Digital Transformation Drives Regional Innovation Ecosystem Resilience: A Study Based on the Dynamic QCA Method

Abstract

In an era marked by volatility, uncertainty, complexity, and ambiguity, constructing resilient regional innovation ecosystems is identified as a critical strategic imperative for achieving high-quality development and advancing sustainable development goals. Drawing on the Technology–Organization–Environment (TOE) integrative framework, this study examines six antecedent conditions of ecosystem resilience from the perspective of digital transformation: digital infrastructure, digital innovation capacity, digital human capital, digital government governance, digital attention, and digital finance. A sample of 48 prefecture-level cities from the Beijing–Tianjin–Hebei, Yangtze River Delta, and Pearl River Delta urban agglomerations in China between 2018 and 2022 is selected. Through the application of dynamic Qualitative Comparative Analysis (QCA), the study explores the multiple configurations across temporal and spatial dimensions through which technological, organizational, and environmental factors contribute to enhancing regional innovation ecosystem resilience. The results indicate that ecosystem resilience is jointly driven by multiple interacting factors, and no single condition is found to be necessary. Four distinct causal pathways are identified as sufficient to enhance resilience: (1) a triadic synergy of technology, organization, and environment; (2) a technology-driven, talent-supported configuration; (3) a technology-driven, government-supported configuration; and (4) a dual technology–environment-driven model. While none of the configurations exhibit consistent temporal effects, some are influenced by unobserved factors in specific years. Moreover, cities do not converge on a single dominant configuration when achieving high levels of ecosystem resilience.

1. Introduction

Regional innovation ecosystems are regarded as foundational units of the national innovation system, responsible for integrating innovation resources, facilitating knowledge flows, and enabling technological diffusion [1,2]. By fostering collaboration and resource sharing among diverse actors, these ecosystems support the transformation of scientific outputs into tangible value, strengthening endogenous economic growth and regional competitiveness [3]. Amid ongoing restructuring of the global geopolitical and economic order, uncertainty and systemic volatility in the external environment are intensifying. Weak global recovery, escalating risks of technological decoupling, and recurring geopolitical conflicts jointly impose unprecedented external shocks and developmental pressure on regional innovation ecosystems [4,5]. Resilience enhancement is therefore considered critical for enabling these ecosystems to adapt to internal and external disruptions and for maintaining the flexibility required to sustain high-quality economic development. This raises a pressing theoretical question: how resilient are China’s regional innovation ecosystems, and how should they be structured and evolve to more effectively withstand future risks?

In recent years, the ongoing advancement of digital transformation across society has enabled frontier technologies—such as the Internet, big data, and artificial intelligence—to penetrate and reshape economic and social systems with unprecedented breadth and depth [6,7]. In this context, digital elements such as data, computing power, and platforms are increasingly regarded as strategically significant resource endowments [8]. Driven by this digital wave, regional innovation ecosystems—characterized by openness, dynamism, and multi-actor coupling—are undergoing profound structural changes. Core elements including knowledge, talent, capital, and data are being rapidly redistributed and recombined within the system, enhancing collaborative innovation and value co-creation among diverse actors [9]. These ecosystems are evolving toward greater resilience, adaptability, and intelligence. Simultaneously, geographic and organizational boundaries are becoming increasingly blurred, facilitating the emergence of deeply nested, multi-level innovation networks that integrate global and local dynamics. These transformations contribute to the development of more tightly connected innovation communities [10]. Empirical evidence shows that digital transformation, by integrating value creation activities across technological, economic, and environmental dimensions, effectively promotes the coordinated and sustainable evolution of regional innovation ecosystems [11]. Within this digital context, factors such as digital infrastructure development, technological innovation, talent cultivation, and digital finance have emerged as key variables influencing ecosystem performance and have received widespread scholarly attention [12,13,14,15]. Given the above, a systematic investigation into how digital transformation affects regional innovation ecosystems—and how it enhances their resilience to external uncertainties and shocks—has become a research agenda of both theoretical and practical importance. In particular, identifying the key influence pathways and multi-factor interaction mechanisms is strategically significant for strengthening ecosystem capacity to sustain innovation vitality and core functions under disruptive conditions. This issue is not only central to achieving high-quality regional development, but also directly affects China’s competitive position in the global innovation landscape [16].

At present, research on the resilience of regional innovation ecosystems in China remains at an early stage [17], primarily focusing on two aspects: the assessment of system resilience and the identification of its influencing factors. In terms of resilience assessment, several representative theoretical frameworks have been proposed. For example, Roundy et al. (2017) [18] argue that diversity and coherence constitute the foundational conditions of innovation ecosystem resilience, and the tension between the two serves as a source of system heterogeneity. Lv et al. (2018) [19] emphasize the joint role of stability and adaptability, as well as their interaction, in shaping innovation resilience, which enables organizations to maintain innovation efficiency while improving adaptability to environmental changes. From the perspective of comparative ecology and resilience theory, Liang et al. (2020) [20] develop a monitoring framework for regional innovation ecosystem resilience comprising four dimensions: diversity, buffering capacity, fluidity, and evolvability. Regarding influencing factors, knowledge and technology are widely recognized as critical drivers of resilience in regional innovation ecosystems [21]. Moreover, cultural and policy factors exert significant influence on the development and sustainability of such resilience [18]. Yang et al. [22], focusing on regional digital innovation ecosystems, apply an indicator-based approach to analyze how configurations of governance niches—such as knowledge development, knowledge diffusion, market formation, resource allocation, and the legal environment—affect system resilience. Although these studies provide a valuable foundation, several limitations remain. First, most assessments of regional innovation ecosystem resilience rely on qualitative analysis, making it difficult to capture multidimensional differences in resilience and their linkages to system dynamics. Second, the existing literature tends to focus on the “net effects” of individual antecedent variables and often examines only sufficiency within causal relationships, thus failing to uncover the more complex causal structures underlying resilience generation. Finally, there is a lack of holistic analytical frameworks that explore how multiple factors—such as technology, organization, and environment—interact and co-evolve. As a result, potential interactions and synergies among different antecedents are often overlooked.

