How Entrepreneurship Drives Digital Transformation: A Moderated Mediation Model Based on the Attention-Based View

Abstract

As a key component of sustainable economic development, digital transformation has become a fundamental driver for developing and upgrading the modern economic system. While existing research has identified resources and dynamic capabilities as foundational elements, a critical yet underexplored factor lies in the cognitive foundations that enable firms to strategically direct and leverage these assets. Based on 19,062 observation samples of more than 3000 listed companies in Shanghai and Shenzhen stock markets from 2010 to 2023, this paper constructs a theoretical framework of entrepreneurship, organizational attention and digital transformation from the Attention-Based View, and examines a moderated mediation model of the relationship between entrepreneurship and digital transformation. The results show that entrepreneurship significantly promotes digital transformation; organizational attention to “cooperation orientation” and “future orientation” plays a mediating role in it; and the regional innovation atmosphere positively strengthens the “cooperation orientation” path, facilitating the diffusion of innovative knowledge and technologies within the region. Meanwhile, online media reports negatively regulate the “future orientation” path, reflecting that short-term public pressure may weaken enterprises’ attention to long-term sustainable technology investment. In addition, different dimensions of entrepreneurship have varied effects on digital transformation. Heterogeneity analysis revealed significant variations across ownership type, scale, region, industry competition intensity, and technological intensity. This study expands the theoretical mechanism of entrepreneurship and digital transformation from the perspective of attention allocation, and provides theoretical and empirical foundation for fostering a strategic cognitive orientation and advancing digital transformation.

1. Introduction

With the accelerated integration of artificial intelligence, big data, and cloud computing, major economies around the world have regarded digital transformation as a key force driving economic growth. The Overall Layout Plan for Digital China Construction (2023) emphasizes that Digital China serves as a vital engine for advancing Chinese modernization in the digital era. Consequently, digital transformation has evolved from an “optional choice” to an “imperative task”, becoming a critical pathway for enterprises to enhance strategic flexibility and achieve sustainable competitive advantage [1,2].While digital transformation fuels the development and structural upgrading of the digital economy [3], it also confronts businesses with paradoxical challenges related to capability, collaboration, and governance. These largely stem from cognitive biases among entrepreneurs [4], who often treat digital transformation as a technological upgrade rather than a strategic shift. As a result, digital investments frequently fail to translate into sustainable advantage, reflecting a lack of deep understanding and strategic commitment to the essence of digital transformation.

Academic discussions on “what drives digital transformation and how to overcome its challenges” continue to deepen based on the Resource-Based View (RBV) or the Dynamic Capability Theory (DCT) [5,6,7,8,9]. While neither the RBV nor the DCT help to explain different outcomes through resource and capability, they rest on an implicit assumption: before firms can deploy resources or capabilities, entrepreneurial leaders must first identify digital opportunities amid complex information, view them as a critical driver for enhancing sustainable development capabilities, and maintain sustained focus on them. Despite extensive research on digital transformation grounded in resource-based and dynamic capability perspectives, limited attention has been paid to the cognitive and attentional mechanisms through which entrepreneurship translates into concrete digital strategic actions. This study directly addresses this gap by applying the Attention-Based View (ABV) to explain how entrepreneurial cognition shapes organizational attention and, in turn, drives digital transformation outcomes.

The ABV posits that organizations cannot attend to all aspects within complex information environments. Entrepreneurs must exercise selective attention regarding which issues deserve focus, how many resources to allocate, and for how long [10,11]; they then translate this attention into strategic actions through formal structures, communication channels, and resource allocation rules [12]. Research further found that the values and cognition of entrepreneurs determine the direction of attention allocation, thereby influencing the firm’s ultimate behavioral choices [13]. This implies that entrepreneurship can guide organizations to maintain long-term focus on the strategic significance and implementation pathways of digital transformation, particularly toward sustainable outcomes such as circular economy practices and socially inclusive business models.

This study situates the investigation within the distinctive context of the Chinese, where a nationally multi-dimensional construct of entrepreneurship encompasses patriotism, innovation, integrity, social responsibility, and global vision; this form of entrepreneurship can continually enhance an enterprise’s long-term resilience and value-creating capacity through the reshaping of business models and organizational structures [8,14]. Specifically, this paper posits that entrepreneurship serves as a powerful antecedent that reconfigures organizational attention along two pathways essential for digital transformation: a cooperation orientation and a future orientation. To further uncover the boundary conditions of this process, the study introduces the regional innovation atmosphere and online media reports as moderating variables to examine how institutional environments influence the effectiveness of these attentional pathways on digital transformation. Ultimately, we construct a moderated mediation model that systematically explains how entrepreneurship affects digital transformation through the selective allocation of managerial attention toward cooperation and future-oriented priorities, while revealing how external environmental and informational factors contextually moderate this process.

The theoretical contributions are mainly reflected in the following three aspects:

(1) This study shifts the focus from “Resources and Capabilities” to “Cognitive Antecedents”. Existing research about digital transformation are primarily based on the RBV and DCT, emphasizing the role of organizational resources and capabilities. This study introduces the ABV, focusing on the cognitive processes that precede strategic action [11,15]. It reveals how entrepreneurship, as a strategic cognition, drives digital transformation through the allocation of organizational attention, thereby uncovering a deeper underlying mechanism. Furthermore, drawing on the research regarding communication channels as a form of organizational attention [16], and Joseph et al. (2024), who synthesized the ABV literature and highlight that managerial attention toward cross-departmental relationships (Bartel and Rockmann, 2024) and long-term perspectives (Boynton, 2024) is crucial for enhancing strategic resilience and technological change effectiveness [17,18,19], we propose and validate two key pathways: “cooperation orientation” and “future orientation”.

(2) This study extends the ABV from internal attention structures to external contexts. The classic ABV emphasizes how internal structures govern attention [11]; we extend this by proposing that the effectiveness of attention configuration is highly dependent on external contexts. This shifts the analytical focus to the organization–environment interface, directly responding to the calls to incorporate external “discourse” into attention research [15,16]. Additionally, findings on the impact of media attention on digital transformation remain inconsistent. Chen and Zhang (2025) found that media attention promotes digital transformation [20]. In contrast, we reveal its potential inhibitory effect: online media reports weaken the positive relationship between future orientation and digital transformation. The discovery adds important boundary conditions and empirical evidence for understanding the role of media in complex, long-term transformations.

(3) This study enriches entrepreneurship theory through contextualization. By employing a multi-dimensional construct of entrepreneurship rooted in China’s institutional and cultural background, this study responds to the scholarly perspective that entrepreneurship research should focus on contextual characteristics [21]. Through revealing the differential impact of its sub-dimensions on digital transformation, as well as their heterogeneous effects across different firms and industries, it provides new empirical evidence and theoretical refinement for understanding entrepreneurship within the Chinese context.

The remainder of this paper is organized as follows: Section 2 presents the theoretical foundation and research hypotheses. Section 3 describes the research design, including data sources, variable measurement, and model specification. Section 4 reports the empirical results. Section 5 conducts further analyses, including sub-dimensional verification and heterogeneity analysis. Section 6 concludes key findings, managerial implications, limitations, and directions for future research.

2. Theoretical Analysis and Hypothesis Development

2.1. Core Theory: Attention-Based View of the Firm

The Attention-Based View posits that the core driver of organizational behavior lies in how decision-makers allocate their limited attentional resources [11]. This theory emphasizes that whether an issue is noticed or ignored directly determines subsequent strategic actions, meaning that the cognitive framework of decision-makers shapes the direction of attention allocation [15]. In recent years, the ABV has evolved significantly, shifting its core perspective from a static view of “attention as a scarce resource” to a dynamic understanding of “attention as a process constructed through communicative practice” [16]. Therefore, we state that organizational attention refers to the process by which an organization allocates its limited cognitive resources to specific issues within a given time and context.

