The Impact of the Digital Economy on Low-Carbon, Inclusive Growth: Promoting or Restraining

Abstract

Based on panel data from 30 provinces in China, this paper uses a two-way fixed effect model to empirically test the influence of regional digital economy development on the level of low-carbon, inclusive growth. The empirical study shows that: (1) The digital economy has a significant inverted U-shaped impact on China’s regional low-carbon, inclusive growth. It shows that regional digital economic development has a significant inverted U-shaped impact on low-carbon, inclusive growth (the inflection point is 0.3081), and it was found that most of the observations fall on the left side of the inverted U shape. (2) The inverted U-shaped influence has significant heterogeneity in the regional location, information degree, and factor productivity level. (3) The digital economy promotes low-carbon, inclusive growth mainly by improving the overall efficiency of source allocation, but low-carbon, inclusive growth may be curbed by distorting the allocation of capital elements. (4) Via dimensionality reduction analysis, we found that the inverted U-shaped impact of digital applications and digital finance on regional low-carbon, inclusive growth is more obvious. In addition, we also found that the inverted U-shaped impact of regional digital economic development on low-carbon ecology and social inclusiveness is more obvious. This study provides an important reference value for relevant departments to formulate low-carbon, inclusive development policies from the perspective of regional digital economic development.

1. Introduction

Climate change has aroused global attention to carbon emission reduction. Governments have attached importance to the role of low-carbon development in the formulation of national economic and energy policies [1]. According to the statistics of the World Energy Statistical Yearbook (70th edition), global carbon dioxide emissions have been increasing continuously since 2013. Global carbon emissions are expected to increase by 30% by 2030 compared to 2010 levels [2]. Therefore, carbon emissions have become an important topic for economists all over the world. For example, Adebayo et al. [3] explored the asymmetric impact of renewable energy consumption and trade openness in Sweden on carbon emissions. They found that there is heterogeneity in the impact of trade openness and renewable energy on carbon dioxide emissions. However, in most quantiles, the impact of economic growth on carbon dioxide emissions is negative. Rjoub et al. [4] explored the relationship between carbon emissions, life expectancy, and economic growth in Turkey from the perspective of energy consumption, human health, and environmental hazards. Odhiambo [5] discussed the relationship between carbon emissions and economic growth in South Africa and found that energy consumption plays a positive role in promoting carbon emissions and economic growth. Lee and Brahmasrene [6] used EU national data to examine the long-term equilibrium relationship between tourism, carbon dioxide emissions, economic growth, and foreign direct investment. They found that tourism, carbon dioxide emissions, and foreign direct investment were all significantly positively correlated with economic growth, but that tourism and foreign direct investment help to reduce carbon dioxide emissions. In recent years, research on carbon emissions has been increasing [7,8,9,10,11,12], indicating that carbon emission reduction has become a key issue for the global climate problem [13,14,15,16,17,18]. As the number one country in global resource consumption, China has not effectively disposed of its heavy dependence on energy and the environment. After reform and opening-up, China’s economy has been growing for 40 years, which has created a “Chinese miracle” in the history of economic growth. However, the rapid development of its economy has led to excessive energy consumption and accelerated greenhouse gas emissions, so environmental problems have become increasingly prominent and cannot be ignored [19]. The emissions of all kinds of major pollutants have gradually exceeded the environmental carrying capacity, and problems such as increased carbon emissions have become the main obstacles to low-carbon sustainable development. As China’s economic development enters a new normal, the intensification of environmental pollution, the tightening of factor endowments, the reduction of labor dividends and the decline of capital income make the rough mode of development unsustainable. Therefore, how to assume the responsibility of a big country and strive to achieve carbon emission reduction has become an important task for China’s future development.

The rise of the digital technology revolution provides unprecedented opportunities and challenges for the economic development of our country. At present, China’s economy is gradually changing from a rough development of speed and quantity to a connotative development model of low-carbon sustainable development [20,21]. Promoting the deep integration of the digital economy and the real economy, and speeding up the transformation of the model of economic development, will become an important starting point for sustainable economic development in the new era [22]. Low-carbon development has been a wide concern for the government, the public, and academia [23]. Thus, whether to maintain low-carbon growth while taking into account socially inclusive development is an urgent problem before the government. Some scholars have tried to make valuable explorations on how to use the digital economy to accelerate the reconstruction of economic development and governance models and how to optimize industrial structure to promote green and sustainable development [24,25].

However, the existing literature pays less attention to the relationship between digital economy development and low-carbon, inclusive growth, which provides a research opportunity for the innovation of this paper. The innovation of this paper mainly focuses on three issues: First of all, there is little literature to comprehensively assess the impact of regional digital economic development on economic growth, social inclusiveness, and low-carbon ecology. There is room for discussion on this research topic. Secondly, there is a relative lack of research on the path of the digital economy affecting low-carbon, inclusive growth, and the literature based on the perspective of the digital economy paradox is even rarer. The discussion of these two issues is helpful in enriching the existing theoretical literature. Finally, we think that digital technology is a double-edged sword, and we should pay attention to its negative externalities while it plays its positive role. The discussion of heterogeneity not only helps to clarify the conditionality of the impact of the digital economy on low-carbon, inclusive growth but also provides detailed empirical evidence for the formulation of carbon emission reduction policies in various regions of China. Therefore, a discussion of the above three issues has important policy value for effectively releasing digital dividends to promote low-carbon, inclusive growth.

2. Literature Review

Based on the paradox of digitization and the realistic perspective of resource allocation, we explore the specific impact of regional digital economy development on low-carbon, inclusive growth. There are mainly two branches of the literature that are closely related to this article. The first branch of the literature discusses the relationship between the digital economy and energy consumption. Most studies support the idea that the development of the digital economy, as represented by the Internet, 5G, and blockchain, is conducive to low-carbon sustainable development [24,25,26,27,28,29,30]. For example, Lange et al. [24] found that the increasing application of information and communication technology (ICT) in the economy and society as a whole will stimulate the hope of reducing energy demand and emissions. Ma Q et al. [25] studied whether the digital economy can replace material resources to become a feasible source of green economy, and it was found that the digital economy can restrain consumption-oriented carbon emissions. Thus, it is beneficial to the development of the green economy. Wu J and tran n K. [26] studied the problem of sustainable energy development. They found that the application of blockchain technology helps to solve the problem of energy sustainability. Based on an analysis of the impact of ICT investment in Japan and South Korea, Khayyat et al. [27] found that ICT investment can replace parts of labor and energy. The above research shows that digital technology can help reduce energy consumption. Fan Xin and Yin Qiushu [28] confirmed that the development of digital finance helps to promote entrepreneurship and technological innovation, thereby improving energy efficiency and green total factor productivity. Cheng Wenxian and Qian Xuefeng [29] found that the impact of digital economy development on China’s industrial green total factor productivity shows a marginal increasing effect. By investigating data from OECD countries, Schulte et al. [30] found that the development of ICT helps to reduce energy demand. However, some scholars believe that the application of digital technology may increase total energy consumption, thereby increasing carbon emissions [31]. Other studies have also found that energy consumption and carbon emissions are increasing with the production and use of more and more ICT devices [32,33]. Therefore, the application of digital technology may also lead to more energy consumption, resulting in higher carbon dioxide emissions [34].

Another branch of the literature discusses the relationship between the digital economy and green development. The main conclusions can be divided into two aspects. First of all, the development of the digital economy with data as the key factor of production can replace traditional factors, which can alleviate the problem of environmental pollution through technological innovation, optimizing industrial structure, and improving the enthusiasm of the government and the public to participate in the cause of environmental protection. For example, Liu Pengcheng and Liu Jie [35] found that informatization can effectively alleviate the problem of urban environmental pollution through technological progress, population agglomeration, and industrial structure. Xie Chunyan et al. [36] found that the Internet can reduce environmental pollution through four mechanisms: government environmental regulation, environmental monitoring technology, public participation in environmental protection activities, and the development of the environmental protection industry. Li Shouguo and Song Baodong [37] believe that the development of the Internet will improve energy use efficiency and the level of technological innovation, and the resulting positive effects will completely absorb the possible negative effects, which will help to alleviate the problem of environmental pollution. Generally speaking, we believe that the digital economy can promote sustainable economic development by optimizing resource allocation, promoting human capital [38], and realizing green progress and cleaner production.

However, some scholars hold a different point of view. They believe that the digital economy will not only bring economic growth but also cause environmental pollution [39,40,41]. Their core view is that the growth of energy demand, driven by the digital economy, may lead to a greater pollution effect than its emission reduction effect, thus aggravating the deterioration of the environment [42]. In addition, a small number of scholars believe that the relationship between the digital economy and green development is uncertain [43]. The existing literature has not reached a consistent conclusion on the effect of carbon emission reduction on the digital economy. Differing from the views of most scholars, we believe that the relationship between the digital economy and low-carbon, inclusive growth may not be a simple linear relationship.

