Correlation Effects, Driving Forces and Evolutionary Paths of Cross-Industry Transfer of Energy Consumption in China: A New Analytical Framework

Abstract

This paper constructs a modified hypothesis extraction method (MHEM)–structural decomposition analysis (SDA)–structural path decomposition (SPD) analytical framework and employs the 2018–2022 Chinese input–output tables to discuss sectoral consumption correlations, driving forces of consumption, and the transmission paths of carbon energy (CE), oil and gas energy (OGE) and electric energy (EE). The results of the study indicate that energy-exporting sectors are primarily energy production or conversion industries, while energy-importing sectors are mainly in the construction sector. China’s energy consumption has shown consistent year-on-year growth, with the primary driving force being the intensity of energy consumption and the secondary factor being per capita demand. The consumption of all three types of energy is primarily directed toward domestic consumption and capital formation. Regarding energy consumption transmission paths, the first-order path with the largest overall impact on CE is “electricity, gas, and water supply sector → domestic consumption”, while higher-order paths are primarily subpaths of “electricity, gas, and water supply sector → capital formation”. For OGE, the main supply and transfer path is “coke, refined petroleum, and nuclear fuel sector → domestic consumption”, along with its subpaths. In contrast, EE transmission is more balanced, with a high demand for electricity across all sectors.

1. Introduction

Global energy consumption has increased due to population growth, industrialization, and urban expansion in the past decade [1]. Energy consumption plays a crucial role in national development, and regional disparities in production levels and energy demand result in variations in how energy is acquired and utilized. These differences further influence overall economic growth, environmental protection, and quality of life [2]. A shortage of energy within an economy can negatively impact various aspects of economic and social development [3]. Consequently, many economies are actively working to transform their energy consumption structures to achieve more efficient and sustainable energy utilization patterns [2,4,5].

With the rapid advancement of China’s economy, energy demand is increasing across all production sectors [6]. However, China is not a foremost producer of traditional energy sources and relies heavily on imports, which has intensified pressure on its energy consumption as demand continues to rise [7]. The energy consumption structure in China urgently needs optimization, particularly due to the still-high share of traditional energy sources such as coal and crude oil. This situation presents significant challenges to the transformation and escalation of the energy consumption structure and achieving the goal of carbon neutrality [8]. China needs to prioritize energy consumption and environmental protection issues, optimize its energy consumption structure, and enhance coordination in energy use across various industrial sectors [9]. Therefore, studying the evolution of national energy consumption structures and proposing policy recommendations based on these findings is crucial for optimizing energy consumption and ensuring sustainable energy growth in each country [9,10,11].

In 2018, the Chinese government released the amended Law of the People’s Republic of China on Energy Conservation, which addresses various aspects of energy conservation management, including a review system for fixed-asset investment terms and the phase-out of outdated, energy-inefficient products and the implementation of an energy conservation evaluation. The government aims to achieve significant progress in energy conservation while exploring a more rational and efficient energy consumption structure. Therefore, studying the consumption and transmission processes of energy on China’s industrial and demand sides is of great practical importance, as it can inform energy policy formulation and the optimization of industrial structure. Currently, there are three main types of energy consumed by countries: carbon energy (CE) (e.g., coal, coke), oil and gas energy (OGE) (e.g., oil, gas), and electric energy (EE) [12,13,14]. By analyzing the overall trends and transmission paths of these three energy types, this paper holds significant practical value for both macro energy policy development and the selection of micro-level energy utilization paths. Using the input–output model, this paper constructs the MHEM-SDA-SPD analytical framework to empirically assess China’s energy consumption from 2018 to 2022 across three levels: overall, production stage, and industrial chain. The study identifies the correlation effects, driving forces, and transmission paths of energy consumption, providing precious insights for the evolution of energy allocation policies and the advancement of industrial energy efficiency.

The marginal contributions of this paper include the following three aspects: (1) Unlike single approaches for assessing the driving forces and transmission paths of energy consumption, we have constructed a combined MHEM-SDA-SPD analytical framework to conduct a detailed analysis of China’s energy consumption. (2) We classify energy consumption into three categories—CE, OGE, and EE—to effectively analyze the consumption patterns of various energy types across different sectors. This classification provides a theoretical foundation for optimizing energy management decisions and elevating energy efficiency. (3) This paper discusses from the input–output perspective and analyzes the three kinds of energy consumption correlation, consumption driving forces, and transmission paths of each industrial sector, which can effectively analyze the links between the various sectors of the economy and make up for the shortcomings of linear regression models in this regard.

2. Literature Review

There is a broad consensus on the crucial role of energy consumption structure in national development [4,6,9]. Numerous scholars have executed in-depth studies on the evolution of energy consumption structures and their associated impacts. Xu et al. [15] used a simple energy consumption decomposition model to analyze the energy consumption structure to analyze China’s carbon dioxide emissions and predicted that China’s peak carbon dioxide emissions may peak in 2030 and 2025 under the planned energy structure scenario and the low-carbon energy structure scenario, respectively. Zhang et al. [7] found that technological progress contributes to simplifying the energy consumption structure; however, crude oil price fluctuations weaken this effect and heighten financial risks by using provincial panel data from China. Yasmeen et al. [11] examined the effect of energy consumption structures in OECD member countries during the 1996–2015 business cycle. Their findings indicate that energy demand varied between economic growth and recession periods, with institutions and foreign investment playing key moderating roles. Brodny and Tutak [4] conducted an exhaustive analysis of the structure and volume of energy consumption in the industrial sectors of EU countries from 1995 to 2019, providing precious insights for energy restructuring and policy implementation in the region. Azam et al. [16] explored the influence of institutional quality on energy consumption and environmental trends over time in 66 developing countries. Their study found that institutional quality positively influenced oil and fossil energy consumption, but it did not improve environmental quality in developing countries over time. While the research has fully probed the role of energy consumption structure in national advancement and its associated impacts, few studies have focused on analyzing the current state and flow of energy consumption structures in the production and demand sectors of the economy. This study seeks to fill that gap by addressing this aspect, thereby complementing existing research.

