Increasing Quantity or Improving Quality: Can Soil Pollution Control Promote Green Innovation in China’s Industrial and Mining Enterprises?

Abstract

This paper uses the provisions of the Soil Pollution Prevention and Control Action Plan for industrial and mining enterprises as a quasi-natural experiment and constructs a difference-in-differences (DID) model to study its effect on increasing the quantity and improving the quality of green innovation based on a panel of 453 industrial and mining enterprises in Shanghai and Shenzhen A-shares in China from 2011 to 2020. The results show that the Soil Plan can significantly promote the increase of green innovation and the quality of industrial and mining enterprises, and the results are still valid after a series of robustness tests. The mechanism test shows that the Soil Plan promotes the quality of green innovation by alleviating the financing constraints of industrial and mining enterprises, but the impact on their incremental quantity is not significant. The heterogeneity analysis shows that the effect of the Soil Plan on the quality of green innovation is stronger in the sample with a more independent board of directors and a digital mine transformation. This paper enriches the results in the field of soil pollution prevention and industrial and mining enterprises, and has important implications for the practice of promoting green innovation in Chinese enterprises.

1. Introduction

The world has reached a consensus that both economic development and environmental protection are needed. In recent years, among a series of environmental issues, the world is most concerned about the greenhouse effect, and has invested significant manpower and money. However, it is a harsh fact that we have overlooked some hidden environmental problems, such as soil pollution. From April 2005 to December 2013, China conducted its first national soil pollution survey. The 2014 National Soil Pollution Bulletin showed that 16.1% of the Chinese soil was polluted, 82% contained toxic inorganic pollutants, the most common being heavy metals such as cadmium, mercury, arsenic, chromium, and lead. Industrial plant waste and mining operations is a main human cause of widespread soil pollution (Chen et al., 2014) [1]. Thus, the impact of mining activities on the ecological environment has aroused widespread concern (Wang et al., 2022; Deng et al., 2022) [2,3].

Green innovation can reduce the emission of pollutants and has an important role in improving the environment (Xu et al., 2021; Lin and Ma 2022; Ali et al., 2022) [4,5,6]. On the one hand, based on the existing literature, green strategies (Sun and Sun, 2021) [7], market orientation (Du and Wang, 2022) [8], corporate social responsibility (Yuan and Cao, 2022) [9], etc., have a positive impact on green innovation of enterprises. On the other hand, as the Porter hypothesis suggests, a strict and flexible environmental regime can promote innovation of enterprises (Porter and Van der Linde, 1995) [10]. Compared with other factors, environmental regimes are formal and mandatory, and their impact on green innovation is more stable. In China, the positive impact of environmental regimes on green innovation has been widely demonstrated (Liu et al., 2020; Yao et al., 2021; Gao and Wang, 2021; Du et al., 2021) [11,12,13,14]. Furthermore, scholars have distinguished between the quantity and quality of green innovation, examining the different effects of these factors (Rao et al., 2022; Huang et al., 2022; Wang et al., 2022) [15,16,17].

Specifically for industrial and mining enterprises, we are curious whether environmental regimes can have a positive impact on green innovation. Aron and Molina (2020) [18] found that in Peru environmental regulation created more pressure on industrial and mining enterprises to promote their green innovation. Qi et al. (2019) [19] found that in China the construction of green mines facilitates green innovation in mining enterprises. Furthermore, Zhou et al. (2021) [20] found that the positive impact of environmental regulation on green innovation of mining enterprises mainly came from the role of environmental legitimacy guarantees. This implies that the Porter hypothesis is also valid in industrial and mining enterprises. However, few scholars have further explored the different impacts of a concrete form of environmental regime on the quantity and quality of green innovation in industrial and mining firms, as well as their potential mechanisms. This provides the motivation for our study.

China is increasingly known for its ambitions towards an ‘ecological civilization’ and a circular economy (Nechifor et al., 2020) [21]. Since the 18th Party Congress, the construction of ecological civilization has been incorporated into the five-sphere integrated plan, and China has put forward the scientific assertion that “Lucid waters and lush mountains are invaluable assets”. Meanwhile, the “Soil Pollution Prevention and Control Action Plan” (The Soil Plan) implemented in 2016 provides a good quasi-natural experiment for our study. More specific arrangements are made for soil environmental protection in industrial and mining enterprises. As a concrete form of environmental regime, the Soil Plan imposes new constraints on industrial and mining enterprises, and whether it will further affect green innovation has not been fully explored. The purpose of this paper is to investigate the impact of the Soil Plan on the quantity and quality of green innovation in industrial and mining enterprises, and its potential mechanisms.

Thus, based on a panel of 453 industrial and mining enterprises in Shanghai and Shenzhen A-shares in China from 2011 to 2020, the paper investigates whether the Soil Plan can promote green innovation to increase quantity and improve quality. Meanwhile, based on information asymmetry theory, it explores whether financing constraints play a mechanism effect in this. The paper found that: (1) the Soil Plan significantly contributed to the increase in green innovation and quality of industrial and mining enterprises in the pilot areas, and the findings still hold after a series of robustness tests. (2) The mechanism test found that the Soil Plan significantly contributed to the increase in green innovation and quality of industrial and mining enterprises by alleviating their financing constraints but did not contribute to the increase in green innovation. (3) Further study found that the effect of the Soil Plan on the increase in green innovation and quality was stronger in the sample with a more independent board of directors and the transformation of digital mines.

This paper makes three contributions: First, it is a quasi-experimental study of the policy effects of the Soil Plan based on a difference-in-differences (DID) model, which enriches the results related to soil pollution prevention and control and industrial and mining enterprises; the current research results related to soil pollution prevention and control mainly focus on the policy interpretation of the background, provisions and significance (Qu et al., 2016; Wang et al., 2016) [22,23]. Few empirical studies have been conducted in the literature specifically on the effects of the Soil Plan on the treatment of enterprises.

