*Article* **How Do FDI and Technological Innovation Affect Carbon Emission Efficiency in China?**

**Qizhen Wang <sup>1</sup> and Suxia Liu 2,\***


**Abstract:** China's economic development is characterized by openness, and trade and investment are important engines for promoting economic development. China's economy is now in a transitional period, during which excessive carbon emission reduction would inevitably hinder economic development. In this context, improving carbon emission efficiency is an effective way to achieve sustainable development. This paper deals with the relationships among foreign direct investment, technological innovation and carbon emission efficiency. Our research findings include the following. First, carbon efficiency shows regional differences. East China has the highest mean value of carbon emission efficiency, followed by central China and west China over the sample period. Second, FDI exerts both direct and indirect impacts on carbon emission efficiency through technological innovation, which confirms the intermediate effect of technological innovation. Finally, sub-sample analysis indicates that the impact of FDI and technological innovation on carbon emission efficiency show regional heterogeneity. According to these findings, we offer policy recommendations as follows. The government should stimulate independent innovation, promote technological progress in renewable energy and green energy, and attract environmentally friendly foreign investment to improve carbon emission efficiency and boost green development.

**Keywords:** FDI; technological innovation; carbon emission efficiency; carbon emission

**Citation:** Wang, Q.; Liu, S. How Do FDI and Technological Innovation Affect Carbon Emission Efficiency in China?*Energies* **2022**, *15*, 9209. https://doi.org/10.3390/en15239209

Academic Editors: Grzegorz Mentel and Sebastian Majewski

Received: 22 October 2022 Accepted: 2 December 2022 Published: 5 December 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

Environmental problems are among the major challenges facing the world in the 21st century. As the largest energy consumer, China faces great pressure in reducing carbon emissions (CE). At the 75th UN General Assembly in 2020, China put forward the commitment to reach peak carbon emissions by 2030 and realize carbon neutrality by 2060. Two methods can be adopted to reduce carbon emissions. One is to reduce the use of fossil fuels, and the other is to improve carbon emission efficiency (CEE) so that maximum economic benefits are generated with less resource consumption and less environmental cost. There is an inverse relationship between CE and CEE. The less the CE, the higher the CEE. China is currently undergoing economic transformation towards green development, a low-carbon economy, and sustainable development. Excessive emission reduction would inevitably hinder economic development. Against this background, improving CEE is an effective means to accomplish sustainable development.

China has made opening up a basic state policy since 1978. A great deal of foreign companies have relocated to the Chinese market and accelerated China's modernization. On the whole, China's use of foreign direct investment (FDI) has been expanding in scale and optimizing in structure. FDI has two-sided influence on China's economy. On the one hand, it brings advanced technology and management experience, optimizes industries, and promotes technological innovation (TI). On the other hand, it increases energy consumption, expands the utilization of resources, and increases carbon emissions and environmental pollution.

FDI can affect regional CEE. On the one hand, it directly affects regional CEE. Firstly, the effect depends on the TSE, structure effect and scale effect of FDI inflow. The TSE and structure effect positively affect CEE and negatively affect CE, while the scale effect negatively affects CEE and positively affects CE. Secondly, the effect is also decided by the stage of local economic development. In the initial stage, the host country would weaken environmental protection in order to develop the economy. In addition, the lack of environmental awareness and technology leads to low intensity of environmental regulations (ERG), which attracts a large amount of foreign investment with high CE. This leads to the agglomeration of FDI enterprises featuring high energy use and high CE, which results in the decline of CEE. At the stage of rapid economic growth, environmental deterioration and improved environmental awareness impel local governments to increase the intensity of ERG and give priority to environmentally friendly foreign enterprises. Local TI is improved through the demonstration effect, TSE, and market competition effect of FDI inflow, which boost technological progress in environmental protection and raise CEE.

On the other hand, FDI indirectly affects regional CEE in different ways, including TI. FDI brings advanced environmental protection technology and pollution control experience. Through the demonstration effect and competition effect, it encourages enterprises to optimize production, improves environmental protection technology, and increases CEE. This covers upstream and downstream enterprises as well as related industries. Advanced experience in environmental protection technology and pollution control is spread among related industries and enterprises through the personnel flow effect and industrial correlation effect. As the technologies become more and more mature, they will be spread to other regions through the spatial spillover effect and eventually lead to overall improvement of environmental protection technology and CEE in the host country. It is worth noting that foreign investment may squeeze out domestic investment, so the TSE of FDI enterprises may be weakened.

