Does China’s Regional Digital Economy Promote the Development of a Green Economy?

Abstract

As countries worldwide are pursuing green development, assessing whether the digital economy as a new economic engine can help us achieve new breakthroughs is of great research value. China, being the largest resource consumer in the world but with a rapidly developing digital economy, can offer us a special view on this question. Using China’s provincial panel data from 2010 to 2020, this study comprehensively measures the development of the digital economy from four dimensions and empirically examines the impact of digital economy development on the green economy based on the super efficiency SBM-GML model. The results show that: first, digital economy development has a significant positive effect on promoting a green economy; second, there are regional differences in both the digital economy and the green economy in China, with the development in the southern region better than that in the northern region; third, the environmental regulation has a double-threshold effect on the relationship that we assessed. The findings in this study highlight the importance of digital economic development in driving the growth of the real economy and are of value to the development of a green economy in the new period.

1. Introduction

The global economy suffered a heavy blow following the pandemic of COVID-19. However, China was the only one to achieve positive economic growth, with a GDP increase of 2.3% in 2020. This was closely related to the vigorous development of the digital economy. Epidemic prevention and control, distance education, remote medical system, and e-commerce all paved the ways for economic recovery. According to the data from the Chinese Academy of Information Technology, the global digital economy accounted for 43.7% of the global GDP in 2020, becoming an important impetus for global economic development. The digital economy has been regarded as a “new engine” for economic growth. On the other hand, green economy development has also become a crucial strategic goal in the global context. Nonetheless, as the largest resource consumer in the world today, China is still unable to effectively wean itself off its heavy dependence on energy and the environment [1]. The neoclassical economic growth theory argues that the growth of total factor productivity is an important source of sustainable economic growth [2]. As the pollution emission caused by economic development has formed a huge obstacle to sustainable development, the growth of green total factor productivity, the target of pursuing green development, is becoming a new trend for not only China, but also the rest of the world. China has proposed to achieve a target of 10% of GDP for core digital economic output by 2025, by which time the digital economy will then be expected to take up half of China’s economy. At present, many prickly economic structural problems in China are rooted in the distorted allocation of resource elements caused by the rough development in the early stage. Therefore, whether the digital economy offers a new path to the green economic development is undoubtedly a meaningful and practical problem of great research value.

Extant studies have found that the development of the digital economy has a positive driving effect on firms, industries, and macroeconomy [3,4]. The main functions of the digital economy are centered around optimizing the industrial structure, reducing transaction costs, improving the efficiency of economic organization and allocation, reducing pollution emissions, and promoting economic transformation [5,6,7]. In terms of the green economic development with the dual goals of achieving economic growth and improving the ecological environment [8], green and high-quality development has become a growing concern in recent academic research [9,10]. As Agion et al. (1998) [11] suggest, if a constant stream of innovation is maintained the productivity of innovation over time can drive the economic equilibrium point outward and capture more green output in a new stage of development. Therefore, technological innovation is a fundamental way to achieve green development under limited ecological and environmental constraints [12,13]. Current research provides us with good theoretical and empirical evidence to further the research on the digital economy and the green economy.

However, there are some limitations in extant studies. Most literature on the digital economy focuses on its development trends and strategies. The indicator system constructed for it usually ignores the time frame and availability of digital data, making it difficult to conduct in-depth empirical research [14,15]. Scholars often select single or very few representative indicators to measure the development of the digital economy [16,17,18], lacking a comprehensive perspective. In terms of the literature on the green economy, current research mainly focuses on the connotation and measurement of the green economy [19,20,21]. Scholars generally use the non-parametric DEA or extended DEA model to measure the green economy, which has some defects in multi-dimensional dynamic comparison in time and space [22,23]. Moreover, existing studies tend to overlook the effect of environmental regulation on the relationship of the digital economy and the development of the green economy.

To fill in the gaps in extant research, our study aims to first improve current measurement approaches on both the digital economy and the green economy. Then, we attempt to investigate the impact of the digital economy as a new type of production factor on the green economy. Furthermore, we examine whether there is a nonlinear “threshold” of the green characteristics of the digital economy. By answering these questions, our study may contribute to the following aspects: First, the development level of a digital economy is assessed more comprehensively based on four dimensions of digital industrialization, data value, industry digitalization, and digital governance, which matches the real world more closely, and regional differences in digital economy development levels are also examined. Second, both the DEA-SBM model and the Malmquist–Luenberger productivity index are used to classify and calculate the static and the dynamic total factor productivity. Third, a kernel density map is used to present the regional differences in China in both the digital economy and the green economy. Fourth, we verify the network effect of the digital economy on the green economy by building a threshold effect model and considering multiple thresholds. The findings and the implications in this study, on one hand, help to analyze the mechanism and characteristics of the digital economy on regional green economic development, thereby providing theoretical enlightenment to academic societies; on the other hand, they also bring practical evidence to help other countries make the most of the opportunities offered by the digital economy and support them in hastening the green transformation, strengthening the digital economy and keeping up with the fast-paced world.

The remaining sections of the study are as follows: Session 2 is the literature review; Session 3 presents a theoretical analysis and proposes research hypotheses; Session 4 describes the research methodology; Session 5 provides the empirical results and discussion; and the final section presents conclusions and policy recommendations.

