific and technological innovation ability has been continuously improved, and the investment of R&D personnel is the first and the investment of funds is the second in the world, it is undeniable that China still has the problem of low conversion rate in the transformation of scientific and technological innovation. Overall, the supporting capacity of scientific and technological innovation for China’s economic and social development is still insufficient, and the contribution rate to economic growth is far lower than that of developed countries. The weak ability of scientific and technological innovation has become the “Achilles heel” of China’s high-quality economic development [5]. In this context, exploring the impact of innovation on high-quality development and evaluating the effectiveness of China’s innovation driven economic development mode transformation have obvious policy meaning and theoretical value.

To some extent, high-quality economic development is reflected in the continuous improvement of total factor productivity. Total factor productivity is both the purpose and means of high-quality economic development. Therefore, improving total factor productivity and realizing high-quality economic development are highly unified in essence [6]. Total factor productivity is essentially a kind of resource allocation efficiency, and innovation can bring technological progress to improve the efficiency of resource allocation. Therefore, innovation can provide power to improve total factor productivity, that is, innovation can drive high-quality economic development. As a key subject of the market economy, enterprises should play the role of innovation subjects to improve the efficiency of the national economy and the transformation of economic development mode. In conclusion, effectively improving the total factor productivity of enterprises is the key to the high-quality development of China’s economy [7], and it is also an important component to achieve the transformation goal of China’s innovation driven economic development mode.

There are abundant studies on the effect of innovation. For example, Hall et al. (2010) found that technological innovation helps to improve the productivity of enterprises [8]; Jennifer et al. (2010) used the data of the United States, Japan and Europe and found that innovation helps to improve regional economic resilience [9]; Josheski et al. (2011) found that there is a positive correlation between innovation level and economic growth in G7 countries [10]; Bristow and Healy (2018) found that the stronger the innovation capacity, the stronger the European countries’ ability to resist the crisis and recover from the crisis [11]. For the research of Chinese scholars, Li (2013) found that the improvement of technological innovation capacity will not only reduce local carbon emissions, but also have a positive spatial spillover effect on their neighboring regions [12]; Lv and Li (2015) found that innovation can promote the high-quality development of China’s economy [13]; Yuan (2016) found that innovation can promote the green development of China’s economy [14]; Zhou et al. (2016) found that the improvement of innovation level has a significant positive impact on the enterprise’s performance [15]; Liu et al. (2018) found that innovation has an energy-saving effect [16]; Shi and Zhao (2018) found that technological innovation can promote the upgrading of industrial structure [17]; Liu et al. (2020) found that the increasing innovation investment can improve the production efficiency of enterprises [18]; Wu et al. (2021) found that innovation can significantly promote the total factor productivity of enterprises [19]; Cheng and Jin (2022) found that the improvement of innovation ability can significantly enhance the resilience of urban economy [20]; Ya et al. (2022) found that technological innovation has significantly promoted the decline of enterprise environmental costs [21]; Chen and Wu (2022) found that innovation can also significantly improve the export tendency of enterprises [22]. However, there are some problems. For example, most of these studies are carried out on the premise that innovation is high-quality innovation [23], and their innovation indicators are mostly proxy variables of high-quality innovation. Therefore, it is biased to take the impact of high-quality innovation as the effect of overall innovation, and few studies have analyzed the innovation effect from the perspective of innovation quality. In recent years, the central government of China has been emphasizing the transformation of innovation activities from the pursuit of quantity to the improvement of quality, which reflects that the strong stimulation of China’s capital policy in the past has led to a certain of strategic and quantity oriented low-quality innovation. Due to the neglect of the impact of low-quality innovation, the conclusions of existing research may not fully reflect the real situation of the effect of innovation in China. In view of this, this paper establishes an econometric model to explore the impact of China’s innovation at the enterprise level. Meanwhile, this study also helps to answer the following research questions:

Does innovation always improve the total factor productivity and thus promote the high-quality development of enterprises?

Are there any essential differences in the impact of innovation of different quality?

Does innovation structure affect the overall performance of innovation?

The first question will be answered by the benchmark analysis along with the endogenous test and robustness test. The results of the mechanism test of this study can answer the second and the third question.

Compared with the existing research, the contribution of this paper is mainly in the following three aspects: first, through the theoretical model and empirical test, this paper jointly reveals the negative effect of low-quality innovation inhibiting the improvement of enterprise productivity, that is, we find out the root of the overall inhibitory effect of innovation in China. The results of this paper form an important supplement to the existing research and have some theoretical significance. Second, this paper not only explores the effect of innovation on enterprise productivity at the overall level, but also distinguishes the effect of different quality innovation on enterprise productivity at the level of innovation structure. It not only verifies the inhibition of low-quality innovation, but also identifies China’s current low-quality biased innovation structure. Hence it provides empirical evidence for the necessity of promoting the innovation structure shift to high-quality innovation. Third, this paper explores the structural problems of the effect of innovation on enterprise productivity and puts forward targeted countermeasures. Under the background of the effect of long-term structural factors and short-term cyclical factors, this study has important practical significance for promoting high-quality economic development and serving the construction of an innovative country.

The rest of the paper is organized as follows. Section 2 is the literature review. Section 3 provides the theoretical model and research hypotheses. Section 4 introduces the research design, the sample selection and data source, model setting and variable definition. Section 5 performs the empirical results and the analysis of these results. Section 6 discusses the transmission mechanism, the heterogeneity tests, and so on. Section 7 discusses how the findings of this study are comparable or in contrast to the findings of other relevant studies. Section 8 gives the conclusions and insights of this study.

2. Literature Review

2.1. Indicators of Innovation

There are many studies on innovation. The existing literature can be divided into two categories: the first is the study of classified heterogeneous effects. For example, Li and Zheng (2016) distinguished innovative from substantive innovation and strategic innovation [24]; Hu and Yu (2022) divided innovation into breakthrough innovation and gradual innovation [25]; Shao and Wu (2021) divided innovation into utilization innovation and exploratory innovation [26]. The second is general research without classification, and its dimensions include two cases: (1) research with a single dimension, such as innovation output [27,28,29], innovation input [30], etc.; (2) indicators of two dimensions, such as Li et al. (2021) [23], Wen and Wang (2021) [31], who measure innovation from two dimensions of input and output; and Dong and Zhang (2021), who measure innovation from two dimensions of output and efficiency [32]. In terms of measurement methods, the input of innovation is mostly measured by the ratio of R&D expenses to operating revenue, but there are differences between current revenue and previous revenue, and there are also measurement methods in the form of dummy variables [33]; Innovation output is often measured by the number of patents, but there are choices between invention patents and all patents, application patents and authorized patents. In a word, there are abundant definitions of innovation in the existing studies, which provide important references for the research of this paper.

