Impacts of Environmental Pollution and Digital Economy on the New Energy Industry

Abstract

This study explores the impacts of environmental pollution and the digital economy on the new energy industry with panel data on 30 Chinese provinces from 2005 to 2020. Mean group regression was performed, and fully modified OLS and dynamic OLS were conducted to check the robustness of the results. The authors reached two conclusions: (1) environmental pollution exerts significant negative impacts on the new energy industry. For every 1% increase in environmental pollution, the level of development of the new energy industry drops by 0.1658%. In other words, environmental pollution levels down the new energy industry. (2) The digital economy produces significant positive impacts on the new energy industry. For every 1% increase in the digital economy, the level of development of the new energy industry rises by 0.4262%. That is, the digital economy levels up the new energy industry. Our policy recommendations place equal stress on both the prevention and the control of environmental pollution, strengthening digital infrastructure, enhancing the government’s digital governance and service capabilities, protecting consumer rights, and replacing conventional energy with new energy.

1. Introduction

History has shown us that high energy consumption makes a huge impact on the environment [1]. When traditional fossil fuels such as coal and oil are burned, they release carbon dioxide (CO₂) and sulfur dioxide (SO₂). CO₂ remains in the atmosphere for a long time and causes the greenhouse effect, while SO₂ forms atmospheric aerosols that may cause deaths if inhaled too much [2]. To prevent and control environmental pollution, many countries have launched innovation-driven strategies. They develop efficient new energy with low pollution and low emissions, seek energy transformation, and strive to build a low-carbon energy system [3]. New energy has become the fundamental means to solve problems caused by environmental pollution and meet growing energy demands for energy [4].

Energy, mostly inefficient fossil fuels, drives the rapid growth of China’s economy since the reform and opening up [5]. In recent years, new energy as a core substitute for fossil fuels is attracting increasing attention. In January 2022, China’s National Development and Reform Commission and National Energy Administration jointly released the “Opinions on Improving Institutional Mechanisms and Policy Measures for Green and Low-Carbon Energy Transition”, in which they stressed the need to properly handle the complementation, coordination or substitution among different types of energy at different stages of economic transformation, drive an optimized combination of coal and new energy, and coordinate the transition to green energy.

China’s new energy industry develops rapidly, with constantly growing installed capacity and amount of electricity generated. According to statistics from the National Energy Administration, in 2021, its installed capacity for new energy accounted for around 30% of total installed capacity, and the amount of electricity generated from new energy sources took up 18.5%, which was up 10.8% from 2015. In 2021, its installed capacity for wind power, solar power and biomass energy all ranked first in the world, and nuclear power ranked third. The amounts of electricity generated from these four sources were 3.5 times, 8.3 times, 3.1 times and 2.4 times those of 2015. As one of the emerging industries, the new energy industry is the main development direction of China’s energy sector. There are already studies on its driving forces. Xu and Lin (2018) suggested that the impacts of economic growth on the new energy industry in eastern China are greater than those in central China and western China [4]. Lin and Xu (2018) concluded that energy consumption structure has positive impacts on the new energy industry in the short term, but limited impacts on it in the long run [6].

Meanwhile, supported by technological innovations, China’s digital economy develops rapidly. According to the White Paper on Global Digital Economy (2022), China, the US and Europe form a tripartite pattern of the digital economy. In 2021, China’s digital economy was the second largest in the world. As shown by the White Paper on China’s Digital Economy (2022), it maintained an average annual growth of 15.9% since 2012, significantly higher than its average GDP growth during the same period. Driven by the “Broadband China” strategy, 5G and industrial internet, the size of China’s digital economy reached 45.5 trillion CNY in 2021, creating a nominal year-on-year growth of 16.2%.

Some have discussed the impacts of the new energy industry on environmental pollution [7], but environmental pollution may push back and promote the development of this industry. China is expanding and strengthening its digital economy, which through digital technologies enables technological progress in production, improves technological innovation, and increases the efficiency and precision of new energy. Empowered by digital technologies, manufacturers adjust their energy input to a greater extent, thereby generating substitution effects and increasing energy input [8]. Therefore, it is of great significance to study the impacts of environmental pollution and the digital economy on the new energy industry. This study probes into this topic in the context that China attaches great importance to sustained ecology, promotes the development of the digital economy, and vigorously develops the new energy industry. It provides a theoretical basis for the Chinese government and even governments around the world to formulate policies on the new energy industry, which is the unique value of this study.

