The Synergistic Effect of Digital Consumption on Pollution and Carbon Reduction

Abstract

As an emerging environmentally sustainable consumption paradigm, digital consumption plays a pivotal role in synergistic pollution and carbon reduction. Based on provincial-level panel data from China covering from 2013 to 2024, this study adopts the two-way fixed effect model, the mediating effect model, and the threshold model to examine the impact and mechanisms of digital consumption in promoting the synergistic effect of pollution and carbon reduction and provide a theoretical framework and practical guidance. The results demonstrate that digital consumption exerts a statistically significant synergistic effect on pollution and carbon reduction, even after endogenous and robustness tests. This effect is more pronounced in regions with supportive digital policy environment, favorable carbon control policy, lower degrees of industrialization, and higher levels of environmental protection. Industrial structure upgrading and green technology innovation are two primary transmission channels through which digital consumption enhances the synergistic effect of pollution and carbon reduction. This effect is significantly stronger when industrial structure upgrading and green technology innovation exceed the threshold values of 3.6253 and 2.0374, respectively.

1. Introduction

Promoting green and low-carbon development is important for modern industrial systems and high-quality development. The 2030 Agenda for Sustainable Development, adopted by the United Nations in 2015, puts forward 17 Sustainable Development Goals (SDGs). Among them, goal 12 aims to ensure sustainable consumption and production patterns, calling on all countries to minimize the use of natural resources and toxic materials throughout the lifecycle of products and services and to reduce waste and pollutant emissions. China has always prioritized the implementation of SDGs, accelerating green social and economic social development, and has achieved remarkable progress in green and low-carbon development. However, structural and other problems in ecological and environmental protection have led to insufficient endogenous momentum. Therefore, it is necessary to transition from a high-energy-consumption and high-pollution development model to a green economy, as well as a circular economic model [1]. China also emphasizes coordinated governance of pollution and carbon reduction and has made significant strategic interventions. The Ministry of Ecology and Environment and six other government departments jointly issued the “Implementation Plan for Synergistic Pollution and Carbon Reduction,” emphasizing the need to explore effective models for synergistic pollution and carbon reduction. The Fourth Plenary Session of the 20th Central Committee of the Communist Party of China proposed strengthening the coordination of pollution and carbon reduction, multi-pollutant control, and regional governance. Because carbon emissions and environmental pollutant emissions originate from the same sources [2], promoting synergistic efficiency in pollution and carbon reduction has become an important strategic direction.

For digital economic and social transformation, consumption must be restructured in terms of its forms, mechanisms, and effects [3]. With the deep integration of digital technologies in consumption, such as the internet, big data, cloud computing, and artificial intelligence, digital consumption is profoundly transforming consumption patterns. As an emerging consumption model, digital consumption is dynamically evolving. Belk [4] argues that digital consumption is an economic activity that targets digital products or services for consumption. Istrate et al. [5] propose that digital consumption encompasses consumption activities conducted through digital channels, as well as all new consumption formats that emerge with the update and iteration of digital technologies. Digital consumption is the product of the deep integration of the technological revolution with the social consumption sector [6]. It is based on digital terminals and relies on digital technology support to promote effective matching between the supply and demand structure of digital products and digital services [7]. As such, this study defines digital consumption as a consumption model grounded in data and digital technologies, which optimizes the processes and experiences of traditional consumption through digital means, offering digital products and services to consumers.

To accelerate the development and expansion of digital consumption, the Ministry of Industry and Information Technology has started constructing pilot cities for information consumption, supporting new formats and models for information consumption. The National Development and Reform Commission issued the “Measures for Restoring and Expanding Consumption,” proposing to expand digital consumption and promote green consumption. The General Office of the Communist Party of China Central Committee and the General Office of the State Council have issued the “Special Action Plan to Boost Consumption,” emphasizing a digital consumption enhancement initiative. With strong support from the government, China’s digital consumption is flourishing. According to data from the National Bureau of Statistics, in 2024, China’s total retail sales of consumer goods and online retail sales reached CNY 48.33 trillion and CNY 15.23 trillion, accounting for 35.82% and 11.29% of its GDP. In January 2026, the National E-commerce Work Conference found that the overall scale of digital consumption in China exceeded CNY 23.8 trillion, maintaining its position as the world’s largest online retail market for 13 consecutive years. The sustained development of digital consumption will optimize both the demand and supply sides, further supporting ecological protection. As an emerging environmentally sustainable consumption paradigm, digital consumption has become deeply integrated into all areas of economic and social development, playing an important role in green and low-carbon development, opening up new pathways, and advancing the synergistic effect of pollution and carbon reduction.

The existing literature has primarily examined the impact of information consumption pilot policies on carbon emissions reduction, carbon productivity, and green development. Wilson et al. [8] argue that digital consumer innovation has significant but varying potential emission reduction benefits in the fields of transportation, food, home, and energy. Liu et al. [9] found that the information consumption pilot policy has significant positive direct effects and spatial spillover effects on reducing carbon emissions, mainly achieved through technological innovation and industrial upgrading. Deng and Qian [10] found that the information consumption pilot policy has improved the efficiency of technological applications and resource allocation by promoting technological innovation and enhancing resource agglomeration. It has also prompted industrial upgrades, accelerating low-carbon industry development and improving carbon productivity. Wang et al. [11] determined that the information consumption pilot policy promotes urban green development through an agglomeration effect, green transportation, and information technology.

