From Digitalization to Sustainability: Does Supply Chain Digitalization Enhance Corporate Green Transformation Performance?

Abstract

Supply chain digitalization (SCD) plays a critical role in accelerating corporate green innovation, reducing carbon emissions, and enhancing corporate green transformation performance (CGTP). Drawing on the practice-based view, this study examines how SCD influences CGTP and through which mechanisms, using data from Chinese A-share listed firms spanning 2013 to 2021. Applying a double machine learning approach, the results demonstrate that SCD significantly enhances CGTP. Further, heterogeneity tests reveal that this positive effect is more pronounced in firms exposed to higher technological uncertainty, led by executives with stronger green cognition, and operating in less competitive markets. Mechanism tests suggest that SCD enhances firms’ sensing, seizing, and reconfiguring capabilities, thereby facilitating CGTP. The findings enrich the understanding of digital transformation and sustainability linkages and provide practical insights for managers and policymakers seeking to leverage SCD-driven strategies to promote corporate green transformation and sustainable development.

1. Introduction

The world is facing unprecedented environmental challenges, prompting governments and international organizations to accelerate the transition toward a green economy. As the primary actors in economic activity, enterprises play a crucial role in this transformation. Corporate green transformation not only reflects firms’ social responsibility but also represents a strategic pathway to sustainable development and enhanced competitiveness [1]. Corporate green transformation refers to shifting corporate strategies, production processes, technologies, and management practices toward resource efficiency and environmental friendliness, thereby integrating energy conservation and emission reduction with economic growth [2,3]. With the introduction of the “dual carbon” goals, firms face increasing pressure as well as motivation to pursue green transition. With this process, the supply chain plays a pivotal role as it links production, distribution, and consumption. Notably, supply chain carbon emissions are estimated to be 26 times higher than firms’ direct operational emissions and account for approximately 75% of total industrial emissions [4]. Therefore, reducing carbon emissions across supply chains is critical for improving corporate green transformation performance (CGTP).

Digital technologies are profoundly reshaping supply chains, driving supply chain digitalization (SCD) and creating new opportunities for low-carbon transition. However, the existing literature provides mixed evidence regarding the effect of SCD on CGTP. Some highlight its positive impacts, suggesting that SCD optimizes resource allocation [5], enhances information integration among supply chain partners [6], and reduces information asymmetry. These improvements, in turn, significantly enhance the efficiency of green innovation [7,8], accelerate low-carbon transformation [9], and strengthen their ESG performance [10]. In contrast, other research warns that digitization may increase energy consumption [11,12], as the digital sector itself is resource-intensive. Heavy investments in digital technologies do not always yield the expected environmental benefits, giving rise to a “digitalization paradox” [13]. For instance, intelligent supply chains rely on numerous electronic devices and sensors that consume substantial electricity and may hinder green development [14]. These conflicting findings highlight the need for further investigation. Hence, the first research question of this study is: How does SCD affect CGTP?

Corporate green transformation is a complex and dynamically evolving process influenced by both internal resources and external environmental uncertainty [15]. Firms must continuously reconfigure resources and capabilities to effectively sense, seize, and transform opportunities and constraints—an idea consistent with dynamic capabilities theory [16]. As a key driver of corporate transformation [17], digitalization is profoundly reshaping business operations, production processes, and supply chain management [5]. To navigate this transformation effectively, firms need dynamic capabilities that enable them to sense, absorb, and integrate new technologies with existing resources [18], thereby achieving superior green transformation performance. Accordingly, the second research question is as follows: Can SCD enhance dynamic capabilities and thereby accelerate CGTP?

Moreover, the essence of SCD lies in the integration and application of digital technologies within supply chain management [19]. The effectiveness of SCD also depends on contextual factors. Its success hinges on alignment with firms’ internal technological environment, organizational decision-making processes, and external market dynamics [10]. Under high technological uncertainty, SCD can enhance information transparency, enabling firms to better navigate risks and exploit opportunities [6]. Executives’ environmental cognition also shapes strategic decisions and resource allocation for green digital transformation [20]. Furthermore, market competition can either incentivize or constrain SCD adoption, influencing green outcomes. Yet few studies have examined how these heterogeneous factors—technological uncertainty, executive green cognition, market competition—moderate the SCD-CGTP relationship. Thus, the third research question is as follows: Under varying conditions of technological uncertainty, executive green cognition, and market competition, how does SCD influence CGTP?

To address these questions, this study leverages China’s Supply Chain Innovation and Application Pilot Policy launched in 2018 (hereafter referred to as “Pilot Policy”) as a quasi-natural experiment for SCD. The policy covers 55 cities and 269 firms nationwide, encouraging pilot firms to establish collaborative supply chain platforms, promote digitalization and intelligent upgrades, and develop data-driven smart supply chains supported by big data technologies [6]. It also explicitly targets the creation of a green supply chain system encompassing all stages of production and operations to promote cost reduction, efficiency, energy conservation, environmental protection, and green development. Several studies have treated this policy as an exogenous event to assess the impact of SCD on firms’ green innovation, environmental performance, and ESG outcomes [1,21], providing a sound basis for identifying its influence on CGTP.

Drawing on the practice-based view (PBV), this study uses data from Chinese A-share listed companies from 2013 to 2021 and applies a double machine learning (DML) approach to examine the mechanisms and boundary conditions through which SCD affects CGTP. The results show that SCD significantly enhances CGTP, with dynamic capabilities serving as a key mediating mechanism. Moreover, the positive effect of SCD is more pronounced among firms facing higher technological uncertainty, exhibiting stronger executive green cognition, and operating in less competitive markets.

This study makes three main contributions. First, grounded in PBV theory, it empirically confirms that SCD enhances CGTP, enriching the theoretical understanding of SCD’s role in CGTP. Second, it identifies dynamic capabilities—especially sensing, seizing, and reconfiguring—as the key channels though which SCD facilitates CGTP, offering a capability-based explanation for digital sustainability performance. Third, the heterogeneity analysis reveals that technological uncertainty, executive green cognition, and market competition condition the effectiveness of SCD on CGTP, providing actionable insights for firms and policymakers to tailor digital transformation strategies to varying environmental contexts. Collectively, these findings advance the theoretical discourse at the intersection of digital supply chain management and green transformation, offering both valuable theoretical and practical implications for achieving sustainable and digitally enabled supply chain management.

2. Literature Review

2.1. Practice-Based View

This study adopts the PBV rather than the traditional resource-based view (RBV) as its theoretical foundation. The RBV conceptualizes SCD as a firm-specific digital resource—such as advanced digital technologies and data infrastructure—whose rarity, value and inimitability constitute sources of sustained competitive advantage [17,22,23]. However, this perspective tends to overlook the importance of SCD as a set of imitable and transferable practices embedded in firms’ daily operations [24]. In contrast, PBV highlights that firms enhance their performance through the accumulation and refinement of routine, repetitive, and replicable practices [25]. Viewing SCD as a set of organizational practices rather than merely a technological resource provides a more nuanced understanding of its role in corporate green transformation. SCD-related practices—such as digital information sharing, data-driven decision-making, and intelligent collaboration—though enabled by digital technologies, fundamentally represent behavioral routines that are continuously enacted, refined, and diffused among organizational members [26]. These learnable and imitable practices can promote green innovation, resource optimization, and process reconfiguration [27], thereby advancing corporate green transformation.

