Does China’s Low-Carbon Pilot Policy Promote Foreign Direct Investment? An Empirical Study Based on City-Level Panel Data of China

Abstract

As an important driving force of China’s economic growth, foreign direct investment (FDI) may be affected by China’s low-carbon pilot policy. Therefore, this paper regards the low-carbon pilot policy as a quasi-natural experiment, and uses the difference-in-difference (DID) model and the panel data of 189 cities in China from 2011 to 2018 to explore the actual impact and intermediary mechanism of low-carbon pilot policy on FDI. The study found that low-carbon pilot policy has a significant promotion effect on FDI, and industrial optimization and upgrading is an important way. At the same time, we construct the difference-in-difference-in-difference (DDD) model, and discuss the heterogeneity of policy effect caused by resource endowments and the individual characteristics of government officials in the process of policy implementation. The results indicate that resource-rich cities can enhance the promotion effect of low-carbon pilot policy on FDI. Similarly, when the mayor of the pilot city is a female, or obtains a master’s degree or a doctorate degree, or majored in non-economics, respectively, the policy effect will be more obvious. In addition, in order to verify the reliability of the research conclusions, this paper also uses a placebo test and data truncation to conduct a series of robustness tests.

1. Introduction

Since the reform and opening up, China’s economy has made remarkable achievements. In the context of economic globalization and trade liberalization, with the deepening of China’s opening up, the huge market, abundant resources, cheap labor, and stable investment environment have created many dividends for foreign-invested enterprises, making China one of the most attractive host countries. As an important driving force of economic development in China, the investment of foreign-invested enterprises in China has generally maintained a growth trend [1], as shown in Figure 1. Undoubtedly, foreign direct investment (FDI) has brought a lot of capital, green technology, and advanced management experience to China [2]. However, in the context of increasing downward pressure on the economy, China’s economic growth is still mainly dependent on traditional high-pollution industries. Whether the introduction of FDI will further aggravate the energy crisis, environmental pollution and ecological degradation of China has attracted widespread attention [3,4].

The BP Statistical Review of World Energy 2020 pointed out that, in 2019, China’s net primary energy growth accounted for more than three-quarters of net global growth, and the country also produced 9.8258 billion tons of carbon dioxide emissions. As one of the important contributors to energy consumption and carbon emissions, China plays a major role in global environmental governance, and has begun to explore a low-carbon, resource saving, and environmentally friendly development model [5]. Under the guidance of the concept of sustainable development, the Chinese government has successively issued laws, policies and plans related to environmental protection, hoping to establish an ecological civilization system with the government as the leader, enterprises as the main stakeholders, and the public as the main participants [6]. For example, the Law of the People’s Republic of China on the Prevention and Control of Atmospheric Pollution, establishes an energy saving and environmentally friendly society, and blue-sky defensive war strategy. These measures indicate that China’s environmental governance methods have gradually developed from “treatment after pollution” to “pollution prevention”, “ecological efficiency”, and “product life cycle (PLC)” [7].

It is worth noting that while the central government has made many efforts to protect the environment at the national level, most environmental protection policies are developed and implemented by local governments [8,9]. Practice shows that environmental decentralization has become the mainstream political arrangement for environmental governance [10]. At present, the construction of low-carbon cities has become the core strategy for addressing climate change around the world [11]. In this context, in order to actively respond to the international community’ efforts on climate change governance and fulfill the carbon emission reduction commitments made in the Copenhagen and Paris conferences [12], the Chinese government, following the multi-nested pilot demonstration mechanism of “pilot-diffusion-promotion” [13], launched three batches of low-carbon pilot projects in 2010, 2012, and 2017, respectively [14]. In view of China’s actual national conditions and historical experience, “policy pilot” (referred to as “pilot”) or “policy experiments” is not only a unique policy testing and innovation mechanism in China’s governance practice, but also an important means and policy form for carrying out national reform and improving national governance [11]. As an exploratory policy experiment, the low-carbon pilot project aims to combine national goals for climate change governance with the low-carbon behavior of local governments, summarize the practical experience of local low-carbon transformation, and explore the policy options and the government action that can be replicated and promoted on a large scale [15]. Based on this, the Twelfth Five-Year Plan for National Economic and Social Development of the People’s Republic of China (The Twelfth Five-Year Plan) listed the low-carbon pilot policy as a key governance measure for controlling greenhouse gas emissions [11].

Compared with non-pilot cities, the pilot cities will set more stringent, detailed, and targeted emission reduction targets and emission reduction tasks [16], mainly including low-carbon agriculture (e.g., biogas technology, solar collectors), low-carbon buildings (e.g., environmentally friendly building materials, water-source heat pumps, ground-source heat pumps), low-carbon transportation (e.g., new energy vehicles), and carbon sinks (e.g., afforestation, biological carbon sequestration). Driven by the low-carbon city pilot policy, during the “Twelfth Five-Year Plan” period, China’s energy consumption per unit of GDP dropped by 18.2%, and the proportion of non-fossil energy consumption increased by 2.6% [17]. As of 2019, during the “Thirteenth Five-Year Plan” period, China’s energy consumption per unit of GDP dropped by 13.2%, and carbon dioxide emissions per unit of GDP dropped by 18.2%. This implies that low-carbon pilot policy is playing an increasingly important role in improving energy efficiency, developing new energy sources, and controlling total carbon emissions, which will help to improve the ecological environment, accelerate sustainable development, and fulfill emission reduction commitments. However, as one of the important measures to improve the city’s ability to deal with climate change [18], it is still unclear how the implementation of the low carbon pilot policy will affect FDI.

Due to the late start of China’s low-carbon pilot projects, the existing literature mainly focuses on the concept, planning, and cases of low-carbon cities, but lacks quantitative research on the complex relationship between low-carbon pilot policy and FDI. Therefore, in order to fill the research gap, this paper employs the difference-in-difference (DID) model to explore the actual impact and intermediary mechanism of low-carbon pilot policy on FDI. In addition, we also construct the difference-in-difference-in-difference (DDD) model to discuss the heterogeneity of policy effect from the perspective of policy implementation. In general, this paper may provide certain theoretical support and empirical evidence for the government to adjust policy measures to balance environmental and economic benefits.

