Resource Economy

Assessing the Impact of Digital Technologies on Energy Efficiency: The Role of OFDI and Virtual Agglomeration

  • SHEN Yang , 1, 2 ,
  • HAN Mengyu 1 ,
  • ZHANG Xiuwu , 1, 2, *
  • 1. Institute of Quantitative Economics, Huaqiao University, Xiamen, Fujian 361021, China
  • 2. School of Statistics, Huaqiao University, Xiamen, Fujian 361021, China
*ZHANG Xiuwu, E-mail:

Received date: 2023-06-15

  Accepted date: 2023-09-02

  Online published: 2023-12-27

Supported by

The Natural Science Foundation of Fujian Province(2022J01320)

The National Science Fund for Distinguished Young Scholars(72103067)


Improving energy efficiency is crucial for achieving the carbon peaking and carbon neutrality goals. The digital economy, which is characterized by big data, artificial intelligence, the internet of things, and a new generation of mobile Internet, has quietly penetrated all aspects of the economy and society, profoundly changing the means of production and lives of human beings. Digital technologies have great potential to improve the global energy system’s security, productivity, efficiency, and sustainability. Based on the panel data of 30 provinces in mainland China from 2006 to 2021, this study divided energy efficiency into total and single factor energy efficiency. The two-way fixed-effect model and the Driscol-Kraay method were used to adjust the standard error test in order to examine the impact of digital technology represented by industrial robots on energy efficiency and its path mechanism. Studies have shown that digital technology can significantly improve total factor energy efficiency and reduce energy intensity per unit of GDP. This conclusion was found to be still valid after the robustness test using feasible generalized least squares, time-varying difference in difference and fixed effect space Durbin model. The results of the mechanism test show that digital technology can improve energy efficiency by increasing the degree of industrial virtual agglomeration and the channels of foreign direct investment. This paper provides a valuable discussion on how information technology advances can improve energy efficiency in the era of the digital economy. The conclusions will help relevant market players to formulate policies and measures and corporate strategies to improve energy efficiency. At the same time, it also deepens the theoretical understanding and mechanism path of digital technology’s impact on energy consumption.

Cite this article

SHEN Yang , HAN Mengyu , ZHANG Xiuwu . Assessing the Impact of Digital Technologies on Energy Efficiency: The Role of OFDI and Virtual Agglomeration[J]. Journal of Resources and Ecology, 2024 , 15(1) : 117 -129 . DOI: 10.5814/j.issn.1674-764x.2024.01.010

1 Introduction

Energy is an indispensable material basis for national development and security, and a necessary driving force for the sustainable development of the national economic system. Since the Industrial Revolution, the widespread use of fossil fuels has caused several global environmental, ecological, and climatic problems, such as the greenhouse gas effect, air pollution, and acid rain. There are negative externalities in human economic behavior, which cause damage to the environment in the pursuit of economic development. According to the National Bureau of Statistics (NBS), China’s GDP will reach about 120 trillion yuan in 2022, an increase of 3% year-on-year, and its economic aggregate will account for about 18% of the global economy. The rapid expansion of economic scale and the rapid advancement of industrialization have led to a sharp increase in energy consumption, and the extensive development mode and low energy efficiency have become major obstacles to eco nomic transformation and upgrading. According to the Statistical Bulletin of National Economic and Social Development 2022 released by the NBS, China’s total energy consumption in 2022 was 5.41 billion t of standard coal, an increase of 2.9% year-on-year, of which coal consumption accounted for 56.2% of the total energy consumption. Clean energy (natural gas, hydropower, nuclear, wind, and solar) accounted for only 25.9%. Although carbon dioxide emissions per 10000 yuan of GDP fell by 0.8%, fossil fuel and electricity consumption are still growing, and the proportion of non-fossil energy consumption is still deficient. According to the “BP Statistical Yearbook of World Energy 2022” released by British Petroleum, China’s total primary energy consumption in 2022 was 159.39 EJ, accounting for 26.39% of the world’s total primary energy consumption. The data in that Yearbook also mentioned that China’s energy carbon emissions are 10.523 billion t, accounting for nearly 30% of the global carbon emissions. Energy enables the stable development of the economy. However, a large amount of energy consumption also causes environmental pollution and the ecological problems of insufficient resource-carrying capacity, which restricts the sustainable development of the economy.
Human society is entering a new round of scientific and technological revolutions represented by the biological sciences, information science, quantum science, nanoscience, energy technology, and artificial intelligence. The multifaceted breakthrough and integration of new technologies have promoted the rise of new industries, business forms, and models, triggering the reconstruction of the modern industrial system and the transformation of social productivity. The organic integration of the new generation of information technology and energy infrastructure to jointly promote the digital transformation of energy is an important measure for improving total factor productivity. The International Energy Agency (IEA) believes that digital transformation will completely disrupt the global energy system, providing a unique opportunity for sustainable energy development. For example, by optimizing the allocation of production factors, digital technology can promote energy optimization, cost optimization, risk prediction, and decision control of traditional industries, and significantly improve the energy efficiency of energy-intensive industries such as transportation, construction, and manufacturing. The Bloomberg New Energy Finance’s (BNEF) “net zero hypothesis” points out that to achieve the goal of keeping the global temperature rise within 2 ℃ set by the Paris Climate Agreement, global solar, wind energy, and battery energy storage will require an investment of USD 15.1 trillion by 2050, and the power grid will need an investment of USD 14 trillion. However, even that level of investment will not meet the needs of this goal. BNEF’s estimates show that despite the record global investment (USD 755 billion, which is a 27% increase from 2020) in low-carbon energy in 2021, more is needed to get the world to net zero emissions by 2050. The wide application of digital technology has injected new momentum into the structural transformation and model innovation of the energy industry, and it is also a vital force in promoting energy security and sustainability. More significant investment in the digital transformation of energy systems is also necessary, with the total investment in energy infrastructure needing to increase by USD 92 trillion to USD 173 trillion by 2050 to ensure that net zero emissions are achieved.
The international community has widely recognized the transformative effect of digital technology on the energy system. For example, The U.S. actively adopts the strategy of “Technology wins”. In 2019, the Advanced Energy Research Project Agency (ARPA-E) of the Department of Energy announced the theme of “Intelligent Design promotes energy conservation and emission reduction and achieves major technological improvements and provided USD 15 million in funding to carry out research projects on the application of digital technologies in the energy field. Energy-related departments will explore the introduction of artificial intelligence and machine learning into energy technology and product design research and development, and maintain the global leadership of the USA in energy and digital technology. The Chinese government has repeatedly emphasized the promotion of digital transformation in the energy sector, seized the historic opportunity of combining the digital technology revolution with the energy revolution, and worked hard to build a clean, low-carbon, safe, and efficient modern energy system. In March 2023, the “Several Opinions on Accelerating the Development of Digital and Intelligent Energy” issued by the National Energy Administration mentioned that it is necessary to promote the actual integration of digital technology into all aspects of energy production, transportation, storage, marketing, and use, and to build a digital and intelligent innovation application system for all aspects of the energy system. This will enable the accelerated transformation of the energy system operation and management mode to comprehensive standardization, profound digitalization, and high intelligence, and drive both an increase in the proportion of new energy in the energy system and the improvement of total factor productivity. Against the background of the digital economy, this is a problem that academic circles need to study deeply in order to promote the improvement of energy efficiency through the integration and interaction of modern information technology progress and the energy industry. The existing literature includes a lot of research on the connotation, measurement methods and influencing factors of energy efficiency (Lin and Du, 2013; Filippini and Hunt, 2016; Lin and Tan, 2016; Ohene-Asare et al., 2020; Chen et al., 2021; Xu et al., 2023). Among them, in the context of the digital economy, the mechanism by which digital technology helps enterprises with digital transformation and industrial intelligence to improve energy efficiency has also been widely studied (Zhao et al., 2021; Huang and Chen, 2023). Digital technologies can improve the productivity and sustainability of energy systems. In the 1970s, the power sector benefited from emerging digital technologies that facilitated grid operations. Cutting-edge digital technologies are reshaping energy end-use patterns (transportation, buildings, industry), changing supply-side business models (coal, oil, gas, electricity), and integrating more excellent market value across various energy boundaries (renewables and grid, residential and utilities).
One important consideration is that there are two direct and indirect channels for the impact of digital technology on energy consumption, that is, there are two possibilities for digital technology to increase energy consumption and improve energy efficiency (Lange et al., 2020). Specifically, there is an increased energy consumption of digital equipment in the process of investment and continuous operation, while the development of digital technology can also eliminate redundant waste in the production process, integrate the production process and information flow of enterprises, and improve the efficiency of energy management. In addition, digital technologies may increase energy dependence among consumers and producers, creating an energy “rebound effect”. The potential innovation points of this study are threefold. First, this study systematically discusses the importance and direct role of digital technology in improving energy efficiency (EE) against the background of a new round of scientific and technological revolution. The second is an evaluation of the impact of digital technology on SFEE and TFEE based on the market behavior of enterprises introducing and installing industrial robots. Third, this research innovatively introduces virtual agglomeration (VA), which represents the digital evolution of the industrial spatial organization form, and re-examines the new path and mechanism of digital technology in improving efficiency from the perspective of new industrial agglomeration. In summary, this study provides a valuable discussion on the role of modern information technology in improving EE in the context of the digital economy. The construction of an intermediate channel has enriched the perspective and content of the energy research field. The conclusions drawn can help enterprises and administrative departments to take relevant actions on time in order to promote the digital transformation of energy data and ultimately promote the smooth realization of the “dual carbon” goal.

