The Comprehensive Impact of Outward Foreign Direct Investment on China’s Carbon Emissions

Abstract

Outward foreign direct investment (OFDI), as an important carrier of global technology and industrial transfer, will significantly impact the home country’s environment. Therefore, using data from 30 Chinese provinces gathered between 2004 and 2019, we empirically analyze the impact of OFDI on China’s carbon emissions across two dimensions: total carbon emissions and carbon emission efficiency. In addition, when the previous studies explored the impact of OFDI on carbon emissions, there were few studies on the synergistic emission reduction effect of OFDI. Therefore, based on sorting out previous research, we incorporated OFDI, technological progress, industrial structure upgrading, international trade, and carbon emissions into the same analytical framework. Based on the classic fixed model, we introduce the interaction term further to explore the synergistic emission reduction effect of OFDI. Our model suggests that OFDI has increased total carbon emissions, but the associated reverse technology spillover has improved carbon emission efficiency. We also found a synergistic emission reduction effect between OFDI and technological progress, international trade, and industrial structure upgrading. This synergistic effect suppresses the growth of total carbon emissions and improves carbon emissions efficiency. Robustness testing confirmed these results. This research also provides a relatively novel perspective for China to achieve the goals of “carbon peaking” and “carbon neutrality”.

1. Introduction

Rising levels of greenhouse gas emissions melt glaciers, cause rising sea levels, increase physical morbidity and mortality, and result in more extreme weather, each of which threatens global sustainable development. Reducing carbon emissions is essential to improving the global climate [1] (Gong et al., 2022). As the largest developing country in the world, China’s rapid economic growth and associated energy consumption have caused a massive increase in carbon dioxide emissions, severely impacting the environment and people’s lives. Global energy-related carbon dioxide emissions rose 6 percent to 36.3 billion tons in 2021, a new record, according to an analysis by the International Energy Agency (IEA). China alone emits more than 11.9 billion tons of carbon dioxide, 33% of the global total. For that reason, China’s progress in reducing emissions has important implications for China and the possibility of improving the global climate.

Previous studies have conducted extensive analyses of factors affecting CO₂ emissions. They found that economic growth, industrial structure, urbanization, technological progress, and energy consumption are important factors affecting carbon emissions. However, many scholars believe that technological progress is the main way to reduce carbon emissions and that technological progress comes from R&D input and FDI [2] (Jiang et al., 2020). Scholars have also conducted rich discussions on the relationship between FDI and carbon emissions. However, most of these explorations focus on inflow FDI (IFDI) and there are few studies on the impact of outflow FDI (OFDI) on carbon emissions. Furthermore, among researchers there is no consensus on the impact of OFDI on carbon emissions. Moreover, the impact path of OFDI on carbon emissions is not yet clear. According to some, an increase in OFDI promotes the development of secondary industries with high energy consumption and high carbon emissions in the home country. In other words, OFDI may expand the scale of goods produced in the home country destined for consumption in other countries (the “polluted paradise” theory) [3] (Zhang et al., 2021). Therefore, OFDI will exacerbate the growth of carbon emissions in the home country. Other scholars believe that the home country can obtain advanced technology in foreign markets through OFDI, which is conducive to improving the domestic technology level and energy utilization efficiency, thereby improving the environmental status of the country (“pollution halo” theory). At the same time, the home country can also transfer its high energy consumption and high carbon emission industries outward through OFDI, thereby reducing domestic carbon emissions [4] (Yang et al., 2021).

In addition, some studies concerning the motives of transnational investment show that improving the technological level of emerging and developing countries is inseparable from the growth of OFDI. The investment flows from developing countries toward developed countries is essentially a strategic investment behavior aimed at absorbing technology [5] (Hao et al., 2021). For China, as China’s economy has grown, the scale of China’s overseas investment has expanded. According to the <World Investment Report 2021> released by the United Nations Conference on Trade and Development (UNCTAD), China’s OFDI was the highest in the world, reaching USD 153.7 billion starting in 2020. Therefore, OFDI, as an important carrier of global technology and industry transfer, will significantly impact the success of China’s carbon emission reform [6] (Zhang et al., 2022b).

It should be noted that most of the prior studies on carbon emissions focus on the influencing factors and prediction analysis of the total amount of carbon emissions, often ignoring the impact on carbon emissions efficiency. However, carbon emission efficiency is an indicator that can reflect a region’s carbon emission performance level. Improving energy utilization efficiency and productivity represented by carbon emission efficiency is an important way to promote economic decarbonization [7] (Chen et al., 2022). Therefore, improving carbon emission efficiency is an effective means to achieve “carbon peaking” and “carbon neutrality” in the long run.

In this context, a comprehensive examination of the impact of China’s OFDI on carbon emissions will be meaningful. Therefore, this study uses data from 30 provinces in China gathered from 2004 to 2019 to empirically analyze the impact of OFDI on China’s carbon emissions across two dimensions: total carbon emissions and carbon emission efficiency. In addition, when analyzing the impact of OFDI on carbon emissions, previous studies often ignored the synergistic emission reduction effect between OFDI and other factors. Therefore, based on sorting out previous research, this paper incorporates OFDI, technological progress, industrial structure upgrading, international trade, and carbon emissions into the same analytical framework. Furthermore, by introducing the interaction terms between OFDI and these factors, we further explored the synergistic emission reduction effect of OFDI.

This work has the following contributions to the current literature. First, most existing studies focus on the impact of OFDI on total carbon emissions. This study compares and analyzes the impact of major explanatory variables such as OFDI on carbon emissions from the two dimensions of total carbon emissions and carbon emissions efficiency. Second, this paper introduces the interaction terms between OFDI and other carbon emission factors. Additionally, it measures the synergistic emission reduction effect of OFDI and technological progress, industrial structure upgrading, and international trade through the moderating effect (interaction term). Some contributions have been presented to improve the mechanism and theoretical analysis of the impact of OFDI on carbon emissions. The comparative analysis results of this study can provide valuable references not only for China but also for policymakers in other emerging countries.