While the literature includes research on the relationship between the information consumption pilot policy and green and low-carbon development, further research is needed on the synergistic effect of pollution and carbon reduction in digital consumption. As such, this study systematically examines how digital consumption supports the synergistic effect of pollution and carbon reduction, based on provincial-level panel data from China from 2013 to 2024, adopts the two-way fixed effect model, the mediation effect model, and the threshold model for quantitative analysis. According to the Notice of the “General Office of the Ministry of Commerce on Implementing the Action Plan to Boost Digital Consumption,” digital consumption mainly refers to consumers’ consumption behaviors in the digital environment, including the digitalization of products and services, as well as the digitalization of content and channels. Based on this definition and drawing on the relevant literature, this study constructs an evaluation index system for digital consumption from four dimensions of digital consumption (development potential, digital consumption development guarantee, digitalization level of consumption content, and digitalization level of consumption pattern) and then uses a fixed-base range entropy weight method to measure digital consumption level, enhancing existing evaluation frameworks and providing data for subsequent quantitative research. The study also systematically analyzes the synergistic effect of digital consumption on pollution and carbon reduction and reveals how digital consumption supports this effect from the perspectives of industrial structure upgrading and green technology innovation, expanding the breadth and depth of the literature. Finally, the study empirically examines the impact, heterogeneity, and pathways of digital consumption in advancing the synergistic effect of pollution and carbon reduction to provide policy support for fully leveraging this effect.

2. Theoretical Analysis and Research Hypotheses

According to consumer sovereignty theory and demand-driven production theory, consumer demand reshapes supply-side production methods, technological choices, and resource allocation. As an emerging environmentally sustainable consumption paradigm, digital consumption is not merely the online migration of distribution channels but rather a systematic reconstruction of supply–demand relationships in traditional social reproduction. In the paradigm of digital consumption, consumers’ ability to release demand signals is significantly enhanced, entailing a logical chain of “demand induction–circulation optimization–supply transformation.” From the production side, the green consumption demand of digital consumption helps drive enterprises to accelerate the production of green and intelligent products, significantly alleviating the environmental burden on the production chain [12]. User preferences and market demand data accumulated through digital consumption serve as new production factors that provide reverse feedback to the supply side, giving rise to C2B flexible customization and the production-by-sales model, compelling enterprises to break away from path dependence on high-energy-consuming, high-pollution manufacturing, reducing energy and material consumption during the production process, promoting optimal resource allocation and circular utilization, injecting momentum into green development at the source, and realizing the synergistic effect of pollution and carbon reduction on the production side. From the circulation side, digital consumption provides sustained market demand and diverse application scenarios for digital technology innovation, promoting the deep penetration and application of emerging technologies in the consumer field. Through supply chain optimization, intelligent logistics scheduling, and green supply chain management, it reduces resource waste and environmental pollution. Meanwhile, the rapid iteration and personalized upgrading of digital consumption demands drive supply chains to transform from traditional “linear series” to modern “networked collaboration,” enhancing supply chain digitalization, enabling efficient operation across all segments, and effectively reducing inventory backlog and capital occupation [13], achieving cost reduction and efficiency improvement [14] and promoting the synergistic effect of pollution and carbon reduction on the circulation side.

From the consumption side, the deep integration of digital technology with consumption patterns has driven digital, personalized, and customized changes [15], while the transition to online platforms has reduced environmental pollution and carbon emissions associated with traditional offline consumption. At the same time, digital consumption leverages e-commerce platforms and social media channels to enable consumers to efficiently access information on green products and environmental knowledge, enhancing awareness of green consumption, raising the level of green consumption, and achieving the synergistic effect of pollution and carbon reduction on the consumer side. In summary, digital consumption achieves environmental dividends through the collaborative digital and low-carbon restructuring of the “consumption–circulation–production” chain. Based on this analysis, this study proposes the following hypothesis:

Hypothesis 1 (H1).

Digital consumption plays a positive role in the synergistic effect of pollution and carbon reduction.

Digital consumption represents not only a shift in consumption patterns but also a systemic reshaping of the demand structure. This transformation leads to the synergistic effect of pollution and carbon reduction by driving industrial upgrading. Digital consumption breaks through the limitations of traditional consumption models, promotes deep integration of online and offline consumption, and brings together traditional industries and information technology [16]. On the one hand, the rapid development of digital consumption drives the continuous upgrading of demand structure toward service-oriented, lightweight, and low-carbon models, forcing the supply side to break away from path dependence on high-energy-consumption, high-emission, and resource-based industries. This allows for advanced coordination of the digital service industry, the high-tech manufacturing industry, and the green low-carbon industry. The proportion of low-energy-consumption and high-value-added industries, such as online services, digital content, smart cultural tourism, and remote services, continues to increase. Consumption upgrading drives industrial structure upgrading and industrial structure upgrading, supporting green transformation. On the other hand, digital consumption also has economic effects. As consumers increasingly choose green products, digital services, and low-carbon lifestyles through digital platforms, this demand-side change will guide the efficient flow of production factors, such as capital, labor, and technology, from traditional high-energy-consuming and high-polluting industries to digital, intelligent, and green industries through price signals and market choices. This promotes industrial structure upgrading and the synergistic effect of pollution and carbon reduction. Xue et al. [17] found that information consumption pilot policies significantly promote the development of producer services and modern services and drive upgrades in service industry structures by enhancing capital allocation efficiency and facilitating technological progress.

Hypothesis 2 (H2).

Digital consumption promotes the synergistic effect of pollution and carbon reduction through industrial structure upgrading.