2 Theoretical mechanism and research hypothesis

2.1 Direct impact of digital technology on EE

Regarding the production side of energy enterprises, digital technology and intelligent technology help in realizing real-time monitoring of the production process, as well as reducing production costs, energy transportation loss rates, and production failures (Dalla’Ora et al., 2022). For example, artificial intelligence technology can automatically detect and warn of faults, ensure the stability of energy transmission, and prevent safety accidents. Predictive maintenance functions play a crucial role in the energy industry because humans cannot predict every failure, and artificial intelligence technology can effectively identify energy equipment corrosion, cracks, inadequate insulation, and other defects, thus achieving an early warning function. With the deepening of the integration of digital technologies, automatic early warning monitoring and control at the millisecond level will be expected in the future. For example, intelligent solutions help improve supply chain link efficiency, such as enterprise production and operation, and improve the efficiency of energy resource allocation (Fu et al., 2023). Supply chains for specific energy sectors, such as the energy and gas industry, are complex systems that involve sales decisions by oil suppliers/distributors, market prices, refining operations, gantry operations, and product transportation. Digital technology can help managers in the day-to-day production and operation of auxiliary enterprise analysis. Ancillary features include, but are not limited to, management decisions such as optimizing energy selling prices, creating smart warehouses, maintaining inventory, handling transportation operations for replacement assets, risk hedging, and reducing lead times, all of which help managers quickly take relevant actions to reduce overall operating costs and achieve EE goals. From the consumer side, intelligent solutions improve the energy consumption pattern, change the terminal consumption pattern, and save resources (Xue et al., 2022). With the increasing maturity of Internet technology and the growing strength of Internet platforms, workers can choose to work at home, which is conducive to saving energy consumption in office settings and reducing the energy consumption caused by commuting. The wide application of digital technology in public transport, online car booking, and private cars can reduce the empty driving rate of public transportation and online car booking, reduce the waiting time of private cars at traffic lights, and optimize the choices of travel routes through real-time sharing of road information to alleviate traffic jams and reduce unnecessary energy consumption.
Energy enterprises can achieve the real-time collection of production data and the accurate management of production energy consumption, which helps the enterprises to customize energy use solutions based on supply-side demand in order to avoid excessive services and improve personal and household energy utilization. The accelerated development of digital energy and information technology and multi- functional collaborative management platform technologies has gradually broken the barriers between entities in different fields such as coal, oil and gas, electricity, communications, and automobiles, and the information flow between different industries has initially realized greater interconnection. According to the Energy Information Administration (EIA) of USA, nearly half of energy users in USA have smart meters installed in their homes. These meters can provide data about personal energy consumption. The data can predict upcoming energy use levels and help customers better regulate their consumption, such as finding the cheapest time to charge an electric car or run an air conditioner or optimizing energy storage. In summary, this study puts forward the first research hypothesis:
Hypothesis 1 (H1): Digital technology will improve energy efficiency.

2.2 The role of outward foreign direct investment

The home country company absorbs and transforms the cutting-edge foreign technology brought by OFDI, applies it to the production and manufacturing link and ultimately forms a competitive market advantage with spillover technology as the core (Song and Wang, 2019). Companies participating in global trade activities may have a more vital awareness of new technologies and more up-to-date knowledge. They will be motivated to keep up with foreign trading partners in technological innovation. Through various channels, they can absorb the host country’s advanced technology and management experience in energy conservation and emission reduction, and realize the reverse spillover of the host country's technology (Zhong and Moon, 2023). For example, enterprises in the home country embedded in the R&D resource-intensive areas and related industrial clusters of the host country can absorb the advanced environmental protection technologies of the host country using resource sharing and technology cluster mechanisms; and then, through the flow of talents and the feedback mechanism of advanced technological achievements, a win-win situation of economic development and environmental protection forms in the home country (Kogut and Chang, 1991; Gong and You, 2022; Ma and Gao, 2022).
OFDI can optimize the utilization of resources by accelerating resource sharing and technical cooperation, which accelerate the overall low-carbon process and energy-saving technology innovation in the home country, and will ultimately improve EE. Based on using the host country's advanced scientific and technological resources to improve their own technical levels and innovation abilities, overseas subsidiaries share cutting-edge patents, management experience, and upstream and downstream channels with the parent company through information transmission, the flow of R&D personnel, feedback of R&D results and product flow, to promote the improvement of the parent company's technical level and achieve reverse technology spillover at the enterprise level. After the parent company fully absorbs these advanced technologies to achieve economies of scale, they will be passed on to the upstream and downstream enterprises in the same industry through the demonstration role, resulting in a demonstration effect and competitive behavior in the domestic market. On the one hand, this allows other enterprises in the same industry to learn, imitate and re-innovate the advanced technology and products of the parent company, and at the same time, it encourages the upstream and downstream affiliated enterprises that provide supporting services to the parent company to continuously improve their technical level and the efficiency and quality of production factors supply. On the other hand, the absolute competitiveness of the parent company in the product market will cause competitive pressure on its peers, prompting other enterprises to take the initiative for improving their technological innovation capabilities and even eliminate inefficient enterprises by the law of “survival of the fittest” in order to improve the allocation efficiency of energy factors. The technological upgrade brought by OFDI through reverse technology spillover includes innovation in management, technology, and system, which systematically impacts EE.
Digital technology optimizes the way that enterprises in the home country obtain information, enabling them to make use of online platforms and big data to obtain information, expand their search scope and matching efficiency, break information barriers, weaken information asymmetry, fully grasp overseas market information, reduce search costs and information costs, and effectively reduce their transaction costs. The end result is greater ease of investment. Moreover, the interconnectivity of digital infrastructure and the breadth of digital technology coverage will significantly reduce the cost of commodity transportation, information acquisition, and dialogue exchange in international trade activities, thereby enhancing the host country’s ability to attract China’s OFDI, and ultimately changing the structure of China’s OFDI and promoting the increase of China’s vertical OFDI scale to the host country. Therefore, digital technology improves OFDI by improving the convenience of investment and reducing the channels of trade costs. Based on these considerations, this study puts forward the second research hypothesis:
Hypothesis 2 (H2): Digital technologies can improve EE by increasing access to OFDI in the home country.

