*3.3. Data Sources and Variable Descriptive Statistics*

Considering the data availability and the actual needs of the research, this paper selected the Chinese provincial panel data between 2003 and 2017 to empirically investigate the impact of environmental regulation on green total factor productivity. To be specific, our sample covers 30 provinces, municipalities, and autonomous regions in mainland China from 2003 to 2017. Due to a shortage of the portion of required data, Hong Kong, Macao, Taiwan, and Tibet are excluded here. The sample dataset used in the study was mainly derived from the China Statistical Yearbook, the China Energy Statistical Yearbook, the China Environment Statistical Yearbook, the China Industrial Statistical Yearbook, the China Industrial Economic Statistical Yearbook, the China Science and Technology Statistical Yearbook, and each province's Provincial Statistical Yearbook for each sample year. The methods of moving the averages and interpolations were applied to supplement the missing data in some years and regions. In total, a balanced sample set of 450 observations was created for each region during this 15-year period. Furthermore, all variable values in the study were transformed into logarithmic forms to reduce heteroscedasticity. In order to eliminate the price effect, we deflated all the nominal variables in this study into real variables, by a GDP deflator, into the 2003 constant price. The descriptive statistics of all aforementioned variables are summarized in Table 1 below.


**Table 1.** Descriptive statistics of the variables from 2003 to 2017.

#### **4. Empirical Results Analysis**

#### *4.1. Unit Root Test and Multicollinearity*

In order to eliminate the spurious regression problem and to ensure that the estimation results are accurate and reliable, the stationarity test of all variables was implemented before the regression analysis. The test methods consisted of LLC, IPS, Fisher-ADF, and Fisher-PP tests. As shown in Table 2, the test results showed that almost all of the variables can pass more than three significance tests simultaneously, indicating that the raw data sequence of each variable was stationary. In addition, the variance inflation factor (VIF) was used to test the multicollinearity problem. The results of the VIF test indicated that the VIF values were all less than 10 and ranged from 1.24 to 2.22, showing that there was no multicollinearity among the variables. Therefore, the regression analysis was performed.


**Table 2.** Unit root and VIF test results.

Note: \*\*\*, \*\*, and \* represent mean significance at the 1%, 5%, and 10% levels, respectively.

#### *4.2. The Spatial-Temporal Dynamic Evolution of Regional GTFP in China*

Based on the panel data of 30 provinces in China over the period from 2003 to 2017, this study adopted a directional distance function and the GML index to measure the regional GTFP growth, as shown in Figures 3 and 4.

**Figure 3.** Trends of regional average GTFP growth index in China from 2003 to 2017.

**Figure 4.** Spatial distribution of regional average GTFP growth rate in China from 2003 to 2017.

Figure 3 shows the overall trend of China's GTFP from 2003 to 2017. The average annual growth rate of GTFP dropped from 7.8375 in 2003 to 0.4687 in 2017, indicating that the GTFP presented an overall downward trend during these 15 years. To be specific, the GTFP declined from 2003 to 2007, followed by a slight rise from 2008 to 2010 and further fluctuates until 2013, although these variation features are not very obvious. During the 11th and 12th Five-Year Plans, the government enhanced the governance of energy conservation and the emission reduction of enterprises, gradually strengthening the responsibility of local governments to control their environmental pollution. Due to the long-term dependence on the extensive economic development model and the occurrence of the financial crisis in 2008, China has implemented a series of stimulus programs to promote infrastructure investment and heavy industries. As a result, the extensive production conditions caused by resource consumption and environmental pollution have not substantially improved. In 2013, the average growth rate of China's GTFP rose to 6.1834, illustrating that environmental supervision has brought about some positive effects. However, since 2013, when China's economy entered a new normal period, the pressure of the growth slowdown caused a decline in GTFP, which decreased to 0.4687 in 2017.

In terms of changes in different regions, the levels of economic development in the eastern, central, and western regions were different, and the changes in GTFP varied accordingly (see Figures 3 and 4). Specifically, between 2008 and 2014, the eastern region experienced rapid economic growth and the contribution of GTFP increased from 0.9425 in 2008 to 3.662 in 2014, after which it began to decline rapidly. From 2006 to 2012, the GTFP growth in the western region fluctuated little. Since 2013, the GTFP growth rate in the western region has shown a downward trend, declining from 9.9725 in 2013 to 3.7044 in 2017. From 2005 to 2015, the GTFP growth in the central region fluctuated little, showing the steady contribution of GTFP to economic development. However, the GTFP growth rate rose to 4.7981 in 2016 and then began to decline rapidly, falling to 0.8975 in 2017.

