Foreign Direct Investment, Technology Innovation and Carbon Emissions: Evidence from China

Abstract

FDI is a critical factor influencing carbon emissions and a significant driver of China’s economic development. However, achieving sustainable economic development remains a major challenge for China. Hence, this paper aims to explore how to foster positive interactions between FDI and carbon emissions. Specifically, we first analyze the mechanism of FDI on carbon emissions from a theoretical perspective. Then, using panel data from 27 provinces in China, an empirical analysis is conducted. In the empirical analysis, we use the panel regression models to analyze the impact of FDI on carbon emissions. Additionally, a configuration analysis method is employed to examine the interactive relationship between FDI quality and carbon emissions. The conclusions of this paper are as follows. Overall, FDI significantly inhibits carbon emissions across provinces, with this effect strengthening as the scale and quality of FDI increase. Heterogeneity analysis shows that the inhibitory effect of FDI on carbon emissions is more pronounced in provinces with high technological financial support and stringent environmental regulations. Mediation analysis indicates that technological innovation serves as a mediator between FDI and carbon emissions, which means that FDI could promote “the quality improvement and the quantity increase” of green technological innovation to reduce carbon emissions. Furthermore, the configuration analysis shows that the carbon reduction effect of FDI results from the combined influence of various factors. Among those, FDI’s export capacity is a key factor. The findings above enhance our knowledge of the environmental effects of FDI from the perspective of FDI quality. Moreover, these explorations also offer new insights and strategies for China’s pursuit of sustainable economic development.

1. Introduction

Since the 21st century, the greenhouse effect has gradually become one of the most severe global issues [1]. On 3 April 2023, the Inter-governmental Panel on Climate Change (IPCC) released the synthesis report of its Sixth Assessment Report, “Climate Change 2023”. The report indicates that global warming is inevitable, the window of opportunity is closing, but hope remains [2]. This report issued a red alert for climate change and identified carbon dioxide as the primary driver. China, the largest developing country and the highest carbon emitter globally, has implemented various measures to combat climate change in recent years. However, balancing climate action with economic growth still presents a significant challenge in China’s pathway to achieve sustainable development [3].

The reform and opening-up policy initiated in 1978 opened up China, resulting in rapid economic growth [4]. In an open economic environment, the scale of foreign direct investment (FDI) in China continuously expands. According to the Report on Foreign Investment in China by the Ministry of Commerce, China continues to see growth in foreign capital absorption in 2022. The actual use of foreign capital reaches USD 189.13 billion, ranking among the highest in the world. Thanks to a stable business environment and a vast domestic market, China has consistently been a hotspot for global transnational investment. For three consecutive years, its share of actual foreign capital use exceeds 10% of global transnational investment.

FDI has played a significant role in China’s economic development. However, the effects of FDI on the environment are uncertain. Even though FDI promotes economic development in China, it may also lead to increased carbon emissions. If we regard the environment as a production factor, countries with low environmental protection efforts typically possess abundant environmental resources, whereas countries with high environmental protection efforts may face a corresponding scarcity. As a result, countries with low environmental protection efforts will capitalize on their abundant environmental resources to specialize in the production of pollution-intensive goods [5,6]. Given that the environmental standards established by developed countries are generally higher than those in developing nations, pollution-intensive industries are likely to migrate to developing countries, causing adverse environmental impacts and increased carbon emissions. For China, as a developing country, the substantial influx of FDI may indeed lead to higher carbon emissions. In reality, China’s environmental standards are lower than those of developed nations, and the transfer of pollution-intensive industries to China has been observed [7]. Specifically, the varying quality of foreign-invested enterprises and the decentralization of local governments lead to inevitable competition among local officials. This competition drives local governments to attract liquid resources by any means necessary to boost local economies. As a result, they lower the quality standards for foreign investment to attract more capital. While this approach fosters rapid economic growth, it could adversely affect ecological sustainability due to the uncertain environmental impact of FDI. Fortunately, the government has recognized this issue [8]. Since 2012, China’s economy has entered a period of new normal, which brings an opportunity for the development of a green and low-carbon economy and environmental protection. Consequently, the standards for attracting foreign investment have shifted from purely capital-driven investors to those that enhance the combination of innovative elements and improve industrial chains, shifting from a focus on scale and short-term efficiency to an emphasis on innovation-driven growth. In 2021, the Ministry of Commerce issued the “14th Five-Year Plan for the Utilization of Foreign Investment”. The plan emphasizes promoting high-level opening-up and more effective use of foreign investment. Furthermore, it sets a vision target for 2035, aiming for a significant improvement in both the level and quality of foreign investment. Under this policy, China achieved both qualitative and quantitative improvements in FDI, in which high-tech industries emerged as key growth areas. As illustrated in Figure 1, while the overall scale of FDI in China, in actual use, continues to grow, foreign investment in China’s high-tech industries also experienced a substantial increase. Foreign investment in high-tech sectors came in at USD 683.5 billion in 2022, up 31% year-on-year and accounting for 36.1% of the total. Conversely, in 2015, the FDI in high-tech industries amounted to just USD 17.7 billion, accounting for only 14.3% of the total [9]. So, when China’s approach to attracting foreign investment changes, what impact will FDI have on the environment in China? Specifically, what effects will it have on carbon emissions? Can China foster a positive interaction between FDI and carbon emissions through the attraction of high-quality FDI?

In the current context, green and sustainable development is the main theme. The report of the 20th National Congress of the Communist Party of China clearly stated the overall goal of China’s development. It points out that by 2035, China should significantly improve the ecological environment and basically achieve the goal of a beautiful China. In this context, studying the relationship between FDI scale, quality and carbon emissions is, therefore, of great significance.

Thus, the purpose of this paper is to explore how China could achieve a positive interaction between FDI and carbon emissions in the macro context of sustainable development. In other words, by incorporating FDI quality and other policy factors, we aim to investigate China’s pathway toward sustainable development. Specifically, to clarify how FDI quality affects its environmental outcomes, we apply threshold models and dynamic panel qualitative comparative analysis (PD-QCA). The threshold model is used to explore the nonlinear relationship between FDI and carbon emissions. Additionally, the PD-QCA method enables us to assess FDI’s impact on carbon emissions with a specific focus on the interaction between various contributing factors.

The possible marginal contributions of this paper are as follows. First, we enhance the understanding of the mechanisms by which FDI affects carbon emissions. It explores how FDI quality impacts carbon reduction effects, thus supplementing existing research. Specifically, we constructed an index evaluation system for FDI quality and used the entropy weight method to measure it. Then, this index is incorporated as a threshold variable in the model to test the nonlinear relationship between FDI and carbon emissions. This approach deepens insights into the environmental effects of FDI. Second, this paper broadens the scope of studies on the impact of FDI on carbon emissions. Most of the existing literature studies focus on linear effects using regression models. Moreover, they often neglect the interaction among factors affecting carbon emissions. We employ the PD-QCA method to analyze how various factors, including FDI, interact to influence carbon emissions. This approach not only enriches existing research content but also presents new findings. Finally, this paper inspires new policy directions. The findings indicate that in provinces with strong technological financial support and environmental regulations, FDI’s inhibitory effect on carbon emissions is more significant. Meanwhile, the carbon reduction effect of FDI results from the combined influence of factors such as FDI scale and quality, with export capacity playing a central role. These findings suggest that local governments could achieve sustainable development by introducing high-quality FDI with robust export capabilities and adjusting policy support, which provides new perspectives and strategies for China to achieve sustainable economic development.

The remainder of this paper is structured as follows. Section 2 is the literature review. In this section, we review the literature related to FDI, carbon emissions and QCA research methods. Section 3 provides a theoretical framework of the relationship between FDI and carbon emissions and proposes our research hypotheses. Section 4 describes the regression model design and data sources. In Section 5, we analyze the empirical results. Section 6 further discusses the empirical results by performing the PD-QCA method. Finally, Section 7 is the conclusion, implication and research limitation. In this section, we lay out the main conclusion of this study and make relevant policy implications. Moreover, the limitations and directions for further research are also included.

2. Literature Review

2.1. The Impact of FDI on Carbon Emissions

The impact of FDI on the environment and carbon emissions in host countries is uncertain. Mainstream views are divided into two categories. The first is the “pollution haven” hypothesis, represented by Copeland and Taylor [10]. The hypothesis posits that in order to avoid the high costs of environmental governance, developed countries tend to transfer high-pollution and high-energy-consuming industries to developing countries. As a result, the environmental pollution and carbon emissions increase in the host countries. A large number of studies support this idea. Wu (2021) noted in his research that due to the cross-border transfer of global value chains, low value-added manufacturing, particularly high-carbon industries, gradually moves from developed countries to developing countries, and the final products are imported from these countries for consumption in developed nations, resulting in increasingly serious environmental problems in developing countries [11]. Some scholars also point out that the increased FDI leads to an increasing rate of environmental degradation, especially if the environmental regulations are inadequate or non-existent [12,13]. However, if developing countries decide to enforce environmental regulations, there will be a corresponding reduction in FDI inflows and, hence, economic growth [14,15].

The second is the “pollution halo” hypothesis. This hypothesis suggests that FDI brings advanced production philosophy and green technologies to host countries, thereby reducing environmental pollution and carbon emissions. By incorporating institutional quality, scholars find that FDI inflows significantly reduce carbon emissions, supporting the “pollution halo” effect of FDI [16].

