Can the Growth of the Digital Economy Be Beneficial for Urban Decarbonization? A Study from Chinese Cities

Abstract

An environmentally friendly city is a livable home for the future. Can the rapidly developing digital economy help decrease carbon emissions and realize a low-carbon and clean city promptly? This study focuses on examining how multi-dimensional digital economic growth has influenced CO₂ emissions across 280 Chinese cities from 2011 to 2019. Findings discover that (1) An “n”-type curve nexus exists between CO₂ emissions and the digital economy in Chinese cities, which means that digital economy expansion initially strengthens CO₂ emissions, but at a certain level, it can help achieve the target of urban decarbonization; (2) The digital economy’s influence on CO₂ emissions is spatially spilled and regionally heterogeneous, and by means of economies of scale and industrial composition upgrades, it can help the city to lower carbon emissions and benefit the low carbonization of neighboring cities. However, based on the “rebound effect”, the intermediary role of technological effects in reducing emissions in the short term is not apparent; (3) The expansion of trade openness and appropriately stringent environmental rules in line with national conditions are beneficial to lower CO₂ emissions in the city and the surrounding cities in the short term. It is recommended that policy makers actively promote the development of the digital economy, strengthen exchanges and cooperation between cities, narrow the gap between cities, and actively learn the advanced management concepts of surrounding cities through the development of economies of scale and industrial structure transformation to accomplish the target of “carbon neutrality” sooner rather than later.

1. Introduction

Due to the continuous refinement of the international division of labor, developing countries are embedded in the production process of the global value chain with low value-added and low-tech processing methods, and these highly polluting processing and production activities have caused many environmental issues [1,2]. The CDIAC database illustrates that in the past 20 years, China’s carbon emission growth rate has exceeded 200%, which makes China’s carbon emissions rank first in the world (CDIAC). In recent years, environmental problems such as flooding in urban subways, floods in Henan Province, and prolonged drought in Southwest China have attracted the attention of the government. The Chinese government actively responded to the slogan of “not invaluable assets, but lucid waters and lush mountains”, and proposed the “dual carbon” target.

Since China proposed the notion of the digital economy (hereinafter referred to as DE) at the G20 summit, the development of China’s DE has subtly changed traditional trade methods. Coupled with the fact that COVID-19 has stimulated the high-speed growth of the DE, lots of transactions are biased towards e-commerce, and digital transaction methods such as WeChat and Alipay have greatly simplified complicated transaction processes, thereby helping to reduce carbon emissions [3,4]. Under the premise of sustainable development, financial inclusion has gradually evolved into a positive response strategy to improve constraints brought about by environmental issues, and promoting digital financial inclusion is conducive to reducing urban CO₂ emissions [5]. The Chinese economy’s growth and the industrial structure of the DE can bring about a reduction in China’s carbon emissions. Among them, the DE can accelerate the upgrading of industrial composition, thereby reducing CO₂ emissions. Instead, it has promoted economic growth, resulting in an increase in carbon emissions [4,5]. Li et al. revealed an “n” -type nexus between a country’s digital economic growth and carbon emissions. CO₂ emissions are increasing in the initial stage of digitization, and when the level of digitization improves to a higher stage, CO₂ emissions start dropping after attaining a maximum value. It is also evident that differences in income levels affect the digital economy’s impact on carbon emissions. This curve trend is more pronounced in high-income regions than in low-income regions [6].

China, which currently ranks first in the world in terms of carbon emissions, is actively taking effective measures to curtail environmental degradation. Actively achieving the carbon neutrality goal is also a challenge for China. Meanwhile, China’s carbon emissions control has affected other countries to a certain extent. Therefore, exploring China’s policies is also a reference for other countries [4]. A focus on urban research can be summed up as the realization that low-carbon urban development is the leading direction of urban development in the future. Some scholars have verified that the digital economy’s growth can contribute to reducing urban carbon emissions and protect the environment. Due to the “mobility of spatial capital factors” and the “emulation effect” between cities, the urban digital economy’s growth will also influence or be affected by adjacent areas. Based on this, the digital economy’s spatial effect on urban CO₂ emissions is worth exploring [4,5].

A review of previous studies found that there is currently no unified standard for digital economic accounting indicators, and current research mainly involves the linear relationship between the digital economy and carbon emissions. In fact, from the residual fitting graph of China’s urban digital economy and carbon emissions, it can be found that there is a nonlinear relationship between the two. In addition, few scholars have examined the spatial spillover effects of the digital economy on carbon emissions and regional heterogeneity from the perspective of spatial econometrics. This study made some supplements on the basis of previous studies. By collecting data from 293 prefecture-level cities, removing cities with missing values during data processing, and finally obtaining a sample size of 280 (as shown in

Table A2. A Study Sample of Chinese Cities.

280 City Names

Ankang, Anqing, Anshan, Anshun, Anyang, Baicheng, Baise, Baishan, Baiyin, Baoding, Baoji, Baoshan, Baotou, Bayannaoer, Bazhong, Beihai, Beijing, Bengbu, Benxi, Binzhou, Cangzhou, Changchun, Changde, Changsha, Changzhi, Changzhou, Chaoyang, Chaozhou, Chengde, Chengdu, Chenzhou, Chifeng, Chizhou, Chongqing, Chongzuo, Chuzhou, Shantou, Yantai, Dalian, Dandong, Daqing, Datong, Dazhou, Deyang, Dezhou, Dingxi, Dongguan, Dongying, Ezhou, Fangchenggang, Foshan, Fushun, Fuxin, Fuyang, Fuzhou, Fuzhou, Ganzhou, Guang’an, Guangyuan, Guangzhou, Guigang, Guilin, Guiyang, Guyuan, Haikou, Handan, Hangzhou, Hanzhong, Harbin, Hebi, Hechi, Hefei, Hegang, Heihe, Hengshui, Hengyang, Heyuan, Heze, Hezhou, Hohhot, Huaian, Huaibei, Huaihua, Huainan, Huanggang, Huangshan, Huizhou, Huludao, Huzhou, Jiamusi, Ji’an, Jiangmen, Jiaozuo, Jiaxing, Jiayuguan, Jieyang, Jilin, Jinan City, Jinchang City, Jincheng, Jingdezhen, Jingmen, Jingzhou, Jinhua City, Jining City, Jinzhong, Jinzhou, Jiujiang, Jiuquan, Jixi, Kaifeng, Karamay, Kunming, Laibin City, Laiwu, Langfang City, Lan’Zhou City, Leshan city, Lianyungang City, Liaocheng, Liaoyang, Liaoyuan City, Lijiang City, Lincang, Linfen City, Linyi City, Liu’an, Liupanshui City, Liuzhou, Longnan, Longyan, Loudi, Luliang, Luohe, Luoyang City, Luzhou, Ma’anshan, Maoming City, Meishan, Meizhou, Mianyang, Mudanjiang, Nanchang, Nanchong, Nanjing, NanNing City, Nanping, Nantong city, Nanyang, Neijiang, Ningbo, Ningde, Ordos, Panjin, Panzhihua City, Pingdingshan, Pingliang, Pingxiang, Putian City, Puyang City, Qingdao, Qingyang, Qingyuan, Qinhuangdao, Qinzhou, Qiqihar City, Qitaihe, Quanzhou, Qujing, Quzhou, Rizhao, Sanmenxia, Sanming, Sanya, Shanghai, Shangluo, Shangqiu, Shangrao, Shanwei, Shaoguan City, Shaoxing, Shaoyang, Shenyang city, Shenzhen, Shijiazhuang City, Shiyan City, Shizuishan, Shuangyashan, Shuozhou, Siping, Songyuan City, Suining, Suizhou, Suqian, Suzhou City, Suzhou City, Tai’an, Taiyuan, Taizhou, Taizhou, Tangshan, Tianjin, Tianshui, Tieling, Tongchuan, Tonghua, Tongliao, Tongling, Urumqi, Weifang, Weihai, Weinan, Wenzhou city, Wuhai, Wuhan, Wuhu, Wulanchabu, Wuwei, Wuxi, Wuzhong City, Wuzhou, Xiamen City, Xi’an, Xiangtan City, Xianning, Xianyang, Xiaogan City, Xingtai city, Xining, Xinxiang City, Xinyang, Xinyu City, Xinzhou, Xuancheng, Xuchang, Xuzhou, Ya’an city, Yan’an City, Yancheng, Yangjiang, Yangquan, Yangzhou, Huangshi, Lishui, Yibin City, Yichang City, Yichun (Heilongjiang province), Yichun(Jiangxi province), Yinchuan, Yingkou, Yingtan City, Yiyang City, Yongzhou, Yueyang City, Yulin City, Yulin City, Yuncheng, Yunfu City, Yuxi, Zaozhuang City, Zhangjiajie, Zhangjiakou, Zhangjiang City, Zhangye City, Zhangzhou, Zhaoqing, Zhaotong City, Zhengzhou City, Zhenjiang, Zhongshan City, Zhongwei, Zhoukou, Zhoushan, Zhuhai city, Zhumadian City, Zhuzhou, Zibo, Zigong, Ziyang City, Zunyi City