The deep integration of digital consumption and green technology innovation has become an important driving force for reducing pollution and carbon emissions and green and low-carbon economic and social transformation. On the one hand, digital consumption expands the market space for green products, effectively reducing the cost of green technology research and application. Moreover, user data generated by the expansion of digital consumption not only provide enterprises with abundant innovation resources but also help traditional enterprises integrate data elements into their production and research and development processes [18], encouraging enterprises to pursue green technological innovation. Green technological innovation accelerates the widespread adoption of clean energy and drives the transformation of traditional high-energy-consumption, high-emission industries toward green and low-carbon development. On the other hand, according to the theory of demand-induced innovation [19], the continuous expansion of digital consumption generates demand for environmentally friendly, low-carbon, and intelligent consumption models, creating new application scenarios and market spaces for green technologies. Digital consumption data are fed back to the production end through digital platforms, leading enterprises to green technology innovation and guiding green technology research and development in energy conservation, pollution and carbon reduction, and resource recycling. Furthermore, digital consumption helps promote digital technological innovation. It not only reduces information asymmetry between manufacturers and consumers regarding product supply and demand but also alleviates the information disconnection between traditional enterprises and research institutions in the process of technology research and development. Consumer demand compels enterprises to engage in green technology innovation [20], supporting the synergistic effect of pollution and carbon reduction.

Hypothesis 3 (H3).

Digital consumption promotes the synergistic effect of pollution and carbon reduction through green technology innovation.

3. Model Construction and Variable Measurement

3.1. Model Construction

To examine the synergistic effect of pollution and carbon reduction in digital consumption, the two-way fixed effects model is constructed as follows:

$$Y_{it}={\beta }_0+{\beta }_{1}D{C}_{it}+{\beta }_2{X}_{it}+{\delta }_i+{\eta }_t+{\epsilon }_{it}$$

(1)

where i denotes the province and t denotes the year. Y_it is the dependent variable, representing carbon emissions, pollution emissions, and the synergistic effect of pollution and carbon reduction. DC_it is the explanatory variable, indicating digital consumption. X_it represents the control variables. δ_i and η_t represent province and year fixed effects, and ε_it is the random error term.

To examine the impact of digital consumption on the synergistic effect of pollution and carbon reduction, based on the study by Li et al. [21], the following model is used:

$$M_{it}={\theta }_0+{\theta }_{1}D{C}_{it}+{\theta }_2{X}_{it}+{\delta }_i+{\eta }_t+{\epsilon }_{it}$$

(2)

where M_it represents the mediating variables, indicating industrial structure upgrading and green technology innovation. The other symbols are consistent with model (1).

To further investigate whether the impact of digital consumption on the synergistic effect of pollution and carbon reduction is subject to threshold effects of industrial structure upgrading and green technology innovation, based on the study by He et al. [22], the following threshold model is used:

$$Coo{r}_{it}={\pi }_0+{\pi }_{1}D{C}_{it}I\left(M_{it}\le \gamma \right)+{\pi }_{2}D{C}_{it}I\left(M_{it}>\gamma \right)+{\pi }_3{X}_{it}+{\delta }_i+{\eta }_t+{\epsilon }_{it}$$

(3)

where Coor_it represents the synergistic effect of pollution and carbon reduction, I(•) is the indicator function, and γ is the threshold value to be estimated. π₁ and π₂ represent the regression coefficient of digital consumption on the synergistic effect of pollution and carbon reduction when threshold variables are in different intervals. The other symbols are consistent with model (1). The model can be expanded into double or triple threshold models.

3.2. Variable Measurement

3.2.1. Dependent Variables

To analyze carbon emissions (lnCO₂), pollution emissions (lnSO₂), and the synergistic effect of pollution and carbon reduction (Coor), following the approach of Wang et al. [23], this study uses the natural logarithms of carbon dioxide and sulfur dioxide emissions. Referring to Li et al. [24], this study adopts the coupling coordination degree model to measure the synergistic effect of pollution and carbon reduction based on the range-normalized lnCO₂ and lnSO₂. Because both lnCO₂ and lnSO₂ are negative indicators, a smaller measured value indicates a better synergistic effect of pollution and carbon reduction. The coupling coordination degree model is constructed as follows:

$$C={\displaystyle \frac{2\sqrt{\ln C{O}_2\times \ln S{O}_2}}{\ln C{O}_2\times \ln S{O}_2}}$$

(4)

$$T=0.5\ln C{O}_2+0.5\ln S{O}_2$$

(5)

$$Coor=\sqrt{C\times T}$$

(6)

3.2.2. Explanatory Variable

To analyze digital consumption (DC), this study constructs an evaluation index system and uses a comprehensive evaluation method to compile an index to characterize the level of digital consumption based on four dimensions: digital consumption development potential, digital consumption development guarantee, digitalization level of consumption content, and digitalization level of consumption pattern (see

Table 1. Evaluation index system for digital consumption.

Indicators	Secondary Indicators
digital consumption development potential	Digital consumption development potential covers four indicators: per capita disposable income of residents, per capita consumption expenditure of residents, per capita GDP, and per capita retail sales of social consumer goods.
digital consumption development guarantee	Digital consumption development guarantee covers four indicators: broadband internet subscribers per 100 people, mobile internet users per 100 people, mobile phone penetration rate, and ratio of enterprises with e-commerce trading activities.
digitalization level of consumption content	Digitalization level of consumption content covers five indicators: ratio of e-commerce sales to total retail sales of consumer goods, ratio of online retail sales to total retail sales of consumer goods, ratio of express delivery revenue to GDP, ratio of software business revenue to GDP, and ratio of information technology service revenue to GDP.
digitalization level of consumption pattern	Digitalization level of consumption pattern covers four indicators: per capita express delivery volume, digital payment coverage, digital payment usage intensity, and degree of payment digitization.

