Climate Policy and Foreign Direct Investment: Evidence from a Quasi-Experiment in Chinese Cities

Abstract

International investment is sensitive to environmental policies, and developing countries are worried about the withdrawal of foreign capital when adopting climate policies. This study treats the pilot policy of low-carbon cities as a quasi-experiment and uses urban panel data from 2006 to 2019 to investigate how climate policy affects foreign direct investment (FDI). Results show that climate policy has significantly reduced the FDI in Chinese cities but has promoted the quality of FDI. The regulatory pressure and innovation incentives brought about by climate policy change the quantity and quality of FDI in cities. Climate policy mainly reduces FDI in large cities and improves the quality of FDI in small cities. Moreover, the quality of FDI in capital outflow cities decreases, whereas that in capital inflow cities increases, thereby suggesting a potential transfer of FDI from large to small cities. In addition, the climate policy in innovative cities improves the quantity and quality of FDI but has a negative impact on FDI in non-innovative cities.

1. Introduction

As a major challenge to human sustainability in the 21st century, countries all over the world should work together to mitigate climate change. International investment activities are closely related to climate policy and to its emission reduction effect. Accordingly, a large number of politicians and scholars have paid attention to how climate policy affects international investment behavior, especially in developing and emerging countries [1]. The developed countries have therefore put pressure on developing and emerging countries to reduce their fossil energy consumption and greenhouse gas emissions [2]. Energy and resources are important determinants of FDI location selection. Developing countries worry that strict climate policies would lead to a decline in the attractiveness of international capital, thereby leading to a decline in foreign direct investment (FDI) and economic growth. Accordingly, these countries tend to adopt loose climate policies, hence explaining why the energy and environmental costs in developing countries are always low, which can attract capital inflows from developed countries [3]. Therefore, international cooperation on climate change should be optimized, and the enthusiasm of developing and emerging countries in implementing climate policies should be promoted. Specifically, these countries should engage in responsible international investment and enhance the positive response of their international investments to climate policies.

China is the largest developing and carbon emitting country in the world, and it has been actively undertaking international responsibilities to address climate change. For instance, in 2013, China released its National Strategy for Adaptation to Climate Change. The country also set its goal to achieve carbon peak by 2030 and carbon neutralization by 2060. In fact, China has made great achievements in reducing its carbon emissions, and the carbon emission intensity in 2020 was 48.4% lower than that in 2005. The pilot policy of low-carbon cities is an important practice in reducing carbon dioxide emissions in China, aiming to enhance the adaptability of the country to global climate change and international climate policies [4,5]. Despite its great achievements in carbon dioxide emission reduction, China’s strict climate policies have also caused public concern regarding their economic consequences. For example, increasingly strict environmental and climate policies have increased the environmental costs for enterprises, which may squeeze out FDI. In fact, as their climate policies become stricter, the FDI absorbed by China is also decreasing, hence suggesting that international investment activities are in conflict with strict climate policies, which is also a common concern among developing countries.

This study takes the pilot policy of low-carbon cities implemented in China in 2010 as the proxy variable of strict climate policy and uses the panel data of Chinese cities from 2006 to 2019 to investigate the impact of climate policy on the quantity and quality of FDI. Results show that a strict climate policy decreases the FDI in pilot cities, which supports the classic view of environmental cost advantage. These results imply that China has suffered adverse economic consequences while fulfilling its climate governance responsibilities. However, the climate policy has significantly improved the quality of its FDI, and China has mitigated its economic losses by attracting higher-quality FDI. In addition, climate policy has a heterogeneous impact on FDI according to differences in urban characteristics, and innovation is conducive to mitigating the adverse impact of climate policy on FDI.

Our research mainly contributes to the existing literature from the following aspects. First, this study identifies the causality between climate policy and FDI. On the one hand, we focus on climate policy rather than traditional environmental policy. On the basis of the literature on the relationship between environmental regulation and FDI [3], this study enriches the relevant theories on the internalization of externalities of global pollutants or climate damage. On the other hand, this study uses the difference-in-difference (DID) method to overcome the potential endogeneities or biases in measuring climate policy. Second, this study examines the changes in the quantity and quality of FDI. Previous studies have found that environmental policies lead to the decline of the FDI scale without paying attention to the improvement of its quality [6,7]. In this way, this study provides novel insights and inspiration for sustainable international investment research from the FDI quality perspective. Third, this study examines the heterogeneity of population size and innovation and finds that FDI may be transferred from large cities to small ones and that innovation can mitigate the adverse effects of climate policy.

2. Background, Literature, and Theoretical Discussion

2.1. Background of Climate Governance in China

China has formulated and implemented a series of strategies, regulations, policies, standards, and actions to address climate change, thereby offering significant contributions to global climate governance. China has also built an executive organizational framework for climate governance to ensure the implementation of its climate policies. In 2007, China established a national leading group on climate change and energy conservation and emission reduction, with the Premier of the State Council serving as the team leader. Moreover, all provinces in China have established leading groups on climate change, energy conservation, and emission reduction. In 2018, China established the Ministry of Ecology and Environment, which aims to address climate change and strengthen the coordination between climate change and ecological environment protection. In 2021, to guide and coordinate its carbon peak carbon neutralization efforts, China established a leading group for carbon peak carbon neutralization.

China actively explores the mode of low-carbon development and encourages local governments, industries, and enterprises to explore low-carbon development paths according to their local conditions. One of these paths is to build a number of low-carbon pilot demonstration cities. Since 2010, China has launched three batches of low-carbon cities, whose carbon emission intensity has decreased faster than the national average level, thereby forming a batch of distinctive low-carbon development models. This pilot policy of low-carbon cities was proposed by the central government, whereas the specific policies and measures were designed by local governments. Therefore, this low-carbon development path centers on independent exploration and contribution. Given that environmental goals are among the main indicators of the assessment system for Chinese government officials since 2012, local governments and officials have enough enthusiasm to explore a low-carbon development model. In addition, due to the uniqueness of industrial information, local governments have information advantages when formulating fierce policies to encourage low-carbon development [5], which can effectively help them achieve their dual goals of low-carbon development and economic benefits.