The remaining part of this paper is organized as follows: Section 2 reviews the literature on the impact of environmental regulation on FDI. Section 3 introduces the model specification, main hypothesis, and variables. Section 4 presents the benchmark regression results, mediation effect test and robustness tests. Section 5 performs a series of heterogeneity analysis. Section 6 summarizes the conclusions and provides policy recommendations.

2. Literature Review

Environmental regulation is closely related to industrial transfer, international trade, and FDI. Due to the differences in study area, industry category, and host and home countries’ socio-economic conditions, scholars have put forward two completely different views on the relationship between environmental regulation and FDI on the basis of the Pollution Heaven Hypothesis and the Porter Hypothesis. These will provide valuable enlightenment for this paper.

The first view is that the host country’s environmental regulation leads to capital outflows by reducing the rate of return on capital [19]. Scott (2005) divided the Pollution Heaven Hypothesis into a series of logical steps and pointed out that high-intensity environmental regulation adjusts the market trade structure and investment behavior by changing the production costs of enterprises, which will urge pollution-intensive enterprises to transfer from developed countries with strict environmental regulation to developing countries with relatively lax environmental regulation [20]. Most studies using panel data or instrumental variable method have confirmed the existence of pollution heaven hypothesis [21]. Mulatu et al. (2010) investigated the impact of environmental regulation on industrial location, and found that pollution-intensive industries (e.g., printing and dyeing, industrial chemicals) are concentrated in countries with lax environmental regulation (i.e., Greece, Belgium), while clean industries (i.e., radio, TV, and communication) are concentrated in countries with stringent environmental regulation (i.e., Finland, Switzerland) [22]. In addition, environmental regulation has a breakthrough effect on the resource curse [23].

The second view is that strict environmental regulation could increase strategic capital inflows and improve overall social welfare [24]. As mentioned in the Porter Hypothesis, appropriate environmental regulation can promote innovative activities, and improve the comprehensive utilization rate of resources and return rate of investment cost [25]. When the expected benefits outweigh the costs caused by environmental protection, the capital market may produce the “race to the top” effect, which will attract high-quality FDI. Furthermore, the green production processes and clean technologies that consciously or unconsciously spread with the inflow of FDI can bring external economies to the host country [26], thus producing technology spillover effects. Therefore, stricter environmental regulation will drive the development of emerging industries [27], such as new energy industry and green ecological industry [28].

The third view is that despite anecdotal evidences, empirical studies do not provide definitive results on the relationship between the two. Some studies even found that environmental regulation has no effect on FDI [29,30]. For example, Eskeland and Harrison (2003) pointed out that under the complementary effect of capital and pollution control, environmental regulation may increase or decrease FDI [31]. A study by Grossman and Krueger (1995) on the activity pattern of Mexican border processing plants also reached a similar conclusion [32].

To sum up, the existing literature usually uses a single indicator to measure environmental regulation, such as environmental taxes, environmental laws, environmental permit fees, pollution control costs, energy subsidies, inverse of the emission intensity per unit of GDP, and the environmental standard slack, which may reduce the accuracy of empirical results. Taking into account the complexity and comprehensiveness of the low-carbon pilot policy [33], this study will examine the impact of low-carbon pilot policy on FDI from a holistic perspective.

3. Methodology and Data

3.1. Benchmark Regression Model Construction

The low-carbon pilot has become a key measure for China to address climate change at the local level, as well as becoming one of the city-level mitigation strategies for carbon emissions, which is of great practical significance for achieving the goal of “carbon peak by 2030 and carbon neutral by 2060” [15]. At the same time, by combing the existing literature, we know that FDI is sensitive to the change of environmental regulation to a certain extent. Therefore, in the context of economic globalization, it is necessary to test the impact of low-carbon pilot policy on FDI.

In recent years, the DID model has been widely used in quantitative evaluation of the implementation effect of public policy or pilot project. On the one hand, the DID model can capture the unobservable individual heterogeneity, alleviate the omitted variable bias, and weaken the impact of other factors on FDI. At the same time, DID can also solve endogeneity to a certain extent, so as to more accurately estimate the “net effect” of low-carbon pilot policy on FDI [34]. Therefore, this paper applies the DID model to examine the actual impact and intermediary mechanism of a low-carbon pilot policy on FDI. The benchmark regression model of this paper is shown in Equation (1). In order to weaken the problem of heteroscedasticity, we also perform logarithmic processing on some variables.

$$lnFD{I}_{it}={\alpha }_0+\beta Pos{t}_t\times Treate{d}_i+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it}.$$

(1)

Among them, $FD{I}_{it}$ is the dependent variable, which denotes the FDI of city $i$ in year $t$; $i=1,2,\dots ,N$, $t=1,2,\dots ,T$. $Pos{t}_t$ denotes the time dummy variable; $Treate{d}_i$ denotes the city dummy variable. $Z_{it}$ denotes control variables. The interaction term $Pos{t}_t\times Treate{d}_i$ of $Pos{t}_t$ and $Treate{d}_i$ is the core independent variable, which represents implementation of low-carbon pilot policy. ${\nu }_t$ denotes year fixed effects, controlling yearly factors common to cities, such as exchange rates, business cycles, macroeconomic fluctuations, monetary policy shocks, credit spread adjustments and changes in the financing environment. ${\mu }_i$ denotes city fixed effects, controlling time-invariant city characteristics, such as geographic conditions, climatic characteristics, and resource endowments [35]. ${\epsilon }_{it}$ denotes the error term. Note that this paper focuses on the coefficient $\beta$, which is used to measure the “net effect” of the low-carbon pilot policy on FDI.

3.2. Research Hypothesis

3.2.1. Impact of Low-Carbon Pilot Policy on FDI

China’s low-carbon pilot projects are jointly controlled by the central and local governments. Based on the top-level design concepts, the central government has formulated clear policy direction and general policy goals. Meanwhile, the Ministry of Ecology and Environment will conduct dynamic monitoring and comprehensive assessment of the life cycle of low-carbon pilot projects. Under the supervision of the central government and the constraints of the cadre assessment mechanism, the local governments will introduce systematic low-carbon action plans, formulate mandatory carbon emission standards and specific emission reduction targets, identify key emission reduction projects, and implement specific environmental protection measures [11,36]. Therefore, as environmental governance costs increase, low-carbon pilot policy may inhibit FDI.