). Digital consumption development potential reflects consumers’ latent purchasing power and willingness to consume, which is crucial for the development of digital consumption. Digital consumption development guarantee encompasses the digital infrastructure system that serves digital consumption, primarily focusing on information, logistics, and production services, providing technical support for the development of digital consumption. The digitalization level of consumption content is manifested in the digital presentation of products and services, a prerequisite for the development of digital consumption. The digitalization level of consumption pattern is reflected in the use of digital technologies to digitize payments, transactions, and other consumption processes, serving as a crucial vehicle for the development of digital consumption. A fixed-base range entropy weight method is used to measure digital consumption levels.

3.2.3. Mediating Variables

Industrial structure upgrading is measured based on the ratio of the value added of the tertiary sector to that of the secondary sector. Green technology innovation is measured based on the number of green and low-carbon patent applications per 10,000 people.

3.2.4. Control Variables

Drawing on the literature, this study selects government support intensity (Fis), environmental regulation (ER), foreign trade openness (Open), and education investment intensity (Ed) as control variables. These are measured by the proportion of fiscal expenditure to GDP, the proportion of investment in industrial pollution control to GDP, the proportion of total imports and exports to GDP, and the proportion of education expenditure to fiscal expenditure.

3.3. Data Sources

This study uses data from 30 Chinese provinces (excluding Tibet, Hong Kong, Macao, and Taiwan) from 2013 to 2024 as the sample for empirical analysis. Specifically, carbon dioxide emissions data are sourced from the China Emissions Accounting Database (CEADs) and green and low-carbon patent applications are sourced from the China National Intellectual Property Administration. Digital payment coverage, digital payment usage, and payment digitization levels are sourced from the Digital Inclusive Finance Index, compiled by the Center for Digital Finance at Peking University, and other data are sourced from the China Statistical Yearbook, Provincial Statistical Yearbooks, and the National Bureau of Statistics website. Missing data are supplemented using trend extrapolation or moving average methods. All statistical data analyses in this study are performed using Stata software (Version 18, StataCorp LLC., College Station, TX, USA). The descriptive statistics for each variable are shown in

Table 2. Descriptive statistics.

Variables	Obs.	Mean	S.D.	Min	Max
lnCO2	360	5.6381	0.7356	3.6757	6.8535
lnSO2	360	2.5846	1.3927	−2.2073	5.1029
Coor	360	0.7712	0.1677	0	0.9893
DC	360	0.1837	0.1294	0.0163	0.7682
ISU	360	1.5281	0.8273	0.6879	5.8809
GLP	360	0.4114	0.5603	0.0110	4.4773
Fis	360	25.1210	10.5569	10.4928	73.7555
ER	360	0.0925	0.1150	0.0007	1.0848
Open	360	24.3190	22.9054	0.6960	124.4513
Ed	360	16.1055	2.7272	9.5452	22.5861

. For dependent variables, standard deviations of lnCO₂ and lnSO₂ are relatively large, indicating that there are significant regional differences in carbon emissions and pollution emissions across provinces. The mean Coor value is relatively high, indicating that the synergistic effect of pollution and carbon reduction is inefficient, but the inter-regional gap is relatively small. Regarding the explanatory variable, the mean value of DC is relatively small, revealing that the level of digital consumption across provinces is still in the cultivation stage and has considerable space for improvement. Regarding the mediating variables and control variables, the standard deviations of these variables are relatively large, suggesting that there are significant regional differences in the level of economic and social development. The variance inflation factor (VIF) was used to test for multicollinearity among variables. The maximum VIF was 5.50 and the minimum VIF was 1.36, which are all below the critical value of 10, indicating that there is no multicollinearity problem.

4. Results

4.1. Benchmark Results

Table 3. Benchmark regression results.

Variables	(1) lnCO2	(2) lnCO2	(3) lnSO2	(4) lnSO2	(5) Coor	(6) Coor
DC	−0.9085 ** (0.3496)	−0.9937 ** (0.4123)	−6.2241 *** (1.0160)	−6.5310 *** (1.0477)	−0.9956 ** (0.4229)	−1.1301 ** (0.5094)
Fis		0.0002 (0.0035)		−0.0015 (0.0140)		0.0008 (0.0019)
ER		−0.0037 *** (0.0009)		0.0010 (0.0014)		−0.0001 (0.0002)
Open		−0.0026 (0.0020)		0.0009 (0.0083)		−0.0012 (0.0013)
Ed		0.0143 ** (0.0061)		0.0714 ** (0.0276)		0.0076 * (0.0041)
Constant	5.6342 *** (0.0242)	5.5366 *** (0.1647)	4.4098 *** (0.0998)	3.2129 *** (0.6324)	0.5743 *** (0.0121)	0.7812 *** (0.0751)
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	360	360	360	360	360	360
R2	0.4291	0.5404	0.9392	0.9437	0.6084	0.6240

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

reports the results of the benchmark regression. Regardless of whether control variables are included, digital consumption helps reduce carbon emissions and pollution emissions and promotes the synergistic effect of pollution and carbon emission reduction at the 5% significance level. After controlling for all variables, for every 1 percentage point increase in digital consumption level, carbon emissions and pollution emissions decrease by 0.9937 and 6.5310 percentage points, while the synergistic effect of pollution and carbon reduction increases by 1.1301 percentage points. This indicates that digital consumption has a significant synergistic effect on pollution and carbon reduction, validating Hypothesis 1. The sustainable development of digital consumption can gradually penetrate the entire chain of production, distribution, and consumption. It achieves the synergistic effect of pollution and carbon reduction by effectively reducing energy and material consumption, improving resource allocation efficiency, precisely matching supply and demand, optimizing intelligent logistics, strengthening green supply chain management, and accelerating the development of new business forms and models.