With the continuous evolution of the ABV, it proposed four specific directions through which communication shapes attention: communication practices, strategic vocabularies, rhetorical tactics, and text [16]. Accordingly, this study regards corporate annual reports as an institutionalized formal communication channel and a carrier of attention [16,22]. The textual content of these reports goes beyond mere information conveyance, and reflects the structured output of organizational attention [12]; it embodies the attentional perspective of top management, crystallizes the attentional engagement invested in drafting and review, and externalizes the ultimate attentional selection within a strategic cycle [15]. Therefore, analyzing annual report texts enables us to identify stable patterns of attentional allocation and provides a theoretical foundation for measuring organizational attention, through cooperation orientation and future orientation. To clearly present the multi-layered insights offered by the ABV for this research, its key evolutionary path and core propositions are summarized in

Table 1. Theoretical evolution and integration of the ABV.

Evolutionary Stage	Core Focus	Core Conceptualization	Key Theoretical Advancements
Foundational Stage Ocasio (1997) [11]	Firm as an attention allocation system	A scarce resource governed by structure.	The three principles of ABV were proposed, establishing attention as the core variable for explaining firm behavior.
Application Stage Ocasio & Joseph (2005) [22]	Strategy process as attention flow in channels	A resource flowing within networks of operational and governance channels.	ABV was applied to strategic processes, and dynamic structural concepts like “channel networks” and “coupling” were introduced.
Integration Stage Ocasio (2011) [15]	The multifaceted analysis of the attention concept	A three-dimensional construct: Perspective, Engagement, Selection.	A clear framework for measuring and studying different qualities of attention, linking micro-cognition and macro-structure, was provided.
Dynamic Stage Ocasio et al. (2018) [16]	Communication as the process of shaping attention	An actively constructed process through communicative practices, not merely an allocated outcome.	The focus shifted from static “structure” to dynamic “practice”, explaining the attentional roots of radical strategic change.

.

To understand the unique explanatory power of the ABV, drawing on the comprehensive analysis on building resilience [23], we compare it with other theories such as the RBV and DCT. Their core differences are summarized in

Table 2. Contrasting theoretical core: ABV, RBV, and DCT.

Theory	RBV	DCT	ABV
Core Focus	Competitive advantage stems from resource attributes.	Sustained advantage stems from dynamic capabilities.	Organizational behavior depends on attention allocation—what issues decision-makers focus on or ignore—shaping the strategic agenda and actions.
Core Mechanism of Adaptation and Change	Resource accumulation and deployment: acquiring and configuring unique resources to build barriers.	Capability iteration and recombination: achieving renewal through the "sense-seize-transform" learning and reconfiguration process.	Attention reconfiguration: guiding organizational cognitive and behavioral focus by changing communication channels, strategic discourse, and agenda-setting.
Core Explanation for Digital Transformation	Whether transformation is possible depends on possessing key digital resources.	How to sustain transformation relies on dynamic capabilities to continually adjust digital strategy and integrate old and new systems.	The reason why transformation is initiated and how it is directed: attention is directed to digital issues and long-term opportunities; differences in transformation paths stem from variations in attention structures.

below.

2.2. Theoretical Integration and Research Framework

Based on the core concepts of the ABV, this study moves beyond explanatory frameworks centered on “what is possessed” (resources) and “what can be done” (capabilities); instead, we delve into the cognitive and procedural dimensions of “what is focused on” and “how focus is directed” to uncover the intrinsic drivers of digital transformation. This approach provides a novel theoretical lens for understanding the variations in digital transformation paths across different firms.

The ABV posits that attention allocation is central in connecting the strategic cognition of decision-makers with organizational action [11]; Cho and Hambrick (2006) further argued in their research that the top management team guides organization attention, thereby driving strategic change, and proposed a theoretical model with “attention” as the core mediator [12]. Building on the core demands of digital transformation for cross-boundary collaboration and long-term evolution, we further propose that entrepreneurship, as a form of higher-order strategic cognition, will drive digital transformation by allocating the attentional focus to cooperation orientation and future orientation. These two types of attentional orientation constitute the operational manifestation and transmission pathway of organization attention within this specific strategic context of digital transformation.

Furthermore, responding to the call to examine the interaction between external contexts and attentional processes [15,16], we introduce two moderating variables to account for the influence of the external environment. The regional innovation atmosphere and online media reports are theorized to moderate the strength of the relationship between the attentional orientations and digital transformation outcomes.

The final integrated research framework is shown in Figure 1, illustrating the logical chain of “cognition—attention allocation—action”, with this path being moderated by the external contexts. This framework is not only rooted in the development of the ABV but also through the introduction of the communication channel perspective, and provides a dynamic and operable theoretical model for understanding the divergent paths of digital transformation, representing the core innovation of this study.

2.3. Hypothesis Development

2.3.1. Direct Effect: Entrepreneurship and Digital Transformation

Entrepreneurship theory posits that entrepreneurs guide the future development of enterprises through their comprehensive qualities such as personal judgment, individual values and willpower [24]. Scholars primarily explain entrepreneurship from two perspectives; the first emphasizes the individual traits, such as their propensity for innovation, proactiveness, and willingness to take risks [8,25], and the second focuses on the corporate level, where entrepreneurship represents a concentrated expression of strategic cognition that enables firms to navigate uncertain environments, respond to market challenges, reconfigure production elements, and drive sustainable growth [9,26]. Entrepreneurship plays a profound role in shaping strategic decision-making, organizational change and value creation [27].

Digital transformation is the strategic process through which firms use digital technologies to reshape operation and business processes [1,28], representing a key strategic action for sustainable development in the expanding global digital economy [29], and becoming a strategic priority for entrepreneurs [30,31]. Thus, from an ABV, entrepreneurship determines not only what resources a firm possesses or what capabilities it has, but first and foremost determines what the firm focuses on and what it deems important. Therefore, we propose that a strong entrepreneurship can drive the formation of a forward-looking attentional perspective, elevating digital transformation to a strategic priority within the organizational agenda.

H1. 

Entrepreneurship has a positive impact on digital transformation.

2.3.2. Mediation Effect: Organizational Attention

The latest ABV literature review emphasized that allocating attention toward cross-departmental relationships and long-term perspectives is crucial for strategic adaptation [17,18,19], this theoretical insight aligns precisely with the dual attentional demands of digital transformation:

First, digital transformation is a systematic change across departments and organizations [32]. Its high failure rate often stems from organizational barriers [33], highlighting that success depends not merely on technology adoption but on internal and external cooperation [34,35,36]. This requires the organization to focus its strategic attention on collaborative solutions for how to achieve transformation. Cooperation orientation serves as a critical mechanism for translating entrepreneurship into digital outcomes by promoting the integration and exchange vital for systemic change [17,28].

Second, it is a future-oriented, dynamic and continuous process from decision to realization [1]. Characterized by long-term nature, complexity, and return uncertainty [6,37,38], it demands that firms focus on long-term value and sustainable development. This requires the organization to focus its strategic attention on the future-oriented issues of why transformation is needed. Entrepreneurship focuses attention on technological opportunities and long-term value [39]. This is supported by evidence that entrepreneurship can instill long-term orientation in family businesses [40], and that advancing digital transformation relies on decision-makers’ integrating opportunities and strengths through a long-term lens [41].

Entrepreneurship, as a critical driver in formulating long-term strategy [42], guides organizational action by influencing how leaders selectively focus on external signals and priorities like sustainable growth [18,43]. Empirically, entrepreneurship fosters digital-centric management, enhances cross-boundary information sharing [44], and aligns corporate goals with national strategies to spur long-term investments [42]. From an ABV, entrepreneurship does not influence digital transformation merely through the availability of resources, but through the selective allocation of managerial attention toward cooperation and future-oriented priorities, which constitute the immediate drivers of digital strategic action.

Therefore, we propose that entrepreneurship drives digital transformation by concretely shaping two dimensions of attention:

H2. 

Cooperation orientation mediates the positive relationship between entrepreneurship and digital transformation.

H3. 

Future orientation mediates the positive relationship between entrepreneurship and digital transformation.

2.3.3. Moderating Effect: Regional Innovation Atmosphere and Online Media Reports

According to the ABV, the translation of attentional focus into strategic action is also contingent upon external contexts [15,16]; therefore, we examine how external contexts moderate the effects of the two attention dimensions in driving digital transformation.