However, under the special background of China’s continuous efforts to achieve a win-win situation between urban low-carbon transformation and economic growth, the promotion effect of low-carbon pilot policy on FDI is also worth looking forward to, mainly discussing from the following three points.

Firstly, China is in the process of rapid industrialization and urbanization, and economic growth still depends on resource-intensive industries, in order to effectively limit the carbon emission behavior of the enterprises, the environmental protection department will implement strict carbon emission verification, carbon assessment and cleaner production audits, order those enterprises that do not meet emission requirements to rectify by a deadline, and shut down enterprises that are unable to meet their governance deadlines [10]. This will inevitably have a greater impact on local traditional industries with high-energy consumption, high-emission, and high-pollution, such as manufacturing, real estate, mining, and power industries [22,36,37]. As Asghari (2013) mentioned in the research, strengthening environmental regulation in the host country will bring more cost growth to local enterprises [38]. With the rising costs of environmental governance, these enterprises may be forced, or even voluntarily withdrawn from the market, which will lead to a larger market gap. However, for foreign-invested enterprises, the long-term fierce competition environment and strict environmental regulations of their home countries enable them to effectively deal with the pressure of low-carbon pilot policy [39]. Therefore, under the possible market situation where demand exceeds supply, foreign-invested enterprises can greatly reduce production costs by virtue of advanced production technology, equipment, human capital, and management experience, thus obtaining huge technological and market dividends.

Secondly, in order to promote regional economic growth and stimulate market vitality in the environment of emission reduction, local governments will guide FDI into low-carbon industries through special fund subsidies, financial support (e.g., carbon trading market, carbon emission reduction futures, and options market), and financial preferential services. At the same time, under the guidance of low-carbon pilot policy, financial institutions will innovate low-carbon financial products, and increase credit support for low-carbon projects, low-carbon products, and low-carbon technologies. These preferential measures will offset pollution control costs and environmental control costs will be offset, thus greatly reducing the production costs of foreign-invested enterprises and stimulating their investment enthusiasm [17].

Finally, in view of the important role of low-carbon consumers in promoting the construction of low-carbon cities, local governments will strengthen low-carbon knowledge publicity and public opinion guidance, encourage low-carbon living (e.g., use clean energy), build low-carbon consumption patterns (e.g., purchase low-carbon products), and advocate low-carbon travel (e.g., use public transport, and new energy vehicles) [40]. As residents’ low-carbon awareness gradually transforms into green lifestyles and low-carbon consumption behaviors, a new environmentally friendly consumer market has gradually formed. In the low-carbon consumption environment vigorously created by local governments, in order to obtain more low-carbon product market shares, foreign-invested enterprises will have the motivation to use their advantages in green production processes and low-carbon equipment to produce low-carbon products.

The investment decision of foreign-invested enterprises depends on the trade-off between expected costs and expected benefits. When the expected benefits are higher than the expected costs, foreign-invested enterprises will make additional investments. Therefore, the actual impact of low-carbon pilot policy on FDI will be the result of the above-mentioned multi-factors game. Based on the above analysis, we predict that under the multiple incentives of technological dividends, potential market profits, and preferential policies, the low-carbon pilot policy may promote FDI. Therefore, we propose the first hypothesis H1 of this paper.

Hypothesis 1 (H1).

China’s low-carbon pilot policy will promote FDI.

3.2.2. Intermediary Mechanism of Policy Effect

Although the above analysis points out that low-carbon pilot policy may promote FDI, the specific impact mechanism is still unclear. At this stage, China’s economic aggregate and traditional industrial capacity continues to expand, and the supporting role of high-tech industries has not yet been highlighted. This means that the heavy-duty characteristics of the industrial structure are difficult to fundamentally change in a short period of time. Therefore, as an important part of China’s economic growth, the traditional high-pollution industries are still the key target of environmental supervision. In order to achieve a win-win situation between low-carbon development and economic growth, pilot cities have introduced a series of policies and measures to accelerate the transformation and upgrading of traditional high-pollution industries. This will bring good development opportunities and a good investment environment for foreign-invested enterprises. Based on this, this paper predicts that a low-carbon pilot policy can attract FDI through industrial optimization and upgrading, which is mainly reflected in the following three aspects.

(1)

Promoting the transformation and upgrading of traditional high-pollution industries, and continuously optimizing the foreign investment environment. In order to accelerate the low-carbon transformation of high-carbon industries, pilot cities will strictly control the production capacity of traditional industries and accelerate the elimination of backward industrial production capacity and technical equipment, such as iron and steel, cement, coal, smelting and casting, chemicals, and building materials. In the process of vigorously promoting the transformation and upgrading of traditional high-pollution industries, local governments will introduce a series of preferential policies and supporting services, increase support for low-carbon industries and projects, and improve the proportion of investment in advanced technology and green production equipment. These measures will create a favorable investment environment for foreign-invested enterprises;

(2)

Giving full play to the demonstration role of low-carbon transformation of industrial enterprises and the agglomeration effect of low-carbon industrial parks, and striving to guide FDI into low-carbon projects. With the support of national and provincial policies and funds, pilot cities will give priority to approving low-carbon demonstration projects, build low-carbon industrial parks, support the promotion and application of low-carbon products and low-carbon technologies, and promote the low-carbon transformation of industrial enterprises. The resulting agglomeration effect can further improve the production efficiency, market competitiveness and expected profits of foreign-invested enterprises, thus gradually forming a virtuous circle of attracting foreign investment;

(3)

Pilot cities will gradually establish systems of low-carbon product certification and carbon labeling, formulate standards for production and sale of low-carbon product, and guide residents to low-carbon consumption, thus creating a new market for low-carbon product. With advanced production technology and equipment, as well as low-carbon production lines, foreign-invested enterprises are more competitive in the low-carbon market. Therefore, in the face of development opportunities brought about by the optimization and upgrading of traditional industries, foreign-invested enterprises will actively transform production decision-making and investment fields, accelerate the adjustment of resource allocation and product structure, and optimize production processes, thus gaining more market share and profits.

