1. Introduction
Climate change has been found as a major challenge facing human beings currently, and coping with climate change turns out to be a global consensus [
1,
2,
3]. China has been the largest carbon emitter and plays a vital role in global climate governance. In response to the Paris Conference on Climate Change in 2015, at important meetings (e.g., the General Debate of the United Nations General Assembly and the Climate Ambition Summit in 2020), the Chinese government has proposed for the first time to achieve a carbon peak by 2030, and carbon neutrality by 2060. To be specific, carbon intensity, as a vital indicator, needs to fall to 65% of the 2005 base year by 2030. Carbon intensity reveals the resource utilization efficiency and carbon emission efficiency in economic development while reflecting the production technology efficiency level of a country or region to a certain extent [
4,
5]. Emission-reduction and cleaner production policies based on carbon intensity facilitate the formation of a backward-forced and long-term mechanism to boost China’s economic transformation [
6]. In 2019, China’s carbon intensity was 48% lower than in 2005, and 17% below the 2030 target [
7]. Since China has vigorously adjusted its industrial structure over the past decade, which has led to significantly reduced carbon intensity, there has been limited room for reducing carbon intensity through industrial restructuring by 2030 [
8]. Accordingly, China is facing greater challenges to achieve its carbon intensity target in 2030.
The emission reduction effects of different industries play an important role in achieving the carbon peaking and carbon neutrality goals. The carbon intensity of energy-intensive industries (EIICI) is significantly higher than that of other industrial sectors [
9,
10], which is recognized as the top priority of China’s future emission reduction actions. In accordance with the “National Economic and Social Development Statistical Bulletin of China in 2019”, energy-intensive industries include six sub-sectors including non-metallic minerals, petroleum coking, electric power, chemicals, steel, and non-ferrous metals. Since 2000, China has entered a stage of rapid urbanization development. Activities (e.g., urban development, infrastructure construction, and the transformation of residents’ lifestyles) all require the support of considerable basic products (e.g., cement, steel, petrochemicals, as well as electricity) [
11]. The extent of CO
2 emissions from energy-intensive industries continues to increase. From 2000 to 2015, six energy-intensive industries increased their carbon emissions at an average annual rate of 8.54%. However, energy-intensive industries have much higher emissions than other industries due to their huge demand for primary or secondary energy to produce products. China’s six energy-intensive industries account for more than 50% of the country’s energy-related CO
2 emissions [
12]. The proportion of the total industrial output value of energy-intensive industries in the entire industrial sector has remained at nearly 30% in the long term, whereas the CO
2 emissions accounted for approximately 80% of the entire industrial sector [
9]. The EIICI in 2019 was 2.5 times that of the entire industrial sector. From the perspective of China’s development in the 10 years ahead, it will take a relatively long time for China to complete industrialization and urbanization, which requires energy-intensive industries to continue to provide energy and raw materials. From a foreign perspective, with the advancement of “The Belt and Road” construction [
13], whether it is the supporting ship construction and port construction required by the “Maritime Silk Road”, or the railway plans, airport projects, and highways driven by the “Land Silk Road”, the growing demand in the above international markets will significantly stimulate the expansion of China’s energy-intensive industries. Judging by the development trend at home and abroad, China’s energy-intensive industries will occupy an even larger share in the long term. Meanwhile, if EIICI is maintained at a high level for a long time, the greenhouse gases it produces will directly endanger people’s health and quality of life [
10,
14]. Thus, to achieve the goal of reducing carbon intensity in 2030, China should formulate effective policy measures in energy-intensive industries [
15].