Input–output analysis is a crucial way of studying the balance between sectors in the national economy. It can provide insights into economic development trends and effectively analyze intersectoral economic linkages within the IO table [17,18]. Additionally, input–output analysis can be integrated with other factors such as investment, environmental impact, and resource allocation to examine their consumption and flow across various sectors [19,20,21,22,23]. This approach is increasingly applied to analyze energy consumption at industrial, regional, and national levels, offering a theoretical foundation for optimizing energy consumption structures and formulating energy policies [24,25,26,27].

In input–output analysis, numerous scholars have applied various methods to examine the consumption of relevant factors within the input–output table, all of which are effective in analyzing the consumption correlation effects and flows across different sectors [17,19,21,25,26,27]. This paper focuses on analyzing the consumption correlations and flows of three energy sources across different economic sectors in China. At the methodological level, we first employ the MHEM method to probe the intrinsic effects of energy consumption in each sector. Next, this paper utilizes the SDA method to investigate the driving forces behind energy consumption. Finally, we apply the SPD method to discuss the flow of energy consumption across sectors in the input–output table. By combining these three methods, we can more effectively assess the energy consumption correlations and flows within the economic system, providing a theoretical basis for adjusting the energy consumption structure and formulating energy policies. This paper focuses on analyzing the consumption correlations and flows of three energy sources across different economic sectors in China. At the methodological level, we first employ the MHEM method to probe the intrinsic effects of energy consumption in each sector. Next, this paper utilizes the SDA method to investigate the driving forces behind energy consumption. Finally, we apply the SPD method to discuss the flow of energy consumption across sectors in the input–output table. By combining these three methods, we can more effectively assess the energy consumption correlations and flows within the economic system, providing a theoretical basis for adjusting the energy consumption structure and formulating energy policies. The research framework of this paper is shown in Figure 1.

3. Methodology and Data

Due to input–output analysis, this section outlines the implementation of the MHEM-SDA-SPD analytical framework, providing a foundation for the subsequent empirical analysis. Additionally, we discuss the data sources and sector classification used in the paper.

3.1. Modified Hypothetical Extraction Method

The hypothetical extraction method (HEM) can analyze the resulting diversifications in the total output and assess the correlations between the extracted sector and others by assuming the migration of a sector from the economy [28,29]. We employ the modified hypothetical extraction method (MHEM) to build a correlation effect model of energy consumption across various sectors in China. MHEM can measure the vertical integration consumption (including direct and indirect consumption) and analyze the linkage effect from four components, i.e., MIE, MME, MNBL, and MNFL. Compared to traditional HEM, MHEM effectively analyzes the combination of technological capabilities and vertically integrated consumption across all sectors, enabling a deeper understanding of how one sector impacts others within the economy [30]. MHEM is now widely employed to examine energy consumption correlations at both national and regional industrial levels and to formulate policy recommendations based on the findings [28]. The implementation process of MHEM is as follows:

Assume that there are n sectors in the economy $\theta$. Sector i’s input to sector j is represented by $x_{ij}$, sector i’s final demand is represented by $y_i$, sector i’s final demand vector is represented by Y = ($y_i$), sector i’s output vector is represented by X = ($x_i$), and sector i’s total output is represented by $x_j$. Let $a_{ij}$ be the coefficient of direct consumption:

$$a_{ij}={\displaystyle \frac{x_{ij}}{X_j}}$$

(1)

The matrix of consumption coefficients is denoted by A = ($a_{ij}$); the economy θ could be written as follows:

$$X=AX+Y={\left(I-A\right)}^{-1}Y$$

(2)

In the research process, the economic system can be regarded as composed of sector ${\theta }_k$ and sector ${\theta }_{-k}$; then, the economy $\theta$ can be denoted as:

$$\theta =\left[\begin{array}{c}{\theta }_{k,k} ,{\theta }_{k,-k} \\ {\theta }_{-k,k}, {\theta }_{-k,-k} \end{array}\right]$$

(3)

According to the above theory, the economy before the extraction of sector ${\theta }_k$ can be denoted as:

$$\left[\begin{array}{c}{X}_k\\ X_{-k}\end{array}\right]=\left[\begin{array}{c}{A}_{k,k} , A_{k,-k}\\ A_{-k,k} , A_{-k,-k} \end{array}\right]\left[\begin{array}{c}{X}_k\\ X_{-k}\end{array}\right]+\left[\begin{array}{c}{Y}_k\\ Y_{-k}\end{array}\right]=\left[\begin{array}{c}{\beta }_{k,k} , {\beta }_{k,-k}\\ {\beta }_{-k,k} , {\beta }_{-k,-k} \end{array}\right]\left[\begin{array}{c}{Y}_k\\ Y_{-k}\end{array}\right]$$

(4)

where $\left[\begin{array}{c}{\beta }_{k,k} , {\beta }_{k,-k}\\ {\beta }_{-k,k} , {\beta }_{-k,-k} \end{array}\right]={(I-A)}^{-1}$.