2.2. Literature Review

Since the 1950s, the location determinants of international investment have always been an important topic in international economics. Information technology is the most important driving force behind global economic and trade cooperation, while the motives of FDI include resource seeking, market seeking, and technology seeking [8]. Information and communication technologies have greatly reduced trade costs, thereby deepening international industrial division and cooperation in recent decades [9]. FDI is an important form of international industrial division and cooperation. Some studies have identified several determinants of FDI location in developing countries, Asian countries, and Arab countries, one of which is resource endowment [10,11,12,13]. The loss of location advantages, such as factor cost, economies of scale, and internal advantages, may lead to the transfer of foreign capital from one location to other advantageous regions [14]. When the cost of resource factors or environment rises, multinational enterprises will transfer their production activities across regions [15,16]. Therefore, developing countries tend to adopt strategic environmental policies to attract international investment.

According to the classical view of international economics, given their comparative advantages in energy, resources, and other factors, developing countries specialize in energy- or resource-intensive production activities [17]. Much empirical evidence has shown that environmental regulation or the rise of environmental costs leads to the decline of FDI, which supports the comparative advantage theory of environmental factors [3,18]. Nevertheless, some studies have obtained opposite conclusions and used Porter’s hypothesis or institutional theory to explain the rise of FDI with rising environmental costs [19,20]. Most of these studies have focused on traditional resources or the environmental regulation of local pollution given the global issue of energy and carbon emissions and the presence of many constraints on international economic and trade rules [21,22]. In addition, environmental regulations may lead to the substitution of quality to quantity of FDI.

The classical view that the comparative advantage of environmental cost determines FDI has also been criticized by some scholars in development economics or international political economics [23]. These scholars hold that the FDI from developed to developing countries mainly aims to grab environmental resources, thereby causing serious environmental damage, and that developing countries would be locked in low value-added, high-pollution production links [24]. With depleted natural resources, these regions would become less attractive to transnational enterprises [25]. Some scholars suggest that developing countries should actively adopt environmental or industrial policies to strengthen the selection effect of FDI so as to maintain the attractiveness of these investments at the dynamic level [19,26]. For example, environmental quality strengthens urban competitiveness through the reservoir effect of human capital, thereby explaining the comparative advantage of FDI in terms of human capital [27].

The term “climate policy” refers to any polices and initiatives aimed at adapting to and mitigating climate change [28]. Climate policies are usually designed to reduce carbon dioxide and other greenhouse gases and are intended to strengthen energy security and regulation [4]. Greenhouse gas emission is a behavior with global, universal, long-term, and uncertain externalities [29]. Carbon policy is the most widely discussed type of climate policy, and a large number of documents have evaluated the effects of low-carbon policy on energy efficiency, green innovation, and economic performance. Nevertheless, only a few studies have examined the potential effects of these policies on the economy and climate governance from the international investment perspective.

Climate policy has similar characteristics to environmental policy; however, the former emphasizes the control of global pollutants, such as greenhouse gases, rather than regional pollutants [29]. In other words, climate policy is aimed at reducing the pollution emissions of global externalities, and its actual effect on climate governance is closely related to international economic and trade behavior. If international investments avoid climate policies and transfer to countries with less restrictive climate policies, then the actual effect of climate policy on global climate governance would be greatly reduced. This assumption may also lead to the relaxation of climate policies in developing and emerging countries, which may spell failure for global climate governance cooperation [30]. However, international investment is not necessarily squeezed out by climate policies. Multinational enterprises may pursue green knowledge spillovers or achieve long-term sustainable development goals by investing in areas with strict climate policies [31].

Empirical studies have established an important relationship between environmental regulation and FDI. Most scholars support the view that environmental regulation leads to the withdrawal of foreign capital [3,8]. Nevertheless, others have found that environmental regulations may help attract FDI. While a good environmental quality can attract FDI, environmental regulations promote the quality of such investments [19,20]. However, identifying the relationship between environmental regulation and FDI is always challenged by endogenous problems [32]. Some studies have shown that climate policies may lead to changes in international investment, but such effect may differ between high- and low-risk countries [33]. Moreover, unlike environmental regulations, climate policies have global and uncertain externalities [29], hence indicating the presence of a complex relationship between climate policy and international investment that has not yet been examined in the literature.

Many studies have linked international investment behavior to climate issues yet mainly investigate the impact of international investment on climate change factors, such as energy consumption and carbon emissions [33]. Climate policies are always a global issue, and developing countries always lack the will to implement strict climate policies [34,35]. For example, greenhouse gas emissions would not cause serious consequences for a specific region but can challenge the global climate. However, international cooperation on climate governance has been continuously strengthened, and the ability to cope with climate change brings a competitive advantage in international economic and trade activities. Therefore, some studies have examined the impact of climate policies on regional competitiveness and show that these policies have a significant positive impact on regional productivity and green innovation [4,36,37]. Although one study has examined how low-carbon policies promote FDI through industrial upgrading [38], the changes in FDI quality also warrant examination. Accordingly, this study attempts to assess the potential effects of climate policies from the perspective of international investment cooperation and provide inspiration for global climate governance cooperation.

2.3. Theoretical Discussion

The climate policy in this study is an initiative for low-carbon development, and local governments promote urban low-carbon development by strengthening carbon regulation and increasing scientific and technological fiscal expenditure. The rise of regulation costs and innovation incentive measures both promote the upgrading of industrial structure and innovative development, and affect foreign direct investment in cities. Figure 1 shows the mechanism by which climate policy affects foreign direct investment.

Although the relationship between climate policy and FDI has been disputed in the literature, the presence of better location substitution is widely believed to trigger the negative impact of climate policies on FDI scale. This study investigates low-carbon pilot policies at the urban level in China. These policies represent an attempt by the Chinese government to explore options for low-carbon development. In order to achieve the goal of low-carbon policy, the government implements stricter carbon regulation policies and stronger innovation incentive policies. Multinational enterprises investing in China may invest directly in cities without pilot policies, which would not affect their labor force, market, and other objectives. Given that many alternative locations are available in China, the carbon regulation policies would reduce the FDI scale. In addition, the innovation incentive policies can generate green technology innovation, human capital reservoir, and industrial structure upgrading effects [27,36,37], which would increase technology seeking FDI and human capital seeking FDI, thereby improving FDI quality. The scale of FDI in cities declines due to regulatory costs, while the incentive of scientific and technological fiscal funds to innovation attracts high-quality FDI. The following hypothesis is therefore proposed:

Hypothesis 1 (H1).