It is a systematic project for China to achieve its carbon intensity target by 2030, and this arduous national task must be broken down into provinces. However, China has a vast territory, and there are large regional differences in natural resource endowments, historical foundations, and technical conditions [
16]. As a result, there are significant spatial differences in carbon intensity and its influencing mechanism and this imbalance process is increasing [
17]. Accordingly, the conclusions obtained only from the research at the overall level of the country are difficult to apply to the development of different regions, which has become a significant issue that has attracted wide attention from the government and scholars. Therefore, focusing on reducing the EIICI, using the panel data of 30 provinces in China from 2000 to 2019, this paper characterizes the temporal and spatial differences of EIICI. On this basis, further considering the spatial spillover effect, several spatial econometric models are established to reveal the driving factors causing the differences. Hopefully, the research results can provide a reference for government departments to formulate and implement differentiated and targeted regional policies.
2. Literature Review
Understanding the spatiotemporal differentiation of carbon emission (intensity) is a solid basis for the scientific formulation of differentiated regional emission reduction policies. Extensive studies have been conducted on drivers (e.g., economic level, technological progress, urbanization, and industrial structure) using various methods (e.g., IDA, SDA, and STIRPAT) from different spatial scales (e.g., global, national, and regional scales), and these studies have achieved fruitful results.
There have been differences in carbon emission (intensity) at different spatial scales, which is a common phenomenon and has still been a hotspot of international concern over the past few years. At the global level, the research objects primarily consist of multiple nations [
18,
19], OECD nations [
20], BRICS nations [
21], developing nations and developed nations [
22], etc. The results show that due to the differences in the development stage, economic development level, energy structure, industrial structure, and other factors of various nations, there is a huge gap in carbon emission (intensity), nations with high-income levels and urbanization rates, and a high proportion of service industries tend to have lower carbon intensity. Since China has been the largest carbon emitter worldwide, its carbon emission changes and emission reduction effects have aroused wide attention. In recent years, there has been an increasing number of studies on China’s carbon intensity from the perspective of spatial differences and spatial effects. Regardless of provincial or regional differences, China’s carbon intensity exhibits significant spatial differentiation characteristics [
23,
24], which are revealed by the characteristics of “high in the west and low in the east”, “high in underdeveloped areas, as well as low in developed areas” [
25]. In addition, from the perspective of spatial effects, the spatial agglomeration characteristics of China’s carbon intensity are significant, and a positive spatial autocorrelation characteristic is identified [
26]. From the perspective of different industries, some scholars have studied the spatiotemporal pattern of the carbon intensity of various industrial sectors in China from the manufacturing industry, power industry, petrochemical industry, and cement industry [
27,
28,
29]. As revealed by the results, there are different degrees of spatial differentiation in various industrial sectors. Besides, several studies were conducted on CO
2 emissions or CO
2 efficiency in six energy-intensive industries. As reported by Yang et al. [
30], from 2007 to 2018, the carbon emissions of China’s energy-intensive industries were primarily distributed in the eastern coastal areas of China, and spatial correlation characteristics were found. Using the provincial data from 2005 to 2017, Zhu et al. [
31] reported that the carbon emission efficiency of China’s energy-intensive industries had significant spatial heterogeneity and spatial agglomeration, with relatively high efficiency in eastern provinces and lower in western regions.
