The Impact of Foreign Direct Investment on Carbon Emissions in Economies Along the Belt and Road

Abstract

With China’s emergence as a major global economy, its involvement in tackling climate change and fostering sustainable growth has garnered considerable focus. What impact does global direct investment have on carbon emissions within Belt and Road economies? This study innovatively utilizes a quantile regression model to analyze the varied impacts of international direct investment across distinct carbon emission quantiles, further delving into the conditional probability distribution of the dependent variable to provide a strong theoretical basis for precise policy-making by relevant departments and integrating time and space delays in examining the effects of carbon reduction strategies within the Belt and Road Initiative. Furthermore, this study aims to concentrate its research efforts on the host nations. Findings from this study indicate that global direct investments could escalate carbon emissions in economies with lower carbon emissions; yet, with the rise in the host nation’s carbon emissions, the ripple effect of international direct investments in green technology becomes increasingly evident. Empirical evidence indicates that global direct investment in Belt and Road economies demonstrates a significant mitigating effect on carbon emissions, thereby amplifying the decarbonization benefits associated with such cross-border capital flows.

1. Introduction

The rise in carbon emissions leading to climate change has emerged as a worldwide issue. Starting with the Paris Agreement’s aim for worldwide “carbon neutrality” and culminating in COP28 (the twenty-eighth Conference of the Parties to the United Nations Framework Convention on Climate Change), marking the first instance of an accord on the Paris Agreement’s global inventory, inclusive of the “movement from fossil fuels” mentioned in the document, an increasing number of economies are proactively seeking strategies for harmonized economic and environmental growth. With China’s emergence as a global economic powerhouse, its involvement in managing climate change and fostering sustainable growth has garnered international interest. This study centers on the variance in carbon emissions following the receipt of China’s OFDI (Outward Foreign Direct Investment) by the participating economies in the Belt and Road Initiative. The focus of this document is on the idea of creating a collective future for humanity through economic collaboration along the Belt and Road, offering both theoretical and practical backing for China’s approach to green development. Concurrently, examining the Belt and Road’s external direct investment sheds light on global skepticism regarding China’s pollution transmission in the “China environmental threat theory” and upholds China’s favorable reputation in the worldwide governance framework.

International direct investment has a profound impact on environmental sustainability, potentially promoting green transformation or exacerbating environmental problems.

On the positive side, it first promotes the development of green industries.

• Investing in sustainable energy forms like solar, wind, and hydroelectric power can significantly lessen dependency on fossil fuels and decrease carbon emissions.
• Secondly, the rise of ESG (environment, society, and governance) investment, which steers capital towards environmentally friendly enterprises based on environmental, social, and governance standards, is another positive aspect.
• Thirdly, investment in nature conservation projects and the circular economy helps to restore ecosystems.
• Moreover, investment can drive the construction of green infrastructure through policy. China’s “dual carbon” goals have driven trillions of Yuan in green investment.

On the negative side, first, investment in traditional high-pollution industries such as coal, oil, and heavy industry still accounts for a large proportion. Second, resource development is driven by short-term interests. For example, excessive investment in industries such as mining, logging, and industrial agriculture leads to deforestation (such as in the Amazon region), water resource depletion, and soil degradation, threatening ecological balance. Third, externalization of environmental costs. Some enterprises, in order to reduce costs, neglect environmental protection measures (such as illegal discharge of pollutants), shifting the environmental burden onto society. Developing countries often become “pollution havens”, taking on the transfer of high-pollution industries. Fourth, financial speculation and resource bubbles. For instance, speculative investment in scarce resources (such as lithium and cobalt) may lead to over-exploitation, damaging the local environment, such as the consumption of water resources in lithium mining in South America.

Present studies on how China’s investments in Belt and Road nations affect the environment primarily concentrate on the provincial or urban scale, seldom considering the host country as the primary research subject. Consequently, this research delves into how Foreign Direct Investment (FDI) impacts the environment in China, viewed from the host nation’s standpoint, thereby enhancing the pertinent scholarly research to some degree. In addition, China’s FDI is mainly concentrated in developing countries, and these countries are generally dominated by low-level manufacturing and immature processing technology. Comparative analysis of China’s carbon emission patterns across developed and developing national contexts reveals that such assessments may incentivize environmentally sustainable investment practices in emerging economies, thereby facilitating the reconciliation of economic development imperatives with ecological preservation objectives.

A significant portion of the pertinent research on eco-friendly development within the Belt and Road Initiative is concentrated on the provinces or cities of China, for example, the studies of Yang and Wang (2018) [1], Zhang (2018) [2], and Liu et al. (2022) [3], while the research on host countries is limited. This study will utilize recent data on Chinese direct investments along the Belt and Road to enhance current research by examining the determinants and levels of carbon emissions in the Belt and Road economies. This study aims to establish a comprehensive analytical framework for systematically examining the theoretical linkages between China’s Outward Foreign Direct Investment (OFDI) and its implications for carbon emission trajectories and environmental quality under the Belt and Road Initiative (BRI), thereby informing strategic policy formulation to advance sustainable, low-carbon development pathways across BRI partner economies. Anticipated advancements and breakthroughs in this study primarily encompass the following elements: (1) The research angle transitions to the host nation, centering on the ecological effects on the economies along the “Belt and Road” following the receipt of China’s OFDI. (2) Quantile regression analysis, considering the varying carbon emissions of nations along the “Belt and Road” route, is utilized to examine China’s OFDI’s effect on economies with diverse carbon emission levels from a broader viewpoint. (3) Liu (2022) developed a dynamic economic growth and environmental pollution model to mitigate the delayed impact of carbon dioxide emissions over time and space, employing the Difference-in-Differences (DID) approach to maintain this paper’s structured integrity and prevent inherent issues [4]. (4) Different from the traditional pollutant measurement criteria, this paper adopts a robustness test to improve the reliability and comprehensiveness of this study.

