Does Green Investment Improve the Comprehensive Performance of Enterprises? A Study on Large and Medium-Sized Steel Enterprises in China

Abstract

In recent years, enterprise green investment is becoming more and more important for improving enterprise environment and promoting enterprise development. Combined with the measurement of slack (SBM) model and the “super efficiency” model, based on the panel data of China’s large and medium-sized steel enterprises from 2009 to 2020, this paper uses the data envelopment analysis (DEA) method to construct a corresponding model to evaluate the comprehensive performance of enterprises, and further studies the impact of enterprise green investment on the comprehensive performance. The results show that: (1) Enterprise green investment has a significant positive effect on enterprise comprehensive performance; (2) Government supervision has a significant positive effect on enterprise comprehensive performance, and the influence of enterprise green investment on comprehensive performance is negatively regulated by government supervision; (3) Enterprise green investment has a heterogeneous effect in promoting comprehensive performance. In terms of scale, the promotion effect is more significant on large enterprises than medium-sized enterprises. In terms of ownership, the green investment of state-owned enterprises has no significant effect on comprehensive performance, while the green investment of private enterprises has a significant effect on the comprehensive performance. In terms of region, green investment has no significant effect on the comprehensive performance in eastern China while the green investment in the center of China and western China has a significant role in promoting comprehensive performance. The most important implication of this study is that enterprise green investment is an effective way to achieve comprehensive green transformation of enterprises.

1. Introduction

In recent years, China’s economy has developed rapidly and achieved remarkable results [1]. However, with the rapid economic growth, it also brought a series of serious environmental pollution problems, which is due to low-cost manufacturing production and lax environmental supervision [2,3,4]. According to the “Global Environmental Performance Index Report” released by Yale University in 2018, China’s air quality ranks last in the world. In 2019, a bulletin on the state of China’s ecological environment disclosed that there were 180 cities (more than half of the total number of cities) whose air quality seriously exceeded the standard; the total time of serious pollution in domestic cities is up to 1666 days, an increase of 88 days over 2018. Therefore, strengthening China’s environmental governance is a major problem to be solved urgently.

Enterprises are the main body of the national economy. While making contributions to national economic development, enterprises also consume a lot of energy and produce a lot of environmental pollution [5,6]. According to environmental statistics data, sulfur dioxide, nitrogen oxides and industrial smog emissions of enterprises across the country account for 88.15%, 67.60% and 83.65% of the total emissions, respectively, which seriously damages China’s air quality [7]. Therefore, the Chinese government encourages enterprises to reduce emissions through green production. In recent years, the emergence of green investment has attracted more and more attention of enterprises. The improvement of pollution through green investment has gradually become an important choice for enterprises [8,9,10]. However, at this stage, a large number of enterprises in China generally have the problem of insufficient green investment scale. The development level of green investment is low, green governance is relatively backward, and the effect of green investment is not ideal [11,12,13]. China’s environmental pollution problem has not been well solved, which also hinders the realization of the goal of green and high-quality development. Therefore, reasonably expanding the scale of enterprise green investment is an important measure to realize the green transformation of China’s economy.

As the largest manufacturing country, China’s share of the world’s steel production has risen sharply year by year, and the steel industry has gradually become a major industry that promotes China’s economic growth [8]. However, this result is driven by the high consumption and high emission of low-cost manufacturing industry, which also leads to many serious environmental pollution problems [14,15,16,17]. Since almost all steel production causes serious environmental pollution [18], most steel enterprises solve this problem through green investment. Firstly, green investment can enable enterprises to reduce pollutant emissions through green technology. Enterprises can produce on the premise of meeting the government’s environmental protection requirements. Secondly, enterprise green investment can enhance enterprise social responsibility, better public image can greatly promote enterprise brand reputation, and then promote enterprise financial performance; Finally, under the strict environmental protection supervision, green investment can enable enterprises to avoid environmental punishment caused by environmental problems, thereby reducing enterprise operating costs and improving enterprise comprehensive performance. Most of the existing literatures focuses on the impact of green investment on enterprise environmental governance or environmental performance [19,20,21,22]. A small number of literatures discuss the impact of green investment on enterprise financial performance [23,24], but few literatures discuss the impact of green investment on enterprise comprehensive performance (including financial and environmental aspects). Based on the DEA method of undesired output, this paper takes the pollutant emission of enterprises as undesired output and the industrial added value of enterprises as desired output to comprehensively measure the economic and environmental performance of enterprises, and further discusses the impact of green investment on comprehensive performance.

This paper has made several contributions. Firstly, based on the financial and environmental data of enterprises, this paper uses the SBM super efficiency DEA model of undesired output to measure the comprehensive performance of enterprises. Compared with the unilateral performance measurement in the previous literature, this paper considers both environmental protection and corporate profits to achieve sustainable development of enterprises. Secondly, this paper analyzes the role of government regulation based on the perspective of enterprises, which provides some experience for the government to properly adjust macro-control. Finally, based on the perspective of heterogeneity, this paper analyzes the impact of different types of enterprises’ green investment on comprehensive performance, which provides strong experience and evidence for the study of the diversity of enterprises’ sustainable development. The purpose of this paper is to provide theoretical reference for China’s steel enterprises to implement green transformation.

