The Impact of Ownership Structure on Technological Innovation and Energy Intensity: Evidence from China

Abstract

Environmental pollution and climate warming have become global issues affecting human life, and the burning of fossil fuels is a major source of greenhouse gases. Ownership structure is related to energy efficiency and a change in ownership structure has a significant potential for energy saving. However, few papers have studied the impact of ownership structure on energy intensity from the perspective of technological innovation in the past. Based on panel data from 29 Chinese provinces from 2005 to 2020, we systematically investigate the impact of industrial department ownership structure on energy intensity and study the function of technological innovation in this relationship from the perspective of ownership heterogeneity by using empirical models including ordinary least squares, two-way fixed effects and random effects. The empirical results of this study reveal three findings. First, as the proportion of state-owned industrial enterprises increases by one unit, energy intensity increases by 0.803 units. However, as the proportion of Hong Kong, Macao and Taiwan-invested industrial enterprises, private industrial enterprises and foreign-invested industrial enterprises increases by one unit, energy intensity decreases by 0.847 units and 0.549 units. Second, R&D activities, FDI, capital intensity and exports can significantly reduce energy intensity, but imports can increase energy intensity. Third, the ownership structure can affect energy intensity by influencing R&D expenditure. The increase in the proportion of state-owned industrial enterprises can reduce R&D expenditure, but results in the opposite situation in private firms. Foreign-invested enterprises can reduce energy intensity by making more use of the parent company’s technology. Based on the above empirical results, we propose suggestions to reduce energy intensity, which can provide reference for government to formulate more effective energy policies and realize sustainable development.

1. Introduction

Although energy is a powerful driver of economic development, excessive energy consumption can lead to a large amount of CO2 emissions and environmental pollution problems and thereby limit the sustainable development of society [1,2,3]. With global climate change and energy depletion, how to improve energy efficiency is a key question that we need to think about. Renewable energy will become an important source of electricity generation in the future, and increasing investment in renewable energy is an effective way to mitigate global warming [4,5,6]. The development of renewable energy sources in the future will require strong government support as well as technological breakthroughs, such as biofuels [7]. Only when R&D expenditure is above a certain level can it be beneficial to the development of the renewable energy industry, thereby reducing the use of non-renewable energy [8]. Past studies suggest that innovation in technology may depend on the ownership structure. However, the impact of ownership structure on technological innovation and energy intensity has rarely been studied in the past, and previous studies of ownership have tended to focus on topics other than technological progress, such as business performance, corporate governance [9,10], state and market institutional logics [11]. This is a surprising research gap.

The economic system and policy objectives of the state may have some influence on the relationship between ownership structure and energy efficiency. Liljeblom et al. (2020) find that state control has a negative effect on performance in Russia, but that firms with state ownership will employ more people than privately controlled firms [9]. Steffen et al. [12] find that state-owned utilities have a higher propensity to invest in renewable energy than private firms, particularly in countries with strict climate policies in the European Union (EU). The economic system of China is different from that of the EU and Russia. Both the Chinese and Russian markets are characterized by high levels of ownership concentration and state ownership plays an important role in economic development. However, the Russian economy differs from China’s. The logic of the state is dominant in China, and we can describe the Chinese economy as a socialist market economy with Chinese characteristics. The proportion of state-owned shares in China’s industrial enterprises is relatively high, and the decision-making of state-owned enterprises usually involves the will of the government, which not only pays attention to profit maximization, but also pays attention to social responsibility [13]. In addition, through the analysis of the data, we find that the proportion of assets and the size of Chinese state-owned enterprises is significantly higher than that of private enterprises, but their per capita sales and per capita profits are significantly lower than those of private firms. This suggests that further research is needed on the effect of ownership structure in China’s industrial sector on energy intensity.

There are some differences in the results of the current empirical analysis in the comparison between state ownership and foreign ownership. On the one hand, foreign-invested enterprises have a lower energy intensity than their local counterparts [14]. Rokhmawati [15] argues that foreign ownership not only reduces greenhouse gas emissions, but also improves the competitiveness of firms. On the other hand, it was also found that the increase in foreign-invested enterprises may generate higher pollution emissions. Herrerias et al. [16] believe that both foreign investment and non-state-owned investment play a leading role in the decline of energy intensity in various regions of China, but there is no evidence that state-owned capital has made a positive contribution. Earnhart and Lizal [17] analyze the effect of ownership structure on the environmental performance of firms in the Czech Republic from 1993 to 1998 and conclude that state-owned enterprises had better environmental performance and note that a centralized form of ownership could significantly improve environmental performance. Reference [18] argues that strict environmental regulations have some extent hindered the development of FDI firms in China. Using panel data for 170 countries from 1990 to 2018, Muhammad and Khan [19] found that FDI increases CO2 emissions.

In addition, China is the largest energy consumer and carbon emitter. In 2020, China’s total energy consumption and carbon dioxide emissions exceed those of the United States. In terms of energy intensity, in 2020, the energy intensity (ratio of total energy consumption to regional GDP) falls to 0.7253 tons of standard coal/10,000 RMB, which is still higher than that of developed countries despite the overall downward trend. The 14th Five-Year Plan of China (2021–2025) has been put forward in China. The reform of state-owned enterprise mixed system and high-quality development of private economy will bring about great changes in China’s ownership structure. Figure 1 illustrates that China’s energy intensity shows a downward trend from 2005 to 2020.

Therefore, in order to explore the impact of ownership structure on energy intensity in developing countries, and the function of technological innovation between the two, our study takes the Chinese industrial market as a typical sample. China’s industrial sector is characterized by the co-existence of private, state-owned, Hong Kong, Macao and Taiwan-invested and foreign-invested enterprises. Therefore, our study extends the theory to the different types of ownership structure, and divides ownership ones into foreign ownership, Hong Kong, Macao and Taiwan ownership (foreign ownership for short), state ownership and private ownership to explore the impact of different ownership economic components on energy intensity, respectively. Compared to previous studies, this work is novel in that it uses a rich provincial dataset, considers the importance of time and regional factors in the analysis of energy consumption in China, and allows for a high level of firm heterogeneity to be taken into account. The results reveal that industrial firms of different ownership have different impacts on energy intensity and technological innovation, which has certain theoretical and applied value. The impact of R&D, FDI, capital intensity and international trade on energy intensity is also explored on this basis. Finally, we propose effective energy policy suggestions to reduce industrial energy intensity, help China and other countries achieve sustainable development goals and alleviate the contradiction between huge industrial energy consumption and sustainable development.