In light of the aforementioned practical context and theoretical gaps, this study adopts the widely used Technology–Organization–Environment (TOE) framework—originally developed for technology adoption research—as the theoretical foundation. A sample of 48 cities within China’s three major urban agglomerations (Beijing–Tianjin–Hebei, Yangtze River Delta, and Guangdong–Hong Kong–Macao Greater Bay Area) is selected for the period 2018–2022. Employing a dynamic panel Qualitative Comparative Analysis (dynamic panel QCA), the study investigates the complex causal relationships between TOE-based configurations of antecedent conditions and the resilience of regional innovation ecosystems. By identifying typical configurational pathways that contribute to enhanced resilience, this study aims to address existing theoretical shortcomings and enrich the empirical foundation for resilience research under the digital transformation context. Specifically, the study seeks to answer the following research questions: (1) How do the complex and nonlinear interactions among technological, organizational, and environmental factors influence the resilience performance of regional innovation ecosystems across different contexts? (2) Which antecedent conditions serve as necessary or sufficient factors for high or non-high levels of regional innovation ecosystem resilience? (3) How do these configurational patterns of antecedent conditions dynamically evolve over time and across regions?

The potential contributions of this study are primarily reflected in the following three aspects: first, in contrast to previous research that explores regional innovation ecosystem resilience from a single-dimensional perspective, this study adopts the TOE framework and constructs an analytical framework from a configurational perspective. This approach provides a new explanatory paradigm for related theoretical research on how digitalization influences regional innovation ecosystem resilience. Second, by employing dynamic panel QCA, this study incorporates the temporal dimension into the identification of resilience-driving mechanisms, overcoming the limitations of traditional studies based on static cross-sectional data in revealing nonlinear causal relationships. This enhances the temporal explanatory power and cross-regional applicability of the research findings. Third, the results of this study offer theoretical support and practical guidance for different regions to formulate differentiated policies and optimize innovation strategies based on their specific development characteristics, thus holding significant practical value and policy relevance.

2. Literature Review and Research Framework

2.1. Literature Review

2.1.1. The Conceptual on RIER

The concept of resilience originated in physics, where it is used to describe an object’s ability to return to its original state after deformation under external forces. The understanding of resilience has evolved from “engineering resilience” to “ecological resilience” and then to “evolutionary resilience” [23]. In the social sciences, the study of resilience has gradually shifted from focusing on the construction of hard infrastructure to addressing governance in social ecosystems, covering areas such as urban planning [24], regional economics [25], and organizational management [26]. As research on resilience has deepened, scholars have begun integrating resilience theory with innovation ecosystems, thus advancing the study of innovation ecosystem resilience [27]. Since the concept of the “innovation ecosystem” was first introduced in the United States in 2003, it has received widespread attention in academic research. The definition of innovation ecosystems has primarily evolved from two perspectives: the network perspective and the ecological perspective. From the network perspective, Adner (2006) [28] conceptualizes the innovation ecosystem as a loosely coupled network centered around a focal firm or platform. Within this system, multiple actors engage in nonlinear collaborative interactions, which, compared to traditional bilateral partnerships, enable more rapid and efficient value co-creation. From an ecological perspective, Iansiti et al. (2004) [29] extend the concept of “ecosystem” from ecology into innovation studies. They define the innovation ecosystem as a complex system composed of innovation organizations occupying distinct ecological niches. This view emphasizes the symbiotic relationships and co-evolutionary dynamics among actors, which collectively drive system-wide value creation. Overall, regional innovation ecosystems are regarded as an evolved form of regional innovation systems. They encompass diverse innovation actors and environmental elements within a specific spatial domain. Through the continuous flow and interaction of resources—such as capital, knowledge, and information—these systems develop interdependent and co-evolving structures. They highlight dynamic resource coupling between actors and their environments, enabling organizational restructuring and behavioral adaptation that foster self-optimization and sustained innovation output [30,31,32]. In terms of the connotation of regional innovation ecosystem resilience, Liang et al. (2020) [20] and Liang & Li (2023) [27], from a state perspective, constructed a resilience analysis framework for regional innovation ecosystems through a comparative study of the evolutionary processes under ecological and resilience theory. They proposed that resilience includes four dimensions: diversity, fluidity, buffering capacity, and evolvability. Tang et al. (2023) [33], taking a process perspective, argue that regional innovation ecosystem resilience is reflected in a system’s ability to resist, absorb, recover from shocks, and transform to create new development paths [23]. Considering the relevant definitions of “resilience” in ecology and economic management, as well as the openness, symbiosis, and dynamics of regional innovation ecosystems [34], this study defines regional innovation ecosystem resilience (RIER) as the system’s capacity to stabilize and upgrade its functions by restructuring and optimizing internal resource allocation through the cyclical flow of innovative resources—such as materials, energy, and information—during external shocks. This process enables self-adjustment and transformation of the ecosystem.