This paper expands research on environmental pollution, the digital economy and the new energy industry. Its main contributions are shown as follows: This study expands on research on environmental pollution, the digital economy and the new energy industry. Firstly, this study, based on data on 30 Chinese provinces from 2005 to 2020, and taking environmental pollution and the digital economy as the main factors influencing the new energy industry, is the very first research study on the impacts of environmental pollution and the digital economy on the new energy industry. Secondly, along with the rapid development of China’s digital economy, data empowerment can reduce environmental pollution and promote the development of the new energy industry. Therefore, it is of great significance to discuss the impacts of environmental pollution and the digital economy on the new energy industry. Thirdly, robust second-generation econometric tests were conducted, which covered cross-sectional dependence tests, second-generation panel unit root tests, cross-sectionally augmented IPS tests, slope heterogeneity tests, and mean group tests. To ensure the robustness of our results, we also performed fully modified OLS and dynamic OLS. Lastly, our research conclusions provide theoretical and empirical references for the new energy industry.

The rest of this paper is organized as follows. Section 2 provides a literature review. Section 3 presents the models and data used. Section 4 shows discussions of empirical results. Section 5 offers conclusions and policy implications.

2. Literature Review

2.1. Research on the Relationship between Environmental Pollution and the New Energy Industry

Environmental pollution refers to all kinds of harmful substances released to nature through direct or indirect channels which destruct the ecosystem and endanger human health [9]. Most of the existing literature has focused on the impacts of energy consumption or renewable energy consumption on environmental pollution. Zhao et al. (2021) indicated that energy consumption increases environmental pollution in China [10]. Ibrahim et al. (2022) concluded that environmental technologies and renewable energy curb the surge of carbon emissions [11]. Li et al. (2022) worked on the interplay among renewable energy, energy innovation and environmental pollution. They concluded that new energy and energy innovation validly reduce environmental pollution [12]. The development of renewable energy such as solar energy, wind energy and hydropower has significant implications for addressing climate change [13,14]. Rahman and Alam (2022) showed that in Australia, the use of industrial and non-renewable energy increased CO₂, while the square of industrial and renewable energy use reduced CO₂ emissions [15]. Li et al. (2022) found that the consumption of renewable energy presents an inverted-N relationship with haze pollution and carbon emissions. The overall spatial spillover effect of REC is larger than its local effect [16]. Isık et al. (2023) pointed out that the consumption of renewable energy accelerated the reduction of CO₂ emissions [17].

According to the Classification of Strategic Emerging Industries (2018) published by China’s National Bureau of Statistics, the new energy industry is represented by photovoltaic solar energy, wind energy, nuclear energy, biomass energy and smart grid. Few studies linked environmental pollution with the new energy industry, most of which centered on the impacts of the latter on the former or carbon emissions. Most scholars believed the new energy industry is green and technology-intensive, whose advantages are green, efficient and energy-saving compared with traditional industries [18]. Developing new energy has become a fundamental way to solve the contradiction between environmental pollution and growing energy consumption [19,20]. Fang et al. (2023) probed into the relationship between the new energy industry and carbon emissions and identified a reverse U-shaped relationship between them [21]. Nuclear energy as a replacement for fossil fuels plays a key role in achieving a carbon peak [22,23]. Yi and Yu (2020) held that accelerating the building of new energy industrial parks can improve regional performance in pollution control [24]. However, the new energy industry, as a capital and technology-intensive industry, requires large investments in R&D and energy transformation in the early stages, so it poses higher costs than traditional fossil fuels [19]. Li et al. (2023) showed that the expansion of the new energy industry in the early stages leads to an increase in carbon emissions [18].

2.2. Research on the Relationship between the Digital Economy and the New Energy Industry

The G20 Digital Economy, Development and Cooperation Initiative released at the G20 Hangzhou Summit in 2016 proposed that the digital economy is a broad range of economic activities that include using digitized knowledge and information as key factors of production, modern information network as an important activity space, and effective use of information and communication technology as an important driver of productivity growth and economic structural optimization. The digital economy is the main form of the economy following the agricultural economy and the industrial economy [25]. There are plenty of studies on its impacts on energy consumption and energy efficiency, but no consensus has been reached. Some suggested that digital development effectively reduces energy demand and improves energy efficiency [26,27]. Technological advancements in the digital sector improve the automation, connectivity and flexibility of production, manufacturing and consumption [28], which is conducive to improving energy efficiency [29,30]. Xu et al. (2022) found that digitization increases information asymmetry, reduces resource classification, improves the efficiency of resource allocation, eases distortion of industrial structure, and makes it easier for industries to adopt lean management in production and intelligently monitor energy consumption in real time, thereby improving energy efficiency [31,32,33]