2. Literature Review and Research Hypotheses

2.1. Enterprise Green Investment and Enterprise Comprehensive Performance

According to Eyraud’s definition, green investment is a sustainable investment arrangement integrating environmental, economic and other social factors [25]. Many studies have shown that enterprise green investment can improve the quality of the environment and enhance their own competitiveness, and then improve the comprehensive performance of enterprises [26,27,28]. The main reasons are as follows: Firstly, through green investment, enterprises can not only reduce energy consumption, improve resource efficiency, but also seek renewable energy as an alternative energy source, which can reduce the emission of pollutants from enterprises, so as to reduce the pollution to the environment. Although some people believe that environmental protection investment may increase the cost of enterprises to some extent, which is not conducive to improving the competitiveness of enterprises, others believe that energy conservation and emission reduction is also a manifestation of improving energy efficiency and saving resource consumption, which makes enterprises have corresponding cost advantages [29,30]. Sustainable enterprise green investment can not only realize the dual growth of environment and economy [31,32], but also the enterprises that take the lead in green investment will have preconceived competitive advantages in green transformation. In the long run, enterprise green investment has an impact on the improvement of comprehensive performance. Secondly, green investment also reflects enterprise social responsibility. Enterprises save energy and protect the environment through green investment and maintain the green reputation of their brands with their own green practices, which can effectively improve the influence and competitiveness of their products in the market, thereby improving their comprehensive performance [33,34,35]. Morever, through the social responsibility of environmental protection, enterprises can meet the needs of stakeholders such as customers, consumers and the government through green investment, and obtain the support of stakeholders to improve enterprise performance [36,37]. Some research shows that enterprises with high environmental awareness tend to have higher profitability through green investments than enterprises with low environmental awareness [38]. In specific practice, green investment can not only change the operation mode of enterprises, but also reduce production costs and significantly improve the enterprise’s reputation, thereby bringing profits to enterprises. Through green investment, enterprise can incorporate environmental issues into enterprise strategies, which can help to consolidate the competition of an enterprise’s advantage [39]. Finally, by expanding the scale of green investment, enterprises can reduce the costs caused by environmental penalties [40,41], which will also lead enterprises to pay more attention to related institutions and social sustainability while focusing on the economy. Therefore, enterprises are more willing to participate in green investment [42]. Based on the analysis, we obtain the following hypothesis:

Hypothesis 1 (H1).

Enterprise green investment has a significant positive effect on enterprise comprehensive performance.

2.2. Enterprise Green Investment, Government Supervision and Enterprise Comprehensive Performance

Generally speaking, compared with strict environmental supervision, enterprise green investment in an environment of relatively weak government supervision is generally based on the spontaneous and voluntary behavior of enterprises. Although strict government supervision may have an adverse impact on the revenue of enterprises in the short term, in the long run, government supervision is conducive to regulating various misconduct of enterprises in the industry and has a positive impact on the total factor productivity of enterprises [43]. The spontaneous green investment of enterprises is not only conducive to the long-term sustainable development of enterprises, but can also release the signal of actively undertaking social responsibility, which can promote the green reputation of enterprises and improve their comprehensive performance [44].Social responsibility to meet environmental requirements is conducive to improving the income of enterprises [21], strict environmental supervision may reduce pollution emissions and improve environmental performance of enterprises [21,45,46,47]. However, the purchase of environmental protection equipment and the investment of green technology require a lot of capital and technology investment, which will make many enterprises with insufficient funds fall into a state of financial difficulty, and make enterprises face certain business risks [29].

In addition, green investment usually only creates benefits for others, but investors must bear all the costs, which will lead to the lack of sufficient motivation for enterprises to invest in new technologies to create public interest. Under high pressure, enterprise green investment has little positive impact on the improvement of enterprise performance and is only forced to meet environmental supervision [48,49,50]. Based on the analysis, we obtain the following hypothesis:

Hypothesis 2 (H2).

Government supervision has a significant positive effect on enterprise comprehensive performance, and the influence of enterprise green investment on comprehensive performance is negatively regulated by government supervision.

3. Methodology, Data and Model

3.1. Data and Sample Selection

Based on the perspective of heterogeneity, it is difficult for different enterprises to accurately measure their comprehensive performance with a unified standard based on different environments, different industry attributes and different demand markets. Therefore, taking enterprises in a fixed industry as the research object can reduce the white noise caused by different industry attributes [51]. As a pillar industry of the national economy, steel enterprises have a great impact on the environment for a long time. With the country and people paying more and more attention to environmental protection, green transformation is the inevitable development of steel enterprises. Therefore, this paper chose steel enterprises as its research sample, which is more representative.

Data is an important part of empirical research. This paper is based on the panel data of 50 large and medium-sized enterprises in China’s steel industry from 2009 to 2020. The data includes the financial information, pollutant discharge information and related production data of the enterprise, which can accurately measure the various indicators of the enterprise. The data comes from China Iron and Steel Association.

3.2. Measuring Enterprise Comprehensive Performance

3.2.1. Measuring Methods

Due to the high pollution characteristics of steel enterprises, this paper defines the comprehensive performance of enterprises as the economic performance with pollutant emission as the unexpected output. Reasonable and accurate evaluation of enterprise comprehensive performance is very important to study the conflict between enterprise financial income and environmental pollution. According to the relevant research on quantifying the comprehensive performance of high pollution enterprises [52,53], the integration of resources and environmental factors into comprehensive performance evaluation has always been the focus of academic attention.

DEA is a widely used efficiency measurement method. Especially in the evaluation of enterprise efficiency, the DEA method has great advantages over the parameter estimation method. When conducting micro enterprise research, the use of parameter estimation will lead to a large deviation in the input–output matching parameters of different enterprises, and it is impossible to use a unified standard to measure enterprise performance. The DEA method only needs to consider the input and output units, not even the unit of data, so as to avoid the error in parameter estimation. Therefore, in the research of micro enterprises, the DEA method is more commonly used than parameter estimation. According to the relevant research on the quantification of the comprehensive performance of high polluting enterprises [53], the integration of resources and environmental factors into the comprehensive performance evaluation of enterprises has always been the focus of academic attention. Due to the characteristics of high consumption, high emission and high pollution, the research on iron and steel enterprises should not only consider financial performance, but also fully consider the related problems such as resource consumption and pollutant emission. Therefore, the ordinary DEA efficiency model will no longer be applicable to such research, and the DEA model with unexpected output needs to be introduced to deal with such problems.