2.2. Innovation and Total Factor Productivity

Many studies show that innovation has a positive impact on total factor productivity. Among them, Li and Wang (2016) found that R&D investment had a positive effect on the promotion of TFP based on the data of Chinese industrial enterprises [34]; Chen et al. (2018) found that innovation had strengthened the role of trade liberalization in improving the total factor productivity of enterprises through the mediation effect [35]; Zhang and Meng (2020) found that the increase of R&D investment was positively related to the improvement of total factor productivity [36]; Sheng et al. (2020) found that with the promotion of exploratory innovation and the continuous emergence of achievements, innovation could continuously promote the growth of enterprise total factor productivity [37]; Wu et al. (2021) found that innovation and green innovation played a significant role in promoting the total factor productivity of enterprises [19]; Song et al. (2020) found that financial innovation significantly promoted the improvement of total factor productivity too [7]. However, a small number of studies, such as Li (2010) and Gao (2017), found that innovation investment did not promote the growth of total factor productivity but inhibited it [38,39].

2.3. Literature Commention

Reviewing the literature, we can find that the mainstream understanding of the existing research is that innovation promotes the improvement of total factor productivity, but it has the following defects: first, the impact of innovation on productivity may be different under different conditions. Most of the existing research concludes that enterprise innovation is high-quality innovation and ignores the existence of low-quality innovation or even the dominant position of low-quality innovation. Second, the only few studies on the negative correlation between innovation and total factor productivity growth also have problems such as incomplete index measurement and insufficient sample representation, which may not fully explain the actual situation of China. Based on above, this paper absorbs the “essence” of the existing research, and further distinguishes the heterogeneous impact of high-quality innovation and low-quality innovation on total factor productivity based on studying the impact of overall innovation on total factor productivity based on the two dimensions of input and output, so as to accurately grasp the innovation effect at the overall and local levels.

3. Theoretical Model and Research Hypothesis

With reference to the Schumpeter growth model of multiple sectors in Aghion and Howitt (1992) [40] and Yi and Liu (2015) [41], the production functions are set as follows:

$$y_t={\displaystyle {\int }_0^1{A}_t}{\left(i\right)}^{1-\alpha }{q}_t{\left(i\right)}^{\alpha }di, \alpha \in \left(0,1\right)$$

(1)

where, $q_t(i)$ is the input of intermediate product i, $A_t(i)$ is the productivity parameter, which measures the quality of intermediate products producing finished products. It can change with the category of intermediate products, and Y is the final output.

Due to the fierce competition among enterprises, when the innovation levels among enterprises are equal, there is no technology monopoly rent. When the innovation levels among enterprises are different, the innovative enterprises in the technology monopoly position can obtain excess profits in the competition. Referring to the research of Acemoglu et al. (2003) [42], the balanced profit form of intermediate product enterprises can be expressed as:

$${\pi }_t(i)=\gamma A_t(i), \gamma =(1-\alpha ){\alpha }^{(1+\alpha )(1-\alpha )}$$

(2)

Suppose that the productivity in period t is $A_t$ and it changes at a exogenous rate of $\delta$, namely:

$$A_t=\delta A_{t-1}$$

(3)

When enterprises carry out high-quality innovation, $\delta >1$; when enterprises carry out low-quality innovation, $\delta <1$.

Referring to the research of Aghion and Howitt (2009) [43], it is assumed that enterprises carrying out high-quality innovation generally belong to the technological frontier department, and their productivity level is $A_t(i)=A_t$, while enterprises carrying out low-quality innovation belong to the department whose technical level lags behind the frontier department, and their productivity is $A_t(i)=A_{t-1}$.

Since innovation is not equivalent to real productivity, there is a probability P for the transformation of innovation achievements:

$$P_{i,t}=\varphi \left(\frac{c_{i,t}}{A_{i,t}^{\ast }}\right)=\lambda {\left(\frac{c_{i,t}}{A_{i,t}^{\ast }}\right)}^{\sigma }$$

(4)

where $A_{i,t}^{\ast }$ is the productivity after the enterprise carries out high-quality innovation, $A_{i,t}^{\ast }=\delta A_{i,t}$ and $\delta >1$; when the enterprise carries out low-quality innovation, then $A_{i,t}^{\ast }=\delta A_{i,t-1}$ and $\delta <1$.

Further, it can be deduced that the R&D capital that enterprises need to invest in carrying out innovation activities is:

$$c_{i,t}={\left(\frac{P_{i,t}}{\lambda }\right)}^{\frac{1}{\sigma }}{A}_{i,t}^{\ast }$$

(5)

Among them, $\lambda$ is the efficiency of the R&D department, and the elasticity is $\sigma$ and $\sigma \in (0,1)$. It is not difficult to understand that the innovation behavior of the enterprise will have an impact on the allocation of R&D capital. Therefore, $\lambda$ can be regarded as a function of the enterprise’s innovation activity T, and $\frac{\partial \lambda }{\partial T}>0$ when the enterprise carries out high-quality innovation and $\frac{\partial \lambda }{\partial T}<0$ when the enterprise carries out low-quality innovation.

In addition to R&D expenses, enterprises also have to pay certain sunk costs to expand financing channels and deal with risks. Therefore, the total cost of enterprises carrying out innovation activities is as follows:

$$C_{i,t}=(1+s){\left(\frac{P_{i,t}}{\lambda }\right)}^{\frac{1}{\sigma }}{A}_{i,t}^{\ast }$$

(6)

According to the research [24], it can be judged that when enterprises carry out low-quality innovation in order to meet government supervision and seize policy preferences, the innovation investment of enterprises is difficult to form effective productivity, and only increases the sunk cost, so $\frac{\partial s}{\partial T}>0$. When an enterprise carries out substantive high-quality innovation activities, it will change the scientific and technological content of products and the proportion of technical elements, improve the efficiency of resource allocation, reduce the cost of enterprise financing and reduce the external risk of high-quality innovation, which will help to reduce the sunk cost of innovation activities, so $\frac{\partial s}{\partial T}<0$.

Under the equilibrium state, the final income of enterprises carrying out high-quality innovation and low-quality innovation activities can be expressed as:

$$R_1=P_{i,t}\gamma A_{t+1}(i)-(1+s){(\frac{P_{i,t}}{\lambda })}^{\frac{1}{\sigma }}{A}_{i,t}^{\ast }$$

(7)

$$R_2=P_{i,t}\gamma A_t(i)-(1+s){(\frac{P_{i,t}}{\lambda })}^{\frac{1}{\sigma }}{A}_{i,t-1}^{\ast }$$

(8)

Assuming that there are only these two enterprises in the market, the final income after they carry out innovation activities can be expressed as:

$$R=R_1+R_2$$

(9)

Because the innovation activities of enterprises are carried out under the principle of profit maximization, there exists the following formula:

$$\frac{\partial R}{\partial P_{i,t}}=0$$

(10)

Then, the probability of transformation of innovation achievements under equilibrium state can be obtained:

$$P_{i,t}={\left(\frac{\gamma {\lambda }^{\frac{1}{\sigma }}\sigma }{(1+s)}\right)}^{\frac{\sigma }{1-\sigma }}$$

(11)