One reason for the opposing conclusions about the impact of FDI on carbon emissions may be the heterogeneity of the research subjects [17]. On the one hand, the heterogeneity of host countries is a factor. For example, the relationship between FDI and carbon emissions can be influenced by a country’s level of economic development. In economically developed regions, FDI tends to reduce carbon emissions, but this is not the case in less developed regions [18]. The stringency of environmental regulations also plays a crucial role. In countries with strict environmental regulations, FDI inflows reduce carbon emissions, while in countries with lax regulations, FDI inflows increase carbon emissions [19,20]. On the other hand, the heterogeneity of FDI itself is also a factor. For instance, high-quality FDI can help mitigate carbon emissions, whereas low-quality FDI does not have the same effect [21,22].

Specifically, do FDI inflows increase carbon emissions in China? Through the study of industry panel data, scholars identify that the inflow of FDI significantly mitigates carbon emissions in China’s industrial sectors. Moreover, the inhibitory effect of FDI on carbon emissions is more significant in high-output, low-energy-consumption and high-tech industries [23]. However, in other sectors, particularly transportation, the inflow of FDI does not mitigate carbon emissions; instead, it leads to an increase in emissions. Through further analysis, scholars attribute this phenomenon to FDI’s ability to promote capital-intensive technological progress, thereby reducing carbon emissions [24]. Taking into account the heterogeneity at the enterprise level, Zhang (2024) found, through the analysis of panel data from Chinese A-share listed companies, that managers’ climate concerns can significantly reduce corporate carbon emissions and that this relationship is positively moderated by investor attention [25]. This conclusion suggests that attracting foreign-invested enterprises with strong climate concerns may help mitigate carbon emissions in China. It also indicates that if foreign investors focus more on enterprises after the introduction of foreign investment, corporate managers are likely to allocate more resources to carbon reduction, ultimately leading to a more effective mitigation of carbon emissions.

Some scholars also use provincial or prefecture-level panel data to study the relationship between FDI and carbon emissions in China. Through their research, Gao (2024) found that FDI inhibits the improvement of carbon productivity and significantly reduces China’s overall carbon emission levels [26]. Comparable findings are observed at the prefecture level. Meanwhile, scholars also identified a regional heterogeneity in the mitigating effect of FDI on carbon emissions. For example, Lin (2022) and Wu (2023) pointed out that in the eastern coastal cities, this effect is notably more pronounced [27,28]. In further research, scholars discovered that regional heterogeneity is primarily influenced by the environmental policies of different regions. Under varying environmental policies, the carbon emission behaviors of various enterprises, including foreign-invested firms, display distinct characteristics. Hou (2023) found, through the analysis of enterprise-level data, that low-carbon city pilots help reduce corporate carbon emissions [29].

2.2. The Nonlinear Relationship Between FDI and Carbon Emissions

As research methods continue to advance, more and more scholars find that the relationship between FDI and carbon emissions is dynamic. In other words, the relationship between FDI and carbon emissions is not simply linear but more complex and uncertain [30]. In empirical studies of China, Wang (2021) discovered that FDI’s impact on carbon emissions in China’s industrial sector initially decreases and then increases. Similarly, there is an inverted U-shaped nonlinear relationship between FDI and carbon emissions at the provincial level in China [31]. As FDI’s share of GDP grows, its role in driving carbon emissions across China’s provinces initially intensifies but later diminishes. Similarly, in their study of G7 countries, Ren (2024) found that the relationship between FDI and carbon emissions evolves over time. Their research indicates that from 1971 to 1995, there was a negative correlation between FDI and carbon emissions in these countries. In contrast, from 2000 to 2015, the relationship becomes significantly positive. Further research reveals that even after accounting for the moderating effects of factors like trade openness on the FDI-carbon emissions relationship, the conclusion remains valid, underscoring the dynamic nature of the relationship between FDI and carbon emissions [32].

Regarding the nonlinear relationship between FDI and carbon emissions, many scholars argue that the nonlinear relationship is driven by the presence of certain threshold conditions. Whether FDI inflows promote or inhibit carbon emissions mainly depends on whether the host country meets certain economic and social development thresholds. Under these threshold conditions, the impact of FDI on carbon emissions could take various forms, such as U-shaped or inverted U-shaped relationships. The recent literature shows that scholars often choose threshold variables such as human capital levels, technological innovation levels, environmental regulation intensity and economic development levels [33,34,35]. For example, Huang (2018) and Zhang (2020) employed the local technology absorption level as the threshold variable [33,35]. Their study discovered that the positive spillover effect of FDI on carbon intensity increases as the number of local full-time R&D personnel rises. Meanwhile, some scholars explored the nonlinear relationship between FDI and carbon emissions by incorporating environmental regulations into their analysis. Their conclusion shows that under the moderating effect of environmental regulation, FDI significantly reduces carbon emission intensity.

2.3. The QCA Method

As a set-theoretic method based on Boolean algebra, QCA aims to analyze the relationships between configurations of elements and outcome variables. It is designed to address the causal complexity in social phenomena [36]. Specifically, first, compared to traditional methods of causal inference, configuration analysis can consider the combinations of multiple variables simultaneously, while traditional approaches often focus on linear relationships between individual variables. This makes configuration analysis more effective for analyzing complex phenomena, such as social, economic or environmental systems [37]. Second, configuration analysis emphasizes conditional relationships, indicating that the effects of variable combinations may differ across various contexts [38]. In other words, it allows for the same outcome to result from different combinations of variables, contrasting with traditional causal inference, which typically seeks a single causal pathway. This diversity better reflects real-world situations. In recent years, numerous studies have used QCA as an empirical analysis method. Management is the field where QCA has grown the fastest, and its use in international trade research is also rapidly increasing. For example, QCA has been applied in empirical studies on various international trade-related topics. These topics include compliance in trade disputes [39], trade agreements [40], logistics performance [41], trade facilitation [42], embodied carbon in trade [43], FDI entry modes [44] and trade liberalization [45].

Depending on the consideration of the time dimension, QCA methods could be divided into non-dynamic and dynamic QCA. Dynamic QCA methods could directly examine configurations at different points in time, including multi-period QCA and panel data QCA(PD-QCA). Multi-period QCA divides the study period into several segments, derives configuration results for each period and compares these results across periods [46]. PD-QCA was first put forward by Garcia-Castro and Ariño in 2016 [47]. It expands the original consistency concept into three types: pooled consistency (POCONS), between consistency (BECONS) and within consistency (WICONS). Compared to non-dynamic QCA, dynamic QCA methods, especially PD-QCA, effectively address issues arising from the “time blindness” of traditional QCA methods. This improvement is significant in both theory construction and empirical testing [48].

By reviewing existing research, it can be found that scholars have conducted extensive studies on how FDI affects China’s carbon emissions. Differences in research methods, data and models have led scholars to different conclusions. Some argue that FDI increases carbon emissions, while others contend that it reduces them. Additionally, some scholars also believe that the relationship between FDI and carbon emissions is nonlinear.

Despite the varying conclusions among scholars, it is noteworthy that existing research has thoroughly examined the relationship between FDI and carbon emissions, providing a solid empirical foundation for further study. However, some deficiencies remain. First, most studies use ARDL models, fixed effects models and spatial econometric models to examine the relationship between FDI and carbon emissions relationship. There is little use of QCA models to analyze potential complex causal relationships under multiple influencing factors. Second, most scholars typically choose threshold variables based on the economic and social characteristics of the host country. There is less focus on the nonlinear relationship between FDI and carbon emissions influenced by FDI’s own characteristics. Particularly, the impact and mechanism of FDI quality on carbon emissions are often overlooked. Third, most scholars classify regions by geographical location (east, west, south, north and central) to perform grouped regression analysis. However, this approach may be overly simplistic, as it fails to fully consider differences in economic development levels and policy environments across regions. And the potential influence of these differences on the FDI’s impact on the carbon emissions relationship is also overlooked.

To overcome the limitations of prior research, we explore the following areas. First, we incorporate the QCA method alongside regression analysis. This approach allows us to explore the complex causal relationships between FDI and carbon emissions influenced by multiple factors. Second, we use FDI quality as a threshold variable in threshold regression. This approach allows us to explore the nonlinear relationship between FDI and carbon emissions. Finally, we take the regional differences in economic development and policy environments into account. Specifically, we investigate how factors like technological fiscal support and environmental regulation impact the relationship between FDI and carbon emissions.

3. Theoretical Framework and Hypothesis Development

3.1. The Relationship Between FDI and Carbon Emissions

The impact of FDI on carbon emissions is uncertain. On the one hand, developed countries are bound by stricter environmental targets. To minimize domestic environmental management costs and maximize profits, they tend to relocate high-pollution industries to developing countries via FDI. This transfer leads to increased environmental pollution and higher carbon emissions in these developing countries. Meanwhile, driven by economic development goals, developing countries often lower ecological regulations to attract foreign investment. This reduction in standards leads to the rapid expansion of domestic polluting industries. As a result, more severe environmental pollution and higher carbon emissions appear [49].

On the other hand, FDI can bring advanced production concepts and green technologies to host countries, resulting from “demonstration effects” and “technology spillover effects” [50]. Enterprises in the host country can study and adopt the business strategies, management models and R&D technologies of foreign companies. This enhances their environmental awareness and supports the development of sustainable green growth models. As a result, their environmental performance improves, promoting green and low-carbon development while reducing environmental pollution.

However, it is important to recognize that the relationship between FDI and carbon emissions is not merely linear but dynamic. The impact of foreign capital inflows on carbon emissions in the host country can be either promotive or inhibitory. This depends on the host country’s policy environment and the quality of the FDI. Therefore, this paper proposes the following hypothesis:

Hypothesis 1 (H1). 

FDI can impact carbon emissions in China, and this impact is dynamic.