We conducted research on the relationships among FDI, TI, and CEE to test whether FDI has a pollution paradise effect or a pollution halo effect on China's environment. Compared with existing literature, major contributions of this paper are as follows. (1) Previous literature has carried out a lot of research on CE, energy efficiency, etc., but the research on CEE is less. In the research on China's CEE, more attention is paid to the spatio-temporal heterogeneity of CEE. Few studies pay attention to the factors affecting CEE. (2) Previous literature pays attention to CEE and its influencing factors, but as one of the important factors, FDI is rarely considered; the impact of FDI on CEE through TI is even less. (3) We conduct a sub-sample analysis to check the regional heterogeneity of the effect of FDI on CEE through TI, having expanded the present research.

#### **2. Literature Review**

#### *2.1. FDI and Carbon Emissions*

Scholars have conducted many studies on FDI and CE. They put forward two main hypotheses. The "pollution paradise" hypothesis (PPH) assumes that developed countries have strict environmental standards. In order to obtain more profits, some pollutionintensive enterprises will relocate to developing countries which have a low intensity of ERG. The inflow of foreign investment not only influences economic growth [1], but also increases environmental pollution. Therefore, FDI is an important variable in PPH [2–4]. Bae et al. [5] explored the factors influencing CE in 15 post-Soviet countries. They found that FDI exerted a positive effect on CEE, which indicated that PPH existed in these countries. Therefore, it was very important for the countries to attract FDI in order to obtain renewable energy, since most of them were not developed and faced high costs in deploying new technologies. Meanwhile, they should focus on the detrimental influence of FDI before implementing regulations on attracting FDI inflows. Arain et al. [6] found that FDI and CE were closely related in terms of wavelet scales, which showed close relations between FDI and CE in the short term. The coherence analysis revealed that the relationship was

significant. FDI was favorable for China's economic growth because renewable technologies and low-carbon technologies were adopted by multinational corporations. On the other hand, these multinational corporations led to the increase of CO2 emissions. Therefore, the government should introduce appropriate policies to develop the economy while also reducing CE. Zakarya et al. [7] found that FDI positively affected carbon emissions in emerging countries, verifying the existence of PPH. Behera and Dash [8] came to the same conclusion.

The "pollution halo" hypothesis (PHH) assumes that the inflow of FDI is beneficial for optimizing energy structure and generating green technology spillovers, thereby leading to the decline of CE in the host country [9,10]. Zhu et al. [11] found that FDI had a heterogeneous effect on CE. The coefficient of FDI was positive but insignificant at the 5th quantile, whereas other coefficients were significantly negative at high quantiles. Also, the influence of FDI on CE was significantly negative in highly polluted countries, which supported the halo effect hypothesis. One of the plausible reasons was that multinationals had more advanced technologies compared with enterprises in highly polluted countries. When investing in high-emission countries, multinationals brought clean technology and innovative skills which helped improve environmental quality, while in low-emission countries, they tended to invest in non-polluting sectors. Hence, the inflow of FDI had an insignificant impact on CE.

Many scholars assumed FDI's effect on CE was unstable [12]. Yildrim [13] found that the PPH was valid in Mozambique, Oman, and the United Arab Emirates, while PHH was invalid in Zambia, Iceland, Panama, and India. Khurram et al. [14] found that a positive shock in FDI increased CE, especially in the long term. A negative shock in FDI had an insignificant impact on CE. However, the negative shock had a positive impact on CE in the short run. This indicated that FDI and CE had a non-linear relationship, which was consistent with the study of Zia et al. [15]. Zia et al. [15] employed an autoregressive distribution lag model to study PPH in Pakistan. They also found that the increase in FDI positively affected CE, and the decrease of FDI insignificantly affected CE.

Regarding carbon emission efficiency, Wu et al. [16] employed data envelopment analysis (DEA) to investigate the influence of energy subsidies on CEE based on the role of FDI in competition. They found that fiscal decentralization positively affected CEE, which meant local governments had strong financial resources and low willingness to enrich the tax base through economic growth when the fiscal decentralization index was high, which led to high inputs in low-carbon industries and improvement in CEE. The interaction between fiscal decentralization and government competition negatively affected CEE, which meant that local governments were not willing to save energy and reduce emissions in order to attract FDI inflow. This results in the decline of CEE. It is evident that FDI inflows make local government lower their environmental standards, which is not conducive to improving CEE [17].

#### *2.2. FDI and Technological Innovation*

The earliest research on how FDI affects TI began with the theory of TSE of foreign direct investment on host countries proposed by MacDougall [18]. This theory holds that FDI can produce technology spillovers on relevant industries in the host country. Keller and Yeaple [19] made a comparative analysis on international TSE on U.S. enterprises through imports and FDI. According to them, FDI allowed domestic firms to obtain great productivity gains, accounting for approximately 14 percent of total productivity growth in the U.S. Imports had technology spillovers to domestic firms, but the effect was weaker than that of FDI.