2. Literature Review

As the extensive literature shows, energy consumption is influenced by a variety of macroeconomic factors. Economic growth is one of the most important factors in energy consumption [1,3,19]. Although the development of the Internet increases total energy consumption, it can reduce energy intensity [20]. Energy intensity is a measure of the energy efficiency of a region and technological progress is one of the most important factors in affecting energy intensity [21,22]. Financial development can promote technological progress, which in turn can influence energy consumption [23]. In addition, foreign direct investment [24,25], capital intensity [26,27,28,29], international trade activities [30,31,32,33] and other influences such as firm size and artificial intelligence [34] are also important factors in changing energy intensity. For example, Luan et al. [35] argue that research and development (R&D) plays a significant controlling role in reducing energy intensity and increasing the entry of foreign firms can reduce energy intensity.

2.1. R&D Activities (R&D)

R&D plays a significant role in determining the decrease in energy intensity. R&D can help to achieve clean energy technology development and plays two key roles in reducing energy consumption without affecting economic development. R&D can increase the rate of production without increasing energy demand [36]. Because higher R&D inputs mean the invention of better equipment and machines that increase production efficiency with minimal energy consumption, R&D investments lead to a decrease in energy intensity [36]. For example, the application of industrial robots significantly reduces industrial energy intensity by increasing output and reducing energy consumption, and this effect is mainly achieved through technological progress [34]. A 1% increase in R&D capital stock can lead to a decrease in energy intensity of about 0.07% [37]. From the overview of these studies, it is easy to note that energy intensity is heavily influenced by technological progress. More investment in R&D can decrease energy intensity by promoting technological innovation. Differences in technological innovation and productivity can be caused by differences in ownership structure. To motivate our paper, we place special emphasis on the role of technological innovation in this influence. Our paper uses R&D funding to measure technological innovation.

2.2. Foreign Direct Investment (FDI)

FDI flows to developing countries can significantly reduce energy intensity. This is because foreign-invested enterprises possess a large number of advanced environmental protection technologies and more efficient management experience, and therefore have higher productivity and energy efficiency compared to the local counterparts [38]. This effect is attributed to technology diffusion, and the higher the level of local R&D investment, the higher the capacity of local firms to absorb energy-efficient technologies from FDI [24]. Some researchers argue that in low-income developing countries with lax environmental regulations, foreign firms will adopt more environmentally friendly and cleaner technologies than local firms [25,39]. Foreign investment contributes to economic growth in developing countries accompanied by clean technology transfer and can help developing countries achieve their sustainable development goals [40]. We measure FDI by the actual amount of foreign investment.

2.3. Capital Density (Capital)

The role of capital in economic and productivity growth has been widely discussed in the existing literature. Several studies show that capital in the economy is an important factor in the production function and the increase in the capital stock is beneficial to economic growth [26]. Intangible capital plays an increasing role in reducing the energy intensity of the sector compared to tangible capital, because as it is used more and more, intangible capital allows more units of industrial added value to be generated with constant energy input by influencing productivity, which in turn leads to a reduction in energy intensity [27].

Economic growth can also require large amounts of energy consumption, which can result in CO2 emissions. However, Kennedy [28] argues that when obsolete assets are replaced by new investments, the construction and use of new assets to produce goods, although they consume energy, they do not require an increase in total energy consumption if they are more efficient than the old end-of-life assets. Zubair et al. [29] studied the relationship between gross fixed capital formation and reduction in carbon emissions in Nigeria and found that for every 5% increase in fixed capital, Nigeria’s carbon emissions decreased by 1.52%. This paper uses the ratio of total industrial assets to the average number of workers employed to measure Capital density.

2.4. International Trade Activities (Export and Import)

There is evidence of a causal relationship between trade (exports or imports) and energy consumption. An increase in the scale of export trade contributes to greater economies of scale, reduces idle workers, conserves energy use and improves energy efficiency [30]. The expansion of exports increases economic activity in export-oriented and related sectors, which in turn increases energy demand [31]. Chen et al. [32] argue that international trade contributes to energy efficiency because it is an important channel for knowledge spillovers. Technological innovation and the expansion of export trade are conducive to energy efficiency because technological innovation facilitates the development of clean energy technologies, which play an important role in reducing energy consumption [30]. Peters et al. [33] find that exporting firms, especially high-productivity firms, are more likely than non-exporting firms to invest in innovative activities such as R&D, patents and the introduction of new products and production processes. They and Huang, 2021 [41] further explain that the reason why exports affect the energy efficiency of enterprises is that exports stimulate enterprises to expand production, and enterprises are more likely to increase investment in innovation, so as to improve energy efficiency.

In terms of imports, transporting imported goods consumes energy, and if the imported goods are energy-intensive, such as cars, machinery or equipment, then an increase in imports can lead to an increase in energy consumption [42]. Therefore, international trade affects energy intensity. This paper uses the value of imports and exports as a measure in USD billion.

As shown in Figure 2, we propose the research framework. First, the literature review is conducted to analyze the research gaps and introduce the research background. Second, we collect data and set the model. Then, the empirical analysis is carried out. Finally, we put forward countermeasures and suggestions.

3. Methodology and Data

3.1. Empirical Model and Variables

Referring to existing international studies and considering the availability of data, we believe that the main factors affecting energy intensity are technological progress (such as R&D, number of patent applications, etc.), FDI and international trade, etc. Therefore, this paper uses R&D expenditure [36], FDI [24], Capital density [27], export [31] and import [42] as control variables, denoted by LNR&D, LNFDI, LNCapital, LNExport and LNImport in Equation (1). The panel data from 2005 to 2020 is constructed and includes 29 provinces in China. These data can be obtained from the China Industrial Statistical Yearbook, the China Energy Statistical Yearbook and the China Statistical Yearbook. Our regression model form is the following Equation (1).

$$\begin{gathered} {\mathrm{EI}}_{\mathrm{it}}={\mathsf{\alpha }}_0{\mathrm{a}}_0+{\mathrm{a}}_1{\mathrm{OWN}}_{\mathrm{it}}+{\mathsf{\alpha }}_2{\mathrm{a}}_2\mathrm{LNR}\&{\mathrm{D}}_{\mathrm{it}}+\sum {\mathsf{\alpha }}_3{\mathrm{LNFDI}}_{\mathrm{it}}+{\mathsf{\alpha }}_4{\mathrm{LNCapital}}_{\mathrm{it}}+ \\ {\mathsf{\alpha }}_5{\mathrm{LNExport}}_{\mathrm{it}}+{\mathsf{\alpha }}_6{\mathrm{LNImport}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}}+{\mathsf{\delta }}_t+{\mathsf{\mu }}_i \end{gathered}$$

(1)

where t represents time, i represents province, EI_it represents the regional energy intensity of province i in year t; OWN_it is the ownership structure of province i in year t. Ownership structure mainly includes state ownership, private ownership and foreign ownership, which are denoted by StateO, PrivaO and ForeiO, respectively. R&D_it is R&D expenditure, which is used to measure technological innovation and is a control variable. LNFDI_it, LNCapital_it, LNExport_it and LNImport represent other control variables selected in this paper. $\mathsf{\epsilon }$_it is a random disturbance term. The data of the control variable are processed by taking logarithms. $\mathsf{\mu }$_i denotes individual fixed effects for region i that do not vary with time; $\mathsf{\delta }$_t denotes control time fixed effects. Hong Kong, Macao, Taiwan, Xinjiang and Tibet are removed because the data for these provinces are not counted. The sample size of our study is 464.