4.2. Endogeneity and Robustness Tests

4.2.1. Endogeneity Tests

As endogeneity issues can lead to biased results, the instrumental variable method was used for endogeneity testing. First, following the approach of He et al. [25], the interaction term between national mobile internet users in the previous year and the number of telephones per 10,000 people in each province in 1984 was used as the instrumental variable for digital consumption. The test results are shown in columns (1) to (3) of

Table 4. Results of endogeneity tests.

Variables	(1) lnCO2	(2) lnSO2	(3) Coor	(4) lnCO2	(5) lnSO2	(6) Coor
DC	−1.6918 *** (0.1839)	−10.3036 *** (1.4632)	−2.1433 *** (0.2443)	−1.0216 *** (0.2000)	−6.0442 *** (0.8450)	−1.1756 *** (0.2284)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
F-value of the first stage	169.250			2569.828
K-P rk LM	13.497 ***			40.427 ***
Cragg–Donald Wald F	173.078			4358.573
N	336	336	336	330	330	330
R2	0.5214	0.9334	0.5257	0.5394	0.9332	0.6898

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. Second, the one-period lag in digital consumption was taken as another instrumental variable. The test results are shown in columns (4) to (6) of Table 4. The F-value of the first-stage estimations are all significantly higher than the rule-of-thumb value of 10 and the 10% critical value (16.38), satisfying the relevance condition. The under-identification test (K-P rk LM) and the weak instrumental variable test (Cragg–Donald Wald F) reject the null hypotheses of insufficient identification of instrumental variables and weak instrumental variables, validating the effectiveness of the instrumental variables. After addressing the endogeneity issue, digital consumption still has a significant synergistic effect on pollution and carbon reduction, further confirming the reliability of benchmark regression results.

4.2.2. Robustness Tests

This study employed four methods for robustness testing. First, the natural logarithm of per capita carbon dioxide emissions and chemical oxygen demand emissions was used as substitute variables for carbon emissions and pollution emissions, and then regression was conducted again. The test results are shown in columns (1) to (3) of

Table 5. Results of robustness tests.

Variables	Replacing the Dependent Variables			Replacing the Explanatory Variable
	(1) lnCO2	(2) lnSO2	(3) Coor	(4)ln CO2	(5) lnSO2	(6) Coor
DC	−0.1154 *** (0.0356)	−4.8288 *** (0.8422)	−0.6838 *** (0.1722)	−1.2670 ** (0.4955)	−8.9150 *** (1.5682)	−1.2552 ** (0.5830)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	360	360	360	360	360	360
R2	0.5581	0.8818	0.6122	0.5404	0.9477	0.5717

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. Second, the equal weight weighting method was used to recalculate digital consumption levels, and then regression was conducted again. The test results are shown in columns (4) to (6) of Table 5. Third, to avoid the influence of outliers on the regression results, all variables were winsorized at the 1st and 99th percentiles, and then regression was conducted again. The test results are shown in columns (1) to (3) of

Table 6. Results of robustness tests.

Variables	Winsorization			Lagged Variables
	(1) lnCO2	(2) lnSO2	(3) Coor	(4) lnCO2	(5) lnSO2	(6) Coor
DC	−0.9786 * (0.4966)	−6.7158 *** (1.2665)	−1.0080 ** (0.4774)
L.DC				−0.9487 ** (0.4323)	−5.6855 *** (0.9858)	−1.1404 ** (0.5454)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	360	360	360	330	330	330
R2	0.5192	0.9445	0.5585	0.5224	0.9335	0.6617

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. Fourth, considering the time lag effect of digital consumption on the synergistic effect of pollution and carbon reduction, regression was re-estimated using a one-period lag of digital consumption. The test results are shown in columns (4) to (6) of Table 6. The results indicate that digital consumption still has a significant synergistic effect on pollution and carbon reduction, confirming the robustness of the benchmark regression results.

4.3. Heterogeneity Analysis

4.3.1. Heterogeneity in the Digital Policy Environment

To investigate whether there is heterogeneity in the digital policy environment regarding the impact of digital consumption on the synergistic effect of pollution and carbon reduction, this study used the mean number of “Broadband China” pilot cities within provinces as the dividing line. All samples were further divided into two groups: regions with supportive digital policy environments and regions with general digital policy environments. Group regressions were conducted. The results are shown in columns (1) to (6) of

Table 7. Results of heterogeneity in digital policy environments.

Variables	Supportive Digital Policy Environment			General Digital Policy Environment
	(1) lnCO2	(2) lnSO2	(3) Coor	(4) lnCO2	(5) lnSO2	(6) Coor
DC	−1.3057 *** (0.2571)	−7.0963 *** (1.0137)	−1.5495 ** (0.5856)	−0.1610 (0.5526)	−4.3552 *** (0.8365)	−0.1057 (0.2074)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	192	192	192	168	168	168
R2	0.5420	0.9428	0.7769	0.6129	0.9622	0.3572

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. It can be observed that in regions with supportive digital policy environments, the synergistic effect of digital consumption on pollution and carbon reduction is stronger. Regions with supportive digital policy environments have more developed digital infrastructure, which is more conducive to absorbing and retaining digital talent. The deeper application of digital technology provides an important technological foundation for the vigorous development of digital consumption, further facilitating the realization of the synergistic effect of digital consumption on pollution and carbon reduction.