Notes: This paper collects data based on 293 prefecture-level cities, removes the cities with missing values during data processing, and finally obtains a sample size of 280.

of Appendix A), this study examines how multi-dimensional digital economic growth has influenced CO₂ emissions across 280 Chinese cities from 2011 to 2019. First, we measured the comprehensive development level of the digital economy from two aspects: Internet development and digital financial inclusion. Through principal component analysis, the data of multiple indicators are standardized and then dimensionally reduced to obtain the comprehensive development index of the digital economy. Next, we assessed the causal nexus between urban carbon emissions and the digital economy by using regression, constructing various spatial weight matrixes to examine spatial spillover effects, and verified whether there were differences between regions. The findings reveal that the city’s digital economy initially boosts carbon emissions and then reduces emissions after reaching a certain level. Furthermore, there are spatial spillover effects and regional heterogeneity about how digital economic growth has influenced CO₂ emissions; approaches such as economies of scale and industrial composition upgrades can help the city lower carbon emissions and benefit the low carbonization of neighboring cities. However, based on the “rebound effect”, the intermediary role of technology effects in reducing emissions in the short term is not evident. Last, expanding trade openness and developing appropriately stringent environmental rules can be beneficial in the short term to lower carbon emissions in one’s own city and neighboring cities.

The contributions can be summarized as follows: (1) Multi-dimensional digital economy indicators were applied in this study to assess the development of the digital economy in Chinese cities; (2) Using Stata software to construct weight matrices of different scales by using city latitude and longitude, this study analyzes how the digital economy affects carbon emissions spatially by means of three approaches; (3) Selecting the dual fixed SDM, this study analyzes the spatial dependence between cities and the differences between regions, so as to offer a theoretical foundation for decision makers to realize ecologically clean cities at an early date.

The organizational structure of this paper is as follows: Section 2 is the literature review, and Section 3 and Section 4 are the model, data description, and empirical results, respectively. Section 5 is the conclusion and discussion.

2. Literature Review

Green and low carbon is a better way for urban transformation and development. Regarding urban carbon emissions reduction, researchers primarily analyze them from two perspectives: measurement approaches as well as impact factors. This paper introduces the variable of the digital economy, and primarily reviews the literature from two aspects: the influencing factors of CO₂ emissions and the digital economy’s environmental effects.

2.1. CO₂ Emissions Influencing Factors

The essence of urban decarbonization is to lower CO₂ emissions and improve the neutralization capacity of urban carbon sequestration. Scholars primarily examine environmental effects from factors such as foreign direct investment, industrial structure, economic growth, population, income, environmental rules, degree of marketization, energy structure, and energy price changes [3,4,6,7,8,9,10,11,12,13,14,15]. The growth of foreign direct investment can lower carbon emissions, and outward direct investment can also contribute to reducing domestic carbon emissions, but export trade will increase domestic carbon emissions to a certain extent. The relatively less lax environmental rules in advanced economies make it possible to move energy-intensive and polluting production activities to developing economies at a lower expense. Thus, increased FDI can create “pollution sanctuaries” that can put stress on a country’s environment [4]. Levinson points out that branch plants of large companies appear to be more sensitive than all plants to environmental regulations, which are binding on manufacturers. Industrial structure, GDP, total population, and built-up areas are significantly correlated with carbon dioxide emissions and emissions efficiency [7,9,16]. Changes in energy prices in key provinces have a direct influence on CO₂ emissions, and changes in energy prices in neighboring provinces have an indirect impact [13,14]. There is an inverted U-shaped relationship between traffic density and urban smog pollution in large and medium cities, while there is no significant relationship in small cities [17]. Income, energy consumption, industrialization, urbanization, foreign direct investment, and financial inclusion increase regional CO₂ emissions, while increasing trade openness appears to reduce CO₂ emissions [1,3,18,19].

Scholars’ research on China’s carbon emissions is mainly concentrated on different industries such as industry, production, life, and transportation [4,9,10,14,16,17]. The main source of emissions in most Chinese cities is industrial energy consumption, with indirect carbon dioxide emissions predominating. Moreover, the total carbon emissions and per capita emissions in central urban areas are higher than those in the surrounding areas of most prefecture-level cities. Furthermore, the extent to which spatial dependencies and technological spillovers reduce pollution is difficult to counteract by directly causing carbon emissions. Improving total factor energy efficiency and carbon emission efficiency will help narrow the gap between various sub-industries of light industry [10,14,16,17,20,21]. Most of China’s environmental innovation measures, such as technological innovation and upgrading of environmental regulations, have effectively reduced carbon emissions [3,12]. Energy consumption and foreign direct investment inflows and outflows have boosted economic growth in Asian countries while increasing their own carbon emissions [11,15].

2.2. Digital Economy and CO₂

COVID-19 has exacerbated the digital economy’s growth, making it an engine of economic growth. During the age of the digital economy, using digital finance to shape energy conservation and emission reductions is a long-term mechanism. Governments of various countries should take early measures to lower carbon emissions from the digital economy, promoting the digital economy’s growth in order to accomplish the world’s sustainable development targets [6,22]. Digitalization and greening have increasingly become the main directions of China’s economic development, and only a benign interaction between the two can promote China’s economy to accomplish higher-quality as well as more sustainable development. The digital economy has upgraded the way consumers consume, giving importance to achieving a “dual carbon” target [15,23]. Digital financial inclusion in financially developed and industrialized cities has obviously affected urban decarbonization. Internet penetration did not significantly affect individuals’ sustainable consumption behavior when factors at the individual and national levels were constant, but significantly promoted the shift in environmental attitudes towards sustainable behaviors [4,24].

Li et al. demonstrated that the digital economy’s impact on carbon emissions presents an inverted “U”-shaped curve. During digitalization’s early period, carbon emissions are on the rise, but when the level of digitalization increases to a certain degree, especially when the production level of enterprises is stable and technological advancements can promote the green economy’s growth, carbon emissions begin to show a downward trend after attaining the maximum value. This “n”-shaped nexus of how the digital economy affects CO₂ emissions is also reflected in differential income levels. High-income areas tend to have a more pronounced inverted “U”-shaped nexus than low-income areas [6]. Research denotes that the growth of digital financial inclusion in China is beneficial to lowering carbon emissions, and coupled with the mobility of spatial capital factors, digital financial inclusion’s growth in cities can also contribute to the decarbonization of neighboring cities [4,6]. In addition, the digital economy indirectly reduces carbon emissions by upgrading the industrial structure, while the digital economy indirectly increases carbon emissions by promoting economic growth [4].