Regional innovation atmosphere refers to the regionally embedded ecosystem of institutional support, policy frameworks, service networks, and digital infrastructure that collectively promote generation, diffusion, and utilization of knowledge through formalized and networked interactions [45]. From an ABV, such an ecosystem functions as a powerful external context that shapes organizational attention through both bottom–up stimulation and the provision of legitimized attentional carriers [15]. Significant disparities in marketization, economic development, and cultural advancement exist across Chinese regions [46], leading to varied strengths of this “collaborative attention ecology”. Studies show that knowledge accumulation within a regional innovation atmosphere significantly influences both corporate innovation behavior and regional economic development [47]. A well-developed innovation atmosphere fosters corporate innovation through cultural alignment, vibrant innovation ecosystems, and shared public resources [48]; this enables resource-rich regions with vibrant innovation atmospheres to more readily achieve cross-departmental and cross-organizational knowledge sharing, translating it into digital transformation momentum [49]. Therefore, we propose the following:

H4. 

Regional innovation atmosphere positively moderates the relationship between cooperation orientation and digital transformation.

Online media reports convey information about corporate decision-making and strategic initiatives to the market from a third perspective; they can monitor how enterprises respond to national policies, adjust their strategic behaviors accordingly, and showcase their future plans to the public [50]. They can also release key information to the market and stakeholders, thereby exerting a supervisory, constraining, and guiding influence on corporate behavior and stakeholder decisions, thus serving an external governance function [51,52]. Market pressure theory posits that intense media attention imposes significant market pressure and immediate performance scrutiny on the management of listed companies. In the context of digital transformation, media reports can provide a verification channel for digital environment information [53]. However, this informational function is accompanied by a significant attention-allocation effect. From the perspective of organizational attention, the market pressures and immediate public scrutiny are carried by intensive media reporting [54], constituting a powerful external stream of attentional stimuli [15]. Given the inherent scarcity of organizational attention [11], these short-term issues engage in a fundamental competition for resources with the long-term strategic focus of "future orientation." The result is not merely a crowding out of resources but rather an inducement for decision-makers to prioritize the allocation of cognitive resources and practical actions toward projects with shorter cycles and higher certainty, in response to external short-term pressures. Therefore, we propose the following:

H5. 

Online media reports negatively moderate the impact of future orientation on digital transformation.

3. Study Design

3.1. Data Sources

The sample is constructed based on the following steps: (1) we obtain the initial sample from the Chinese Listed Companies Entrepreneurship Database developed by the Peking University Open Research Data Platform [55], which is the data of samples from companies listed on the Shanghai and Shenzhen stock exchanges between 2010 and 2023; (2) this initial sample was then merged with data from the CNRDS using firm identifiers and fiscal years as the matching criteria; (3) we exclude firm-year observations with missing values for any of the core variables, and as such the final panel dataset comprises 19,060 firm-year observations over 3000 companies across 12 industries.

3.2. Variable Design and Measurement

3.2.1. Dependent Variable: Digital Transformation

Some researchers have found that the application of text analysis allows for a more precise evaluation of digital transformation [56,57,58], so we measure it using the “Digital Transformation” indicator obtained directly from the CNRDS database, this indicator is constructed by CNRDS through systematic text analysis of listed companies’ annual reports, which involves counting the frequency of keywords related to digital technologies and applications (e.g., AI, cloud computing, big data). Following common practice to mitigate right-skewness [58], we take the natural logarithm of the keyword frequency after adding 1.

3.2.2. Independent Variable: Entrepreneurship

Consistent with research Bu (2022) and Wang et al. (2025), this study focuses on entrepreneurship at the corporate level [9,59]. We employ the corporate entrepreneurship dataset from the Peking University Open Research Data Platform, which provides a comprehensive measurement framework. As detailed in

Table 3. Composition of entrepreneurship indicators for listed companies.

Primary Indicator	Weight	Secondary Indicator	Evaluation Content
Patriotism	20.00%	Tax Contribution (10.00%)	Absolute Value of Total Tax Paid (10.00%)
		Common Prosperity (10.00%)	Labor Income Share (10.00%)
Innovation	30.00%	Innovation Potential (10.00%)	Entrepreneur’s Education Level (10.00%)
		Innovation Investment (10.00%)	R&D Expenditure (5.00%)
			R&D Human Resources Input (5.00%)
		Innovation Performance (10.00%)	Patent Output per Capita (10.00%)
Integrity	25.00%	Integrity (12.50%)	Information Disclosure Distortion (12.50%)
		Law-Abiding (12.50%)	Legal Violations (12.50%)
Social Responsibility	15.00%	Social Donation (7.50%)	Contribution Value (7.50%)
		Employment Stability (7.50%)	Number of New Employees Added (7.50%)
Global vision	10.00%	Global Depth (5.00%)	Ratio of Overseas Sales to Total Sales (5.00%)
		Global Breadth (5.00%)	Number of Overseas Subsidiary Locations (5.00%)

, the framework comprises five first-level captures of entrepreneurial strategic cognition, including patriotism, innovation, integrity, social responsibility, and global vision. These dimensions collectively delineate the highest-order strategic cognition, reflecting entrepreneurs’ capacity to recognize key external trends, assess long-term value, and engage in proactive contemplation of sustainable development.

3.2.3. Mediator Variable: Organizational Attention

According to the ABV, organization attention is shaped and expressed through organizational communication channels, where the repeated and prominent use of specific vocabulary reflects institutionalized strategic foci filtered through internal attention structures [15,16]. As one of the most authoritative and structured formal communication channels, the textual content of annual reports serves as a key observable carrier of such attention allocation [60]. Therefore, this study adopts the frequency of specific thematic words in annual reports as the measurement for two dimensions of organization attention. Systematic variation in the relative emphasis on specific thematic words provides a quantifiable signal, revealing how organizational attention is structurally allocated. This approach enables us to capture deep-seated attentional differences that are more fundamental than superficial disclosure styles.

(1) Cooperation orientation operationalizes the focus on strategic answers—the prioritization of action alternatives and solutions centered on collaboration, partnerships, and ecosystem engagement. The frequency of words like “win-win, sharing, cooperation, and synergy” indicates an organizational focus on cooperation [61], and it is measured using the CNRDS database “Annual Report Cooperation Culture” indicator.

(2) Future orientation operationalizes the focus on strategic issues—the noticing and encoding of long-term opportunities, trends, and potential disruptions in the environment. The frequency of words like “next year, intend to, promising, expectation, and future” reflects an emphasis on future and forward-looking opportunities [62], and it is measured using the CNRDS database “Annual Report Future Matters” indicator.

Following Liu et al. (2025) and Hu et al. (2021), we take the natural logarithm of the keyword frequency after adding 1 [62,63].

3.2.4. Moderator Variables

(1) Regional innovation atmosphere: Drawing on research by Wu and Zhang (2021), patents can serve as a suitable quantitative indicator for regional innovation [64]. Accordingly, we adopt the “Provincial Patent Application Status” indicator from the CNRDS database, which aggregates the total applications for three types of patents—invention, utility model, and design. Following the methodology of Kong et al. (2020), this variable is quantified as the natural logarithm of the total number of annual patent applications [65].

(2) Online media reports: Compared to traditional media, online media utilizes the internet as a communication medium and possesses stronger dissemination power and influence, which is capable of fully leveraging media effects within a short time [66]. Wu and Zheng (2021) also noted that traditional media are gradually being replaced by online media [67]. Drawing on related the research of Wang et al. (2021), we use the “Quantitative Statistics of Online News on Listed Companies” indicator from the CNRDS database; based on the total number of online media reports each listed company receives annually, we take the natural logarithm of the keyword frequency after adding 1 [66].

3.2.5. Control Variables

We control for a series of variables at both the corporate and governance levels that may influence corporate digital transformation following [59,68].

The definitions of all variables are provided in

Table 4. Variables and definitions.