The pilot policy of low-carbon cities, or climate policy, can significantly reduce the scale and improve the quality of FDI.

The controversy regarding the relationship between climate policy and FDI is related to Porter’s hypothesis. If such a hypothesis is supported, then climate policies will force enterprises to carry out technological innovation, and a large number of enterprises will actively carry out innovation activities to gain new regional competitive advantages, which would attract FDI. However, the Porter effect of climate policy is related to urban innovation capacity or innovation support. If the urban innovation support is strong, then the Porter effect would be greater, thus attracting technology- or human-capital-seeking FDI. These additional investments may further offset the decline in resource-seeking FDI. Some development economics theories also support the idea that developing countries adopt more active industrial or environmental policies to cultivate competitive advantages in an open economy [39,40]. Innovation is the most important way to mitigate the adverse impact of rising regulatory costs on FDI. The difference of urban innovation characteristics leads to the heterogeneous effect of climate policy. The following hypothesis is then proposed:

Hypothesis 2 (H2).

Climate policy has significantly improved the quantity and quality of FDI for innovative cities but mainly has a negative impact on FDI for non-innovative cities.

3. Design of Empirical Research

3.1. Samples and Data Sources

This study uses the panel data of Chinese cities from 2006 to 2019 as the analysis sample. These data exclude those urban samples with missing key variables. A total of 109 pilot cities and 124 control cities are eventually included in the sample. A pilot list of low-carbon cities was obtained from the website of the National Development and Reform Commission of China. The first low-carbon pilot policy in China was implemented in 2010, and three batches of pilot cities were approved within the same year. Data on urban economic and social characteristics were obtained from the China Urban Statistical Yearbook.

3.2. Empirical Model

The pilot policies of low-carbon cities are used in a quasi-experiment to evaluate the impact of climate policies on FDI using the differences-in-differences (DID) method. The specific econometric model is formulated as

$${\mathrm{lnPFDI}}_{\mathrm{it}}=\mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{Cab}\_{\mathrm{City}}_{\mathrm{i}}\times {\mathrm{After}}_{\mathrm{t}}+Control{s}_{\mathrm{it}}\gamma +{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}},$$

(1)

where the subscripts i and t refer to city and year. The dependent variable (lnPFDI) is the logarithm of FDI per capita, which represents the scale of FDI. The core explanatory variable is the DID item (Cab_City × After), which is the interaction item of the group dummy variable (Cab_City) and year dummy variable (After). Controls refers to a vector of a series of control variables. This study also controls urban fixed effect (${\mathsf{\mu }}_{\mathrm{i}}$) and time fixed effect (${\mathsf{\lambda }}_{\mathrm{t}}$). This model is actually a classic DID regression for panel data. The urban fixed effect controls the characteristics of cities that do not change with time, including the characteristics of group dummy variables, whereas the time fixed effect captures the dummy variable characteristics of policy shocks.

By taking the quality of FDI as a dependent variable, an econometric model of the impact of climate policy on the quality of FDI is built as follows:

$$\mathrm{FDI}\_{\mathrm{Quality}}_{\mathrm{it}}=\mathsf{\alpha }+{\mathsf{\beta }}_2\mathrm{Cab}\_{\mathrm{City}}_{\mathrm{i}}\times {\mathrm{After}}_{\mathrm{t}}+Control{s}_{\mathrm{it}}\mathsf{\gamma }+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}},$$

(2)

where FDI_Quality is the dependent variable for measuring the quality of FDI. The definitions of the other variables are the same as those in Formula (1).

In addition, this study uses FDI technology spillovers to measure the quality of FDI, and the econometric model is as follows.

$$\begin{gathered} {\mathrm{Technology}}_{\mathrm{it}}=\mathsf{\alpha }+{\mathsf{\beta }}_3\mathrm{Cab}\_{\mathrm{City}}_{\mathrm{i}}\times {\mathrm{After}}_{\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{Cab}\_{\mathrm{City}}_{\mathrm{i}}\times {\mathrm{After}}_{\mathrm{t}}\times {\mathrm{lnPFDI}}_{\mathrm{it}}+{\mathsf{\beta }}_5{\mathrm{lnPFDI}}_{\mathrm{it}} \\ +Control{s}_{\mathrm{it}}\mathsf{\gamma }+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}, \end{gathered}$$

(3)

where Technology refers to the city’s technology and is measured by the green total factor productivity (GTFP) and the logarithm of the number of patents (lnPat). The marginal effect of FDI on urban technology is considered to be the quality of FDI. The greater the effect, the higher the quality of FDI. If the coefficient β₄ is significantly greater than zero, it indicates that climate policy has improved the technology spillover effect of FDI and the quality of FDI.

3.3. Variable Definition and Description

As this study focuses on the change of FDI scale and FDI quality, the dependent variables include FDI scale (lnPFDI) and FDI quality (FDI_Quality). The scale of FDI (lnPFDI) is measured by the logarithm of the total amount of FDI and the proportion of urban registered population, whereas the quality of FDI (FDI_Quality) is measured by the logarithm of the ratio of the total amount of FDI to the number of FDI enterprises. Some studies have evaluated the quality of FDI from technology spillovers [24]. Similar to Wang and Luo (2020) [41], this study measures the quality of FDI using the total investment of a single foreign enterprise. A higher FDI of a single foreign enterprise corresponds to a higher quality of FDI. In addition, this study adopts the green total factor productivity (GTFP) and the logarithm of the number of patents (lnPat) as the dependent variable to measure the technology spillover of FDI. The definitions of the main variables in this study are presented in

Table 1. Definition of related variables.

	Variables	Definition
Dependent variables	lnPFDI	FDI scale measured by the logarithm of FDI per capita
	FDI_Quality	FDI quality measured by the amount of FDI of foreign enterprises
	GTFP	Green total factor productivity, measured by data envelopment analysis
	lnPat	The logarithm of the number of city patents
Core explaining variable	Cab_City × After	Interaction of the group and year dummy variables
Control variables	lnPGDP	Logarithm of regional GDP per capita
	Population	Proportion of total population and administrative area
	Industry	Proportion of added value of the secondary industry
	lnWage	Logarithm of average wage of urban residents
	lnFiscal	Logarithm of public fiscal expenditure per capita
	lnTech	Logarithm of science-and-technology-related public expenditure
	Environment	Harmless treatment rate of domestic waste
	Internet	Logarithm of employees in the information technology service industry
	Child_Rate	Proportion of primary and secondary school students in the population

.