3.2. FDI, Technological Innovation and Carbon Emissions

The inflow of FDI, particularly when it includes advanced technology, can boost technological innovation at both industry and national levels [15,51]. On the one hand, FDI not only brings in capital but often involves technology transfer. On the other hand, foreign enterprises can indirectly influence the technological innovation of host country firms. Specifically, this influence could occur through channels such as the demonstration effect, spillover effect, competition effect and industry linkages.

Foreign firms may face limitations, such as technology levels, that prevent the realization of demonstration effects and spillover effects. However, despite these limitations, they can still drive technological innovation in local firms through the competition effect [52]. Specifically, the entry of foreign capital captures premium resources in the host country’s market, inevitably intensifying competition. As competition increases, it stimulates technological innovation among local firms [53]. The driving mechanism behind this phenomenon is primarily manifested in the two aspects outlined below. On the one hand, innovation could reduce costs and offer firms a competitive advantage. On the other hand, becoming a technological leader could allow firms to earn higher profits than competitors. It is these potential excess profits that motivate local firms to innovate. At last, the drive for innovation ultimately raises the country’s overall technological level.

Technological innovation, particularly in green technologies, could significantly lower carbon emissions. Meanwhile, it is also an effective tool for achieving economic sustainable development [54]. In terms of industrial layout, technological innovation drives improvements in production processes and enhances efficiency. It fosters healthy competition, gradually eliminating outdated production methods. This could facilitate the advancement of high-tech and emerging green industries. As a result, the industrial structure could be optimized, leading to a reduction in carbon emission intensity. For enterprises, increasing innovation capabilities, especially in developing green patents, boosts resource efficiency and facilitates cleaner production [55]. Moreover, enhancing technological innovation awareness and capacity helps transform auxiliary sectors like transportation and construction [56]. This transformation enhances efficiency, modernization and sustainability throughout the industry chain. Finally, it effectively drives sustainable development while reducing overall carbon emissions. Therefore, this paper proposes the following hypothesis:

Hypothesis 2 (H2). 

FDI can reduce carbon emissions by enhancing technological innovation.

3.3. FDI Quality and Carbon Emissions

Low-quality FDI inflows are generally believed to increase carbon emissions in the host country. In contrast, high-quality FDI inflows help mitigate them and are usually seen as a key driver for sustainable development. Specifically, low-quality FDI enters the host country’s market primarily to avoid high environmental management costs in their home country [57]. These investments typically target high-energy-consuming and high-pollution industries to achieve short-term profits. Moreover, by prioritizing cost avoidance and profit maximization, these firms generally neglect pollution control and energy-saving measures. Instead, they focus on expanding production, which exacerbates environmental pollution and increases carbon emissions in the host country.

High-quality FDI inherently includes advanced production technologies and robust management systems. The entry of high-quality FDI introduces these technologies and management systems. This entry could boost the production efficiency of host country enterprises through “demonstration”, “spillover” and “competition” effects. As a result, local businesses are encouraged to pursue green transformations [58]. Moreover, high-quality FDI usually has a stronger sense of environmental responsibility [59]. Both the green transformations and the stronger environmental awareness will contribute to sustainable development and carbon reduction. Thus, this paper proposes the following hypothesis:

Hypothesis 3 (H3). 

The quality of FDI can influence how FDI impacts carbon emissions.

Based on the above analysis, this paper suggests that FDI can directly impact carbon emissions. And FDI can also exert an indirect influence by enhancing technological innovation. Specifically, FDI can help mitigate carbon emissions by enhancing both the quantity and quality of green technology innovation in China. Moreover, the effectiveness of FDI in affecting carbon emissions is influenced by the quality of the FDI itself. As FDI quality influences the situation, the impact of FDI on carbon emissions will show dynamic changes. With improvements in FDI quality, its mitigation effect on carbon emissions will become more pronounced and significant. The theoretical model of this paper is illustrated in Figure 2. In Figure 2, H1, H2 and H3 represent the three hypotheses proposed in this paper, where (+) indicates a promoting effect, (−) indicates a suppressive effect, and (+/−) suggests that the effect remains uncertain and requires validation through additional empirical analysis.

4. Research Method and Data

4.1. Research Method Design

4.1.1. Benchmark Regression Model

To examine the direct impact of FDI on carbon emissions in China’s provinces, based on the analytical model by Xu (2022) [59], this paper first sets the baseline model as follows:

$${LC}_{it}={\beta }_0+{\beta }_1{FDI}_{it}+{\beta }_2{LABOR}_{it}+{\beta }_3{GDP}_{it}+{\beta }_4{TAX}_{it}+{\beta }_5{GOI}_{it}+{{\beta }_6{HCL}_{it}+{\beta }_7{ISR}_{it}+\mu }_{ij}$$

(1)

In Equation (1), the dependent variable ${Lc}_{it}$ denotes the carbon emissions level of province i at time t. The key explanatory variable ${FDI}_{it}$ reflects the actual level of foreign capital utilized in province i at time t. Control variables ${LABOR}_{it}$, ${GDP}_{it}$, ${TAX}_{it}$, ${GOI}_{it}$, ${HCL}_{it}$ and ${ISR}_{it}$ represent the labor productivity, economic development level, tax burden, degree of government intervention, human capital level and industrial structure rationalization in province i at time t, respectively. The coefficients ${\beta }_i$ capture the effect of each variable, while ${\mu }_{ij}$ denotes the random disturbance term.

4.1.2. Mediation Mechanism Test Model

Secondly, to explore the mechanism by which FDI influences carbon emissions, this paper constructs a mediation effect model. The model is developed within the framework of Equation (1) based on prior theoretical analysis and the existing research [60,61]. By incorporating technological innovation as a mediating variable, we analyze the pathways through which FDI affects carbon emissions. The mediation effect model is constructed as follows:

$${TEC}_{it}={\beta }_0+{\beta }_1{FDI}_{it}+{\beta }_2{Control}_{it}+{\mu }_{ij}$$

(2)

$${LC}_{it}={\beta }_0+{\beta }_1{FDI}_{it}+{\beta }_2{TEC}_{it}+{\beta }_3{Control}_{it}+{\mu }_{ij}$$

(3)

In Equations (2) and (3), ${TEC}_{it}$ serves as the mediating variable, denoting the quantity ($Tec-N$) and quality ($Tec-Q$) of green technological innovation, respectively. ${Control}_{it}$ indicates the control variables.

4.1.3. Threshold Regression Model

Finally, considering the possibility of a nonlinear relationship between FDI and carbon emissions under the influence of FDI scale and quality, this paper constructs a threshold regression model. The model is developed based on Hansen’s threshold model [62,63], as shown below:

$${LC}_{it}={\beta }_0+{\beta }_1{FDI}_{it}I(q<r)+{\beta }_2{FDI}_{it}I(r\le q)+{\beta }_4{Control}_{it}+{\mu }_{ij}$$

(4)

Equation (4) depicts a single-threshold model, where the threshold variable $q$ denotes both the FDI scale ($FDI-s$) and FDI quality ($QFDI$). $I(\ast )$ is the indicator function, and $r$ is the threshold value. ${\beta }_1$ and ${\beta }_2$ are the coefficients indicating the impact of FDI on carbon emissions when the threshold variable satisfies $q\le r$ and $q>r$, respectively.

4.2. Variable Selection

4.2.1. Explained Variable: Province Carbon Emission Level ($LC$)

In this paper, the dependent variable, carbon emission level, is measured in terms of carbon emissions. The data are primarily obtained from the China Carbon Emission Database (CEADs). This database focuses on establishing a multi-scale carbon emission accounting framework. CEADs calculates carbon dioxide emissions for 30 provincial-level administrative regions in China using the accounting methods of the Intergovernmental Panel on Climate Change (IPCC). It offers detailed carbon accounting data broken down by socio-economic sectors and energy types and qualities. The CEADs inventory covers 45 production sectors and 2 residential sectors across 17 types of fossil fuels. This coverage ensures comprehensive and detailed data that accurately reflect regional carbon emission levels in China [64,65,66].

4.2.2. Core Explanatory Variable: FDI ($FDI$)

The level of actual foreign capital utilization in this paper is measured by the flow of non-financial FDI utilized in each province during the year [15], serving as the core explanatory variable.

4.2.3. Mechanism Variable

Technological Innovation ($TEC$). This paper uses technological innovation as a mediating variable based on the existing research [67,68]. It encompasses both the quantity and quality of green technological innovation. The inflow of FDI, especially technology-intensive FDI, aids in elevating the level and capacity for technological innovation across industries and the entire nation. Green technological innovation is an effective strategy for reducing environmental pollution and achieving economic sustainable development. It could significantly decrease carbon emissions. Furthermore, the enhancement of technological innovation by FDI in the host country occurs progressively. It improves both quantity and quality of the innovation. Accordingly, this paper considers both the quantity and quality of green technological innovation as mediating variables. The quantity of green technological innovation ($TEC-N$) is measured by the number of green patents obtained by each province annually. The quality of green innovation ($TEC-Q$) is defined as the ratio of green invention patents to green utility model patents obtained by each province annually [50].

Scale of Actual Foreign Capital Utilization ($FDI-S$). While FDI flow influences carbon emissions, the stock of FDI also plays a crucial role [69]. The carbon emission levels of provinces are affected by both current and past FDI. This is because the technical and environmental characteristics of FDI could vary at different stages. Previous FDI could impact how current FDI affects carbon emissions. Therefore, this paper uses the non-financial foreign capital actually utilized as a threshold variable to examine the nonlinear relationship between FDI and carbon emissions.