2. Literature Review

2.1. Research on the Digital Economy

Since Tapscott (1996) [24] introduced the term “digital economy” in 1996, the research has broadly gone through three phases: information economy, internet economy, and new economy. The early definitions focused on covering digital technology productivity, emphasizing digital technology industries and their market-based applications such as communication equipment manufacturing, IT service industries, and digital content industries [25,26]. As the research progressed, the focus gradually shifted to the interpretation of the economic functions of digital technologies and the transformation of production relations by digital technologies [27].

The majority of studies on the digital economy are qualitative at the moment as there is no direct measurement and recognized indicator system for the digital economy [21,28]; moreover, there is little quantitative research literature, with most of it centered around national and provincial levels [29]. The digital economy is a distinct economic form which virtually completes commodity service transactions. The advancement of information and communication technology (ICT) is inextricably linked to the development of the digital economy. ICT provides the necessary basic guarantee for the digital economy to penetrate into all walks of life, change the way of self-operation, and improve efficiency due to the rapid development of information technology [30,31]. With the rapid development of the digital economy, scholars began to pay attention to the evolution of laws and characteristics of the digital economy and conducted extensive research on digital economy measurement methods. Buhkt and Heeks (2017) [21] believe that the evolution of the digital economy and the evolution of ordinary economic activities maintain an inherent consistency. ICT and digital technology development are central to the digital economy. The combination of these digital technologies and manufacturing services has the potential to transform traditional manufacturing and service industries’ production processes, as well as implement production service process reengineering. The main trend of the future development is “digital economy” plus “physical manufacturing”. The development of the digital economy will hasten the formation of new business models, including the platform economy and the sharing economy, with certain green characteristics.

2.2. Research on the Green Economy

According to the World Air Quality Report (2021), air pollution in China’s cities is moderate. Some cities’ pollution indexes have been included in the key pollution monitoring cities. Environmental protection in China’s cities remains strict. Cities, as important spatial carriers of economic agglomeration, population concentration, and economic growth, are also significant sources of pollution and greenhouse gas emissions. With the dual goals of achieving urban economic growth and improving the ecological environment [8], green and high-quality urban development have become a growing concern in recent academic research [9].

The idea of total factor productivity (TFP), which originated from Solow (1957) [2], considers the part of technological progress other than the growth of factor inputs as an indicator of the quality of economic development. Traditional total factor productivity measures the impact of capital and labor inputs on output. It excludes pollution emissions, the ecological environment, and other factors, and thus cannot comprehensively and reasonably assess the quality of economic development. However, the GTFP is consistent with today’s new concept of the green economic development because it fully accounts for pollution emissions as well as other unexpected output and ecological environment factors [32,33]. Some scholars have also developed a non-radial and non-angular slacks-based measure (SBM) model based on the relaxation measure, to measure the green efficiency value which includes unexpected outputs, thus identifying and enhancing the GTFP measurement in current research [34,35,36,37].

2.3. The Relationship of the Digital Economy and the Green Economy

In general, academic research on the impact of the digital economy on the GTFP is still rare. Some scholars found that the development of the digital economy contributes to a reduction in carbon emissions and pollution, and its natural green properties can bring economic efficiency [12]. In the view of Nguyen et al. (2020) [38], the development of information and communication technology not only encourages economic growth, but also improves the productivity of G20 countries with regard to carbon emissions. According to Ben Lahouel et al. (2021) [39], a smooth transformation regression model was used to analyze whether the application of ICT technology can improve the relationship between carbon emissions and macroeconomic growth, and whether it has a positive effect on green economic growth. As an important driving force of economic transformation at this stage, the digital economy will certainly have an impact on the realization of regional green development, but current empirical studies that properly assess the role of the digital economy on green development are relatively scarce.

In terms of the mechanism of the digital economy affecting the green economy, scholars believe that, firstly, the development of the digital economy has improved the efficiency of information interaction, which improves the efficiency of regional economic operation and thus achieves green urban development [40,41]. Secondly, the digital economy has improved the level of innovation, and the speed and cost of knowledge spillover and knowledge interaction in the innovation network is faster and lower, which has promoted technological development and thus industrial green transformation [42]. Thirdly, the digital economy strengthens the interaction and cooperation between regions, forming a circular economy system, reducing pollutant emissions and thus promoting the green transformation of the region [43].

In summary, although there are some valuable studies on the digital economy or the green economy, few studies discuss the relationship of the two considering a comprehensive measurement of the digital economy and the impact brought by environmental regulation. Therefore, this study attempts to enrich the research on this area.

3. Theoretical Analysis and Research Hypotheses

3.1. The Impact of Digital Economy Development on the GTFP

In light of the new norm of economic development, the conventional development model of the industrial economy, which is characterized by high pollution and high energy consumption, needs to be altered as soon as possible. As a general trend, green and high-quality development is becoming prevalent, and the digital economy could assist companies to change their rough development mode to a green development mode, and the impact mechanism of the digital economy on the green economy can be illustrated from three aspects:

Firstly, the development of the digital economy can bridge the economic gaps between different regions. The GTFP varies greatly from region to region due to factors such as population size, development rate, and environmental regulations. Nevertheless, digital technology can provide a new approach to inter-regional cooperation and some virtual industries can even form industrial clusters in the cloud, effectively improving the efficiency of economic operations and contributing to the green development of cities [44].

Secondly, the digital economy reduces energy consumption and improves the efficiency of using resource factors. With data as the key production factor, the digital economy can significantly reduce the excessive consumption of physical resources and energy by traditional industrial production. Through the effect of energy conservation and emission reduction, the accelerated adjustment of factor structure, the improvement in efficiency utilization, and the promotion of a green and high-quality social economy can be achieved [45].

