Spatio-Temporal Evolution and Drivers of Carbon Emission Efficiency in China’s Iron and Steel Industry

Abstract

Improving the carbon emission efficiency (CEE) of the iron and steel industry (ISI) is crucial for China to achieve the goal of carbon peak and carbon neutrality. This study employed the undesirable SBM and Dagum Gini coefficient to measure the ISI’s CEE and analyzed the spatial heterogeneity among three regions of China. This study also used the Tobit model to clarify the influencing factors. The conclusions show that (1) the CEE in eastern provinces is the highest, the central ones rank second, while the western ones rank the worst; the promoting effect of Technical Change is greater than that of Efficiency Change. (2) ISI’s CEE shows a positive spatial correlation and an apparent spatial heterogeneity. The CEE gap between the regions contributes most to the CEE difference among provinces. The regional CEE gap within the western region is the largest, with a maximum difference of 0.520 in the Dagum Gini coefficient. Furthermore, the total CEE gap shows a narrowing trend from 2009 to 2020, with the Dagum Gini coefficient decreasing from 0.414 in 2009 to 0.357 in 2020. (3) Industrial structure, enterprise scale, foreign direct investment, and technology level positively correlate with ISI’s CEE; the marginal impacts are 0.6711, 0.1203, 0.0572, and 3.5191, respectively. While energy intensity, environmental regulation, and product structure negatively correlate with it, the marginal impacts are 0.0178, 1.4673, and 0.2452, respectively.

1. Introduction

The international community has set climate change mitigation targets through multilateral agreements, such as the Paris Agreement. As one of the world’s largest carbon emitters, China’s carbon emissions from its iron and steel industry (ISI) contribute directly to global greenhouse gas emissions and significantly impact the realization of international climate goals. China’s crude steel output reached 1.032 billion tons in 2021, accounting for 53% of the world’s total production [1]. China’s iron and steel industry (ISI) provides an essential basis for industrial modernization and is crucial to economic growth [2]. China is a significant producer of crude steel globally [3]. Owing to the large total ISI output and the energy consumption structure dominated by fossil fuels, the total carbon emissions are tremendous, leading to massive pressure on China’s carbon emissions reduction target and global climate change [4].

Furthermore, a conflict exists between promoting economic growth and conserving energy [5]. Still, ISI has enormous energy-saving and carbon-reduction potential [6]. Therefore, improving carbon emission efficiency (CEE) is a momentous approach to protect energy security, accelerate economic growth, and diminish carbon emissions [7]. Moreover, the Chinese government has determined to increase its renewable energy share to 25% and 80% by 2030 and 2060 [8]; enhancing ISI’s CEE has become a significant challenge to developing a low-carbon economy and producing more steel products.

Previous studies have estimated how to improve CEE. Some emphasized improving energy efficiency [9]. Zhang et al. (2022) explored the emission pathways of ISI to 2050, respectively [10]. Adopting breakthrough technologies is more critical than using energy-saving technologies to decarbonize the ISI in the long run. Furthermore, many influencing factors affect the ISI’s CEE at present, such as technical development [11], energy intensity [12], and environmental regulation [13].

In summary, scholars have increasingly focused on the ISI’s CEE. Dynamic changes and spatial heterogeneity are seldom evaluated in the research on CEE improvement potential. Furthermore, fewer scholars have considered the factors that influence ISI, such as enterprise scale and endowment structure. These may be the research direction for further study on improving CEE. Therefore, this study intends to address the following three objectives: (1) Zhang et al. (2014) found that the development status of ISI and the program design differ in different provinces [14]. This study examined which provinces behave well in ISI’s CEE. (2) Few studies have combined the factors influencing ISI’s CEE with time-space effects [15]. This study aimed to determine what leads to these CEE spatial differences and what contributes to the CEE gap among regions. (3) To improve CEE, the first step is to accurately identify the factors that correlate with it [16]. This study aimed to discover more factors that influence ISI.

To reach the three objectives above, we used the panel data of 29 China’s provinces from 2009 to 2020 and specifically conducted the following study, as shown in Figure 1. (1) We adopted the undesirable SBM model to calculate CEE and decompose it into GML, TC, and EC to analyze CEE changes. (2) We used ArcGIS 10.8 software to explore the CEE spatially and temporally and calculate the Dagum Gini coefficient to measure its spatial heterogeneity. (3) We utilized the Tobit model to clarify the eight factors relevant to ISI’s CEE. (4) We proposed corresponding suggestions based on the conclusions and results.

The possible innovations and contributions are as follows: (1) Compared with the CEE revealed by previous studies at the macro level or the whole industry level, this study focused on the ISI and explored the factors that affect its CEE. The research conclusion can provide a theoretical basis and data support for the government to improve ISI’s input of production factors. (2) This study extended the research content to the geospatial dimension, revealing the spatial differentiation. The research results clearly show CEE’s temporal and spatial evolution characteristics, which provides an empirical basis for proposing the regional coordinated development of ISI. (3) Based on the existing research, this paper expands the spatial heterogeneity analysis and calculates the inter-provincial and intra-provincial CEE gap using the Dagum Gini coefficient. The research results will help Chinese municipal governments identify the root causes of ISI’s inter-provincial carbon emission gap and formulate ISI’s inter-provincial coordinated emission reduction policies.

The remaining sections are listed below: Section 2 is the literature review; Section 3 introduces the methods, models, data, and a description of the data sources; Section 4 and Section 5 conduct the analysis and present the discussions; and Section 6 is the conclusions, policy recommendations, and limitations.

2. Literature Review

2.1. Measurement of CEE

Research on CEE includes single-factor and full-factor CEE. The single-factor CEE requires only one input factor. Per capita carbon dioxide emissions [17] and carbon intensity [18] are two common means.

Lu et al. (2016) proposed a method that considers the energy consumed in producing one unit of a steel product. [19]. Sun et al. (2021) also employed this approach to calculate the ISI’s CEE [20]. Many scholars chose energy consumption per output value as an indicator to reflect the ISI’s energy intensity. Energy intensity is easy to define and measure. However, as a single-factor approach to measure CEE, it ignores other essential input production factors, such as capital and labor.