2. Literature Review and Research Hypothesis

While the relationship between economic growth and the environment has always attracted scholarly attention, there are still relatively few studies concerning the comprehensive impact of OFDI on carbon emissions.

Previous studies have coalesced around two opposing views concerning the impact of OFDI on carbon emissions. Some scholars believe that OFDI helps reduce carbon emissions. They believe that as the scale of OFDI expands, it can bring advanced production technology and management experience to improve the energy efficiency and productivity of the home country [8] (Wang et al., 2020). Second, the reverse technology spillover effect of OFDI is conducive to improving the home country’s green technology innovation capability and promoting the ecological environment and sustainable development [9] (Tan et al., 2021). Third, through OFDI, enterprises in the home country can transfer some high-polluting and energy-intensive production lines abroad, thereby reducing domestic carbon emissions [10] (Gao et al., 2018). Alternatively, other scholars believe that OFDI will increase carbon emissions. They argue that OFDI exacerbates environmental pollution by expanding the economic scale of the home country and increasing industrial output and energy consumption [11] (Hao et al., 2020). Second, in the fierce international market competition, to improve competitiveness, multinational enterprises have to reduce investment in improving energy efficiency in order to reduce costs. At the same time, enterprises may blindly pursue economies of scale, continuously increase investment in equipment, and expand production, which promotes the growth of the proportion of the secondary industry with high pollution and high energy consumption in the home country [12,13] (Hanif et al., 2019; Zhang et al., 2014). Third, if OFDI mainly concentrates on leasing and business services, it is often impossible to transfer high-polluting, high-energy-consuming industries to the home country [14] (Wang et al., 2019). As far as China is concerned, compared with Western developed countries, China’s current technological level and energy utilization efficiency are still relatively low. Therefore, the ability to absorb the reverse technology spillover of OFDI is limited. Thus, the economic expansion and the increase in the proportion of high-carbon-emitting industries caused by OFDI will bring more carbon emissions [15] (Yi et al., 2018). However, the reverse technology spillover of OFDI can improve the regional technology level, which in turn improves the regional carbon emission efficiency [16] (Pan et al., 2020). Accordingly, we propose Hypothesis 1:

H1: 

OFDI will promote the growth of total carbon emissions and increase the efficiency of carbon emissions.

In addition, scholars widely recognize technological progress as an important way to reduce carbon emissions. For example, Ref. [17] Guo et al. (2022) found that both the direct and indirect effects of technological innovation significantly suppressed the increase in carbon emissions. Similarly, Wang et al. (2022) Ref. [18] also believe that technological innovation (especially high-efficiency recycling technology, zero-carbon energy technology, and negative emission technology) is an important way to achieve carbon neutrality. Some scholars have further explored the relationship between OFDI, technological progress, and carbon emissions. The previous research believes that OFDI mainly promotes technological progress in the home country through the following ways, thereby reducing carbon emissions. First, the “endogenous growth theory” believes that technological innovation is the driving force for the transformation and development of enterprises. However, global R&D innovation activities are highly concentrated in advanced economies, while emerging and developing economies have limited innovation capacity and underinvestment in frontier technologies. Therefore, investing in advanced economies will create more access to frontier technologies, thereby promoting technological innovation in the home country [19] (Li et al., 2016). High-carbon companies can use technological innovation to treat pollutants emitted into the atmosphere (such as carbon decomposition and carbon capture) to achieve the goal of reducing carbon dioxide emissions. Secondly, OFDI is conducive to international strategic technology cooperation and mergers, acquisitions of multinational enterprises, and obtaining more technology spillovers. Improving enterprise energy efficiency and environmental protection technology must be promoted [20] (Wang and Wu., 2016). Third, OFDI can strengthen international human capital flows and enterprises can promote technological progress in home countries through human capital spillovers. To sum up, technological progress may restrain the growth of carbon emissions and there may be a synergistic emission reduction effect between OFDI and technological progress. Accordingly, we propose Hypothesis 2:

H2: 

Technological progress is conducive to restraining the growth of total carbon emissions and improving carbon emissions efficiency and OFDI and technological progress have a synergistic emission reduction effect.

In addition, scholars have also conducted much exploration on the impact of OFDI on the upgrading of industrial structures. For example, the “marginal industry expansion theory” proposed by [21] Kojima (1978) believes that OFDI can transfer industries or marginal industries that are already relatively disadvantaged in the country. It will help the investing country save resources and concentrate on developing initiatives and emerging industries with comparative advantages, thereby promoting the upgrading of the domestic industrial structure. Ref. [22] Cantwell and Tolentino (1990) proposed the “Technology Innovation Industry Upgrading Theory” that developing countries can promote the accumulation of technology in the home country through OFDI, improve the technological level of enterprises and then promote the upgrading of the industrial structure of the home country. At the same time, the upgrading of industrial structures is also another important factor affecting the level of carbon emissions. For example, Ref. [23] Yang et al. (2022) proposed that since China’s industrial structure entered a new stage in 2012, upgrading the industrial structure has had a pronounced inhibitory effect on carbon emissions. Therefore, building a green and low-carbon economic system is an important way for China to achieve the goals of “carbon peaking” and “carbon neutrality” [24] (Tong et al., 2022). In addition, Ref. [11] Hao et al. (2020) proposed that the reverse technology spillover effect of OFDI optimizes the domestic industrial structure, thereby reducing the domestic environmental pollution. Similarly, Ref. [25] Dong et al. (2021) proposed that multinational corporations in emerging economies cannot ignore investment in technologically underdeveloped countries, which will help transfer marginal industries in the home country and promote the upgrading of the industrial structure and sustainable development of the home country. Accordingly, we propose Hypothesis 3:

H3: 

The upgrading of industrial structures is conducive to restraining the growth of total carbon emissions and improving the efficiency of carbon emissions. There is a synergistic emission reduction effect between OFDI and industrial structure upgrading.

