Unlocking the potential of supply chain digitalization for enhancing enterprise green transformation performance: evidence from China

Abstract

The supply chain is a significant source of carbon emissions for enterprises, and effectively managing this carbon footprint is a key strategy for achieving dual carbon goals and promoting sustainable development. In 2018, China launched the supply chain innovation and application pilot policy to promote supply chain digitalization (SCD), yet its actual impact on enterprise green transformation performance (EGTP) remains unclear. This study uses panel data of Chinese A-share listed companies from 2013 to 2023 as the research sample. Grounded in the practice-based view and dynamic capability theory, we employ a double machine learning method for empirical analysis. Our findings reveal that SCD significantly improves EGTP. The heterogeneity analysis further shows that this effect is more pronounced in non-state-owned enterprises, high supply chain integration, high industry competition, and non-high-polluting industries. The mechanism analysis shows that SCD promotes EGTP by enhancing an enterprise’s adaptive, absorptive, and innovative capabilities. Further analysis demonstrates that the spillover effects of SCD have a more pronounced impact on upstream suppliers’ EGTP. These insights offer theoretical and practical guidance for enterprise managers and policymakers seeking to leverage SCD in sustainable environmental practices.

Introduction

In the contemporary global economic context, enterprises, as pivotal drivers of economic growth, encounter significant and pressing challenges in improving their environmental performance. They are required to directly confront the severe risks associated with climate change and increasing pressure from resource limitations. Simultaneously, they are obligated to meet environmental commitments by demonstrating measurable advancements in energy efficiency and reductions in carbon emissions, thereby demonstrating their environmental accountability to stakeholders (Li et al. 2024; Zhang et al. 2024). In this context, the development of a sustainable model that prioritizes enterprise green transformation performance (EGTP) as a fundamental competitive advantage has emerged as a strategic imperative for global enterprises (Chen et al. 2021; Jan et al. 2024). As micro-level agents in both environmental protection and economic development, enterprises are the primary sources of resource consumption and pollutant emissions (Guo et al. 2024). Research indicates that supply chains account for 60% of global carbon emissions, with approximately 75% of industrial sector emissions originating from supply chain operations. Crucially, supply chain-related emissions can be 26 times greater than those generated by an enterprise’s direct operations (Blanco 2021; Chen et al. 2024; Singh et al. 2022). These findings demonstrate that enterprises focusing exclusively on internal carbon performance enhancement while neglecting supply chain-wide decarbonization efforts will struggle to achieve substantial environmental improvements. Consequently, investigating how to drive enterprises’ green transformation effectively through supply chain interventions to attain superior environmental performance has emerged as a pressing practical challenge for businesses.

Meanwhile, amid the advancement of digital technologies, enterprises are navigating waves of digital transformation (Feng and Wang 2024). The integration of digital technologies into supply chain operations and management is referred to as supply chain digitalization (Buyukozkan and Gocer 2018; Zhao et al. 2023). However, scholars have yet to reach a consensus on the impact of supply chain digitalization (SCD) on green development. Some scholars have argued that SCD may increase corporate carbon emissions. For instance, Liu et al. (2023) emphasized that smart supply chains require extensive electronic devices and sensors that consume significant amounts of electricity and other resources. This suggests that digital industries built on digital technologies are inherently energy-intensive (Wang et al. 2023). Thus, the adoption of digital technologies may elevate carbon emissions and weaken the effectiveness of environmental governance, leading to a green paradox (Li 2022). In contrast, other studies indicate that SCD can foster green innovation (Han and Wei 2025), mitigate information asymmetry (Meng and Lin 2025), and enhance resource allocation efficiency (Wang et al. 2025), thereby reducing carbon emissions. Overall, the impact of digitalization on the environmental performance of enterprises has both inhibitory and promotional forces. That is to say, whether SCD supported by digital technology can promote EGTP improvement remains to be further tested.

In 2018, the Chinese government implemented the Supply Chain Innovation and Application Pilot (SCIAP) policy. This policy requires enterprises to adopt a green development approach encompassing the entire process, full chain, and all aspects of production, promoting the application of new supply chain technologies and models. It aimed to drive the digitalization and intelligentization of industrial supply chains (Han and Wei 2025; Li et al. 2025). The pilot policy sparked extensive academic discussion, with related literature treating it as a quasi-natural experiment to examine the impact of the SCD on corporate competitiveness (Wang and Li 2024), green innovation performance (Luo et al. 2024; Xu et al. 2023), and sustainability performance (Chen et al. 2025). These studies provide a foundation for identifying the influence of SCD on EGTP. Based on this, this study considers the SCIAP as a quasi-natural experiment. Based on panel data of China’s A-share listed companies from 2013 to 2023, we systematically examine the impact, mechanism of action, and spillover effect of SCD on EGTP. The empirical results demonstrate that SCD significantly enhances EGTP, with notable heterogeneity and spillover effects. Moreover, SCD primarily drives EGTP by strengthening enterprises’ dynamic capabilities.

Compared with the existing literature, this study makes several theoretical contributions. First, it expands the theoretical perspective of how SCD affects environmental performance. Current research has primarily examined the relationship between SCD and green innovation or carbon emissions, with limited direct investigation of its role in enabling the EGTP. Second, this study extends the application of dynamic capability theory to explain the relationship between SCD and environmental performance. Our findings address gaps in the current literature regarding the mechanisms through which SCD influences EGTP, while broadening the understanding of how dynamic capabilities facilitate sustainability outcomes in digital environments. Third, we reveal the differential characteristics of digital spillover effects in vertical supply chains, which provides new empirical evidence for understanding SCD.

Literature review

The literature that is most closely related to this study encompasses three research streams. The first is the research on the impact of SCD on economic performance. The second is research on the factors influencing green transformation. The third is the research on the impact of SCD on environmental performance.

Research on the economic consequences of the SCD

In recent years, corporate management, government agencies, and policymakers have increasingly recognized digital transformation as a core element in building sustainable business models. Saeedikiya et al. (2025) highlighted that digital transformation represents an ongoing societal structural change that leverages digital technologies to create new value and achieve a sustained competitive advantage. As digital transformation reshapes supply chain management paradigms, SCD is becoming increasingly significant. Buyukozkan and Gocer (2018) emphasized SCD as an intelligent technological system that integrates hardware, software, and network resources to optimize service value, accessibility, and cost-effectiveness, thereby enabling agile and sustainable supply chain operations. Similarly, Son et al. (2021) characterized SCD as the integration of digital technologies across supply chain processes to establish a comprehensive system featuring digitalization, networking, and intelligent integration of supply chain processes. Additionally, Luo et al. (2024) described this as the application of digital tools in supply chain management. Collectively, these scholarly perspectives reveal that SCD refers to the application of digital technologies by enterprises to digitally transform end-to-end supply chain processes, establishing a data-driven, intelligently collaborative, and efficient decision-making management model. Its fundamental objectives include optimizing supply chain efficiency, reducing costs, enhancing transparency and flexibility through technological means, and ultimately, improving corporate competitiveness and sustainable development.

Extensive research has consistently demonstrated the significant positive impact of SCD on corporate performance. Specifically, Sharma et al. (2022) found that digital supply chain networks facilitate the adoption of low-carbon practices, thereby enhancing sustainable performance. Concurrently, Wang and Prajogo (2024) established that digital transformation in supply chains strengthens both supply chain agility and innovation capability, leading to improved corporate performance. Further supporting these findings, Chen et al. (2025) demonstrated that SCD contributes to sustainable performance by simultaneously boosting innovation willingness and reducing operational costs. Regarding innovation performance, Li et al. (2024) showed that digitalization enhances information acquisition capabilities and resource integration efficiency through improved data sharing, creating optimal conditions for collaborative innovation within supply chains. In terms of green innovation, SCD can positively influence corporate green innovation through multiple pathways, including alleviating financing constraints (Jiang et al. 2025) enhancing internal control capabilities (Luo et al. 2024) and improving supply chain management efficiency (Ma et al. 2024). These findings underscore the multifaceted positive effects of SCD on various dimensions of performance.