After the extraction of sector ${\theta }_k$, assuming that there is no economic activity between sector ${\theta }_k$ and sector ${\theta }_{-k,}$, the matrix form of its economic system can be denoted as:

$$\left[\begin{array}{c}{X}_k^{∆}\\ X_{-k}^{∆}\end{array}\right]=\left[\begin{array}{c}{A}_{k,k} , 0\\ 0 , { A}_{-k,-k} \end{array}\right]\left[\begin{array}{c}{X}_k^{∆}\\ X_{-k}^{∆}\end{array}\right]+\left[\begin{array}{c}{Y}_k\\ Y_{-k}\end{array}\right]=\left[\begin{array}{c}{\left(I-A_{k,k}\right) }^{-1} , 0\\ 0 , { \left(I-A_{-k,-k}\right) }^{-1}\end{array}\right]\left[\begin{array}{c}{Y}_k\\ Y_{-k}\end{array}\right]$$

(5)

Equations (4) and (5) show the difference in the total output of the economy after the extraction of sector ${\theta }_k$:

$$X-X^{∆}=\left[\begin{array}{c}{X}_k-X_k^{∆}\\ X_{-k}-X_{-k}^{∆}\end{array}\right]=\left[\begin{array}{c}{\beta }_{k,k}-{\left(I-A_{k,k}\right) }^{-1} , { \beta }_{k,-k}\\ {\beta }_{-k,k} , {\beta }_{-k,-k}-{ \left(I-A_{-k,-k}\right) }^{-1} \end{array}\right]\left[\begin{array}{c}{Y}_k\\ Y_{-k}\end{array}\right]=\left[\begin{array}{c}{\gamma }_{k,k} , {\gamma }_{k,-k}\\ {\gamma }_{-k,k} , {\gamma }_{-k,-k} \end{array}\right]\left[\begin{array}{c}{Y}_k\\ Y_{-k}\end{array}\right]$$

(6)

Furthermore, the energy consumption of different sectors in China is represented by the matrix DC, which is introduced in this study. DC = ${\left[{\omega }_1,{\omega }_2,{\dots \omega }_i\dots {\omega }_j\right]}^{\prime }$, and sector i’s direct energy consumption, is represented by ${\omega }_i$.

Sector i’s direct energy consumption coefficient is denoted by $Q_i$, while $Q_j$ is the direct energy consumption coefficient. Then, according to Formula (1), we can get:

$$Q_j={\displaystyle \frac{{\omega }_j}{X_j}}$$

(7)

The complete consumption coefficient vector BX can be computed as:

$$BX=Q_j{\left(I-A\right)}^{-1}$$

(8)

Then, the vertical integration consumption of sector j is:

$${VIC}_j=\sum _{i=1}^n{Q}_i{c}_{ij}{y}_j$$

(9)

${VIC}_j$ denotes the indirect and direct energy consumption of sector j to achieve final demand, $c_{ij}$ is the element of the Leonitef inverse matrix, $y_j$ represents the final demand of industry j. VIC combines the direct energy consumption coefficient, the complete energy consumption coefficient, and the final demand and uses numerical values to represent the industrial consumption characteristics, which can better analyze the indirect and direct use of energy in various sectors.

Additionally, MHEM combines HEM and VIC to decompose the total association effect into four independent association factors [31,32]: MIE represents energy consumption in the ${\theta }_k$ sector. MME calculates the energy consumed by the ${\theta }_{-k}$ sector to purchase the products produced by the ${\theta }_k$ sector as intermediate inputs, which then flow into the ${\theta }_k$ sector again and finally form the final demand $Y_k$. MNFL represents energy consumption by the ${\theta }_k$ sector’s products to be purchased by the ${\theta }_{-k}$ sector to produce the final demand $Y_{-k}$. MNBL reflects the indirect and direct energy consumption of ${\theta }_k$ in the process of ${\theta }_k$ sector purchasing products of ${\theta }_{-k}$ sector to obtain the final demand $Y_k$. With the above theoretical support, the calculation process of MIE, MME, MNFL, and MNBL could be realized by the following equations:

$$MIE=\widehat{Q_k}{\left(I-A_{k,k}\right) }^{-1} Y_k$$

(10)

$$MME=\widehat{Q_k}\left[{\beta }_{k,k}-{\left(I-A_{k,k}\right) }^{-1}\right] Y_k$$

(11)

$$MNFL=\widehat{Q_k}{\beta }_{k,-k}{Y}_{-k}$$

(12)

$$MNBL=\widehat{Q_{-k}}{\beta }_{-k,k}{Y}_k$$

(13)

In summary, based on the calculation methods for the four factors outlined above, the calculation methods for direct consumption and vertically integrated consumption can also be expressed as follows: VIC = MIE + MME + MNBL, DC = MIE + MME + MNFL.

3.2. Structural Decomposition Analysis

Structural decomposition analysis (SDA) is a methodology that examines changes in the structural relationships between variables in an economy [33]. SDA helps to certify the relative contributions of different factors, offering deeper insights into the relationship between economic activities and environmental impacts [34]. We apply the SDA method to discuss changes in energy consumption across industries and further explore the structural factors influencing energy consumption.