With the continuous improvement of globalization, the impact of OFDI on international trade has received extensive attention. Some scholars believe that OFDI will decrease after some domestic industries relocate outward, which will adversely affect the export and investment of the home country [26] (Al-Sadiq et al., 2013). Other scholars believe that OFDI can help enterprises enter new markets, which is beneficial to reduce the cost of imported products and obtain foreign advanced technology, thereby improving the home country’s competitiveness in trade and investment. Most studies believe that the reverse technology spillover effect and the marginal industrial transfer effect are the two main ways that OFDI affects the international trade level of the home country [27] (Ahmad et al., 2016). For example, the research results of [28] Li et al. (2021) show that OFDI can promote the improvement of the international trade status of home countries by promoting technological progress and improving the quality of trade networks and the OFDI impact on emerging countries is more significant than that of developed countries. In addition, a growing body in the literature indicates that international trade is another critical factor affecting carbon emissions [29] (Wang and Zhang., 2021). For example, [30] Wang and Yi. (2022) proposed that international trade can introduce advanced technologies, improve energy efficiency, and reduce carbon emissions. Conversely, [31] Jun et al. (2020) propose that the development of international trade expands the scale of domestic production, increasing a country’s pollution footprint. As far as China is concerned, [5] Hao et al. (2021) proposed that China has become an essential link in the global production chain. It is involved in the manual manufacture of many semi-finished and finished products. This contributes to an increase in carbon emissions. However, through international trade, Chinese companies can learn and imitate the technology of imported low-carbon products, promote the development of low-carbon products and technologies in China, and reduce China’s carbon emissions. In addition, [32] Chen et al. (2021) found that the trade products of most coastal provinces in China gradually shifted from early high-pollution products to the tertiary industry and high-tech industrial products. This shift has prompted a change in international trade from promoting carbon emissions to suppressing carbon emissions. Similarly, [33] Qi et al. (2020) proposed an “inverted U-shaped” relationship between international trade and carbon emissions. That is to say, international trade will increase carbon emissions initially, but when international trade reaches a critical value, it will suppress the increase in carbon emissions. To sum up, OFDI may promote the improvement of the international trade level of the home country through reverse technology spillover and the promotion of industrial structure upgrading. In addition, technological spillovers and energy efficiency gains from developed international trade may dampen carbon emission growth. Accordingly, we propose Hypothesis 4:

H4: 

International trade is conducive to curbing the growth of total carbon emissions and improving the efficiency of carbon emissions and OFDI and international trade have a synergistic emission reduction effect.

To sum up, scholars have engaged in a rich exploration concerning the possibility of reducing carbon emissions from different perspectives. In the long run, improving carbon emission efficiency is an important way to achieve the “dual carbon” goal. Still, existing research often only focuses on the impact of OFDI on total carbon emissions and there are few studies on the impact of OFDI on carbon emissions efficiency. In addition, the existing research on the synergistic emission reduction effect of OFDI and other factors is still insufficient. Therefore, this study comprehensively analyzes the effects of OFDI on total carbon emissions and carbon emission efficiency. Furthermore, we use the moderating effect (interaction term) to measure the synergistic effect of OFDI and other factors on carbon emissions. The attempts to contribute to the existing research are through these. Figure 1 is a roadmap of the synergistic emission reduction effects of OFDI.

3. Methodology Specification and Variable Description

3.1. Model Construction

To systematically analyze the impact of OFDI on the total carbon emission and carbon emission efficiency, this paper constructs the functional form of the correlation between OFDI, industrial structure upgrading, international trade, technological progress, and carbon emission (total carbon emission and carbon emission efficiency) as follows.

$$\mathrm{CO}{2}_{\mathrm{it}}=\mathrm{f}\left({\mathrm{OFDI}}_{\mathrm{it}},{ \mathrm{ISU}}_{\mathrm{it}},{\mathrm{TRADE}}_{\mathrm{it}},{\mathrm{TECH}}_{\mathrm{it}}\right)$$

(1)

$${\mathrm{CTFP}}_{\mathrm{it}}=\mathrm{f}\left({\mathrm{OFDI}}_{\mathrm{it}},{ \mathrm{ISU}}_{\mathrm{it}},{\mathrm{TRADE}}_{\mathrm{it}},{\mathrm{TECH}}_{\mathrm{it}}\right)$$

(2)

To alleviate the problem of heteroscedasticity, the variables of the above function (except CTFP and ISU) were converted to linear models by taking the logarithm. Therefore, Equations (1) and (2) can be rewritten as:

$$\mathrm{LnCO}{2}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{LnOFDI}}_{\mathrm{it}}+{\mathsf{\alpha }}_2{\mathrm{ISU}}_{\mathrm{it}}+{\mathsf{\alpha }}_3{\mathrm{LnTRADE}}_{\mathrm{it}}+{\mathsf{\alpha }}_4{\mathrm{LnTECH}}_{\mathrm{it}}+{\mathsf{\beta }\mathrm{X}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(3)

$${\mathrm{CTFP}}_{\mathrm{it}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1{\mathrm{LnOFDI}}_{\mathrm{it}}+{\mathsf{\gamma }}_2{\mathrm{ISU}}_{\mathrm{it}}+{\mathsf{\gamma }}_3{\mathrm{LnTRADE}}_{\mathrm{it}}+{\mathsf{\gamma }}_4{\mathrm{LnTECH}}_{\mathrm{it}}+{\mathsf{\beta }\mathrm{X}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(4)

