Green Eco-Innovation and Supply of Critical Metals: Evidence from China

Abstract

Ensuring a green supply of critical metals is essential to achieve high-quality economic development and ecological security. Based on data from 2000 to 2019 on five critical metals in China (copper, aluminum, nickel, antimony, and titanium), this study employs a series of econometric methods, such as fixed-effects regression and moderating effects, to examine the influence of green eco-innovation on primary and recycling supply of critical metals, as well as the underlying mechanisms. The findings indicate: (1) Green eco-innovation has an inverted U-shaped impact on the primary supply of bulk critical metals, and it is currently in the declining phase of the curve. (2) Green eco-innovation has a U-shaped impact on the recycling supply of critical metals and the primary supply of minor critical metals. The former is currently in the rising phase of the curve, while the latter is in the declining phase. (3) The impact of green eco-innovation on the supply of critical metals is stronger in industrially underdeveloped regions. (4) The improvement of energy efficiency, optimization of energy structure, and strengthening of environmental regulations enhance the impact of green eco-innovation on the supply of critical metals. Finally, the optimization of the energy mix is conducive to securing the supply of key metals. This study provides a theoretical basis for utilizing green eco-innovation to achieve a green supply of critical metals and enriches the theoretical research on green eco-innovation.

1. Introduction

Critical metal resources serve as the essential material foundation for the development of modern society and play a crucial role in the normal functioning of the economy [1,2,3]. Critical metals such as copper, aluminum, nickel, and titanium are widely applied in high-tech industries such as semiconductors, healthcare, defense, and aerospace [4,5]. In 2021 alone, Chia’s consumption of refined copper and electrolytic aluminum reached 13.87 million tons and 39.87 million tons, respectively, representing an increase of nearly 5% compared to the previous year and eight times the consumption of the second-ranking country [3]. Furthermore, the development of clean energy sectors also relies heavily on critical metals. For instance, the construction of wind turbines and water and solar power plants requires a significant amount of copper and rare earth elements [6,7], while lithium and nickel are key raw materials for power batteries [8]. It is foreseeable that as the “dual carbon” strategic goals are gradually pursued, and strategic emerging industries continue to develop, various sectors will place higher demand on the supply of critical metals (CMS). Against this backdrop, enhancing the level of critical metal resource supply becomes a vital path for China’s pursuit of high-quality economic development.

However, the critical metal industry, being an energy-intensive sector, exhibits the characteristics of “high energy consumption and high pollution” [3,9]. According to statistics, the energy consumption in China’s copper, aluminum, and nickel industries exceeded 44.40 million tons, 118.38 million tons, and 30.36 million tons of standard coal in 2019, respectively. The total carbon dioxide emissions exceeded 500 million tons, nearly doubling compared to 2006. Continued emissions of significant amounts of carbon dioxide and other harmful substances will pose a substantial threat to society and even life on Earth. Therefore, ensuring CMS while achieving green and sustainable development has become an important global research topic.

Among the various approaches to achieving green supply, green eco-innovation, including clean energy development, pollution control, and green recycling, has emerged as a dominant factor [10,11]. Numerous studies have found that green eco-innovation can enhance production capacity while reducing carbon dioxide and other harmful emissions [12,13]. For example, improving energy efficiency can enhance the utilization efficiency of polluting energy sources in production and daily life [14], while green recycling can reduce energy and resource consumption, as well as pollutant emissions, thereby enhancing the green utilization efficiency of metal resources [15,16]. In light of this, governments worldwide are actively promoting green eco-innovation to achieve a green supply of critical metals. The United Nations has put forward the Sustainable Development Goals (SDGs), which call for the development of clean energy and the improvement of the climate environment. The European Commission emphasizes the development and promotion of green and low-carbon innovations in mining and smelting [17]. The Chinese government, especially in the “14th Five-Year Plan” and the “National Report on Mineral Resource Protection and Comprehensive Utilization”, encourages green development, promotes harmonious coexistence between humans and nature, strengthens green eco-innovation and its dissemination, and promotes the application of advanced technologies in the development and utilization of mineral resources [9]. Therefore, it is necessary to conduct in-depth research from the perspective of green eco-innovation to understand its impact on CMS and provide valuable insights for achieving green CMS.

Therefore, what specific impacts does green eco-innovation have on CMS? Are the impacts different for primary and recycling CMS? Are there differences in the impacts on bulk and minor CMS? Are there variations in the impacts on CMS across different regions? What are the mechanisms behind the impacts on CMS? Exploring these questions is of great theoretical and practical significance for China to achieve green CMS through green eco-innovation, ensuring the security of critical metal resources, and achieving the strategic goals of carbon peaking and carbon neutrality. This study is based on panel data of five critical metals in China, namely copper, aluminum, nickel, antimony, and titanium, from 2000 to 2019. It first investigates and summarizes the impacts of green eco-innovation on different types of CMS and their differences. Then, it analyzes the impacts of green eco-innovation on CMS in different regions. Finally, it explores the mechanisms behind the impacts of green eco-innovation on CMS through moderating effects.

The contributions of this paper are as follows. Firstly, this study reveals the impacts of green eco-innovation on CMS for the first time, effectively coupling green eco-innovation with CMS and filling the research gap in related fields. Secondly, this paper compares the impacts of green eco-innovation on different types of CMS from the perspective of bulk and minor critical metals, both primary and recycling supply, providing ample evidence for existing research. Thirdly, this paper analyzes the impacts of green eco-innovation on CMS in different regions from a regional perspective. This analysis is helpful for local governments to formulate more targeted policies based on regional development differences, thus having high practical value. Lastly, this paper explores the mechanisms behind the impacts of green eco-innovation on CMS from the perspectives of energy efficiency, energy structure, and environmental regulations, revealing the nonlinear characteristics involved. It helps to understand the complexity of the impacts of green eco-innovation and provides a theoretical basis for ensuring CMS security. The main variables taken in this paper and their definitions are shown in

Table 1. Variables and their definitions.

VarName	Unit	Define	Mean	Min	Max
S	t	Supply of critical metals	220,988.451	1.000	8,667,658.000
GEI	item	Green Eco-Innovation	3738.925	3.000	67,258.000
EE	%	Energy efficiency	1.136	0.221	4.170
ES	%	Energy structure	0.139	0.066	0.258
ER	%	Environmental regulations	0.175	0.069	0.836
AS	%	Advanced industrial structure	5.234	0.518	1.037
FD	%	Foreign trade dependence	0.163	0.000	0.021
FI	%	Financial investments	0.628	0.046	0.201
PC	%	Pollution control	141.646	0.060	18.640
UR	%	Urbanization rate	0.896	0.233	0.488
PS	item	Population size	4971.096	516.000	12,489.000
C	yuan	Production cost	17,788.847	155.530	134,083.080
RGDP	yuan	Per capita GDP	32,219.555	2661.557	164,220.000
RD	%	Research and development investment	0.012	0.002	0.074
PerS	t/person	Per capita supply of critical metals	80.852	0.000	4164.840
PerGEI	item/person	Per capita green ecological patents	0.726	0.010	15.820
X	-	Control variable	-	-	-
Y	-	Moderating variables	-	-	-
μ	-	Province fixed	-	-	-
λ	-	Time fixed	-	-	-
ε	-	Disturbance error term	-	-	-

.

The article is organized in the following manner. Section 2 presents an overview of the relevant research and the development of the hypotheses. Section 3 outlines the variables, data sources, and methodologies employed in this study. Section 4 presents the findings and analysis. Lastly, Section 5 summarizes the study and presents corresponding policy recommendations.

2. Literature Review and Research Hypothesis

2.1. Literature Review

2.1.1. The Role of Green Eco-Innovation in the Field of Resources and Environment

Green eco-innovation is an integral part of innovation, closely related to the progress and creativity of industrial production, and these factors have a crucial impact on environmental pollution [18]. With the development of the green eco-field, green eco-innovation contributes to addressing significant environmental challenges faced by humanity, such as climate change [19,20].

Numerous studies have shown that green eco-innovation can promote economic growth while protecting environmental quality. For instance, Chen et al. [21] explored the relationship between eco-innovation levels and economic growth rates, finding that regions with better ecological resource management tend to have higher levels of economic growth. Shang et al. [22] found that the overall ecological efficiency of Chinese cities has gradually improved, and the spillover effect of green innovation is a decisive factor leading to convergence in urban ecological effects. Lin and Ma [23] studied prefecture-level cities in China from 2006 to 2017 and discovered that green eco-innovation can reduce urban carbon emissions through industrial structure upgrading. Zhang et al. [24] observed that green innovation has short-term and long-term effects on economic growth and the ecological footprint, and green eco-innovation can achieve optimal efficiency while minimizing its impact on the ecological footprint.