However, some studies show that digital development speeds up the popularization of digital infrastructure, promotes the emergence and application of new communication devices, and stimulates people’s consumption of electronic appliances, thus increasing dependence on energy [34,35]. Ren et al. (2021) held that digital development increases China’s energy consumption because more devices and servers are needed to accommodate large amounts of data [36]. If improperly managed, these devices may lead to a significant net increase in energy demand [37]. Some studies indicate that digitization reduces energy demand in the short term, but increases it in the long run due to the rebound effect [38].

Some other studies probed into the impacts of the digital economy on energy structure and concluded that digitization reduces energy consumption and optimizes energy structure [31,39]. For example, Madaleno et al. (2022) showed that digital technology promotes energy transformation [40]. Chen (2022) believed the digital economy is a mode of green economy mainly applied in traditional economic sectors, which is characterized by high efficiency, high output and low emissions. The digital economy features widespread application of digital technologies such as the internet, blockchain and big data [41]. Digital technologies are characterized by high electricity consumption, which can boost the development of clean energy [42]. Digitization effectively addresses technological challenges brought by clean energy, stimulates the development of clean energy, and thus optimizes energy structure [43,44]. Shahbaz et al. (2022) demonstrated that the digital economy stimulates the transition to renewable energy by improving the government’s governance capabilities [45]. Zheng and Wang (2021) concluded that digital development facilitates the development of renewable energy in developing countries [46]. Wang et al. (2023) discussed the impacts of the digital economy on renewable energy generation in Asian countries from 2003 to 2019 and found that the digital economy exerts positive impacts on renewable energy generation [47]. Emerging technologies can greatly reduce the cost of new energy [48].

The above research sheds some light on this study, but has some shortcomings. First, the majority of them discuss the impacts of energy consumption or renewable energy consumption on environmental pollution or carbon emissions, with few focusing on the impacts of environmental pollution on the new energy industry. Secondly, there are many studies on the impacts of the digital economy on energy consumption, energy efficiency and energy structure, but research on its impacts on the new energy industry is missing. Finally, in terms of research methods, the existing literature usually assumes that the impact of environmental pollution and the digital economy on the new energy industry is the same among all individuals, so as to establish the panel data model of the homogeneity regression coefficient, rarely considering the differences among different individuals. There are fewer studies considering the cross-sectional correlation between individuals of panel data. Therefore, this study fills the gaps in the literature and estimates the impact of environmental pollution on the new energy industry combined with the digital economy. Finally, regarding research methods, the existing literature usually assumes that environmental pollution and the digital economy exert the same impacts on all players of the new energy industry, and builds the panel data model of homogeneity regression coefficient on this basis, rarely considering differences among industry players. Fewer studies probe into cross-sectional correlation among individuals in panel data. This study fills the gaps in this regard by estimating the impacts of environmental pollution and the digital economy on the new energy industry. Mean group estimation was used. It is assumed that all the coefficients and perturbation terms across panels are different, and the least squares estimation method is applied to each individual to obtain their regression coefficients. All the regression coefficients were averaged, which revealed a robust consistency of long-run coefficients. It is not only an important supplement to existing research, but also provides a useful reference for developing the digital economy and the new energy industry and for protecting the environment in China and even the world at large.

3. Method and Data

3.1. Econometric Model

This study mainly examines the impacts of environmental pollution and the digital economy on the new energy industry. In addition to environmental pollution and the digital economy, energy tax policies also have an impact on it. Tax on the new energy industry can be lowered to promote the development of the industry. In addition, environmental regulation is also an important factor affecting the development of the industry. Therefore, reasonable government regulation and restraint means can be taken to strengthen supervision and control, so as to give full play to the role of environmental regulation in the new energy industry. The following model was built:

LNE_it = f (LEP_it, LDE_it, LFIX_it, LEG_it, LCO_2it)

(1)

in which i represent province, t time (from 2005 to 2020), LNE, LEP, LDE, LFIX, LEG and LCO₂ the logarithms of the new energy industry, environmental pollution, the digital economy, investments in fixed assets, environmental regulations and carbon dioxide.