The total factor CEE research has become more common [21]. Many scholars used DEA and SFA [22] to evaluate the ISI’s CEE. However, Zeng (2019) argued that the SFA model has practical limitations, while the DEA model considers the radial and non-radial relaxation on efficiency [23]. DEA provides more accurate results and can deal with the problem of undesired outputs. Wu and Lin (2022) employed the SBM model and the Malmquist indices to estimate the dynamic ISI’s CEE [24]. Moreover, Feng et al. (2018) used DEA to measure the total factor CEE [25].

2.2. Influencing Factors of CEE

Wang and Wang (2022) concluded that carbon trading policies make a sustained contribution to the ISI’s carbon emission reduction [26]. Arens et al. (2012) found that energy efficiency could impact the ISI’s energy consumption [27]. Hasanbeigi et al. (2014) discovered that the final energy intensity in China was about 50% more than that of the U.S. in ISI [28]. Morfeldt and Silveira (2014) concluded that the Swedish ISI declined energy intensity by increasing natural gas utilization [29]. Rojas-Cardenas et al. (2017) studied the technical structure and discovered that electric arc furnaces helped decrease carbon intensity and increase CEE [30]. Talaei et al. (2020) studied product structure, green supply chain, and waste recycling [31].

In summary, this study made innovations in three aspects. (1) Many studies concentrated on reducing carbon emissions [32]. However, the emission reduction potential of energy efficiency is limited [33]. Furthermore, only some studies focused on minimizing ISI’s redundant input. Using the SBM model, we focused on the promotion of utilization efficiency and carbon emission reduction. (2) While the literature on CEE using time-space analysis gradually matures, the spatial heterogeneity of the ISI requires further research; this study introduces the Dagum Gini coefficient to carry out spatial heterogeneity analysis. (3) Most scholars have analyzed the influence of environmental regulation and energy intensity on CEE. However, studies have not been conducted on iron and steel companies’ foreign direct investment and enterprise scale. Therefore, we studied some innovative factors influencing CEE and found different results. The articles and methods that we mainly refer to are shown in

Table 1. Referential method.

Theme	Research Method	Authors and Year	Literature
Measurement of ISI’s CEE	SBM model	Wu and Lin (2022) [24]	Environmental regulation and its influence on energy environmental performance: evidence on the Porter Hypothesis from China’s iron and steel industry.
		Feng et al. (2018) [25]	Decomposition of energy efficiency and energy-saving potential in China: A three-hierarchy meta-frontier approach.
		Zeng et al. (2019) [23]	Analysis of regional differences and influencing factors on China’s carbon emission efficiency in 2005–2015.
Influencing factors of ISI’s CEE	Panal data analysis	Huang et al. (2021) [34]	Biased technical change and its influencing factors of iron and steel industry: Evidence from provincial panel data in China.
Spatial- temporal analysis	Dagum Gini coefficient	Li (2020) [35]	The Road Map of China’s Steel Industry.

.

3. Model Construction and Variable Definition

3.1. Model Construction

3.1.1. Undesirable SBM Model

The traditional input–output model does not consider slack variables, leading to efficiency evaluation distortions; Tone and Sahoo (2003) held that the SBM model could address this issue by taking slack variables into consideration. Undesirable SBM focuses on undesirable types of outputs, i.e., when assessing the efficiency of decision-making units, undesirable outputs are considered at the same time. Compared with other models, undesirable SBM has a wider scope of application, and the efficiency assessment is more comprehensive and closer to reality, which can provide more accurate decision support. The model is the following [36]:

$$\begin{array}{l}\rho =min\frac{1-\frac{1}{u}{{\displaystyle \sum }}_{i=1}^u\frac{s_i^{-}}{x_{i_0}^{ }}}{1+\frac{1}{t_1+t_2}\left({{\displaystyle \sum }}_{r=1}^{t_1}\frac{s_r^g}{y_{r_0}^g}+{{\displaystyle \sum }}_{r=1}^{t_2}\frac{s_r^b}{y_{r_0}^b}\right)}\\ s.t.\left\{\begin{array}{c}{x}_0=X\lambda +s^{-}\\ y_0^g=Y^g\lambda -s^g\\ y_0^b=Y^b\lambda -s^b\\ {\displaystyle {\displaystyle \sum }_{j=1}^n}{\lambda }_j=1\\ \lambda \ge 0\\ s^{-}\ge 0,s^g\ge 0,s^b\ge 0\end{array}\right.\end{array},$$

where x_i (i = 1, 2, …, u), y^g_r (r = 1, 2, …, t₁), and y^b_r (r = 1, 2, …, t₂) represent the input, desired output, and undesired output of the DMU; λ is the weighting variable; u is the number of inputs; t₁ and t₂ are the quantity of desired and undesired outputs, respectively; s⁻ is the excess input; s^g is the insufficient desired output; s^b is the excess undesired output; ρ is the CEE value, which is strictly decreasing based on s⁻, s^g and s^b; n is the number of DMUs. For 0 ≤ ρ ≤ 1, when ρ = 1 and s⁻ = s^g = s^b =0, the evaluated unit is effective. When ρ is smaller than 1 or s⁻ ∙s^g ∙s^b ≠ 0, the evaluated unit is in a DEA invalid state. Therefore, it is necessary to improve CEE.

3.1.2. GML, EC, and TC Indices

This study decomposes the global Malmquist–Luenberger (GML) index into the Efficiency Change (EC) and Technical Change (TC) indices, providing a more comprehensive understanding of the CEE in a dynamic view. While the ML exponent may lead to a “no solution” outcome in linear programming, the GML index can analyze the change in total factor productivity from a global perspective by constructing the global production possibility set. This set measures the maximum number of products produced with a given resource input [37]. The GML, EC, and TC indices are calculated using the following formulae:

$$GM{L}^{K,K+1}=\frac{CE{E}^g\left(x^{K+1},y^{g,K+1},y^{b,K+1}\right)}{CE{E}^g\left(x^K,y^{g,K},y^{b,K}\right)}=EC\times TC$$

$$EC=\frac{CE{E}^{K+1}\left(x^{K+1},y^{g,K+1},y^{b,K+1}\right)}{CE{E}^K\left(x^K,y^{g,K},y^{b,K}\right)}$$

$$TC=\frac{CE{E}^g\left(x^{K+1},y^{g,K+1},y^{b,K+1}\right)}{CE{E}^{K+1}\left(x^{K+1},y^{g,K+1},y^{b,K+1}\right)}\frac{CE{E}^K\left(x^K,y^{g,K},y^{b,K}\right)}{CE{E}^g\left(x^K,y^{g,K},y^{b,K}\right)}$$

The GML index shows the improvement of CEE, g means global production possibility set, and K is the time period.