The influencing mechanism behind carbon emission (intensity) spatiotemporal differences is complex and is caused by a variety of drivers. The factors studied by the existing research mainly include energy structure, industrial structure, economic output, population size, energy intensity, rate of urbanization, technological advancement, foreign trade, etc. Most studies show that economic growth will lead to increased carbon emissions [
32,
33]. On the one hand, the increase in residents’ income will prompt residents to live a high-energy-consuming lifestyle, resulting in increased carbon emissions; on the other hand, it will also prompt residents to realize the limitations of resources and the environment and make efforts to adopt energy-saving technologies to reduce carbon emission intensity [
34]. Urbanization has maintained a positive correlation with carbon emissions for a long time and is an essential factor in promoting China’s carbon emissions [
35]. However, there is a certain debate about the direction of the effect of economic level and urbanization rate on carbon intensity. Some scholars believe that both of them have a negative correlation with carbon intensity, while some scholars believe that the two have a non-linear effect in addition to a direct linear effect on carbon intensity [
36]. Technological progress is considered to be one of the most important factors in reducing carbon intensity. Typical domestic and foreign technologies include the reduction of fuel combustion in the process, processing conversion, and end use, as well as technologies such as carbon recovery, low-carbon energy, and carbon sinks [
37,
38,
39]. Scholars generally believe that the industrial structure and energy structure contribute most to China’s carbon intensity effect, and that structural adjustment is still the main direction of China’s future emission reduction [
40]. In addition, energy intensity is the main factor driving the reduction of China’s carbon intensity [
41]. The impact of foreign direct investment on the host country’s carbon intensity is twofold; carbon intensity can either be increased through the transfer of pollution-intensive industries or reduced by technology spillovers that make related production processes cleaner [
42]. Guan et al. [
43], Xu et al. [
44], Ouyang and Lin [
27], and other studies have shown that industrial activity is a key factor in increasing CO
2 emissions. Xie et al. [
45] found that energy-intensive industries such as power production, petroleum processing, coking, chemical products, metal smelting and rolling, and non-metallic mineral products have the most prominent impact on carbon emissions. For the above energy-intensive industries, Lin and Long [
46] argue that energy intensity and energy mix may be beneficial for reducing CO
2 emissions from China’s chemical industry. Yang et al. [
30] found that growth in GDP and population per capita increases CO
2 emissions from energy-intensive industries, while the share of service industries leads to a reduction in CO
2 emissions. Scholars have used various methods to analyze the drivers of carbon intensity under different research backgrounds, which mainly included Index decomposition analysis (IDA) [
47,
48], Structural decomposition analysis (SDA) [
49,
50], and econometric regression analysis based on the IPAT model or STIRPAT model [
51,
52].
The extensive studies have provided a very useful reference for the text, whereas the research on energy-intensive industries still has deficiencies. First, existing studies lack the exploration of the spatial differences in the carbon intensity of energy-intensive industries, so it is difficult to provide help for the formulation of scientific regional decomposition plans in the process of emission reduction in this key industry. Second, spatial elements often have spillover effects and the first law of geography also highlights that the closer the spatial distance between things, the stronger the interdependence will be [
53]. Most of the existing research has recognized regions as separate individuals, and the spatial dependence and spatial heterogeneity of carbon intensity have been rarely discussed from the perspective of geography and spatial interaction. In this way, it is difficult to reveal the spatial correlation and provide a scientific basis for proposing emission reduction measures for joint prevention and control. To fill this research gap, this study first finds the characteristics of spatiotemporal differences in EIICI and then uses the spatial econometric method based on the STIRPAT model to investigate the drivers and spatial spillover effects of EIICI.
4. Spatiotemporal Differences of EIICI
- (1)
From 2000 to 2019, the EIICI showed a significant phased downward trend
The EIICI in 2000 reached up to 8.70 t/104 yuan since China was in the middle stage of industrialization around the year 2000. At that time, China’s economic growth was largely dependent on large-scale consumption of resources and the environment, the level of technical management was relatively low, and the development method was relatively extensive, thus leading to large-scale energy consumption and CO2 emissions. As the economic level and the low-carbon development path have been continuously improved, the EIICI has been reducing due to several factors (e.g., industrial upgrading, technological progress, as well as environmental regulation). According to the rate of descent, there are two distinct stages. The first stage is between 2000 and 2009, during which the EIICI decreased rapidly, with an average annual decrease of 0.54 t/104 yuan. By 2010, the EIICI quickly declined to 3.32 t/104 yuan. The second stage is from 2010 to 2019, with an average annual decrease of 0.12 t/104 yuan. By 2019, its intensity will be adjusted to 2.13 t/104 yuan, nearly 1/4 of that in 2000.