Some studies showed that TSE of FDI positively affected TI in the host country [20,21]. Munteanu [22] divided knowledge spillovers into two categories: supplemental spillovers and complementary spillovers. Both play an important role in increasing the value of technology transfer through FDI and are influenced by learning ability and technology gap. TSE and propagation effects, especially for the knowledge of technology, significantly influence economic growth at both horizontal and vertical levels. Fernandes and Paunov [23] found that FDI in services positively affected the TFP of manufacturing firms in Chile. It was evident that the horizontal spillovers of FDI were weaker than its vertical spillovers. The reason was that foreign-owned firms did not intend to produce technology spillovers to domestic firms within the industry. They were more willing to produce technology spillovers to downstream providers and upstream clients. To explore the influence of different technological sources on energy conservation technology, Yang et al. [24] analyzed six basic technological sources. They found that forward TSE of FDI competition forced domestic firms to improve energy-saving technology, while backward TSE, horizontal TSE, learning by export, and innovation had no significant effect on energy-saving technology. Negash et al. [25] found that Chinese-invested firms had higher productivity than local firms in Ethiopia. Hence, Chinese-invested firms brought advanced technology, capital, and knowledge to local firms, which stimulated the latter's TI. Whether local firms could gain significant positive TSE from Chinese-invested firms was decided by their absorptive capacity.

Some studies demonstrated that FDI negatively affected TI in the host country [26,27]. Feng et al. [28] found that FDI negatively affected China's urban innovation. Therefore, the government is advised to optimize the structure of FDI: central and west regions should attract high-quality FDI instead of expanding the size of FDI and should not lower the intensity of ERG in order to attract foreign investments. What's more, local government should implement strict environmental regulations to attract multinationals with advanced technology and energy-saving ability and restrict enterprises with weak technological strength and high energy consumption. Hu et al. [29] classified FDI into labor-based FDI and capital-based FDI. The research results demonstrated that both types of FDI negatively affected green total factor productivity (GTFP), but only the effect of labor-based FDI was significant. The plausible reason was that labor-intensive industries produced less carbon emissions and featured cleaner production [30], and labor-based FDI may improve the quality of human capital and generate TSE in the host country. Meanwhile, there were weak ERG in labor-intensive industries, and these industries had no strong technological innovation, which led to the decline of GTFP. The intensity of ERG on capital-intensive industries was higher, so capital-based FDI brings less pollution. Hence, the influence of capital-based FDI on GTFP in the host country was uncertain.

Some scholars also found that FDI had mixed effects on TI in the host country. Anwar and Nguyen [31] investigated the effect of FDI on TFP and found that it was different in different regions. For instance, FDI produced positive horizontal TSE in the northeast, but negative horizontal TSE in the Red River Delta. FDI produced positive backward TSE in the Red River Delta, but negative backward TSE in the Mekong River Delta. Hu et al. [32] found that labor-based FDI had a significantly negative TSE, while capitalbased FDI had a significantly positive TSE in industries with a low intensity of ERG. In industries with a high intensity of ERG, labor-based FDI had an insignificantly negative TSE, and capital-based FDI still had a significant TSE. This indicated that the negative TSE of labor FDI was decided by the intensity of ERG. The "pollution haven" hypothesis was valid in industries with a low intensity of environmental regulations.

#### *2.3. Technological Innovation and Carbon Emissions*

Many studies focused on environmental variation and its determinants [33–36]. Among them, the link between technological innovation and CE has been abundantly investigated. Most studies assume that technological innovation reduces CE [37–40]. Using firm level data, Lee and Min [41] examined the impact of investment in green R&D on the environment and financial performance. They found that green R&D negatively affected CE, but positively affected financial performance. Kong et al. [42] found that all efficient energy technologies could greatly improve pulp and paper technology. Therefore, advanced technologies should be used to decrease energy consumption and CE in the industry. Sgobbi et al. [43] found that the improvement of technological efficiency was very important to tidal energy. Improving efficiency was more important than reducing costs, because technological upgrades can increase energy supply and reduce CE. Zeeshan et al. [44] found that R&D negatively affected CE. In the short run, there was an insignificant negative relationship between R&D and CE. As far as China is concerned, spending on R&D was also related to CE.

Few studies concluded that technological innovation increased CE. R&D plays a major role in new technologies and new products. Through R&D, more and more competitive products are produced. Shaari et al. [45] employed the panel DOLS and FMOLS to analyze the relations between R&D and CO2 emissions. The results of FMOLS suggested that R&D positively affected CO2 emissions. The results of DOLS suggested that R&D positively affected CE. This implies that expenditures on R&D should be reduced to improve environmental quality and boost economic growth.