The above research has accumulated many valuable conclusions. For example, on the one hand, the existing research leads us to find that the carbon reduction effect of the digital economy has two sides. The digital economy may reduce carbon emissions by improving energy efficiency. It is also possible to increase total energy consumption and, thus, carbon emissions. Let us realize that the two sides of the externalities of the digital economy are also worthy of attention, which also provides a research perspective for this paper. On the other hand, the existing research also provides much of its basis in the literature. However, there are still the following shortcomings. First, the literature on digital economy is mostly qualitative research, while quantitative research is worthy of progressive discussion because there are few quantitative studies based on the two-sided effect of carbon emission reduction in the digital economy. Second, at present, there is little literature that explores the impact of digital economic development on regional low-carbon, inclusive growth and its mechanism from the perspective of resource distortion and puts the three factors into a unified framework for analysis. Third, the existing literature does not pay due attention to phenomena such as digital inequality and mismatch of resources, which may be caused by the digital economy in a region, let alone explore their negative effects on economic growth.

Given the above shortcomings, the marginal contribution of this paper has three aspects. First, index innovation. This paper brings social inclusiveness into the framework of the relationship between low-carbon ecology and economic growth. This paper constructs the level of low-carbon, inclusive growth in China from the three dimensions of economic growth, social inclusiveness, and low-carbon ecology. Compared to previous studies, this paper makes a more comprehensive assessment of the impact of digital economic development on low-carbon, inclusive growth. The second is perspective innovation. We explore the nonlinear relationship between the digital economy and low-carbon, inclusive growth, which not only helps to fully understand the carbon reduction effects of the digital economy but also expands the related research topics. This paper extends the mechanism path analysis because the index of the digital economy is composed of four aspects in this paper: digital foundation, digital information, digital application, and digital finance. The comprehensive index of low-carbon, inclusive growth consists of three aspects: economic growth, social inclusiveness, and low-carbon ecology. Most of the previous studies have discussed the influence of the first-level comprehensive indicators. Therefore, through dimension reduction analysis, this paper analyzes the carbon reduction channels of the digital economy. It enriches the existing theoretical literature and provides new insights for improving the carbon reduction effect of the digital economy. The third is content innovation. We explore the conditional impact of the digital economy on low-carbon, inclusive growth from the dimensions of key regional location, factor productivity, and information level, which provide detailed empirical evidence for local policymakers.

3. Theoretical Mechanism

The digital economy has high green value, but whether its development is necessarily conducive to regional low-carbon and inclusive growth is worthy of further discussion. Based on the paradox of the digital economy, this paper discusses its influence and mechanisms with respect to low-carbon, inclusive growth.

Digital Economy and Low-Carbon, Inclusive Growth

On the one hand, as a kind of innovative economy, the digital economy has a positive effect on carbon emission reduction. In the process of blending with traditional industries, the digital economy continuously promotes the development of high-quality industries by improving the efficiency of industrial–technological innovation [44]. The digital economy can provide digital technology upgrades for high energy consumption or high pollution industries, which can not only improve energy efficiency but also help to promote an enterprise’s green technology innovation in order to achieve regional low-carbon inclusive growth. Specifically, there are three aspects.

First, the digital economy can improve the resource allocation of high energy consumption, high pollution, and high emission industries. It can help to open up new channels for the flow of factor resources. Specifically, the digital economy can replace and eliminate traditional industries with high pollution and high energy consumption through the substitution effect. Through the enabling effect, we can apply new technology to improve the efficiency of traditional industries. It will also help to promote the transformation of industries with high energy consumption and high pollution. The digital economy can also break market boundaries by alleviating information asymmetry, reducing search, transaction, matching, and replication costs and lowering transaction barriers. It is conducive to the flow of resources in a larger space, optimizes the allocation of resources, and promotes sustainable development [45].

Second, the digital economy helps to give birth to economies of scale and scope and is conducive to the formation of industrial agglomeration, which can improve the efficiency of resource allocation through labor reservoir, intermediate input contribution, and knowledge spillover [46]. It is helpful to improve the efficiency of market operation [47]. In addition, agglomeration improves the efficiency of the division of labor and cooperation among enterprises and promotes the automation of the production process, thus reducing enterprise energy consumption and improving energy efficiency [48]. The improvement of the degree of economic agglomeration can reduce the cost of environmental regulation and environmental publicity and improve their efficiency, thus effectively restraining carbon emissions to a certain extent and contributing to the development of low-carbon, inclusive growth [20].

Third, digital technology accelerates the spread of the concept of green health, gradually affects people’s productivity and lifestyles, promotes an increase in green consumption demand, and then leads to changes in human preferences and economic and social systems [49]. The development of the digital economy, as represented by the Internet and big data, makes the market more transparent. Enterprises are forced to innovate continuously and pay attention to ecological protection in order to reduce negative effects on the environment. If enterprises want to develop in the market for a long time, they need to take into account both economic and environmental benefits. With extensive economic development, all kinds of environmental problems in China emerge one after another, which deepens people’s desire for green waters and green mountains. Therefore, green consumption drives green innovation and contributes to low-carbon and inclusive growth.

On the other hand, the digital economy has a negative impact on carbon emission reduction.

Some scholars have found that some digital platforms use digital technology to form monopolies, which are not conducive to the sustainable development of the regional economy [50,51,52,53,54]. Therefore, digital technology may lead to market monopolies and market failure, hindering the free flow of resources, which is not conducive to green technology innovation and industrial upgrading. For example, Liu Zhibiao and Kong Linchi [55] believe that the exercise of monopolistic market power by platform enterprises may lead to market disintegration, which will hinder the healthy development of the market economy. Aghion et al. [56] found that unreasonable education mechanisms and irregular labor markets hurt the healthy development of the job market, thus restraining the healthy development of the economy. Acemoglu [57], found that the excessive use of artificial intelligence technology may distort the proportion of capital–labor factors, which reduces production efficiency and is not conducive to economic growth.

In addition, the development of the factor market is relatively slow, and there are double distortions in the allocation and price of the commodity market and factor market, which will not be conducive to improving green total factor productivity [58]. According to previous studies, the unbalanced development of the digital economy may lead to new digital inequality and the “digital divide” phenomenon, leading to the “HONGXI effect”, which may lead to a wide gap between regional urban and rural development, and then lead to other unfair phenomena [59]. The development of the digital economy may accelerate the flow of all kinds of factors from peripheral cities to central cities, aggravating the agglomeration advantage of big cities and producing the “Matthew effect” of the strong getting stronger and the weak getting weaker, further distorting the allocation of the factors. it’s not conducive to regional low-carbon development [21]. Some scholars have found that more than a certain level of urban agglomeration may bring negative effects that increase the intensity of carbon emissions. Because of the intensification of urban agglomeration, there may be a congestion effect [60] and environmental pollution [61]. In addition, some scholars believe that the growth in energy demands brought about by the development of the digital economy leads to the pollution effect exceeding the emission reduction effect. Therefore, it aggravates environmental pollution because the application of digital equipment and infrastructure directly increases the demand for energy [62]. In addition, the development of the digital economy will not only generate more demand for ICT through the resulting economic growth but also increase the availability of resources while breaking down technical barriers. This will further aggravate the depletion of resources, which, in turn, exacerbates carbon emissions and environmental pollution [63]. Through the analysis, the following hypotheses are put forward.

H: The digital economy has an inverted U-shaped impact on low-carbon, inclusive growth. Specifically, the digital economy can promote energy conservation and emission reduction by optimizing the allocation of regional resources in order to achieve regional low-carbon, inclusive growth. When the development level of the digital economy is too high, it may cause a resource mismatch to restrain low-carbon, inclusive growth.

To visually show how the digital economy affects low-carbon, inclusive growth, we briefly draw the framework of this article. As shown in Figure 1.

4. Econometric Model and Data Description

4.1. Econometric Model

This paper discusses the impact of the digital economy on low-carbon, inclusive growth. It belongs to the causal identification test. The five classical econometric models of the causal identification test include: controlled regression, regression discontinuity design, difference-in-difference, fixed effects regression, and instrumental variable. The statistical requirement of control regression is that, given the interference variable C, the distribution of the potential result, Y, should be conditionally independent of the independent variable, X. RDD and DID are mostly used to evaluate the effectiveness of policies, which are not suitable for the analysis of this paper. The instrumental variable method is suitable for the test of mutual causality, but the mutual causality in this paper is not obvious. In contrast, the fixed effect regression method is the most suitable for this paper. In addition, we also perform a random effect and fixed effect model selection test, i.e., a test of overidentifying restrictions: fixed vs. random effects cross-section timeseries model: Sargan–Hansen statistic 155.726, chi-sq (7), p-value = 0.0000. The results of the random effect were rejected and the fixed effect model was supported.

