Digital transformation (DT) is a pivotal driver in enhancing the transfer of scientific, knowledge, and technological achievements (STAT), thereby fostering the growth of knowledge-based productive forces. While digitalization represents a complex, multi-dimensional process, the mechanisms facilitating its impact on STAT remain underexplored in the context of innovation systems. Using a panel data set of Chinese A-share listed firms from 2001–2023, this study analyzes this relationship and investigates its operational pathways. The results reveal that DT substantially accelerates STAT, with consistent and stable outcomes that are robust to rigorous evaluation through the Heckman two-stage procedure and instrumental variable estimations. Mechanistically, this research reveals that DT facilitates the STAT process by augmenting patent portfolio quality, refining resource allocation efficiency, promoting the platform ecosystem embedding effect, and enhancing dynamic capabilities—four channels that reflect its role in strengthening knowledge absorption and innovation governance. The analysis of heterogeneous treatment effects demonstrates systematically more potent impacts in enterprises with streamlined cost structures, higher growth potential, intensified industrial competition, and state-owned ownership structures. Enterprises operating in environments with strong financial regulation and high market competition show a more pronounced role of DT in promoting STAT. These findings advance the theoretical and empirical scholarship on digital innovation and knowledge transformation while informing enterprise strategies that leverage digital capabilities to scale innovation commercialization.
The Resolution from the 20th National Congress of the Communist Party of China (CPC) Central Committee in 2022 stipulated, "Strengthen enterprise-driven depth integration of industry, academia, and research, emphasize goal orientation, and enhance the level of sci-tech achievement transformation and industrialization." During an official visit to Heilongjiang Province on September 8, 2023, General Secretary Xi Jinping highlighted, "Take enterprises as the core carriers for the commercialization of sci-tech achievements and improve the conversion rate of sci-tech achievements into practical applications." This statement has since led to concerted efforts to advance the transfer of scientific, knowledge, and technological achievements (STAT). Subsequently, the Third Plenary Session of the 20th CPC Central Committee in 2024 provided additional guidance, highlighting the necessity to "advance reforms in the systems for converting sci-tech achievements," "strengthen the development of the national technology transfer system," and "allow sci-tech personnel greater autonomy in the distribution of benefits from STAT." The Session's communiqué elaborates on strategies for STAT, covering critical areas such as mechanism reform, the development of technology transfer systems and platforms, government incentives and procurement policies, the cultivation of technology manager teams, and incentives for research institutions and sci-tech personnel. These areas cover most of the key aspects of STAT, reflecting China's recognition of STAT as a core driver for advancing sci-tech innovation and Chinese-style modernization. The 2024 National Science and Technology Conference emphasized accelerating STAT, converting more achievements from "samples" to "products," and promoting their industrialization.
During a symposium on central China's revitalization in 2024 Xi Jinping outlined the following key initiatives: cementing enterprises' central role in innovation, deepening industry-academia-research ties, establishing innovation consortia, and accelerating the application of research findings. This highlights STAT as a critical node that determines whether technological innovation can generate economic and social value. Only by successfully transforming new technologies and inventions into new production processes and products can their market value be realized (Guindalini et al., 2021), a prerequisite for pursuing innovation-driven development. Propelled by these policies, China has witnessed robust growth in innovation and a significant rise in research and development (R&D) spending, totaling RMB 3.3278 trillion in 2023. The average income from industrialized invention patents per enterprise was RMB 8.296 million per patent in the same period. However, despite the substantive quantitative and qualitative growth of scientific, knowledge, and technological achievements, a negligible proportion of the large number of patents have been converted into market applications. Such drastically low transformation efficiency remains a pressing problem that restricts the improvement of China's innovation capacity and economic development (Li, 2012). To secure policy support, many enterprises overstate their R&D expenditure data; however, a large volume of R&D achievements fail to be transformed. Currently, high-quality innovation achievements are scarce: many remain in the laboratory stage, characterized by imbalance between supply and demand. These challenges have severely hindered the STAT potential of enterprises (Sharma et al., 2022). Against this backdrop, identifying a universal, replicable, and scalable empirical model for enterprises' STAT, and enhancing the intensity, breadth, and depth of such transformation, is a high-priority need.
Fueled by progress in artificial intelligence (AI), big data (BD), blockchain, cloud computing (CC), and mobile internet technologies, digital transformation (DT) has emerged as a significant and practical approach for enabling businesses to bring R&D results to market, garnering growing interest from researchers. This transformative process leverages digital technologies to reshape organizational structures, production processes, and operational mechanisms, facilitating a fundamental shift from traditional industrial management systems to digital paradigms (Liang et al., 2017). As STAT is marked by significant uncertainty, considerable irreversibility, and long durations, enterprises increasingly rely on efficient and cost-effective digital solutions for its improvement. Advances in BD and information technologies have significantly enhanced the accuracy of information acquisition, processing, and transmission. These improvements facilitate better matching between technology innovators and end-users, thereby reducing information asymmetry among market participants. The growth of digital platforms has also broadened the scope of technology markets, overcoming geographical and institutional boundaries, and generating new opportunities for transactions and STAT. Despite these advancements, China continues to face significant challenges in STAT. Traditional models, which have long been dominated by universities and research institutions, are increasingly inadequate in the current landscape. They neither align with enterprise-led industry-university-research collaboration nor meet the demands of modern day industrialization and the development of knowledge-based productive forces (Shen et al., 2025). Successful transformation is contingent on overcoming two major obstacles: the "Valley of Death" and the "Darwinian Sea," which involve challenges such as funding shortages, technology-market mismatches, and the development of viable industrialization models. DT is widely regarded as a potential solution to address these barriers. In this context, this study seeks to answer the following questions: Can DT drive STAT in enterprises? What mechanisms underlie this relationship? How does its effect vary across different contexts? We examine how DT shapes the process of STAT by examining the mediating factors of enterprise patent quality, resource allocation efficiency, platform ecosystem embedding, and dynamic capabilities. This research thus contributes a micro-foundational lens to the literature, yielding insights for both theory and practice on enhancing transformation efficacy.
Defined by the integration of digital technologies and data as key productive resources, DT is acknowledged for its ability to improve the efficiency and value of enterprise STAT. However, the precise mechanisms through which this impact is mediated are yet to be fully understood, necessitating in-depth academic research. Two main areas of literature are relevant to this topic. The first highlights the beneficial role of DT in promoting the innovation performance of enterprises, with thorough analyses of the mechanisms involved (Liu et al., 2024a). It covers all stages from product development to market launch. The second stream focuses on factors influencing STAT (Farah & Amara, 2025). Mian et al. (2016) analyze how policy incentives, including fiscal and tax support, coupled with network resources and risk management, collectively shape technology commercialization outcomes among startups, underscoring the importance of integrated policy approaches. Other studies identify market demand, knowledge accessibility, Knowledge Transfer Offices and R&D investment as critical determinants of knowledge transfer and commercialization success (Wang et al., 2013; Compagnucci & Spigarelli, 2024; Wang et al., 2025). The majority of studies are focused on specific regions owing to the scarcity of detailed micro-level data and methodological limitations. Some adopt a macroeconomic approach to evaluate how the digital economy influences the commercialization of high-tech products (Wan et al., 2023; Liu et al., 2024b), while others investigate the role of DT in the context of university technology transfer (Ma & Li, 2022). For instance, Hu et al. (2024) explore how developments in China's technology transaction markets affect enterprises' innovation performance through technology acquisitions.
Notwithstanding the contributions of these studies, few address the mechanisms through which DT influences STAT from micro-enterprise perspective, resulting in this relationship being inadequately defined. Existing research offers limited synthesis of insights pertaining to digital technologies and BD, failing to provide actionable guidance for enterprises. Empirical analyses that incorporate mediating channels such as patent quality, resource allocation efficiency, platform ecosystem embedding, and dynamic capabilities are few and far between, a research gap that this study aims to fill. By building on these foundations, this study empirically examines the mechanisms through which DT affects STAT, with emphasis on patent quality and resource allocation efficiency. Our results elucidate the heterogeneous impacts and implementation challenges of DT.
