Research on the Impact of Supply Chain ESG on Enterprises’ Green Innovation Performance

First of all, according to different channels, we summarize the high-frequency green innovation performance evaluation indexes statistically, and combined with the characteristics of the logistics industry, eliminate the missing values of the data. After data cleaning, a total of 22 indicators with more than 20 citations were screened out, and an evaluation system of enterprise green innovation performance indicators with 4 guideline levels and 23 indicators was established on the basis of the quantizability of the indicators after the initial screening, and the specific indicators are shown in Table 1.

Due to the different units and meanings of the original data of the indicators, the growth rate of the indicators of some enterprises has a negative value, in order to eliminate the influence of the outline and order of magnitude of the different indicators on the evaluation results, this paper firstly carries out the dimensionless processing of the data of the indicators, and then uses entropy value method to calculate the weights of indicators of each level, and the steps of its calculation are as follows: (1) Wj=(1−Hj)/(n−∑i=1nHj) \[{{W}_{j}}=(1-{{H}_{j}})/(n-\underset{i=1}{\overset{n}{\mathop \sum }}\,{{H}_{j}})\]] (2) Hj=−1lnm∑j=1mfij⋅lnfij \[{{H}_{j}}=-\frac{1}{\ln m}\underset{j=1}{\overset{m}{\mathop \sum }}\,{{f}_{ij}}\cdot \ln {{f}_{ij}}\]] (3) fij=Aij/∑j−1nAij \[{{f}_{ij}}={{A}_{ij}}/\underset{j-1}{\overset{n}{\mathop \sum }}\,{{A}_{ij}}\]]

Where, m is the evaluation enterprise, n is the evaluation index, W_j is the weight value of the j th index, H_j is the entropy of the jth index, and A_ij is the standardized data matrix.

Compared with the subjective scoring and assignment method, the TOPSIS evaluation method focuses more on the comparative observation of the comprehensive ranking between different objects, and the method is more inclusive of the evaluation indexes with a wider range, which can either completely adopt objective data to construct the evaluation system, or cut in from the subjective perspective, combining the characteristics of the industry, which is more targeted. The specific processing steps are as follows: (4) Zij=yij∑i−1myij2 \[{{Z}_{ij}}=\frac{{{y}_{ij}}}{\sqrt{\underset{i-1}{\overset{m}{\mathop \sum }}\,y_{ij}^{2}}}\]] (5) Xij=Wj⋅Zij \[{{X}_{ij}}={{W}_{j}}\cdot {{Z}_{ij}}\]]

Ideally interpreted as: (6) Ideal solutionXij={ maxxijBeneficial propertiesminxijCost Attributes Negative Ideal SolutionXij={ minxijBeneficial PropertiesmaxxijCost Attribute \[\begin{array}{*{35}{l}} Ideal~solution\,{{X}_{ij}}=\{\begin{matrix} \max {{x}_{ij}} & Beneficial~properties \\ \min {{x}_{ij}} & Cost~Attributes \\ \end{matrix} \\ Negative~Ideal~Solution\,{{X}_{ij}}=\{\begin{array}{*{35}{l}} \min {{x}_{ij}} & Beneficial~Properties \\ \max {{x}_{ij}} & Cost~Attribute \\ \end{array} \\ \end{array}\]] (7) Distance to ideal solution:di=∑j−1n(xij−minxj)2Distance to negative ideal solution:di=∑j−1n(maxxj−xij)2 \[\begin{array}{*{35}{l}} Distance~to~ideal~solution:\,{{d}_{i}}=\sqrt{\underset{j-1}{\overset{n}{\mathop \sum }}\,{{({{x}_{ij}}-\min {{x}_{j}})}^{2}}} \\ Distance~to~negative~ideal~solution:\,{{d}_{i}}=\sqrt{\underset{j-1}{\overset{n}{\mathop \sum }}\,{{(\max {{x}_{j}}-{{x}_{ij}})}^{2}}} \\ \end{array}\]] (8) Cjn=di0(di0+din) \[C_{j}^{n}=\frac{d_{i}^{0}}{(d_{i}^{0}+d_{i}^{n})}\]]

where y_ij is the dimensionless index value.