From this perspective, PBV captures the dynamic mechanisms of capability formation that RBV often overlooks. The replicable and diffusive nature of SCD practices suggests that the development of green capabilities depends not only on internal resource endowments but also on continuous learning and experiential transmission through practice [28]. Moreover, such practices evolve through ongoing iteration and institutional embedding, reflecting how external policies, industry standards and market norms shape organizational behavior. Consequently, PBV transcends the static, resource-based logic of RBV and better explains the dynamic, practice-oriented processes underlying green transformation. Furthermore, the current study employs Pilot Policy as a quasi-natural experiment to capture SCD. This policy requires firms to undertake a series of SCD-related practices, such as developing intelligent supply chain systems, promoting digital transformation and upgrading activities [1,29]. These practices are inherently learnable and transferable, and replicable, aligning closely with PBV’s assumption regarding the diffusion and institutionalization of practices. Therefore, PBV provides an appropriate theoretical lens for explaining how firms enhance CGTP through the implementation, learning and refinement of SCD practices.

2.2. Supply Chain Digitization and Corporate Green Performance

SCD refers to the integration of digital technologies into supply chain activities to improve operational efficiency across functions such as demand forecasting, research and development, production and logistics. This integration gives rise to a new paradigm—digitally driven supply chain—that reshapes traditional supply chain management models [30]. Existing studies on this topic primarily focus on two interrelated strands: (1) the determinants of corporate green transformation, and (2) the influence of SCD on corporate green transformation.

From the perspective of determinants, prior research has identified both internal and external drivers of corporate green transformation. Internally, corporate culture [15], human and technological resources [31], and digitalization [32,33] are key enablers. Externally, institutional support, government policy, and the broader socio-economic environment significantly shape firms’ green transformation strategies [34,35,36].

A growing body of literature explores the relationship between SCD and corporate green performance, encompassing dimensions such as ESG performance, green technological innovation, and carbon emission reduction. Empirical evidence suggests that SCD enhances internal operational efficiency, facilitates supply chain information transparency, and strengthens external supervision, thereby improving ESG outcomes [21]. Furthermore, SCD improves supply chain financing capacity, alleviates financial constraints, attracts investor attention, and raises environmental awareness, which in turn promote green technological innovation [19,37,38]. By integrating digital technologies, firms can strengthen internal controls, accelerate technological innovation, and optimize financing structures, collectively contributing to lower carbon emissions [5,39,40]. Moreover, SCD enables firms for supplier integration and collaboration, fostering greater resource efficiency and higher levels of green investment [6]. It also promotes supply chain resilience, improving corporate green productivity [23].

Nevertheless, SCD may also generate adverse environmental effects. From a resource crowding-out perspective, the adoption of digital technologies demands substantial investments in capital, technology, and human resources. Such intensive investments may divert resources away from environmental management or green projects, thereby constraining short-term improvements in environmental performance [22]. Similarly, from an energy rebound perspective, digitization, particularly through data centers and artificial intelligence training, entails high energy consumption, potentially offsetting its carbon emissions [41,42]. For example, digital infrastructure deployment [43] and intelligent manufacturing [14] have been associated with increased carbon emissions. Furthermore, the digitalization paradox highlights that investments in digital technologies do not always yield the anticipated performance outcomes [44]. Improper technology implementation, poor management, or incompatibility with existing systems can reduce efficiency and lead to higher energy use or electronic waste, generating adverse environmental impacts [45]. Consequently, even as firms invest heavily in digital transformation, they may face heightened environmental pressures rather than relief.

2.3. Research Gap

Although an emerging body of research has explored the relationship between SCD and CGTP, several significant gaps remain [23]. First, most existing studies are grounded in the RBV, emphasizing firm-specific digital resources such as technological infrastructure and data assets. However, research adopting the PBV remains limited. This theoretical bias constrains a deeper understanding of how SCD, as a set of transferable and imitable practices, contributes to a firm’s green transformation and sustainable performance improvement. Second, the existing literature predominantly examined specific facets of environmental performance, such as ESG outcomes, carbon emissions, or green innovation, rather than assessing firms’ overall CGTP in a systematic manner. Consequently, the holistic influence of SCD on the broader process of corporate green transformation remains underexplored. Third, the internal mechanisms through which SCD affects green performance have been narrowly investigated. Existing studies primarily highlight technological innovation or financial factors, while the role of internal organizational capabilities—such as sensing, seizing, and reconfiguring capacity—has received comparatively little attention. Moreover, the impacts of SCD may vary across technological, organizational, and environmental contexts, yet empirical research providing cross-contextual or heterogeneity analysis remains sparse. Finally, from a methodological standpoint, most empirical investigations rely on traditional statistical techniques such as structural equation modeling or multiple regression analysis. While informative, these approaches are limited in addressing complex causal relationships and potential confounding factors. As a result, the true effects of SCD on CGTP may be underestimated or obscured by confounding factors.

To address these gaps, this study employs a quasi-natural experimental design using China’s SCD policy pilots as an exogenous shock and draws on a double DML approach to systematically examine the impact of SCD on CGTP and explore its underlying mechanisms. This approach allows for the identification of causal relationships while effectively controlling for multiple confounding factors. It also uncovers the key pathways through which SCD drives CGTP, thereby providing more robust and evidence-based support for theoretical advancement, management decisions and sustainable policy development.

3. Hypothesis Development

3.1. The Interplay Between SCD and CGTP

A core tenet of the PBV is that firms’ capabilities and competitive advantages arise not merely from the possession of unique resources but from the specific practices developed through daily operations and interactions within the organization [24]. These practices—formed through repetition, learning, adaptation, and knowledge sharing—reflect how organizational members collectively generate and embed routines that enhance firm performance in specific contexts. In the context of green transformation, SCD can be conceptualized as a critical organizational practice that fosters CGTP. SCD encompasses a set of replicable and adaptive digital practices that enable firms to optimize resource use, promote innovation and strengthen collaboration across the supply chain. First, SCD enhances firms’ green innovation capability. By digitizing supply chain processes, firms accumulate digital assets in technology, data and talent [23]. Firms with richer digital resources can facilitate more effective access to the frontier knowledge on environmental technologies and sustainable practices [46], thereby providing an impetus to advance green technological innovation [7].

Second, SCD contributes to resource optimization and environmental management. Digital technologies enable real-time tracking, monitoring and analysis of operational data throughout production and distribution processes [47]. Through precise data analytics, firms can identify sources of energy use, waste generation, and emissions [7], make timely adjustments to their green strategies, and promote resource recycling, circular use of materials and production efficiency [8], thereby significantly reducing energy consumption and improving overall environmental performance [48].

Third, SCD facilitates collaborative sustainability practices among supply chain partners. By integrating internal and external information flow, SCD enhances inter-organizational coordination, communication and transparency [49]. Such information integration and exchange enable supply chain stakeholders to jointly identify and address environmental challenges, thereby advancing green supply chain practices [50]. Moreover, as digital supply chains evolve, they gradually form interconnected and interdependent networks which enhance firms’ capabilities to monitor carbon-related behaviors, map carbon footprints, and ultimately strengthen green supply chain management [51].

In summary, by improving green innovation, resource efficiency, and supply chain collaboration, SCD plays an essential role in advancing corporate green transformation. Thus, we posit:

H1. 

SCD positively influences CGTP.

3.2. The Role of Dynamic Capability

According to PBV, organizational practices influence firm performance through both mediating and moderating effects [25]. Dynamic capability theory emphasizes that firms must continually sense opportunities and threats, seize them through strategic and operational actions, and reconfigure resources to sustain competitive advantage in changing environments [16]. Building on this perspective, this study proposes that the sensing, seizing, and reconfiguring dimensions of dynamic capability mediate the relationship between SCD and CGTP. The underlying mechanisms are elaborated as follows.