By further deducing the relationship between enterprise innovation behavior and the transformation probability of innovation achievements, the impact of innovation on enterprise productivity is identified as follows:

$$\frac{\partial P_{i,t}}{\partial T}=\frac{\sigma }{1-\sigma }{\left(\frac{\gamma {\lambda }^{\frac{1}{\sigma }}\sigma }{(1+s)}\right)}^{\frac{2\sigma -1}{1-\sigma }}\frac{{\Phi }_T^{\prime }}{{(1+s)}^2}$$

(12)

$${\Phi }_T^{\prime }=\gamma (1+s){\lambda }^{\frac{1-\sigma }{\sigma }}{\lambda }_T^{\prime }-\gamma \sigma {\lambda }^{\frac{1}{\sigma }}{s}_T^{\prime }$$

(13)

Obviously, when enterprises carry out high-quality innovation, due to $s_T^{\prime }<0$, ${\lambda }_T^{\prime }>0$, so, ${\Phi }_T^{\prime }>0$, $\frac{\partial P_{i,t}}{\partial T}>0$; when enterprises carry out low-quality innovation, due to $s_T^{\prime }>0$, ${\lambda }_T^{\prime }<0$, so, ${\Phi }_T^{\prime }<0$, $\frac{\partial P_{i,t}}{\partial T}<0$; if all enterprises in the market are regarded as a whole, there are both low-quality innovation and high-quality innovation, then $s_T^{\prime }$ and ${\lambda }_T^{\prime }$ have some uncertainty, that is, ${\Phi }_T^{\prime }$ and $\frac{\partial P_{i,t}}{\partial T}$ cannot be determined too.

The above results show that carrying out high-quality innovation can improve enterprise productivity, while carrying out low-quality innovation will inhibit the improvement of enterprise productivity, but it is uncertain whether the overall innovation level of all enterprises in the market will promote or inhibit the improvement of enterprise productivity. Based on the above analysis, this paper makes the following alternative assumptions:

Hypothesis 1 (H1).

Innovation promotes the improvement of total factor productivity of enterprises overall and promotes the high-quality development of enterprises.

Hypothesis 2 (H2).

Innovation inhibits the improvement of total factor productivity of enterprises overall and hinders the high-quality development of enterprises.

4. Methodology

4.1. Research Design

The present research is aimed to examine whether innovation improves the total factor productivity and thus promotes the high-quality development of Chinese enterprises. This study used the panel data of A-share listed companies collected from CSMAR and CNRDS database. Learning from the practice of some classical literature, we set the benchmark model of this study and also selected and defined the corresponding variables. Referring to the research of [7], we arranged several empirical tests: first, through the benchmark regression, the core problem of this paper will be tested, that is, what is the impact of innovation on the high-quality development of Chinese enterprises; secondly, a series of endogenous tests and robustness tests will be conducted to ensure the benchmark regression results reliable; then, we will test the mechanism: How innovation affects the high-quality development of Chinese enterprises and explores its internal logic. In addition, this paper tries to relax the general assumption of homogeneity in model setting and selection through a heterogeneity test, so as to have a more reasonable understanding of the results. Finally, through additional tests, it explores whether it is rational for Chinese enterprises to carry out low-quality innovation, and this may prove the necessity and importance of promoting high-quality independent innovation in China from the opposite side. The results obtained from the empirical process were used to draw conclusions. A description of the research design is provided in Figure 1.

4.2. Sample Selection and Data Source

This paper selects the A-share listed companies in Shanghai and Shenzhen from 2013 to 2020 as the research object. Drawing from the research of [44,45], the initial sample is screened as follows: (1) exclude the listed companies in the financial industry, the financial institutions are special, their report structure and main accounting items are different from those of other industries. Generally, it is unnecessary to compare them to other companies; (2) exclude listed companies marked as ST, ST* and PT in the sample period, because ST, ST* and PT are special cases for the stock transactions of listed companies with abnormal financial or other conditions. Companies with such marks are not of general research significance and they will also cause extreme value problems; (3) exclude the company samples with missing research variables, in order to ensure that the results are not distorted and that enterprises with serious data loss are generally eliminated. After the above processing, 12,600 company annual observations of 1575 listed companies were finally obtained. The data of company level in this paper are from the CSMAR database, and some supplementary data about patents are from the CNRDS database. In order to avoid the potential influence of extreme values, this paper also shrinks the tail of all continuous variables by 1% up and down.

4.3. Model Setting and Variable Definition

The goal of Innovation Driven Development Strategy is to create a new engine of economic growth and promote the transformation of the economic development mode to high-quality development. Therefore, improving the innovation ability and innovation rate of enterprises requires implementing the innovation driven strategy at the micro level, and they are the measures to implement the transformation of economic development mode to high-quality development [46]. At the same time, the high-quality development of enterprises is highly unified with the improvement of total factor productivity. Therefore, this paper explores the effect of innovation on the high-quality development of enterprises by studying the effect of innovation on the total factor productivity of enterprises, and evaluates the effectiveness of the transformation of innovation driven economic development mode. So, referring to the research of Cao et al. (2022) [47] and Wu et al. (2022) [48], this paper constructs the following model:

$$TF{P}_{i,t}={\alpha }_0+{\alpha }_{1}Innovatio{n}_{i,t}+\Sigma Control{s}_{i,t}+{\epsilon }_{i,t}$$

(14)

The definition of dependent variable (TFP): the accounting methods of enterprise total factor productivity include the OP method, LP method, WRDG method, MrEst method, etc. Among them, OP method can well alleviate the endogenous problem of traditional methods. Therefore, this paper draws lessons from the practices of scholars such as Olley and pakes (1996), Lu and Lian (2012) [49,50], and uses the OP method to estimate enterprise total factor productivity as a benchmark:

$$\ln Y_{i,t}={\beta }_0+{\beta }_1\ln K_{i,t}+{\beta }_2\ln L_{i,t}+{\beta }_3\ln In{v}_{i,t}+{\epsilon }_{i,t}$$

(15)

where Y represents the output of the enterprise and which is measured by the operating income of the enterprise; K is the capital stock of the enterprise, which is measured by the net value of fixed assets of the enterprise; L is the labor input of the enterprise, which is measured by the number of employees of the enterprise; Inv is the current investment of the enterprise, which is measured by the capital expenditure of the enterprise.

As the OP method needs to meet that the investment is greater than 0 and the investment increases monotonically, some samples will be lost, while the LP method using the intermediate input as the instrumental variable is more flexible. Therefore, according to the research of Levinsohn and Petrin (2003) [51], this paper takes the total factor productivity estimated by the LP method as the alternative variable for robustness test.

The definition of independent variable (Innovation): referring to the research of Balsmeier et al. (2017) [52], Chang et al. (2019) [53] and Li et al. (2021) [23], this paper measures the overall innovation of enterprises from two dimensions: innovation input (R&D, using R&D input) and innovation output (Patent, using patent application).