There were also some mixed or different results regarding the relationship between TI and CE. Demir et al. [46] analyzed this relationship using the ARDL approach. They found that the relationship between home patents and CE followed an inverted U-shape in Turkey. This meant that home patents were positively related to CO2 emissions in the early stages of economic development. When economic development reached a certain level, home patents were negatively related to CO2 emissions. Like Demir et al. [46], Gu et al. [47] found the relationship between energy-technological progress and CE in China was in an inverted U-shape. Yii and Geetha [48] found that technological innovation negatively affected CE in the short term and positively affected the latter in the long term in Malaysia. Dauda et al. [49] revealed that technological innovation negatively affected CE in the G6 countries but positively affected CE in the MENA (Middle East and North Africa) and BRICS (Brazil, Russia, India, China, and South Africa) countries. Erdo ˘gan et al. [50] studied the influence of innovation on CE in G20 countries. They found that the increase in innovation resulted in the reduction of CE in the industrial sector and resulted in the increase of CE in the construction sector.

As for the connection between energy and environment, CEE which reflects the level of green development has attracted a lot of attention. Zhang and Chen [51] found that technological progress negatively affected CEE. This is the same as the conclusion of Wang et al. [52] and Huang et al. [53]. The reason behind this conclusion was the rebound effect. Usually, technological progress positively affected energy efficiency and negatively affected CE. But technological progress would lead to production expansion and more energy consumption, so carbon efficiency declined because of the rebound effect. In addition, technological progress changed lifestyles and led to the use of a large number of electronic products, which greatly increased electrical energy consumption and thus improved CE. Hence, the positive effect of technological progress may be offset by a negative rebound effect. Different from Zhang and Chen [51], Yan et al. [54] found that technological progress was a main driving force for improving CEE. Santra [55] studied the effect of environmental innovation on energy efficiency and CEE in BRICS countries. They concluded that green technological innovation reduced energy absorption and CE and improved energy efficiency and CEE in each member country.

#### *2.4. Research Gap*

CO2 is a kind of heat-trapping gas and the largest contributor global warming. Hence, curbing or reducing CO2 emissions is crucial to sustainable development [56]. According to PPH and PHH, FDI is a double-edged sword that can both increase and decrease CE, which is decided by the sum of TSE, the structure effect, and the scale effect of FDI in different stages of economic development. There are abundant studies on the relationship between FDI and CE, but few deal with the relationship between FDI and CEE. Is there a linear or non-linear relationship between the two variables? If FDI is related to CEE, what is the degree of FDI's impact on CEE? Few studies have focused on the channels through which CEE is affected, and possible channels of technological innovation have been ignored. In addition, regional analysis is also missing. Since China has many regions, heterogonous

characteristics should be investigated. In view of this situation, this paper examines the effect of FDI on CEE, including direct and indirect channels, and takes into account regional differences. The results are quotable for policymakers to increase CEE through FDI.

#### **3. Methods and Materials**

#### *3.1. Super Efficiency DEA Method*

Data envelopment analysis (DEA) is a special tool based on linear programming proposed by Charnes et al. [57]. It is mainly used to evaluate the efficiency of a decision-making unit (*DMU*) using an input–output method. There are two traditional DEA methods: CCR-DEA model [57] and BCC-DEA model [58]. Both can be used to calculate the efficiency score and test the effectiveness of the efficiency score for each DMU. One of their disadvantages is that they cannot compare and analyze different DMUs when DMUs are effective at the same time. In addition, they do not take into consideration the impact of random errors and are easily affected by sample data. Hence, the efficiency score may be biased. When the error item is large, the results estimated by the DEA model will be serious biased. The super efficiency DEA model which was proposed by Anderson and Petersen [59] is different from traditional DEA models and can compare different effective DMUs.

After calculating efficiency with the traditional DEA method, we obtain two kinds of efficiency values: those less than 1 and those is equal to 1. The former means that the efficiency value is invalid, while the latter shows the efficiency value is valid. The traditional DEA method is unable to compare and analyze effective *DMUs* because all efficiency values are 1. In fact, there are differences among effective *DMUs*. According to Anderson and Petersen [59], the formula for this study is:

$$\begin{cases} \min(\theta) - \varepsilon (\sum\_{i=1}^{m} s\_i^- - \sum\_{r=1}^{s} s\_r^+) \\ \text{s.t.} \sum\_{j=1}^{n} x\_{ij} \lambda\_j + s\_r^- = \theta x\_{ij\_0} \\ \quad \sum\_{j=1}^{n} y\_{ij} \lambda\_j - s\_r^+ = y\_{ij\_0} \\ \quad \lambda\_j, s\_r^-, s\_r^+ \ge 0, i = 1, 2, \dots, m \\ j = 1, 2, \dots, j\_0 - 1, j\_0 + 1, \dots, n \end{cases} \tag{1}$$

This is a super efficiency model used to evaluate *j*<sup>0</sup> decision-making unit *DMU*0. It removes the evaluated unit *DMU*<sup>0</sup> from the reference set and obtains its own value by referring to the frontiers of other *DMUs*. This fills in the blank of the traditional DEA model, which cannot be used for further analysis when the efficiency value is 1. The multi-input–multi-output evaluation system has *n DMUs*, including *m* input indicators and *s* output indicators. *xij* is the *i*-th input of the *j*-th *DMU*, and *yij* is the *i*-th output of the *j*-th *DMU*. *θ* is the super efficiency value of *DMU j*0. *ε* indicates the non-Archimedean infinitesimal, and *s*<sup>−</sup> and *s+* are slack variables.