In general, the commonly used estimation methods for panel data are OLS models, FE models and RE models. To avoid the problem of heteroscedasticity, robust standard errors are added to all methods. OLS, two-way FE and RE methods are used for baseline regression and the robustness of our findings is further verified using stepwise regression.

3.2. Measurement of Energy Intensity and Ownership Structure

Ownership structure is a reflection of the share, position, function and interrelationship of various ownership components in the overall national economy. Because state-owned enterprises and other non-state-owned enterprises have different levels of complexity and limitations in environmental pollution management, they have different levels of impact on energy intensity. In this paper, the ratio of the main business income of state-owned industrial enterprises to the main business income of industrial enterprises is chosen to measure the state ownership of the industry. Similarly, private ownership and foreign ownership are measured in the same way. The higher the ratio, the more widely distributed it is in the industry. State-owned industrial enterprises (SOEs) cover the original state-owned and state-holding industrial enterprises.

Energy intensity, usually measured by energy consumption per unit of GDP, is a key indicator of an economy’s energy efficiency. Low energy intensity means that a unit of energy translates into more GDP.

3.3. Descriptive Statistics

As mentioned earlier, the purpose of this study is to investigate the relationship between ownership structure of the industrial sector and energy intensity. Industrial sector ownership structure is the explanatory variable and energy intensity is the explanatory variable. The explanatory variables are first treated to abate the effect of price changes on regional GDP using the consumer price index, with data obtained from the National Bureau of Statistics of China.

Table 1. Description of variables.

Variable	Explanation (Unit)	Variable	Explanation (Unit)
EI	Energy intensity (tce/104 Yuan RMB)	StateO	State ownership. Ratio of main business income of state-owned industrial enterprises to that of industrial enterprises (%)
LNR&D	R&D expenditure in logarithms (108 Yuan RMB)	PrivaO	Private ownership. Ratio of the main business income of private industrial enterprises to the main business income of industrial enterprises (%)
LNFDI	Foreign direct investment. Total investment in foreign-invested enterprises (10 billion USD)	ForeiO	Foreign ownership. Ratio of the main business income of Hong Kong, Macao and Taiwan and foreign-invested industrial enterprises to the main business income of industrial enterprises (%)
LNCapital	Capital density in logarithms (104 Yuan RMB/per worker)	X1	Ratio of the assets of state-owned industrial enterprises to the assets of industrial enterprises (%)
LNExport	Exports in logarithms (100 million USD)	X2	Ratio of the assets of private industrial enterprises to the assets of industrial enterprises (%)
LNImport	Imports in logarithms (100 million USD)	X3	Ratio of the assets of Hong Kong, Macao and Taiwan and foreign-invested industrial enterprises to the assets of industrial enterprises (%)

Notes: State-owned enterprises (SOEs) cover the original state-owned and state-holding industrial enterprises.

provides an explanation of the variables and

Table 2. Descriptive statistics of variables (2005–2020).

Variable	Observes	Mean of Variables	Standard Deviation	Minimum Value	Maximum Value
EI	464	1.1782	0.6664	0.2635	4.1841
StateO	464	0.3823	0.1816	0.0959	0.8359
PrivaO	464	0.2571	0.1275	0.0301	0.6357
ForeiO	464	0.1982	0.1585	0.0112	0.6564
X1	464	0.4992	0.1752	0.1400	0.8396
X2	464	0.1746	0.0918	0.0261	0.4423
X3	464	0.1768	0.1318	0.0217	0.5911
LNR&D	464	4.6789	1.5004	−0.1315	7.8240
LNFDI	464	6.9374	2.9379	2.0109	19.4305
LNCapital	464	4.5681	0.6526	3.1595	6.4981
LNExport	464	5.0665	1.8648	−7.9020	8.9308
LNImport	464	5.0384	1.8253	−8.3349	8.6115

presents descriptive statistics for the variables studied.

From the data obtained, the energy intensity of each province shows a clear downward trend from 2005 to 2020, but there are significant differences in the level of energy intensity among provinces within the same year. As of 2020, Ningxia has the highest energy intensity at 3.1234 tons of standard coal/RMB 10,000. Beijing has the lowest energy intensity at 0.2635 tons of standard coal/RMB 10,000. In terms of ownership structure, the proportion of state-owned enterprises in each province shows a clear downward trend, while the proportion of private enterprises and foreign-funded enterprises gradually increases. In 2020, the proportion of state-owned industrial enterprises in Qinghai Province is the highest, reaching 0.7282 and the lowest value is 0.1485 in Zhejiang Province.

4. Empirical Results

4.1. Baseline Results

First, descriptive statistics of variables is conducted and empirical methods such as ordinary least squares (OLS), fixed-effects regression (FE) and random-effects regression (RE) are used to discuss the correlation between ownership structure and energy intensity. At the same time, time effects and regional effects are controlled.

As shown in

Table 3. Baseline estimation.