3.2. Variable Definition and Data Sources

3.2.1. Carbon Emission Efficiency

Referring to Ding et al. (2019), we established the input–output system to measure CEE, as shown in

Table 2. Input-output factors of ISI.

Input-Output Factors	Meaning	Unit
Input	Total fixed asset	CNY 100 million
	Year-end Employees	10,000 people
	Total energy consumption	10,000 tons of standard coal
Desired output	Total ISI output	CNY 100 million
Undesired output	Carbon emissions	10,000 tons

[38]. This study selected the ISI’s total fixed assets, employees, and energy consumptions as capital, labor, and energy input indicators. This study also selected China’s ISI output and its carbon emissions as the desired and undesired outputs, respectively.

3.2.2. Influencing Factors

We referred to the following studies and selected eight influencing factors on China’s ISI’s CEE, as shown in

Table 3. Influencing factors.

Indicators	Definitions	Unit
Industrial structure (INS)	Proportion of the total output value of the ISI of the province’s total GDP.	%
Endowment structure (ES)	The total fixed assets multiplied by the number of employees in the ISI	10,000 CNY/ISI employee
Enterprise Scale (SC)	The ratio of the total output value to the number of enterprise units in the ISI	10,000 million CNY/ISI enterprise
Foreign direct investment (FDI)	Foreign direct investment per capita in each province	10,000 CNY/person
Energy intensity (EI)	The energy consumption per unit of gross output of the ISI	tons standard coal/CNY
Environmental regulation (ER)	The ratio of the expenditures on the treatment of exhaust gases to the ISI’s GDP	%
Product structure (PS)	Ratio of the crude iron and steel to the iron and steel products	%
Technology level (TEC)	The internal expenditures on R&D by each province as a proportion of its GDP	%

[39].

(1) Industrial structure (INS): Danson (1982) found that the ISI’s factor allocation is related to upstream and downstream industries [40]. The ISI output value accounts for much of China’s industry’s total output value. The industrial structure can reflect a province’s industrialization level, so it is an important influencing factor of the ISI’s CEE. We chose this variable to study whether it has a crucial impact on CEE.

(2) Endowment structure (ES): Schlicht (2016) concluded that the appropriate capital-labor ratio helps construct a reasonable industrial development model and factor input measurement [41]. The capital-labor ratio determines the ES of the ISI. High ES means that a relatively higher share of capital is used to produce iron and steel products in a region. This proportion of input factors will indirectly affect the CEE. We select ES to study its relationship with CEE.

(3) Enterprise scale (SC): Feng (2019) found that diversified, competitive markets increase the flow of production factors [42]. We suppose that, to a certain extent, the expansion of enterprise scale and the improvement of intensive level can improve the utilization efficiency of production factors, including energy, thus reducing carbon emissions per steel output and improving CEE. We selected SC to reflect ISI’s competitive markets.

(4) Foreign direct investment (FDI): Wang and Jia (2019) concluded that FDI influences the domestic capital markets and brings technology spillover effects [43]. Foreign direct investment (FDI) has brought advanced low-carbon technologies and production equipment. These technologies help improve the efficiency of resource and energy utilization. We studied whether this will positively influence CEE.

(5) Energy intensity (EI): EI measures the energy consumption ratio to output. Because carbon emissions are mainly produced during energy consumption, improving EI has a direct impact on CEE. Reducing energy intensity through improving energy efficiency, optimizing energy structure, and developing clean energy will help reduce carbon emissions.

(6) Environmental regulation (ER): Li et al. (2019) believed that proper environmental regulations help stimulate innovation and the energy-saving potential [44]. On the one hand, ER may increase enterprises’ production costs. The rising cost may force enterprises to reduce their investment in technological innovation and energy substitution. On the other hand, ER may also force enterprises to reduce emissions and adopt more environmentally friendly and efficient production methods.

(7) Product structure (PS): Zhang (2021) found that a high PS ratio leads to low CEE as the production of crude steel generates more carbon emissions [45]. The proportion of crude steel can measure the productivity level of the ISI industry. The production of high-quality steel products indicates the industrialization level of the ISI. By contrast, a high quantity of crude steel production may mean that the ISI has difficulty handling the manufacturing technique and finds it hard to increase CEE.

(8) Technology level (TEC): Xu and Lin (2017) believed that iron and steel companies need to acquire advanced technologies [46]. Scientific research investment promotes innovation and drives the ISI’s scientific and technological progress. For example, by developing new energy-saving technologies and processes, energy consumption and waste can be effectively reduced, thus reducing carbon emissions. We chose TEC to reflect the ISI’s technology level.

3.3. Data Sources

The sample of this study was the panel data of 29 provinces in China from 2009 to 2020, excluding Tibet, Hainan, Hong Kong, Macao, and Taiwan, considering principles of comprehensiveness and availability of data. We mainly searched the China Labour Statistical Yearbook, China Industry Economy Statistical Yearbook, China Energy Statistical Yearbook, China Statistical Yearbook and various local statistical yearbooks for data, as shown in

Table 4. Statistics for China’s ISI.

Variable	Maximum	Minimum	Average	STD
CEE	1.000	0.018	0.394	0.293
INS	0.671	0.006	0.120	0.111
ES	416.090	13.105	73.366	43.952
SC	39.349	0.971	6.922	5.876
FDI	7906.430	2.720	490.173	1063.531
EI	17.571	0.045	1.852	2.108
ER	0.178	0.001	0.016	0.020
PS	1.379	0.007	0.788	0.249
TEC	0.064	0.005	0.017	0.011

. For missing values and outliers in the data, we used linear interpolation and the mean value method to correct them.

4. Results and Analysis

4.1. Static and Dynamic Analysis

We used Origin to make the heat map of CEE from 29 provinces, as Figure 2 shows. Beijing (BJ), Tianjin (TJ), Hebei (HE), Shanghai (SH), Jiangsu (JS), and Zhejiang (ZJ) have the highest CEE.