Type	Variable	Symbol	Definition	Measurement
Dependent Variable	Digital Transformation	$\mathit{lnDig}$	Degree of digital transformation	$\mathrm{Ln}$(Annual Report Digitalization Keyword Frequency + 1)
Independent Variable	Entrepreneurship	Ent	Corporate entrepreneurship	Total Score of five dimensions
Mediator Variables	Cooperation Orientation	$\mathit{lnCoop}$	The frequency keys words of cooperation culture, add 1, and then take the natural logarithm	Ln (Annual Report Cooperation Culture Keyword Frequency + 1)
	Future Orientation	$\mathit{lnFut}$	The frequency keys words of Future matters, add 1, and then take the natural logarithm	Ln (Annual Report Future-Related Disclosure Keyword Frequency + 1)
Moderator Variables	Regional Innovation Atmosphere	$\mathit{lnPatent}$	The total number Patents of Province applied for, and take the natural logarithm	Ln (Provincial Annual Number of Patent Applications)
	Online Media Reports	$\mathit{lnMedia}$	The total number of Online Media Reports, add 1, and then take the natural logarithm	Ln (Annual Total Number of Online Media Reports + 1)
Control Variables	Enterprise age	listage	Difference between the sample observation year and the firm’s establishment year	Observation Year − Establishment Year
	Sales Growth	Growth	The ratio of the sales revenue growth	(Current year Sales revenue/Prior year sales revenue) − 1
	Current Ratio	Current ratio	Current assets divided by current liabilities	Current Assets/Current Liabilities
	Debt-to-Asset Ratio	Leverage	The ratio of liability to total assets	Total Liabilities/Total Assets
	Return on Assets	Roa	The value that net income divided by total assets	Net Income/Total Assets
	Enterprise Size	$\mathit{lnSize}$	The natural logarithm of total assets	ln (Total Assets)
	Proportion of Directors’ Shareholding	Bshare	Total shareholding percentage of all directors.	Total Shares Held by All Directors/Total Outstanding Shares
	Board size	Board	The total number of board members	The Total Number of Board Members
	Proportion of independent directors	Idirector	Proportion of independent directors on the board	Number of Independent Directors/The total number of board members
	Ownership Concentration	Top10	Shareholding ratio of the top ten shareholders	(∑(Shares held by top 10 shareholders)/Total outstanding shares
	CEO Duality	Duality	Duality of CEO and Chairman roles	Equals 1 if CEO and Chairman are the same person, otherwise 0
	State-Owned	Soe	Nature of the firm	Equals 1 if the firm is state-owned, otherwise 0

.

3.3. Research Model

To rigorously test the hypotheses proposed in Section 2, we employ a set of fixed-effects panel regression models, controlling for year, industry, and province fixed effects to mitigate unobserved heterogeneity. The models are constructed as follows.

3.3.1. Baseline Model: Testing the Direct Effect

To examine the direct effect proposed in H1, following the methodologies of Wu et al. (2021) and Xiao et al. (2022) [57,69], we construct the following baseline fixed-effects model (1):

$${\mathit{lnDig}}_{i,t}={\alpha }_0+{\beta }_1{\mathit{Ent}}_{i,t}+{\beta }_2{\mathit{Controls}}_{i,t}+\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}$$

(1)

where ${\mathit{ln}\left(\mathit{Dig}\right)}_{i,t}$ represents the digital transformation level of the enterprise; ${\mathit{Ent}}_{i,t}$ represents total entrepreneurship score of five dimensions. ${\mathit{Controls}}_{i,t}$ represent control variables; ${\mathit{Year}}_t$, ${\mathit{Industry}}_i$ and ${\mathit{Province}}_k$ represent year, industry, and province fixed effect. The intercept ${\alpha }_0$ measures the baseline digital transformation level. The coefficient ${\beta }_1$ measures the marginal effect of a one-unit increase in entrepreneurship, conditional on other variables. The vector ${\beta }_2$ contains the marginal effects of the control variables. The term ${\epsilon }_{i,t}$ is the idiosyncratic error. All standard errors are clustered at the firm level.

3.3.2. Mediation Models: Testing the Organizational Attention Pathways

To examine the mediating effects proposed in H2 and H3, the mediating effect models for cooperation orientation are Model (2a) and (2b),the mediating effect models for future orientation are Model (3a) and (3b):

$${\mathit{lnCoop}}_{i,t}={\gamma }_0+{\gamma }_1{\mathit{Ent}}_{i,t}+{\gamma }_2{\mathit{Controls}}_{i,t}+\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}$$

(2a)

$${\mathit{lnDig}}_{i,t}={\delta }_0+c{\mathit{Ent}}_{i,t}+b{\mathit{lnCoop}}_{i,t}+{\theta }_1{\mathit{Controls}}_{i,t}+\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}$$

(2b)

$${\mathit{lnFut}}_{i,t}=u_0+u_1{\mathit{Ent}}_{i,t}+u_2{\mathit{Controls}}_{i,t}+\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}$$

(3a)

$${\mathit{lnDig}}_{i,t}=v_0+c^{\prime }{\mathit{Ent}}_{i,t}+b^{\prime }{\mathit{lnFut}}_{i,t}+V_1{\mathit{Controls}}_{i,t}+\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}$$

(3b)

3.3.3. Moderated Mediation Models: Testing the Boundary Conditions

To examine the moderating effects in H4 and H5, the moderated mediation effect models are Models (4) and (5):

$$\begin{array}{cc}\hfill {\mathit{lnDig}}_{i,t}& \hfill ={\theta }_0+{\theta }_1{\mathit{Ent}}_{i,t}+{\theta }_2\ln {\mathrm{Coop}}_{i,t}+{\theta }_3{\mathrm{lnPatent}}_{i,t}+{\theta }_4\left(\ln {\mathrm{Coop}}_{i,t}\times {\mathrm{lnPatent}}_{i,t}\right)+{\theta }_5{\mathit{Controls}}_{i,t}\\ & +\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}\hfill \end{array}$$

(4)

$$\begin{array}{cc}\hfill {\mathit{lnDig}}_{i,t}& \hfill ={\mathsf{\varphi }}_0+{\mathsf{\varphi }}_1{\mathit{Ent}}_{i,t}+{\mathsf{\varphi }}_2{\mathit{lnFut}}_{i,t}+{\mathsf{\varphi }}_3{\mathit{lnMedia}}_{i,t}+{\mathsf{\varphi }}_4\left({\mathit{lnFut}}_{i,t}\times {\mathit{lnMedia}}_{i,t}\right)+{\mathsf{\varphi }}_5{\mathit{Controls}}_{i,t}\\ & +\sum {\mathit{Year}}_t+\sum {\mathit{Industry}}_i+\sum {\mathit{Province}}_k+{\epsilon }_{i,t}\hfill \end{array}$$

(5)

4. Empirical Analysis

4.1. Descriptive Statistics

Table 5. Descriptive statistics of variables.

Variables	Observations	Mean	Standard Deviation	Median	Minimum	Maximum
$\mathit{lnDig}$	19,062	1.94	1.53	1.79	0.00	6.40
$\mathit{Ent}$	19,062	44.92	6.26	45.07	14.40	71.59
$\mathit{lnFut}$	19,062	5.36	0.41	5.41	0.00	6.65
$\mathit{lnCoop}$	19,062	3.77	0.54	3.81	0.00	6.85
$\mathit{lnMedia}$	19,062	4.95	0.98	4.84	0.00	10.21
$\mathit{lnPatent}$	19,062	12.17	1.14	12.26	4.87	13.75
listage	19,062	19.95	5.90	20.00	8.00	37.00
Growth	19,062	13.27	29.21	9.52	−48.12	140.76
Current ratio	19,062	2.43	2.19	1.72	0.37	14.16
Leverage	19,062	0.41	0.19	0.41	0.06	0.87
Roa	19,062	3.34	6.53	3.58	−25.91	19.90
Bshare	19,062	0.13	0.18	0.01	0.00	0.94
Board	19,062	9.85	2.87	9.00	0.00	29.00
Idirector	19,062	0.39	0.10	0.38	0.00	1.00
Top10	19,062	0.56	0.15	0.57	0.04	1.00
Duality	19,062	0.30	0.46	0.00	0.00	1.00
Soe	19,062	0.31	0.46	0.00	0.00	1.00
$\mathit{lnSize}$	19,062	22.32	1.24	22.14	20.09	26.09

reports the descriptive statistics for all variables. The degree of digital transformation has a mean of 1.94 and a standard deviation of 1.53, with values ranging from 0 to 6.40, reflecting considerable variation across sample firms. Entrepreneurship shows a mean of 44.92 and a standard deviation of 6.26, ranging from 14.4 to 71.59, similarly indicating significant differences among the sample enterprises. The mean for future orientation and cooperation orientation is 5.36 and 3.77, with standard deviations of only 0.41 and 0.54, indicating a relatively concentrated distribution. In contrast, online media reports and regional innovation atmosphere exhibit moderate dispersion. The mean value of SOEs is 0.31, indicating that the majority of the samples are Non-SOEs. Overall, the sample firms provided sufficient data support for subsequent empirical analysis.