To verify the research hypothesis and further test whether there is a synergistic emission reduction effect between OFDI and industrial structure upgrading, international trade, and technological progress, the interaction term between OFDI and these factors is introduced. This leads to the following model:

$$\begin{gathered} \mathrm{LnCO}{2}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{LnOFDI}}_{\mathrm{it}}+{\mathsf{\alpha }}_2{\mathrm{ISU}}_{\mathrm{it}}+{\mathsf{\alpha }}_3{\mathrm{LnTRADE}}_{\mathrm{it}}+{\mathsf{\alpha }}_4{\mathrm{LnTECH}}_{\mathrm{it}}+{\mathsf{\alpha }}_5{\mathrm{LnOFDI}}_{\mathrm{it}}{*\mathrm{ISU}}_{\mathrm{it}}+{\mathsf{\alpha }}_6{\mathrm{LnOFDI}}_{\mathrm{it}} \\ {*\mathrm{LnTRADE}}_{\mathrm{it}}+{\mathsf{\alpha }}_7{\mathrm{LnOFDI}}_{\mathrm{it}}{*\mathrm{LnTECH}}_{\mathrm{it}}+{\mathsf{\beta }\mathrm{X}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}} \end{gathered}$$

(5)

$$\begin{gathered} {\mathrm{CTFP}}_{\mathrm{it}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1{\mathrm{LnOFDI}}_{\mathrm{it}}+{\mathsf{\gamma }}_2{\mathrm{ISU}}_{\mathrm{it}}+{\mathsf{\gamma }}_3{\mathrm{LnTRADE}}_{\mathrm{it}}+{\mathsf{\gamma }}_4{\mathrm{LnTECH}}_{\mathrm{it}}+{\mathsf{\gamma }}_5{\mathrm{LnOFDI}}_{\mathrm{it}}{*\mathrm{ISU}}_{\mathrm{it}}+{ \mathsf{\gamma }}_6{\mathrm{LnOFDI}}_{\mathrm{it}} \\ {*\mathrm{LnTRADE}}_{\mathrm{it}}+{\mathsf{\gamma }}_7{\mathrm{LnOFDI}}_{\mathrm{it}}{*\mathrm{LnTECH}}_{\mathrm{it}}+{\mathsf{\beta }\mathrm{X}}_{\mathrm{it}}+{\mathsf{\epsilon }}_{\mathrm{it}} \end{gathered}$$

(6)

where $\mathrm{LnCO}{2}_{\mathrm{it}}$ is the total carbon emission of province $\mathrm{t}$ in year $\mathrm{i}$ in China; ${\mathrm{CTCP}}_{\mathrm{it}}$ is the carbon emission efficiency of province $\mathrm{t}$ in year $\mathrm{i}$ in China; ${\mathrm{LnOFDI}}_{\mathrm{it}}$ is the stock indicator of OFDI; ${\mathrm{ISU}}_{\mathrm{it}}$ is the industrial structure upgrading indicator; ${\mathrm{LnTRADE}}_{\mathrm{it}}$ is the international trade indicator; ${\mathrm{LnTECH}}_{\mathrm{it}}$ is the technological progress indicator; ${\mathrm{X}}_{\mathrm{it}}$ is a series of control variables, including urbanization level and level of economic growth; ${\mathsf{\epsilon }}_{\mathrm{it}}$ is the error term.

3.2. Variable Description and Data Source

3.2.1. Explained Variables

(1)

Total carbon emissions (CO₂): The total carbon emissions data used in this paper are from the China Carbon Emission Accounts Datasets (CEADs). These inventories were compiled in a combined accounting approach of scope 1 (Intergovernmental Panel on Climate Change territorial emissions from 17 types of fossil fuel combustion and cement production by 47 socioeconomic sectors) and scope 2 (emissions from purchased electricity and heat consumption).

(2)

Carbon emission efficiency (CTFP): Under the current situation of enormous energy consumption and severe environmental pollution, incorporating energy and environmental issues into the analytical framework of total factor productivity growth is important. In addition, carbon emission total factor productivity (green TFP) is an indicator that can reflect the dynamic changes in carbon emission efficiency [7] (Chen et al., 2022). Therefore, this study relied on the biennial Malmquist–Luenberger productivity index (BMLPI) calculation method used by [34] Wang et al. (2020), which changes the undesirable output into carbon emissions according to the research content of this paper and calculates the CTFP—using CTFP as a proxy indicator for carbon emission efficiency. The specific calculation formula is shown in Formula (7). $\overrightarrow{{\mathrm{D}}^{\mathrm{B}}}$ is a biennial weighted Russell directional distance function. We choose the direction vector as $\mathrm{g}=\left(-\mathrm{x}, \mathrm{y},-\mathrm{b}\right)$, which reflects the production activity that seeks to contract inputs and undesirable outputs and expand desirable outputs. When the value of the BMLPI is greater than (less than, equal to) 1, it indicates that the CTFP is increasing (declining, unchanged).

$$\mathrm{BMLPI}=\frac{1+\overrightarrow{{\mathrm{D}}^{\mathrm{B}}}\left({\mathrm{K}}^{\mathrm{t}},{\mathrm{L}}^{\mathrm{t}},{\mathrm{E}}^{\mathrm{t}},{\mathrm{Y}}^{\mathrm{t}},{\mathrm{C}}^{\mathrm{t}};{\mathrm{g}}^{\mathrm{t}}\right)}{1+\overrightarrow{{\mathrm{D}}^{\mathrm{B}}}\left({\mathrm{K}}^{\mathrm{t}+1},{\mathrm{L}}^{\mathrm{t}+1},{\mathrm{E}}^{\mathrm{t}+1},{\mathrm{Y}}^{\mathrm{t}+1},{\mathrm{C}}^{\mathrm{t}+1};{\mathrm{g}}^{\mathrm{t}+1}\right)}$$

(7)

The data for calculating CTFP input and output indicators are all from the China Statistical Yearbook. The specific definitions are as follows.

① Capital stock (K). This study selected the fixed capital stock of each province to perform its calculations. Refer to the perpetual inventory method of [35] Luo et al. (2021) to calculate the capital stock of each province (unit: CNY 100 million).

$${\mathrm{K}}_{\mathrm{it}}={\mathrm{K}}_{\mathrm{it}-1}*\left(1-\partial \right)+{\mathrm{I}}_{\mathrm{it}}$$

(8)

where ${\mathrm{K}}_{\mathrm{it}}$ is the capital stock of the current year, ${\mathrm{K}}_{\mathrm{it}-1}$ is the capital stock of the previous year in the region, $\partial$ is the capital depreciation rate, and ${\mathrm{I}}_{\mathrm{it}}$ is the total investment in fixed assets in the region for that year.

