Assessing Embodied Carbon Emission and Its Intensities in the ICT Industry: The Global Case

With the intensification of globalization, the information and communication technologies (ICT) sector, as one of the emerging industrial sectors, has played an important role in reducing carbon emissions in regional trade and improving the energy efficiency of traded products. This article uses a multi-regional input-output (MRIO) model to explore the flow of embodied carbon emissions and embodied carbon emission intensities in the ICT sector triggered by trade in 15 major economies around the world from 2000 to 2014. The article further discusses the emission-reduction strategies of each ICT sector. The results show that: (1) The overall embodied carbon emissions of ICT have demonstrated a significant downward trend, but the proportion of embodied carbon emissions caused by trade is increasing; (2) The embodied carbon emissions in the ICT sector of most developed countries in 2014 are significantly lower than in 2000, but in developing countries such as China and India they show an upward trend during this period; (3) The ICT sector’s export embodied carbon emission intensity in Indonesia is much higher than that of other countries, and the embodied carbon emissions of ICT exported by Mexico have increased significantly; (4) The manufacture of computer, electronic and optical products (S1) is the main contributor to the ICT sector’s embodied carbon emissions, while import embodied carbon emission intensities among publishing activities (S2) are increasing significantly in most countries. The conclusion of this paper has important implications for how to reduce ICT’s embodied carbon emissions of major countries.


INTRODUCTION
With global warming and frequent disasters, countries are increasingly starting to realize the severity of climate change (Sun et al., 2020;Hou et al., 2021). Studies have shown that carbon dioxide (CO 2 ), as the most important greenhouse gas, has always been the main influencing factor of global warming, and has caused the global temperature to rise by nearly 1 • C since the preindustrial era (Peters et al., 2019;Tiwari et al., 2021). The continuous increase in CO 2 emissions is the result of an increase in global energy demand associated with the growth in global trade (Li and Hewitt, 2008;Lu et al., 2020). Peters and Hertwich (2008) indicate that carbon emissions caused by globally traded products accounted for 21.5% of global emissions. Wu et al. (2020) also point out that the transfer of emissions embodied in interregional trade comprises around 40% of global carbon emissions.
Many scholars are also actively exploring corresponding emission reduction strategies, including the use of clean energy or renewable energy to replace fossil fuels (Dong et al., 2017, 2018, Dong K. et al., 2020, optimizing the industrial structure (Dong et al., 2019, Dong B. et al., 2020Zhu and Zhang, 2021), and increasing energy efficiency (He et al., 2021;Xu and Tan, 2021). Some scholars have also proposed that technological innovation can help industrial sectors achieve low-carbon development while taking into account economic improvement (Zhang et al., 2016;. As a new type of industry, the information and communication technologies (ICT) industry can partially replace high-energy-consuming products, and its success in this respect and its ability to reduce carbon emissions and use energy efficiently have received widespread attention from scholars. Moyer and Hughes (2012) believe ICT can improve economic productivity, reduce energy intensity, and exert downward pressure on renewable energy costs. Global Energy Statistical Yearbook (2014) also points out that ICT have an important role to play in reducing the energy intensity and increasing energy efficiency of the economy.
However, ICT are facing issues of uneven development among regions, and even a digital divide (Anwar, 2019;Wang et al., 2021). The process of globalization and the imbalance of regional development will cause the rapid promotion of international trade. The ICT sector will flow from regions with developed industries to regions with immature ICT technology. Vogiatzoglou (2009) points out that exports of ICT products have persistently accounted for 15% of global commodity exports in the past decade. Meanwhile, trade products have also triggered the flow of carbon emissions. Since the establishment of the World Trade Organization (WTO) in 1995, policy makers have considered the impact of international trade on the economy, and have increasingly investigated its impact on the environment (Honma and Yoshida, 2020). As an important component of international trade, ICT not only promotes economic growth, but also causes a large amount of carbon emission transfer and flow (Azam et al., 2021). With the liberalization of international trade, some scholars have also highlighted an increasing transfer of CO 2 emissions from developed countries to developing countries (Schaeffer and de Sá, 1996).
Although many scholars have studied the relationship between the ICT industry and carbon emissions (Moyer and Hughes, 2012;Amri et al., 2019;Shahnazi and Dehghan Shabani, 2019;Zhou et al., 2019;Nguyen et al., 2020), few scholars have conducted in-depth research on carbon emissions caused by the flow of products in the ICT sector caused by global trade. As the ICT industry is one of the important global development sectors in the future , it is necessary to measure its embodied carbon emissions flow and embodied carbon emission intensities, and provide corresponding cross-regional emission reduction strategies. Therefore, this paper uses a multi-regional input-output (MRIO) model to explore the flow and network relationships of embodied carbon emissions and intensities by trade in the ICT sector of the world's major economies, and proposes emission reduction strategies for each country.
Above all, three main features distinguish this study from previous studies and contribute to filling the gap in the literature. First, we explore the embodied carbon emissions of the global ICT industry and the embodied carbon emissions of imports and exports of major economies, which is necessary to understand the developing trend and emission reduction direction of the global ICT industry; Second, we measure the mutual flow and intensities of embodied carbon emissions among major economies in the world, which can better evaluate the embodied carbon emissions in export and import, and conducive to the policy formulation and negotiation of cross-regional carbon emission reduction; Third, the embodied carbon emissions and intensities of various ICT industries have also been calculated by this study, which shows heterogeneity in the ICT sector's carbon emissions and provides a good reference for government to formulate emissionreduction strategies for enterprises. This paper is structured as follows: Section "Literature review" organizes the literature on the relationship between embodied carbon emissions, intensities and ICT sectors. Section "Methodology and data" presents the methodology and data. Section "Results and discussion" analyzes the global carbon emissions and intensities in 15 major economies. Section "Further discussion" provides further analysis of the emissions and intensities of each ICT sector. Section "Conclusions and policy implications" addresses the policy implications and concludes the paper.

