Study on Relationship between Environmental Regulation and Green Total Factor Productivity from the Perspective of FDI—Evidence from China

Abstract

The existing literature has paid less attention to the key role of FDI in the realization of Porter’s hypothesis. Based on China’s provincial panel data from 2007 to 2020, this paper uses the SBM function and the Luenberger productivity index to measure the green total factor productivity (GTFP) of China’s provinces while considering energy consumption and undesired output. Using FDI as the mediating variable and threshold variable, this paper studies the relationship between environmental regulation (ENV), FDI, and GTFP. The research results show that: (1) There is a U-shaped relationship between ENV and GTFP. China is currently on the left side of the U-shaped inflection point. Further improving the intensity of ENV can promote the growth of GTFP. (2) The mediating effect of FDI is significantly established. ENV and FDI show a significant U-shaped relationship. Further development of FDI can promote GTFP. (3) FDI has a threshold effect. When FDI is at a low level of development, ENV and GTFP are negatively correlated and insignificant; when FDI is at a high level of development, ENV can significantly promote the growth of GTFP. The marginal effect on green TFP increases with FDI growth. The results of this paper show that FDI is a necessary prerequisite for the Porter hypothesis to become a reality in China. The research of this paper has important reference value for realizing the sustainable development of environment, economy, and society.

1. Introduction

China has made great achievements in economic development through more than 40 years of reform and opening up. However, China’s economic growth mainly relies on large-scale government-led investment and a large number of factor inputs. Some scholars believe that the construction of an environmental governance system should focus on the significant differences between the private economy and the state-owned economy. Under the dual pressures of green development and economic transformation, FDI may face more difficulties, and the impact of ENV on FDI cannot be ignored.

Neoclassical economics believes that ENV will increase the production cost of enterprises and reduce the competitiveness of enterprises, thereby offsetting the positive effects of environmental protection on society and negative effects on economic growth. However, Porter challenged it in 1991 and proposed the Porter hypothesis [1]. The hypothesis holds that ENV in the true sense will not increase the cost of enterprises but can trigger innovation, generate net income, and then improve the competitive advantage of enterprises. In 1995, Porter and Linde [2] perfected this hypothesis and further explained the mechanism that environmental protection enhances competitiveness through innovation. Through the method of case study, they pointed out that the relationship between ENV and economic development cannot simply be divided into two opposites. They believe that stricter but well-designed ENV (especially market-based environmental taxes, pollution emission permit trading mechanisms, etc.) can stimulate innovation and improve product quality, thereby partially or even fully offsetting the costs of complying with environmental regulations and making manufacturers more competitive in the international market. The Porter hypothesis provides a new perspective for us to re-understand the relationship between environmental quality and economic development.

In 2021, China’s actual use of FDI will reach US$173.48 billion. FDI is an important carrier for China to be deeply embedded in the global value chain system, and it is also an important link for foreign high-quality elements to promote China’s sustainable development. The massive influx of FDI has brought great opportunities for China’s industrial upgrading. Therefore, it is worth paying attention to how to make better use of FDI and actively guide FDI to be invested in key areas that promote high-quality economic development and green development in China.

The productivity effect and environmental effect of FDI has always been a hot research topic. In the early stage, the main focus was on the impact of FDI and its spillover effects on total factor productivity and the environment. In recent years, with the attention of various countries on ENV issues, the research literature on ENV on GTFP has gradually increased, and, at the same time, attention has been paid to the impact mechanism of FDI on green development. In terms of the impact of FDI and its technology spillover effects on productivity, some studies believe that FDI provides rich capital, advanced management experience, and an efficient production chain for the economic development of the host country, which will promote enterprises to invest in R&D and carry out technological innovation activities. Thereby, it can improve the total factor productivity of the host country [3,4,5,6]. Some studies believe that the advanced technology brought by FDI can improve the utilization level of energy and resources and the cleanliness of production through the demonstration effect and correlation effect [7,8]. Some studies believe that through the competitive effect, FDI can stimulate the host country’s enterprises to improve the level of clean technology innovation and reduce energy consumption and pollutant emissions [9,10,11,12]. Some studies also believe that FDI can promote the GTFP of the host country [13,14,15,16]. However, other studies support the polluted paradise hypothesis. Scholars such as Rafindadi pointed out that in order to develop the economy, developing countries have lowered the requirements for environmental protection, thus attracting developed countries to transfer high-polluting industries to these developing countries, aggravating the environmental pollution of the host countries, and not being conducive to the improvement of GTFP [17,18,19]. Scholars such as Kukulski believe that the impact of FDI on the GTFP of the host country is affected by the level of ENV, level of financial development, level of investment, and heterogeneity of innovation [20,21,22,23,24]. Yuan Yijun pointed out that only the combined effect of FDI and ENV can promote the improvement of GTFP [25].

Although some studies have verified the impact of ENV on GTFP, few studies have mentioned the moderating role of FDI in the relationship between ENV and GTFP, which is an important issue for China and other developing countries to achieve sustainable development. Is there still a U-shaped relationship in the time dimension between ENV and GTFP? However, this paper does not want to simply verify whether there is a U-shaped relationship. Further, we believe that FDI has a certain moderating effect.