To sum up, there is little literature on the environmental effects of the digital economy at the city level. Existing studies basically analyze the direct or indirect effects of the digital economy on carbon emissions at the industrial and technological levels. Based on the creation of a weight matrix, this paper further evaluates the spatial impact of the digital economy on CO₂ emissions through scale, technology, and structural effects. This will help policymakers more comprehensively understand a city’s digital economy’s environmental impact and the spatial effects of neighboring cities on the city, so as to provide policymakers with constructive policy recommendations.

3. Model and Data

3.1. Model

Under the theoretical framework of the Cobb-Douglas production function [25], this study analyzes the impact of the urban digital economy on carbon emissions. Based on the influence factors of carbon emissions mentioned in a large number of literature [3,4,6,7,8,9,10,11,13,14,15], this study selects urban population, urban per capita GDP, trade openness, and environmental rules as control variables, and considering that there is a nonlinear curve in the impact of the digital economy on carbon emissions, we square the $lndigital$ to create the following econometric Equation (1):

$$lnC{O}_2{}_{it}={\phi }_0+{\phi }_{1}lndigita{l}_{it}+{\phi }_{1}lndigita{l}^2{}_{it}+{\phi }_2{V}_{it}+{\beta }_i+{\gamma }_t+{\epsilon }_{it},$$

(1)

where, $lnC{O}_2{}_{it}$ are the carbon emissions of Chinese cities, $lndigita{l}_{it}$ represents two digital economy measurement indicators, namely internet development and digital financial inclusion. By drawing the residual fitting value graph, it is found that there is a nonlinear relationship between the digital economy and carbon emissions of Chinese cities, so the $lndigita{l}_{it}$ variable is squared. As a control variable, $ln{V}_{it}$ primarily includes urban population, urban GDP per capita, trade openness, and environmental rules. Moreover, ${\phi }_0$ is a constant term, ${\beta }_i and {\gamma }_t$ stand for the city and year which are fixed, respectively.

The test of spatial autocorrelation Moran index indicates that spatial autocorrelation exists in carbon emissions among 280 cities, which indicates that CO₂ emissions in this area are not only influenced by independent variables in the region, but also may be affected by independent variables in neighboring regions [26,27]. Therefore, to further assess how the urban digital economy spatially affects CO₂ emissions, the Spatial Durbin Model (SDM) is selected [28,29] to establish Equation (2) and set the mediated variable in it to evaluate whether the digital economy of Chinese cities has direct and indirect effects on carbon emissions in space.

$$\begin{array}{r}lnC{O}_2{}_{it}=\rho {\displaystyle \sum _{j=1}^{280}}{W}_{ij}lnC{O}_2{}_{jt}+{\delta }_0+{\delta }_{1}lndigita{l}_{it}+{\delta }_{2}ln{V}_{it}\hfill \\ +{\delta }_3\left(lndigita{l}_{it}\ast ln{G}_{it}\right)+{\delta }_4\left(lndigita{l}_{it}\ast ln{T}_{it}\right)\hfill \\ +{\delta }_5\left(lndigita{l}_{it}\ast ln{S}_{it}\right)+{\sigma }_1{\displaystyle \sum _{j=1}^{280}}{W}_{ij}lndigita{l}_{jt}\hfill \\ +{\sigma }_2{\displaystyle \sum _{j=1}^{280}}{W}_{ij}ln{V}_{jt}+{\sigma }_3{\displaystyle \sum _{j=1}^{280}}{W}_{ij}\left(lndigita{l}_{jt}\ast ln{G}_{jt}\right)\hfill \\ +{\sigma }_4{\displaystyle \sum _{j=1}^{280}}{W}_{ij}\left(lndigita{l}_{jt}\ast ln{T}_{jt}\right)\hfill \\ +{\sigma }_5{\displaystyle \sum _{j=1}^{280}}{W}_{ij}\left(lndigita{l}_{jt}\ast ln{S}_{jt}\right)+{\epsilon }_{it}\hfill \end{array}$$

(2)

Here, $lnC{O}_2{}_{it}$ are the carbon emissions of Chinese cities, $lndigita{l}_{it}$ represents two digital economy measurement indicators, namely internet development and digital financial inclusion. The terms ${\sum }_{j=1}^{280}{W}_{ij}lnC{O}_2{}_{jt}$ and ${\sum }_{j=1}^{280}{W}_{ij}lndigita{l}_{jt}$, respectively, represent the influence of neighboring cities’ carbon emissions on their own city and the influence of neighboring cities’ digital economic growth on their own city’s digital economic growth. As a control variable, $ln{V}_{it}$ primarily includes urban population, urban GDP per capita, trade openness, and environmental rules. The terms $ln{G}_{it} (Chinese$ city GDP), $ln{S}_{it}$ (ratio of the secondary industry to the tertiary industry), and $ln{T}_{it}$ (number of green patents) denote scale, structure, and technology effects, respectively. Moreover, both ${\phi }_0 and {\delta }_0$ are constant term, ${\beta }_i and {\gamma }_t$ stand for the city and year are fixed, respectively.

3.2. Data

3.2.1. Independent Variables

This study refers to the research conducted by Liu et al. to create a multi-dimensional indicator [2] to assess the comprehensive development level of the digital economy from two levels of digital financial inclusion and Internet development. Meanwhile, the digital financial inclusion data adopt the China Digital Financial Inclusion Index [30] jointly compiled by the Digital Finance Research Center of Peking University and Ant Financial, and the Internet development data use the Internet and mobile phone penetration rates of Huang et al. to measure the development of the Internet at the city level, as well as the indicators of different dimensions such as digital industry development and innovation ability calculated by other scholars [31,32]. Then, we adopt the principal component analysis approach to standardize the indicators of multiple dimensions, and then perform dimensionality reduction to obtain the final comprehensive digital economy indicator. The specific classification accounting is exhibited in

Table 1. Calculation method of the digital economic indicator.

Digital Economy Metrics	Specific Indicators	Indicator Definition
Digital infrastructure	Broadband Internet Basics	Amount of households with mobile Phone users (per 10,000)
	Mobile Internet Basics	Amount of households with broadband Internet (per 10,000)
Digital industry development	E-commerce industry development	Amount of urban e-commerce parks (pieces)
	Information Industry Foundation	Amount of computer and information Internet practitioners (10,000 people)
	Telecom industry yield	Total telecom commerce (10,000 yuan)
Digital innovation capability	Support for innovation element	R&D expenditure (10,000 yuan)
	Innovation yield level	Amount of digital economy patents per 10,000 people (pieces)
	High-tech penetration	Listed companies’ digital high-tech penetration applications
Digital financial inclusion	Coverage width	Coverage Width Index
	use depth	depth Index
	Digitization degree	Digitization Degree Index

Note: The calculation method of digital economic indicators here refers to the research method of Xu et al., (2022) [32]. Specifically quoted in Table 1 “Evaluation indicators for measuring the level of development of the digital economy” in the paper on page 114 of the Journal of Geographical Research, Volume 41, Issue 1.

.

Through the CSMAR database, we collected the core keywords about the digital economy’s growth, clearly describing the digital economy’s development tendency in the past ten years through text mining. From e-commerce and fintech to digital finance, big data cloud computing, blockchain, etc., we can know from Figure 1 that China’s digital economy has achieved leapfrog development.