Second, based on the theory of information asymmetry and financing constraints, this paper explores the mechanisms underlying the ability of the Soil Plan to improve the quality of green innovation, in contrast with other environmental protection systems, and enriches the research findings in the field of environmental protection systems. The Porter hypothesis has been confirmed in the literature, and some scholars have further focused on the issue of innovation quality, but there is little evidence that China’s environmental protection system can improve the quality of green innovation on an incremental basis.

Third, considering the heterogeneity, this paper explores the role of board independence and digital mine transformation in promoting green innovation and quality in enterprises based on a micro-enterprise perspective, and proposes corresponding policy recommendations. The existing literature has already provided theoretical interpretations of the effects of the implementation of the Soil Plan, but this paper provides specific recommendations for its further implementation.

2. Institutional Background and Theoretical Mechanisms

2.1. Institutional Background

Soil pollution has been a concern in China since the 1970s, but efforts have not been effective. As shown in

Table 1. The stages and Characteristics of Soil Pollution Control.

Stage	Time	Representative Documents	Features
Stage 1	The 1970s	Article 10 of Chapter 2 of the Environmental Protection Act (1979), entitled “Protection of the Natural Environment”, requires that “the land shall be used rationally by local conditions”.	Principled and general legislation, without supporting specific provisions, is not very operational in implementation.
Stage 2	The 1980s–early 21st century	Article 30 of the Mineral Resources Law (1986), Chapter IV “Exploitation of Mineral Resources”, requires that “the land shall be used sparingly in the exploitation of mineral resources”. The Agriculture Law (1993), the Law on the Prevention and Control of Environmental Pollution by Solid Waste (1995), the Measures for the Administration of Pollution Source Monitoring (1999) and the Law on the Prevention and Control of Radioactive Pollution (2003) also contain some provisions referring to soil environment issues.	The provisions related to soil pollution prevention and control are scattered among other individual laws on pollution prevention and control, lacking a systematic and complete system of soil pollution prevention and control.
Stage 3	The 18th Party Congress to the present	Soil Pollution Prevention and Control Action Plan (2016), Soil Pollution Prevention and Control Act (2018).	Specific prevention and control measures are proposed for different sectors and pollution problems, which are systematic, targeted and specific.

, in the first and second phases (the 1970s to the early 21st century), China regarded soil pollution prevention and control as selective, so local governments tended to act according to their means and enterprises seeking to maximize their profits were less likely to take substantive action. In 2013, the Standing Committee of the 12th National People’s Congress included the Soil Pollution Prevention and Control Law in its legislative plan, marking the official launch of China’s special legislation on soil pollution prevention.

In 2016, the Soil Plan was successfully launched, becoming the first programmatic document for soil pollution prevention and control in China. In order to prevent local governments and industrial and mining enterprises from selectively ignoring soil pollution prevention and control, the provisions of the Soil Plan are very systematic and targeted, with Article 18 in Chapter 6 specifying in detail how to prevent soil pollution from industrial and mining industries.

Since 2017, in 13 provinces including Inner Mongolia, Jiangxi and Henan, the strict prevention of soil contamination by mineral resources development resulted in the pilot provinces accounting for 41.9% of the 31 provinces in Chinese mainland. As shown in

Table 2. China’s industrial and mining industry and industrial solid wastes in 2016.

Pilot Area	Province	Mining Employees (Million People)	General Industrial Solid Wastes (Million Tons)	Hazardous Solid Wastes (Million Tons)
Yes	Inner Mongolia	0.166	309.86	2.066
Yes	Jiangxi	0.058	125.35	1.105
Yes	Henan	0.452	173.72	0.760
Yes	Hubei	0.065	104.73	0.923
Yes	Hunan	0.081	61.56	4.868
Yes	Guangdong	0.028	82.70	2.867
Yes	Guangxi	0.031	94.27	2.300
Yes	Sichuan	0.186	136.20	2.475
Yes	Guizhou	0.134	90.77	0.495
Yes	Yunnan	0.136	172.89	3.221
Yes	Shaanxi	0.349	107.92	0.748
Yes	Gansu	0.107	67.06	1.226
Yes	Xinjiang	0.160	118.36	2.786
No	Beijing	0.045	6.04	0.178
No	Tianjin	0.044	15.67	0.267
No	Hebei	0.228	346.71	1.020
No	Shanxi	0.911	408.23	0.798
No	Liaoning	0.247	209.12	2.007
No	Jilin	0.131	62.10	2.629
No	Heilongjiang	0.279	82.05	0.622
No	Shanghai	0.001	17.97	0.668
No	Jiangsu	0.085	130.87	4.359
No	Zhejiang	0.006	55.62	3.863
No	Anhui	0.231	142.42	1.494
No	Fujian	0.021	69.25	0.827
No	Shandong	0.574	263.50	5.167
No	Hainan	0.005	3.33	0.044
No	Chongqing	0.056	25.20	0.589
No	Tibet	0.005	8.48	0.000
No	Qinghai	0.035	175.60	1.282
No	Ningxia	0.053	44.83	0.541

, first, the development level of industrial and mining in the pilot area is high, there are more employees. Second, the production of general industrial solid wastes and hazardous solid wastes in the pilot area is also high. This shows that the pilot area is highly representative and can reflect the contradiction between the development of China’s industrial and mining industry and soil pollution. Additionally, the prevention and control targets involve several sub-sectors, including mining, non-ferrous metal smelting, petroleum processing, coking, electroplating, etc.