However, scholars have found that green eco-innovation may not always have positive impacts. Ali et al. [25] found no significant positive correlation between green eco-innovation and economic development. Khan et al. [26] argue that eco-innovation may lead to increased overall energy and resource consumption by promoting the use of more efficient products, ultimately resulting in higher carbon emissions. Additionally, some researchers have suggested that the relationship between green eco-innovation, economic growth, and carbon emissions may be nonlinear [27,28]. Furthermore, green eco-innovation is influenced by factors such as environmental regulations, market turbulence, research and development investments, market size, energy structure, and energy efficiency [29,30].

2.1.2. Green Eco-Innovation and Critical Metals

As one of the essential materials for the development of high-tech industries, such as new energy vehicles and aerospace, critical metals have always been a focal point. Technological innovation, among other factors, can safeguard the critical metal industry by influencing production scale, energy structure, and energy efficiency [2,31]. However, the production of critical metals inevitably brings about certain pollution issues [9]. Based on this, scholars have measured the efficiency of green development in the metal industry. Feng et al. [32] used MFA and DEA to measure the changes in China’s metal industry’s green total factor productivity from 2000 to 2015 and its influencing factors. They found that technological innovation was the most critical driving factor, while low-scale efficiency acted as a restraining factor. Luo et al. [33] calculated the green development index and its influencing mechanism in China’s mining industry and found that technological innovation was crucial for the development of green mining. With the increasing emphasis on environmental issues in society as a whole, achieving green CMS has become a vital challenge and focal point for sustainable development.

Green eco-innovation serves as the driving force behind the green transformation of industries, such as the metal industry, contributing to sustainable development [34]. For instance, Jadhao et al. [35] found that green eco-innovation enables the recovery of critical metals from electronic waste, such as printed circuit boards, thereby improving the recycling rate of metals. Li et al. [36] discovered a U-shaped relationship between green eco-innovation and the value chain of the metal industry. They highlighted that green eco-innovation can promote value chain upgrading by reducing pollution costs and facilitating the development of green products. Furthermore, the development of green eco-innovation in renewable energy, pollution control, and recycling has tangible effects on the metal industry. Jiskani et al. [37] revealed that green and climate-smart mining can achieve clean mineral extraction by protecting the environment and reducing ecological footprints, transforming traditional mining systems into more sustainable models. Xiong et al. [38] conducted a study on 430 heavy metal enterprises and found that the development of digital ecosystems, under the agglomeration effect of enterprises, can reduce pollution in the metal industry. Feng et al. [9] found that the application of green eco-innovation in the mineral industry, such as renewable energy, can significantly enhance the efficiency of green resource utilization, with regional heterogeneity present.

In summary, most studies have focused on the impact of green innovation on pollution and green utilization efficiency, with limited research on CMS. Firstly, there is a lack of research exploring the relationship between green eco-innovation and the critical metal industry, as well as its impact on CMS. Secondly, these studies have paid little attention to the specific effects of green eco-innovation on certain CMS and have not delved into the differences in primary and recycling supply. Thirdly, there is a scarcity of research on the mechanisms through which green eco-innovation affects CMS. To address these gaps, this study utilizes panel data from 2000 to 2019 for five critical metals in China, namely copper, aluminum, nickel, antimony, and titanium. The study first examines and summarizes the impacts of green eco-innovation on various CMS and their differences. It then analyzes the effects of green eco-innovation on CMS in different regions. Finally, by introducing energy efficiency, energy structure, and environmental regulations as moderating variables, the study explores the mechanisms through which green eco-innovation influences CMS.

2.2. Research Hypothesis

As the challenges of climate change become increasingly severe, industries across various sectors, particularly the mining and smelting industries, are placing greater emphasis on green eco-innovation. For example, Luo et al. [33] found that green eco-innovation plays a significant role in promoting sustainable development in the mining industry. However, metal resources are often limited. As the extraction volume gradually expands, green eco-innovation may, in turn, inhibit the increase in supply. This effect is particularly pronounced for critical metals, such as copper and aluminum, where mining technologies are already well-developed [1]. The extraction and selection of minor critical metals are often interrelated with bulk metals, such as antimony, which is frequently associated with tin ores [39]. The mining process for these metals is often more complex and requires advanced technologies to achieve stable extraction. Therefore, early-stage green eco-innovation may be unfavorable for ensuring their supply. On the other hand, minor critical metals like antimony and titanium are widely used in eco-friendly sectors, such as aerospace and new energy vehicles, which contributes to the green development of the environment [40,41]. In terms of recycling supply, unlike primary supply, it is more environmentally friendly [42]. In the process of green eco-innovation, an increasing number of discarded critical metals are being recycled, and the pollutants generated during the recycling process are being identified and treated [43]. Therefore, green eco-innovation is beneficial for ensuring the recycling of the CMS. However, it is worth noting that the technology required for recycling supply is often more advanced [15]. This implies that in the early stages of development, green eco-innovation may be unfavorable for increasing the recycling of the CMS. Hence, we propose hypotheses H1 and H2.

H1: 

There may be a nonlinear relationship between green eco-innovation and the CMS.

H2: 

The impact of green eco-innovation on different types of CMS may vary.

In the context of energy constraints, changes in energy factors have a significant impact on green innovation in a country or region. In industrial production, improving energy efficiency is an important measure for achieving green development [9]. For example, enhancing energy efficiency can reduce energy consumption and thereby decrease the emissions of carbon dioxide and other pollutants [44]. However, some studies have found that improvements in energy efficiency may lead to substitution and crowding-out effects, which can weaken the role of green eco-innovation [45]. Furthermore, green eco-innovation is beneficial for optimizing the energy structure, such as increasing the proportion of clean energy sources like wind and solar power [46]. Unlike traditional fossil fuels, the use of clean energy sources does not generate significant amounts of carbon dioxide, thereby avoiding harm to the ecological environment. Under their influence, it is possible to reduce reliance on fossil fuels effectively and ensure a sustainable CMS within the framework of green development [9]. Hence, we propose hypothesis H3.

H3: 

Improvements in energy efficiency and optimization of the energy structure will moderate the impact of green eco-innovation on the CMS.

The study also introduces environmental regulations as a moderating variable to discuss the impact mechanism of green eco-innovation on the CMS. Many studies have found that appropriate and reasonable environmental regulations can enhance the efficiency of industry resource allocation and contribute to the improvement of innovation levels, leading to a win-win situation for both the environment and the industry [47]. Singh et al. [48], by comparing the effects of national regulations on innovation in Japan and the European Union, found that stricter environmental regulations have a more effective promoting effect on innovation. Mahmood et al. [30] discovered that policy tools, such as environmental regulations, play a positive role in driving eco-innovation and green growth. However, for high-energy-consuming industries, several studies suggest that stringent environmental regulations may lead to increased investment in environmental protection and technological transformation, resulting in higher production costs and a lack of innovation motivation [49]. For instance, Brunnermeier and Cohen [50] found that government regulations do not necessarily enhance the green innovation capacity of the manufacturing industry. Kneller and Manderson [51] found that environmental management policies increase governance costs for industrial manufacturing firms, leading to crowding-out effects. By studying the effects of various segmented environmental regulations, Zhang et al. [52] discovered that voluntary environmental regulations inhibit green innovation. Therefore, we propose hypothesis H4.

H4: 

Strengthening environmental regulations will moderate the impact of green eco-innovation on the CMS.

3. Research Methods and Data

3.1. Variable Descriptions

3.1.1. Independent Variable

The core explanatory variable was provincial-level green eco-innovation (GEI). Following the method proposed by Du et al. [19], this study used the number of patents related to green ecological environment as an indicator of green eco-innovation. The data were obtained from the Chinese National Intellectual Property Administration (CNIPA).

3.1.2. Dependent Variable

The dependent variable was the primary production and recycling production (S) of five critical metals: copper, aluminum, nickel, antimony, and titanium. These metals include both bulk and minor metals, as well as primary and by-product metals, allowing us to thoroughly examine the differences in the impact of green eco-innovation on different types of metal supply. Following the approach of Zhang et al. [3], we excluded provinces with zero annual production. The data were sourced from the “China Nonferrous Metals Industry Statistical Yearbook”.

3.1.3. Control Variable

Advanced industrial structure (AS). Upgrading industrial structures contributes to the improvement of CMS [2]. In this study, the ratio of the production value between the tertiary industry and the secondary industry was used to measure the advancement of industrial structure. The data were sourced from the “China Statistical Yearbook”.