To sum up, scholars have conducted various discussions on the direct impact of ENV→GTFP. However, the question of how FDI affects the realization of Porter’s hypothesis is less involved and lacks in-depth research. Current research by other scholars also lacks an analysis of the mediating effect on the relationship between ENV and GTFP from the perspective of FDI. Therefore, this paper not only verifies the nonlinear relationship between ENV and GTFP but also analyzes the mediating effect of FDI in the relationship between ENV and GTFP. In this way, it can be verified whether the influence mechanism of ENV→FDI→GTFP is established. On this basis, this paper further analyzes the threshold effect of FDI in the relationship between ENV and GTFP. This paper fully considers the moderating effect and threshold effect of FDI in the relationship between ENV and GTFP, thereby helping us to better understand the actual role of FDI in ENV and GTFP. Figure 1 briefly explains the research objectives and research logic of this paper. This paper has a certain reference value for verifying whether the Porter hypothesis is established in China.

2. Theoretical Analysis and Research Hypothesis

ENV is an important part of social regulation. The areas of ENV mainly include air pollution, water pollution, use of toxic substances, hazardous waste disposal, and noise pollution. Michael Porter pointed out that an appropriate ENV intensity can stimulate technological innovation. Existing studies mainly focus on verifying the impact of ENV on productivity and production costs [26,27,28,29,30,31]; or focus on verifying the impact of ENV on overall innovation (i.e., the synthesis of production technology innovation and pollution control technology innovation) and pollution control technology innovation [32,33,34,35]; or verify the impact of ENV on GTFP [36,37,38]. Among them, in the literature on the impact of ENV on GTFP, although the final results will vary depending on the selection of specific indicators, most research results believe that ENV will have a positive impact on GTFP.

ENV means that the government formulates environmental protection laws, policies, and guidelines, and intervenes and restrains the production activities of enterprises through coercive public power to achieve sustainable development. ENV can stimulate technological innovation to a certain extent, thereby promoting GTFP. When ENV is at a low level of intensity, companies often respond by paying pollution discharge fees or purchasing environmental protection treatment equipment, which does not improve the ability of independent innovation and hinders the promotion of GTFP. With the increasing intensity of ENV, the existing technical level of enterprises cannot meet the requirements of ENV, which forces enterprises to innovate in technology and adjust their operation and management models, thereby promoting the improvement of GTFP.

In fact, the impact of ENV on GTFP is the result of a comprehensive comparison of positive and negative impacts. The impact of positive and negative effects is not synchronized, and negative effects tend to have an impact in the current period. It takes a relatively long time to improve GTFP itself, which determines that the innovation compensation effect of ENV lags behind the negative effect of compliance cost. Thus, in the short run, ENV reduces GTFP. However, in the long run, the innovation compensation effect offsets the negative effect of compliance costs, and ENV can improve GTFP. Therefore, in the temporal dimension, there may be a U-shaped relationship between ENV and GTFP.

Based on the above analysis, this paper proposes Hypothesis 1.

Hypothesis 1 (H1).

There is a U-shaped relationship between ENV and GTFP.

Some studies have identified FDI as an important factor affecting the environment and GTFP. Existing studies have not come to a consistent conclusion on this issue. Some studies have concluded that the impact of FDI on the environment or GTFP is negative, summarizing the view that FDI causes environmental degradation in host countries as the pollution paradise hypothesis. Considering factors such as saving environmental governance costs, multinational companies tend to transfer pollution-intensive industries or production links to countries with low levels of ENV (mainly developing countries), thereby aggravating the environmental pollution of host countries [39,40,41].

However, other studies have concluded that FDI helps to improve the environmental quality of the host country. FDI will demonstrate and disseminate clean production technology and management concepts in the host country and promote the improvement of the host country’s environmental quality. Since foreign companies have a better environmental performance than local companies, they will bring about a pollution halo effect that improves the host country’s environment [42,43,44,45,46,47,48,49,50,51].

Since China’s reform and opening up, its economic structure has undergone a transformation from domestic demand-dependent to export-oriented. Lin Boqiang and Zou Chuyuan called the development model of the international economy transmitted to China through international trade as the World-China model. This process not only boosted China’s rapid economic growth but also resulted in serious trade-induced environmental pollution. China’s current ENV policy is formulated in response to the increasingly serious environmental pollution in the economic development process of World-China, which means that there is a correlation effect between FDI inflow and the formulation of China’s ENV. Inspired by the above studies, this paper argues that FDI has a moderating effect on the relationship between ENV and GTFP.

This paper believes that the positive moderating effect of FDI on the relationship between ENV and GTFP is mainly reflected in the following aspects:

(1) After learning from the ENV of developed countries, local governments can effectively correct their rigid and inefficient environmental policies, and constantly try to use more appropriate policies and management methods to solve environmental problems and promote the improvement of GTFP.

(2) The inflow of FDI funds will help increase the host country’s investment in environmental protection and technological innovation, improve the ability of green technology innovation, and then promote the improvement of GTFP.