Referring to Zhang et al. [35], we decompose the final demand Y into:

$$Y=MTDH$$

(14)

Among them, the matrix M is a (n × 3)-dimensional matrix, which expresses the product structure of the final demand. Matrix T is a (3 × 1) dimensional matrix, representing different categories of final demands. The scalar numbers D and H stand for the per capita allocation of final demand and the total population at the end of the year in the economic system, respectively.

Due to Equation (7), the formula for energy consumption can be changed to:

$$e=\widehat{Q}LMTDH$$

(15)

This paper uses different subscripts to represent IO tables of different years, 0 is the base year, and t is the corresponding analysis year. Therefore, the energy consumption during the analysis period can be expressed as follows:

$$\begin{gathered} \Delta \mathrm{e}=\widehat{Q_t}{L}_t{M}_t{T}_t{D}_t{H}_t-\widehat{Q_0}{L}_0{M}_0{T}_0{D}_0{H}_0 \\ =e\left(\Delta \widehat{Q}\right)+e\left(\Delta L\right)+e\left(\Delta M\right)+e\left(\Delta T\right)+e\left(\Delta D\right)+e\left(\Delta H\right) \end{gathered}$$

(16)

In the above formula, the change in relevant factors is represented by Δ. Further, we use the two-pole decomposition method to compute the structural situation of energy consumption:

$$\begin{gathered} \Delta e={\displaystyle \frac{1}{2}}\Delta \widehat{Q}\left(L_0{M}_0{T}_0{D}_0{H}_0+L_t{M}_t{T}_t{D}_t{H}_t\right) \\ +{\displaystyle \frac{1}{2}}\left(\widehat{Q_0}\Delta L{M}_t{T}_t{D}_t{H}_t+\widehat{Q_t}\Delta L{M}_0{T}_0{D}_0{H}_0\right) \\ +{\displaystyle \frac{1}{2}}\left(\widehat{Q_0}{L}_0\Delta M{T}_0{D}_0{H}_0+\widehat{Q_t}{L}_t\Delta M{T}_t{D}_t{H}_t\right) \\ +{\displaystyle \frac{1}{2}}\left(\widehat{Q_0}{L}_0{M}_0\Delta T{D}_t{H}_t+\widehat{Q_t}{L}_t{M}_t\Delta T{D}_0{H}_0\right) \\ +{\displaystyle \frac{1}{2}}\left(\widehat{Q_0}{L}_0{M}_0{T}_0\Delta D{H}_t+\widehat{Q_t}{L}_t{M}_t{T}_t\Delta D{H}_0\right) \\ +{\displaystyle \frac{1}{2}}\left(\widehat{Q_0}{L}_0{M}_0{T}_0{D}_0+\widehat{Q_t}{L}_t{M}_t{T}_t{D}_t\right)\Delta H \end{gathered}$$

(17)

Meanwhile, we split the final demand to analyze the effect of different demand categories on energy consumption growth [36]. This study divides the final requirements into three parts: $Y=Y_{dc}+Y_{cf}+Y_{ne}$. Among them, $Y_{dc}$(n × 1), $Y_{cf}$(n × 1) and $Y_{ne}$(n × 1) represent the vectors of domestic consumption, capital formation, and net export, respectively. We still apply the above SDA method to analyze the impact on final demand.

3.3. Structural Path Decomposition

Structural path decomposition (SPD) treats the economic system as a multi-level network and identifies the effect of each path on energy consumption by breaking down the contributions of different production chains or paths [37]. Based on SDA, SPD provides a more effective way to trace the specific sources of energy use.

Assume L is the Leontief inverse matrix:

$$L={\left(1-A\right)}^{-1}=I+A+A^2+A^3+\cdots ,$$

(18)

Based on Equation (15), the energy consumption equation can be transformed into:

$$\begin{gathered} e=\widehat{Q}LMTDH=e=\widehat{Q}\left(I+A+A^2+A^3+\cdots \right)MTDH, \\ =\widehat{Q}MTDH+\widehat{Q}AMTDH+\widehat{Q}{A}^{2}MTDH+\cdots , \end{gathered}$$

(19)

Then, decompose the consumption path of each level:

$$\begin{gathered} \Delta e=\left(\Delta \widehat{Q}\right)MTDH+\widehat{Q}\left(\Delta M\right)TDH+\widehat{Q}M\left(\Delta T\right)DH+\widehat{Q}MT\left(\Delta D\right)H+\widehat{Q}MTD\left(\Delta H\right), \\ +\left(\Delta \widehat{Q}\right)AMTDH+\widehat{Q}\left(\Delta A\right)MTDH+\widehat{Q}A\left(\Delta M\right)TDH+\widehat{Q}AM\left(\Delta T\right)DH+\widehat{Q}AMT\left(\Delta D\right)H+\widehat{Q}AMTD\left(\Delta H\right), \\ +\left(\Delta \widehat{Q}\right)AAMTDH+\widehat{Q}\left(\Delta A1\right)AMTDH+\widehat{Q}A\left(\Delta A2\right)MTDH+\widehat{Q}AA\left(\Delta M\right)TDH+\widehat{Q}AAM\left(\Delta T\right)DH+\widehat{Q}AAMT\left(\Delta D\right)H+\widehat{Q}AAMTD\left(\Delta H\right)+\cdots , \end{gathered}$$

(20)

Among them, $\Delta A$ represents the change in direct consumption coefficient, $\Delta A1$ and $\Delta A2$ means the same as $\Delta A$.