Foreign trade dependence (FD). Foreign trade is one of the important pathways for innovation diffusion and development [53]. In this study, the total import and export trade volumes were used to measure foreign trade dependence. The data were obtained from the statistical yearbooks of various provinces and cities.

Financial investments (FI). Government investment is an important source of funding for supporting industry development and can have a significant impact on industry growth [54]. In this study, the proportion of fiscal expenditure to GDP was used to measure financial investments. The data were sourced from the “China Regional Economic Statistical Yearbook”.

Pollution control (PC). Investments in pollution control in a region can impact both the local ecosystem and the production activities of high-emission industries [55]. In this study, the ratio of investment in industrial pollution control to industrial value added was used to measure the intensity of pollution control. The data were sourced from the “China Environmental Statistical Yearbook”.

Urbanization rate (UR). The process of urbanization is closely related to the supply of metals. Furthermore, the current development of urbanization places increasing importance on the construction of clean energy systems, which imposes higher demands on CMS [2]. This study measured urbanization using the proportion of urban population. The data were sourced from the “China Statistical Yearbook”.

3.1.4. Moderating Variables

In the exploration of mechanisms, we believe that energy efficiency (EE), energy structure (ES), and environmental regulation (ER) can moderate the impact of green eco-innovation on CMS.

This study measured energy efficiency using the GDP generated per unit of energy consumption. We measured energy transformation using the proportion of non-fossil energy consumption. Drawing from the approach of Pargal and Wheeler [56], this study measured environmental regulation. The data for the above variables were sourced from the “China Industrial Statistics Yearbook” and the “China Energy Statistics Yearbook”.

3.2. Data Sources and Descriptive Statistics

The data for the supply of critical metals used in this article were obtained from the China Nonferrous Metals Industry Statistical Yearbook. As the data are currently updated only until 2019, the dataset collected for this study describes the variables of 30 provinces in mainland China from 2000 to 2019. The study period was limited by data availability, and all variables were converted to constant prices based on the prices of the year 2000. The descriptive statistics are presented in

Table 2. Descriptive statistics.

VarName	Mean	SD	Min	Median	Max
lnS (Cu-p)	9.2340	4.212	0.00	10.93	14.11
lnS (Al-p)	12.0686	3.261	0.00	12.74	16.02
lnS (Ni-p)	6.7843	3.950	0.00	7.63	11.94
lnS (Sb-p)	7.8232	3.329	0.00	8.54	12.13
lnS (Ti-p)	5.9537	4.164	0.00	8.02	10.43
lnS (Cu-s)	9.7001	3.363	0.00	10.60	13.65
lnS (Al-s)	4.5792	5.383	0.00	0.00	13.69
lnGEI	6.8830	1.742	1.10	6.80	11.12
lnEE	−0.0172	0.549	−1.51	−0.03	1.57
lnES	−2.0081	0.266	−2.71	−2.02	−1.35
lnER	−1.8257	0.384	−2.67	−1.85	−0.18
lnC	6.1890	1.460	2.29	6.30	8.92
lnAS	−0.0095	0.287	−0.66	−0.00	1.66
lnFD	−1.8027	0.921	−4.48	−2.07	0.63
lnFI	−1.6929	0.423	−3.06	−1.68	−0.46
lnPS	8.3573	0.604	6.25	8.40	9.43
lnUR	−0.7606	0.301	−1.46	−0.74	−0.11
lnRGDP	10.0503	0.861	7.89	10.18	12.01
lnRD	−4.5733	0.599	−6.20	−4.57	−2.60
lnPerS	−0.0663	4.929	−9.31	1.58	8.33
lnPerGEI	−1.4743	1.594	−5.19	−1.49	2.76

. The correlations between variables did not exceed 0.6, and all variables had a variance inflation factor (VIF) of less than 4, indicating that there was no severe multicollinearity in the model.

3.3. Econometric Regression Models

3.3.1. Benchmark Regression Model

The benchmark regression model was constructed first. As can be seen from the previous article, the primary bulk CMS is closely related to the resource endowment, and in the early stages of supply, technological innovation favors the expansion of supply. However, reserves are finite, so after a certain level of technological innovation, the supply will instead be reduced. The primary minor CMS and the recycling CMS often require more advanced and sophisticated technologies. Moreover, they are still in the early stages of technological innovation. Only when technology has reached a certain level of development is it conducive to expanding supply [1,3]. To summarize, the effect of GEI on CMS should be nonlinear. Therefore, referring to Borenstein et al. [57] and Nickell [58] and Song et al. [59], we used a fixed effects model with fixed provinces in constructing the baseline regression model:

$$\ln S_{pt}={\alpha }_0+{\alpha }_1\ln GE{I}_{pt}+{\alpha }_2\ln GE{I}^2{}_{pt}+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}$$

(1)

where p and t stand for province and year, respectively. $\ln S_{pt}$ represents the supply of critical metals. $\ln GE{I}_{pt}$ represents green eco-innovation, which is the core independent variable. ${\mu }_P$ and ${\lambda }_t$ denote province fixed effects and time fixed effects, respectively. ${\epsilon }_{pt}$ is the disturbance error term.

From the above, it is clear that apart from GEI, advanced industrial structure (AS), foreign trade dependence (FD), financial investments (FI), pollution control (PC), and urbanization rate (UR) were also closely related to the supply of critical metals. Urbanization rate (UR) is also closely related to the supply of critical metals. In view of this, this paper incorporates these factors as control variables in Equation (1) and expresses the variables of the model in a more specific way by formulating Equation (1) as follows:

$$\left\{\begin{array}{l}\ln S_{pt}={\alpha }_0+{\alpha }_1\ln GE{I}_{pt}+{\alpha }_2\ln GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}\\ \theta X={\theta }_1\ln FD+{\theta }_2\ln AS+{\theta }_3\ln FI+{\theta }_4\ln UR+{\theta }_5\ln PC\end{array}\right\}$$

(2)

3.3.2. Identification

The explanatory variable green eco-innovation may suffer from endogeneity issues. The development of green eco-innovation, including clean energy innovation, is closely intertwined with CMS, which can give rise to a potential reverse causal relationship. Hence, referring to Zhang and Song, this study constructs a simultaneous equation model to ascertain the existence of a reverse causal relationship between green eco-innovation and CMS.

$$\ln S_{pt}={\alpha }_0+{\alpha }_1\ln GE{I}_{pt}+{\alpha }_2\ln GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}$$

(3)

$$\ln GE{I}^2{}_{pt}={\beta }_0+{\beta }_1\ln S_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}$$

(4)

Equation (3) represents the determination equation of green eco-innovation on CMS, which is consistent with Equation (1). Equation (4) represents the determination equation of CMS on green eco-innovation. Considering that research and development investment (RD), per capita GDP (RGDP), and production cost (C) are closely related to green eco-innovation, these variables are included as control variables.

Then, green eco-innovation often exhibits a certain degree of time lag. The birth of new innovation and its application in society usually take time to have an impact on production [2]. Hence, this study utilized a lagged effects model to investigate the existence of a delayed impact between green eco-innovation and CMS. We lagged lnGEI by one order and constructed a lagged effect model to study the effect of lnGEI in period t − 1 on period t’s lnS.

$$\ln S_{pt}={\alpha }_0+{\alpha }_{1}L.\ln GE{I}_{pt}+{\alpha }_{2}L.\ln GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}$$

(5)

In this econometric model, $L.\ln GE{I}_{pt}$ denote $\ln GE{I}_{pt}$ lags of first order.