Combined with each province’s self-developed action plan for soil pollution prevention and control, the provisions directly related to industrial and mining enterprises are summarized as follows: (1) Clear responsibility for soil pollution prevention and control. By the principle of “who pollutes, who treats, who benefits, who is responsible”, the enterprises concerned are urged to take preventive and control measures against soil pollution promptly, and to comprehensively rectify the historical tailings ponds. (2) Strengthen soil environmental assessment and monitoring. Key regulatory enterprises will conduct annual soil environmental risk assessment and radiation monitoring, and the results will be made public. (3) Strengthen the supervision of mineral pollution sources. Implement special emission limits for key pollutants and promote the adoption of clean production processes and technologies by relevant enterprises. (4) Strengthen pollution penalties. The relevant enterprises that fail to complete the above requirements are ordered to make corrections and fined, and those that refuse to make corrections are subject to continuous daily penalties by the original penalty amount, and those with serious circumstances are ordered to cease production and shut down.

At present, the Soil Plan has begun to show results (Li et al., 2019) [24]. The Ecological Environment Bulletin 2020 shows that China’s soil pollution prevention and control efforts have achieved substantial results. China has basically achieved the two core objectives of soil pollution prevention and control, i.e., the safe rate of contaminated arable land has reached about 90% and the safe utilization rate of contaminated land has reached more than 90%.

2.2. Theoretical Mechanisms

MM theory assumes that capital markets are perfect, that information is perfectly symmetrical and that there are no transaction costs, and that the interest rate on liabilities is risk-free. As a result, exogenous and endogenous financing can be fully substituted, and the investment behavior of a company depends only on the investment project and is not related to the capital structure. However, the assumptions of MM theory do not hold in reality. Based on the theory of information asymmetry and the theory of transaction costs, the theory of preferential financing suggests that exogenous financing costs more than endogenous financing. Therefore, the higher the information asymmetry between the firm and the external financing institution, the higher the cost of external financing (Myers and Majluf, 1984) [25], and the difference between it and the cost of internal financing is the financing constraint.

Based on Schumpeterian innovation theory, the innovative behavior of firms is strongly influenced by the availability of finance. Compared with ordinary innovation, green innovation has higher inputs, longer cycles and more uncertain returns, and is therefore more affected by the issue of financing constraints. It has been shown that environmental information disclosure is significantly and negatively related to the cost of financing (García-Sánchez et al., 2019; Li et al., 2022) [26,27]. Furthermore, financing constraints caused by environmental protection problems would inhibit enterprises from carrying out green innovation (Zhai et al., 2022) [28]. In the past, China’s emission monitoring system for air and water pollutants has been relatively well developed, but it is not mandatory to test and disclose information on the emission of soil pollutants. In addition, only a handful enterprises actively disclose information related to environmental management (Kuo et al., 2012) [29]. Enterprises with poorer environmental performance tend to disclose more low-quality qualitative information and avoid separate disclosure and quantitative disclosure. Therefore, there is a serious information asymmetry between industrial and mining enterprises and exogenous financiers on the soil environment, and stronger financing constraints make them less willing to innovate green (Liu et al., 2021) [30].

With the gradual expansion of green credit and green bonds by Chinese banks and other financial institutions, higher requirements are being placed on the financing qualifications of heavily polluting enterprises. However, the development of green finance has also provided more financing opportunities for enterprises with better environmental performance. With the implementation of the Soil Plan, relevant industrial and mining enterprises are required to conduct regular soil environmental monitoring and risk assessment, and to disclose the assessment results together with information on the discharge of special pollutants, fines for non-compliance and suspension of work. Therefore, the Soil Plan can alleviate the problem of information asymmetry related to the soil environment to a certain extent, reduce financing costs, and then promote green innovation. Based on this, we propose the following hypothesis:

Hypothesis 1.

The Soil Plan can significantly increase the total number of green patents for industrial and mining enterprises.

In China, patent innovations can be further divided into design, utility models and invention patents. Among these, utility model patents are also known as “small inventions”. According to the relevant provisions of China’s Patent Law, a utility model patent can be applied for if it is a practical, new technical solution for the shape, construction or combination of products, and the product is protected under the law. It can be found that utility model patents require far less innovation and technology than patents for inventions, and therefore their development costs and risks are lower. In the early days, to stimulate innovative behavior, China introduced many policies to encourage enterprises to undertake research and development of utility model patents. In recent years, the number of patent applications and grants in China has grown at a rapid pace, with utility model patents accounting for over 50% of the total. However, there is a growing concern among scholars that utility model patents, while encouraging “learning innovation”, may also have a dampening effect on “independent innovation”. As Hu et al. (2017) [31] suggest that non-innovation related motives for acquiring patents may have played an important role in the patenting surge. Furthermore, utility model patents have become a strategic act for Chinese enterprises to seek support, and innovation that only increases quantity but not quality does not promote sustainable development of enterprises (Jiang and Bai, 2022) [32].

For green patents, China usually uses the green patent IPC classification number for matching, which is divided into two categories: green utility model patents and green invention patents. For industrial and mining enterprises, compared with green utility model patents, green invention patents can fundamentally promote their green development. On the one hand, green invention patents can give better play to the advantages of more efficiency, reduce energy consumption by transforming the original process, improve the recycling rate of tailings and slag, and thus reduce soil pollution caused by mine solid waste. On the other hand, green invention patents can give full play to their unique advantages, and gradually replace backward processes and products through green supply, green production, green recycling, etc., so as to form a unique green competitiveness. However, China’s green innovation sector is facing the problem of increasing quantity rather than improving quality. Many companies have engaged in “green speculation”, trying to create a green image through green utility model patents to alleviate the financing constraints they face. It has been documented that policies such as green credit in China have not had the positive impact of improving the quality on green innovation (Liu and Dong, 2022) [33].

Xiang et al. (2020) [34] found that environmental disclosure plays an important role in promoting sustainable development of enterprises by enhancing their environmental awareness. Ren et al. (2022) [35] found that mandatory CSR disclosure in China helps promote the filing of more green utility model patents and green invention patents. Furthermore, Huang et al. (2020) [36] found that firms, when they innovate, can exacerbate information asymmetry in the short term. Subsequently, to meet the information needs of market investors, innovative firms disclose more patent-related earnings forecasts. Similarly, investors may pay attention to the actual environmental protection effect of green innovation. Compared with other environmental protection systems, the Soil Plan provides for mandatory soil environmental testing and information disclosure, which is conducive to improving the quality of information disclosure by industrial and mining enterprises, thereby discouraging their undesirable motives of seeking support through green utility model patents. Based on this, we propose the following hypothesis.

