Introduction

A new wave of industrial revolution has emerged since the 21st century, alongside the advancement of modern information technology. Green development in manufacturing has become a prevailing global trend and direction, as highlighted by the international community (Henao et al., 2024). Strategies such as Germany’s Industry 4.0, the U.S. “Reindustrialization,” and China’s “Made in China 2025” have emerged in response. As the world’s leading manufacturing nation, China, under the current backdrop of ecological civilization construction and green development, is actively accelerating the green transformation of its manufacturing. This transformation is creating historic new opportunities for manufacturing enterprises and making significant contributions to protecting the Earth, our shared home. The green transformation and upgrading of the manufacturing industry (GTUM) is manifested not only in the shift from low-end to high-end industrial development but, more crucially, through cutting back on energy use and emissions of pollutants, thereby enhancing the overall sustainability of the manufacturing industry, and thus realizing green manufacturing. As an important part of the “Made in China 2025” strategy, green manufacturing is a modern manufacturing model with low consumption, low emissions, and high efficiency. It is a crucial basis and marker for assessing the effectiveness of the manufacturing sector’s green transformation. Green manufacturing is essentially about embedding green development concepts and management requirements throughout the entire life cycle, considering factors like industrial structure, energy resources, and ecological environment throughout the entire development process of the industry. This deep transformation of manufacturing modes drives traditional industries towards green transformation. Furthermore, the Chinese government actively advocates for the acceleration of the development of a manufacturing powerhouse and a digital China, aiming to promote the transformation of the manufacturing sector towards higher-end, smarter, and greener development. In this process, the key lies in fostering a deep integration of the digital economy (DEC) along with the physical economy, injecting new vitality into GTUM, and delineating a clear direction for high-quality development of manufacturing.

Amidst the latest wave of technological revolution and industrial transformation, DEC acts as a central force in the new cycle of global industrial innovation and development. It has introduced a series of innovations in the organizational, production, and innovation models of traditional manufacturing, playing a superimposed role in promoting the manufacturing sector’s high-quality development. Specifically, digital technologies represented by artificial intelligence and the Internet of Things bring substantial opportunities for greening product design, production, and supply chains in manufacturing. These technologies help optimize production processes, enhance efficiency, and reduce energy consumption and carbon emissions (Wang et al., 2020). However, the development of DEC and the realization of GTUM require specific market conditions. China is rapidly constructing a national unified large market (ULM) that is fully open, efficient standardized, fair competitive. This will facilitate better connectivity between domestic and international markets, and further unleash the market dividends of a large country. The construction of China’s ULM amplifies the scale economic effects of DEC, accelerates the diffusion of digital technologies, and increases innovation gains. Consequently, resource allocation and energy conservation are enhanced, and emissions are reduced, acting as an “accelerator” for GTUM (Bian et al., 2019; Liang et al., 2024). Further research into the relationship between these two aspects should also consider the impact of constructing ULM. The present study offers a novel pathway for China to leverage DEC and the advantages of an ultra-large scale market to enhance the level of GTUM. It also provides valuable insights for the green development of industries globally.

Scholarly study on DEC and GTUM has attracted much attention from scholars, focusing primarily on two aspects. First, the impact of DEC on industrial structure of traditional manufacturing has been a key study area. Some scholars have identified a restructuring effect of DEC on manufacturing, which significantly facilitates the upgrading of traditional manufacturing by optimizing value, innovation, supply, and service chains (Zhao et al., 2023). By integrating digital technologies, the manufacturing sector can automate and smarten production processes, thus enhancing production efficiency and reducing costs, further optimizing the industrial structure of manufacturing (Zhou et al., 2018). Furthermore, digital finance contributes to the service-oriented transformation of manufacturing by boosting enterprise innovation intensity and elevating the level of digitalization (Chen and Zhang, 2021). Second, DEC influences the green development of manufacturing. Given the increasing environmental pollution from industrial production, harmonizing the relationship between industrialization and ecological sustainability has become crucial. Information technology effectively transforms traditional manufacturing (Bai et al., 2024; Yuan et al., 2020). Integrating digital technologies into green innovation and production processes can significantly enhance manufacturing firms’ green innovative performance and competitiveness (Lin and Xie, 2024). Digital transformation promotes green manufacturing development by driving innovation in green manufacturing technologies and reducing energy consumption and pollution (Li et al., 2023). Moreover, DEC facilitates green manufacturing development through channels such as human capital, digital innovation, and industrial upgrading. Furthermore, DEC has a spatial spillover effect on manufacturing’s green development (Liu et al., 2023). These studies provide valuable references and inspiration for this study.

Existing literature primarily focuses on the influence of DEC on the service-oriented upgrading of manufacturing or its environmental effects. However, there is relatively little research concentrating on green manufacturing. Few studies have examined this issue from a market perspective, connecting DEC development with the green transformation and GTUM. Therefore, from China’s unified large market construction perspective, this study investigates the effects of DEC on GTUM and its transmission mechanisms. The potential contributions of this research are manifold:

Research perspective. It explores the relationship between DEC and GTUM from the external environment of the ULM construction, offering a new perspective for studies on how DEC impacts industrial structure upgrading; Research content. This study posits that green manufacturing represents the direction and the result of GTUM. It defined “GTUM” as a synthesis of the levels of industrial structure upgrading and greening in manufacturing. The study further explores the underlying logic of how DEC promotes GTUM through three channels such as market segmentation, market potential, and market supply and demand, thereby enriching the existing literature; Research methodology. This study employs traditional econometric models, integrates geographic coordinate systems, and leverages advanced technologies like Python machine learning to enrich the research methods concerning the relationship between DEC and GTUM and their mechanisms.