The development status of the comprehensive indicators of the digital economy, as well as financial inclusion in 2011 and 2019, are exhibited in Figure 2. We can know that both the digital economy and financial inclusion have developed rapidly in recent years, from the perspective of geographical distribution, cities with high digital economy competitiveness are still primarily gathered in the economically developed eastern areas as well as southern regions. Beijing, Shanghai, and Shenzhen are comprehensive leading cities. From the perspective of city scale, medium-sized and large cities are important focal points for the transformation and upgrading of industries driven by the digital economy. Except for the eastern coastal cities and first-tier cities, the digital economy’s growth and digital inclusive finance in provincial capitals, special economic zones, and demonstration cities are noticeably faster than that in surrounding areas, indicating that government policies have played a promoting role to a certain extent.

3.2.2. Dependent Variable

Carbon emissions data of 280 cities are derived from the CEADs database of Chinese urban carbon emissions indicators created by Shan et al. [33,34,35] to assess the effect of the digital economy on carbon emissions. The distribution of carbon emissions of the 280 cities in Figure 3 implies that the carbon emissions of most cities are increasing year by year, but the growth rate is not notable, and the emissions of individual cities have tended to decline in recent years; however, the decrease is not conspicuous. In addition, the distribution of carbon emissions among cities is uneven, mainly concentrated in the special economic zone of Chongqing, the southeast coast, north China, and old industrial bases in northeast China.

3.2.3. Control Variables

Urban population: Human production and living activities emit a large amount of greenhouse gases, and population growth and population aggregation directly increase energy consumption. These phenomena are all factors that affect the increase in carbon emissions [5,17,36,37].

Per capita GDP: Since the beginning of the industrial revolution era, human beings tend to over-exploit resources and carry out large-scale production activities at the expense of the environment. These phenomena have caused environmental degradation including increased carbon emissions. Many scholars have mentioned that per capita GDP is the benchmark for measuring urban development. While pursuing economic development, human beings promote the increase in atmospheric carbon content by consuming energy [5,38].

Trade openness: The ratio of foreign investment to GDP is used to measure the degree of foreign trade openness. For one thing, “technology spillovers” point out that foreign-investment enterprises with clean technology and excellent management can reduce environmental pollution in their own countries. Furthermore, in order to transfer the environmental pressure of developed countries, industrial activities with high energy consumption and relatively serious pollution are established in developing countries, thus causing the pollution paradise effect [5,39,40,41,42,43].

Environmental rules: Relatively strict environmental rules are often conducive to reducing the entry of polluting enterprises or production activities into the country, thereby reducing carbon emissions. The “following effect” and “competition effect” between cities indicate that managers like to imitate neighboring cities when formulating policies and even develop their economies to surpass their neighbors despite environmental degradation [5,7,13,44].

Green patent ($ln{T}_{it})$: According to the research of some scholars, the number of green patents is an indicator used to measure technology. Through technological innovation, enterprises can not only reduce labor costs and production costs, but also produce clean and energy-saving products, greatly reducing pollution [4,5,8,38]. Xie et al. proposed the “rebound effect”, pointing out that technological progress improves energy efficiency while increasing energy demand, thereby stimulating more production activities and increasing CO₂ emissions [5,45].

Industrial composition ($ln{S}_{it})$: This paper refers to the approach of Wang et al., using the value obtained by dividing the secondary industry by the tertiary industry [4] to evaluate the changes in industrial composition. Due to the fast growth of the ratio of secondary industry, many production activities stimulated the demand for energy, thereby improving the carbon content of the atmosphere. Guided by the policy of pursuing “not invaluable assets, but lucid waters and lush mountains”, the Chinese government and enterprises have continuously adjusted the industrial structure and virtually improved the environment.

3.2.4. Data Sources

Digital economy measurement data adopted in this paper and the control variable data in this study are primarily from the China Urban Statistical Yearbook. The digital financial inclusion data are taken from the CDFI Index measured by the Digital Finance Research Center of Peking University and Ant Financial Group (Digital Economy Open Research Platform. Available online: https://tech.antfin.com/research/data; accessed on 7 August 2022). The digital patents are collected from SIPO (Patent Search and Analysis. Available online: http://pss-system.cnipa.gov.cn/; accessed on 7 August 2022) and the digital high-tech applications in listed companies comes from CDER Database (China Stock Market & Accounting Research Database. Available online: https://cn.gtadata.com/; accessed on 8 August 2022) provided by CSMAR. The carbon emission data of Chinese cities are from the CEADs database (Carbon Emission Accounts & Datasets. Available online: https://www.ceads.net/data/city; accessed on 13 June 2022).

Table A1. Statistical description.

Variable	Meaning	Source	Obs	Mean	Std. Dev.	Min	Max
lnCO2	Carbon emissions	CEADs database	2520	3.136	0.74	1.276	5.424
lnfinancial	Digital finance	PKU_DFIIC (https://tech.antfin.com/research/data, accessed on 7 August 2022)	2520	5.012	0.518	2.834	6.017
lnwidth	Width of digital economy		2520	4.893	0.531	0.631	5.61
lndepth	Depth of digital economy		2520	4.974	0.508	1.456	5.786
lndegree	Degree of digital economy		2520	5.147	0.609	0.993	6.365
lnT	Technology patent	SIPO database	2520	3.768	0.914	1.099	5.075
lndigital	Digital economy	China Urban Statistical Yearbook	2520	8.614	0.939	5.801	12.803
lnpop	City population		2520	5.886	0.71	2.97	9.315
lnG	Urban GDP		2520	16.361	0.975	12.764	19.54
lnpgdp	Urban per capita GDP		2520	10.474	0.809	6.638	13.814
lnS	Industrial structure		2520	–0.169	0.423	−2.175	1.427
lnopenness	Trade openness		2520	−6.289	1.367	−14.848	−1.649
lnregulation	Environmental rules		2520	4.281	0.64	−4.605	4.787

Note: Logarithms were taken for all variables.

is placed in the appendix, which presents the definitions of core variables and control variables, data sources, and descriptions of statistics. There are a total of 2520 observations in this study. Although there are differences in the carbon emissions of various cities, comparing the average, maximum, and minimum values, there is little difference in carbon emissions between cities. The comprehensive indicators of the digital economy are generally higher than the digital financial inclusion index, and the internal gaps in the three different dimensions of the digital economy indicators are also obvious, which reflects the uneven development of the digital economy among cities.

4. Results

4.1. Direct Effect

4.1.1. Regression Test

Through Equation (1), we assess how the digital economy affects urban carbon emissions from comprehensive indicators of Internet development as well as digital financial inclusion. The POLS result indicates that $lndigital$ in

Table 2. The digital economy’s impact on CO₂ emissions.

	Digital Economy Indicator			Financial Indicator
	POLS	FE	RE	POLS	FE	RE
lndigital	0.642	0.156 ***	0.140 ***
	(0.345)	(0.033)	(0.036)
lndigital2	−0.028	−0.008 ***	−0.006 **
	(0.019)	(0.002)	(0.002)
lnpop	0.534 ***	0.006	0.026 ***	0.679 ***	0.005	0.028 ***
	(0.059)	(0.007)	(0.007)	(0.041)	(0.006)	(0.007)
lnpgdp	0.389 ***	−0.000	0.009 *	0.465 ***	−0.002	0.008 *
	(0.042)	(0.004)	(0.004)	(0.033)	(0.004)	(0.004)
lnopenness	0.057 **	0.001	0.003 *	0.060 ***	−0.000	0.002
	(0.018)	(0.001)	(0.001)	(0.018)	(0.001)	(0.001)
lnregulation	−0.133 ***	−0.005 *	−0.006 *	−0.139 ***	−0.007 **	−0.007 **
	(0.034)	(0.002)	(0.003)	(0.035)	(0.002)	(0.003)
lnfinancial				−0.853 ***	0.188 ***	0.162 ***
				(0.200)	(0.031)	(0.034)
lnfinancial2				0.095 ***	−0.017 ***	−0.014 ***
				(0.021)	(0.003)	(0.004)
cons	−6.574 ***	2.379 ***	2.209 ***	−2.884 ***	2.652 ***	2.480 ***
	(1.615)	(0.158)	(0.173)	(0.531)	(0.084)	(0.096)
City fixed		YES			YES
Year fixed		YES			YES
N	2520	2520	2520	2520	2520	2520
R-sq	0.612	0.028		0.604	0.083
F	81.656	10.860		86.190	33.667
p	0.000	0.000	0.000	0.000	0.000	0.000

Note: ***, **, and * of standard errors in parentheses are expressed within 0.01, 0.05, and 0.1, respectively.

is not conspicuous for lnCO₂, and the results of FE and RE are both notably positive at the 1% level. Based on the optimal selection principle of the model, this study selects fixed effects.