- (2)
At the provincial scale, the EIICI shows a significant downward trend, but there are gradually expanding spatial differences
In general, China can be divided into three major regions (
Table 2). As indicated by the calculation results of the coefficient of variation (CV) from 2000 to 2019, the CV of EIICI has progressively increased from 0.42 to 0.64, which implies that the regional gap tends to expand (
Figure 1). To describe the relative differences in EIICI between regions, this study uses the World Bank’s classification method for regional economic development levels [
63] and adopts 50%, 100%, and 150% of the average EIICI of each province in each year as node values. It is divided into four types: high intensity, medium-high intensity, medium-low intensity, and low intensity. In 2000, there were six high-intensity type areas, including Shaanxi, Guizhou, Anhui, Shanxi, Chongqing, and Inner Mongolia, all of which are located in the central and western regions, and the overall spatial distribution was banded. Besides, there were five medium-high intensity types, including Jilin, Hebei, Heilongjiang, Guangxi, and Ningxia, all of which are located in the central and western regions except for Hebei. The number of low-intensity type areas was relatively small, with only the five areas of Beijing, Tianjin, Shanghai, Guangdong, as well as Xinjiang. Except for Xinjiang, the other four are coastal provinces with developed economies. The other 14 provinces were the low-medium intensity type areas, characterized by a contiguous distribution in space. In 2010, the EIICI of Jiangsu, Zhejiang, Hebei, Anhui, and Chongqing was reduced, while the EIICI of Yunnan, Henan, Ningxia, Xinjiang, and Hainan increased. To be specific, Xinjiang and Hainan shifted the most. Xinjiang has shifted from a low-intensity type to a medium-high intensity type, and Hainan has shifted from a medium-low intensity type to a high-intensity type. In 2019, the spatial distribution pattern of provinces and regions has changed, and the number of four types, including high intensity, medium-high intensity, medium-low intensity, and low-intensity provinces, has been adjusted to 6, 7, 10, and 7, respectively. The high-intensity type areas all belong to the northern provinces, and the southern provinces are mostly two types of low-intensity and low-intensity areas (
Figure 2).
- (3)
At the regional scale, the EIICI shows a pattern of regional differences in the coexistence of “high in the west and low in the east” and “high in the north and low in the south”
For the differences between the three major regions in China, the CV of EIICI in the eastern, central, and western regions from 2000 to 2019 increased from 0.29, 0.25, and 0.39 to 0.46, 0.55, and 0.57, respectively, thus revealing that the regional differences in China’s EIICI have formed a pattern in which the west was larger than the central region and the eastern region, and the differences within the region were also expanding. In 2000, the EIICI of the eastern, central, and western regions reached 7.05 t/104 yuan, 11.99 t/104 yuan, and 13.20 t/104 yuan, respectively. In 2019, the EIICI of the eastern, central, and western regions was regulated to 1.57 t/104 yuan, 2.95 t/104 yuan, and 3.15 t/104 yuan, respectively, and the central and western regions were 1.88 times and 2.01 times higher than the eastern region, respectively. Furthermore, from the perspective of the differences between the north and the south, ten northern provinces of Xinjiang, Gansu, Inner Mongolia, Ningxia, Shaanxi, Shanxi, Hebei, Liaoning, Jilin, and Heilongjiang are taken in this study as a whole to examine the difference in EIICI between the northern provinces and the rest of the southern provinces. In 2000, the average EIICI of northern provinces and other provinces was 15.73 t/104 yuan and 7.70 t/104 yuan, respectively, the former was 2.04 times that of the latter. In 2019, the average EIICI of the two was regulated to 4.24 t/104 yuan and 1.66 t/104 yuan, respectively, and the gap between the former and the latter increased to 2.55 times. Thus, China’s EIICI shows significant “high in the west and low in the east” and “high in the north and low in the south” patterns, while the “difference between the north and the south” is higher than the “difference between the east and the west”.