Thirdly, the highly technological nature of the digital economy has driven continuous technological innovation and development. The digital economy is the breeding ground for new industries, new formats, and new business models that can improve the GTFP through the promotion of green technological innovation [46]. In the digital economy, barriers to the flow of information, data, technology, and talent among regions have been greatly reduced and enterprises have been encouraged to develop green technologies through spillover effects and achieve green high-quality development [47,48].

Based on the above analysis, the following assumption is proposed:

Hypothesis 1. 

The development of the digital economy could improve the GTFP across regions.

3.2. The Threshold Effect of Environmental Regulation

The mechanism of the impact of the digital economy on the GTFP is complex, and environmental regulations may affect the green benefits of the digital economy in three aspects:

Firstly, the rapid development of the digital economy cannot be achieved without the support of the institutional environment. This will result in an insecure legal basis for the operation of the digital economy, as well as a lack of corresponding soft facilities for the digital economy when the institutional environment is not perfect. Moreover, the innovative features of the digital economy are unable to be fully utilized under such an institutional environment. This has a negative effect on enterprises using the digital economy to reshape production processes because it is not conducive to integrating traditional supply chains. In order to support the growth of the digital economy many countries have published relevant regulations. GDPR, or the General Data Protection Regulation of the EU, came into force in 2018, providing legal support for the rapid development of the European digital economy and providing a legal basis for citizens accessing digital information. Japan’s Ministry of Economy and Industry issued a White Paper on Trade in 2018 which put the digital economy in a strategic position. Therefore, a good institutional environment enables the digital economy to operate smoothly, reduces legal disputes, further standardizes the digital economy, lowers the uncertainty risk for enterprises using the digital economy, encourages enterprises to use the digital economy for green innovation, and achieves the GTFP improvement [49].

Secondly, it has been shown that the development of the digital economy can partially offset the negative impact of environmental regulations on the GTFP. The combination of digital economy development and environmental regulation can better utilize the promotion effect of the digital economy on green total factor productivity [50]. The intensity of environmental regulation and the GTFP have an inverted U-shaped relationship [51]. The moderate government regulation of the environment can promote the innovation of green technology, improve the efficiency of green technology use, and obtain higher returns through technological innovation, thus promoting the improvements in the GTFP while the high intensities of environmental regulation may lead to changes in the prices of production factors and make enterprises pay more for environmental management, which will offset part of the benefits gained from technological innovation and reduce their subsequent R&D investment, thus inhibiting any improvements in green total factor productivity [52]. Therefore, the effect of digital economy development on the GTFP is influenced by the intensity of environmental regulations, and the relationship between the two may be non-linear.

Thirdly, the process of environmental regulation on its own has a non-linear character, especially in the process of green economy inputs and outputs [53]. The “Porter hypothesis” argues that environmental regulation, while initially increasing the operating costs of firms also “pushes back” the costs of green economy inputs, but it will also “force” enterprises to upgrade their technology. Consequently, after the environmental regulation reaches a certain intensity, the efficiency of the green economy will be improved at the level of the whole society. After environmental regulations reach a certain intensity, the efficiency of the green economy will be enhanced at the society-wide level [54]. Environmental regulations will have a non-linear effect on digital development.

Based on the above analysis, the following assumption is proposed:

Hypothesis 2. 

As long as other factors are controlled, environmental regulation has a threshold effect on the impact of the digital economy on the GTFP.

4. Variable Selection and Measurement

4.1. Dependent Variable

Green total factor productivity (GTFP) is the dependent variable. According to the traditional concept of total factor productivity, environmental factors such as energy consumption and pollutant emissions are considered, which can highlight the concept of green development. Based on the SBM-GML index model, this study uses Chung’s (1997) [34] method to calculate the GTFP of 30 provinces and municipalities in China.

The first step is to define the environmental technology model. The model outputs include expected output vector $y$ and unexpected output vector $b$. The input factor vector is $x$, and the output set is as follows:

$$Y\left(x\right)=\left\{\left(y,b\right):\max y\left(x,b\right)\right\} , x\in R{N}^{+}$$

(1)

In Equation (1) the production factor $x$ can achieve the maximum output under certain technical conditions, including expected and unexpected results. Three prerequisites should be met for the environmental technology model to be valid: first, $y$ and $b$ must be zero sum; second, $y$ and $b$ are jointly weakly disposable; third, $y$ is strongly disposable. The relative efficiency equation of each decision can be calculated by constructing the production possibility boundary of the environmental technology model:

$$\overrightarrow{D_0}\left(x,y,b;g\right)=sup\left\{\beta :\left(y,b\right)+\beta g\in Y\left(x\right) \right\}$$

(2)

Equation (2) reflects the preference characteristics of $y$ and $b$, taking $g$ as the direction vector, $g=\left(y,-b\right)$. When the production factor $x$ is relatively constant, the output that vector $y$ can achieve along the $g$ direction is the maximum output $Y\left(x\right)$. Considering “expected output“ as “good output“ and “unexpected output“ as “bad output” such as pollution, linear programming is used to maximize “good production” and minimize “bad output”.