2.1.2. Research on Measurement Methods of RIER

In the academic literature, four main approaches are commonly used to quantify and evaluate resilience. The first is the core variable method, which incorporates a key variable—such as innovation performance or GDP—into an econometric model to capture relative changes before and after a shock. This method provides a direct assessment of resilience by observing the deviation of the variable. For instance, Martin et al. (2016) [35] propose a sensitivity index approach that evaluates economic resilience across UK cities by measuring fluctuations in employment as the core variable. Similarly, Wang et al. (2023) [36] employ an information entropy index to quantify the resilience of regional innovation ecosystems, reflecting the system’s capacity to maintain information flow under external shocks. The second is the composite indicator approach, which constructs an index system composed of variables that reflect different dimensions of resilience, typically selected based on theoretical frameworks. Briguglio (2014) [37], for example, develops a composite index to measure economic resilience in over 80 national and regional economies, focusing on four dimensions: macroeconomic environment, market efficiency, governance, and social stability. From the perspective of evolutionary resilience, Martin et al. (2015) [38] conceptualize resilience through four key dimensions—resistance, recovery, reorientation, and renewal. This multidimensional framework has gained broad recognition and serves as the foundation for many subsequent empirical studies and index refinements. Liang and Li (2023) [27] developed a resilience measurement system for regional innovation ecosystems based on diversity, evolvability, fluidity, and buffering capacity, and constructed a resilience index using the coupling coordination model. The third method is the input–output method, which quantifies resilience from an input–output perspective using relevant models. For instance, Ying et al. [39] evaluated the resilience of coastal county cities along the East Sea using an efficiency model. The fourth method is the network simulation method, which conceptualizes network disruptions and resilience based on the structural relationships of entities in a resilience network. Shen et al. [40] constructed a global trade model for critical mineral resources using multi-layer complex network theory, conducted risk simulations, and evaluated the resilience of network nodes. Despite the diversity of these research methods, studies on the measurement of innovation ecosystem resilience remain limited, and the methods are still under development. This study, building on the research of Liang et al. (2020) [20] and Liang & Li (2023) [27], constructs a RIER measurement system based on four dimensions: diversity, evolvability, fluidity, and buffering capacity. Specifically, diversity refers to the richness and variety of actors and innovation activities within the ecosystem, enhancing knowledge interaction and ensuring sufficient response capacity to external shocks. Evolvability involves optimizing and reorganizing resource allocation to improve the innovation input-output ratio, thus enhancing overall ecosystem performance. Fluidity refers to the rapid movement and interaction of innovation factors, promoting tighter system links and filling gaps post-shock. Buffering capacity covers the accumulation and complexity of resources across various levels, providing the system with the “ammunition” needed to withstand risks and extend its “window of opportunity” during shocks.

2.1.3. The Conceptual on Digital Transformation

The concept of digital transformation (DT) initially emerges from practices within the private sector. For example, Tobias and colleagues define DT as a systemic change driven by digital technologies, emphasizing its profound impact on both internal management and external value creation within enterprises [41].

With the widespread adoption of digital technologies across industries and sectors, the meaning of DT has continuously expanded beyond the enterprise level to encompass societal behavior and governance systems. As a result, it has gained increasing attention and application at national, regional, and market levels [42]. In China’s 14th Five-Year Plan, the government explicitly proposes that “DT should drive the transformation of production, lifestyles, and governance systems in an integrated manner.” This policy articulation marks a shift in DT from a firm-level managerial tool to a core component of national development strategy, positioning it as a critical pathway for enabling systemic social change. In response to practical demands, scholarly research has increasingly explored DT across multiple levels and actors. Two dominant definitional approaches have emerged in the literature: the technological perspective and the strategic perspective. From a technological perspective, DT is understood as the integration and application of digital technologies within organizational operations to improve efficiency and generate economic value [43,44]. This view emphasizes the instrumental and supportive role of digital technologies as the foundational enablers of transformation. In contrast, the strategic perspective conceptualizes DT as a means of reshaping customer value propositions, optimizing business processes and organizational structures, and creating entirely new business models and sources of competitive advantage through digital technologies [45,46,47]. This perspective highlights the strategic role of DT in enabling firms to achieve sustainable growth and maintain long-term market competitiveness. Building on this foundation, Saeedikiya et al. (2025) [48] define DT as “a continuous process of structural change driven by digital technologies, aimed at creating new value and achieving sustained competitive advantage”. Specifically, DT is understood as a long-term, technology-driven evolutionary process characterized by organizational restructuring. Its core lies in the creative integration of digital technologies to simultaneously enhance value creation and sustain competitiveness. Hinings et al. (2018) [49] further distinguish between digital innovation and DT. While digital innovation refers to the development and implementation of new digital products and services, DT is conceptualized as a broader process involving the interplay of multiple digital innovations. This process gives rise to new actors and network structures while fundamentally reshaping institutional arrangements, organizational practices, value systems, and cognitive frames within industries and organizations. These shifts may disrupt, complement, or replace existing “rules of the game”. Within this framework, the authors identify three institutional structures central to DT: digital organizational forms, digital institutional infrastructures, and digital institutional components. Accordingly, DT is not merely the diffusion or adoption of digital technologies; rather, it constitutes a strategic bridge across micro (organizational), meso (industry/regional), and macro (national/institutional) levels. It involves the reconfiguration of innovation resources, the transformation of value creation mechanisms, and the co-evolution of institutional environments [50,51]. From this perspective, digitalization is not only instrumental but also a core strategic variable for enhancing the resilience and adaptability of innovation systems. Based on a synthesis of the existing literature and the integration of institutional and strategic perspectives, this study defines regional DT as a continuous, strategic transformation process at the regional level. It is driven by the systemic embedding of digital technologies and the reconstruction of innovation mechanisms, enabled through multi-level, multi-actor collaboration. This process facilitates the efficient flow of resources and the co-evolution of institutional systems. It encompasses both the physical embedding of technologies and infrastructure, and the cognitive shifts in institutions, organizations, and behaviors. Ultimately, it reflects a complex evolutionary process involving the coupling of digital technologies, organizational change, and institutional innovation. In empirical research, the degree of regional digital transformation is typically assessed through multiple dimensions. Core indicators include the level of e-government development, digital human capital reserves, internet penetration rates, and the growth of digital finance [52,53,54].

Moreover, a review of the literature reveals ongoing conceptual ambiguity and interchangeable use of the terms digitalization and digital economy within academic discourse. To ensure conceptual clarity and analytical precision, this study systematically distinguishes between the two. Several studies argue that, compared to the industrial economy, the digital economy represents a higher stage of economic development, driven fundamentally by digital technologies. It significantly enhances productivity and embodies a more advanced form of productive forces in the current era [55]. Chen et al. (2022) [56] further define the digital economy as a mode of economic activity centered on digital information as a key resource, relying on internet platforms as primary carriers, and guided by digital technological innovation. It involves a series of emerging business models and economic forms. These definitions suggest that the digital economy is essentially a digitally embedded process within the economic system, and therefore conceptually distinct from the broader notion of digitalization. While both are rooted in the development of digital technologies, they differ significantly in nature and scope. Digitalization refers to the pervasive integration of digital technologies and information networks across various systems. It emphasizes deep technological embedding across urban functions and extends beyond the economic sphere to influence social, cultural, and political domains. In contrast, the digital economy focuses more narrowly on the transformation of economic activities through digital technology. Accordingly, digitalization has broader coverage and deeper transformative potential. This study adopts the more systemic and integrative perspective of digitalization as its conceptual foundation to examine how DT influences the resilience of regional innovation ecosystems.

