1. Introduction
With the rapid evolution of the new scientific and technological revolution, the latest generation of information and communication technology has achieved an integrated breakthrough. Frontier Technology, including Big Data, Cloud Computing, and Artificial Intelligence, is promoting the reform of resource allocation and industrial upgrading with unprecedented breadth and depth [
1]. The Global Digital Economy White Paper (2022) shows that the added value of the digital economy in 47 major countries around the world has reached USD 38.1 trillion, accounting for 45.0% of GDP, indicating the growing importance of digital elements [
2]. In order to reshape the competitive advantage, the manufacturing industry has introduced a large number of digital assets as new production factors and applied digitization on the strategic level [
3,
4]. For example, the industrial digitalization scale of China has achieved USD 5.85 trillion, occupying 81.7% of the digital economy. The digital transformation of the manufacturing industry realizes renewed value creation and remodeling through the reorganization of resources and the innovation of production modes [
5]. Meanwhile, it also exerts a new impact on resource utilization and the environment. At present, facing the increasingly deteriorating climate and environment, carbon emission reduction has become a global consensus. The EU plans to implement a carbon border tax in 2026 and proposes to take the lead in achieving “carbon neutrality” by 2050. The United States returned to the Paris Agreement and proposed to achieve the goal of a 100% clean energy economy and net-zero carbon emission by 2050. China has also responded positively to the international call, suggesting to strive for the peak of carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060 [
6]. As the pillar industry of the economy, the manufacturing industry has the characteristics of high input and high consumption, and it is an important source of greenhouse gas emissions. International Energy Agency (IEA) statistics show that, in 2019, the carbon emissions of the manufacturing industry reached 6254 million tons, accounting for 19% of global greenhouse gas emissions, surpassed in amount only by the energy industry. Manufacturing faces severe pressure on carbon reduction. In particular, the developing economies represented by China have undertaken major manufacturing activities in the world, making it more difficult to complete their energy conservation and emission reduction tasks. The IEA stated that, in 2019, the carbon emissions of developing economies accounted for 58% of the world. Therefore, the issue of carbon emission in the manufacturing industry should be given full attention, and it is essential to explore the path to reducing carbon emissions in manufacturing, especially in developing economies.
Against the background of the rapid development of the digital revolution, more and more countries are paying attention to the application of digital technology in environmental protection. For example, the 2021 China Double Carbon Strategy and Energy Digitization Forum put forward the view that digital technology will accelerate the process of the energy revolution and help the country achieve the goal of carbon peak and carbon neutralization, which affirmed the significant position of digital technology in carbon emission reduction strategy. However, as a new production factor, whether digital input can truly realize environmental sustainability through integration with the manufacturing industry is a question worthy of deep consideration. Generally speaking, digital input should positively affect the ecology and the environment through less resource consumption and the substitution of pollution elements [
7]. However, will resource consumption caused by the manufacturing industry’s digital transformation aggravate the burden on the environment? From the perspective of developing countries, which have been undertaking the work of processing and manufacturing for the long term, could digital input reverse the current situation of high energy consumption and high pollution in the manufacturing industry? It is of great theoretical and practical significance to clarify these issues for achieving national environmental goals and accelerating the process of world carbon emission reduction.
The main structure of this paper is as follows: The second part reviews and combs the relevant literature, including carbon emission and digitization in manufacturing, as well as digital investment and FDI. The third part conducts the theoretical analysis and proposes research hypotheses. The fourth part constructs an econometric model, describes the calculation methods of key variables, and introduces the data sources. The fifth part provides the corresponding empirical test results, including benchmark regression results, robustness test results, mechanism test results, endogeneity test results, and extended analysis. The sixth part further extends the analysis for developing countries based on the hypothesis proposed in the third part. The seventh part summarizes the full text and puts forward corresponding policy suggestions and also discusses the shortcomings and future perspectives of this research.