Based on the above analysis, this paper guesses that China’s low-carbon pilot policy will promote FDI through industrial optimization and upgrading, that is, hypothesis H2. In order to verify the mediation effect, we further construct the mediation effect model, as shown in Equations (2) and (3). Taking into account the research theme and the availability of data, we use the ratio of the added value of the secondary industry to GDP to measure industrial optimization and upgrading. In addition, Figure 2 shows the impact path of the low-carbon pilot policy on FDI.

$$IN{D}_{it}={\alpha }_0+{\beta }_{1}Pos{t}_t\times Treate{d}_i+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it},$$

(2)

$$lnFD{I}_{it}={\alpha }_0+{\beta }_{2}Pos{t}_t\times Treate{d}_i+\gamma IN{D}_{it}+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it}.$$

(3)

Among them, $IN{D}_{it}$ is an intermediary variable, and represents industrial transformation and upgrading. The meaning of other variables is the same as Equation (1).

Hypothesis 2 (H2).

China’s low-carbon pilot policy will promote FDI through industrial optimization and upgrading.

Figure 2. The impact path of low-carbon pilot policy on FDI.

Figure 2. The impact path of low-carbon pilot policy on FDI.

3.3. Variable and Data

Many practices indicate that cities are the pioneers and innovators in reducing carbon emissions, improving ecological environment, and promoting low-carbon economic transformation [41,42]. Considering that China’s first batch of low-carbon pilot regions (2010) were mainly provinces, and the third batch of low-carbon pilot projects (2017) were launched late, therefore, this paper takes the second batch of pilot projects (2012) as the research object. After multiple screenings, we finally select 189 cities in 18 provinces in China, of which the treatment group includes 15 low-carbon pilot cities, and the control group includes 174 non-pilot cities. It is worth noting that in order to eliminate the impact of the first and third batches of pilot projects on the research results, our sample does not include any of the cities in the two batches of pilot projects. Figure 3 shows the spatial distribution of the geographic locations of the pilot and non-pilot cities. In addition, we use the consumer price index (CPI) to convert the price data into constant price data with 2011 as the base period [43], and use trend extrapolation to supplement the missing data [44]. The data required for this study mainly come from the China City Statistical Yearbook and the statistical yearbooks of each prefecture-level city.

Dependent variables: FDI is one of the main forms of modern capital internationalization. In order to promote economic development and expand foreign trade, China has always regarded the introduction of foreign investment as an important link and established strategy of opening up. In recent years, FDI has played an important supporting and pulling role in economic structural upgrade, economic system reform, and macroeconomic management in China. Especially in the current economic downturn environment, how to strengthen the attractiveness of FDI is one of the important tasks of the Chinese government. This paper uses the total amount of foreign investment actually utilized to measure FDI. Figure 4 shows the spatial distribution of FDI of 189 cities in 2011, 2014, and 2018.

Independent Variable: Low-carbon pilot policy ($Post\times Treated$). We assume that once a city becomes a low-carbon pilot city, the city will implement stricter environmental policy, and the policy effect will have relatively clear boundaries. That is, the low-carbon measures are only implemented within the pilot city, which eliminates the possibility of policy spillovers. Given that the government announced the second batch of pilot cities at the end of 2012, this paper considers 2013 as the first year of policy implementation. For pilot cities, if $year\ge 2013$, then, $Pos{t}_t\times Treate{d}_i=1$, otherwise $Pos{t}_t\times Treate{d}_i=0$. For non-pilot cities, $Pos{t}_t\times Treate{d}_i=0$.

Control variables: In order to alleviate the interference of the omitted variables on the regression results and improve the accuracy of parameter estimation, it is necessary to add control variables into the model. In recent years, the key determinants of FDI inflow into host countries have aroused extensive discussions in the academic community. The existing research pointed out that the inflow of FDI may be affected by various factors, such as socio-economic factors [45], labor force and human capital [46], international trade [47], financial development [48], political risk [49], and so on. At the same time, Fratzscher (2012) and Marfatia (2016) argued that the determinants of FDI can be divided into two categories, namely, common factors (i.e., global “push” factors) and “pull” factors (i.e., country-specific “pull” factors), which are more obvious for emerging economies [50,51]. Moreover, the foreign push factors can better explain the flow of FDI to the majority of Asia [51]. In general, the determinants of the inflow of FDI have not yet reached a consensus [52]. Therefore, combining the existing research conclusions, our research topics, and data availability, we finally select five indicators as the control variables of this study, namely city size, trade openness, labor cost, human capital, and maturity of the financial market. The specific analysis is as follows.

(1)

City size: City size will affect the economic output efficiency, resource integration and recycling capacity of the city, which is measured by the number of the household registered population at year-end. As city size expands, the agglomeration of advanced production factors will enhance the diversity of the city’s economic structure, market vitality, technological innovation, and urban functions [53]. The resulting scale effect and positive externalities will effectively reduce the production, financing, and transaction costs of foreign-invested enterprises. In addition, for larger cities, centrally control pollution (e.g., treatment of three wastes) and improve the effect of emission reduction become possible [54];

(2)

Trade openness: As an important factor affecting foreign direct investment [55], trade openness usually manifests itself as market opening starting from the commodity market, which is reflected in all aspects of international trade [56]. Given that multinational companies are more willing to invest in countries and regions with a higher degree of openness, increasing trade openness will help attract a large amount of international capital. This paper uses the ratio of the total export–import volume to GDP to measure trade openness;

(3)

Labor cost can be measured by the average wage of employed staff and workers. Many studies have shown that labor cost is an important factor that affects the investment decision-making of foreign-invested enterprises [57]. On the one hand, higher labor cost means that foreign-invested companies will pay higher production costs, thus stimulating capital to flow into countries and regions with lower labor cost [58]. On the other hand, higher labor cost means higher labor productivity, which helps to attract FDI [57]. Therefore, investors will inevitably face a trade-off between cost and efficiency when making investment decisions;

(4)

Human capital. Abundant human capital and the resulting talent advantages help to promote production, improve production efficiency and management efficiency, and reduce enterprise costs. Therefore, foreign-invested enterprises are more inclined to choose regions with rich human capital when investing. We use the total enrollment of regular higher education institutions to measure human capital. It is worth noting that, referring to the definition of Crane and Hartwell (2019), the term of “talent” mentioned in this study is defined as the combined human capital and social capital that an individual possesses [59];

(5)

Maturity of financial market can be measured by the ratio of loans of the national banking system at year-end to GDP. As an investment hub and intermediary, domestic credit provided by the financial sector has become an important financing factor for attracting FDI [60].