The core interpretation variable is the DID interaction item (Cab_City × After). As the pilot policies of low-carbon cities are implemented in batches, this study can only directly define the interaction item [4] but cannot define the year dummy variable at the time point of policy intervention. When selected as a pilot city, the local government takes a series of specific climate actions to support the development of urban low-carbon cities. Therefore, the pilot policy of low-carbon cities can be used as proxy variables of climate policies.

Referring to relevant literature and FDI location selection theory [19,40], the following control variables are selected from economic scale, factor endowment, population characteristics, public support, and environmental constraints. Economic development (lnPGDP) is measured by per capita GDP. A higher level of economic development corresponds to a greater demand for FDI. Population density (Population) is a logarithmic variable measured by the population per unit of administrative area [24]. A higher population density corresponds to a higher labor abundance. Industrialization (Industry) is measured by the proportion of the secondary industry in the total GDP [26]. Industrial structure is one of the important determinants of FDI location choice. Labor cost (lnWage) is measured by the logarithm of per capita wage. A higher regional labor cost corresponds to a less attractive FDI. Public finance (lnFiscal) is measured by the logarithm of per capita financial expenditure, whereas innovation support (lnTech) is measured by the logarithm of per capita public science and technology expenditure. Environmental regulation (Environment) is an important factor affecting FDI that is measured by the harmless treatment rate of domestic waste. The Internet (Internet) is conducive to reducing the cost of information countermeasures for FDI investment and is measured by the logarithm of employees in the information technology service industry [21]. This study also controls the characteristics of urban population structure (Child_Rate), which is measured by the proportion of primary and secondary school students in the population.

Table 2. Descriptive statistics of the main variables.

Variable	Obs	Mean	Std. Dev.	Min	Max	Correlation 1	Correlation 2
lnPFDI	3262	4.1330	1.6063	−3.5927	7.9905	1.0000	0.5104
FDI_Quality	3262	6.5000	1.2448	0.3362	10.7942	0.5104	1.0000
Cab_City×After	3262	0.2738	0.4460	0.0000	1.0000	0.1987	0.1556
lnPGDP	3262	10.4618	0.7839	8.2845	13.1851	0.6603	0.1736
Population	3262	0.0487	0.0339	0.0005	0.2759	0.3028	0.0002
Industry	3262	0.4818	0.0971	0.1170	0.8068	0.1476	−0.0020
lnWage	3262	10.5849	0.4980	9.1308	12.0622	0.3309	0.2863
lnFiscal	3262	8.5614	0.7999	6.3044	11.6026	0.5079	0.2862
lnTech	3262	10.0329	1.5848	4.2047	15.5293	0.5805	0.2401
Environment	3262	0.8622	0.2378	0.0000	1.0000	0.2711	0.1713
Internet	3262	−0.8814	1.0324	−3.9120	4.4533	0.4402	0.1071
Child_Rate	3262	0.1270	0.0414	0.0571	0.5174	0.0835	−0.0755
GTFP	3262	1.0178	0.0316	0.9601	1.1024	0.1528	0.0823
lnPat	3262	7.0409	1.7452	0.0000	12.4012	0.5740	0.1331

presents the descriptive statistics of relevant variables. Around 27.34% of the observed value shows that the pilot cities are under the influence of policies, thereby suggesting a relatively uniform distribution of the control and treatment groups. In addition, the dependent variables have a certain degree of volatility, and the econometric model can be used to investigate the formation mechanism of FDI. The control variables have a certain correlation with the two dependent variables, hence supporting the rationality of their adoption. However, the DID items show a positive correlation with the scale of FDI, thereby necessitating a further examination of whether the pilot policies of low-carbon cities can increase the FDI scale.

4. Research Results and Analysis

4.1. Results of Benchmark DID Regression

Based on Formulas (1) and (2), we use urban panel data to estimate the impact of pilot policies for low-carbon cities on FDI. The results are shown in

Table 3. Empirical results of benchmark DID regression.

Variables	Dep. Var.: lnPFDI		Dep. Var.: FDI_Quality
	(1)	(2)	(3)	(4)
Cab_City × After	−0.1742 ***	−0.2014 ***	0.1389 ***	0.1186 **
	(0.0538)	(0.0516)	(0.0515)	(0.0509)
Population	−1.0950	3.5371	5.7071 *	7.4920 **
	(3.3236)	(3.2240)	(3.1825)	(3.1784)
Environment	0.0587	0.0343	0.0044	−0.0036
	(0.0808)	(0.0781)	(0.0774)	(0.0770)
Internet	0.1318 ***	0.0903 **	0.0808 **	0.0572
	(0.0419)	(0.0406)	(0.0401)	(0.0400)
Child_Rate	1.7625 *	2.2652 **	2.6912 ***	2.9149 ***
	(1.0440)	(1.0050)	(0.9996)	(0.9907)
lnPGDP		1.2583 ***		0.4756 ***
		(0.1474)		(0.1453)
Industry		−0.7005 *		−0.3441
		(0.4209)		(0.4149)
lnWage		0.2579		0.0028
		(0.1806)		(0.1780)
lnFiscal		0.3868 ***		0.3628 ***
		(0.1177)		(0.1160)
lnTech		0.1708 ***		0.1618 ***
		(0.0323)		(0.0318)
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
Adjusted R2	0.1028	0.1827	0.2677	0.2929
Number of samples	3262	3262	3262	3262

Note: The figures in parentheses are the estimated standard errors. Asterisks are significant at 1% (***), 5% (**), 10% (*) levels.

. The dependent variable in columns (1) and (2) is FDI scale, whereas that in columns (3) and (4) is FDI quality. Columns (1) and (3) do not control for per capita GDP, labor cost, industrial structure, public finance, innovation support, and other control variables because these variables may be directly affected by policy intervention and further affect FDI. If these variables are controlled, then some indirect transmission mechanism of policy intervention on FDI may be excluded.