② Labor (L). This paper selected the number of employees in each province at the end of the year (unit: 10,000) as a proxy variable for the labor input.

③ Energy consumption (E). In terms of energy input, this study converted the coal, oil, natural gas, and electricity consumed in each province into a standard coal equivalent (unit: 10,000 tons of standard coal).

④ Desirable output (Y). This study selected GDP as the desired output. It was calculated at the constant price in 2000 (unit: CNY 100 million).

⑤ Undesirable output (C). The total carbon emissions of each province were selected as the undesirable output.

3.2.2. Explanatory Variables

(1)

Outward foreign direct investment (OFDI). Considering the stability and volatility of stock data, this study selects the natural logarithm of the stock of OFDI in each province in China to measure the OFDI level.

(2)

Industrial Structure Upgrade (ISU). In the intermediate stage of industrial development, industrial structure upgrading is an important way to achieve carbon peaking and carbon neutrality. Therefore, this study chose the ratio of the tertiary industry’s GDP and the secondary industry’s GDP as proxy variables for upgrading the industrial structure.

(3)

International trade (TRADE). According to the existing research, international trade is an important factor that affects carbon emissions. It is measured by the natural logarithm of the total import and export trade volume within each Chinese province [36] (Bai et al., 2022).

(4)

Technological progress (TECH). Technological innovation can reduce carbon emissions by improving the energy efficiency and promoting clean new energy. This study selected the natural logarithm of the technology market turnover to measure technological progress [37] (Tian et al., 2022).

3.2.3. Control Variables

(1)

Urbanization (URB). Urbanization prompts people to pursue a higher standard of living and a continuous increase in energy consumption, which increases carbon dioxide emissions. Therefore, the ratio of urban to the total population within each province is selected to measure the level of urbanization [38] (Zhang et al., 2022a).

(2)

Level of economic growth (RGDP). The rapid growth of China’s economy has required a great deal of energy consumption, which has directly led to increased carbon emissions. This study used 2000 as the base period, removed the impact of prices, and used the real GDP of each province to measure the level of economic growth.

Due to data availability, this article only uses data between 2004 and 2019 for 30 provinces (autonomous regions and municipalities) in China and excludes Tibet, Hong Kong, Macau, and Taiwan. See

Table 1. Definitions of Variables.

Variable	Form	Definition
Total carbon emissions	LnCO2	Log of total carbon emission
Carbon emission efficiency	CTFP	Carbon productivity
Outward foreign direct investment	LnOFDI	Log of OFDI stock
Industrial Structure Upgrade	ISU	Tertiary industry GDP/Secondary industry GDP
International trade	LnTRADE	Log of total imports and exports
Technological progress	LnTECH	Log of technical market turnover
Urbanization	URB	Urban population/Total population
Level of economic growth	LnRGDP	Log of real GDP

Note: Dates are drawn from the China Statistical Yearbook (2004–2017), Statistical Bulletin of China’s Outward Direct Investment (2004–2019), EPS China data, and Carbon Emission Accounts Datasets (CEADs).

for a summary of the variables used in this study.

4. Findings and Discussions

4.1. Estimation of Basic Statistics

The summary of the relevant statistics includes an analysis of the mean, standard deviation, minimum, and maximum value for the variables used in this article. The results are reported in

Table 2. Results of estimation of basic statistics.

Variable	N	Mean	Std. Dev.	Min	Max
LnCO2	480	9.998	0.784	7.406	11.447
CTFP	480	1.243	0.232	0.665	1.953
LnOFDI	480	1.914	0.987	−1.073	4.122
ISU	480	1.164	0.637	0.527	5.234
LnTRADE	480	7.653	1.651	3.522	11.181
LnTECH	480	3.949	1.865	−1.668	8.647
URB	480	0.539	0.142	0.262	0.896
LnRGDP	480	9.006	1.016	6.001	11.055

. Figure 2 reports the trend graph of each variable’s annual summation across the provinces.

Table 2 and Figure 2 show that the mean value of the total carbon emissions is 9.998, the standard deviation is 0.784, and the overall trend is increasing. The mean value of the carbon emission efficiency is 1.243 and the standard deviation is 0.232. The mean value of OFDI is 1.914 and the standard deviation is 0.987. In addition, OFDI and technological progress generally show a significant upward trend. Similarly, the upgrading of industrial structure, international trade, urbanization, and economic growth are also rising in volatility.

4.2. Baseline Regression Results

To observe whether other variables will seriously affect the robustness of the main explanatory variables, this paper adds variables one by one in the regression. Based on the results obtained (and summarized in