The selection of control variables: referring to the research of Li et al. (2020) [54] and Song et al. (2020) [7], this paper mainly controls some variables of corporate financial characteristics and governance characteristics, because they may affect the innovation and total factor productivity of enterprises. Among them, the financial characteristic variables include: the cash asset ratio (Cash), which is characterized by the ratio of cash and cash equivalents to total assets. The larger the ratio, the more sufficient the cash flow of the enterprise, and the more abundant the funds for R&D and efficient production activities; enterprise leverage (Lev) is characterized by the ratio of total liabilities to total assets. The level of enterprise debt ratio is a measure of its financing ability. The greater the ratio, the stronger the financing ability of the enterprise; the current asset ratio (CR) is the ratio of current assets to total assets. The greater the ratio, the stronger the short-term solvency of an enterprise; enterprise size (Size) is represented by the natural logarithm of total assets. Generally speaking, large enterprises have more resource advantages and are easy to form economies of scale; proportion of intangible assets (Intangible) is the proportion of intangible assets in total assets. Generally, intangible assets have a strong correlation with the R&D and innovation behavior of enterprises; number of employees (Employee) is expressed by the natural logarithm of the number of employees in the enterprise. According to the human capital theory, human capital can improve the resource intensity and promote the productivity of enterprises through knowledge substitution channels. The governance characteristic variables include: shareholding ratio of major shareholders (TOP10) is expressed by the total shareholding ratio of the top 10 shareholders. The shareholding ratio of shareholders is an important indicator of corporate internal governance, which directly affects the allocation of resources; separation rate of two rights (Seperation) is expressed by the ratio of control to ownership. The greater the degree of separation of two rights, the higher the agency cost. Company age (Age) is expressed by the natural logarithm of the years of establishment plus 1. The longer the duration of the enterprise, the more abundant the production and operation resources and the more mature the production technology; for nature of equity (Nature), the value of state-owned enterprises is 1, otherwise 0. Most state-owned enterprises are monopoly enterprises, the degree of industry competition is low, and state-owned enterprises also shoulder political and social responsibilities. The specific description of variables are shown in

Table 1. Variable description.

Variable Name	Symbol	Variable Definition
Total factor productivity	TFP	Total factor productivity of enterprises measured by OP/LP method
Innovation	R&D	Ratio of R& D expenses to current operating income
	Patent	Natural logarithm of the number of patent applications plus 1
Cash asset ratio	Cash	Ratio of cash and cash equivalents to total assets
Shareholding ratio of top 10 shareholders	Top10	Total shareholding ratio of top 10 shareholders
Separation rate of two rights	Seperation	Ratio of control to ownership
Enterprise leverage	Lev	Ratio of total liabilities to total assets
Current ratio	CR	Ratio of current assets to total assets
Enterprise scale	Size	Natural logarithm of total assets
Nature of equity	Nature	The value of state-owned enterprises is 1, otherwise is 0
Proportion of intangible assets	Intangible	Proportion of intangible assets in total assets
Number of employees	Employee	Natural logarithm of the number of employees of the enterprise
Enterprise age	Age	Natural logarithm of the years of establishment plus 1

.

5. Empirical Results and Analysis

5.1. Descriptive Statistics

Table 2. Descriptive statistics of main variables.

Variable	Obs	Mean	Std. Dev.	Min	Max
TFP	12,600	6.919	1.057	2.934	11.186
R&D	12,600	4.655	4.481	0.050	26.515
Patent	12,600	1.766	1.617	0	6.166
Cash	12,600	0.147	0.109	0.011	0.554
Top10	12,600	55.452	14.441	23.345	89.155
Seperation	12,600	4.675	7.424	0	28.205
Lev	12,600	0.420	0.199	0.057	0.921
CR	12,600	0.562	0.175	0.149	0.908
Size	12,600	22.350	1.231	14.648	23.208
Nature	12,600	0.344	0.475	0	1
Intangible	12,600	18.944	1.526	0	26.204
Employee	12,600	7.918	1.170	5.398	11.133
Age	12,600	2.884	0.315	1.946	3.497

lists the descriptive statistics of the main variables of this article, in which the average value of enterprise total factor productivity (TFP) is 6.919, the standard deviation is 1.057, which is basically consistent with the relevant indicators of the existing research; the maximum value of innovation investment (R&D) is 26.515, minimum 0.050, with an average of 4.655, the standard deviation is 4.481, indicating that the R&D investment intensity of each company in the sample is quite different; the maximum value of patent is 6.166, the minimum value is 0, and the average value is 1.766, the standard deviation is 1.617, which reflects the differentiation of innovation ability of Chinese enterprises to a certain extent. Other variables are similar to equivalent studies and they will not be repeated here.

5.2. Benchmark Regression

Table 3. Benchmark regression results.

	(1)	(2)	(3)	(4)
	TFP_OP	TFP_OP	TFP_OP	TFP_OP
R&D	−0.0204 ***	−0.0248 ***
	(0.0059)	(0.0041)
Patent			−0.0283 ***	−0.0086 ***
			(0.0043)	(0.0031)
Cash		−0.5048 ***		−0.4871 ***
		(0.0824)		(0.0436)
Top10		−0.0018 **		−0.0004
		(0.0007)		(0.0004)
Seperation		−0.0004		0.0002
		(0.0013)		(0.0008)
Lev		−0.0831		−0.0885 ***
		(0.0677)		(0.0189)
CR		−0.0132 ***		−0.0154 ***
		(0.0035)		(0.0012)
Size		0.6570 ***		0.5399 ***
		(0.0299)		(0.0072)
Nature		−0.0758 **		−0.0062
		(0.0336)		(0.0182)
Intangible		0.0096		0.0010
		(0.0059)		(0.0038)
Employee		−0.2661 ***
		(0.0204)
Age		0.0077 **		0.0155 ***
		(0.0032)		(0.1328)
Year&Indcd	Yes	Yes	No	No
_cons	16.9770 ***	4.0159 ***	17.0908 ***	4.2851 ***
	(0.2269)	(0.5009)	(0.0231)	(0.1328)
N	12,600	12,600	12,600	12,600
R2	0.3701	0.5994

Note: Robust standard errors are shown in brackets, *** and ** are significant at the level of 1% and 5% respectively.

reports the regression results of the impact of innovation on enterprise total factor productivity. The explanatory variable is the enterprise total factor productivity (TFP_OP) measured by OP method. Each regression group is clustered at the company level. The coefficients of R&D in columns (1) and (2) are significantly negative at the level of 1%, indicating that R&D investment inhibits the improvement of total factor productivity of enterprises. The coefficient of Patent in columns (3) and (4) is also significantly negative at the level of 1%, indicating that patent applications cannot promote the improvement of total factor productivity of enterprises too. In conclusion, the results in Table 3 show that innovation inhibits the improvement of total factor productivity and hinders the high-quality development of enterprises overall. This result supports hypothesis H2. At the same time, the results also reflect that the transformation of China’s innovation driven economic development mode has not achieved practical effects.