Hypothesis 2.

The Soil Plan can significantly increase the green invention patents of industrial and mining enterprises.

3. Study Design

3.1. Model Design

To test Hypotheses 1 and 2, we constructed a difference-in-differences (DID) model with mixed fixed effects, as follows.

$$\begin{array}{l}\mathrm{Patent}/{\mathrm{Invention}}_{\mathrm{i},\mathrm{t}+1}\\ ={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{Time}}_{\mathrm{t}}\times {\mathrm{Treat}}_{\mathrm{i}}+{\mathsf{\beta }}_2{\mathrm{Controls}}_{\mathrm{i},\mathrm{t}}+{\mathrm{Year}}_{\mathrm{t}}+{\mathrm{Industry}}_{\mathrm{i}}\\ +{\mathrm{Province}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}\end{array}$$

Model (1) is a mixed Poisson regression model due to the “law of rare events” for patent data. Where Patent/Invention_i,t+1 denotes the total number of green patents and the number of green invention patents in the next year, respectively, and Time_t × Treat_i denotes policy variables, and Controls_i,t denotes the above control variables. In addition, a series of fixed effects are controlled for in the model, with Year_t denotes year fixed effects, Industry_i denotes industry fixed effects, and Province_i denotes province fixed effects. ε_i,t denotes the random disturbance term. The coefficient β₁ in model (1) should be significantly positive if the Soil Plan can promote green innovation in industrial and mining enterprises to increase quality.

3.2. Variable Selection and Measurement

3.2.1. Explanatory Variables

Drawing on the research results of Tan et al. (2022) [37] and Tan and Zhu (2022) [38], we measured the quantity of green innovation in enterprises by the sum of the number of green patent applications and granted patents and the quality of green innovation in enterprises by the sum of the number of green invention patent applications and granted patents Invention.

First, the patent applications and grants of the sample companies were obtained from the State Intellectual Property Office of China (SIPO), including the results of independent inventions and joint inventions of group companies. Second, the “International Patent Green Classification List” developed by the World Intellectual Property Organization (WIPO) in 2010 was compared and matched with the green patents by their IPC classification numbers. Third, all the matched green patents were summed up to obtain the Patent, and among them, the green invention patents were summed up to obtain an Invention.

3.2.2. Explanatory Variables

In 2016, the Soil Plan was successfully introduced, of which Chapter 6, Article 18 stipulates that soil pollution by mineral resources development should be strictly prevented in areas where mineral resources development activities are concentrated, such as Inner Mongolia, Jiangxi, Henan, Hubei, Hunan, Guangdong, Guangxi, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu and Xinjiang provinces (districts).

Therefore, we define Time as a dummy variable for whether the policy is implemented or not, where Time = 1 if the time is 2016 and later, and Time = 0 otherwise. Treat is defined as a dummy variable for the pilot region of the policy, where Treat = 1 if the province where the enterprise is registered is the pilot region, and Treat = 0 otherwise. Time × Treat which is the cross-product term of the two is the policy variable of this paper.

3.2.3. Control Variables

Concerning previous research findings related to corporate green innovation, we selected a series of control variables, mainly including corporate financial indicators, governance structure, and regional economic factors. The specific variables and definitions are detailed in

Table 3. The definitions of variables.

Variable Name	Variable Definitions	Variable Measures
Ltime	Number of years on the market	Natural logarithm of the number of years on the market
Size	Business size	Natural logarithm of total assets
ROA	Return on total assets	Net profit/total assets
LEV	Gearing ratio	Liabilities/total assets
Manfee	Management efficiency	Administrative expenses/operating income
Occupy	Major shareholder occupancy	Other receivables/total assets
Board	Board size	Natural logarithm of the number of board members
Ind_r	Board independence	Number of Independent Directors/Number of Board of Directors
Top10	Concentration of shareholding	Shareholding of top ten shareholders
Age	The average of the senior management team	Mean age of the executive team
Gender	Gender Ratio of the Executive Team	Percentage of men in the senior management team
Gdp	Regional economic level	Natural logarithm of regional GDP
Env_f	Regional environmental finance	Regional environmental protection financial expenditure
Resource	Regional Resource Endowment	Number of people employed in the mining industry in the region

.

3.3. Sample and Data Sources

We selected the initial sample of industrial and mining enterprises in China’s Shanghai and Shenzhen A-shares from 2011 to 2020. As there is no official document defining the classification of industrial and mining enterprises, we refer to the industries mentioned in Article 18 of Chapter 6 of the Soil Plan. It compares the secondary categories in the 2012 Industry Classification of the Securities and Futures Commission for selection. In the end, all secondary categories under mining (B), five secondary categories under manufacturing (C), and two secondary categories under electricity, heat, gas, and water production and supply (D) were selected as industrial and mining industries. The specific secondary sectors are detailed in

Table 4. The classification of industry and mining.

Class I Sections	Category Name	Secondary Sectors	Category Name
B	Mining	06	Coal mining and washing
		07	Oil and gas extraction
		08	Ferrous metal mining
		09	Non-ferrous metal mining and processing
		10	Non-metallic mineral extraction
		11	Mining support activities
		12	Other mining industries
C	Manufacturing	25	Petroleum processing, coking and nuclear fuel processing industries
		30	Non-metallic mineral products industry
		31	The Ferrous metal smelting and rolling processing industry
		32	The Non-ferrous metal smelting and rolling processing industry
		33	Metal products industry
D	Electricity, heat, gas, and water production and supply	44	Electricity, heat production and supply industry
		45	Gas production and supply industry

.