Table 3. Regression results concerning total carbon emissions.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
LnOFDI	0.148 *** (5.39)	0.105 *** (4.05)	0.113 *** (4.32)	0.115 *** (4.46)	0.112 *** (4.62)	0.098 *** (3.91)	0.116 *** (4.50)	0.043 * (1.69)	0.041 * (1.68)
ISU		−0.254 *** (−8.43)	−0.268 ** (−8.68)	−0.281 *** (−9.27)	−0.117 *** (−3.27)	−0.099 ** (−2.68)	−0.032 (−0.72)	−0.079 ** (−2.25)	−0.084 ** (−2.45)
LnTRADE			−0.039 ** (−1.96)	−0.035 * (−1.80)	−0.077 *** (−4.02)	−0.085 *** (−4.38)	−0.085 *** (−4.37)	−0.079 *** (−4.27)	−0.076 *** (−4.24)
LnTECH				−0.044 *** (−4.53)	−0.055 *** (−5.97)	−0.056 *** (−6.09)	−0.056 *** (−5.98)	−0.058 *** (−6.60)	−0.059 *** (−6.96)
URB					2.484 *** (7.51)	2.353 *** (7.02)	1.990 *** (5.52)	1.704 *** (5.09)	1.573 *** (4.85)
LnRGDP						0.240 ** (2.07)	0.254 ** (2.21)	0.268 ** (2.43)	0.297 ** (2.76)
LnOFDI $\times$ ISU							−0.039 ** (−2.63)
LnOFDI $\times$ LnTRADE								−0.025 *** (−6.55)
LnOFDI $\times$ LnTECH									−0.024 *** (−8.46)
Constant	9.353 *** (346.92)	9.632 *** (232.04)	9.903 *** (68.68)	10.011 *** (69.98)	9.045 *** (48.57)	7.191 *** (7.88)	6.525 *** (6.98)	6.474 *** (7.21)	6.296 *** (7.22)
Adj-R2	0.823	0.848	0.850	0.856	0.873	0.874	0.876	0.886	0.892
Cross-section fixed	YES	YES	YES	YES	YES	YES	YES	YES	YES
Period fixed	YES	YES	YES	YES	YES	YES	YES	YES	YES
N	480	480	480	480	480	480	480	480	480

Note: T-values are shown in parentheses; * 10% significant level; ** 5% significant level; and *** 1% significant level.

and

Table 4. Regression results concerning carbon emission efficiency.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
LnOFDI	0.175 *** (5.83)	0.203 *** (6.80)	0.188 *** (6.28)	0.187 *** (6.28)	0.189 *** (6.44)	0.168 *** (5.55)	0.130 *** (4.24)	0.235 *** (7.67)	0.220 *** (7.28)
ISU		0.165 *** (4.78)	0.191 *** (5.44)	0.197 *** (5.60)	0.094 ** (2.16)	0.122 ** (2.74)	−0.017 (−0.32)	0.098 ** (2.30)	0.108 ** (2.53)
LnTRADE			0.074 *** (3.27)	0.072 *** (3.20)	0.098 *** (4.24)	0.086 *** (3.64)	0.084 *** (3.67)	0.079 *** (3.50)	0.078 *** (3.44)
LnTECH				0.019 * (1.74)	0.027 ** (2.38)	0.025 ** (2.26)	0.022 ** (2.06)	0.027 ** (2.58)	0.028 ** (2.64)
URB					−1.559 *** (−3.89)	−1.756 *** (−4.33)	−0.999 ** (−2.33)	−0.968 ** (−2.39)	−1.042 ** (−2.57)
LnRGDP						0.362 ** (2.59)	0.332 ** (2.43)	0.327 ** (2.45)	0.310 ** (2.30)
LnOFDI $\times$ ISU							0.081 *** (4.60)
LnOFDI $\times$ LnTRADE								0.031 *** (6.57)
LnOFDI $\times$ LnTECH									0.022 *** (6.18)
Constant	0.974 *** (33.14)	0.793 *** (16.65)	0.278 * (1.70)	0.230 (1.39)	0.837 *** (3.70)	−1.961 * (−1.77)	−0.967 (−0.87)	−0.827 (−0.76)	−0.652 (−0.60)
Adj-R2	0.352	0.359	0.374	0.379	0.340	0.409	0.437	0.463	0.458
Cross-section fixed	YES	YES	YES	YES	YES	YES	YES	YES	YES
Period fixed	YES	YES	YES	YES	YES	YES	YES	YES	YES
N	480	480	480	480	480	480	480	480	480

Note: T-values are shown in parentheses; * 10% significant level; ** 5% significant level; and *** 1% significant level.

), when the explanatory variables were gradually added, each variable’s coefficients, signs, and significance did not change significantly, indicating that the regression results were relatively stable. The empirical results show that OFDI increases total carbon emissions but can also promote carbon emission efficiency. This confirms research Hypothesis 1. This may be because the development of OFDI directly encourages increases in the scale of high-carbon emissions industries and increases total carbon emissions. However, the reverse technology spillover from OFDI also improves China’s technological level, energy efficiency, and carbon emission efficiency.

In addition, the regression coefficient of industrial structure upgrading on carbon emissions was negative and significant at the 1% level. Concurrently, there was a significant positive correlation with the carbon emission efficiency. This means that with the adjustment of China’s industrial structure, the proportion of the tertiary industry has gradually increased and the proportion of the secondary industry with high energy consumption has gradually decreased. At the same time, the evolution of the industrial structure reflects the changes in China’s economic growth mode. In the process of industrial structure adjustment, China’s energy utilization efficiency is constantly improving. Therefore, upgrading the industrial structure not only inhibits the growth of China’s total carbon emissions but also promotes the improvement of carbon emission efficiency. Ref. [4] Yang et al. (2021) also came to a similar conclusion.

Interestingly, international trade inhibits the growth of total carbon emissions and improves carbon emission efficiency. The reason may be that international trade promotes technological exchanges and cooperation between China and developed countries. In addition, in the global market competition, developing low-carbon products in China is facilitated by learning and imitating high-end products in the international market [5] (Hao et al., 2021). China’s foreign trade structure has also gradually shifted to low-carbon emission industries, such as the tertiary and high-tech industries, thus restraining the growth of carbon emissions.

The results in columns (6) of Table 3 and Table 4 show that an increase of 1 standard unit of technological progress will reduce the total carbon emission by 0.056 units and increase the carbon emission efficiency by 0.025 units. In other words, technological progress can effectively restrain the growth of total carbon emissions and promote improvements to carbon emission efficiency. Because technological progress is an important way to achieve carbon emission reduction, enterprises can promote technological innovation through R&D investment and technical cooperation to reduce carbon dioxide emissions [17] (Guo et al., 2022).

Higher urbanization levels increase total carbon emissions while impairing the growth of carbon emission efficiency. This may be because China is still in the early stages of urbanization. Energy consumption continues to be the primary source of carbon emissions. The high energy consumption that characterizes urbanization’s early stages has brought considerable economic benefits but has also caused significant environmental damage.