2. Literature Review

2.1. International Direct Investment and Environmental Impact

Research into how global direct investments affect the environment originates from initial studies exploring the link between economic expansion and environmental integrity. Grossman and Krueger (1991) innovatively introduced the Kuznets curve in economics into the field of environmental research, suggesting a strong correlation between environmental pollution and the level of national economic growth, along with the idea that an “inverted U-shaped” curve in environmental Kuznets [5] can be synthesized. With the continuous acceleration of economic globalization, the discussion of environmental pollution has gone beyond the traditional analysis framework and has begun to introduce influential factors such as international direct investment. Currently, in the debate about the link between global direct investment and environmental contamination, two predominant perspectives prevail: “pollution refuge” and “pollution halo”.

1. Hypothesis of a sanctuary for pollution. According to the “pollution haven” theory, companies in their native countries are motivated to relocate their resource-heavy industrial sectors to nations with more lenient environmental rules, resulting in environmental degradation in the receiving country and becoming “pollution havens” [6,7].

Halicioglu (2009), focusing on Turkey, established a dynamic empirical link among carbon emissions, energy use, earnings, and international commerce, deducing that foreign trade primarily influences Turkey’s carbon dioxide levels [8]. Bu and Wagner (2016) found that the heterogeneity of one’s own environmental protection ability and the size of an enterprise determine the investment location of an enterprise [9]. In short, enterprises with strong environmental protection ability should be invested in regions with strict environmental control, while enterprises with weak environmental protection ability should be invested in regions with looser environmental control. This finding is further support for the concept of “pollution havens” [10]. The research by Shahbaz et al. (2018) delves into how foreign direct investments affect CO₂ emissions, incorporating public investment in financial growth and R&D within France’s energy sector for carbon emission purposes [11]. The study’s findings indicate a substantial beneficial effect of Foreign Direct Investment on reducing carbon emissions in France. The research by Singhania and Saini (2021) verifies the presence of “pollution havens” in both advanced and emerging nations, especially noticeable in the developing world [12]. They point out that foreign capital tends to invest in resource-intensive and pollution-intensive ways, which exacerbates environmental problems in developing countries. Based on this finding, they argue that strict environmental access regimes should be put in place to prevent foreign investment from further exacerbating environmental problems.

For domestic research on this hypothesis, Sha and Shi (2006) developed a five-year panel analysis starting in 1999 to evaluate the environmental effects of Foreign Direct Investment [13]. Findings indicate that Foreign Direct Investment (FDI) has markedly escalated the emissions of industrial exhaust gases in China, adversely affecting the ecological system. Furthermore, Guo (2013) compiled panel data spanning from 2002 to 2010 across various sectors and carried out a practical analysis of foreign direct investment in China’s secondary industry [14]. The study reveals a positive link between China’s carbon emissions and the rise in Foreign Direct Investment (FDI).

2. Hypothesis of “pollution halo.” According to the “pollution halo” theory, the home country’s investment in the host nation yields not just economic gains but also aids in enhancing the host country’s environment [8,15]. D Popp’s (2011) research highlights that the eco-friendly technologies developed by advanced nations can be replicated in developing countries via global trade and overseas investments, so that they can possess these advanced technologies and realize the improvement of environmental quality [16].

Yang and Wang (2018) conducted a balanced panel study on 11 provinces and cities along the Yangtze River from 2006 to 2016 and found that two-way FDI can reduce total CO₂ emissions [1]. However, the maximization of this effect depends on moderate environmental regulation. Li and Liu (2012) focused their research on the factors leading to industrial carbon emissions and found that FDI spillover within and between industrial sectors had a positive impact on carbon emissions [17]. Meng et al. (2016) investigated how Foreign Direct Investment (FDI) affects ecological advantages, incorporating both temporal and regional evaluations [18]. It was highlighted that, following China’s joining of the WTO, Foreign Direct Investment (FDI) exhibited regional variances in its ecological advantages. For example, the ecological benefits could be significantly amplified in the eastern and central regions, confirming the positive effects of FDI on the regional ecosystem. In the study of China’s OFDI by Liu and colleagues (2022), it was determined that, considering aspects like direct impact, scale influence, structural impact, and technological influence, Chinese companies can markedly reduce carbon emissions in nations along their routes, thanks to their more eco-friendly technology and effective management systems [3].

2.2. The Impact of OFDI on the Belt and Road Initiative

Within the framework of worldwide economic unification, the role of international direct investment is increasingly significant as a key manifestation of the technology spillover phenomenon. Consequently, academics focus intently on China’s direct investments in the Belt and Road economies.

A significant portion of the existing scholarly work concentrates on how China’s OFDI economically aids the host nations, framed within the “One Belt, One Road” model. Wang et al. (2016), by concentrating on infrastructure development along the Belt and Road, verified that building the Belt and Road markedly enhances the economic advantages for nations along this route, but also reduces the excess capacity of China’s steel and other industries, thus playing a driving role in China’s economy and achieving a “win-win effect” [19]. In his study, Zhang (2018) highlighted that the “Belt and Road” OFDI has markedly enhanced the job rates of secondary and tertiary sectors in nations along the Belt and Road [2]. Especially for economies with low per capita GDP, China’s investment has a significant effect on poverty reduction. The results underscore the beneficial impact of China’s OFDI in enhancing economic expansion, generating employment, and diminishing poverty in B&R nations.

Concentrating on the link between OFDI and the environment within the Belt and Road Initiative framework, Meng’s (2019) research reveals that relocating the service sector to coastal regions as part of the Belt and Road Initiative has markedly decreased direct CO₂ emissions [20]. Lei et al. (2016) developed a joint framework addressing economic expansion and carbon emissions, with the expectation [21] that the interplay between them would progressively intensify going forward.