EI	OLS	RE		FE
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
StateO	0.987 *	0.815 **	0.802 **	0.770 ***	0.634 ***	0.549 ***	0.790 ***	0.790 ***	0.802 ***
	(0.523)	(0.338)	(0.339)	(0.175)	(0.179)	(0.170)	(0.176)	(0.176)	(0.168)
LNR&D	−0.193 *	−0.139 **	−0.100		−0.120 ***	−0.121 ***	−0.139 ***	−0.135 ***	−0.100 **
	(0.109)	(0.0592)	(0.0745)		(0.0385)	(0.0354)	(0.0379)	(0.0384)	(0.0388)
LNFDI	−0.0194 **	0.00778 *	−0.0818			−0.102 **	−0.0926 **	−0.0906 **	−0.0818 **
	(0.00723)	(0.00470)	(0.0805)			(0.0408)	(0.0371)	(0.0377)	(0.0353)
LNCapital	0.0793	−0.210 ***	−0.487 ***				−0.457 ***	−0.464 ***	−0.487 ***
	(0.103)	(0.0755)	(0.168)				(0.133)	(0.133)	(0.130)
LNExport	0.376 **	−0.0541	−0.115					−0.00612	−0.115 ***
	(0.149)	(0.0656)	(0.0700)					(0.00620)	(0.0378)
LNImport	−0.424 ***	0.0451	0.110						0.110 ***
	(0.141)	(0.0680)	(0.0727)						(0.0379)
Constant	1.710 ***	2.469 ***	3.478 ***	0.941 ***	1.465 ***	2.039 ***	3.518 ***	3.536 ***	3.478 ***
	(0.421)	(0.272)	(0.697)	(0.0940)	(0.196)	(0.292)	(0.528)	(0.524)	(0.509)
Year fixed effects	NO	NO	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region fixed effects	NO	NO	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	464	464	464	464	464	464	464	464	464
R2	0.602			0.948	0.950	0.952	0.955	0.955	0.956

Notes: *, ** and *** indicate significant at the 10%, 5% and 1% levels, respectively, with robust standard errors in parentheses.

, the FE regressions (models (4) to (9)) are the primary results. In general, ownership structure has a significant impact on industrial energy intensity. Reducing the proportion of state-owned enterprises and increasing the proportion of foreign-owned enterprises is conducive to reducing energy intensity.

As can be seen from Table 3, with the addition of control variables one by one, the effect of industrial ownership structure on energy intensity is significantly positive at the 1% level and the magnitude and significance levels of the estimated coefficients in models (1) to (8) are similar to those in model (9), indicating that the results are not sensitive to the estimation method. The coefficients of ownership structure in models (1), (2) and (3) in Table 3 are significantly positive. In model (1), a one-unit increase in the proportion of state-owned enterprises can increase energy intensity by about 0.987 units. In model (2), a one-unit increase in the proportion of state-owned enterprises is associated with an increase in energy intensity of 0.815 units. As shown in model (3) in Table 3, after adding Region fixed effects and time-fixed effects into the RE model, the coefficient of ownership structure decreases from 0.815 to 0.802. The positive coefficient of the variable of ownership structure in the results of (9) in Table 3 indicates that for every one-unit increase in the share of state-owned enterprises, the energy intensity increases by about 0.802 units in the FE model.

With regard to ownership structure, the coefficient of the State ownership is positive and significant, implying that an increase in the proportion of state-owned enterprises in the industrial sector significantly increases energy intensity. There are two possible explanations for this. First, politically connected state-owned enterprises have more political rent-seeking behaviors, which leads to the lack of competition, distorted resource allocation among enterprises and excessive consumption of primary energy by State-owned enterprises [43,44]. In addition, energy efficiency subsidies have a stronger incentive effect on non-State-owned enterprises than state-owned enterprises facing soft budget constraints. Second, due to China’s special economic system, State-owned enterprises mainly control industrial sectors that are closely linked to the lifeblood of the national economy, such as the production and supply of electric power and heat power, which are mostly energy-intensive.

Table 4. Fifteen high-energy-consuming industrial subsectors in China in 2020 and the share of state-owned enterprises.

Industrial Sector	Energy Consumption (10,000 tce)	Ratio of Assets of State-Owned Industrial Enterprises to Assets of Industrial Enterprises (%)	Ratio of Main Business Income of State-Owned Industrial Enterprises to that of Industrial Enterprises (%)
Smelting and Pressing of Ferrous Metals	66,851	44.25	32.16
Manufacture of Raw Chemical Materials and Chemical Products	56,723	29.81	20.58
Manufacture of Non-metallic Mineral Products	35,387	20.12	12.54
Processing of Petroleum, Coal and Other Fuels	35,267	47.69	53.47
Production and Supply of Electric Power and Heat Power	32,076	84.01	88.52
Smelting and Pressing of Non-ferrous Metals	25,460	37.29	34.39
Mining and Washing of Coal	8953	77.17	61.1
Manufacture of Metal Products	6385	14.23	9.3
Manufacture of Computers, Communication and Other Electronic Equipment	5120	16.01	8.91
Manufacture of Automobiles	4078	41.12	40.41
Manufacture of General Purpose Machinery	3982	18.91	12.48
Extraction of Petroleum and Natural Gas	3743	87.69	82.41
Production and Supply of Water	1923	81.46	71.2
Mining and Processing of Ferrous Metal Ores	1728	71.73	39.6
Production and Supply of Gas	1525	50.86	50.57
Total	289,201

Notes: State-owned enterprises (SOEs) cover the original state-owned and state-holding industrial enterprises.

shows the proportion of state-owned enterprises in the 15 key energy consuming sectors.

As shown in Table 4, these 15 industrial sectors are mostly high-energy-consuming and high-polluting industries sectors, with energy consumption accounting for 86.95% of total industrial energy consumption. No matter from the assets or operating income proportion, the 15 industrial sectors have a relatively high proportion of state-owned industrial enterprises. In addition, there are differences in the economic efficiency of industrial enterprises under forms of different ownership, and the reform of the mixed ownership system of Chinese state-owned industrial enterprises still needs to be deepened. The proportion of assets of state-owned industrial enterprises to total industrial assets is greater than the proportion of operating income of state-owned enterprises to total industrial operating income. This also indicates that state-owned industrial enterprises are less capable of generating income than non-state-owned enterprises despite their larger assets. Therefore, the expansion of SOEs will probably lead to an increase in energy intensity.

As can be seen from the results of (9) in Table 3, A 1% increase in the level of R&D is associated with a 0.001-unit reduction in energy intensity, proving once again that technological progress is conducive to reducing energy intensity. R&D can be an important tool for improving environmental performance, particularly in terms of reducing the energy intensity and carbon intensity of firms. This is because enterprises with higher innovation capabilities tend to adopt more advanced production technologies and efficient production processes, especially R&D activities and patented research and development, which enable firms to achieve higher stages of energy efficiency [45].

For every 1% increase in the level of capital intensity, energy intensity decreases by 0.00487 units. On the one hand, an increase in capital intensity can lead to an increase in productivity, allowing more units of industry value added to be generated with constant energy inputs, which in turn results in a decrease in energy intensity. On the other hand, intangible assets play an increasingly important role in reducing energy intensity [27]. Total energy consumption does not increase when new assets are built and used to produce goods if they are more efficient [27].