Based on the defects and limitations of the traditional DEA model, the (SBM) super efficiency model is gradually becoming a research hotspot [54]. The model mainly focuses on the following problems. When multiple efficiency values are 1 (DEA effective), the traditional model cannot further distinguish the efficiency difference of 1 unit, but SBM allows the efficiency value to be greater than 1, and scientific research can further accurately distinguish the unit efficiency value. Therefore, this method is also called “super efficiency” [55]. Combined with the defect that the traditional model cannot calculate the undesired output, this paper intends to combine the unexpected output model with the SBM model to form a model that can overcome two problems at the same time, as follows:

$$\min \mathsf{\rho }=\frac{1+\frac{1}{\mathrm{m}}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{A}}_{\mathrm{i}}^{-}/{\mathrm{x}}_{\mathrm{ik}}}{1-\frac{1}{{\mathrm{p}}_1+{\mathrm{p}}_2}\left({{\displaystyle \sum }}_{\mathrm{r}=1}^{{\mathrm{p}}_1}{\mathrm{A}}_{\mathrm{r}}^{+}/{\mathrm{y}}_{\mathrm{rk}}+{{\displaystyle \sum }}_{\mathrm{t}=1}^{{\mathrm{p}}_2}{\mathrm{A}}_{\mathrm{r}}^{\mathrm{b}-}/{\mathrm{b}}_{\mathrm{rk}}\right)}$$

(1)

$$\mathrm{S}.\mathrm{t}.{\displaystyle \sum }_{\mathrm{j}=1,\mathrm{j}\ne \mathrm{k}}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{ij}}{\mathsf{\lambda }}_{\mathrm{j}}-{\mathrm{A}}_{\mathrm{i}}^{-}\le {\mathrm{x}}_{\mathrm{ik}}$$

(2)

$${\displaystyle \sum }_{\mathrm{j}=1,\mathrm{j}\ne \mathrm{k}}^{\mathrm{n}}{\mathrm{y}}_{\mathrm{ij}}{\mathsf{\lambda }}_{\mathrm{j}}-{\mathrm{A}}_{\mathrm{r}}^{+}\ge {\mathrm{y}}_{\mathrm{rk}}$$

(3)

$${\displaystyle \sum }_{\mathrm{j}=1,\mathrm{j}\ne \mathrm{k}}^{\mathrm{n}}{\mathrm{b}}_{\mathrm{tj}}{\mathsf{\lambda }}_{\mathrm{j}}-{\mathrm{A}}_{\mathrm{t}}^{b-}\le {\mathrm{b}}_{\mathrm{rk}}$$

(4)

$$1-\frac{1}{{\mathrm{p}}_1+{\mathrm{p}}_2}\left({\displaystyle \sum }_{\mathrm{r}=1}^{{\mathrm{p}}_1}{\mathrm{A}}_{\mathrm{r}}^{+}/{\mathrm{y}}_{\mathrm{rk}}+{\displaystyle \sum }_{\mathrm{t}=1}^{{\mathrm{p}}_2}{\mathrm{A}}_{\mathrm{r}}^{\mathrm{b}-}/{\mathrm{b}}_{\mathrm{rk}}\right)>0$$

(5)

$${\mathsf{\lambda }}_{\mathrm{j}},{\mathrm{A}}_{\mathrm{i}}^{-},{\mathrm{A}}_{\mathrm{r}}^{+}\ge 0$$

(6)

$$\mathrm{i}=1,2,\cdots \mathrm{m};\mathrm{r}=1,2,\cdots \mathrm{p};\mathrm{j}=1,2,\cdots \mathrm{n};\left(\mathrm{j}\ne \mathrm{k}\right)$$

(7)

In model (1), each unit contains an input vector (m types of inputs), an output vector (${\mathrm{p}}_1$ types of outputs) and an undesired output vector (${\mathrm{p}}_2$ types of outputs), including the input vector is $\mathrm{x} \in { \mathrm{R}}^{\mathrm{m}}$, the expected output vector is $\mathrm{y}\in { \mathrm{R} }^{\mathrm{p}1}$, and the undesired output vector is $\mathrm{b}\in { \mathrm{R}}^{\mathrm{p}2}$. S represents the slack of input and output, ${\mathrm{A}}^{-}$ represents the redundancy of the input variables, ${\mathrm{A}}^{+}$ represents the shortage of expected output variables, ${\mathrm{A}}^{\mathrm{b}-}$ represents the excess of undesired output variables, $\lambda$ is the weight vector, and $\rho$ represents the efficiency score.

Referring to Duanmu’s practice [2], this paper intends to use the necessary inputs (such as water resource consumption, fixed assets, number of employees, and energy consumption) required by the production of iron and steel enterprises as inputs, the industrial added value of enterprises as expected outputs, and the industrial wastes (such as waste gas, waste water, and waste residue) generated in the production process of enterprises as unexpected outputs, The efficiency value calculated by DEA of SBM with unexpected output is taken as the comprehensive performance of the enterprise. The framework is shown in Figure 1.

3.2.2. Description of Measurement Variables

In this paper, fixed assets, number of employees, energy consumption and water consumption are used as input variables, industrial added value is used as expected output, and the discharged waste residue, waste gas and wastewater are selected as undesired output.

This paper considers the selection of variables in this way. Firstly, fixed assets and number of employees are the basic variables to reflect the enterprise scale and comprehensive situation, using them as input variables can directly reflect the basic situation of enterprise such as the fixed assets investment and the human resources investment. At the same time, considering the characteristics of high energy consumption in the steel industry, resource consumption and water consumption are included in the research as the resource input to calculate the comprehensive performance. Secondly, because the comprehensive performance we discussed is based on environmental protection, this paper regards the main waste emissions (industrial waste gas, wastewater and waste residue) in steel production as undesired output. Finally, this paper examines the comprehensive performance of the combination of environment and finance of steel enterprises and takes the industrial added value of the enterprise as the expected output. The results are detailed in

Table 1. Interpretation of variables.

Variables		Interpretation
Inputs	Water consumption	Total annual water consumption
	Fixed assets	the original value of the fixed assets—the accumulated depreciation
	Number of employees	Average number of employees per year
	Energy consumption	Total annual resource consumption (converted into coal consumption)
Undesired outputs	Waste residue	The total amount of waste residue discharged by enterprises in the year
	Waste gas	The total amount of waste gas discharged by enterprises in the year
	Waste water	Total amount of waste water discharged by enterprises in the year
Expected output	industrial added value	= Gross industrial output value (current price, new regulations) $-$ industrial intermediate input $+$ value-added tax payable in the current period

and

Table 2. Descriptive statistics of input and output variables.