#### *3.2. Data Resources*

The data are from the China Statistical Yearbook, China Economic Network Statistical Database, China Science and Technology Statistical Yearbook, World Bank, United Nations Conference on Trade and Development (UNCTAD) Database, and China's Carbon Emissions Database from 2004 to 2019. Tibet, Macao, Hong Kong, and Taiwan are not included owing to incomplete data.

#### 3.2.1. Carbon Emission Efficiency

The super efficiency DEA model is employed to estimate CEE. Based on previous research literature and taking into account the characteristics of this study, the input variables are capital (*X*1), labor (*X*2), and energy consumption (*X*3), and the output variables are GDP (*Y*1) and CO2 (*Y*2). The input and output variables are presented in Table 1.


**Table 1.** Index system of regional carbon efficiency.

In Table 1, capital refers to the stock of fixed capital, which is the weighted sum of the previous investment flows measured at constant prices. Total current capital is equal to total capital of the previous period—depreciation + current capital, which is expressed as *Kjt = Kjt*−*1*(1 − *δjt*) + *Ijt*. *Kjt* is the capital stock of province *j* in period *t*, *Kjt*−<sup>1</sup> means the capital stock of province *j* in period *t* − 1, *δjt* is the depreciation rate of province *j* in year *t*, and *Ijt* represents the investment in province *j* in year *t* by the prices of the current year. The base year is 2004. Labor input is the number of employees in each province, and energy consumption is the total energy consumption of each province. The GDP is calculated based on the GDP deflator of 2004, and CO2 refers to the CO2 emissions of each province.

The values of CEE are shown in Figure 1. It is clear that the value of CEE in east China was the highest before 2012, and central China had the highest value after 2012. Both national and regional values showed an increasing trend, and the value of CEE showed regional imbalances. Since 2004, the national average value of CEE has been growing. It reached a peak in 2015 and then decreased. The CEE value of east China had a downward trend, while central China had an upward trend. The lowest value of CEE was in west China, but the value showed an upward trend.

**Figure 1.** Average CEE values, 2004–2019.

#### 3.2.2. Other Variables

In this study, FDI is an investment behavior in which investors in one country use their capital for production or operations in other countries and exert control. FDI (*fdi*) is a main explanatory variable of this study, which is measured by the total amount of foreign direct investment by logarithm. Technological innovation (*rd*) is the application of new knowledge, new technology, and new processes by enterprises to improve product quality, develop new products, and ultimately occupy a certain market share and realize market value, which refers to the R&D expenditure of each province during the sample period. Based on previous studies [60,61] and the characteristics of this study, we add the degree of openness (*open*), industrial structure (*industry*), and fiscal expenditure (*fine*) as control variables to our models. *open* is measured by the proportion of the import and export in GDP. Industrial structure (*industry*) is equal to the proportion of the secondary industry in the total output. Fiscal expenditure (*fine*) is the ratio of fiscal expenditure to the GDP.

#### *3.3. Econometric Models*

The inflow of FDI helps expand local economy. However, it also increases energy consumption and CO2 emissions in the host country. Since FDI enterprises usually have better technological strengths than local enterprises, its inflow will cause technology spillover to local enterprises, which improves local industrial technology and increases CEE. Therefore, FDI will directly and indirectly affect CEE, and the indirect effect is through TI. In order to test the relationships among FDI, technological innovation, and CEE, the following three models are built:

$$
\omega a\_{it} = \gamma\_0 + \gamma\_1 r d\_{it} + \gamma\_2 f d\_{it} + \gamma\_3 o pen\_{it} + \gamma\_4 \|dustry\_{it} + \gamma\_5 f inc\_{it} + \varepsilon\_{it} \tag{2}
$$

$$
\cos\alpha\_{it} = \beta\_0 + \beta\_1 f d i\_{it} + \beta\_2 op n\_{it} + \beta\_3 industry\_{it} + \beta\_4 func\_{it} + \theta\_{it} \tag{3}
$$

$$r d\_{it} = a\_0 + a\_1 f d\_{it} + a\_2 o pen\_{it} + a\_3 industry\_{it} + a\_4 func\_{it} + \tau\_{it} \tag{4}$$

where *i* is province and *t* is year. *αi*, *βi*, (*i* = 0, 1, ... , 4), and *γi*, (*i* = 0, 1, ... , 5) are regression coefficients. *εit*, *θit*, *τit* are residual terms. *co* refers to carbon emission efficiency, *rd* is technological innovation, and *fdi* is foreign direct investment. *open*, *industry*, and *fine* are control variables, referring to the degree of openness, industrial structure, and fiscal expenditure, respectively. Equations (2)–(4) jointly verify the intermediate effect of technological innovation [62].