Following Owen et al. [38], this paper classifies different types of energy consumption paths for easier analysis. As shown in Formula (20), each row represents a distinct decomposition path. For the analysis, we set the path decomposition level to three levels.

The first-level decomposition path is:

$${ P}_{1st}=\left(\Delta \widehat{Q_i}\right)M_{ik}{T}_{k}DH,\dots ,{\widehat{Q_i}M}_{ik}\left({\Delta T}_k\right)DH,\dots ,{\widehat{Q_i}M}_{ik}TD\left(\Delta H\right)$$

(21)

The second-level decomposition path is:

$$P_{2nd}=\left(\Delta \widehat{Q_i}\right)A_{ij}{M}_{jk}{T}_{k}DH,\dots ,{\widehat{Q_i}{A}_{ij}M}_{jk}\left({\Delta T}_k\right)DH,\dots ,{\widehat{Q_i}{A}_{ij}M}_{jk}TD\left(\Delta H\right)$$

(22)

The third-level decomposition path is:

$$P_{3rd}=\left(\Delta \widehat{Q_i}\right)A_{ij}{A}_{jl}{M}_{lk}{T}_{k}DH,\dots ,{\widehat{Q_i}{A}_{ij}{A}_{jl}M}_{lk}\left({\Delta T}_k\right)DH,\dots ,{\widehat{Q_i}{A}_{ij}{A}_{jl}M}_{lk}TD\left(\Delta H\right)$$

(23)

In the above formula, the change in relevant factors is represented by Δ, with the corresponding formula indicating the energy path consumption resulting from these changes. Through path decomposition, we can more effectively analyze the channels of energy consumption and perform a detailed factor analysis.

3.4. Data

The data applied in this study include energy consumption data for various sectors, input–output tables, population data, and the price index. The China Energy Statistical Yearbook is the source of the energy statistics for every sector. For consistency and ease of comparison, all energy consumption is converted into million tons of standard coal (MtSC). Since the input–output table from the National Bureau of Statistics of China is not available for consecutive years and is less convenient for analysis, this study utilizes the continuous annual input–output tables for the years 2018 to 2022 published by the Asian Development Bank (ADB) to ensure data quality. The IO tables are based on international standards for department classification and are suitable for use by scholars in academic research. Additionally, the input–output table is denominated in U.S. dollars. To account for exchange rate fluctuations, we convert the input–output data into RMB, applying the average annual exchange rate. To mitigate the effects of inflation, the input–output table is deflated with 2010 as the base year. The China Statistical Yearbook provided the study’s price deflation index for various industries and demographic data. Given the discrepancies between sector classifications in the IO table and the energy data, we categorize the IO table into 21 sectors based on available data. For a more complete sector classification, please refer to

Table A1. Sector classification directory.

Serial Number	Industry Title
S1	Agriculture, hunting, forestry, and fishing
S2	Mining and quarrying
S3	Food, beverages, and tobacco
S4	Textiles and textile products
S5	Leather, leather products, and footwear
S6	Wood and products of wood and cork
S7	Pulp, paper, paper products, printing, and publishing
S8	Coke, refined petroleum, and nuclear fuel
S9	Chemicals and chemical products
S10	Rubber and plastics
S11	Other nonmetallic minerals
S12	Basic metals and fabricated metal
S13	Machinery, nec
S14	Electrical and optical equipment
S15	Transport equipment
S16	Manufacturing, nec, recycling
S17	Electricity, gas, and water supply
S18	Construction
S19	Retail, hotels and restaurants
S20	Transportation and post and telecommunications
S21	Comprehensive service industry

Note: S21 includes the following sectors: financial intermediation sector, real estate activities sector, Renting of M&Eq and other business activities sector, public administration and defense; compulsory social security sector, education sector, health and social work sector, other community, social, and personal services sector, private households with employed persons sector.

in Appendix A.

4. Empirical Analysis

In view of the analytical framework developed in the previous section, this section empirically examines the correlation, driving forces, and key transmission paths of energy consumption across various industrial sectors in China, providing a corresponding analysis.

4.1. Sectoral Correlation of Energy Consumption

China’s energy consumption between 2018 and 2022 is shown in Figure 2. Total energy consumption in China has consistently enhanced, with the consumption of all three types of energy maintaining a growth trend. The proportion of CE consumption has remained high, exceeding 50%. Over the five-year period, CE consumption grew by 404.47 MtSC, OGE consumption rose by 198.26 MtSC, and EE consumption increased by 159.41 MtSC. Among the three energy types, CE consumption experienced the largest increase, indicating that CE remains the dominant source of China’s energy consumption, while OGE and EE consumption have shown relatively stable growth. This shows that China’s demand for traditional energy is still huge, and traditional energy plays a major role in China’s energy supply. It also highlights how important it is for China to step up its efforts to promote energy-saving measures.

Using the calculation formula for vertically integrated consumption and the energy data obtained from each sector, we analyzed the three types of energy consumption across sectors. The sum of direct consumption (DC) and vertically integrated consumption (VIC) across all sectors is equal for the economy. This is because the total energy consumption, both indirect and direct, used by various sectors within the economic system to meet their final demands is equal to the summation of the actual energy consumption of all industries. From the perspective of an individual sector, if VIC > DC, it suggests that the sector receives energy consumption from other sectors, making it a net energy importer. Conversely, if VIC < DC, the sector is a net energy exporter, supplying energy consumption within the system.