The steel enterprises in eastern provinces have solid industrial bases and sound economic development levels. Moreover, local governments should take adequate measures to improve CEE. For instance, the JS government started enforcing a limitation on the amount of low-quality steel produced by iron and steel enterprises in 2016. After that, the production and sale of low-quality steel made from scrap metal throughout the province further diminished, and an ineffective and backward iron and steel production capacity was gradually eliminated. Consequently, JS’s CEE progressed rapidly during this period.

He also performs well in the ISI’s CEE. This is because it uses cleaner energy sources to partially replace traditional coal combustion and enacts a stricter carbon emission policy. Advanced technology and strict regulation contribute to carbon reduction.

SH has high CEE, indicating that it handles input and output well. Regarding raw materials, it imports iron ore mainly from Australia by taking advantage of its ports. It significantly reduces carbon emissions during transportation; in terms of the labor force, SH implements various welfare policies for talent. It attracts many high-quality human resources. Therefore, it can maintain a high-quality capital and labor input.

Furthermore, its energy efficiency is also very high because it invests heavily in science and technology to conduct low-carbon research. Moreover, SH also strengthens cooperation with scientific research institutions and universities, which explains its strong scientific and technological innovation capabilities. The reason why ZJ, TJ, and BJ have high CEE is similar to those of JS, HE, and SH, respectively.

We also conclude from Figure 2 that NM and SD had high CEE in 2010 and 2012, respectively. For NM, this is probably because it absorbed a lot of steel production equipment or facilities transferred from BJ and TJ. At the same time, SD introduced many foreign capitals to the ISI from overseas because it has many ports. Western provinces such as Qinghai (QH) and Ningxia (NX) have the lowest CEE. They are generally backward in ISI development, and neither these western provinces nor their neighboring provinces can effectively form a virtuous circle of CEE promotion. Although many western iron and steel companies can access sufficient natural resources, they need more meticulous carbon emission management, resulting in a low energy utilization rate. Moreover, insufficient coke combustion during steel production may lead to waste energy consumption and low CEE. They may also result in carbon black particles, smog, and the “greenhouse effect” that causes global temperature to grow [47].

This study uses ArcMap 10.8 to describe the evolution of CEE, as shown in Figure 3. Henan (HA) and Hubei (HB) perform well in CEE. HA’s and HB’s steel enterprises have paid great attention to product innovation and recently invested in extensive research and development expenses. Therefore, they outperform other central provinces in employing low-carbon techniques and higher-efficiency production methods. Therefore, their CEE is the highest in the central region.

In contrast, Heilongjiang (HL) and Jilin (JL) are restricted by the slow transformation of industrial structures and low production capacity. They invest less in technology innovation because they need more funds. Moreover, their CEE performance could be better, with severe talent loss and too much-outdated capital stock in the ISI.

Eastern ISI gradually transformed its steel facilities in central and western provinces after 2009, resulting in a shift in carbon emissions footprint. Eastern ISI enhanced CEE by adjusting its allocation of production factors and reducing carbon emissions. However, as the demand for steel products still exists, total carbon emissions have remained unchanged but only shifted. The growth of carbon emissions in the central and western provinces was higher. Therefore, their CEE improvement is slower than that of the eastern provinces.

Figure 3. Evolution of CEE.

Figure 3. Evolution of CEE.

We used Origin to make the radar map and figure out how each province’s CEE changed from 2009–2020, as shown in Figure 4. The GML indices of Gansu (GS) and Inner Mongolia (NM) are the highest. This is because Baotou Company made a massive investment in the green production of NM’s ISI. Since 2009, it has invested a lot in upgrading steel equipment to improve the technology level. Meanwhile, Baotou Company strengthened its ties with the local railway transportation industry. It helps reduce carbon emissions generated during transportation. Furthermore, Baotou Company actively promotes the ISI’s energy efficiency and labor productivity. Therefore, NM also has a high EC index.

Figure 4. Radar map of CEE’s dynamic changes. Note: GML, EC and TC represent Global-Malmquist-Luenberger, Efficiency change and Technical Change indices, respectively.

Figure 4. Radar map of CEE’s dynamic changes. Note: GML, EC and TC represent Global-Malmquist-Luenberger, Efficiency change and Technical Change indices, respectively.

We also conclude from Figure 4 that TC is the main driving force of CEE improvement. The effect of EC is smaller than that of TC because the increase of TC in most provinces is greater than that of the EC, especially for western provinces such as QH, NX, and XJ. This demonstrates that the western ISI makes more efforts in technological progress than that of the other regions, thus greatly enhancing TC.

For example, the Qinghai (QH) government makes great efforts to encourage the introduction of more advanced steel production equipment from other provinces. Therefore, its TC index increases greatly. However, its EC is below 1, meaning it does not put much emphasis on improving the efficiency of ISI production. The CEE improves mainly by enlarging output but not by adjusting the proportion of factor input. The situations for Guizhou, Heilongjiang, and Xinjiang are similar.

ISI in western provinces can catch up with the East in CEE through quick TC improvement only if the rising trend is maintained, but at present, they should emphasize improving EC. Gansu (GS) set an excellent example with a relatively high EC, probably because its local government pays much attention to enhancing the resource utilization efficiency of ISI factors. For example, the government urges native iron and steel enterprises to apply renewable energy more. These enterprises further improved EC by strengthening the utilization efficiency of renewable energy, such as solar and wind energy.

4.2. Spatial Heterogeneity Analysis

This study analyzed the spatial autocorrelation of the CEE changes by introducing Moran’s I. Moran (1950) described the global Moran index as the measure of the autocorrelation of a given variable in space [48]. The Moran scatter plot has four quadrants, with the position of each province depending on the CEE of the ISI as well as the neighboring provinces. Moran’s I is mainly calculated by utilizing the spatial adjacent matrix. The local Moran index requires the division of different provinces into quadrants by combining the Moran scatter plot method to finally determine the spatial distribution characteristics of CEE in each province [49]. The calculation formula for Moran’s I is as follows:

$$Moran’s I=\frac{N}{{{\displaystyle \sum }}_{i=1}^N{{\displaystyle \sum }}_{j=1}^N{W}_{ij}}\frac{{{\displaystyle \sum }}_{i=1}^N{{\displaystyle \sum }}_{j=1}^N{W}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{{{\displaystyle \sum }}_{i=1}^N{(x_i-x_j)}^2},$$

where x_i and x_j represent the CEE of the province i and province j; N is the quantity of spatial distances formed by i and j; and W is the spatial adjacent matrix.