The step method proposed by Wen and Ye [63] is adopted to estimate the intermediate effect of FDI. Firstly, the significance of coefficient *β*<sup>1</sup> in Equation (3) is tested. If it is significant, subsequent tests will be conducted. Otherwise, the test will be terminated. Secondly, the significance of *α*<sup>1</sup> in Equation (4) and *γ*<sup>1</sup> in Equation (2) is tested. If both coefficients are significant, it proves the existence of the intermediate effect. If one of them is insignificant, the Sobel test will be conducted. When the null hypothesis of the Sobel test (*H*0:*α*2*γ*<sup>3</sup> = 0) is rejected, the intermediate effect is supported. Finally, the significance of *γ*<sup>2</sup> in Equation (2) is checked. If it is significant, there is a partial intermediate effect; otherwise, there is a complete intermediate effect. The intermediate effect can be calculated by *α*<sup>1</sup> ∗ *γ*1, and the direct effect can be estimated by *γ*2. Therefore, the total effect is *α*<sup>1</sup> ∗ *γ*<sup>1</sup> + *γ*2.

We first check whether there is serious multicollinearity. If the data are highly correlated, it will lead to distorted regression results and an inaccurate estimation result. Table 2 reports the results of the correlation analysis of the panel data. The maximum coefficient of correlation between *rd* and *fdi* is 0.614, and that of the correlation between *fine* and *fdi* is −0.566. Therefore, there are no serious problems of multicollinearity, and we can conduct a regression analysis.


**Table 2.** Results of the correlation analysis.

Table 3 presents what we get from descriptive statistical analysis. There are 480 observed values. The mean of all variables except for *open* is greater than the standard error. The mean value of the variable *open* is 0.053, and the standard error is 0.076. This means the value is smaller than the standard error, which shows that this variable is a little scattered. Since the sample size is greater than 30, the regression result will not be affected. This table gives a list of the maximum and minimum values of each variable. The largest value of these maximum values is the variable *rd*, but the smallest value of these maximum values is the variable *open*. The largest value of these minimum values is the variable *rd*, but the smallest value of these minimum values is the variable *fdi*. The variable *rd* has the largest gap between the maximum value and the minimum value, being 8.386, and the variable *co* has the smallest difference, standing at 0.207.

**Table 3.** Descriptive statistical analysis.


#### **4. Results and Discussion**

*4.1. Analysis of the Impact of FDI and Technological Innovation on CEE*

The impact of FDI and TI on CEE is analyzed first. When processing panel data, we should choose whether to adopt the fixed-effect model or the random effect model. The null hypothesis is that "unobservable random variables are not related to all explanatory variables". If the null hypothesis is accepted, the random effect model should be used. If the null hypothesis is rejected, the fixed-effect model should be used. According to Hausman test, we find that panel data is significant at the statistical level of 1%. Hence, the fixed-effect model should be used. Table 4 reports the regression results of the fixed-effect model.


**Table 4.** Impact of FDI and technological innovation on CEE.

Note: \*\*\*, \*\*, and \* indicate significant levels at 1%, 5%, and 10%, respectively; the data in parenthesis are standard errors.

FDI positively influences CEE. When it increases by 1%, CEE increases by 0.003%. Foreign investment brings advanced technologies, standards, and concepts on environmental protection to the host country. Furthermore, enterprises in the host country reduce carbon emissions and increase CEE thanks to the imitation effect and the demonstration

effect. Foreign ideas of environmental protection are of great significance to improving CEE of the host country. Environmental protection not only concerns human health, but also plays a vital role in long-term social development. The spread of ideas on environmental protection allows the host country to recognize the role of the environment, pay attention to the improvement of environmental protection technology, and improve the intensity of environmental regulations. Meanwhile, the inflow of foreign investment improves the technical level of the whole industry, increases energy efficiency, saves energy, and reduces carbon emissions. Fang et al. [60] drew the opposite conclusion in their study of CE of 282 cities in China. They found that FDI negatively affected CEE. Therefore, each city should raise thresholds for environmental access when absorbing the inflow of FDI. In addition to reducing disorderly competition and energy rebound, foreign enterprises could enhance environmental governance and advance technological progress.

Technological innovation positively affects CEE. It is a process of increasing the technology level which brings TSE to enterprises, reduces CE, and increases CEE. Fang et al. [60] also found that the effect of technical development was positive, that is, the use of clean technology improved CEE. R&D expenditure on low-carbon industries could bring technological progress as well as the energy rebound effect. Technology investment may bring technological progress as well as the energy rebound effect. When the former was greater than the latter, R&D expenditure could improve CEE. Our results are also consistent with those of Rizwana et al. [64]. Rizwana et al. [64] studied the Belt and Road economies and found that technological innovation could save energy consumption costs and improve environmental quality.