The data processing process is as follows: (1) exclude samples that were ST and ST* in the current year, (2) exclude samples that were listed for one year or less, (3) exclude samples with gearing ratios greater than 100%, and (4) exclude samples with a large number of missing data on core variables. A total of 453 sample companies with a total of 3024 observations were eventually obtained for this paper.

The financial data of the companies were obtained from CSMAR and Wind databases. Data on corporate green patents were obtained by hand searching and matching at the State Intellectual Property Office of China (SIPO), as detailed above. Provincial indicator data were obtained by collating through the China Statistical Yearbook.

4. Empirical Analysis

4.1. Descriptive Statistics

Table 5. Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
Patent	3024	19.506	112.468	0	2402
Invention	3024	10.368	91.708	0	2091
Ltime	3024	2.389	0.699	0.693	3.332
Size	3024	22.785	1.516	16.117	28.636
ROA	3024	0.03	0.163	−1.892	7.445
LEV	3024	0.493	0.202	0.007	0.996
Manfee	3024	0.084	0.405	0.002	18.191
Occupy	3024	−0.019	0.174	−9.302	0.702
Board	3024	2.296	0.199	0	2.944
Ind_r	3024	0.37	0.051	0.231	0.667
Top10	3024	0.606	0.159	0.013	0.986
Age	3024	0.502	0.031	0.395	0.59
Gender	3024	0.86	0.098	0.375	1
Gdp	3024	10.199	0.82	6.553	11.587
Env_f	3024	67.54	43.457	21.611	553.688
Resource	3024	0.195	0.239	0.001	1.03

reports the descriptive statistics of the main variables in this paper. the mean value of Patent is 19.506, indicating that the average Chinese industrial and mining company applies for or grants a total of about 20 green patents a year. The mean value of Invention is 10.368, which is more than half of the mean value of Patent, indicating that Chinese industrial and mining enterprises have gradually become aware of the importance of green invention patents in recent years. The descriptive statistics show that both Patent and Invention are consistent with “the law of rare events “, and the Poisson regression model can be selected for testing. The results of the descriptive statistics for the other variables are shown in Table 5.

4.2. Baseline Regression

4.2.1. Baseline Regression Analysis

Table 6. The baseline regression.

	1	2	3	4
	Patent	Patent	Invention	Invention
Time × Treat	0.658 **	0.451 **	1.031 ***	0.655 ***
	(0.321)	(0.200)	(0.303)	(0.208)
Ltime		−0.211 **		−0.306 ***
		(0.090)		(0.100)
Size		0.706 ***		0.625 ***
		(0.053)		(0.056)
ROA		0.739 **		1.369 ***
		(0.338)		(0.275)
LEV		−0.214		0.519
		(0.389)		(0.356)
Manfee		−1.417		−8.443 ***
		(2.751)		(2.502)
Occupy		−4.596 **		−9.498 ***
		(1.863)		(1.431)
Board		0.561		0.208
		(0.412)		(0.480)
Ind_r		0.757		−1.710
		(2.085)		(1.728)
Top10		−1.305 *		−0.216
		(0.696)		(0.328)
Age		6.273 ***		12.410 ***
		(2.274)		(2.794)
Gender		−1.874 ***		−1.779 ***
		(0.449)		(0.647)
Gdp		−0.649		−0.639
		(0.451)		(0.554)
Env_f		0.002		0.003
		(0.003)		(0.003)
Resource		−0.511		−0.225
		(0.736)		(0.853)
_cons	3.890 ***	−9.818 **	2.944 ***	−11.481 **
	(0.312)	(4.559)	(0.395)	(5.710)
Year	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
Obs	3024	3024	3024	3024

Note: Robust standard errors in brackets, ***, **, * indicate significant at the 0.01, 0.05, 0.1 levels, respectively.

reports the results of the benchmark regressions. As can be seen from columns 1 and 3, Time × Treat is significantly and positively correlated with Patent and Invention, controlling for year, industry and province fixed effects. Columns 2 and 4 show that Time × Treat is still significantly positively correlated with Patent and Invention after the inclusion of control variables. The above results indicate that “the Action Plan for Soil Pollution Prevention and Control” has significantly promoted the green innovation of industrial and mining enterprises in the pilot areas, and has a more significant impact on the quality of green innovation. Therefore, Hypotheses 1 and 2 are confirmed.

4.2.2. Dynamic Effects Test

In order to ensure the unbiased estimation of the difference-in-differences model, the assumption of parallel trends needs to be satisfied. For robustness reasons, we refer to the work of Jacobson et al. (1993) [39] and construct the following model based on an event study approach to test for dynamic effects.

$$\begin{array}{l}\mathrm{Patent}/{\mathrm{Invention}}_{\mathrm{i},\mathrm{t}+1}\\ ={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{{\displaystyle \sum }}_{\mathrm{t}=2011}^{2019}{\mathrm{Year}}_{\mathrm{t}}\times {\mathrm{Treat}}_{\mathrm{i}}+{\mathsf{\beta }}_4{\mathrm{Controls}}_{\mathrm{i},\mathrm{t}}+{\mathrm{Year}}_{\mathrm{t}}\\ +{\mathrm{Industry}}_{\mathrm{i}}+{\mathrm{Province}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}\end{array}$$

Referring to the findings of Nunn and Qian (2011) [40], Figure 1 plots the regression results for Patent and Invention at 95% confidence intervals, respectively, using the first year of the sample period (2011) as the base period. Figure 1 indicates that there was no significant difference between the experimental group and the control group before the implementation of the Soil Plan, and the hypothesis of a parallel trend was met. After 2016, the confidence intervals of the regression coefficients are significantly different from 0. This indicates that the experimental group started to produce significant policy effects after the implementation of the Soil Plan, which further supports Hypotheses 1 and 2.

Meanwhile, the regression results for Patent and Invention did not reach 95% significance in 2016. A reasonable explanation is that although the Soil Plan policy was released in 2016, the treatment of mineral pollution was only formally implemented in 2017. At the same time, green innovation has a certain cycle, and although the study conducted a one-period lag, the response period is still short.