(3) FDI inflow can have a positive impact on the growth of GTFP of the host country through the competition effect, factor reconfiguration effect, demonstration effect, etc., resulting in a pollution halo effect.

(4) FDI inflow can bring advanced production technology to the host country, improve the utilization rate of clean technology and energy efficiency in the production process of the host country, improve the pollution control level of enterprises and regions in the host country, reduce the emission of undesired output, and promote green development, thereby increasing GTFP.

Based on the above analysis, this paper proposes Hypothesis 2:

Hypothesis 2 (H2).

FDI can play a positive moderating effect on the relationship between ENV and GTFP.

3. Materials and Methods

3.1. Empirical Model Design

3.1.1. Model Construction of the Direct Effect and Mediation Effect

Theoretically, the strategic choice of enterprises to deal with ENV policies mainly depends on the cost-benefit analysis. When the intensity of ENV is low, pollution fines are not enough to attract the attention of enterprises, and the cost and risk of green technology research and development are huge, and enterprises will choose terminal treatment or pay pollution fines directly. The resource crowding effect is not conducive to the improvement of GTFP. When the intensity of environmental rules is high, the innovation compensation effect can compensate or even exceed the resource crowding effect, and ultimately promote the improvement of GTFP. The quadratic term of the intensity of ENV should be added to the measurement equation to reflect this positive or negative impact.

This paper refers to scholars such as Jefferson [52] and Zhang Cheng [53], who added the quadratic term of explanatory variables to the model to test whether there is a U-shaped relationship. The model is constructed as follows:

$${\mathrm{GTFP}}_{\mathrm{it}}{=\mathsf{\beta }}_0{+\mathsf{\beta }}_1{\mathrm{ENV}}_{\mathrm{it}}{+\mathsf{\beta }}_2{\mathrm{ENV}}_{\mathrm{it}}{}^2{+\mathsf{\beta }}_3{\mathrm{GOV}}_{\mathrm{it}}{+\mathsf{\beta }}_4{\mathrm{EDU}}_{\mathrm{it}}{+\mathsf{\beta }}_5{\mathrm{RD}}_{\mathrm{it}}{+\mathsf{\beta }}_6{\mathrm{INF}}_{\mathrm{it}}{+\mathrm{u}}_{\mathrm{i}}{+\mathsf{\epsilon }}_{\mathrm{it}}$$

(1)

Among them, GOV represents fiscal expenditure. EDU represents human capital. RD represents the level of research and development expenditure, and INF represents the level of infrastructure construction. This paper will further test the indirect impact of ENV on GTFP through FDI. Referring to the research methods of Wen Zhonglin [54], the mediating effect model is set as follows:

$${\mathrm{GTFP}}_{\mathrm{it}}{=\mathsf{\beta }}_0{+\mathsf{\beta }}_1{\mathrm{ENV}}_{\mathrm{it}}{+\mathsf{\beta }}_2{\mathrm{ENV}}_{\mathrm{it}}{}^2{+\mathsf{\beta }}_3{\mathrm{GOV}}_{\mathrm{it}}{+\mathsf{\beta }}_4{\mathrm{EDU}}_{\mathrm{it}}{+\mathsf{\beta }}_5{\mathrm{RD}}_{\mathrm{it}}{+\mathsf{\beta }}_6{\mathrm{INF}}_{\mathrm{it}}{+\mathrm{u}}_{\mathrm{i}}{+\mathsf{\epsilon }}_{\mathrm{it}}$$

(2)

$${\mathrm{FDI}}_{\mathrm{it}}{=\mathsf{\alpha }}_0{+\mathsf{\alpha }}_1{\mathrm{ENV}}_{\mathrm{it}}{+\mathsf{\alpha }}_2{\mathrm{ENV}}_{\mathrm{it}}{}^2{+\mathsf{\alpha }}_3{\mathrm{OP}}_{\mathrm{it}}{+\mathsf{\alpha }}_4{\mathrm{URB}}_{\mathrm{it}}{+\mathsf{\alpha }}_5{\mathrm{IND}}_{\mathrm{it}}{+\mathrm{u}}_{\mathrm{i}}{+\mathsf{\epsilon }}_{\mathrm{it}}$$

(3)

$${\mathrm{GTFP}}_{\mathrm{it}}{=\mathsf{\gamma }}_0{+\mathsf{\gamma }}_1{\mathrm{ENV}}_{\mathrm{it}}{+\mathsf{\gamma }}_2{\mathrm{ENV}}_{\mathrm{it}}{}^2{+\mathsf{\gamma }}_3{\mathrm{FDI}}_{\mathrm{it}}{+\mathsf{\gamma }}_4{\mathrm{GOV}}_{\mathrm{it}}{+\mathsf{\gamma }}_5{\mathrm{EDU}}_{\mathrm{it}}{+\mathsf{\gamma }}_6{\mathrm{RD}}_{\mathrm{it}}{+\mathsf{\gamma }}_7{\mathrm{INF}}_{\mathrm{it}}{+\mathrm{u}}_{\mathrm{i}}{+\mathsf{\epsilon }}_{\mathrm{it}}$$

(4)

Referring to the research methods of Wen Zhonglin [54], if the coefficients β₁, β₂, α₁, and α₂ of ENV in the equation are all significant, and the coefficient γ₃ of FDI is also significant, it means that the mediating effect is significantly established.