FDI Quality ($QFDI$). The quality of FDI influences its environmental impact. High-quality FDI could raise the technological capabilities of local enterprises in the host country through technology spillover and competition effects. This improvement, in turn, improves the host country’s ecological environment [70]. Therefore, this paper uses FDI quality (QFDI) as a threshold variable to examine the nonlinear relationship between FDI and carbon emissions.

Currently, there is no consensus in academia on how to measure FDI quality. Drawing from the operational characteristics of foreign-invested enterprises, this paper follows the approach of Shangguan (2022) [71]. Specifically, we evaluate FDI quality through its profitability, technological level, export capacity and management level. The details are as follows. First, the profitability of FDI. Profitable foreign enterprises achieve better performance, which could lead to increased tax revenue for local governments. This increase in revenue prompts greater investment in ecological improvements. Additionally, such enterprises generate more profits, providing more capital for reinvestment and technological R&D. This investment enhances local industry technology, boosts production efficiency and improves the ecological environment. Therefore, FDI profitability is a key indicator of FDI quality. This paper measures FDI profitability using the ratio of the profit margin of FDI in the industrial sector to the cost-to-profit margin of industries above a certain scale. Second, the technological level of FDI. High-tech FDI could affect host country enterprises through competition and demonstration effects. As a result, their technological innovation capabilities and total factor productivity improve. Thus, the technological level of FDI is a key indicator of FDI quality. This paper measures the technological level of FDI using the ratio of labor productivity in the FDI industrial sector. Labor productivity is represented by the per capita output value of finished products in the industrial sector. Third, the export capacity of FDI. Foreign enterprises with strong export capacity entering the host country could expose local enterprises to larger global markets. This exposure boosts their technological levels through competition. Furthermore, transactions between multinational corporations and their foreign subsidiaries, along with cross-border trade and factor flows, aid in the diffusion and spillover of technology and knowledge. This process enhances the management and technological levels of host country enterprises. Consequently, FDI’s export capacity is a crucial indicator of FDI quality. This paper measures FDI export capacity using the ratio of the FDI industry’s export volume to the total regional export volume. Finally, the management level of FDI. Higher management levels in foreign enterprises result in more comprehensive incentive and constraint mechanisms. These factors better ensure enterprise vitality and competitiveness, stimulating the competition effect among local firms in the host country. Therefore, FDI’s management level is a key indicator of FDI quality. This paper measures the management level of FDI using the ratio of the asset contribution rate of the FDI industrial sector to that of industries above a certain scale.

4.2.4. Control Variable

Labor Productivity ($LABOR$). Increases in labor productivity typically coincide with technological advancements and innovations. This results in more efficient resource use and reduces energy consumption and carbon emissions per unit of output. Specifically, given the same level of input, higher labor productivity allows for greater value creation per worker, with less resource consumption. This increase in efficiency leads to lower carbon emissions. Based on the existing research [72,73], this paper measures labor productivity in each province by the ratio of regional GDP to the number of employed persons. And the variable’s estimated coefficient is expected to be negative.

Economic Development Level ($GDP$). According to the Environmental Kuznets Curve, economic development is a key factor influencing carbon emissions. Typically, as a region’s economic development level increases, so does its energy demand, which, in turn, leads to higher carbon emissions. Based on the existing research [74], this paper measures each province’s economic development level using the regional GDP of each province. And the variable’s estimated coefficient is expected to be positive.

Tax Burden Level ($TAX$). The effect of the tax burden on carbon emissions is uncertain. On the one hand, a high tax burden can increase the cost of carbon emissions, encouraging companies to adopt energy-efficient and emission-reducing practices. It also generates additional tax revenue for local governments to invest in environmental improvements. On the other hand, high taxes may place economic pressure on companies, impeding their technology development and green production efforts. Based on the existing research [75,76], this paper measures each province’s tax burden level by the ratio of tax revenue to regional GDP. And the estimated coefficient of this variable is currently uncertain.

Government Intervention Degree ($GOI$). We measure the degree of government intervention in each province by the ratio of fiscal expenditure to regional GDP [77]. Fiscal expenditure impacts carbon emissions in two primary ways: first, local governments regulate environmental quality, a public good, through fiscal spending. Second, fiscal expenditure affects environmental quality indirectly by driving economic growth. Currently, local fiscal spending in China is primarily directed towards promoting economic growth. It often flows into capital- and labor-intensive industries that are prone to causing environmental pollution. The estimated coefficient of this variable is expected to be positive.

Human Capital Level ($HCL$). This paper employs the share of the employed population with higher education to assess the human capital level across various provinces, following methodologies from existing research [78,79]. Generally, an increase in human capital level contributes to mitigating carbon emissions, as enhancing human capital is essential for improving regional innovation capabilities [80]. Human resources play a critical role in technology transfer, and effectively harnessing green technology innovation to influence carbon emissions necessitates a higher level of human capital [81]. The variable’s estimated coefficient is expected to be negative.

Industrial Structure Rationalization ($ISR$). This paper employs the Theil index to assess the degree of industrial structure rationalization across various provinces, drawing on methodologies from existing research [82]. Industrial structure rationalization reflects the level of inter-industry correlation; a higher degree indicates enhanced coordination among industries, improved integration of input and output structures and more efficient resource utilization [83,84]. The effect of enhanced industrial structure rationalization on carbon emissions manifests is evident in its capacity to foster the efficient allocation of social resources, enhance production factor efficiency—especially in energy utilization—and elevate total factor productivity. Consequently, this leads to reduced energy consumption in production processes and, ultimately, lower carbon emission levels. The variable’s estimated coefficient is expected to be negative.

4.3. Data Source and Descriptive Statistics

4.3.1. Data Source

This paper uses panel data from 2009 to 2021 for 27 provinces and autonomous regions in mainland China for empirical analysis. Regarding the selection of the research object and research period, first, the primary reason for choosing the period from 2009 to 2021 as the research interval is that significant changes have occurred in both the targets and methods of attracting foreign investment in China during this time. On the one hand, China has started to prioritize the quality of foreign investment and its environmental impacts; on the other hand, the approach to attracting foreign investment has transitioned from providing policy incentives to facilitating investments. Specifically, regarding the targets for foreign investment, before 2012, China’s approach primarily focused on efficiency. Since 2012, China’s economy has entered a period of new normal, which brings an opportunity for the development of a green and low-carbon economy and environmental protection. Consequently, the standards for attracting foreign investment have shifted from purely capital-driven investors to those that enhance the combination of innovative elements and improve industrial chains, shifting from a focus on scale and short-term efficiency to an emphasis on innovation-driven growth. At the same time, the policies for attracting foreign investment also changed. The unified Enterprise Income Tax Law was implemented in China on 1 January 2008, and the Anti-Monopoly Law was also officially enacted that year. China started to adopt globally accepted practices, shaping a fair competitive market and regulatory environment through legislation. The methods for attracting foreign investment shifted from policy incentives to investment facilitation. Therefore, significant changes have occurred in both the targets and methods for attracting foreign investment in China since 2008. In this context, it is justified for this paper to select the period 2009–2021 to examine the relationship between FDI scale, FDI quality and carbon emissions in China. In addition, regarding the choice of research period, another reason for selecting data after 2009 is that actual foreign investment data for Tibet before 2009 are missing. To ensure the integrity and completeness of the research subjects, this paper selects the period 2009–2021. Secondly, regarding the selection of research subjects, the rationale for selecting the 27 provincial-level regions in mainland China, excluding the four municipalities directly governed by the central government, is that people are the primary participants in economic activities. There are considerable differences in population size between municipalities and other provincial-level regions. Moreover, because of the unique status of municipalities, their policymaking differs from that of other provinces. To mitigate the impact of these peculiarities on the research outcomes, this paper selects the 27 provincial-level regions in mainland China, excluding the four municipalities, for analysis.

The carbon emission data for each province are primarily sourced from the CEADs. Data on the actual utilization of foreign capital, regional GDP and technology expenditure for each province are mainly derived from the China Business Statistics Yearbook. Data on employment figures, tax revenue and general fiscal expenditure are mainly derived from the China Regional Economic Statistics Database. The employment population data of each industry and the proportion of people with higher education in the employment population are mainly derived from China’s Labor Economy Database. Additionally, information on operating revenue, main business costs, total assets and employment figures for foreign-invested enterprises and industrial enterprises above a certain scale is mainly obtained from the China Industrial Economy Statistical Yearbook. All data from the statistical yearbooks and databases are accessed via the EPS Data Platform. Green patent data are primarily obtained from the China Research Data Service Platform (CNRDS). And the detailed data sources are also shown in Supplementary Material Table S1: Data description and source.

4.3.2. Descriptive Statistics

Table 1. The descriptive statistics of major variables.

Variable	N	Min.	Max.	Mean.	S. D.	P50
LC	351	35.46	2099.79	396.72	317.46	287.21
FDI	351	1	2,353,187	172,331.8	317,553.9	63,596
LABOR	351	21,242.4	239,284.8	86,058.78	37,398.49	81,451.8
GDP	351	441.36	124,370	24,347.02	22,241.58	17,289.2
TAX	351	0.0419	0.1389	0.0739	0.0176	0.0716
GOI	351	0.0964	1.3337	0.283	0.2041	0.2265
HCL	351	0.4446	30.6000	15.5016	5.6625	15.3797
ISR	351	0.0082	0.4860	0.1841	0.1001	0.1692
Tec-Q	351	0	0.8333	0.2353	0.1207	0.2167
Tec-N	351	1	45,359	3088.0456	5557.2988	1178
QFDI	351	0.0244	0.8739	0.1915	0.1006	0.1789

shows the descriptive statistics for each variable. Carbon emission level ($LC$) ranges from 35.46 to 2099.79, with an average of 396.72, highlighting considerable variation in carbon emission levels among provinces during the sample period. The actual utilization of foreign capital (${FDI}_{it}$) has a maximum value of 2,353,187, an average of 172,332 and a median of 63,596. The FDI exhibits significant gaps between its maximum and minimum values, and its median is relatively low. This suggests that the overall scale of foreign capital utilization is quite low across provinces during the sample period. At the same time, there are substantial differences in FDI scales among provinces. This disparity is similarly observed in other control variables and green technological innovation levels. Moreover, the disparity is particularly notable in technological innovation levels, with the number of green patents (Tec-N) having a maximum of 45,359. The minimum is only one.