4.3. Robustness Tests

4.3.1. PSM-DID

When using a difference-in-differences model, the selection of the experimental group should be randomized. In order to ensure robustness, this section draws on the findings of Heckman et al. (1997) [41] and uses propensity score matching (PSM) so that the experimental and control groups do not differ significantly from each other in any other way as far as possible.

As shown in

Table 7. The logit propensity score estimation.

Variables	Coef.	P
Ltime	0.596 ***	0.000
Size	−0.058	0.282
ROA	0.133	0.772
LEV	−0.407	0.206
Manfee	−1.992 *	0.057
Occupy	−2.863 **	0.016
Board	0.303	0.375
Ind_r	1.164	0.312
Top10	−0.663 *	0.087
Age	−1.217	0.565
Gender	−0.386	0.521
Gdp	0.847 ***	0.000
Env_f	0.010 ***	0.000
Resource	−1.899 ***	0.000
_cons	−10.513 ***	0.000
R2	0.091

Note: Robust standard errors in brackets, ***, **, * indicate significant at the 0.01, 0.05, 0.1 levels, respectively.

, first, logit regression is performed on policy variables through all control variables, with seven variables, Ltime, Manfee, Occupy, Top 10, Gdp, Env_f, and Resource, being significantly correlated, indicating a possible effect on the choice of the experimental group. Second, the seven variables mentioned above were selected to calculate propensity score values for 1:1 nearest neighbor matching. Third, as can be seen from column 1 of

Table 8. Robustness tests.

	1		2			3		4		5
	Patent	Invention	Time × Treat	Patent	Invention	Patent	Invention	Patent	Invention	Patent	Invention
Time × Treat	0.862 ***	1.018 ***		2.716 ***	2.192 **	0.309 **	0.484 ***	0.428 **	0.650 ***	0.427 **	0.652 ***
	(0.323)	(0.377)		(0.991)	(1.117)	(0.122)	(0.141)	(0.202)	(0.210)	(0.205)	(0.212)
Ln (1 + Earthquake)			0.110 ***
			(0.007)
Gas										0.003	0.003
										(0.004)	(0.005)
Water										−0.035 **	−0.012
										(0.014)	(0.014)
_cons	5.562	2.136	−4.873 ***	−0.751	−3.801	−16.852 ***	−23.463 ***	−8.046 *	−9.151	−9.593 **	−11.066 *
	(12.132)	(12.329)	(0.500)	(6.416)	(8.100)	(3.689)	(4.130)	(4.571)	(5.803)	(4.652)	(5.842)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs	704	704	3024	3024	3024	3024	3024	2514	2514	3024	3024
R2			0.657

Note: Robust standard errors in brackets, ***, **, * indicate significant at the 0.01, 0.05, 0.1 levels, respectively.

, only 704 observations remained after the PSM. Controlling for fixed effects and control variables, Time × Treat remained significantly positively correlated with Patent and Invention, and the findings remained robust.

4.3.2. Instrumental Variables

Although the pilot provinces are largely randomized, it is still not possible to completely exclude the influence caused by other potential factors. In this paper, the cumulative number of earthquakes in each province since 2001 was chosen as the instrumental variable to eliminate endogenous factors that may influence the selection of pilot provinces through a two-stage regression (Cai et al., 2016) [42]. First, mineral resources are mainly distributed in areas with active crustal movements, and provinces with a higher number of earthquakes tend to have a faster development of industrial and mining industries and are more likely to be selected as pilot areas for mineral pollution control. Meanwhile, the number of earthquakes is determined by the geological structure and crustal movements. Therefore, the cumulative number of earthquakes in each province is also consistent with the assumption of exogenous, based on satisfying the correlation of instrumental variables.

As can be seen from column 2 of Table 8, first, a one-stage regression of the cumulative number of earthquakes, which takes the natural logarithm, was performed on Time × Treat. Second, the new Time × Treat was fitted to the regression results, and then a two-stage regression was performed on Patent and Invention, which were still significantly positively correlated. This indicates that the findings are still robust after eliminating most of the potential endogenous factors.

4.3.3. Replacing the Regression Model

The Poisson regression model is used for the baseline regression as there is “the law of rare events” in the patent data. Meanwhile, there is a risk of excessive dispersion in the patent data, so we use a negative binomial regression for robustness testing. As can be seen from column 3 in Table 8, Time × Treat is still significantly and positively correlated with Patents and Invention. Therefore, the study results remain robust.

4.3.4. Exclude Samples without Patents

As the baseline regression uses a mixed effects model, the existence of a large number of samples that have not been applying for or obtaining green patents during 2011–2020 may have some impact on the regression results. Therefore, in this section, after excluding the zero patent sample, the remaining 354 sample companies and 2514 observations were re-regressed for analysis. As can be seen from column 4 in Table 8, Time × Treat is still significantly and positively correlated with Patents and Invention. Therefore, the study results remain robust.

4.3.5. Control of Other Environmental Governance

In 2013, the implementation of the Air Plan put forward the target of reducing the emission intensity of major air pollutants in key industries by more than 30% by the end of 2017. In 2015, the implementation of the Water Plan required that by the end of 2017, industrial agglomerations should become centralized, and sewage treatment facilities should be completed as required. The industrial and mining industry is also one of the key sectors for the regulation of waste gas and wastewater, and therefore needs to further control the impact of waste gas and wastewater treatment on its green innovation.

The amount of investment in industrial waste gas and wastewater treatment in the previous year (billion RMB) by province was added to the baseline regression model as a control variable. As can be seen from column 5 in Table 8, Time × Treat remains significantly positively correlated with Patent and Invention after controlling for the effects of exhaust and wastewater treatment. Therefore, the study results remain robust.