Finally, in order to avoid the influence of the size of each province on green eco-innovation and CMS, this paper introduced per capita CMS (PerS) and per capita green eco-innovation (PerGEI) as new explanatory variables. The following model was constructed:

$$\ln P\mathrm{er}{S}_{pt}={\alpha }_0+{\alpha }_1\ln P\mathrm{er}GE{I}_{pt}+{\alpha }_2\ln P\mathrm{er}GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}$$

(6)

3.3.3. Influence Mechanism Test

To verify whether energy efficiency, energy structure, and environmental regulation have an impact on the relationship between green eco-innovation and CMS, the following moderation effects model was constructed.

$$\left\{\begin{array}{l}\ln S={\alpha }_0+{\alpha }_1\ln GE{I}_{pt}+{\alpha }_2\ln GE{I}^2{}_{pt}+{\beta }_1\ln E{E}_{pt}+{\beta }_2\ln E{E}_{pt}*\ln GE{I}_{pt}+{\beta }_3\ln E{E}_{pt}*\ln GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}\\ \ln S={\alpha }_0+{\alpha }_1\ln GE{I}_{pt}+{\alpha }_2\ln GE{I}^2{}_{pt}+{\beta }_1\ln E{S}_{pt}+{\beta }_2\ln E{S}_{pt}*\ln GE{I}_{pt}+{\beta }_3\ln E{S}_{pt}*\ln GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}\\ \ln S={\alpha }_0+{\alpha }_1\ln GE{I}_{pt}+{\alpha }_2\ln GE{I}^2{}_{pt}+{\beta }_1\ln E{R}_{pt}+{\beta }_2\ln E{R}_{pt}*\ln GE{I}_{pt}+{\beta }_3\ln E{R}_{pt}*\ln GE{I}^2{}_{pt}+\theta X+{\mu }_P+{\lambda }_t+{\epsilon }_{pt}\end{array}\right\}$$

(7)

Equation (7) represents the nonlinear moderation effect model, and derivation of Equation (7) yields the turning point of the U-curve:

$${(\ln GEI)}^{*}=\frac{-{\alpha }_1-{\beta }_2\ln Y}{2{\alpha }_2+2{\beta }_3\ln Y}$$

(8)

where Y means a moderating variable. To illustrate how the turning point changes depending on the moderating variable, we derive this equation for the moderating variable:

$$\frac{\delta (\ln GEI)}{\delta (\ln Y)}=\frac{{\alpha }_1{\beta }_3-{\alpha }_2{\beta }_2}{2{({\alpha }_2+{\beta }_3\ln Y)}^2}$$

(9)

Since the denominator is strictly greater than 0, the direction in which the turning point moves depends on the sign of the numerator:

$$Ci={\alpha }_1{\beta }_3-{\alpha }_2{\beta }_2$$

(10)

Equation (10) is used to determine the direction of the inflection point shift. In the nonlinear moderation effect model, the moderating variable primarily serves to induce inflection point shifts and alter the steepness of the curve. As shown in the Figure 1 below, when Ci > 0, as Y decreases, the inflection point shifts to the left. As Y increases, for a U-shaped curve, when ${\beta }_3$ > 0, the curve becomes steeper.

4. Empirical Results

4.1. Changes in National Supply of Critical Metals

From Figure 2, Figure 3 and Figure 4, we can see that the primary supply and secondary supply of five critical metals (copper, aluminum, nickel, antimony, and titanium) and the application for green eco-innovation basically showed an upward trend in China from 2000 to 2019. This is consistent with Song et al. [2] that critical metals play an increasingly important role in modern production and life, and securing the supply of critical metals has become an important task for all countries. Among them, the primary supply of aluminum has seen the biggest boost, growing from 2.67 million tons in 2000 to 36 million tons in 2019, an increase of nearly 13 times. As we know, in industrial production, aluminum is widely used in automotive, aerospace, and other industries, which leads to the expanding demand for aluminum and affects its supply. The chart also shows that the recycling supply of aluminum increased more sharply between 2013 and 2014. This may be due to China’s 12th Five-Year Plan, which encourages the development of advanced recycling technologies, which in turn has contributed to the increase in aluminum recycling supply [2].

In addition, Figure 5 illustrates a map of the distribution of CMS. As can be seen from the figure, CMS is highly regional in nature. This is because, unlike demand, CMS is dependent on the resource endowment of each region, and it is closely related to the metal mineral content of each region.

4.2. Benchmark Regression Analysis

Table 3. Benchmark regression analysis.

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
		Cu-p	Al-p	Ni-p	Sb-p	Ti-p	Cu-r	Al-r
Variables	Parameters	lnS	lnS	lnS	lnS	lnS	lnS	lnS
lnGEI	α1	2.379 *	3.303 ***	3.741	−3.507 **	−4.610 *	−2.224	−2.068 ***
		(1.259)	(0.655)	(3.178)	(1.385)	(2.615)	(1.495)	(0.778)
(lnGEI)2	α2	−0.146 **	−0.275 ***	−0.328 *	0.189 **	0.327 *	0.134 *	0.150 ***
		(0.0641)	(0.0348)	(0.194)	(0.0852)	(0.171)	(0.0779)	(0.0403)
lnFD	θ1	0.892	1.521 ***	1.062	1.834 ***	1.723	−1.913 **	−1.207 ***
		(0.569)	(0.383)	(1.428)	(0.625)	(1.146)	(0.882)	(0.436)
lnAS	θ2	0.380	−1.733 **	−5.184 *	1.358	2.305	−1.303	−2.030 **
		(1.144)	(0.854)	(3.115)	(1.810)	(2.339)	(1.562)	(0.913)
lnFI	θ3	−0.910	−1.432	−2.739	4.069 *	−3.462 *	5.061 **	−0.0129
		(1.212)	(0.920)	(2.034)	(2.084)	(1.816)	(2.076)	(0.974)
lnUR	θ4	6.393 ***	8.970 ***	6.978	8.926 **	8.931 *	12.25 ***	7.686 ***
		(2.358)	(2.074)	(11.03)	(4.229)	(4.887)	(2.803)	(1.837)
lnPC	θ5	−0.277	−0.372 *	−0.0512	0.849 *	−1.514 ***	−0.124	0.152
		(0.273)	(0.211)	(0.745)	(0.471)	(0.512)	(0.401)	(0.214)
Constant	α0	6.364	12.33 ***	1.255	41.80 ***	27.04 *	36.42 ***	15.79 ***
		(6.503)	(4.348)	(21.60)	(9.342)	(13.73)	(8.671)	(4.416)
Year effect		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province effect		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations		460	420	140	160	120	300	540

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

displays the impact of green eco-innovation on CMS. Columns (1)–(5) reflect the effects of green eco-innovation on the primary supply of copper, aluminum, nickel, antimony, and titanium. Columns (6)–(7) reflect the effects of green eco-innovation on the recycling supply of copper and aluminum. All equations control for year fixed effects and provincial fixed effects.

The coefficients of (lnGEI)² on the primary supply of copper (Cu-p), aluminum (Al-p), and nickel (Ni-p) were −0.146, −0.275, and −0.328, respectively, all statistically significant. Green eco-innovation exhibits a significant inverted U-shaped effect on Cu-p, Al-p, and Ni-p. This implies that as the capability of green eco-innovation strengthens, the supply of Cu-p, Al-p, and Ni-p initially increases and then decreases. This is because, in the initial stage of supply, the continuous development of green eco-innovation helps increase the metal supply. However, mineral reserves are limited, and the mining process often involves pollution, which is not conducive to green development. Therefore, when green eco-innovation reaches a certain stage of development, it turns to inhibiting metal supply. This effect is more pronounced for metals such as copper and aluminum, which have larger supplies and wider usage [1]. Further analysis reveals that the turning points for Cu-p, Al-p, and Ni-p were 8.147, 6.005, and 5.702, respectively. And the lnGEI for these three metals was 10.115, 11.303, and 7.548 in 2019. This indicates that the current green eco-innovation in Cu-p, Al-p, and Ni-p is located on the descending phase of the inverted U-shaped curve, corresponding to the descending phase. This finding is consistent with the discovery of Song et al. [2] that some sub-sectors of metal in China have experienced overcapacity.

The impact coefficients of (lnGEI)² on the primary supply of antimony (Sb-p) and titanium (Ti-p) are 0.189 and 0.327, respectively, and both are statistically significant. This indicates that green eco-innovation has a U-shaped effect on Sb-p and Ti-p. Unlike bulk metals such as copper, aluminum, and nickel, antimony, and titanium are minor metals and associated minerals. Their mining processes are more complex and require a certain level of innovation development to be beneficial for supply [3]. Moreover, antimony and titanium are widely used in the construction of aerospace, new energy vehicles, and other clean energy sectors, which contribute to ecological green development [40,41]. Additionally, the turning points for Sb-p and Ti-p were 9.277 and 7.084, and the lnGEI values for these metals in 2019 were 8.869 and 6.580, respectively. This indicates that the current green eco-innovation for Sb-p and Ti-p is on the left side of the U-shaped curve. Sb-p and Ti-p are still in the initial stage of supply, where green eco-innovation cannot yet promote an increase in their supply. However, as green eco-innovation continues to develop, the supply of Sb-p and Ti-p will gradually show an upward trend.