3.1.2. Threshold Regression Model

This paper refers to the panel threshold regression model proposed by Hansen [55]. Its advantage is that on the one hand, it can estimate the threshold value, and at the same time, it can also perform a significance test on the endogenous threshold effect. The idea is to incorporate a certain threshold value into the regression model as an unknown variable, construct a piecewise function, and conduct empirical estimation and testing of the threshold effect and the corresponding threshold value. This paper selects FDI as the threshold variable. Assuming that there is a double threshold, the threshold effect model is set as follows:

$${\mathrm{GTFP}}_{\mathrm{it}}{=\mathsf{\eta }}_0{+\mathsf{\eta }}_1{\mathrm{ENV}}_{\mathrm{it}}·\mathrm{I}\left({\mathrm{FDI}<\mathsf{\theta }}_1\right){+\mathsf{\eta }}_2{\mathrm{ENV}}_{\mathrm{it}}·\mathrm{I}\left({\mathsf{\theta }}_1{<\mathrm{FDI}<\mathsf{\theta }}_2\right){+\mathsf{\eta }}_3{\mathrm{ENV}}_{\mathrm{it}}·\mathrm{I}\left({\mathrm{FDI}>\mathsf{\theta }}_2\right){+\mathsf{\eta }}_4{\mathrm{X}}_{\mathrm{it}}+\mathsf{\epsilon }$$

(5)

Among them, GTFP represents the GTFP. ENV represents the environmental rule. I() represents the indicative function of the threshold. X_it represents the control variables of the GTFP, and the selection of the control variables of the GTFP are the same as above.

Equation (1) is used to verify the direct effect between ENV and GTFP. Equations (2)–(4) are used to verify the mediating effect of FDI in the relationship between ENV and GTFP. Equation (5) is to verify the threshold effect of FDI in the relationship between ENV and GTFP. Their purpose and relationship are shown in Figure 2.

3.2. Variable Selection

3.2.1. Explained Variable: GTFP

The three main calculation methods of GTFP include the Solow residual value method, stochastic frontier analysis method, and data envelopment analysis method. The data envelopment analysis method has been used by many researchers because it does not require a functional form and is simple and easy to implement. More importantly, the non-parametric directional distance function in the data envelopment analysis method can consider both good outcomes (economic development) and bad outcomes (environmental pollution). It embodies the essence of green development of reducing damage to the natural environment or improving the condition of natural assets while maintaining sustained economic growth. Therefore, this paper uses the data envelopment analysis method to measure the GTFP. In the data envelopment analysis method, the SBM directional distance function overcomes the defect of the inaccurate estimation caused by over-input or under-output. Therefore, this paper chooses the non-radial and non-oriented SBM directional distance function and the Luenberger productivity index to measure the GTFP of Chinese provinces [56].

The relevant original data are explained as follows: First, the input factors include labor, capital, and energy input, and the data are from China Statistical Yearbook, China Labor Statistical Yearbook, and China Energy Statistical Yearbook. The labor input is based on the year-end number of urban employed persons in each province as an indicator. The capital investment is calculated by the perpetual inventory method. The depreciation rate refers to the method of Wu Yanrui, and the capital investment estimated by Zhang Jun et al. is used as the capital investment of the year [57,58]. The energy input is based on the total energy consumption of each province. Second, the expected output is based on the provincial GDP at constant prices in 2007, and the data comes from the China Statistical Yearbook. Third, undesired outputs include waste water, waste gas, and solid waste, which are, respectively, indexed by the industrial waste water discharge, industrial waste gas discharge, and industrial solid waste generation in each province. The relevant data are from the China Environmental Yearbook.

3.2.2. Core Explanatory Variables: ENV

Appropriate ENV can promote product innovation and process innovation and help to promote GTFP. However, if the intensity of ENV is too high, it will crowd out production resources and inhibit the improvement of GTFP. The mainstream methods of measuring ENV in the existing literature can be divided into three categories: the first type uses the cost of abatement or investment in pollution control to represent the degree of strictness of ENV, such as Ederington et al. and Levinson and Zhang Cheng et al. [59]. The second category uses pollution reductions, emissions, or emissions intensity to measure ENV, such as Cole and Elliott, Fu Jingyan, and Zhou Hao [60,61]. The third category measures the strictness of ENV by constructing a comprehensive index, such as Zhao Xikang, Fu Jingyan, and Li Lisha [62,63]. The first method measures ENV from the perspective of cost, which will be affected by the environmental governance capacity and technical level of a country (region) itself and cannot truly reflect the strictness of ENV. The second type of approach measures ENV from an effect perspective but does not guarantee that these effects are entirely derived from ENV rather than other policies [64]. Relatively speaking, the comprehensive index method can more accurately and comprehensively reflect the strictness of ENV. Therefore, referring to the method of Zhao Xikang [62], this paper constructs an ENV index system to measure the ENV of each province in China.