Technology spillovers from FDI and exports have a significant dampening effect on the rise in energy intensity. FDI has a significant negative impact on energy intensity and a 1% increase in FDI is associated with a 0.000818-unit decrease in energy intensity. This supports to some extent the results of previous studies [46,47]. FDI is conducive to bringing advanced management experience and optimizing the industrial structure, thus improving the efficiency of technological innovation and reducing energy intensity. This means that the lower the share of state-owned enterprises and the higher the share of private enterprises and foreign-invested enterprises, the more conducive it is to reducing energy intensity.

Exports also have a significant negative impact on energy intensity. In theory, exporters are more willing to adopt energy-efficient technologies in order to maximize their profits and improve their competitiveness on the international market. Moreover, increasing environmental awareness and tightening green trade barriers may force firms to improve their technological innovation capabilities, thus reducing energy intensity. This is the same conclusion that has been drawn from existing studies, namely that international trade affects a country’s energy intensity. Trade openness is closely related to energy efficiency, and exports can improve energy efficiency, because as firms export, especially to developed countries, they can learn advanced technologies [48]. In addition, larger exporters are likely to be more productive and energy efficient due to their larger size, and the interaction between energy efficiency and exports is more significant; likewise, exports are also positively correlated with innovation, which is beneficial to energy efficiency [49].

In contrast, the effect of imports on energy intensity is significantly positive. This is because transporting imported goods requires energy consumption, especially the import of energy-intensive products such as cars, machinery or equipment, and the use of imported goods brings about an increase in energy consumption [42]. At the same time energy is an important driver of economic development and China needs to import large amounts of primary energy each year, which increases China’s energy consumption. For example, China’s energy imports amount to 124.85 million tons of standard coal in 2020, accounting for 24.59% of total energy consumption.

4.2. Endogeneity and Robust Stability Test

4.2.1. Further Tests

Based on the heterogeneous characteristics of enterprise ownership, we use private ownership and foreign ownership as explanatory variables to explore the impact of ownership structure on energy intensity. Private ownership refers to the share of operating income of private industrial enterprises in the total operating income of the industrial sector, whereas foreign ownership refers to the share of operating income of industrial enterprises invested by Hong Kong, Macao and Taiwan and foreign-invested industrial enterprises in the total operating income of the industrial sector.

The results shown in

Table 5. (a) Further tests. (b) Further tests.

(a)
EI	FE				RE
	(1)		(2)		(3)		(4)
PrivaO	−0.847 ***				−0.798 ***
	(0.201)				(0.164)
ForeiO			−0.549 **				−0.875 ***
			(0.221)				(0.185)
LNR&D	−0.105 ***		−0.163 ***		−0.132 ***		−0.233 ***
	(0.0374)		(0.0396)		(0.0298)		(0.0259)
LNFDI	−0.0869 **		−0.0901 **		0.0140 ***		0.0121 ***
	(0.0369)		(0.0393)		(0.00332)		(0.00330)
LNCapital	−0.476 ***		−0.315 **		−0.237 ***		−0.188 ***
	(0.136)		(0.130)		(0.0381)		(0.0360)
LNExport	−0.104 ***		−0.112 ***		−0.0668 **		−0.0702 **
	(0.0355)		(0.0371)		(0.0289)		(0.0292)
LNImport	0.0968 ***		0.112 ***		0.0552 **		0.0669 **
	(0.0360)		(0.0370)		(0.0275)		(0.0279)
Constant	4.057 ***		3.604 ***		3.048 ***		3.236 ***
	(0.549)		(0.523)		(0.132)		(0.127)
Year fixed effects	Yes		Yes		NO		NO
Region fixed effects	Yes		Yes		NO		NO
Observations	464		464		464		464
R2	0.956		0.954
(b)
EI	FE				RE
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
StateO	0.363 *	0.598 ***		0.745 ***	0.313 *	0.674 ***		0.724 ***
	(0.199)	(0.194)		(0.165)	(0.173)	(0.160)		(0.133)
PrivaO	−0.775 ***	−0.489 **	−1.034 ***		−0.780 ***	−0.327 *	−1.031 ***
	(0.260)	(0.245)	(0.208)		(0.213)	(0.196)	(0.162)
ForeiO	−0.647 ***		−0.831 ***	−0.315	−0.991 ***		−1.150 ***	−0.664 ***
	(0.183)		(0.185)	(0.215)	(0.203)		(0.184)	(0.184)
LNR&D	−0.108 ***	−0.0881 **	−0.122 ***	−0.114 ***	−0.137 ***	−0.104 ***	−0.153 ***	−0.152 ***
	(0.0399)	(0.0389)	(0.0386)	(0.0414)	(0.0295)	(0.0300)	(0.0282)	(0.0296)
LNFDI	−0.0641 **	−0.078 **	−0.0645 **	−0.0761 **	0.0143 ***	0.0104 ***	0.0162 ***	0.0099 ***
	(0.0318)	(0.0341)	(0.0321)	(0.0338)	(0.00338)	(0.00337)	(0.00322)	(0.00322)
LNCapital	−0.490 ***	−0.526 ***	−0.455 ***	−0.459 ***	−0.244 ***	−0.252 ***	−0.236 ***	−0.231 ***
	(0.135)	(0.135)	(0.135)	(0.133)	(0.0360)	(0.0376)	(0.0359)	(0.0362)
LNExport	−0.108 ***	−0.110 ***	−0.105 ***	−0.115 ***	−0.0685 **	−0.0680 **	−0.0681 **	−0.0697 **
	(0.036)	(0.0363)	(0.0355)	(0.0379)	(0.0278)	(0.0284)	(0.0279)	(0.0283)
LNImport	0.105 ***	0.104 ***	0.102 ***	0.113 ***	0.0672 **	0.0598 **	0.0667 **	0.0679 **
	(0.036)	(0.0365)	(0.0357)	(0.0375)	(0.0265)	(0.0270)	(0.0266)	(0.0270)
Constant	3.893 ***	3.758 ***	4.080 ***	3.465 ***	3.119 ***	2.615 ***	3.364 ***	2.741 ***
	(0.562)	(0.552)	(0.555)	(0.511)	(0.187)	(0.163)	(0.129)	(0.155)
Year fixed effects	Yes	Yes	Yes	Yes	NO	NO	NO	NO
Region fixed effects	Yes	Yes	Yes	Yes	NO	NO	NO	NO
Observations	464	464	464	464	464	464	464	464
R2	0.958	0.957	0.957	0.957

Notes: *, ** and *** indicate significant at the 10%, 5% and 1% levels, respectively, with robust standard errors in parentheses.

a are consistent with expectations. The increase in the number of Hong Kong, Macau and Taiwan-invested enterprises and foreign-invested enterprises has a positive and significant effect in reducing energy intensity. The increase in the share of private enterprises dampens the rise in energy intensity, and private enterprises play a greater role in reducing energy intensity than foreign enterprises. This is due to the fact that after China’s reform and opening up, private enterprises have achieved higher performance and have become more competitive compared with state-owned enterprises and foreign-invested enterprises [50]. The proportion of private enterprises in China is much larger than that of foreign enterprises, and the innovation ability of private enterprises is increasing.