In addition, the results in columns (6) of Table 3 and Table 4 also tell us that every 1% increase in the level of economic development will increase the total carbon emission by 0.24% and the carbon emission efficiency by 0.362%. This is consistent with the environmental Kuznets curve hypothesis that economic growth and environmental pollution have an “inverted U-shaped” relationship. Environmental degradation intensifies as the economy grows until an inflection point is reached, after which ecological degradation eases. Although China is the largest developing country, it still lags behind developed countries in terms of technology level and energy efficiency. China’s economic growth has not yet reached this “inflection point”. As a result, China’s economic growth brings more total carbon emissions. However, with the continuous development of China’s economy, China’s investment in R&D is also increasing. This has contributed to China’s technological progress and increased productivity. Therefore, economic growth not only expands the total amount of carbon emissions but also promotes the development of carbon emission efficiency.

It is worth noting that the interaction terms of industrial structure upgrading, international trade, and technological progress all have significant impacts on total carbon emissions and the coefficients are negative. However, each of these contributes significantly to the growth of carbon efficiency at the 1% level. It shows that OFDI has a synergistic emission reduction effect with industrial structure upgrading, international trade, and technological progress. Hypotheses 2, 3, and 4 were verified. The reasons may be: (1) The reverse technology spillover brought by China’s OFDI promotes domestic technological progress, which inhibits the growth of carbon emissions. For example, Chinese MNCs set up subsidiaries in knowledge-intensive clusters and feedback technologies acquired by subsidiaries through R&D investments and technology acquisitions to their parent companies. Through technology spillovers, parent companies can promote productivity improvements and the development of low-carbon technologies, thereby reducing carbon emissions. (2) OFDI promotes the upgrading of China’s industrial structure through the “marginal industrial expansion theory” and “technological innovation industrial upgrading theory”, thereby reducing domestic environmental pollution [11] (Hao et al., 2020). For example, with the development of China’s economy, labor costs have also increased. Many labor-intensive and energy-intensive industries have transferred production lines to areas with relatively low labor costs through OFDI. China can further concentrate domestic resources to develop emerging and profitable industries, promote upgrading the industrial structure and curb carbon emissions. (3) OFDI is conducive to optimizing China’s international trade structure. Chinese enterprises enter new markets through OFDI, which is conducive to the introduction of advanced production technologies and energy-saving technologies. In the international trade market competition, they can learn from foreign high-tech products, promote the development of low-carbon products in China and reduce carbon emissions. [39] (Wang and Wang., 2021). For example, [32] Chen et al. (2021) found that in most coastal provinces of China, the trade products in the region gradually shifted from early high-pollution products to the tertiary industry and high-tech industrial products, which led to the impact of international trade on carbon emissions from positive to negative.

4.3. Endogenous Problem Analysis

To more accurately estimate the impact coefficient of OFDI on carbon emissions, this study uses the instrumental variable method to analyze the possible endogeneity problems. According to [40] Li and Ouyang (2020), the first-order time lag and second-order time lag of OFDI are used as instrumental variables and the two-stage least squares (2SLS) method is used for estimation. The results are shown in

Table 5. 2SLS regression results.

Variables	First LnOFDI	Second LnCO2	First LnOFDI	Second CTFP
L.LnOFDI	0.824 *** (16.98)		0.824 *** (16.98)
L2.LnOFDI	−0.109 ** (−2.41)		−0.109 ** (−2.41)
LnOFDI		0.098 ** (2.65)		0.230 *** (5.17)
ISU	0.031 (0.63)	0.043 (1.03)	0.031 (0.63)	0.073 (1.46)
LnTRADE	0.030 (1.24)	−0.080 *** (−3.96)	0.030 (1.24)	0.063 ** (2.58)
LnTECH	0.013 (1.03)	−0.048 *** (−4.73)	0.013 (1.03)	0.019 (1.52)
URB	−0.265 (−0.60)	2.612 *** (7.11)	−0.265 (−0.60)	−2.612 *** (−5.92)
LnRGDP	0.300 * (1.88)	0.0220 (1.62)	0.300 * (1.88)	0.548 *** (3.35)
Constant	−2.234 (−1.54)	7.188 *** (5.71)	−2.234 (−1.54)	−3.753 ** (−2.48)
Adj-R2	0.929	0.799	0.929	0.375
Cross-section fixed	YES	YES	YES	YES
Period fixed	YES	YES	YES	YES
Underidentification test		266.168 [0.0000]		266.168 [0.0000]
Weak identification test		255.390		255.390
Sargan statistic		2.494 [0.1143]		0.009 [0.9228]
N	420	420	420	420

Note: T-values are shown in parentheses; * 10% significant level; ** 5% significant level; and *** 1% significant level. The values in square brackets are the p values corresponding to the statistical tests.

. The under-identification test and weak identification test results show that instrumental variables are correlated with endogenous variables and there is no weak instrumental variable. The Sargan test result p value is greater than 0.1, indicating that the selected instrumental variables are exogenous and this study’s panel IV estimation is effective. The 2SLS regression results show that, after reducing the endogeneity problem through the instrumental variable method, OFDI still promotes the growth of total carbon emission and improved carbon emission efficiency. The research hypothesis was verified again. The results for the other highlighted variables are also consistent with the benchmark regression results, further demonstrating the robustness of the results.

4.4. Robustness Test

In this study, the robustness test was carried out by replacing the explained variables. We refer to the practice of [16] Pan et al. (2020) using the method proposed by the Intergovernmental Panel on Climate Change (IPCC) to recalculate the total carbon emissions. The calculation formula is shown in Formula (9).

$${\mathrm{CO}}_2={{\displaystyle \sum }}^{ }{\mathrm{E}}_{\mathrm{i}}\times {\mathrm{EF}}_{\mathrm{i}}$$

(9)

where $\mathrm{i}$ is the energy type, ${\mathrm{E}}_{\mathrm{i}}$ is the energy consumption, and ${\mathrm{EF}}_{\mathrm{i}}$ is the carbon emission coefficient. Carbon emission coefficients are determined according to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories.