The effects of $lndigital$ and the square of $lndigital$ on ${lnCO}_2$ were remarkably positive and obviously negative at the 1% level, respectively, which could be explained as an “n”-shaped correlation between digital economy ($lndigital)$ and CO₂ emissions $\left({lnCO}_2\right)$. The effect of $lnfinancial$ on ${lnCO}_2$ also has a similar curve nexus; internet development as well as digital inclusive finance, both contribute to leading to a rise in CO₂ emissions in cities, but when they develop to a certain level, this environmental effect begins to subside and gradually enters an environmental protection state. Whether it is the development of the Internet or digital financial inclusion, when the development level is low, people tend to prefer the traditional work mode, which cannot significantly reduce carbon emissions. With an improvement in the level of development, when people realize the convenience of the Internet and the high efficiency brought by digital inclusive finance to simplify transactions or work processes, a high-level digital economy shows advantages that are conducive to the development of low-carbon cities. This reminds us that accelerating the digital economy’s growth will help cities decarbonize.

Population ($lnpop$), urban per capita GDP ($lnpgdp$) and trade openness ($lnopenness$) are all unfavorable to urban low-carbon transformation, but the impact coefficient is small, which is similar to the research of scholars [5,17,36,37,38,39,40,41,42]. However, for every 10% increase in $lnregulation$, ${lnCO}_2$ decreases by 0.005, which indicates that the government’s formulation of appropriate and strict import and export regulations will be beneficial to lower CO₂ emissions in the short term [5,7,13].

4.1.2. Robustness

This paper chooses two methods to test the model robustness: (1) Replace the independent variables ($lndigital$ and $lnfinancial$) with the three dimensions of the digital economy (width, depth, and development level); (2) Add new variables including technology ($lnT$) and an industry component ($lnS$) on the basis of the original model.

In the regression test, fixed effects were selected, so

Table 3. Instrumental variable and the adding new variable method.

	Substitution Variable Method						Adding New Variable Method
	Width		Depth		Degree		Digital economy		Financial indicator
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
	POLS	FE	POLS	FE	POLS	FE	POLS	FE	POLS	FE
lnwidth	−0.034	0.071 ***
	(0.172)	(0.019)
lnwidth2	0.006	−0.005 **
	(0.021)	(0.002)
lnpop	0.674 ***	0.001	0.681 ***	0.008	0.674 ***	0.007	0.568 ***	0.010	0.674 ***	0.002
	(0.042)	(0.006)	(0.042)	(0.006)	(0.042)	(0.007)	(0.060)	(0.007)	(0.042)	(0.006)
lnpgdp	0.464 ***	−0.004	0.482 ***	−0.000	0.470 ***	−0.002	0.418 ***	0.004	0.474 ***	−0.006
	(0.034)	(0.004)	(0.033)	(0.004)	(0.033)	(0.004)	(0.045)	(0.004)	(0.033)	(0.004)
lnopenness	0.058 ***	−0.001	0.062 ***	−0.000	0.057 ***	0.001	0.055 ***	0.002	0.059 ***	−0.001
	(0.018)	(0.001)	(0.018)	(0.001)	(0.017)	(0.001)	(0.018)	(0.001)	(0.018)	(0.001)
lnregulation	−0.140 ***	−0.005 **	−0.131 ***	−0.004 *	−0.137 ***	−0.004	−0.135 ***	−0.004	−0.137 ***	−0.005 **
	(0.035)	(0.002)	(0.033)	(0.002)	(0.035)	(0.002)	(0.034)	(0.002)	(0.035)	(0.002)
lnT	0.020	0.006 ***	0.067 ***	0.005 ***	0.033 ***	0.009 ***	0.016	0.011 ***	0.034 **	0.008 ***
	(0.013)	(0.001)	(0.014)	(0.002)	(0.012)	(0.001)	(0.013)	(0.001)	(0.014)	(0.002)
lnS	0.123 *	−0.030 ***	0.165 **	−0.030 ***	0.140 **	−0.024 ***	0.103	−0.014 ***	0.141 **	−0.030 ***
	(0.067)	(0.005)	(0.066)	(0.004)	(0.065)	(0.005)	(0.064)	(0.004)	(0.067)	(0.005)
lndepth			−0.512 **	0.100 ***
			(0.205)	(0.026)
lndepth2			0.039	−0.008 **
			(0.024)	(0.003)
lndegree					−0.178	0.047 ***
					(0.175)	(0.016)
lndegree2					0.016	−0.003 *
					(0.020)	(0.002)
lndigital							0.645 *	0.173 ***
							(0.341)	(0.032)
lndigital2							−0.031	−0.009 ***
							(0.019)	(0.002)
lnfinancial									−0.350	0.229 ***
									(0.253)	(0.033)
lnfinancial2									0.033	−0.022 ***
									(0.029)	(0.004)
_cons	−4.773 ***	2.926 ***	−3.619 ***	2.777 ***	−4.412 ***	2.957 ***	−6.963 ***	2.214 ***	−4.027 ***	2.568 ***
	(0.594)	(0.073)	(0.497)	(0.080)	(0.583)	(0.075)	(1.610)	(0.157)	(0.659)	(0.092)
City fixed		YES		YES		YES		YES		YES
Year fixed		YES		YES		YES		YES		YES
N	2520	2520	2520	2520	2520	2520	2520	2520	2520	2520
R-sq	0.609	0.106	0.616	0.093	0.610	0.066	0.616	0.059	0.610	0.109
F	68.046	32.981	67.123	28.450	64.325	19.770	62.633	17.528	62.352	34.073
p	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000

Note: ***, **, and * of standard errors in parentheses are expressed within 0.01, 0.05, and 0.1, respectively.

does not present the results of RE. First, both $lnwidth$ and its square, $lndepth$ and its square, and $lndegree$ and its square have an “n”-type relationship with ${lnCO}_2$, which is the same as the results in Table 2. For inclusive finance, it once again confirmed that in the long run, accelerating the digital economy’s growth will contribute to lowering CO₂ emissions. After adding new variables, the impact of $lndigital$ and $lnfinancial$ on ${lnCO}_2$ is still an inverted “u”-shaped curve, which implies that the model is robust. Moreover, increasing the number of green patents will not be beneficial to lower carbon emissions in the short term. However, the adjustment of industrial composition will notably contribute to carbon emissions reduction.

4.2. Spatial Effects

4.2.1. Moran Index

Using the adjacency matrix (two cities are next to each other, the value is 0, if they are not next to each other, the value is 1), this study constructs the Moran index for CO₂ from 2011 to 2019. At the 1% level, the p-values of the Moran index in

Table 4. Moran’s I of global spatial autocorrelation.

Variables	I	z	p-Value *
C2011	0.018	3.243	0.001
C2012	0.018	3.239	0.001
C2013	0.014	2.722	0.003
C2014	0.015	2.754	0.003
C2015	0.016	3.022	0.001
C2016	0.017	3.137	0.001
C2017	0.015	2.823	0.002
C2018	0.016	2.974	0.001
C2019	0.015	2.892	0.002

* 1-tail test.

are all significant, illustrating that CO₂ emissions among Chinese cities are spatially dependent. The Z indicator dropped from 3.243 in 2011 to 2.892 in 2019, which can be described as a slight decrease in the regional correlation of carbon emissions.