Sensing capability refers to a firm’s ability to perceive changes in external markets, technological advancements, and shifts in customer demand with agility and accuracy [16,18]. First, SCD enhances a firm’s ability to integrate real-time environmental and operational data and establish early warning systems. Through the application of technologies such as the Internet of Things (IoT) and blockchain, firms can obtain critical environmental information from suppliers in real time, including data on raw material production, energy consumption, and waste generation [7]. This transparent data flow enables firms to detect potential environmental risks and emerging green opportunities at an early stage [9]. Meanwhile, improved insight into environmental trends allows firms to better sense shifts in consumer preferences toward green products and sustainable services [52], enabling firms to proactively adjust their green strategies, incorporate eco-design principles into product and process innovation.

Second, digital collaboration platforms and information-sharing mechanisms enable firms to more effectively scan their business ecosystems and identify potential partners, technologies and resources for green innovation. This includes locating suppliers and customers capable of providing eco-friendly inputs, sharing environmental technologies, or co-developing sustainable solutions [53]. Through SCD, firms can identify and integrate supply chain partners with strong green capabilities [49], thereby improving the overall environmental performance of the supply chain. In addition, digital simulation tools enable firms and their partners to jointly evaluate the environmental impacts, benefits and potential risks of new green production processes, as well as optimize them before implementation [54]. Such mechanisms facilitate joint learning and risk sharing across the supply chain, significantly enhancing firms’ sensitivity and environmental responsiveness and foresight. Accordingly, we propose:

H2a. 

SCD facilitates CGTP by strengthening firms’ sensing capability.

Seizing capability refers to a firm’s ability to take swift and effective action upon recognizing new opportunities, translating them into innovative products, processes, or business models [16,55]. Within the context of green transformation, SCD strengthen this capability by facilitating rapid resource integration, collaborative innovation and operational agility. First, SCD enhances the rapid integration of green technologies and resources. Through the establishment of digital supplier relationship management systems [17], firms can efficiently identify and connect with suppliers possessing specialized green technologies or environmentally sustainable production capabilities. This ability allows firms to promptly capture and leverage external green resources, transforming them into competitive advantages in the development of green products and services.

Second, SCD promotes collaborative innovation among supply chain partners. By strengthening information sharing and coordination across the supply chain [56], SCD facilitates joint research, co-development, technological practice sharing, and knowledge exchange across supply chain partners. Through such digitally enabled collaboration, firms can seize cross-organizational opportunities for green transformation and collectively advance sustainable practices [57].

Finally, SCD enhances supply chain flexibility and agility, enabling firms to adopt adaptive, on-demand green manufacturing models that improve the efficiency of new product development and innovation [58]. This capability allows firms to respond rapidly to evolving environmental regulations and consumer demands. As a result, firms become better positioned to capitalize on green market opportunities in dynamic and uncertain environments. Thus, we posit:

H2b. 

SCD facilitates CGTP by strengthening firms’ seizing capability.

Reconfiguring capability refers to a firm’s ability to fundamentally adjust and realign its organizational structures, processes, and business models in response to perceived opportunities and acquired resources [16,18]. Within the context of green transformation, SCD functions as a crucial enabler for reshaping firms’ sustainability strategies and operational frameworks. First, SCD empowers firms to continuously assess and adapt sustainability strategies. Through IoT technologies and big data analytics, firms can obtain real-time information about environmental and demand changes across the supply chain, allowing them to reconfigure and adapt internal capabilities more swiftly [59] and demonstrating greater organizational flexibility [18]. Furthermore, SCD supports the development of transparent and traceable circular supply chains, allowing firms to track product lifecycle data, optimize reverse logistics, promote waste recycling, and shift production processes from linear to circular models [60]. These changes enhance resource utilization efficiency and improve environmental performance.

Second, SCD drives innovation in green business models. By leveraging digital platforms, firms can extend product lifecycles, minimize resource consumption [7], and restructure their value creation logic to achieve sustainable competitive advantages [27]. Additionally, SCD enhances supply chain transparency and visibility by providing real-time data and advanced analytical capabilities. This improvement not only strengthens firms’ ability to manage and recover from green-related risks but also enhances overall supply chain resilience [30]. Consequently, firms are better equipped to sustain superior green performance [23]. Therefore, we posit:

H2c. 

SCD facilitates CGTP by strengthening firms’ reconfiguring capability.

3.3. SCD and CGTP Under the Heterogeneous Conditions

According to PBV, differences in firm performance are influenced by a company’s ability to implement operational practices, as well as by institutional and industry-specific factors [25]. Accordingly, this study investigates how SCD affects CGTP under heterogeneous conditions, focusing on technological uncertainty, executive green cognition, and market competition.

At the technical level, uncertainty shapes how firms adopt and apply emerging green technologies. Under conditions of high technological uncertainty, enterprises encounter rapidly evolving technological environments and unpredictable market demands. When technological pathways remain unclear, comprehensive and accurate information becomes crucial for identifying the most likely pathways to achieve green goals [9]. SCD significantly enhances the transparency of internal and external information through real-time data sharing, integrated information systems, and advanced data analytics [49]. This transparency enables firms to quickly identify opportunities for green innovation, assess environmental risks, and optimize their transformation decisions [38]. Furthermore, high technological uncertainty often accompanies substantial risks associated with R&D investment. Digitalization mitigates these risks by improving data-driven decision-making and facilitating collaborative innovation among supply chain stakeholders [49]. Through such digital collaboration, firms gain access to diverse external knowledge sources and transform them into capabilities for green development. Thus, we posit:

H3a. 

Under conditions of higher technological uncertainty, the positive impact of SCD on CGTP is stronger.

At the organizational level, executives’ green cognition—their awareness, understanding and prioritization of environmental sustainability—plays a decisive role in shaping internal green practices and capability development. Executives with strong environmental awareness tend to embed sustainability principles into corporate strategy and operational practices. During the process of SCD, such executives prioritize the adoption of digital technologies that directly support green goals and allocate the corresponding resources [23]. For instance, these executives are more inclined to invest in digital systems that minimize energy consumption and waste generation rather than focusing solely on operational efficiency. Their deeper understanding of the potential synergies between digital transformation and green innovation and environmental management enables them to translate environmental awareness into concrete organizational practices [61]. Moreover, executives’ green cognition helps shape an organizational culture that values green transformation [62], effectively communicating a green vision, reducing internal resistance, and motivating employees to adopt green digital processes when promoting SCD. This cultural alignment in turn facilitates the widespread implementation of digital practices in the supply chain, thereby achieving overall green transformation efficiency. Thus, we posit:

H3b. 

Under conditions of higher executives’ green cognition, the positive impact of SCD on CGTP is stronger.

At the environmental level, market competition constitutes a critical external condition for corporate green transformation, influencing firms through market pressures and differences in policy responses. Both SCD and green transformation demand substantial investments in financial, technological, and human resources. However, firms’ willingness to invest varies significantly under different levels of market competition. In markets with lower competition intensity, firms experience lower survival pressure and enjoy greater financial flexibility and strategic autonomy. These conditions enable them to invest in SCD more confidently [6]. As a result, such firms are more likely to treat SCD as a strategic lever for enhancing long-term environmental performance. Conversely, in highly competitive markets, firms are more likely to concentrate resources on projects that deliver immediate results, such as cost control and market share expansion, which leads to insufficient investment in long-term and high-risk initiatives like digitalization and green transformation [63]. Accordingly, we propose:

H3c. 

Under conditions of lower market competition, the positive impact of SCD on CGTP is stronger.

Based on the above hypothesis, we develop a research conceptual framework as presented in Figure 1.