4.2. Baseline Regression

We employed stepwise regression analysis to examine the direct effect, with the baseline regression results presented in

Table 6. Baseline regression analysis.

Variables	(1)	(2)	(3)	(4)
	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0516 ***	0.0460 ***	0.0222 ***	0.0202 ***
	(0.003)	(0.004)	(0.003)	(0.003)
Constant	−0.8544 *	0.2139	−1.8866 ***	−1.7691 ***
	(0.489)	(0.501)	(0.387)	(0.384)
Observations	19,062	19,062	19,060	19,060
Controls	YES	YES	YES	YES
Year	NO	YES	YES	YES
Industry	NO	NO	YES	YES
province	NO	NO	NO	YES
F	45.24	24.85	17.25	15.80
Adj R2	0.0914	0.131	0.492	0.498

Note: 1. * p < 0.1, *** p < 0.01. 2. The coefficient for controls is not reported. 3. Standard errors are reported in parentheses. The same for the below.

. The results indicated that there is a significant positive correlation between entrepreneurship and digital transformation; an effect that remains robust across different model specifications. After including these controls, the coefficient stabilizes at 0.0202. All models are significant at the 1% level.

The significant positive coefficient provides strong support for H1. This confirms our core proposition that entrepreneurship, as a strategic cognition, is a critical antecedent for digital transformation. This finding extends the ABV by suggesting that in the digital era, intangible strategic cognition is as crucial as driving technological adoption. It aligns with Bu (2022) and indicates that fostering a holistic entrepreneurship is foundational for firms navigating disruptive change [9].

4.3. Addressing Endogeneity

The relationship between entrepreneurship and digital transformation may be endogenous due to reverse causality. Omitted variables could also bias the results. To mitigate these concerns, our baseline regression model includes year, industry, and province fixed effects, with standard errors clustered at the firm level.

4.3.1. Instrumental Variable Method

First, we use the average Ent of other firms in the same city and year (mean_Ent) as an instrumental variable (IV). While mean_Ent is correlated with a focal firm’s entrepreneurship through knowledge spillovers and demonstration effects, it is unlikely to directly influence the firm’s digital transformation outcomes. As shown in

Table 7. Instrumental variable method (mean_Ent).

Variables	The First Stage	The Second Stage
	$\mathit{Ent}$	$\mathit{lnDig}$
IV (mean_Ent)	0.1669 *** (0.026)
$\mathrm{Ent}$		0.1558 ** (0.042)
Constant	−17.5771 *** (1.995)
Observations	18,184	18,184
Controls	YES	YES
Year	YES	YES
Industry	YES	YES
province	YES	YES
Adj R2	0.397	−0.300
Kleibergen–Paap rk LM		43.965 ***
Kleibergen–Paap Wald rk F		42.157

Note: ** p < 0.05, *** p < 0.01.

, the first-stage results confirm a strong positive correlation between mean_Ent and Ent (p < 0.01). The Kleibergen–Paap rk LM statistic rejects underidentification, and the Kleibergen–Paap Wald rk F statistic exceeds the Stock–Yogo critical value, ruling out weak instrument concerns. In the second stage, entrepreneurship remains positively associated with digital transformation at the 5% level, supporting the robustness of our baseline findings.

Second, we employ one-period lag (L.Ent) and two-period lag (L2.Ent) of entrepreneurship as alternative IV. The corresponding 2SLS results are presented in

Table 8. Instrumental variable method (L.Ent and L2. Ent).

Variables	The First Stage	The Second Stage	The First Stage	The Second Stage
	$\mathit{Ent}$	$\mathit{lnDig}$	$\mathit{Ent}$	$\mathit{lnDig}$
$\mathrm{L}.\mathrm{Ent}$	0.5489 *** (0.011)	0.0405 *** (0.005)
$\mathrm{L}2. \mathrm{Ent}$			0.4384 *** (0.013)	0.0470 *** (0.008)
Constant	−3.5997 *** (0.900)		−4.1857 *** (1.201)
Observations	14, 359	14, 359	11, 839	11, 839
Controls	YES	YES	YES	YES
Year	YES	YES	YES	YES
Industry	YES	YES	YES	YES
province	YES	YES	YES	YES
Adj R2	0.588	0.0206	0.513	0.0124
Kleibergen–Paap rk LM		658.098 ***		446.385 ***
Kleibergen–PaapWald rk F		2698.455		1140.902

Note: *** p < 0.01.

. The first-stage results show that both lagged terms are positive correlations with current entrepreneurship at the 1% level. The Kleibergen–Paap rk LM statistics (658.098 and 446.385) are significant at the 1% level, rejecting the underidentification hypothesis. Furthermore, the Kleibergen–Paap Wald rk F statistics (2698.455 and 1140.902) comfortably exceed the Stock–Yogo critical value of 16.38, effectively ruling out weak instrument concerns. The second-stage results indicate that entrepreneurship continues to exert significant influence on digital transformation at the 1% level.

4.3.2. Heckman Two-Stage Method

To address potential sample self-selection bias, we implement the Heckman two-stage method. In the first stage, we construct a dummy variable (Treat_Ent), which equals 1 if Ent exceeds the industry-year average, and 0 otherwise. Using a probit model with Treat_Ent as the dependent variable, we include all baseline control variables and an IV—the average Ent of other firms in the same industry and year (mean_Ent_ind). The Imr is estimated from this stage. In the second stage, the Imr is incorporated into the main regression as an additional control. As reported in

Table 9. Heckman two-stage method.

Variables	The First Stage	The Second Stage
	Treat_Ent	$\mathit{lnDig}$
IV (mean_Ent_ind)	−0.2225 ***
	(0.0125)
		0.0336 ***
Ent		(0.0058)
Imr		−0.0995
		(0.0966)
Observations	19,040	9901
Constant	−7.3279 (1.3678)	−1.0999 (0.8760)
Controls	YES	YES
Year	YES	YES
Industry	YES	YES
province	YES	YES
Wald chi2	1266.21 ***
Adj R2		0.5033

Note: *** p < 0.01.

, the first-stage results confirm a statistically significant correlation between IV and Treat_Ent at the 1% level, supporting the validity of IV. In the second stage, the coefficient of the Imr is insignificant, suggesting no severe self-selection bias. More importantly, after including the Imr, the effect remains consistent with baseline findings, further reinforcing the robustness of our conclusions.

4.4. Robustness Tests

4.4.1. Propensity Score Matching

We employ propensity score matching (PSM) to conduct endogeneity tests and mitigate potential sample self-selection bias. Following the approach of Sun et al. (2025), treatment and control groups are defined based on the annual industry-average level of entrepreneurship [70]. The nearest neighbor matching method (1:1) is applied to construct counterfactual outcomes.

The balance test results in

Table 10. Balance test for propensity score matching.