The regression equation is:

LNE_it = δ_1it + δ_2itLEP_it + δ_3itLDE_it + δ_4itLFIX_it + δ_5itLEG_it + δ_6iLCO_2it + ε_it

(2)

in which δ means parameter and ε error term.

3.2. Econometric Methodology

We first performed cross-sectional dependency (CSD) tests, slope heterogeneity (SH) tests, second-generation unit root tests on stationarity, and cointegration tests, then evaluated the impacts of environmental pollution and the digital economy on the development of China’s new energy industry using mean group (MG) panel regression.

3.2.1. CSD Tests and SH Tests

It is important to check CSD before performing a metrological analysis. Negligence of CSD may lead to biased panel data estimates and inaccurate test statistics. The CSD test used in this paper was developed by Pesaran (2004) [49]. Its formula is [50]:

$$CSD=\sqrt{\frac{2T}{N(N-1)}}\left({\displaystyle \sum _{i=1}^{N-1}{\displaystyle \sum _{j=i+1}^N\rho ij}}\right)N(0,1)i,j$$

(3)

where N denotes sample size, i time period, and ρ_ij the pairwise correlation coefficient of the OLS estimate of each cross-section i.

SH is another problem with panel data that should be solved. In this paper, Pesaran and Yamagata’s (2008) tests were conducted to check SH in the model [51].

$$\tilde{\Delta }adj=N^{\frac{1}{2}}\left(\left.\frac{N^{-1}\tilde{S}({\tilde{z}}_{it})}{\sqrt{\mathrm{var}\tilde{S}({\tilde{z}}_{it}})}\right)\right.$$

(4)

where the mean $E({\tilde{z}}_{it})=k$, and variance $\mathrm{var}({\tilde{z}}_{it})=2k(T-k-1/T+1)$.

3.2.2. Unit Root Test

Ensuring the stationarity of panel data is also very important for econometric analysis because non-stationary variables can result in spurious regression. Therefore, unit root tests on panel data are necessary. Since first-generation unit root tests cannot solve the CSD problem, second-generation unit root tests were carried out to check data stationarity. Among other methods, the cross-sectionally augmented DF (CADF) unit root test (Pesaran, 2007) is the most commonly used. Its regression equation is shown below [52]:

$$\Delta Y_{i,t}={\alpha }_i+b_i{Y}_{i,t-1}+c_i{\overline{Y}}_{t-1}+d_i\Delta {\overline{Y}}_t+{\epsilon }_{i,t}$$

(5)

$\overline{Y}=\frac{1}{N}{\displaystyle {\sum }_{i=1}^N}{Y}_{i,t}$, $\Delta {\overline{Y}}_t=\frac{1}{N}{\displaystyle {\sum }_{i=1}^N}\Delta Y_{i,t}$, in which ${\epsilon }_{i,t}$ is an error term.

By conducting CADF tests, we obtained the results of country-specific horizontal sections and cross-sectionally IPS (CIPS) statistics. CIPS was performed to obtain overall panel statistics based on cross-section averages [53].

The equation for the unit root test of CIPS panel data is:

$$CIPS=\frac{1}{N}{\displaystyle \sum _{i=1}^{N}CAD{F}_i}$$

(6)

3.2.3. Cointegration Test

Long-term cointegration among variables was checked with the second-generation Westerlund test. In dealing with CSD among variables, the Westerlund test is more effective than any other test [54]. It has 4 statistics (Gt, Ga) and panel statistics (Pt, Pa). Since our data is a mixture of I(0) and I(1) variables, we also performed first-generation unit root cointegration tests proposed by Pedroni [55,56], Johansen-Fisher and Kao (1999) [57,58].

3.2.4. MG Estimator

To evaluate the impacts of environmental pollution and the digital economy on the new energy industry, the MG estimator was adopted [59]. It is used to calculate long-run parameters with the mean of long-run coefficients of the ARDL model for each province [60], as performed by Kusairi et al. (2019) [61]:

$$Y_{it}={\alpha }_i+{\gamma }_i{Y}_{it-1}+{\beta }_i{X}_{it}+u_{it}$$

(7)

where $Y_{it}$ is the dependent variable, $X_{it}$ the explanatory variable, and ${\beta }_i$ the estimated coefficient of a certain province.