4.3.2. Heterogeneity in Carbon Control Policy

To investigate whether there is heterogeneity in carbon control policy regarding the impact of digital consumption on the synergistic effect of pollution and carbon reduction, this study classified provinces implementing the carbon emission trading pilot as regions with favorable carbon control policy. The remaining provinces were regarded as regions with average carbon control policy. Group regressions were conducted, and the results are shown in columns (1) to (6) of

Table 8. Results of heterogeneity in carbon control policy.

Variables	Favorable Carbon Control Policy			Average Carbon Control Policy
	(1) lnCO2	(2) lnSO2	(3) Coor	(4)ln CO2	(5) lnSO2	(6) Coor
DC	−1.3268 *** (0.2699)	−8.2171 *** (0.6975)	−1.9929 ** (0.6674)	−0.1155 (0.4723)	−4.1560 *** (1.0573)	−0.2261 *** (0.0730)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	84	84	84	276	276	276
R2	0.7237	0.9582	0.7225	0.5706	0.9489	0.8395

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. In regions with favorable carbon control policy, the synergistic effect of digital consumption on pollution and carbon reduction is stronger. Regions implementing carbon emission trading pilots have institutional and policy advantages. Through strict regulation, cost constraints, and profit incentives, these regions have driven enterprises to shift from passive emission reduction to proactive carbon abatement and can force enterprises to carry out technological innovation and production optimization, driving the transformation of the industrial structure towards low energy consumption and high added value. Therefore, against this backdrop, the carbon emission reduction potential on the demand side unleashed by digital consumption can better facilitate the synergistic effect of pollution and carbon reduction through carbon regulation policies.

4.3.3. Heterogeneity in Industrialization Level

To investigate whether there is heterogeneity in industrialization level regarding the impact of digital consumption on the synergistic effect of pollution and carbon reduction, this study used the mean proportion of industrial added value to GDP as the dividing line. All samples were further divided into two groups, regions with higher industrialization level and regions with lower industrialization level, and group regressions were conducted. The results are shown in columns (1) to (6) of

Table 9. Results of heterogeneity in industrialization level.

Variables	Higher Industrialization Level			Lower Industrialization Level
	(1) lnCO2	(2) lnSO2	(3) Coor	(4) lnCO2	(5) lnSO2	(6) Coor
DC	−0.5839 (0.5902)	−4.2523 *** (1.3048)	−0.3503 ** (0.1435)	−1.3300 ** (0.5554)	−5.9497 *** (0.6244)	−1.5488 ** (0.5796)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	204	204	204	156	156	156
R2	0.7127	0.9609	0.8788	0.4012	0.9498	0.6648

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. In regions with lower industrialization levels, the synergistic effect of digital consumption on pollution and carbon reduction is stronger. Regions with higher industrialization levels often exhibit path dependence on traditional production methods and development models, have relatively insufficient innovation capacity, feature relatively monolithic industrial structures, and are subject to high-carbon infrastructure and industrial lock-in effects, resulting in a slower pace of industrial transformation and upgrading. In contrast, regions with lower industrialization levels typically have more rational and advanced industrial structures and stronger innovation capacity, which is more conducive to fully leveraging the synergistic effect of digital consumption on pollution and carbon reduction.

4.3.4. Heterogeneity in the Level of Environmental Protection

To investigate whether there is heterogeneity in the level of environmental protection and the impact of digital consumption on the synergistic effect of pollution and carbon reduction, this study used the mean proportion of environmental protection expenditure to fiscal expenditure as the dividing line. All samples were further divided into two groups, regions with higher levels of environmental protection and regions with lower levels of environmental protection, and group regressions were conducted. The results are shown in columns (1) to (6) of

Table 10. Results of heterogeneity in the level of environmental protection.

Variables	Higher Level of Environmental Protection			Lower Level of Environmental Protection
	(1) lnCO2	(2) lnSO2	(3) Coor	(4) lnCO2	(5) lnSO2	(6) Coor
DC	−1.6231 *** (0.3656)	−8.1563 *** (0.8973)	−2.1043 *** (0.2737)	−0.3895 (0.3585)	−4.1831 *** (1.1086)	−0.4158 *** (0.1146)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	144	144	144	216	216	216
R2	0.6355	0.9492	0.8540	0.4867	0.9514	0.6125

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. In regions with higher levels of environmental protection, the synergistic effect of digital consumption on pollution and carbon reduction is stronger. Regions with higher levels of environmental protection are usually equipped with more complete environmental information disclosure systems, stricter pollutant discharge standards, and stronger law enforcement. These institutional conditions can amplify the environmental spillover effect of digital consumption. Moreover, the public’s green environmental preference forces enterprises to achieve environmental protection goals through green technological innovation and digital means from the demand side, which is more conducive to fully leveraging the synergistic effect of digital consumption on pollution and carbon reduction.

4.4. Mechanism Analysis

Table 11. Mediation effect tests.

Variables	(1) ISU	(2) ISU	(3) GLP	(4) GLP
DC	2.2077 ** (0.7827)	2.0952 ** (1.3770)	4.5803 *** (1.7500)	4.2708 ** (1.7500)
Control variables	No	Yes	No	Yes
Province and year fixed effects	Yes	Yes	Yes	Yes
N	360	360	360	360
R2	0.6751	0.7397	0.7514	0.7565

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

reports the results of the mediation effect tests. Columns (1) and (2) present the results for the mechanism of industrial structure upgrading. Regardless of whether control variables are included, digital consumption can significantly promote industrial structure upgrading, validating Hypothesis 2. Digital consumption drives the supply side to break away from path dependence on high-energy-consumption, high-emission, and resource-based industries through the demand side and promotes the optimization and efficient utilization of resources, promoting the synergistic effect of pollution and carbon reduction. There exists a transmission mechanism of “digital consumption–industrial structure upgrading–the synergistic effect of pollution and carbon reduction.”