2.1.4. Research on the Impact of DT on RIER

A growing body of regional-level research demonstrates that DT significantly enhances regional innovation performance. Varian (2010) [57] highlights that advancements in information technology effectively facilitate innovation spillover effects. Luo et al. (2023) [58], by constructing an interprovincial Internet Development Index, find that the internet improves regional innovation efficiency through multiple channels, including the acceleration of human capital accumulation, financial development, and industrial integration across provinces. Yang et al. (2024) [59] reveal that government-led digital transformation exerts a positive impact on regional green innovation. In this relationship, green finance development and the agglomeration of green talent serve as mediating factors, while policy attention to green development and technology research plays a significant moderating role. Based on provincial panel data from 2013 to 2020, Tian et al. (2023) [60] employ a dynamic spatial Durbin model to quantify the effects of digitalization on regional technological innovation. Their findings indicate that digitalization generates strong spatial spillover effects, enhancing innovation not only within a province but also in neighboring regions. In addition, Wen et al. (2020) [61] argue that the digital economy promotes regional innovation capacity through micro-level firm evolution effects, meso-level industrial diffusion effects, and macro-level scale effects, which are empirically validated using interprovincial panel data. Zhou et al. (2020) [62], using panel data from 73 counties in Zhejiang Province, empirically confirm the positive role of regional digital infrastructure in boosting innovation performance.

In summary, existing studies provide robust evidence for the positive influence of DT on regional innovation capacity and performance from various analytical dimensions. However, several important limitations remain. First, DT has not yet been examined at the systemic level, and research on its mechanisms within regional innovation ecosystems remains scarce. Second, a comprehensive theoretical framework connecting DT and the resilience of innovation ecosystems has yet to be established; current studies remain fragmented. Third, most empirical research focuses on the effect of a single variable on regional innovation, neglecting the interactive mechanisms and synergistic effects among multiple factors. As a result, the complex causal structures underpinning DT and ecosystem resilience remain insufficiently explored.

2.2. Theoretical Framework

The TOE framework was initially proposed by Tornatzky [63] to describe the adoption and application process of technological innovation at the organizational level and has gradually been widely applied to explain organizational behaviors related to technology integration and adoption. The TOE framework consists of three dimensions: technology, organization, and environment. The technology dimension focuses on the characteristics of the technology itself and its relationship with the organization; the organization dimension emphasizes organizational features, such as resources and structure; and the environment dimension addresses external factors, including the economic, social, and institutional environment [64]. As academic research has deepened, scholars have expanded and refined the TOE framework based on different research contexts, broadening its application from a micro-level organizational perspective to include macro and meso-level contexts, such as innovation performance, public health governance, and industrial digitalization [65,66].

DT emphasizes the deep embedding of digital technologies across various urban operational domains, exerting transformative impacts on social, cultural, political, and economic systems. In the context of localized DT practices in China, significant regional disparities in performance outcomes highlight the necessity of employing multiple driving mechanisms to strengthen and refine regional innovation ecosystems. From a multi-factorial and conjunctural perspective, the resilience of regional innovation ecosystems is influenced by multiple internal dimensions of DT, each with its own developmental logic. These logics interact and intertwine, collectively shaping the level of resilience. Against this backdrop, analyzing the configurations of technological, organizational, and environmental conditions underlying DT becomes particularly relevant and practically meaningful. Some scholars have already applied the Technology–Organization–Environment (TOE) framework to examine regional innovation and development in China, yielding valuable insights [67,68]. However, the empirical application of the TOE framework within the Chinese context remains underdeveloped and lacks contextual adaptation [64]. Given China’s unique institutional arrangements, how the TOE framework—originally developed from Western practice—performs in explaining Chinese cases requires further investigation. Specifically, how the integration of technological, organizational, and environmental dimensions manifests distinct Chinese characteristics is a key question to be explored. Moreover, the internal meanings and synergies among these dimensions need to be deeply examined across different contexts, taking into account the heterogeneity of research subjects and their institutional environments. This inquiry constitutes a central focus of the present study. In sum, the TOE framework proves analytically suitable for examining the resilience of regional innovation ecosystems in China. To investigate the pathways through which resilience is enhanced under DT, it is essential to unpack the relationships between the three TOE dimensions and the core elements of DT, while also critically assessing their contextual applicability in the Chinese setting.

2.2.1. Technological and RIER

In terms of technological conditions, this study focuses on two secondary elements: digital infrastructure and digital innovation technologies. Among these, digital infrastructure refers to the foundational digital technologies, platforms, and related services and facilities that support the operations of enterprises or industries. First, digital infrastructure transcends spatial limitations in information dissemination and accelerates information flows across regions [13]. It effectively mitigates information asymmetry caused by geographic distance and cultural differences, thereby enhancing the capacity of innovation actors to access innovation resources. Moreover, it facilitates the cross-regional flow of talent and capital [69], contributing to the diversity and buffering capacity of regional innovation ecosystems. Second, as a novel platform for data interaction, transmission, and sharing, digital infrastructure plays a critical role within the ecosystem. Through integration and synergy with other production factors, it generates penetration effects, substitution effects, and synergistic effects, thereby improving the quality of factor supply and optimizing allocation structures. This process enhances the efficiency of internal coordination between system resources and capabilities [70], strengthens the system’s capacity for self-renewal and transformation, and provides essential support for the development of RIER.