Foreign-invested enterprises usually have more advanced and efficient environmental technologies that can bring positive technological spillover effects to a region and promote regional economic growth, thus having a negative impact on energy intensity. Most domestic private enterprises in China are sole proprietorships. For survival and better performance, competition among private enterprises is usually fierce, and the market elimination mechanism of the winners and losers requires private enterprises to continuously enhance product innovation. In order to maximize revenue, private enterprises must respond to market changes and technological innovations in a timely manner, which determines that private enterprises have a higher awareness of technological innovation and management efficiency. In short, private enterprises have higher output per unit of energy consumption and have lower energy intensity. Therefore, improving the market structure and encouraging foreign firms to enter the Chinese market is an ambitious goal to achieve control of industrial energy intensity.

We further put the three ownership types into Equation (1) for regression together, and group the three variables in pairs for regression to test the impact of ownership structure on energy intensity. The results are shown in Table 5b.

The results show that the proportion of state-owned enterprises, the proportion of private enterprises and the proportion of foreign enterprises all have significant impact on energy intensity. Overall, state ownership contributes to the rise in energy intensity, whereas private ownership and foreign ownership dampened it. The direction of the coefficients of ownership structure is consistent with the baseline regression.

4.2.2. Alternative Ownership Structure Measurement

The use of stepwise regression above ensures that our results are somewhat accurate. To further verify the accuracy of our results, we use an alternative measure of ownership structure. State ownership is measured by the percentage of SOEs assets to total industrial sector assets, denoted by X1. Similarly, private ownership and foreign ownership are measured by the percentage of assets of private industrial enterprises to total assets of the industrial sector, and the proportion of total assets of foreign-invested enterprises and Hong Kong, Macao and Taiwan-invested enterprises to total assets of the industrial sector, denoted by X2 and X3, respectively.

The results are presented in

Table 6. Robust stability test.

EI	(1) FE	(2) RE	(3) FE	(4) RE	(5) FE	(6) RE
X1	0.524 **	0.641 ***
	(0.245)	(0.180)
X2			−0.626 *	−0.606 **
			(0.358)	(0.246)
X3					−1.199 ***	−1.609 ***
					(0.210)	(0.232)
LNR&D	−0.125 ***	−0.156 ***	−0.123 ***	−0.171 ***	−0.172 ***	−0.226 ***
	(0.0373)	(0.0291)	(0.0372)	(0.0293)	(0.0373)	(0.0247)
LNFDI	−0.0933 **	0.0105 ***	−0.0950 **	0.0130 ***	−0.0764 **	0.0121 ***
	(0.0395)	(0.00332)	(0.0401)	(0.00343)	(0.0377)	(0.00321)
LNCapital	−0.409 ***	−0.209 ***	−0.390 ***	−0.206 ***	−0.329 ***	−0.225 ***
	(0.138)	(0.0372)	(0.139)	(0.0378)	(0.126)	(0.0353)
LNExport	−0.107 ***	−0.0622 **	−0.109 ***	−0.0669 **	−0.111 ***	−0.0770 ***
	(0.0369)	(0.0295)	(0.0361)	(0.0295)	(0.0363)	(0.0285)
LNImport	0.102 ***	0.0525 *	0.102 ***	0.0553 **	0.112 ***	0.0780 ***
	(0.0372)	(0.0280)	(0.0366)	(0.0281)	(0.0368)	(0.0272)
Constant	3.424 ***	2.522 ***	3.761 ***	2.996 ***	3.683 ***	3.460 ***
	(0.488)	(0.183)	(0.551)	(0.129)	(0.515)	(0.133)
Year fixed effects	Yes	NO	Yes	NO	Yes	NO
Region fixed effects	Yes	NO	Yes	NO	Yes	NO
Observations	464	464	464	464	464	464
R2	0.9539		0.9537		0.9554

Notes: *, ** and *** indicate significant at the 10%, 5% and 1% levels, respectively, with robust standard errors in parentheses.

. Robustness tests, controlling for area and time effects and using robust standard errors, produce results that again demonstrate a significant relationship between different ownership components and energy intensity. With the reform and opening up of the economy, the assets of China’s private enterprises are constantly expanding. Unlike state-owned enterprises, a rise in the share of private enterprise assets can reduce energy intensity, but state-owned enterprises can bear more of the policy burden, such as through hiring more workers [9,50]. A one-unit increase in the proportion of state-owned enterprise assets leads to a 0.524-unit increase in energy intensity. A one-unit increase in the share of assets of private enterprises and assets of foreign-invested enterprises inhibits energy intensity from rising by 0.626 and 1.199 units. The results in Table 6 are still robust and stable when they are tested by another ownership structure index.

4.2.3. Lagged One Period of the Explanatory Variables

Our paper conducts regression by all lagged explanatory variables by one period, and the results are shown in

Table 7. Results of variables lagged by one period.