$$\begin{gathered} \overrightarrow{D_0^t}\left(x^t,y^t,b^t;y^t,-b^t\right)=max\overrightarrow{D_0}\left(x,y,b;g\right)=max\beta \\ s.t.{\displaystyle \sum }_{k=1}^K{\lambda }_k^t{y}_{ks}^t\ge \left(1+\beta \right)y_{ks}^t,{\displaystyle \sum }_{k=1}^K{\lambda }_k^t{b}_{km}^t\ge \left(1+\beta \right)b_{ks}^t, \\ {\displaystyle \sum }_{k=1}^K{\lambda }_k^t{x}_{kn}^t\ge x_{kn}^t,{\lambda }_k^t\ge 0 \end{gathered}$$

$$s=1,2,\cdots S;m=1,2,\cdots M;n=1,2,\cdots N;k=1,2,\cdots K$$

(3)

In Equation (3), $s$ represents the types of expected output $y$, $m$ represents the types of unexpected output $b$, n represents the types of input production factors $x$, k represents the decision-making units in the model, $t$ represents the period, and ${\lambda }_k^t$ represents the weight of each cross-sectional observation. The Malmquist–Luenberger $\left(ML\right)$ productivity index from period $t$ to $t+1$ is constructed as follows:

$$M{L}_t^{t+1}={\left\{\frac{1+D_0^t\left(x^t,y^t,b^t;g^t\right)}{1+D_0^t\left(x^{t+1},y^{t+1},b^{t+1};g^{t+1}\right)}\times \frac{1+D_0^{t+1}\left(x^t,y^t,b^t;g^t\right)}{1+D_0^{t+1}\left(x^{t+1},y^{t+1},b^{t+1};g^{t+1}\right)}\right\}}^{\frac{1}{2}}$$

(4)

The $ML$ index can be further decomposed into the EFFCH index and the TECH index, representing the efficiency change and the technological change, respectively, from period $t$ to $t+1$, written as:

$$EFFC{H}_t^{t+1}=\frac{1+D_0^{t+1}\left(x^t,y^t,b^t;g^t\right)}{1+D_0^{t+1}\left(x^{t+1},y^{t+1},b^{t+1};g^{t+1}\right)}$$

(5)

$$TEC{H}_t^{t+1}={\left\{\frac{\left[1+D_0^{t+1}\left(x^t,y^t,b^t;g^t\right)\right]}{\left[1+D_0^t\left(x^t,y^t,b^t;g^t\right)\right]}\times \frac{\left[1+D_0^{t+1}\left(x^{t+1},y^{t+1},b^{t+1};g^{t+1}\right)\right]}{\left[1+D_0^t\left(x^{t+1},y^{t+1},b^{t+1};g^{t+1}\right)\right]}\right\}}^{\frac{1}{2}}$$

(6)

Combining Equations (4) and (6), the GTFP increases when ($ML>0$), and otherwise declines. In Equation (5), $EFECH>0$ means the output growth is caused by efficiency changes. In Equation (6), $TECH>0$ is the output growth caused by technological progress. Finally, linear programming is used to solve the ML productivity index and the GTFP is obtained. Specifically, 30 provinces and municipalities are treated as independent decision-making units (DMUs). The estimated input indicators include labor, capital, and energy. The output indicators are divided into expected output and unexpected output. The labor input is measured by the total number of employees at the year end of each DMU. The energy input index is expressed by the energy consumption of each DMU. The capital input index is calculated by the “perpetual inventory method” with reference to Zhang et al. (2004) [55]. The expected output is the actual gross regional product of each DMU, and the unexpected output includes sulfur dioxide (SO₂) emissions and chemical oxygen demand (COD) emissions in wastewater. Finally, the annual GTFP of 30 DMUs from 2010 to 2020 was calculated with Stata 16. Due to the lack of data, Tibet, Hong Kong, Macao, and Taiwan are not included in the research.

4.2. Independent Variables

The development level of the digital economy (DEI) is the independent variable. As the measurement of the digital economy has evolved from a single indicator to a comprehensive index, the study adopts the method proposed by Yang et al. (2021) [56] and constructs an evaluation system by integrating the “four modernizations” of the digital economy—digital industrialization, data valorization, industrial digitalization, and digital governance, referring to the White Paper on the Development of China’s Digital Economy (2020) [57] and Statistical Classification of Digital Economy and its Core Industries (2021) [58].

The specific measurement indicators are shown in

Table 1. The evaluation system of China’s digital economy.

First Tier Indicators	Second Tier Indicators	Third Tier Indicators	First Tier Indicators	Second Tier Indicators	Third Tier Indicators
Digital industrialization	Infrastructure	a. Internet broadband access users	Data valorization	Trading market	k. Big Data exchange
		b. Internet broadband access ports	Industrial digitalization	Agricultural marginal contribution of digital technology	l. Added value of the primary industry
		c. Length of long-distance optical cable		Industrial marginal contribution of digital technology	m. Added value of the secondary industry
		d. Length of optical cable lines		Service industry marginal contribution of digital technology	n. Added value of the tertiary industry
	Software and services	e. Software product revenue	Digital governance	Multi-agent participation	o. Number of government websites
		f. Software service revenue			p. Number of government microblogs
	Internet	g. Number of webpages			q. Amount of government news in public media
		h. Number of domain names
		i. Number of Internet users
		j. Internet penetration

. Data on the added value of the primary, second, and tertiary industries are derived from the Urban Statistical Yearbook [59], and other data are taken from the Statistical Report on the Development of China’s Internet [60]. The principal component analysis method is applied to construct the comprehensive indicator for the digital economy development level of 30 provinces and municipalities in China, and multiple variables are reduced to a few representative variables (principal components) in order to avoid interference caused by subjective random factors. It can also eliminate the differences in dimensions and magnitudes among variables, thus overcoming the limited informational content provided by a single variable and simplifying the analysis of complex statistical data. Finally, the comprehensive score model is used to calculate the development level of the digital economy in each province or municipality.