2.3 The role of virtual agglomeration

Marshall proposed three sources of positive externalities of industrial agglomeration, namely, labor reservoir, intermediate input sharing, and knowledge spillover. The impact of VA on EE is also inseparable from these three channels. The application of digital technology is conducive to expanding the scope of information interaction among enterprises, broadening the geographical spatial pattern of industrial agglomeration, and strengthening the influence of agglomeration on the regional economy. The boundary-free organization theory holds that information, resources, and ideas can quickly cross boundaries between businesses, enabling managers to respond quickly to environmental changes. When the daily operations of enterprises are no longer bound by geography and physics, the production potential will be maximized under the support of digital technologies such as AI, cloud computing, the Internet of Things (IoT), and blockchain. With the acceleration of technological innovation iteration, virtualization features are becoming increasingly prominent, reducing the congestion effect caused by geographical agglomeration and promoting the vertical and horizontal expansion of the energy industry chain.
Advanced technology is fundamentally changing the development paradigm of traditional industrial organizations and promoting the continuous evolution and renewal of industrial development models. Industrial clusters gradually break the barriers of physical boundaries and transform from geographic spatial agglomerations to network VAs with the real-time exchange of data and information as the core (Wang et al., 2018). This model is intended to transform the demand parties and related enterprises from geographic space agglomeration to cloud-level collaborative agglomeration, reduce transaction costs by shortening the information exchange distance of each production link, and finally realize dynamic, flexible production. As a new trend of industrial organization and a new agglomeration economic model in the era of the digital economy, VA promotes the blurring of industry boundaries and the virtualization of industrial clusters (Duan and Zhan, 2023). With the open ecosystem of digital platforms as the carrier, VA can integrate all aspects of social reproduction, such as production, exchange, distribution, and consumption. The spillover effect of agglomeration is no longer limited by geographic proximity, and the information between upstream and downstream enterprises and end consumers can be transmitted and communicated quickly, accurately, and immediately (Tan and Xia, 2022). The spillover effects of closer network connections even outweigh the effects of physical agglomeration in real society (Wang and Liang, 2022).
The positive externality of VA is more from the joint effect of cross-network externality and general network externality, which is regarded as an “invisible community on the Internet”. It not only has the positive externality of traditional geographical agglomeration but also helps to break the geographical space limitation and provides a unique advantage in optimizing production factor allocation, knowledge, and information-sharing linkages (Song and Lu, 2017). Digital networks provide docking channels for the circulation of resources, equipment and waste materials, and provide complementary resource integration information for industrial production networks. The most typical example is that industrial agglomeration will also form a closed material flow between various types of enterprises, which means that the waste “abandoned” by one enterprise may be an essential raw material for another enterprise. At the cyberspace level, company search costs will be reduced, which will not only improve resource recycling but also help improve regional energy efficiency. For example, under the traditional market model, the logistical costs are reflected in the “iceberg cost” of commodities and then reflected in the form of costs in commodity prices. Physical space distance brings transportation costs and information interaction costs due to team cooperation, while the psychological distance from a geographical distance brings a “trust” cost to cooperation. Although VA cannot directly save the cost of goods transportation like traditional industrial clusters, it can reduce the unnecessary costs by optimizing transportation routes, providing goods on demand, improving supply chain efficiency, and helping companies to collect, process and analyze information more effectively through virtual operation, which reduces the costs related to finding and evaluating potential trading partners and improving logistics efficiency and output efficiency, ultimately achieving the goal of improving EE (Chen, 2017; Yang et al., 2023). As another example, VA changes the mode of enterprise organization product innovation and technology research and development. The agglomeration of industrial clusters in cyberspace will compress the “distance” of the division of labor and cooperation, making it possible to specialize in the division of labor nationwide. The energy industry chain under VA pursues a competitive advantage rather than a comparative advantage of resource endowment. The pursuit of internal production optimization is a closed process of value creation, and internal resources easily restrict its value creation ability. In a virtual industry cluster, enterprises can obtain more external resources through network interaction to compensate for the lack of internal resources. The enterprise introduces the non-core business into the professional force through third-party outsourcing and makes up for its shortcomings through the division of labor and cooperation, so that it can focus on developing the leading business. Finally, the tacit knowledge, which was previously difficult to communicate or systematically express, is encoded and decoded in digital media, VR/AR, and other next-generation information technologies, which allows that information to be transmitted in virtual space and transcend the limitations of physical space. This provides a new path for enterprises to use resources better, share knowledge and optimize innovation strategies. By leveraging the cross-border penetration capabilities of the Internet and the IoT in connecting the factors of production, a high degree of integration between offline and online is promoted on a broader scale, which greatly expands the storage space of resources throughout society and constantly changes the efficiency of factor allocation and input mix of business production, which can influence EE (Hanelt et al., 2021).
Hypothesis 3 (H3): Digital technologies can improve EE through channels that promote the virtual clustering of industries.