Variables	Units	Mean	Std. Dev	Min	Max
Water consumption	Ten thousand ${\mathrm{m}}^3$	2532.23	2153.54	228.56	11,531.41
Fixed assets	Billion Yuan	224.11	212.12	7.93	1242.22
Number of employees	Thousand people	20.49	19.20	1.90	151.08
Energy consumption	Ten thousand tons	440.23	362.06	24.51	2102.23
Waste residue	Ten thousand tons	477.12	457.25	1.64	2771.31
Waste gas	Hundred million ${\mathrm{m}}^3$	1541.23	1435.21	0.06	7976.02
Waste water	Million ${\mathrm{m}}^3$	509.34	531.23	1.20	3760.80
Industrial added value	Million Yuan	6112	9.82	−4360	40,024

.

3.2.3. Measurement Results and Variable Description

In this paper, the waste gas, wastewater and waste residue of steel enterprises are regarded as undesired outputs, the SBM super-efficiency model is used to measure the comprehensive performance of the enterprise, so as to prepare for the study of the impact of enterprise green investment on the comprehensive performance of the enterprise. The detailed results are shown in

Table 3. Summary statistics.

Variables	Variable Symbol	Obs	Mean	Std. Dev	Min	Max
Comprehensive performance	Cp	600	0.5382	0.8823	0.0088	1.7923
Green investment	lnGi	600	8.8923	1.7723	3.2189	12.9023
Government supervision	lnGs	600	7.8235	0.8441	5.1878	9.8423
Enterprise scale	lnScale	600	15.2540	0.9528	11.956	17.6682
Economic density	lnEd	600	7.3921	1.2588	3.2581	10.9023
Education level	Edu	600	9.1632	0.9723	5.9300	12.6300
Total operating cost	lnTc	600	15.0346	0.9936	12.0184	17.9826
Main business income	lnMbi	600	15.2281	1.6522	3.2189	12.9816

.

The explanatory variable in this paper is the comprehensive performance of enterprises. Based on the SBM super efficiency DEA model, this study takes asset investment, labor input, energy consumption and water consumption as input units, three wastes discharge as undesired output and industrial added value as expected output. The calculated efficiency value is used to reflect the comprehensive performance of enterprises under the dual requirements of environment and finance. The higher the efficiency value, the better the comprehensive performance of the enterprise.

The explained variable is green investment. Considering that the effect of enterprise green investment has a certain lag, this paper uses the logarithm of the enterprise’s annual total environmental protection investment (unit: 10,000 yuan) lagging one period to measure the enterprise’s green investment scale.

The moderating variable is government supervision. This paper uses the logarithm of the annual sewage charges paid by enterprises and fines due to environmental pollution (unit: ten thousand yuan) to reflect the strength of government supervision.

The control variables include the following: (1) The scale of the enterprise (lnScale). lnScale refers to the logarithm of the total assets of the enterprise at the end of the year. (2) Total operating cost (lnTc). lnTc refers to the logarithm of the total cost of business operations (including selling goods and providing services). (3) Main business income (lnMbi) refers to the logarithm of the main operating income obtained by the enterprise. (4) Education level (Edu) represents the average years of education of the labor force in the region. (5) Economic density (lnEd) is the logarithm of GDP divided by regional area (calculated by province) to reflect the economic situation of the region. Table 3 shows summary statistics for all variables.

3.3. Model Building

In order to make an empirical analysis of the above hypothesis, this paper constructs relevant econometric models with reference to the practices of Cai et al. (2016) to test the relationship between enterprise green investment and enterprise comprehensive performance [56]. The models are as follows:

$${\mathrm{Cp}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{lnGi}}_{\mathrm{it}-1}+{\mathsf{\alpha }}_2{\mathrm{lnScale}}_{\mathrm{it}}+{\mathsf{\alpha }}_3{\mathrm{lnTc}}_{\mathrm{it}}+{\mathsf{\alpha }}_4{\mathrm{lnMbi}}_{\mathrm{it}}+{\mathsf{\alpha }}_5\mathrm{lnEd}+{\mathsf{\alpha }}_6{\mathrm{Edu}}_{\mathrm{it}}+{ \mathsf{\epsilon }}_{\mathrm{it}}$$

(8)

In order to further test the impact of enterprise green investment on enterprise comprehensive performance under different government supervision environments, this paper adds government supervision variables, constructs the interaction term between government supervision and enterprise green investment to examine the moderating relationship between enterprise green investment and enterprise comprehensive performance by government supervision.

$${\mathrm{Cp}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{lnGi}}_{\mathrm{it}-1}+{\mathsf{\beta }\mathrm{lnGs}}_{\mathrm{it}}+{\mathsf{\gamma }\mathrm{lnGi}}_{\mathrm{it}-1}{\ast \mathrm{lnGs}}_{\mathrm{it}}+{\mathsf{\alpha }}_2{\mathrm{lnScale}}_{\mathrm{it}}+{\mathsf{\alpha }}_3{\mathrm{lnTc}}_{\mathrm{it}}+{\mathsf{\alpha }}_4{\mathrm{lnMbi}}_{\mathrm{it}}+{\mathsf{\alpha }}_5\mathrm{lnEd}+{\mathsf{\alpha }}_6{\mathrm{Edu}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(9)

where i represents the enterprise and t represents the time (2009–2020). ${\alpha }_0~{\alpha }_5$, are parameters to be estimated. ${\mathsf{\epsilon }}_{\mathrm{it}}$ is the random disturbance terms.

4. Discussion and Validation of Results

4.1. Basic Test

Since panel data are used in this study, this paper intends to use the fixed effect model to test the relationship between green investment and enterprise comprehensive performance. The regression results are shown in

Table 4. Basic test.