2.3. Economic Impact on Environmental Sustainable Development

In 1987, the United Nations introduced the idea of sustainable development, primarily urging every societal segment to evolve in a manner that caters to the current generation’s requirements while maintaining the ability of future generations to achieve their goals, and to formulate distinct strategies for the forthcoming trajectory of human society. Currently, most research starts with the improvement of the concept of sustainable development (Vos, 2007) [22] and progressively broadens to encompass various aspects like society, economy, and ecology for a thorough examination of its operational mechanisms and investigation of the factors that influence it.

This research domain amalgamates numerous scholarly articles, reports, and associated studies focusing on the convergence of two principal environmental concerns: air pollution and sustainable development. The studies undertaken by Dominski et al. (2021) [23] and Sivarethinamohan et al. (2021) [24] have attracted widespread attention. They emphasize the serious harm of air pollution to health, especially its impact on the respiratory and cardiovascular systems, thereby highlighting the urgency of reducing pollution to benefit public health. Concurrently, the study by Aix et al. (2022) [25] elucidates critical insights into the intricate relationship between atmospheric pollutants and climate dynamics, demonstrating how greenhouse gases such as carbon dioxide (CO₂) and methane (CH₄) exacerbate radiative forcing—a key driver of global warming—while underscoring their profound adverse effects on both terrestrial and aquatic ecosystems, as well as public health outcomes. Regarding research on sustainability, the pioneering contribution of Vilcassi and Thurston (2023) [26] lies in proposing a research framework for understanding sustainable development and highlighting the importance of harmonizing economic expansion, social fairness, and ecological conservation. Additionally, the research by Makri and Stilianakis’ study (2008) [27] delves into the “tragedy of the commons” idea, emphasizing the way unchecked use of resources results in environmental deterioration, including air pollution.

Another part of the literature on environmental sustainability is related to digital financial policies. Scholarly works on digital financial strategies primarily investigate the diverse roles of financial technology in influencing economic growth, financial inclusivity, and ecological sustainability. Research conducted by Rasoulinezhad (2020) [28] and Xi and Wang [29] explored how digital finance can enhance fund accessibility. Concurrently, financial services have the potential to foster entrepreneurial spirit, thus encouraging economic expansion, particularly in developing economies. Furthermore, Al-Smadi (2023) [30] underscored the transformative role of mobile money and digital payment systems in enhancing financial inclusivity, demonstrating their potential to mitigate socioeconomic disparities by providing accessible, low-cost financial infrastructure to underserved populations. In parallel, Ouyang et al. (2023) [31] and Tang and Geng (2024) [32] examined the contributions of digital financial innovations—including mechanisms such as green bonds and carbon credit market—to environmental sustainability, emphasizing their efficacy in catalyzing capital allocation toward renewable energy deployment, energy efficiency optimization, and other sustainability-oriented projects. These advancements are posited to accelerate the transition toward decarbonized economic systems while attenuating the cascading impacts of anthropogenic climate change.

Consequently, considering China’s unique national circumstances and traits, scholarly works on environmental integrity and sustainable progress in China have extensively examined the intricate interplay among swift industrialization, urbanization, economic expansion, and the deterioration and exhaustion of environmental resources. The works by Xin et al. (2023) [33] and Fang et al. (2023) [34] have thoroughly evaluated China’s environmental hurdles, such as air and water contamination, soil deterioration, and biodiversity loss, highlighting their effects on public health, ecological equilibrium, and socio-economic growth. Moreover, the studies by Bai et al. (2023) [35] and Islam and Wang (2023) [36] conducted a study on how effective environmental strategies and rules are in reducing pollution and fostering eco-friendly practices, emphasizing the significance of enhancing institutional capabilities, implementing enforcement strategies, and engaging stakeholders to reach environmental objectives. Additionally, Adebayo and Ullah (2023) [37] investigated the capabilities and effects of eco-friendly technologies, circular economic models, and sustainable city planning approaches in alleviating environmental issues, advocating for improved resource utilization efficiency and enhanced resilience to climate change.

3. Conceptual Research Framework

Building upon the foundational work of Grossman and Krueger, as well as subsequent scholarly discourse on the complex environmental externalities associated with foreign direct investment (FDI), this analysis conceptualizes the impact of Outward Foreign Direct Investment (OFDI) on host-country carbon emissions through a tripartite theoretical framework, operationalized via the systematic analytical framework illustrated in Figure 1.

3.1. Scale Effects

The magnitude of OFDI implies that investments directed towards nations with minimal economic growth will invariably coincide with a rise in internal consumption and production growth, resulting in heightened carbon emissions and environmental contamination, even as the nation’s economic status is on the rise. However, the impact of OFD cannot be generalized. While developing its economy, the host country can reduce industrial pollution emissions through technology introduction and industrial upgrading. Residents will gradually increase their demand for environmental protection and eventually encourage the government to introduce a series of environmental regulations, including carbon emission reduction, to reduce pollution levels. Consequently, the magnitude of OFDI’s impact remains unclear, a factor intimately linked to a nation’s economic progress. The “environmental Kuznets curve” theory offers the strongest backing for this perspective.

3.2. Structural Effects

The influence of OFDI’s structure may also take into account its beneficial or detrimental effects on the environment of the host nation. With the inflow of foreign capital, the industrial structure of the host country may undergo differentiated transformation due to technology spillover: should overseas investments predominantly focus on high-value and tertiary sectors, the host nation’s industrial framework will transition to a more sophisticated model, heavily reliant on capital and technology. Through the above ways, the host country can achieve profit growth and enterprise optimization while reducing the pollution density per unit output and carbon emissions. If foreign capital flows to resource-intensive industries, the environmental pressure faced by the host country may increase due to the increase in the use of resources and energy and the rise in carbon dioxide emissions.