Descriptive statistics show significant disparities in the economic and social development levels among the provinces, which is indeed the case. In terms of economic development level, using gross regional product (GRP) as an example, in 2021, Guangdong was the province with the highest GRP among the 27 provinces and autonomous regions, reaching USD 1.9277 trillion—approximately 60 times that of the lowest-ranked Tibet, which had a GRP of only USD 32.2 billion. Specifically, there are notable disparities in economic development among these 27 provinces and autonomous regions. According to regional economic classification, they are divided into four areas: northeast, eastern, central and western regions. In 2021, the total GRP of the eastern region reached USD 764.18 billion, which is 8.85 times, 1.97 times and 2.3 times that of the northeast, central and western regions, respectively. In terms of social development levels, using the proportion of employed individuals with higher education as an example, in 2021, Jiangsu Province had the highest proportion at 27.71% among the 27 provinces and autonomous regions, while Guizhou Province had the lowest at only 14.47%. Based on economic location classification, disparities exist in the average proportion of employed individuals with higher education across provinces in different regions. The average proportions in the northeast and eastern regions exceed 23%, while those in the central and western regions are around 20%.

5. Results

5.1. Benchmark Regression

Before performing the baseline regression, we conduct the F-test, LM test and Hausman test to select the appropriate regression model. The results show that the p-values for all tests are less than 0.001. Based on these results, a fixed effects model is employed for the regression. The baseline regression results are presented in

Table 2. Benchmark regression.

Variable	LC
FDI	−21.04 ***
	(−3.49)
LABOR	−316.6 ***
	(−3.25)
GDP	246.5 **
	(2.55)
TAX	231.0 ***
	(4.23)
GOI	149.3 **
	(2.09)
HCL	−13.77 ***
	(−5.21)
ISR	−721.8 ***
	(−4.25)
_cons	2130.1 ***
	(4.86)
fe-year	Yes
fe-province	Yes
N	351
R2	0.307
F	17.60

Note: *** and ** represent the significant level of 1% and 5%, respectively.

.

In the baseline regression, all variable coefficients passed the significance test. The coefficient for the core explanatory variable, actual foreign capital utilization, is negative at the 1% significance level. This indicates that FDI significantly reduces carbon emissions across provinces. This result partly supports Hypothesis 1, confirming that FDI affects provincial carbon emissions negatively. In other words, it means that foreign capital inflow helps mitigate carbon emissions. The reasons may be as follows. First, compared to developed countries, China, particularly in the central and western areas, lags behind multinational corporations in production technology, environmental awareness and management experience. The inflow of FDI introduces advanced environmental management systems, production technologies and higher environmental standards. As a result, it leads to a carbon reduction effect. Second, as the economy develops, China’s industrial structure becomes more complete. The China Foreign Investment Report indicates that during the study period of this paper, infrastructure in the central and western regions has continuously improved. Moreover, the industrial and supply chain support in China has also been enhanced, making these areas new hotspots for attracting foreign investment. It is these improvements that strengthen China’s capacity to adopt advanced technologies from FDI, effectively enhancing its carbon reduction impact. Finally, since the 18th National Congress of the Communist Party of China in 2014, the Chinese government increasingly prioritizes environmental protection. It intensifies its efforts to promote ecological conservation and the development of economic sustainable development. This approach has provided more investment opportunities for international investors. Additionally, China has raised the environmental protection thresholds for foreign investment. While attracting foreign capital, the country has focused on introducing advanced foreign technologies and environmental standards. As a result, this focus effectively restricts the entry of high-pollution, high-energy-consuming FDI.

The coefficient for the control variable labor productivity is negative at the 1% significance level, which aligns with expectations. It indicates that higher labor productivity helps reduce carbon emissions. With the same factor inputs, higher labor productivity results in greater value generated per employed person while consuming fewer resources, ultimately leading to lower carbon emission levels.

The coefficient for economic development level is positive at the 5% significance level. It suggests that economic growth increases carbon emissions. It reflects that there is still a gap between China’s current economic development model and sustainable development.

The coefficient for the tax burden level is positive at the 1% significance level. It indicates that higher tax burdens during the study period led to increased carbon emissions across provinces. This may be because higher taxes impose economic pressure on companies, where the suppressive effect outweighs incentives for green transformation. Moreover, although higher taxes increase government revenue, not all of it is necessarily allocated to environmental protection.

The coefficient for government intervention degree is positive at the 5% significance level, consistent with expectations. It suggests that government spending is mainly directed toward promoting economic growth. In other words, the spending often targets capital- and labor-intensive industries that are more polluting but can drive economic growth rapidly in a short time.

The coefficient for human capital level is negative at the 1% significance level, suggesting that improvements in human capital contribute to the reduction in carbon emissions. Individuals with a certain level of education act as carriers of technical knowledge and possess enhanced capabilities in knowledge acquisition, transformation and utilization. A higher human capital level within a region implies stronger abilities to absorb, apply and transform new technologies, thereby facilitating the conversion of technological innovations into new products and industries, which subsequently reduces carbon emissions.

The estimated coefficient for industrial structure rationalization is negative at the 1% significance level, suggesting that enhancing this rationalization contributes to the reduction of carbon emissions. This finding aligns with prior predictions. Improvements in industrial structure rationalization can lower production costs by enhancing resource allocation and increasing labor productivity, thereby generating scale effects. Additionally, these scale effects can further promote industrial structure rationalization through optimized resource allocation and the development of industrial clusters. As a result, the interplay between industrial structure rationalization and scale effects will reduce the energy consumption per unit of product, ultimately leading to a significant decrease in carbon emissions.

5.2. Endogeneity Test

To address the potential endogeneity issue arising from the bidirectional causal relationship between FDI and carbon emissions, this paper employs two approaches in the regression analysis. The results of these regressions are presented in

Table 3. Endogeneity test.

Variable	(1)	(2)	(3)
	First-Order Lag	Second-Order Lag	2SLS
L.FDI	−21.68 ***
	(−3.57)
L2.FDI		−14.50 **
		(−2.38)
FDI			−56.20 ***
			(−3.33)
_cons	1852.43 ***	1193.54 **
	(3.93)	(2.48)
control	Yes	Yes	Yes
N	324	297
R2	0.260	0.199
F	14.52	9.31
Ward F			50.845
LM statistic			0.0000

Note: *** and ** represent the significant level of 1% and 5%, respectively.

.

5.2.1. Lagged Variables

Considering that FDI may have lagged effects on carbon emissions, we perform regression analysis using lagged values of the core variable. Table 3 presents the regression results for one-period and two-period lags in columns (1) and (2), respectively.

5.2.2. Instrumental Variable Method

To solve the endogeneity problem, we employ the one-period lag of actual foreign capital utilization (L.FDI) as an instrumental variable in the model. Additionally, we use the two-stage least squares method for regression. The results are shown in Table 3, column (3). The rationale is that the lagged FDI is independent of the current random disturbance term, but it could directly influence the current level of actual foreign capital utilization.

The regression results in Table 3 show that after addressing the endogeneity problem, the conclusion that FDI can reduce carbon emissions in Chinese provinces remains valid.

5.3. Robustness Test

To ensure the validity of the conclusions, we test the robustness of the regression results in three aspects based on the baseline regression.

5.3.1. Replacing the Explained Variable

Considering the close relationship between carbon emission levels, carbon production efficiency and per capita carbon emission intensity, we substitute the dependent variable with carbon production efficiency (LC-A) and per capita carbon emission intensity (LC-B) for regression analysis. The regression results are presented in

Table 4. Robustness test.

Variable	(1)	(2)	(3)	(4)
	LC-A	LC-B	LC	LC
FDI	−30.15 ***	−6.145 ***	−16.84 ***	−12.91 **
	(−7.27)	(−3.41)	(−2.91)	(−1.99)
_cons	2279.4 ***	734.2 ***	2064.4 ***	2274.1 ***
	(7.56)	(5.60)	(5.26)	(4.94)
control	Yes	Yes	Yes	Yes
fe-year	Yes	Yes	Yes	Yes
fe-province	Yes	Yes	Yes	Yes
N	351	351	351	324
R2	0.513	0.318	0.325	0.314
F	47.64	21.09	21.81	17.20

Note: *** and ** represent the significant level of 1% and 5%, respectively.

, columns (1) and (2).

5.3.2. Refining the Study Period

To avoid the impact of the COVID-19 pandemic, we adjust the sample period by limiting it to 2009–2019 and then conducting regression analysis. The regression results are presented in Table 4, column (3).

5.3.3. Removing the Influence of Extreme Values

To mitigate the potential impact of extreme values on the regression results, we apply bilateral winsorization at the 1% and 99% levels to all variables, which means replacing values outside the specified range with the corresponding percentile values before re-running the regression analysis. The results are presented in Table 4, column (4).