The impact coefficients of (lnGEI)2 on the recycling supply of copper (Cu-r) and aluminum (Al-r) are 0.134 and 0.150, and they are significant at the 10% and 1% levels. This indicates that green eco-innovation has a U-shaped effect on Cu-r and Al-r, which is similar to the findings of Zhang et al. [3]. Recycling the supply of metals requires advanced technology, and it is only beneficial for supply improvement when the technology reaches a certain level. Moreover, compared to direct mining, the recycling supply is beneficial for protecting the ecological environment. Therefore, as green eco-innovation develops to a certain degree, it positively affects the recycling supply of copper and aluminum. Currently, both Cu-r and Al-r are situated on the right side of the U-shaped curve. This means that green eco-innovation is currently in a phase favorable for Cu-r and Al-r. Hence, H1 and H2 are supported. The specific values for time and province fixed effects are shown in

Table A1. Benchmark regression analysis.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Cu-p	Al-p	Ni-p	Sb-p	Ti-p	Cu-r	Al-r
Variables	lnS	lnS	lnS	lnS	lnS	lnS	lnS
lnGEI	2.379 *	3.303 ***	3.741	−3.507 **	−4.610 *	−2.224	−2.068 ***
	(1.259)	(0.655)	(3.178)	(1.385)	(2.615)	(1.495)	(0.778)
(lnGEI)2	−0.146 **	−0.275 ***	−0.328 *	0.189 **	0.327 *	0.134 *	0.150 ***
	(0.0641)	(0.0348)	(0.194)	(0.0852)	(0.171)	(0.0779)	(0.0403)
lnFD	0.380	−1.733 **	−5.184 *	1.358	2.305	−1.303	−2.030 **
	(1.144)	(0.854)	(3.115)	(1.810)	(2.339)	(1.562)	(0.913)
lnAS	0.892	1.521 ***	1.062	1.834 ***	1.723	−1.913 **	−1.207 ***
	(0.569)	(0.383)	(1.428)	(0.625)	(1.146)	(0.882)	(0.436)
lnFI	−0.910	−1.432	−2.739	4.069 *	−3.462 *	5.061 **	−0.0129
	(1.212)	(0.920)	(2.034)	(2.084)	(1.816)	(2.076)	(0.974)
lnUR	6.393 ***	8.970 ***	6.978	8.926 **	8.931 *	12.25 ***	7.686 ***
	(2.358)	(2.074)	(11.03)	(4.229)	(4.887)	(2.803)	(1.837)
lnPC	−0.277	−0.372 *	−0.0512	0.849 *	−1.514 ***	−0.124	0.152
	(0.273)	(0.211)	(0.745)	(0.471)	(0.512)	(0.401)	(0.214)
_Iyear_2001	0.138	0.195	−2.312	−0.600	0.796	−2.292 **	0.154
	(0.803)	(0.623)	(1.684)	(1.027)	(1.299)	(0.989)	(0.660)
_Iyear_2002	−0.496	0.190	−0.0328	−1.335	0.0657	−1.413	−1.520 **
	(0.838)	(0.658)	(2.017)	(1.123)	(1.336)	(1.030)	(0.683)
_Iyear_2003	−0.823	−0.324	−1.142	−2.493 **	0.844	−0.919	0.0655
	(0.865)	(0.672)	(2.369)	(1.193)	(1.325)	(1.065)	(0.699)
_Iyear_2004	−0.963	−0.612	−2.024	−1.785	0.768	−0.459	−4.931 ***
	(0.891)	(0.705)	(2.814)	(1.281)	(1.390)	(1.094)	(0.714)
_Iyear_2005	−1.394	−0.825	−2.665	−1.860	1.809	−1.178	−5.110 ***
	(0.982)	(0.768)	(3.316)	(1.448)	(1.482)	(1.196)	(0.774)
_Iyear_2006	−1.265	−0.952	−2.825	−2.429	3.862 **	−1.111	−5.195 ***
	(1.078)	(0.854)	(3.780)	(1.677)	(1.696)	(1.312)	(0.846)
_Iyear_2007	−1.400	−0.774	−1.850	−3.760 **	6.134 ***	−1.526	−5.354 ***
	(1.174)	(0.915)	(4.328)	(1.866)	(1.763)	(1.414)	(0.909)
_Iyear_2008	−1.136	−0.871	−0.195	−3.965 *	7.807 ***	−1.067	−5.558 ***
	(1.297)	(1.025)	(4.786)	(2.179)	(1.988)	(1.584)	(1.005)
_Iyear_2009	−0.950	−0.206	0.825	−3.867	7.232 ***	−2.483	−5.982 ***
	(1.495)	(1.149)	(4.953)	(2.572)	(2.240)	(1.865)	(1.144)
_Iyear_2010	−0.802	−0.589	0.910	−4.036	7.424 ***	−2.429	−6.234 ***
	(1.553)	(1.217)	(5.417)	(2.817)	(2.352)	(1.957)	(1.193)
_Iyear_2011	−1.161	−0.878	1.129	−3.380	8.192 ***	−1.887	−6.465 ***
	(1.655)	(1.311)	(5.910)	(3.077)	(2.429)	(2.121)	(1.272)
_Iyear_2012	−1.115	−0.629	1.671	−2.515	7.850 ***	−2.595	−6.393 ***
	(1.818)	(1.442)	(6.309)	(3.432)	(2.758)	(2.312)	(1.408)
_Iyear_2013	−1.746	−0.268	2.237	−3.092	8.569 ***	−2.785	3.278 **
	(1.935)	(1.554)	(6.798)	(3.641)	(2.888)	(2.423)	(1.502)
_Iyear_2014	−0.683	−0.832	2.726	−4.080	8.024 ***	−2.953	3.941 **
	(2.044)	(1.633)	(7.134)	(3.970)	(3.072)	(2.567)	(1.591)
_Iyear_2015	−1.114	−0.895	3.551	−4.741	5.125	−4.453	3.094 *
	(2.255)	(1.768)	(7.400)	(4.373)	(3.441)	(2.887)	(1.760)
_Iyear_2016	−1.172	−0.753	3.163	−4.502	4.410	−5.642 *	3.030
	(2.424)	(1.893)	(7.669)	(4.818)	(3.882)	(3.142)	(1.898)
_Iyear_2017	−1.257	−1.207	2.774	−4.965	3.427	−5.163	2.909
	(2.533)	(1.977)	(8.037)	(5.215)	(4.189)	(3.321)	(1.992)
_Iyear_2018	−1.602	−1.299	2.967	−4.470	2.132	−5.644	2.827
	(2.593)	(2.032)	(8.372)	(5.396)	(4.428)	(3.447)	(2.043)
_Iyear_2019	−1.867	−1.245	3.106	−4.585	2.022	−6.717 *	2.683
	(2.681)	(2.097)	(8.526)	(5.517)	(4.582)	(3.552)	(2.114)
_Iprovince_2	−2.086	0.000	0.000	0.000	0.000	0.610	1.928
	(6.358)	(0.000)	(0.000)	(0.000)	(0.000)	(1.412)	(1.551)
_Iprovince_3	2.188	0.000	0.000	0.000	−1.043	3.803	5.283 **
	(6.906)	(0.000)	(0.000)	(0.000)	(1.596)	(2.920)	(2.172)
_Iprovince_4	9.500	13.00 ***	0.000	0.000	−0.167	0.000	0.759
	(6.130)	(4.159)	(0.000)	(0.000)	(1.950)	(0.000)	(2.329)
_Iprovince_5	9.662 *	13.44 ***	0.000	0.000	0.000	0.000	1.180
	(5.630)	(3.797)	(0.000)	(0.000)	(0.000)	(0.000)	(2.219)
_Iprovince_6	5.586	8.105 *	0.000	0.000	0.000	−1.307	1.987
	(6.577)	(4.319)	(0.000)	(0.000)	(0.000)	(1.980)	(1.732)
_Iprovince_7	3.496	0.000	0.000	0.000	0.000	0.000	−0.918
	(6.142)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(1.878)
_Iprovince_8	1.075	2.524	0.000	0.000	0.000	0.000	1.490
	(6.236)	(4.120)	(0.000)	(0.000)	(0.000)	(0.000)	(2.067)
_Iprovince_9	1.461	0.000	0.000	0.000	0.000	0.000	2.918 **
	(6.622)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(1.164)
_Iprovince_10	6.690	4.857	0.000	0.000	0.000	7.400 ***	3.937 **
	(7.259)	(4.935)	(0.000)	(0.000)	(0.000)	(2.245)	(1.953)
_Iprovince_11	5.541	3.939	0.000	0.000	0.000	7.308 ***	3.632 **
	(7.083)	(4.772)	(0.000)	(0.000)	(0.000)	(2.022)	(1.827)
_Iprovince_12	11.63 *	0.000	0.000	0.000	0.000	5.534 **	3.819 *
	(6.756)	(0.000)	(0.000)	(0.000)	(0.000)	(2.805)	(2.057)
_Iprovince_13	4.203	7.401	0.000	0.000	0.000	0.000	2.200
	(6.745)	(4.610)	(0.000)	(0.000)	(0.000)	(0.000)	(2.060)
_Iprovince_14	11.68 *	0.000	−2.164	42.57 ***	0.000	5.700 *	0.000
	(6.301)	(0.000)	(21.17)	(9.156)	(0.000)	(3.222)	(0.000)
_Iprovince_15	8.651	12.68 **	0.000	0.000	0.000	7.920 ***	2.528
	(7.278)	(4.987)	(0.000)	(0.000)	(0.000)	(2.716)	(2.101)
_Iprovince_16	10.49	16.86 ***	0.000	41.77 ***	4.387 *	6.456 *	3.239
	(7.090)	(4.962)	(0.000)	(10.19)	(2.430)	(3.493)	(2.422)
_Iprovince_17	9.583	12.10 ***	0.000	0.000	0.000	3.471	2.449
	(6.788)	(4.589)	(0.000)	(0.000)	(0.000)	(2.765)	(1.922)
_Iprovince_18	0.000	11.74 **	0.000	47.75 ***	0.000	2.865	2.607
	(0.000)	(4.730)	(0.000)	(9.532)	(0.000)	(3.026)	(2.016)
_Iprovince_19	0.000	0.000	0.000	33.67 ***	0.000	6.806 ***	2.686
	(0.000)	(0.000)	(0.000)	(9.182)	(0.000)	(1.736)	(1.721)
_Iprovince_20	3.955	14.59 ***	1.020	43.89 ***	0.000	0.000	2.992
	(6.579)	(4.583)	(21.92)	(9.392)	(0.000)	(0.000)	(2.122)
_Iprovince_21	2.394	10.77 **	−0.310	0.000	0.000	0.000	3.554 *
	(6.392)	(4.300)	(20.13)	(0.000)	(0.000)	(0.000)	(1.925)
_Iprovince_22	3.782	15.35 ***	−0.995	0.000	2.603	0.000	3.545 *
	(7.014)	(4.829)	(23.16)	(0.000)	(2.205)	(0.000)	(2.112)
_Iprovince_23	0.000	18.30 ***	0.000	43.24 ***	8.406 ***	0.000	4.107
	(0.000)	(4.462)	(0.000)	(9.423)	(2.937)	(0.000)	(2.539)
_Iprovince_24	13.08 **	16.97 ***	1.855	43.75 ***	0.000	3.389	3.885 *
	(6.580)	(4.585)	(22.98)	(9.562)	(0.000)	(3.573)	(2.347)
_Iprovince_25	4.052	13.16 ***	0.000	0.000	0.000	0.000	1.189
	(6.560)	(4.452)	(0.000)	(0.000)	(0.000)	(0.000)	(2.159)
_Iprovince_26	13.63 **	17.60 ***	7.580	37.76 ***	0.000	−1.501	3.117
	(6.187)	(4.341)	(22.23)	(9.264)	(0.000)	(3.919)	(2.515)
_Iprovince_27	0.000	18.43 ***	0.000	0.000	0.000	0.000	0.000
	(0.000)	(3.390)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
_Iprovince_28	0.000	14.33 ***	0.000	0.000	0.000	0.000	0.000
	(0.000)	(3.586)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
_Iprovince_29	6.132	12.83 ***	1.800	0.000	0.000	0.000	2.537
	(5.976)	(4.140)	(19.89)	(0.000)	(0.000)	(0.000)	(2.350)
Constant	6.364	12.33 ***	1.255	41.80 ***	27.04 *	36.42 ***	15.79 ***
	(6.503)	(4.348)	(21.60)	(9.342)	(13.73)	(8.671)	(4.416)
Observations	460	420	140	160	120	300	540