We used Stata 15.0 to calculate the global Moran index and Moran scatter plot, as shown in

Table 5. Global Moran index and significance test.

	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
I	0.096	0.132	0.173	0.117	0.139	0.148	0.077	0.113	0.142	0.148	0.133	0.153
p	0.002	0.001	0.001	0.001	0.001	0.001	0.018	0.001	0.001	0.001	0.001	0.001

and Figure 5. A positive spatial autocorrelation in CEE exists because each year passes the significance test, except for 2015, indicating that the ISI’s CEE exhibits a spatial autocorrelation phenomenon. China’s government made many adjustments to environmental policies before and after 2015, so the frequent policy changes might have led to the spatial autocorrelation coefficient not being significant.

Figure 5. Moran scatter plot in four years.

Figure 5. Moran scatter plot in four years.

Based on the positive spatial autocorrelation in the ISI’s CEE, we studied the spatial distribution characteristics of CEE from Figure 5: (1) High-CEE provinces, such as Beijing, Shanghai, and Tianjin, gather in the eastern region. As they have better transportation, policy guidance and sufficient financial support, it is natural for the eastern region to have high CEE in the ISI. (2) Low-CEE provinces, such as Qinghai, Gansu, and Xinjiang, gather in the western region. This is because the western ISI pays little attention to decreasing carbon emission reductions or optimizing the input ratio of production factors; more emphasis should be put on efficient energy utilization and improving energy efficiency. (3) Central provinces mainly gather in the second and fourth quadrants. The central ISI is at the stage of energy transformation. Its renewables account for over 25% of total energy consumption [50]. The proportion is between high-CEE and low-CEE provinces. Central provinces may move to the first quadrant with continuous investment in developing clean energy.

We used ArcMap 10.8 to reveal the time–space transition of CEE, as shown in Figure 6. Referring to the spatio-temporal leap analysis method of Rey and Janikus (2006), we divided the provinces into two types. Type 1 means that a time–space transition occurred [51]. Most provinces belong to Type 2, meaning no time–space transition happened. China’s ISI’s CEE at the provincial level has changed significantly, especially in the central region of China. Many time–space transitions took place in 2009–2012 and 2012–2016. Since 2016, the ISI’s capital introduction speed has slowed down, and the proficiency of steelworkers has gradually stabilized. Therefore, the number of provinces that made the transition in 2016–2020 is smaller than that in 2009–2020.

This study used the Dagum Gini coefficient to measure and analyze the spatial heterogeneity for China’s ISI. The Dagum Gini coefficient can be elastically adjusted according to the actual situation during calculation, making it more flexible and suitable for different datasets and research subjects. In addition, the Dagum Gini coefficient has a certain degree of robustness, with minimal impact from outliers. This makes it more reliable in handling accurate data and less susceptible to interference from extreme values. Dagum (1980) first used the Gini coefficient to identify the sources of regional disparities [52]. Li also used this method to study the carbon footprint of China’s steel industry [35]. The Dagum Gini coefficient is decomposed as follows:

$$G=G_w+G_b+G_t$$

G_w and G_b represent the CEE gap within and between the regions, respectively. G_t considers the factors both within and between regions and solves the problem that other two indices ignore the overlapping parts of CEE gap among different regions.

We calculated the Dagum Gini coefficient to find the sources of the CEE gap, as shown in Figure 7. The gap between regions is the main reason for the CEE gap. Specifically, the gap between the East and the West is the largest, followed by the Middle and West, and the gap between the East and the Middle is relatively tiny. This is mainly due to the difference in industrialization levels between the three regions. Furthermore, their energy-saving and emission-reduction capabilities are very different.

The contribution of the CEE gap between regions (yellow bar) remains high during 2009–2011 and 2016–2020. In 2009–2011, China government issued the following measures to revitalize its ISI: (1) Coordinate the domestic and international markets. Implement measures to expand domestic demand and stimulate domestic steel consumption. Implement a moderately flexible export tax policy to stabilize the international market share. (2) Strictly control the total amount of steel, eliminate backward production capacity, and inhibit steel projects that expand production capacity. (3) Encourage the leading role of large groups to promote the joint reorganization of enterprises, increase the scale of iron and steel enterprises, and improve their concentration. (4) Intensify technological transformation, increase R&D funds, and promote the technological progress of the ISI. (5) Adjust product structure and improve steel quality. All the strategies above have led to the widening CEE gap between the regions.

Furthermore, in 2016–2020, the Chinese government vigorously implemented a strict plan to reduce excess capacity. It enforced the eastern ISI to improve the utilization efficiency of input resources, especially energies. Therefore, the CEE gap between the eastern and the western widens. It should be noted that the contributions of the CEE gap between regions were low in 2012, 2015, and 2016. In 2012, many central and western provinces introduced advanced steel technologies from the eastern region. This enabled their CEE to improve. For 2015 and 2016, iron and steel enterprises in eastern provinces manufactured too much steel production and needed more resource utilization efficiency. Therefore, the leading edge of CEE for the eastern provinces has declined again. In addition, before 2017, the Chinese government determined to enforce ISI to diminish redundant energy inputs and excess carbon emissions. Therefore, the CEE gap between regions fell in 2012, 2015, and 2016. Moreover, the CEE gap between eastern and western, central and western, and eastern and central ISI shows a “widening first before narrowing” trend, indicating that the ISI in 2020 was more balanced than in 2009.

We also conclude from Figure 7 that the regional CEE gap differs significantly among the three regions. The CEE gap is highest within western ISI (purple line), while it is lowest within eastern ISI (blue line). This is mainly because the ISI development of the three regions is unbalanced and inadequate. Eastern ISI leads in resource endowment, economic development level, and advanced technology. These high-quality resources are well-distributed among eastern provinces, so the CEE gap is the smallest. The western ISI is relatively less developed and mainly adopts the traditional heavy industry development mode. It has slow adjustment of industrial structure, low utilization rate of resources, and low investment in high and new technology.

Furthermore, some western provinces undertook steel facilities transferred from the east. This increased carbon emissions, leading to a more significant CEE gap with the western ISI. Therefore, the gap within it is the largest.