The results presented in Table 4 show that, after applying the three adjustment methods, the significance and direction of the coefficient for the core explanatory variable remain unchanged. This suggests that the baseline regression is robust, confirming that FDI helps reduce carbon emissions in Chinese provinces, further supporting part of Hypothesis 1.

5.4. Mechanism Test

To prove the mechanism by which FDI affects carbon emissions, particularly through its impact on technological innovation, we conduct a mediation effect analysis. The analysis is based on previous analyses, using Equations (2) and (3). The mediation effect is decomposed and tested using 500 Bootstrap simulations.

Table 5. Mechanism test of technology innovation.

Variable	(1)	(2)	(3)	(4)	(5)
	LC	Tec-Q	LC	Tec-N	LC
FDI	−53.73 ***	0.0239 ***	−47.83 ***	1449.5 ***	−54.37 ***
	(−4.35)	(4.26)	(−3.84)	(11.20)	(−4.39)
Tec-Q			−279.1 ***
			(−2.69)
Tec-N					−0.00763 **
					(−2.11)
_cons	−715.8 ***	0.190 **	−641.8 ***	−12,608.5 ***	−2004.6 ***
	(−2.99)	(2.40)	(−0.01)	(−8.85)	(−3.63)
control	Yes	Yes	Yes	Yes	Yes
N	351	351	351	351	351
R2	0.407	0.066	0.419	0.264	0.379
F	33.60	6.070	30.84	125.5	32.26

Note: ***and ** represent the significant level of 1% and 5%, respectively.

presents the regression results of the mediation effect analysis. In columns (2) and (4), the estimated coefficients for FDI are positive at the 1% significance level. It indicates that FDI significantly enhances both the quality and quantity of green technological innovation. In columns (3) and (5), the coefficients for green technological innovation quality and quantity are negative at the 1% significance level. It suggests that improvements in these areas help reduce carbon emissions across provinces. Compared to the baseline regression results in Table 2, the estimated coefficients for FDI in columns (3) and (5) remain negative at the 1% significance level. This finding preliminarily suggests that technological innovation, encompassing both the quality and quantity of green innovation, may partially mediate the impact of FDI on carbon emissions.

We further examine the mediating effect between FDI and carbon emissions, with the decomposition and test results presented in

Table 6. Mediation effect decomposition and bootstrap test.

Variable	Effect Decomposition	Observed Coefficient	Bootstrap Std. Err.	z	p > z	Normal Based	Observed Coefficient
Tec-Q	Direct effect	−44.087	15.122	−2.92	0.004	−73.726	−14.449
	FDI→Lc
	Indirect effect	−7.204	2.961	−2.43	0.015	−13.007	−1.402
	FDI→Tec-Q→Lc
	Total effect	−51.291	16.136	−3.11	0.002	−81.566	−21.017
Tec-N	Direct effect	−54.369	15.706	−3.46	0.001	−85.152	−23.586
	FDI→Lc
	Indirect effect	−6.747	3.434	−1.96	0.049	−13.477	−0.017
	FDI→Tec-N→Lc
	Total effect	−61.116	17.546	−3.48	0.000	−95.505	−26.726

. When using green technological innovation quality as the mediating variable, the estimated coefficient for the indirect effect is −7.204. This coefficient is significant at the 1% level, with a 95% confidence interval that excludes 0. The finding indicates that the mediating effect of green technological innovation quality between FDI and carbon emissions is significant. Moreover, it suggests that FDI could reduce carbon emissions by enhancing the quality of green technological innovation.

When using green technological innovation quantity as the mediating variable, the estimated coefficient for the indirect effect is −6.747. This coefficient is significant at the 5% level, with a 95% confidence interval that excludes 0. This finding indicates that the mediating effect of green technological innovation quantity between FDI and carbon emissions is significant. Moreover, it suggests that FDI could reduce carbon emissions by increasing the quantity of green technological innovation.

In summary, technological innovation mediates the relationship between FDI and carbon emissions. FDI significantly boosts both the quantity and quality of regional technological innovation, and their combined effects lead to a substantial reduction in carbon emissions. This supports Hypothesis 2 of this paper. Furthermore, compared to the pathway through green technological innovation quantity, the effect of FDI on reducing carbon emissions is more pronounced through the increase in green technological innovation quality.

5.5. Threshold Effect

To further investigate how FDI affects carbon emissions and clarify the nonlinear relationship between FDI and carbon emissions, we use Hansen’s threshold model as a reference. Thus, a regression model is constructed using FDI quality and FDI scale as threshold variables, according to Equation (4) for threshold regression. Additionally, the FDI scale is represented by the actual foreign capital utilized in each province. Before regression, the validity of the threshold must be tested, with results shown in

Table 7. Threshold effect test.

Variable	Threshold	Fstat	Prob	Crit10	Crit5	Crit1
FDI-S	Single	14.46	0.028	9.571	12.010	16.907
	Double	7.96	0.234	11.649	15.595	19.760
QFDI	Single	16.00	0.098	15.900	19.292	26.299
	Double	10.85	0.152	11.901	13.941	17.726

. The detailed analysis is as follows:

Table 7 presents the F-values and p-values from the threshold test. The results indicate that the single threshold values for FDI scale and FDI quality are significant at the 5% and 10% levels, respectively. However, the double thresholds are not significant. This suggests that a nonlinear relationship exists between FDI and carbon emissions when using the FDI scale and FDI quality as threshold variables. And the model contains only a single threshold.

Table 8. Threshold effect regression.

Variable	(1)		(2)
	FDI-S		QFDI
	Single		Single
	r ≤ 119,192	r > 11,912	r ≤ 0.2308	r > 0.2308
FDI × I	−16.724 ***	−24.015 ***	−13.565 ***	−20.559 ***
	(−2.78)	(−4.03)	(−2.68)	(−3.85)
_cons	2357.705 ***	2357.705 ***	2042.268 ***	2042.268 ***
	(5.44)	(5.44)	(4.72)	(4.72)
control	Yes	Yes	Yes	Yes
N	351	351	351	351
R2	0.337	0.337	0.330	0.330
F	20.08	20.08	19.49	19.49

Note: *** represents the significant level of 1%.

presents the regression results of the threshold effect analysis. In column (1), with the FDI scale as the threshold variable, the impact of FDI on carbon emissions is divided into two stages based on the threshold value. Both stages have negative estimated coefficients that increase in magnitude and are statistically significant. It indicates that the inhibitory effect of FDI on carbon emissions strengthens as its scale grows. The reason could be attributed to the intensifying competition as the number and size of foreign enterprises in a region increase. Specifically, the competition prompts local companies to innovate independently, introduce technology, or learn from foreign technologies, thereby advancing technological progress. As local firms enhance their technological capabilities, outdated production capacities are phased out. Their ability to absorb and implement green technologies from foreign enterprises improves. It is this process that allows the carbon reduction effects of FDI to be more effectively realized. The finding supports the remaining part of Hypothesis 1, which states that the impact of FDI is dynamic. In other words, as the scale of FDI increases, its inhibitory effect on carbon emissions is enhanced.

In column (2), with FDI quality as the threshold, the impact of foreign direct investment on carbon emissions is divided into two stages based on the threshold value. Both stages have negative estimated coefficients that increase in magnitude and are statistically significant. It indicates that the inhibitory effect of FDI on carbon emissions strengthens as its quality improves. High-quality foreign enterprises have high profitability, technological levels, export capacity and management skills. Specifically, first, the greater the operational and management capabilities of foreign-invested enterprises, the higher their asset contribution rate. This reflects their ability to retain larger profits, which not only facilitates capital reinvestment but also enables investment in research and development, technological innovation and the adoption of more advanced production methods. As a result, they can improve production efficiency, reduce energy consumption per unit and decrease environmental pollution. Secondly, improving technological levels enhances the production efficiency of enterprises, enabling them to achieve greater output with fewer resources, which, in turn, contributes to resource conservation and reduced pollution emissions. Additionally, the entry of high-tech foreign-invested enterprises introduces advanced pollution control technologies and environmental management practices, positioning them as role models in China’s environmental protection efforts. Finally, foreign-invested enterprises with strong export capabilities can drive export growth. Typically, foreign investment is facilitated by multinational corporations, which influence the host country in two key ways: first, they provide resources and access to new markets, enabling local enterprises to explore new export opportunities; second, they offer competitive assets unique to multinational corporations, significantly enhancing the green production capacity of local businesses. Therefore, as the quality of FDI improves, its carbon reduction effect becomes more pronounced. The finding validates Hypothesis 3, which states that FDI quality influences FDI’s impact on carbon emissions. Moreover, the finding also supports the remaining part of Hypothesis 1, which states that the impact of FDI is dynamic. As the quality of FDI increases, its inhibitory effect on carbon emissions is enhanced.

5.6. Heterogeneity Test

Recognizing the substantial differences in development levels across provinces, we use the technology fiscal support level (Tec-Support) and the environmental regulation intensity as the basis for grouping to examine the varying effects of FDI on carbon emissions. Specifically, we measure fiscal support for technology research and development in each province using the science and technology expenditures from general fiscal spending. Meanwhile, the environmental regulation intensity is represented by the ratio of environmental pollution control expenditures to regional GDP.

By using the median of the heterogeneity variables to group, the provinces are categorized into low tec-support and high tec-support groups, as well as low environmental regulation intensity and high environmental regulation intensity groups. The regression results are presented in

Table 9. Heterogeneity test.