3 Study design and data sources

3.1 Variable setting

3.1.1 Explained variable

In terms of the factor bias of technological progress, there are two paths for improving EE. The first path is through neutral technological progress that increases the marginal output of capital, labor, and energy factors in equal proportions and causes year-on-year changes in the use of each factor, thereby improving EE. The second is through limited technological progress that changes the ratio of the marginal output of each factor, causing a change in the factor substitution effect and changing the use of energy factors relative to non-energy factors, in which the process has the effect of saving the use of energy factors and improving TFEE (Li and Li, 2022). The improvement of existing EE is also due to the technological progress caused by the growth of investment in advanced equipment and advanced processes. The technological progress generated by capital or labor alternative energy sources will significantly improve SFEE. However, it is independent of the underlying EE and does not necessarily improve the TFEE. In order to measure the improvement space of EE under a given technical level, it is also necessary to overcome the limitation that the measured SFEE excessively relies on the input of energy factors and largely ignores the roles of other factors (Hu and Wang, 2006). In this study, SFEE and TFEE were used as proxy variables of energy use efficiency.
SFEE is usually measured using energy consumption per unit of gross product (standard tons of coal per 10000 yuan), and its equation is expressed as:
where, TEC represents the total energy consumption, including nine types of energy: coal, coke, crude oil, gasoline, kerosene, diesel, fuel oil, electricity and natural gas.
The undesirable output from the process of energy utilization, that is, environmental pollutants, can be regarded as a social cost, which offsets some of the positive effects of the desired output (Wang and Lu, 2021). TFEE emphasizes the inclusion of energy factors in the factor input variables and also considers the negative impact of unexpected output on energy utilization efficiency, which can reflect the characteristics of cooperation between energy, capital, and labor and is more in line with the connotation of economic Pareto efficiency (Liu and Li, 2023; Liu et al., 2023a; Zhang et al., 2023). In measuring the TFEE, it is necessary to master the form of the production frontier, take it as the efficiency benchmark, and measure the efficiency by the relative distance between the actual output (or input) level and the frontier (Huang et al., 2023). This study adopted the SBM-GML model to measure TFEE because of the advantages of data envelopment analysis in measuring TFEE. The evaluation system of TFEE and its indicators are shown in Table 1.
Table 1 TFEE index system.
Category Index Indicator specification
Input variable Labour force The total number of employed persons in urban units and rural areas
Fixed capital Real capital stock based on 2006
Energy consumption Total of various energy sources used for consumption (tons of standard coal)
Expected output GDP Real GDP based on 2006
Undesirable output Energy carbon emission The IPCC coefficient method was used to measure the total carbon emissions from energy sources

Note: The measurement method of the actual capital stock is the perpetual inventory method, the depreciation rate is set at 10.96%, and its expression is ${{K}_{i,t}}={{K}_{i,t-1}}\left( 1-{{\delta }_{it}} \right)+{{I}_{it}}$, where $\text{ }\!\!\delta\!\!\text{ }$ represents the depreciation rate and ${{I}_{i,t}}$ represents the fixed capital investment in the current period. The formula for calculating the capital stock in 2006 is ${{K}_{0}}={{I}_{2006}}\left( {{g}_{2006-2020}}+\delta \right)$, where ${{g}_{2006-2020}}$ is the average growth rate of fixed asset investment from 2006 to 2020. The equation for calculating carbon emissions from energy sources in the 2006 IPCC Guidelines for National Greenhouse Gas inventories is $C=\underset{j=1}{\overset{9}{\mathop \sum }}\,{{C}_{j}}=\underset{k=1}{\overset{9}{\mathop \sum }}\,{{E}_{i}}\times NC{{V}_{i}}\times CE{{F}_{i}}\times CO{{F}_{i}}\times 44/12$, where ${{C}_{j}}$ is the carbon dioxide emissions generated by j energy, NCV is the average low calorific value of primary energy, CEF is the carbon emission factor provided by the IPCC greenhouse gas inventory, and COF is the carbon oxidation factor.

3.1.2 Core explanatory variable

Digital technology. The dominant feature of current advances in digital technology is the creation of a new type of asset based on a combination of computers, machines, and artificial intelligence. These assets can be produced autonomously with minimal human intervention, and production activities that were previously done only by humans and traditional capital can be carried out by intelligent machines (DeCanio, 2016). As a general-purpose technology, industrial robots have significantly changed enterprises’ production efficiency, organization, and final output, which has broad and profound impacts on the economy, society, and the environment. Relevant research shows that automated production technology represented by industrial robots is becoming a new engine driving global economic growth (Acemoglu and Restrepo, 2018). Compared with traditional automation equipment, industrial robots can be programmed according to work objects and requirements, and can also be integrated into the entire production control network based on information management, data collection, feedback information, and performing operations. The application of robots to production activities is a typical feature of integrating artificial intelligence technology and industry. It has the characteristics of autonomy, adaptability, and automation, and it also has the characteristics of a labor force, but it belongs to the fixed asset investment, and its essence is the process of capital deepening, with both labor and capital attributes, which can be classified as intellectual capital (Liu et al., 2023b). Given that the industrial sector is a major source of energy consumption and carbon emissions, this study used industrial robot installation density to measure the level of digital technology development at the provincial level. According to the research concepts of the existing literature (Acemoglu and Restrepo, 2020; Caselli et al., 2021; Chen et al., 2022; Shen and Zhang, 2023), this study used the employment data of various provinces and industries in China Labor Statistics Yearbook 2007 to match the robot installation data provided by the International Federation of Robotics (IFR) and the robot installation density data at the provincial level were obtained. The calculation process is as follows:
$Robo{{t}_{it}}=\sum\limits_{j=1}^{N}{\frac{emplo{{y}_{ij,t=2006}}}{emplo{{y}_{i,t=2006}}}\times \frac{Robo{{t}_{jt}}}{emplo{{y}_{j,t=2006}}}}$
where, Robot is the installed density of robots, which represents the development level of digital technology; Robotit is the robot installation density of province i in year t; N is a collection of industries involved in manufacturing; employij,t=2006 is the number of employees in industry j of province i in 2006; employi,t=2006 is the total number of employment in province i in 2006; and Robotjt/employj,t=2006 is the robot installation density of each year and industry level.

3.1.3 Mediating variables

Outward foreign direct investment (OFDI). Since the flow data fluctuates significantly in the short term, and this study mainly investigates the long-term relationships between variables, the OFDI stock data of each region were used over the years. Investment stock measures the cumulative amount of foreign investment up to a given point in time, which can better reflect the long-term effect of investment, and there is no net outflow (negative) situation.
Virtual agglomeration. In the process of virtual industrial services, although the enterprise’s digital content is online, the digital content formed by VA is not virtual but carries specific professional knowledge, big data analysis, creative design, virtual derivatives, blockchain endorsement, and other services through digital media. Based on the definition of the connotation and denotation of VA used in existing studies, this study used the method of location entropy to measure it according to the concepts in the existing literature (Ru and Liu, 2022; Liu et al., 2023b). The calculation method is as follows:
where, VA is virtual agglomeration, ICSit and Totalit represent the number and total number of employment in the information transmission, computer service and software industries in year t of city i, respectively. ICS and Total represent employment in all urban information transmission systems, computer services and software industries and total employment, respectively, that is, total employment in industries at the national level and total employment in all industries.

3.1.4 Control variables

A set of provincial-level control variables were added to the benchmark model in this study based on existing studies to mitigate the bias caused by missing variables as much as possible. This set of inter-provincial characteristic variables includes regional economic development level (EDL), measured by the real per capita GDP, but excluding the price factor. The industrial structure (IS) was measured by the proportion of the secondary industry’s added value to the GDP. Transportation infrastructure (TI) was measured by the actual paved area of roads and the total paved area, bridges, and tunnels connected with the road (10000 km2). Macro-control was measured by the proportion of expenditure in the general public budget to GDP. The number of patent applications for inventions was used to measure technological innovation ability (TIA). Urbanization (UR) was measured by the proportion of the urban population to the total population at the end of the year. Foreign direct investment (FDI) was measured using the amount of FDI utilized by each region in the current year and converted into CNY (Chinese yuan) based on the average exchange rate between CNY and USD over the years.