3.3. Technological Effects

Extensive research on “technology spillover” can support the positive impact of technology effects on host country environments. The host nation can cut down on resource usage of the initial production and enhance operational effectiveness by adopting the investor country’s sophisticated manufacturing methods and lines, along with emulating its management approach. Second, after the introduction of foreign clean energy technology, enterprises in the host country should break through the technical bottleneck, achieve profit growth, and enhance the innovation capacity of the whole industry, and should not deny that the early stage of technological progress may exacerbate the damage to the environment, such as the invention of certain electrical appliances. In the long run, however, technology spillovers are more of a positive influence on a global scale.

4. Empirical Testing

4.1. Research Sample Selection

The latest data on OFDI stock collection are from 2022. The latest available data for CO₂ emissions in the WDI database are from 2020. The time span for the selected study is from 2005 to 2020. The primary motive behind selecting 39 Belt and Road economies for this study is to evaluate the accessibility and dependability of the World Bank’s World development indicator data. Concurrently, our dedication lies in thoroughly encompassing the range of “One Belt, One Road” nations. Selection of representative economies from Asia, Europe, and Africa for the Belt and Road Initiative was achieved through rigorous data integrity and ongoing screening of prospective research samples.

Table 1. Lists the sample countries studied.

Regions	Countries
East Asia (2)	China, Mongolia
Southeast Asia (9)	Philippines, Cambodia, Laos, Malaysia, Myanmar, Thailand, Singapore, Indonesia, Vietnam
Central and Eastern Europe (10)	Russian Federation, Czech Republic, Hungary, Poland, Romania, Ukraine, Slovakia, Bosnia and Herzegovina, Bulgaria, Latvia
Central Asia (4)	Kazakhstan, Tajikistan, Uzbekistan, Kyrgyzstan
South Asia (5)	India, Pakistan, Nepal, Sri Lanka, Bangladesh
West Asia and North Africa (9)	Azerbaijan, United Arab Emirates, Egypt, Saudi Arabia, Israel, Kuwait, Iran, Iraq, Jordan

presents an in-depth analysis of the economic outcomes.

4.2. Selection of Variables and Data Explanation

1. Explained variable: carbon emission degree(lnCO2_it). As for the measurement and source of carbon emission data, there are different choices in the existing literature. Notably, Hu and Shao (2022) [38] analyzed carbon emission dynamics using the metrics of total CO₂ output and emissions intensity per GDP unit. Concurrently, scholars, including Liu et al. (2022) [3] and Yu et al. (2023) [39], have adopted per capita CO₂ emissions as a proxy for environmental impact assessment. Aligning with this methodological framework, the present investigation computes per capita CO₂ emissions for Belt and Road Initiative economies by normalizing aggregate carbon emissions against population statistics, leveraging standardized datasets from the World Bank’s World Development Indicators (WDI) database. Following the exclusion of data points significantly absent in the statistical records, the information from 39 Belt and Road economies spanning from 2005 to 2020 is chosen for examination.

2. Primary explanatory factor: the OFDI stock of China (lnOFDI_it). We learnt, from Zhuang Danyu and Fu Lei (2021) [40], to select OFDI stock data in order to avoid the problem of negative flow data in logarithmic processing.

3. Control variable: economic scale (lnGDP_it). The impact of a nation’s economic magnitude on CO₂ emissions must be acknowledged, with a potential inverted “U” curve illustrating their connection in the previously mentioned environmental Kuznets curve. Concurrently, this document employs fixed prices in 2010 dollars to mitigate the impact of price variances.

Level of industrialization (lnIndustry_it). There are many different ways to measure the level of industrialization in academic circles. Take, for instance, the technique of allocating varied importance to the output worth of distinct sectors within the GDP (Zhai and Huang 2023) [23], and the ratio of the tertiary industry’s added value to that of the secondary industry (Hu and Shao, 2022) [38]. Considering the global collaboration between China and various economies along the “Belt and Road” route, primarily focused on the manufacturing sector, this document employs the manufacturing industry’s additional value to denote the degree of industrialization.

The level of receptiveness towards the external world (lnOpen_it). A cumulative measure of imported and exported goods and services relative to the gross domestic product (GDP).

Population density (lnPD_it). Count of individuals per square kilometer of terrain.

Government effectiveness (GE_it). Use “Government Effectiveness: Estimate” from the WDI database. The definition of GE gives the country’s score on the aggregate indicator. The measurement of this metric is in the standard normal distribution unit, with a range between −2.5 and 2.5 (−2.5 is weak; 2.5 is strong). Government effectiveness captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies. This mirrors the caliber of governmental policy development and execution, along with the perceived trustworthiness of governmental policy dedication by the populace. For more details, please refer to the data files for each source available at www.govindicators.org. The indicator was introduced considering that host governments can reduce carbon emissions by introducing emission reduction policies to eliminate lagging production capacity and promote environmental quality improvement through financial support for innovative research and development and guidance for industrial structure upgrading.

Degree of informatization (lnInternet_it). The count of people accessing the Internet as a fraction of the overall population.