The influencing factors of green transformation

Green transformation refers to the process through which enterprises integrate environmental protection and sustainable development into their business strategies, operational models, products, and services. This green development process fundamentally aims to enhance resource utilization efficiency, while reducing pollution and greenhouse gas emissions (Ge et al. 2023). It encompasses the adoption of environmentally friendly practices throughout the production process and supply chain (Guo et al. 2024). This transformation is influenced by internal enterprise factors and external environmental conditions.

From an internal enterprise perspective, the existing literature identifies several critical factors influencing green transformation, including organizational resources and capabilities, corporate culture, and management competence (Wang and Cheng 2024). Specifically, human capital and technological resources (Yang et al. 2023; Zhai and An 2020), corporate risk awareness (Deng et al. 2024), executive environmental concerns (Ding et al. 2023), and digital and intelligent transformation initiatives (Guo et al. 2024; Wang et al. 2023; Xu et al. 2023) have been shown to contribute significantly to enterprises’ green transformation.

From an external perspective, enterprises often face challenges, such as insufficient resources and weak transformation incentives, necessitating appropriate external subsidies and environmental regulatory constraints to motivate green transformation. Research on external factors primarily focuses on national policies, with existing studies examining the impacts of government-guided funds (Sun et al. 2024), carbon emission policies (Ge et al. 2023), and green credit policies (Lu et al. 2025; Zhang et al. 2024) on corporate green transformation. Furthermore, according to the coase theorem, environmental pollution problems mainly stem from market failures, requiring governments to implement reasonable environmental regulations to maintain sustainable corporate development (Sun et al. 2024). Formal environmental regulations, represented by government oversight and enforcement, can directly promote corporate green transformation (Li et al. 2019; Wu et al. 2024; Zhai and An, 2020) or serve as positive moderators to facilitate this process (Lin et al. 2025; Zhai and An 2020). Meanwhile, informal environmental institutions, such as public and media supervision, also influence corporate willingness to pursue green transformation (Liu et al. 2024).

Supply chain digitalization and environmental performance

SCD is closely linked to environmental performance. By leveraging digital technologies to optimize resource allocation, enable intelligent logistics, facilitate green production processes, and enhance supply chain collaboration (Lerman et al. 2022), digitalization effectively reduces resource waste and carbon emissions, thereby positively influencing environmental performance. For instance, 2022 Wang et al. (2025) found that SCD improves carbon performance by strengthening green innovation and internal control capabilities. Similarly, Li et al. (2025) demonstrated that it enhances the supply chain eco-centrality, thereby boosting carbon performance. SCD also promotes digital transformation and green innovation, leading to reduced carbon emissions (Han and Wei 2025). It also alleviates financing constraints and increases external market attention, which further contributes to emissions reduction (Shen et al. 2025).

However, it should be noted that SCD may also have negative environmental impacts. Hittinger and Jaramillo (2019) pointed out that because the digital industry is a high energy-consuming industry, new base stations, data centers, cloud servers, etc., which need to be built for digital transformation, consume a large amount of energy, which may increase carbon emissions and the environmental burden. Similarly, Liu et al. (2023) found that the application of advanced technologies in the SCD may elevate carbon emissions. Feng et al. (2024) further revealed that digital infrastructure contributes to a 10.84% increase in carbon intensity among small enterprises. Additionally, Liu et al. (2024) demonstrated that the establishment of smart factories significantly increases CO₂ emissions in the short term. These findings highlight the potential adverse effects of SCD on the environmental performance.

Research gaps

Table 1 summarizes the relevant literature related to this study. Our analysis revealed several research gaps, warranting further investigation. (1) While existing studies focus primarily on how SCD affects corporate innovation and sustainable performance, there is a notable lack of research examining its impact on EGTP. (2) The literature on the factors influencing corporate green transformation predominantly focuses on internal resources and external institutional policies, with insufficient attention paid to the role of SCD in this process. (3) Existing evidence regarding the environmental impact of digitalization remains inconclusive, demonstrating dual effects (both positive and negative). Furthermore, while some studies have confirmed the positive influence of SCD on green innovation capability, green innovation represents only one critical competency for green transformation, suggesting that other potentially significant capabilities remain unexplored. (4) Methodologically, current research on SCD and performance predominantly employs structural equation modeling or multiple regression analysis to examine variable relationships; however, these methods may be insufficient for precisely estimating causal relationships.

Table 1 Summary of literature review content.

These limitations present opportunities for further study. In this study, we utilize data from Chinese listed companies as our sample, leveraging the SCIAP policy, employing the SMB-GML model to measure EGTP, and then applying a double machine learning approach to estimate both the causal effects of SCD on EGTP and its underlying mechanisms. Furthermore, we conduct heterogeneity analyses to examine differential effects across various contexts and investigate potential spillover effects.

Theoretical basis and research hypotheses

Practice-based view

PBV posits that firms enhance their performance by adopting a series of observable and imitable operational practices (Bromiley and Rau 2014, 2016). In contrast to the resource-based view (RBV), which emphasizes scarce, valuable, and difficult-to-imitate resources as sources of sustainable competitive advantage (Barney 1986), the PBV focuses on how performance differentials emerge through variations in the execution of imitable practices. The existing literature has predominantly employed RBV theory, treating SCD as a valuable, rare, and imperfectly imitable resource to explain its impact on firm performance (Chatterjee et al. 2024; Son et al. 2021; Wang and Prajogo 2024). However, these studies overlooked the fact that SCD can serve as an imitable and transferable practice in corporate operations. SCD can be regarded as a practical activity, as evidenced in the literature (Dwivedi and Paul 2022; Mishra et al. 2024; Yu et al. 2024). For instance, Yu et al. (2024) argued that the adoption of digital technologies by enterprises in the digital transformation of supply chains can be termed as digital supply chain practices. Tian et al. (2023) demonstrated that in digital transformation practices, due to the bounded rationality of managers, they tend to prioritize the implementation of well-known and easily imitable approaches.

In this study, we argue that PBV is more suitable than RBV in explaining SCD. This is because, under the SCIAP policy, 55 cities and 266 enterprises were selected as pilot participants after rigorous screening. The primary task of these pilot enterprises is to establish smart supply chain systems and promote digital transformation and upgrading practices in China’s corporate supply chains. The practices adopted by these pilot firms are highly similar and can be readily learned and imitated by other enterprises (Liu et al. 2023). Second, whereas the RBV emphasizes sustained competitive advantage as the dependent variable, the PBV focuses on firm performance (Bromiley and Rau 2016). Because our study employs EGTP as the dependent variable, it aligns more closely with the theoretical requirements of the PBV.

Supply chain digitalization and dynamic capabilities

Dynamic capabilities refer to an enterprise’s core competencies in sensing, seizing, and reconfiguring internal and external resources to address drastic market changes (Saeedikiya et al. 2024; Teece 2007; Teece et al. 1997). Furthermore, Wang and Ahmed (2007) emphasized dynamic capabilities as a firm’s behavioral orientation to continuously integrate, reconfigure, and upgrade resources and core competencies to adapt to environmental shifts and gain a competitive advantage. Based on the distinctive characteristics of dynamic capabilities, three key dimensions have been identified: adaptive, absorptive, and innovative capabilities. These have been widely adopted in scholarly research (Feng and Wang 2024; Yang et al. 2023). Adaptive capability represents a firm’s core competence in identifying market opportunities and swiftly adjusting resources and strategies to respond to external changes, manifested through strategic flexibility, organizational transformation, and dynamic balancing abilities (Wang and Ahmed 2007). Absorptive capability refers to an enterprise’s capacity to acquire, assimilate, and exploit external knowledge (Feng and Wang, 2024). Innovative capability reflects a firm’s ability to develop new products or explore new markets through strategic orientation and innovation activities, encompassing multidimensional innovations in technology, processes, and markets (Saeedikiya et al. 2025; Yang et al. 2023).