LITERATURE REVIEW
Due to the rise of the ICT sector in recent years and the intensification of digital activities on a global scale, scholars have increasingly started to pay attention to research on the relationship between the ICT sectors and global embodied carbon emissions. This study intends to discuss the extant literature under two strands of research: (1) Studying the relationship between carbon emissions and ICT sectors; and (2) studying the relationship between carbon emission intensity and ICT sectors.

Carbon Emissions and ICT Sectors
Information and communication technologies contributes substantially to global greenhouse gas (GHG) pollutant emissions, but scant research has quantified the impact of ICT products on carbon emissions. From the perspective of the products themselves, the types of ICT products included are very complex, and traditional industries also include many ICT sector products, making quantification difficult (Teehan and Kandlikar, 2013). However, with the development of digital industries and the intensification of the global greenhouse effect, more and more scholars have started to pay attention to the impact of the ICT industry on carbon emissions. Some scholars analyze the embodied carbon emissions of the ICT sector of specific countries. Zhou et al. (2019) summarize China's embodied carbon emissions of ICT and point out that the ICT sector is far from being environment friendly while considering its embodied carbon impacts, which are dozens of times greater than the direct impacts. Shahnazi and Dehghan Shabani (2019) analyze the correlation between the development of the ICT industry and carbon emissions in Iran's provinces, and their results show that an increase in ICT in a province first leads to an increase and then a decrease in CO 2 emissions in other provinces. Due to the prominence of the ICT sectors of some developed countries and the associated advantages of ICT for economic development, these countries have become a good target for many scholars to study ICT and embodied carbon emissions growth (Malmodin et al., 2014;Malmodin and Lundén, 2016). For example, Malmodin and Lundén (2016) conduct a detailed study of embodied carbon emissions of the ICT sector in Sweden from 1990 to 2015. They point out that although the embodied carbon emissions of the ICT sectors in that country have decreased since their peak in 2010, embodied carbon emissions from their manufacturing processes abroad are the largest source of ICT-related carbon emissions in Sweden.
Some scholars have expanded their research to the global region and analyze the relationship between the development of the global ICT industry and the impact of carbon emissions, and quantified the potential reduction of emissions in ICT (Moyer and Hughes, 2012;Bastida et al., 2019). Malmodin and Bergmark (2015) explore possible reductions globally within a 2030 timeframe, and predict a potential reduction of total GHG emissions by 2030 of about 8 Gtonnes CO 2 or 12% of global GHG emissions due to ICT solutions. Asongu (2018) investigates how ICT counteracts the negative influence of globalization on CO 2 emissions in his study of 44 Sub-Saharan African countries over the period 2000-2012. The author finds that ICT can be employed to dampen the potentially negative effect of globalization on environmental degradation through CO 2 emissions. Malmodin and Lundén (2018) analyze carbon emissions of the global ICT sectors for 2010-2015, and point out that the ICT sectors have turned their previously growing carbon footprints into shrinking ones.
However, related research on the embodied carbon emissions of the ICT sector caused by global trade is relatively insufficient. With the rapid development of globalization and the systematic construction of global industrial chains and value chains, this will be a very important component.