Regional digital technological innovation refers to the capacity to transform digital technology inputs into new products, processes, and services [71]. According to endogenous growth theory, technological progress serves as the core driver of sustained economic growth [72]. The widespread application of digital technologies significantly accelerates the flow of key innovation factors—such as technology, human capital, and investment—across regions. It deepens communication and learning among innovation actors, reduces transaction and coordination costs within the innovation process, and facilitates knowledge spillovers and information sharing. These effects not only stimulate innovation demand but also accelerate the emergence of new knowledge and technologies, thereby fostering collaborative technological development and enhancing regional innovation capacity overall [73]. A higher level of regional digital innovation capacity is closely associated with stronger adaptability and transformability within the innovation ecosystem. In highly digitalized systems, external shocks can be addressed proactively through the development of new products, exploration of new markets, or reallocation of resources. Moreover, such systems can undergo post-shock transformations by restructuring existing production factors or generating novel forms of economic activity, thereby breaking established innovation path dependencies [74]. This enables the continuous evolution of system resilience.

2.2.2. Organizational and RIER

In terms of organizational conditions, this study focuses on two secondary elements: digital human capital and digital government governance. Digital human capital refers to the collective pool of professional talent equipped with capabilities in the development and application of information technologies. In the era of the digital economy, the deep integration of traditional industries with digital technologies relies heavily on high-quality human resources capable of adapting to new business models and digital infrastructures [75]. As a core innovation resource within regional innovation ecosystems, digital human capital not only serves as a key carrier of knowledge and technology but also plays a critical role in processes such as knowledge absorption and transformation, as well as technology transfer and diffusion [76]. On the one hand, highly skilled digital professionals directly participate in innovation activities, providing intellectual support for improving regional productivity and advancing industrial upgrading, thereby enhancing the vitality and momentum of the regional innovation ecosystem. On the other hand, the regional agglomeration of digital talent helps increase system diversity, fosters collaboration and interaction among innovation actors, and creates favorable conditions for the emergence and evolution of original innovations.

Digital government governance refers to the systemic transformation of organizational models driven by digital resources and digital capabilities [77]. First, digital government leverages the mining, analysis, and evaluation of massive datasets to enable more precise allocation of innovation resources, thereby enhancing the efficiency and specificity of resource utilization and facilitating the coordination and integration of heterogeneous information [78]. Moreover, the deep embedding of data into the production, distribution, and circulation processes of innovation activities fosters interdepartmental collaboration and collaborative governance, providing the regional innovation ecosystem with greater flexibility and adaptability in the face of external shocks [79]. Second, digital government governance promotes innovation in public service delivery and facilitates transparency and information sharing, thereby creating a more open, efficient, and transparent innovation environment. This not only enhances the diversity and dynamism of innovation actors and encourages the emergence of new business models and organizational forms but also provides strong support for the system’s innovation capacity and its ability to transform [80].

2.2.3. Environmental and RIER

In terms of environmental conditions, this study focuses on two secondary elements: digital attention and digital finance. Digital attention refers to the degree of governmental focus and strategic prioritization of digital development through governance resources [81]. Government-formulated and -implemented digital policies play a critical role in guiding regional innovation. These policies not only clarify the developmental direction of regional innovation ecosystems but also foster collaboration and interaction among innovation actors. First, the development of digital innovation relies heavily on policy support. For instance, by accelerating the construction of essential infrastructure—such as high-speed urban broadband networks and big data centers—governments provide strong technological and data foundations for regional innovation. In addition, policy instruments such as financial subsidies and innovation incubation platforms offer necessary funding and institutional support to innovation actors, thereby promoting the orderly evolution of innovation ecosystems [82]. Second, heightened governmental attention to digitalization contributes to the establishment and improvement of digital governance systems, enhancing operational efficiency in areas such as regulation, oversight, and information disclosure [80]. This not only strengthens the scientific basis and fairness of government decision-making but also improves public service delivery and administrative efficiency, providing a more favorable macro-environment for the sound functioning of regional innovation ecosystems.

Digital finance plays a critical role in strengthening regional innovation ecosystems by enhancing resource allocation efficiency, lowering innovation thresholds, and improving the collaborative capacity of innovation actors. First, the development of digital finance overcomes the spatial limitations of traditional financial systems, providing essential support for the efficient supply of R&D capital. At the same time, it promotes the mobility of R&D talent and the spillover of technological resources, optimizes regional factor input structures, and facilitates interregional technological exchange and progress, thereby improving the overall performance of the innovation ecosystem [83]. Second, with the rapid advancement of digital technologies such as big data and the Internet, emerging financial formats—including internet finance and digital economy platforms—continue to evolve. These platforms foster openness and sharing of financial resources among innovation actors, enhance the efficiency of capital matching, and effectively diversify the risks associated with the innovation process. This, in turn, supports industrial upgrading and promotes the high-quality development of regional innovation ecosystems [84]. Finally, the deepening of digital finance contributes to optimizing the regional innovation market environment, enabling governments to more accurately adjust the intensity and direction of innovation-support policies. This fosters a positive interaction mechanism between market incentives and administrative regulation [85]. Such a bidirectional enabling mechanism generates synergistic effects between institutional and technological innovation, strengthens the coupling efficiency among innovation factors, and helps build a dynamically balanced innovation governance system. As a result, it provides a stable yet forward-looking institutional environment for the sustainable development of regional innovation ecosystems.

In summary, this study proposes six core antecedent conditions that influence the resilience of regional innovation ecosystems under the context of DT. These conditions are systematically identified based on the TOE framework and span across three dimensions: technology, organization, and environment. Adopting a configurational perspective, a dynamic and synergistic TOE-based model is constructed to examine how multiple factors—and their interactions—jointly contribute to the formation and evolution of RIER in complex and changing environments. The analytical framework is illustrated in Figure 1.