2. Literature Review
With the continuous deterioration of the global climate and environment, the issue of carbon emissions has been widely concerning. As an industry with severe energy consumption and pollution, manufacturing is an important source of global greenhouse gas emissions. Existing literature has fully explored the field of carbon emission in the manufacturing industry, and its research is mainly divided into the following two categories: The first category focuses on the measurement of carbon emission in the manufacturing industry, which includes the Logarithmic Mean Divisia Index (LMDI) method [
8], the Multi-Regional Input–Output Method (MRIO) [
9], etc. The second type of literature mainly focuses on the analysis of the influencing factors of carbon emissions. Specifically, in the existing research, some scholars use the Index Decomposition Analysis (IDA) and Structural Decomposition Analysis (SDA) methods to investigate the actual contribution of various factors to carbon emissions in the manufacturing industry. For example, Li et al. took China’s manufacturing industries as research samples and decomposed the impact of various factors, including economic development, population size, and technical progress on carbon emission levels [
6]. While other researchers tend to focus on the impact of a single factor on carbon emissions, such as scale expansion, structural adjustment, and investment changes [
10,
11], the main driving force behind them is still the change in energy or resource consumption. Therefore, improving production efficiency [
12] and promoting technological progress [
13] are effective ways to reduce carbon emissions.
Regarding research on the manufacturing industry’s digital transformation, the early literature mainly focused on the perspective of technology. Scholars used to believe that digital transformation refers to the improvement of production efficiency and enterprise performance through digital technology [
1] or the improvement of the management and decision level by using information technology such as ERP [
14]. In recent years, with the development of new digital technologies such as Big Data and Cloud Computing, the importance of digital technology in enterprise management has been further strengthened, and the connotation of industrial digitization has expanded further. For example, Ilvonen believes that industrial digitization refers to applying digital technology to products, production, and services, promoting the restructuring and transformation of enterprise production and operation mode to gain competitive advantages [
15]. No matter how the definition of industrial digitization changes, the constant is that the existing literature attaches great importance to the role of information or data and regards it as a new production factor, believing that it has an impact on industrial production, organization, and management through its penetration and integration with traditional factors [
5,
16]. This has been established by policies in some countries, such as China, which officially established the status of data as a production factor in 2019. In this paper, we mainly refer to the research of Schallmo and Williams [
17] to define digitalization in the manufacturing industry. They elaborate on digital transformation and break it down into two processes: digitization and digitalization, where digitization represents the process of converting analog information to digital information. In other words, it is the process of acquiring information and data through digital technologies such as information and communications technology (ICT). Digitalization is a fundamental change made to business models based on newly acquired knowledge gained via digitization initiatives [
17]. Clarifying the distinction between these two definitions will be of great benefit to our research. Therefore, by combing the existing literature, we define digitalization in manufacturing as the transformation of business processes such as design, production, and warehousing based on newly formed knowledge obtained through digital methods, which collect data and store them as crucial factors, thus gaining comparative advantage and creating new value. This definition is not limited to early or new digital technologies, but it applies to the analysis in this paper as an inclusive concept, so what we investigate about digital technologies in this paper regards generic digital technologies, as represented by ICT.
As for the research on the impact of digitization on carbon emissions, the existing literature is mainly carried out at the national or regional level [
7,
18,
19,
20]. The relationship between digital technology and energy consumption is complex. The development of digital technology has both positive and negative effects on the environment [
7,
18]. Haseeb et al. investigated the impact of ICT on the environment based on the panel data of BRIC countries and found that ICT positively contributes to environmental quality through internet development and technological progress [
21]. However, Shvakov and Petrova used the data from countries with high digital competitiveness and found that economic digitization will impose a more significant burden on the environment by improving economic expansion and energy consumption [
20]. Furthermore, based on China’s provincial or urban panel data, some researchers have concluded that the development of the digital economy could significantly reduce urban carbon emissions by improving energy structures or promoting technological progress [
22]. Some studies also suggest that the transformation cost brought on by digitization and the consumption of power resources may increase carbon emissions [
23]. In light of the effects of two different directions, some scholars have also found that there is an inverted U-shaped relationship between carbon emissions and ICT [
24]. However, due to restrictions during data acquisition on digitization, the existing studies on the environmental effects of manufacturing’s digitization are mainly qualitative analyses or professional technical reports [
25,
26], which lack empirical examination.