3.2 Econometric model

In order to determine whether digital technology can improve EE, this study constructed the following panel econometrics model combined with the above set of various variables and research H1:
$TFEE{}_{it}={{a}_{0}}+{{a}_{1}}D{{T}_{it}}+\sum\limits_{j=1}^{7}{{{a}_{2}}{{Z}_{ijt}}}+{{\varepsilon }_{it}}+{{\nu }_{i}}+{{\delta }_{t}}$
$SFEE{}_{it}={{b}_{0}}+{{b}_{1}}D{{T}_{it}}+\sum\limits_{j=1}^{7}{{{b}_{2}}{{Z}_{ijt}}}+{{\varepsilon }_{it}}+{{\nu }_{i}}+{{\delta }_{t}}$
where, the subscripts i, t, and j represent province, time, and the j-th control variable, respectively. a0 and b0 represent constant terms, a1 and b1 represent regression coefficients for digital technology, Z is the information set of a series of control variables, vi is the individual fixed effect, δt is the time fixed effect, εit is the random disturbance term subject to the white noise process, and DT is digital technology. SFEE and TFEE represent single-factor energy efficiency and total factor energy efficiency, respectively.
This study used the more commonly used approach of the two mechanisms in order to validate the intermediate path of digital technology to improve EE. The new intermediary effect model proposed by Jiang (2022) focuses on explaining how institutional variables affect EE in the components of theoretical analysis and research hypothesis, and then the impact of digital technology on institutional variables is tested in the empirical analysis component. The main observation is whether the coefficients and significance of the core explanatory variables in the second part of the equation meet the expectations. Compared with the classical panel fixed effect model, the interactive panel fixed effect can better fit the data and fully consider the impacts of various uncertain factors on the real economy and society (Bai et al., 2009; Petrova and Westerlund, 2020). This method has important applications in controlling for missing variables, capturing time-varying features, and improving goodness of fit. To alleviate the endogeneity problem of the mediation effect model, the following two equations were established in this study.
$OFD{{I}_{it}}={{c}_{0}}+{{c}_{1}}D{{T}_{it}}+\sum\limits_{j=1}^{7}{{{c}_{2}}{{Z}_{ijt}}}+{{\varepsilon }_{it}}+{{\nu }_{i}}+{{\delta }_{t}}+\nu _{i}^{\text{T}}{{\delta }_{t}}$
$V{{A}_{it}}={{d}_{0}}+{{d}_{1}}D{{T}_{it}}+\sum\limits_{j=1}^{7}{{{d}_{2}}{{Z}_{ijt}}}+{{\varepsilon }_{it}}+{{\nu }_{i}}+{{\delta }_{t}}+\nu _{i}^{\text{T}}{{\delta }_{t}}$
where, c0 and d0 represent constant terms, c1 and d1 represent regression coefficients of digital technology, c2 and d2 represent the regression coefficient of the control variable, $\text{ }\!\!\nu\!\!\text{ }_{i}^{\text{T}}{{\text{ }\!\!\delta\!\!\text{ }}_{t}}$ is the interactive fixed effect, and Z represents a set of control variables. OFDI and VA represent outward foreign direct investment and virtual agglomeration, respectively.

3.3 Data sources

Following the principles of data availability and comparability, this study selected panel data from 30 provinces in China from 2006 to 2021 as the statistical samples. Note that samples from Tibet Autonomous Region and Hong Kong, Macao, and Taiwan are not included due to missing data values and inconsistent statistical caliber. The primary data of the relevant variables in this study came from the China Statistical Yearbook, China Outbound Direct Investment Statistical Bulletin, China Energy Statistical Yearbook, International Federation of Robotics, EPS data platform, and the statistical yearbooks of provincial and municipal statistics. For the very few missing values, this study used the linear interpolation method to complete the dataset. In order to eliminate the negative effects of outliers and heteroscedasticity, 1% tailing treatment and logarithmic conversion were performed on both ends of all continuous variables. The descriptive statistical analysis of the relevant variables is shown in Table 2.
Table 2 The descriptive statistics of the variables
Variable Code Mean Std. Dev. Min. Max.
Single-factor energy efficiency SFEE 0.1614 0.5543 -0.7118 1.5582
Total-factor energy efficiency TFEE 1.6756 0.7879 0.6170 5.4777
Digital technology DT 3.3912 2.0908 -0.5713 6.8516
Economic development level EDL 10.6067 0.6171 9.1016 11.9658
Transportation infrastructure TI 0.2301 0.0979 0.0936 0.5927
Foreign direct investment FDI 9.6884 0.8474 7.2133 11.4446
Virtual agglomeration VA 6.4491 1.4895 3.1781 9.8798
Industrial structure IS 0.7177 0.9659 -1.5256 2.8241
Technological innovation ability TIA 4.7330 4.8934 0.1099 31.5533
Urbanization UR 7.6893 1.6483 3.5554 11.0551
Outward foreign direct investment OFDI 4.0036 0.2401 3.4177 4.4920

4 Results

4.1 Baseline regression

The results show that both the F test and Hausman test reject the null hypothesis at the 1% level, indicating that the FE model is the most suitable for the sample data in this study. Driscol-Kraay’s (DK) estimation method sets the error structure as heteroscedasticity and a specific order autoregressive. Compared with other estimation methods, this method can obtain consistent standard errors in the control of heteros- cedasticity and autocorrelation. When the time dimension is gradually increased, the standard errors are robust to the general form of sectional correlation and time correlation (Driscoll and Kraay, 1998; Shen et al., 2023b). According to research H1, the two-way fixed effect (TWFE) was used to calculate Eqs. (1) and (2) and the results are shown in Table 3.
Table 3 The results of baseline regression
Variable TFEE SFEE
DT 0.2349***
EDL 0.1151
TI 0.3405***
UR 0.4860
MC 0.1411
IS -0.0314***
TIA -0.2310***
FDI 0.0833**
Individual effect No Yes No Yes
Time effect No Yes No Yes
R2 0.4180 0.7043 0.7176 0.9057
N 480 480 480 480

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively, and the t statistics are reported in parentheses.

The data in Table 3 show that for the results of OLS, which does not include the time-fixed effect or the individual fixed effect, the regression coefficients of DT on TFEE and SFEE are 0.2349 and -0.0833, respectively. Both reject the null hypothesis at the significance level of 1%, which initially confirms research H1, that DT can improve EE. Based on the results of the TWFE, the regression coefficients of DT for the two categories of EE are 0.2267 and -0.2497, respectively. They are significant at 5% and 1%, respectively, indicating that technology can improve TFEE and reduce energy consumption per unit of GDP, that is, DT can improve EE. Thus, H1 is confirmed. DT has significant advantages in reducing the cost of data analysis and improving the speed of information transmission, which helps to improve the optimal combination of production factors such as labor, capital, energy, and technology, accurately allocate factor resources, reduce energy consumption, and ultimately improve EE (Tang et al., 2021).