A growing body of research recognizes dynamic capability theory as a crucial perspective for explaining the positive effects of digital transformation (Deng et al. 2024; Li, 2022; Luo et al. 2024). The application of digital technologies enhances enterprises’ real-time data collection and analysis capabilities, enabling the more precise identification of market demand fluctuations (Sarfraz et al. 2023). Digital tools facilitate process optimization in production, whereas digital transformation promotes the establishment of energy-efficient and environmentally friendly production models, thereby improving resource allocation efficiency (Gao and Huang 2024; Li 2022). However, digital transformation alone does not directly generate additional profits and may even increase organizational costs. Only when properly applied to business processes to cultivate dynamic capabilities (Mikalef et al. 2019) can they create competitive advantages for enterprises (Li 2022). The realization of the value of digital transformation critically depends on the synchronous development of dynamic capabilities, which serve as the key bridging mechanism connecting digitalization and environmental performance (Deng et al. 2024; Li 2022; Yang et al. 2023).

This study adopted the dimensional classification of dynamic capabilities proposed by Wang and Ahmed (2007). From the perspective of short-term adaptive capability, the application of digital technologies enables firms to perceive external environmental changes in real-time, respond swiftly to environmental policies and market fluctuations, and implement internal strategic transformations. This allows firms to identify green opportunities for survival in a rapidly changing environment, thereby demonstrating their short-term adaptive capacities. Regarding the resource reconfiguration effect of absorptive capacity, SCD provides a strong impetus to upgrade human capital within firms, thereby enhancing their supply chain learning capabilities (Liu et al. 2023). It also helps firms efficiently identify and internalize external knowledge and resources (Ngo et al. 2023), reflecting the knowledge and resource reconfiguration effects of their absorptive capacity. Finally, from the perspective of long-term innovation capability, SCD strengthens firms’ ability to integrate cross-domain knowledge and engage in collaborative innovation (Li et al. 2025), highlighting its role in promoting long-term innovation. The above theoretical analysis provides a foundation for the hypotheses proposed in the following sections.

The direct impact of SCD on EGTP

The PBV theory posits that a firm’s practical activities can directly or indirectly influence its performance (Bromiley and Rau 2016; Liu et al. 2023). We argue that SCD optimizes internal factor structures, enables data-driven resource management, and enhances the green performance of external supply chains, thereby improving EGTP. Specifically, first, SCD facilitates the adoption of digital technologies (Nie et al. 2025), increases the proportion of R&D personnel, and shifts labor input from traditional manpower to technology-intensive roles, thereby upgrading both technological and human resources (He et al. 2023; Luo et al. 2024; Shen and Zhang 2024). This upgrading of labor factors allows firms to implement advanced environmental technologies and management practices (Wang and Cheng 2024), integrating various production factors with digital technologies to achieve energy-saving and carbon reduction management (Zhang et al. 2023).

Furthermore, SCD facilitates data-driven operational management within enterprises, thereby enabling optimal resource allocation and waste reduction. By enhancing the integration of digital technologies into production and operational activities, firms can effectively monitor and track resources and energy consumption throughout the production process (Sharma et al. 2022). This capability allows enterprises to identify and trace pollution sources and to achieve greater transparency, intelligence, and sustainability in production operations. Consequently, companies can promptly adjust and refine their environmental strategies to minimize their ecological impacts (Yuan and Pan 2023). Notably, because supply chains represent a significant source of corporate carbon emissions, digital transformation embeds technological solutions into fundamental supply chain operations. This integration significantly improves supply chain visibility and transparency (Dolgui and Ivanov 2022; Zhou et al. 2023) and breaks down the information barriers between different supply chain segments (Fan et al. 2024). Such digital transformation effectively reduces resource consumption in procurement and logistics management, optimizes transportation routes, and lowers carbon emissions (Liao et al. 2024), thereby enhancing the overall sustainability of supply chain operations. These improvements facilitate collaborative environmental governance among stakeholders and ultimately boost green corporate performance (Kumar et al. 2024).

H1: SCD positively impacts EGTP.

The indirect impact of SCD on the EGTP

The PBV theory posits that practical activities generate intermediate outcomes that serve as bridges between practices and firm performance, thereby emphasizing the importance of examining mediating variables (Bromiley and Rau 2014). Furthermore, Bromiley and Rau (2016) argued that PBV highlights how organizations embed learning and adaptation mechanisms through practical activities, thereby developing inimitable specialized capabilities. In this study, we conceptualize dynamic capabilities as a mediating variable to elucidate how SCD enhances dynamic capabilities, which, in turn, improves EGTP.

The mechanism role of adaptive capacity

The ability of enterprises to rapidly adapt and respond is a critical determinant of their survival and success. First, SCD enhances an enterprise’s capacity to adapt to rapid changes in its external environment. Against the backdrop of green transformation, enterprises must dynamically respond to shifts in the market landscape driven by new entrants, regulatory changes, or evolving customer expectations (Chevrollier et al. 2024). By integrating digital technologies, SCD enables enterprises to detect market fluctuations, shifts in customer preferences, and trends in green policies more swiftly (Mak and Max Shen 2021). This facilitates timely strategic adjustments and strengthens an enterprise’s ability to capitalize on green opportunities (Dangelico et al. 2017).

Second, SCD facilitates the dynamic optimization of enterprises’ green operational positioning. By providing data processing and analytical capabilities, digitalization enables enterprises to assess internal resources and energy consumption (Han and Wei 2025), thereby improving the efficiency of resource allocation between supply and demand (Jiang et al. 2025). This allows enterprises to strategically allocate resources to key areas, such as green technology R&D and sustainable production processes, thereby helping them establish more precise green transformation objectives (Gu 2025; Jiang et al. 2025). Consequently, internal resources are better aligned with external green market demand, enhancing both the adaptability and competitiveness of enterprises in terms of sustainable development. Thus, we posit:

H2a: SCD promotes EGTP by enhancing ADC.

The mechanism role of absorptive capacity

Digitalization enhances enterprises’ capacity to absorb and transform knowledge and resources (Saeedikiya et al. 2025), serving as a bridge between digital and green transitions. First, SCD accelerates firms’ knowledge acquisition capabilities (Ngo et al. 2023; Schniederjans et al. 2020). To achieve green transformation, enterprises must adopt greener innovation technologies and resources (Ge et al. 2023). Digitalization in supply chains strengthens network connectivity, facilitating knowledge exchange and technology spillover among firms (Meng and Lin 2025). This enables enterprises to efficiently identify and acquire external green technological knowledge (Han and Wei 2025). Such capabilities help firms master the latest environmentally friendly technologies and best practices more swiftly during their green transition (Albort-Morant et al. 2018).

Second, absorptive capacity encompasses knowledge acquisition, knowledge transformation, and practical application. SCD provides enterprises with pathways to access external green technologies, such as renewable energy, clean production techniques, and energy-saving equipment (Nie et al. 2025). Firms integrate externally acquired green resources and technological knowledge with their green transformation objectives to update their internal knowledge. The use of digital technologies has accelerated the reconfiguration of internal resources and technologies (Schniederjans et al. 2020). Finally, digitalization optimizes corporate talent structures (Feng and Wang, 2024; Shen and Zhang, 2024), thereby enhancing absorptive capacity (Luo et al. 2024). Enterprises with higher absorptive capacities can better comprehend and adapt to process changes driven by digitalization (Yang et al. 2023), align new workflows with their green transformation strategies (Liao et al. 2024), and improve their green transition performance. Based on this, we propose:

H2b: SCD promotes EGTP by enhancing ABC.