4. Research Design

4.1. Sample Selection

This study selected A-share listed companies in China as the research sample. These firms constitute a vital component of the Chinese economy and are characterized by comprehensive financial disclosures and relatively standardized supply chain information, thereby providing a reliable empirical basis for analysis. Moreover, A-share companies span a wide range of industries, enabling the investigation of heterogeneity in SCD and CGTP across different industrial contexts. Data were obtained from multiple authoritative sources. Specifically: (1) SCD data were derived from the matching results of pilot policy datasets and the CSMAR database. (2) CGTP data were sourced from the China Environmental Statistics Yearbook (https://www.stats.gov.cn) and the CSMAR database. (3) Micro-level control variables (e.g., firm size, leverage, and profitability) were extracted from the CSMAR database. (4) Macro-level control variables, such as regional economic development and industrial structure, were obtained from the China Environmental Statistics Yearbook. To ensure data reliability and consistency, the following preprocessing steps were undertaken: (1) Financial enterprises and ST/*ST firms were excluded to avoid bias arising from abnormal financial structures and potential delisting risks. (2) Observations with missing or incomplete data were removed. (3) All continuous variables were winsorized at the 1st and 99th percentiles to mitigate the influence of extreme values.

4.2. Variable Definition

4.2.1. Explained Variable

CGTP is measured using green total factor productivity (GTFP), a widely recognized indicator of evaluating corporate green transformation [64]. Green transformation requires firms to sustain or enhance output levels under environmental constraints by advancing technologies, upgrading production processes, and optimizing factor allocation, thereby reducing resource consumption and environmental costs [23]. GTFP explicitly incorporates undesirable outputs such as pollution emissions and energy consumption into the traditional total factor productivity framework. This integration endogenizes the efficiency of “emission reduction and productivity enhancement” within the production frontier, making GTFP particularly suitable for capturing the dynamic transition from “high consumption and high emissions” to “low consumption and low emissions” [65,66]. Following Wu et al. [66], this study employs the Slack-based Measure Global Malmquist-Luenberger index to estimate CGTP. The model simultaneously accounts for desirable and undesirable outputs, including capital investment, fixed asset investment, labor investment, and urban energy consumption, operating revenue as the desirable output and including industrial SO₂ emissions, wastewater discharge, and soot emissions as undesirable outputs. This approach allows for the quantification of firms’ actual production efficiency under environmental constraints.

4.2.2. Explanatory Variable

SCD is measured using an interaction term (${\mathrm{Treat}}_{\mathrm{i},\mathrm{j}}\times {\mathrm{Post}}_{\mathrm{i},\mathrm{t}}$), which is the product of the dummy variable for the pilot status (${\mathrm{Treat}}_{\mathrm{i},\mathrm{j}}$) and the dummy variable for the pilot time (${\mathrm{Post}}_{\mathrm{i},\mathrm{t}}$). The variable ${\mathrm{Treat}}_{\mathrm{i},\mathrm{j}}$ takes a value of 1 if a firm’s registered location is in a Pilot Policy, and 0 otherwise. The variable ${\mathrm{Post}}_{\mathrm{i},\mathrm{t}}$ is assigned a value of 0 before 2018 and 1 for 2018 and subsequent years.

4.2.3. Control Variable

To reduce omitted-variable bias and ensure robust estimation, this study includes a set of control variables drawn from prior empirical research on digital transformation and corporate environmental performance [35,67,68]. The controls encompass firm-level and regional characteristics that may simultaneously influence both SCD and CGTP. The firm-level variables include firm size, firm age and managerial ownership ratio. At the regional level, regional economic development is controlled to account for differences in digital infrastructure and environmental policy enforcement across provinces.

Table 1. Variable description.

Variable	Metrics	Abbreviation	Measurement
Dependent variable	Corporate green transformation performance	CGTP	Corporate green total factor productivity
Independent variable	Supply chain digitalization	SCD	$({\mathrm{Treat}}_{\mathrm{i},\mathrm{j}}\times {\mathrm{Post}}_{\mathrm{i},\mathrm{t}}$)
Control variable	Firm size	SIZE	Ln (Total assets)
	Asset liability ratio	LEV	Total liabilities/Total assets
	Financial performance	ROA	Net profit/Total assets at beginning of period
	Tobin’s Q value	TOBINQ	Market value of the company/Cost of asset replacement
	Cash holdings	CASH	Net Cash Flow from Operating Activities/Total assets
	Fixed asset ratio	FIXED	Fixed assets/Total assets
	Growth ability	GROWTH	Operating revenue growth rate
	Listing age	AGE	Ln (2021—Company’s listing year)
	Management shareholding ratio	MSHARE	Proportion of total shares owned by directors and supervisors
	Industry concentration degree	HHI	Summed quadratic values of individual enterprises’ market proportions
	Economic development level	EDV	Natural logarithm of GDP per capita
	Industrial structure	INS	Service sector to manufacturing sector ratio

provides the measurement indicators for all the variables used in this study.

4.3. Model Specification and Estimation Method

Traditional policy evaluation studies often employ the Difference-in-Differences (DID) method. However, DID’s inherent model specification bias and linear constraints may lead to estimation errors, particularly when the parallel trends assumption is not strictly satisfied [69]. To address these limitations, this study adopts the DML approach, which offers distinct advantages in causal identification. By incorporating machine learning techniques and regularization algorithms, DML automatically selects effective control variables from a pre-specified high-dimensional set, yielding a subset with high predictive accuracy. This approach not only mitigates the “curse of dimensionality” arising from redundant controls but also alleviates bias caused by a limited number of key control variables [70]. Furthermore, due to the strengths of machine learning algorithms in handling nonlinear data, DML effectively avoids model misspecification issues that often arise when conventional linear regressions are used to capture nonlinear relationships among variables. DML employs cross-fitted machine learning models to examine the relationship between the treatment and outcome variables, thereby producing residuals free from confounding effects. These residuals are then incorporated into the regression model to estimate causal effects. This two-stage analytical procedure substantially reduces specification errors caused by omitted variables or incorrect model structures, thus minimizing the causal inference bias frequently observed in conventional regression analyses.

This study focuses on the determinants of CGTP. In addition to SCD, it is necessary to control for confounding factors such as firm heterogeneity, industry characteristics, and policy environment [1,5]. More importantly, firm-level microdata often exhibits complex nonlinear patterns that traditional linear regressions struggle to capture accurately. Therefore, following Chernozhukov et al. [70], this study employs the Double Machine Learning framework to examine the impact of SCD on CGTP. First, a partially linear model is constructed as follows:

$$CGT{P}_{it}={\theta }_{0}SC{D}_{it}+g\left(X_{it}\right)+U_{it},E\left(U_{it}|SC{D}_{it},X_{it}\right)=0$$

(1)

In Equation (1), i represents the firm, t denotes the year, ${CGTP}_{it}$ stands for corporate green transformation performance, ${SCD}_{it}$ indicates supply chain digitalization, and ${\theta }_0$ is the estimated coefficient of the policy dummy variable. $X_{it}$ represents the multidimensional control variables, whereas the specific form of $g(X_{it})$, denoted as $\widehat{g}\left(X_{it}\right)$, needs to be estimated using machine learning algorithms. $U_{it}$ is the error term with a conditional mean of zero. Direct estimation of Equation (1) produces a biased result.

$${\widehat{\theta }}_0={\left({\displaystyle \frac{1}{n}}{{\displaystyle {\sum }_{i\in I,t\in T}SC{D}_{it}}}^2\right)}^{-1}{\displaystyle \frac{1}{n}}{\displaystyle {\sum }_{i\in I,t\in T}SC{D}_{it}}\left(CGT{P}_{it}-\widehat{g}\left(X_{it}\right)\right)$$

(2)

In Equation (2), with n defined as the sample size, given that ${\theta }_0$ converges to ${\widehat{\theta }}_0$, we can further examine its estimation bias.