Cp	Static Panel
	Model I	Model II	Model III	Model IV
lnGi-1	0.0301 **	0.0211 **	0.0099 ***	0.0218 **
	(2.28)	(2.02)	(2.82)	(2.34)
lnGs			0.2253 *	0.2298 **
			(1.74)	(2.16)
lnGi-1*ln Gs				−0.0337 ***
				(−3.10)
lnScale		1.8012 ***	1.8631 **	1.8279 ***
		(3.39)	(2.41)	(3.12)
lnEd		0.6367 **	0.8271 *	0.7663 *
		(2.31)	(1.93)	(1.83)
Edu		0.0327 *	0.0442	0.0599
		(1.70)	(1.48)	(1.61)
lnToc		−0.5662 ***	−0.5023 ***	−0.6081 ***
		(−3.40)	(−3.64)	(−3.47)
lnMbi		0.3327 ***	0.4270 ***	0.5327 ***
		(3.82)	(3.31)	(3.34)
Constant	0.5235 ***	−1.2712 **	−0.6612 *	−0.8012 **
	(4.02)	(−2.41)	(−1.90)	(−2.61)
Hausman test	17.03	24.41	20.60	22.15
	(0.0000)	(0.0000)	(0.0001)	(0.0000)
Control time	Yes	Yes	Yes	Yes
Control region	Yes	Yes	Yes	Yes
Model selection	FE	FE	FE	FE
${\mathrm{R}}^2$	0.3233	0.3254	0.3398	0.3762

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

Table 4 shows that the coefficients of enterprise green investment in model I-IV are significantly positive, indicating that enterprise green investment can significantly promote the comprehensive performance of enterprises, and hypothesis H1 is verified; The results of model III and model IV show that the government supervision coefficient is significantly positive at the statistical level of 10%, the interaction term between government supervision and green investment is significantly negative, indicating that government supervision can promote comprehensive performance of enterprises. Moreover, the impact of enterprise green investment on enterprise comprehensive performance is negatively regulated by government supervision, that is, in the environment of weak supervision, the promotion effect of enterprise green investment on enterprise comprehensive performance is more significant than that in the environment with strong supervision, which verifies the hypothesis H2 to a certain extent.

4.2. Robust Test

In order to test the robustness of the basic test, this paper uses the maximum likelihood estimation (MLE) for robustness test; the regression results are shown in

Table 5. Robust test1.

Cp	Mle
	Model I	Model II	Model III	Model IV
lnGi-1	0.0362 **	0.0301 ***	0.0188 ***	0.0381 **
	(2.58)	(3.21)	(3.04)	(2.62)
lnGs			0.2892 **	0.2901 ***
			(2.21)	(2.88)
lnGi-1*ln Gs				−0.0551 **
				(−2.47)
lnScale		1.3325 ***	1.4625 **	1.4191 ***
		(3.06)	(2.47)	(3.21)
lnEd		0.3312 *	0.4619	0.4226 *
		(1.80)	(1.45)	(1.77)
Edu		0.0613	0.0829	0.1002
		(1.30)	(1.50)	(1.35)
lnToc		−0.3226 ***	−0. 3183 ***	−0.3445 ***
		(−2.81)	(−3.22)	(−3.01)
lnMbi		0.4417 ***	0.4036 ***	0.3488 ***
		(3.04)	(2.88)	(3.17)
Constant	0.7228 *	−0.8835 *	−0.8056 *	−0.4611 *
	(1.70)	(−1.90)	(−1.82)	(−1.84)
Control time	Yes	Yes	Yes	Yes
Control region	Yes	Yes	Yes	Yes
LR chi	221.24	308.16	228.26	333.52
${\mathrm{R}}^2$	0.3661	0.3383	0.3811	0.3724

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

In this paper, the maximum likelihood estimation method is used to test the robustness, and the time effect and regional effect of the original model are controlled at the same time. The regression results are basically consistent with the basic test. The relationship among green investment, government supervision and comprehensive performance is basically consistent with the basic test (including moderating effect)). Except for individual variables, most of the control variables are consistent with the basic test results, which also verifies the robustness of the model to a certain extent.

In order to test the robustness of the basic test, this paper uses the data of enterprises’ green investment lagging behind two periods for the robustness test. The regression results are shown in

Table 6. Robust test2.

Cp	Mle
	Model I	Model II	Model III	Model IV
lnGi-2	0.0371 **	0.0402 **	0.0277 ***	0.0539 ***
	(2.20)	(2.37)	(3.06)	(2.88)
lnGs			0.2029 *	0.1791 **
			(1.73)	(2.36)
lnGi-2* lnGs				−0.0381 *
				(−1.90)
lnScale		0.6145 **	0.7729 **	0.8036 ***
		(2.02)	(2.36)	(2.81)
lnEd		0.2639 *	0.3711 *	0.3159 *
		(1.76)	(1.82)	(1.79)
Edu		0.0629 *	0.811 *	0.0808
		(1.88)	(1.70)	(1.58)
lnToc		−0.2419 ***	−0.3095 ***	−0.3376 ***
		(−3.60)	(−3.11)	(−3.37)
lnMbi		0.2663 **	0.2319 ***	0.2762 ***
		(2.16)	(3.00)	(3.21)
Constant	06619 ***	0.5278 **	0.6032 ***	0.7714 ***
	(3.21)	(2.42)	(3.17)	(2.80)
Control time	Yes	Yes	Yes	Yes
Control region	Yes	Yes	Yes	Yes
LR chi	211.37	240.28	227.01	249.12
${\mathrm{R}}^2$	0.3282	0.3001	0.3127	0.2812

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

This paper tests the robustness of the model with the enterprise green investment data lagging behind the second period and controls the time effect and regional effect of the original model. The regression results are basically consistent with the basic test. The relationship among green investment, government supervision and comprehensive performance is basically consistent with the basic test (including moderating effect). Except for individual variables, most control variables are consistent with the basic test results, which also verifies the robustness of the model to a certain extent.

4.3. Dynamic Panel Test

Static panel data model is easy to ignore the impact of cultural environment, system and other factors which also play an important role on comprehensive performance. At the same time, considering the possible endogenous problems in the model variables, that is, the causal relationship between the explained variables and the explanatory variables, the potential sample selectivity deviation and the missing variable deviation, this paper introduces the first-order lag term of comprehensive performance to construct the dynamic panel model, and uses the system GMM to correct some errors in the results of the static panel model. The regression results are shown in

Table 7. Dynamic panel test.