4.1.1. Benchmark Model

This paper uses panel data from 30 provinces. First of all, unobservable effects, such as the geographical location of each province, may affect the level of low-carbon, inclusive growth. These factors change with the individual and not with time, but we can control the individual fixed effect in the panel data. Similarly, the provincial macro policy environment and other factors will also have an impact on low-carbon, inclusive growth, such as environmental regulation and changes in government change over time. We control for this by constructing a virtual time variable, i.year. This can more cleanly identify the impact of the digital economy on low-carbon, inclusive growth. Therefore, this paper chooses the two-way fixed effect model for analysis. To verify the hypothesis of this article, this paper constructs the following nonlinear econometric model drawing on the work of Chen Aizhen and Wang Xianbin [64,65]:

$$Ligrowt{h}_{it}={\alpha }_0+{\alpha }_{1}digieco{n}_{it}+{\alpha }_{2}digieco{n}_{it}{}^2+{\alpha }_c{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

In Equation (1), Ligrowth_it represents the level of low-carbon and inclusive growth of region I in the t period. digiecon_it represents the comprehensive level of the regional digital economy. Vector X_it represents the control variable, µ_i represents the time-fixed effect, λ_t represents the individual fixed effect, and ε_it represents the random disturbance term.

4.1.2. Mechanism Model

To test how the digital economy affects low-carbon, inclusive growth through resource allocation, we need to construct a mechanism model. Refer to the practice of Wang Xianbin et al. [65]. First of all, resource allocation is regarded as a dependent variable, and the digital economy and its square term are used to regress the resource allocation index. Then, the resource allocation index is added to model (1) as a control variable for regression. The specific construction model is shown in Equations (2) and (3).

$$Mecha{n}_{it}={\delta }_0+{\delta }_{1}digieco{n}_{it}+{\delta }_{2}digieco{n}^2{}_{it}+{\delta }_c{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

$$Ligrowt{h}_{it}={\gamma }_0+{\gamma }_{1}digieco{n}_{it}+{\gamma }_{2}digieco{n}^2{}_{it}+{\gamma }_{3}Mecha{n}_{it}+{\gamma }_c{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(3)

The Mechan in Equations (2) and (3) are mechanism variables. In this paper, we use the reverse index as the proxy variable. The reverse indicators of resource allocation in this paper are as follows: resource mismatch index (Mismatch), capital factor mismatch index (Miscap), and labor factor mismatch index (Mislab). The other variables are consistent with the explanation of Equation (1) and will not be repeated.

4.2. Variable Definition

4.2.1. Low-Carbon, Inclusive Growth

The measurement of low-carbon growth is relatively mature. Chung et al. [66] constructed a Malmquist–Luenberger (ML) index, which is used to measure the growth of TFP taking into account environmental pollution emissions. However, the ML index method has three obvious defects: the intertemporal mixed DDF may have no solution, the index does not have cyclic characteristics, and the angle and radial efficiency measures are biased [67,68]. Some scholars have studied the concept of inclusive growth [69,70]. However, the measurement of low-carbon, inclusive growth is relatively rare. Based on existing research, this paper includes two aspects of green low-carbon and inclusive growth and constructs a low-carbon, inclusive growth index system from 3 first-level indicators and 54 fourth-level indicators of economic growth, social inclusion, and low-carbon ecology. As shown in

Table 1. Comprehensive Index of Low-carbon, inclusive Growth.

Dimension	Domain	Indicator Name	Measurement Method	Unit	Attribute
Economic growth	Power of economic growth	Investment capital	The stock of fixed asset investment in the whole society/GDP	%	+
		Consumption expenditure	Total retail sales of social consumer goods/GDP	%	+
		Net export	(Total exports minus total imports)/GDP	%	+
	Economic growth potential	Human capital	Number of graduates from general colleges/total population	%	+
		innovate	Investment in research and experimental development/GDP	%	+
			The proportion of turnover in the technology market	%	+
		The openness of foreign investment	Foreign direct investment/GDP	%	+
		Upgrading of industrial structure	The ratio of the output value of the tertiary industry to that of the secondary industry	%	+
		Economic stability	Economic volatility	%	−
			CPI	%	−
	Quantity of economic growth	Per capita GDP	Per capita GDP	RMB	+
		labor productivity	GDP/Number of employed persons	RMB/one person	+
		GDP growth rate	GDP growth rate	%	+
Social inclusion	Fair opportunity	Fair right to exist	Sex ratio of male to female	/	−
			Number of live births in maternal health care	one	+
		Fair employment opportunities	Employment rate	%	+
		Fair educational opportunities	Ten thousand people have educational resources.	Every ten thousand people	+
			Illiteracy and semi-illiteracy rate	%	−
	Social security equity	Unemployment insurance	Unemployment insurance coverage	%	+
		Industrial injury insurance	The coverage rate of industrial injury insurance	%	+
		Birth security	Maternity insurance coverage	%	+
		Medical insurance	Coverage of basic medical insurance	%	+
			Ten thousand people own the number of beds in hospital health institutions	Bed/10,000 people	+
		Old-age security	Basic pension coverage rate	%	+
			Elderly dependency ratio	%	−
		Housing security	Per capita residential area	Square meters/one person	+
			The proportion of housing investment in the whole society	%	+
	Improvement of public facilities	Improvement of rigid public facilities	Transport length (road, rail, and river transport)	Kilometers	+
			Long-distance optical cable line length	Kilometers	+
			Public toilets for every 10,000 people	one	+
		Improvement of soft public facilities	Phone coverage (mobile phone + landline phone)	%	+
			Software business revenue	RMB 100 million	+
			Per capita ownership of library collections	Book/one person	+
	Achievement sharing	Sharing level between urban and rural areas	Urban–rural income ratio	/	−
			Urban–rural consumption ratio	/	−
		Regional sharing level	Per capita GDP of each province per capita GDP of the whole country	/	+
			The average consumption level of residents in each province/average consumption level of the whole country	/	+
		Poverty reduction effect	Gini coefficient of resident incomes	/	−
			Engel coefficient	%	−
			The proportion of minimum guaranteed residents	%	−
Low carbon ecology	Low carbon production	Production consumption	Energy consumption per unit output value	Tons/ten thousand RMB	−
			Industrial water consumption per unit output value	Cubic meters/ten thousand RMB	−
			Industrial electricity consumption per unit output value	Kilowatt-hour/RMB	−
			The proportion of resource tax	%	−
		Production emission	Completed investment intensity of industrial pollution control	%	+
			Carbon dioxide emissions per unit output value	Tons/ten thousand RMB	−
			The comprehensive utilization rate of industrial solid waste	%	+
	Low carbon consumption	Green life	Per capita household water consumption	Tons/one person	−
			Harmless treatment rate of municipal solid waste	%	+
			Handling area of special vehicle equipment for city appearance and sanitation per unit	Ten thousand square meters/per set	+
		Green travel	Public transport for every 10,000 people (number of buses, trams, light rail operators)	Cars/per ten thousand people	+
			Ten thousand people have a number of taxis	Cars/per ten thousand people	+
			Total passenger transport volume of public transport	100 million person-times	+
	Ecologically livable	The greening of the city	Green coverage rate in the built-up area	%	+
			Urban per capita public green space area	Square meters per person	+
			Per capita has park green space area	Square meters per person	+
		Tourism culture	Tourism resources (proportion of tourism income)	%	+
			Cultural resources (books + periodicals + print numbers of general newspapers)	100 million copies	+
	Ecological endowment	Coal resources	Coal reserves	100 million tons	+
		Water resources	The total amount of water resources	100 million cubic meters	+
		Forest resources	The proportion of nature reserve area	%	+
			forest coverage	%	+
	Ecological governance	Area for prevention and control of soil erosion	Area for prevention and control of soil erosion	Hectare	+
		Afforestation area	Afforestation area	Hectare	+
		Investment in prevention and control of geological disasters	Investment in prevention and control of geological disasters	RMB 100 million	+
		Prevention and control of forest diseases, pests, and rats	Total prevention and control rate of forest diseases, pests, and rodents	%	+

Note: Due to a lack of data, this paper calculates the green development comprehensive index and subsystem index of 30 provinces except Tibet, Hong Kong, Macao, and Taiwan. The last column is in Table 1. The positive and negative attributes indicate that there is a positive and negative relationship between the index and the comprehensive index. It is also to prepare for the use of the entropy method for calculation.