FDI is conducive to enhancing TI of the host country. When it increases by 1%, technological innovation increases by 1.054%. Some FDI goes to OEM production, that is, producing, processing, and assembling products in China and finally exporting the products to other countries. In this process, advanced foreign technical standards and environmental protection standards are followed by local enterprises, bringing about technology spillover and improving CEE in China. Other FDI goes to R&D institutions in China. In order to gain more profits, multinationals attach great importance to developing technologies, which causes technology spillovers to Chinese enterprises. The technology spillover is realized through the demonstration effect of products and the flow effect of R&D personnel.

Among the control variables, *open* positively affects CEE. It is calculated as the ratio of international trade to GDP. The larger the ratio, the greater the degree of openness. When China is open enough, it is fully connected to the world. This means it can get advanced environmental protection technologies from developed countries, which is conducive to its energy conservation and CEE. *open* negatively affects R&D expenditure. The possible reason is that most of China's product exports win by quantity, with low technological content, so the exports do not produce strong technology spillover to domestic R&D. Industrial structure, which equals to the ratio of secondary industry output to the total output, negatively and insignificantly affects CEE. This is because the secondary industry consumes a large amount of energy and generates a lot of CE, which counts against CEE. This is consistent with the conclusion of Fang et al. [60]. According to Fang et al. [60], industrial structure was negatively related to CEE, because CE mainly came from the secondary industry. Fiscal expenditure positively affects CEE. After the reform and opening up is implemented, local governments in China have focused on developing the economy. They no longer sacrifice the environment for economic growth due to river pollution, ecological degradation, and human health threats caused by ignoring environmental protection. They are changing from an extensive development mode to emphasis on environmental protections. Financial expenditure on environmental protection has been increased, reducing CE and increasing CEE.

#### *4.2. Robustness Test*

In order to test whether the above regression results are consistent and stable when some parameters change, a robustness test was conducted. Three methods can be used for the robustness test: variable replacement, method replacement, and change of sample size. The Tobit regression method is used in this section, and the results are listed in Table 5. FDI positively affects CEE. When FDI increases by 1%, CEE rises by 0.003%. Technological innovation positively affects CEE. When technological innovation increases by 1%, CEE increases by 0.002%. FDI positively affects R&D expenditure. When FDI increases by 1%, R&D expenditure rises by 1.141%. Hence, the regression results are robust.


**Table 5.** Robustness test.

Note: \*\*\*, \*\*, and \* indicate significant levels at 1%, 5%, and 10%, respectively; the data in parenthesis are standard errors.

#### *4.3. Intermediate Effect Test*

Based on Equations (2)–(4), the stepwise method is used to test the intermediate effect of technological innovation. Table 6 reports the test results with both fixed-effect regression and Tobit regression. First, the coefficient *β*<sup>1</sup> in Equation (3) indicates that the influence of FDI on CEE is significant. *β*<sup>1</sup> is 0.004 in both the fixed-effect regression and the Tobit regression, so subsequent tests can be performed. Secondly, the significance of coefficient *α*<sup>1</sup> in Equation (4) and coefficient *γ*<sup>1</sup> in Equation (2) is tested. It is found that both coefficients are significant in the fixed-effect regression and the Tobit regression, which proves the existence of an intermediate effect.

**Table 6.** Intermediate effect test.


Note: \*\* indicates the significant level at 5%.

In the fixed-effect regression, the direct effect of FDI on CEE is 0.0028, and the indirect effect is 0.0017, of which the indirect effect accounts for 37.78%. This shows that FDI directly affects CEE and indirectly affects the latter through TI. In the Tobit regression, the direct effect of FDI on CEE is 0.0026, and the indirect effect is 0.0018, of which the indirect effect accounts for 40.91%. This not only confirms that FDI affects CEE both directly and indirectly, but also indicates that the conclusion is robust. As for carbon emission efficiency, He et al. [65] also proved the existence of an intermediate effect. In this study, we used the stepwise method to analyze the relationships among FDI, technological innovation, and

carbon efficiency, which is different from the research of He et al. [65]. He et al. [65] used the panel threshold model to examine the relationships among technological innovation, market forces, and carbon efficiency. The intermediate variable in this study is technological innovation, but it was market forces in the research of He et al. [65]. Although the intermediate variables are different, we all find that there is an intermediate effect in the research of CEE.