Theoretical analysis and research hypotheses

DEC and GTUM

With the continuous development of a new generation of digital technologies, traditional manufacturing industries are increasingly adopting emerging digital technologies, and DEC has become a new driving force for GTUM (Liu et al., 2023; Wang et al., 2024b). First, digital technology and the Internet have enabled the manufacturing industry to digitize production. Real-time monitoring and management of energy usage and pollutant emissions during production is made possible by technologies like big data analytics, the Internet of Things, and ecological innovation (Henao et al.,. 2024). These technologies support the green transformation of the whole manufacturing chain and help to optimize production processes. Second, DEC offers robust data analysis and forecasting capabilities, enabling manufacturing firms to understand better and optimize resource efficiency (Kristoffersen et al., 2020). Through data analysis, companies can identify and improve segments of high energy consumption, optimize production plans, and reduce waste generation. Data analysis reduces resource wastage and promotes pollution control and end-of-pipe emissions reduction. Furthermore, DEC facilitates the development of green supply chain management (Wang et al., 2024). Digital supply chain management allows companies to choose and collaborate with suppliers that are green-certified and sustainably managed, thus driving the green transformation of the entire supply chain (Feng et al., 2022). Lastly, DEC provides a platform for innovation and collaborative cooperation in GTUM. The application of digital technologies enhances the convenience of research and development (R&D) in green technologies, creating more environmentally friendly and sustainable products and production methods (Zheng et al., 2023). Additionally, DEC promotes collaborative cooperation among companies, governments, and research institutions, accelerating GTUM (Dahlman et al., 2016).

Hypothesis 1: DEC promotes GTUM.

DEC, ULM construction, and GTUM

The development of DEC and GTUM is closely linked to ULM construction, which provides significant support and momentum. According to the “Opinions of the CPC Central Committee and the State Council on Accelerating the Construction of a National Unified Large Market,” accelerating the establishment of a fully open, efficient standardized, and fair competitive national unified market is beneficial for transforming China’s market from a large to a strong one. Orienting toward a large domestic market to restructure the manufacturing industry chain promotes high-quality manufacturing development, and enhances the “dual circulation” of domestic and international markets (Guo and Tian, 2021). In this context, exploring how DEC influences GTUM necessitates a consideration of the impact stemming from ULM construction. The national unified large market is characterized by its market unity, immense scale, and robust functionality (Liu and Kong, 2021). Moreover, the opposite of market unity is market segmentation. Building a unified national market lies in breaking through market segmentation and promoting the efficient movement of resources and commodity elements over a broader area. The immense scale refers to the market’s super-scale advantage capacity. Accelerating the construction of a national unified market helps unleash the potential of the strong domestic market and fully leverages the advantages of its large scale. Robust functionality means the market efficiently transmits signals, adjusts supply-demand matching, and guides the effective distribution and optimal linkage of resources, goods, and factors (Guo and Tian, 2021). Therefore, based on Liu and Kong (2021) and market economic theory, ULM construction can be considered from three aspects: market segmentation, market potential, and market supply and demand.

Reducing market segmentation effects

DEC is a new economic form primarily carried by modern information networks endowed with inherent advantages such as cross-temporal and spatial information distribution, openness, and sharing. First, the spillover effects arising from the diffusion of digital technologies and network externalities can break the limitations of time and space, enabling the free flow and efficient allocation of production factors, and reducing market segmentation (Wan et al., 2023). Second, based on the theory of boundlessness, the development of digital technologies and the Internet has dismantled the constraints of traditional markets. Such development blurs the boundaries of traditional industries, propelling them toward digitalization, networking, and intelligent transformation and expanding the effective boundaries of the market. Lastly, the widespread application of digital technologies and the resulting information integration effects facilitate interregional communication and collaboration, minimizing informational friction. In turn, such application enhances the diffusion and reach of digital technologies, significantly strengthening economic ties among regions and promoting market integration (Wang et al., 2023). Furthermore, market segmentation can lead to market fragmentation for production factors, impeding the free movement of capital and technology across geographical boundaries. Consequently, it constrains industrial structural upgrading and the achievement of pollution reduction, ultimately reducing the efficiency of resource allocation (Ren et al., 2021a). Efficient resource utilization encourages cost reduction, technical innovation, and waste minimization, enhancing the supply chain sustainability and leading the manufacturing industry toward greener, more sustainable practices (Zheng et al., 2023; Zhang et al., 2024). Moreover, market segmentation weakens regional division of labor and collaboration, resulting in diseconomies of scale that worsen environmental pollution and increase challenges associated with GTUM (Bai et al., 2004).

Hypothesis 2a: DEC promotes GTUM by reducing market segmentation effects.

Stimulating market potential effects

First, according to the theory of economies of scale, DEC utilizes the advantages of digital technologies and the Internet to make production processes more precise and controllable, facilitating resource integration and sharing. Consequently, businesses can adjust and expand their production scales more efficiently, thereby reducing average costs, enhancing the benefits of economies of scale, and stimulating market potential. Second, based on Metcalfe’s Law, DEC has established convenient online trading platforms that eliminate geographical barriers to transactions, collect, analyze, and utilize massive amounts of consumer data and accurately capture market demands. This approach allows for the discovery of a broader base of potential customers, unleashing market potential (Rabinovich and Bailey, 2004). Finally, according to the theory of industrial agglomeration, digital technologies encourage the clustering of related industries, enhancing inter-firm information exchange and technology sharing and fostering knowledge spillovers and technological innovation, thus increasing production efficiency (Chang et al., 2023). Consequently, more products and services are provided, market size expands, and market potential is stimulated. When the potential for market demand is high, it reduces the cost of product R&D per unit, encouraging enterprises to invest in production equipment and technological improvements and leading to greater agglomeration effects and more intense market competition. Additionally, expanding market potential through the advantages of a large domestic market can reduce business operational and transportation costs. This enhances labor productivity and facilitates technological innovation and improves energy efficiency, thereby accelerating the MGTU process in manufacturing.