Bartlett and KMO tests are shown in

Table 2. KMO test and Bartlett test results.

Bartlett Test		Kaiser-Meyer-Olin
Chi square	6901.783	0.839
Df	136
Sig.	0.000

. Principal component analysis indicates that the variables in the digital economy evaluation system have structural validity and are relatively independent to some extent.

As shown in

Table 3. Principal component analysis results.

Component	Characteristic Value	Variance	Cumulative Variance Contribution	Component	Characteristic Value	Variance	Cumulative Variance Contribution
1	7.8880	5.1215	0.4640	10	0.2755	0.0760	0.9632
2	2.7665	0.5857	0.6267	11	0.1995	0.0459	0.9750
3	2.1809	1.2839	0.7550	12	0.1536	0.0246	0.9840
4	0.8969	0.2228	0.8078	13	0.1290	0.0447	0.9916
5	0.6741	0.0984	0.8474	14	0.0843	0.0599	0.9966
6	0.5757	0.1519	0.8813	15	0.0244	0.0028	0.9980
7	0.4239	0.0380	0.9062	16	0.0216	0.0090	0.9993
8	0.3859	0.0784	0.9289	17	0.0126	.	1.0000
9	0.3075	0.0320	0.9470

, SPSS 25 is used to extract the principal components of the 17 variables listed in Table 1. Among the 17 variables, three have a characteristic value greater than 1 and the cumulative variance contribution rate is greater than 75%, which represents most of the information in the sample.

The weight of each original variable in the principal component is shown in

Table 4. Weight of each original variable in the principal component.

Variables	Comp1	Comp2	Comp3	Variables	Comp1	Comp2	Comp3
a	0.3252	0.1889	−0.0839	j	0.2040	−0.2972	−0.1901
b	0.3316	0.1339	−0.1146	k	0.1981	−0.3239	0.0101
c	0.0977	0.3988	−0.1151	l	0.0558	0.3209	0.1606
d	0.2942	0.2188	−0.1857	m	0.1147	0.2013	0.4587
e	0.2904	−0.241	0.1663	n	0.2750	0.0541	0.2644
f	0.2965	−0.2112	0.1551	o	0.0438	0.2231	0.5107
g	0.2038	−0.3918	0.1635	p	0.2979	0.0440	−0.2598
h	0.2700	−0.1661	0.0902	q	0.2144	0.1487	−0.4101
i	0.3035	0.2052	0.0814

. The original variables are scored using principal component calculation method and the results are shown in

Table 5. Digital economy development index of provinces in China from 2010 to 2020.

	Year	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Province
Guangdong		0.427	0.497	0.466	0.532	0.582	0.662	0.819	0.850	0.872	0.871	1.000
Jiangsu		0.403	0.457	0.432	0.449	0.485	0.572	0.670	0.754	0.710	0.707	0.763
Shandong		0.323	0.363	0.348	0.426	0.429	0.455	0.5321	0.586	0.589	0.545	0.621
…		…	…	…	…	…	…	…	…	…	…	…
Qinghai		0.016	0.018	0.023	0.026	0.030	0.033	0.041	0.044	0.048	0.048	0.053
Hainan		0.006	0.011	0.008	0.010	0.014	0.018	0.032	0.038	0.039	0.054	0.049
Ningxia		0.001	0.004	0.003	0.005	0.004	0.013	0.022	0.033	0.035	0.030	0.042

.

With regard to the comprehensive score model, the development level of the digital economy in each province or municipality is calculated and standardized using the min-max method. In most provinces in China, the digital economy develops at an increasing rate. Guangdong, Jiangsu, and Shandong had the highest levels of digital economy development in 2020. The three regions with the lowest level of development in the digital economy are Qinghai, Hainan, and Ningxia (see Table 5 for details (Due to space limitations, the results of the development level of digital economy in other provinces are not provided.)). As a whole, the development level of the digital economy in the eastern region is higher than that in the central and western regions.

4.3. Threshold Variables

Environmental regulation (Envi): environmental regulation is used to explore the impact of provincial environmental governance on the GTFP. Green economy development is generally believed to be enhanced by powerful environmental governance. Is there, however, the possibility of “more governance, more pollution”? To investigate this problem, Levinson’s environmental regulation index is used to evaluate the intensity of the environmental governance of various provinces in China [61].

4.4. Control Variables

The control variables include population size as well as the traffic infrastructure, economic development, opening-up, and government support levels. The population size (Urban) includes the population of each province [62]. The level of transportation infrastructure (Infra) is calculated by dividing the sum of railway mileage, highway mileage, and inland waterway mileage by the land area [63]. The level of economic development (Pgdp) is measured by the per capita GDP of each province or municipality [64]. The opening-up level (Trad) is measured by the proportion of import and export trade volume in GDP of each province [65]. The level of government support is measured by the proportion of provincial fiscal expenditure in GDP [66].

4.5. Data Sources and Descriptive Statistics

The data used in this study are taken from the China Statistical Yearbook, the Statistical Yearbook issued by the provincial statistics bureaus, the China Fixed Assets Investment Yearbook, the China Energy Statistical Yearbook, and the China Environmental Statistical Yearbook over the years. The descriptive statistics of the variables involved are shown in

Table 6. Descriptive statistics of variables.