3. Research Design

3.1. Research Methods

In recent years, configurational theory and Qualitative Comparative Analysis (QCA) have been increasingly applied in the fields of management and economics—particularly in the study of innovation ecosystems—to address complex, system-level problems from a holistic perspective [86]. However, traditional static QCA methods rely primarily on cross-sectional data to examine causal relationships at a single point in time. This approach often suffers from temporal selection bias and overlooks the configurational effects that unfold over time, limiting its ability to capture the dynamic trajectories of multiple interacting factors [87]. Given that the resilience of regional innovation ecosystems is inherently a dynamic process, analyzing causal configurations based solely on data from a single year fails to reveal time-dependent causal mechanisms. To overcome this limitation, this study draws on the theoretical and methodological advances proposed by Garcia-Castro and Ariño [88] and employs a dynamic QCA approach based on panel data from Chinese cities. The study focuses on identifying how technological, organizational, and environmental (TOE) conditions interact over time under the context of DT. R language is used to model configurational changes and explore the influence of temporal effects. By aggregating panel data across multiple time periods, the study constructs representative temporal configurations, enabling an integrated analysis that accounts for both causal complexity and temporal evolution. This approach not only preserves the strengths of QCA—such as its ability to identify causal asymmetry and equifinality—but also leverages the richness of panel data for capturing time-sensitive patterns and dynamic variation. Specifically, the analysis is conducted from three perspectives: between-group comparison, within-group variation, and overall aggregation. It systematically explores the formation and evolution of configurational outcomes across different cases and within individual cases over time. In addition, the study introduces the metric of consistency-adjusted distance to quantify the degree of configurational change along both temporal and cross-sectional dimensions. This allows for a more comprehensive understanding of how resilience configurations evolve in response to DT.

3.2. Sample Selection and Data Sources

This study selects 13 prefecture-level cities from the Beijing–Tianjin–Hebei urban agglomeration, 26 cities from the Yangtze River Delta, and 9 cities from the Pearl River Delta as the research sample, using data from the period 2018 to 2022. The rationale for this selection is as follows: first, the period from 2018 to 2022 marks a critical phase in the rapid development of China’s digital economy. Analyzing this timeframe helps to reveal the underlying mechanisms linking regional digitalization processes with innovation ecosystem resilience. Second, the selected cities exhibit significant internal heterogeneity in terms of policy environment, economic development levels, and resource endowments. This diversity facilitates the identification and synthesis of multiple pathways and mechanisms through which RIER may be enhanced. Third, these cities have accumulated a substantial volume of policy documents, data resources, and textual materials related to DT and regional innovation, providing a solid empirical and textual foundation for in-depth case analysis. The relevant data are primarily obtained from authoritative sources, including the “China City Database” on the EPS (Economy Prediction System) platform, the China City Statistical Yearbook, individual municipal statistical yearbooks and bulletins, as well as government work reports released by the National Bureau of Statistics and official websites of local governments.

3.3. Variable Measurement

3.3.1. Result Variable

The outcome variable in this study is the resilience of regional innovation ecosystems. Although resilience is inherently a dynamic and continuous process, its level at a specific point in time reflects the region’s overall capacity to resist and recover from external shocks, adapt and adjust to changing conditions, and undergo transformation and upgrading. Following the approach of Liang et al. (2020) [20], a composite evaluation index system is constructed based on four key dimensions: diversity, transformability, mobility, and buffering capacity (see

Table 1. Measurement of result variables.

Tier l Indicators	Tier 2 Indicators	Tier 3 Indicators	Indicator Properties
Diversity	Diversity of talents	The proportion of employees with tertiary education or above	+
	Diversity of enterprises	Number of above-designated-size industrial enterprises	+
	Diversity of colleges and universities	Number of regular higher education institutions	+
Evolvability	Innovation input	R&D personnel full-time equivalent	+
		New product development expenses	+
	Innovation output	Number of domestic patent applications	+
		Sales revenue of new products	+
Fluidity	Capital flow	Fixed-asset investment	+
		Utilization of foreign direct investment	+
	Technical flow	Value of contract exportation from domestic technical markets	+
		Value of contract inflows to domestic technical markets	+
	Information flow	Number of broadband subscribers’ port of internet	+
		Business volume of telecommunication services	+
	Freight flow	Freight traffic	+
Cushioning	Economic resources	Per capita GDP	+
	Knowledge resources	Invention patent ownership per 10,000 people+	+
	Natural environment resources	Per capita water resources	+
		Industrial sulfur dioxide emissions per 10,000 people	-
	Social environment resources	Number of books in public libraries per 10,000 people	+
		Number of beds in medical and health institutions per 10,000 people	+

). In addition, drawing on the methodology used by Chen et al. [16], this study adopts the entropy-weighted TOPSIS method to measure the resilience level of regional innovation ecosystems. The secondary indicator data for RIER are obtained from the China City Database of the EPS (Economy Prediction System) platform, the China Urban Statistical Yearbook (2018–2022), as well as statistical yearbooks and bulletins published by individual prefecture-level cities.

3.3.2. Condition Variable

Technological factors comprise two dimensions: digital infrastructure and digital technological innovation. To evaluate digital infrastructure, this study follows the methodology proposed by Huang et al. [89], selecting three indicators: the number of internet users per 100 people, the number of mobile phone users per 100 people, and the per capita volume of telecommunication services. These indicators are used to capture the development level of digital infrastructure across prefecture-level cities. An entropy-weighted method is then applied to compute a composite index, which quantifies the digital infrastructure level of each region. The data are obtained from the China Urban Statistical Yearbook (2018–2022). For the measurement of digital technological innovation, this study adopts the approach of Deng et al. [90] and focuses on innovation outputs. Specifically, the number of authorized patents related to digital technologies is used to assess the digital innovation capacity of local regions. This indicator provides a direct reflection of the intensity of digital innovation activities and the effectiveness of technology commercialization. The data are sourced from the China National Research Data Services Platform (CNRDS).

Organizational factors include two aspects: digital human capital and digital government governance. Regarding digital human capital, this study follows the method proposed by Zeng et al. [91] and uses the ratio of employees in the software and information technology services industry to the total number of employees in a region as an indicator. This reflects the reserve of professional and technical talent during the process of regional DT. The data are sourced from the China Statistical Yearbook and the China Urban Statistical Yearbook (2018–2022). In terms of digital government governance, this study draws on the research of Zhao et al. [92] and adopts the digital service capability scores of local governments from the “China Local Government Internet Service Capability Development Report” as the indicator to assess the capability and level of local governments in digital governance.