The mechanism of innovation capacity

Innovation is multifaceted and emerges when enterprises implement a digital structural transformation (Saeedikiya et al. 2025). First, SCD enhances an enterprise’s willingness to innovate. By facilitating increased interaction among enterprises, customers, and suppliers, it provides access to external technological innovation resources and capabilities (Li et al. 2024; Luo et al. 2024). Moreover, digital technologies enable firms to rapidly acquire green technologies and market information, while effectively reducing transaction costs (Luo et al. 2024). The critical management data collected through SCD not only lower operational and innovation costs (Wang and Li 2024), but also alleviate financing constraints (Shen et al. 2025). These combined effects encourage enterprises to increase R&D investment and attract high-end talent, ultimately creating a positive feedback loop for innovation (Luo et al. 2024).

Second, SCD enhances an enterprise’s innovation capabilities in terms of technology, services, and processes. Specifically, it drives firms to adopt information technologies, such as big data and artificial intelligence, to achieve technological innovation (Mishra et al. 2024). Digital tools also improve customer service experiences and facilitate service innovation (Buyukozkan and Gocer 2018). Furthermore, digital technologies increase the flow efficiency of innovation elements across supply chains, fostering collaborative innovation among supply chain members (Shen et al. 2025). This enhances green supply chain coordination and governance capabilities, and accelerates environmental sustainability (Chen et al. 2025). Finally, SCD optimizes resource allocation and strengthens internal management, thereby promoting sustainable green innovation (Jiang et al. 2025). This enables breakthroughs in green product and process innovation (Dangelico et al. 2017), creating long-term, sustainable competitive advantages for enterprises. Based on this, we propose:

H2c: SCD promotes EGTP by enhancing INC.

The research framework of this study is presented in Fig. 1.

Fig. 1

Theoretical framework.

Research design

Model construction

Policy effect evaluation studies primarily rely on traditional models, such as the difference-in-differences model. However, these methods face challenges including model specification bias and constraints from linearity assumptions. This is particularly problematic when the parallel trend assumption, a fundamental prerequisite, cannot be satisfied, potentially leading to biased estimates if applied directly. Compared with fixed-effects models and instrumental variable methods for estimating treatment effects, double machine learning (DML) has distinctive advantages in causal inference. In this study, on the one hand, EGTP is influenced not only by SCD but may also be affected by multiple factors at the firm level, industry characteristics, and policy environments. Therefore, it is crucial to control these confounding variables (Wang and Cheng 2024). DML effectively addresses this challenge by employing machine learning algorithms and regularization techniques to select a set of high-precision control variables from a set of predetermined high-dimensional variables. This approach achieves simultaneous control of high-dimensional covariates while effectively avoiding the negative impacts of the curse of dimensionality and multicollinearity on the estimation accuracy (Chernozhukov et al. 2018).

On the other hand, complex nonlinear relationships often exist among enterprise-level micro variables. The use of traditional linear regression models for estimation may lead to model specification errors, thereby compromising the reliability of the research findings. By contrast, the DML method overcomes the limitations of linear relationships between variables and captures their true associations more accurately, thereby improving the precision of causal inference. This approach not only ensures an unbiased estimation of treatment effects but also effectively addresses common machine learning issues, such as regularization bias and overfitting, through techniques such as cross-fitting and orthogonalization (Wang and Cheng 2024). Given these advantages, this study adopts the framework proposed by Chernozhukov et al. (2018) to construct a partially linear DML model to investigate the mechanism by which SCD influences corporate green transformation. The specific procedure is as follows.

$$EGT{P}_{it}={\theta }_{0}SC{D}_{it}+g({X}_{it})+{U}_{it},E({U}_{it}|SC{D}_{it},{X}_{it})=0$$

(1)

In Eq. (1), i represents the enterprise, t represents the year, and $EGT{P}_{it}$ represents the enterprise green transformation performance. The variable $SC{D}_{it}$ represents the supply chain digitization, and ${\theta }_{0}$ corresponds to the coefficient $SC{D}_{it}$ estimated using machine learning methods. ${X}_{it}$ represents a multidimensional control variables and the specific form $\hat{g}({X}_{it})$ of $g({X}_{it})$ needs to be estimated using machine learning algorithms, where ${U}_{it}$ represents the error term with a conditional mean of zero. The direct estimation of Eq. (1) can be obtained coefficient estimates ${\hat{\theta }}_{0}$. Directly applying machine learning algorithms to estimate Eq. (1) yields a biased estimator ${\hat{\theta }}_{0}$ for ${\theta }_{0}$.

$${\hat{\theta }}_{0}={\left(\frac{1}{n}{\mathop{\sum}\limits_{i\in I,t\in T}SC{D}_{it}}^{2}\right)}^{-1}\frac{1}{n}\mathop{\sum}\limits_{i\in I,t\in T}SC{D}_{it}(EGT{P}_{it}-\hat{g}({X}_{it}))$$

(2)

In Eq. (2), where n represents the sample size and I represents the population observations, the estimator introduces bias as ${\theta }_{0}$ converges to ${\hat{\theta }}_{0}$. The estimation bias is given by.

$$\begin{array}{c}\sqrt{n}({\hat{\theta }}_{0}-{\theta }_{0})={\left(\frac{1}{n}{\mathop{\sum}\limits_{i\in I,t\in T}SC{D}_{it}}^{2}\right)}^{-1}\frac{1}{\sqrt{n}}\mathop{\sum}\limits_{i\in I,t\in T}SC{D}_{it}{U}_{it}\\ +\,{\left(\frac{1}{n}{\mathop{\sum}\limits_{i\in I,t\in T}SC{D}_{it}}^{2}\right)}^{-1}\frac{1}{\sqrt{n}}\mathop{\sum}\limits_{i\in I,t\in T}SC{D}_{it}[g({X}_{it})-\hat{g}({X}_{it})]\end{array}$$

(3)

In Eq. (3), $a={\left(\frac{1}{n}{{\sum}_{i\in I,t\in T}SC{D}_{it}}^{2}\right)}^{-1}\frac{1}{\sqrt{n}}{\sum }_{i\in I,t\in T}SC{D}_{it}{U}_{it}$ follows a normal distribution with a mean of 0, and $b={\left(\frac{1}{n}{{\sum }_{i\in I,t\in T}SC{D}_{it}}^{2}\right)}^{-1}\frac{1}{\sqrt{n}}{\sum }_{i\in I,t\in T}SC{D}_{it}[g({X}_{it})-\hat{g}({X}_{it})]$ represents the regularization bias. Under these conditions, the convergence rate of $\hat{g}({X}_{it})$ to $g({X}_{it})$ is ${n}^{-{\varphi }_{g}}$, where ${\varphi }_{g}$ < 1/2. To correct the regularization bias, orthogonalization is applied to $SC{D}_{it}$ and constructs an auxiliary regression model as follows.

$$SC{D}_{it}=m({X}_{it})+{V}_{it},E({V}_{it}|{X}_{it})=0$$

(4)

In Eq. (4), $m({X}_{it})$ represents the regression function of the treatment variable on the high-dimensional control variables. Similarly, its specific functional form $\hat{m}({X}_{it})$ needs to be estimated using machine learning algorithms, where $V_{it}$ denotes the error term with a conditional mean of zero.