In addition to the above analysis, some studies pointed out that economic policy uncertainty may increase corporate risk management, hinder corporate investment, and thus have a negative impact on FDI inflows [61,62]. However, limited by data availability, the discussion of economic policy uncertainty is still mainly focused on the global sample at the national level, which also makes it difficult to comprehensively and accurately measure the economic policy uncertainty of Chinese provinces in different years [63]. Based on this, although we would like to consider the possible impact of economic policy uncertainty, we currently do not have sufficient capacity to implement this idea. Therefore, in this study, we do not discuss the economic policy uncertainty.

Table 1. Variables and measurement.

Predicted Relationship	Symbol	Variable	Measurement
Dependent Variables	LNFDI	Foreign Direct Investment	Total Amount of Foreign
			Investment Actually Utilized
Independent Variable	$Post\times Treated$	Low-carbon City Pilot	Pilot Cities
	LNSIZE	City Size	Household Registered Population at Year-end
Control Variables	OPEN	Trade Openness	Total Export-import Volume/GDP
	LNWAGES	Labor Cost	Average Wage of Employed Staff and Workers
	HUMCAP	Human Capital	Total Enrollment of Regular Higher Education Institutions
	LOANS	Maturity of Financial Market	Loans of National Banking System at Year-end/GDP
Mediating variable	INDUS	Industrial optimization and upgrading	Added Value of the Second Industry/GDP

,

Table 2. Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
LNFDI	1512	11.94433	1.609704	2.892627	15.77754
$Post\times Treated$	1512	0.0595238	0.2366807	0	1
LNSIZE	1512	5.946699	0.645489	3.400197	7.297091
OPEN	1512	14.45804	21.1027	0.0505215	230.3771
LNWAGES	1512	10.74825	0.2549521	9.75314	12.67803
HUMCAP	1512	10.49283	1.413236	0	13.80897
LOANS	1512	89.55015	60.97874	6.585535	782.3959

and

Table 3. VIF value of each variable.

Variable	VIF	1/VIF
Human Capital	1.67	0.600411
City Size	1.47	0.681131
Labor Cost	1.26	0.794691
Maturity of Financial Market	1.19	0.842696
Trade Openness	1.09	0.919785
$Post\times Treated$	1.03	0.973528
Mean VIF	1.28

show the variable settings, descriptive statistics, and the variance inflation factor (VIF) values, respectively. All VIF values less than 10 indicate that there is no multicollinearity between the variables.

4. Empirical Results and Analysis

4.1. Benchmark Regression Results

Firstly, this paper performs model regression on Equation (1), and the results are shown in

Table 4. Results of benchmark regression and intermediary mechanism.

Variable	DID				Tests of Intermediary Mechanism
	FDI				INDUS	FDI
	(1)	(2)	(3)	(4)	(5)	(6)
$Post\times Treated$	0.3776 ** (0.1699)	0.0088 (0.1310)	0.2614 ** (0.1151)	0.2691 ** (0.1139)	1.7002 *** (0.5371)	0.2188 * (0.1145)
INDUS						0.0296 *** (0.0087)
_cons	11.9219 *** (0.0427)	−0.4738 (1.6095)	12.9985 *** (0.1574)	2.2535 (4.3561)	−33.9648 (31.9480)	3.2595 (4.0350)
Control Variable	NO	Control	NO	Control	Control	Control
year-fixed effects	NO	NO	YES	YES	YES	YES
city-fixed effects	NO	NO	YES	YES	YES	YES
R-squared	0.0031	0.3385	0.8287	0.8308	0.9228	0.8334
N	1512	1512	1512	1512	1512	1512

Note: *, **, and *** indicate significance levels of 10%, 5%, and 1%, respectively.

. Columns (3) and (4) control the year fixed effects and the city fixed effects, while columns (1) and (2) do not. Comparing columns (3) and (4), when the control variables are not added, the coefficient of $Post\times Treated$ is significant positive at the 5% level (see column (3)). On this basis, this paper further adds control variables to obtain the results of benchmark regression model (see column (4)). It can be seen that the coefficient of $Post\times Treated$ increases from 0.2614 to 0.2691, and is still significant at the 5% level, which verified the main hypothesis H1 of this paper. That is, the FDI in the policy-affected cities experienced higher growth than that in the unaffected cities. This implies that on average, the implementation of China’s low-carbon pilot policy can effective promote FDI.

In the short term, low-carbon pilot policy may increase the costs of environmental governance of foreign-invested enterprises. However, in the long run, in order to encourage foreign-invested enterprises to invest in low-carbon projects and low-carbon products, the pilot cities have provided foreign-invested enterprises with sufficient funds, talents, technology, financial products, and complete supporting services, which will offset the environmental governance costs of foreign-invested enterprises. On the other hand, in the process of comprehensively promoting the construction of low-carbon cities, pilot cities vigorously guide residents to low-carbon life, low-carbon consumption, low-carbon travel, leading to a low-carbon consumer market with huge potential gradually taking shape. Compared with local traditional high-pollution enterprises, foreign-invested enterprises have more market competitiveness in the production and sales of low-carbon products by virtue of advanced technology, equipment, and management experience. With the increase in expected profit, in order to seize market opportunities, foreign-invested enterprises will have the motivation to increase investment. In general, the implementation of a low-carbon pilot policy can create a better investment environment for foreign-invested enterprises, as well as promising development prospects, which will significantly attract high-quality FDI.

It is worth noting that the coefficient of $Post\times Treated$ in column (4) of Table 4 is only significant at the 5% level, which may be related to the structure of FDI in China. According to the data in the China Statistical Yearbook, FDI in China is mainly concentrated in the secondary and tertiary industries. In the context of China’s rapid urbanization and industrialization, compared with the tertiary industry, the secondary industry has a stronger role in supporting and promoting economic growth. Therefore, when formulating low-carbon pilot policies, local governments may pay more attention to the secondary industry and provide more preferential measures for the optimization and upgrading of the secondary industry. In other words, the low-carbon pilot policy has a relatively weak promotion effect on the tertiary industry, which will weaken the promotion effect of the low-carbon pilot policy on FDI to a certain extent. While, in the long run, with the in-depth implementation of low-carbon pilot policy, its policy effects on FDI will become more significant.