Cp	Systerm GMM
	Model I	Model II	Model III	Model IV
Cp-1	0.3518 **	02284 ***	0.2491 ***	0.3004 ***
	(2.21)	(2.90)	(3.01)	(3.41)
lnGi-1	0.2835 **	0.2612 *	0.2462 *	0.3015 *
	(2.26)	(1.92)	(1.84)	(1.78)
lnGs			0.1519	0.2512
			(1.40)	(1.52)
lnGi-1*ln Gs				−0.0619
				(−1.50)
lnScale		0.5018 *	0.3996 **	0.5501 *
		(1.75)	(2.21)	(1.89)
lnEd		0.2382	0.3091	0.3318
		(1.28)	(1.07)	(1.43)
Edu		0.0771	0.0619	0.0338
		(1.21)	(0.81)	(1.29)
lnToc		−0.1826 ***	−0.2009 **	−0.2126 **
		(−3.01)	(−2.32)	(−2.58)
lnMbi		0.3072 **	0.3119 ***	0.3218 ***
		(2.61)	(3.29)	(3.11)
Constant	0.5881	0.6193 *	0.5088	0.5952 **
	(1.19)	(1.92)	(1.44)	(2.47)
AR1	0.0182	0.0331	0.0401	0.0319
AR2	0.0602	0.0681	0.0811	0.0599
Sargan	42.1163	55.2552	38.3766	40.117
P	0.0661	0.0581	0.0681	0.0639

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

As shown in Table 7, the p values of AR (2) in model I-model IV are greater than 0.05, indicating that the model has no second-order autocorrelation and the random interference term has no sequence correlation. The sargan test values for judging the excessive identification of tool variables are greater than 0.05, indicating that the tool variables selected by the model are effective. The results show that the dynamic panel regression results are generally consistent with the basic test, in which the lag term of comprehensive performance has a significant positive impact on it. Green investment and government supervision have a significant positive impact on comprehensive performance, and the impact of green investment on comprehensive performance is significantly negatively regulated by government supervision.

4.4. Heterogeneity Test

4.4.1. Scale Heterogeneity Test

Due to the different scale of enterprises, having different financial situations, environmental protection concepts and marketing, this paper further studies the impact of green investment on comprehensive performance under different enterprise sizes. The selected samples are divided into large and medium-sized enterprises according to enterprise size. The regression results are shown in

Table 8. Scale heterogeneity test.

Cp	Large Enterprises		Medium-Sized Enterprises
	Model I	Model II	Model I	Model II
lnGi-1	0.1862 ***	0.1557 ***	0.3081	0.3936 *
	(3.01)	(2.84)	(1.53)	(1.74)
lnGs		0.0581 **		0.0660 *
		(2.21)		(1.73)
lnGi-1*ln Gs		−0.0718		−0.0839 ***
		(−1.49)		(−2.81)
lnScale	0.6881 ***	0.5095 ***	0.4289 **	0.3398 **
	(3.46)	(3.11)	(2.44)	(2.61)
lnEd	0.3371 **	0.3294 *	0.2581 *	0.3027 *
	(2.12)	(1.78)	(1.86)	(1.80)
Edu	0.0409	0.0612	0.0498	0.0636
	(1.41)	(1.20)	(1.55)	(1.14)
lnToc	−0.3889 ***	−0.3151 ***	−0.3443 ***	−0.2996 ***
	(−3.31)	(−3.02)	(−3.13)	(−3.35)
lnMbi	0.3971 ***	0.2884 ***	0.4219 ***	0.3994 ***
	(3.42)	(3.71)	(3.02)	(3.12)
Constant	−0.6182 **	−0.4619 **	−0.6012 *	−0.5081 *
	(−2.00)	(−2.31)	(−1.84)	(−1.92)
Hausman test	24.39	22.81	20.18	19.96
	(0.0000)	(0.0000)	(0.0001)	(0.0000)
Control time	Yes	Yes	Yes	Yes
Control region	Yes	Yes	Yes	Yes
Model selection	FE	FE	FE	FE
${\mathrm{R}}^2$	0.3617	0.3552	0.2991	0.2869

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

The results in Table 8 show that, firstly, the impact of green investment of large enterprises on comprehensive performance is more significant than that of medium-sized enterprises; secondly, government supervision has a significant promoting effect on both large and medium-sized enterprises; Finally, the moderating effect of government supervision is only significant for medium-sized enterprises, but not significant for large enterprises.

4.4.2. Ownership Heterogeneity Test

Since enterprises with different ownership often make inconsistent decisions, this paper further studies the impact of green investment on comprehensive performance under different ownership. The selected samples are divided into state-owned enterprises and private enterprises. The regression results are shown in

Table 9. Ownership heterogeneity test.

Cp	State-Owned Enterprise		Private Enterprise
	Model I	Model II	Model I	Model II
lnGi-1	0.3183	0.4227	0.3129 ***	0.3002 ***
	(1.41)	(1.29)	(3.81)	(3.39)
lnGs		0.0509 *		0.0731 **
		(1.70)		(2.02)
lnGi-1*ln Gs		−0.0437		−0.0734 ***
		(−1.28)		(−3.06)
lnScale	0.3072 *	0.2815 *	0.3710 **	0.3291 ***
	(1.82)	(1.79)	(2.47)	(3.11)
lnEd	0.2821 **	0.3419 *	0.3006 **	0.3813
	(2.49)	(1.78)	(2.21)	(1.52)
Edu	0.0499	0.0406	0.0511 *	0.0329
	(1.41)	(1.56)	(1.82)	(1.46)
lnToc	−0.2619 ***	−0.2891 ***	−0.3389 ***	−0.4211 ***
	(−2.81)	(−3.19)	(−2.72)	(−3.03)
lnMbi	0.3916 ***	0.3041 ***	0.6283 ***	0.4594 ***
	(3.36)	(3.12)	(2.78)	(3.29)
Constant	−0.8813 **	−0.4571 *	0.5884	−0.3901 **
	(−2.43)	(−1.93)	(1.49)	(−2.30)
Hausman test	20.31	21.79	18.62	22.81
	(0.0006)	(0.0001)	(0.0000)	(0.0000)
Control time	Yes	Yes	Yes	Yes
Control region	Yes	Yes	Yes	Yes
Model selection	FE	FE	FE	FE
${\mathrm{R}}^2$	0.2889	0.3017	0.3628	0.3611

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

Table 9 shows that, firstly, the green investment of state-owned enterprises have no significant effect on the comprehensive performance, while the green investment of private enterprises have a significant role in promoting the comprehensive performance. Secondly, government supervision has a significant promoting effect on both state-owned enterprises and private enterprises. Finally, the moderating effect of government supervision is only significant for private enterprises, but not significant for state-owned enterprises.