Hypothesis 2b: DEC promotes GTUM by stimulating market potential effects.

Matching market supply and demand effects

The inclusive characteristics of DEC effectively promote rapid growth and dynamic matching of supply and demand. From the demand side, the widespread adoption of digital terminals and network externalities have increasingly amplified the impact of e-commerce platforms serving business users and individual consumers, thereby stimulating diversified consumer demand. Utilizing big data to track consumer behavior in real time allows for a scientific and rational analysis of consumer preferences, emphasizing personalized and customized production features (Ma et al., 2021). From the supply side, the application of digital technologies has increased enterprise output and enabled firms to engage in more diversified activities at lower costs, stimulating scope and scale economic effects and providing strong support for flexible production (Cullen and Farronato, 2021). Therefore, the emergence of new technologies has increased the effectiveness of information, offering optimized pathways for addressing matching issues between supply and demand in the market. According to Moore’s Law and Gilder’s Law, the marginal costs of DEC decrease in the long term. The application of digital technologies and increased transparency in information significantly exceed the costs borne by the supply and demand sides for DEC. It facilitates market supply-demand matching, allocating more funds toward R&D, energy conservation, and emission reduction (Qinqin et al., 2023) and thus driving GTMU. Moreover, market mechanisms enable rational pricing and the free flow of production factors such as green technologies (Wang et al., 2021). As the demand for green products continually increases, enterprises introduce advanced production equipment and improve green production technologies. The accessibility of green technology can enhance resource utilization efficiency and green development capabilities (Williams, 2011), thereby promoting GTMU.

Hypothesis 2c: DEC promotes GTUM by matching market supply and demand effects.

Non-linear impact of DEC on GTUM

From temporal dynamic changes of this perspective, the development of DEC in the early stage requires a large amount of energy-intensive infrastructure to serve as the basis for its operation and application due to the large investment and long cycle of development, which causes energy consumption (Ren et al., 2021b). However, as DEC rapidly evolves and increasingly integrates with traditional industries, it brings about economies of scale, enhances energy efficiency, reduces production costs, and strengthens the effectiveness of energy conservation and emission reduction (Liu et al., 2023). Therefore, DEC exhibits non-linear effects, which align with Metcalfe’s Law: as digital platforms and online social networks grow, the number of users increases rapidly, significantly enhancing network value, attracting more users, and creating a virtuous cycle where the value generated by DEC grows exponentially (Pouri and Hilty, 2021). Yu et al., (2024) also pointed out that DEC exhibits a “U-shaped” pattern for local carbon emission efficiency and an inverted “U-shaped” pattern for neighboring regions. Zhao and Guo (2023) explore the non-linear relationship between DEC and energy intensity, suggesting that this effect will significantly reduce energy intensity as urbanization accelerates and industrial structures upgrade. Moreover, the manufacturing industry can establish greener value chains through digital platforms and the sharing economy model, achieving resource sharing and recycling. This innovative business model may bring non-linear impacts, facilitating resource integration and efficient allocation, thus promoting a low-carbon production approach and further driving GTUM (Dwivedi and Paul, 2022).

Hypothesis 3: The impact of DEC on GTUM exhibits non-linear effects (Fig. 1).

Fig. 1
figure 1

Mechanism diagram of the impact of DEC on GTUM.

Full size image

Research design

Model specification

Panel benchmark regression model

$$GTU{M}_{it}={\alpha }_{0}+{\alpha }_{1}DE{C}_{it}+{\alpha }_{2}{X}_{it}+{u}_{i}+{\delta }_{t}+{\varepsilon }_{it},$$
(1)

In Eq. (1), DECit and GTUMit represent DEC and GTUM levels in province i and time t, respectively. Xit represents a series of control variables. μi denotes province-fixed effects, δt represents time-fixed effects, and εit denotes the random error term.

Furthermore, to explore the mechanisms through which DEC influences GTUM, this study introduces a series of mechanism variables and constructs the following model:

$${M}_{it}={\beta }_{0}+{\beta }_{1}DE{C}_{it}+{\beta }_{2}{X}_{it}+{u}_{i}+{\delta }_{t}+{\varepsilon }_{it},$$
(2)

Equation (2) represents the DEC regression equation for the mediating variables (Mit), where other variables are the same as in Eq. (1).

Random forest model

Building on Hypothesis 3, this study employs a machine learning-random forest model to analyze the real non-linear effects of DEC on GTUM. This approach is justified for two major reasons. First, the development of DEC aligns with Metcalfe’s Law and is likely to exhibit non-linear characteristics. Second, traditional statistical models have limitations in accuracy. To achieve higher precision, machine learning models, including black box models, are utilized to investigate the real non-linear impacts of DEC on GTUM, ensuring that the overall model results possess high accuracy and generalizability. The following model is constructed:

$$GTU{M}_{it}=f(DE{C}_{it},{X}_{it},{u}_{i},{\delta }_{t},{\varepsilon }_{it}),$$
(3)

where f (⋅) represents the non-linear model constructed using the random forest method; other variables are the same as in Eq. (1).