4.3. Static and Dynamic Analysis

Tobin (1955) described the Tobit model to effectively process panel data [53]. The Tobit model can handle situations where the dependent variable is truncated or censored due to experimental design or other reasons. In other models, handling truncated data may lead to loss of information or bias, while the Tobit model can more fully utilize the observable part of the data to provide a more accurate estimation of causal relationships. Additionally, the Tobit model exhibits a certain degree of flexibility in dealing with different types of truncated data, allowing for adjustments and extensions to accommodate various causal relationship testing needs. By referring to previous research, this study used the Tobit model to analyze eight factors relevant to the ISI’s CEE, in which some of the indicators were smoothed using the natural logarithm to make the variance fluctuations relatively stable:

$$\begin{array}{c}CEE=b_0+b_{1}INS+b_{2}lnES+b_{3}lnSC+b_4\ln FDI+b_{5}EI\\ +b_{6}ER+b_{7}PS+b_{8}TEC+e_{it}\end{array}$$

b and e_it are the coefficient value and random error, respectively.

We used Stata 15.0 to conduct the Tobit model on the eight influencing factors, as shown in

Table 6. Tobit model regression results.

Items	Coefficient of Regression
Intercept	0.0707
	(0.76)
INS	0.6711 ***
	(7.46)
lnES	−0.0169
	(−0.79)
lnSC	0.1203 ***
	(5.76)
lnFDI	0.0572 ***
	(5.90)
lnEI	−0.0178 **
	(−2.10)
ER	−1.4673 *
	(−1.87)
PS	−0.2452 ***
	(−4.10)
TEC	3.5191 **
	(2.05)

Note: * p < 0.10, ** p < 0.05, *** p < 0.01. The z value is in parentheses.

. All regression coefficients in the model pass the significance test, except for lnES.

Then, the analysis of the influencing factors is as follows.

(1) Industrial structure. INS passed the significance test at a 1% significance level. Industrialization increases the demand for steel products, accelerating the ISI’s expansion [54]. What is more, the higher ISI’s output-to-GDP ratio indicates a high level of industrialization. The ISI in eastern provinces adopts more efficient steel production methods and does well in minimizing input while reducing carbon emissions. With the development of industrialization, the ISI’s fixed assets stock gradually stabilized. Enterprises mainly operate on the original infrastructure for large-scale production. Therefore, capital redundancy decreases, and CEE improves. Upgrading industrial structures can strengthen regional cooperation and help build a low-carbon industrial system. During 2016–2020, Chinese local governments took measures to strengthen cooperation with other provinces. They wished to jointly promote the ISI development and application of advanced technologies and form low-carbon industrial chains. The ISI in the central and western provinces also absorbed more of the eastern region’s steel capital and production facilities during that period. The above all contributed to realizing resource sharing and complementary advantages and improving the overall CEE of China.

(2) Endowment structure. The regression coefficient is less negative and does not pass the significance test. The differences among individual provinces cause it. For example, BJ and XJ have similar ES while distinct carbon reduction potential and CEE performance. Moreover, although ES is not correlated with CEE in statistics, the regression results still have reference values. The increase in the ES indicates the process of capital deepening. It is not necessarily a good thing, especially regarding CEE improvement. On the one hand, high ES indicates that the capital inputs, such as machinery and equipment, may be too many. It may lead to capital redundancy. With the deepening of industrialization in various provinces of China, the demand for the number of steelworkers decreases. Moreover, with the continuous upgrading of automatic machinery and equipment, the labor force required to produce the same steel output is reduced. In addition, in 2009 and 2016, the China Municipal Government vigorously promoted the policy of de-capacity. This led to the demand for ISI workers further decreasing, accelerating the rise of ES. On the other hand, the increase in ES may bring some drawbacks to CEE. According to factor endowment theory, if the value of the industrial endowment structure is closer to the capital-labor ratio of the economy, then the ISI can obtain higher profits because it is more in line with the comparative advantage determined by the endowment structure. The rapid increase in ES indicates that enterprises may deviate from the appropriate capital-labor ratio of the ISI. More labor input is needed to expand steel production. Therefore, the ISI needs to optimize the allocation of input factors to improve output and CEE.

(3) Enterprise scale. lnSC has a positive regression coefficient, indicating that high SC benefits CEE. Large iron and steel enterprises usually have a high level of specialization. They are more efficient in the integration and utilization of resources. When an enterprise’s human, material, and financial resources are concentrated, its management capacity and capital scale are also enhanced. This can lead to high efficiency of the whole production process. This high efficiency helps increase the steel output and reduces carbon dioxide emissions. In addition, to cut costs, larger companies are more conducive to optimizing the allocation of input factors and reducing energy wastage. Furthermore, larger iron and steel companies focus more on diminishing carbon emissions and shouldering corporate social responsibility. Increasing enterprise scales could improve their technological innovation ability to a certain extent. Therefore, in terms of energy conservation and emission reduction, the marginal cost of large enterprises is relatively low. Therefore, SC positively correlates with the CEE.

(4) Foreign direct investment. lnFDI positively correlates with CEE. In 2009, the Chinese government implemented a policy to promote the development of the ISI. The steel output proliferated by encouraging foreign companies to invest in its ISI. There, China’s total CEE dramatically rose in 2009–2012. In 2016, with the help of FDI, China’s iron and steel enterprises introduced cleaner technologies. These advanced technologies help further reduce carbon emissions. China’s ISI significantly promotes technological upgrading by introducing advanced low-carbon technologies and cleaner production equipment. With the global economy’s decline and the ISI’s profit rate, steel enterprises paid more attention to cost savings. Driven by their profit-seeking nature, foreign steel enterprises transferred steel factories to China because the labor price in China’s ISI was relatively low. Due to the introduction of external capital, CEE improvement in 2009–2012 showed a sharp rise compared with 2009–2010. Moreover, significant foreign direct investment indicates a high degree of opening up level, which benefits the introduction of advanced steel production equipment or the transfer of redundant capital investment.