4.3. Construction of Model

To analyze the influence of China on carbon emissions from Belt and Road economies, the following regression model has been developed:

$${lnCO2}_{it}={\beta }_0+{\beta }_1{lnOFDI}_{it}+\theta {lnX}_{it}+{\mu }_{it}$$

(1)

In Formula (1), t represents the country and i represents the year; the variable lnCO2_it in question signifies the level of carbon emissions in the economies situated along the Belt and Road, and the core explanatory variable lnOFDI_it symbolizes the extent of China’s immediate investment in the Belt and Road economies. In addition, X_it represents the control variable matrix, θ is the corresponding coefficient matrix, and the term μ_it for random error. If the coefficient β₁ to be estimated is statistically significantly greater than zero, it can be concluded that China’s direct investment in the “Belt and Road” economies has increased the region’s carbon emissions; if it is significantly less than zero, the implication is the opposite. By taking logarithms of the data, the possible heteroscedasticity values in the regression are reduced to facilitate discussion of the results. Specifically speaking, the logarithmic form of variables is mainly adopted to reduce the heteroscedasticity of the data in the model and eliminate the dimensional problem at the same time, so that the coefficients between variables fall within a reasonable range.

In this paper, Stata16 version 16.1 is used to perform panel regression analysis on the data; the detailed outcomes are presented in

Table 2. Results of descriptive statistical analysis of variables.

Variables	Variable Description	Sample Number	Mean	Standard Deviation	Minimum	Maximum
lnOFDIit	Stock of outward direct investment/USD 10,000	576	10.31	2.464	2.303	15.60
lnCO2it	Carbon dioxide emissions per kiloton	576	1.115	1.258	−2.316	3.443
lnGDPit	GDP/constant dollar	576	8.468	1.173	6.321	11.02
lnIndustryit	Manufacturing value added/constant USD	571	3.323	0.664	−0.168	4.315
lnOpenit	Total imports and exports of goods and services/GDP	576	4.124	0.814	0.149	5.839
lnPDit	Population density/number of people per kilometer area	576	4.538	1.434	0.499	8.983
lnInternetit	Number of people using the Internet/per	564	3.244	1.274	−2.730	4.605
GEit	Government effectiveness/−2.5~2.5	576	0.113	0.795	−1.731	2.470

.

4.4. Regression Analysis

1. Preliminary test: In order to avoid problems such as pseudo-regression, decline in accuracy, and contrary to expectations in the regression analysis, the following processing is carried out: Initially, the aim is to investigate the presence of multicollinearity issues among variables in the panel model, thereby preventing regression result variances caused by linear relationships.

Table 3. Results of variance inflation factor test.

Variable	VIF	1/VIF
GEit	5.30	0.188677
lnGDPit	4.36	0.229167
lnIndustryit	3.18	0.314618
lnOpenit	2.81	0.355928
lnInternetit	1.99	0.503152
lnPDit	1.45	0.688039
lnOFDIit	1.14	0.876760
Mean VIF	2.89

lists the test results of the variance inflation factor (VIF). Findings indicate that all variable VIF values fall below 5, with the exception of government effectiveness (GE), which is marginally above 5, signifying the model’s minimal multicollinearity.

Next, a matrix test of correlation coefficients is conducted to more effectively assess the interconnections among variables. Results of the tests are displayed in

Table 4. Correlation coefficient matrix.

Variables	ln_OFDI	ln_CO2	ln_GDP	ln_Industry	ln_Open	ln_PD	ln_Internet	GE
lnOFDIit	1.000
lnCO2it	0.044	1.000
lnGDPit	0.034	0.855	1.000
lnIndustryit	0.181	0.077	0.002	1.000
lnOpenit	0.013	0.106	0.124	0.717	1.000
lnPDit	0.015	−0.244	0.050	−0.049	0.078	1.000
lnInternetit	0.140	0.688	0.679	−0.043	0.084	−0.006	1.000
GEit	0.043	0.557	0.794	−0.175	0.145	0.358	0.600	1.000

. According to the correlation coefficient matrix, the interrelation coefficient among all variables falls below 0.8, signifying a feeble connection between them.

Ultimately, the Hausman test is conducted to ascertain if the panel data are influenced by fixed or random effects. Results from the tests are displayed in

Table 5. Results of the Hausman test.

Ho: Difference in Coefficients Not Systematic
chi2(7) =108.39
Prob > chi2 = 0.0000

. A p-value of 0.0000 is assigned to dismiss the null hypothesis, followed by choosing the fixed effect for regression analysis.

2. Initial regression analysis: The data in

Table 6. Regression results of China’s OFDI and carbon emissions.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnOFDIit	0.007	0.008	0.004	−0.003	0.004	−0.014	−0.009	0.046 ***
	(0.021)	(0.011)	(0.011)	(0.011)	(0.010)	(0.009)	(0.009)	(0.013)
lnGDPit		0.918 ***	0.913 ***	0.927 ***	0.935 ***	0.795 ***	0.879 ***
		(0.023)	(0.023)	(0.024)	(0.020)	(0.025)	(0.038)
lnIndustryit			0.138 ***	0.278 ***	0.161 ***	0.237 ***	0.152 ***	0.657 ***
			(0.042)	(0.060)	(0.051)	(0.049)	(0.056)	(0.073)
lnOpenit				0.155 ***	−0.059	0.110 ***	−0.055	0.347 ***
				(0.049)	(0.041)	(0.039)	(0.043)	(0.058)
lnPDit					0.251 ***	0.234 ***	0.205 ***	0.356 ***
					(0.016)	(0.015)	(0.018)	(0.023)
lnInternetit						0.191 ***	0.201 ***	0.382 ***
						(0.023)	(0.023)	(0.031)
GEit							0.180 ***	0.879 ***
							(0.061)	(0.056)
Constant	1.046 ***	6.734 ***	7.112 ***	6.993 ***	5.998 ***	5.363 ***	6.246 ***	1.313 ***
	(0.226)	(0.229)	(0.257)	(0.257)	(0.223)	(0.224)	(0.372)	(0.250)
N	576	576	571	571	571	559	559	559
R2	0.000	0.732	0.730	0.735	0.817	0.545	0.840	0.685
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note: *** indicate significant at 1%.

illustrate the foundational regression outcomes for CO₂ emissions from economies along the “Belt and Road” following China’s OFDI. The models represent the projected outcomes from incrementally incorporating control variables like economic size, industrialization level, openness to the outside world, population density, government effectiveness, and informatization level.