The empirical results in Table 3 reveal that the pilot policy of low-carbon cities have significantly reduced the scale of FDI but significantly increased its quality. From columns (1) and (2), the estimated coefficients of the DID term (Cab_City × After) are significantly negative at the 1% level, thereby indicating that strict climate policy significantly inhibits urban FDI. Meanwhile, the coefficient from columns (3) and (4) is significantly positive, thereby indicating that strict climate policy increases the quality of FDI. The empirical results in Table 3 are consistent with Hypothesis 1, thereby indicating that the pilot policy of low-carbon cities make cities pay more attention to the quality than the scale of FDI. Although Zhao and Wang (2021) [38] found a positive correlation between climate policy and FDI, this study shows that climate policies change FDI from quantity expansion to quality improvement. Despite increasing environmental costs, adopting policies to actively respond to climate change also promotes urban technological innovation and industrial structure upgrading, hence enabling cities to absorb higher-quality foreign capital.

We also obtain some interesting findings about the control variables that are mostly consistent with the theoretical explanation. For example, internet facilities can help absorb FDI by reducing information barriers. As the world enters the digital economy era, digital facilities are becoming increasingly important for international investment and trade activities [10]. The population density and proportion of primary and secondary school students represent the population structure with abundant labor, which is conducive to attracting more FDI, thereby indicating that labor advantage is an important factor in FDI location selection [42]. Public financial expenditure and innovation support help absorb FDI, which suggests that the government plays an important role in attracting foreign capital investment. A higher level of economic development corresponds to a higher attraction of FDI. In fact, this is a mystery of capital flow. International capital flows to places with abundant capital. One explanation is that the developed regions are mainly capital-intensive industries, thus increasing their demand for capital. The impact of environmental regulations, industrialization, and labor cost on FDI is not clear, which can be ascribed to the impact of these variables on FDI with a variety of complex relationships.

Table 4. Empirical results of climate policy affecting FDI quality.

Variables	Dep. Var.: GTFP			Dep. Var.: lnPat
	(1)	(2)	(3)	(4)	(5)	(6)
lnPFDI × Cab_City × After	0.0034 ***	0.0037 ***	0.0032 ***	0.0353 ***	0.0258 **	0.0529 ***
	(0.0009)	(0.0010)	(0.0011)	(0.0120)	(0.0123)	(0.0129)
lnPFDI	0.0001	0.0001	0.0003	−0.0041	−0.0167 *	−0.0322 ***
	(0.0008)	(0.0008)	(0.0008)	(0.0100)	(0.0097)	(0.0099)
Cab_City × After	0.0129 ***	0.0116 **	0.0134 **	−0.0306)	0.0115	−0.0953
	(0.0050)	(0.0052)	(0.0054)	(0.0640)	(0.0642)	(0.0657)
lnPGDP	−0.0037	−0.0034	−0.0412	0.7897 ***	0.3144 ***	2.5855 ***
	(0.0043)	(0.0059)	(0.0283)	(0.0557)	(0.0728)	(0.3453)
lnPGDP × lnPGDP			0.0017			−0.1049 ***
			(0.0013)			(0.0156)
Control Variables	No	Yes	Yes	No	Yes	Yes
City/Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Adjusted R2	0.0283	0.0270	0.0273	0.8619	0.8740	0.8759
Number of samples	3262	3262	3262	3262	3262	3262

Note: The figures in parentheses are the estimated standard errors. Asterisks are significant at 1% (***), 5% (**), 10% (*) levels.

shows the results of model (3), which is to evaluate the impact of climate policy on the quality of FDI by examining the technology spillover effects of FDI. It can be seen that the impact of FDI on urban technology is uncertain, which has been discussed in the literature [23]. The relationship between FDI and technological progress in developing countries is complex, which depends on the quality of FDI. In all columns of Table 4, the coefficients of lnPFDI× Cab_City×After are significantly positive at the 5% level, indicating that climate policy has significantly improved the technology spillovers. Therefore, it can be considered that climate policy has significantly improved the quality of FDI.

4.2. Results of Entropy Balancing DID Regression

A challenge in the classical DID estimation results lies in the non-randomness of pilot cities, which leads to huge differences between two groups of cities. Previous studies have mainly adopted propensity score-matching (PSM) to overcome such differences. However, given that cities have no initiative in the selection of pilot policies, the factors that affect the propensity score cannot be easily described. This study then adopts PSM and finds that the probability of different cities becoming pilot ones is relatively close. Therefore, the PSM method is deemed unsuitable for this work. Instead, this study adopts the entropy balancing method for the sample weighting and matching in order to make the characteristics of covariates as consistent as possible. The entropy balancing method not only does not need to rely on the decision-making model of participating projects but also tries to retain the information of similar samples by giving different weights to different samples [4,43]. This method adds a weight matrix to the cities in the control group, assigns a greater weight to the cities in the control group whose characteristics are close to those of the covariates for the treatment group, and assigns weights to the cities in the control group for a comparison with the pilot cities.

Table 5. Empirical results of entropy balancing.

Variables	Treat			Control
	Mean	Variance	Skewness	Mean	Variance	Skewness
lnPGDP	10.4300	0.5155	0.6465	10.4300	0.6922	0.4002
Population	0.0481	0.0014	2.6690	0.0481	0.0012	0.3387
Industry	0.5005	0.0073	−0.7209	0.5005	0.0079	0.0533
lnWage	10.3600	0.0681	0.3062	10.3600	0.0692	0.0848
lnFiscal	8.4810	0.4056	0.9661	8.4810	0.3637	0.2504
lnTech	10.0700	1.9050	1.0240	10.0700	0.9913	0.4903
Environment	0.8395	0.0560	−1.8520	0.8394	0.0392	−1.4840
Internet	−0.8936	1.1950	1.1150	−0.8939	0.7737	1.1310
Child_Rate	0.1332	0.0024	2.8410	0.1332	0.0008	0.6406

Note: This is the result of entropy balancing in 2010.

shows the characteristics of covariates between the control group cities and low-carbon pilot cities after entropy balancing. We calculate the weight of each city in the control group according to the entropy balancing in 2010. Table 5 shows that after entropy balancing, the characteristics of the control group cities are very close to those of the pilot cities. At the mean level, the covariant characteristics of these two groups of cities are almost identical. Although some gaps can be observed between variance and skewness, the overall gap is small, thereby indicating that the characteristics of the higher-order moments are relatively close. Therefore, entropy balancing overcomes the differences between the two groups of cities before the policy intervention. Moreover, the samples weighted by entropy balancing generally conform to the parallel trend assumption, hence overcoming the problem that such assumption cannot be easily met in a DID design.