The estimated results of the control variables in Table 3 show that there is a positive correlation between enterprise size (Size) and total factor productivity, which is significant at the statistical level of 1%. Generally speaking, large-scale enterprises have more operability in resource allocation. They can improve and reform relevant departments through the continuous specialization of division of labor, so as to realize the specialization of product procurement and sales and the optimization of R&D environment, which improves the productivity of enterprises. This result is consistent with the research of Sun and Wang (2014) [55]. The company age (Age) is also positively correlated with the total factor productivity, and it is significant at the statistical level of 5%. Companies with a long history of establishment have rich experience in resource allocation, their resource allocation efficiency will naturally be higher, which can improve the total factor productivity. This result is consistent with the research of Tian and Twite (2011) [56]. The proportion of intangible assets (Intangible) is positively correlated with total factor productivity, but it is not statistically significant. The possible reason is that the quality of innovation output represented by intangible assets in Chinese enterprises is low, which does not play a significant role in promoting the improvement of total factor productivity. This result also confirms the findings of the following research, that is, low-quality innovation occupies the dominant position in enterprise innovation. Both the cash asset ratio (Cash) and the current asset ratio (CR) are negatively correlated with the total factor productivity of the enterprise at the statistical level of 1%, which is different from the expectation. The possible reason is that this part of the fund is idle or occupied and has not been invested in the R&D activities of the enterprise, resulting in efficiency loss. The shareholding ratio of major shareholders (TOP10) is negatively correlated with the total factor productivity of enterprises at the statistical level of 5%, which is consistent with the research of Chen and Qi (2014) [57], and this result is in line with the “tunneling effect” hypothesis. The coefficient values of the enterprise leverage (Lev) and separation rate of two rights (Seperation) are not statistically significant., indicating that their impacts on enterprise total factor productivity are not very important. The nature of equity (Nature) is negatively related to the total factor productivity of enterprises, which is in line with the reality that China’s state-owned enterprises shoulder political and social responsibilities, and also confirms the necessity of China’s continuous implementation of the reform of state-owned enterprises. There is a significant negative correlation between the number of employees (Employee) and the total factor productivity of the enterprise at the statistical level of 1%, which is also different from the expectation. The possible reason is that the number of employees only reflects the number of human capital but does not reflect the quality of human capital. At the same time, the result also shows that Chinese enterprises still have some serious human capital mismatch. In keeping with what Ma et al. (2018) [58] found, the mismatch of human capital led to the loss of 1.79% and 1.63% of China’s actual total output in 2007 and 2013. Therefore, it is necessary to increase the investment in vocational education and training to improve the quality of human capital of no-skilled labor, so as to jointly improve the total factor productivity of enterprises.

5.3. Endogenous Test

5.3.1. Instrumental Variable Method

In the benchmark regression, this paper controls some factors that can affect innovation and enterprise total factor productivity, but it can not avoid the results being affected by some unobservable factors. This missing variable problem will lead to the bias of the estimation results in this paper. At the same time, there may be a reverse causal relationship between innovation and productivity. To alleviate the above concerns, this paper further adopts the instrumental variable method.

Based on the research of Zhang et al. (2017) [59] and Song et al. (2020) [7], this paper uses the mean value of the innovation level of the same industry where the enterprise is located as the instrumental variable (IV1 and IV2). Firstly, there can be a high correlation between the average innovation level of the same industry in the same region and the innovation level of the target enterprise. Secondly, the average innovation level of the same industry in the same region is difficult to directly affect the total factor productivity of the target enterprise. Therefore, the IV selected in this paper meets the instrumental variables’ requirements of correlation and externality.

Columns (1)–(4) of

Table 4. Endogenous problem test.

	(1)	(2)	(3)	(4)	(5)	(6)
	2SLS	2SLS	2SLS	2SLS	GMM	GMM
	R&D	TFP_OP	Patent	TFP_OP	TFP_OP	TFP_OP
R&D		−0.0325 ***			−0.0258 ***
		(0.0021)			(0.0010)
IV1	0.7130 ***
	(0.0139)
Patent				−0.0097 ***		−0.0001
				(0.0047)		(0.0039)
IV2			0.9593 ***
			(0.0100)
L.TFP_OP					0.5357 ***	0.6184 ***
					(0.0263)	(0.0452)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
N	12,600	12,600	12,600	12,600	12,600	12,600
R2		0.7958		0.6852
AR(1)					0.0000	0.0000
AR(2)					0.0230	0.0610
Hansen					0.0000	0.0000

Note: Robust standard errors are shown in brackets, *** is significant at the level of 1% respectively.

report the results of the two-stage regression of instrumental variables, of which columns (1) and (3) are the results of the first stage regression and columns (2) and (4) are the results of the second stage regression. The results of the first stage show that the regression coefficients of IV1 and IV2 are significantly positive at the level of 1%, indicating that the higher the average innovation level of the same industry in the same region, the higher the innovation level of the target enterprise, which supports the assumption of the correlation of instrumental variables. The two-stage regression results show that the coefficients of R&D and Patent are significantly negative at the level of 1%, indicating that the results of this paper still exist after alleviating the possible endogeneity. In addition, the tests not listed also show that there is no problem of weak instrumental variables.

5.3.2. GMM Dynamic Panel Analysis

The total factor productivity of enterprises has a certain sequence correlation. In order to alleviate this problem, we draw from the ideas of Xiao and Xue (2019) [60], Song et al. (2020) [7] and further use the System GMM regression to test the robustness of the conclusions in this paper. The regression results are shown in columns (5) and (6) of Table 4. Among them, the results of test statistics show that the instrumental variables are effective and the System GMM is applicable; the coefficients of R&D and Patent are negative, and the coefficient of R&D is significant at the level of 1%. The results roughly show that after considering the sequence correlation of total factor productivity (controlling L. TFP_OP and its endogeneity), the conclusions above are still valid.

5.4. Robustness Check

5.4.1. Change the Measurement Method of Total Factor Productivity (TFP)

The total factor productivity of enterprises used in the previous regression is calculated by the OP method. Here, the LP method is used to re-calculate the enterprise’s total factor productivity (TFP_LP), and then substitute the TFP_LP into Equation (14) for regression. The results are shown in panel A of

Table 5. Robustness test.

Panel A: Change the measurement method of total factor productivity (TFP)
	(1)	(2)	(3)	(4)
	TFP_LP	TFP_LP	TFP_LP	TFP_LP
R&D	−0.0184 ***	−0.0248 ***
	(0.0062)	(0.0041)
Patent			−0.0336 ***	−0.0173 ***
			(0.0040)	(0.0035)
Controls	No	Yes	No	Yes
Year&Indcd	Yes	Yes	No	No
N	12,600	12,600	12,600	12,600
R2	0.3045	0.5632
Panel B Key variables lag by one period
	(1)	(2)	(3)	(4)
	TFP_OP	TFP_OP	TFP_OP	TFP_OP
L.R&D	−0.0125 ***	−0.0122 ***
	(0.0044)	(0.0034)
L.Patent			−0.0084 *	−0.0092 ***
			(0.0044)	(0.0047)
Controls	No	Yes	No	Yes
Year&Indcd	Yes	Yes	No	No
N	11,025	11,025	11,025	11,025
R2	0.3122	0.5415
Panel C Only the sample of manufacturing enterprises is retained
	(1)	(2)	(3)	(4)
	TFP_OP	TFP_OP	TFP_OP	TFP_OP
R&D	−0.0181 ***	−0.0243 ***
	(0.0055)	(0.0057)
Patent			−0.0204 ***	−0.0090 ***
			(0.0044)	(0.0031)
Controls	No	Yes	No	Yes
Year&Indcd	Yes	Yes	No	No
N	6424	6424	6424	6424
R2	0.3559	0.6168

Note: Robust standard errors are shown in brackets, *** and * are significant at the level of 1% and 10% respectively.