First, given any black box model, a partial dependence function is defined as follows:

$${\hat{f}}_{{x}_{s}}({x}_{s})={E}_{{x}_{c}}[\hat{f}({x}_{s},{x}_{c})]=\int \hat{f}({x}_{s},{x}_{c})dP({x}_{c}),$$
(4)

where xs represents the feature variable of interest, and xc represents all other variables. By integrating over xc, a function \(\hat{f}({x}_{s})\) that depends only on xs is obtained. This function is known as the partial dependence function, which allows for interpreting the effect of the single variable xs. In practical terms, the estimate of \({\hat{f}}_{{x}_{s}}\) is obtained by averaging the training data, using the following formula:

$${\hat{f}}_{{x}_{s}}({x}_{s})=\frac{1}{n}\mathop{\sum }\limits_{j=1}^{n}\hat{f}({x}_{s},{x}_{x}^{(i)}),$$
(5)

where \({{x}_{c}}^{(i)}\) represents the actual values of the features in the feature, and space n denotes the sample size, excluding set s in the dataset. Plotting the relationship between different values of the feature variable and the predicted values yields the partial dependence plot. This plot illustrates the marginal effects of DEC on ETFP.

Variable setting

Explained variable

GTUM is not only reflected in the upgrading trend from lower to higher levels within the industry but also by reduced energy consumption and enhanced environmental protection, reflecting a comprehensive internal improvement within the manufacturing sector, and thus realizing green manufacturing. Green manufacturing is a modern manufacturing model with low consumption, low emissions, and high efficiency. It serves as a core embodiment of GTUM. Based on policy documents such as “Made in China 2025” and “Guidelines on Deepening the ‘Internet + Advanced Manufacturing’ to Develop the Industrial Internet,” as well as the study on GTUM by Xie and Han, (2022), this study considers that “GTUM” is a composite of the degree of industrial structure upgrading and greening of the manufacturing. Following Li et al., (2019), the GTUM index is constructed and measured using an improved entropy method, as shown in Table 1. Among them, the efficiency dimension represents the degree of upgrading of manufacturing, and the energy consumption and environmental protection (ECEP) dimension represents the degree of greening of the manufacturing industry.

Table 1 Evaluation indicator system for GTUM.
Full size table

Core explanatory variable

Drawing from Wang et al., (2024), this study constructs a comprehensive evaluation index system for DEC development across four dimensions: the development environment of DEC, digital industrialization, industrial digitalization, and digital governance. As shown in Table 2.

Table 2 Evaluation indicator system for DEC.
Full size table

Mechanism variables

(1) Market segmentation (Merc_Seg): This study uses employs the relative price method to calculate the index of factor market segmentation, taking inspiration from Bian et al., (2019). First, three-dimensional panel data concerning time, region, and various factors are required, and the first-order difference of the price loop index measures the relative price.

$$\,\Delta {Q}_{{ijt}}^{k}={ln}\left(\frac{{P}_{{it}}^{k}}{{P}_{{jt}}^{k}}\right)-{ln}\left({P}_{{it}}^{k}-\frac{1}{{P}_{{jt}}^{k}}-1\right),$$
(6)

Second, considering that the calculation of relative prices is affected by the differences brought about by the factors’ own characteristics, the de-meaned method is used to deal with it, and the relative price fluctuation \({q}_{{ijt}}^{k}\) is calculated by the formula:

$${q}_{{ijt}}^{k}=\left|\Delta {Q}_{{ijt}}^{k}\right|-\left|{\overline{\Delta {Q}_{t}^{k}}}\right|,$$
(7)

where \({q}_{{ijt}}^{k}\) is independent of the characteristics of the factors themselves and solely reflects the degree of market segmentation.

Finally, the variance of the portion of the relative price change of factors between each two regions is calculated and combined by province, resulting in a market segmentation index for each province and neighboring provinces:

$${var}\left({q}_{{nt}}\right)=\left[{\sum }_{i\ne j}{var}\left({q}_{{ijt}}\right)/{\rm{N}}\right],$$
(8)

We select the price indices of eight commodity categories such as food, clothing and footwear, and daily necessities for calculation, where N is the number of neighboring provinces.

(2) Market potential (Merc_Pot): Following the method of Midelfart et al., (2000), incorporating the impact of its distance on the economy of the region, market potential is calculated as follows:

$${{Merc{\_}Pot}}_{i}=\mathop{\sum }\limits_{\text{k}=1}^{\text{k}}\frac{{I}_{k}}{{D}_{{ik}}}+\frac{{I}_{i}}{{D}_{{ii}}},$$
(9)

where Merc_Poti represents the market potential of provinces (or cities), and k represents provinces other than province i, I is measured using total retail sales of consumer goods, and Dik represents the spherical distance between provinces, calculated using the coordinates of the center of mass of each province and the distance formula:

$${D}_{{ik}}=\Omega \times \arccos (\cos \left({\alpha }_{o}{\alpha }_{v}\right){\cos {\beta }_{o}\cos \beta }_{v}+\sin {\beta }_{o}\sin {\beta }_{v}),$$
(10)

Self-distance \({D}_{{ii}}=(\frac{2}{3})\sqrt{{S}_{i}/\pi }\), where S is the land area of the province.

(3) Market supply-demand matching (Merc_Sd): Drawing from Hitt et al., (1997) and Chang and Wang, (2007), supply-demand matching refers to the situation where consumers find providers whose products meet quality and price requirements. Producers create diverse products based on market demand, and a company’s level of product diversity is highly correlated with its R&D activities. Therefore, the extent of market supply-demand matching is also reflected i in the sales of new products companies develop. Merc_Sd can be calculated using the sales income from new goods of industrial companies more than a certain size within the region (in trillions), with the values logarithmically transformed.