The long-run parameter of province i is [61]:

$${\theta }_i=\frac{{\beta }_i}{1-{\gamma }_i}$$

(8)

MG estimator is expressed as follows [61]:

$$\stackrel{⌢}{\theta }=\frac{1}{N}{\displaystyle \sum _{i=1}^N{\theta }_i}\stackrel{⌢}{\alpha }=\frac{1}{N}{\displaystyle \sum _{i=1}^N{\alpha }_i}$$

(9)

MG allows variation in intercepts, short-run parameters and variance of error terms across the group. It ensures robust consistency for long-run coefficients [55].

Fully modified OLS and dynamic OLS were adopted to test robustness. The former can overcome spurious regression and endogeneity, and produce reliable estimates of small samples. The latter can correct the possible endogeneity of independent variables [62].

3.3. Variable Measurement and Data Sources

In light of data availability, we selected panel data from 30 provinces in China’s Mainland from 2005 to 2020. Details of variables and data sources are described below.

3.3.1. Dependent Variable

The core explanatory variable in this study is the new energy industry. According to the classification of strategic emerging industries by China’s National Bureau of Statistics (2018), new energy covers nuclear power, wind power, solar power and biomass energy. Drawing on Fang et al. (2023) [21], this study presents the level of development of the new energy industry with the proportion of electricity generated from new energy (nuclear power, wind power, solar power and hydropower) to the total amount of energy generated.

3.3.2. Key Independent Variables

One of the key explanatory variables is environmental pollution, which is represented by SO₂ emissions/GDP following the methodology used by Zhao et al. (2021) [10]. Another key explanatory variable is the digital economy. As there is no readily available data for the digital economy, we built an integrated index system to evaluate it based on three sub-indicators: digital infrastructure, digital industrialization and digital governance. The entropy weight method was used to calculate the digital economy index of the 30 provinces. The index system is shown in

Table 1. Indicators of the digital economy.

Dimension	Indicator	Measure	Attribute
Digital infrastructure	Broadband	No. of households connected to broadband (10,000)	+
	Mobile phone	No. of mobile phone users (10,000)	+
Digital industrialization	Telecommunications business	Per capita Telecom business revenue (CNY)	+
Digital governance	R&D intensity	Funding for R&D/GDP (%)	+
	Number of authorized patents	No. of authorized invention patents, utility model patents and appearance design patents per 10,000 people	+

.

3.3.3. Control Variables

Control variables are environmental regulation and CO₂ emissions. Environmental regulation is represented by the proportion of investments in industrial governance to the costs of goods sold. CO₂ emissions were calculated based on fossil fuels consumed as shown in each province’s energy balance table, through heating value, carbon content and oxidation factors with reference to existing research [63].

3.4. Data Sources

Data for this study mainly comes from the China Energy Statistical Yearbook, the China Statistical Yearbook, the China Statistical Yearbook on Environment, the China Statistical Yearbook on Science and Technology, the China Statistical Yearbook on Land and Resources, and statistical yearbooks of 30 Chinese provinces. The missing data were filled through interpolation; that is, according to the data characteristics of the sample, the missing data was calculated from the average of adjacent data. The basic attributes of variables are shown in

Table 2. Basic characteristic values of variables.

Variable	Symbol	Mean	Max.	Min.	Standard Deviation	Observations
New energy industry	NE	24.5263	91.8528	0.0278	23.5185	480
Environmental pollution	EP	63.8844	686.1844	0.0488	85.8298	480
Digital economy	DE	0.1754	0.8005	0.0111	0.1378	480
Investment in fixed assets	FIX	13,553.64	59,185.99	329.81	12,590.39	480
Environmental regulation	EG	0.1542	0.9034	0.0025	0.1451	480
Carbon dioxide emission	CO2	10.8142	54.3051	1.9709	8.1288	480

.

4. Empirical Results and Discussions

4.1. Test of Multicollinearity

The variance inflation factor was used to test multicollinearity among variables. The results are shown in

Table 3. VIF tests for multicollinearity.

Variable	VIF	1/VIF
LEP	3.29	0.3036
LDE	5.97	0.1675
LFIX	2.89	0.3465
LEG	2.18	0.4579
LCO2	1.20	0.8307
Mean	3.11

. The VIF values of all variables are less than 10, which indicates there is no multicollinearity among variables [64].

4.2. Cross-Sectional Dependence Test and Slope Heterogeneity Test

Table 4. Results of Pesaran (2015) [65] CSD test.