Columns (3) and (4) present the results for the mechanism of green technology innovation. Regardless of whether control variables are included, digital consumption can significantly promote green technology innovation, validating Hypothesis 3. Digital consumption expands the market space of green products, effectively reduces the marginal cost of green technology research and application, and creates new application scenarios and market space for green technology, stimulating endogenous momentum for green technology innovation and improving the level of green technology innovation, ultimately achieving the synergistic effect of pollution and carbon reduction. There exists a transmission mechanism of “digital consumption–green technology innovation–the synergistic effect of pollution and carbon reduction.”

4.5. Threshold Effect Analysis

We further employed the threshold model to examine the threshold effects of digital consumption on the synergistic effect of pollution and carbon reduction. The results of the threshold effect test are shown in

Table 12. Threshold effect test.

Variables	Model	F-Value	p-Value	Critical Value
				1%	5%	10%
ISU	Single threshold	323.45	0.0000	85.0670	47.9290	35.5355
	Double threshold	9.57	0.8740	1300.0000	893.8864	598.4503
	Triple threshold	7.13	0.7100	918.5957	549.8789	340.9593
GLP	Single threshold	201.70	0.0000	71.2677	34.9942	27.6618
	Double threshold	7.42	0.3960	541.0978	123.2407	14.5493
	Triple threshold	5.59	0.8000	982.7952	380.6380	275.3503

. At the 10% significance level, both industrial structure upgrading and green technology innovation pass the single-threshold test. This indicates that the promoting effect of digital consumption on the synergistic effect of pollution and carbon reduction does not follow a simple linear relationship. Instead, it is influenced by the level of industrial development and green technology innovation, demonstrating significant threshold effects.

Based on the results of the threshold effect test, the regression results of digital consumption on the synergistic effect of pollution and carbon reduction are shown in

Table 13. Threshold model regression results.

Variables	(1) Coor	(2) Coor
DT · I(ISU ≤ 3.6253)	−0.4549 *** (0.1213)
DT · I(ISU > 3.6253)	−1.4982 *** (0.0598)
DT · I(GLP ≤ 2.0374)		−0.6107 *** (0.2034)
DT · I(GLP > 2.0374)		−1.0761 *** (0.1563)
Control variables	Yes	Yes
Province and year fixed effects	Yes	Yes
N	360	360
R2	0.8051	0.7620

Note: ***, **, and * indicate significance at the level of 1%, 5%, and 10%. The values in parentheses are robust standard errors.

. Column (1) of Table 13 presents the results of the threshold effect for industrial structure upgrading. When the level of industrial development is below 3.6253, the coefficient of digital consumption is −0.4549, and when the level of industrial development exceeds 3.6253, the coefficient of digital consumption is −1.4982. This indicates that with the continuous improvement of industrial development levels, the effect of digital consumption on pollution and carbon reduction is strengthened. When the level of industrial development is low, the industrial structure is dominated by traditional extensive manufacturing and resource-based industries, with a low degree of modernization. The penetration and application scenarios of digital consumption are limited, and the driving force of demand is insufficient, resulting in weaker synergistic effects of pollution and carbon reduction. When the level of industrial development crosses the threshold, the industrial system gradually becomes more sophisticated, the proportion of high-tech industries, green and low-carbon industries, and modern service industries continues to rise, and the resilience and greening level of industrial and supply chains are significantly enhanced. At this stage, digital consumption can penetrate the entire chain of production, distribution, and consumption. By guiding demand structure upgrades, driving supply-side optimization, and improving resource allocation efficiency, it significantly amplifies the synergistic effect of pollution and carbon reduction.

Column (2) of Table 13 presents the results of the threshold effect for green technology innovation. When the level of green technology innovation is below 2.0374, the coefficient for digital consumption is −0.6107; when the level of green technology innovation exceeds 2.0374, the coefficient of digital consumption is −1.0761. This indicates that as the level of green technology innovation continues to rise, the synergistic effect of digital consumption on pollution and carbon reduction is significantly strengthened. When green technology innovation is at a relatively low level, the depth and efficiency of green technology embedded in various aspects of digital consumption are insufficient, which limits the synergistic effect of pollution and carbon reduction. When the level of green technology innovation crosses the threshold, technologies such as big data and artificial intelligence are deeply integrated with green processes, enabling digital consumption to significantly enhance the synergistic effect of pollution and carbon reduction by matching supply and demand, optimizing intelligent logistics, and promoting green products.

5. Conclusions and Policy Implications

5.1. Conclusions

Based on provincial-level panel data from China covering 2013 to 2024, this study empirically examines the synergistic effect of digital consumption on pollution and carbon reduction. It also examines heterogeneity in digital policy environments, the level of rule of law and industrialization level, and the mediating and threshold effects of industrial structure upgrading and green technological innovation. First, digital consumption has a significant synergistic effect on pollution and carbon reduction. For every 1 percentage point increase in digital consumption level, the synergistic effect of pollution and carbon reduction increases by 1.1301 percentage points. Second, the impact of digital consumption on the synergistic effect of pollution and carbon reduction exhibits heterogeneity. In regions with a supportive digital policy environment, favorable carbon control policy, lower degrees of industrialization, and higher level of environmental protection, the synergistic effect of digital consumption on pollution and carbon reduction is stronger. Third, the transmission mechanisms of “digital consumption–industrial structure upgrading–the synergistic effect of pollution and carbon reduction” and “digital consumption–green technology innovation–the synergistic effect of pollution and carbon reduction” are validated, and the promoting effect of digital consumption on the synergistic effect of pollution and carbon reduction demonstrates nonlinearity with increasing marginal effects.