Since the reform and expansion, the central region has given priority to the development of heavy industry to achieve rapid economic growth. Due to the lack of physical capital, human capital, and advanced technology, the effect of the industrial policy on economic growth was restricted to a great extent. Therefore, the local government relaxed the punishment for resource damage and environmental pollution, trying to sacrifice the environment in exchange for the rapid economic growth. However, this extensive development model is doomed to be unsustainable. Although environmental governance was strengthened in 2015, the region is still unable to get out of the development pattern characterized by high pollution and high energy consumption.

The western region is rich in natural resources and energy and has a strong economic development potential. With the implementation of the Western Development Strategy and the latecomer advantage, some developed areas have witnessed rapid economic growth, which will inevitably accelerate energy consumption and environmental pollutants emission, resulting in a poorer performance of GTFP long-term.

### *4.3. The Analysis of Baseline Empirical Results*

The purpose of this paper is mainly to test whether the nonlinear U-shaped relationship between environmental regulation and GTFP exists or not. Furthermore, considering the background of the Chinese market-oriented reform, governance transformation is also introduced into the model to investigate its impact on GTFP. Specifically, we first used the feasible generalized least squares (FGLS) method, the fixed effect (FE) method, and the random effect (RE) method to explore the effects of environmental regulation and governance transformation on GTFP, respectively. The estimation results are presented in columns 1–6 of Table 3. As we expected above, the coefficients of the quadratic term of environmental regulation were positive, and so were the coefficients of governance transformation. However, the fitting degrees of these six models were all less than 0.4, indicating that these three estimation methods were not able to explain causality effectively. In addition, if the explanatory variables are endogenous, the FGLS method, the FE method, and the RE method may lead to inconsistencies in parameter estimations. Therefore, in order to effectively overcome the endogeneity problem, we used the system generalized method of moments (SYS-GMM) to estimate the models by introducing instrumental variables. The difference generalized method of moments (DIFF-GMM) was also used to ensure that the regression results were robust. The estimation results of the DIFF-GMM method and SYS-GMM method are shown in columns 7 and 8 and columns 9 and 10 of Table 3, respectively. It should be noted that the AR tests, which are used to test the autocorrelation of the residual term, showed that there was a first-order autocorrelation but there was no evidence of a second-order autocorrelation. Meanwhile, the Hansen overidentification tests, which are usually adopted to examine the validity of the instrumental variables, indicated that the null hypothesis cannot be rejected; namely, the instrumental variables were jointly effective. Therefore, the specifications of the dynamic panel data models in this study are reasonable.


**Table 3.** The regression results of impact of environmental regulation on GTFP for full samples.

Note: standard errors in parentheses. \*\*\*, \*\*, and \* represent mean significance at the 1%, 5%, and 10% levels, respectively.

Firstly, the coefficients of the first-order lag term of GTFP were all significantly positive at the 1% level in Models (7)–(10), showing that the growth of GTFP is a dynamic accumulation process with significantly positive feedback and path dependency. The SYS-GMM estimation results of the environmental regulation showed that the coefficients of the linear term of environmental regulation were all significantly negative, while the coefficients of the quadratic term were significantly positive, indicating that there is a nonlinear Ushaped relationship between environmental regulation and GTFP. As a comparison, the DIFF-GMM estimation results showed that the regression results were robust. Therefore, Hypothesis 1 (H1) is verified. The feasible explanation for this nonlinear relationship is that when environmental supervision is less strict, the pollution control costs and the extensive production are higher, and the green economic efficiency is lower. However, with the improvement of environmental regulation stringencies, enterprises must continue to increase environmental investment to carry out green technology innovation, which will gradually compensate for the compliance costs and contribute to green productivity in the long term. Furthermore, taking Model (10) as an example, the inflection point value of environmental regulation is 3.065, which is greater than the mean level of 2.259, shown in Table 1, indicating that the intensity of environmental regulation in China is still located on the left side of the U-shaped curve, meaning that the promotional effect of environmental regulation on GTFP has not been achieved fully.

Secondly, the coefficients of governance transformation were all positive and significant in Model (8) and Model (10), showing that governance transformation can promote the improvement of GTFP. Hypothesis 2 (H2) is, thus, verified. The possible explanation for this is that with the establishment and improvement of the socialist market economic system, the market-oriented governance transformation plays an increasingly important role in the efficiency of the resource allocation among enterprises. Therefore, resources will flow into high efficiency enterprises over time, which contributes to the enhancement of the internal production efficiency of enterprises by the specialized division of labor

and the adoption of advanced green technology. This will then improve the overall green productivity of the whole society in the long term.