$$\sqrt{n}\left({\widehat{\theta }}_0-{\theta }_0\right)={\left({\displaystyle \frac{1}{n}}{{\displaystyle {\sum }_{i\in I,t\in T}SC{D}_{it}}}^2\right)}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}SC{D}_{it}}{U}_{it}+{\left({\displaystyle \frac{1}{n}}{{\displaystyle {\sum }_{i\in I,t\in T}SC{D}_{it}}}^2\right)}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}SC{D}_{it}}\left[g\left(X_{it}\right)-\widehat{g}\left(X_{it}\right)\right]$$

(3)

In Equation (3), let $a={\left({\displaystyle \frac{1}{n}}{{\displaystyle {\sum }_{i\in I,t\in T}{SCD}_{it}}}^2\right)}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}{SCD}_{it}}{U}_{it}$, which is normally distributed with a zero mean, and $b={\left({\displaystyle \frac{1}{n}}{{\displaystyle {\sum }_{i\in I,t\in T}{SCD}_{it}}}^2\right)}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}{SCD}_{it}}\left[g\left(X_{it}\right)-\widehat{g}\left(X_{it}\right)\right]$. Given the high-dimensional model specification, it is necessary to introduce a regularization term. This approach, while preventing excessive variance in the estimator, may introduce bias in the estimation of $g\left(X_{it}\right)$. At this point, the convergence rate of $\widehat{g}\left(X_{it}\right)$ towards $g\left(X_{it}\right)$ is $n^{-{\phi }_g}$, where ${\phi }_g$ < 1/2. To address this issue, an orthogonal method must be introduced to correct for bias.

Specifically, the first regression Formula (4) is used to obtain the estimated ${\widehat{V}}_{it}={SCD}_{it}-\widehat{m}\left(X_{it}\right)$ of the residual term V:

$$SC{D}_{it}=m\left(X_{it}\right)+V_{it},E\left(V_{it}|X_{it}\right)=0$$

(4)

Here, ${\stackrel{⌢}{V}}_{it}$ is equivalent to the instrumental variable of $SC{D}_{it}$, and the final estimate of ${\theta }_0$ is as follows:

$${\widehat{\theta }}_0={\left({\displaystyle \frac{1}{n}}{\displaystyle {\sum }_{i\in I,t\in T}{\widehat{V}}_{it}SC{D}_{it}}\right)}^{-1}{\displaystyle \frac{1}{n}}{\displaystyle {\sum }_{i\in I,t\in T}{\widehat{V}}_{it}}\left(EGT{P}_{it}-\widehat{g}\left(X_{it}\right)\right)$$

(5)

The estimation error at this time is as follows:

$$\sqrt{n}\left({\widehat{\theta }}_0-{\theta }_0\right)={\left[E(V_{it}^2)\right]}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}{V}_{it}}{U}_{it}+{\left[E(V_{it}^2)\right]}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}\left[m(X_{it})-\widehat{m}\left(X_{it}\right)\right]}\left[g\left(X_{it}\right)-\widehat{g}\left(X_{it}\right)\right]$$

(6)

In Equation (6), ${\left[E(V_{it}^2)\right]}^{-1}{\displaystyle \frac{1}{\sqrt{n}}}{\displaystyle {\sum }_{i\in I,t\in T}{V}_{it}}{U}_{it}$ follows a normal distribution with a mean of 0. Because two error terms are introduced, the product of $\left[m\left(X_{it}\right)-\widehat{m}\left(X_{it}\right)\right]\left[g\left(X_{it}\right)-\widehat{g}\left(X_{it}\right)\right]$ converges at the speed of $n^{-({\phi }_m+{\phi }_g)}$. Here, ${\phi }_m$ and ${\phi }_g$ respectively represent the convergence rates of m and g to $\widehat{m}$ and $\stackrel{⌢}{g}$. Although the convergence speeds of both are relatively slow, the error term converges to zero more quickly after being transformed into interaction terms, thereby ensuring the unbiasedness of the coefficient estimates.

5. Empirical Findings

5.1. Summary Statistics

Table 2. Descriptive analysis.

Variable	N	Mean	SD	Min	Max	VIF
CGTP	16,634	0.999	0.016	0.918	1.082	—
SCD	16,634	0.340	0.474	0.000	1.000	1.147
SIZE	16,634	22.240	1.249	19.910	26.480	1.916
LEV	16,634	0.404	0.193	0.046	0.901	1.689
ROA	16,634	0.042	0.068	−0.363	0.250	1.687
TOBINQ	16,634	2.182	1.389	0.818	11.420	1.270
CASH	16,634	0.050	0.063	−0.150	0.252	1.362
FIXED	16,634	0.199	0.140	0.004	0.689	1.154
GROWTH	16,634	0.170	0.328	−0.552	2.499	1.154
AGE	16,634	2.108	0.758	0.693	3.367	1.526
MSHARE	16,634	0.158	0.293	0.000	20.170	1.180
HHI	16,634	−0.798	0.245	−1.020	−0.216	1.075
INS	16,634	1.136	0.523	0.284	3.397	1.090
EDV	16,634	10.890	0.509	9.275	12.100	1.026

presents the summary statistics and correlation analysis. The VIF values for all predictors remained below 1.916, suggesting no significant multicollinearity concerns in the regression model.

5.2. Regression Results

The empirical analysis in this study was performed using Stata 18. Following Gao et al. [65], the random forest algorithm was employed as the primary machine learning method, with all hyperparameters set to their default values to ensure model stability and reproducibility.

Table 3. Baseline regressions.

Variable	(1)	(2)	(3)	(4)
	CGTP	CGTP	CGTP	CGTP
θ0	0.0024 *** (0.0007)	0.0049 *** (0.0007)	0.0023 *** (0.0007)	0.0045 *** (0.0006)
Control	Yes	Yes	Yes	Yes
Control2	No	No	Yes	Yes
Firms/Year FE	Yes	Yes	Yes	Yes
N	16,634	16,634	16,634	16,634

Note: Standard errors are in parentheses; *** p < 0.01.

summarizes the key regression results. Columns (1) and (2) report the estimates from partial linear model and the interaction model incorporating first-order control variables, respectively. Columns (3) and (4) present the regressions of the linear and interaction models, respectively, with the addition of the second-order control variables. The findings consistently show that SCD significantly improves CGTP, validating H1, confirming that SCD serves as an effective driver of firms’ green transformation.

5.3. Robustness Test

This study employed several robustness testing approaches. First, to ensure that the research conclusions were not affected by the measurement method selection, an alternative method for measuring green transformation performance was adopted. Following Shi and Geng [71], a composite green transformation performance index was constructed by integrating three indicators: CGTP, corporate green patent grants, and the environmental (E) score of ESG ratings, using the entropy weight method. The regression results for the reconstructed dependent variable are presented in Column (1) of

Table 4. Robustness test result.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	CGTP	CGTP	CGTP	CGTP	CGTP	CGTP
θ0	0.0139 *** (0.0052)	0.0023 *** (0.0007)	0.0012 *** (0.0004)	0.0051 *** (0.0003)	0.0019 *** (0.0004)	0.0025 *** (0.0005)
DML model	RF	RF	GBDT	SVM	Lasso	NN
Control/Control2	Yes	Yes	Yes	Yes	Yes	Yes
Firms/Year FE	Yes	Yes	Yes	Yes	Yes	Yes
N	16,634	16,634	16,634	16,634	16,634	16,634

Note: Standard errors are in parentheses; *** p < 0.01.

.

Second, to control for the potential confounding effects of concurrent policies, relevant policy variables were incorporated as additional controls. Because the sample period overlaps with the implementation of both the National Big Data Comprehensive Pilot Zone and Broadband China initiatives, dummy variables were included for these policies in the baseline regression model. The results presented in Column (2) of Table 4 indicate that the main findings remain unchanged.

Third, the random forest algorithm was substituted with gradient boosting, support vector machines, lasso regression, and neural network models. Columns (3) through (6) of Table 4 presented the results. These robustness checks collectively demonstrate the stability of the main regression results, providing further empirical support for Hypothesis H1.