Firstly, an ENV comprehensive evaluation system is built. Combined with the availability of data, this paper constructs a comprehensive ENV evaluation index system from three levels of environmental protection status, process, and results. See

Table 1. ENV comprehensive evaluation index system.

Evaluation Object	Basic Indicators	Specific Indicators
Environmental Protection Status	Number of regulations	The total number of local regulations and administrative regulations
	Number of environmental protection organizations	Number of institutions of regional environmental protection system at the end of the year
	Environmental investment	The proportion of industrial pollution control investment in industrial added value
Environmental protection process	Water pollution treatment rate	COD removal rate in industrial wastewater
	Air pollution treatment rate	Industrial SO2 removal rate
	Solid waste treatment rate	Comprehensive utilization rate of industrial solid waste
Environmental results	Wastewater discharge intensity	COD emitted per unit of industrial added value
	Exhaust emission intensity	SO2 emitted per unit of industrial added value
	Solid waste emission intensity	Solid waste discharged per unit of industrial added value

for details.

Secondly, the ENV comprehensive evaluation index is constructed. For positive indicators, that is, the indicators are positively correlated with the measured degree, including the number of regulations, number of environmental protection agencies, investment in environmental protection, rate of water pollution treatment, rate of air pollution treatment, and rate of solid waste treatment. The standardized calculation method is as follows:

$${\mathrm{V}}_{\mathrm{it}}{}^{\mathrm{s}}=\frac{{\mathrm{V}}_{\mathrm{i}\left(\mathrm{t}\right)}-{\mathrm{V}}_{\min \left(0\right)}}{{\mathrm{V}}_{\max \left(0\right)}-{\mathrm{V}}_{\min \left(0\right)}}$$

(6)

For the reverse indicators, that is, the indicators are negatively correlated with the measured degree, including the wastewater discharge intensity, exhaust emission intensity, and solid waste discharge intensity. The standardized calculation method is as follows:

$${\mathrm{V}}_{\mathrm{it}}{}^{\mathrm{s}}=\frac{{\mathrm{V}}_{\max }-{\mathrm{V}}_{\mathrm{i}\left(\mathrm{t}\right)}}{{\mathrm{V}}_{\max \left(0\right)}-{\mathrm{V}}_{\min \left(0\right)}}$$

(7)

Among them, V_it ^s represents the standardized value of the ith index in the t year. V_i represents the original data of the ith index in the t year in a certain region. V_min(0) is the smallest value among the raw data of the i-th indicator in all provincial base years (2007) and V_max(0) is the largest value.

Finally, the ENV comprehensive evaluation index can be obtained by weighting and summarizing the standardized basic indexes. Considering that there is little difference in the pollution emissions of different provinces in China, this paper adopts the equal weighted linear method to construct the ENV comprehensive evaluation index. The calculation formula is:

$${\mathrm{ER}}_{\mathrm{t}}=\frac{1}{3}\left(\frac{1}{{\mathrm{N}}^1}{\displaystyle \sum _{{\mathrm{N}}^1=1}^{{\mathrm{n}}^1}{\mathrm{V}}_{\mathrm{it}}{}^{\mathrm{s}}+\frac{1}{{\mathrm{N}}^2}{\displaystyle \sum _{{\mathrm{N}}^2=1}^{{\mathrm{n}}^2}{\mathrm{V}}_{\mathrm{it}}{}^{\mathrm{s}}+\frac{1}{{\mathrm{N}}^3}{\displaystyle \sum _{{\mathrm{N}}^3=1}^{{\mathrm{n}}^3}{\mathrm{V}}_{\mathrm{it}}{}^{\mathrm{s}}}}}\right)=\frac{1}{\mathrm{N}}{\displaystyle \sum _{\mathrm{N}=1}^{\mathrm{n}}{\mathrm{V}}_{\mathrm{it}}{}^{\mathrm{s}}}$$

(8)

Among them, N represents the total number of all basic indicators. N¹, N², and N³ represent the total number of basic indicators of the environmental protection status, environmental protection process, and environmental protection structure, respectively. N = N¹ + N² + N³ = n.

3.2.3. Mediating Variables: FDI

From the above analysis, it can be seen that FDI may play a role in promoting the improvement of GTFP. This paper uses the proportion of foreign direct investment in GDP to measure FDI.

3.2.4. Control Variables

When examining the direct effect of FDI on GTFP, the following control variables are selected: fiscal expenditure (GOV), which provides necessary public goods and services through focused financial support in key areas such as innovation, education, and environmental protection, to promote economic transformation and promote the improvement of total factor productivity. This paper uses the proportion of local fiscal expenditure to GDP to measure GOV. Human capital (EDU) reflects the accumulation of knowledge and skills and is the basis for technological innovation and high-quality development, contributing to increased productivity.