First, it offers micro-level evidence on how DT facilitates STAT, specifically through patent transfers. Studies have mainly highlighted the impact of DT on technological R&D and patent production (Ma & Li, 2022), but fallen short of addressing STAT outcomes. Yang et al. (2021) put forth theoretical propositions that suggest that digital technology integrates front-end and back-end value chains to improve commercialization efficiency. Distinct from this, we examine how DT influences the intermediate mechanisms that shape enterprises' willingness and capacity to transfer scientific, knowledge, and technological achievements, rather than focusing on their ultimate market outcomes. We employ patent transfer data to directly identify the STAT activities and apply a fixed-effects model for the empirical analysis. Second, this paper identifies and analyzes four fundamental mechanisms to explain how DT supports the STAT process: the enhancement of patent quality, the improvement of resource allocation efficiency, the promotion of platform ecosystem embedding effect, and the enhancement of dynamic capabilities. Third, by focusing on platform ecosystem embedding and patent quality dimensions, this research complements existing region- and industry-level studies, while providing empirical insights on the challenges of low STAT rates among Chinese enterprises. Fourth, to address the common challenge of quantifying STAT, this study utilizes patent assignment records from the China Research Data Services Platform (CNRDS), offering a more precise measure of enterprises' STAT activities. Studies have typically relied on indirect proxies such as patent applications (Hong, 2008), patent grants (Guerrero & Menter, 2024), new product output value (Gaglio et al., 2022), and licensing income (Agrawal & Henderson, 2002; Li & Zhang, 2020). Our measure improves upon these by capturing actual market transactions of patent rights. Lastly, the findings show significant heterogeneity effects: the impact of DT is stronger in enterprises operating in highly competitive sectors and those with higher growth prospects. This indicates that the advantages of DT depend on the competitive environment and the organization's ability to grow. Overall, the results provide useful insights for businesses aiming to integrate DT with their STAT strategies in different contexts.
Theoretical analysis and research hypothesesThe connection between DT and STATThe Law of the People's Republic of China on Promoting the Transformation of Scientific and Technological Achievements, enacted in 2015, provides a firm foundation to accelerate the translation of research achievements into practical productivity. It defines the transformation process as involving experimentation, development, application, and dissemination of results derived from sci-tech activities, ultimately leading to new products, processes, materials, or industries. DT, a complex socio-technical process entailing the large-scale adoption of digital technologies (Forman & van Zeebroeck, 2019), can facilitate STAT. Studies reveal that DT reshapes business processes and value creation mechanisms, breaking down traditional information barriers and resource constraints. By eliminating bottlenecks in innovation chains, it enhances the efficiency of converting scientific achievements into productive outputs (Yang et al., 2021). Digital platforms contribute significantly by integrating innovation resources, optimizing allocation efficiency, and promoting knowledge flows, thereby shortening the technology commercialization cycle. At a regional level, DT supports enterprises by improving information systems, enabling managerial innovation, and refining business models (Fan et al., 2025). Such ecosystems provide enterprises with enhanced access to innovation resources and market opportunities.
Internet-based digital tools reduce information asymmetry and transaction costs, fostering deeper collaboration between enterprises, universities, and research institutes (Linzalone et al., 2020). For instance, digital learning platforms help dissolve organizational and spatial boundaries, enabling cross-organizational knowledge integration and improving communication efficiency. DT also promotes the internalization and recombination of knowledge, spurring cross-department cooperation and enhancing novelty knowledge commercialization (Martínez-Caro et al., 2020). Enterprises leveraging digital integration mechanisms can achieve more efficient technological breakthroughs and market applications. The iterative nature of digital innovation also accelerates the conversion of R&D outcomes into diversified products while reducing uncertainties related to transformation delays. Through interconnected digital devices and intelligent systems, enterprises can perceive external shifts better, identify market opportunities in real time, and implement rapid technological adaptations (Liu et al., 2024a). Real-time feedback from consumers, suppliers, and competitors further provides actionable market intelligence, supporting product optimization and enhancing external innovation capacity (Tan et al., 2015). With DT, enterprises address precision demand, streamline development pathways, and significantly shorten the time from research to market.
Empirical evidence indicates that enterprises employ DT to obtain market information and identify latent customer needs. This capacity facilitates faster technology integration, innovation experimentation, and new product development, thereby accelerating STAT (Mauerhoefer et al., 2017). By utilizing digital analytics (DA), AI, and Internet of Things (IoT) technologies, DT facilitates thorough analysis and precise prediction of market trends, offering businesses real-time and dynamic insights. These features shorten the time between R&D and market implementation, while boosting the likelihood of successful STAT. Digital tools also enable enterprises to track market changes more cost-effectively, quickly respond to shifts in demand and regulations, and overcome information obstacles typical of conventional business models. Businesses can thus enhance resource allocation efficiency and optimize STAT strategies to achieve the best economic results. Specifically, digital platforms support real-time tracking of market trends, competitor behavior, and regulatory developments, optimizing R&D resource allocation and fostering proactive transformation strategies. Such data-driven decision-making reduces uncertainties stemming from information asymmetry and enhances the precision and adaptability of resource management. By integrating multi-source data, including market, behavioral, and technical metrics, enterprises can develop predictive models that offer scientifically grounded decision support. DT thus helps enterprises accurately discern market demands and user preferences (Tan et al., 2015), enabling targeted refinement of technologies and reducing market risks during STAT. Research also suggests that digital enterprises are better equipped to access information on technological outputs, mitigate information asymmetry between stakeholders, and improve the accuracy of technology valuation (Martínez-Caro et al., 2020). Digital platforms consolidate patent data, technical literature, and application cases, creating transparent information channels that enhance match-making efficiency and value realization. Blockchain technology further ensures the authenticity and traceability of technological information, bolstering trust and efficiency in transactions. IoT and CC technologies optimize information dissemination, while AI and blockchain transform scientific, knowledge, and technological achievements into identifiable, tradable digital assets. These developments promote broader knowledge sharing and collaboration among innovators. DA of innovation and transformation processes helps construct multi-dimensional matching mechanisms, shifting traditional linear models into cyclic, efficient systems (Xing et al., 2023). By integrating data from innovation, industrial, and capital chains, DT fosters a closed-loop ecosystem for STAT. We therefore hypothesize that:
Hypothesis 1 DT drives STAT in enterprises.
Patent quality encompasses several dimensions, including technological substance, degree of innovation, practical feasibility, and robustness of legal protection. Enhancing the quality of patents elevates their market value and competitive advantage, thereby increasing their potential for commercial application and adoption. This, in turn, facilitates STAT within enterprises. High-quality patents typically exhibit greater innovativeness, offer solutions to persistent technical challenges, and demonstrate broader applicability, all of which contribute to higher STAT success rates. Digital technologies help improve patent quality by strengthening R&D capabilities, refining innovation management processes, and streamlining patent applications. These enhancements subsequently promote STAT. From a micro-enterprise perspective, DT mitigates external information asymmetry and enhances internal information utilization, key mechanisms for improving resource allocation efficiency. Jiang and Li (2024) note that DT facilitates the growth of total factor productivity through more efficient resource allocation. Resultant from the impact of DT on STAT, heightened resource allocation efficiency allows enterprises to deploy financial, human, and material resources more effectively. Digital tools also enable enterprises to discern market demands and emerging technological trends more accurately, thereby directing resources toward the most promising R&D and commercialization projects. This leads to improved resource utilization, lower operational costs (Hitchen et al., 2017), and ultimately, more successful STAT.
Following from this, this study examines how DT supports STAT through four principal channels: improved patent quality, resource allocation efficiency, promoted platform ecosystem embedding, and enhanced dynamic capabilities. Fig. 1 presents a framework of the underlying mechanisms linking DT to STAT.
DT, patent quality, and STAT
The quality of sci-tech achievements serves as a critical mediator in the pathway from DT to their successful commercialization within enterprises (Ferraris et al., 2017). High-quality patents affect the rate at which enterprises commercialize technological achievements, are typically characterized by strong market relevance and commercial potential, and are more readily transferable and acceptable in the market, thereby enhancing the likelihood of successful STAT. Effective STAT thus relies on a steady supply of high-value patents embodying technical novelty and economic viability, two dimensions that significantly influence commercialization rates (Van Norman & Eisenkot, 2017; Wang et al., 2024). A key obstacle to STAT in China is the low technical and economic value of many technological outputs.