Using the weighted samples of control group cities and pilot cities obtained via entropy balancing, a DID regression is performed to test the impact of the pilot policies of low-carbon cities on the scale and quality of urban FDI. The estimated results are shown in

Table 6. Empirical results of entropy balancing DID regression.

Variables	Dep. Var.: lnPFDI		Dep. Var.: FDI_Quality
	(1)	(2)	(3)	(4)
Cab_City × After	−0.0659	−0.1370 ***	0.2037 ***	0.1471 ***
	(0.0503)	(0.0468)	(0.0471)	(0.0457)
Population	1.7218	10.1384 ***	7.9775 ***	13.1377 ***
	(3.1674)	(3.0251)	(2.9661)	(2.9523)
Environment	−0.0647	−0.0342	−0.0340	−0.0110
	(0.0845)	(0.0787)	(0.0791)	(0.0768)
Internet	0.1950 ***	0.1050 ***	0.0708 *	0.0138
	(0.0402)	(0.0376)	(0.0377)	(0.0367)
Child_Rate	0.3717	1.4829	1.6312 *	2.4306 ***
	(0.9605)	(0.9024)	(0.8994)	(0.8807)
lnPGDP		1.5443 ***		0.8293 ***
		(0.1330)		(0.1298)
Industry		0.1807		0.5015
		(0.3913)		(0.3819)
lnWage		0.8093 ***		0.3701 **
		(0.1570)		(0.1532)
lnFiscal		0.3285 ***		0.4039 ***
		(0.0956)		(0.0933)
lnTech		0.1119 ***		0.0915 ***
		(0.0306)		(0.0299)
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
Adjusted R2	0.1002	0.2332	0.2997	0.3518
Number of samples	3262	3262	3262	3262

Note: The figures in parentheses are the estimated standard errors. Asterisks are significant at 1% (***), 5% (**), 10% (*) levels.

. In general, the pilot policies for low-carbon cities have reduced the scale of urban FDI but improved its quality, which is consistent with the empirical results of classic DID in Table 3. Except for Column (1), the coefficients of DID term in the other columns are significant, and their symbols have not changed. The T value of the DID term in Column (1) is −1.31, which means that this DID term still has a negative impact on the scale of FDI with a large probability. Therefore, the empirical results of entropy balance DID support Hypothesis 1. In other words, a strict climate policy squeezes out FDI but also promotes an improvement in FDI quality. From the absolute value of the coefficients, the coefficients of the DID term in Columns (1) and (2) in Table 6 are smaller than those in the corresponding column in Table 3. Furthermore, the coefficients of the DID term in Columns (3) and (4) in Table 6 are larger than those in the corresponding column in Table 3. In sum, the empirical results of entropy balancing DID regression reveal that the pilot policy of low-carbon cities has a less negative impact on the scale of FDI but a more positive impact on the quality of FDI. These empirical results support Hypothesis 1, that is, the location choice of FDI is driven by non-renewable energy [13], and climate policies restrict the use of non-renewable energy, hence driving resource-seeking FDI to transfer to other regions. Such location choice may also attract some technology oriented FDI, which would drive cities to focus on expanding the scale to improving the quality of FDI.

4.3. Robustness of DID Regression

The parallel trend before the policy intervention is the premise of DID design. If the two groups of cities before the policy intervention do not meet the parallel trend, then the estimated effect may include the potential trend change characteristics of these two groups. Therefore, this study tests the parallel trend of these two groups of cities and draws a test chart of such trend, as shown in Figure 2. The left figure shows the parallel trend test result when the dependent variable is FDI scale, whereas the right figure shows the parallel test result when the dependent variable is FDI quality. All parallel trend tests are based on entropy balancing DID.

Figure 2 shows that the FDI characteristics of the two groups of cities meet the parallel trend assumption and that the DID design can be used to evaluate the FDI effect of policy intervention. Before the implementation of the pilot policy, no significant policy effect can be detected, thereby indicating that the scale and quality of FDI in the two groups of cities meet the parallel trend hypothesis. No significant policy effect was also observed in the first few years of the 2010s. As the parallel trend test assumes that 2010 is the time point for the first policy implementation, a large number of cities have not implemented pilot policies at this time. The second vertical dotted line shows that most pilot cities have implemented pilot policies, thereby explaining the significant policy effects observed after 2012. Specifically, the pilot policy has reduced the scale of urban FDI yet increased its quality. These effects are consistent with the results of the DID regression.

To further verify the robustness of the DID results, this study also performs a placebo test. Specifically, we randomly select some non-pilot cities as pilot cities, place the remaining non-pilot cities to a control group, and then calculate the DID coefficient and its p value obtained by randomization grouping. We obtain 1000 groups of coefficients and p values by performing randomized sampling for 1000 times. The distribution of coefficients and p values is shown in Figure 3. The left and right figures show the placebo test result when the dependent variable is FDI scale and FDI quality, respectively. The coefficients of the DID term obtained from the randomized samples do not have significant policy effects. These test results support Hypothesis 1, which identifies the causal relationship between climate policy and the scale and quality of FDI.

4.4. Results of Mechanism Analysis

The policy of low-carbon pilot cities is an initiative to promote low-carbon development, so it is also necessary to examine the specific measures that this initiative affects foreign direct investment. This study selects environmental regulation (ER) and financial expenditure of science and technology (lnFiscal) as mechanism variables to investigate specific measures of climate policy affecting FDI. This study selects environmental regulation and financial expenditure of science and technology as mechanism variables to investigate specific measures of climate policy affecting FDI. Environmental regulation is defined as the relevant description of environmental governance in the government work report. Science and technology financial expenditure indicates the degree of support or subsidy provided by local governments to science and technology activities.

Table 7. Empirical results of mechanism analysis.

Variables	(1) ER	(2) lnFiscal	(3) lnPFDI	(4) FDI_Quality
Cab_City × After	0.1367 *	0.0625 ***	−0.2022 ***	0.1169 **
	(0.0715)	(0.0127)	(0.0516)	(0.0509)
lnFiscal			0.3888 ***	0.3660 ***
			(0.1176)	(0.1160)
ER			−0.0091 ***	0.0131
			(0.0013)	(0.0130)
Control variables	Yes	Yes	Yes	Yes
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
Adjusted R2	0.1766	0.9196	0.1824	0.2937
Number of samples	3262	3262	3262	3262

Note: The figures in parentheses are the estimated standard errors. Asterisks are significant at 1% (***), 5% (**), 10% (*) levels.

shows the results of mechanism analysis.