The coefficient of ln_OFDI in model (1) is positive, but not significant, when other control variables are not added. Thus, control variables are sequentially incorporated into the subsequent regression models. Upon inclusion of ln_GDP, the empirical analysis reveals that this variable demonstrates a consistently positive and statistically significant coefficient at the 1% level, consistent with the prevailing academic consensus that economic growth correlates positively with carbon dioxide emissions prior to reaching a carbon peak. The coefficient for ln_Industry suggests that a 1% increase in industrialization corresponds to a 0.152% elevation in CO₂ emissions, exhibiting a statistically significant effect at the 1% level. This indicates that the acceleration of the industrialization process does lead to an increase in carbon emissions. The ln_Open coefficient is negative, which confirms that trade among economies along the Belt and Road is beneficial. Although the correlation coefficient is not significant, it indicates that trade and openness can help reduce environmental pollution (Zhang et al., 2021) [40]. The ln_PD coefficient is significantly negative at the 1% level, and CO₂ emissions will decrease by 0.205% when population density increases by 1%. This may be because cities with higher population density are typically associated with greater economic development and technological innovation, leading to a reduction in carbon emission intensity. (Shang et al., 2023) [41] At a 1% significance level, the ln_Internet coefficient shows a positive trend, possibly due to the information and communication sector’s high electricity demand, resulting in elevated carbon dioxide emissions. Consequently, China ought to focus more on eco-friendly and low-carbon growth through collaborative information efforts with economies on the “Belt and Road” initiative. The GE coefficient turns out to be negative and to successfully meet the 1% significance threshold. A plausible reason could be that governments have fostered low-carbon growth through the encouragement of low-carbon product consumption, prompting businesses to launch eco-friendly products, and integrating low-carbon development tactics with market strategies.

In view of the fact that carbon emissions mainly come from the industrial sector, Zhuang and Fu (2021) [6] refer to the industrialization level as an alternative indicator of economic scale and include it into the analysis of model (8) to directly compare the impact of industrialization process along the Belt and Road on carbon emissions of economies. Findings reveal a notable and positive ln_Industry coefficient, suggesting a 0.657% rise in carbon dioxide emissions for each 1% increment in industrialization. Currently, the ln_OFDI coefficient remains notably negative, underscoring China’s substantial impact on reducing CO₂ emissions from economies along these routes, aligning with the strategic objective of constructing an eco-friendly “Belt and Road”.

Regarding particular choices, China maintains eco-friendly environmental standards for investing in infrastructure construction and advocates for green, low-carbon growth. In the field of science and technology, China actively encourages enterprises with the “green and low-carbon” label to “go global”. For example, the Karot hydropower station built by China in cooperation with Pakistan adopted the world’s strictest environmental and social responsibility standards, provided more than 2000 jobs during the construction period, and provided over 3.1 billion KWH of clean energy for the local area, advocating for the synchronized growth of regional energy and industry. In 2016, the Teda Cooperation Zone, jointly built by China and Egypt, adopted green environmental protection as its standard and became a new platform for Chinese companies to explore low-carbon and environmentally friendly commercial applications, such as “seawater desalination” and “desert greening” in the local area, injecting new vitality into the Belt and Road Initiative. These instances clearly illustrate China’s collaborative efforts with Belt and Road nations in establishing an eco-friendly Belt and Road, thereby reinforcing the empirical findings of this study.

3. Quantile Regression (QR): It is considered that the traditional regression model based on the least squares estimation method has a bias in the estimation result due to the heteroscedasticity. Therefore, a quantile regression model is introduced in this paper. Even when the independent variable has different effects on different parts, the model can still describe the conditional distribution characteristics of the dependent variable more comprehensively. In addition, the coefficient of quantile regression is more stable and less susceptible to extreme value perturbations than that of the least squares method. The outcomes of the quantile regression are displayed in

Table 7. Quantile regression results of China’s OFDI and carbon emissions.

	QR 10%	QR 20%	QR 30%	QR 40%	QR 50%	QR 60%	QR 70%	QR 80%	QR 90%
lnOFDIit	0.033 **	0.046 ***	0.027 *	0.011	0	−0.029 *	−0.053 ***	−0.049 ***	−0.048 ***
lnGDPit	0.762 ***	0.917 ***	0.942 ***	0.916 ***	0.903 ***	0.861 ***	0.824 ***	0.761 ***	0.746 ***
lnIndustryit	0.623 ***	0.359 ***	0.099	−0.016	−0.067	0.062	0.132 *	0.272 ***	0.192 ***
lnOpenit	0.031	0.006	−0.014	0.054	0.089	0.008	−0.042	−0.162 ***	−0.096 *
lnPDit	−0.19 ***	−197 ***	−0.215 ***	−0.213 ***	−0.223 ***	−0.219 ***	−0.214 ***	−0.178 ***	−0.162 ***
lnInternetit	0.281 ***	0.217 ***	0.177 ***	0.18 ***	0.172 ***	0.173 ***	0.201 ***	0.158 ***	0.13 ***
GEit	0.017	−0.174	−0.188 *	−0.242 ***	−0.233 ***	−0.151 *	−0.126 *	−0.093 ***	−0.164 ***
Constant	8.58 ***	8.63 ***	7.27 ***	6.65 ***	6.22 ***	5.53 ***	4.99 ***	4.33 ***	4.09 ***

Note: ***, **, and * indicate significant at 1%, 5%, and 10% levels, respectively.

and Figure 2.