, the entropy method is used to calculate the comprehensive index of low-carbon, inclusive growth (Ligrowth). In addition, the entropy weight TOPSIS method is used to calculate the low-carbon, inclusive growth index (Tligrowth) for the robustness test.

4.2.2. Digital Economy Development Index

As the measurement method of the regional digital economy is relatively perfect, we refer to the practice of Guoge Yang [51,71]. The comprehensive index system of the digital economy is constructed from four aspects: digital infrastructure, digital resources, digital application, and digital finance. The comprehensive development index of the digital economy (digiecon) is calculated with the entropy method.

4.2.3. Mechanism Variable

The digital economy mainly affects regional carbon emission performance by affecting the allocation of resources. Therefore, to analyze the specific mechanism of the digital economy affecting the development of low-carbon transformation, this paper constructs the asset mismatch index as the reverse index of resource allocation efficiency. This paper draws lessons from the work of Shao Ting and other scholars [72] to calculate the asset mismatch index. The calculation of the resource mismatch index formula is as follows:

$$Mismatch=\frac{f_i}{f}$$

(4)

where fi indicates the financing cost of industrial enterprises above the provincial scale, and f represents the average financing cost of industrial enterprises above the scale. If $Mismatch$ < 0, it means that there is no resource allocation distortion in the province; when $Mismatch$ > 0, it means that enterprises will obtain loans at a higher cost of capital. The larger the capital cost, the higher the cost of capital, that is, the misallocation or distortion of resources. In addition, to make an in-depth analysis of the role of capital and labor factors, we draw lessons from the work of Chen Yongwei [73] and Hu Weimin, and Cui Shuhui [74] to construct the capital mismatch index (Miscap) and the labor mismatch index (Mislab) for further testing.

4.2.4. Control Variable

The control variables of this paper are as follows: (1) Regional innovation level (innovate) measured by the ratio of the number of regional researchers to the total population of each province. Innovation is the endogenous driving force of economic growth, and green technology innovation is also an important means of carbon emission reduction. We believe that regional innovation is conducive to low-carbon development. (2) The economic development (develop) is characterized by the per capita GDP of each province. Economic development is directly related to regional social inclusiveness, economic growth, and carbon emission efficiency. In an area with a high level of the economy, on the one hand, more income is spent on environmental governance and green technology research and development, which is conducive to low-carbon green development. On the other hand, China’s economic growth is mainly extensive, and the higher the level of industrialization is, the more serious the carbon emission problem may be. Therefore, the relationship between regional economic development level and low-carbon, inclusive growth is uncertain. (3) The financial development (finance) is measured by the ratio of deposit and loan amounts to provincial GDP. Financial development helps to support low-carbon and inclusive growth. This paper argues that there is a positive correlation between the level of financial development and low-carbon development. (4) The material capital level (material) is measured by the ratio of infrastructure investment to GDP. China’s economic development is mainly extensive. Investment in fixed assets promotes the development of heavy industry and high energy-consuming industries. Therefore, we believe that there may be a negative correlation between this indicator and low-carbon, inclusive growth. (5) The level of urbanization (urban) is measured using the amount of urban population as the proxy variable. On the one hand, the high level of urbanization shows that the population is dense and the industrial industry is more developed. Industrialization may lead to greater carbon emissions. On the other hand, the high level of urbanization indicates that the regional economy is developed: People have a high income level and have higher requirements for the environment. At the same time, the government has more revenue for environmental governance. Therefore, it is conducive to low-carbon and inclusive growth.

4.2.5. Other Variables

(1) Tool variable (instru). Based on endogenous consideration and the work of Yang Guoge [51] and others, the digital circuits in the history of each province are used as tool variables and multiplied by the number of Internet users in the previous year to construct panel tool variable data as tool variables of the regional digital economic index for regression. (2) Missing variables. The education level of each province (education) and the level of informatization of each province (information). These two variables are not directly related to low-carbon, inclusive growth, but there may be an indirect relationship. Consider our alternative missing variables. The level of education (education) is expressed by the ratio of undergraduate students and above students in each province to the total population of each province. The higher the ratio, the more the number of college students and the higher the level of education in the province. Information level (information): we use the ratio of the total number of post and telecommunications businesses in each province to the total population at the end of the year. The higher the ratio, the closer the connection between the region and the outside world. It shows that the higher the degree of informatization in this place.

4.3. Data Sources and Descriptive Statistics

Table 2. Descriptive statistics.

Variables	Mean	p50	sd	Min	Max	N
Ligrowth	0.286	0.262	0.087	0.168	0.602	270
Tligrowth	0.275	0.243	0.089	0.173	0.550	270
digiecon	0.210	0.142	0.197	0.0310	0.867	270
digiecon 2	0.0830	0.0200	0.155	0.00100	0.752	270
innovate	0.0910	0.0640	0.0760	0.0220	0.604	270
develop	5.198	4.376	2.657	1.619	16.45	270
finance	1.437	1.334	0.441	0.671	2.577	270
material	17,000	14,000	12,000	1436	57,000	270
urban	57.63	55.59	12.18	34.96	89.60	270
Misresource	0.037	0.043	0.113	−0.459	0.462	270
Miscap	0.275	0.236	0.214	0.001	1.471	270
Mislab	0.374	0.300	0.336	0.001	2.126	270

shows the descriptive statistics results of this article. P50 represents a 50% quantile value. The 50% quantile of the digital economy and low-carbon, inclusive growth is less than its average. It shows that the levels of the digital economy and low-carbon, inclusive growth in Chinese provinces are low. It can be seen that the average value of the comprehensive index of low-carbon, inclusive growth is 0.286 and the variance is 0.087. It shows that the overall level of inclusive growth is low, and the gap between regions is large, indicating that the development of some regions is still accompanied by serious carbon emissions and pollution. Other places have begun to change their mode of economic growth. Similarly, the average value of the regional digital economic index is 0.210 and the variance is 0.197. It can be seen that the digitization level of each region in China is generally low, and there is a great gap between regions. Other variables show the differences in regional development. This paper takes 30 provinces from 2011 to 2019 as the research sample. The original data come from the China Statistical Yearbook and the statistical yearbooks of 30 provinces.

5. Empirical Result

5.1. Benchmark Regression

Table 3. Benchmark regression.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
digiecon	0.0146	0.0110	0.0474 ***	0.0441 ***	0.0453 ***	0.0277 **
	(0.0091)	(0.0073)	(0.0156)	(0.0167)	(0.0165)	(0.0124)
digiecon2			−0.0213 **	−0.0170	−0.0158	−0.0205 ***
			(0.0107)	(0.0114)	(0.0118)	(0.0073)
innovate		0.0033		0.0024	0.0017	0.0032
		(0.0039)		(0.0039)	(0.0032)	(0.0038)
develop		0.0126		0.0191 ***	0.0223 ***	0.0125
		(0.0075)		(0.0071)	(0.0054)	(0.0074)
finance		0.0112 *			0.0207 ***	0.0113 **
		(0.0055)			(0.0075)	(0.0055)
material		−0.0090 **			0.0020	−0.0083 **
		(0.0040)			(0.0041)	(0.0040)
urban		0.0406 **				0.0369 *
		(0.0198)				(0.0194)
Time fixed	Yes	Yes	No	No	No	Yes
Province fixed	Yes	Yes	No	No	No	Yes
Constant	0.2831 ***	0.3046 ***	0.2850 ***	0.2948 ***	0.3099 ***	0.3044 ***
	(0.0040)	(0.0145)	(0.0134)	(0.0107)	(0.0130)	(0.0140)
Observations	270	270	270	270	270	270
R-squared	0.0967	0.2997	0.1137	0.1295	0.2080	0.3170
Number of id	30	30	30	30	30	30

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

reports the return of the digital economy to low-carbon, inclusive growth. Most existing studies believe that there is a linear relationship between the digital economy and low-carbon, inclusive growth. We report the linear regression results of both columns (1) and (2). The results show that, with or without control variables, the development of the digital economy hurts low-carbon, inclusive growth, but the statistical level is not significant. Column (3) shows the results of the nonlinear estimation of low-carbon, inclusive growth in the digital economy. By adding the square term and all the control variables to control the time and individual effects, the final estimated results are shown in column (6). It shows that the first-order term coefficient of the digital economy is 0.0277 and is positive at the 1% statistical level. The quadratic coefficient of the digital economy is −0.0205, and is negative at the 1% statistical level. It shows that the impact of the digital economy on regional low-carbon, inclusive growth is two-sided. In other words, the initial stage of the development of the digital economy helps to promote the development of regional low-carbon transformation; when the level of the digital economy is too high, it may restrain regional low-carbon, inclusive growth. This conclusion is consistent with existing research conclusions [51].