5.4. Endogeneity Test

First, an endogeneity test was conducted for sample selection bias. Given the Pilot Policy’s non-random nature, potential endogenous issues may arise from sample selection. To address this selection bias, the PSM-DID method was employed. Initially, the control variables were used as matching variables for the kernel matching. During the matching process, control group firms that failed to match were excluded. The probability of each firm being selected for the supply chain innovation pilot project was estimated using a logit model.

Table 5. Descriptive statistics of treatment and control groups.

Variable	Treatment Group					Control Group
	Count	Mean	Sd	Min	Max	Count	Mean	Sd	Min	Max
CGTP	9658	1.000	0.017	0.918	1.082	6976	0.998	0.015	0.918	1.068
SIZE	9658	22.282	1.295	19.914	26.477	6976	22.174	1.180	19.914	26.477
Lev	9658	0.409	0.195	0.046	0.901	6976	0.398	0.190	0.046	0.901
ROA	9658	0.041	0.068	−0.363	0.250	6976	0.043	0.069	−0.363	0.250
TobinQ	9658	2.203	1.407	0.818	11.422	6976	2.154	1.362	0.818	11.422
CASH	9658	0.047	0.063	−0.150	0.252	6976	0.054	0.064	−0.150	0.252
FIXED	9658	0.178	0.137	0.004	0.689	6976	0.228	0.140	0.004	0.689
GROWTH	9658	0.172	0.333	−0.552	2.499	6976	0.166	0.320	−0.552	2.499
AGE	9658	2.092	0.764	0.693	3.367	6976	2.130	0.748	0.693	3.367
MSHARE	9658	0.165	0.345	0.000	20.171	6976	0.149	0.199	0.000	1.692
HHI	9658	−0.792	0.249	−1.020	−0.216	6976	−0.806	0.240	−1.020	−0.216
INS	9658	1.193	0.581	0.284	3.397	6976	1.056	0.416	0.284	3.397
EDV	9658	10.865	0.493	9.275	12.076	6976	10.915	0.529	9.275	12.101

presents the descriptive statistics for the treatment and control groups. Regression analysis was subsequently applied to the matched pairs. The statistically significant positive estimate in Column (1) of

Table 6. Endogeneity test result.

Variable	(1)	(2)
	CGTP	CGTP
θ0	0.0013 *** (0.0005)	0.0143 * (0.0074)
Control/Control2	Yes	Yes
Firms/Year FE	Yes	Yes
N	16,617	15,884

Note: Standard errors are in parentheses; * p < 0.1, *** p < 0.01.

suggests that Hypothesis H1 remains robust to selection bias.

The instrumental variable (IV) is constructed as the product of the number of landline telephones per 100 inhabitants in each city in 1984 and the lagged broadband Internet users per 100 inhabitants at the prefecture-level city. The rational is twofold. First, the historical prevalence of landline telephones captures the legacy of regional communication infrastructure, which could shape subsequent SCD. This satisfies the relevance condition of the IV. Second, given that landline telephones primarily served basic communication needs and have lost practical significance in digital era, they are unlikely to exert any direct impact on CGTP, thereby meeting the exogeneity requirement. Furthermore, because the landline data are cross-sectional and thus unsuitable for direct use in panel analysis, a panel-compatible IV was developed by interacting landline penetration rates with lagged broadband adoption levels. The IV regression results, reported in Column (2) of Table 6, show that the coefficient of SCD remains significantly positive. This finding confirms that the main conclusion—SCD significantly enhances CGTP—remains robust even after addressing potential endogeneity concerns, thereby reinforcing the validity of Hypothesis H1.

5.5. Heterogeneity Analysis

To further explore the boundary conditions of the relationship between SCD and CGTP, this study conducts a series of heterogeneity analyses based on technological uncertainty (TU), executives’ green cognition (EGC), and market competition (MC).

First, following Huang and Mu [72], technological uncertainty is measured and firms are classified into low and high groups according to annual industry medians.

Table 7. Heterogeneity test.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Low TU	High TU	Low EGC	High EGC	Low MC	High MC
θ0	0.0011 (0.0007)	0.0021 ** (0.0008)	0.0011 (0.0009)	0.0022 *** (0.0009)	0.0022 ** (0.0008)	−0.0006 (0.0007)
Control/Control2	Yes	Yes	Yes	Yes	Yes	Yes
Firms/Year FE	Yes	Yes	Yes	Yes	Yes	Yes
N	7814	8820	9815	6819	8258	8376

Note: Standard errors are in parentheses; ** p < 0.05, *** p < 0.01.

reports the regression results. In Column (1), representing firms with low technological uncertainty, the coefficient of SCD on CGTP is 0.0008 and statistically insignificant. In contrast, Column (2) reports the results for firms operating under high technological uncertainty, where the coefficient of SCD on CGTP is 0.0021 (p < 0.01). The finding indicates that the positive effect of SCD on CGTP is more pronounced when firms face greater technological uncertainty, thereby supporting Hypothesis H3a.

Second, EGC is measured following Zhou et al. [73]. Firms are divided into low-EGC and high-EGC groups based on annual industry median values. The corresponding regression results are displayed in Columns (3) and (4) of Table 7. For firms with low EGC, the coefficient of SCD on CGTP is 0.0011, which is statistically insignificant. Conversely, for firms with high EGC, the coefficient of SCD on CGTP is 0.0022 (p < 0.05). The result suggests that SCD exerts a stronger positive impact on CGTP when executives demonstrate higher environmental awareness, lending empirical support to Hypothesis H3b.

Finally, following Deng et al. [74], MC is measured using the industry Lerner index, which inversely reflects the intensity of market competition. Firms are categorized into high-competition (lower Lerner index) and low-competition (higher Lerner index) groups based on the annual industry median. The results are shown in Columns (5) and (6) of Table 7. In less competitive markets, the coefficient of SCD on CGTP is 0.0021 (p < 0.05), while in highly competitive markets, the coefficient is −0.0006 and statistically insignificant. These findings indicate that the positive effect of SCD on CGTP is more substantial in markets characterized by lower competitive intensity, thereby supporting Hypothesis H3c.

5.6. Mechanism Test

The study utilizes the mechanism-testing approaches of Wang et al. [5] and Li et al. [75] to analyze the mechanisms through which sensing, seizing, and reconfiguring capacities mediate the relationship between SCD and CGTP. The constructed model is as follows.

$$Me{d}_{it}={\theta }_{0}SC{D}_{it}+g\left(X_{it}\right)+U_{it},E\left(U_{it}|SC{D}_{it},X_{it}\right)=0$$

(7)

$$CGT{P}_{it}={\theta }_{0}SC{D}_{it}\times Me{d}_{it}+g\left(X_{it}\right)+U_{it},E\left(U_{it}|SC{D}_{it}\times Me{d}_{it},X_{it}\right)=0$$

(8)

In Models (7) and (8), $Me{d}_{it}$ represents the dynamic capability. First, following Kindermann et al. [76], we use digital orientation as a proxy for a firm’s sensing capability. Digital orientation reflects a firm’s adoption of digital technologies and tools, data analytics capabilities, and responsiveness to market and environmental dynamics [76]. These factors collectively capture a firm’s ability to identify opportunities and threats. Following Fan et al. [18], we quantify digital orientation by identifying digital-related keywords in corporate annual reports and calculating the logarithm of their word frequency plus one. As shown in Column (1) of

Table 8. Mechanism analysis.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Sensing	CGTP	Seizing	CGTP	Reconfiguring	CGTP
SCD	0.1894 *** (0.0236)		0.0206 *** (0.0065)		0.0608 * (0.0364)
SCD × Sensing		0.0004 *** (0.0001)
SCD × Seizing				0.0020 * (0.0012)
SCD × Reconfiguring						0.0001 ** (0.0000)
Control/Control2	Yes	Yes	Yes	Yes	Yes	Yes
Firms/Year FE	Yes	Yes	Yes	Yes	Yes	Yes
N	16,634	16,634	13,961	13,961	13,411	13,411

Note: Standard errors are in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01.