The specific results of energy consumption are presented in Figure 3. For CE, the electricity, gas, and water supply sector (S17) has the highest direct consumption, with a consumption of 1814.12 MtSC, significantly higher than other sectors. This indicates that most CE consumption is applied in the production of other energy sources. The sector with the largest vertically integrated consumption is the construction sector (S18), consuming 1183.94 MtSC. For OGE, the coke, refined petroleum, and nuclear fuel sector (S8) has the highest direct consumption, with a consumption of 997.38 MtSC, again much higher than other sectors. The construction sector (S18) has the largest vertical integration consumption, at 452.14 MtSC. EE consumption is more balanced compared to the other two energy sources. The basic metals and fabricated metal sector (S12) has the largest direct consumption, with 200.23 MtSC, significantly higher than other sectors. The construction sector (S18) has the largest vertical integration consumption, at 254.78 MtSC. These results demonstrate that most energy consumption is concentrated in the energy production or conversion sectors. As the construction industry is highly energy-intensive, its energy needs are substantial, relying heavily on energy input from the energy production sectors [39].

Figure 4 demonstrates the energy consumption and transfer of each sector. The innermost circle in the figure represents the proportion of the four correlation factors for S1, while the outermost circle shows the proportion for S21. In the consumption process, MNBL is the major contributor to both CE and OGE across sectors. In EE consumption, MNFL and MNBL both play significant roles, with MME accounting for the smallest proportion among the three energy consumption processes. These results indicate that China’s energy consumption is primarily driven by production processes, ultimately fulfilling final demand. The heavy reliance of social production on energy is a widespread issue, and nations are working towards reducing energy consumption through cleaner production methods. The analysis also includes energy consumption data for 2018, 2019, 2020, and 2021, which shows minimal deviation from the 2022 results. Due to space constraints, these results are omitted from the article.

4.2. Analysis of Driving Forces of Energy Consumption

SDA is a powerful tool for analyzing the causes of energy consumption and identifying the impact of different factors [40]. In view of the derivation of Equation (17), this study decomposes energy consumption into six economic driving forces to examine how various factors have influenced energy consumption over time. The decomposition results for the three types of energy consumption are presented in Figure 5. Overall, energy consumption for all three types of energy—CE, OGE, and EE—has increased from 2018 to 2022, which aligns with the quick progress of China’s economy. The decomposition analysis reveals that the main driving forces for energy consumption in all three categories are the consumption intensity (ΔQ) and per capita demand effects (ΔD). Energy consumption intensity reflects the efficiency with which energy is used during economic development, and is closely linked to China’s sustained focus on economic and industrial growth over the past several decades. As energy use and economic development have expanded in China, individual energy demand has also risen, making the per capita demand effect increasingly significant. Annual energy consumption trends reveal a consistent upward trajectory, driven by ongoing industrialization and technological advancement, both of which demand substantial energy inputs [41]. This rising demand highlights the need to secure a sufficient energy supply for both industrial production and residential consumption [42]. These findings highlight that China’s future energy consumption will continue to rise, necessitating significant energy support to maintain sustainable development.

Decomposing final demand is essential for understanding the specific types of demand driving energy consumption [43]. We examine the influence of changes in diverse final demand types on energy consumption. Figure 6 shows the energy consumption resulting from different demand categories from 2018 to 2022. CE consumption is mainly due to domestic consumption ($Y_{dc}$) and net exports ($Y_{ne}$), and to a lesser extent capital formation ($Y_{cf}$). OGE and EE consumption is mainly derived from exports ($Y_{ne}$), followed by domestic consumption ($Y_{dc}$) and the least from capital formation ($Y_{cf}$).

Figure 7 illustrates the evolution of energy consumption in different sectors under the influence of different economic factors. The contribution to the growth in CE consumption comes from the electricity, gas, and water supply sector (S17), with mostly decreases in other sectors. The leading sectors of OGE’s consumption growth are the coke, refined petroleum, and nuclear fuel sector (S8, 51.5%), the chemicals and chemical products sector (S9, 18.76%), and the electricity, gas, and water supply sector (S17, 12.41%); other sectors showed less change. This suggests that most of the growth in CE and OGE is used for energy conversion to provide energy that can be directly used by other sectors. The contribution of EE consumption growth is more pronounced in the chemicals and chemical products (S9, 10.52%), the basic metals and fabricated metal sector (S12, 14.1%), and the comprehensive service industry sector (S21, 18.2%), and relatively even in other sectors. The demand for electricity in various sectors is gradually increasing, which requires China to maintain continuous growth in power transmission to ensure the normal use of energy.