(5) Energy intensity. EI affects CEE negatively. The increase in output may accompany the increase in total energy consumption. However, carbon dioxide emissions are also increasing. Iron and steel enterprises pursue minimum input costs by using cheap but highly polluting fossil energies, releasing much carbon emissions. Moreover, excessive energy consumption indicates the redundancy of fuel input and low energy efficiency. In addition, heat loss and insufficient combustion of fossil energy all lead to low CEE. China’s de-capacity actions in 2016 aimed to minimize the ISI’s carbon emissions and reduce the use of highly polluting fossil fuels. The series of actions helps decrease the energy intensity. China kept pushing the ISI to reduce significantly excessive energy consumption and adjust the industrial structure during 2009–2020, thus lowering energy intensity and improving CEE. Therefore, optimizing fossil fuel allocation and increasing the proportion of cleaner resources is critical for carbon emission goals.

(6) Environmental regulation. ER negatively correlates with CEE. ER may dampen the improvement of CEE. It is even more complex when China’s ISI has shallow profit margins and a poor market outlook. Strict environmental regulations may restrain enterprises from carrying out scientific research or new product innovation. Iron and steel companies remain discouraged and need more incentives to improve their CEE. However, ER is still essential because if government policies are too loose, it will reduce the supervision of enterprises. Moreover, that will hurt the overall situation. Moreover, the production halt due to the pandemic in 2019 and 2020 made it even harder for enterprises to maintain sufficient funds for their operations. In this case, the government’s strict policies to regulate carbon emissions undoubtedly increase the cost burden of their operations. The government needs to take more effective measures to strengthen environmental management and improve CEE further. Although this punishment can help to reduce carbon emissions in the short term, it may limit the development of steel enterprises in the long term. Instead of unthinkingly charging excessive environmental pollution fees for the ISI, it can provide economic incentives. In 2013, China further proposed stricter regulations and inhibited excessive dependence on traditional energy sources, and CEE declined a little during 2012–2016. CEE gradually improved when China encouraged enterprises to reduce excessive capacity during 2016–2020 by providing subsidies and policy support. The iron and steel companies in the eastern provinces have more incentives and sufficient funds to reduce carbon emissions. Under these circumstances, the local government can provide preferential tax policies and subsidies to deal with environmental problems.

(7) Product structure. PS negatively affects CEE. In processing crude steel into iron and steel products, increasing energy consumption and adjusting the proportion of various metal elements is necessary. In most cases, crude steel production will release more carbon emissions. Furthermore, a high crude steel ratio indicates that the technology level of ISI is relatively low because neither the production conditions nor the requirements for equipment for crude steel production are demanding. As a result, crude steel production may lead to more waste of resource input, lower steel output, and thus lower CEE. The ISI needs to adopt more environmentally friendly and energy-saving production technologies to produce more high-quality steel products and lower PS. It must also strengthen industry management and policy guidance to promote green transformation and sustainable development. The eastern ISI performs better in these areas; it has a low PS and higher CEE.

(8) Technology level. TEC positively correlates with CEE. A high technology level will bring many technological advancements. Therefore, TEC has a positive effect on CEE improvement. By investing in scientific research, iron and steel enterprises perform better at refining and optimizing production processes, using advanced methods such as carbon capture and storage and inventing more energy-efficient equipment. At the same time, scientific research investment can support the development of waste gas treatment devices and recovery technologies and reduce carbon emissions in waste treatment and discharge. In addition, the high level of internal scientific research expenditures indicates that a province’s ISI is operating well and has enough funds for scientific research investment and technological innovation. The increase in scientific research investment can promote the usage of cleaner energy and can reduce the demand for fossil energy. The innovation in the ISI also accelerates the transformation of the energy consumption structure and improves energy utilization efficiency. In 2009 and 2016, China encouraged investment in scientific research in the ISI. Therefore, the TC of each province significantly improved. This proves that the level of science and technology is positively related to CEE.

5. Discussion

The Chinese government proposed the carbon peak and carbon neutrality target in 2020. This study analyzed the changes and influencing factors of the ISI’s CEE from multiple dimensions. We aimed to address the contradiction between promoting the development of ISI and reducing carbon emissions. Based on the previous studies, our findings are as follows.

First, our results are consistent with Lin and Wu’s (2020), which identified that EC and TC are driving factors for efficiency performance improvement using the research method of meta-frontier non-radial Malmquist directional distance function [55]. Furthermore, we also found that the current GML is mainly driven by TC, which is supported by Chen et al. (2022) [56]. We conclude that the CEE improvement in the western provinces is mainly due to the TC increase, which aligns with Niu et al.’s (2022) results of the stochastic frontier model analysis [57]. The TC and CEE indices drastically increase by introducing more advanced technology, expanding enterprise scale, and increasing ISI endowment structure.

Next, we focused on the CEE gap and the reason for it in the second part. The most advanced production equipment and resources are concentrated in the top CEE provinces [58]. Moreover, these high-CEE provinces gather in the East. Our results show that the eastern and central ISIs have almost the same performance in terms of the EC and TC indices (Figure 3). However, a huge CEE gap remains (Figure 1), indicating that the central ISI has reached a CEE bottleneck, especially for Heilongjiang and Jilin. Although Heilongjiang and Jinlin had abundant capital accumulation in the ISI from 1950 to 1960, they did behave poorly in terms of CEE from 2009 to 2020. We conclude that the CEE gap is due to the different levels of economic development, distances, resource endowment, and various economic development models of ISI. Furthermore, we conclude that the CEE gap is significant because many low-CEE provinces are fossil fuel-exporting provinces while the eastern ones are fuel-importing provinces [59].

Finally, many scholars have concentrated on the specific manufacturing techniques used in the ISI. We classified them based on their technology level and proved their importance from a new perspective. For TEC, the fund for ISI research can support the optimization research of the transportation process. The carbon emissions caused by transporting steel materials between provinces are inevitable. Researchers, therefore, can design a more environmentally friendly transportation process. For example, by designing more energy-efficient ways of purchasing and transporting raw materials, carbon emissions in logistics can be reduced. Therefore, TEC is essential for the CEE improvement of China’s ISI.

Huang (2021) concluded that endowment structure (ES) may curb further improvement in the ISI’s CEE [34]. Moreover, Benjamin’s supposed industrial structure (INS) [60] negatively impacts the metallurgy industry’s CEE. Our research findings support the former and propose a different one that is different from the latter. The reason for the inconsistency in conclusions may be that Benjamin’s study was conducted in the metallurgical industry, which is broader in scope than the ISI. It includes heavy, light, and thin metallurgy in addition to ISI. Furthermore, Benjamin’s study data cover the period 1991–2014, which is quite different from today’s industrial structure development, which may also affect the consistency of the results.