, the regression analysis reveals that SCD exerts a significant positive effect on sensing capability. Column (2) reveals that the interaction term coefficient between SCD and sensing capability is 0.0004 (p < 0.01), confirming that sensing capability serves as an effective mechanism through which SCD improves CGTP, thereby supporting H2a.

Second, following Yang et al. [77] and Fan et al. [18], we measure seizing capabilities using standardized annual R&D intensity and the proportion of technical personnel. These indicators capture firms’ capacity to transform potential opportunities into innovative outcomes. Higher R&D investment enhances technological advancement, while a greater share of technical personnel strengthens firms’ ability to evaluate and capitalize on emerging technological and market opportunities [18]. Column (3) of Table 8 presents the regression results of SCD on seizing capability and shows that SCD significantly enhances a firm’s seizing capability. Column (4) reveals that the interaction term coefficient between SCD and seizing capability is 0.0020 (p < 0.1), confirming that seizing capability serves as an effective mechanism through which SCD improves CGTP, thereby supporting H2b.

Finally, following Luo et al. [78], the logarithm of the sum of corporate investment and intangible assets is used to represent reconfiguration capability, as these resources reflect a firm’s potential to flexibly adjust and reallocate assets to support strategic transformation. The results in Column (5) of Table 8 present the regression results of SCD on reconfiguration capability, indicating that SCD significantly enhances a firm’s reconfiguration capability. Column (6) reveals that the interaction term coefficient between SCD and reconfiguration capability is 0.0001 (p < 0.05), confirming that reconfiguration capability serves as an effective mechanism through which SCD improves CGTP, thus supporting H2c.

6. Discussion

6.1. Relationship Between SCD and CGTP

Answering RQ1, how does SCD affect CGTP? The empirical results demonstrate that SCD exerts a significant positive influence on CGTP, thereby confirming H1. This finding is consistent with the growing body of literature linking SCD to sustainable development outcomes. Previous studies on firms’ environmental implications of SCD present mixed evidence, including both optimistic and skeptical perspectives [7,14,60]. The critical view suggests that SCD may entail potential drawbacks, such as substantial upfront investments, data security risks, and potential increases in energy consumption or carbon emissions, which could undermine short-term environmental performance. However, the results of this study corroborate more recent empirical findings that highlight the net positive effect of SCD on firms’ green transformation efforts [6,23].

The plausible explanation for this positive relationship is that SCD, through the optimization of information flows and data analytics, enhances firms’ ability to identify design and develop green products, processes, and business models, thereby significantly promoting green innovation [38,49]. Moreover, SCD facilitates more transparent and efficient management of environmental footprints and social responsibilities [21]. It also enables more efficient resource allocation and leaner production, leading to increased green investments [6] and green productivity [23]. Therefore, it can be confirmed that advancing SCD significantly enhances CGTP. These findings deepen our understanding of the digital–green nexus in supply chains and offer valuable implications for both scholars and practitioners seeking to advance sustainable supply chain management.

6.2. Discussion on the Dynamic Capabilities

Answering RQ2, does dynamic capability mediate the impact of SCD on CGTP? The empirical findings confirm that the three dimensions of dynamic capabilities—sensing, seizing, and reconfiguring—serve as important mediating mechanisms through which SCD promotes CGTP. This study’s distinctive contribution lies in its more granular exploration of how each dimension of dynamic capability operates within the digit-green nexus. While prior studies have generally emphasized that SCD enhances firms’ innovation potential, absorptive capability, adaptability, and supply chain resilience, thereby facilitating low-carbon transformation or improving green productivity [9,23], this research provides more targeted evidence. It empirically verifies that the three core dynamic capabilities function as distinct and critical yet complementary channels through which SCD drives corporate green transformation.

Specifically, the results show that SCD significantly enhances firms’ sensing capabilities, enabling them to more effectively identify potential opportunities and threats during the process of green transformation, thereby supporting H2a. Digital technologies and platforms allow for comprehensive data collection, integration, and real-time analysis across the entire supply chain [18]. Through these tools, firms can monitor environmental indicators, such as emissions and energy consumption in real time [60], while big data analytics enhance market sensing by tracking changes in consumer preferences for green products and evaluating the maturity of emerging green technologies [49]. These data-driven insights provide firms with timely and precise market intelligence, allowing them to anticipate and adapt to shifts in both consumer demand and regulatory landscapes. This aligns closely with the sensing dimension of dynamic capabilities theory, highlighting how SCD creates an informational foundation for proactive and informed corporate green transformation strategies.

Moreover, SCD substantially improves firms’ seizing capabilities, confirming H2b. Beyond identifying green opportunities, digitalization empowers firms to capture and exploit them more efficiently and effectively. Digital tools enable firms to conduct scenario analyses and impact simulations for alternative green transformation strategies, thereby facilitating evidence-based decision-making on investment [18]. In parallel, blockchain technologies enhance transparency and traceability across supply chains [79], enabling firms to verify supplier compliance with green standards and ensure the authenticity of green products. Additionally, SCD supports process optimization and accelerates the design and development of green products [7]. Collectively, these mechanisms allow firms to transform perceived opportunities into concrete investments and innovations, embodying the “seizing” function of dynamic capabilities in the digital era.

Finally, the findings demonstrate that SCD enhances firms’ reconfiguring capabilities, which are critical sustaining green transformation, thereby supporting H2c. Reconfiguring capability enables firms to flexibly realign resources, structures and strategies in response to evolving demands of green transformation [9]. Consistent with prior research, this study finds that digitally enabled supply chains are more agile in reallocating resources and restructuring supplier networks, which enhances supply chain resilience and green performance [23]. Furthermore, SCD drives organizational restructuring and business process reengineering to align with emerging green business models. It also facilitates deeper resource integration and collaboration with supply chain partners, supporting the eco-design and simulation of innovative and sustainable new business processes. Through these pathways, SCD strengthens firms’ ability to reconfigure both tangible and intangible assets, ultimately increasing their green investment and transformation capacity [6].

6.3. Discussion on Heterogeneity

Answering RQ3, how does the impact of SCD on CGTP vary across different technological, organizational, and environmental contexts? The heterogeneity analysis reveals that the positive effect of SCD on CGTP is more pronounced under conditions of higher technological uncertainty, stronger executive green cognition, and lower market competition, thereby supporting H3a, H3b, and H3c, respectively.

This finding aligns with previous research suggesting that in rapidly changing technological environments, firms face heightened uncertainty, which compels them to actively acquire new knowledge and adopt emerging technologies and skills in order to quickly introduce new products [72]. High technological uncertainty often entails information asymmetry and incompleteness, which hinder effective decision-making. SCD mitigate these challenges by providing advanced data analytics and information-sharing platforms that effectively breaks down traditional information barriers, facilitates data flow and transparency along the supply chain, and enhances inter-organizational coordination [80]. Consequently, firms are increasingly inclined to employ SCD to address these challenges. Moreover, under high technological uncertainty, firms must develop stronger sensing capabilities to identify emerging green technologies and market trends. SCD supports this by enabling firms to leverage advanced analytical tools, such as deep learning algorithms, to extract insights from vast amounts of unstructured data, forecast technological trajectories, and thereby strengthen their ability to sense and seize green technology opportunities [9].