4.3. Critical Path Analysis of Energy Consumption

SPD helps identify the key driving forces of energy consumption and tracks how these paths evolve across different industrial sectors [44]. In this part, we utilize the SPD method to discuss the consumption paths of the three primary energy sources—CE, OGE, and EE—during the period from 2018 to 2022. The results of the CE consumption critical path analysis are shown in Figure 8. In terms of path transmission factors, consumption intensity (ΔQ) is the key influence on CE energy consumption, occupying 15 out of 30 key decomposition paths. From a final demand perspective, the majority of CE energy consumption is directed toward capital formation ($Y_{cf}$), which occupies 16 critical paths. This suggests that a significant portion of CE consumption is used to support energy security for capital accumulation. In terms of conduction paths, the energy transmission for CE starts primarily with the electricity, gas, and water supply sector (S17) and the basic metals and fabricated metal sector (S12), indicating that CE energy is still more of a key element in the production of other products. The final sector of this energy consumption is largely the construction sector (S18), which requires large amounts of energy and materials to support the construction process. Further analysis of the transmission paths reveals that the electricity, gas, and water supply sector (S17) goes mostly to domestic consumption ($Y_{fc}$) (ranked 3rd, 4th, 8th, and 22nd) and the basic metals and fabricated metal sector (S12) goes mostly to net exports ($Y_{ne}$) (ranked 5th, 7th, 9th, 10th, 21st, and 30th) in both the first-order path and the second-order path. Of the third-order paths, the electricity, gas, and water supply sector (S17) mostly constitutes capital formation ($Y_{cf}$), and its main transmission path is: “electricity, gas, and water supply sector (S17) → other nonmetallic minerals sector (S11) → construction sector (S18) → capital formation ($Y_{cf}$)” (ranked 11th, 17th, and 18th) and “electricity, gas, and water supply sector (S17) → basic metals and fabricated metal sector (S12) → construction sector (S18) → capital formation ($Y_{cf}$)” (ranked 20th and 23rd). The influences of these third-order paths are mainly caused by ΔQ, ΔA2, ΔD, and ΔA1, suggesting an indirect effect of CE energy supply on fixed capital formation, where CE consumption drives the construction sector, which in turn contributes to capital formation. CE is not readily usable in its raw form and must undergo multiple transformation processes before being transmitted to other sectors and converted into final products, which aligns with our empirical findings. CE remains the dominant source of energy consumption in China. Therefore, measures must be implemented to reduce environmental pollution during the conversion process to support sustainable development. These include limiting the construction of new coal-fired power plants, upgrading existing units, and phasing out outdated production capacities.

The results of the OGE energy consumption critical path analysis are demonstrated in Figure 9. In terms of path transmission factors, consumption intensity (ΔQ) and per capita demand effect (ΔD) are the key influences on CE energy consumption, occupying 11 and 8 out of 30 key decomposition paths. From a final demand perspective, the destination of OGE energy consumption is mostly domestic consumption ($Y_{dc}$), which occupies the 20 critical paths. In terms of conduction paths, the starting sectors of the conduction paths of OGE energy are mostly the coke, refined petroleum, and nuclear fuel sector (S8), which occupy 15 critical paths. This indicates that OGE energy is predominantly consumed in the extraction or refining process, which also supports the production of other energy sources, such as various types of oil. The endpoints of the pathway conduction are mostly in the construction sector (S18) and the comprehensive service industry sector (S21), indicating that OGE is mostly used in the construction and service industries after product conversion. In terms of path transmission statistics, the direct paths “coke, refined petroleum, and nuclear fuel sector (S8) → domestic consumption ($Y_{dc}$)” (ranked 7th, 10th, and 15th) and “coke, refined petroleum, and nuclear fuel sector (S8) → net exports ($Y_{ne}$)” (ranked 3rd, 9th, 19th, and 20th) occur most frequently. In the high-order transmission path starting from the coke, refined petroleum, and nuclear fuel sector (S8), the transmission endpoint is the comprehensive service industry sector (S21) and the destination is mostly domestic consumption ($Y_{dc}$) (ranked 2nd, 5th, 16th, 18th, 26th, and 30th), while the transmission endpoint is the construction sector (S18) and the destination is mostly capital formation ($Y_{cf}$) (ranked 4th, 8th, 13th, 17th, 22nd, 25th, and 27th). These results show the great role that OGE plays in the construction of facilities and in daily production and life, and that it will remain an important source of energy for daily use until new energy sources are fully utilized. OGE is transferred to other sectors via extraction and refining processes, thereby supporting the production activities of downstream industries. China depends heavily on imports of oil and natural gas, which serve as primary energy sources for both industrial and residential use. Promoting the rational use of energy remains essential and can be achieved through energy-saving and emission-reduction measures, such as upgrading oil extraction and refining equipment and adopting electrified terminals.

The results of the critical path analysis of EE energy consumption are shown in Figure 10. In terms of path transmission factors, consumption intensity (ΔQ) remains a key influence on electricity energy consumption, occupying 15 of the 30 key decomposition paths. From a final demand perspective, the roles played by domestic consumption, capital formation, and net exports in electricity consumption do not differ much, occupying 9, 10 and 11 critical paths, respectively, suggesting that electricity is relatively balanced on the demand side. In terms of conduction paths, the 30 critical paths are mostly first-order conduction paths, which also proves that electricity utilization is highly direct. The more common direct paths are: “comprehensive service industry sector (S21) → domestic consumption ($Y_{dc}$)” (ranked 3rd, 4th and 5th) and “basic metals and fabricated metal sector (S12) → net export ($Y_{ne}$)” (ranked 6th, 7th, and 11th). The most common final sector for secondary and tertiary pathways is the construction sector (S18), and the destination is mostly capital formation ($Y_{cf}$) (ranked 1st, 2nd, 12th, 16th, 20th, and 28th). Electricity, as the most widely used energy source in daily life, exhibits substantial demand across all sectors. Our analysis indicates that electricity consumption along the critical path is significant across all sectors and plays a vital role in satisfying diverse types of final demand. Both China’s industrial development and residential life impose a high demand for electricity. Therefore, it is essential to expand the supply of clean electricity—such as hydropower and wind power—while ensuring overall energy security. China has made continuous progress in these areas to reduce its reliance on conventional energy sources.