DT supports STAT by facilitating the transfer and assimilation of knowledge and capabilities. It enables information application in production processes and enhances information dissemination through digital networks, thereby accelerating knowledge diffusion within enterprises and promoting the generation of high-quality innovations (Paunov & Rollo, 2016). While DT impacts patent quality positively, its full potential depends on the depth of integration with enterprise operational systems. The embedding of digital technologies into core business processes represents a crucial pathway for improving patent quality (Fang & Liu, 2024). By fostering more dynamic technology markets and enhancing patent quality, DT strengthens STAT capacity (Zhou et al., 2024).
A persistent challenge remains the misalignment between technological outputs and market needs in China. Patent quality, reflected in the novelty, inventiveness, and practical applicability of technologies, significantly shapes STAT success. High-quality achievements are essential for effective market transformation. DT energizes brings fresh energy to both the supply and demand aspects of the STAT ecosystem. On the supply side, it elevates patent quality and improves the output of technological achievements. As policy incentives alone cannot fundamentally improve patent quality, technologies such as BD enhance enterprises' abilities to acquire and create new knowledge (Martínez-Caro et al., 2020), allowing researchers to derive market-informed insights from large-scale data analysis (Zhan et al., 2018). AI and machine learning further improve decision-making in R&D and experimental design, increasing the accuracy and reliability of technological outputs. Digital platforms also enable global collaboration, opening new avenues for research cooperation and enhancing innovation efficiency. We therefore hypothesize that:
Hypothesis 2 DT facilitates STAT in enterprises by improving patent quality.
Schumpeter's theory of innovation suggests that innovation results from the recombination of production factors. DT facilitates this recombination, leading to more efficient resource allocation. STAT depends on the aggregation and effective deployment of resources, with synergistic innovation across all factors being crucial to its improvement. Public platforms dedicated to commercialization attract and concentrate technological and project resources, accelerate the application of cutting-edge results, and leverage key innovation elements such as technology, talent, and capital. DT further enables resource sharing, horizontal coordination, and vertical integration across enterprises, increasing the dynamism of the STAT process. By enhancing resource allocation efficiency, enterprises can utilize information more effectively, identify consumer preferences with greater accuracy, and align the transformation of technological achievements with diverse market needs. The high productivity resulting from DT offers enterprises enhanced resource flexibility, allowing more adaptive and efficient resource deployment. Digital tools, such as DA and AI-powered management systems, enable the integration of extensive data sets, enhancing the transformation and utilization of technologies (Saura, 2021). Through detailed analyses of market and operational data, enterprises can forecast demand variations, optimize production schedules, and prevent inventory surpluses. Such capabilities enhance resource allocation efficiency, accelerate decision-making in STAT, and improve organizational agility in response to shifting market conditions. The convergence of multiple digital technologies strengthens capabilities in information gathering, processing, and sharing. This helps enterprises assimilate diverse knowledge, optimize resource configurations, and stimulate internal innovation (Jiang & Li, 2024). Digital access to academic and patent literature also promotes knowledge dissemination, increasing the likelihood of knowledge being transformed into technological innovations. Lv et al. (2024) show that DT advances innovation capacity and improves resource allocation efficiency, albeit with heterogeneous effects across enterprises. Digital tools alleviate maturity mismatch in enterprise finance, extend beyond mere IT investment to optimize enterprise-wide resource allocation, and generate high returns and operational flexibility. They also reduce information asymmetry, improve transparency, diminish inefficient capital allocation, and boost both investment efficiency and market adaptability (Jiang & Li, 2024).
Superior resource allocation allows enterprises to focus on developing and transferring high-quality patents. Digital applications dissolve traditional innovation boundaries through advanced technological integration, promoting open innovation. They enable efficient merging of external information and technologies with internal resources, facilitating the generation of innovative outputs. Digital technologies improve inter-departmental collaboration and information sharing, increasing the execution efficiency of collective tasks (Briscoe & Rogan, 2016). This reduces redundancy and waste, leading to more optimal resource allocation. Such precise and agile resource management helps direct resources toward core STAT activities, minimizes misallocation, and accelerates both the transformation of technological achievements and new product development (Mubarak et al., 2021). In effect, digital applications allow enterprises to rapidly acquire new knowledge, transcend conventional operational models, and enhance innovation output through improved resource allocation. Underpinned by digital technologies, enterprises can use collaborative platforms to obtain real-time market and industry information, enable frequent information and technology exchange, broaden resource endowments, and cultivate an open innovation environment that supports the efficient flow and allocation of innovation factors. The integration of DT with innovation processes increases the efficiency and transparency of information flows, reduces internal information asymmetry, aids managers in resource integration, and improves environmental sensing, leading to more precise decision-making in critical STAT choices (Gupta et al., 2018). Overall, DT helps enterprises overcome resource constraints, diversify resource allocation methods, and reduce vulnerabilities in core technologies. Accordingly, we propose:
Hypothesis 3 DT promotes STAT in enterprises by increasing the efficiency of resource allocation.
In the digital economy, platform ecosystems represent a significant paradigm for innovation within business ecosystems. They facilitate STAT among enterprises, making them worthy of academic inquiry. Platform ecological embedding refers to the process by which an enterprise integrates into a platform ecosystem to share and utilize resources within it. Through interaction and collaboration with other participants, the enterprise leverages complementary resources to generate synergistic effects. DT serves as the core enabler of platform ecosystems, strengthening connectivity and cooperation among participants (Liu et al., 2023). Technologies such as BD, AI, CC, and IoT allow organizations to acquire, process, and share knowledge and resources more efficiently (Qiao et al., 2024). Digital platforms support efficient knowledge flows between innovators, offering effective pathways for enterprises to access external digital expertise (Liu et al., 2023). This enhanced connectivity enables enterprises to transcend traditional innovation approaches and engage in co-innovation within diverse, multi-actor ecosystems. For example, in healthcare, digital platforms integrate professionals from management, medicine, and information technology to jointly develop knowledge strategies, thereby advancing STAT (Zhao & Canales, 2021).
Platform ecosystem embedding further facilitates STAT by providing an open innovation environment and substantial resource support, which accelerates their commercialization and practical application (Valkokari et al., 2022; Li et al., 2023a). Embedded in a digital platform ecosystem, an enterprise engages in wide-ranging interactions with other actors, which helps it overcome resource constraints and accumulate the necessary assets for substantive innovation (Li et al., 2023a). For instance, platform service portfolio management plays a critical role in social digital platforms. Its core function—developing both standardized technology service portfolios and customized solutions—directly enables STAT within collaborative product development ecosystems (Bao et al., 2024). Platform ecosystems also enhance innovation management efficiency and technological innovation capabilities by promoting inter-enterprise collaborative innovation through industrial chain coordination, open innovation-chain sharing, and the integration of industrial and innovation chains.
The strength of relationships within a platform's innovation network also positively influences knowledge transfer and innovation performance among enterprises (Fritsch & Kauffeld-Monz, 2010). Stronger network ties increase the fluidity of knowledge resources between organizations, thereby creating more favorable conditions for STAT (Van Wijk & Nadolska, 2020). Embedding in a platform ecosystem reduces information asymmetry by providing real-time, transparent information. This lowers search, evaluation, and negotiation costs for both technology suppliers and seekers, consequently facilitating funding acquisition for STAT projects such as green innovation (Qiao et al., 2024). Empirical research on Chinese enterprises shows that digital economic development significantly enhances enterprise productivity, with platform ecosystem integration primarily mediating this relationship (Liu, 2025). Through DT, enterprises can more effectively utilize data-driven organizational structures and business models to build digital ecosystems involving diverse partners, platforms, and stakeholders. This capability supports long-term sustainable development and effective STAT (Xie & Wu, 2024). Accordingly, we propose:
Hypothesis 4 DT promotes STAT in enterprises by enhancing the extent of their platform ecosystem embedding.
Dynamic capabilities theory contends that enterprises sustain competitive advantage by continuously integrating, building, and reconfiguring internal and external resources in response to rapidly changing environments (Bodendorf & Franke, 2024). In the digital era, these capabilities are essential for enterprise survival and growth, as they enable organizations to sense digital opportunities, respond flexibly to market shifts, and effectively translate technology into commercial value (Oliveira et al., 2024). Recent empirical evidence confirms that DT significantly strengthens an enterprise's dynamic capabilities. For example, Chen et al. (2025b) observe that DT within manufacturing enterprises drives technological innovation by enhancing these capabilities in the context of the digital economy and Industry 4.0. Hu and Sun (2025) demonstrate the distinct mediating effect of dynamic capabilities in the relationship between DT strategy and STAT.