Variable	(1)	(2)	(3)	(4)
	Low Tec-Support	High Tec-Support	Low Environmental Regulation Intensity	High Environmental Regulation Intensity
FDI	−9.328 *	−27.18 ***	−2.486	−21.24 ***
	(−1.67)	(−2.93)	(−0.57)	(−2.70)
_cons	2589.4 ***	−677.5	231.9	3683.0 ***
	(4.38)	(−1.33)	(0.98)	(4.40)
control	Yes	Yes	Yes	Yes
N	176	175	180	171
R2	0.405	0.291	0.309	0.469
F	14.29	8.815	9.451	17.76

Note: *** and * represent the significant level of 1% and 10%, respectively.

.

The regression results suggest that FDI can inhibit carbon emissions in all groups. This phenomenon may be attributed to several factors: first, the government’s increased financial support for technological research and development helps alleviate financing pressures on local enterprises and other innovators. By fostering innovation and providing funding, this support reduces innovation risks and stimulates technological advancements. As a result, local enterprises are better positioned to absorb and implement advanced technologies from FDI, effectively enhancing the carbon reduction effect of FDI. Secondly, the level of environmental regulation has both direct and indirect effects on the pollution emissions behavior of local enterprises, including foreign-invested firms. Stricter environmental standards increase the costs associated with wastewater and gas emissions, potentially driving high-pollution, high-energy-consuming companies out of the market. Additionally, stringent regulations compel enterprises to innovate technologically, enhancing productivity and adopting greener production methods. Furthermore, such regulations can limit the entry of high-pollution foreign-invested enterprises into the market.

6. PD-QCA

From the previous analysis, it is evident that as FDI quality improves, its inhibitory effect on carbon emissions increases significantly. This effect is also more pronounced in provinces with high technological support and stringent environmental regulations. To investigate the specific mechanisms by which FDI quality, technological support levels and environmental regulation intensity drive the carbon reduction effect of FDI, we use PD-QCA for configuration analysis.

The relationships between the antecedent variables and the outcome variable are shown in Figure 3. Firstly, each antecedent variable could independently influence carbon emissions, or they could jointly impact carbon emissions when acting in combination. For instance, as noted earlier in the heterogeneity analysis, the inhibitory effect of FDI on carbon emissions is more pronounced in provinces with higher technological fiscal support. Moreover, aside from their impact on carbon emissions, these antecedent variables could also interact with one another. For example, with stricter environmental regulations, the quality of foreign direct investment attracted may improve. Since higher environmental compliance costs tend to deter the entry of low-quality foreign investments. In previous research, we found that FDI significantly reduces carbon emissions in China’s provinces. However, it remains unclear whether this effect is exclusively caused by FDI or results from the combined influence of various antecedent variables. This issue requires further validation through configuration analysis.

6.1. Variable Selection and Calibration

In QCA, each condition and each outcome are viewed as a set, and each case has membership scores within these sets. Therefore, before applying the PD-QCA method, all data need to be calibrated to assign membership degrees to each variable. This section employs the direct calibration method, as used by Llopis–Albert [85]. It selects the 95th, 50th and 5th percentiles as the full membership point, crossover point and full non-membership point, respectively. These points are used to calibrate variables into set data within the [0, 1] range.

Table 10. Calibration points and description of variables.

Variable	Description	Full Membership	Intersection	Non-Membership
LC	Carbon Emission Level	1020.8233	287.2132	60.2172
FDI	FDI Inflow	841,337.5000	63,666.0000	1574.5000
QFDI-E	FDI Export Capability	0.2696	0.0913	0.0011
QFDI-M	FDI Management Capability	0.0058	0.0055	0.0054
QFDI-P	FDI Profitability	0.0097	0.0086	0.0082
QFDI-T	FDI Technological Characteristics	0.1325	0.0614	0.0324
Environmental Regulation Intensity		0.0098	0.0025	0.0007
Tec-Support		378.8036	49.7600	6.0189

provides relevant descriptions and specific calibration anchors for each antecedent condition.

6.2. Necessary Condition Analysis

Before conducting configuration analysis, a necessity analysis of individual antecedent conditions is required to determine whether they are necessary for the occurrence of the outcome. A condition is considered necessary if its consistency level exceeds 0.9; otherwise, it is not [86]. We mainly focus on the pathways to achieving carbon reduction, so the analysis will be limited to the low outcome group (low carbon emissions) in the subsequent sections.

The data in

Table 11. Analysis of necessary conditions.

Condition Variable	Consistency	Coverage
FDI	0.68	0.796
~FDI	0.694	0.578
QFDI-E	0.734	0.745
~QFD-E	0.603	0.564
QFDI-M	0.661	0.698
~QFDI-M	0.729	0.658
QFDI-P	0.648	0.679
~QFDI-P	0.737	0.669
FDI-T	0.639	0.679
~QFDI-T	0.753	0.676
Environmental Regulation Intensity	0.625	0.664
~Environmental Regulation Intensity	0.714	0.641
Tec-Support	0.739	0.799
~Tec-Support	0.64	0.567

show that the consistency indicators for each single condition variable are below the 0.9 threshold needed to establish a necessary condition. This suggests that each condition could contribute to reducing carbon emissions in provinces to some degree. However, none of them is sufficient on its own to create an effective pathway for carbon reduction. Instead, achieving the carbon reduction effect is the result of the combined influence of various conditions.

6.3. Configuration Analysis

Using calibrated sample data, we conduct a configuration analysis, resulting in complex, intermediate and parsimonious solutions. The intermediate solution serves as the primary reference, with the nested relationship between the intermediate and parsimonious solutions as a secondary reference. If an antecedent condition appears in both the parsimonious and intermediate solutions, it is deemed a core condition, signifying high importance for the outcome. If it only appears in the intermediate solution, it is considered an auxiliary condition, indicating relatively lower importance. The final configuration analysis results are presented in

Table 12. Paths for low carbon emission level.

Condition Variable	Path1	Path2	Path3	Path4	Path5	Path6
FDI	●	⬤				⬤
QFDI-E	⬤		⬤	⬤	⬤
QFDI-M		⬤	⬤	●		⬤
QFDI-P		●	●	●		●
QFDI-T			●			●
Environmental Regulation Intensity				●	⬤	●
Tec-Support	⬤	●	●	⬤
Consistency	0.923	0.959	0.952	0.969	0.93	0.975
Coverage	0.478	0.433	0.359	0.311	0.307	0.24
Overall Consistency	0.901
Overall Coverage	0.654

Note: The symbol “” represents the missing of the condition, and “●” represents the existence of the condition. A large circle indicates a core condition, a small circle indicates a peripheral condition, and an empty space indicates that the condition has an optional or negligible impact on the outcome.

.

Table 12 shows that there are six pathways for FDI-driven carbon reduction, with consistencies of 0.923, 0.959, 0.952, 0.969, 0.93 and 0.975 and an overall solution consistency of 0.901. Both individual and overall solution consistencies exceed 0.9, far surpassing the acceptable minimum of 0.75. Additionally, the overall coverage is high. The result indicates that the analysis in this section is valid.

The configuration analysis results indicate that almost all variables have acted as core conditions for reducing carbon emissions across the pathways.

Specifically, the results of configuration analysis reveal four combinations of conditions that enabled the realization of the carbon reduction effect from FDI within the statistical interval. These combinations consist of six pathways through which various provinces utilized FDI to achieve carbon reduction during the statistical period. The first type is driven by the export capabilities of FDI and technological fiscal support, represented by Path1 and Path4. In these pathways, the export capabilities of FDI and local technological fiscal support serve as core conditions that play a dominant role, while the scale of FDI and other quality characteristics serve as marginal conditions that provide additional support. This indicates that these provinces have attracted foreign enterprises with strong export capabilities and, in the process, have intensified their fiscal support for technology research and development. Consequently, they have continuously improved their capacity to absorb advanced technologies from foreign enterprises and enhance their own independent research, development and innovation capabilities. Ultimately, this has led to a positive interaction between FDI and carbon emissions, effectively mitigating local carbon emissions. The second type is driven by the scale of FDI and its management capabilities, represented by Path2. In Path2, the scale of FDI and the management capabilities of foreign enterprises serve as core conditions that play a leading role, while the profitability of foreign enterprises and local technological fiscal support serve as marginal conditions that provide additional support. This indicates that these provinces have achieved a positive interaction between FDI and carbon emissions by attracting foreign enterprises with strong management capabilities, effectively mitigating local carbon emissions while attracting foreign investment. The reason may be that foreign enterprises with higher management capabilities have more complete incentive and constraint mechanisms, which better assure various measures to maintain their vitality and competitiveness, thereby enhancing competition among local enterprises. By attracting a number of foreign enterprises with strong management capabilities, these provinces have effectively fostered healthy competition and gradually eliminated outdated production capacity, resulting in improvements to the local ecological environment and a reduction in carbon emissions. The third type is driven by the quality of foreign direct investment, represented by Path3. In this pathway, the export capabilities and management capabilities of FDI serve as core conditions that play a leading role, while the profitability and technological characteristics serve as marginal conditions that provide additional support. This indicates that these provinces have achieved a positive interaction between FDI and carbon emissions by attracting high-quality FDI that excels in these four dimensions, effectively mitigating local carbon emissions. The fourth type is driven by environmental regulation, represented by Path5 and Path6. In these pathways, the quality characteristics of FDI show varying degrees of deficiency; however, under the influence of environmental regulations, these provinces have still achieved a positive interaction between FDI and carbon emissions. This may be because, although these provinces have not managed to attract high-quality FDI, environmental regulations effectively limit the entry of low-quality, high-pollution and high-emission foreign investments.