Control variables

To control for other factors that may influence GTUM in various regions, this study references existing literature and selects the following control variables: Regional economic development level (PGDP), measured by per capita gross domestic product (GDP); Foreign Direct Investment (FDI), measured by the ratio of actual utilized foreign direct investment to the regional GDP; Human capital level (Human_Cap), measured by the average years of education per capita; Degree of government intervention (Gover_Inter), measured by the ratio of general budget expenditures to regional GDP; and Transportation infrastructure (Infra_Trans), measured by road network density.

Data sources

Given the absence of data for Tibet and considering that data for manufacturing sub-sectors are only published at the provincial level, this study employs panel data from 2011 to 2020, integrating provincial data with manufacturing sub-sectors in China. Missing values for some indicators were filled in using linear interpolation. The data sources include the “China Statistical Yearbook,” “China Industrial Yearbook,” provincial statistical yearbooks and statistical bulletins, the Baidu Index, the Guoyan Network, and the EPS Data Platform. Descriptive statistics for the variables are shown in Table 3.

Table 3 Descriptive statistics of variables.
Full size table

The descriptive statistics of key variables from Table 3 indicate that GTUM in the sample has a certain foundation, but there is still much room for improvement. The standard deviation suggests that the dispersion of GTUM values around the mean is relatively low across regions. The minimum and maximum values reveal that some regions are only at the preliminary stages of GTUM, whereas others have achieved higher levels of GTUM. The average value from DEC shows that the overall development degree of DEC is moderate to low, and the relatively small standard deviation indicates that the development level of DEC in these regions is relatively concentrated. The range of extreme values from 0.071 to 0.610 demonstrates that while some areas are highly developed in terms of DEC, others are comparatively underdeveloped.

Empirical results

Baseline results

First, to test the influence of DEC on GTUM. Based on the Hausman test, we selected the fixed effects model to regress Eq. (1) and used the OLS regression results for comparison. Table 4 presents the regression results of the fixed effects model in columns (1) and (2), and the OLS regression results in columns (3) and (4).

Table 4 Baseline regression results.
Full size table

The results from Table 4 indicate that, even without control variables, the coefficient for DEC is significantly positive in Columns (1) and (3), confirming that DEC development positively impacts GTUM. After incorporating control variables in Columns (2) and (4), DEC remains positive and significant, consistently supporting the significant influence of DEC on GTUM and confirming Hypothesis 1. DEC not only injects new vitality into traditional industries but also promotes the formation of a green manufacturing model by optimizing supply chain management and reducing pollution emissions (Liu et al., 2023). The deep implementation of next-generation information technology in the industrial sector enhances the green production capabilities of the manufacturing industry, gaining a competitive advantage in the market. This advancement is essential for addressing contemporary environmental challenges.

Furthermore, regression analysis was conducted based on the two major dimensions of GTUM. As indicated in Column (5) of Table 4, DEC is positively insignificant, indicating that the effect of DEC on enhancing the efficiency dimension of GTUM is unclear. However, DEC in Column (6) is significantly positive, indicating that DEC significantly promotes the optimization of the ECEP dimension of GTUM. Compared to the efficiency dimension, DEC has a greater impact on the energy and environmental dimensions of manufacturing. The underlying reason may be that the industrial chain of some of China’s manufacturing sectors is predominantly concentrated in the downstream, the foundation is generally weak, and the effects of DEC on enhancing manufacturing efficiency have not been fully realized. Meanwhile, applying new technologies such as big data and the Internet has improved resource utilization efficiency in the production and consumption phases and strengthened environmental awareness, significantly promoting the reduction of ECEP in manufacturing (Zhang and Zhang, 2024).

Robustness tests

Endogeneity treatment test

To control potential endogeneity issues as much as possible and enhance the robustness of our research conclusions, this study draws on the approach by Huang et al., (2019) for constructing instrumental variables. It generates panel data by multiplying the number of fixed telephones per hundred people in 1984 for each area by the previous year’s national income from information service technology (DEC_Phone), which acts as an instrumental variable for DEC in each region. The lagged one period of DEC (L.DEC) is also chosen as an instrumental variable. Table 5 shows the results of a two-stage least squares regression analysis.

Table 5 Endogenous test results.
Full size table

Columns (1) and (2), and Columns (3) and (4) in Table 5 represent the two-stage regression results obtained using DEC_Phone and L.DEC as instrumental variables, respectively. The Kleibergen-Paap k LM statistic tests for underidentification of instrumental variables. The p-values reported in parentheses are all less than 0.01, rejecting the null hypothesis of underidentification of the instrumental variables. The Kleibergen-Paap Wald rk F statistics are 32.019 and 122.055, respectively, suggesting no weak instrumental variable problem. The instrumental variables DEC_Phone and L.DEC and the development of DEC have an important positive connection, according to the first-stage regression findings. The second-stage regression results indicate that, after using instrumental variables, DEC is significantly positive, consistent with the baseline regression results. Thus, the core hypothesis still holds even after considering potential endogeneity issues.

Other robustness tests

The following robustness tests are conducted to verify the reliability of the core hypothesis:

  1. (1)

    Following the “2017 National Economic Industry Classification” for categorizing manufacturing sectors, this study reevaluates the level of GTUM by using the ratio of total output value of clean industries to that of pollution industries, represented as GTUM1. Column (1) of Table 6 shows that DEC promotes GTUM, underscoring the reliability of this finding.