Variable Type	Variable Name	Code	Obs	Mean	Std. Dev.	Min	Max
Dependent variable	Green total factor productivity	GTFP	330	1.0179	0.0896	0.5536	1.4072
Independent variable	Digital economy	DEI	330	0.2262	0.1776	0	1
Threshold variable	Environmental regulation	Envi	330	1.0791	1.0257	0.0202	7.9264
Control variables	Population size (in ten thousands)	Urban	330	882	1.158	364	2.821
	Transport infrastructure	Infra	330	1.3525	1.0389	0.0737	5.7447
	Economic development	Pgdp	330	4.6657	0.2050	4.1099	5.2172
	Opening-up	Trad	330	0.0194	0.0157	0.0001	0.0829
	Government support	GT	330	0.2418	0.1032	0.0135	0.6284

.

4.6. Model Construction

Referring to Cheng and Qian (2021) [49], the benchmark model used to examine the impact of the digital economy on the GTFP is shown as Equation (7):

$$GTF{P}_{it}={\alpha }_0+{\alpha }_{1}DE{I}_{it}+{\beta }_{1}contro{l}_{it}{}^{ }+{\mu }_i+{\nu }_t+{\epsilon }_{it}$$

(7)

where $GTF{P}_{it}$ represents the GTFP of province i in period t period, $DE{I}_{it}$ is the digital economy, ${\alpha }_0$ is the coefficient, $contro{l}_{it}$ represents all control variable, and ${\mu }_i$ and ${\nu }_t$ are the urban and time-fixed effects, respectively.

Based on Hansen’s (1999) non-dynamic panel threshold model [67], the basic model is modified with environmental regulation ($Env{i}_{it}$) as the threshold variable in order to further measure the network effect of the digital economy development on the GTFP:

$$GTF{P}_{it}={\gamma }_0+{\gamma }_{1}DE{l}_{it}·I\left(Env{i}_{it}\le \lambda \right)+{\gamma }_{2}DE{l}_{it}·I\left(\lambda <Env{i}_{it}\right)+{\gamma }_{3}contro{l}_{it}+{\mu }_i+{\nu }_t+{\epsilon }_{it}$$

(8)

where $I (·)$is the indicator function, which takes 1 when the conditions in parentheses are met; otherwise, it takes 0. Equation (8) assumes that there is only one threshold, but it can be extended to a multi-threshold measurement model.

5. Analysis and Discussion

5.1. Regression Result Analysis

5.1.1. Empirical Facts

According to Lorna et al. (2021) [68], the development gap between urban and rural areas due to digital exclusion should be taken seriously; otherwise, the already-fast-growing urban areas will become ‘faster, fastest’, leaving rural areas behind. To find out if there are differences in the development of the digital economy and the green economy in different regions of China by drawing on the methodology of Zhou et al. (2021) [48], the development of the digital economy and the green economy in different regions of China is dynamically presented and analyzed by mapping the three-dimensional kernel densities.

Figure 1 and Figure 2 show the development level, the evolution of distribution, the ductility, the polarized trend of the digital economy, and the GTFP in the three-dimensional kernel density maps. From this, the digital economy and the GTFP in different regions of China exhibit the following characteristics.

Firstly, the distribution center of the digital economy and the GTFP moved to the right, indicating an upward trend in the development level of the digital economy and the GTFP in different provinces.

Secondly, the digital economy of provinces or municipalities in China has a long right tail, which means that there are provinces with very high levels of digital economy development. Despite this, the right tail of the GTFP does not appear to be obvious, indicating that the GTFP across provinces in China is relatively centralized.

Thirdly, there was only one peak point in the distribution of the digital economy in the provinces of China from 2010 to 2020, and there was no clear peak point from 2012 to 2016, indicating that the polarized trend of China’s regional digital economy development is not evident during this period. A single peak point of the GTFP can be found during 2010–2012 and 2018–2020, suggesting that the development of the GTFP in provinces of China does not show a significant polarized trend during this period. However, from 2012 to 2018, the GTFP changed from multi-peak to double-peak, and the span of kurtosis gradually widened, which indicates that the GTFP gap between different provinces also expanded.

Figure 3 and Figure 4 illustrate the change in the digital economy development and the GTFPs in 2014, 2017, and 2020 in 30 provinces or municipalities in China by spidergrams, classifying them into northern and southern parts for analysis and comparison.

It can be observed that the digital economy development and the GTFPs have generally increased in provinces or municipalities from 2014 to 2020. It is noteworthy, however, that there is a large gap between the levels of the digital economy across regions, whereas the development trend of the GTFP is relatively flat. Beijing, Henan, and Shandong (in the northern region) and Jiangsu, Sichuan, and Guangdong (in the southern region) are the representatives with a high level of digital economy. Contrarily, Ningxia and Hainan are the least developed, or even not developed at all, followed by Jilin and Tianjin, which has obvious weaknesses in the digital economy development.

Specifically, Figure 3 shows that there is an evident gap across different places, which are large and distributed irregularly. The digital economy development level in the south, however, is significantly higher than that in the north. The digital economy in the south began earlier and the digital infrastructure is better, so it is more likely to attract capital, technology, and talent, resulting in a good circulation of the development. A digital divide exists between regions due to differences in technological capabilities [69].