4.2 Robustness test

In order to verify the robustness of the baseline regression results, this study used two methods of replacing the econometric model and the core explanatory variables. First, the comprehensive feasible generalized least squares (FGLS) method was used to correct the potential autocorrelation, heteroscedasticity, and cross-sectional correlation of short panel data to obtain more effective estimators. Secondly, the spatial location information of different provinces was incorporated into the model using the spatial Durbin model. The rapid development of DT is based on accelerating the flow of factors and optimizing the input-output combination. Due to the many economic connections among Chinese cities, the impact of DT on cities is not independent but has a strong spatial correlation. Similarly, advances in energy-using technology in a region or industry may improve the overall energy efficiency in the region or adjacent regions through technology spillover, the industrial chain, or market imitation. The third method is replacing the proxy indicators of DT. In order to implement the national big data strategy and accelerate the creation of a new economic and social development platform supported by big data, the “Action Outline for Promoting Big Data Development” issued by The State Council in 2015 proposed that pilot work related to big data should be carried out in-depth to achieve the integration of big data-related infrastructure and the convergence and utilization of data resources. Promoting the use and sharing of data resources is the primary task of establishing big data pilot zones (Guo et al., 2023; Shen et al., 2023a).
As seen in Table 4, the test results of these three methods all show that DT can significantly improve the TFEE and reduce the EE per unit of GDP. The signs and directions of these result coefficients are consistent with the results of the bidirectional fixed-effect model. That is, the conclusion that DT can improve EE is robust.
Table 4 The results of baseline regression
Variable TFEE SFEE
DT 0.1313***

Note: *** indicates significance at the 1% level, the z statistics are reported in parentheses for the FGLS and SDM models, the t statistics are reported in parentheses for the DID models. Time effect, individual effect and control variables have been controlled.

4.3 Endogeneity Test

Although this study assumes that DT is exogenous to EE and tries to control the external variables affecting EE as much as possible, the model design still needs to address the problem that missing essential variables, such as EE, may be affected by other factors such as environmental regulations, resource endowments, and consumer preferences. In addition, one cannot rule out that DT may be endogenous; that is, there could be a reverse causality endogenous relationship, such as provinces with higher EE may have a complete digital infrastructure and more advanced management experience, pay more attention to the digital transformation of enterprises and flexible production of products, and their DT research and development and application are more convenient. To eliminate the potential endogenous problem, this study used the instrumental variable method to deal with it. Referring to the ideas of previous studies, this study used the number of post offices per 10000 people in each region in 1984 as the instrumental variable (Zhao et al., 2020; Shen et al., 2023b). The instrumental variable selected in this study was the cross-section data for 1984, but that data does not change with time. Therefore, following the solution of the existing literature, this study introduced a variable (the number of Internet broadband access ports) that changes with time and interacts with the historical data of cross-section to form the panel data, and then the instrumental variable of this study was constructed (Nunn and Qian, 2014; Zhao et al., 2020). In addition, to verify the robustness of the test results of the instrumental variable method, this study also used the generalized space two-stage least square method (GS2SLS) to estimate the sample data. This method takes the higher-order spatial lag term of the explanatory variable as the tool variable and estimates the spatial panel model based on the 2SLS method, which can control the spatial spillover effect and endogeneity of DT and EE simultaneously (Baltagi and Liu, 2014).
As seen in Table 5, the LM and Wald F test results of the 2SLS method show that the instrumental variables selected in this study are influential through the under-recognized and weak instrumental variable tests. The results of both 2SLS and GS2SLS show that DT can significantly improve TFEE and SFEE, and the estimation results using the instrumental variables are consistent with the baseline regression results, which verifies the robustness of the baseline regression results.
Table 5 Results of the endogeneity test
Variable TFEE SFEE
Second stage GS2SLS First
Second stage GS2SLS
DT 0.6030***
Instrumental variable 1.3304***
LM test 119.829*** 119.829***
F test 155.334 155.334

Note : The z statistics are reported in parentheses for the 2SLS model, and the t statistics are reported in parentheses for the GS2SLS model. ***, ** and * represent significance at the levels of 1%, 5% and 10%, respectively. Time effect, individual effect and control variables have been controlled.

4.4 Mechanism test

In order to examine the channel mechanism by which DT improves EE, combined with research hypotheses 2 and 3, this study used the interactive fixed effect model for verification.
The data in Table 6 show that the regression coefficient of DT for VA is 0.1121, and it has passed the significance test at the 10% level. These results suggest digital technologies can improve EE through channels facilitating VA. Thus, H2 is tested. The integration and application of big data, the industrial Internet of Things, and 5G digital technologies help to accelerate the agglomeration of new production factors (data) in virtual cyberspace and the entire flow among various subjects, breaking the dependence of traditional industries on geographical space and promoting the formation of close connections between enterprises and enterprises and between enterprises and consumers in the network information space. On the one hand, the flow and agglomeration of production factors of digital infrastructure in the virtual space network provide a material carrier, which helps the real-time circulation and exchange of factors on the network at low cost and high efficiency, overcoming the problem of information asymmetry, driving the flow of factors to areas with more development space, and optimizing the allocation of factors across the core of the division of labor structure. On the other hand, DT enables VA entities not only to publish output information quickly but also to instantly obtain any number of intermediate inputs that exist in the market. At this time, the market effect of intermediate inputs will be amplified infinitely, reducing external transaction costs (Ru and Liu, 2022). Therefore, DT integrates the resources of various participants in the service ecosystem, realizing the dynamic allocation of production, services, and resources in the virtual space, as well as the value co-creation between service providers and users, ultimately contributing to the improvement of EE.
Table 6 The results of the mechanism test
Variable VA OFDI
DT 0.1121*(1.82) 0.6555**(2.15)
Interaction effect Yes Yes

Note: * and ** indicate significance at the 10% and 5% levels, respectively, and the t statistics are reported in parentheses. Time effect, individual effect and control variables have been controlled.

The data in Table 6 also show that the regression coefficient of the influence of DT on OFDI is 0.6555, and it is significant at the 5% level. These results suggest that digital technologies can improve EE by increasing the channels for home-country companies to invest overseas. Thus, H3 is verified. First of all, with the help of DT, it is easier for enterprise management to obtain the marketing information of subsidiaries and departments in different countries and information that accurately captures idle resources, thus reducing the stickiness of enterprise costs (Warren et al., 2015). The application of various digital technologies accelerates the exchange of information elements in the supply chain and the connections of resources. Enterprises can make timely and reasonable responses according to the information fed back by digital technologies, which can improve the efficiency of capital utilization, reduce the possibility of idle resources and reduce the stickiness of financial costs. The automatic control of the business processes promoted by DT reduces the probability of management's self-interested manipulation and helps curb the cost risk caused by opportunism. Secondly, DT can improve business efficiency and promote OFDI. Digital platforms represented by Amazon, Dunhuang, eBay, Twitter, and Zoom can assist management in actively pushing information directly to various market participants scattered around the world, including investors, creditors, and suppliers, under the condition of efficiently and accurately processing and outputting adequate information, thus improving the matching efficiency between the demand side and the supply side of platform-based service enterprises (Liu et al., 2015; Warren et al., 2015). The de-intermediation operation mode of DT can reduce the transaction costs of overseas mergers and acquisitions and help enterprises sort out existing resources and redistribute them, improve resource utilization efficiency, and reduce management costs. Finally, DT can enhance the adaptability of enterprises to the international market and promote OFDI. DT can not only improve the internationalization tendency of enterprises by reducing the transaction costs of information exploration, international communication, and logistics, but it can also improve the correlations between enterprises themselves and the upstream and downstream enterprises in the supply chain through information sharing and encourage enterprises to implement internationalization strategies. Enterprises can use digital infrastructure to improve their information processing capabilities, enhance their understanding of international markets, enhance their ability to perceive opportunities in dynamic and complex international markets and enhance the flexibility of global supply chains (Elia et al., 2015).