5.2. Policy Implications

Based on the above conclusions, the following policy implications are proposed:

(1)

Broaden the application scenarios of digital consumption and stimulate its endogenous momentum. First, accelerate the construction of digital infrastructure and increase financial support and policy inclinations for underdeveloped and rural areas to promote the balanced development of digital infrastructure construction, enhancing the breadth and depth of network coverage for digital consumption scenarios. Second, focus on promoting the deep integration of digital technologies with traditional consumption models; actively carry out digital consumption expansion and quality improvement actions; fully utilize advanced information technologies to innovate in online and offline integrated consumption scenarios; and create immersive and experiential consumption spaces. Vigorously develop new forms of service consumption to build a diversified digital consumption ecosystem; and enhance the digitalization level of service consumption.

(2)

Promote the synergistic effect of digital consumption on pollution and carbon reduction across all stages of production, circulation, and consumption. First, accelerate the establishment of digital consumption platforms and green data centers to help enterprises utilize consumer big data, user profiles, and platform demand forecasts to achieve flexible customization and on-demand production, reducing resource waste and improving resource utilization efficiency. Support enterprises in building green digital factories; and provide fiscal interest subsidies, tax incentives, and special subsidies to enterprises that adopt energy-saving processes, low-carbon materials, and circular technologies. Second, establish a unified national carbon inclusive platform and green consumption credit system, expand the coverage of green consumption vouchers and subsidies, reduce value-added tax and consumption tax on green digital products and low-carbon digital services, and guide consumers to purchase green products and use low-carbon services to foster green consumption preferences and behavioral patterns. Explore the establishment of a green information platform for digital products to achieve standardized disclosure and convenient access to information on product carbon footprints, environmental labels, energy efficiency ratings, and other relevant information.

(3)

Implement differentiated regional development policies. First, regions with supportive digital policy environments should establish demonstration models and innovative green consumption and regulatory models, while regions with general digital policy environments should prioritize consolidating digital infrastructure and improving digital supporting facilities. Second, regions with favorable carbon control policy should strengthen the transmission of carbon price signals, incorporate digital consumption into the carbon quota accounting system, and leverage digital consumption to advance precision pollution control and tap carbon reduction potential, while regions with average carbon control policy should accelerate the improvement of carbon constraint rules, expand the coverage of carbon-inclusive systems, and guide the green transformation of digital business formats. Third, regions with higher industrialization levels should deepen the integration of digital consumption and green manufacturing based on industrial foundations, promoting digital and green transformation of traditional industries, while regions with lower industrialization levels should actively cultivate green emerging industries and prioritize the development of low-carbon digital service consumption. Finally, regions with higher levels of environmental protection should leverage institutional advantages to explore innovative models for deep integration of digital consumption and environmental protection, while regions with lower levels of environmental protection should first strengthen foundational environmental regulation and capacity building.

(4)

Leverage digital consumption to drive industrial structure upgrading and green technological innovation. First, leverage the demand-driven role of digital consumption; accelerate the construction of a digital green collaborative industrial ecosystem; and promote the transfer of labor, capital, technology and other factors from high-pollution, hig-energy-consumption, and low-efficiency traditional industries to low-carbon and high-efficiency industries, such as digital services, green manufacturing, and high-tech industries. Implement the special action of digital energy conservation and carbon reduction transformation, use industrial internet and big data to achieve real-time control of energy consumption and emissions, and upgrade industry from extensive to intensive. Second, establish a linkage mechanism between the green demand for digital consumption and the direction of green technology research and development; set up a special fund for green technology research driven by digital consumption; encourage leading e-commerce platforms, payment platforms, and life service platforms to collaborate with universities and research institutes in establishing green technology innovation consortia to carry out collaborative innovation targeting pollution and carbon reduction needs in the digital consumption sector; and provide tax incentives and supportive policies, such as additional tax deductions for research and development expenses, payroll tax reductions, and research and development subsidies to promote research and application of green and low-carbon technologies.

5.3. Limitations and Prospects

Although this study systematically examines the direct effects, heterogeneity, and mechanisms of digital consumption on the synergistic effect of pollution and carbon reduction, limitations remain. First, the measurement of the synergistic effect of pollution and carbon reduction mainly relies on the coupling coordination degree model and the data envelopment analysis model, and both methods have advantages. If the DEA model could be adopted for efficiency measurement and subsequent regression analysis is conducted, the conclusions would be more reliable. Second, the sustainable effects of digital consumption largely depend on increasingly stricter environmental and carbon policies. This study primarily conducts a heterogeneity analysis from the perspectives of digital policy environments, carbon control policy, degree of industrialization, and environmental protection. However, there is insufficient research on the interaction mechanism between environmental and carbon control policies and digital consumption in enhancing the synergistic effect of pollution and carbon reduction. Third, this study verified the mechanisms of digital consumption on the synergistic effect of pollution and carbon reduction from the two pathways of industrial structure upgrading and green technology innovation, while other mechanisms, such as market demand and green total factor productivity, are not further explored.