- (4)
EIICI has significant positive spatial autocorrelation
The global Moran’s I index is adopted to measure the degree of spatial autocorrelation of EIICI from 2000 to 2019, and the random permutation method is employed to construct a normal distribution to examine its significance (
Table 3). As indicated by the results, the global Moran’s I in each year is all positive, and the statistic Z index is all significant at the 1% level (
p values all less than 0.01), thus revealing a significant positive spatial autocorrelation of EIICI in China, i.e., the EIICI of this unit positively impacts the adjacent units. In contrast, the adjacent units will have a positive effect on this unit as well. From the perspective of time evolution, the Moran’s I from 2000 to 2019 showed a fluctuating upward trend, increasing from 0.282 to 0.377, indicating that the agglomeration degree of China’s EIICI has increased, and the types of areas with similar EIICI are more spatially inclined in agglomeration distribution. Thus, spatial effects should be added to improve the accuracy of model estimation in the panel data regression model adopted to explore the drivers below. Furthermore, the local spatial agglomeration characteristics of EIICI are significant. The provinces with high-high agglomeration are largely distributed in Guangxi–Guizhou and northern China, while the provinces exhibiting low-low agglomeration are concentrated in the eastern coastal zone.
7. Implications and Future Study
The above results have high implications for policy. First, the dominant drivers affecting the EIICI should be determined, as well as the key direction of emission reduction policies. They emphasize the role of technological innovation in reducing EIICI and combine independent innovation with the introduction of advanced foreign industrial technology and management experience. In order to adjust the energy consumption structure of high-energy-consuming industries and increase the share of clean energy consumption, the government needs to provide more incentives (such as increasing environmental taxes and increasing clean energy subsidies). Given the characteristics of most high-energy-consuming enterprises with small scale and scattered locations, we should fully exploit the scale effect, optimize the industrial space layout and facilitate the agglomerated development of energy-intensive industries. It is necessary to strengthen environmental control, formulate strict emission standards, and appropriately increase punitive damages for companies exceeding the standards. Second, targeted emission reduction measures should be formulated based on the reality of regional differences. In the eastern coastal areas, independent innovation and the introduction of foreign advanced technologies should be facilitated, the application of innovative technologies in industrial production should be boosted, and ecological industrial clusters and a circular economy should be created. The central and western regions and northern provinces are required to largely optimize the energy consumption structure and progressively reduce the high consumption and high emission development mode dominated by coal. In the meantime, the transfer of energy-intensive industries from the eastern region should be actively and effectively dealt with to avoid becoming a “pollution heaven”. Local governments in the central and western regions should strengthen environmental supervision and rationally facilitate industrial layout under the premise that the carrying capacity of resources and the environment is evaluated. Third, stress should be placed on the spatial spillover effect of EIICI, and regional coordinated development and control strategies should be formulated. In particular, when formulating economic development policies, the governments should consider the transfer of carbon emissions to emphasize the inhibition of local EIICI, as well as to consider the effect on surrounding areas. To reduce the overall carbon intensity of the region, administrative boundaries should be broken, and a cooperative governance model of overall planning, resource sharing, industrial collaboration, and information sharing should be implemented in adjacent regions.
This study still has some limitations, and some issues should be explored in depth. First, it can be analyzed according to smaller spatial scales and basic units. Impacted by data limitations, only provincial data are selected in this study. To improve the precision and accuracy of the research, further research using prefecture-level cities or counties as the basic spatial unit should be conducted. Second, stress should be placed on the research on the sub-sectors of energy-intensive industries. Since there are significant differences in the factors that determine the carbon intensity of various industries, in-depth discussions should be conducted from the sub-sectors of energy-intensive industries in the future, which is conducive to improving the pertinence of energy conservation and emission reduction in different regions. Third, new models and novel methods should be adopted to analyze the influencing mechanism of key drivers and the interaction between multiple drivers. This study primarily aims to reveal the influence direction, relative strength, and spillover effect of different factors. Future research needs to further consider the bilateral causal relationship between the dependent and independent variables to address the endogeneity problem. In the study of long-term series, it is necessary to further analyze the influence of factors such as economic purchases on the results. A geospatial weighted regression model should be used to explore the impact intensity of different factors in each province, in order to reveal the differences in the driving mechanisms between regions. Finally, by conducting a regional comparative study, it is investigated whether the relevant calculation results can be verified in typical case areas.