5 Conclusions and policy implications

5.1 Conclusions

Science and technology determine the future of energy, and science and technology create future energy. Based on a statistical sample of 30 provinces in China from 2006 to 2021, this study used the solid bidirectional fixed effect model and the Driscol-Kraay method to adjust the standard error, and it objectively evaluated the role of DT represented by industrial robots in improving EE and its potential pathway mechanism. Similar to the existing literature that analyzes how technological advances affect EE, the results show that DT can significantly improve the TFEE and reduce the energy consumption per unit of GDP. This conclusion is still robust after the regression of FGLS and SDM models, the use of the national big data total pilot area as the proxy variable of DT and the test of the DID model. In addition, the instrumental variable method and GS2SLS method used in this study have solved the endogenous problems, and the results show that DT can still significantly improve EE. The mechanism test results show that DT can promote VA and increase OFDI to improve EE. This research provides a valuable discussion on the role of DT in the energy field against the background of the new wave of technological revolution. The conclusions obtained are helpful for enterprises and city managers by providing experience for reference in the digital transformation of energy.

5.2 Policy enlightenment and suggestions

The results of this study are useful for promoting the digital transformation of the energy sector, focusing on building a high-quality digital grid and improving the digital capabilities of the energy industry. The digitalization of the energy industry is a digital upgrade of the energy industry chain and supply chain, the crux of which is a highly intelligent digital grid. This effort should continue to deepen digital grid technology, adhere to the needs of the energy industry as a guide, organize resources from all parties in the digital grid field, continue to improve the architecture of the digital grid, and strengthen the standard leading and compilation to achieve synergy between continued innovation, standard creation, and industrial application. Enterprises and government departments should focus on researching the integration of the digital grid into the national integrated arithmetic network and accelerating the construction of the national arithmetic network infrastructure. Promote the in-depth integration of new energy technologies and information technology, strengthen cloud computing services, and layout and build national hub nodes of the national integrated computing network. Form a distributed and open sharing network based on renewable energy, build a national energy internet with extra-high voltage grids as core nodes and coordinate the development of grids at all levels, and change the energy development mode that is now based on over-reliance on coal transportation and the development mode of unbalanced power. Comprehensively improve the intelligent interactive capabilities of the distribution grid and promote the use of distributed energy for widespread access, electric vehicles, energy storage, smart meters, and smart homes.
Relying on the digital economy, integrate and utilize the R&D strength of the whole industry chain. City managers and entrepreneurs should seize the opportunities brought about by the development of the digital economy, promote innovation in production technologies, business models, and industrial formats, closely integrate data advantages with the population advantages, market advantages, and institutional advantages of traditional manufacturing industries, deeply promote action plans based on DT, promote intelligent production and high-end industries, and advocate entrepreneurship and craftsmanship. Leveraging data resources to complete the effective docking of upstream and downstream demand in the manufacturing industry chain, enterprise manager should increase the factors of production to capital. Use the advantages of digitalization and the Internet to change traditional “high investment and low return” production mode of enterprises, integrate manufacturing modules and downstream service modules, and provide personalized and accurate products and services. Focusing on the essential positioning of energy security, administrative departments and various types of companies should promote the integrated application of intelligent manufacturing key technology and equipment, core support software, the industrial Internet and other systems, promote manufacturing service cloud platforms, intelligent connected products and enabling tools and systems, and improve TFEE by enhancing industrial VA.
Education department should strengthen the construction of a digital talent team and accelerate the construction of a conforming talent team. By combining absorption and training, we will fill the gap of composite digital talents with multidisciplinary knowledge of oil and gas, economics, law, industrial policy, etc., as well as excellent practical abilities and management technology. In the process of digital transformation, it is necessary to fully mobilize the enthusiasm, initiative, and creativity of talents in all aspects and actively participate in and lead this change. It is necessary to continuously absorb talents from various fields and adopt a multi-professional integration organization model, that is, artificial intelligence experts, mathematicians, software engineers, and oil and gas professional engineers should be closely combined to establish a multi-professional collaborative working group, so that DT and the oilfield business can be seamlessly connected. It is necessary to formulate a corresponding talent training strategy, organize DT training extensively, and build a composite talent team that is proficient in the energy business and understands DT. Industrial enterprises should increase the re-education of digital skills for existing talents, improve their digital thinking and management ability, innovate the talent management mode, stimulate talent potential and vitality, and tap into relevant digital technical talents, so as to ensure that modern information technology can play a better role in improving EE.
We thank the reviewers for their valuable comments on earlier drafts of this manuscript that helped us improve its quality. The authors are thankful for the financial support of Fundamental Research Funds from the Central Universities in Huaqiao University.


We thank the reviewers for their valuable comments on earlier drafts of this manuscript that helped us improve its quality. The authors are thankful for the financial support of Fundamental Research Funds from the Central Universities in Huaqiao University.
Acemoglu D, Restrepo P. 2018. The race between man and machine: Implications of technology for growth, factor shares, and employment. American Economic Review, 108(6): 1488-1542.


Acemoglu D, Restrepo P. 2020. Robots and jobs: Evidence from US labor markets. Journal of Political Economy, 128(6): 2188-2244.

Bai J. 2009. Panel data models with interactive fixed effects. Econometrica, 77(4): 1229-1279.


Baltagi B H, Liu L. 2014. Instrumental variable estimation of a spatial autoregressive panel model with random effects. Economics Letter, 111(2): 135-137.


Caselli M, Fracasso A, Traverso S. 2021. Robots and risk of COVID-19 workplace contagion: Evidence from Italy. Technological Forecasting and Social Change, 173: 121097. DOI: 10.1016/j.techfore.2021.121097.

Chen C, Frey C B, Presidente G. 2022. Automation or globalization? The impacts of robots and Chinese imports on jobs in the United Kingdom. Journal of Economic Behavior & Organization, 204: 528-542.


Chen X. 2017. Virtual transformation of industrial cluster. China Industrial Economics, 357(12): 78-94. (in Chinese)

Chen Y, Wang M, Feng C, et al. 2021. Total factor energy efficiency in Chinese manufacturing industry under industry and regional heterogeneities. Resources, Conservation and Recycling, 168: 105255. DOI: 10.1016/j.resconrec.2020.105255.

Dalla’Ora N, Alamin K, Fraccaroli E, et al. 2022. Digital transformation of a production line: Network design, online data collection and energy monitoring. IEEE Transactions on Emerging Topics in Computing, 10(1): 46-59.


DeCanio S J. 2016. Robots and humans-complements or substitutes? Journal of Macroeconomics, 49: 280-291.


Driscoll J C, Kraay A. 1998. Consistent covariance matrix estimation with spatially dependent panel data. The Review of Economics and Statistics, 80(4): 549-560.


Duan X, Zhang Q. 2023. Industrial digitization, virtual agglomeration and total factor productivity. Journal of Northwest Normal University (Social Sciences), 60(1): 135-144. (in Chinese)

Elia S, Giuffrida M, Mariani M M, et al. 2015. Resources and digital export: An RBV perspective on the role of digital technologies and capabilities in cross-border e-commerce. Journal of Business Research. 132: 158-169.


Filippini M, Hunt L C. 2016. Measuring persistent and transient energy efficiency in the US. Energy Efficiency, 9: 663-675.

Fu S, Liu J, Tian J, et al. 2023. Impact of digital economy on energy supply chain efficiency: Evidence from Chinese energy enterprises. Energies, 16(1): 568. DOI: 10.3390/en16010568.