According to the calculation, the threshold of the digital economy is 0.3081. According to statistical analysis, the results show that the development of the digital economy in 75.9% of the observed areas in China can promote low-carbon, inclusive growth. On the surface, the overall level of the digital economy in China is in the early stage of development. The development of the digital economy in 24.1% of the observed areas may inhibit low-carbon, inclusive growth. From the provincial point of view, digitally and economically developed areas such as Beijing, Shanghai, Zhejiang, and Guangdong have had a high level of digital economy in recent years. This is in line with the actual regional situation. In recent years, Alibaba, Tencent, and other Chinese ecommerce giants have monopolized behavior, and the two head offices are in Zhejiang and Guangdong. The existing monopoly problems of digital leading enterprises such as Tencent and Ali are also consistent with the conclusions of this paper [52,53,54,75], which proves the effectiveness of the conclusions of this study to a certain extent. Therefore, on the whole, the development of the digital economy in most areas will help to enhance the level of low-carbon, inclusive growth. In some developed areas, there may be some problems with platform monopolies [75]. Therefore, we should strengthen the supervision of digital platforms to prevent platform monopolies from distorting the allocation of resources.

To verify the existence of an inverted U-shaped relationship, we use UTEST to test the nonlinear relationship. According to the UTEST command, the nonlinear relationship test results show that the p-value is equal to 0.0032. It shows that the original hypothesis is significantly rejected at a 1% statistical level, and the extreme point is 0.3081, falling into the confidence interval [−0.9110, 3.3375]. Therefore, the overall U-shaped relationship test is established. For intuitive display, we draw a U-shaped diagram, as shown in Figure 2. The initial slope (4.7311) is a positive rising stage (the low level of the low digiecon digital economy). The T value is 3.0055. After exceeding the extreme point, the slope of the descending stage is positive −11.7560. The T value is −2.9289 (up digiecon digital economy, high-level segment). According to the test, the inverted U-shaped relation test is established.

In Table 3, the coefficients of the regional R&D level and the economic development level are all positive. It shows that these factors play a positive role in improving the performance of carbon emissions. However, it is not statistically significant, which may be because these factors are not the direct factors affecting carbon emissions. It shows that the higher the level of financial development, the more local funds support low-carbon, inclusive growth. The high level of urbanization indicates that the regional economy is developed. People have a high-income level and have higher requirements for the environment. At the same time, the government has more revenue for environmental governance. Therefore, it is conducive to low-carbon and inclusive growth. Most of China’s infrastructure investment is in high energy-consuming industries, such as iron and steel building materials. Although this promotes rapid economic growth, promotes employment, and increases resident incomes, it also leads to problems, such as widening the income gap between urban and rural areas and increasing carbon emissions. The estimated result is negative, although the absolute value of the coefficient is very small (0.0083). However, it has been shown that the negative impact of infrastructure investment outweighs the positive impact in recent years. On the whole, the estimated results of the control variables are consistent with the actual situation.

5.2. Robustness Test

5.2.1. Endogenous Treatment

Considering the endogenous problems caused by the omitted variables and two-way causality, we use tool variable development to deal with them. Drawing lessons from the work of Guoge Yang and others, the digital circuit in the history of each province is used as a tool variable and estimated with the two-stage least square method (2SLS) [51,76]. Based on the consideration of robustness, generalized method of moments estimation (GMM) is used to test endogenesis. The estimated results are shown in

Table 4. Endogenous test results.

	(1)	(2)	(3)	(4)
Variables	First Stage	First Stage	The Second Stage	GMM
digiecon			6.6641 ***	0.9314 ***
			(2.2035)	(0.2976)
digiecon2			−11.3613 ***	−0.6084 **
			(2.1380)	(0.3018)
L.lowctfp	0.0240 *	0.0211	0.0579
	(0.0147)	(0.0193)	(0.1633)
innovate	−0.0223	−0.0305	−0.1339	−0.0584
	(0.0326)	(0.0376)	(0.3332)	(0.0803)
develop	0.2187 **	0.1504 ***	0.0040	−0.2546 *
	(0.0674)	(0.0377)	(0.7013)	(0.1338)
finance	−0.0126	−0.0019	−0.1147	0.4505 ***
	(0.0409)	(0.0292)	(0.2918)	(0.1651)
material	−0.0257	0.0127	0.2607	0.2366
	(0.0454)	(0.0228)	(0.2425)	(0.1496)
urban	0.1908 *	−0.0185	−1.5525	−0.2902 ***
	(0.1039)	(0.0603)	(1.0632)	(0.0878)
Time-fixed	Yes	Yes	Yes	Yes
Province-fixed	Yes	Yes	Yes	Yes
Constant	0.1355 **	0.0411	0.7689	2.0453 ***
	(0.0674)	(0.0513)	(0.5326)	(0.5040)
instru	0.1527 *	0.0673 *
	(.0838)	(0.0570)
instru2	−0.0697 *	−0.0163 *
	(0.0395)	(0.0256)
F value	647.39	146.77
AR2				0.059
Hansen				1
Observations	240	240	240	270
R-squared	0.1397	0.5462	0.0069
Number of id	30	30	30	30

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

.

Columns (1) and (2) report the results of the first phase of estimates. It can be seen that the tool variable (instru) and its square term (instru2) are significant at the statistical level of 10%. The F value passed the correlation test, meeting the requirements of relevance. The estimated results of column (3) show that the impact of the digital economy on low-carbon, inclusive growth is consistent with the conclusion of benchmark regression. In addition, column (4) reports the results estimated using the GMM method. The results show that the conclusion of the GMM test is still robust. The AR2 value and the Hansen value pass the relevance test. Therefore, after solving the endogenous problems, the digital economy still has an inverted U-shaped effect on low-carbon inclusive growth, indicating that there is a “double-edged sword” effect. This is consistent with the benchmark estimate.

5.2.2. Replacement Variables and Data Smoothing

We replace the dependent variable measure method for re-testing. The results of the test after replacing the dependent variables are shown in column (1) of

Table 5. Robustness test.

	(1)	(2)	(3)	(4)
Variables	Replace Dependent Variable	Smoothing Treatment	Missing Variable Processing	Negative Binomial Regression
digiecon	0.0224 *	0.0273 **	0.0263 **	0.0328 ***
	(0.0124)	(0.0124)	(0.0114)	(0.0080)
digiecon2	−0.0217 **	−0.0204 **	−0.0198 ***	−0.0169 **
	(0.0086)	(0.0077)	(0.0071)	(0.0080)
innovate	0.0033	0.0008	0.0051	0.0012
	(0.0042)	(0.0097)	(0.0033)	(0.0026)
develop	0.0098	0.0137 *	0.0127 *	0.0159 ***
	(0.0091)	(0.0077)	(0.0068)	(0.0034)
finance	0.0150 *	0.0094 *	0.0113 *	0.0118 ***
	(0.0074)	(0.0055)	(0.0056)	(0.0032)
material	−0.0063	−0.0078 *	−0.0066	−0.0066 **
	(0.0075)	(0.0039)	(0.0040)	(0.0028)
urban	0.0276	0.0390 *	0.0190	0.0361 ***
	(0.0254)	(0.0204)	(0.0215)	(0.0062)
education			0.0145
			(0.0086)
information			−0.0018
			(0.0018)
Time-fixed	Yes	Yes	Yes	Yes
Province-fixed	Yes	Yes	Yes	Yes
Constant	0.2868 ***	0.3063 ***	0.3026 ***	0.3087 ***
	(0.0148)	(0.0158)	(0.0129)	(0.0095)
Observations	270	270	270	270
R-squared	0.2146	0.3060	0.3514
Number of id	30	30	30	30

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

. The estimated results show that the digital economy still has an inverted U-shaped impact on low-carbon, inclusive growth. In addition, considering the interference of some data outliers, the estimation is biased. For this reason, this paper performs extreme value smoothing for the data in the 1% and 99% quantiles. The estimated results are still robust.

5.2.3. Solving Missing Variables and Replacement Estimation Methods

Based on the consideration of missing variables, we add the provincial education level (education) and information level (information) to the model as a control. The estimated results are shown in column (3) of Table 5. The estimated results are still robust. In addition, considering that the dependent variables are nonnegative, we use the negative binomial method for re-estimation, and the result is shown in column (4). It shows that the digital economy still has an inverted U-shaped impact on low-carbon, inclusive growth.

5.3. Heterogeneity Test

5.3.1. Factor Productivity

According to the research of Yang Ligao and other scholars [77], different factor input types affect the development level of enterprises; thus, this will also affect the carbon emission level of enterprises. This paper believes that the level of regional factor utilization will affect the level of low-carbon, inclusive growth in each region. We take the average factor productivity as the standard and carry out a group test according to the level of labor, capital, and energy productivity. The estimated results are shown in

Table 6. Factor grouping test.