4.2. Tests of Intermediary Mechanism

The above analysis shows that China’s low-carbon pilot policy can indeed promote FDI. Next, we will use Equations (2) and (3) to test the mediation effect of industrial optimization and upgrading, and the results are shown in columns (5) and (6) of Table 4. For column (5), the coefficient of $Post\times Treated$ is significant at the 1% level, which means that a low-carbon pilot policy can significantly promote industrial optimization and upgrading. For column (6), the coefficient of $Post\times Treated$ is significant at the 10% level, and the coefficient of $INDUS$ is significant at the 1% level. This implies that industrial optimization and upgrading can significantly promote FDI. At the same time, comparing the coefficients of $Post\times Treated$ in column (4) (i.e., 0.2691) and column (6) (i.e., 0.2188) of Table 4, we find that when industrial optimization and upgrading is added to the regression model, the coefficient of $Post\times Treated$ decreases, which implies that industrial optimization and upgrading has a partial mediating effect. In other words, the implementation of low-carbon pilot policy can not only directly promote the growth of FDI, but also indirectly attract FDI through industrial optimization and upgrading. Therefore, H2 of this paper has been verified

At present, due to the insufficient leading and supporting role of China’s high-tech industries in economic development, it is difficult to change the heavy-duty characteristics of the industrial structure in a short period of time. Therefore, in order to balance economic and environmental benefits, the pilot cities are committed to accelerating the optimization and upgrading of traditional industries, eliminating outdated industrial production capacity, technologies and equipment, and increase support for low-carbon industries and projects, thus promoting the capital market into a period of “good money driving out bad money”. At the same time, the demonstration role of low-carbon transformation of industrial enterprises and the agglomeration effect of low-carbon industrial parks will also help to further improve the market competitiveness of foreign-invested enterprises. Facing the development opportunities of industrial optimization and upgrading and the competition mechanism of survival of the fittest, foreign-invested enterprises will have the motivation to actively change production decisions, optimize production processes, and improve production efficiency, which will help to gradually form a virtuous circle of attracting high-quality FDI. In general, industrial optimization and upgrading is indeed an important way for a low-carbon pilot policy to promote FDI.

4.3. Robustness Test

4.3.1. Parallel Trend Test

In fact, the use of the DID model needs to satisfy an important prerequisite, namely, the parallel trend assumption. Specifically, before the policy shock, the FDI of the treatment group (i.e., pilot cities) and the control group (i.e., non-pilot cities) should have similar changing trends. Firstly, we perform the parallel trend test on the benchmark regression model (i.e., Equation (1)), as shown in Figure 5a. Two years before the start of the pilot project, the regression coefficients are not significantly different from 0, indicating that the samples passed the parallel trend test. After the policy shock, except for the third year, the regression coefficients show an overall upward trend, which implies that the promotion effect of the low-carbon pilot policy on FDI is increasing year by year. Secondly, for the mediating effect model, since Equation (2) takes industrial transformation and upgrading as the dependent variable, a parallel trend test is also required. Figure 5b indicates that the sample also passed the parallel trend test.

4.3.2. Placebo Test: The Influence of Random Factors

Considering the data limitations, changes of FDI in the benchmark regression model may come from some random factors. Therefore, we need to test whether random factors interfere with the regression results. Drawing on the existing literature [64], this paper assumes that there are random factors ${\vartheta }_{it}$, and the changes of FDI caused by ${\vartheta }_{it}$ can be expressed by Equation (4). Based on this, the estimated value of coefficient of $Post\times Treated$ in Equation (1) can be expressed by Equation (5). For Equation (5), two cases may indicate that the benchmark regression model is robust, that is, $\mathsf{\Phi }=0$ (no random factors) or $\xi =0$ (random factors do not interfere with the regression results). Given that $\Phi$ is difficult to measure, this paper mainly focuses on the case of $\xi =0$. Therefore, this paper randomly changes the policy shock time (i.e., $Pos{t}^{random}$), generates the false policy variable (i.e., $Pos{t}^{random}\times Treated$), and constructs the new regression model (see Equation (6)). In order to ensure that the false policy variable has no theoretical impact on FDI, we conduct 1000 computer simulations for the above random process, that is, ${\beta }_1^{random}=0$. Based on this, we plot the probability distribution of t-values (see Figure 6a). As we expect, the values of ${\widehat{\beta }}_1^{random}$ are distributed around 0, and the average value of 1000 simulations is about 0.0006. Compared with the benchmark regression result of 0.2691, ${\widehat{\beta }}_1^{random}$ can be regarded as 0, that is, ${\widehat{\beta }}_1^{random}=0$. According to Equation (5), we can get $\xi =0$, which implies that random factors do not interfere with the regression results. Therefore, the benchmark regression model is robust.

$$\mathsf{\Phi }=\frac{cov(Post\times Treated,{\vartheta }_{it}|Z)}{var(Post\times Treated|Z)},$$

(4)

$$\widehat{\beta }=\beta +\xi \mathsf{\Phi }\left(Post\times Treated,{\vartheta }_{it}|Z\right),$$

(5)

$$lnFD{I}_{it}={\sigma }_0+{\beta }_1^{random}Pos{t}_t^{random}\times Treate{d}_i+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it}.$$

(6)

4.3.3. Placebo Test: The Influence of Other Policies in the Same Period

Regarding the benchmark regression model, we still have such doubts, namely, does changes of FDI come from other policies in the same period? To this end, this study randomly generates pilot cities, constructs the new treatment group, and establishes the new regression equation (see Equation (7)). Furthermore, in order to ensure that the low-carbon pilot policy will not affect the new treatment group, we repeat the above random process 1000 times and plot the probability distribution of t-values (see Figure 6b). Among them, the values of ${\widehat{\beta }}_2^{random}$ are distributed around 0, and ${\overline{\widehat{\beta }}}_2^{random}=0.0005$. Therefore, compared with the benchmark regression result of 0.2691, we can consider ${\widehat{\beta }}_2^{random}=0$. This shows that other policies in the same period hardly affect the benchmark regression results, that is, the benchmark regression model is robust.