. Throughout panel A, it is easy to see that the regression results are consistent with the benchmark regression.

5.4.2. Key Variables Lag by One Period

The R&D process takes some time, and R&D activities will not only affect the current productivity but also have a lag effect. Therefore, this paper introduces the lag of innovation input (L.R&D) and innovation output (L.Patent) into the model. The regression results are shown in panel B of Table 5. It is not difficult to see from panel B that the consistent results with the benchmark regression still exist.

5.4.3. Only the Sample of Manufacturing Enterprises Is Retained

Generally speaking, the manufacturing industry is the industry most in need of innovation, and it is also the most active industry in technological innovation. Based on this, this paper only retains the data of manufacturing enterprises to test the robustness of the benchmark results. The regression results are shown in panel C of Table 5, and the results consistent with the benchmark regression can still be obtained.

6. Further Study

6.1. Mechanism Analysis

Drawing from the ideas of Li and Zheng (2016) [24], this paper takes utility model patents and design patents as the proxy variables of low-quality innovation (L-Patent) and invention patents as the proxy variables of high-quality innovation (H-Patent), and respectively substitutes them into Equation (14) for regression. The results are shown in columns (2) and (3) of

Table 6. Mechanism test.

	(1)	(2)	(3)	(4)	(5)
	TFP_OP	TFP_OP	TFP_OP	MISS	TFP_OP
Patent	−0.0086 ***			0.0131 *	−0.0041
	(0.0031)			(0.0076)	(0.0044)
L-Patent		−0.0337 ***
		(0.0046)
H-Patent			0.0141 ***
			(0.0046)
MISS					−0.0289 ***
					(0.0071)
Controls	Yes	Yes	Yes	Yes	Yes
Year&Indcd	No	No	No	Yes	Yes
N	12,600	12,600	12,600	12,600	12,600
R2				0.0301	0.5653

Note: Robust standard errors are shown in brackets, *** and * are significant at the level of 1% and 10% respectively.

, where the coefficient of L-Patent is −0.0377, significant at the level of 1%, and the coefficient of H-Patent is 0.0141, significant at the level of 1%.

Through the results of columns (1)–(3), it is not difficult to find that low-quality innovation inhibits the improvement of the enterprise’s total factor productivity, while high-quality innovation promotes the improvement of the enterprise’s total factor productivity, and innovation has an inhibitory effect on enterprise productivity on the whole because the inhibitory effect of low-quality innovation should mask the promoting effect of high-quality innovation. The results also reflect that the current innovation structure of China is that low-quality innovation occupies the dominant position of enterprise innovation. Since the 18th National Congress of the Communist Party of China, in order to accelerate the realization of the strategic goal of innovation driven development, the government has intensively issued capital policies to stimulate innovation, resulting in the proliferation of low-quality innovation [61], forming a pattern in which low-quality innovation occupies the dominant position of innovation. After years of policy adjustment, the number of low-quality innovation has decreased to a certain extent [23], but the innovation structure has not changed substantially, Low-quality innovation still occupies the leading position of enterprise innovation.

The above analysis shows that the overall inhibition of innovation is caused by the dominant inhibition effect of low-quality innovation masking the promotion effect of high-quality innovation, but the specific transmission path is not clear. Total factor productivity is also a kind of resource allocation efficiency in essence, and resource mismatch is an important reason for the decline of resource allocation efficiency. Therefore, in order to test the transmission mechanism, this paper sets the following model:

$$MIS{S}_{i,t}={\alpha }_0+{\alpha }_{1}Innovatio{n}_{i,t}+\Sigma Controls+{\epsilon }_{i,t}$$

(16)

$$TF{P}_{i,t}={\alpha }_0+{\alpha }_{1}Innovatio{n}_{i,t}+{\alpha }_{2}MIS{S}_{i,t}+\Sigma Controls+{\epsilon }_{i,t}$$

(17)

Referring to Richardson (2006) [62], an investment model is established to fit the normal investment level of the enterprise. The absolute value of the residual (MISS) of the model represents the deviation degree between the actual investment and the normal investment, that is, the degree of resource mismatch. The greater the MISS, the more serious the resource mismatch. We use MISS as the intermediary variable into Equations (16) and (17), and the regression results are shown in columns (4) and (5) of Table 6. According to the three-step principle of intermediary effect, it is not difficult to find that resource mismatch plays a complete intermediary effect, that is, the inhibition of innovation is caused by a resource mismatch. Combined with the results of columns (1)–(3), it can be considered that those low-quality innovations have caused a serious resource mismatch, resulting in the loss of production efficiency.

6.2. Heterogeneity Test

Based on regional heterogeneity, ownership heterogeneity and time heterogeneity, this paper divides the samples into different groups to further investigate the cross-sectional differences of the impact of innovation on total factor productivity.

6.2.1. Regional Heterogeneity

There are many enterprises in the eastern region of China, and the industry competition is also very fierce over there. The enterprises tend to carry out low-quality innovation activities to compete for scarce resources. However, the development of enterprises in the central and western regions of China is relatively backward. Some enterprises are still in the stage of factor driven or investment driven, and the overall demand for innovation is not high. In order to test regional heterogeneity, this paper divides the samples into a eastern region sample group and a central and western region sample group according to the location of the company. The grouping regression results are shown in panel A of

Table 7. Heterogeneity test.

Panel A Regional heterogeneity
	east	Central and Western	east	Central and Western
	(1)	(2)	(3)	(4)
R&D	−0.0266 ***	−0.0177 **
	(0.0042)	(0.0072)
Patent			−0.0151 ***	0.0072
			(0.0038)	(0.0053)
Controls	Yes	Yes	Yes	Yes
Year&Indcd	Yes	Yes	No	No
N	8983	3617	8983	3617
R2	0.5894	0.6225
Panel B Equity heterogeneity
	state-owned enterprise	Non state-owned enterprises	state-owned enterprise	Non state-owned enterprises
	(1)	(2)	(3)	(4)
R&D	−0.0154 **	−0.0277 ***
	(0.0065)	(0.0042)
Patent			−0.0050	−0.0120 ***
			(0.0047)	(0.0040)
Controls	Yes	Yes	Yes	Yes
Year&Indcd	Yes	Yes	No	No
N	4334	8266	4334	8266
R2	0.6128	0.6189
Panel C Temporal heterogeneity
	early stage (2013–2016)	later stage (2017–2020)	early stage (2013–2016)	later stage (2017–2020)
	(1)	(2)	(3)	(4)
R&D	−0.0239 ***	−0.0178 ***
	(0.0075)	(0.0052)
Patent			−0.0148 ***	−0.0018
			(0.0046)	(0.0040)
Controls	Yes	Yes	Yes	Yes
Indcd	Yes	Yes	No	No
N	6300	6300	6300	6300
R2	0.1323	0.1072

Note: Robust standard errors are shown in brackets, *** and ** are significant at the level of 1% and 5% respectively.