	(1)	(2)	(3)	(4)	(5)	(6)
	Low Productivity			High Productivity
Variables	Capital	Labor	Energy	Capital	Labor	Energy
digiecon	0.0368 **	0.0094	0.0202 *	0.0111	0.0087	0.0061
	(0.0146)	(0.0288)	(0.0126)	(0.0131)	(0.0052)	(0.0260)
digiecon2	−0.0555 **	−0.0072	−0.0154 **	−0.0243 *	−0.0088 *	0.0430
	(0.0268)	(0.0117)	(0.0074)	(0.0125)	(0.0050)	(0.0900)
innovate	0.0017	0.0239	0.0021	−0.0433 *	0.0220	0.0010
	(0.0023)	(0.0188)	(0.0042)	(0.0207)	(0.0134)	(0.0044)
develop	0.0176	−0.0026	0.0410 ***	0.0108	0.0071	0.0388
	(0.0122)	(0.0086)	(0.0102)	(0.0075)	(0.0057)	(0.0409)
finance	0.0111 **	0.0041	0.0152 **	0.0158	0.0041	0.0231 *
	(0.0042)	(0.0112)	(0.0056)	(0.0102)	(0.0057)	(0.0117)
material	−0.0034	−0.0087	−0.0016	0.0092 ***	−0.0049	−0.0033
	(0.0040)	(0.0143)	(0.0035)	(0.0029)	(0.0062)	(0.0083)
urban	0.0380 **	0.0216 *	0.0039	0.0703 ***	0.0383 *	−0.0235
	(0.0180)	(0.0118)	(0.0204)	(0.0132)	(0.0204)	(0.0405)
Time-fixed	Yes	Yes	Yes	Yes	Yes	Yes
Province-fixed	Yes	Yes	Yes	Yes	Yes	Yes
Constant	0.2886 ***	0.3552 ***	0.3031 ***	0.3735 ***	0.3167 ***	0.3118 ***
	(0.0140)	(0.0302)	(0.0208)	(0.0065)	(0.0112)	(0.0507)
Observations	199	71	213	57	171	99
R-squared	0.3384	0.5271	0.3753	0.5658	0.4254	0.4165
Number of id	25	18	26	12	23	17

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

. The estimated results of columns (1) to (6) in Table 6 are shown. Compared to the high productivity group, the digital economy on low-carbon, inclusive growth in areas with low capital and energy productivity is more obvious.

In the group with low capital factor productivity, the first term coefficient of the digital economy is (0.0368) and is significant at a 5% statistical level. The quadratic term coefficient of the digital economy is −0.0555, and it is significant at a 5% statistical level. In the group with lower energy factor productivity, the coefficient of digiecon (0.0202) is significant at a 10% statistical level. The coefficient of digiecon2 (−0.0154) is significant at a 5% statistical level. In the group with lower labor factor productivity, the impact of the digital economy on low-carbon, inclusive growth is not statistically significant. In the group with higher productivity, the one-time item of the digital economy is not significant. The secondary term of the digital economy is significant at a 10% statistical level in areas with higher capital and labor factor productivity, while the statistical level is not significant in areas with high energy production. The economic significance shows that the digital economy in areas with low capital and energy productivity is more significant. From the size of the coefficient, compared to the areas with lower energy productivity, the digital economy has a greater promoting effect on low-carbon, inclusive growth in areas with lower capital. Similarly, on the right side of the extreme point, the digital economy has a relatively greater inhibitory effect on low-carbon, inclusive growth in areas with low capital productivity. In other words, the use of digital technology to improve energy and capital efficiency can effectively promote low-carbon, inclusive growth. However, beyond a certain degree of development, digital technology will increase total energy consumption and lead to capital misallocation, which will also inhibit low-carbon, inclusive growth. In the group with higher factor productivity, the inhibitory effect of the digital economy on low-carbon, inclusive growth occurs in the group with higher capital and labor factor productivity. This may be related to the imbalance of regional capital and labor factors. This conclusion is consistent with the findings of Acemoglu et al. [78]. Through statistical observations, it is found that the current levels of capital and energy factor productivity in most areas are low, and 91.4% of the observations are on the left side of the extreme point. Therefore, in regions with low factor productivity, the digital economy will help to improve the level of low-carbon, inclusive growth.

5.3.2. Information

The level of informatization represents the level of regional development to a certain extent, so the difference in the level of informatization will also affect its learning and absorption of digital technology. In this paper, the ratio of mail quantity to regional GDP is used to measure the level of informatization, and the average value is used as the standard for high and low grouping. The results of grouped regression are shown in

Table 7. Grouping test of information and regional.

	(1)	(2)	(3)	(4)
	Informatization		Region
Variables	Low Level	High Level	Coastal Area	Inland
digiecon	0.0063	0.0918 ***	0.0479 **	0.0356 **
	(0.0059)	(0.0217)	(0.0203)	(0.0141)
digiecon2	−0.0076	−0.0474 ***	−0.0284 **	−0.0686 **
	(0.0047)	(0.0135)	(0.0120)	(0.0293)
innovate	0.0056 ***	−0.0059	−0.0245	0.0026
	(0.0016)	(0.0042)	(0.0295)	(0.0024)
develop	0.0185	0.0078	0.0322 **	0.0372 ***
	(0.0109)	(0.0075)	(0.0112)	(0.0108)
finance	0.0041	0.0198 **	0.0128	0.0114 ***
	(0.0063)	(0.0076)	(0.0121)	(0.0033)
material	−0.0090 **	0.0092	−0.0096	−0.0023
	(0.0039)	(0.0074)	(0.0111)	(0.0028)
urban	0.0237	−0.0056	0.0951 *	0.0303 **
	(0.0231)	(0.0203)	(0.0429)	(0.0140)
Time-fixed	Yes	Yes	Yes	Yes
Province-fixed	Yes	Yes	Yes	Yes
Constant	0.2935 ***	0.3406 ***	0.3463 ***	0.2885 ***
	(0.0151)	(0.0106)	(0.0157)	(0.0153)
Observations	188	82	90	153
R-squared	0.3092	0.5198	0.5410	0.4907
Number of id	29	29	10	17

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

. The estimated results from columns (1) and column (2) are displayed. In the group with a higher level of informatization, the primary term coefficient of the digital economy (0.0918) is significant at a 1% statistical level. The quadratic coefficient of the digital economy (−0.0474) is significant at a 1% statistical level. Compared to the group with a lower level of informatization, the digital economy has a more significant impact on the low-carbon, inclusive growth of the group with a higher level of informatization. This shows the digital economy has a greater promoting effect on low-carbon, inclusive growth in areas with a higher level of informatization. Beyond the extreme point, the digital economy has a greater inhibitory effect on low-carbon, inclusive growth in areas with higher informatization. According to statistical analysis, the current level of informatization in about 70% of the areas is low. Therefore, to speed up the development of informatization is to make full use of digital technology to improve the level of regional informatization. The government should provide support to improve the construction of information infrastructure and to formulate information development policies. However, at the same time, we should also do a good job in the supervision of the digital economy and actively deal with the negative impacts of the development of the digital economy.

5.3.3. Regional

We know that the economic development of China’s coastal areas has a natural opening-up advantage over inland areas. If the basis of economic development is different, will the impact of the digital economy on low-carbon, inclusive growth be affected by the region? To answer this question, we divided China’s provinces into coastal and inland areas for testing. The estimated results are shown in columns (3) and (4) in Table 7. The grouping results show that the digital economy is significantly below the 5% statistical level, whether in the coastal or inland areas. It shows that the digital economy on low-carbon, inclusive growth is less affected by regional location. However, from the grouping coefficient, we can see that the primary term coefficient of the digital economy in inland areas is 0.0356. The first-order term coefficient of the digital economy in coastal areas is 0.0479. It shows that the digital economy has a greater promoting effect on low-carbon, inclusive growth in coastal areas. Compared to the absolute value of the quadratic coefficient of the digital economy, on the right side of the inverted U-shaped, the digital economy has a greater inhibitory effect on low-carbon, inclusive growth in inland areas. This may be due to underdeveloped economic development in inland areas. Regional development is guided by GDP. After improving the level of digitization, the application of digital technology is beneficial to improving labor productivity and energy efficiency. However, increased income may have increased the total energy consumption, thus increasing the total carbon emissions. This is consistent with existing research conclusions [31,34]. According to statistics, in the inland group, 98.03% of the observed areas are located on the left side of the extreme point, indicating that the digital economic level of most inland areas is in the early stages of development. Therefore, government departments should vigorously develop a digital economy that is conducive to regional low-carbon, inclusive growth. At the same time, the development of the digital economy brings higher economic benefits. However, due attention has not been paid to the impact of the real economy and the widening of the income gap between urban and rural areas [79]. Therefore, based on improving the level of opening up and the digitization of the western region, the government should prevent the negative impact of the digital economy.