Variables	CSD-Statistics	p-Value	corr	abs(corr)
LNE	40.68	0.000	0.488	0.540
LEP	82.29	0.000	0.986	0.986
LDE	81.40	0.000	0.976	0.976
LFIX	76.35	0.000	0.915	0.915
LEG	39.31	0.000	0.471	0.485
LCO2	63.91	0.000	0.766	0.793

presents the results of CSD tests on the 30 Chinese provinces, namely, all variables passed significance tests. This rejects the null hypothesis of no CSD. In other words, there is CSD among variables [65]. The economies of the 30 Chinese provinces are interrelated and interdependent. If one province is economically impacted, other provinces will also suffer shocks.

According to

Table 5. Pesaran and Yamagata (2008) [51] SH test results.

	Value
$\tilde{\Delta }$	6.278 *** (0.000)
$\tilde{\Delta }$ Adjusted	8.371 *** (0.000)

Note: the null hypothesis is slope homogeneity. *** p < 0.001. p-values are reported in parentheses.

, SH tests demonstrate $\tilde{\Delta }$ and $\tilde{\Delta }$ adjusted rejected the null hypothesis that slope coefficients are homogeneous, and accepted the alternative hypothesis that slope coefficients are heterogeneous.

4.3. Panel Unit Root Test and Cointegration Test

Table 6. CIPS panel unit root test results.

Variables	At Level	First Differences
LNE	−4.497 *** (0.000)	-
LEP	−2.409 *** (0.008)	-
LDE	−0.691 (0.245)	−1.391 * (0.082)
LFIX	−1.525 * (0.064)	-
LEG	−3.104 *** (0.000)	-
LCO2	−0.682 (0.248)	−1.749 ** (0.040)

Note: * p < 0.1. ** p < 0.05. *** p < 0.001. p-values are reported in parentheses.

reports the results of the second-generation unit root CIPS test following Pesaran (2007) [52]. Except for environmental pollution and CO₂ emissions, variables such as the level of development of the new energy industry and the digital economy are stationary at the level. Environmental pollution and the 1st difference of CO₂ emissions are stationary. Based on the results of CIPS tests, cointegration tests were conducted to further examine whether there is a long-term relationship among variables [50].

Firstly, Westerlund [54,66] cointegration tests were performed to check the long-term relationship among variables. According to

Table 7. Results of Westerlund cointegration tests.

Westerlund Test (2005) [66]	Statistic	p-Value
Variance ratio	1.6872	0.0458
Westerlund test (2007) [54]
Statistics	Value	Z-Value	p-Value
Gt	−3.518	−2.997	0.001
Ga	−1.133	10.778	1.000
Pt	−13.039	2.228	0.987
Pa	−2.410	8.057	1.000

, the variance ratio of Westerlund (2005) passed the significance test, which indicates a long-term cointegration among variables. Gt statistics also passed the significance test, which rejected the null hypothesis of no cointegration at the 1% level of significance. This means all variables are cointegrated.

In addition, we examined the long-term co-integration among variables using first-generation unit root tests proposed by Pedroni (2004), Johansen-Fisher and Kao (1999). The results are reported in

Table 8. Results of panel cointegration tests.

Pedroni’s Residual Cointegration Test
Within-Dimension	Statistics	p-Value	Weighted Statistics	p-Value
Panel v-stat	−1.4556	0.9273	−4.1694	1.0000
Panel rho-stat	3.0795	0.9990	1.3884	0.9175
Panel PP-stat	−3.9680	0.0000	−6.4967	0.0000
Panel ADF-statistic	−5.2942	0.0000	−6.4964	0.0000
Group rho-stat	3.6217	0.9999
Group PP-stat	−10.1229	0.0000
Group ADF-statistic	−7.9358	0.0000
Johansen-Fisher panel co-integration test
Cross sections	Statistics	p-Value	Statistics	p-Value
None	1112.0	0.0000	687.3	0.0000
At most 1	541.5	0.0000	286.7	0.0000
At most 2	313.8	0.0000	185.6	0.0000
At most 3	184.9	0.0000	104.1	0.0004
At most 4	143.8	0.0000	108.7	0.0001
At most 5	116.3	0.0000	116.3	0.0000
Kao’s (1999) Residual Cointegration Test			Statistics	p-Value
Modified Dickey–Fuller t			5.0484	0.0000
Dickey–Fuller t			7.6473	0.0000
Augmented Dickey–Fuller t			9.6703	0.0000
Unadjusted modified Dickey–Fuller t			7.5270	0.0000

Note: the null hypothesis is that variables are not cointegrated, based on the hysteretic selection of SIC.