$$lnFD{I}_{it}={\sigma }_0+{\beta }_2^{random}Pos{t}_t\times Treate{d}_i^{random}+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it}.$$

(7)

4.3.4. The Influence of the Dependent Variable Outliers

Specifically, if some cities have better conditions for attracting investment, such as superior geographical location, strong capital, advanced technology, and sufficient talents, then the FDI of these cities may be much higher than other cities. On the contrary, if some cities are technologically backward, lack talents, and have inconvenient transportation, then their FDI may be much lower than other cities. Therefore, in order to eliminate the influence of potential outliers on the regression results, this paper shortens the dependent variable by 1%, and then conducts regression analysis on the 1% to 99% quantiles of FDI. The results are shown in column (1) of

Table 5. Data truncation for robustness test.

FDI	Data Truncation
	(1)	(2)	(3)
$Post\times Treated$	0.2204 * (0.1156)	0.2280 * (0.1293)	0.3081 ** (0.1354)
_cons	5.3422 (3.4856)	5.3648 (3.5469)	5.4201 (3.5473)
Control Variable	Control	Control	Control
year-fixed effects	YES	YES	YES
city-fixed effects	YES	YES	YES
R-squared	0.8572	0.8385	0.8198
N	1481	1422	1346

Note: *, ** indicate significance levels of 10%, 5%, and 1%, respectively.

. We can see that after eliminating the outliers, the coefficient of $Post\times Treated$ is 0.2204, and still reaches the 10% significance level, that is, the low-carbon pilot policy still has a significant positive effect on FDI. In addition, we also shorten the dependent variable by 5% and 10%, respectively, and the regression coefficients of $Post\times Treated$ are still significantly positive at the levels of 10% and 5%. The results are shown in columns (2) and (3) of Table 5. Therefore, the above results indicate that the benchmark regression model is still robust after eliminating the outliers.

5. Heterogeneity Analysis

5.1. Differences in Resource Endowments

China has a vast territory and different regions have different initial resource endowments, which may affect the actual effect of policy implementation. Therefore, we construct the DDD model to discuss the heterogeneity of policy effect brought about by different resource endowments, as shown in Equation (8). Taking into account the actual situation in China, compared with the Eastern and Central regions, the Western region of China is rich in resources, especially land and clean energy (e.g., hydropower, solar energy, wind energy, and geothermal energy). Based on this, for the convenience of analysis, we regard cities in the Western region as resource-rich cities, and the cities in the Eastern and Central regions as resource-poor cities.

$$lnFD{I}_{it}={\alpha }_0+{\beta }_{1}Pos{t}_t\times Treate{d}_i\times Resource+{\beta }_{2}Pos{t}_t\times Treate{d}_i+{\beta }_{3}Pos{t}_t\times Resource+{\beta }_{4}Treate{d}_i\times Resource+{\beta }_{5}Resource+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it}.$$

(8)

Among them, $Resource$ is a dummy variable. When the city is rich in resources, $Resource=1$, otherwise, $Resource=0$. The meaning of other variables is the same as Equation (1).

The regression results are shown in column (1) of

Table 6. Heterogeneity analysis of resource endowments and individual characteristics of government officials.

FDI	Resource Endowment	Individual Characteristics of Government Officials
	Western Region	Gender	Educational	Major
	(1)	(2)	(3)	(4)
$Post\times Treated\times Resource$	0.7793 ***
	(0.2977)
$Post\times Treated\times OFFI{C}_1$		0.6520 **
		(0.2638)
$Post\times Treated\times OFFI{C}_2$			0.4143 *
			(0.2343)
$Post\times Treated\times OFFI{C}_3$				0.3995 ***
				(0.1412)
_cons	1.4448	2.6832	3.1901	5.6229
	(4.3472)	(4.3054)	(3.2239)	(3.5542)
Control Variable	Control	Control	Control	Control
year-fixed effects	YES	YES	YES	YES
city-fixed effects	YES	YES	YES	YES
R-squared	0.8346	0.8317	0.8311	0.8213
N	1512	1512	1512	1512

Note: *, **, and *** indicate significance levels of 10%, 5%, and 1%, respectively.

. The coefficient of $Post\times Treated\times Resource$ is 0.7793, and is significant at the 1% level, which indicates that in the process of low-carbon city construction, a low-carbon pilot policy implemented by resource-rich cities have a stronger promotion effect on FDI. With the implementation of a low-carbon pilot policy, the resource-rich cities, that is, the cities in the Western region in this study, will actively undertake the transfer of production capacity in the Eastern and Central regions, continuously improve the integration and deep processing capabilities of resources, and accelerate the industrial chain and value chain upgrade to the middle and high end, thus improving the use efficiency and output efficiency of resource, and gradually transforming resource advantages into actual productivity. In order to gain pioneering advantage and seize market share, foreign-invested enterprises will be more willing to make large-scale investments in resource-rich cities. Therefore, in the long run, the advantage of resource endowment of the city will help to strengthen the promotion effect of low-carbon pilot policy on FDI.

5.2. Differences in Individual Characteristics of Government Officials

Government leaders play an important role in formulating and implementing low-carbon pilot policy. In the final analysis, urban governance and policy implementation mainly rely on the “rule by man” by government officials [17]. Therefore, this paper will further analyze the impact of the individual characteristics of government officials, namely, gender, educational background, and major, on the policy effect of low-carbon pilot policy. Similar to Section 5.1, we construct the DDD model, including the individual characteristics of government officials (see Equation (9)), and the regression results are shown in Table 6.

$$lnFD{I}_{it}={\alpha }_0+{\beta }_{1}Pos{t}_t\times Treate{d}_i\times OFFI{C}_j+{\beta }_{2}Pos{t}_t\times Treate{d}_i+{\beta }_{3}Pos{t}_t\times OFFI{C}_j+{\beta }_{4}Treate{d}_i\times OFFI{C}_j+{\beta }_{5}OFFI{C}_j+\sum \lambda Z_{it}+{\nu }_t+{\mu }_i+{\epsilon }_{it}.$$

(9)

Among them, OFFIC is a dummy variable, representing the mayor’s gender, educational background, and major, respectively, and $j=1, 2, 3$. Specifically, when the mayor of a city is a female, $OFFI{C}_1=1$, otherwise, $OFFI{C}_1=0$. When the mayor obtains a graduate degree, $OFFI{C}_2=1$, otherwise, $OFFI{C}_2=0$. When the mayor majors in non-economics, $OFFI{C}_3=1$, otherwise, $OFFI{C}_3=1$. The meaning of other variables is the same as Equation (1).