(1) For ES, this study finds that the ISI’s endowment structure, i.e., the capital-labor ratio, may hurt CEE improvement. The increase in ES indicates the excess ISI capital investment. It may cause redundant input and more carbon emissions, thus decreasing CEE. From an economic perspective, high ES leads to the diminishing marginal output of capital and even the distortion of input factors in allocation. Therefore, exorbitant ES is averse to improving CEE. For example, if a province’s ES is high, its ISI production may be more dependent on a large amount of capital input, and at the early stage, its CEE may be improved with a significant increase in output. However, at the later stage, with the decreasing marginal return of capital, excessive capital input may become redundant, and CEE may decline.

(2) For INS, it indicates the industrialization level of a province. Therefore, higher INS means a province has a higher carbon emission reduction capacity. Moreover, for higher INS provinces, the input can be further minimized through a more rational allocation of the input of production factors. Upgrading industrial structures can promote the transformation of ISI’s energy structure from high-carbon to clean energy. With renewable energy development, carbon emissions in steel production have been effectively controlled. The transformation of energy structure caused by this industrial structure optimization reduces dependence on fossil fuels, improves energy utilization efficiency, and reduces carbon emissions.

Furthermore, continuously optimizing the ISI industrial structure can also improve energy efficiency. Traditional ISI’s production process is often accompanied by input redundancy, output insufficiency, and energy waste, while industrial upgrading can improve energy efficiency through technological innovation and improvement. For example, the application of artificial intelligence can realize fine control of the production process and reduce energy waste. Technological progress will help to further reduce carbon emissions per unit output in the long run, thus improving ISI’s CEE and benefiting the sustainability of ISI.

6. Conclusions and Recommendations

6.1. Conclusions

This study combined the undesirable SBM model and Dagum Gini coefficient to analyze China’s ISI’s CEE in a dynamic view and verify its spatial heterogeneity. This study also employed the Tobit model to clarify its influencing factors. We reached the following conclusions:

(1) CEE varies significantly among provinces. The eastern ISI has the highest CEE, the central ranks second, while the western performs the worst. The western ISI has the highest GML and TC. Moreover, the effect of TC on GML is greater than that of the EC, indicating that Technical Change is the main contributing factor to the promotion of CEE for China’s ISI.

(2) The ISI’s CEE shows an apparent spatial heterogeneity. High-CEE provinces gather in the eastern region, while low-CEE gather in the western. The regional CEE gap within the western region is the largest. Furthermore, the main contributor to the spatial heterogeneity is the CEE gap between regions, which shows a narrowing trend.

(3) Industrial structure, enterprise scale, foreign direct investment, and technology level all positively correlate with CEE, while endowment structure, energy intensity, product structure, and environmental regulation have adverse effects on it.

6.2. Policy Recommendations

Improving the ISI’s CEE helps reduce carbon emissions and promote the ISI’s output in China; this study proposes the following feasible policy recommendations:

(1) Iron and steel enterprises in more developed areas should give full play to their advantages in industrial structure and lead the industrial structure adjustment in neighboring provinces with lower CEE levels. The eastern ISI can transfer the iron and steel facilities and equipment to the central and western regions to promote their industrial structure adjustment. In this way, the eastern ISI can decrease its endowment structure. In a word, provincial governments should promote the sharing, integration, and coordination of resources in the iron and steel industry in adjacent areas, making allocating resources more effective. Moreover, to improve CEE, steel enterprises need to increase funds for scientific research and promote their scientific and technological development. Local governments can achieve this by providing subsidies for steel production enterprises that increase investment in green innovation, develop and apply environmental protection technologies, and reduce carbon emissions. The green innovative technologies include clean energy, energy saving and emission reduction, carbon capture and storage technology, etc. The usage of these technologies can help reduce carbon emissions and improve CEE.

(2) Iron and steel enterprises should expand their scale to obtain higher economies of scale. The expansion of enterprise scale is conducive to developing the industrial chain and increasing the added value of products. By perfecting industrial clusters, enterprises become more competitive in the market. Enterprises with more substantial competitiveness are more willing to pay attention to environmental protection and sustainable development. Therefore, they adopt environmental protection technologies and production methods to improve CEE. Moreover, the promotion of product-added value is often accompanied by technological innovation and industrial upgrading. Large-scale and well-run iron and steel enterprises in a province should actively focus on cross-province mergers and acquisitions and adjust their scale to reorganize poorly managed enterprises with eliminated production capacity. This way, high-quality production capital and human resources are gradually transferred to high-CEE iron and steel enterprises. Moreover, steel enterprises should increase cooperation with foreign investors. Introducing cleaner production technologies can further improve energy efficiency and CEE.

(3) In reducing energy intensity, iron and steel enterprises should strive to increase the proportion of cleaner energy in total energy consumption and reduce excessive dependence on traditional energy. Furthermore, optimizing the product structure is also the key to improving the ISI’s CEE. Local government should encourage enterprises to produce as little crude steel and more steel products as possible. Furthermore, the government must formulate more effective environmental rules and policies to regulate the iron and steel enterprises. For example, the ISI in the eastern provinces has more incentives to reduce carbon emissions and participate in trading carbon emission rights. The government can guide steel enterprises to participate by perfecting the establishment of a carbon emission trading market and formulating relevant laws and regulations. The government can also encourage CEE improvement by subsidizing iron and steel enterprises that actively save energy and reduce emissions. In addition, the environmental regulations should be adapted to local conditions according to the development of the ISI in different regions.

6.3. Limitations and Prospects

This paper has the following limitations and can be expanded. First, there are missing values and errors in the data, which may affect the results of the empirical tests even though they are corrected using linear interpolation and the mean value method. Second, due to the length limitation of the paper, we have yet to extend our study of CEE influencing factors to the spatial dimension. After verifying the spatial heterogeneity of the ISI’s CEE, we suppose that the industrial structure of ISI may also influence the CEE of neighboring provinces. Future researchers can focus on the spatial spillover effect of the influencing factors. Third, a contradiction exists between the increase in asset investment and the phasing out of excess production capacity in the ISI [35]. However, capital investment in the ISI does not align with China’s carbon emission reduction goals. Therefore, the key to further improving CEE lies in phasing out phasing out excess ISI production capacity. Increasing the actual production capacity to improve CEE may be further studied.