In this paper, EDU is represented by the average education level of employees in the region. Referring to the research method of Wu Yanrui [57], the calculation method of EDU is as follows: EDU_it = (6PS_it + 9JS_it + 12SS_it + 15SC_it + 16HE_it + 19PG_it)/P_it, where PS_it, JS_it, SS_it, SC_it, HE_it, and PG_it are the number of employed persons with an education level of primary school, junior high school, high school, college junior college, undergraduate, postgraduate, and above in i province, respectively. P_it is the total number of employed persons. The data comes from the China Population and Employment Statistical Yearbook. R&D investment (RD) is measured by the proportion of RD expenditure in GDP. The level of infrastructure development (INF) is expressed as the geometric mean of the length of road transport lines per square kilometer of land and the length of railway transport lines per square kilometer of land.

When conducting the mediation effect test, in addition to the above control variables, the following control variables are also selected: trade openness (OP), which is expressed by the ratio of total import and export to regional GDP; and the urbanization level (URB), which is measured by the proportion of urban population to the total local population. The industrial structure (IND) is measured by the proportion of the output value of the tertiary industry in GDP. The description of the variables in this paper is shown in

Table 2. The description of the variables.

Variable Name	Symbol	Calculation Method
Green total factor productivity	GTFP	SBM-Luenberger Index
Environmental regulation	ENV	ENV Index
Foreign direct investment	FDI	FDI/GDP
Government spending	GOV	Local fiscal expenditure/GDP
Human capital	EDU	Years of education per capita
R&D level	RD	R&D spending/GDP
Infrastructure construction level	INF	The geometric mean of the length of road transport lines per square kilometer of land and the length of railway transport lines per square kilometer of land
Trade openness	OP	Total import and export/GDP
Urbanization level	URB	Proportion of urban population
Industrial structure	IND	The proportion of the tertiary industry

.

3.3. Data Source

This paper uses the panel data of 30 provinces in China from 2007 to 2020 as the sample for the empirical study. The data used are compiled and calculated according to the China Statistical Yearbook, China Industrial Economic Statistical Yearbook, China Environmental Yearbook, and various local statistical yearbooks in China. Based on the availability of data, the sample excludes Tibet, Hong Kong, Macau, and Taiwan.

4. Empirical Results

4.1. Direct Effect Analysis

In order to ensure the reliability of the research results, this paper uses the LLC method, HT method, and IPS method to perform unit root tests on the relevant variables (

Table 3. Unit root test.

Method	GTFP	ENV	GOV	EDU	RD	INF
LLC	−12.315 ***	−8.389 ***	−9.027 ***	−6.382 ***	−10.652 ***	−8.267 ***
HT	−13.357 ***	−3.865 ***	−9.256 ***	−6.234 ***	−10.217 ***	−5.672 ***
IPS	−3.578 ***	−8.597 ***	−7.384 ***	−5.867 ***	−8.692 ***	−3.479 ***

Note: *** means p < 0.01.

). After testing, all variables are stationary series. In order to test the multicollinearity among the variables, this paper tests the VIF and Pearson correlation coefficients of the variables in the model, and it is found that there is no multicollinearity problem between the variables.

In this paper, the Wald test, Woodridge test, and Pesaran test show that the p value strongly rejects the null hypothesis. The between-group heteroskedasticity, within-group autocorrelation, and spatial correlation must be handled in the regression process. In this paper, comprehensive FGLS is used for estimation, and the regression results are shown in

Table 4. Regression results of the direct effect of ENV on GTFP.

KERRYPNX	GTFP
	(1)	(2)	(3)	(4)
ENV	−0.045 *** (−8.32)		−0.041 *** (−7.57)
ENV2	0.021 *** (5.67)		0.016 *** (5.85)
ENV(-1)		−0.032 * (−1.72)		−0.027 * (−1.75)
ENV2(-1)		0.007 * (1.69)		0.005 * (1.83)
GOV			−0.005 (−0.93)	−0.007 (−0.52)
EDU			0.011 *** (3.82)	0.009 *** (3.76)
RD			0.073 ** (2.37)	0.065 ** (2.12)
INF			0.009 (0.87)	0.013 (0.96)
constant	1.284 *** (3.74)	1.325 *** (4.07)	0.837 *** (4.32)	1.429 *** (5.61)
R2	0.81	0.79	0.92	0.87

Note: T values are in brackets; *, **, and *** indicate significance at the 10%, 5%, and 1% significance levels, respectively.

. To ensure the robustness of the conclusions, Model 1 does not consider the influence of control variables. Due to the possible interaction between ENV and GTFP, or the endogeneity problem caused by missing variables, this paper uses the laggeg variable of ENV for reference. The regression results are as in Model 2. ENV(-1) represents one period lagged of ENV. ENV²(-1) represents one period lagged of ENV². In this paper, relevant control variables are added to the model, and the final regression result is obtained as Model 3. Model 4 uses the lag of ENV and control variables as explanatory variables.

From the regression results of Model 3 in Table 4, the ENV coefficient is significantly negative at the 1% significance level, and the ENV² coefficient is significantly positive at the 1% significance level, indicating that there is a significant U-shaped relationship between ENV and GTFP. To test the robustness of the U-shaped relationship, this paper also refers to the research method of Lind and Mehlum [65]. After testing, it was again verified that there is a U-shaped relationship between ENV and GTFP.