Specifically, first, a machine learning algorithm is employed to estimate the auxiliary regression $\hat{m}({X}_{it})=E(SC{D}_{it}|{X}_{it})$, yielding the orthogonalized regressor ${\hat{V}}_{it}=SC{D}_{it}-\hat{m}({X}_{it})$, which removes the influence of the control variables. Here, ${\hat{V}}_{it}$ effectively serves as an instrumental variable for $SC{D}_{it}$. Second, the same machine learning approach was used to estimate $\hat{g}({X}_{it})$, and the primary regression equation was adjusted to $EGT{P}_{it}-\hat{g}({X}_{it})={\theta }_{0}SC{D}_{it}+{U}_{it}$, ultimately obtaining an unbiased coefficient estimator.

$${\hat{\theta }}_{0}={\left(\frac{1}{n}\mathop{\sum}\limits_{i\in I,t\in T}{\hat{{\rm{V}}}}_{{\rm{it}}}SC{D}_{it}\right)}^{-1}\frac{1}{n}\mathop{\sum }\limits_{i\in I,t\in T}{\hat{{\rm{V}}}}_{{\rm{it}}}(EGT{P}_{it}-\hat{g}({X}_{it}))$$

(5)

Similarly, Eq. (5) can be approximated as.

$$\begin{array}{lll}\sqrt{n}({\hat{\theta }}_{0}-{\theta }_{0})={[E({V}_{it}^{2})]}^{-1}\frac{1}{\sqrt{n}}\mathop{\sum}\limits_{i\in I,t\in T}{V}_{it}{U}_{it}\\\quad\quad\quad\quad\quad\quad\quad +\,{[E({V}_{it}^{2})]}^{-1}\frac{1}{\sqrt{n}}\mathop{\sum}\limits_{i\in I,t\in T}[m({X}_{it})-\hat{m}({X}_{it})][g({X}_{it})-\hat{g}({X}_{it})]\end{array}$$

(6)

In Eq. (6), ${[E({V}_{it}^{2})]}^{-1}\frac{1}{\sqrt{n}}{\sum }_{i\in I,t\in T}{V}_{it}{U}_{it}$ follows a normal distribution with a mean of zero. Here, $[m({X}_{it})-\hat{m}({X}_{it})][g({X}_{it})-\hat{g}({X}_{it})]$ converges at a speed of ${n}^{-(\varphi _{m}+\varphi _{g})}$, where $\varphi _{m}$ and $\varphi _{g}$ represent the convergence rates of m and g to $\hat{m}$ and $\hat{g}$, respectively, thereby ensuring an unbiased result.

Variable selection and measurement

The dependent variable is EGTP, which is measured by corporate green total factor productivity (Gao et al. 2024; Wu et al. 2022). This is because green total factor productivity integrates multiple dimensions, including economic output, resource utilization efficiency, and environmental impact, thereby reflecting production efficiency and environmental governance effectiveness. We employ the SBM-GML model to calculate corporate green total factor productivity (Wu et al. 2022), as this model effectively handles undesirable outputs and provides a more accurate assessment of real production efficiency under environmental constraints. The indicators used in the model are as follows: Capital input indicators include capital investment, fixed asset investment, and labor input, which reflect the allocation of production factors. The energy consumption indicator represents the total energy consumption of the city in which the firm is located. The desirable output indicator is measured by the enterprise’s main business revenue and serves as a proxy for economic output. Undesirable output indicators include industrial SO₂ emissions, industrial wastewater discharge, and industrial soot emissions as output measures.

The independent variable is SCD. The dummy variable interaction term (${{\rm{Treat}}}_{{\rm{i}},{\rm{j}}}\times {{\rm{Post}}}_{{\rm{i}},{\rm{t}}}$) is used to measure whether the interaction term (${{\rm{Treat}}}_{{\rm{i}},{\rm{j}}}\times {{\rm{Post}}}_{{\rm{i}},{\rm{t}}}$) is the product of the ${{\rm{Treat}}}_{{\rm{i}},{\rm{j}}}$ pilot and pilot time ${{\rm{Post}}}_{{\rm{i}},{\rm{t}}}$. If the registered address of the enterprise is a pilot city for supply chain innovation and application, the ${{\rm{Treat}}}_{{\rm{i}},{\rm{j}}}$ value is 1; otherwise, it is 0. Before 2018, the value of ${{\rm{Post}}}_{{\rm{i}},{\rm{t}}}$ was 0 during the pilot period, and it was 1 in 2018 and beyond.

In the selection of control variables, the DML method can effectively analyze high-dimensional control variables. To ensure the accuracy of the policy effect estimation, and based on references to relevant literature (Chen et al. 2025; Ge et al. 2023; Han and Wei 2025), macro-level variables such as environmental regulation, economic development, and industrial structure were selected, considering data availability. Meso-level variables such as industry competition intensity were also included. Finally, micro-level variables, including firm size, age, and managerial ownership ratio, were incorporated. Table 2 lists the measurement indicators for the relevant variables.

Table 2 Variable description.

Data sources and sample selection

Given the accessibility and completeness of the variable data, we selected Chinese A-share listed companies from 2013 to 2023 as the initial sample. This determination is driven by the current unavailability of complete 2024 core variable data, as key data sources from the 2024 China Environmental Statistical Yearbook have not yet been updated. During the sample screening process, financial firms, ST and *ST companies, and observations with missing key variables were excluded to ensure the reliability and validity of the findings. After filtering, we obtained an unbalanced panel dataset consisting of 3973 listed companies, covering 22047 firm-year observations. The data come primarily from the CSMAR database, Wind database, China Environmental Statistical Yearbook. To mitigate the influence of outliers, all continuous variables were winsorized at the 1% and 99% levels.

Empirical analysis

Descriptive statistics

Table 3 presents the descriptive statistics and correlation results for relevant variables. The variance inflation factor (VIF) for all variables was below 1.673, indicating no severe multicollinearity issues.

Table 3 Descriptive analysis of variables.

Baseline regressions

Table 4 presents the regression results of the DML analysis. Following the methodology of Gao et al. (2024), the random forest algorithm was employed for prediction with a sample splitting ratio of 1:4. Columns (1) and (2) of Table 4 display the estimation results for the partial linear and interactive models, respectively, incorporating the first-order terms of the control variables. Columns (3) and (4) present the corresponding results after including the second-order terms for the control variables. All baseline regressions are statistically significant at the 1% level, indicating that the SCD effectively enhances the EGTP. Thus, H1 is supported.

Table 4 Baseline regressions.

Robustness test

We employed the following robustness tests: First, we altered the measurement of EGTP by adopting the natural logarithm of green patent applications plus one, following Yuan et al. (2025). The regression results presented in Column (1) of Table 5 confirm that the findings remain robust even after modifying the measurement of the dependent variable. Second, we addressed the potential confounding effects of parallel policies. The selected sample might have been influenced by concurrent policies, such as the national big data comprehensive pilot zone, broadband China, and low-carbon city pilot initiatives. To mitigate these policy effects, we incorporated dummy variables for the pilot programs as control variables. The regression results, shown in Column (2) of Table 5, demonstrate that SCD continues to exert a significant positive impact on EGTP after accounting for concurrent policy interferences.

Table 5 Robustness test.

Third, we reconfigured the machine learning algorithms in the DML framework by replacing the previously used random forest algorithm with gradient boosting, ridge regression, support vector machines, and neural network algorithms. Columns (3) to (6) of Table 5 present the results. The regression outcomes in Table 5 demonstrate that, despite minor variations in the coefficients of the SCD impact on EGTP compared to the baseline regression, the results retain their statistical significance across all robustness checks. This reinforces the conclusion that SCD exerts a positive influence on EGTP, and further validates our findings.

Addressing endogeneity

To mitigate potential endogeneity issues, we employed propensity score matching (PSM) and instrumental variable (IV) approaches.