Variables	Unmatched Matched	Mean		%Reduct		t-Test		V(T)/V(C)
		Treated Control		%Bias	|Bias|	t	p > |t|
listage	U	20.163	19.729	7.4		5.08	0.000	1.01
	M	20.162	20.001	2.7	62.9	1.93	0.053	1.04
Growth	U	14.867	11.552	11.4		7.84	0.000	0.95 *
	M	14.804	15.315	−1.8	84.6	−1.21	0.228	0.87 *
Current ratio	U	2.3166	2.5478	−10.6		−7.30	0.000	0.77 *
	M	2.3160	2.2967	0.9	91.6	0.63	0.526	0.84 *
Leverage	U	0.4266	0.3991	14.4		9.95	0.000	0.93 *
	M	0.4267	0.4278	−0.6	95.9	−0.42	0.678	0.92 *
Roa	U	3.9900	2.6382	20.7		14.35	0.000	0.74 *
	M	3.9738	3.9969	−0.5	98.3	−0.27	0.785	1.02
Bshare	U	0.1168	0.1426	−14.3		−9.86	0.000	0.84 *
	M	0.1169	0.1178	−0.5	96.6	−0.35	0.727	0.95 *
Board	U	9.9903	9.6912	10.4		7.20	0.000	1.02
	M	9.9902	10.023	−1.1	89.1	−0.78	0.436	0.92 *
Idirector	U	0.3884	0.3876	0.8		0.57	0.570	0.94 *
	M	0.3860	0.3807	0.8	0.1	0.58	0.564	0.94 *
Top10	U	0.5702	0.5549	10.0		6.92	0.000	1.08 *
	M	0.5700	0.5738	−2.5	74.9	−1.72	0.086	0.94 *
Duality	U	0.2871	0.3117	−5.4		−3.71	0.000
	M	0.2873	0.2801	1.6	70.9	1.12	0.263
Soe	U	0.3476	0.2655	17.9		12.30	0.000
	M	0.3471	0.3540	−1.5	91.6	−1.01	0.311
lnSize	U	22.693	21.925	65.1		44.70	0.000	1.62 *
	M	22.688	22.689	−0.0	99.9	−0.02	0.982	0.96

Note: * if variance ratio outside [0.96; 1.04] for U and [0.96; 1.04] for M.

indicated that after matching, the standardized biases of all variables were reduced to below 5%. Key covariates achieved a bias reduction of over 95%, demonstrating excellent matching quality. As further shown in

Table 11. Balance results before and after matching.

Sample	Ps R2	LR chi2	p > |t|	MeanBias	MedBias	B	p	%Var
Unmatched	0.086	2260.87	0.000	15.7	11.0	70.6 *	1.52	80
Matched	0.000	11.15	0.516	1.2	1.0	4.7	0.93	70

Note: * if B > 25%, R outside [0.5; 2].

, key metrics including the pseudo-R², standardized bias, and balance measure B were all substantially improved after matching, confirming the absence of systematic differences between the treatment and control groups. Figure 2 also indicates that a good post-matching quality between the treatment and control groups is based on the method of Kernel density.

Table 12. PSM and ATT regression results.

Variables	(1)	(2)	(3)
	Baseline Regression Analysis	PSM	ATT
$\mathrm{Ent}$	0.0202 *** (0.0027)
High-$\mathrm{Ent}$		0.1264 *** (0.0327)	0.2661 *** (0.0341)
Constant	−1.7688 *** (0.384)	−1.2285 * (0.4959)
Observations	19,060	14,600	19,062
Controls	YES	YES	YES
Year	YES	YES	YES
Industry	YES	YES	YES
province	YES	YES	YES
F	15.8	8.56
Adj R-squared	0.498	0.495

Note: * p < 0.1, *** p < 0.01.

reports the results of the PSM estimation and the Average Treatment Effect on the Treated (ATT) of entrepreneurship on digital transformation. The results show that both the PSM and ATT estimates are consistent with the baseline regression, but with significantly larger values, indicating that even after controlling for selection bias, entrepreneurship still has a significant positive effect on digital transformation.

4.4.2. Replacement Dependent Variable

Following the approach of Chen et al. (2023) and Zhang et al. (2021), we represent the dependent variable replaced by two dimensions: the digital-related word frequency of MD&A ($\mathrm{lnDig}1$) and the intangible assets related to digital technology ($\mathrm{lnDig}2)$ [71,72]. As shown in

Table 13. Regression results of replacement method.

Variables	(1)	(2)
	lnDig 1	lnDig 2
$\mathrm{Ent}$	0.0188 *** (0.002)	0.0272 *** (0.004
Constant	−1.0635 *** (0.348)	−7.3927 *** (0.535)
Observations	19,060	15,737
Controls	YES	YES
Year	YES	YES
Industry	YES	YES
province	YES	YES
F	13.29	216.7
Adjusted R2	0.486	0.471

Note: *** p < 0.01.

, the regression coefficients are still significantly positive at the 1% level, mitigating potential bias resulting from reliance on a single indicator.

4.4.3. Altering the Sample Period

To test the reliability of sample selection, we employ an econometric approach by changing the sample period and conducting regression using data from 2010 to 2019 (pre-pandemic). As shown in Column (1) of

Table 14. Regression results with restricted sample window.

Variables	(1)	(2)
	$\mathit{lnDig}$	lnDig
$\mathrm{Ent}$	0.0209 ***	0.0179 ***
	(0.004)	(0.003)
Constant	−2.1096 ***	−2.2599 ***
Observations	(0.467)	(0.408)
Controls	10,217	14,181
Year	YES	YES
Industry	YES	YES
province	YES	YES
Constant	YES	YES
F	11.09	13.11
Adj R-squared	0.501	0.339

Note: *** p < 0.01.

, after altering the sample period, the regression coefficient remains significantly positive, confirming that the conclusions are robust.

4.4.4. Excluding Specific Industry Samples

Considering that some samples belong to highly digitalized industries where firms may inherently exhibit elevated digitalization levels, we exclude samples from sectors including “Computer, Communication, and Other Electronic Equipment Manufacturing”, “Software and Information Technology Services”, “Internet and Related Services”, and “Computer Application Services”. The results are also shown in Column (2) of Table 13. After removing these specific industry samples, the regression coefficient remains significantly positive, further demonstrating the robustness of our conclusions.

4.5. Tests for Mediating and Moderating Mechanisms

4.5.1. Mediation Mechanism Test

Table 15. Regression results of the mediating mechanisms.

Variables	(1)	(2)	(3)	(4)
	$\mathit{lnCoop}$	$\mathit{lnDig}$	$\mathit{lnFut}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0054 *** (0.001)	0.0170 *** (0.003)	0.0019 *** (0.001)	0.0194 *** (0.003)
$\mathrm{lnCoop}$		0.6010 *** (0.037)
$\mathrm{lnFut}$				0.4579 *** (0.057)
Constant	1.6393 *** (0.158)	−2.7543 *** (0.376)	3.8756 *** (0.094)	−3.5438 *** (0.431)
Observations	19,060	19,060	19,060	19,060
Controls	YES	YES	YES	YES
Year	YES	YES	YES	YES
Industry	YES	YES	YES	YES
province	YES	YES	YES	YES
F	30.42	34.30	42.69	20.00
Adjusted R2	0.389	0.526	0.634	0.504

Note: *** p < 0.01.

presents the results of the stepwise regression analysis on the two mediating variables. Column (1) shows that entrepreneurship has a significantly positive effect on cooperation orientation. In Column (2), after incorporating cooperation orientation, the coefficient decreases but remains significant, indicating partial mediation in the relationship with digital transformation. Similarly, Column (3) reports a significant positive effect of entrepreneurship on future orientation. Column (4) further confirms the partial mediating role of future orientation, as the coefficient of entrepreneurship declines yet stays significant when this variable is included.

These results support H2 and H3, empirically validating that entrepreneurship directs organizational attention toward collaborative ecosystems and long-term strategic horizons. This finding empirically responds to the proposition of Ocasio and Joseph (2005) regarding the ABV by linking micro and macro perspectives in strategy processes [22].

More importantly, these attention orientations transcend their immediate role in digitalization. These findings resonate with recent evidence reported by Martell (2025), who demonstrates that strategic foresight strengthens organizational sustainability by enhancing anticipatory capacity, fostering innovation, and aligning managerial decision-making with long-term corporate social responsibility objectives [73]. From this perspective, the orientation of organizational attention functions not only as a digital transformation mechanism but also as a sustainability-enabling capability. Consequently, entrepreneurship, by shaping these attentional foci, equips firms not just to adopt new technologies, but to build the cognitive foundations for enduring resilience and sustainable growth in the digital era.

Based on the approach of Hu et al. (2021) [62], the results of the Sobel tests are summarized in

Table 16. Sobel test for mediating variables.