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

of Appendix A.

In the controlled variables, the impacts of trade dependency and urbanization level on most CMS are positive, indicating that economic development and urbanization processes will expand the demand for critical metals, thereby driving an increase in supply. On the other hand, financial investments and pollution control have a negative impact on most CMS. This is because metal production often involves certain levels of pollution, and governments are more inclined to support environmentally friendly industries. The increasing investments in environmental governance also have a negative impact on the metal industry, which belongs to the resource-intensive and highly polluting sectors.

4.3. Robustness Analysis

4.3.1. Simultaneous Equation Estimation Results

In order to identify the potential reverse causal relationship between green eco-innovation and CMS, this study employs simultaneous equation estimation. The results, presented in

Table 4. Results of simultaneous equation estimation.

		(1)	(2)	(3)	(4)
		Cp and Ap and Np		Sp and Tp and Cr and Ar
VARIABLES	Parameters	lnS	(lnGEI)2	lnS	(lnGEI)2
lnGEI	α1	7.724 *		−14.22
		(4.297)		(9.475)
(lnGEI)2	α2	−0.517 **		1.138 **
		(0.263)		(0.566)
lnS	β1		−27.04		0.193
			(23.82)		(0.417)
lnFD	θ1	2.577 ***	29.77	−3.373
		(0.991)	(26.00)	(2.330)
lnAS	θ2	0.737	−55.03	−5.953 **
		(1.776)	(53.88)	(2.704)
lnFI	θ3	−1.574 *		5.580 **	−3.871 **
		(0.939)		(2.516)	(1.676)
lnUR	θ4	1.281	213.7	19.93 *	−4.215
		(4.293)	(187.5)	(11.13)	(3.156)
lnPC	θ5	−0.608 *		0.811	−0.166
		(0.363)		(0.668)	(0.321)
lnRD	θ6		−31.48
			(37.61)
lnRGDP	θ7				2.161
					(1.656)
lnC	θ8				0.0615
					(0.815)
Constant	α0/β0	−21.97	144.2	63.40	18.05
		(18.01)	(106.6)	(44.73)	(14.16)
Year effect		Yes	Yes	Yes	Yes
Province effect		Yes	Yes	Yes	Yes
Observations		1020	1020	1120	1120

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1. Cp, Ap, and Np represent primary supply of copper, aluminum, and nickel; Sp, Tp, Cr, and Ar represent primary supply of antimony and titanium and recycling supply of copper and aluminum, same below.

, unveil the following findings.

Columns (1) and (3) display the impact of green eco-innovation on CMS, while columns (2) and (4) report the reverse impact. It can be observed that the coefficients of CMS on green eco-innovation are not statistically significant, indicating that green eco-innovation is not influenced by CMS. Moreover, after considering the potential reverse causal relationship between the independent and dependent variables, the coefficients of lnGEI² on CMS were −0.517 and 1.138, and these coefficients passed the 5% significance level test. Therefore, even after considering the reverse causal relationship between green eco-innovation and CMS, the conclusions remain valid.

4.3.2. Results Excluding Lagged Effects

Given the time lag between the invention and implementation of a new innovation, this study considered the lagged effect of green eco-innovation on CMS [2].

As shown in

Table 5. Lagged effect results.

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
		Cu-p	Al-p	Ni-p	Sb-p	Ti-p	Cu-r	Al-r
Variables	Parameters	lnS	lnS	lnS	lnS	lnS	lnS	lnS
L.lnGEI	α1	3.620 ***	3.538 ***	4.200	−4.049 ***	−5.907 **	−2.805 *	−1.706 **
		(1.319)	(0.695)	(3.620)	(1.474)	(2.828)	(1.527)	(0.784)
(L.lnGEI)2	α2	−0.204 ***	−0.301 ***	−0.336 *	0.266 ***	0.442 **	0.153 *	0.134 ***
		(0.0690)	(0.0374)	(0.217)	(0.0948)	(0.179)	(0.0816)	(0.0419)
lnFD	θ1	1.206 **	1.633 ***	1.089	1.897 ***	1.562	−2.269 **	−0.921 **
		(0.579)	(0.393)	(1.476)	(0.637)	(1.181)	(0.876)	(0.432)
lnAS	θ2	1.276	−1.126	−6.217 *	1.796	4.123	−1.468	−0.578
		(1.183)	(0.915)	(3.435)	(1.983)	(2.484)	(1.601)	(0.916)
lnFI	θ3	−2.079	−2.589 **	−2.392	3.920 *	−6.293 **	5.358 **	−1.336
		(1.532)	(1.223)	(4.159)	(2.172)	(2.638)	(2.118)	(1.164)
lnUR	θ4	6.197 **	9.418 ***	8.365	12.32 ***	10.41 *	14.77 ***	7.358 ***
		(2.585)	(2.291)	(12.49)	(4.625)	(5.560)	(3.040)	(1.936)
lnPC	θ5	−0.286	−0.371 *	−0.0953	0.828 *	−1.333 **	−0.109	0.249
		(0.286)	(0.224)	(0.763)	(0.463)	(0.543)	(0.407)	(0.218)
Constant	α0	−3.430	9.540 *	2.709	37.77 ***	27.05 *	34.57 ***	14.28 ***
		(8.057)	(5.116)	(25.01)	(9.472)	(15.89)	(9.331)	(5.154)
Year effect		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province effect		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations		437	399	133	152	114	285	513

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

, the coefficients of the lagged (lnGEI)² on the supply of Cu-p, Al-p, and Ni-p were −0.204, −0.301, and −0.336, respectively, indicating an inverted U-shaped effect. The coefficients for the impact on the supply of Sb-p and Ti-p were 0.266 and 0.442, respectively, indicating a U-shaped effect. The coefficients for the impact on the supply of Cu-r and Al-r were 0.153 and 0.134, respectively, also indicating a U-shaped effect. All of these results were statistically significant and consistent with the baseline regression. Therefore, even after considering the lagged effect of innovation, the conclusion remains valid.