(1) Given the non-random nature of the supply chain innovation pilot selection, sample selection bias could introduce endogeneity. We addressed this issue using PSM to reduce selection bias. Specifically, we treated the control variables as matching covariates, removed unmatched control group observations during the matching process, and estimated each firm’s probability of being selected as a supply chain innovation pilot using a logit model. We then applied nearest-neighbor 1:1 and 1:4 matching and kernel matching, and reanalyzed the matched samples. The regression results presented in Columns (1) to (3) of Table 6 show significantly positive coefficients, confirming that SCD continues to exert a statistically significant positive effect on EGTP after accounting for sample selection bias.

Table 6 Robustness test.

(2) This study employed the product of each city’s number of telephones at the end of 1984 and the previous year’s internet penetration rate as an instrumental variable (IV). The rationale for the IV selection is twofold. First, the number of telephones reflects the historical communication infrastructure in a firm’s location, which may influence subsequent SCD implementation, thus satisfying the relevance condition. Second, telephone services primarily provided traditional communication for the public and have gradually declined in usage, meaning they are unlikely to directly affect the EGTP, thereby meeting the exogeneity requirement. Because the telephone data are cross-sectional and cannot directly serve as an IV for panel data, we constructed a panel-compatible instrument by multiplying the 1984 telephone count by the lagged internet penetration rate at the city level. The IV regression results, shown in Column (4) of Table 6, demonstrate a significantly positive coefficient for the interaction terms. This confirms that, even after accounting for endogeneity concerns, SCD maintains its positive effect on EGTP, further supporting the robustness of our baseline regression results.

Heterogeneity analysis

PBV posits that organizational performance derived from practices varies because of heterogeneity in organizational structures and industry characteristics (Bromiley and Rau 2014). Accordingly, this study examines the differential effects of SCD on EGTP by analyzing two organizational characteristics (ownership type and supply chain integration) and two industry characteristics (industry competition intensity and industry pollution level) from both organizational and industrial perspectives.

(1) This study conducted a heterogeneity analysis based on ownership type. We categorized the samples into state-owned enterprises (SOEs) and non-state-owned enterprises (non-SOEs). Columns (1) and (2) of Table 7 present the subgroup regression results. Column (1) shows that, for SOEs, the regression coefficient of SCD on EGTP is 0.0010 (p = 0.135). Column (2) indicates that, for non-SOEs, the estimated coefficient is 0.0011 (p = 0.024). This suggests that implementing SCD in non-SOEs yields a more pronounced improvement in EGTP. This disparity may stem from SOEs’ stronger policy-driven motivations and resource advantages but relative deficiencies in managerial flexibility and market responsiveness. By contrast, non-SOEs facing greater market competition pressures are more inclined to leverage SCD to enhance EGTP to gain competitive market advantages (Zhong et al. 2023).

Table 7 Firm heterogeneity test.

(2) A heterogeneity analysis of supply chain integration was conducted. Supply chain integration is a critical capability that coordinates and reallocates resources to enhance efficiency and achieve a competitive advantage. By strategically leveraging digital technologies, organizations can strengthen their adaptability to evolving market conditions (Wang et al. 2024). This study measures supplier integration using the ratio of procurement expenditures from the top five suppliers to total annual procurement, and customer integration using the ratio of sales revenue from the top five customers to total annual sales. Supply chain integration is defined as the average of customer and supplier integration (Ma et al. 2024). Firms are then classified into high- and low-integration groups based on their annual industry median. Column (3) of Table 7 represents the low supply chain integration group, and the regression coefficient for the impact of SCD on EGTP is 0.0007 (p = 0.179). In Column (4), representing the high supply chain integration group, the coefficient is 0.0016 (p = 0.003). This discrepancy may arise because high supply chain integration fosters closer collaboration between firms and their core supply chain partners. The application of digital technologies accelerates data sharing and coordination across supply chain segments, thereby facilitating green development. By contrast, under low supply chain integration, corporate partnerships with supply chain actors become more fragmented. Even if firms possess digital capabilities, they may struggle to effectively incentivize green efforts among supply chain partners, thereby limiting the environmental benefits of digitalization (Han and Wei 2025).

(3) This study conducted a heterogeneity analysis of the intensity of industry competition. We measure industry competition using the Herfindahl-Hirschman Index (HHI) and divide the sample into high- and low-competition groups based on the HHI. Column (1) of Table 8 shows the high-competition group, where the regression coefficient for the impact of SCD on EGTP is 0.0019 (p = 0.000). Column (2) shows the low-competition group, with a coefficient of 0.0007 (p = 0.215). This discrepancy may stem from the fact that firms in highly competitive industries face greater market pressure to avoid obsolescence, compelling them to leverage digital technologies for green innovation more efficiently, thereby significantly enhancing their transformation performance.

Table 8 Industry heterogeneity test.

(4) This study conduct a heterogeneity analysis of the polluting industries. With reference to the classification method for polluting enterprises proposed by Meng and Lin (2025), we divided the sample into high and non-high pollution groups. Columns (3) and (4) of Table 8 present the regression results. Column (3) shows that the impact of SCD on EGTP is close to zero and statistically insignificant in high-pollution industries. In contrast, column (4) reveals a regression coefficient of 0.0014 (p = 0.007) for non-high-pollution industries. This indicates that SCD plays a more pronounced role in promoting the EGTP in industries with low pollution levels. A plausible explanation is that carbon emissions in non-high pollution industries primarily stem from auxiliary activities, such as logistics, office operations, and equipment functioning (Han and Wei 2025). Digitalization can directly reduce energy consumption by optimizing resource allocation and minimizing redundant transportation. In comparison, carbon emissions from high-pollution industries are concentrated in energy-intensive production processes (Li et al. 2021). Although digitalization enhances supply chain management efficiency, it cannot replace core fossil fuel-dependent production techniques, resulting in limited emission reduction effects.

Mechanism analysis

This study employs the causal inference mediation test method proposed by Jiang (2022) to examine the mediating roles of adaptive, absorptive, and innovation capacities in the relationship between SCD and EGTP. The random forest was retained as the predictive algorithm for the DML approach. The test results are presented in Table 9.

Table 9 Mechanism analysis.

(1) Adaptive capability (ADC) is measured using the negative values of the coefficient of variation of the three main expenditures of R&D, capital, and advertising of the sample enterprises in the year to reflect the flexibility of the enterprise’s resource allocation, and then measure the adaptability of the enterprises (Feng and Wang 2024; Yang et al. 2023). The negative transformation ensures that the direction of variation aligns with the adaptive capability, where greater flexibility corresponds to greater adaptability. Column (1) of Table 9 presents the results of the mechanism test for the effect of SCD on adaptive capability, confirming that SCD enhances this capability, thereby validating hypothesis H2a.

(2) Absorptive capability (ABC) is measured as the proportion of employees with a bachelor’s degree or higher (Peng et al. 2022; Yang et al. 2023). This metric was chosen because the impact of SCD on performance depends on human capital, and firms with stronger talent pools are better equipped to assimilate digital technologies (Jiang et al. 2025). Column (2) of Table 9 shows the mechanism test results for SCD effect on absorptive capability, demonstrating a positive influence and supporting hypothesis H2b. Finally, innovative capability (INC) is quantified by standardizing annual R&D intensity and the percentage of technical personnel (Yang et al. 2023). Column (3) of Table 9 reports the mechanism test results for SCD’s impact on innovative capability, indicating that SCD strengthens this capability, thus confirming hypothesis H2c.