The Moran index scatter diagram in Figure 4 is used to describe the degree of spatial difference in carbon emissions of various cities within the variable urban agglomeration. The X-axis in the figure (that is, the z value in Figure 4) is the dispersion of the variable y, and the Y-axis (that is, the $Wz$ value in Figure 4) is the spatial lag variable of the variable y. The scatter plot reflects the spatial agglomeration effect of carbon emissions of urban agglomerations. The carbon emission distribution of each city is divided into four situations: (1) The first quadrant: the carbon emissions of the city itself and surrounding cities are both high (HIGH-HIGH); (2) The second quadrant: the carbon emission of the city itself is lower than that of surrounding cities (LOW-HIGH); (3) The third quadrant: the carbon emissions of the city itself and surrounding cities are both low (LOW-LOW); (4) The fourth quadrant: the city’s own carbon emissions are higher than those of surrounding cities (HIGH-LOW).

The Moran index scatter points in Figure 4 are primarily concentrated in the second and fourth quadrants (the observed values in this area are lower or higher than the surrounding areas), and the characteristics of HIGH-LOW or LOW-HIGH indicate that the observations gathered in these areas show a negative space dependency. However, there are also some distributed in the first and third quadrants (with high or low observations in this area and surrounding areas), and the HIGH-HIGH or LOW-LOW distribution characteristics denote that the observations clustered in these cities exhibit positive spatial dependencies.

4.2.2. Spatial Spillover Effects

To evaluate how the digital economy spatially affects CO₂ emissions, this study uses Stata software to construct three spatial weight matrices based on the longitude and latitude data of 280 cities in China, namely the inverse distance matrix W1, economic distance matrix W2, and the square of the reciprocal distance W3. If the three matrices are represented by formulas, they are as follows:

$$W{1}_{ij}=\{\begin{array}{c}\frac{1}{D_{ij}}, i\ne j\\ 0, i=j\end{array},\mathit{distance} \mathit{from} i \mathit{to} j \mathit{is} \mathit{reciprocated};$$

$$W{2}_{ij}=\{\begin{array}{c}\frac{1}{\left|GD{P}_i-GD{P}_j\right|}, i\ne j\\ 0, i=j\end{array},\mathit{GDP} \mathit{difference} \mathit{from} i \mathit{to}j\mathit{is} \mathit{taken} \mathit{the} \mathit{absolute} \mathit{value} \mathit{and} \mathit{then} \mathit{reciprocated};$$

$$W{3}_{ij}=\{\begin{array}{c}{\left(\frac{1}{D_{ij}}\right)}^2, i\ne j\\ 0, i=j\end{array},\mathit{distance} \mathit{from} i \mathit{to} j \mathit{is} \mathit{reciprocated} \mathit{and} \mathit{then} \mathit{squared}.$$

Since the calculated Moran index explains that there is spatial autocorrelation between Chinese cities, this paper further uses the Wald and LR tests to find that the results have not failed the 1% significance test. This implies that the SDM model will not degenerate into the SLM model or the SEM model; the Hausman test denoted that the fixed effects model was used, so this study finally chose the SDM double fixed effect model. Adding different spatial weight matrices, this paper constructs Equation (2) and evaluates whether there is a spatial spillover effect of the impact of digital economic development in Chinese cities on carbon emissions through three approaches: patented technology, the economic dimension, and industrial composition.

Table 5. Spatial spillover effects of the inter-city digital economy on CO₂.

	Direct Effect (SDM)			Indirect Effect (SDM)			Total Effect (SDM)
Matrix	W1	W2	W3	W1	W2	W3	W1	W2	W3
lndigital	0.258 ***	0.264 ***	0.253 ***	0.404	0.232 **	−0.023	0.662	0.496 ***	0.230 **
	(0.036)	(0.035)	(0.036)	(0.409)	(0.096)	(0.101)	(0.414)	(0.105)	(0.108)
lndigital*lnT	0.000	0.000	0.000	0.001	0.000	0.001	0.001	0.000	0.001
	(0.000)	(0.000)	(0.000)	(0.002)	(0.001)	(0.001)	(0.002)	(0.001)	(0.001)
lndigital*lnG	−0.016 ***	−0.016 ***	−0.015 ***	−0.033	−0.016 ***	−0.001	−0.049 *	−0.032 ***	−0.016 **
	(0.002)	(0.002)	(0.002)	(0.026)	(0.006)	(0.006)	(0.026)	(0.007)	(0.007)
lndigital*lnS	−0.002 ***	−0.002 ***	−0.002 ***	−0.004	−0.002	−0.001	−0.006	−0.004 **	−0.003 **
	(0.000)	(0.000)	(0.000)	(0.005)	(0.001)	(0.001)	(0.005)	(0.002)	(0.001)
lnpop	0.163 ***	0.166 ***	0.160 ***	0.343	0.175 ***	0.019	0.505 **	0.341 ***	0.179 ***
	(0.021)	(0.021)	(0.021)	(0.243)	(0.060)	(0.058)	(0.246)	(0.066)	(0.063)
lnpgdp	0.154 ***	0.158 ***	0.151 ***	0.322	0.157 ***	0.010	0.477 **	0.315 ***	0.161 ***
	(0.020)	(0.020)	(0.020)	(0.239)	(0.057)	(0.057)	(0.242)	(0.063)	(0.061)
lnopenness	−0.001	−0.001	−0.001	−0.009	−0.013 ***	−0.006 *	−0.010	−0.014 ***	−0.006 *
	(0.001)	(0.001)	(0.001)	(0.012)	(0.003)	(0.003)	(0.012)	(0.003)	(0.003)
lnregulation	−0.007 ***	−0.006 ***	−0.006 ***	−0.069 **	−0.011 **	−0.020 ***	−0.076 ***	−0.017 ***	−0.026 ***
	(0.002)	(0.002)	(0.002)	(0.027)	(0.005)	(0.007)	(0.027)	(0.005)	(0.007)
City fixed	yes	yes	yes	yes	yes	yes	yes	yes	yes
Year fixed	yes	yes	yes	yes	yes	yes	yes	yes	yes
N	2520	2520	2520	2520	2520	2520	2520	2520	2520
R-sq	0.060	0.077	0.094	0.060	0.077	0.094	0.060	0.077	0.094

Note: ***, **, and * of standard errors in parentheses are expressed within 0.01, 0.05, and 0.1, respectively.

reflects that the urban digital economy ($lndigital)$ has a spatial spillover effect on CO₂ emissions through three approaches. First, in terms of direct effects, in the short term, patented technology ($lnT$) of digital economy ($lndigital)$ has a positive but insignificant impact on CO₂ emissions, which further confirms the “rebound effect” described by Xie et al. and other scholars. Technological progress improves energy efficiency while increasing energy demand, thereby stimulating more production activities and raising CO₂ emissions [5,45]. The digital economy ($lndigital)$ can reduce CO₂ emissions through the economies’ size ($lnG$) and industrial composition adjustment ($lnS$). This has the following explanation: (1) For every 0.01 point decrease in the size of the city’s economy ($lnG$), the reduction in the impact of the digital economy ($lndigital)$ on the city’s CO₂ ranges from 0.015 to 0.016, and for every 0.01 point decline in the city’s industrial composition adjustment ($lnS)$, the reduction in the impact of the digital economy ($lndigital)$ on the city’s CO₂ emissions is about 0.002; (2) The digital economy ($lndigital$) of the city affects the CO₂ emission of neighboring cities through economies of scale or the adjustment of industrial components, and in turn affects the CO₂ emissions of the city due to the mobility of capital factors.