Dynamic capabilities facilitate STAT through several mechanisms. First, DT enhances an enterprise's sensing capacity, honing its ability to detect market trends, technological frontiers, and potential innovation opportunities (Chen et al., 2025b). By utilizing DA and AI, enterprises can more accurately identify technological achievements with commercial potential and understand market demands, thereby providing strategic direction for STAT efforts (An et al., 2026). In a healthcare case study, Oliveira et al. (2024) note that design thinking empowers this sensing capacity through a deep understanding of internal and external problems, a key component of DT. Second, DT improves an enterprise's ability to acquire and integrate external knowledge and resources. Dynamic capabilities allow enterprises to form strategic alliances rapidly, engage in collaborative innovation, and absorb external technologies into their innovation systems (Qi et al., 2024). National technology transfer centers, for instance, promote digital innovation in enterprises through technology spillovers, collaborative innovation, and market scale effects (Xiao et al., 2024). A strong absorptive capacity further enables enterprises to assimilate and re-innovate acquired scientific, knowledge, and technological achievements effectively (Ning et al., 2023). Finally, DT empowers enterprises to reconfigure internal resources, processes, and structures to meet the demands of STAT (Akhtar et al., 2025). This includes adapting R&D processes, optimizing production lines, and forming cross-functional teams. Digital tools facilitate more flexible resource allocation, shorten time-to-market, and lower transfer costs (Faro et al., 2024), and employing action design research, emphasize that dynamic capabilities are crucial for balancing agility and resilience during sustained DT. In sum, DT enhances enterprises' absorptive capacity—its ability to identify, assimilate, and apply new knowledge—which in turn promotes the transfer of innovations (Ning et al., 2023). Extending this logic, Zhou and Kuang (2025) find that dynamic capabilities mediate the relationship between DT and enterprise strategic aggressiveness. We therefore propose:
Hypothesis 5 DT promotes STAT in enterprises by enhancing their dynamic capabilities.
In China, STAT primarily occurs through the licensing, transfer, investment (equity participation), and authorization of technology. This entire process aligns with the international concept of "technology transfer." Under China's Patent Law, technology transfer includes the assignment of patent rights, transfer of patent application rights, patent licensing, and the transfer of technical secrets. Given its association with formal changes in ownership and high technological mobility, patent assignment is widely adopted as an indicator of STAT activity. This study relies on established methods used by Agrawal and Henderson (2002), who employed the count of licensed patents as an indicator, and Belderbos et al. (2014), who emphasized the distinct commercialization mechanisms of collaborative versus non-collaborative patents. Following Hu and Wang (2023), we assess STAT using three different metrics: the number of non-collaborative patent transfers, collaborative patent transfers, and their sum. These correspond, respectively, to patents solely applied for and later assigned to third parties, jointly applied for and subsequently transferred, and their sum. To avoid bias from self-transactions, observations where the assignor and assignee are identical are excluded. The natural logarithm of the total number of patent assignments is used to measure the enterprise's STAT level.
Explanatory variableThis study adopts the text-based methodology introduced by Xiong et al. (2025) to assess the level of enterprise DT. Specifically, we identify and count keywords related to digital technology in the annual reports of public listed firms, and calculate the logarithm of the total keyword frequency plus one (Leng & Zhang, 2024; Du et al., 2025). Our theoretical framework defines DT through two aspects: foundational technologies and practical applications, measured by five sub-indicators: AI, BD, CC, DA, and the application of digital technology (ADT). The data are acquired from the Management Discussion and Analysis (MD&A) sections of annual reports, which provide detailed and reliable information about enterprises' digital efforts (Guo et al., 2023). A composite DT index constructed using these data serves as the empirical proxy for DT level.
Control variablesFollowing Guo et al. (2023) and Du et al. (2025), this study includes firm-level control variables, and applies year and industry fixed effects. A summary of all variable definitions is provided in Table 1. The controls are firm age (FirmAge), management shareholding ratio (Mshare), ownership percentage of the largest shareholder (Top1), asset turnover ratio (ATO), leverage ratio (Lev), cash flow ratio (Cashflow), and return on assets (ROA).
Variable construction.
To explore the effect of DT on enterprise STAT, this study establishes the following model:
In this specification, STATi, t represents the STAT level for firm i in year t, DTi, t indicates the DT level of firm i in year t, and Controli, t is a vector of firm-level control variables. Year and Industry fixed effects are incorporated to account for time-related and industry-specific differences. The error term εi, t captures all unobserved factors. A positive and statistically significant estimate of α1 would validate the research hypothesis. We estimate all models using a two-way fixed effects approach (by year and industry) to control for unobserved heterogeneity, and we report t-statistics calculated with standard errors clustered at the firm level.
Mediating effect modelThe theoretical analysis suggests that DT enhances STAT by improving patent quality, the efficiency of resource allocation, the extent of platform ecosystem embedding, and dynamic capabilities. Following Chen et al. (2023), this study examines these mechanisms by constructing specialized models to test the mediating effects of each pathway.
The subscripts i and t represent the firm and the year, respectively. Mediatori, t represents mechanism variables, while all other variables follow the notation of the baseline regression model (1).
Data sourcesThis research uses a sample of A-share firms listed on Chinese stock exchanges covering the years 2001 to 2023. Data on patent transfers were sourced from the Patent Transfer Research Database provided by the CNRDS, which includes detailed information such as application numbers, assignors, assignees, and transfer execution dates. The dataset was constructed by matching patent transfer records from 2001 to 2023 with corresponding listed companies using unique identifiers. It involved the following steps: 1. excluding financial companies and those labeled as ST, *ST, or PT; 2. deleting companies that have been delisted; 3. excluding records with missing key variable data; and 4. applying winsorization to all continuous variables at the 1st and 99th percentiles to reduce the impact of outliers. The resulting dataset included 10,987 firm-year observations. Financial information was sourced from the Wind and the China Stock Market and Accounting Research databases, while company annual reports were obtained from the official websites of the Shenzhen and Shanghai Stock Exchanges. Basic patent details—such as application numbers, applicants, grant dates, and filing dates—were gathered from the China National Intellectual Property Administration, covering patents filed and granted between 2001 and 2023.
Analysis of empirical resultsDescriptive statisticsTable 2 shows that the variable STAT, which measures the level of STAT, ranges from 0 to 5.509, with an average value of 1.576. Over half of the enterprises have values below this average, reflecting the generally low STAT levels observed in China. The DT variable ranges from 0 to 6.380, with an average of 1.720. This distribution suggests that a majority of enterprises trail behind the sample average in DT. This pattern aligns with findings from the 2023 Enterprise Digital Transformation Index Report, which indicates that only 2 % of Chinese enterprises are classified as "digital transformation leaders." The consistency between these independent sources highlights the ongoing challenges and modest pace of digital adoption among Chinese enterprises. The control variables demonstrate distributions that are broadly consistent with prior studies and are therefore not discussed in detail here. Methodologically, the coherence between our sample statistics and established macro-level trends supports the representativeness of the data and its relevance to current industrial conditions.
Descriptive statistics and correlations.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
Table 2 shows the Pearson correlation matrix for the key variables. There is a positive and statistically significant correlation (p < 0.01) between DT and STAT, which supports a meaningful association between these constructs. Among the control variables, Lev shows the highest correlation with STAT, followed by FirmAge, suggesting that both financial structure and organizational maturity influence enterprises' commercialization performance. To evaluate possible multicollinearity in the initial regression model, variance inflation factors (VIFs) were analyzed. The highest VIF observed is 1.49, and the average VIF is 1.2, both significantly lower than typical critical limits. This suggests that multicollinearity is not a significant issue in the model, reinforcing the reliability of the estimated findings.
Baseline regression analysisColumn (1) of Table 3 shows that DT has a significant positive impact (coefficient = 0.118, p < 0.01) on STAT when control variables are not included. This positive association becomes stronger (coefficient = 0.131, p < 0.01) after adding control variables in Column (2), providing preliminary evidence in favor of Hypothesis 1.