5. Further Analysis

5.1. Mechanism Testing

In the existing literature, there are three main approaches to measuring corporate financing constraints. The first is through a single indicator related to the financing characteristics of the firm, such as the size and age of the firm, the gearing ratio and interest expenses. The second is through the “investment-cash flow” sensitivity model. The third is to construct a financing constraint model by using a list of financial indicators, such as the WW index constructed by Whited and Wu (2006) [43] based on an improved KZ index. The WW index is different from other indices in that it removes endogenous factors such as Tobin’s Q value and takes into account the characteristics of the industry to which the firm belongs. Therefore, the WW index is considered to be highly effective as a proxy variable for financing constraints and is widely used in empirical studies. We draw on the research findings of Whited and Wu (2006) [43] to measure the financing constraints of firms by constructing a WW index. The specific formula is as follows.

$$\begin{array}{l}{\mathrm{WW}}_{\mathrm{i},\mathrm{t}}=-0.091\times {\mathrm{Cashflow}}_{\mathrm{i},\mathrm{t}}-0.062\times {\mathrm{Divpos}}_{\mathrm{i},\mathrm{t}}+0.021\times {\mathrm{Ldebt}}_{\mathrm{i},\mathrm{t}}\\ -0.044\times {\mathrm{Size}}_{\mathrm{i},\mathrm{t}}+0.102\times {\mathrm{Indsg}}_{\mathrm{i},\mathrm{t}}-0.035\times {\mathrm{Sgrowth}}_{\mathrm{i},\mathrm{t}}\end{array}$$

${\mathrm{Cashflow}}_{\mathrm{i},\mathrm{t}}$ is the ratio of the firm’s net cash flow to total assets for the period. ${\mathrm{Divpos}}_{\mathrm{i},\mathrm{t}}$ is a dummy variable that takes 1 if the firm pays dividends in the current period and 0 otherwise. ${\mathrm{Ldebt}}_{\mathrm{i},\mathrm{t}}$ is the ratio of long-term liabilities to total assets for the period. ${\mathrm{Size}}_{\mathrm{i},\mathrm{t}}$ is the natural logarithm of the firm’s total assets for the period. ${\mathrm{Indsg}}_{\mathrm{i},\mathrm{t}}$ is the current sales revenue growth rate of the enterprise’s industry. ${\mathrm{Sgrowth}}_{\mathrm{i},\mathrm{t}}$ is the growth rate of sales revenue of the enterprise in the current period. ${\mathrm{WW}}_{\mathrm{i},\mathrm{t}}$ The smaller the ratio, the weaker the financing constraint faced by the enterprise in the current period.

Table 9. Financing Constraints.

	1			2
	WW	Patent	Invention	Patent	Invention
Time × Treat	−0.058 ***	0.443 **	0.652 ***	0.467 ***	0.190
	(0.018)	(0.197)	(0.203)	(0.170)	(0.180)
WW		0.054	−0.271 *
		(0.204)	(0.159)
Time × Treat × Gfinance				−0.127	0.936 **
				(0.488)	(0.376)
Gfinance				0.482 *	0.547 *
				(0.269)	(0.292)
Time × Gfinance				−0.165	−0.359
				(0.202)	(0.237)
Treat × Gfinance				0.022	−0.868 ***
				(0.372)	(0.240)
_cons	−0.977 *	−9.025 *	−12.879 **	−10.941 **	−13.518 **
	(0.507)	(4.639)	(5.628)	(4.268)	(5.324)
Controls	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes
Obs	3024	3024	3024	3024	3024
R2	0.308

Note: Robust standard errors in brackets, ***, **, * indicate significant at the 0.01, 0.05, 0.1 levels, respectively.

reports the results of the mechanism tests. As shown in column 1, Time × Treat is significantly negatively related to Financing Constraints WW, controlling for fixed effects and control variables. By adding financing constraint as a control variable to the benchmark model for regression, financing constraint WW is not significantly positively related to Patent, but significantly negatively related to Invention. Meanwhile, the regression coefficients of Time × Treat and Invention decreased (see column 4 in Table 5), indicating that the Soil Plan can promote the quality of green innovation in industrial and mining enterprises by alleviating the financing constraints, but the increment of green innovation is not caused by the financing constraints.

In order to ensure the robustness of the mechanism test, the role of green financing constraints is further considered. Referring to the research results of Liu and Dong (2022) [33], the dummy variable Gfinance was set as a green credit restricted industry based on the Green Credit Guidelines promulgated by China and the A class environmental and social risk industries in the Key Evaluation Indicators for Green Credit Implementation promulgated by the former Banking Regulatory Commission. Specifically, Gfinance was set to 1 when the industries to which the sample companies belonged were coal mining and washing, oil and gas extraction, ferrous metal mining, non-ferrous metal mining, non-metallic mining, mining auxiliary activities, other mining, and electricity and heat production and supply, and 0 otherwise. The Gfinance is cross-multiplied with the policy variable Time × Treat, the time variable Time, and the pilot variable Treat, and the cross-multiplied term is added to the baseline model.

As shown in column 2 of Table 9, Time × Treat × Gfinance is not significantly negatively correlated with Patent but is significantly positively correlated with Invention, indicating that the Soil Plan, under the effect of green credit, can significantly promote the quality of green innovation in risky A industries, but does not have an impact on the increment of green innovation. On the other hand, it also has a certain degree of impact on the quality of green innovation. On the other hand, this also complements the findings of Liu and Dong (2022) [33] that green credit policies need to be symmetrical in terms of environmental information to work better.

5.2. Heterogeneity Analysis

5.2.1. Board Independence

The independent directorship is seen as one of the most important measures to effectively monitor corporate behavior. The literature has found that independent directors can also play an important role in strengthening corporate governance and maintaining capital market stability in China (Zhu et al., 2016; Melis and Rombi, 2021) [44,45]. More specifically, Khan et al. (2022) [46] and Fu (2019) [47] found that independent directors have a facilitating effect on the environmental disclosure and innovation of listed companies. Therefore, industrial and mining companies with a higher proportion of independent directors may be required to disclose higher quality information about the soil environment, which helps to alleviate information asymmetry with the outside world.