. Pedroni cointegration tests showed 6 among the 11 data passed the significance test and all p-values were significant in Johansen-Fisher and Kao tests. To sum up, the results of the three tests are consistent, which provides strong support for the existence of long-term cointegration among variables.

4.4. Results of MG Estimation

We conducted MG regression as the benchmark regression, the results of which are shown in

Table 9. MG estimation results.

Variables	Coef.	Std. Err.	Z	p-Value
LEP	−0.1658 **	0.0768	2.16	0.031
LDE	0.4262 *	0.2384	1.79	0.074
LFIX	0.2606 *	0.1459	1.79	0.074
LEG	0.1011 *	0.0516	1.96	0.050
LCO2	−0.5124 **	0.257	1.99	0.046
Constant	3.0436 **	1.2537	2.43	0.015

Note: * p < 0.1.** p < 0.05.

. The coefficient of environmental pollution is significantly negative, which indicates environmental pollution lowers the level of development of the new energy industry. The coefficient of the digital economy is also significantly negative, which means the digital economy effectively promotes the growth of the new energy industry. The coefficient of investments in fixed assets is significantly positive. This demonstrates that increasing investments in fixed assets in China boosts the development of the new energy industry. The coefficient of environmental regulation is also significantly positive. This shows that strengthening pollution control reduces the negative impacts of environmental pollution on the new energy industry. In addition, the coefficient of CO₂ is significantly negative. That is, reducing carbon emissions is an important way to foster the development of the new energy industry. In addition to environmental pollution and the digital economy, energy tax policies have an impact on the new energy industry. Tax on the industry can be lowered to promote the development of the industry. Environmental regulation is also an important factor affecting the development of the new energy industry. Reasonable government regulation and restraint means can be taken to strengthen supervision and control, so as to give full play to the role of environmental regulation in promoting the development of the new energy industry.

4.5. Robustness Testing

4.5.1. Robustness Check 1: Replace Key Variable

The key explanatory variable environmental pollution was replaced with the proportion of nitrogen oxide emissions to GDP as its proxy variable to conduct MG estimation again. The results are displayed in

Table 10. Robustness check I.

Variables	Coef.	Std. Err.	Z	p-Value
lnEP	−0.1966 **	0.0967	2.03	0.042
lnDE	0.5704 ***	0.1942	2.94	0.003
lnFIX	0.2830	0.2038	1.39	0.165
lnEG	0.0837 *	0.0434	1.93	0.054
lnCO2	−0.3210	0.3322	0.97	0.334
Constant	1.0361	1.52841	0.68	0.498

Note: * p < 0.1.** p < 0.05. *** p < 0.001.

. The coefficient of environmental pollution is still significantly negative, namely, that environmental pollution levels down the new energy industry. The coefficients of other variables are consistent with those in Table 9. The key explanatory variable of the digital economy and control variables passed significance tests. That is to say, the regression results in Table 9 are robust [45].

4.5.2. Robustness Check 2: Alternative Estimation

We used fully modified OLS and dynamic OLS estimators to further check the robustness of results, whose results are shown in

Table 11. Robustness check II.

Tests	FMOLS		DOLS
Variables	Coef.	p-Value	Coef.	p-Value
lnEP	−0.1804 ***	0.0003	−0.1728 **	0.0411
lnDE	0.2704 ***	0.0071	0.3221 **	0.0461
lnFIX	0.3584 ***	0.0000	0.2936 **	0.0158
lnEG	0.0564 **	0.0475	0.0324	0.4997
lnCO2	−0.4295 **	0.0113	−0.2623	0.2804
R2	0.9305		0.9276

Note: ** p < 0.05. *** p < 0.001.

. The estimations showed that the coefficient of the key explanatory variable environmental pollution is significantly negative, while that of the digital economy is significantly positive. This confirms the robustness of our baseline regression.

4.5.3. Robustness Check 3: Replace Control Variable

We also carried out robustness tests using per capita investments in industrial pollution control as a proxy for control variable environmental regulation. The MG regression results are presented in

Table 12. Robustness check III.