Secondly, the following findings stem from the indirect effect of SDM: (1) For every 0.01 point reduce in the economy size ($lnG$) of the neighboring city, the impact of the neighboring city’s digital economy ($lndigital)$ on the CO₂ emissions of the city will drop by about 0.016; (2) The digital economy ($lndigital)$ of the neighboring city affects its own CO₂ emissions through economy size ($lnG$), and finally affects the CO₂ emission of the city through the circulation system, implying that the digital economy ($lndigital)$ of the neighboring city can affect the CO₂ emission of the city through economy size. This is because when regional governments formulate development policies, in addition to considering the country’s macro and micro policies, they also imitate or compete with neighboring cities, resulting in spatial dependence. It further confirms the “following effect” and “competition effect” mentioned by Levinson and other scholars [5,7,13].

In addition, increasing the population size ($lnpop$) and urban per capita GDP ($lnpgdp$) is detrimental to the decarbonization of the city itself, and it is not conducive to the decarbonization of surrounding areas, while expanding trade openness ($lnopenness$) and developing appropriately stringent environmental regulations ($lnregulation$) can help decarbonize the city and its neighbors [5,36,38,39,40,41,46,47,48].

4.2.3. Robustness

This study chooses digital financial inclusion ($lnfinancial$) to replace $lndigital$ to verify the robustness of the model.

Table 6. Robustness test.

	Direct Effect (SDM)			Indirect Effect (SDM)			Total Effect (SDM)
Matrix	W1	W2	W3	W1	W2	W3	W1	W2	W3
lnfinancial	0.384 ***	0.387 ***	0.388 ***	0.425	0.176 *	0.016	0.809 **	0.564 ***	0.404 ***
	(0.034)	(0.034)	(0.034)	(0.407)	(0.101)	(0.098)	(0.411)	(0.111)	(0.105)
lnfinancial*lnT	0.001 **	0.001 **	0.001 **	0.002	0.000	0.001	0.003	0.001	0.002 *
	(0.000)	(0.000)	(0.000)	(0.004)	(0.001)	(0.001)	(0.004)	(0.001)	(0.001)
lnfinancial*lnG	−0.022 ***	−0.022 ***	−0.022 ***	−0.045	−0.012*	−0.005	−0.068 **	−0.034 ***	−0.027 ***
	(0.002)	(0.002)	(0.002)	(0.030)	(0.008)	(0.007)	(0.030)	(0.008)	(0.008)
lnfinancial*lnS	−0.004 ***	−0.004 ***	−0.004 ***	−0.006	−0.002	−0.002	−0.010	−0.006 **	−0.006 **
	(0.001)	(0.001)	(0.001)	(0.008)	(0.002)	(0.002)	(0.008)	(0.002)	(0.002)
lnpop	0.151 ***	0.153 ***	0.153 ***	0.254	0.105 **	0.026	0.405 **	0.258 ***	0.179 ***
	(0.015)	(0.015)	(0.015)	(0.175)	(0.050)	(0.042)	(0.177)	(0.054)	(0.045)
lnpgdp	0.140 ***	0.143 ***	0.141 ***	0.229	0.079 *	0.019	0.369 **	0.221 ***	0.160 ***
	(0.014)	(0.014)	(0.014)	(0.168)	(0.043)	(0.040)	(0.170)	(0.047)	(0.043)
lnopenness	−0.002	−0.002	−0.002	−0.002	−0.012 ***	−0.005 *	−0.003	−0.013 ***	−0.007 **
	(0.001)	(0.001)	(0.001)	(0.013)	(0.003)	(0.003)	(0.013)	(0.003)	(0.003)
lnregulation	−0.004 **	−0.004 **	−0.004 **	−0.057 **	−0.010 **	−0.021 ***	−0.061 **	−0.014 ***	−0.026 ***
	(0.002)	(0.002)	(0.002)	(0.027)	(0.005)	(0.007)	(0.027)	(0.005)	(0.007)
City fixed	yes	yes	yes	yes	yes	yes	yes	yes	yes
Year fixed	yes	yes	yes	yes	yes	yes	yes	yes	yes
N	2520	2520	2520	2520	2520	2520	2520	2520	2520
R-sq	0.027	0.408	0.413	0.027	0.408	0.413	0.027	0.408	0.413

Note: ***, **, and * of standard errors in parentheses are expressed within 0.01, 0.05, and 0.1, respectively.

denotes that the replaced variable $lnfinancial$ can still affect the decarbonization of cities and surrounding areas in three ways; namely, $lnT$,$lnG,$ and $lnS$, which implies that digital financial inclusion’s growth is also spatially dependent on urban CO₂ emissions [5,32]. Results for other control variables are consistent with those in Table 5, which further confirms the robustness and effectiveness of the model.

4.2.4. Heterogeneity

Due to the geographical differences in the digital economy’s growth, 280 cities were classified into northeastern, western, eastern, and central areas from the perspective of geographical distribution (the standard of zoning refers to the geographical distribution of China), and then assessed on whether there are regional spatial spillover differences in the impact of the digital economy ($lndigital)$ on CO₂ emissions. The comparison results in

Table 7. Heterogeneity test by region.