Figure 4 shows that the overall growth of the GTFP in China is relatively steady. Other than Tianjin and Xinjiang, the GTFPs in other provinces or municipalities are on the rise, higher than the historical average by 2020. Liaoning, Shandong, and Henan (in the north) and Zhejiang and Guangdong (in the south) have the highest levels of GTFP. Comparing the provinces we can observe that the GTFPs in the northern part are higher than those in the southern part. Both the development trends of the digital economy and the GTFP have their own characteristics and deserve some attention in future studies. The development of the digital economy can be an effective tool to accelerate the development of economies that are relatively backward, as verified in VÁZQUEZ’s (2021) study on the digital economy in Spanish rural areas [70].

5.1.2. Regression Analysis

The Impact of the Digital Economy on the GTFP

Table 7. Benchmark regression results and subregional regression results.

Variables	(1)	(2)	(3)	(4)
	FE	RE	North	South
DEI	0.067 **	0.031 *	0.324 ***	0.418 ***
	(2.111)	(1.901)	(5.681)	(4.215)
Dens	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***
	(−5.321)	(−3.911)	(−5.553)	(−2.692)
Infra	0.032 ***	0.020 ***	0.018 **	0.056 ***
	(4.672)	(5.221)	(2.226)	(4.427)
Pgdp	−0.028 *	−0.052 ***	−0.118 ***	−0.008
	(−1.053)	(3.223)	(-2.788)	(−0.236)
Trad	−0.860 ***	−0.246 *	−1.442 ***	−0.243
	(−4.644)	(−1.734)	(-6.575)	(−0.768)
GT	0.130 **	0.067 *	0.233 ***	0.025
	(2.433)	(1.753)	(2.749)	(0.339)
Constant	0.945 ***	0.715 ***	1.258 ***	0.831 ***
	(9.191)	(10.661)	(7.634)	(6.498)
Observations	330	330	165	165
Number of id	30	30	15	15
R-squared	0.423	0.401	0.558	0.420

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

presents the benchmark regression results of the digital economy on the GTFP. Both the results of the fixed effects (FEs) and the random effects (REs) are reported. The FE model with a better fitting effect and more robust results is then selected based on the Hausman test, AIC criteria, and R². The coefficient of DEI in Column (1) is statistically positive.

For control variables the coefficient of Urban is significantly negative, indicating that the population size inhibits improvements in the GTFP. This may be because, on the one hand, population gathering brings about industrial agglomeration which promotes economic development. However, on the other hand, it will result in an increase in resource consumption and the emissions of various pollutants, which will adversely affect the environment and hinder improvements in the GTFP. The transportation infrastructure (Infrastructure) can significantly improve the GTFP, reflecting the importance of transportation facilities for factor flow to a certain extent. The coefficient of Pgdp is significantly negative, indicating that economic development improvements have a negative impact on the GTFP, which confirms that China’s economy continues to be at the expense of the environment. There is a significant negative correlation between Trad and the GTFP, indicating that the level of opening-up hinders improvements in the GTFP, proving the existence of the pollution paradise hypothesis. Government expenditure plays a positive role in promoting the GTFP, as does the growth of government support (GT).

Analysis by Regions

The study splits the entire sample into two regions, the north and the south, to investigate whether the mechanism of the digital economy affects the GTFP differently in different regions, i.e., the heterogeneity analysis. Table 7 shows that the impact of the digital economy on the GTFP in northern and southern China is clearly heterogeneous. The digital economy has contributed positively to the GTFP in both the north and the south, as shown by the results in Columns (3) and (4). However, the coefficient of 0.418 in the south is higher than that of 0.324 in the north.

5.1.3. Threshold Effect Analysis

In

Table 8. Threshold effect test results.

Dependent Variable	Threshold Variable	No. of Thresholds	F Value	p Value	BS Times	Threshold Value			Critical Value
						I	II	III	1%	5%	10%
GTFP	Envi	Single	2.14	0.0086	300	7.6688			12.8614	10.8064	9.3574
		Double	19.34	0.0033	300	7.6688	7.9464		15.8282	10.6819	8.5331
		Triple	10.48	0.1700	300	7.6688	7.9464	8.8827	24.6793	17.8675	13.2923

the threshold effect test results indicate that the double thresholds of the GTFP are all significant at the 1% level, while the triple thresholds are not significant. Therefore, it is presumed that the GTFP has a double threshold.

Table 9. Regression results of threshold effect.

Threshold Variable	Threshold Interval	Coefficient	p Value
Envi	$W\le 7.6688$	0.0993	0.112
	$7.6688<W\le 7.9464$	0.2896	0.000
	$W>8.8827$	0.0535	0.271

shows that when the single threshold value is less than 7.6688, the coefficient of Envi on the GTFP is 0.0993, which is a positive, but not statistically significant. In the case where the threshold value is greater than 7.6688 but less than or equal to 7.9464, the coefficient increases to 0.2896, which is significantly positive at the 1% level. This demonstrates that the impact of the digital economy on the GTFP is significantly enhanced, and the network effect is achieved. Hypothesis 2 above is verified.

5.1.4. Endogeneity Analysis

In terms of data sources, the study uses officially published data, which are more reliable and can reduce the endogenous problems caused by measurement errors. The GMM and 2SLS instrumental variable methods are employed to mitigate the interference of potential endogenous problems with conclusions. The GMM regression result is presented in Column (1) in

Table 10. Endogenous test results.