The results in Table 7 show that the changes in the scale and quality of FDI are jointly determined by environmental regulation and scientific and technological fiscal expenditure caused by climate policy. It can be seen that cities that have adopted low-carbon pilot policies have significantly strengthened environmental regulations and increased science and technology subsidies. In addition, environmental regulation has a significant negative impact on the scale of FDI, while it has an insignificant impact on the quality of FDI. The fiscal expenditure of science and technology has a significant positive impact on the scale and quality of FDI, which means that local governments can mitigate the adverse impact of environmental costs on FDI by encouraging urban innovation.

4.5. Heterogeneity of Urban Characteristics

In order to investigate the heterogeneity effect of cities, the sample is divided into large and small cities according to their number of urban registered population. Results of urban scale heterogeneity are shown in Columns (1) to (4) in

Table 8. Heterogeneous results based on urban scale.

Variables	Dep. Var.: lnPFDI		Dep. Var.: FDI_Quality
	(1) Large	(2) Small	(3) Large	(4) Small	(5) Outflow	(6) Inflow
Cab_City × After	−0.2985 ***	−0.0648	0.0419	0.1712 **	−0.2239 ***	0.3833 ***
	(0.0534)	(0.0701)	(0.0526)	(0.0682)	(0.0727)	(0.0605)
lnPGDP	0.5933 ***	1.7571 ***	(0.1286)	0.9131 ***	1.5789 ***	0.7261 ***
	(0.1923)	(0.1950)	(0.1895)	(0.1897)	(0.2171)	(0.1726)
Industry	−0.7295	0.1289	−0.5416	0.9262 *	−3.2016 ***	2.2400 ***
	(0.5484)	(0.5415)	(0.5404)	(0.5266)	(0.6239)	(0.5252)
lnWage	1.2650 ***	0.3524	0.5402 **	0.0949	−0.2173	0.3272
	(0.2212)	(0.2223)	(0.2180)	(0.2162)	(0.2082)	(0.2321)
lnFiscal	0.3566 ***	0.3740 **	0.4124 ***	0.4747 **	0.5243 ***	0.3633 ***
	(0.0832)	(0.1898)	(0.0820)	(0.1846)	(0.1224)	(0.1359)
lnTech	0.0370	0.1194 ***	0.0533	0.0923 **	0.0564	0.1240 ***
	(0.0413)	(0.0428)	(0.0406)	(0.0416)	(0.0434)	(0.0421)
Population	19.6666 ***	4.5657	25.5719 ***	6.5709	32.3483 ***	9.3816 ***
	(4.4741)	(4.9116)	(4.4087)	(4.7764)	(9.3421)	(3.2283)
Environment	−0.1406	−0.0291	−0.1828 *	0.0332	0.0312	−0.0688
	(0.0969)	(0.1126)	(0.0955)	(0.1095)	(0.0914)	(0.1280)
Internet	0.0837 **	0.0296	−0.0052	−0.0460	0.0245	0.1041 **
	(0.0381)	(0.0649)	(0.0376)	(0.0631)	(0.0622)	(0.0465)
Child_Rate	1.1533	1.3786	2.6884 ***	1.2816	6.5135 ***	−0.1397
	(0.8537)	(1.6192)	(0.8412)	(1.5746)	(1.3708)	(1.2492)
City fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Adjusted R2	0.4171	0.1905	0.585	0.2633	0.4067	0.3443
Number of samples	1293	1969	1293	1969	1387	1875

Note: The figures in parentheses are the estimated standard errors. Asterisks are significant at 1% (***), 5% (**), 10% (*) levels.

. The dependent variable in Columns (1) and (2) is the scale of FDI. The pilot policy of low-carbon cities has significantly reduced the scale of FDI in large cities but has no significant impact on the scale of FDI in small cities. Columns (3) and (4) show the changes in FDI quality and reveal that the pilot policy of low-carbon cities has significantly improved the quality of FDI in small cities but has no significant impact on the quality of FDI in large cities. This difference may be due to the different paths of pursuing low-carbon transformation between these two types of cities. Large cities can achieve low-carbon transformation through industrial structure change. For example, they actively develop their tertiary industry and reduce their investment in energy-intensive sectors so as to realize a low-carbon transformation of cities. The economic development of small cities largely depends on manufacturing or energy-intensive sectors. The technology of these sectors can be improved by introducing high-quality FDI investment so as to achieve a low-carbon transformation of urban development [5]. This result has two implications. On the one hand, the pilot policy of low-carbon cities may lead to the transfer of FDI between cities. On the other hand, the pilot policy has a greater negative impact on the scale of FDI in large cities but has a greater positive impact on the quality of FDI in small cities.

The cities in the sample are further divided into capital outflow and capital inflow cities for a heterogeneity analysis. Specifically, the regions where the residents’ year-end deposit balance of financial institutions is greater than the loan balance of financial institutions are defined as capital outflow cities, whereas the other regions are defined as capital inflow cities. The sub-sample analysis results divided by capital inflow and capital outflow are shown in Columns (5) and (6) in Table 8. The pilot policies for low-carbon cities significantly promote the quality of FDI flowing into cities but reduce the quality of FDI flowing out of cities. This finding is in line with our expectations. Cities with capital inflows have a stronger demand for capital, especially low-carbon transformation through FDI, which requires higher-quality FDI. Given that the demand for FDI from capital outflow cities has declined, the possibility of high-quality FDI inflow is low. Another explanation is that FDI flows from big to small cities, thereby improving the quality of FDI in small and capital inflow cities.

At the static level, the strict climate policy increases the production cost, which in turn reduces the attraction of the region to FDI. However, climate policies may also produce the Porter effect, which can dynamically attract FDI and improve its quality through indirect mechanisms, such as technological innovation [44]. Therefore, technological innovation plays an important role in the impact of climate policy on FDI, hence underscoring the need to further investigate the heterogeneity effect brought about by urban innovation characteristics. The cities are then sub-divided according to two innovation-related variables, and the estimated results of these sub-samples are presented in

Table 9. Heterogeneous results based on urban innovation.