Carbon Emission Intensities and ICT Sectors
Energy intensity and carbon emission intensity are two important indicators for studying the environmental impact of industries. Some scholars are concerned about the interaction between the ICT industry and energy intensity. And the Global Energy Statistical Yearbook (2014) points out that the ICT industry plays an important role in reducing energy intensity and improving industrial efficiency, but few scholars pay attention to the energy intensity of the ICT industry itself. Wang and Han (2016) study the impact of ICT on energy intensity in China, and point out that the impact of ICT investment in energy intensity is significantly negative in western and central regions in China.
However, relatively speaking, the intensity of carbon emissions caused by energy is more representative and has been widely studied by scholars (Pan et al., 2019;Liu et al., 2021a). However, research on carbon emission intensity and ICT-related research is relatively rare, and we can only find some relevant evidence in some research on the industrial framework or emerging technology industry. For example, Liu et al. (2021b) have shed light on the effect of Artificial Intelligence (AI) on carbon emission intensity. Their results show that AI tends to have a greater effect in reducing carbon emission intensity in the labor-intensive and tech-intensive industries. Chen et al. (2019) investigate the impact of ICT on CO 2 emission intensity by using Internet penetration and mobile phone penetration as proxies in China, and point out that the reduction effect of Internet penetration on CO 2 emission intensity in China's eastern and central provinces is more obvious. In contrast with the reduction effect of Internet penetration on CO 2 emission intensity in China's eastern provinces, it is greater in China's central provinces.
Few scholars have demonstrated concern about the intensive impact of global embodied carbon emissions. Although Jiang and Liu (2015) quantify the flow of embodied carbon emissions from the global ICT industry and trade, they do not highlight the impact of trade intensity under the global value chain system, and the research period is relatively lagging. This article will build a research framework on embodied carbon emissions and carbon emission intensities of global ICT trade in the context of global value chains. The paper will also analyze the volume and technical advantages and disadvantages of major economies' embodied carbon emissions, and recommend cross-regional emission-reduction strategies.
Furthermore, we will absorb the research experience of the above-mentioned scholars, based on previous studies on carbon emissions and carbon emission intensities in the ICT industry, and expand the analysis of embodied carbon emissions and the measurement of embodied carbon emission intensities from the perspective of global value chains in major economies around the world. Furthermore, this study will explore in detail the dynamic changes of the embodied carbon emissions and intensities of each ICT sector.

Multiregional Input-Output Model
The basic methodology used in this study is the multi-regional input-output (MRIO) model from the perspective of both the environment and the global value chain. This model put forward by Moses (1955), can depict economic links among regions. The row equivalent of the MRIO can be expressed as: Where x r i is denoted as the total output of sector i in region r. a rs ij is the direct requirement coefficient of sector i in region r for sector j in region s, which can be calculated by a rs ij = t rs ij /x s j , and t rs ij is the intermediate use of sector i in region r for sector j in region s; y rs i is the final demand of region s by sector i in region r. And the basic model of MRIO can be expressed in matrix form: And we define F = f r i 1×RN as the carbon emission intensity generated by the unitary output of sector i in region r. Embodied carbon emissions of an economy can be expressed as: , the embodied carbon emissions from sector i in region r to sector j in region s can be calculated as: Therefore, the element of domestic embodied carbon emissions diagonal matrix DEC= DEC r i RN×RN can be calculated as: The element of export embodied carbon emissions from sector i in region r to sector j to region s can make up the export embodied carbon emissions matrix EEC = EEC rs ij RN×RN , which can be represented as: The element of import embodied carbon emissions from sector i in region r to sector j to region s can make up the import embodied carbon emissions matrix IEC = IEC rs ij RN×RN , which can be represented as: Next, we define V = v r i 1×RN as the value-added coefficients generated by the unitary output of sector i in region r, which can be calculated by v r i = va r i /x r i , where va r i denotes the value added of sector i in region r. According to the global value flow analysis method (Fan et al., 2021), the element of embodied value-added cross countries matrix EVA = EVA rs ij RN×RN can be calculated as: And the element of embodied carbon emission intensity of sector i in region r flowing into sector j in region s can make up the matrix ECI = ECI rs ij RN×RN , which can be calculated as: Similarly, the domestic embodied carbon emission intensity of sector i in region r can make up the diagonal matrix DECI = DECI r i RN×RN , which can be calculated as: The export embodied carbon emission intensity of sector i in region r can make up the diagonal matrix EECI = EECI r i RN×RN , which can be calculated as: And the export carbon emission intensity from region r to region s can be calculated as: The import embodied carbon emission intensity of sector i in region r can make up the diagonal matrix IECI = IECI r i RN×RN , which can be calculated as: And the import carbon emission intensity from region r to region s can be calculated as:

Data
Here we will use the MRIO model to measure the embodied carbon emissions of global ICT sectors. The World Input-output tables coming from the World Input-output database (WIOD) are released in 2016. The 2016 version covers 43 countries for the period 2000-2014 (Timmer et al., 2015). We have selected 15 major countries as the research objects of this study, and other countries are classified as other countries (ROW) 1 , the country names and abbreviations are shown in Table 1. In order to make comparative prices, the multiregional input-output tables are processed at constant 2010 prices, and use a double-shrink method to select the price indexes for each sector's products (Miller and Peter, 2009). The price indexes are selected as the consumer price index of countries from the World Bank (2020). And the data of the 56 sectors are classified according to the International Standard Industrial Classification Revision 4 (ISIC Rev.4). We select five sectors as ICT departments, including one ICT manufacturing sector and four ICT service sectors, which are shown in Table 2.
The direct carbon emissions come from Corsatea et al. (2019), and the data fit perfectly with WIOD and are applied to this study.

Domestic, Export, and Import Embodied Carbon Emissions and Intensities of Major Countries
Embodied Carbon Emissions  (Wang H. et al., 2020;Sahoo et al., 2021). Nevertheless, the growth of these countries' ICT industries is reflected in their domestic and export products, which have increased embodied carbon emissions. The China Academy of Information and Communications Technology  Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import Domestic Export Import

Embodied Carbon Emission Intensities
The carbon emission intensities of trade reflect to a certain extent the technological differences in a country's trade products and can provide intuitive emission reduction strategies. From the perspective of embodied carbon emission intensities (Figure 1 This reflects to a certain extent that the progress of production technology has promoted the process of carbon emission reduction in the ICT sector. However, it must be noted that India's domestic embodied carbon emission intensity and export embodied carbon emission intensity of ICT have both increased slightly, which requires the attention of local government departments. From the selected countries, Indonesia's export embodied carbon emission intensity of ICT is much higher than that of other countries. Even though there is a significant decrease in 2014 compared to 2000, the export embodied carbon emission intensities of 0.19 kg/USD should still receive the attention of the authorities. For example, key considerations for Indonesia will be to find ways of optimizing the energy structure of products in the export sector and the industrial energy consumption intensity of its ICT sector. The country will also need to improve its energy poverty environment (Setyowati, 2021). In 2000, Russia's domestic and export embodied carbon emission intensities of ICT are relatively high, reaching 0.15 kg/USD and 0.19 kg/USD respectively, but in 2014 the export embodied carbon emission intensities decrease to 0.07 kg/USD and 0.09 kg/USD, respectively. This shows that the export of ICT products may not reduce the carbon intensity to achieve emission reduction due to technical restrictions, which has caused the significant spillover effect 2 of export carbon emissions (Liu and Liu, 2019;Wang and Li, 2019). Relatively speaking, the embodied carbon intensities of European countries, such as Spain, France, and the United Kingdom, are at the low level. The ICT's embodied carbon intensities of their exports to other countries are all about 0.01 kg/USD, reflecting the technological advantages and emission reduction driving forces of ICT products in these European countries.

FURTHER DISCUSSION
After investigating the embodied carbon emissions and embodied carbon emission intensities of the ICT sector in various countries, we find that there is relatively little understanding of the internal ICT sector, and the internal ICT manufacturing products or services that have caused the increase in embodied carbon emissions are worthy of attention. Therefore, we continue to explore the impact of each ICT sector.   embodied carbon emissions, from 7.7 and 2.8% in 2000 to 5.6 and 2.3%, respectively. Figure 5 shows the embodied carbon emissions and intensities of ICT sectors in each country in 2000 and 2014. From a quantitative point of view, the embodied carbon emissions of the sector of manufacture of computer, electronic and optical products (S1) in most countries are much higher than those of other sectors, and the amount in 2014 is less than that in 2000. The import and export embodied carbon emissions of manufacture of computer, electronic and optical products (S1) in China have both risen significantly, and Brazil's import embodied carbon emissions have increased. The embodied carbon emissions of manufacture of computer, electronic and optical products (S1) in other countries show a downward trend from 2000 to 2014. What is more anomalous is that in 2000, the embodied carbon emission of the telecommunications (S4) sector is the highest in India at 0.141 Mt. In 2014, the computer programming, consultancy and related activities; information service activities (S5) sector has the highest embodied carbon emissions at 0.806 Mt.