Variable	(1)	(2)	(3)
	GMM	First Stage	Second Stage
DEI	0.084 ***	0.041 **	0.045 **
	(3.063)	(2.115)	(1.992)
Other Variables	control	control	control
AR (2)	0.510	0.222	0.998
Hansen	0.151		0.125
Wald Chi2	90.32 ***
F		64.668 ***	14.51 ***
Kleibergen–Paap rk LM statistic			38.365 ***
Kleibergen–Paap rk Wald F statistic			64.668 ***
N	330	330	330

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

. The coefficient of DEI on the GTFP is significantly positive and passed AR (2) and the Hansen tests, indicating that the DEI promotes the GTFP. The outputs of mobile phones and microcomputer equipment are chosen as the instrumental variables in the 2SLS test, and Columns (2) and (3) in Table 10 show the regression results of the first and second stages, respectively. As can be seen in Column (2), the instrumental variable shows a significant positive correlation with DEI, and the F test is significant at 1%. From Column (3), the DEI coefficient on the GTFP is significantly positive at 1%; thus, the development of the DEI can result in an improvement in the GTFP. Considering endogenous factors the DEI plays a significant role in promoting the GTFP, which is consistent with the previous results and confirms the hypotheses.

5.2. Discussion

This session discusses our findings to link the empirical evidence with the research questions regarding the impact of the digital economy on the GTFP, the differences between regions, and the role of environmental regulation on the relationship of the digital economy and the GTFP.

5.2.1. The Impact of Digital Economy Development on the GTFP

According to the results reported in Table 7, the development of the digital economy can significantly improve the GTFP. Our findings verify the positive economic consequences, such as changing the traditional commercial logic and improving the industrial efficiency brought by the digital economy [40].The findings of Xiao and Chi (2021) [41] were tested only by theoretical arguments, but not by data facts. Our study also supports the work of Zhou et al. (2021) [48], which also analyzes the impact of the digital economy on the Chinese inter-provincial GTFP and has a slightly higher impact coefficient than our results when using single-dimension indicators of the digital economy.

The findings verify that the digital economy can promote a green and high-quality economy, which corresponds to the new concept of green development.

5.2.2. The Differences between Regions

The three-dimensional kernel density maps show the dynamic trends of hierarchical differentiation in the digital economy and the GTFPs among Chinese provinces or municipalities during various times. The overall digital economy and the GTFP of each province in China exhibit an upward trend, but there are also gaps among provinces and the development disparity tends to widen.

The regional differences in the development of the digital economy and the green economy are verified by sub-regional regression analysis. Due to China’s vast territory, there are significant differences in economic development, transportation infrastructure, opening-up, and other resource endowments among different regions, which may have an impact on the relationship between the digital economy and the GTFP. The results show that the development of the digital economy in China is characterized by heterogeneity; the development of the digital economy has a greater impact on the GTFP in the south, which is consistent with the north–south differences in the digital economy, and the GTFP shown in the findings validates the scholars’ previous research results [43] and also enriches the study effects in terms of three-dimensional dynamic analysis.

5.2.3. The Threshold Effect of Environmental Regulation

Threshold analysis indicates that if the improvement in environmental governance exceeds the threshold, institutional advantages can be utilized as a means of stimulating enterprises to accelerate technological research and development and improving the positive impact of the digital economy on the GTFP. Conversely, if the environmental regulation is below the threshold, entities may be faced with uncertain institutional conditions. The rent-seeking power of enterprises exceeds that of the technological research and development. The digital economy will not be used to improve the GTFP, but rather to reduce external uncertainty by using more resources for repetitive production. Therefore, the digital economy must be accompanied by a good environmental governance system to exert a significant positive effect on GDP.

The existence of a threshold effect of environmental regulation on the relationship between the digital economy and the GTFP also corroborates the study by Cheng and Qian (2021) [49], who obtain a single threshold of environmental regulation. However, our study found a double threshold and an interval range of environmental regulation, thus offering a clearer policy room; in other words, below or above the thresholds will have a negative impact on the economy. Such findings provide a better reference value for the government to formulate environmental policies.

6. Conclusions and Policy Recommendations

This study uses the provincial panel data of China from 2010 to 2020 to build a digital economy evaluation system from the four aspects of integrated “four modernizations”—digital industrialization, data valorization, industrial digitalization, and digital governance—to comprehensively measure the development level of the digital economy, empirically examine the impact of the digital economy on the GTFP, and obtain the following conclusions: (1) China’s digital economy plays a significant positive role in promoting the GTFP; (2) the digital economy development varies across different regions, and the development in the southern provinces is higher than that of northern provinces; (3) the development of the digital economy has the characteristics of a network effect. There are double thresholds for the impact of environmental regulation on the GTFP. If the improvement in environmental governance exceeds the threshold, institutional advantages can be brought into play.

Based on the conclusions, the following policy recommendations are proposed: First, governments at all levels should prudently and meticulously plan the development of the digital economy to create conditions enabling green economic development. Second, considering the heterogeneity of the digital economy in different regions, it is necessary to implement differentiated development policies adapted to local conditions, thus promoting the development of the digital economy and exploring new drivers for the growth of a green economy. Third, governments at all levels should enhance environmental governance, strictly execute environmental protection laws, clearly develop energy conservation and environmental protection industries, and thereby promote the development of a green economy. Future research based on our study can consider the introduction of the Durbin model to analyze the spatial spillover effects of the digital economy, as well as analyze the capital and labor factor mismatch to further explore the mechanism for the differential effects of the digital economy on the development of the green economy.