Mediating Variables	Coefficient Estimate					Result
	Total Effect (c)	First-Stage Path (a)	Second-Stage Path (b)	Direct Effect (c′)	Indirect Effect (a × b)
$\mathrm{lnCoop}$	0.038 *** (0.003)	0.006 *** (0.001)	0.701 *** (0.049)	0.034 *** (0.003)	0.004 *** (0.001)	Partial Mediation Effect
$\mathrm{lnFut}$	0.038 *** (0.003)	0.003 *** (0.001)	0.580 *** (0.069)	0.036 *** (0.003)	0.002 *** (0.001)

Note: *** p < 0.01.

. When cooperation orientation serves as the mediator, path coefficients a and b are significantly positive, indicating its role as a partial mediator in the relationship between entrepreneurship and digital transformation. Similarly, when future orientation serves as the mediator, path coefficients a and b are also exhibited as significantly positive, confirming its function as a partial mediator. H2 and H3 were supported.

4.5.2. Moderation Mechanisms Test

Table 17. Regression results of the moderating mechanisms.

Variables	(1)	Variables	(2)
	$\mathit{lnDig}$		$\mathit{lnDig}$
$\mathrm{lnCoop}$	−0.3605 (0.248)	$\mathrm{lnFut}$	0.7402 *** (0.166)
$\mathrm{lnPatent}$	−0.2454 ** (0.098)	$\mathrm{lnMedia}$	0.4795 *** (0.166)
c.$\mathrm{lnCoop}$ # c.$\mathrm{lnPatent}$	0.0802 *** (0.021)	c.$\mathrm{lnFut}$# c.$\mathrm{lnMedia}$	−0.0636 ** (0.031)
Constant	0.1967 (1.229)	Constant	−4.5875 *** (0.980)
Observations	19,060	Observations	19,060
Controls	YES	Controls	YES
Year	YES	Year	YES
Industry	YES	Industry	YES
province	YES	province	YES
F	30.96	F	21.08
Adjusted R2	0.527	Adjusted R2	0.508

Note: ** p < 0.05, *** p < 0.01.

presents the regression results examining the moderating mechanisms. As shown in Column (1), the interaction term between cooperation orientation and regional innovation atmosphere is positively significant at the 1% level, indicating that a stronger regional innovation atmosphere enhances the mediating effect of cooperation orientation on digital transformation. In Column (2), the interaction term between future orientation and online media reports is significantly negative at the 5% level, suggesting that extensive online media reports weaken the mediating effect of future orientation on digital transformation.

These results provide empirical support for H4 and H5. The positive moderating effect of the regional innovation atmosphere aligns with institutional and ecosystem theories. It demonstrates that a supportive external environment with rich knowledge spillovers and partners complements and amplifies internally driven cooperative strategies, making them more effective in achieving digital outcomes. However, the negative interaction in Column (2) offers a critical theoretical insight. It suggests that intense online media scrutiny may induce “short-termism pressure.” To manage reputational demands under the spotlight, firms might prioritize quick, visible digital outputs over the patient capacity-building nurtured by future orientation. This result stands in contrast to the finding of Chen and Zhang (2025), who reported a positive role of media attention in promoting digital technology adoption [20]. Instead, our finding lends empirical support to the “media hindrance hypothesis” [54], which posits that external scrutiny can distort long-term strategic investments. This finding enriches the literature on media governance by highlighting its potential “dark side” in complex, long-term transformations, demonstrating that external attention can sometimes crowd out strategic foresight.

5. Future Study

5.1. Analysis of Individual Sub-Dimensions of Entrepreneurship

Table 18. Baseline regression results for the sub-dimensions of entrepreneurship.

Variables	(1)	(2)	(3)	(4)	(5)
	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$
Patriotism	0.0089 *** (0.002)
Innovation		0.0150 *** (0.002)
Integrity			0.0005 (0.001)
Global vision				−0.0014 (0.001)
Social responsibility					0.0026 ** (0.001)
Constant	−1.3354 *** (0.414)	−2.1329 *** (0.381)	−1.9832 *** (0.386)	−2.0131 *** (0.387)	−1.8407 *** (0.389)
Observations	19,060	19,060	19,060	19,060	19,060
Controls	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES
Industry	YES	YES	YES	YES	YES
province	YES	YES	YES	YES	YES
F	12.82	18.12	10.74	10.74	11.09
Adjusted R2	0.495	0.504	0.494	0.494	0.494

Note: ** p < 0.05, *** p < 0.01.

reports the regression results of the different entrepreneurial dimensions on digital transformation. First, Innovation has a significantly positive impact on digital transformation, indicating that the willingness to innovate is a core driver of digital transformation. This finding reinforces prior evidence [59]. Second, Patriotism also shows a significant positive impact, suggesting that entrepreneurs with strong patriotic sentiment are more inclined to align with national strategic initiatives and actively invest in building digital capabilities, which verifies the findings of Yue et al. (2024) and Liu et al. (2025) [74,75]. Furthermore, Social responsibility shows a positive coefficient at the 5% level, indicating that enterprises emphasizing social responsibility are more proactive in promoting digital transformation. This may be driven by their goals to enhance resource efficiency or improve transparent communication with communities through digital technology. Lastly, Integrity and Global vision did not demonstrate statistical significance. This could be because international strategies might divert resources otherwise allocated for digital transformation, while Integrity represents a baseline norm that all enterprises are expected to adhere to, thus failing to generate significant differential impacts.

5.2. Heterogeneity Analysis

Drawing on existing research, we conduct heterogeneity tests from both firm-level and industry-level to further elucidate how entrepreneurship influences corporate digital transformation.

5.2.1. Firm-Level Heterogeneity Analysis

First, with reference to the study by Xie et al. (2021) [76], firms are classified into large and SMEs based on the industry median size. As shown in

Table 19. Regression results of firm size heterogeneity.

Variables	(1) Large Firm	(2) SMEs
	$\mathit{lnDig}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0198 *** (0.003)	0.0313 *** (0.004)
Constant	1.0860 *** (0.234)	0.5812 ** (0.254)
Observations	9546	9505
Controls	YES	YES
Year	YES	YES
Industry	YES	YES
province	YES	YES
F	5.778	8.006
Adjusted R2	0.482	0.515

Note: ** p < 0.05, *** p < 0.01.

, entrepreneurship in large firms drives digital transformation more powerfully, which may be attributed to their resource advantages and greater risk-bearing capacity, while entrepreneurship in SMEs remains a significant effect, though its impact is comparatively weaker than that observed in large enterprises.

Second, drawing on studies by Li and Wang (2025) [77], we categorized the sample firms into three groups: Non-SOEs, Central SOEs, and Local SOEs. The regression results in

Table 20. Regression results of heterogeneity in firm ownership nature.

Variables	(1) Non-SOEs	(2) Central SOEs	(3) Local SOEs
	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0215 *** (0.003)	0.0225 ** (0.009)	0.0120 ** (0.005)
Constant	−1.8767 *** (0.499)	−2.3971 ** (1.060)	−2.5429 *** (0.788)
Observations	13,185	2411	3457
Controls	YES	YES	YES
Year	YES	YES	YES
Industry	YES	YES	YES
province	YES	YES	YES
F	12.68	3.137	5.192
Adjusted R2	0.500	0.544	0.495

Note: ** p < 0.05, *** p < 0.01.

reveal significant ownership heterogeneity in how entrepreneurship drives digital transformation. Entrepreneurship has the strongest effect in Non-SOEs, likely due to their market-oriented and profit-driven governance. The effect is weaker in both Central and Local SOEs, as they are more influenced by government macro-policies. However, entrepreneurship proves more effective in Central SOEs than in Local SOEs, owing to their stronger market competitiveness and greater innovation autonomy, whereas Local SOEs face stricter oversight from Local SOEs.

Furthermore, sample enterprises are categorized into four regions—Eastern, Central, Western, and Northeastern. As shown in

Table 21. Regression results of region heterogeneity.