	Direct Effects (Central City)			Indirect Effects (Central City)			Total Effects (Central City)
Matrix	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM
lndigital	−0.148	−0.153	−0.178 *	−2.018 ***	−0.845 ***	−1.252 ***	−2.166 ***	−0.998 ***	−1.430 ***
	(0.092)	(0.096)	(0.093)	(0.448)	(0.286)	(0.206)	(0.447)	(0.315)	(0.218)
lndigital*lnT	0.000	0.000	0.000	0.002	−0.000	0.001	0.003	0.000	0.001
	(0.000)	(0.000)	(0.000)	(0.002)	(0.001)	(0.001)	(0.002)	(0.001)	(0.001)
lndigital*lnG	0.008	0.009	0.010 *	0.132 ***	0.056 ***	0.080 ***	0.140 ***	0.065 ***	0.090 ***
	(0.006)	(0.006)	(0.006)	(0.028)	(0.017)	(0.013)	(0.028)	(0.019)	(0.014)
lndigital*lnS	−0.003 ***	−0.003 ***	−0.003 ***	−0.017 ***	−0.004	−0.007 ***	−0.020 ***	−0.006 **	−0.010 ***
	(0.001)	(0.001)	(0.001)	(0.004)	(0.002)	(0.002)	(0.004)	(0.003)	(0.002)
N	702	702	702	702	702	702	702	702	702
R-sq	0.009	0.003	0.011	0.009	0.003	0.011	0.009	0.003	0.011
	Direct Effects (Eastern City)			Indirect Effects (Eastern City)			Total Effects (Eastern City)
Matrix	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM
lndigital	0.430 ***	0.424 ***	0.435 ***	−1.293 ***	−0.410*	−0.474 ***	−0.863 **	0.014	−0.039
	(0.064)	(0.065)	(0.063)	(0.332)	(0.212)	(0.161)	(0.339)	(0.223)	(0.180)
lndigital*lnT	0.00031	0.00029	0.00025	0.002	0.001	0.001	0.003 *	0.001*	0.001
	(0.000)	(0.000)	(0.000)	(0.002)	(0.001)	(0.001)	(0.002)	(0.001)	(0.001)
lndigital*lnG	−0.025 ***	−0.025 ***	−0.026 ***	0.077 ***	0.024 *	0.028 ***	0.052 ***	−0.001	0.003
	(0.004)	(0.004)	(0.004)	(0.020)	(0.012)	(0.010)	(0.020)	(0.013)	(0.011)
lndigital*lnS	−0.001	−0.001	−0.001	0.006	0.002	0.005 **	0.006	0.001	0.004 *
	(0.001)	(0.001)	(0.001)	(0.005)	(0.002)	(0.002)	(0.005)	(0.002)	(0.002)
N	783	783	783	783	783	783	783	783	783
R-sq	0.187	0.178	0.238	0.187	0.178	0.238	0.187	0.178	0.238
	Direct Effects (Northeast City)			Indirect Effects (Northeast City)			Total Effects (Northeast City)
Matrix	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM
lndigital	0.618 *	0.143	−0.014	17.955 **	4.870 ***	2.722	18.573 **	5.014 ***	2.707
	(0.317)	(0.193)	(0.217)	(7.372)	(1.182)	(1.656)	(7.648)	(1.323)	(1.821)
lndigital*lnT	0.00037	0.00037	−0.00010	0.016	0.009 ***	0.003	0.016	0.010 ***	0.003
	(0.001)	(0.001)	(0.001)	(0.015)	(0.003)	(0.005)	(0.016)	(0.004)	(0.005)
lndigital*lnG	−0.036 *	−0.007	0.003	−1.105 **	−0.298 ***	−0.158	−1.141 **	−0.305 ***	−0.155
	(0.020)	(0.012)	(0.014)	(0.458)	(0.072)	(0.105)	(0.475)	(0.081)	(0.116)
lndigital*lnS	0.018 ***	0.009 ***	0.013 ***	0.270 ***	0.054 ***	0.094 ***	0.288 ***	0.064 ***	0.107 ***
	(0.004)	(0.001)	(0.002)	(0.102)	(0.008)	(0.019)	(0.106)	(0.009)	(0.021)
N	297	297	297	297	297	297	297	297	297
R-sq	0.032	0.206	0.101	0.032	0.206	0.101	0.032	0.206	0.101
	Direct Effects (Western City)			Indirect Effects (Western City)			Total Effects (Western City)
Matrix	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM	W1_SDM	W2_SDM	W3_SDM
lndigital	0.133 ***	0.137 ***	0.141 ***	0.036	−0.038	−0.057	0.169	0.098	0.084
	(0.047)	(0.046)	(0.046)	(0.358)	(0.071)	(0.119)	(0.371)	(0.087)	(0.134)
lndigital*lnT	0.00037	0.00027	0.00032	0.003	0.001 **	0.001	0.003	0.002 **	0.001
	(0.000)	(0.000)	(0.000)	(0.002)	(0.001)	(0.001)	(0.003)	(0.001)	(0.001)
lndigital*lnG	−0.008 ***	−0.009 ***	−0.009 ***	−0.006	0.002	0.002	−0.014	−0.006	−0.007
	(0.003)	(0.003)	(0.003)	(0.023)	(0.005)	(0.008)	(0.024)	(0.006)	(0.009)
lndigital*lnS	−0.003 ***	−0.002 ***	−0.003 ***	0.007	0.001	0.001	0.0040	−0.0010	−0.0013
	(0.001)	(0.001)	(0.001)	(0.006)	(0.001)	(0.002)	(0.006)	(0.001)	(0.002)
N	738	738	738	738	738	738	738	738	738
R-sq	0.000	0.001	0.001	0.000	0.001	0.001	0.000	0.001	0.001

Notes: ***, **, and * of standard errors in parentheses are expressed within 0.01, 0.05, and 0.1, respectively. The double fixed effects of the SDM model are used here, that is, both city and year are fixed.

are summarized as follows: (1) Technical level: digital economy ($lndigital)$ in the central as well as eastern areas has no visible influence on CO₂ emissions through patented technologies, while the impact of digital economy ($lndigital)$ in the northeastern and western regions on CO₂ emissions has a technological “rebound effect” [5,45]; (2) Scale of the economy: digital economy ($lndigital)$ will increase the CO₂ emissions of the region and its neighbors with an expansion in the economic size of the central region. This may be explained by the fact that the digital economy in the central region is in a stage of rapid development, and the construction of digital infrastructure has indirectly increased CO₂ emissions. Moreover, because the core cities in the central region have significant advantages, it has caused a “siphon effect” [32]. However, the digital economy ($lndigital)$ in the other three regions will help the decarbonization of the region and its neighbors with the help of the economy’s size, indicating that the formation of economies of scale is also an effective way to lower CO₂ emissions; (3) Industrial composition level: the digital economy ($lndigital)$ in the western as well as central areas can contribute to achieving the target of urban low-carbonization with the help of industrial upgrading. Compared with the remaining three areas, the eastern area has a high degree of development at the level of digital infrastructure and digital industries, coupled with the blessing of innovative talents and capital, which makes the environmental effect of the fine-tuning of the industrial structure in the eastern region less obvious. The level of the digital economy in northeast China is relatively low, and the energy generated by economic production activities and the effects of digital empowerment are offset, resulting in insignificant environmental effects [32].

5. Conclusions and Policy Implications

5.1. Conclusions

This study primarily assesses how the digital economy of 280 cities affected carbon emissions from 2011 to 2019. For research purposes, we conducted a detailed analysis of the direct effects, the mediated effects, and the spatial spillover effects of different dimensions of China’s urban digital economy on carbon emissions. Conclusions can be described as follows.

• An “n”-type curve nexus exists between CO₂ emissions and the digital economy in Chinese cities, which means that the city’s digital economy initially boosts carbon emissions and then reduces emissions after reaching a certain level;
• There are spatial spillover effects and regional heterogeneity about how digital economic growth has influenced CO₂ emissions, and approaches of economies of scale and industrial composition upgrades can help the city lower carbon emissions and benefit the low carbonization of neighboring cities. However, based on the “rebound effect”, the intermediary role of technology effects in reducing emissions in the short term is not evident;
• There are differences between regions. The digital economy will increase the CO₂ of the region and its neighbors with the expansion in the economic size of the central region. But the other three regions are quite the opposite. The digital economy in the central and western regions can be beneficial to accomplish the target of urban low-carbonization with the help of industrial upgrading. Compared with the other three regions, the eastern region has a high degree of development at the level of digital infrastructure and digital industries, coupled with the blessing of innovative talents and capital, which make the environmental effect of the fine-tuning of the industrial structure in the eastern region less obvious;
• Expanding trade openness and developing appropriately stringent environmental rules can be beneficial in the short-term to lower CO₂ emissions in one’s own city and neighboring cities.

5.2. Policy Implications

Based on the conclusions drawn from the empirical analysis, this study proposes the following constructive suggestions.

First, the digital economy needs to develop to a certain level in order to contribute to the low-carbon development of cities. To realize a low-carbon and clean city at an early date, policy makers can first keep up with the accelerating pace of the Internet and formulate appropriate policies to accelerate a city’s inclusive finance and digital economy’s growth.

Second, based on the characteristics of inter-city spatial dependence, the government can imitate the surrounding cities that have implemented the low-carbon goal well, learn advanced management concepts and development models from them, allocate resources reasonably, and encourage enterprises to put green and sustainable development ideas into action.

In addition, policymakers need to balance the gap between cities and strengthen cooperation and exchanges between cities to form economies of scale or upgrade the industrial structure to accomplish the target of “carbon neutrality” sooner rather than later. Even if the region quickly becomes an eco-city, the fluidity of space can lead to high-emitting neighbors affecting the region.

Finally, governments and businesses should actively expand trade openness, as well as formulate appropriate and stringent environmental regulations in line with national conditions, which can effectively contribute to carbon reductions in the short term.

Considering that previous studies did not form a unified indicator when calculating the digital economy indicators, although this study examines the development level of the digital economy in Chinese cities from multiple dimensions, it inevitably has some limitations. This study did not consider the phenomenon of carbon transfer when analyzing the impact of the urban digital economy on carbon emissions. In fact, the formation of carbon footprints in the process of digital development is also an obvious phenomenon. A more comprehensive evaluation of the development of a city or country’s digital economy and an identification of the indicators that can be used to demonstrate more effectively the level of the digital economy’s development in a city or region, deserve follow-up research.