    Table 6 Robustness test results.
    Full size table
  2. (2)

    GTUM is reassessed using the green total factor productivity of manufacturing (GTUM2), calculated using the slacks-based measure model. Taking each manufacturing sector in each province as a decision-making unit and considering the combination of production input factors and output factors in manufacturing. For inputs, this includes capital, labor, and energy, where energy input is represented by total energy consumption, capital input is represented by the fixed asset value of industrial enterprises above designated size, and the number of employees in urban manufacturing units represents labor input. Regarding outputs, the expected output is represented by the added value of manufacturing. In contrast, the undesired outputs include industrial smoke and dust emissions, industrial sulfur dioxide, industrial chemical oxygen demand, and industrial ammonia nitrogen. As shown in Column (2) of Table 6, DEC is still significantly positive, confirming that the core hypothesis of this study is robust.

  3. (3)

    Following Wang et al., (2020), this study remeasures the level of GTUM using the ratio of the added value of technology-intensive manufacturing to capital-intensive manufacturing, denoted as GTUM3. This ratio reflects whether the manufacturing industry structure is evolving towards “servitization” and “technology intensity,” indicating a shift toward more efficient and cleaner production and promoting the development and application of green technologies, thereby facilitating GTUM. As shown in Column (3) of Table 6, the conclusion that DEC still promotes GTUM is robust.

  4. (4)

    The transportation infrastructure variable is replaced with the urbanization level variable (Urban), represented by the ratio of the urban population to the total regional population. The findings in Table 6’s Column (4) demonstrate that DEC is significantly positive, confirming that the conclusion that DEC still promotes GTUM is robust.

Mechanism analysis

This study analyzes the impact mechanisms of DEC development on GTUM from the perspective of ULM construction, focusing on market segmentation, market potential, and market supply-demand matching. The regression results are shown in Table 7.

Table 7 Mechanism test results.
Full size table

(1) Testing the reducing market segmentation effects. Analysis of Column (1) of Table 7 reveals that the development of DEC significantly influences the reduction of market segmentation. This outcome can be attributed to the advantages of DEC in breaking the time and space limitations, as per theories on network externalities and market boundarylessness. By enabling the free flow and efficient allocation of production factors, DEC expands the market’s effective boundaries, thereby diminishing market segmentation. Building on the theoretical analysis section and existing research, the study demonstrates that reducing market segmentation facilitates the formation of a national ULM, enhances resource allocation efficiency, leads to industrial structural upgrades and carbon emission reductions, and promotes GTUM (Bai et al., 2004), thus validating Hypothesis 2a.

(2) Testing the stimulating market potential effects. The results from Column (2) of Table 7 indicate that the development of DEC significantly positively impacts stimulating market potential. According to the theory of economies of scale and Metcalfe’s Law, DEC utilizes the advantages of digital technology and the Internet to explore potential customers fully, and enterprises can adjust the scale of production more efficiently, provide more products and services, and stimulate market potential. Meanwhile, according to the previous theoretical analysis section and existing research, the higher potential of market demand will reduce the production and operation costs and transportation costs of enterprises, strengthen the technological innovation of enterprises, and promote the manufacturing industry toward green transformation (Melitz and Ottaviano, 2008). Therefore, by stimulating market potential, DEC can drive GTUM, confirming Hypothesis 2b.

(3) Testing the market supply and demand matching effects. The results from Column (3) of Table 7 show that DEC’s development significantly positively impacts market supply-demand matching. A plausible explanation for this is based on market supply-demand mechanisms and Moore’s Law: the widespread application of digital technologies fosters diversified consumer demand and drives increases in enterprise output and cost reductions, triggering scope and scale economic effects. Furthermore, according to the previous theoretical analysis section and existing research, matching market supply and demand can enhance the sales capacity of enterprises; reduce inventories; and facilitate enterprises to use more funds for R&D, energy saving, and emission reduction; improve the environmental performance of enterprises; and promote GTUM (Wang et al., 2023; He et al., 2023). Therefore, DEC can drive the GTUM through market supply-demand matching, confirming Hypothesis 2c.

Heterogeneity analysis

Existing research indicates that the development of DEC can facilitate GTUM. To further systematically unveil the variations in the effects of DEC-driven GTUM across different regions, this study conducts a heterogeneity analysis focusing on three aspects: external environmental factors, internal driving force factors, and societal pressure factors. Specifically, there are significant regional differences in the degree of openness to foreign markets, capabilities for technological innovation, and public environmental awareness. These variations may result in differing impacts of DEC on GTUM across different areas, necessitating further discussion.

(1) External environmental factors (Degree of trade openness). External environmental factors determine a region’s capability to access external resources and technology. Regions with a higher degree of trade openness can more easily adopt internationally advanced green technologies and environmentally friendly products, enabling green production practices. Due to technical barriers and market entry restrictions, regions with limited trade openness may have difficulty obtaining the external resources and market opportunities needed for green transformation. Drawing from previous studies, this study measures the degree of trade openness (Trade_Open) by the ratio of actual foreign direct investment to GDP, dividing the sample into regions with high and low trade openness based on the median value. Regression results presented in Columns (1) and (2) of Table 8 indicate that, from the perspective of external environmental factors, DEC plays a more significant role in promoting GTUM in regions with a higher degree of trade openness. However, this effect is less pronounced in regions with lower trade openness, highlighting the importance of transnational cooperation and knowledge sharing. By actively participating in international collaborations, regions can facilitate the exchange of advanced digital technologies and environmental practices (Buckley et al., 2002), accelerating the process of GTUM and supporting China’s economy to realize the efficient and smooth operation of internal and external cycles.