As can be seen from the results of Models 1–4, no matter whether the control variables are added or not, or the laggeg variable is used and there is no significant difference in the results. This shows that the possible risk factors such as endogenous or omitted variables do not have a serious impact on the results, and the research conclusions have certain robustness and reliability. Model 3 shows that there is a significant U-shaped relationship between the ENV and GTFP, with an inflection point of 1.25. As far as China’s current situation is concerned, the average intensity of ENV is 0.362, which is still on the left side of the inflection point in general. It shows that only by continuing to increase the intensity of ENV can it promote further improvement of China’s GTFP. The above analysis shows that Hypothesis 1 is not rejected.

4.2. Analysis of the Mediation Effect

Model 5 in

Table 5. Results of the mediating effect of FDI.

	GTFP	FDI	GTFP
	(5)	(6)	(7)
ENV	−0.041 *** (−7.57)	−0.031 * (−1.83)	−0.028 *** (−7.63)
ENV2	0.016 *** (5.85)	0.014 ** (2.27)	0.012 *** (5.82)
FDI			0.068 *** (5.33)
GOV	−0.005 (−0.93)		0.037 (0.57)
EDU	0.011 *** (3.82)		0.006 ** (2.14)
RD	0.073 ** (2.37)		0.018 *** (3.57)
INF	0.009 (0.87)		0.006 (0.28)
OP		0.024 *** (2.97)
URB		0.047 (0.97)
IND		0.081 (1.24)
constant	0.837 *** (4.32)	0.634 *** (2.89)	0.912 *** (4.32)
R2	0.92	0.86	0.93

Note: T values are in brackets; *, **, and *** indicate significance at the 10%, 5%, and 1% significance levels, respectively.

represents the direct impact of ENV on GTFP. The coefficients of ENV and its quadratic term are both significant at the 1% significance level, and the mediation effect test can be continued.

Model 6 in Table 5 verifies the impact of ENV on FDI. The coefficient of EVN is significantly negative at the significance level of 10%, and the coefficient of EVN² is significantly positive at the significance level of 5%, indicating a U-shaped relationship between EVN and FDI. To test the robustness of this result, this paper also refers to the research method of Lind and Mehlum (2010). After testing, it was found that the results still showed a U-shaped relationship between ENV and FDI.

In the range of low ENV intensity, the proportion of ENV cost to total cost is always low. Since FDI pursues profit maximization, the low proportion of ENV costs is not enough to motivate enterprises to develop green technologies. The increase in regulation costs will only occupy production resources and is not conducive to the development of FDI. In the range of high environmental intensity, ENV stimulates process innovation and product innovation, which will offset the cost increase and ultimately promote FDI. After calculation, the inflection point of the U-shaped curve is 0.045, and China is still on the left side of the inflection point, indicating that China should further enhance the intensity of ENV in order to promote the development of FDI.

Model 7 in Table 5 examines the impact of FDI on GTFP. The results show that FDI can significantly promote the improvement of GTFP. The main reason is that foreign enterprises are more aware of market competition. In order to subvert the existing market position, it is more eager to carry out green technology innovation. In the process of green technology innovation, FDI is more flexible in talent selection, resource mobilization, and incentive mechanisms. The increase in FDI will help to improve GTFP more effectively.

To sum up, the mediating effect of FDI is established. The influence path of ENV→FDI→GTFP is obviously established. Increasing the intensity of ENV can further promote the development of FDI and promote the improvement of GTFP. The above analysis shows that Hypothesis 2 is not rejected.

4.3. Analysis of Threshold Effect

Through the above analysis, it is found that ENV has an U-shaped nonlinear effect on GTFP, and the mediating effect of FDI is also established. In order to analyze the difference in the impact of ENV on the green total factors under different development levels of FDI, this paper analyzes the threshold effect of FDI.

In order to further test whether there is a threshold effect, this paper uses FDI as the threshold variable to conduct a threshold regression analysis. Referring to the panel threshold regression model method proposed by Hansen [55], this paper assumes that there is no threshold value (null hypothesis) or there is a threshold value (alternative hypothesis); there is only one threshold value (null hypothesis) or there are two threshold values (alternative hypothesis); there are only two threshold values (null hypothesis) or there are three threshold values (alternative hypothesis) to test under three conditions. The p-value was estimated by the bootstrap method suggested by Hansen [55], and the double threshold passed the 1% significance test (

Table 6. Threshold test.

Threshold Variable	Threshold Number	F	p	Bootstrap Times	1%	5%	10%
FDI	Single threshold	36.457	0.000	500	19.325	10.732	7.338
	Double threshold	15.625	0.000	500	13.927	7.321	5.129
	Triple threshold	3.256	0.121	500	8.366	5.365	3.753

Note: F-statistics and p-values were obtained from 500 bootstrap replicates.

), so this paper selects the double threshold effect model for analysis.

The double-threshold regression results are shown in

Table 7. Results of the threshold effect.