Compared with developed countries, developing countries’ imperfect infrastructure and a relatively late start in the digital economy have led to the immature development of digital technology, which results in the digital dependence on developed countries [
27]. Furthermore, technology transfer accompanied by industry undertakings will also lead digital input to flow inward [
28]. Therefore, a considerable part of developing countries’ digitalized input to the manufacturing industry comes from developed countries. Compared with the digital input from home, the digital input from developed economies has both high technical content and monopoly characteristics [
27]. Meanwhile, the digital input inevitably contains part of the digital investment, which could significantly improve the technical level of manufacturing enterprises but may also lead to the problem of cost increase [
29]. Especially when a digital investment comes from other economies, it shall also possess the attributes of foreign direct investment. Existing studies have usually regarded the impact of FDI on the environment as positive (Pollution Halo Hypothesis) or negative (Pollution Shelter Hypothesis) and generally consider a linear relationship between FDI and the carbon emissions of host countries and prove the above two hypotheses based on this [
30,
31]. However, when the investment has digital elements, the relationship between them may become more complicated. As an investment with highly knowledge-intensive characteristics, digital investment includes the integration of advanced manufacturing technology, intelligent technology, and information technology, and possesses a significant technology spillover effect [
32]. Simultaneously, considering the cost brought on by digital transformation [
33] and the productivity paradox of information technology [
34], the Pollution Haven and Pollution Halo effect may both be reflected in digital investment. Therefore, there may be a complex nonlinear relationship between digital input from developed countries and carbon emissions in developing countries. To investigate the environmental effects of input digitization more comprehensively, this paper will further study the relationship between digital input from developed countries and the carbon emission intensity of developing countries.
Therefore, after combing the relevant literature, this paper finds that, first, studies on the impact of digitalization on carbon emissions are mainly conducted on the national or regional level, and few scholars focus on the impact of digitalization on carbon emissions in the manufacturing sector, while the way digital technology represented by ICT works on the manufacturing industry is quite different. Even if the research is performed, the scholars only analyze it qualitatively and lack empirical studies. Second, in the manufacturing sector, large quantities of papers have examined the economic effects and industrial upgrading effects of ICT, but there is a dearth of studies on the environmental effects of digitalization. Third, the existing literature pays less attention to the differences between the impact of different sources of input on production and the environment. The inputs are usually regarded as a whole while conducting the research. Therefore, based on the deficiencies in the above research, this paper takes carbon emission intensity as the research object and uses the empirical method to investigate the impact of input digitization on manufacturing’s carbon emissions from multiple perspectives. The marginal contribution of this paper is as follows: Firstly, by focusing on the lack of quantitative research on the environmental benefits of digitization, this paper empirically examines the carbon reduction effect of input digitization in manufacturing for the first time and proves it through rigorous intermediary mechanisms, which provides empirical support for theoretical research on the promotion of carbon emission reduction with digitization. Secondly, the paper not only tests the carbon emission reduction effect of input digitization on the global level but also verifies the relationship between digital input and the carbon emission intensity of host countries from the perspective of developing countries, which traces the impact of different sources of input and clarifies the path for carbon emission reduction from different angles. Thirdly, the paper conducts various heterogeneity analyses from multiple angles and examines the dynamic effects and spillover effect of digital inputs’ impact on carbon emission reduction to enhance the comprehensiveness and robustness of the research.