Table 8 Heterogeneity analysis results.
Full size table

(2) Internal driving factors (Level of technological innovation). Technological innovation is the core internal driver of GTUM, supported fundamentally by DEC. Regions with higher levels of technological innovation can access and adopt the latest digital technologies earlier, making it easier for their manufacturing sectors to achieve smart manufacturing. This study enhances production efficiency and reduces energy consumption, contributing to green production. Additionally, the clustering effect of technology accelerates the dissemination and application of green technologies, driving GTUM. Drawing from previous studies, this study measures the level of technological innovation (Techn_Nov) by the ratio of R&D expenditure to GDP, dividing the sample into two groups based on the median value. From the perspective of internal driving force factors, DEC plays a more substantial role in fostering GTUM in places with greater levels of technical innovation, according to the regression findings shown in Columns (3) and (4) of Table 8. However, this effect is less evident in regions with lower levels of technological innovation. This finding suggests that regions with higher levels of technological innovation can better leverage the advantages of DEC to accelerate GTUM and should continue to increase their investment in innovation without letting up to promote the realization of a digitally driven GTUM throughout the entire society (Wang et al., 2024b).

(3) Social pressure factors (Public environmental awareness). Public environmental awareness has a direct impact on the social atmosphere for GTUM. Concern for environmental issues is one of the key manifestations of public preference for the environment. Heightened public environmental awareness can promote consistency in environmentally friendly behaviors, such as increased demand for green products and vigilance over the quality of living environments(Wang et al., 2024a), thereby creating external societal pressure. Drawing on Yu et al., (2023), this study measures public environmental awareness (Penv_Tend) using the Baidu search index for “environmental pollution,” dividing regions into high and low concern areas based on the median value. Group regression analysis, with empirical results shown in Table 8, Columns (5) and (6), shows that from the perspective of societal pressure factors, in regions with high public environmental awareness, DEC more effectively promotes GTUM. However, this effect is not apparent in regions with low public environmental awareness, suggesting that the empowerment of GTUM by DEC requires the support of public environmental engagement.

Further discussion: non-linear effects

Following the approach of Escribano and Wang (2021), this study validates Hypothesis 3 by employing a machine learning model to analyze the true non-linear effects of GTUM development. Unlike linear regression models, decision tree-based learning algorithms provide accuracy, stability, and interpretability, offering better mapping of non-linear relationships. Partial dependence plots, a visualization tool for analyzing machine learning models, allow for the display of the influence of individual features on prediction outcomes while controlling other features constantly. This plot reveals the relationships among variables in the predictive model. Therefore, this study constructs a random forest model with a regression decision tree as the basic learner and minimum mean squared error as the optimization criterion for selecting the split nodes, whereby the biased partial dependence Fig. 2 is drawn.

Fig. 2
figure 2

Partial dependence plot for GTUM on DEC.

Full size image

Figure 2’s partial dependence function illustrates the impact of DEC development on GTUM. The internal scale on the horizontal axis marked “|” represents the percentiles of DEC development levels. The temporal dynamic changes of this influence reveals that this partial dependence relationship is inhibited first and then promotes the characteristics of the “U-shaped” change, U-shaped inflection in DEC development index of 0.3. In the early stages of DEC development (when the DEC index is less than 0.3), the function overall shows a decreasing trend, indicating a negative impact of DEC on GTUM. Specifically, in the 6/10–8/10 and 8/10–9/10 percentile regions, the function exhibits a fluctuating increasing trend in stages, suggesting that DEC has a slight promotional effect on GTUM. However, as DEC enters a relatively mature phase (when the DEC index reaches 0.3), its impact on GTUM shifts to a positive and steadily enhancing role. This change indicates that the early accumulation in DEC begins to exhibit exponential growth, and costs are significantly reduced, increasingly highlighting DEC’s driving effect on GTUM. Therefore, Hypothesis 3 is validated, indicating that the impact of DEC in promoting GTUM conforms to Metcalfe’s Law and exhibits non-linear characteristics (Pouri and Hilty, 2021).

A possible explanation for this phenomenon is that companies face high costs associated with digital production equipment and infrastructure at the outset of digital transformation. Especially for some small and medium-sized enterprises, the challenge of changing from a traditional to a digital production model is vast (Ren et al., 2021b). Consequently, during the early stages of DEC development, GTUM progresses slowly and with difficulty. Consequently, during the early stages of DEC development, GTUM progresses slowly and with difficulty. However, as DEC matures, the marginal costs of digital technologies decrease. Companies gradually adapt to the demands of digital transformation, and their production processes also become cleaner and more environmentally friendly (Ren et al., 2023), accelerating the pace of GTUM.

Furthermore, ULM construction is incorporated into the partial dependence function of DEC development on GTUM, as shown in Fig. 3. Among them, Fig. 3a–c display the effects of the interactions between DEC and market segmentation (DEC* Merc_Seg), DEC and market potential (DEC* Merc_Pot), and DEC and market supply-demand matching (DEC* Merc_Sd) on GTUM, respectively. It examines how the real non-linear effects of DEC development on GTUM would differ under different market segmentation, different market potentials, and different degrees of matching market supply and demand.

Fig. 3: Partial dependence function of DEC and ULM construction on GTUM.
figure 3

Specifically, (a–c), respectively, showcase the impact of the interaction terms between DEC and market segmentation, DEC and market potential, and DEC and market supply–demand matching on GTUM.

Full size image

Figure 3a illustrates that under the intervention of the factor of market segmentation, as the level of development of DEC and the degree of market segmentation increase, the overall partial dependence function exhibits a fluctuating downward trend. This increase with the “U-shaped” change in Fig. 2 regarding the impact of DEC on GTUM reveals that, in the case of considering the existence of a higher market segmentation, DEC significantly lowers the level of GTUM. The relationship in Fig. 3a shows the change characteristics of the left side of the “U-shaped” inflection point. One reason for this is that the early phases of DEC growth are dependent on energy-intensive businesses, including data centers and electronic equipment, which increases the strain on the environment. Besides, higher levels of market segmentation mean reduced inter-industrial communication, leading to longer production chains and redundant supply chains, which increase resource consumption and hinder GTUM.