Variables	(1) Eastern	(2) Central	(3) Western	(4) Northeastern
	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0209 *** (0.003)	0.0144 ** (0.007)	0.0217 *** (0.006)	0.0217 * (0.012)
Constant	−2.0440 *** (0.470)	−1.6827 * (0.973)	−0.5128 (1.143)	−2.9972 (2.327)
Observations	13,495	2729	2133	693
Controls	YES	YES	YES	YES
Year	YES	YES	YES	YES
Industry	YES	YES	YES	YES
province	YES	YES	YES	YES
F	12.30	2.787	2.384	3.448
Adjusted R2	0.493	0.505	0.541	0.530

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

, in Eastern China, where the economy is more developed and the business environment is favorable, entrepreneurship exerts the strongest positive effect, significantly driving digital transformation. In Western China, entrepreneurship also shows a significant positive impact, amplified by policy support and digital infrastructure, yielding pronounced marginal benefits and demonstrating a notable latecomer advantage. In Central China, the effect remains positive but is weaker than in Eastern and Western China, largely due to challenges such as industrial restructuring and talent outflow, which constrain the full realization of digital potential. In Northeastern China, the influence of entrepreneurship is positive but the weakest among the four regions, primarily due to the substantial transitional risks that limit its effectiveness.

5.2.2. Industry-Level Heterogeneity Analysis

First, drawing on Zhu and Wang (2023) [78], industries are classified as high- or low-technology-intensive based on their R&D expenditure. As shown in

Table 22. Heterogeneity regression based on industry technology intensity.

Variables	(1) High-Technology-Intensive	(2) Low-Technology-Intensive
	$\mathit{lnDig}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0208 *** (0.004)	0.0141 *** (0.003)
Constant	−0.9856 * (0.511)	−2.5818 *** (0.495)
Observations	9590	9461
Controls	YES	YES
Year	YES	YES
Industry	YES	YES
province	YES	YES
F	8.437	9.673
Adjusted R2	0.507	0.521

Note: * p < 0.1, *** p < 0.01.

, entrepreneurship promotes digital transformation in both high- and low-technology-intensive industries, but the effect is more pronounced in high-technology-intensive industries. The fact is that the core competitiveness stems from R&D and the application of new technologies, where entrepreneurship directly facilitates the implementation of digital technologies. Additionally, high-technology-intensive enterprises typically possess more technical talent and IT infrastructure, enabling entrepreneurship to be translated more efficiently into digital initiatives. In contrast, low-technology-intensive industries may be more influenced by competition or customer demand, with entrepreneurship playing a greater role in following rather than leading technological change.

Secondly, drawing on the research of Liu et al. (2025) [79], we use the median Herfindahl–Hirschman Index (HHI) of industries as the classification criterion to divide the sample into low-competition and high-competition industries. The regression results in

Table 23. Heterogeneity regression based on industry competition intensity.

Variables	(1) High-Competition	(2) Low-Competition
	$\mathit{lnDig}$	$\mathit{lnDig}$
$\mathrm{Ent}$	0.0205 *** (0.004)	0.0194 *** (0.004)
Constant	−1.4829 *** (0.612)	−1.9948 (0.484)
Observations	9396	9664
Controls	YES	YES
Year	YES	YES
Industry	YES	YES
province	YES	YES
F	8.224	9.498
Adjusted R2	0.555	0.395

Note: *** p < 0.01.

demonstrate that the promoting effect of entrepreneurship is more pronounced when firms operate in highly competitive industries. This can be primarily attributed to the fact that intense industry competition and rapid technological iteration require entrepreneurship to swiftly guide enterprises through digital transformation to enhance their core competitiveness. In contrast, in low-competition industries, while entrepreneurship remains influential, its impact on digital transformation is relatively less effective compared to high-competition industries, partly due to the greater influence of macroeconomic policies and resource allocation patterns.

6. Research Conclusions and Policy Recommendations

6.1. Research Conclusions

This study develops and empirically tests an integrated theoretical framework. Moving beyond traditional explanations that primarily focus on resources or capabilities, it reveals the cognitive mechanisms and boundary conditions through which entrepreneurship drives digital transformation from the ABV.

(1) While existing research is predominantly grounded in the RBV and DCT, this study introduces the ABV; beyond confirming the positive role of entrepreneurship, we advance theory by demonstrating that attention allocation constitutes a critical micro-level mechanism linking entrepreneurial cognition with sustainable digital transformation, thereby illuminating the “cognitive black box “from values to digital action.

(2) It extends the ABV from internal structures to external contexts, revealing the contingency of attention configuration effectiveness. By introducing and validating the moderating role of external contexts, this study shifts the analytical perspective to the "organization-environment" interface. This not only responds to the call to incorporate external "discourse" into attention research but also provides important boundary conditions and empirical evidence for understanding contradictory scholarly findings regarding the role of media in complex, long-term transformations.

(3) Grounding the study in the Chinese context enriches relevant theory by deconstructing the multi-dimensional construct of entrepreneurship. By employing a multi-dimensional construct of entrepreneurship rooted in China’s institutional and cultural background, this study addresses scholarly calls to focus on the contextual characteristics of entrepreneurship. The empirical results reveal differential impacts of its sub-dimensions on digital transformation: patriotism, innovation, and social responsibility exhibit significant positive driving effects, whereas global vision and integrity show no significant impact. This finding not only uncovers the unique value-driven logic of digital transformation in Chinese enterprises but also, by demonstrating heterogeneous effects across firms with different ownership types, sizes, regions, and industries, offers new theoretical refinement and empirical evidence for understanding the complexity and strategic priorities of entrepreneurship within the Chinese context.

6.2. Policy Recommendations

6.2.1. Government Level

First, governments should foster an open, transparent business environment and implement targeted incentives to create a socio-cultural atmosphere that encourages entrepreneurial engagement in digital innovation. This will solidify the ideological foundation for China’s digital economy strategy and promote high-quality, sustainable growth.

Second, regional governments should cultivate a high-quality innovation ecosystem. By providing accessible digital technology platforms and achievements, they can amplify the positive role of collaborative synergy on digital transformation and circular economic practices.

Third, governments should promote the adoption of authoritative, long-term-oriented digital and sustainability evaluation metrics. This can help mitigate the negative substitution effect of short-term media pressure on corporate future orientation and encourage patient investment in digital infrastructure.

Fourth, policy support should be prioritized for SMEs, SOEs (especially local SOEs), firms in less developed regions, and those in low-competition or low-technology-intensive industries. Efforts should be made to foster entrepreneurship within these entities and create a supportive environment for their inclusive and sustainable digital transformation.

6.2.2. Corporate Level

First, entrepreneurs should steer organizational attention toward cooperation and future orientation. This facilitates digital transformation through internal coordination, external resource sharing, long-term strategic investment, and the allocation of patient capital toward sustainable digital initiatives.

Second, enterprises should proactively counter “short-termism” amplified by media reports. This includes strengthening investments in patient capital, allowing room for experimentation in digital technology adoption, and balancing external communication with substantive capability building. Transparent disclosure of digital progress can help safeguard long-term R&D and talent development from being undermined by short-term performance expectations.

6.3. Limitations and Future Research

First, the sample of this study is geographically limited, focusing solely on Chinese listed companies. Although this design provides a representative perspective on digital transformation in China, the generalizability of the conclusions to other institutional contexts requires cautious validation. As highlighted by Hanelt et al. (2021), national characteristics are fundamental contextual conditions influencing digital transformation such as legal frameworks, regulations, and digital infrastructure [37]. Moreover, Tsui (2004) [80] pointed out that high-quality indigenous research based on specific contexts can reveal nuanced mechanisms and boundary conditions, contributing to theoretical development and providing a foundation for subsequent comparative studies. Therefore, future research could explore how the impact of entrepreneurship on digital transformation varies across different institutional environments through cross-country comparisons.

Second, regarding the research model, this study primarily employs the ABV to examine how entrepreneurship influences digital transformation. Although this offers a clear theoretical perspective, it does not encompass all potential mechanisms. Future research could integrate complementary perspectives such as strategic renewal theory to further explore how entrepreneurship affects digital transformation through organizational learning and resource reconfiguration. Additionally, incorporating factors such as digital leadership, organizational culture, and ecosystem partnerships would contribute to constructing a more comprehensive analytical framework.