Finally, among the control variables, the coefficients of R&D investment were significantly positive, showing that R&D investment can promote the growth of GTFP. The credible explanation for the result is that the increase in R&D investment is beneficial to the enhancement of the innovation capacity of enterprises through the promotion of technological progress, thus improving production efficiency. The coefficients of FDI were always positive but not significant, indicating that FDI cannot significantly improve GTFP. This is due to the fact that local governments pay more attention to the quantity of FDI but ignore the quality of FDI. As a result, the introduction of foreign-invested enterprises with a high production efficiency and green technology is insufficient, which affects the improvement of regional productivity. However, export trade dependence and the factor endowment structure have a negative impact on GTFP. The conceivable explanation is that the low-level export product quality and the unreasonable resource allocation models hinder the transformation of the extensive economic development characterized by high input, high emissions, and low efficiency and, thus, inhibit the growth of GTFP.

### *4.4. The Analysis of Regional Heterogeneity Results*

Although the regression results of the full sample show a nonlinear U-shaped relationship between environmental regulation and GTFP, governance transformation can improve GTFP. Due to China's vast territories and the differences in the level of economic development among regions, whether the benchmark regression results are still tenable in different regions is, therefore, unknown. In order to solve this problem, the study divided the full sample into three sub-samples according to geographical locations and economic characteristics, which are the eastern region, the central region, and the western region. Then, we used the SYS-GMM method to estimate three sub-samples. The results are presented in Table 4.

The regression results of Models (1)–(6) in Table 4 show that the effects of environmental regulation on GTFP have regional differences. Specifically, the U-shaped relationship between environmental regulation and GTFP was significant in the eastern and western regions, but it was uncertain in the central region. The reasons can be summarized into three aspects. Firstly, cities in the eastern region have established third-party pollution control mechanisms and have obtained remarkable achievements. Compared to the other regions, industrial enterprises in the eastern region are more adaptable to stricter environmental standards. Secondly, following the findings by Song et al. [49], government subsidies will impair the positive effect of environmental regulation on technological innovation. In the central region, the R&D activities of environmental control are heavily dependent on government investments, so that the enterprise benefits from technological innovation investment are not enough to compensate for the additional production costs caused by environmental regulation. In addition, in order to stimulate economic development, local governments compete to set lower environmental regulatory standards to attract investment projects. For this reason, environmental regulation policies become a mere formality. Thirdly, in the western region, owing to the relatively backward economic development level, many enterprises are able to receive more favorable policies under the national implementation of the Western Development Strategy. With the strengthening of environmental regulation, more green funds provided by local governments are used to support enterprises to carry out green technology innovations, which improves the green production efficiency in the long term.


**Table 4.** Regression results of different regions with SYS-GMM.

Note: Standard errors in parentheses. \*\*\*, \*\*, and \* represent mean significance at the 1%, 5%, and 10% levels, respectively. The eastern region consists of Liaoning, Hebei, Tianjin, Beijing, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian, Guangdong, and Hainan. The central region consists of Heilongjiang, Jilin, Shanxi, Anhui, Jiangxi, Henan, Hubei, and Hunan. The western region consists of Shanxi, Gansu, Ningxia, Qinghai, Xinjiang, Sichuan, Yunnan, Guangxi, Guizhou, Chongqing, and Inner Mongolia.

Meanwhile, the regression coefficients of governance transformation in the eastern region, central region, and western region were 0.0202, 0.0621, and 0.0215, respectively, indicating that the promotional effect of governance transformation on GTFP in the central region is greater than that in the eastern region and western region. Because of the implementation of the Central Rise Policy, the state and the local governments have issued a series of policies to promote economic development, which stimulates the development of private enterprises in the central region. Under such circumstances, market-oriented governance transformation can effectively strengthen the efficiency of resource allocation among enterprises, which inspires the vitality of private enterprises with a high efficiency and promotes the realization of radical innovation, as well as improving the production efficiency in this region [44,46].

In addition, the influence of other control variables on GTFP also present regional differences. For example, the regression coefficients of export trade dependence in the eastern region were significantly negative, while those in the central and western regions were not significant. The regression coefficients of the factor endowment structure were significantly negative in the eastern and western regions, while those in the central region were not significant. The regression coefficients of FDI were significantly positive in the eastern region, while those in the central and western regions were not significant, mainly because the quality of FDI introduced in the eastern region is improved constantly and GTFP is also promoted. Moreover, the impact of R&D investment on GTFP in all regions was significantly positive, indicating that R&D investment contributes to the enhancement of enterprise innovation capacities, as well as improved production efficiency over time.