EI	(1) FE	(2) RE	(3) FE	(4) RE	(5) FE	(6) RE
StateO	0.653 ***	0.803 ***
	(0.169)	(0.136)
PrivaO			−0.705 ***	−0.799 ***
			(0.214)	(0.169)
ForeiO					−0.360	−0.638 ***
					(0.242)	(0.203)
LNR&D	−0.102 **	−0.133 ***	−0.115 ***	−0.153 ***	−0.154 ***	−0.234 ***
	(0.0395)	(0.0284)	(0.0379)	(0.0285)	(0.0404)	(0.0260)
LNFDI	−0.116 **	−0.0681 ***	−0.125 ***	−0.0747 ***	−0.134 ***	−0.0661 ***
	(0.0459)	(0.0240)	(0.0457)	(0.0245)	(0.0463)	(0.0253)
LNCapital	−0.253 *	−0.0745 *	−0.244	−0.0471	−0.122	−0.0280
	(0.147)	(0.0424)	(0.154)	(0.0428)	(0.145)	(0.0412)
LNExport	−0.0874 **	−0.0229	−0.0797 **	−0.0210	−0.0836 **	−0.0269
	(0.0373)	(0.0290)	(0.0348)	(0.0293)	(0.0353)	(0.0299)
LNImport	0.0868 **	0.0184	0.0770 **	0.0132	0.0858 **	0.0246
	(0.0370)	(0.0275)	(0.0349)	(0.0278)	(0.0350)	(0.0284)
Constant	2.828 ***	2.243 ***	3.376 ***	2.776 ***	3.014 ***	2.909 ***
	(0.591)	(0.145)	(0.641)	(0.121)	(0.597)	(0.125)
Year fixed effects	Yes	NO	Yes	NO	Yes	NO
Region fixed effects	Yes	NO	Yes	NO	Yes	NO
Observations	435	435	435	435	435	435
R2	0.956		0.956		0.954

Notes: *, ** and *** indicate significant at the 10%, 5% and 1% levels, respectively, with robust standard errors in parentheses.

. As shown in Table 7, the regression results of the explanatory variables lagged by one period are still significant, which indicates that ownership structure has a significant impact on energy intensity after eliminating certain potential endogeneity problems.

Compared with Table 3 and Table 5, the absolute value of the coefficient of ownership structure decreases slightly. The regression coefficients of the proportion of state-owned enterprises are significantly positive at the statistical level of 1% in columns (1) and (2). The regression coefficients of the proportion of private enterprises are significantly negative at the level of 1% in columns (3) and (4). The regression coefficients of the proportion of foreign enterprises are significantly negative at the statistical level of 1% in column (6).

4.2.4. Regression with Reduced Tails

To exclude the effect of extreme values, we use reduced-tail regression for further testing. The results of the reduced-tail regression of the explanatory variables at the 1% level can be seen in

Table A1. Regression with Reduced Tails.

EI	(1) FE	(2) RE	(3) FE	(4) RE	(5) FE	(6) RE
StateO	0.675 ***	0.761 ***
	(0.157)	(0.123)
PrivaO			−0.598 ***	−0.636 ***
			(0.189)	(0.160)
ForeiO					−0.657 ***	−0.944 ***
					(0.182)	(0.176)
LNR&D	−0.133 ***	−0.159 ***	−0.123 ***	−0.165 ***	−0.204 ***	−0.283 ***
	(0.0429)	(0.0299)	(0.0413)	(0.0329)	(0.0412)	(0.0291)
LNFDI	−0.115 ***	0.00638 **	−0.120 ***	0.0107 ***	−0.113 ***	0.00976 ***
	(0.0337)	(0.00302)	(0.0350)	(0.00311)	(0.0352)	(0.00305)
LNCapital	−0.206 **	−0.161 ***	−0.188 *	−0.159 ***	−0.0507	−0.104 ***
	(0.100)	(0.0349)	(0.109)	(0.0366)	(0.0946)	(0.0340)
LNExport	−0.0795 **	−0.0642 **	−0.0850 ***	−0.0754 **	−0.0721 **	−0.0549 *
	(0.0318)	(0.0290)	(0.0315)	(0.0296)	(0.0310)	(0.0295)
LNImport	0.0857 **	0.0512 **	0.0749 **	0.0424	0.0892 **	0.0599 **
	(0.0350)	(0.0258)	(0.0346)	(0.0263)	(0.0345)	(0.0262)
Constant	2.794 ***	2.387 ***	3.232 ***	2.929 ***	2.884 ***	3.071 ***
	(0.484)	(0.143)	(0.520)	(0.126)	(0.466)	(0.121)
Year fixed effects	Yes	NO	Yes	NO	Yes	NO
Region fixed effects	Yes	NO	Yes	NO	Yes	NO
Observations	464	464	464	464	464	464
R2	0.959		0.958		0.958

Notes: *, ** and *** indicate significant at the 10%, 5% and 1% levels, respectively, with robust standard errors in parentheses.

. The results show that the significance of the regression results is generally consistent with the description of the core explanatory variables in the baseline regression, which proves the reliability of the results after avoiding the influence of outliers.

4.3. Mediating Effects

To further investigate the transmission channels through which industry ownership structure affects energy intensity, we adopt the stepwise regression method and use technological progress as a mediating variable to illustrate the causal relationship between the two, where Equation (1) has been tested previously and Equations (2) and (3) are tested next. The results are presented in

Table 8. Mediation effect test.

Dependent Variables	(1) EI	(2) LNR&D	(3) EI	(4) LNR&D	(5) EI	(6) LNR&D
Independent Variables
StateO	0.894 ***	−0.920 ***
	(0.167)	(0.225)
PrivaO			−0.966 ***	1.131 ***
			(0.193)	(0.250)
ForeiO					−0.400 *	−0.918 ***
					(0.207)	(0.301)
LNFDI	−0.0784 **	−0.0336	−0.0836 **	−0.0308	−0.0918 **	0.0109
	(0.0339)	(0.0509)	(0.0354)	(0.0502)	(0.0390)	(0.0451)
LNCapital	−0.476 ***	−0.109	−0.466 ***	−0.0969	−0.279 **	−0.221 *
	(0.131)	(0.117)	(0.136)	(0.118)	(0.133)	(0.117)
LNExport	−0.141 ***	0.267 ***	−0.131 ***	0.253 ***	−0.156 ***	0.267 ***
	(0.0373)	(0.0627)	(0.0344)	(0.0604)	(0.0359)	(0.0578)
LNImport	0.131 ***	−0.214 ***	0.117 ***	−0.197 ***	0.145 ***	−0.207 ***
	(0.0375)	(0.0578)	(0.0354)	(0.0560)	(0.0365)	(0.0542)
Constant	3.023 ***	4.532 ***	3.662 ***	3.777 ***	2.891 ***	4.367 ***
	(0.441)	(0.472)	(0.504)	(0.519)	(0.460)	(0.498)
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Region fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Observations	464	464	464	464	464	464
R2	0.955	0.984	0.955	0.984	0.952	0.984

Notes: *, ** and *** indicate significant at the 10%, 5% and 1% levels, respectively, with robust standard errors in parentheses.