In this paper, A-share listed companies in a city from 2014 to 2023 are selected as research samples, and in order to ensure the reliability of the data, the selected samples are further processed in accordance with the following principles: all samples of listed companies labeled as ST and *ST by the CSRC are excluded, samples of listed companies in the financial and insurance sectors are excluded, and the observations with serious omissions in the data are removed; and the samples with negative debt ratios are excluded. After the above processing, 18687 valid observations are finally obtained. At the same time, in order to eliminate the possible influence of extreme values on the regression results, this paper carries out 1% shrinkage treatment for continuous variables.

To explore the relationship between performance and green innovation performance, this paper constructs the following model 1: (9) Green_IP=α0+α1ESG+α2Lev+α3Cashhld+α4Tobinq+α5Idr+α6Ec+α7Size+α8Age+∑​Year+∑​Ind+ε \[\begin{array}{*{35}{l}} Green\_IP & ={{\alpha }_{0}}+{{\alpha }_{1}}ESG+{{\alpha }_{2}}Lev+{{\alpha }_{3}}Cashhld \\ {} & +{{\alpha }_{4}}Tobinq+{{\alpha }_{5}}Idr+{{\alpha }_{6}}Ec+{{\alpha }_{7}}Size+{{\alpha }_{8}}Age \\ {} & +\mathop{\sum }^{}Year+\mathop{\sum }^{}Ind+\varepsilon \\ \end{array}\]]]

where Green_IP is the explanatory variable, representing green innovation performance, and ESG is the explanatory variable, representing ESG performance. α₁ – α_s denotes the coefficient of each control variable, Year denotes the year, Ind is the industry fixed effect, and ε is the estimated residual. If the coefficient α₁ is positive, it reflects that ESG performance promotes firms’ green innovation performance.

In order to further explore the moderating relationship between enterprise internal control on ESG performance and green innovation performance, this paper constructs model 2 on the basis of model 1: (10) Green_IP=β0+β1CL+β2CL*ESG+β3ESG+β4Lev+β5Cashhld+β6Tobinq+β7Idr+β8Ec+β9Size+β10Age+∑​Year+∑​Ind+ε \[\begin{array}{*{35}{l}} Green\_IP & ={{\beta }_{0}}+{{\beta }_{1}}CL+{{\beta }_{2}}CL*ESG+{{\beta }_{3}}ESG+{{\beta }_{4}}Lev \\ {} & +{{\beta }_{5}}Cashhld+{{\beta }_{6}}Tobinq+{{\beta }_{7}}Idr+{{\beta }_{8}}Ec+{{\beta }_{9}}Size \\ {} & +{{\beta }_{10}}Age+\mathop{\sum }^{}Year+\mathop{\sum }^{}Ind+\varepsilon \\ \end{array}\]]

The model focuses on the coefficient β₂ of the cross-multiplier term between ESG performance and firms’ internal controls, which, if negative, reflects that firms’ internal controls inhibit the interaction between ESG performance and green innovation performance.

Through the Hausman test, this paper chooses the individual time two-way fixed effects model, and the results of the benchmark regression and robustness test are shown in Table 4. In the benchmark regression, the regression coefficient for corporate ESG performance and green innovation performance is 0.421, which is significantly positively correlated at the 1% level. In this paper, the robustness test using the method of replacing explanatory variables still holds. Therefore, corporate ESG performance has a significant positive impact on the performance of green innovations, and Hypothesis 1 is confirmed.

Good ESG performance indicates that enterprises not only focus on the improvement of their own profitability, but also take the protection of the ecological environment as a development goal, actively fulfill their social responsibility, and integrate it into the corporate governance process. The enterprise implements the development concept of innovation, coordination and green, realizes the growth of sales revenue of green products, and achieves energy saving and reduction of pollutant emissions through green process technology and process improvement and other measures, which can achieve the purpose of cost reduction and efficiency enhancement, and ultimately realizes the comprehensive enhancement of the company’s economic performance, technological innovation performance and environmental performance.

Meanwhile, the regression test results show that the regression coefficient between ESG performance and green innovation performance of the larger enterprise size group is larger than that of the smaller enterprise size group (i.e., 0.409>0.311), indicating that the ESG performance of larger enterprises has a more pronounced positive impact on green innovation performance.