Environmental factors consist of two dimensions: government digital attention and digital finance. Following the approach of Zeng et al. (2024) [91] and Bi et al. (2022) [93], this study constructs a digitalization-related vocabulary and measures textual digital attention using the ratio of digital-related keyword frequency to the total word count in the text. The keyword dictionary covers two main categories: digital technologies and digital applications. Specific terms include technologies such as “big data” (e.g., virtual reality, heterogeneous data, digital twins), “cloud computing” (e.g., quantum computing, cloud platforms), and “blockchain” (e.g., digital currency, smart contracts), as well as digital practices in key sectors such as industry (e.g., intelligent manufacturing, industrial digitalization, smart homes), agriculture (e.g., smart agriculture, unmanned farming), and services (e.g., mobile payments, internet healthcare). The textual data of government work reports is sourced from the official websites of prefecture-level municipal governments. For digital finance, this study utilizes data from the Digital Inclusive Finance Index, jointly developed by the Digital Finance Research Center at Peking University and Ant Financial Group [94]. The index evaluates the level of digital financial development across regions from three dimensions: coverage breadth, usage depth, and digitalization level, providing a comprehensive assessment of how well digital finance supports regional economies and innovation activities.

3.4. Calibration

The application of dynamic QCA requires all variables to be calibrated by assigning set memberships to each case. Since there are no universally accepted external or theoretical standards for determining the levels of DT or RIER, these variables are typically assessed based on their relative performance within the sample. Moreover, resilience itself is inherently a relative concept. Therefore, following established methodological practices in the literature [95], this study adopts the direct calibration method. Based on the distribution of each variable within the full sample, the 75th percentile, 50th percentile (median), and 25th percentile are set as the full membership, crossover point, and full non-membership thresholds, respectively. This calibration approach has been widely applied in prior QCA research [96]. All values are then transformed into fuzzy sets ranging from 0 to 1 according to the predefined anchors. To avoid cases with a membership score of exactly 0.5—which cannot be included in the analysis—this study follows Zhang & Du (2019) [97] and replaces 0.5 with 0.501 to ensure analytical validity. The calibration details and descriptive statistics for all variables are presented in

Table 2. Variable calibration.

Result Variables	Completely Affiliated	Crossing Point	Completely Unaffiliated	Standard Deviation	Mean	Maximum	Minimum
RIER	0.1450525	0.07057	0.0417475	0.151892426	0.128777583	0.71536	0.01553
Digital Infrastructure	66.7092775	48.206605	34.576895	23.1600587	52.8748215	171.59	16.44612
Digital Technological Innovation	1688.75	484	154.5	6958.072609	2871.8	49,454	7
Digital Human Capital	0.022361175	0.012060014	0.008659018	0.023724741	0.021693669	0.114770116	0.003589792
Digital Government Governance	81.94	74.635	68.155	10.18759505	73.835	91.83	41.13
Digital Attention	6711.25	5941	5161	1301.10722	5869.804167	9862	66
Digital Finance	312.3025	287.34	264.3375	33.08275946	286.5985417	361.07	214.77

.

4. Empirical Results

4.1. Necessity Analysis of Individual Conditions

Before conducting configurational analysis, it is necessary to assess whether individual antecedent conditions constitute necessary conditions for the outcome variable. This is performed using two key metrics: consistency and coverage. Generally, when the consistency score of a condition exceeds 0.9, it may be considered a necessary condition for the occurrence of the outcome. In QCA panel data analysis, particular attention must also be given to the consistency-adjusted distance. A value below 0.2 indicates a high level of precision in the summary consistency; however, a value above 0.2 may suggest the presence of temporal or regional effects [87,91]. In such cases, further investigation is required to explore how the necessity of specific conditions varies across time or space, thereby enhancing the understanding of underlying causal mechanisms.

4.1.1. Pooled Consistency Analysis

Table 3. Analysis of the necessary conditions.

Condition Variables	Y				~Y
	Aggregate Consistency	Aggregate Coverage	BECONS Adjusted Distance	WICONS Adjusted Distance	Aggregate Consistency	Aggregate Coverage	BECONS Adjusted Distance	WICONS Adjusted Distance
X1	0.787	0.793	0.098541362	0.364361599	0.338	0.344	0.197082724	0.707290163
~X1	0.349	0.343	0.159405144	0.714434508	0.797	0.791	0.057965507	0.414372015
X2	0.872	0.904	0.08115171	0.285773803	0.285	0.299	0.266641332	0.750156234
~X2	0.323	0.31	0.185489622	0.707290163	0.908	0.878	0.066660333	0.357217254
X3	0.622	0.672	0.057965507	0.585836297	0.406	0.443	0.092744811	0.728723198
~X3	0.485	0.447	0.046372406	0.607269332	0.7	0.651	0.03767758	0.492959811
X4	0.672	0.657	0.423148201	0.342928564	0.461	0.455	0.608637823	0.528681536
~X4	0.442	0.448	0.591248171	0.528681536	0.653	0.667	0.388368897	0.357217254
X5	0.644	0.645	0.199980999	0.428660705	0.456	0.46	0.191286173	0.550114571
~X5	0.461	0.456	0.266641332	0.500104156	0.648	0.648	0.147812043	0.364361599
X6	0.707	0.718	0.504299911	0.278629458	0.389	0.399	0.866584329	0.571547607
~X6	0.408	0.399	0.878177431	0.528681536	0.725	0.714	0.507198186	0.264340768

reports the results of the necessity analysis for each antecedent condition under both high and non-high levels of RIER. The results indicate that, under high resilience, the aggregated consistency scores for all six antecedent conditions are below 0.9, suggesting that no single condition qualifies as a necessary condition for achieving high resilience. However, under the non-high resilience outcome, the condition of non-high digital technological innovation exhibits an aggregated consistency score exceeding 0.9, along with an aggregated coverage score above 0.5, indicating that this condition demonstrates strong necessity. To further validate whether non-high digital technological innovation constitutes a necessary condition for non-high RIER, this study adopts the approach suggested by Schneider et al. [98] and performs a supplementary test using an X–Y scatterplot (see Figure 2). As shown in Table 3, although this condition shows a relatively high aggregated consistency score, it ultimately fails to pass the formal necessity test.