In general, the transmission pathways of the three energy sources differ. CE energy consumption predominantly follows advanced path decomposition, whereas OGE and EE energy consumption are mostly characterized by primary path decomposition. First-order pathways represent the direct routes of energy consumption within a sector, while higher-order pathways (second-order and beyond) track the flow of energy consumption from upstream to downstream sectors. Analyzing these pathways can provide valuable insights for implementing targeted energy consumption intervention policies [45]. In terms of driving forces, consumption intensity plays a significant role in structural path decomposition, accounting for over 45% of the total. From a final demand perspective, the majority of energy consumption is directed towards domestic consumption and capital formation. The energy demand in the consumer market primarily stems from direct energy supply and product consumption, while the demand in capital formation—particularly in fixed capital formation industries such as construction and manufacturing—remains substantial [46]. This is clearly reflected in the analysis results. In conclusion, energy consumption is a critical driver of national development. Identifying the key factors and pathways of energy consumption is highly beneficial for facilitating energy utilization and formulating relevant policies.

5. Conclusions

As a large energy-consuming country, focusing on the development and evolutionary path of China’s energy consumption will help us to grasp the key triggers of energy consumption and formulate appropriate policies to optimize the energy consumption structure. In the paper, we construct the MHEM-SDA-SPD analytical framework to analyze the correlation structure, driving forces, and structural paths of China’s energy consumption in view of China’s 2018–2022 IO table data to investigate the structural relationship of energy consumption among industrial sectors. The results of the study show the following: First, by means of the MHEM method, we analyzed the flow of energy consumption in each sector. Most of the three energy outputs originate from the energy production or conversion sector, while the energy input sector is mostly the building sector. Then, we utilized the SDA model to discuss the driving forces of energy consumption. We found that China’s energy consumption shows year-on-year growth, while consumption intensity and per capita demand effects are the crucial triggers of energy consumption. Finally, this study applies the SPD model to study the critical path of energy consumption. The findings indicate that the destination of energy consumption is mostly domestic consumption and capital formation. On the energy dissipation conduction path, the starting sectors for CE and OGE energy consumption are mainly the electricity, gas, and water supply sector and the coke, refined petroleum, and nuclear fuel sector; path-ending sectors are mostly the construction sector and the comprehensive service industry sector. Specifically, the first-order path with the largest overall CE impact is “electricity, gas, and water supply sector → domestic consumption”, while the higher-order paths are mostly subpaths of “electricity, gas, and water supply sector → capital formation”. The main supply and transfer path for OGE is “coke, refined petroleum, and nuclear fuel sector→ domestic consumption” and its subpaths. EE consumption is relatively balanced between sectors, the demand for electricity in all sectors is large, and its conduction path is mostly a first-order path, indicating the directness of electricity use, which can be widely used in various sectors of production and residential life.

In view of the research above, we suggest the following policy recommendations to support the sustainable development of energy consumption in China:

(1)

Promote the structural transformation of energy consumption and increase the proportion of clean energy. Statistical data clearly show that China’s reliance on traditional energy sources continues to grow, despite its limited domestic energy reserves. To address this, the government should optimize the shift towards clean energy sources such as hydroelectric, wind, and solar power. This transition could be achieved by offering subsidies to clean energy enterprises, investing in the progress of clean energy infrastructure, and incentivizing households to adopt cleaner energy solutions. Supporting the scaling up of these renewable sources would help reduce China’s dependence on traditional energy and promote more sustainable energy consumption.

(2)

Improve energy efficiency in energy-intensive industries and maximize energy utilization. Energy-efficient production is a key strategy actively adopted by industrialized nations to optimize energy utilization. The empirical results indicate that both the construction and manufacturing sectors exhibit substantial energy consumption. To further enhance energy efficiency in China, high-energy-consuming sectors such as industry and construction should prioritize the utilization of energy-saving technologies and practices. For example, promoting the development of energy-efficient industrial equipment and advancing green building technologies can significantly decrease energy consumption in these sectors. In addition, the transportation sector should focus on expanding the application of new energy vehicles and developing electrified rail systems, which would help to decrease energy waste and facilitate overall energy efficiency.

(3)

Strengthen government regulation and policy support. Given the growing energy consumption, achieving sustainable energy-saving development is a pressing challenge for the Chinese government. To address this, the government can establish stringent energy-efficiency standards and set entry thresholds for energy-intensive industries, thereby phasing out outdated production capacities and optimizing the allocation of energy resources. Additionally, policy support can play a critical role in incentivizing energy-saving practices. The government can offer subsidies or tax incentives to encourage enterprises to apply low-energy-consumption technologies and practices, facilitating the transition to more energy-efficient production methods and promoting long-term sustainability.

Our study also has some limitations. We analyzed the consumption of three common energy sources in the industrial chain, but we were not able to obtain data on the application of clean energy sources such as solar and wind in the industrial sector to complement this paper. Additionally, our research focuses on a limited set of driving forces. Future studies could introduce more diverse factors and internal structural factors to explore their impacts on energy consumption, offering a more comprehensive foundation for the structural transformation of energy consumption and policy formulation.