	Double Threshold Regression Results
	Variable	Coefficients
Threshold variable	ENV (FDI < 0.549)	−0.0068 (−1.27)
	ENV (0.549 < FDI < 0.627)	0.0536 ** (6.32)
	ENV (FDI > 0.627)	0.0872 *** (5.86)

Note: T values are in brackets; **, and *** indicate significance at the 10%, 5%, and 1% significance levels, respectively.

. In the range of low FDI development levels (FDI < 0.549), ENV is negatively correlated with GTFP, and the regression coefficient is not significant, indicating that at the stage of low FDI levels, ENV is not conducive to the improvement of GTFP. In the middle stage of FDI development (i.e., 0.549 < FDI < 0.627), the coefficient of ENV on GTFP is 0.0536, which is significant at the 1% significance level. When FDI is in the high-level development range (FDI > 0.627), the coefficient of ENV on GTFP is 0.0872, which is significant at the 1% significance level. This shows that further promoting the development of FDI can promote green technology innovation, thereby offsetting the cost of ENV, so that ENV can promote the improvement of GTFP. Additionally, with the continuous development of FDI, the promotion effect of ENV on GTFP is also increasing. This further shows that FDI is an important factor to promote the realization of Porter’s hypothesis, and further development of FDI can achieve a win-win situation for both the environment and the economy.

5. Conclusions

How to build a sound environmental governance system in China to achieve high-quality economic development and high-level protection of the ecological environment has become an important issue to be solved urgently. It is of great significance to study the impact of ENV on China’s FDI and GTFP, and how to coordinate the relationship between the three. Based on China’s provincial panel data from 2007 to 2020, this paper used the SBM function and the Luenberger productivity index to measure the GTFP of China’s provinces while considering energy consumption and undesired output. Taking FDI as an intermediary variable, it verified that the transmission mechanism of ENV→FDI→GTFP is real and effective. On this basis, this paper further used the threshold effect model to explore the moderating role of FDI between ENV and GTFP.

The research results show that:

(1) There is a significant U-shaped relationship between ENV and GTFP, and China is still on the left side of the inflection point. Further improvement of ENV will contribute to the improvement of GTFP.

(2) The mediating effect of FDI is established. The transmission mechanism of ENV→FDI→GTFP is remarkably real and effective. ENV and FDI show a significant U-shaped relationship, and China is currently on the left side of the inflection point. Further improvement of ENV can stimulate the innovative vitality of FDI and promote the further development of FDI.

(3) FDI has a threshold effect in the relationship between ENV and GTFP. When FDI is at a low level of development, the effect of ENV on the improvement of GTFP is not significant. Only when FDI is at a high level of development can ENV significantly promote GTFP. The threshold effect results in this paper show that FDI is an important prerequisite for the realization of the Porter hypothesis.

The policy implications of this study are as follows:

(1) The government needs to choose the ENV tools reasonably. ENV is an art, and the effect of regulation depends not only on the intensity of ENV, but also on the method of ENV. Controlled ENV such as environmental standards and emission quotas lack sufficient incentives for enterprises due to their strong compulsion. However, incentive environmental tools such as emissions trading, environmental subsidies, and technology subsidies provide continuous incentives for technological innovation of enterprises, which is conducive to improving GTFP. The government should rationally use various ENV tools to promote technological innovation and green development and achieve a win-win situation of environmental sustainability and economic development to realize the Porter hypothesis as soon as possible.

(2) China and other developing countries should fully consider the quality of FDI when attracting investment, avoid a large inflow of foreign capital with high pollution and high energy consumption, strengthen the level of environmental control, and guide FDI and local environmental control levels to form a positive interaction and promote green development and sustainable development through the Porter effect.

(3) FDI should be guided to achieve high-level development through ENV. FDI pays more attention to profit, and it is easy to invest and enter high-polluting, high-emission, and low-efficiency industries. China and other developing countries should guide FDI to achieve high-quality development by formulating reasonable ENV policies and play the role of FDI in promoting GTFP. Further formulation of policies to attract and stimulate further development of FDI should be undertaken. China should comprehensively consider the development stage and goals of the local environment and economy and formulate reasonable FDI incentive policies to further stimulate the vitality of green technology innovation in society.

(4) The relationship between ENV, FDI, and GTFP should be handled and balanced. The government should not only achieve economic development but also achieve sustainable development of the environment and society. Sustainable development is long-term development that meets the needs of the present without compromising the ability of future generations to meet their needs. They are an inseparable system, not only to achieve the purpose of economic development but also to protect the natural resources and environment such as the atmosphere, fresh water, ocean, land, and forest on which human beings depend, so that future generations can develop and live in peace and contentment. Only the realization of sustainable and long-term development is real development.

For further research in the future, this paper proposes study of the impact of ENV on GTFP in different countries of the world, and what different mediating roles FDI plays in the relationship between ENV and GTFP in these different countries. On this basis, this paper proposes to further study the heterogeneity of the relationship between ENV and GTFP in developed and developing countries and compare the different mediating roles of FDI in the relationship between ENV and GTFP.