Further analyses

SCD integrates suppliers and customers into a firm’s digital supply chain network (Sharma et al. 2022), providing channels for the transfer of information, knowledge, and technology among supply chain members. The adoption of SCD practices by focal firms may influence the environmental orientation of upstream and downstream enterprises throughout the supply chain. SCD enhances visibility and transparency, enabling firms to utilize digital tools to track environmental metrics in raw material procurement and production processes (Kosmol et al. 2019). Focal firms can directly monitor upstream suppliers’ environmental compliance through digitalization and embed green standards into procurement contracts, incentivizing suppliers to adopt green supply chain management and mitigate their environmental impacts (Sun et al. 2024). For instance, Apple mandates that its core suppliers achieve 100% clean energy usage, reinforcing their green initiatives, and potentially accelerating their transition toward sustainability. In contrast, for downstream customers, although focused enterprises can influence the market by designing and promoting green products, the green transformation of downstream customers relies on market demand and consumer awareness, which are usually beyond the direct influence of the enterprise’s digital system. Although core enterprises can provide green products and services, if the market demand for these products is insufficient, customers’ green transformation process will also be restricted.

To test these hypotheses, we collected data on the top five customers and suppliers of the focal firms. Following Guo et al. (2024), we constructed annual firm-customer (supplier) datasets. This study retained only samples where customers (suppliers) were A-share listed companies, excluding ST and *ST firms and observations with missing data, resulting in 722 valid observations after the screening. We used the focal firm’s SCD as the independent variable and the EGTP of upstream suppliers and downstream customers as the dependent variables in separate regression analyses. The results are presented in Table 10. Column (1) indicates that the implementation of SCD in the focus enterprises promoted the EGTP of suppliers. Column (2) shows that the implementation of SCD in the focus enterprises has no significant impact on downstream enterprises, confirming that the SCD in the focus enterprises has a more significant impact on the EGTP of suppliers.

Table 10 Supply chain spillovers.

Discussion

Discussion on the direct effects of SCD

The findings of study demonstrate that SCD significantly enhances the EGTP. These results reaffirm the notion that digital transformation facilitates green development (Wang et al. 2025). With the continuous innovation and application of digital technologies, digitalization has been recognized as a core pathway driving corporate green initiatives (Guo et al. 2024; He et al. 2023). Consistent with prior research, several studies indicate that SCD can reduce corporate carbon emissions (Han and Wei 2025; Wang et al. 2025), promote green technological innovation (Gu 2025; Jiang et al. 2025), and improve corporate energy efficiency (Li et al. 2025), all of which positively impact environmental performance. Our research advances these findings, as previous studies have primarily focused on the effects of internal corporate digital transformation on green transition, lacking direct evidence of the impact of SCD on environmental performance. Furthermore, compared with carbon emissions and green technology innovation, the EGTP more comprehensively reflects synergistic improvements across multiple dimensions, including resource utilization, environmental governance, and economic benefits.

Discussion on the mechanisms of dynamic capabilities

This study demonstrated that SCD can enhance EGTP through adaptive, absorptive, and innovative capacities. Consistent with the prior literature, Yang et al. (2023) found that digital transformation strengthens firms’ innovative and absorptive capacities, thereby improving low-carbon technological innovation. Additionally, Peng et al. (2022) argued that digitalization influences corporate green transformation via dynamic capabilities, though their study did not examine the specific dimensions of dynamic capabilities or provide an in-depth explanation of their internal mechanisms. Similarly, Yuan and Pan (2023) suggested that the adoption of digital technologies enhances supply chain dynamic capabilities and contributes to the development of a circular economy. Alyasein et al. (2025) argued that SCD enhances innovation capabilities and improves corporate agility. Previous studies suggest that dynamic capabilities play a crucial mediating role in the relationship between digitalization and performance. However, most of these studies focused on a single dimension of dynamic capabilities or treated them as a unified construct, failing to capture the distinct roles of multidimensional components. Adaptive capacity enables firms to respond swiftly to external market changes (Yang et al. 2023), absorptive capacity facilitates the internalization and application of green technologies, and innovative capacity drives long-term sustainable advantages (Saeedikiya et al. 2025), which collectively constitute dynamic capabilities. Our study reveals the differential and complementary effects of adaptive, absorptive, and innovative capacities on the relationship between SCD and EGTP. This multidimensional perspective provides a deeper theoretical framework for understanding the complex mechanisms by which digitalization drives corporate green transformation.

Discussion on heterogeneity

Heterogeneity analysis reveals that the impact of SCD on EGTP is more pronounced in non-SOE enterprises, firms with high supply chain integration, highly competitive industries, and non-high-polluting industries.

(1) SCD exerts a more significant effect on EGTP in non-SOEs than in SOEs. This finding aligns with Zhong et al. (2023), who argued that SOE management tends to adopt conservative strategies, lacking the initiative and motivation for green transformation. In contrast, non-SOEs enterprises exhibit greater sensitivity to technological innovation and efficiency improvements, making them more likely to leverage the SCD to facilitate the adoption of green technologies.

(2) The impact of SCD on EGTP is more pronounced in firms with high supply chain integration. Our findings align with those of Han and Wei (2025), who suggested that in highly integrated supply chains, the close linkages between various segments enable digitalization to consolidate resources more effectively, optimize processes, and reduce energy waste, thereby enhancing green performance. Moreover, high supply chain integration reflects strong collaborative relationships between firms and supply chain partners, which can actively moderate the relationship between SCD and innovation capability (Alyasein et al. 2025). Such integration also helps to establish robust connections with external stakeholders, addressing key challenges in digital transformation while meeting green expectations (Gao and Huang 2024).

(3) The impact of SCD on EGTP is more pronounced under conditions of high competition. This aligns with the findings of Wang et al. (2024), who argued that competitive industries amplify the effect of digitalization on green technology innovation. In highly competitive environments, firms must leverage differentiation and cost control to gain a competitive advantage. Consequently, enterprises increasingly rely on digitalization to reduce transaction costs, drive green innovation, and enhance resource utilization efficiency (Li et al. 2025). Moreover, intense competitive pressure compels firms to simultaneously maintain sustainable operations and cultivate differentiation through green initiatives to align with the preferences of both the public and capital markets (Deng et al. 2024).

(4) The impact of SCD on the EGTP is more pronounced in non-high-polluting industries than in high-polluting industries. Although counterintuitive, this finding is consistent with that of previous research. Wang et al. (2023) suggested that digital technology adoption can induce an energy rebound effect, where the green performance gains from digitalization in high-polluting firms may be offset by increased energy demand. Similarly, Han and Wei (2025) pointed out that non-polluting industries can achieve significant emission reductions by digitizing and optimizing indirect emission links, such as logistics and office work. However, the core emissions of high-polluting industries come from high-carbon locked production processes, and it is difficult to change their energy consumption by merely enhancing the SCD. Similarly, Guo et al. (2024) contend that, owing to the inherently high pollution nature of their production, such firms encounter greater obstacles in leveraging digital infrastructure for green transformation.

Discussion on the spillover effects of SCD

The spillover effect indicates that SCD positively impacts the environmental performance of upstream suppliers. Previous studies have presented divergent perspectives on the spillover effect. Han and Wei (2025) found that SCD reduces suppliers’ carbon emissions but has a minimal effect on customers. This view is supported by Bian and Luo (2025), who argue that customer digitalization promotes green innovation among suppliers. However, Guo et al. (2024) observed that digital infrastructure primarily stimulates the green transformation of downstream customers and has insignificant effects on upstream suppliers. Our research further substantiates that focal firms’ SCD exerts a more pronounced influence on upstream suppliers’ environmental performance.

Conclusion and implications

Conclusion

To further investigate the inconsistencies in the relationship between SCD and corporate environmental performance, this study integrates the PBV with dynamic capability theory to explore their interplay. Using a sample of Chinese listed companies from 2013 to 2023, we demonstrate that SCD has a significantly positive impact on EGTP. Notably, these positive effects are more pronounced in non-state-owned enterprises, firms with high supply chain integration capabilities, highly competitive industries, and non-high-polluting industries. Mechanistic analysis reveals that SCD enhances EGTP by strengthening adaptive, absorptive, and innovative capacities. Furthermore, SCD has positive spillover effects by improving the EGTP of upstream suppliers. This study provides novel theoretical insights into the academic literature on SCD and environmental performance.