3. Theoretical Analysis and Research Hypothesis
Generally speaking, the input digitization of manufacturing has significantly positive effects on the environment through the following three mechanisms:
The first mechanism is to achieve carbon reduction by improving production efficiency. With the introduction of digital technology, data acquisition, storage, and transmission costs are significantly reduced [
35]. Manufacturing enterprises can arrange production activities more reasonably based on data and information. According to the research of Bartel et al., the use of ICT in the manufacturing industry will improve the efficiency of all stages of the production process by reducing setup times, run times, and inspection times. Therefore, digital technology will realize the reorganization of business processes, and significantly improve production and operation processes, hence reducing management costs and improving production efficiency [
36], which is consistent with the logic of Process Reengineering Theory. After combining traditional production factors such as capital and labor with data, this new element may achieve several times greater efficiency than the previous production capacity [
37], thus reducing the capital and resource consumption under the same production quantity and leading to a reduction in the resource consumption of per unit output. Previous studies have also proved that the improvement of TFP has a significant inhibitory effect on carbon emissions because of efficient resource utilization [
12,
38]. Specifically, the increased productivity and rationalization of production organization resulting from digital inputs help to direct the efficient allocation of resources, which leads to increased energy efficiency and lower carbon emissions [
39]. In addition, digital technologies allow manufacturing enterprises to track consumers’ personalized needs and build exclusive databases and then carry out personalized product customization and marketing from the consumer side [
40]. As mentioned above, digital technologies have greatly reduced setup costs, while setup costs are a bigger fraction of unit costs for customized products, so the reduction in unit costs due to a reduction in setup time is greater for customized products than for commodity products [
36]. This helps manufacturing enterprises shift from the large-scale standardized production of homogeneous products to a customized production mode that meets the heterogeneous needs of consumers, improving the added value as well as production efficiency [
40]. Furthermore, the transformation of the production mode also greatly reduces the resource waste caused by invalid inventory and benefits the environment. This means that the transformation of the production mode brought about by digital input is different from the simple scale expansion caused by traditional technological innovation, but it produces the environmentally friendly innovation of production and organization. Moreover, according to Technical Innovation Theory, innovation is the re-combination of production factors, which means introducing a new combination of production conditions and production factors that did not exist before into the production system. The introduction of ICT makes data a completely new production factor. Massive amounts of data are gathered in the process of economic development, and they help traditional production factors to achieve product innovation and production method breakthroughs via the fast feedback of market information and process reorganization [
41]. Furthermore, the data’s characteristics of easy storage and copying allow them to break through the limitations of existing production factors and provide support for enterprises to achieve product innovation [
4]. Innovation in products and production leads to the improvement of production efficiency, which in turn results in efficient use of resources and carbon neutralization.
The second mechanism is to achieve carbon reduction by improving energy efficiency. Firstly, the introduction of digital technology helps collect massive amounts of market information. The fast feedback of demand information can help enterprises realize the constant adjustment of production structures to manage energy consumption accurately and improve energy efficiency [
42]. Secondly, what needs to be mentioned is that energy has high substitution flexibility, and the possibility lies in the substitution between energy and other production factors. In particular, capital has a considerable substitution space for energy, especially physical capital, showing a stronger substitutive relationship. Furthermore, with the increase in neutral technological progress parameters, the substitution becomes stronger [
43]. This indicates that physical capital, such as manufacturing machines with higher technology content, will achieve greater substitution for energy. The introduction of digital elements has improved manufacturing production technology and derived digital manufacturing equipment such as numerical control machines, which has greatly reduced carbon emissions by improving production efficiency and replacing a large number of polluting elements such as energy. In addition, as a new production factor, data have the characteristics of cleanness, low cost, and easy sharing compared with traditional resources, which helps to reduce pollution emissions further and change the energy consumption structure in the process of element substitution [
44]. Specifically, ICT offers various functions such as the substitution of virtual processes for physical processes, system monitoring with censoring tools, data transmission and processing, and efficient equipment control. ICT enhances the decoupling of economic activities from energy use through these functions [
45]. Finally, ICT enables manufacturing enterprises to access the internet and the native area network in the production process and optimize energy parameters through the local industrial energy network. Meanwhile, as mentioned above, the rationalization of the production process also brings about the optimization of the energy use structure [
46]. Martynenko believes that digitalization delivers the ecological modernization of production, which can save resources and secure industry and societal sustainability [
47].