(1)

Gender of the mayor: In column (2) of Table 6, the coefficient of $Post\times Treated\times OFFI{C}_1$ is 0.6520, and is significant at the 5% level. This implies that compared with cities where the mayor is a male, the policy effect of a female-administered city is greater. It is a very interesting finding. According to existing research, we suppose that it may be because, on average, female leaders may be more assertive and better at managing resources and public goods [65], thus achieving better results in maintaining social stability and improving social outcomes [66]. In addition, compared with male leaders, female leaders may be more inclined to democratic and participatory leadership styles, which makes them have the natural advantage of becoming charismatic leaders. During the implementation of the low-carbon pilot policy, by shaping the image of exceptional competence, female leaders will more easily mobilize the initiative of policy executors [67], and more effectively improve the management efficiency and organizational efficiency of the local government. Therefore, with the efficient and continuous release of the low-carbon policy’ effectiveness, the investment motivation of foreign-invested enterprises will be further stimulated;

(2)

Educational background of the mayor: Observing column (3) of Table 6, when the mayor obtains a master’s degree or a doctor’s degree, the coefficient of $Post\times Treated\times OFFI{C}_2$ is 0.4143, and is significantly positive at the 10% level. This indicates that compared with cities where mayors obtain bachelor’s degrees, mayors who obtain graduate degrees will more effectively play the promotion effect of low-carbon pilot policy on FDI. Environmental governance is a long-term, complex, and dynamic system engineering, which involves the entire process management of pre-prevention, mid-term supervision, and post-governance, and is closely related to economic development, social harmony, and residents’ health. Therefore, in the context of low-carbon city construction, after long-term and more systematic training and learning, leaders can make better use of environmental policy instruments (e.g., finance, taxation, and subsidies), fully release policy dividends, and continuously stimulate the market vitality, thus creating a more attractive investment environment;

(3)

Mayor’s major: For column (4) of Table 6, the coefficient of $Post\times Treated\times OFFI{C}_3$ is significantly positive at the 1% level, indicating that in cities where mayors majored in non-economics, a low-carbon pilot policy can promote FDI more significantly. As mentioned in previous studies, during the period in power, mayors who majored in economics may pay more attention to short-term economic benefits [68] and prefer to achieve regional economic growth goals by using direct policy measures. Therefore, the low-carbon pilot policy, as an environmental regulation that indirectly promotes economic growth, may be ignored to a certain extent. However, for mayors who majored in non-economics, how to obtain sustainable economic benefits without destroying the ecological environment has been put on the agenda. Under the background of accelerating the construction of low-carbon cities, pilot cities can actively strive for national and provincial funds, and vigorously attract FDI through strong preferential measures, so as to make up for environmental governance costs and pollution control costs.

6. Conclusions and Policy Recommendations

This paper regards China’s low-carbon pilot policy as a quasi-natural experiment, and employs the DID model to analyze the actual impact and intermediary mechanism of low-carbon pilot policy on FDI. The results indicate that China’s low-carbon pilot policy can significantly promote FDI, and industrial optimization and upgrading is an important way. Based on this, we construct the DDD model to discuss the heterogeneity of the policy effect of low-carbon pilot policy. Firstly, we find that resource endowment helps to enhance the promotion effect of low-carbon pilot policy on FDI. Secondly, considering that the effect of policy implementation greatly depends on the “rule by man” by government officials, this study further explores the heterogeneity of policy effect that may be brought about by the personal characteristics of government officials (i.e., gender, educational background, and major). We find that when the mayor of the pilot city is a female, or obtains a master’s degree or a doctorate degree, or majored in non-economics, respectively, the promotion effect of a low-carbon pilot policy will be more obvious. In addition, a series of robustness tests are carried out to verify the reliability of the research conclusions.

This study has obtained some interesting and meaningful conclusions, while there are some limitations. Limited by the availability of data, this paper does not consider the global network of economic policy uncertainty. In fact, controlling for some measure of policy certainty is valuable for studying the relationship between low-carbon pilot policy and FDI. Therefore, in the future research, we will try to explore the actual impact of economic policy uncertainty on FDI in the context of low-carbon city construction. In addition, as China’s low-carbon pilot project is still at the early stage, in order to further explore the implementation effect of the low-carbon pilot policy, we also consider studying the possible environmental, economic, and social effects caused by low-carbon pilot policy from the perspective of policy instrument, which may provide multidimensional policy recommendations for governments to achieve the win-win situation between economic growth and low-carbon transformation.

According to the findings of this study, we propose the following policy recommendations.

(1)

Under the special background of increasing downward pressure of economy, local governments should give full play to the leverage of low-carbon pilot policy, comprehensively use environmental policy tools (e.g., finance, taxation, trade, government procurement), and encourage foreign-invested enterprises to invest in low-carbon projects and participate in the development and utilization of clean energy, low-carbon technologies, and low-carbon products. At the same time, local governments should accelerate the construction of green industrial parks and closed loops of the entire industry chain, vigorously promote the transformation and upgrading of traditional industries, and make full use of cluster advantages and scale effect to create new growth poles for foreign-invested enterprises;

(2)

Pilot cities, especially those with good resource endowments, should seize the opportunity of low-carbon city construction, use resource advantages to develop advantageous industries, promote the construction of key emission reduction projects, and encourage foreign-invested enterprises to open branches and build large-scale factories. It is worth noting that urban environmental governance and economic development are highly dependent on government officials. Therefore, it is necessary to strengthen the training of the working ability of government officials, encourage government officials to exchange experience and continue their further studies, and continuously improve their urban governance ability and comprehensive quality. In addition, the important role of female leaders in the construction of low-carbon cities also needs more attention and discussion.