Theoretical implications

This study has significant implications in these fields.

(1) This study overcomes the limitations of traditional research perspectives. Prior studies on corporate green transformation have predominantly focused on the impact of internal digitalization (Wang et al. 2023; Xu et al. 2023), overlooking the pivotal role of supply chain-level digitalization. In reality, corporate green transformation is not an isolated endeavor, but is closely interconnected with upstream and downstream supply chains. We demonstrate that SCD enhances EGTP, thereby extending previous research that concentrated on its effects on green innovation (Li et al. 2025) and carbon emission (Han and Wei 2025). Furthermore, unlike earlier studies that explained SCD through a resource-based view (Chatterjee et al. 2024; Wang and Prajogo 2024), this study adopts a PBV to confirm the positive impact of SCD on EGTP. This expands the environmental performance perspective of the SCD (Chen et al. 2025; Han and Wei 2025; Jiang et al. 2025). These findings provide policymakers with empirical evidence to assess the environmental benefits of SCD.

(2) The findings of the study enrich the mechanisms by which SCD influences EGTP, offering a significant addition to the literature on the interplay between digital transformation and sustainable performance. While prior studies have indicated that digitalization facilitates corporate green transformation by enhancing green technology innovation (Hou et al. 2023; Luo et al. 2024), alleviating financing constraints (Guo et al. 2024; Luo et al. 2024), and improving energy efficiency (Hou et al. 2023), the exploration of its intrinsic driving mechanisms remains limited. Technology adoption is a prerequisite for achieving sustainable performance; however, superior outcomes depend on effective interactions between technological resources and capabilities (Syed et al. 2022), which ultimately fosters green competitive advantage (Yang et al. 2023; Yuan and Cao 2022). By introducing dynamic capabilities into the study of the relationship between SCD and EGTP, this study provides a more robust theoretical foundation for understanding the synergy between digital transformation and sustainability in supply chains.

(3) This study confirms the existence of spillover effects on how SCD influences the EGTP of upstream enterprises, providing crucial empirical support for the theory of collaborative green development in supply chains. While prior theoretical frameworks predominantly focused on intra-firm green transformation or direct collaborative mechanisms between upstream and downstream enterprises, this study identifies digitalization as a factor that generates spillover effects within supply chains. This expands our understanding of how digitalization facilitates supply chain spillover effects during corporate green transformation processes (Guo et al. 2024).

Management implications

These findings provide insights for corporate managers and policymakers. For enterprises,

(1) Managers should prioritize digital infrastructure development as a core strategy for driving SCD. The success of digital transformation in supply chains relies on a robust digital infrastructure, including hardware, software, data platforms, and network architectures. Without these foundational elements, the application of digital technologies may remain superficial and fail to enhance green and sustainable performance genuinely. During the digital transformation of supply chains, enterprises should adopt a phased approach to ensure that the digital infrastructure aligns with their business needs. Additionally, managers must mitigate the energy rebound effects that may arise from SCD when making related investments in the future.

(2) Managers should prioritize the development of dynamic capabilities along with digital transformation to achieve superior green performance. In terms of adaptive capacity, enterprises can leverage digital technologies to analyze trends in carbon emission reduction policies, proactively plan clean energy transitions for production lines, and flexibly adjust production scales and resource allocation to respond swiftly to market demands. Regarding absorptive capacity, firms should establish internal digital knowledge management platforms and cultivate talent pools that specialize in digital technologies. Externally, they should facilitate learning exchanges with supply chain partners, to enhance tacit knowledge acquisition and dissemination. For innovation capability, companies should implement evaluation systems centered on the EGTP, accelerate the development of open innovation channels and supply chain collaborative innovation mechanisms, and translate innovation outcomes into quantifiable environmental benefits and competitive advantages.

(3) Enterprises should tailor their SCD strategies based on their intrinsic attributes and industry characteristics to achieve superior sustainable performance. Non-state-owned enterprises should leverage their decision-making agility to increase investment in SCD and accelerate the integration of green principles with digital technologies. Firms with high supply chain integration should capitalize on resource synergies and utilize digitalization to dismantle information silos across supply chain segments, thereby ensuring the seamless alignment of green standards and processes throughout the entire chain to enhance holistic green transformation efficacy. Enterprises operating in highly competitive industries must employ digitalized supply chains to swiftly detect shifts in market demand for sustainability, enabling the precise development of green products and services to secure market advantages through differentiated green offerings. For non-high-pollution industries, the relatively lower pressure for transformation presents an opportunity to proactively adopt digital supply chain strategies and embed digital solutions throughout the entire product life cycle, from design and production to recycling, to establish a sustainable low-carbon development model and enhance long-term competitiveness.

(4) The spillover effects of SCD on the EGTP of upstream enterprises present new opportunities for collaborative green transitions. Focus enterprises can establish digital supply chain platforms and implement green incentive policies to assume a leadership role, driving upstream enterprises to proactively align with green standards, advance internal process digitalization, and participate in joint supply chain innovation initiatives.

For governments, when formulating relevant policies, it is imperative to encourage and support SCD when formulating relevant policies. Such policies should explicitly highlight the significance and potential value of integrating digitalization with green transition performance in supply chains, thereby guiding enterprises to prioritize this dual relationship in their strategic planning. Furthermore, governments must pay close attention to disparities in the digitalization levels of enterprises within supply chains. The government should further enhance the establishment of pilot demonstration projects for supply chain innovations. Leveraging the leading and driving roles of pilot enterprises is essential for disseminating advanced practical experience in supply chain digitization. Additionally, proactive resource support should be provided to facilitate digital and green transformation of enterprise supply chains.

Limitations and future research

This study has several limitations.

(1) Regarding the boundary effect, given that the PBV theory emphasizes the potential moderating effects on the disparity between practices and performance, future research could examine the moderating roles of executives’ risk preferences and managerial myopia (Feng and Wang 2024; Zhang et al. 2024). Top management holds decision-making authority and resource allocation power in corporate strategic management, which significantly influences digital transformation. Additionally, this study investigated only the mediating role of internal dynamic capabilities. However, digitalization may also impact supply chain dynamic capabilities such as supply chain visibility (Yu et al. 2024), agility, and collaboration (Zhou et al. 2023). These factors may be critical for enhancing EGTP, warranting further exploration of their roles in future studies.

(2) The digital infrastructure serves as a critical foundation for the application of digital technologies and SCD (Nie et al. 2025). Prior research has demonstrated that digital infrastructure positively influences innovation capabilities (Saeedikiya et al. 2024; Tian and Lu 2023), digital technology adoption (Yang et al. 2023), green transformation (Guo et al. 2024), and renewable energy innovation (Song and Chen 2025). However, recent studies have presented different perspectives. For instance, Nie et al. (2025) argued that, as SCD intensifies, the contribution of digital technologies to net-zero emissions declines. This suggests that excessive digitalization may lead to technological redundancy and energy consumption, highlighting a “crowding-out effect” on technological synergy. Thus, future research should identify the critical threshold phenomenon of digital infrastructure to enhance the EGTP. Additionally, digital infrastructure functions as a boundary resource (Lior et al. 2020), warranting further exploration of how small and medium-sized enterprises can leverage these resources to achieve SCD.

(3) Regarding the selection of the research samples and methodologies, our study used data from China. However, significant disparities exist in digital development levels, policy environments, and economic structures across countries and regions (Nie et al. 2025). Future research should expand the sample scope to validate the cross-cultural applicability of our conclusions. Additionally, while our data originated from large-scale samples, employing case studies in future studies could elucidate the relationship between SCD and EGTP. By contrast, case studies enable an in-depth examination of transformation pathways in representative firms or industries, thereby uncovering the intrinsic evolutionary mechanisms through which SCD drives green transformation.