4.4.3. Regional Heterogeneity Test

Due to the differences in economic conditions, factor endowments, labor conditions and market mechanism in various regions, the impact of green investment on comprehensive perfornance is also unbalanced. the nationwide regression test is likely to ignore some regional characteristics. Therefore, it is necessary to subregions to further examine its heterogeneity characteristics. The selected samples are divided into east, center and west; the regression results are shown in

Table 10. Regional heterogeneity test.

Cp	East		Center		West
	Model I	Model II	Model I	Model II	Model I	Model II
lnGi-1	0.3061	0.3624	0.2847 *	0.3068 **	0.3491 ***	0.3814 ***
	(1.30)	(1.21)	(1.89)	(2.52)	(3.09)	(2.81)
lnGs		0.0636		0.0419		0.3381 **
		(1.52)		(0.98)		(217)
lnGi-1*ln Gs		−0.2638		−0.0712		−0.0626 ***
		(−1.33)		(−1.42)		(3.26)
lnScale	0.3016 *	0.3142 **	0.3619 **	0.3511 *	0.4217 **	0.4022 **
	(1.90)	(2.37)	(2.21)	(1.82)	(2.51)	(2.29)
lnEd	0.3611 *	0.4228 *	0.4047 **	0.3615 *	0.3393 **	0.4072 **
	(1.90)	(1.81)	(2.06)	(1.80)	(2.28)	(2.52)
Edu	0.3179	0.3038	0.1728	0.1571	0.2414	0.2073
	(1.31)	(1.42)	(1.45)	(1.32)	(1.43)	(1.56)
lnToc	−0.3218 ***	−0.2825 ***	−0.3513 ***	−0.3262 ***	−0.3472 ***	−0.4162 **
	(−3.20)	(−3.47)	(−3.62)	(−3.37)	(−4.12)	(−2.41)
lnMbi	0.3252 ***	0.3128 ***	0.4032 ***	0.3417 ***	0.3771 ***	0.4035 ***
	(3.24)	(3.41)	(3.34)	(3.13)	(3.34)	(3.52)
Constant	−0.4192 *	0.4533	−0.3812	−0.2915 *	−0.2208 **	−0.3417
	(−1.73)	(1.51)	(−1.34)	(−1.78)	(−2.16)	(1.16)
Hausman test	20.34	19.61	22.52	20.27	17.32	20.21
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Control time	Yes	Yes	Yes	Yes	Yes	Yes
Control region	Yes	Yes	Yes	Yes	Yes	Yes
Model selection	FE	FE	FE	FE	FE	FE
${\mathrm{R}}^2$	0.2905	0.3011	0.2731	0.3215	0.3098	0.3171

Note: *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively, and the numbers in brackets are “t” of the estimated coefficients.

.

Table 10 shows that the promotion of green investment on comprehensive perfornance is significant in the central and western regions, but not significant in the east region.

5. Final Remarks

5.1. Ansvering the Research Question

As for the performance measurement of environmental protection, most of the existing studies are based on the DEA method. Chariri estimates the financial performance of enterprises through the DEA method, and further studies the impact of green investment on financial performance of enterprises [37]; Jiang et al. calculate the innovative performance of enterprises through DEA, and studies the impact of mandatory environmental regulation on innovative performance [9]; Long et al. measures the economic and environmental performance of enterprises based on DEA, and excavates the factors to improve the economic and environmental performance of enterprises by studying the innovation behavior of enterprises [19]. Trumpp and Guenther (2017) measure the environmental performance and financial performance of enterprises through the DEA method, and further study the inverse U between the two [16]. Therefore, this paper takes the steel enterprises with high energy consumption and high emission as the sample to study the impact of enterprise green investment on comprehensive performance.

Firstly, this study empirically tests the impact of enterprise green investment on enterprise comprehensive perfornance. It is found that enterprise green investment can significantly promote enterprise comprehensive perfornance, which is basically consistent with the research conclusions. The purpose of enterprise green investment is to improve the environment without affecting or even promoting the financial performance of enterprises [31]. On the one hand, green investment invests a lot of funds and resources into the purchase of environmental protection and energy-saving equipment and the development of green technology. By reducing enterprise energy consumption, improving resource utilization efficiency and seeking renewable energy, green investment can effectively reduce pollutant emissions and improve environmental quality [21]. On the other hand, although the green investment of enterprises may increase the operating costs of enterprises due to non-profit in the short term, the benefits of green investment to enterprise development are obvious in the long run. First of all, green investment can help enterprises avoid the government’s punishment on pollution emissions and avoid damage to the reputation of enterprises [40]. Secondly, green investment can release the signal that enterprises are keen to bear social responsibility and bring good reputation to enterprises [56,57]. Finally, under the increasingly strict environmental supervision, enterprise green investment is more sustainable than ordinary investment, which is conducive to the long-term development of enterprises.

Secondly, the government controls the pollutant emission of enterprises through the supervision and environmental protection system, so as to improve their environmental protection effect. However, too strict government supervision will also cause some enterprises to damage their financial profits because they cannot bear a lot of environmental protection costs and will damage their comprehensive performance to a certain extent. On the one hand, some studies show that government supervision can significantly improve the green investment of enterprises. Strict government supervision has prompted a large number of enterprises to increase green investment, reduce the damage to the environment through energy conservation and emission reduction and the development of alternative new energy, and then improve the environmental protection performance of enterprises [58]. On the other hand, too strict government supervision makes a large number of enterprises forced to accept green investment, but green investment is difficult to generate profits in the short term, resulting in a significant increase in enterprise production costs (especially small and medium-sized enterprises). In this case, even if the enterprise environment is improved, the financial profits are reduced or even bankrupt, and the comprehensive performance of the enterprise has not been fundamentally improved [59].