#### *4.5. The Analysis of the Robustness Test Results*

In order to ensure the reliability and validity of the baseline regression results, this paper used four ways to perform robustness tests. First, we used the dynamic panel threshold model to estimate the full sample. To be specific, according to the threshold variable, namely, the level of environmental regulation, the full sample was classified into a high group and a low group. The SYS-GMM method was then used to estimate the two groups simultaneously. Second, we adjusted the measurement pattern of environmental regulation. Specifically, the ratio of the total investment of industrial pollution control to the operating costs of industrial enterprises was selected as a substitute variable for environmental regulation. Then, the regression analysis was based on the adjusted full sample data. Third, the two-step SYS-GMM method was used to estimate the full sample. Compared with the one-step SYS-GMM, the two-step SYS-GMM can improve the estimation efficiency, as the weight matrix of the instrumental variables can be modified by the residual matrix of the one-step SYS-GMM. Fourth, we adjusted the sample interval for estimation. Specifically, the samples in 2003 and 2017 were excluded to eliminate the impact of the sample time selection on the estimation results. Thus, the provincial panel data from 2004 to 2016 were used for the re-estimation. The robustness test results are shown in Table 5.

**Table 5.** Robustness test results of impact of environmental regulation on GTFP.


Note: Standard errors in parentheses. \*\*\*, \*\*, and \* represents mean significance at the 1%, 5%, and 10% levels, respectively.

Columns 1 and 2 in Table 5 report the results of the panel threshold regression model, indicating that there is a nonlinear U-shaped relationship between environmental regulation and GTFP. Specifically, when the level of environmental supervision is lower than the threshold value (3.15), environmental regulation inhibits the growth of GTFP. If the level of environmental supervision is greater than 3.15, environmental regulation can facilitate the improvement of GTFP. Furthermore, the coefficients of governance transformation were both significantly positive in the low and high groups. Columns 3 and 4 in Table 5 show the regression results when the measurement of environmental regulation is replaced, demonstrating that the coefficients of the quadratic term of environmental regulation were positive and significant, confirming the nonlinear U-shaped relationship between environmental regulation and GTFP. The coefficients of governance transformation were significantly positive. Columns 5 and 6 in Table 5 show the estimation results of the twostep SYS-GMM, illustrating that the relationship between environmental regulation and GTFP is nonlinear and U-shaped. The coefficients of governance transformation were also significantly positive. Columns 7 and 8 in Table 5 report the regression results of

adjusting the sample interval, indicating that the nonlinear U-shaped relationship between environmental regulation and GTFP exists. The coefficients of governance transformation were positive. In addition, the regression results of the four robustness tests showed that the signs and significance of the coefficients of other control variables were consistent with the baseline empirical results. In conclusion, the robustness tests verified the existence of the nonlinear U-shaped relationship between environmental regulation and GTFP and that governance transformation can promote the growth of GTFP. Thereby, the baseline regression results of this study are robust and reliable.

#### *4.6. Further Discussion*

With the deepening of the socialist market economic reform, the role of the market mechanism in resource allocation has been increasingly strengthened. In essence, governance transformation reflects a market-oriented resource allocation reform, indicating that the governance model is changing from an administrative governance to an economic governance. Regarding the increasingly serious environmental problems, it is unknown if the market-oriented governance transformation can improve the effect of environmental regulation on GTFP and if the governance transformation can facilitate the realization of the Porter hypothesis and its influencing mechanisms. In order to solve these problems, the interaction term between governance transformation and environmental regulation was added into the model to investigate the influence of governance transformation on the Porter hypothesis. The regression results are shown in Table 6.

**Table 6.** The regression results of impact of governance transformation on the Porter hypothesis.


Note: Standard errors in parentheses. \*\*\*, \*\*, and \* represent mean significance at the 1%, 5%, and 10% levels, respectively.

The estimation results reported in columns 1–8 show that the coefficients of the interaction term between governance transformation and environmental regulation are significantly positive, indicating that, with the improvement of the environmental regulation intensity, governance transformation plays an increasingly important role in promoting the growth of GTFP. In other words, governance transformation can accelerate the realization of the improvement effects of environmental regulation on GTFP. Hypothesis 3 (H3) is, thus, verified. The reason for the result is probably that as market plays an increasingly important role in resource allocation, the efficiency of resource allocation among enterprises can be improved, which contributes to the inspiration of the innovative enthusiasm of private enterprises in the face of stricter environmental regulations. Radical innovation can, therefore, be realized quickly, and this will promote productivity over the long term. In addition, the signs and significance of the coefficients of environmental regulation, governance transformation, and the other control variables are also consistent with the baseline regression results.