Embodied Carbon Emissions and Intensities of Each ICT Sector in Various Countries
From the perspective of carbon emission intensities, the export carbon emission intensity of the publishing activities (S2) sector in many countries has increased significantly by 2014. For example, Australia increases its export carbon emission intensity of the S2 sector from 0.02 kg/USD in 2000 to 0.15 kg/USD in 2014. Germany increases from 0.03 kg/USD in 2002 to 0.08 kg/USD in 2014. The import carbon emission intensities of the other four sectors have shown varying degrees of downward trends. The export carbon emission intensities of ICT sectors in most countries have shown a significant downward trend, reflecting the rapid progress of ICT products exported by these countries. For example, the export carbon emission intensity of China's manufacture of computer, electronic and optical products (S1) sector decrease from 0.12 kg/USD to 0.05 kg/USD in 2000, and the export carbon emission intensities of the S1 sector in the United States decrease from 0.06 kg/USD to 0.02 kg/USD. India and Indonesia are relatively special; their sector of manufacture of computer, electronic and optical products (S1) has shown an upward trend in export carbon emission intensity, indicating that the export technology of its ICT manufacturing sector has not yet shown much carbon emission reduction potential.

CONCLUSION AND POLICY IMPLICATIONS Conclusion
This article explores the embodied carbon emissions and carbon emission intensities of 15 major countries around the world, and further discusses the changes in various industrial sectors in each country. Accordingly, the main conclusions are as follows: ( and intensities of various ICT industry sectors, the sector of manufacture of computer, electronic and optical products (S1) is the main embodied carbon component of the ICT sectors, both globally and locally for each country. However, with the passage of time, the percentage of embodied carbon emissions in the ICT manufacturing industry (S1) has been decreasing year by year. In comparison, the embodied carbon emissions of the ICT service industries (S2-S5) have been increasing year by year. As far as countries are concerned, the embodied carbon emissions of India's ICT service industry are higher than those of the manufacturing industry, which is reflected in the sectors of telecommunications (S4) and computer programming, consultancy and related activities; information service activities (S5). In addition, the import embodied carbon emission intensities of the publishing activities (S2) sector in most countries have increased significantly, reflecting the technical disadvantages of imported products in this sector. The embodied carbon emission intensities of exports of various countries and ICT sectors have shown a significant downward trend.

Policy Implications
(1) Although the embodied carbon emissions of the global ICT sector are not high, the sector has already demonstrated its technological advantages and has shown a special advantage in reducing carbon emissions. However, the promotion of trade carbon emissions needs to attract the attention of local governments. It should be noted that the control of the embodied carbon emission intensities of exports can reduce the growth of embodied carbon emissions, and that the ICT sector can further reduce its carbon emissions.
(2) Developing countries such as China, India, and Brazil need to pay special attention to the significant increase in embodied carbon emissions from the ICT sector. China needs to pay attention to the embodied carbon emissions of the export sector, while India needs to pay attention to the embodied carbon emission intensity changes of exports, and realize that maintaining a downward trend is conducive to achieving emission reductions. This requires above countries to enhance the competitiveness of their enterprises in carbon reduction technologies. The Indonesian government needs to focus on the high embodied carbon emission intensity of exports, which is much higher than other countries and is the main driving force behind the embodied carbon emissions of the country's ICT sector.
(3) From the perspective of regional coordinated emission reduction, the embodied carbon emissions of the ICT sector in the Asia-Pacific region remain high, mainly in China, Japan, and South Korea. A coordinated development policy in the Asia-Pacific region could help accelerate the process of regional carbon emission reduction. The growth of embodied carbon intensity in Mexico's exports and the high level of embodied carbon intensity in Indonesia have both reminded local governments to strengthen the regulation of ICT exports, improve the energy efficiency of these products, and learn more from the experience of European countries in achieving low-carbon intensity development. (4) Regarding the ICT sector, carbon emissions caused by trade in the manufacturing industry account for a relatively high proportion, while carbon emissions caused by trade in the service industry are on the rise. This requires countries to optimize the export carbon intensities of ICT-manufactured products and strengthen supervision of the export of ICT service industries. The Indian government should pay particular attention to the export of ICT services, mainly the embodied carbon emissions of telecommunications (S4) and computer programming, consultancy and related activities; information service activities (S5).

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
XD: conceptualization, validation, investigation, resources, data curation, and funding acquisition. QJ: conceptualization, writing -review and editing. JW: methodology, software, visualization, writing -original draft, writing -review and editing. All authors contributed to the article and approved the submitted version.