Finally, this study further investigates the heterogeneity of samples in different ranges and categories. Based on the heterogeneity of scale, the impact of large enterprises’ green investment on comprehensive performance is more significant than that of medium-sized enterprises, but the moderating effect of government supervision is only significant for medium-sized enterprises. The main reasons are as follows: on one hand, green investment (especially green technology and new energy development) requires a lot of capital investment. Large enterprises have higher efficiency and more obvious effect than small and medium-sized enterprises with their own scale advantages and sufficient funds. Moreover, large enterprises (especially the high energy consumption and high emission enterprises, such as the iron and steel industry) produce more pollutants than small and medium-sized enterprises. As an investment way of energy conservation and emission reduction, green investment has a more obvious effect in enterprises with more pollutants. On the other hand, in order to avoid environmental punishment and gain higher social reputation, large enterprises often spontaneously and voluntarily improve the environment through green investment, while small and medium-sized enterprises often make green investment to meet national policies due to capital shortage and outdated technology, especially when government supervision or environmental protection policies become more stringent, Small and medium-sized enterprises are forced to make a lot of green investment, which is likely to lead to losses and even bankruptcy, the enterprise comprehensive performance has not been improved and may even be reduced. Based on the heterogeneity of ownership, private enterprises green investment can promote comprehensive performance, while state-owned enterprises do not. The moderating effect of government supervision is only significant for private enterprises, but not significant for state-owned enterprises. The main reason is that state-owned enterprises not only play a role in driving economic development in China, but also many undertake the function of leading national policies and macro-control. Therefore, the implementation of green investment in state-owned enterprises is not only from the perspective of enterprise development, but also from the perspective of complying with national policies, and the effect is not significant in a short period of time. The implementation of green investment by private enterprises should be considered more based on their own actual conditions. If it is not too strict in terms of environmental protection policy restrictions, the green investment of private enterprise is based on the premise of promoting their own development, and the impact on comprehensive performance is relatively significant. Based on the heterogeneity of region, the impact of green investment on comprehensive perfornance is significant in the central and western regions, but not significant in the east region. The main reason is that the eastern region has a good economic environment, and the emission of pollutants is more serious than that in the central and western regions. Under the high-intensity environmental regulation in eastern China, most enterprises comply with the strict environmental policy, which has a great negative impact on the financial performance of enterprises. Therefore, the effect of green investment is not significant. While the environmental foundation of the central and western regions is relatively good, the government regulation and environmental regulation are relatively loose; most enterprises’ green investment are based on their own development. The promotion relationship of comprehensive performance is significant.

5.2. Conclusions

Based on the panel data of 50 large and medium-sized steel enterprises in China from 2009 to 2020, this paper constructs static and dynamic panel models, respectively, for regression analysis and robustness tests, and further tests the moderating effect and the heterogeneity. The results show that:

Firstly, enterprise green investment in steel industry has a significant positive effect on comprehensive performance as a whole. Secondly, government supervision has a significant positive effect on comprehensive performance in the steel industry, and the relationship between green investment and comprehensive performance is negatively affected by government supervision. Thirdly, based on the heterogeneity of scale, the promotion of large enterprises green investment on comprehensive performance is more significant than that of medium-sized enterprises, and the moderating effect of government supervision is only significant for medium-sized enterprises; based on the heterogeneity of ownership, private enterprises green investment can promote comprehensive performance, while state-owned enterprises cannot; based on the heterogeneity of region, the promotion of green investment on comprehensive performance is significant in the central and western regions, but not significant in the east region.

5.3. Recommendations

This paper focuses on the relationship between enterprise green investment and comprehensive performance, and comprehensively examines the relationship between them from multiple aspects of heterogeneity. Through research, we found that green investment by enterprises can not only improve the enterprise environment, reduce enterprise emissions, but also improve the comprehensive performance of enterprises, and achieve the goal of joint development of the enterprise economy and environment. This can not only comprehensively promote the green transformation of enterprises, implement long-term sustainable development, but also improve the environmental health of the region where enterprises are located to a certain extent, and also promote regional economic growth; high-quality sustainable development of the regional economy can be realized. Therefore, enterprises should take into account both environmental and economic aspects and strive to maximize performance. Based on the full text analysis, this paper puts forward the following recommendations:

Firstly, in terms of green investment, Chinese enterprises are still in their infancy. Most enterprises are often constrained by short-term economic conditions and are not active in green investment. At the same time, due to the pressure of the government and society, most enterprises often ignore the benefits and competitiveness brought by green investment. On one hand, enterprises should enhance their awareness of environmental protection and use green investment to promote their sustainable development. On the other hand, the government should not only encourage enterprises’ green investment, but also introduce relevant preferential policies and financial subsidies to support enterprises’ green investment. Enterprise green investment needs a lot of capital and technology. Some small and medium-sized enterprises are facing a series of problems such as shortage of funds and backwards technology. Excessive green investment will bring some difficulties to enterprises. The government should support the green investment of small and medium-sized enterprises by reducing taxes or supporting corresponding subsidies during the transition period.

Secondly, local governments formulate environmental systems according to local economic and environmental conditions. On the one hand, with the increasingly serious environmental pollution, the government should strengthen environmental constraints and supervision to ensure environmental and ecological health. On the other hand, the formulation of environmental protection policies should also proceed from reality. Excessive regulation will lead to low efficiency of most enterprises and may even lead to damage and bankruptcy of enterprises. The comprehensive performance of the enterprise cannot be greatly improved.

In this paper, we only use data from China and do not conduct a comparative study with enterprises from developed economies. Therefore, in other economies with different political and economic systems, the interpretation of the results should be cautious, and further discussion may be needed. In addition, the driving mechanism of enterprises green investment in the same industry should also be explored. Future research can explore the factors affecting enterprises’ independent green investment.