Benchmark regression results.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
To correct for possible sample selection bias, we use the Heckman two-step approach. In the first step, a Probit model is estimated to calculate the Inverse Mills Ratio (IMR). This model includes a dummy variable (dumDT) which takes the value 1 if the firm has adopted DT and 0 if not. Following Ren et al. (2023), we incorporate the instrument IV7, defined as the average level of DT among other enterprises within the same city, industry, and year. This variable satisfies the relevance condition, as enterprises' DT levels are often influenced by local and sectoral peers, while also meeting the exclusion restriction since peer digitalization is unlikely to directly affect the focal enterprise's STAT. The IMR is included as an extra control variable in the second-stage regression. As indicated in Table 4, IV7 has a significant positive relationship with dumDT in the first stage, confirming the instrument's strength. After accounting for IMR, the second-stage results are qualitatively consistent with our baseline estimates, indicating the robustness of the main findings.
Endogeneity tests: Heckman.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
Another potential source of endogeneity is reverse causality. While DT can facilitate STAT, enterprises with stronger STAT capabilities—often reflected in greater R&D investment and higher innovative output—may also exhibit a greater propensity to invest in DT. This study employs a two-stage least squares (2SLS) method with two different instrumental variables to more rigorously identify the causal relationship between DT and STAT. Table 5 shows the 2SLS results using these two instruments. Based on Lewbel (1997),1 the first instrumental variable (IV1) is created by cubing the difference between an enterprise's DT level and the average DT level in its industry and province. Columns (1) and (2) present the findings using IV1. In the first stage, IV1 has a positive and statistically significant coefficient at the 1 % level. The instrument's validity is supported by the Kleibergen–Paap rk LM test (p < 0.01), which rejects under-identification, and the Cragg–Donald Wald F-statistic of 6719.611, which is well above the Stock–Yogo critical value of 16.38, indicating no weak instrument issues. In the second stage, the coefficient for DT remains positive and significant at the 1 % level, aligning with our main hypothesis. Following Gopalan et al. (2022), Columns (3) and (4) use the second instrument, IV8, defined as the DT level of industry leaders. The first-stage result for IV8 is also positive and significant at the 1 % level. The Cragg–Donald Wald F-statistic of 1274.403 surpasses the Stock–Yogo critical value (16.38 at the 10 % level), confirming the instrument's strength and relevance. The second-stage estimates also reveal a positive and highly significant coefficient for DT (p < 0.01), further validating Hypothesis 1.
Endogeneity tests: Instrumental variables.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
To address the potential time lag in the impact of DT on STAT, this study applies a lag structure to the DT variable. This methodological adjustment also helps alleviate potential reverse causality concerns. The sustained effect of DT on STAT is examined by extending the observation window. Models (2) through (5) in Table 6 show the results including lagged values of the main explanatory variable from one to four time periods. DT continues to have a positive and statistically significant impact (at the 1 % level) on STAT throughout all lag intervals, without any notable decrease over time. These findings suggest that DT has a sustained and cumulative influence on enterprises' STAT, corroborating the main hypothesis. Specifically, DT enhances continuous improvements in resource allocation efficiency, innovation capability, and knowledge dissemination, which collectively support the market application and STAT over both short and long terms. By reducing information asymmetry, increasing transparency in technology transactions, and enabling cross-domain collaboration, DT contributes to a lasting increase in patent transfers. This highlights its importance for both short-term efficiency improvements and the long-term enhancement of enterprises' innovation environments.
Robustness test: Extended observation window.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
STAT often involves considerable time lags between initial inputs and eventual outcomes. Reflecting this long-cycle characteristic, we analyze the delayed impact of DT on STAT over one- and two-period horizons. The findings (Table 7) show that DT has a positive and statistically significant (at the 1 % level) effect on STAT in both following periods. Notably, the magnitude and significance of the coefficients increase relative to the contemporaneous results, suggesting that the beneficial effect of DT strengthens over time.
Robustness test: Considering the long-term nature of the STAT process.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
Two alternative econometric strategies are employed to ensure the robustness of findings. First, since patent transfers represent a count variable with a high proportion of zeros and a skewed positive distribution, we re-estimate the model using a Tobit specification to account for censoring and distributional characteristics. Second, to address potential omitted variable bias more rigorously than standard two-way fixed effects, we incorporate high-dimensional fixed effects by fully interacting time and industry indicators. Results from these approaches, presented in Table 8, align closely with the baseline estimates. DT retains a statistically and economically significant positive effect on STAT, supporting the main hypothesis.
Robustness test: Alternative measurement methods.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
Enterprises may at times overstate their DT efforts to appeal to capital markets. To reduce potential bias from such strategic disclosure, following Ricci et al. (2020), we exclude observations reporting zero digitalization intensity. This restriction improves the credibility of the sample's disclosure practices. As indicated in Table 9, the estimated impact of DT continues to be positive and statistically significant, aligning with the baseline findings and supporting Hypothesis 1.
Robustness test: Excluding the influence of certain factors.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
We undertake a final robustness test by employing alternative measures for both the explained and explanatory variables. We address potential measurement issues in patent-based outcomes by constructing a binary variable indicating whether an enterprise's number of patent assignments exceeds the sample median (1 if above, 0 otherwise). As indicated in Column (1) of Table 10, DT significantly raises the likelihood of belonging to the high-patent-transfer category, reinforcing the initial results. We also adopt a refined measure of DT (DT_C) following Fang and Liu (2024), which incorporates 139 digital-related keywords across six dimensions including strategic orientation and technology adoption. To mitigate right-skewness, we use the log-transformed word frequency (plus one). The findings in Column (2) of Table 10 confirm a strong positive correlation between DT_C and STAT, supporting the reliability of the conclusions.
Robustness test: Variable replacement.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
This study uses the International Patent Classification (IPC) system to evaluate patent quality. The Chinese IPC system is organized into five hierarchical levels: section, class, subclass, main group, and subgroup. Following Zhang and Li (2010), we measure patent quality by knowledge breadth, which is defined as the total knowledge complexity within a company's patent portfolio, based on the assumption that a broader knowledge scope indicates higher quality. Using only the number of primary classification codes may overestimate quality and mask internal heterogeneity. To enhance measurement precision and discriminative validity, the measurement of knowledge breadth is refined at the primary group level by applying a Herfindahl-index algorithm.
Among these variables, Zimt stands for the cumulative number of invention and utility model patent applications submitted by firm i within technology group m up to year t, while Zit represents the total number of patent applications filed by the same firm across all technology groups. As the value of Patentknowledgeit increases, the knowledge breadth covered by the firm's patents expands significantly, which, in turn, enhances patent quality through the effect of technological diversity.
Masur (2010) demonstrates that the patent examination process effectively filters out low-quality applications, leading to higher average quality among granted patents compared to filed ones. Granted patents thus constitute a more accurate reflection of an enterprise's innovative capability and patent value.2 In line with recent empirical work (Zhou et al., 2024), the results show that DT significantly enhances the quality of both patent applications and granted patents (Table 11). Improvements in enterprises' technological innovation efficiency depend on the deliberate adoption of digital technologies that align with operational and managerial needs. Tailored digital solutions enable substantive enhancements in patent quality. A 1 % increase in the quality of patent applications and granted patents corresponds to a 0.010 % and 0.012 % increase, respectively, in the STAT rate. This suggests that higher patent quality, reflecting greater innovation capacity, facilitates STAT, with granted patents exhibiting a stronger marginal effect. These results corroborate those of (Zhuo & Chen, 2023), who find that DT in the new digital economy enhances the quality, originality, and applicability of enterprise sci-tech outputs, thereby promoting the commercialization of patented technologies and their subsequent market success. Our empirical findings thus support Hypothesis 2, indicating that DT promotes STAT through improvements in patent quality.
Mechanism test: Patent quality.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
DT plays a role in optimizing innovation governance by significantly mitigating over-investment and improving resource allocation accuracy. These mechanisms subsequently enhance the efficiency of channeling resources toward STAT. Building on the methods of Biddle et al. (2009) and Hirshleifer et al. (2013), this paper assesses resource allocation efficiency (lneff) by determining the optimal amount of investment for a company in a specific year and then measuring the degree of excessive investment. Over-investment reflects the allocation of excessive resources to low-return projects, resulting in resource waste, elevated opportunity and capital costs, and increased managerial complexity and financial risks. These factors collectively reduce resource allocation efficiency and often lead enterprises to adopt more conservative STAT strategies. Column (2) of Table 12 shows that DT significantly decreases over-investment at the 1 % significance level, demonstrating a beneficial impact on the efficiency of resource allocation. This allows companies to allocate resources more precisely to STAT, reduce wasteful investments, and improve the alignment between supply and demand during the transformation process. These results resonate with (Jiang & Li, 2024) finding that DT helps correct resource misallocation and improves allocation efficiency, thereby fostering innovation output and innovation governance. These empirical findings support Hypothesis 3, demonstrating that DT facilitates STAT by enhancing resource allocation efficiency.