Variables	Coef.	Std. Err.	Z	p-Value
lnEP	−0.1839 **	0.0771	2.38	0.017
lnDE	0.3923 *	0.2353	1.67	0.096
lnFIX	0.2048	0.1495	1.37	0.171
lnEG	0.0688 **	0.0336	2.05	0.041
lnCO2	−0.5086 *	0.2703	1.88	0.060
Constant	3.0378 **	1.3187	2.30	0.021

Note: * p < 0.1.** p < 0.05.

. The coefficient of environmental pollution was greater than 0, while that of the digital economy was smaller than 0, both passing significance tests. The coefficient of environmental regulation is significantly positive, which confirms the robustness of our benchmark regression.

5. Conclusions and Policy Implications

5.1. Conclusions

China is developing a digital economy, but there is a lack of attention to the impacts of the digital economy and environmental pollution on the new energy industry. This paper explores the impact of environmental pollution and the digital economy on the development of the new energy industry based on panel data from 30 provinces in China from 2005 to 2020. MG regression estimation was carried out. Fully modified OLS and dynamic OLS were employed to check robustness. We reached the following conclusions:

(1)

Environmental pollution has significant negative impacts on the new energy industry. For every 1% rise in environmental pollution, the level of development of the new energy industry drops by 0.1658%. In other words, environmental pollution hinders the growth of the new energy industry, and the results remained robust after the measure of environmental pollution was changed;

(2)

The digital economy exerts significant positive impacts on the new energy industry. For every 1% rise in the digital economy, the level of development of the new energy industry moves up by 0.4262%. That is, the digital economy boosts growth in the new energy industry. Data empowerment should be emphasized to boost technological innovation and thus improve the development of the new energy industry;

(3)

FMOLS and DOLS were conducted to further a robustness test. Both showed that the coefficient of environmental pollution was significantly negative, while the coefficient of the digital economy was significantly positive, which confirmed the robustness of baseline regression.

5.2. Policy Implications

In order to foster the development of the new energy industry, we proposed the following implications for policymakers based on the empirical results of this study.

5.2.1. Equally Emphasize the Prevention and the Control of Environmental Pollution

To this end, the following measures can be taken. (1) Increase investments in pollution control. Improved pollution management can not only reduce pollution, but also effectively enhance environmental regulation. The empirical analysis herein shows that environmental regulation promotes the development of the new energy industry. Therefore, increasing investments in pollution control can boost the development of the industry while reducing pollution. (2) Enhance the environmental awareness of enterprises and the public. In addition to soft measures such as publicity, hard measures such as laws, regulations and fines can be used to curb pollution. (3) Include environmental pollution in the evaluation of local government performance. In this way, local governments would control environmental pollution while pursuing economic growth, and thus preventing pollution.

5.2.2. Improve the Digital Economy

(1) Strengthen the construction of digital economy infrastructure. Due to the existence of an income gap, there is a large gap in digital infrastructure between urban and rural areas. Cities have sound digital infrastructure, but there is still room for improvement in rural digital infrastructure. For example, a small proportion of rural residents are able to access broadband. Therefore, it is necessary to strengthen digital infrastructure in China’s rural areas. One of the methods is to lower the price of broadband to increase rural residents’ access to broadband. (2) Improve governance of the digital economy. It is necessary to enhance the government’s digital governance and service capabilities and protect consumer rights. (3) Improve technological innovations. The digital economy is supported by technology. To improve its development, efforts should be made to improve technological innovation, strengthen research on key technologies, and promote the commercialization of technological achievements.

5.2.3. Optimize Energy Structure and Substitute New Energy for Traditional Energy

Traditional fossil fuels such as coal and oil are the main sources of carbon emissions. As mentioned above, CO₂ levels down the new energy industry. Therefore, it is important to increase the efficiency of fossil fuels on the one hand, and expand the use of renewable energy such as hydropower and wind power on the other hand.

5.2.4. Promote the Development of the New Energy Industry in Rural Areas

China’s rural areas are rich in wind, solar, biomass and other new energy resources, but there is a low utilization rate of them. Technical support should be given to developing new energy in vast rural areas following the basic principles of adaptation to local conditions and rational development, thereby increasing the supply of new energy. In addition, China has a large rural new energy market. The government should improve preferential policies, increase capital investment, and encourage rural residents to use new energy by means of subsidies to drive the development of the new energy industry.

This study has some limitations. It only considers 30 provinces in China and does not explore the impact of the digital economy and environmental pollution on the new energy industry in each city. Future research can examine the situation of each city. In addition, data are limited to the years 2004 to 2020. However, it provides a new dimension for future research.