. It can be seen from panel A that the coefficients of R&D are all significantly negative at the level of 1%, but the absolute value of the coefficient of the sample group in the eastern region is greater (0.0266 > 0.0177). Patent’s coefficient is significantly negative in the eastern sample group, while it is positive and not significant in the central and western sample group. The results of panel A are consistent with expectations, and the inhibitory effect of innovation is stronger among enterprises in the eastern region with fierce industry competition.

6.2.2. Ownership Heterogeneity

Most state-owned enterprises are large-scale and have a certain monopoly position. At the same time, state-owned enterprises also shoulder social and political responsibilities. The probability and necessity of low-quality innovation of state-owned enterprises are low; by contrast, non-state-owned enterprises not only face fierce industry competition but also have financing constraints. They need to carry out strategic innovation to meet the preference of government innovation incentives, so as to obtain various policy preferences. Therefore, non-state-owned enterprises have a higher tendency to carry out low-quality innovation. In order to test the heterogeneity of ownership, the sample is divided into state-owned enterprise group (Nature = 1) and non-state-owned enterprise group (Nature = 0) according to the nature of ownership. The grouping regression results are shown in panel B of Table 7. It can be seen from panel B that the coefficients of R&D are all significantly negative, but the absolute value of the coefficient of non-state-owned enterprise group is greater (0.0277 > 0.0154). Patent’s coefficients are all negative too, and the coefficient of non-state-owned enterprise group is significant at the level of 1%, while the coefficient of state-owned enterprise group is not significant. The results of panel B show that the inhibitory effect of innovation is stronger and more significant in non-state-owned enterprises, which is consistent with expectations.

6.2.3. Temporal Heterogeneity

In the early stage of the implementation of the innovation driven strategy, governments at all levels will intensively introduce innovation incentive policies. Substantive innovation is risky and difficult to succeed, but in order to obtain government subsidies and tax incentives, enterprises are motivated to carry out a large number of strategic and symbolic innovation, resulting in the proliferation of low-quality innovation. Because the increase of low-quality innovation does not produce actual and effective productivity and causes serious waste of resources, the government will implement various measures to correct the negative effects of innovation incentive policies and improve the overall innovation quality later. Therefore, this paper expects that the inhibitory effect of innovation in the implementation of innovation driven strategy will be stronger in early time. Based on this, the sample is divided into an early (2013–2016) sample group and a late (2017–2020) sample group according to the time distribution. The grouping regression results are shown in panel C of Table 7. It can be seen from panel C that the coefficients of R&D are all significantly negative at the level of 1%, but the absolute value of the coefficients in the early sample group is greater (0.0239 > 0.0178). Patent’s coefficients are all negative, of which the coefficient of the early sample group is significant at the level of 1%, but the coefficient of the late sample group is not significant. The results of panel C show that the inhibitory effect of innovation is strong in the early stage of the implementation of innovation driven strategy, which is also consistent with expectations. At the same time, the results of panel C also reflect that China’s overall innovation structure tends to shift to high-quality innovation, and the effectiveness of the transformation of innovation driven economic development mode may appear later.

6.3. Is It Rational for Enterprises to Carry Out Low-Quality Innovation?

The previous research has proved that low-quality innovation is not conducive to the improvement of enterprise’s productivity. Do managers who promote enterprise innovation engage in low-quality innovation activities based on the motivation of maximizing personal interests? In order to test this possibility, this paper sets the following model:

$$Salar{y}_{i,t}={\alpha }_0+{\alpha }_{1}Innovatio{n}_{i,t}+\Sigma Control{s}_{i,t}+{\epsilon }_{i,t}$$

(18)

where Salary is executive compensation (the natural logarithm of the sum of executive compensation), and the regression results of this model are shown in

Table 8. Test of rationality of low-quality innovation behavior.

	(1)	(2)	(3)
	Salary	Salary	Salary
Patent	0.0114 **
	(0.0056)
L-Patent		0.0099 *
		(0.0058)
H-Patent			0.0130 **
			(0.0062)
Controls	Yes	Yes	Yes
Year&Indcd	Yes	Yes	Yes
N	12,007	12,007	12,007
R2	0.3619	0.3618	0.3619

Note: Robust standard errors are shown in brackets, ** and * are significant at the level of 5% and 10% respectively.

.

The results in column (1)–(3) show that innovation has a positive impact on executive compensation, but the promotion effect of high-quality innovation on executive compensation is stronger than that of low-quality innovation. The results show that it is an irrational and short-sighted behavior for enterprises to engage in low-quality innovation in order to cater to government regulation and grab policy benefits. Low-quality innovation can neither optimize self-interest nor provide sustainable productivity for enterprises, while high-quality innovation can achieve the “win-win” goal.

7. Discussion

The results of the empirical estimation show that innovation inhibits the improvement of total factor productivity of enterprises and hinders the high-quality development of enterprises overall. Therefore, the transformation of China’s economic development mode driven by innovation has not achieved actual results. Overall, we found that innovation has a restraining effect on the total factor productivity of enterprises. This finding is in contrast with previous studies on innovation promoting the productivity of enterprises (such as [16,19]). The possible reason is that these studies used a model specification and methodology different from the approach used in this paper and did not incorporate the impact of innovation structure. However, our results are also consistent with some previous findings. For example, the author of [38] found that China’s domestic R&D did not play its due role in promoting the improvement of TFP and even hindered it. The author intuitively believes that the low intensity of basic research in China, combining with the policy of market for technology, led to this result. The author of [39] found an innovation mystery in the process of China’s economic development: the investment in technological innovation continues to increase, but the growth of total factor productivity continues to decline. The author attributes the above findings to the fact that the change of the combination relationship of Government-Market in China has led to the intensification of the market distortion between different factors such as labor, capital and land. As the factor marketization is blocked, the mere increase of investment in technological innovation may not lead to the continuous improvement of TFP. The author of [63] points out that the distortion of the factor market leads to the factor mismatch among enterprises, and this distorts the decision-making of enterprises’ innovation and withdrawal from the market. The factor mismatch effect is greater than the factor reset effect, and it offsets the factor reset effect of enterprise innovation, thus reducing the total factor productivity. These studies have analyzed the possible causes of relevant problems to some extent, but the impact of innovation structure may better explain such problems.