4.3.3. Inverted U Test

The results of the U-test are presented in

Table 6. Inverted U-shaped impact test.

	Cp and Ap and Np		Sp and Tp and Cr and Ar
	Lower Bound	Upper Bound	Lower Bound	Upper Bound
Interval	1.098	11.116	1.098	11.116
Slope	3.157	−0.966	−0.350	1.265
t-value	5.130	−1.947	−0.398	1.919
p > |t|	1.75 × 10−7	0.025	0.045	0.027
Extreme point	8.767		3.270

. It can be observed that the turning points of the impact of green eco-innovation on the two types of supply were 8.767 and 3.270, respectively. The range of values for green eco-innovation was [1.098, 11.116], and both of these turning points fell within this range. Additionally, the slope had a negative sign within the interval for all cases. Therefore, we can conclude that both the inverted U-shaped and U-shaped impacts are reliable.

4.3.4. Results of Excluding Demographic Effects

Considering that the scale of development varies across provinces due to demographic factors, we replaced the explanatory variable with per capita CMS and the core explanatory variable with per capita green eco-innovation. The results obtained are shown in

Table 7. Results of excluding demographic effects.

		(1)	(2)
		Cp and Ap and Np	Sp and Tp and Cr and Ar
Variables	Parameters	lnPerS	lnPerS
lnPerGEI	α1	−1.102 ***	0.0544
		(0.341)	(0.456)
(lnPerGEI)2	α2	−0.439 ***	0.181 **
		(0.0562)	(0.0729)
lnFD	θ1	−1.208 *	−0.257
		(0.690)	(0.421)
lnAS	θ2	1.191 ***	−3.277 ***
		(0.298)	(0.966)
lnFI	θ3	−1.749 **	1.689
		(0.724)	(1.064)
lnUR	θ4	0.539	7.863 ***
		(1.845)	(2.324)
lnPC	θ5	−0.222	−0.0582
		(0.171)	(0.225)
Constant	α0	1.940	6.406 **
		(2.337)	(3.122)
Year effect		Yes	Yes
Province effect		Yes	Yes
Observations		1020	1120

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

.

As can be seen in Table 7, the effect of per capita eco-innovation capacity on per capita CMS is consistent with the benchmark regression. There is a U-shaped effect on the primary supply of copper, aluminum, and nickel and an inverted U effect on the primary supply of antimony and titanium, and the recycling supply of critical metals. Thus, we can be confident that the results obtained after excluding the effect of demographic factors are trustworthy.

4.4. Impact of Population Size on the Supply of Critical Metals

Considering that population size may have some impact on CMS, this paper added the population size of each province as a variable to study the impact of green eco-innovations on CMS. As seen in

Table 8. Results of adding population size.

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
		Cu-p	Al-p	Ni-p	Sb-p	Ti-p	Cu-s	Al-s
Variables	Parameters	lnS	lnS	lnS	lnS	lnS	lnS	lnS
lnGEI	α1	2.387 *	3.194 ***	4.497	−2.680 *	−4.198	−2.284	−2.035 ***
		(1.253)	(0.652)	(3.222)	(1.492)	(2.774)	(1.499)	(0.777)
(lnGEI)2	α2	−0.126 *	−0.286 ***	−0.404 **	0.123	0.314 *	0.130 *	0.132 ***
		(0.0644)	(0.0348)	(0.202)	(0.0963)	(0.174)	(0.0782)	(0.0413)
lnFD	θ1	0.548	−1.135	−5.172 *	1.780	2.819	−1.697	−2.125 **
		(1.140)	(0.878)	(3.106)	(1.826)	(2.598)	(1.665)	(0.912)
lnAS	θ2	0.963 *	1.499 ***	1.302	1.884 ***	1.735	−1.900 **	−1.147 ***
		(0.567)	(0.380)	(1.436)	(0.624)	(1.151)	(0.883)	(0.436)
lnFI	θ3	−1.247	−1.210	−2.925	4.882 **	−3.374 *	5.186 **	0.286
		(1.215)	(0.917)	(2.033)	(2.149)	(1.834)	(2.086)	(0.985)
lnUR	θ4	4.471 *	10.17 ***	−1.919	11.40 **	8.624 *	13.68 ***	9.224 ***
		(2.491)	(2.109)	(12.98)	(4.541)	(4.953)	(3.483)	(2.012)
lnPC	θ5	−0.233	−0.455 **	0.0149	0.904 *	−1.587 ***	−0.161	0.138
		(0.272)	(0.211)	(0.745)	(0.470)	(0.538)	(0.405)	(0.214)
lnPS	θ6	−6.643 **	7.357 ***	−13.91	9.491	4.919	3.316	4.275 *
		(2.901)	(2.829)	(10.77)	(6.539)	(10.59)	(4.808)	(2.313)
Constant	α0	58.22 **	−45.25 **	102.5	−36.73	−16.91	10.53	−16.76
		(23.55)	(22.56)	(81.27)	(54.90)	(95.66)	(38.53)	(18.15)
Year effect		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province effect		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations		460	420	140	160	120	300	540

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

, after adding the population size factor, the impact of green eco-innovations on CMS was consistent with the benchmark regression, except for the primary supply of antimony.

From the effect of population size on CMS, we can also see that population size had a significant effect only on the primary supply of copper and aluminum and the recycling supply of aluminum. The impacts were significantly negative on the primary supply of copper and significantly positive on the primary and recycling supply of aluminum. Their impact coefficients were −6.643, 7.357, and 4.275, respectively. This means that for every 1% increase in population size, the primary supply of copper decreased by 6.643%, and the primary and recycling supply of aluminum increased by 7.357% and 4.275%, respectively. This could be due to the fact that the aluminum industry has a stronger dependence on population, so as the population rises, it boosts the supply of aluminum to some extent.

4.5. Heterogeneity Analysis

The above results were derived from the complete sample; however, the influence of green eco-innovation on CMS may differ among various regions. Hence, conducting sub-sample regressions becomes imperative to assess the variations. Different regions have significant differences in industrial levels and stages of development, which may result in variations in the development of green eco-innovation and its impact on CMS across regions [60]. In light of this, we divided the 30 provinces into industrially developed and industrially underdeveloped regions and performed heterogeneity regressions. The results are presented in

Table 9. Heterogeneity analysis results.

		(1)	(2)	(3)	(4)
		Industrially Developed Regions	Industrial Underdeveloped Areas	Industrially Developed Region	Industrial Underdeveloped Areas
		Cp and Ap and Np		Sp and Tp and Cr and Ar
Variables	Parameters	lnS	lnS	lnS	lnS
lnGEI	α1	0.941 *	4.339 ***	−2.598 ***	−0.196
		(0.508)	(0.509)	(0.636)	(0.653)
(lnGEI)2	α2	−0.0535 *	−0.254 ***	0.155 ***	0.00124
		(0.0280)	(0.0308)	(0.0351)	(0.0343)
Constant	α0	8.529 **	−12.29 ***	23.40 ***	0.122
		(3.316)	(2.897)	(5.711)	(8.949)
Year effect		Yes	Yes	Yes	Yes
Province effect		Yes	Yes	Yes	Yes
Control variables		Yes	Yes	Yes	Yes
Observations		1940	1300	1480	1860

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

.