. Next, the mediating effect is argued by using the R&D measure of technological innovation as a mediating variable. Regression model form is the following Equations (2) and (3):

$$\begin{gathered} {\mathrm{EI}}_{\mathrm{it}}={\mathsf{\beta }}_0{a}_0+{\mathsf{\beta }}_1{\mathrm{OWN}}_{\mathrm{it}}+{\mathsf{\beta }}_2{\mathrm{LNFDI}}_{\mathrm{it}}+{\mathsf{\beta }}_3{\mathrm{LNCapital}}_{\mathrm{it}}+{\mathsf{\beta }}_4{\mathrm{LNExport}}_{\mathrm{it}}+ \\ {\mathsf{\beta }}_5{\mathrm{LNImport}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}}+{\mathsf{\delta }}_{t}t+{\mathsf{\mu }}_i \end{gathered}$$

(2)

$$\begin{gathered} \mathrm{LNR}\&{\mathrm{D}}_{\mathrm{it}}={\mathsf{\gamma }}_0{a}_0+{\mathsf{\gamma }}_1{\mathrm{OWN}}_{\mathrm{it}}+{\mathsf{\gamma }}_2{\mathrm{LNFDI}}_{\mathrm{it}}+{\mathsf{\gamma }}_3{\mathrm{LNCapital}}_{\mathrm{it}}+{\mathsf{\gamma }}_4{\mathrm{LNExport}}_{\mathrm{it}}+ \\ {\mathsf{\gamma }}_5{\mathrm{LNImport}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}}+{\mathsf{\delta }}_{t}t+{\mathsf{\mu }}_i \end{gathered}$$

(3)

The results of applying the stepwise regression to Equation (2) and Equation (3) are shown in Table 8. The results verify our hypothesis that an increase in the proportion of state-owned industrial enterprises will reduce R&D spending, which in turn affects energy intensity. In contrast to this, an increase in the share of private firms contributes to an increase in R&D expenditure. Therefore, private ownership reduces energy intensity.

Although the total assets and R&D expenditures of Chinese state-owned enterprises are the largest, the ratio of R&D expenditures to total assets of state-owned enterprises is not more than 1.32% in 2020. The ratio of R&D expenditure to total assets of Hong Kong, Macau, Taiwan and foreign-invested enterprises is 1.207%, whereas the ratio of R&D expenditure to assets of private enterprises is 1.6367%, which is much higher than the first two types of enterprises.

The results of model (1) in Table 8 show that the total magnitude of the effect of state ownership on energy intensity is 0.894, the direct effect of state ownership on energy intensity is 0.802 (coefficient of ownership structure in model (9) in Table 3) and the indirect part of the effect through technological progress is 0.092 (coefficient of ownership structure in model (1) in Table 8 multiplied by the coefficient of the R&D variable in model (9) in Table 3).

It is worth mentioning that the increase in Hong Kong, Macau and Taiwan-invested enterprises and foreign-invested enterprises does not increase R&D expenditure, because foreign-invested enterprises are opening up China’s mainland market by virtue of their technological advantage; they utilize the clean technology of its parent company and therefore the R&D investment is not high.

5. Recommendations and Limitations

With its dominant state-owned economy, unique system of governance and intact industrial system, China provides a unique environment to test the impact of different types of economic composition on energy intensity and technological innovation. More generally, our paper expands the related theories of ownership structure, energy intensity and technological innovation, which has practical significance for energy conservation and CO2 emission reduction. China has entered a new stage of development, with the goal of achieving carbon neutrality. Therefore, the next step in the development of Chinese industry should consider strengthening energy management and improving relevant policy support and exerting a good influence of energy policy on technological innovation and energy intensity. In order to reduce energy intensity, the following countermeasures are proposed in conjunction with the results of the empirical analysis.

First, the government should continue to steadily promote the reform of mixed ownership of enterprises, encourage state-owned enterprises to introduce domestic private capital and encourage state-owned enterprises to absorb the advanced technology and production methods of private enterprises. State-owned enterprises should focus on functions such as serving the people’s livelihood and public services, accelerate the optimization of the industrial layout and structural adjustment of the state-owned economy, give full play to the strategic support role of the state-owned economy, improve the innovation capacity of state-owned enterprises, and increase government subsidies to fully mobilize non-state-owned enterprises, especially private enterprises, to research and develop cutting-edge technologies. Second, the government should encourage exports, especially by increasing policy subsidies to encourage exporters to undertake R&D activities, improve the technological content of export products, encourage enterprises to use more energy efficient equipment and machinery, and increase capital per capita to promote production efficiency. Third, the government should play a guiding role in improving policies to incentivize technological innovation, creating a favorable market environment for innovation and supporting local enterprises in China to establish product and equipment development centers in China in cooperation with multinational companies to facilitate technology spillover.

However, our work also has some limitations that can be addressed in the future. First, there are differences in energy intensity and ownership structure between industrial subsectors, such as high-energy-using industries and low-energy-using industries. Therefore, in future research, we can further break down the industries and analyze the heterogeneity of the impact of ownership on industrial energy intensity across industrial subsectors. Second, this paper uses relevant data at the industrial level to demonstrate the impact of ownership structure on energy intensity. At the enterprise level, the impact of private capital and state-owned capital on the energy efficiency and technological innovation of an enterprise can be examined.

6. Conclusions

Differently from previous studies, we explain the logic of ownership structure affecting energy intensity from the perspective of technological innovation. Our research has uncovered a number of new conclusions that the differences in the innovation behavior of state-owned and non-state-owned enterprises have an impact on China’s energy intensity. First, the share of the state-owned economy in the industrial sector is significantly and positively associated with energy intensity, and the increased share of private enterprises, Hong Kong, Macao and Taiwan-invested enterprises and foreign-invested enterprises can reduce energy intensity which supports the existing studies [16,24]. A one-unit increase in the share of the state-owned economy leads to a 0.802-unit increase in energy intensity, and a one-unit increase in the share of private enterprises leads to an energy intensity decrease of approximately 0.847 units. A one-unit increase in the share of Hong Kong, Macau and Taiwan-invested enterprises and foreign-invested enterprises would result in a corresponding decrease in an energy intensity of 0.875 units. Second, R&D, FDI, capital intensity and exports can significantly reduce energy intensity, but import trade can increase energy intensity. The results further support existing research [29,38,41]. Third, technological innovation depends on the ownership structure, and R&D expenditure is an important manifestation of autonomous innovation in the private enterprises. Specifically, the proportion of state-owned enterprises has a negative impact on R&D expenditure, whereas the proportion of private firms in the industrial market has a positive impact on the increase in R&D expenditure. Foreign-invested enterprises have a negative impact on total R&D expenditure, with low levels of R&D, due to greater use of their parent company’s technology to reduce energy intensity. The empirical results contrast with the results of previous studies on European countries [12].