The lnOFDI coefficient successfully meets the significance threshold of 1% at 20%, 70%, 80%, and 90%, and 5% at 10%, according to Table 6. In the traditional regression model (7), no direct explanation can be given for the result with a negative but not significant lnOFDI coefficient. However, by introducing quantile regression, the trend changes in the lnOFDI coefficient under the influence of different control variables can be directly reflected: the coefficient between 10% and 40% is positive, the coefficient between 50% is 0, and the coefficient between 10% and 20% is significant and then becomes insignificant. The coefficients between 60% and 90% of the sub-sites are negative and have significance from 70%. A possible explanation for this phenomenon is that economies with lower levels of carbon emissions themselves have stronger environmental regulations and commensurate levels of green development technology. Consequently, the ripple effect of green technology on China’s investments in comparable economies is minimal, and its influence on carbon emissions is negligible, potentially resulting in a rise in carbon emissions. As carbon dioxide emissions rise, China’s OFDI elasticity shifts from positive to negative, signifying that economies on these routes have successfully reduced their carbon emissions through the absorption of Chinese investments, with this reduction effect being more pronounced in economies where carbon emissions exceed 60%.

In conclusion, China’s commitment to the Belt and Road economies epitomizes the idea of synchronized economic and environmental growth, underscoring the ethos of collaboration and collective building of the Green Belt and Road. This stands in stark contrast to the malevolent conjectures of certain Western news outlets regarding the creation of “pollution refuges” as a result of China’s shift away from heavily polluting sectors.

4.5. Difference-in-Difference Method

The difference before and after the implementation of the policy is controlled by the differential model so as to effectively evaluate the effect of the policy. In the research, an experimental group and a control group need to be selected. At a certain moment, a particular policy will impact the experimental group, whereas the control group will remain unaffected by it. Prior to and following this specific moment, a comparison was made between the developmental alterations in the experimental and control groups. Should the alterations in the test group distinctly vary from those in the control group, one can deduce that the policy’s impact is substantial.

Based on the aforementioned guidelines, 2013 is selected as the timeline for the “Belt and Road” project, nations participating in the “Belt and Road” project are considered the test group, and those not involved are regarded as the control group. The difference-difference model is as follows:

$${lnY}_{it}={\beta }_0+{\beta }_1{treat}_i·{post}_t+\gamma {lnX}_{it}+{\epsilon }_{it}$$

(2)

In Formula (2), the independent variable is lnY_it. the primary explanatory factor is treat_i·post_t, and its corresponding coefficient is β₁. treat_i is used to divide the processing group and control group according to whether countries along the Belt and Road are involved (1 for countries along the Belt and Road; 0 if it is not). post_t represents the policy implementation period of the dummy variable. Set 2013 and subsequent years to ${\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}}_{\mathrm{t}}$1, and pre-2013 years to 0. β₀ is the constant term, X_it represents the matrix of control variables, and ${\epsilon }_{it}$ denotes the term for random errors. Data are selected based on their availability, and the time span is set from 2005 to 2020. In addition to the original economies along the “Belt and Road”, the pertinent data pertaining to different economies or regions are chosen.

To confirm the validity of the parallel trend hypothesis, carbon dioxide emissions from economies situated along the Belt and Road were analyzed against those not on the Belt and Road, both before and after the Initiative’s initiation, serving as a control group and a treatment group. A minimal variance in carbon emissions between the experimental and control groups signals that the requirements for the parallel trend test have been met. Consequently, alterations in the sample trends preceding and following the introduction of the “Belt and Road” policy in 2013 were analyzed.

Through the regression of the sample data using the differentially applied method, we pay attention to the positive and negative cross-term coefficients of the treatment group to judge whether the Belt and Road Initiative influences the carbon footprint of economies situated along these routes. The outcomes of the regression study are presented in

Table 8. Regression results of difference-in-difference method.

	ln_CO2	ln_CO2
did	0.021 ***	0.010 ***
	(0.002)	(0.002)
lnOFDIit		0.066 ***
		(0.004)
lnGDPit		0.135 ***
		(0.026)
lnIndustryit		0.269 ***
		(0.032)
lnOpenit		0.246 ***
		(0.032)
lnPDit		0.304 ***
		(0.040)
lnInternetit		0.198 ***
		(0.012)
GEit		0.091 ***
		(0.031)
Constant	1.130 ***	8.417 ***
	(0.157)	(0.644)
Observations	1216	1088

Note: *** indicate significant at 1% level.

.

Once control variables are factored in, the negative regression coefficient for the cross and multiplier term at a 1% significance level suggests China’s outbound investment via the “Belt and Road” route is continuously committed to reducing pollution together with economies along the route, and fervently advocates for the adoption of strategies to reduce carbon emissions, embodying the robust principle of eco-friendly development. This result further validates China’s efforts in reconciling economic growth with environmental protection.

4.6. Robustness Test

Emissions of methane, nitrogen oxide, and overall greenhouse gases were chosen as substitute measures to thoroughly evaluate China’s OFDI’s effect on the economies along these routes and to conduct a thorough analysis of air quality and environmental states. These metrics serve as substitutes for the regression analysis of the initial variables, aiming to achieve stronger empirical findings. The outcomes of the regression study are presented in

Table 9. Regression results of robustness test.

Variables	ln_Methane	ln_Nitrousoxide	ln_TGG
lnOFDIit	0.020 ***	0.017 ***	0.075 ***
	(0.016)	(0.017)	(0.018)
lnGDPit	0.243 ***	0.692 ***	0.006
	(0.069)	(0.072)	(0.074)
lnIndustryit	1.821 ***	1.110 ***	1.604 ***
	(0.099)	(0.104)	(0.107)
lnOpenit	1.547 ***	1.024 ***	1.396 ***
	(0.075)	(0.078)	(0.081)
lnPDit	0.018	0.003	0.033
	(0.032)	(0.033)	(0.034)
lnInternetit	0.166 ***	−0.055	−0.057
	(0.043)	(0.045)	(0.046)
GEit	0.423 ***	0.961 ***	0.544 ***
	(0.108)	(0.113)	(0.116)
Constant	10.806 ***	13.832 ***	10.375 ***
	(0.661)	(0.692)	(0.714)
N	607	607	607
R2	0.637	0.453	0.495

Note: Standard errors in parentheses, *** represents p < 0.01.