Panel A: Heterogeneity of innovative city construction
Variables	Dep. Var.: lnPFDI		Dep. Var.: FDI_Quality
	(1) Yes	(2) No	(3) Yes	(4) No
Cab_City × After	0.1762 ***	−0.3575 ***	0.4110 ***	−0.0721
	(0.0638)	(0.0616)	(0.0669)	(0.0592)
Control variables	Yes	Yes	Yes	Yes
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
Adjusted R2	0.3003	0.2559	0.4685	0.3513
Number of samples	826	2436	826	2436
Panel B: Heterogeneity of government support for innovation
Variables	Dep. Var.: lnPFDI		Dep. Var.: FDI_Quality
	(5) High	(6) Low	(7) High	(8) Low
Cab_City × After	0.1226 **	−0.3018 ***	0.3043 ***	0.0362
	(0.0561)	(0.0822)	(0.0593)	(0.0785)
Control variables	Yes	Yes	Yes	Yes
City fixed effect	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes
Adjusted R2	0.1239	0.2011	0.3493	0.2223
Number of samples	1585	1677	1585	1677

Note: The figures in parentheses are the estimated standard errors. Asterisks are significant at 1% (***), 5% (**) levels.

. First, the sample is divided according to whether the city has implemented an innovation demonstration city policy as shown in Panel A in Table 9. Second, the sample is divided according to the level of the city’s scientific and technological financial expenditure as shown in Panel B in Table 9.

Table 9 shows that the characteristics of urban innovation significantly affect the relationship between climate policy and FDI. Meanwhile, Panel A shows that the pilot policy of low-carbon cities has a significant positive impact on the quality and scale of FDI in innovative demonstration cities, thereby confirming that innovative demonstration cities actively promote the Porter effect of climate policy, thus increasing the scale and quality of FDI by absorbing high-quality FDI. Meanwhile, in non-innovation demonstration cities, climate policy significantly reduces the scale of FDI and has no significant negative impact on the quality of FDI. The same conclusions can be derived for innovation support. For cities with strong innovation support, the climate policy increases the scale and quality of FDI but mainly has a negative impact on FDI for cities with weak innovation support. Pan et al. (2022) [4] argued that climate policy leads to urban low-carbon innovation, which is an effective means to achieve the dual goals of climate governance and economic benefits. Therefore, Hypothesis 2 is empirically supported.

5. Conclusions

Global climate change has brought great challenges to human sustainable development, and low-carbon development is a necessary way to cope with climate change [45]. However, developing and emerging countries are always hesitant to adopt climate policies mainly because they worry that strict climate policies would increase their production costs and reduce their investment enthusiasm [46,47]. This study uses panel data of Chinese cities from 2006 to 2019 and adopts the DID method to investigate how the pilot policy of low-carbon cities affect the quality and scale of urban FDI. In addition, this study innovatively adopts the entropy balancing DID method to improve the reliability of its findings. This study also examines how the characteristics of city size and urban innovation affect the relationship between climate policy and FDI. The main conclusions are as follows.

First, the scale and quality of FDI in pilot cities have changed significantly after policy intervention. Specifically, climate policy has significantly reduced FDI in Chinese cities but improved its quality. Specifically, the regulatory pressure and innovation incentives brought by climate policy change the quantity and quality of FDI in cities. The empirical results of the entropy balancing DID regression are consistent with those of the classic DID regression. However, in the entropy balancing DID regression, the pilot policy of low-carbon cities has a less negative impact on the scale of FDI but has a greater positive impact on its quality. Results of both the parallel trend test and placebo test support the reliability of these conclusions. The findings of this study are also in line with the classical international investment theory, that is, the rising cost of resource factors reduces the scale of FDI. This study also discovers that a strict climate policy can shift the focus of FDI from quantity to quality, which can help developing countries break away from the low-end lock of the value chain. Therefore, climate policy does not have a single negative impact on international investment and has some beneficial aspects. International investment coordination organizations should strengthen the positive response of international investment to climate policies and promote responsible international investment.

Second, urban heterogeneity, urban size, urban innovation, and other characteristics affect the relationship between climate policy and FDI. The pilot policy of low-carbon cities mainly reduces FDI scale in large cities and improves the quality of FDI in small cities. The pilot policies for low-carbon cities significantly promote the quality of FDI flowing into cities but reduce the quality of FDI flowing out of cities. In addition, the climate policy in innovative cities improves the quantity and quality of FDI but has a negative impact on FDI in non-innovative cities. This finding validates the Porter hypothesis because the production of the Porter effect depends on the innovative characteristics of cities. In other words, innovation is an effective path to solve the static adverse effects of climate policy. The heterogeneity effect is in line with the theoretical view of FDI investment motivation. Specifically, climate policy has reduced resource-seeking FDI but induced an urban innovation effect that, in turn, increases technology-seeking FDI.

Our findings have important implications for China and other developing countries as well as for political negotiations on global climate governance. First, developing countries should pay attention to the adverse effects of climate policies on their production costs, investment, and foreign capital inflows, and global climate negotiations should fully consider the demands of economic development and improvement of people’s quality of life in developing countries. Second, developing countries should be actively encouraged to take climate action initiatives to cope with changes in global climate policies and economic and trade rules [47]. Third, developing countries should attach importance to the adverse effect of climate policies, create a better innovation environment, and encourage green technology innovation so as to improve their climate competitiveness in dynamic competition. Fourth, these findings also offer some recommendations for the implementation of the green Belt and Road Initiative (BRI). Countries in the BRI region have different climate policies, which may bring new challenges to international investment. Multinational enterprises should be guided to carry out responsible investment in this region to avoid the shock of climate policies. The COVID-19 pandemic has driven many countries in the BRI region to relax their climate policies. China and other countries should strengthen their green investment and cooperation in the BRI region and actively take measures to mitigate climate change [48]. For example, these countries should formulate climate standards for international investment and design their process arrangement of climate policies.

Some limitations of this work need to be considered in future research. First, the quality of FDI can be measured by using the FDI data of different industries instead of the size of a single investment. Second, the regional transfer of FDI capital across cities can be further verified. Third, given that the COVID-19 pandemic has had a profound impact on global climate governance and international investment, the sample period can be extended to the latest year when data are available. Fourth, due to the large differences between the climate policies of different countries, the effects of climate policies on FDI across countries should be compared.