#### **5. Conclusions**

How to protect the ecological environment and promote high-quality economic development has gradually become a research hotspot in recent years. Under such circumstances, this paper uses the GML method to measure the GTFP of 30 provinces in China from 2003 to 2017 and investigates the impact and mechanisms of environmental regulation on GTFP by using a dynamic panel model and the SYS-GMM method. The main conclusions are drawn as follows. First, under the dual pressures of severe resource constraints and environmental protection, GTFP is an accurate and meaningful indicator of the measurement of the level of economic development. Second, there is a nonlinear U-shaped relationship between environmental regulation and GTFP, indicating that the Porter hypothesis is verified in China. Most notably, the intensity of environmental regulation is still located on the left side of the U-shaped curve, which means that the promotional effect of environmental regulation on GTFP has not yet been realized fully. Meanwhile, the nonlinear U-shaped relationship shows significant regional differences. It is the western region that presents the highest level of significance, followed by the eastern region, but it is insignificant in the central region. Third, governance transformation can significantly improve GTFP by promoting enterprise technological innovation. Fourth, governance transformation can accelerate the realization of the Porter hypothesis by inspiring the innovation enthusiasm of private enterprises. In other words, governance transformation contributes to the achievement of the realization of the improvement effects of environmental regulation on GTFP. Moreover, R&D investment can significantly improve GTFP, while the impacts of export trade dependence and the factor endowment structure on GTFP are negative and significant. The influence of FDI on GTFP is insignificant.

In order to uphold the guidance of green development and to promote high-quality economic development by preserving the ecological environment, based on the above conclusions, we propose the following policy recommendations.

Firstly, China should firmly hold the conviction that lucid waters and lush mountains are invaluable assets and should always adhere to an ecological priority and green development. Specifically, the Chinese government should closely monitor energy conservation, emission reductions, and ecological environment protection, as well as optimizing the ecological environment of factor allocation and enhancing the value creativity of factor resources, such as labor, capital, and energy. Meanwhile, the state and local governments should formulate a diversified green government performance appraisal system, especially for the official promotion evaluation mechanism, in which the weight of environmental indicators should be considered. Moreover, local governments should strengthen exchanges and cooperation in environmental governance and should establish inter-jurisdiction joint preventions and control coordination mechanisms to improve the ecological environment.

Secondly, the Chinese government should establish a long-term, institutionalized, and standardized environmental regulation policy system that is coordinated with regional economic development as soon as possible. To be specific, the central government should further strengthen the overall planning of environmental regulation across regions and establish a coordinated and unified environmental supervision system, such as setting up an inter-regional transfer payment system characterized by an ecological compensation mechanism. At the same time, the government should vigorously expand environmental regulation tools and comprehensively use legal, economic, technical, and administrative means to establish a fair and diverse environmental regulation tool system. In addition, the state and local governments should enhance the coordination between regional environmental policies and regional economic development policies, as well as implementing flexible and differentiated environmental regulation policies to improve the effectiveness of environmental regulation over time.

Thirdly, China should strive to create a fair and trustworthy market environment and give full play to the decisive role of the market in resource allocation. Specifically, the government should accelerate the market-oriented reform of the factor price in the capital market, labor market, and land market, as well as improving the transparency of market transactions to eliminate the factor price distortion. Meanwhile, the government should strengthen anti-monopoly policies, reduce entry constraints to monopoly industries, and promote the reform of the profit distribution systems of state-owned enterprises. In addition, the government should avoid excessive intervention in the economy, eliminating the differential treatment in credit supplies, interest rates, and market access, as well as improving the official administrative efficiency and public service levels.

Finally, in order to realize the green development of China's economy, we should use the coordination effect of technological innovation, FDI, trade openness, and factor endowment. Specifically, (1) the state and local governments should increase their investment in R&D and should provide financial support for green technology innovation by setting up green development funds. In particular, the government should improve the incentive mechanisms for the commercialization of technological innovation achievements. (2) In order to fully exploit the spillover effect and pollution halo effects of FDI, the local governments should transform the FDI attraction mode from quantity to quality to attract high-quality technological FDI and green FDI based on local economic development circumstances. (3) China should continue to expand and optimize the foreign trade structure. It is important to accelerate the transition of the export trade from quantity to quality, such as reducing the proportion of export volume in low value-added and high pollution industries. Meanwhile, the import volume of new technologies, especially the green technologies, should be enlarged. (4) China should improve the factor market allocation and promote the free flow of capital, labor, and other factors. For example, the government should strengthen financial marketization reforms to effectively reduce the capital cost for enterprises and to deepen the Hukou (household registration) system reform to optimize labor allocations between urban and rural areas.