For the primary supply of bulk critical metals, the impact of green eco-innovation on industrially developed and underdeveloped regions was consistent with the benchmark regression, showing an inverted U-shaped relationship. For the supply of recycling and minor critical metals, the impact of green eco-innovation on industrially developed and underdeveloped regions was also consistent with the benchmark regression, showing a U-shaped relationship. However, it can be observed that the impact of green eco-innovation on underdeveloped regions was stronger. Of these, the impact of green eco-innovations was not significant for recycling and minor critical metals. For bulk critical metals, the significance was also significantly weaker. This may be attributed to the fact that underdeveloped regions are mostly located in the northwest and southwest of the country, where the utilization of clean energy, such as wind and solar power, is relatively high, leading to faster development of green eco-innovation. For industrially developed regions, although they have gradually placed more emphasis on the development of green eco-innovation in recent years, they have been slower in transitioning from traditional energy-dependent development patterns. As a result, the impact of green eco-innovation on industrially underdeveloped regions is even more significant.

4.6. Mechanism Analysis

This study investigated the moderating mechanisms of energy efficiency, energy structure, and environmental regulations on CMS. As seen in

Table 10. Moderating effects of energy efficiency, energy structure, and environmental regulation.

		(1)	(2)	(3)	(4)	(5)
		Primary Supply				Recycling Supply
		Bulk Metals			Minor Metals
Variables	Parameters	lnS	lnS	lnS	lnS	lnS
lnGEI	α1	10.64 ***	2.707 ***	6.260 **	−25.51 ***	9.192 **
		(2.706)	(0.860)	(2.515)	(7.155)	(4.231)
(lnGEI)2	α2	−0.664 ***	−0.180 ***	−0.493 ***	1.477 ***	−0.539 *
		(0.190)	(0.0605)	(0.162)	(0.460)	(0.277)
lnEE	β1		−8.525 ***
			(2.109)
lnEE*lnGEI	β2		1.726 ***
			(0.570)
lnEE*(lnGEI)2	β3		−0.129 ***
			(0.0408)
lnES	β1	−10.27 **			50.52 ***	−15.87 **
		(4.870)			(12.59)	(7.796)
lnES*lnGEI	β2	3.934 ***			−10.54 ***	4.359 *
		(1.483)			(3.631)	(2.278)
lnES*(lnGEI)2	β3	−0.220 **			0.612 **	−0.285 *
		(0.106)			(0.242)	(0.156)
lnER	β1			−6.251
				(4.429)
lnER*lnGEI	β2			2.122 *
				(1.251)
lnER*(lnGEI)2	β3			−0.184 **
				(0.0830)
lnFD	θ1	1.442 ***	1.209 ***	1.395 ***	2.089 ***	−0.254
		(0.325)	(0.339)	(0.333)	(0.662)	(0.526)
lnAS	θ2	−0.694	−1.565 **	−0.808	2.774 *	−2.485 **
		(0.708)	(0.720)	(0.715)	(1.590)	(1.197)
lnFI	θ3	−1.811 **	−1.846 **	−1.223 *	0.528	1.750
		(0.723)	(0.747)	(0.724)	(1.479)	(1.381)
lnUR	θ4	4.412 ***	5.917 ***	4.196 ***	8.914 ***	2.768 **
		(1.535)	(1.567)	(1.599)	(3.302)	(1.237)
lnPC	θ5	−0.445 **	−0.0977	−0.0172	−1.300 ***	−0.988
		(0.188)	(0.181)	(0.190)	(0.393)	(0.720)
Constant	α0	−14.43	3.147	−5.579	144.2 ***	−51.60 **
		(9.425)	(3.886)	(9.739)	(27.16)	(20.14)
Ci		0.271	−0.038	−0.105	−0.044	−0.270
Year effect		Yes	Yes	Yes	Yes	Yes
Province effect		Yes	Yes	Yes	Yes	Yes
Observations		1020	1020	1020	280	840

Standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

the increase in energy efficiency, optimization of energy structure, and strengthening of environmental regulations led to steeper curves in the primary supply. This implies that with the expansion of these moderating variables, the impact of green eco-innovation on primary CMS will strengthen.

Regarding the influence of energy efficiency and energy structure on the supply of primary bulk critical metals, the optimization of energy structure delays the decline phase in the inverted U-shaped curve for the supply of primary bulk critical metals. This suggests that optimizing the energy structure is beneficial for ensuring an adequate supply of primary bulk critical metals. However, it is worth noting that an increase in energy efficiency shifts the inflection point of primary supply towards the left. This may be due to the substitution and crowding-out effects caused by improved energy efficiency, leading to reduced demand for green eco-innovation [45]. This phenomenon is more pronounced for bulk critical metals, which are energy-intensive, and ultimately results in a delayed promotion of green eco-innovation on supply. Simultaneously, the optimization of energy structure advances the ascending phase in the U-shaped curve for the supply of recycling critical metals and the supply of primary minor critical metals, indicating its positive impact on ensuring the supply of recycling critical metals and the supply of primary minor critical metals. H3 is confirmed.

Environmental regulations resulted in a Ci less than 0 for the supply of bulk critical metals, indicating that the strengthening of environmental regulations leads to an earlier decline in primary supply. This suggests that with the reinforcement of environmental regulations, the increase in CMS may be hindered to some extent. H4 is confirmed.

5. Conclusions and Recommendations

Ensuring CMS while achieving sustainable development has long been a significant challenge faced by countries worldwide. Green eco-innovation provides a solution to this challenge. In this context, this study is based on panel data of primary and recycling supplies of five critical metals (copper, aluminum, nickel, antimony, and titanium) in China from 2000 to 2019. It examines and analyzes the impacts of green eco-innovation on different types of CMS and their variations. It also identifies the effects of green eco-innovation on CMS in different regions. Lastly, by introducing moderating variables such as energy efficiency, energy structure, and environmental regulations, it explores the mechanism through which green eco-innovation influences CMS.

The research findings are as follows. Firstly, green eco-innovation has an inverted U-shaped impact on the primary supply of bulk critical metals such as copper, aluminum, and nickel, and they are currently in the declining phase. Secondly, there is a U-shaped impact of green eco-innovation on the recycling supply of copper and aluminum, as well as the primary supply of minor critical metals such as antimony and titanium, but with some differences. The impact of green eco-innovation on the primary supply of minor critical metals is still in the declining phase of the U-shaped curve, while the impact on the recycling supply of copper and aluminum is already in the rising phase of the U-shaped curve. Thirdly, the influence of green eco-innovation on industrially underdeveloped regions is significantly stronger than on industrially developed regions. Lastly, the improvement of energy efficiency, optimization of energy structure, and strengthening of environmental regulations enhance the impact of green eco-innovation on CMS. Where improvements in energy efficiency and optimization of the energy structure are conducive to securing CMS, stronger environmental regulations can be detrimental to securing CMS.

Based on the above conclusions, the following policy recommendations are proposed in this paper. Firstly, it is important to promote a balance between green eco-innovation and CMS. For bulk critical metals’ primary supply, identifying the turning point of green eco-innovation can effectively prevent excessive or blind investments. For minor critical metals’ primary supply, the government should actively support and encourage the development of green eco-innovation. This can be achieved by providing research and development funding, establishing innovation funds, and setting up green technology incubators, among other measures, to collectively promote the research and application of green technologies and sustainable production methods [2].

Second, it is important to promote metal recycling and resource circularity. The government should formulate and implement effective metal recycling policies and resource utilization mechanisms. This can include establishing recycling infrastructure, providing incentives and rewards for recycling, and strengthening research and development of recycling technologies. Financial support and tax incentives can also be provided to encourage businesses to adopt environmentally friendly technologies and equipment. Promoting the adoption of a circular economy model by businesses can reduce reliance on primary supply and increase the recycling and re-use rates of critical metals [16]. Additionally, the government can establish partnerships with relevant industries to jointly promote the development of metal recycling and reduce dependence on primary supply.

Third, it is essential to develop policy measures that are tailored to each specific region. The government should require industrialized regions to vigorously develop green eco-innovation, including establishing green technology parks and incubators to foster local green innovation industries. Support for the research and application of renewable energy should be increased to promote the substitution of clean energy for traditional energy sources. Furthermore, encouraging businesses to adopt renewable energy and clean technologies can help reduce reliance on finite resources [9]. For industrially underdeveloped regions, the government should also continue to encourage the development of green eco-innovation. This will contribute to enhancing the impact of green eco-innovation on CMS in different regions and promote sustainable development.

Last, the government should take measures to promote improved energy efficiency and optimized energy structure. This includes encouraging businesses to adopt energy-efficient technologies and equipment, actively promoting the development and utilization of clean energy, and reducing reliance on traditional energy sources. These actions can safeguard the primary CMS and reduce environmental pressure. Additionally, the strengthening of environmental regulations may have certain negative impacts on CMS. Therefore, when formulating environmental regulations, the government needs to strike a balance between environmental protection and resource supply. It should adopt scientifically and reasonably balanced environmental regulatory strategies to avoid excessive restrictions on industrial development and CMS [49].