The third mechanism is to achieve carbon reduction by improving information transmission. Negroponte believes that digital technology reduces the cost of information storage, replication, and transmission and improves the traceability of content in innovation activities [
35]. Other studies also prove that digital technology has dramatically reduced the cost of information transmission and improved the speed and scope of information dissemination [
48]. According to Spillover Theory, the spread of information will produce extensive externalities and spillover effects. For example, while testing the role of knowledge dissemination, Wiel et al. introduced technology spillover variables into the production function and found that information technology spillover can significantly and continuously promote the improvement of productivity [
49]. Furthermore, Absorptive Capacity (ACAP) Theory indicates that information technology can enhance firm innovation by facilitating the creation of patent inventions and the introduction of new products and services into the market. ICT can help create new knowledge by merging, categorizing, reclassifying, and synthesizing existing knowledge [
45]. Existing studies have shown that there is a significant substitution relationship between technological progress and carbon emissions [
50,
51]. Technological innovation brought about by digital elements can improve carbon emission efficiency by improving production efficiency and environmentally friendly technology [
52]. Inevitably, technological progress may also lead to the rebound effect, which creates new demands for energy due to industry expansion and partially counteracts energy efficiency improvement [
53,
54]. However, as mentioned above, the innovation in production patterns brought about by digital technology is different from the disordered expansion led by traditional technological advances. Furthermore, digital input tends to promote resource-saving, biased technological progress, and the development of environmental protection technology [
55], which indicates that the rebound effect of digital technology shall be relatively weak. Moreover, information technology can promote the transformation and upgrading of traditional industries by reducing material dependence and reshaping the production structure, which greatly promotes carbon reduction [
56].
With the internet and sharing networks, enterprises’ latest technologies for environmental protection or energy innovation will spread more quickly in the industry through the digital channel. Positive environmentally friendly technology spillover is conducive to reducing the industry’s resource consumption and carbon intensity [
57]. The information transparency caused by the development of digital technology will also make enterprises face greater competition and environmental protection pressure, which will push them to upgrade production and environmentally friendly technology by increasing innovation investment. Moreover, the utilization of ICT enables energy-using information to be shared within industries or regions, realizing the energy supply segment’s intensification, digitization, and refinement. This helps alleviate the information asymmetry between the energy supply side and the demand side and makes the energy supply process more reasonable, thus avoiding energy waste and overproduction and improving the efficiency of resource allocation and carbon emission [
48]. Therefore, the improvement of information transmission helps the industry enhance innovation and production efficiency, replace outdated polluting technology, and promote the development of environmental protection techniques, leading to a reduction in resource consumption and pollution emissions [
58].
Based on the analysis above, this paper puts forward the following hypothesis:
Hypothesis 1 (H1): Input digitization can reduce the carbon emission intensity of the manufacturing industry. Specifically, it is mainly achieved by improving production efficiency, energy efficiency, and information transmission.
Although input digitization has many positive effects on the environment, there may be differences in how digital input affects the environment in developing countries. Since the 1960s, developing countries have attracted and undertaken a large number of processing and manufacturing segments with high energy consumption and pollution from developed countries due to abundant energy, labor, land, and other resources, as well as relatively weak environmental regulations. The advanced production technology and management skills of multinational companies will also transfer in the process of the industrial undertaking [
28]. Due to the significant role of digital technology in reducing costs, expanding the market, and improving production efficiency [
59], digital technology and investment have become an essential part of international technology transfer, such as investment in computers, communication equipment, and the internet, as well as the provision of telecommunication and data processing services. Therefore, digital input from developed countries is likely to be a necessary investment or technology transfer provided by multinationals to carry out production activities more effectively. These production activities may belong to labor-intensive or resource-intensive manufacturing industries with high energy consumption. Therefore, this part of the digital input may not be conducive to environmental improvement in developing countries and may even cause further environmental pollution.