6. Further Analysis

6.1. Mechanism Analysis

Models (2) and model (3) test the mediating effect of resource allocation. According to the previous analysis, the development of the digital economy will improve the efficiency of the resource allocation of capital, labor, and other factors, which will contribute to regional low-carbon and inclusive growth. In this paper, the resource distortion index is used as the reverse index of resource allocation. The mismatch index of resources is calculated in Formula (4). The specific estimates are shown in

Table 8. Regression of action mechanism.

	(1)	(3)		(5)	(2)	(4)		(6)
Variables	Mismatch		Miscap		Mislab		Ligrowth
digiecon	−0.1304 **	−0.0053 *		−0.0133 *	0.0269 **	0.0278 **		0.0276 **
	(0.0583)	(0.0599)		(0.0437)	(0.0121)	(0.0124)		(0.0121)
digiecon2	0.0889 **	0.0016 **		0.0728	−0.0200 ***	−0.0205 ***		−0.0198 ***
	(0.0380)	(0.0686)		(0.0577)	(0.0072)	(0.0073)		(0.0063)
Mismatch					−0.0064 **
					(0.0111)
Miscap						−0.0054 **
						(0.0097)
Mislab								−0.0096 **
								(0.0279)
innovate	−0.0152	−0.0448 ***		0.0076	0.0033	0.0029		0.0031
	(0.0165)	(0.0151)		(0.0133)	(0.0037)	(0.0038)		(0.0039)
develop	−0.0425	0.1067		−0.1036 **	0.0128	0.0131 *		0.0135
	(0.0423)	(0.1047)		(0.0492)	(0.0078)	(0.0076)		(0.0090)
finance	−0.0140	0.1073 **		0.0079	0.0113 *	0.0118 **		0.0112 **
	(0.0275)	(0.0499)		(0.0249)	(0.0056)	(0.0056)		(0.0054)
material	0.0311	0.0858		−0.0648 *	−0.0085 **	−0.0078 *		−0.0077 *
	(0.0264)	(0.0796)		(0.0340)	(0.0040)	(0.0043)		(0.0039)
urban	−0.0175	−0.1037		0.1717	0.0370 *	0.0363 *		0.0352 *
	(0.0828)	(0.1702)		(0.1186)	(0.0197)	(0.0198)		(0.0199)
Time-fixed	Yes	Yes		Yes	Yes	Yes		Yes
Province-fixed	Yes	Yes		Yes	Yes	Yes		Yes
Constant	−0.0672	0.3867 ***		0.3904 ***	0.3048 ***	0.3065 ***		0.3007 ***
	(0.0443)	(0.0693)		(0.0621)	(0.0145)	(0.0137)		(0.0152)
Observations	270	270		270	270	270		270
R-squared	0.2870	0.2615		0.4014	0.3189	0.3178		0.3183
Number of id	30	30		30	30	30		30

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

.

Column (1) in Table 8 shows the estimated result of the resource mismatch index (Mismatch) as the explained variable. The result of the first-order term coefficient of the digital economy is −0.1304, and it is significant at a 5% statistical level. The quadratic term coefficient of the digital economy is 0.0889, and it is significant at a 5% statistical level. It shows that the digital economy has a U influence on the resource mismatch index. The economic significance shows that the digital economy can improve the allocation of resources and promote low-carbon and inclusive growth at the initial stage of development. After exceeding a certain threshold, the digital economy will aggravate the distortion of resources and cause a mismatch of resources, thus suppressing low-carbon, inclusive growth. Column (4) is the result of the regression of low-carbon, inclusive growth with the resource mismatch index as a mechanism variable and the digital economy. The estimated results show that the resource mismatch coefficient (−0.0064) is significantly negative at a 1% statistical level. That is, there is a negative correlation between the resource distortion index and low-carbon, inclusive growth; in other words, the improvement of resource allocation efficiency will help to improve the level of regional low-carbon, inclusive growth. To sum up, the digital economy can promote low-carbon, inclusive growth by optimizing the allocation of resources. When the development level of the digital economy is too high, it may lead to resource mismatch and inhibit low-carbon, inclusive growth. This is consistent with the results of previous studies [80,81].

Columns (2) and (3) show the regression results of the capital mismatch index and labor mismatch index as explained variables. The results show that the digital economy has a significant U-shaped influence on the capital mismatch index. The regression result of the labor mismatch index shows that the digital economy (digiecon) is significantly negative. The quadratic term of the digital economy (digiecon2) is not significant. The mismatch index of capital and labor in columns (5) and column (6) is significantly negative. It shows that the digital economy promotes low-carbon and inclusive growth by improving the allocation efficiency of capital and labor. However, exceeding a certain level, the digital economy suppresses low-carbon, inclusive growth mainly through capital misallocation. The distortions to labor resources are not significant. To sum up, resource allocation plays a Mechanism function in the inverted U-shaped influence of the digital economy on low-carbon, inclusive growth. Compared with labor factors, the digital economy suppresses low-carbon, inclusive growth mainly by causing a mismatch of capital factors.

6.2. Dimension Reduction Analysis

The core variables of this paper are all comprehensive indicators. Low-carbon, inclusive growth uses 54 measurable indicators. By reducing the dimension of the data to one dimension (that is, the comprehensive indicator), this allows us to accurately draw and visualize it. However, in this paper, low-carbon, inclusive growth includes three core indicators: low-carbon ecology, social inclusiveness, and economic growth. The contents represented by these three indicators have their emphasis. Therefore, it is necessary for us to reduce 54 dimensions to these 3 dimensions for analysis. It will help to clarify the internal relationship between the digital economy and low-carbon, inclusive growth. Similarly, this paper also analyzes the dimensionality reduction of the digital economy. It has important practical significance to guide the development of the digital economy and support low-carbon, inclusive growth.

6.2.1. Dimension Reduction Analysis of Independent Variables

Learning from the work of Xie Tingting and Gao Lili [82], we will analyze the impact path of the digital economy on low-carbon, inclusive growth through upgrade indicators. The comprehensive index of the digital economy in this paper is composed of four secondary indicators. The secondary indicators are the digital foundation (digibasic), digital resources (digisource), digital applications (digiapp), and digital finance (digifinanc). The estimated results are shown in

Table 9. Independent variable upgrade dimension test.

	(1)	(2)	(3)	(4)
Variables	Digital Foundation	Digital Resources	Digital Applications	Digital Finance
digibasic	−0.9031
	(0.8891)
digibasic2	0.4152
	(0.1766)
digisource		−0.9005 **
		(0.7547)
digisource 2		0.9108 *
		(0.4481)
digiapp			0.2677 ***
			(0.7761)
digiapp2			−0.4603 ***
			(0.2996)
digifinanc				0.0271 **
				(0.0102)
digifinanc2				−0.0001 **
				(0.0001)
innovate	0.0451	0.0148	0.0278	0.0838
	(0.0915)	(0.0715)	(0.0921)	(0.0880)
develop	−0.3625 *	−0.4526 *	−0.3463	0.2607
	(0.1955)	(0.2338)	(0.2191)	(0.3804)
finance	0.0490	0.0329	−0.0042	−0.0097
	(0.1983)	(0.2335)	(0.2168)	(0.1776)
material	0.1090	0.0595	0.1328	0.1831
	(0.2306)	(0.2384)	(0.2202)	(0.2107)
urban	−0.6004	−0.2938	−0.7387	−0.8938
	(0.5864)	(0.6417)	(0.5543)	(0.5364)
Time-fixed	Yes	Yes	Yes	Yes
Province-fixed	Yes	Yes	Yes	Yes
Constant	1.5456 ***	2.3270 ***	1.2133 ***	0.5973
	(0.4349)	(0.3970)	(0.3080)	(0.5663)
Observations	270	270	270	270
R-squared	0.0492	0.0631	0.0952	0.0619
Number of id	30	30	30	30

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

. The coefficient of digital applications is 0.2677, and it is significant at the 1% statistical level. The quadratic term coefficient (−0.4603) is applied to the figure, and it is significant at a 1% statistical level. The coefficient of digital finance is 0.0271, and it is significant at a 5% statistical level. The quadratic coefficient of digital finance is −0.00001, and it is significant at a 5% statistical level. It shows that both digital application and digital finance have a significant inverted U-shaped impact on low-carbon, inclusive growth. Economic significance shows that the digital economy mainly affects low-carbon, inclusive growth through two core elements: digital application and digital finance. Because digital application is the most extensive aspect of digital technology, digital finance is also a part of digital application. When the development of the digital economy exceeds a certain extent, some platform organizations apply digital technology, big data, and other advantages to form monopolies. This may lead to a mismatch of resources, which is not conducive to the development of low-carbon transformation. The coefficient of digital resources (−0.9005) is significant at a 5% statistical level. The quadratic coefficient of digital resources is 0.9108, and it is significant at a 10% statistical level. It shows that digital resources have a U-shaped impact on low-carbon, inclusive growth. Economic significance shows that the abundance of digital resources in the early stage of development is not conducive to low-carbon, inclusive growth. However, when digital resources exceed a certain level, it is conducive to low-carbon, inclusive growth. The reason may be due to what this paper describes as an abundance of digital resources based on the popularity of the Internet, telephones, and fixed-line phones. Thus, in the early days, with the increase in the popularity of landlines, telephones, and the Internet, the demand for plastic materials and electricity may have increased. In turn, this leads to an increase in energy consumption in power and other related industries, thus aggravating carbon emissions. With the gradual popularity of Internet applications, the level of pollution control and the energy efficiency of industries with high energy consumption has improved. This is conducive to the development of low-carbon transformation. The digital base has an inverted U-shaped effect on carbon emission performance, but it is not statistically significant.