Another important result of this study is that the mismatch of resources caused by low-quality innovation causes a serious loss of production efficiency, eroding the promotion of high-quality innovation, so that innovation has a negative effect of inhibiting the improvement of total factor productivity overall. In a word, we found the inhibition effect of low-quality innovation and the promotion effect of high-quality innovation. This conclusion is logically consistent with some studies focusing on the effects of heterogeneous innovation. For example, the research of Ran and Zheng (2021) [64] found that the increase in the number of patents has no significant impact on the high-quality development of regional economy, while the innovation ability represented by technological complexity can significantly promote the high-quality development of regional economy. Jiang and Wang (2015) [65] found that in the comparison of different types of R&D activities, the effect of experimental development is the strongest, followed by basic research and finally applied research. Zhao et al. (2022) [66] found that the transition from utilization innovation to exploratory innovation negatively affects enterprise failure, while the transition from exploratory innovation to utilization innovation has no significant impact on enterprise failure. Li and Zheng (2016) [24] pointed out that invention patents, as the “high-quality” innovation achievements, can bring long-term benefits to enterprises, while the non-invention patents’ original purpose is to obtain other benefits, and itself does not increase the value of enterprises. The above studies have found that heterogeneous innovation has different effects, and our study also uses this logic to explore the internal reasons for the overall inhibitory effect of innovation in China.

In addition to the above findings, this study also explored the different performance of innovation’s inhibitory effect in different time, region and equity samples, mainly because they have different tendencies to carry out low-quality innovation. We found that in the early stage of the implementation of China’s innovation driven strategy, the inhibitory effect of innovation was strong. Tang et al. (2014) [67] pointed out that in the early stage, China’s technological innovation knowledge accumulation was insufficient and the foundation of technological innovation was also weak, so they failed to promote the growth of total factor productivity. These reasons can partly explain our results. We found that the inhibition effect of enterprise innovation in eastern China is stronger than that in central and western China, which is consistent with Yan et al. (2021) [68] who found that there are obvious regional differences in the pollution reduction effect of technological innovation in eastern, central and western China. We also found that the innovation inhibition effect of non-state-owned enterprises is stronger than that of state-owned enterprises, that is, non-state-owned enterprises are more inclined to carry out low-quality innovation, but this finding is in sharp contrast to what Li and Zheng (2016) [24] found (when enterprises are stimulated by industrial policies, state-owned enterprises’ patent applications increase significantly than that of non-state-owned enterprises, especially for the increase of non-invention patents). They believe that the state-owned enterprises are to cater to policies and the government by carrying out strategic innovation, rather than to improve the “quality” of innovation. Non-state-owned enterprises face fierce market competition. In order to win in the market, they will pay attention to improving the “quality” of innovation and will not blindly innovate to cater to policies and obtain government’s support. However, on the contrary, we just believe that state-owned enterprises themselves are inextricably linked with the government, and they have no motivation to seek government’s support, while non-state-owned enterprises need to cater to policies and capture the government. After all, it is well known that “He who gets close to a good tree will have a good shade”. In addition, this study also found that high-quality innovation has a stronger effect on the promotion of executive compensation than low-quality innovation, and low-quality innovation is a kind of irrational and short-sighted behavior. Existing studies generally discuss the crowding out effect of executive compensation on enterprise innovation by explore the impact of executive compensation on enterprise innovation (such as Lu and Zhou (2022) [69]). However, arguing the impact of innovation on executive compensation may effectively encourage the executives to carry out high-quality innovation on the practical level, which is of great practical significance.

8. Conclusions and Insights

At present, the basic contradictions of the society in China have changed, and the contradictions and problems in development are mainly reflected in the quality of development. Therefore, in the new era and new stage, China should still take promoting high-quality development as the theme. Innovation is the first driving force leading development, so adhering to innovation to promote high-quality development is the core of the overall situation of China’s modernization. Taking the A-share listed companies in Shanghai and Shenzhen from 2013 to 2020 as the research object, this paper discusses the impact of innovation on high-quality development at the enterprise level, and also evaluates the effectiveness of the transformation of China’s innovation driven economic development mode. It concludes that the innovation has an inhibitory effect on enterprise’s total factor productivity, that is, innovation hinders the high-quality development of enterprises on the whole. This result also reflects that China’s innovation driven economic development mode transformation has not achieved actual effects, mainly because China is still in the dilemma of low-quality innovation occupying the dominant position of enterprise innovation. The mismatch of resources caused by low-quality innovation leads to a serious loss of production efficiency and erodes the promotion of high-quality innovation, so that innovation shows inhibition on the whole. This inhibitory effect of innovation is stronger in enterprises in the eastern region of China, non-state-owned enterprises and the early implementation of innovation driven strategy, which tend to engage in low-quality innovation. Further research also found that the promotion effect of high-quality innovation on executive compensation is stronger than low-quality innovation, and low-quality innovation is an irrational and short-sighted behavior.

This paper not only provides micro level incremental research evidence for the research about the impact of innovation on high-quality development, but it also evaluates the effectiveness of innovation driven economic development mode transformation. In addition, this paper also has some insights: first, the strong incentive of capital policies in the early stage lead to the proliferation of low-quality innovation, which shows that there are problems in our innovation management. Subsequent decisions should highlight the transformation orientation, establish the concept of quality first and understand that transformation can realize innovation value, forcing the optimization and improvement of innovation work; improve the assessment and evaluation mechanism, establish the management process of pre-evaluation, in-process monitoring and post-evaluation, reducing ineffective and low-quality innovation activities; optimize the innovation incentive policy, reduce and gradually cancel the subsidy and reward in the early stage of innovation activities, and improve the reward of transformation income distribution in the later stage of innovation; strengthen the construction of specialized institutions and talent teams, effectively improve the quality of scientific and technological innovation achievements, and promote the efficient transformation of innovation achievements into real productivity, so as to accelerate the high-quality development of the economy. Second, although the policy correction has reduced the number of low-quality innovation, the low-quality biased innovation structure has not changed, indicating that the increment of high-quality innovation is not enough, and it is difficult to form a good ecosystem of effective supply of high-quality innovation only relying on external regulation. After all, innovation depends on people to promote, and the effective supply of high-quality innovation depends on the entrepreneur as the innovation subject. Based on the “win-win” characteristics of high-quality innovation, we can stimulate the vitality of entrepreneurs as innovation explorers, organizers and leaders through more and more optimized equity incentive mechanisms, making them to carry out more and higher-quality innovation activities. This should speed up the innovation structural change of China by high-quality innovation and promote the continuous transformation of innovation achievements into real productivity, which in turn should realize the transformation and upgrade industrial structure and high-quality economic development.

This article also has the following shortcomings: First, enterprise total factor productivity can measure the high-quality development of enterprises to a certain extent, but it cannot cover the whole connotation of high-quality development of enterprises, which will be an important direction for the author to build a comprehensive index in the next step. Second, using the sample data of listed companies to assess the impact of innovation driven economic development mode transformation has limitations, because the results cannot fully reflect all the effects; after all, some unlisted small and medium enterprises accounted for more than 50% of the country’s tax revenue, more than 60% of GDP, more than 70% of the technical innovation and more than 80% of the labor employment, and their impact matters.