Furthermore, most countries have taken digital protection measures to protect domestic data flow and restrict data and core technology transfers to other countries. Digital barriers and upstream technological monopolies hinder foreign digital input’s knowledge spillover and technology upgrading effect, limiting its contribution to reducing carbon emissions [
60]. Furthermore, excessive industrial transfer accompanied by digital investment may even lead developing countries into a low-end lock-in and being captured in processing and assembling industries with low value-added [
61], further inhibiting the positive environmental effects brought on by digitization.
On the other hand, in consideration of developing countries’ own disadvantages, weak environmental regulations and limited knowledge absorption capacity will also restrict the environmental effects of digital input. The digital input provided by developed countries will improve the production efficiency and economic effect of the manufacturing industry in developing countries, resulting in a rapid expansion of industrial scale, and weak environmental regulations cannot effectively restrain the environmental pollution caused by the expansion of production activities, eventually leading to an increase in resource consumption [
62]. Moreover, the low level of human capital in developing countries leads to their weak ability to absorb advanced digital technologies, while not allowing a rapid shift in production patterns and leading to the rebound effect of technical progress. Therefore, the effects of technology improvement and carbon emission reduction are both restricted in the short term [
63]. In addition, local enterprises’ absorption of and research on transferred digital technologies will generate a large demand for talent, funds, and other innovative elements, which will lead to an increase in costs and, consequently, will not be conducive to environmental improvement in the short term [
64]. Therefore, this paper puts forward the following hypothesis:
Hypothesis 2 (H2): In the short term, digital input from developed countries has an inhibitory effect on reducing the carbon emission intensity of the manufacturing industry in developing countries.
Although digital input from developed countries may come with the motivation of carrying out manufacturing activities effectively, even the minimum technology transfer also provides a technical basis for the manufacturing industry in developing countries. Technology spillovers can be realized through the demonstration and imitation effect, personnel flow effect, and connection effect [
65] to improve environmental quality. Moreover, access to digital technology and internet applications will expose enterprises to foreign patents and advanced technologies, thus causing technology spillover and the driving effect to emerge in local enterprises [
66]. In addition, the improvement of production efficiency brought on by digital input and the companying industrial transfer helps developing countries to embed in the global value chain, which will have a positive scale effect, spillover effect, and competition effect on the domestic manufacturing industry [
61]. It will promote independent R&D innovation and the industrial upgrade of developing countries, consequently having a beneficial influence on the environment in the long term. Simultaneously, the introduction of digital technology may be the beginning of local digital industry, driving the emergence and development of industries such as computer manufacturing, telecommunications services, and other emerging industries, consequently improving the digital level of the domestic manufacturing industry and producing positive environmental effects. Moreover, in the long run, when the R&D investment of enterprises reaches a certain level, digital technology will lead to the rapid improvement of innovation efficiency. As intellectual assets, digital technologies such as digital software and information management systems have meager marginal costs [
67]. After being absorbed and mastered by local enterprises, they can be used cheaply by branches or put into technology transactions with other enterprises. They will produce a significant technology spillover effect and promote technological progress and digital transformation, thus realizing the elimination of pollution elements. Based on the above analysis, this paper puts forward the following hypothesis:
Hypothesis 3 (H3): With the passage of time, the effect of developed countries’ digital inputs on reducing carbon emissions in developing countries’ manufacturing industries will continue to increase and eventually exceed its environmental inhibitory effect, which makes the digital input from developed countries and the carbon emission intensity of developing countries an inverted U-shaped, nonlinear relationship.