#### *4.4. Sub Sample Regression*

We find that some places have a relatively developed economy and more foreign investment of higher quality, which help improve the local technical level and raise the ratio of environmentally friendly enterprises, thus improving CEE and reducing environmental pollution. In other regions, the economy is relatively backward. To ensure economic growth, they sacrifice the ecological environment. Among the foreign enterprises introduced by them, there are a large number of non-environmentally friendly enterprises, which reduces CEE and increases regional environmental pollution. Therefore, regional economy has different effects on FDI, technological innovation, and CEE. China has a developed economy in eastern regions and a backward economy in central and western regions. In this section, the total sample data is divided by region according to geographical location for detailed regression. The results of the regression are listed in Table 7.

**Table 7.** Results of sub-sample regression.


Note: \*\*\* indicates significant levels at 1%; the data in parenthesis are standard errors.

FDI positively affects CEE in east China, with an insignificant impact coefficient. It negatively affects CEE in central and western regions, with an insignificant impact coefficient. Regarding the direction and degree of impact, we can see that the eastern region has attracted a great quantity of FDI inflows with high technological content, which improves local technological innovation, CEE, and the ecological environment. Central and western regions have received many non-environmentally friendly FDI enterprises, which has a slight positive and even negative effect on CEE. FDI positively affects R&D expenditure in all three of the regions, which indicates that there is TSE in China. Nevertheless, whether technology spillover improves local technology innovation and CEE depends on the absorption capacity of each region.

#### **5. Conclusions and Policy Implications**

This paper firstly employs the super efficiency DEA model to obtain the CEE of 30 provinces and cities in China, which reveals that there are regional imbalances. During the whole period, the highest mean value of CEE falls in the eastern region, followed by the central region and the western region. The efficiency shows a downward trend in the central region and the western region but an upward trend in the eastern region. On this basis, we test the relationship among FDI, TI, and CEE with regression models. The results reveal that both FDI and technological innovation positively and significantly affect CEE, and FDI positively and significantly affects technological innovation. The intermediate effect test confirms that FDI directly affects regional CEE and indirectly affects CEE through TI. Finally, we divide the total sample into the eastern region, the central region, and the western region and test the regional heterogeneity of FDI, technological innovation, and CEE. Based on research results, we propose several policy implications which are stated below.

Foreign investment in central and western regions is more likely to transfer pollution. It promotes local economic development but brings about serious environmental problems. The central and western regions should commit to low-carbon development and increase environmental standards to attract environmentally friendly foreign investment. Therefore, in the process of attracting FDI inflow, the central and western regions should perform comprehensive low-carbon planning, construction, and management and urge all parties to meet low-carbon requirements. The central and western regions should focus on economic growth and improve their absorptive capacity so that they can transform FDI technology spillovers into independent innovation and realize sustainable development.

Independent innovation is a key driving factor for high-quality economic development. Promoting independent innovation through FDI technology spillover is an important means for improving China's technology. In general, the government should increase investment in R&D and enhance its patent system to increase FDI spillovers. Considering regional differences, the eastern region, which embraces a high degree of innovation, should focus on independent innovation capability. It should enhance competitive advantages through absorbing high-technology FDI. The central and western regions should strengthen the attraction of foreign investment through improving the innovation market, cultivating technical talents, and expanding innovation subjects, so as to form the innovation catch-up effect. Moreover, some of the areas in the western region should not blindly carry out independent R&D activities. They should invest more in technological transformation, increase economic output based on effective innovation, and then improve carbon emission efficiency.

It is difficult to change the status quo that China's economy will depend on resources for a long time, and carbon emissions will continue to increase. Therefore, it is more realistic to reduce carbon emissions and increase carbon emission efficiency by using renewable energy. It is recommended that the government formulate industrial and macroeconomic policies to support renewable energy and green energy technologies, supervise related R&D activities, and provide financial support for them. Also, the government is advised to encourage domestic and foreign enterprises to cooperate in the development of renewable energy. The renewable energy industry is a high-risk industry characterized by high investment. Foreign enterprises face more risks and challenges due to transnational factors. For example, high financing costs and less experience in the renewable energy industry often hinder their overseas investment. To solve these problems, it is a better choice for domestic and foreign enterprises to jointly invest in this industry.

This study represents more in-depth research on the relationships among FDI, technological innovation, and CEE, but there are still deficiencies and limitations. We suggest the following future research directions. First, alternative estimation methods (considering structural breaks and non-linear/asymmetric) can be adopted to investigate the relationships among the three variables and explore whether future results support empirical research in different panels. Second, the analysis can be performed in the Environmental Kuznets Curve (EKC) framework. Different stages of open economy may have diverse effects on FDI and CEE, which is of great significance for China to formulate targeted energy policies. These analyses should be more fruitful and helpful.

**Author Contributions:** Methodology, Q.W.; formal analysis, S.L.; investigation, S.L.; resources, S.L.; data curation, Q.W.; writing—original draft preparation, Q.W.; writing—review and editing, Q.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** All relevant data have been submitted within this manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