As shown in Fig. 3b, c under the intervention of market potential and supply-demand matching, the overall partial dependence function exhibits a fluctuating upward trend as the DEC development index and market potential increase and the degree of market supply-demand matching becomes higher.

Comparing this with the “U-shaped” change observed in Fig. 2 regarding the impact of DEC on GTUM, in the case of considering the existence of market potential and market supply-demand matching factors, DEC significantly enhances the level of GTUM. In addition, compared to Fig. 3b, the ascending trend of the function in Fig. 3c is steeper, suggesting that the market supply-demand match has a stronger amplifying effect on DEC’s promotion of GTUM relative to the intervention of the market potential factor.

The reason is that greater market potential means a big demand for digital technology applications to optimize the production and manufacturing process and reduce energy consumption and waste. Second, with the new generation of digital technology, mass production and effective matching of supply and demand can improve economic efficiency, and enterprises have more energy to invest in research and development of green technology, which together promotes GTUM and sustainable production. All of the above shows that China’s ULM construction has amplified the effect of DEC on GTUM. The above explanations show that China’s ULM construction has amplified the effect of DEC empowering the GTUM.

Conclusions and Policy Implications

Research conclusions

This study examines the impact of DEC on GTUM and its action mechanism at the theoretical and empirical levels based on the perspective of China’s ULM construction. The results indicate that DEC significantly enhances GTUM, a conclusion that remains robust after a series of empirical tests. The sub-dimensional regression shows that compared to the efficiency dimension of GTUM, DEC has a greater impact on the ECEP dimensions. Mechanism tests show that the development of the DEC primarily promotes GTUM by reducing market segmentation effects, enhancing market potential effects, and matching market supply and demand. Additionally, the effectiveness of DEC in driving GTUM exhibits heterogeneity, with stronger impacts observed in regions with higher degrees of trade openness, technological innovation, and public environmental awareness. Furthermore, machine learning models indicate that the influence of DEC on GTUM aligns with Metcalfe’s Law, displaying non-linear characteristics. Besides, ULM construction amplifies the effect of DEC empowering GTUM. Specifically, under the intervention of market segmentation factors, the development of DEC reduces the level of GTUM. The intervention of market potential and market supply-demand matching plays a significant role in promoting GTUM, and market supply-demand matching has a strong amplification effect on this promotion role.

Policy recommendations

Based on the empirical findings outlined above, this study puts forth the following recommendations.

First is accelerating the digital transformation of production to empower green manufacturing. The government should actively guide enterprises in undergoing digital renovations and upgrades. Manufacturing enterprises need to deepen the application of digital technologies to enhance resource utilization efficiency and production flexibility, reduce market transaction costs, and provide support and assistance for GTUM. More importantly, efforts should be made to expedite the establishment of intelligent manufacturing centers, where collaboration between humans and intelligent machines can lead to the automation, intelligence, and greening of the production process, fully leveraging “Metcalfe’s Law.”

Second is to leverage the synergistic effects between the development of DEC and ULM construction to unleash the potential of a vast market scale. Governments should strive to eliminate barriers caused by market segmentation and size limitations, promoting the formation of a large, nationally unified, competitive, and orderly market. This elimination can be achieved by refining market rules, strengthening market supervision, protecting fair competition, satisfying diverse market demands, and stimulating market vitality, thus providing strong support for the development of DEC and the construction of a modern market system. At the same time., manufacturing enterprises should actively establish cross-regional green innovation alliances to foster collaborative innovation, jointly develop green technologies and solutions, and compensate for the innovation resource limitations of small and medium-sized manufacturing businesses. Additionally, implementing digital and intelligent supply chain management is critical; this involves flexibly adjusting production and logistics arrangements based on market demands to ensure timely product delivery, which is key to enhancing market competitiveness and tapping into market potential.

Third is to adopt a localized approach and formulate differentiated strategies. For regions with strong technological innovation capabilities and high levels of openness, governments should fully utilize both domestic and international market resources. With DEC as a focal point, efforts should be concentrated on building a new industrial system. This includes developing leading industries such as integrated circuits and biopharmaceuticals, creating high-end industrial clusters in the electronics information and automotive sectors, and fostering a group of internationally competitive advanced manufacturing bases to further enhance the effectiveness of DEC in driving GTUM. In areas where public environmental awareness is lower, it is crucial to strengthen online environmental education and awareness campaigns, organize digital nationwide green initiatives, and promote a lifestyle and consumption pattern that is economical, green, low-carbon, and healthy. This will encourage the manufacturing industry to adopt digital technologies more actively to meet societal environmental expectations. Through these differentiated strategies, the positive impact of DEC on GTUM can be maximized.

Research limitations and future directions

Drawing from the context of the digital era and the environmentally conscious evolution within the manufacturing industry, this study reveals the driving role of DEC in GTUM and its action mechanism. Two aspects that can be further improved and thoroughly studied in the future are as follows. First, this study adopts econometric modeling combined with machine learning and other techniques, and further expansion of the application of machine learning models in causal inference is considered in the future. The study also highlights further enhancement and exploration in two key dimensions. Second, while this study explores the impact of DEC on GTUM at macro and meso-levels, future research could investigate this issue from the perspective of micro-level enterprises. For instance, examining how specific digital technologies affect green practices in manufacturing enterprises could enrich the literature on microeconomic impacts.