Notably, FDI export capacity has the most significant impact. It serves as a core condition in four of the six pathways. Moreover, the results also indicate that strong environmental regulation could effectively enhance the inhibitory effect of FDI with strong export capacity on carbon emissions. At the same time, high scientific and technological support could also contribute to this enhancement. This enhancement occurs even when FDI scale or technological characteristics are limited. Although the scale and technology level of FDI may not be sufficient to drive green transformation through competition and spillover effects, its robust export capacity can still motivate local enterprises. This motivation comes from entering larger global markets, encouraging them to improve their technological levels. Specifically, strong environmental regulations further support this process by promoting green innovation and eliminating outdated capacities, while high technological support provides financial backing for innovation. Together, these factors enable the carbon reduction effects of FDI to be realized.

6.4. Robustness Test

We follow Zhong’s approach by increasing the original consistency threshold to test robustness. Specifically, the original consistency threshold is raised from 0.8 to 0.85 while maintaining other criteria [87].

Table 13. Robustness test for QCA.

Condition Variable	Path1	Path2	Path3	Path4	Path5	Path6
FDI	●	⬤				●
QFDI-E	⬤		⬤	⬤	●
QFDI-M		●	●	●		●
QFDI-P		●	●	●		●
QFDI-T						●
Environmental Regulation Intensity				●	●	●
Tec-Support	●	●	●	⬤
Consistency	0.976	0.970	0.987	0.989	0.910	0.982
Coverage	0.439	0.442	0.349	0.367	0.343	0.246
Overall Consistency	0.947
Overall Coverage	0.626

Note: The symbol “” represents the missing of the condition, and “●” represents the existence of the condition. A large circle indicates a core condition, a small circle indicates a peripheral condition, and an empty space indicates that the condition has an optional or negligible impact on the outcome.

shows the result of robustness test. The results show that the configurations under the two consistency thresholds exhibit a clear subset relationship, with only minor changes observed in various indicators. Thus, the conclusions of this study are robust.

7. Conclusions, Implication and Research Limitation

7.1. Conclusions

FDI is a critical factor influencing carbon emissions and a significant driver of China’s economic development. Investigating the extent and mechanisms through which FDI impacts carbon emissions could offer new insights and strategies for China’s pursuit of sustainable economic development. To examine how FDI impacts carbon emissions, we conducted an empirical analysis. The analysis is based on a review of the existing literature. It uses panel data from 27 provincial-level administrative units in mainland China from 2009 to 2021. Concerning the impact of FDI on carbon emissions, the empirical analysis results show that FDI significantly reduces overall carbon emissions in China, with its effect intensifying as the scale grows. Regionally, due to differences in technological fiscal support and environmental regulation levels, FDI’s impact on carbon emissions shows regional variation. In provinces with high technological fiscal support and strong environmental regulations, the inhibitory effect of FDI on carbon emissions is particularly pronounced and effective. The findings are largely in agreement with those of Zhang and Ahmad. We all confirm that FDI could effectively alleviate carbon emissions in China and exhibit regional heterogeneity. However, unlike Ahmad’s research, during the study period we selected, FDI mitigated carbon emissions in all provinces [35,88]. Moreover, concerning the mechanisms through which FDI impacts carbon emissions, the mediation effect analysis shows that technological innovation mediates the relationship between FDI and carbon emissions. FDI impacts carbon emissions by influencing both the quality and quantity of technological innovation. Notably, the increase in green technological innovation quantity has a more pronounced effect on reducing carbon emissions compared to improvements in innovation quality. The findings are largely consistent with Ail’s research. We both confirm that FDI can alleviate carbon emissions through the promotion of technological innovation. However, unlike Ail’s research, we further refine the criteria for technological innovation. It reveals that FDI alleviates carbon emissions by enhancing both the quantity and quality of technological innovations [89]. Additionally, concerning the nonlinear relationship between FDI and carbon emissions, the threshold effect analysis shows that the impact of FDI on carbon emissions has a clear threshold effect. With FDI scale and quality as thresholds, a nonlinear relationship emerges between FDI and carbon emissions. Specifically, as both the scale and quality of FDI increase, its carbon reduction effect becomes progressively stronger. The findings are largely consistent with Fang’s research. While our paper used the quality of FDI as a threshold variable, we still arrived at similar conclusions [70]. Finally, in the further discussion about the mechanisms, the configuration analysis results show that the carbon reduction effect of FDI is driven by the combined influence of several factors. These factors include FDI scale, quality, technological fiscal support and environmental regulation. FDI export capacity stands out as the most significant factor. It serves as the core condition for achieving carbon reduction through FDI. Strong environmental regulations or high technological fiscal support could enhance the effectiveness of FDI with robust export capacity. This approach can effectively mitigate the limitations of FDI’s inhibitory effect on carbon emissions due to scale or technological constraints.

7.2. Implication

Based on the findings above, we advance the following policy implications for China’s pursuit of sustainable economic development.

First, the government should prioritize the quality of FDI by exploring new approaches to attracting and utilizing capital. A favorable business environment is essential for attracting foreign investment; only through continuous optimization of this environment can high-quality FDI be consistently drawn [90,91]. While China has made significant strides in improving its business environment, it still lags behind global standards in certain areas, with some sectors exhibiting discriminatory treatment toward foreign investors. To address this, the government should strengthen and refine policies and regulations, continuously enhancing investment liberalization and ease of access. For instance, the government can enhance the business environment by further reducing the negative list for foreign investment access.

Second, while refining the foreign investment entry system, the government should raise the entry thresholds for foreign investment. It should categorize FDI based on its source and direction and, in formulating targeted and effective investment policies, focus on attracting foreign capital that supports green and low-carbon development while limiting the inflow of high-pollution, high-energy consumption and high-emission enterprises. From a cost-benefit perspective, although increasing the entry criteria for foreign investment and environmental standards may lead to a temporary decline in foreign investment inflows, in the long term, high-quality FDI will introduce more advanced technologies and management practices, reducing environmental governance costs and, undoubtedly, facilitating the achievement of sustainable economic development.

Third, the government should promote foreign-invested enterprises’ participation in exports by fostering a fair market environment and focusing on attracting foreign-invested enterprises with strong export potential. To maximize FDI’s carbon reduction impact, the government should ensure the full implementation of national treatment for foreign investment and remove unreasonable approval procedures and qualification requirements. Specifically, the government should review foreign-invested enterprises’ business licenses and qualification applications based on unified standards, ensuring they have equal access to production factors and industrial policies. At the same time, it should swiftly eliminate policies obstructing fair competition, fully enforce the fair competition review system and enhance anti-monopoly regulation and enforcement. This will encourage both domestic and foreign-invested enterprises to engage fairly in trade and other commercial activities, thereby enhancing foreign-invested enterprises’ export willingness through institutional support.

Finally, local governments should implement a policy mix tailored to regional development needs to maximize the carbon-reduction impact of FDI. On the one hand, local governments should increase fiscal support for R&D and support the steady growth of high-tech industries. This will strengthen local enterprises’ capacity to absorb advanced FDI technologies. Moreover, it will also enhance human capital accumulation in general, which is broadly shown to be a key factor in modern economic growth [92,93]. On the other hand, a well-established legal and regulatory framework is fundamental for effective pollution control. Local governments should improve regulations, policies and standards at all levels, establish a robust environmental pollution control framework and use relevant environmental laws to regulate corporate behavior, thereby prompting enterprises to adopt technological transformations and upgrades. However, it is important to note that environmental regulations may also lead fossil fuel producers to anticipate stricter future regulations. As a result, they may accelerate the extraction and production of fossil fuels, leading to a reduction in fossil fuel prices, increased consumption and a decline in environmental quality [94,95]. Therefore, local governments should conduct comprehensive market analysis before formulating policies and tailoring environmental regulations to the specific characteristics of different industries.

7.3. Limitations and Directions for Future Research

Due to constraints in time and data availability, there are still some limitations in this paper. First, the identification of factors affecting carbon emissions is not exhaustive. The empirical and configuration analysis models did not encompass all relevant factors, such as regional industrial structure and economic development levels, which could influence the carbon reduction effect of FDI. Hence, in the future, we will give full consideration to the impact of industrial structure and economic development levels on the effect of FDI in reducing carbon emissions. Specifically, by treating industrial structure upgrading as a mediating variable, we will examine the mechanism by which FDI influences carbon emissions. Furthermore, we will explore the nonlinear relationship between FDI and carbon emissions by using the stage of economic development as threshold variables. These variables will also be incorporated into the configuration analysis. Second, the sample size is limited. Due to constraints in time and data availability, we only utilized provincial-level data and did not refine the analysis to the prefecture level. This limitation may have caused us to overlook how specific differences in urban development affect the carbon reduction effects of FDI. Thus, in the future, we aim to enhance data sources to broaden the scope of analysis. Specifically, first, by employing panel data at the prefecture level, we aim to gain a more in-depth understanding of the spatial dimension. Second, by extending the temporal scope of the study, we aim to achieve a more comprehensive understanding of the effects of FDI on China’s carbon emissions over time. Third, the scope of the research is limited. Due to constraints in data availability and resources, we focus solely on China to validate the environmental effects of FDI. However, key hypotheses, such as the pollution haven and pollution halo hypotheses, are generally applicable to developing countries as a whole. Thus, restricting our study to China presents certain limitations in verifying these hypotheses. In the future, we will broaden our investigation to explore the impact of FDI on carbon emissions in other developing countries. For instance, we will delve into the environmental challenges faced by countries like Brazil and India and assess the influence of FDI on their carbon emissions. Furthermore, we will conduct a comparative analysis between these findings and the situation in China.