Gong M, You Z. 2022. Environmental technology effect of two-way foreign direct investment interactive development in China. Frontiers in Ecology and Evolution, 10: 954614. DOI: 10.3389/fevo.2022.954614.

Guo B, Wang Y, Zhang H, et al. 2023. Impact of the digital economy on high-quality urban economic development: Evidence from Chinese cities. Economic Modelling, 120: 106194. DOI: 10.1016/j.econmod.2023.106194.

Hanelt A, Bohnsack R, Marz D, et al. 2021. A systematic review of the literature on digital transformation: Insights and implications for strategy and organizational change. Journal of Management Studies, 58(5): 1159-1197.


Hu J, Wang S. 2006. Total-factor energy efficiency of regions in China. Energy Policy, 34(17): 3206-3217.


Huang J, Feng S, Yang Y, et al. 2023. Total factor energy efficiency and its measurement, comparison and validation. Resources Science, 45(2): 281-295. (in Chinese)

Huang S, Chen L. 2023. The impact of the digital economy on the urban total-factor energy efficiency: Evidence from 275 cities in China. Sustainability, 15(4): 3195. DOI: 10.3390/su15043195.

Jiang T. 2022. Mediating effects and moderating effects in causal inference. China Industrial Economics, 410(5): 100-120. (in Chinese)

Kogut B, Chang S J. 1991. Technological capabilities and Japanese foreign direct investment in the United States. The Review of Economics and Statistics, 73(3): 401-413.


Lange S, Pohl J, Santarius T. 2020. Digitalization and energy consumption. Does ICT reduce energy demand? Ecological Economics, 176: 106760. DOI: 10.1016/j.ecolecon.2020.106760.

Li H, Li Y. 2022. Environmental regulation, directed technical change and energy efficiency. East China Economic Management, 36(4): 97-106. (in Chinese)

Lin B, Du K. 2013. The energy effect of factor market distortion in China. Economic Research Journal, 248(9): 125-136. (in Chinese)

Lin B, Tan R. 2016. Ecological total-factor energy efficiency of China’s energy intensive industries. Ecological Indicators, 70: 480-497.


Liu D, Chen S, Chou T. 2015. Resource fit in digital transformation lessons learned from the CBC bank global e-banking project. Management Decision, 49(10): 1728-1742.


Liu J, Li Y. 2023. How can the digital economy enable the improvement of all factor energy efficiency? The Theory and Practice of Finance and Economics, 44(2): 105-113. (in Chinese)

Liu J, Xue Y, Mao Z, et al. 2023a. How to improve total factor energy efficiency under climate change: Does export sophistication matter? Environmental Science and Pollution Research, 30: 28162-28172.


Liu Y, Han Y, Wang X. 2023b. Can industrial intelligentization balance the promoting growth and ensuring living needs? Journal of Quantitative & Technological Economics, 40(6): 69-90. (in Chinese)

Ma G, Cao J. 2022. Spatial heterogeneity impacts of bilateral foreign direct investment on green energy efficiency in China. Frontiers in Environmental Science, 10: 905933. DOI: 10.3389/fenvs.2022.905933.

Nunn N, Qian N. 2014. US food aid and civil conflict. American Economic Review, 104(6): 1630-1666.

Ohene-Asare K, Tetteh E N, Asuah E L. 2020. Total factor energy efficiency and economic development in Africa. Energy Efficiency, 13: 1177-1194.


Petrova Y, Westerlund J. 2020. Fixed effects demeaning in the presence of interactive effects in treatment effects regressions and elsewhere. Journal of Applied Econometrics, 35(7): 960-964.


Ru S, Liu H. 2022. New infrastructure construction, industrial virtual agglomeration and coordinated development of regional economy. Journal of Harbin University of Commerce (Social Science Edition), 187(6): 104-115.

Shen K, Lin J, Fu Y. 2023a. Network infrastructure construction, information accessibility and the innovation boundaries of enterprises. China Industrial Economics, 418(1): 57-75. (in Chinese)

Shen Y, Yang Z, Zhang X. 2023b. Impact of digital technology on carbon emissions: Evidence from Chinese cities. Frontiers in Ecology and Evolution, 11: 1166376. DOI: 10.3389/fevo.2023.1166376.

Shen Y, Zhang X. 2023. Intelligent manufacturing, green technological innovation and environmental pollution. Journal of Innovation & Knowledge, 8(3): 100384. DOI: 10.1016/j.jik.2023.100384.

Song H, Lu Q. 2017. The innovation of supply chain finance pattern based on virtual clusters: A case study of SJET. China Industrial Economics, 350(5): 172-192. (in Chinese)

Song X, Wang H. 2019. Technology spillovers, outward foreign direct investment and total factor energy efficiency: Empirical analysis based on the Yangtze River Delta region. Accounting and Economics Research, 33(2): 111-127. (in Chinese)

Tan H, Xia J. 2022. The theory and mode of digital trade rebuilding industrial agglomeration—From geographical agglomeration to online agglomeration. Research on Financial and Economic Issues, 463(6): 43-52. (in Chinese)

Tang C, Xu Y, Hao Y, et al. 2021. What is the role of telecommunications infrastructure construction in green technology innovation? A firm-level analysis for China. Energy Economics, 183: 105576. DOI: 10.1016/j.eneco.2021.105576.

Wang Q, Lu F. 2021. Economic effects of high-speed rail: Emission reduction and efficiency enhancement. Statistical Research, 38(2): 29-44. (in Chinese)

Wang R, Liang Q. 2022. Realistic foundation and application of virtual agglomeration in digital economy. Journal of Changan University (Social Science Edition), 24(4): 34-52. (in Chinese)

Wang R, Liang Q, Li G. 2018. A new form of the spatial agglomeration under the deep integration of a new generation of information technologyand the real economy. Journal of Management World, 34(2): 13-21. (in Chinese)

Warren J D, Moffitt K C, Byrnes P. 2015. How big data will change accounting. Accounting Horizons, 29(2): 397-407.


Xu B, Ke D, Xu Z. 2023. Does the improvement of fossil energy efficiency promote the upgrading of industrial structure in China? Jiangxi Social Sciences, 43(3): 52-63. (in Chinese)

Xue Y, Tang C, Wu H, et al. 2022. The emerging driving force of energy consumption in China: Does digital economy development matter? Energy Policy, 165: 112997. DOI: 10.1016/j.enpol.2022.112997.

Yang J, Wu R, Yang H. 2023. Digital transformation and enterprise sustainability: The moderating role of regional virtual agglomeration. Sustainability, 15(9): 7597. DOI: 10.3390/su15097597.

Zhang W, Cheng J, Liu X, et al. 2023. Heterogeneous industrial agglomeration, its coordinated development and total factor energy efficiency. Environment, Development and Sustainability, 25: 5511-5537.


Zhao T, Zhang Z, Liang S. 2020. Digital economy, entrepreneurship, and high-quality economic development: Empirical evidence from urban China. Journal of Management World, 36(10): 65-76. (in Chinese)

Zhao Y, Xia S, Zhang J, et al. 2021. Effect of the digital transformation of power system on renewable energy utilization in China. IEEE Access, 9: 96201-96209.


Zhong Y, Moon H C. 2023. Investigating the impact of industry 4.0 technology through a TOE-based innovation model. Systems, 11(6): 277. DOI: 10.3390/systems11060277.