.

The regression results showed that ln_OFDI had a significant negative correlation with methane emissions, nitrogen oxide emissions, and total greenhouse gas emissions, and no positive or negative coefficient changes were observed compared with the baseline regression. To be precise, a 1% rise in China’s OFDI leads to a 0.02% reduction in methane gas emissions, a 0.017% drop in nitrogen oxide gas emissions, a 0.075% decrease in overall greenhouse gas emissions, and a 0.014% fall in total carbon dioxide emissions. As a result, the significant impact of China’s OFDI in improving pollution levels and air quality in the “Belt and Road” economies is somewhat clear, underscoring the model’s robustness.

Compared with previous ones, a significant portion of the pertinent research on eco-friendly development within the Belt and Road Initiative is concentrated on the provinces or cities of China (Yang and Wang, 2018; Zhang, 2018; Liu et al., 2022) [1,2,3], while the research on host countries is limited. This study utilizes recent data on Chinese direct investments along the Belt and Road to enhance current research by examining the determinants and levels of carbon emissions in the Belt and Road economies. This study aims to establish a comprehensive analytical framework for systematically examining the theoretical linkages between China’s Outward Foreign Direct Investment (OFDI) and its implications for carbon emission trajectories and environmental quality under the Belt and Road Initiative (BRI), thereby informing strategic policy formulation to advance sustainable, low-carbon development pathways across BRI partner economies.

Further research strengthening this discussion would involve “regional standardization of laws” and “unobserved factors”. Specifically speaking, regional standardization of laws refers to the process within a certain geographical area where countries or regions gradually eliminate legal differences through consultation, cooperation, and other means to achieve legal coordination and unity. This process aims to promote economic, social, and cultural exchanges within the region, enhance legal efficiency, reduce transaction costs, and thereby advance the regional integration process of carbon emission reduction related to investment. At the same time, unobserved factors may include psychological factors, technical limitations, random events, and other factors that are difficult to quantify. These unobserved factors may sometimes have a significant impact on the results of this study. In-depth research will further delve into the micro level and expand to some unobserved factors.

5. Conclusions and Policy Recommendations

Findings from this study indicate that global direct investments could escalate carbon emissions in economies with lower carbon emissions; yet, with the rise in the host nation’s carbon emissions, the ripple effect of international direct investments in green technology becomes increasingly evident. Empirical evidence indicates that global direct investment in Belt and Road economies demonstrates a significant mitigating effect on carbon emissions, thereby amplifying the decarbonization benefits associated with such cross-border capital flows. Research indicates that China’s immediate investments in Belt and Road economies substantially reduce carbon emissions. This stands in opposition to certain Western media claims suggesting it might result in a “pollution refuge” phenomenon. Additional quantile regression analysis shows a distinct variation in the effects of China’s OFDI across economies with varying carbon emission rates. In particular, economies with reduced carbon emissions might see a rise in their carbon output due to Chinese investments, as these economies have advanced in green technology and implemented more stringent environmental rules. Yet, with the rise in carbon emissions from Belt and Road economies, China’s OFDI elasticity coefficient shifts from positive to negative, thereby reducing carbon emissions. This aligns with China’s strategy of “technology spillover” into the Belt and Road’s less developed regions.

At the same time, through the separate analysis of each variable affecting carbon emissions, we can find the positive impact of government effectiveness on emission reduction and provide a basis for formulating relevant policies. The above research has important practical significance for China to promote the global green transition and achieve climate change goals, and will help build a more environmentally friendly and sustainable international cooperation mechanism. In summary, we propose the following policy proposals to propel the Belt and Road Initiative towards improved quality in green and sustainable development:

• Differentiated outbound investment policies based on carbon emission levels: Based on quantile regression results, the impact of China on economies with varying levels of carbon emissions could be contrary. Therefore, policy design should reflect the above differences. With “green development” as the core, it is imperative to bolster collaboration in science and technology and encourage the conversion and implementation of scientific and technological advancements, thus contributing to the reduction of carbon emissions and fostering the creation of an eco-friendly Belt and Road.
• Pay attention to government efficiency, a key factor in promoting carbon emission reduction: Empirical results show that carbon emissions are significantly inhibited by the improvement of government efficiency. Consequently, China has the potential to enhance governmental engagement and policy discussions, forge strong economic connections along the Belt and Road, and collaboratively intensify technical efforts in reducing carbon emissions within this area.
• Matching carbon emission reduction targets to optimize industrial structure of economies along the route: Observational data indicate that the escalation in CO₂ emissions stems from the industrialization of economies situated along this path. Attaining the goal of reducing carbon emissions requires a harmonious alignment between environmental regulations and the fine-tuning of the industrial framework for outbound investments.
• Vigorously promote the successes of China’s OFDI in enhancing the environment: It is imperative for China to amplify the visibility of its OFDI’s accomplishments in bettering the Belt and Road economies through outbound investment. For example, China’s investment in the Zanatas wind power project has effectively solved the power shortage problem in southern Kazakhstan and set an example for the greening of the energy system. China’s introduction of green technological innovation in the Colombo Port City project in Sri Lanka has effectively protected the ecological environment of the Port City’s coral community. These instances comprehensively illustrate the eco-friendly growth strategy of the Belt and Road Initiative that has been executed by China through investments across various sectors.