Carbon emission intensity (CEI) refers to the carbon emission per unit of GDP (tons per CNY 100 million). It measures the relationship between carbon emissions and economic growth. That is to say, through technological progress and the improvement of energy utilization, the energy consumption per unit of GDP can be reduced (reducing carbon emission intensity), thereby reducing carbon emissions [30] (Wang and Yi., 2022). Therefore, this study uses carbon emission intensity for robustness testing as a surrogate variable for carbon emission efficiency. The regression results are shown in

Table 6. Results of robustness test.

Variables	(1) LnCO2	(2) LnCO2	(3) LnCO2	(4) CEI	(5) CEI	(6) CEI
LnOFDI	0.053 *** (3.86)	0.030 ** (2.20)	0.020 (1.61)	0.074 ** (2.20)	−0.010 (−0.30)	0.021 * (1.67)
ISU	−0.034 (−1.41)	−0.027 (−1.39)	−0.025 (−1.36)	−0.037 (−0.65)	−0.160 *** (−3.34)	−0.037 ** (−2.07)
LnTRADE	−0.047 *** (−4.14)	−0.047 *** (−4.29)	−0.043 *** (−4.24)	−0.105 *** (−4.18)	−0.102 *** (−4.03)	−0.030 ** (−3.15)
LnTECH	−0.021 *** (−4.04)	−0.022 *** (−4.54)	−0.024 *** (−5.24)	−0.057 *** (−4.78)	−0.061 *** (−5.14)	−0.021 *** (−4.81)
URB	1.041 *** (5.21)	0.804 *** (4.31)	0.642 *** (3.69)	0.670 (1.49)	0.908 ** (2.00)	0.502 ** (2.97)
LnRGDP	0.098 (1.57)	0.108* (1.79)	0.130 ** (2.30)	−1.079 *** (−7.22)	−1.084 *** (−7.25)	0.064 (1.15)
LnOFDI $\times$ ISU	−0.001 (−0.17)			−0.080 *** (−4.20)
LnOFDI $\times$ LnTRADE		−0.011 *** (−4.98)			−0.021 *** (−4.07)
LnOFDI $\times$ LnTECH			−0.014 *** (−9.28)			−0.014 *** (−9.22)
Constant	2.971 *** (5.84)	2.980 *** (6.04)	2.868 *** (6.22)	9.916 *** (8.15)	9.762 *** (8.02)	3.481 *** (7.66)
Adj-R2	0.815	0.826	0.848	0.877	0.877	0.837
Cross-section fixed	YES	YES	YES	YES	YES	YES
Period fixed	YES	YES	YES	YES	YES	YES
N	480	480	480	480	480	480

Note: T-values are shown in parentheses; * 10% significant level; ** 5% significant level; and *** 1% significant level.

.

The results in Table 6 show that the interaction terms of OFDI, technological progress, and international trade significantly inhibit the growth of total carbon emissions. In addition, the interaction terms of OFDI, industrial structure upgrade, technological progress, and international trade negatively influence coefficients on carbon emission intensity. That is to say, the synergistic effect of OFDI and these variables are beneficial to reduce the carbon intensity and total carbon emissions. Compared to the baseline regression results, we observe that the impact coefficients of the highlighted variables vary only in size and significance in the statistics concerning carbon emissions. In other words, our results were robust and reliable.

5. Conclusions

As the country with the largest carbon emission in the world, China plays an important global role in reducing carbon emissions and promoting sustainable development. Therefore, this study uses China as a research sample and uses data from 30 provinces in China gathered between 2014 and 2019 to comprehensively analyze the impact of OFDI on carbon emissions. The results show that the direct effect of OFDI promotes the increase in total carbon emissions and the growth of China’s carbon emission efficiency. Technological progress, international trade, and industrial structure upgrading have inhibited the total carbon emission and promoted the improvement of carbon emission efficiency. We also confirmed that the level of urbanization and economic growth contributes to China’s total carbon emissions growth and that the improvement in economic growth has also promoted the carbon emission efficiency. Notably, the synergistic effect between OFDI and technological progress, international trade, and industrial structure upgrading can restrain the growth of total carbon emissions and promote improving carbon emission efficiency. Finally, the findings also support the above results when using panel IV estimates for endogeneity problem discussion and substituting explained variables for robustness testing.

Based on the above empirical analysis results, we present the following recommendations. First, China should continue to adhere to the concept of “sustainable development”, strengthen technological exchanges and cooperation with developed countries, and focus on developing high-tech industries. At the same time, it encourages enterprises to strengthen R&D investment, improve the absorption capacity of OFDI’s reverse technology spillovers, and promote domestic technological progress. In addition, the structure of OFDI should be optimized. Investing in developed countries can guide pollution-intensive enterprises to participate and promote technical exchanges between domestic and foreign enterprises. Second, provide full play to the role of OFDI in promoting the upgrading of the industrial structure and combine the opportunities of the “Belt and Road” to promote the transfer of marginal industries. Concentrate superior resources to develop high-tech and emerging industries, promote the upgrading of domestic industrial structure, and realize the transformation of a “low carbon economy”. Finally, pay attention to the synergistic emission reduction effect of OFDI and international trade. When promoting the development of international trade, strengthen the introduction of high-tech talents. Use the technology spillover of OFDI to promote the development of low-carbon products and technologies, thereby optimizing the structure of international trade.

This study complements the existing literature concerning the comprehensive impact of OFDI on carbon emissions. However, it is somewhat limited. For example, while considering the research sample, this study focuses only on China. Developed countries may show different results, and future research may focus on other countries. Furthermore, in this study, only a panel fixed-effects model was used. Future scholars can regress this problem using alternative models such as the panel vector autoregressive model or panel vector error correction model.