Mechanism test: Enterprise resource allocation efficiency.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
Embedding into platform ecosystems constitutes a key strategic priority for enterprises undergoing DT and pursuing high-quality development. Annual reports, as comprehensive and forward-looking documents, routinely contain disclosures that reflect this strategic orientation. The presence of terminology related to "platform ecosystem embedding" in these reports signals an enterprise's strategic intent, future planning, and projected pathways for engaging with digital ecosystems. The formal and regulated nature of annual reporting also lends credibility to these disclosures, suggesting a commitment to implementation and offering meaningful insight into managerial decision-making. Therefore, measuring the extent of platform ecosystem embedding by extracting relevant term frequencies from listed firms' annual reports is a feasible and methodologically sound approach. In line with the methods adopted by Chen et al. (2025a) and Chen and Li (2024), this study applies Python-based text analysis to annual reports. We capture term frequencies across three dimensions—strategic embedding, platform embedding, and ecological embedding—to quantify the extent of a firm's platform ecosystem embedding. The raw frequency counts for each firm are then log-transformed to normalize the distribution. The specific keywords used in this analysis are summarized in Table 13.
Statistical table of feature words for platform ecosystem embedding.
Drawing on the digital innovation ecosystem research (Liu et al., 2023; Chen et al., 2025a), platform ecosystem embedding reflects not only an enterprise's technological connectivity but also the strategic depth of its participation in open innovation collaboration. Column (2) of Table 14 indicates that the level of DT exerts a strong positive and statistically significant impact on the extent of platform ecosystem embedding (coefficient = 0.786, t = 203.78). This suggests that digitally advanced enterprises are more likely to engage systematically with digital platform networks across strategic, technical, and relational dimensions. Platform ecosystem embedding further acts as a key mediating pathway between DT and STAT. DT directly facilitates STAT, while also strengthening an enterprise's capacity to absorb, integrate, and transform external innovative resources through deeper embedding in three ways: platform embedding (adopting specific digital technologies and tools), strategic embedding (incorporating platform logic into strategic orientation), and ecological embedding (engaging in cross-organizational collaboration and value-network building). Among the control variables, FirmAge shows a consistently positive and significant coefficient across both models, indicating that established firms possess structural advantages in ecosystem integration and STAT. These results support the proposed "DT → platform ecosystem embedding → STAT" mechanism, thus confirming Hypothesis 4 that DT promotes STAT by enhancing the extent of platform ecosystem embedding. They imply that successful DT depends on technology adoption as much as on systematic embedding into platform ecosystems to effectively convert digital resources into transferable and commercializable innovations.
Mechanism test: The extent of platform ecological embedding.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
The predominant use of cross-sectional data through survey-based methods to measure dynamic capabilities fails to capture how these capabilities evolve within enterprises over time. Based on the dynamic capabilities theory (Teece, 2007), we address this limitation by constructing a panel data set adopting the measurement approach outlined by Hu and Sun (2025). Dynamic capabilities are operationalized across three dimensions: innovation capability, absorptive capacity, and adaptive capability. Innovation capability is measured using a composite index based on two indicators: R&D intensity and the proportion of technical personnel. These indicators are first standardized and then summed to derive a firm-level innovation capability score (see Eq. (4)). Absorptive capacity is proxied by R&D expenditure intensity, calculated as the ratio of a firm's annual R&D spending to its operating revenue. Adaptive capability is measured by the flexibility of a firm's resource allocation, reflected in the coefficient of variation (CV) of its annual expenditures across three categories: R&D, capital, and advertising. To ensure that higher values correspond to greater adaptability, the calculated CV is multiplied by –1. Thus, a larger adjusted CV indicates stronger adaptive capability.
The regression results in Table 15 provide evidence of the mediating role of dynamic capabilities in the relationship between DT and STAT. DT shows a significant positive effect on each dimension of dynamic capabilities: absorptive capacity (coefficient = 0.009, t = 13.83), innovation capability (coefficient = 0.046, t = 14.20), and adaptive capability (coefficient = 0.043, t = 23.33). These results indicate that DT acts as a technological enabler and key driver in reshaping internal capabilities and fostering dynamic capabilities within enterprises. A closer examination reveals differentiated pathways through which DT, via enhanced dynamic capabilities, facilitates STAT. Improvements in absorptive capacity enable enterprises to better identify and incorporate external knowledge, accelerating technology integration. Strengthened innovation capability drives more market-oriented R&D outputs, while increased adaptive capacity allows enterprises to reallocate resources more flexibly to navigate uncertainties inherent in the innovation process. Overall, the findings in Table 15 corroborate the theoretical sequence of DT → dynamic capabilities enhancement → STAT, underscoring that cultivating and upgrading dynamic capabilities is essential for enterprises to effectively translate digital resources into sustained innovation performance. The empirical evidence thus validates Hypothesis 5, demonstrating that DT promotes STAT by strengthening enterprises' dynamic capabilities.
Mechanism test: Enterprise dynamic capabilities.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
This section incorporates financial supervision as a contextual moderator into the "DT–STAT" framework. Drawing on Li and Hou (2025), we construct a measure of regional financial supervision intensity defined as the ratio of regional financial regulatory expenditure to the value added of the financial sector (see Eq. (5)).
This ratio accounts for scale differences across regions. We further create a dummy variable, H_Supervision, which equals 1 if a region's supervision intensity exceeds the sample median, and 0 otherwise. Both H_Supervision and its interaction term with DT are included in the regression model. Column (1) of Table 16 reports that the coefficient on DT × H_Supervision is positive and statistically significant at the 1 % level. This suggests that the effect of DT on STAT is stronger in regions with higher financial supervision intensity, consistent with theoretical expectations.
Heterogeneity analysis: Regional financial supervision intensity and market competition environment.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
Market competition dynamically shapes the alignment between technology and market needs, with an enterprise's external environment significantly affecting the acceptance of innovative technologies. We capture this by computing the Herfindahl–Hirschman Index (HHI) based on enterprise sales revenue for each industry-year as a measure of market concentration. A dummy variable, H_HHI, is defined, taking the value of 1 in industries where competition is lower than the median (i.e., higher concentration), and 0 otherwise. We introduce H_HHI and its interaction with DT into the regression model. Column (2) of Table 16 shows that the coefficient on DT × H_HHI is positive and significant at the 1 % level, indicating that DT has a stronger effect on STAT in more competitive markets. The result aligns with theoretical expectations and supports the dynamic capabilities perspective, which posits a synergy between market dynamism and innovation efficiency (Teece, 2007). In highly competitive settings, enterprises may rely more on targeted digital investments to respond rapidly to market shifts, thus gaining near-term advantages in STAT. Overall, DT appears to play a pronounced role in promoting STAT among enterprises situated in environments characterized by both strong financial supervision and intense market competition.
Industry characteristicsThe structural features of competitive and regulated industries influence enterprises' selection of upstream and downstream partners, which subsequently affects external transaction costs (Acemoglu et al., 2010). Enterprises in competitive industries typically have access to multiple alternative partners even after terminating existing supply-chain relationships. However, this flexibility also increases their exposure to opportunistic behavior from transaction counterparts, resulting in higher external transaction costs relative to regulated industries. We therefore expect that DT will reduce external transaction costs more substantially and consequently enhance STAT more effectively in competitive industry settings.
Based on the China Securities Regulatory Commission Industry Classification Guidelines (2012 Edition), we divide industries into regulated and competitive categories. The regression outcomes shown in Columns (1) and (2) of Table 17 reveal that the coefficient for DT is 0.166 (significant at the 1 % level) in competitive industries, which is notably higher than that in regulated industries (coefficient = 0.060, p < 0.01). This indicates that the impact of DT on STAT is stronger in competitive market settings.