In addition, executives green cognition plays a vital role in amplifying the effectiveness of SCD in driving firms’ green transformation Prior studies have shown that executives’ environmental awareness and strategic commitment to sustainability are critical determinants of firms’ green performance [81,82]. Executives with strong green cognition are more likely to promote the deep integration of digital technologies into corporate green strategies, directing technological applications toward optimizing resource efficiency and reducing emissions, while prioritizing investment in green innovation [73]. Such leaders are more inclined to allocate resources to green operations, sustainable supply chain practices, and the development of green technologies. Consequently, when executives green cognition is high, SCD’s contribution to CGTP is magnified.

Finally, the analysis shows that the positive impact of SCD on CGTP is stronger in markets with lower competition intensity. This observation is consistent with prior studies [63], which suggest that firms operating in less competitive environments can focus more on internal management optimization and low-carbon technological innovation without being overly constrained by short-term competitive pressures [83]. In such markets, firms typically possess more abundant resources and a longer-term strategic vision, allowing them to invest more effectively in SCD and green transformation initiatives [5]. Lower competition also reduces the need for short-term financial returns, enabling firms to better leverage the data-driven insights provided by SCD to sense and seize long-term green opportunities.

7. Conclusions and Contributions

7.1. Conclusions

The rapid advancement of digital technologies is transforming traditional supply chains into intelligent and digital supply chains. However, whether SCD fosters green development and generates positive environmental impacts remains underexplored. Drawing on the PBV theory, this study investigates the impact of SCD on CGTP using panel data from Chinese listed companies spanning the years 2013 to 2021 and employs the DML method for empirical validation. The findings reveal that SCD significantly enhances CGTP. Mechanism analysis further demonstrates that this improvement occurs through strengthened sensing, seizing, and reconfiguring capabilities. Moreover, the heterogeneity analysis indicates that SCD has a stronger facilitative effect on CGTP in scenarios characterized by high technological uncertainty, high executive green cognition, and low market competition.

7.2. Theoretical Implications

This study offers several important theoretical contributions. First, it expands the theoretical boundaries for understanding SCD by introducing the PBV perspective. Existing research has primarily relied on the RBV [17,84] and dynamic capabilities theory [5] to interpret SCD, highlighting its positive effects on operational performance [85], supply chain resilience [30], and financial performance [17]. However, few studies have examined the relationship between SCD and green performance from the PBV perspective. Unlike prior studies that regard SCD merely as a digital resource [22], this research emphasizes the dynamic and practice-oriented nature of SCD, challenging the static resource assumption of RBV and uncovering the practice-driven mechanisms that may have been overlooked within that framework. This perspective not only enriches the theoretical connotation of SCD but also provides a more nuanced and dynamic explanatory framework for understanding how SCD fosters CGTP through replicable and adaptive practices [39].

Second, this study advances the literature by identifying dynamic capabilities as the core mechanism linking SCD and CGTP. While prior studies generally highlighted how SCD enhances supply chain resilience and thereby influences corporate greening levels [23], this research explicitly demonstrates that SCD enhances firms’ sensing, seizing, and reconfiguring capabilities, thereby facilitating green transformation. This approach extends the dynamic capabilities framework in the digital context by showing how SCD enhances firms’ strategic flexibility and responsiveness to environmental challenges. By integrating these capabilities into the analytical framework of SCD and green transformation, this study addresses a critical gap of previous research that inadequately explained the underlying mechanisms of digitalization [5,21].

Finally, this study identifies technological uncertainty, executives’ green cognition, and market competition intensity as key contextual contingencies that condition the effectiveness of SCD. This underscores that SCD’s influence is not universal but context-dependent. Whereas previous studies have mainly examined variations in SCD’s influence on green capability across ownership structures, life-cycle stages, industrial pollution levels, or economic development [6,23], this research offers new insights from technological, organizational, and environmental perspectives. These findings contribute to building a more generalizable theoretical framework for understanding the interaction between SCD and greening under varying contexts of both internal and external corporate environments.

7.3. Managerial Implications

This study also provides meaningful insights for practitioners and policymakers. For practitioners, several actionable recommendations emerge. First, firms should regard SCD as a strategic tool to support the green transition but implement it through pilot projects and phased rollouts. Given that the benefits of SCD are context-dependent, it is advisable to start with small-scale trials in selected business units or product lines to assess environmental and economic returns and then scale up gradually after confirming their effectiveness. Second, firms should prioritize capability building over treating SCD as a mere technological upgrade. This study finds that SCD can strengthen firms’ sensing, seizing, and reconfiguring capabilities. Accordingly, digital investments should be pursued in parallel with complementary measures, such as improving employee digital literacy, increasing R&D spending, and enhancing organizational flexibility, to ensure that technology investments ultimately convert into durable organizational capabilities and improved green performance. Finally, firms should tailor their SCD strategies to their specific circumstances. Under conditions of high technological uncertainty, weak market competition, or strong managerial environmental awareness, firms may appropriately increase resource allocation to SCD to better meet green transition requirements and respond to market changes.

For policymakers, several implications arise. First, policymakers should balance fiscal incentives with institutional support by continuously providing subsidies, tax incentives, or targeted grants to help firms build digital infrastructure, lower upfront digitalization costs, and encourage wider participation in SCD pilot projects. Second, they should offer normative guidance for the application of digital technologies in green production and supply chain management: advance carbon reduction monitoring and certification pilots in phases, progressively refine technical standards and evaluation metrics, enhance transparency, and incentivize firms to disclose green supply chain practices and performance. Third, differentiated implementation strategies that recognize SCD’s varying effectiveness across levels of technological uncertainty, managerial green awareness, and industry competition should be adopted. Policymakers should conduct longitudinal evaluations to identify scalable and transferable SCD models and design targeted support accordingly—for example, prioritizing executive training, experience sharing, and technical consulting for firms with weak managerial environmental awareness and devising cost-effective incentive schemes for highly competitive or resource-constrained industries.

7.4. Limitations and Future Research

Despite its contributions, this study has several limitations that suggest promising directions for future research. First, while the results confirm the positive effect of SCD on CGTP, prior studies suggest that excessive or poorly implemented digitalization can lead to diminishing returns or even increase emissions [60]. Future research could thus explore the dark side of SCD, including risks of digital overdependence or corporate greenwashing [86].

Second, although this study identifies dynamic capabilities as key mediating mechanisms, the measures used for sensing, seizing, and reconfiguring are indirect proxies based on secondary data. Future research could employ survey-based or mixed-method approaches to directly capture these capabilities and trace how they evolve over time. Longitudinal case studies could also provide richer insights into the co-evolution of SCD, dynamic capabilities, and green transformation. In addition, future research could further examine the mediating effects of specific supply chain dynamic capabilities, such as supply chain transparency [85], supply chain collaboration [87], and supply chain resilience [30].

Third, this study considers only the PBV and dynamic capability theories; future research could further integrate resource-based, stakeholder, and institutional theories to analyze how SCD influences CGTP. In addition, our study employs a dummy variable to measure the effect of SCD. Future research could further integrate methods such as machine learning and text analysis to quantify SCD [88] and explore the impact of the different dimensions of SCD on CGTP. For instance, it would be valuable to investigate whether the digitalization of procurement, production, or distribution plays a more critical role in this context.

Finally, while this study’s evidence from Chinese A-share listed firms demonstrates a strong positive link between SCD and CGTP, the results should be interpreted with caution due to contextual specificity. China’s institutional environment features strong government guidance and environmental enforcement, which may amplify the green benefits of SCD. Future research could conduct cross-country comparative analyses to test the generalizability of these findings under different market orientations and regulatory regimes. Examining how institutional quality, policy incentives, or cultural norms shape the SCD-CGTP relationship would deepen understanding of how SCD fosters green transformation in diverse contexts.