6.2.2. Dimension Reduction Analysis of Dependent Variables

To explore the impact path in-depth, we also upgrade the low-carbon, inclusive growth indicators. There are mainly three secondary indicators of low-carbon, inclusive growth in this paper, including economic growth (Growth), social inclusion (Inclusive), and low-carbon ecology (Lcarecology). We use the entropy method to measure them and use the digital economy to regress these three variables. The estimated results are shown in

Table 10. Dependent variable upgrade dimension test.

	(1)	(2)	(3)
Variables	Growth	Inclusive	Lcarecology
digiecon	0.0118 *	0.0291 *	0.0359 **
	(0.0245)	(0.0177)	(0.0148)
digiecon2	0.0073	−0.0160 **	−0.0384 ***
	(0.0227)	(0.0119)	(0.0113)
innovate	0.0008	0.0024	0.0037
	(0.0045)	(0.0048)	(0.0052)
develop	0.0002	0.0346 ***	0.0018
	(0.0127)	(0.0099)	(0.0075)
finance	−0.0040	0.0064	0.0228 ***
	(0.0099)	(0.0080)	(0.0082)
material	0.0029	−0.0148 **	−0.0053
	(0.0089)	(0.0070)	(0.0090)
urban	0.0297	0.0738 ***	−0.0031
	(0.0239)	(0.0232)	(0.0330)
Constant	0.2401 ***	0.3442 ***	0.2995 ***
	(0.0159)	(0.0173)	(0.0125)
Time-fixed	Yes	Yes	Yes
Province-fixed	Yes	Yes	Yes
Observations	270	270	270
R-squared	0.4335	0.3923	0.1876
Number of id	30	30	30

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

. The results of the return to social inclusiveness show the one-time term coefficient of the digital economy is 0.0291, which is significant at a 10% statistical level. The quadratic term coefficient of digital economy is −0.0160, and it is significant at a 5% statistical level. The regression results of low-carbon ecology show that the first term coefficient of the digital economy is 0.0359, and it is significant at a 5% statistical level. The quadratic coefficient of the digital economy (−0.0384) is significant at a 1% statistical level. The regression results of economic growth show that the core explanatory variables are not statistically significant. This shows that the inverted U-shaped impact of the digital economy on social inclusiveness and low-carbon ecology is more significant. The economic significance suggests that the initial stage of the development of the digital economy is conducive to social inclusiveness and low-carbon ecological development. The fact is that there are regional differences in the development level of the digital economy. At the initial stage of the development of the digital economy, it helps to narrow the gap between urban and rural areas, but there may be a digital economic gap effect in less developed areas, widening the gap between urban and rural areas and reducing social inclusiveness [79]. Furthermore, in the early stages of development, the digital upgrading of the energy industry helps to reduce carbon emissions, but as the level of digitization increases, energy consumption increases. From the perspective of the whole society, it may aggravate the total carbon emissions, which is not conducive to low-carbon, inclusive growth.

7. Conclusions and Suggestions

7.1. Conclusions

The main conclusions of this paper are as follows: (1) Digital economy has an inverted U-shaped impact on regional low-carbon, inclusive growth. We found that the digital economy is conducive to low-carbon, inclusive growth, but the too high level of the digital economy will also inhibit low-carbon, inclusive growth. According to statistical analysis, it was found that 75.9% of the observed areas in China are in the early stages of development. Therefore, the development of the digital economy is generally conducive to low-carbon, inclusive growth. Of the observed areas, 24.1% are on the right side of the inverted U-shaped, indicating that the high level of digital economic development in some areas is not conducive to low-carbon, inclusive growth. (2) In areas with low capital and energy productivity, the inverted U-shaped impact of digital economy on low-carbon and inclusive growth is more significant. On the left side of the threshold of digital economy, the promoting effect of the digital economy on low-carbon, inclusive growth in areas with low capital and energy productivity is more obvious. However, on the right side of the threshold of digital economy, the digital economy will increase total energy consumption and lead to capital misallocation to curb low-carbon, inclusive growth. In the case of high factor productivity, the development of the digital economy mainly leads to the distortion of capital and labor factors, which suppresses low-carbon, inclusive growth. In the case of a high level of information technology, the inverted U-shaped impact of the digital economy on low-carbon, inclusive growth is more obvious. In areas with a high degree of informatization, the positive effect of the digital economy is greater. Compared toa inland areas, the digital economy has a greater promoting effect on coastal areas. (3) Resource allocation plays an important role in the transmission mechanism between the digital economy and low-carbon, inclusive growth. Specifically, the digital economy can improve the level of low-carbon, inclusive growth by improving the allocation efficiency of capital and labor. However, if the development level of the digital economy is too high, it will mainly cause capital mismatch and then restrain low-carbon, inclusive growth.

7.2. Suggestions

(1)

According to the empirical results, first of all, we should vigorously develop the digital economy and promote regional low-carbon and inclusive growth. At the same time, government departments should constantly improve the market supervision system for the healthy development of the digital economy, such as the establishment of a market supervision committee and other institutions to effectively prevent monopoly and misallocation of resources brought about by digital technology. To provide a fair and fair competitive environment for the healthy development of the digital economy

(2)

We should make full use of the digital economy to improve the level of regional factor productivity. We should vigorously improve the information infrastructure and improve the degree of regional informatization, improve the openness of inland areas, actively carry out exchanges and cooperation with coastal cities, and make full use of the digital economy to promote low-carbon and inclusive growth in inland areas. At the same time, we should actively improve the system and mechanism for the development of the digital economy and improve the laws and regulations of the digital economy. Government departments should actively deal with negative effects caused by the digital economy.

(3)

Capital and talents are indispensable factors of production for economic growth, so the digital economy should be used to improve the efficiency of resource allocation of production factors. After the emergence of the digital economy, data have become an important factor in production. We should make full use of digital resources and digital finance to promote low-carbon transformation and development. We should vigorously promote the construction of data platforms, improve the market for data elements, achieve information disclosure, break the isolated island of information, and promote the opening and sharing of data resources. We should attach importance to the development of rural digital economy and improve the coordination of the development of the urban and rural digital economy, making use of the digital economy to increase rural income. For example, the development of rural ecommerce and webcasts selling goods to increase farmer incomes. We should also narrow the development gap between urban and rural areas and improve the inclusiveness of economic and social development.

There are still some shortcomings in this paper: (1) Digital economy measurement and data: Provincial data contain more comprehensive index information, but they still cannot well describe the development level of the digital economy. Because the measurement of the digital economy is constantly improving, the conclusion of this paper has certain time and space limitations. (2) The solution to endogenous problems: Although this paper tries to solve possible endogenous problems by using tool variables, generalized method of moments estimation (GMM), solving omitted variables, and so on, there are still some shortcomings, such as the digital circuit only being part of the digital economic foundation, which may cause estimation deviation. The authors’ further research will be carried out using the following three aspects. (1) Verification based on more microscopic data, such as cities or enterprises, improving the reliability and universality of the estimation conclusion. (2) The relationship between regions is getting closer and closer, so there is a spatial spillover effect in regional development. The spatial impact of the digital economy on low-carbon, inclusive growth is worthy of further discussion. (3) On the question of endogenesis, in addition to trying to find more appropriate tool variables, in the next step, the authors will consider using urban digitization policies, such as “Broadband China”, as quasi-natural experiments to test exogenous shocks to improve the accuracy and reliability of the estimated results in this paper.