Heterogeneity analysis: Industry characteristics.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
The management expense ratio, determined by dividing management expenses by operating revenue, indicates how efficiently a company controls its internal costs. A dummy variable, L_Mfee, is defined to equal 1 if an enterprise's ratio is below the sample median, and 0 otherwise. This variable and its interaction with DT (DT × L_Mfee) are included in the regression. The interaction term DT × L_Mfee in Column (1) of Table 18 shows a significantly positive coefficient at the 1 % level, indicating that DT has a greater effect on STAT for companies with lower management expense ratios.
Heterogeneity analysis: Management expense ratio and firm growth.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
The relationship between technology and market fit is dynamic and influenced by a firm's growth stage and external environment (Teece, 2007). Firms in early growth stages typically exhibit simpler organizational structures and lower internal control costs, while more mature firms often face greater complexity and higher costs. Therefore, firm growth is negatively associated with internal control costs. We assess firm growth by the revenue growth rate, calculated as (current year revenue / previous year revenue) minus 1. We create a binary variable, H_Growth, which is set to 1 if a firm's growth rate exceeds the sample median, and 0 otherwise. This variable, along with its interaction term with DT (DT × H_Growth), is included in the model. The findings shown in Column (2) of Table 18 indicate that the coefficient for the interaction term is positive and highly significant at the 1 % level, implying that DT has a stronger effect in enhancing STAT in firms experiencing high growth. This result supports the dynamic capabilities theory, which highlights synergies between organizational growth and innovation efficiency (Teece, 2007). Overall, DT facilitates STAT more strongly in firms with lower management expense ratios and higher growth rates.
Firm ownership structureInstitutional frameworks, including reforms introducing mixed ownership of state-owned assets, play an important role in improving the efficiency of STAT, promoting collaboration among industry, academia, and research institutes, and shaping organizational approaches to commercialization. In practice, technology transfers between state-owned enterprises (SOEs) are frequently influenced by macro-level objectives such as national strategic layout, industrial upgrading, and the optimized allocation of state-owned assets, rather than by profit maximization alone. SOEs may also undertake context-appropriate technology transfers to firms in underdeveloped regions to support local economic development and employment, thereby meeting their social responsibilities. Using these institutional and behavioral differences, we categorize firms into SOEs and non-SOEs and conduct the regression analysis separately for each category.
Table 19 reports these results, which show that the DT coefficient is positive and statistically significant in both Columns (1) and (2), although its size varies between SOEs and non-SOEs. This suggests that DT has a greater impact on STAT in SOEs compared to non-SOEs, which aligns with previous studies (Ma & Li, 2022). SOEs benefit from several advantages that enhance their responsiveness to DT. These include larger operational scale, stronger human capital, greater government support, as well as structural strength in ownership arrangement, organizational coordination, and resource sharing. These features improve their absorptive and transformative capacity for knowledge and technology. The public ownership structure facilitates tight-knit innovation networks among state-owned holding companies, subsidiaries, and affiliated entities, allowing efficient sharing and integration of innovation resources such as technologies or intellectual property (IP). This significantly lowers institutional barriers to STAT. By contrast, non-SOEs often encounter obstacles such as proprietary technology protection, information silos, and competitive barriers, which limit internal and external flows of innovation resources and impede effective STAT. The superior access of SOEs to policy support, resource allocation mechanisms, and long-term strategic planning further augments their DT capabilities and STAT efficiency. The collaborative innovation environment and innate knowledge-sharing culture of SOEs create more favorable conditions for STAT, whereas the market-driven nature of non-SOEs constrains the depth and scale of their transformation efforts. Stronger governmental backing and policy preferences for SOEs may reduce the opportunities available to non-SOEs. Leading academic institutions may also prefer to collaborate with SOEs in initiatives such as postdoctoral research stations, which implicitly strengthen knowledge transfer and commercialization channels between universities and SOEs, further amplifying the role of DT in SOEs. It is thus evident that DT exerts a more substantial influence on STAT in SOEs compared to non-SOEs.
Heterogeneity analysis: Ownership structure.
Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.
DT serves as a critical catalyst for advancing STAT, constituting an essential driver for developing knowledge-based productive forces. A thorough examination of this issue can offer valuable insights for managing innovation within enterprises, thus aiding China's goal of becoming a world leader in science and technology innovation. This research uses extensive data from Chinese A-share listed companies from 2001 to 2023 to explore this relationship from various analytical perspectives. The empirical results reveal several important conclusions. First, DT has a substantial and significant positive impact on STAT. This result is consistent after accounting for potential endogeneity through instrumental variable methods and Heckman's two-stage model. It also holds up under further robustness tests such as variable replacements and longer observation periods. Second, the mechanism analysis reveals that DT facilitates STAT through four distinct pathways: enhancing patent quality, optimizing resource allocation efficiency, promoting platform ecosystem embedding extent, and enhancing dynamic capabilities. DT enhances enterprises' knowledge absorption capacity by improving the knowledge breadth and quality of their patent portfolios. Simultaneously, it optimizes resource allocation, demonstrating effective innovation governance. The four pathways that jointly facilitate STAT thus suggest that enterprises should strategically prioritize DT initiatives, with emphasis on developing high-quality patent portfolios and optimizing resource allocation mechanisms. Third, heterogeneity tests reveal that the positive impact of DT on STAT is particularly significant in companies with lower management expenses, greater growth prospects, more intense market competition, and state ownership. It is also pronounced in enterprises operating in environments with strong financial regulation and high market competition. Policy makers may consider targeted support for SOEs and enterprises in competitive industries to maximize the commercialization benefits of DT.
This empirical analysis focuses primarily on resource and capability-based mechanisms internal to the firm. However, the findings also inform broader theoretical concepts. For instance, the platform ecosystem embedding effect driven by DT could potentially restructure entire regional or industrial innovation ecosystems at a macro level. Likewise, the differential impacts of DT on SOEs and non-SOEs points to complex interactions between distinct institutional frameworks, such as market versus administrative logic. While beyond the scope of the current empirical examination, these extended theoretical aspects offer productive avenues for future studies, which could advance in two key directions. First, they can use richer micro-level data, such as records of digital investments within firms or network data on R&D teams, to uncover granular organizational processes underlying the mechanisms identified here. Second, they can employ inter-organizational network data, such as patent citation or R&D collaboration networks, to examine how DT reconfigures knowledge-flow and value-creation networks from an ecosystem perspective. These would extend the present framework toward a more systemic level of analysis.
Theoretical contributionsThis study makes critical contributions to the theoretical literature. First, it helps fill the research gap on digitalization-driven STAT, offering new theoretical and empirical perspectives on factors influencing commercialization performance. It enriches the theoretical foundation of digitalization-supported STAT activities and expands the literature on knowledge transfer. Prior research has primarily concentrated on the R&D phase of innovation (Li et al., 2023b; Misati et al., 2024) or on theoretical discussions (Gaglio et al., 2022; Farah & Amara, 2025). This study fills a critical lacuna on the empirical aspects by examining how DT impacts the STAT process at the micro-enterprise level. It broadens the conversation by focusing on the STAT stage within enterprises, thereby enabling the assessment of the impact of DT on innovation. Second, this study confirms the mediating effect of patent quality, measured in terms of knowledge breadth, in the relationship between DT and STAT, another area that has received little empirical attention. It offers micro-level evidence to tackle the issue of low STAT rates in China by incorporating resource allocation efficiency, measured through over-investment, as an additional mechanism. Both platform ecosystem embedding and enterprise dynamic capabilities function as critical mediating "bridges" in the relationship between DT and STAT. However, current research lacks an in-depth quantitative analysis and robust evaluation of this mediating effect. There is a clear need for further investigation to quantify its magnitude, clarify its characteristics, and address related methodological and contextual challenges. Lastly, the study identifies heterogeneous effects across institutional and market environment, and internal governance structure, such as financial regulation intensity, market competition environment, management efficiency, growth stage, and ownership type, thereby broadening the scope of research on the relationship between DT and STAT. These findings advance our understanding of how digital capabilities transform innovation ecosystems and knowledge commercialization processes, and offer practical guidance for enterprises seeking to align DT with STAT.