The mean value of ESG performance of other firms at the industry level in which the sample firms are located is selected as the instrumental variable, and the two-stage least squares 2SLS is used to conduct the endogeneity test, the results of which are shown in Table 5, where the values in square brackets are the critical values.

The results of the two-stage regression show that the F-value of the first-stage regression is much larger than 10 and significant at the 1% level, and the statistic of the second-stage Under identification test is significant at the 1% level, and the statistic of the Weak identification test is much larger than the corresponding critical value, which indicates that the instrumental variables selected in this paper are appropriate. The regression coefficient of corporate ESG performance and green innovation performance is 1.745, which is significant and positively correlated at 1% level, which is consistent with the findings of the previous study, and hypothesis 1 is verified.

In order to test whether the quality of internal control plays a moderating effect on the relationship between corporate ESG performance and green innovation performance, this paper introduces the quality of corporate internal control and the interaction term between the quality of internal control and the independent variable of corporate ESG performance into the fixed-effects model for regression analysis. The analysis results show that the regression coefficient of the interaction term of internal control quality and ESG performance on corporate green innovation is 0.372 and significant (p<0.01), indicating that the quality of corporate internal control plays a positive moderating role between ESG performance and corporate green innovation. That is, the higher the quality of corporate internal control, the more obvious the promotion effect of corporate ESG performance on green innovation. The above results also indicate that enterprises need to pay attention to the management of internal control, and good internal control has a certain role in promoting the innovation ability of enterprises. Therefore hypothesis 2 is verified.

Due to the slight differences in the level of economic development, market sophistication, and policy planning among regions, there may be some regional differences in the level of corporate green innovation. This paper further divides the sample enterprises into four regional sample groups according to the regions where they are registered, with reference to the four major economic regions of the country, and examines the extent to which the ESG performance of enterprises in the sample groups in different regions influences the green innovation of enterprises. Table 6 shows the analysis of regional heterogeneity.

Firms’ ESG performance has a significant positive effect on green innovation in the eastern region, while the positive effect is not significant in the central and western regions. Notably, corporate ESG performance has a significant negative effect on green innovation performance at the 1% level in the Northeast region. This paper argues that Northeast China is an old industrial zone, and most of the enterprises are heavy industry enterprises or large-scale manufacturing enterprises, which have a great impact on the environment, and the relevant regulations on corporate environmental protection and social responsibility in Northeast China have been issued late in recent years, and the awareness of corporate social responsibility is generally not strong.

Considering that the differences in the nature of corporate property rights will have a certain impact on both corporate social performance and corporate governance, this paper sets a dummy variable for the nature of corporate property rights (Soe), which is 1 if it is a state-owned enterprise and 0 otherwise, to further examine the effect of the nature of property rights on the ESG performance and green innovation of enterprises. Table 7 shows the analysis of property rights heterogeneity.

In the sample group of non-state-owned enterprises (Column 1), ESG performance is significantly and positively related to green innovation performance at the 1% level, while in the sample group of state-owned enterprises, it is significantly and positively related at the 5% level. It also indicates that in non-state-owned enterprises, ESG performance has a more significant contribution to corporate green innovation. The reason for the difference may be that the SOEs themselves bear strong social responsibilities and should actively implement the policies and systems promulgated by the state, so the role of ESG performance in corporate green innovation is relatively small, and the industry competitiveness of SOEs is smaller than that of non-SOEs, and investors’ ESG preferences are less binding on SOEs, which also affects the role of ESG in promoting green innovation in SOEs.

The industries of the sample firms are further categorized into two samples of high environmental sensitivity industries as well as low environmental sensitivity industries1 for regression, and the regression results are shown in Table 8.

The ESG performance of enterprises in highly environmentally sensitive industries is significantly and positively associated with corporate green innovation at the 1% level, while enterprises in low environmentally sensitive industries fail the significance test. The results indicate that the ESG performance of enterprises in highly sensitive industries has a stronger impact on corporate green innovation due to the stronger attention from social stakeholders and public media. For example, if firms in heavily polluting industries improve their ESG performance, their green innovation performance will also increase significantly.

The above analysis shows that environmental heterogeneity, such as region, property rights, and industry, also partially affects the performance of firms in green innovation.