Since 30 June 2003, the SFC has made it mandatory for listed companies to have at least 1/3 of their directors as independent directors. As can be seen from the descriptive statistics in Table 4, the mean value of the proportion of independent directors in the sample companies is only 37%. Therefore, we use 1/3 as the criterion, with no less than 34% of independent directors being considered as having stronger board independence and vice versa. As shown in

Table 10. Independent board.

	Patent		Invention
	Weak	Strong	Weak	Strong
Time × Treat	0.588 ***	0.355	0.543 ***	0.726 ***
	(0.175)	(0.271)	(0.207)	(0.220)
_cons	−15.656 *	−13.927 ***	−32.954 **	−16.951 ***
	(8.789)	(5.385)	(13.092)	(5.875)
Controls	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
Obs	1543	1481	1543	1481

Note: Robust standard errors in brackets, ***, **, * indicate significant at the 0.01, 0.05, 0.1 levels, respectively.

, Time × Treat is significantly and positively correlated with Patents and Invention in the sample with weak board independence. In contrast, in the sample with stronger board independence, Time × Treat is no longer significantly related to Patent, but the positive relationship with Invention is further enhanced. Therefore, the independent directorship system can effectively promote the implementation of the Soil Plan and make the green innovation decisions of industrial and mining enterprises more quality-oriented.

5.2.2. Digital Mine Transformation

The rapid development of China’s digital economy is driving the digital transformation of enterprises, which in turn has a profound impact on their level of quality development. Specifically, Li and Shen (2021) [48] and Xue et al. (2022) [49] found that Chinese heavy polluters rely on digital transformation to promotes their green technology innovation. With the spread of digital mines, Chinese industrial and mining enterprises are also experiencing dramatic changes in digital transformation. On the one hand, digital transformation promotes the sharing and learning of innovative information, which helps industrial and mining enterprises to increase the number of green innovations. On the other hand, digital transformation has built a broader platform for innovation talents and innovation capital, which can fundamentally improve the independent research and development capability of enterprises and help industrial and mining enterprises improve the quality of green innovation.

Drawing on the research findings of Tian et al. (2022) [50], we measure the digital transformation of listed companies by summing the frequency of the appearance of artificial intelligence technology, blockchain technology, cloud computing technology, and big data technology, and digital technology applications in their announcements. As there is a large sample of industrial and mining companies that have not yet undergone digital transformation, this paper only distinguishes whether the sample is transformed or not. If there are one or more relevant descriptions, it is considered to have undergone a digital transformation. Conversely, if none, the period is considered not to have undergone digital transformation. Data on digital transformation were obtained from the “Digital Economy” section of the CSMAR database.

As shown in

Table 11. The transformation of digital mines.

	Patent		Invention
	No	Yes	No	Yes
Time × Treat	−0.125	0.746 ***	−0.043	0.811 ***
	(0.202)	(0.206)	(0.205)	(0.227)
_cons	−13.876 **	−19.596 ***	−26.198 ***	−20.458 ***
	(6.274)	(6.978)	(7.207)	(6.887)
Controls	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
Obs	1904	1120	1904	1120

Note: Robust standard errors in brackets, ***, ** indicate significant at the 0.01, 0.05 levels, respectively.

, Time × Treat was significantly and positively correlated with both Patent and Invention in the sample that had undergone digital transformation. In the sample that did not undergo digital transformation, Time × Treat was not significantly negatively correlated with Patent and Invention. Therefore, the digital mine transformation can effectively promote the implementation of the Soil Plan and improve the quality of green innovation in industrial and mining enterprises based on increasing quantity.

6. Conclusions and Insights

Based on a panel of 453 industrial and mining enterprises listed on the Shanghai and Shenzhen A-shares in China from 2011 to 2020, we use the Soil Plan promulgated by China as a quasi-natural experiment in order to investigate the impact and mechanism of soil pollution prevention and control on the green transformation of industrial and mining enterprises: (1) Soil pollution prevention and control can significantly contribute to the green innovation and quality improvement of industrial and mining enterprises. (2) Soil pollution control promotes the quality of green innovation by alleviating the financing constraints of industrial and mining enterprises, but not the quantity of incremental green innovation. (3) When the board of directors is more independent and the transformation of the digital mine is underway, the effects of the Soil Plan on the green innovation and quality improvement of industrial and mining enterprises are more significant.

The findings of this paper provide a basis for the policy effects of the Soil Plan on promoting green transformation in industrial and mining enterprises, and also identify the mechanisms underlying its promotion of green innovation and quality improvement, providing insights for future development of environmental protection systems and promotion of enterprise innovation and quality improvement. First, the prevention and control of soil pollution have been effective, and China should further implement the relevant provisions of the Soil Plan. In the future, the prevention and control of soil pollution should pay more attention to the green innovation of industrial and mining enterprises, which is the key to achieving green mines.

Second, China should coordinate the Soil Plan with policies such as green credit to guide banks and other financial institutions to reduce their financing constraints on industrial and mining enterprises with better environmental performance. Meanwhile, it should continue to place greater emphasis on the testing and disclosure of soil environmental information, and give full play to the role of green finance as an incentive for high-quality green innovation by alleviating information asymmetries.

Third, China should further guide industrial and mining companies to improve their internal governance structures and increase the independence and oversight of their boards of directors. At the same time, China should increase its support for digital mines, which will help industrial and mining companies to improve their ability to share information and integrate knowledge for green innovation, resulting in digitalization and greening work together.

This paper also has some limitations. First, we use a three-step approach in testing the mechanism, which has been suggested to have some endogeneity problems. Second, this study finds that the Soil Plan has a mechanism to alleviate the financing constraints of industrial and mining enterprises, because it will strengthen the disclosure of soil environmental information under the pressure of compliance. However, the disclosure effect has not been tested in this paper and could be added to future studies.