Policy recommendationsCapitalize on the transformative potential of DT to expedite the STAT advances into market-ready solutionsEnterprises should actively adopt digital technologies to analyze market dynamics, bridge virtual and physical innovation activities, and coordinate diverse digital ecosystems. These abilities would enable companies to recognize market opportunities more precisely, minimize investment risks, and speed up the STAT process. To mitigate information asymmetry and foster a sophisticated innovation ecology, it is essential to strengthen digital platforms for technology transfer. Governments should improve policy frameworks and establish diversified funding mechanisms to lower commercialization costs and enhance operational efficiency. Creating a multi-tiered, field-specific, and open repository for sci-tech projects, closely aligned with industrial needs, will help broaden the sources of STAT. Deepening industry-university-research collaborations, enhancing coordination between central and local governments as well as between large and small enterprises, and integrating public and value-added services are essential to integrating R&D, testing, and market promotion.
Prioritize patent quality improvement as a key strategic channelEnterprises should increase R&D investment, conduct in-depth market research, and strengthen management and collaborative exchanges to improve the practicality and novelty of patents. Strategic patent layouts tailored to operational realities can enhance utilization efficiency and support product quality improvements and new product development. Governments could consider designing a "patent quality-oriented" digital subsidy contract, incorporating indicators such as "post-transfer maintenance duration" and "citation frequency" into fiscal subsidy agreements. An "ex-post reimbursement coupled with performance-based incentives" mechanism could be adopted to mitigate rent-seeking behavior and inefficient transfers. Additionally, blockchain-based smart contracts could be utilized to automatically trigger subsidy disbursement, thereby reducing administrative review costs. Governments should also reinforce IP protection, establish sound incentive mechanisms, and refine patent examination systems to reduce infringement risks and encourage high-quality patent creation. Introducing market-oriented valuation mechanisms into the patent evaluation system will provide a reliable basis for STAT decisions. Optimizing patent administration and establishing digital service platforms for patent transactions can significantly reduce information asymmetry.
Policy pathways for optimizing innovation resource allocationGovernments should refine fiscal support and innovation subsidy policies, enhance targeting in digital resource allocation, and invest in data infrastructure to construct a supportive environment for digital applications. Enterprises should be encouraged to form talent alliances with industry associations and research institutes, co-build shared databases, and strengthen the agglomeration of sci-tech talent. Establishing investment evaluation systems to assess DT projects and prioritizing those with high return potential will boost resource efficiency. Market-oriented institutional mechanisms should be anchored in the reform of the national innovation system, fostering an environment conducive to innovation, enabling enterprises to become primary actors in innovation. Differentiated support policies based on enterprise growth stage and needs, particularly for non-state-owned and innovative enterprises, can alleviate resource constraints and stimulate STAT.
External ecosystem collaboration and internal capability reconfigurationEnterprises should strategically embed themselves within platform ecosystems to access innovative resources, while simultaneously cultivating dynamic capabilities to assimilate and transform those resources. Correspondingly, public policy should adopt a dual-track approach, establishing an institutional environment that systematically enables ecosystem development and stimulates organizational capability building. Efforts should focus on constructing and opening industry-level digital infrastructure, formulating secure and credible data circulation protocols, and fostering third-party service platforms. These steps can reduce the technical and institutional barriers to integrating into external ecosystems. Policy incentives should concomitantly shift from subsidizing hardware and software procurement toward supporting enterprises' participation in ecosystem-based collaborative innovation and dynamic capability development programs. Integrated guidance combining technical and managerial dimensions should also be provided to help enterprises internalize digital tools as sustainable, innovation-oriented organizational capacities. By concurrently establishing platform-ecosystem linkages and empowering the growth of dynamic capabilities, digital potential can be methodically translated into high-quality innovation momentum.
A combination of effective regulation and robust competitionA well-regulated and fully competitive environment is crucial for maximizing the effects of digital innovation. The ideal macro-environment is one that is stable and orderly, yet dynamic and open—achieved by combining effective regulation, which provides clear institutional expectations and uses RegTech to foster innovation, with robust competition, which removes market barriers and maintains vitality. Enterprises, in turn, must develop strategic adaptability. In strongly regulated contexts, they should internalize compliance as trust capital and pursue "regulation-friendly" innovation. In highly competitive markets, they must convert digital tools into core competencies for agile response and differentiation. This enables precise alignment between external conditions and internal strategy. To cultivate such an ecosystem, policy should coordinate efforts across two complementary fronts. First, regulatory frameworks should be differentiated and nuanced. While safeguarding against systemic financial risks, regulators can create "green channels" to accelerate the approval of digital financial products and services that promote technological innovation and green transitions. Expanding regulatory sandboxes allows enterprises to test innovative digital business models within a controlled and time-bound setting, for instance, in supply chain fintech or digital IP pledge platforms, enabling rules to adapt dynamically and balance innovation with risk. Second, fair market access is essential, particularly in platform-based and data-driven sectors. Policies should prevent dominant players from using data or capital advantages to foreclose markets, ensuring that small and medium-sized enterprises (SMEs) and innovators can participate equitably in digital ecosystems. Vigilance against digital "winner takes most" effects, such as through anti-competitive mergers or exclusive agreements, helps preserve market dynamism and innovative dynamism. Policy should also encourage digitally enabled collaboration along industrial and innovation chains, supporting clusters of innovative SMEs. The development of specialized digital industrial parks or innovation communities can foster knowledge spillovers and collective learning, creating healthy networks of cooperation alongside competition.
Limitations and research perspectivesAgainst the backdrop of rapid digitalization, this study empirically demonstrates that DT impacts STAT through a range of mediating pathways. The following limitations of the study warrant further investigation. First, due to data availability constraints, enterprise DT levels are proxied using annual report disclosures, which may not fully capture the actual extent of transformation. Future research could integrate machine learning methods with field survey data to develop more accurate measures. Second, the mechanisms through which DT influences STAT warrant further theoretical expansion and empirical refinement. While this study examines the role of resource allocation efficiency, patent quality, and dynamic capabilities in influencing STAT, future research could extend this analytical framework in several directions. For example, examining the bi-directional relationship between DT and patent quality would offer a more dynamic perspective. Developing a technology-adaptation index in digital contexts could help quantify how well technological characteristics align with evolving market demand. Employing survey and qualitative methods to directly measure institutional perceptions and logic conflicts would add nuance to mechanistic explanations. Integrating data from China's technology transaction networks and Wind IP investment and financing databases to construct multi‑actor dynamic network models would further illuminate how DT reshapes ecosystem structures—such as node centrality, the configuration of knowledge intermediaries, and technology-market liquidity—and thereby influences STAT.
This study focuses on the actual occurrence of enterprise STAT, specifically the outward transfer of such achievements. Given established theoretical conventions, data availability, and the specific context of technology-transfer pathways in China, the use of external patent transfer counts represents a reasonable and methodologically robust proxy for examining enterprises' STAT behavior at this stage. Future research could further distinguish between genuine commercialization and strategic STAT by incorporating more granular commercialization outcome data and implementing micro-level, patent‑specific post‑transfer tracking. This would involve linking China's National Intellectual Property Administration Patent legal status database with patent transfer records to trace each transferred invention patent's trajectory—including whether it is subsequently re-transferred, invalidated, or declared standard‑essential. Indicators such as post‑transfer maintenance duration and post‑transfer citation counts could then serve as quality proxies. A stratified Cox proportional‑hazards model could subsequently be applied to examine the heterogeneous effects of DT on high-quality versus low‑quality commercialization outcomes.
FundingThis work was supported by the Humanities and Social Sciences Youth Scientific Research Project of the Anhui Provincial Department of Education (grant number 2025AHGXSK40155); the General Project of Humanities and Social Sciences Research at Fuyang Normal University (grant number 2025FSSK20); the research outcome of the Social Science Planning Project of Fuyang City (grant number FSK2025062); and the Youth Project of Hunan Provincial Department of Education (grant number 25B0461).
CRediT authorship contribution statementQian Liu: Writing – review & editing, Writing – original draft, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. En Xie: Writing – review & editing, Visualization, Supervision, Resources, Project administration, Methodology. Xin Gao: Visualization, Validation, Supervision, Software, Resources, Project administration.
Lewbel (1997) demonstrates that constructing instrumental variables based on the heteroscedasticity of endogenous variables can mitigate potential endogeneity concerns, thereby circumventing the conventional requirement for exclusion restrictions in instrumental variable estimation.
It should be noted that due to the lack of substantive examination for utility model patents, this study excludes not only invention patents that have been published but not granted but also the part of utility model patents that were marked as "invalid - fully invalidated by judgment" due to appeals after being granted.
articles