4.1.2. Between-Group Analysis

To further examine the causal configurations with inter-group consistency-adjusted distances exceeding 0.2, this study conducts a necessity analysis, as reported in

Table 4. Data between groups with adjusted distances greater than 0.2.

Situations	Causal Combination Situations		Years
			2018	2019	2020	2021	2022
1	X4/Y	Inter-group consistency	0.363	0.478	0.711	0.923	0.881
		Inter-group coverage	0.951	0.657	0.619	0.63	0.635
2	~X4/Y	Inter-group consistency	0.71	0.628	0.435	0.206	0.237
		Inter-group coverage	0.441	0.478	0.493	0.384	0.397
3	~X5/Y	Inter-group consistency	0.404	0.507	0.626	0.412	0.359
		Inter-group coverage	0.443	0.43	0.504	0.473	0.421
4	X6/Y	Inter-group consistency	0.24	0.574	0.787	0.96	0.972
		Inter-group coverage	0.998	0.967	0.855	0.619	0.605
5	~X6/~Y	Inter-group consistency	0.827	0.589	0.408	0.123	0.099
		Inter-group coverage	0.473	0.407	0.367	0.273	0.263
6	X2/~Y	Inter-group consistency	0.238	0.227	0.258	0.323	0.383
		Inter-group coverage	0.275	0.274	0.282	0.31	0.339
7	X6/~Y	Inter-group consistency	0.956	0.948	0.927	0.877	0.832
		Inter-group coverage	0.835	0.839	0.882	0.916	0.937
8	~X6/~Y	Inter-group consistency	1	0.981	0.87	0.409	0.354
		Inter-group coverage	0.566	0.706	0.808	0.911	0.925

. The analysis identifies eight configurations with adjusted distances above 0.2, suggesting the presence of significant temporal effects. Each configuration is evaluated across different years to determine whether any exhibit necessary relationships, defined by consistency scores above 0.9 and coverage above 0.5 (see Figure 2). The results indicate that Configurations 2, 3, 5, and 6 consistently fall below the 0.9 consistency threshold, disqualifying them as necessary conditions. Although Configuration 1 shows both high consistency and coverage in 2021, further examination using an X–Y scatterplot reveals that it does not meet the formal criteria for necessity. Likewise, scatterplots for Configuration 4 (2021–2022), Configuration 7 (2018–2020), and Configuration 8 (2018–2019), shown in Figure 2, indicate that the associated conditions also fail to constitute necessary relationships. Overall, despite some configurations demonstrating high consistency and coverage in specific years, none pass the necessity test upon further inspection. These findings underscore the complexity of RIER governance, suggesting that only through the synergistic interaction of technological, organizational, and environmental factors within specific configurations can substantial impacts on resilience be achieved, while no single condition alone is sufficient.

4.1.3. Within-Group Analysis

As shown in Table 3, the intra-group consistency-adjusted distances for all antecedent conditions exceed 0.2, indicating the potential presence of significant regional effects between these condition variables and the outcome variable. To further validate this, the study conducts an analysis of the intra-group consistency for the 48 cities and finds significant regional differences in the necessity of antecedent conditions across cities. Given that the current dynamic QCA methodology lacks a mature tool for evaluating the variation of necessary conditions under regional heterogeneity, this study employs one-way ANOVA and the Kruskal–Wallis rank-sum test to explore regional characteristics that influence the differences in the necessity of antecedent conditions. Considering that the antecedent variables fail normality tests and homogeneity of variance tests, a non-parametric approach, specifically the Kruskal–Wallis rank-sum test, is used. In the analysis, the sample is divided into three major urban agglomerations—Beijing–Tianjin–Hebei, the Yangtze River Delta, and the Pearl River Delta—to assess whether the consistency of antecedent conditions differs significantly across regions. As shown in

Table 5. Kruskal–Wallis rank-sum test.

Variable	Mean	SD	Chi-Square	p-Value
High Digital Infrastructure	0.75	0.27	4.2337	0.1204
High Digital Technological Innovation	0.84	0.24	1.2439	0.5369
High Digital Human Capital	0.69	0.39	3.3219	0.19
High Digital Government Governance	0.69	0.23	8.3349	0.01549 **
High Digital Attention	0.69	0.29	3.7612	0.1525
High Digital Finance	0.68	0.19	17.576	0.0001525 ***

Note: ***, and ** correspond to the 1%, and 5% significance levels, respectively.

, significant regional differences are observed in the necessity effects of high digital government governance and high digital finance on RIER. This suggests that the regional heterogeneity in government governance capacity and digital financial development leads to divergent mechanisms through which these factors influence system resilience.

4.2. Adequacy Analysis of Conditional Configuration

The core of the QCA method lies in uncovering how different combinations of antecedent conditions jointly influence the occurrence of the outcome variable. The key criterion is whether a given configuration of conditions demonstrates a sufficient relationship, as indicated by its consistency level. Following prior research and based on the specific context of this study [87], the consistency threshold is set at 0.8, the frequency threshold at 4, and the PRI (Proportional Reduction in Inconsistency) threshold at 0.7. During the counterfactual analysis, all six antecedent conditions are systematically coded in terms of their presence or absence, resulting in three enhanced solution types: complex solution, parsimonious solution, and intermediate solution. In line with standard QCA procedures, the intermediate solution is prioritized, with reference to the parsimonious solution, to identify typical configurational paths that satisfy the conditions of sufficiency. The detailed results of the configurational analysis are presented in

Table 6. Results of configuration analysis.

Condition Variables	High RIER				Non-High RIER
	H1	H2	H3	H4	H5	H6	H7	H8
X1	⬤	⬤	•	⬤	⊗	U	U	⊗
X2	⬤	•	⬤	•	•	⊗	⊗	⊗
X3	⬤	⬤			U	U		U
X4			U	⬤	U	•	U	U
X5	⬤		⬤		⊗	⊗	⊗	•
X6		•	•	•	⊗	U	•
Consistency	0.95	0.941	0.969	0.946	0.956	0.955	0.958	0.951
PRI	0.939	0.927	0.949	0.927	0.910	0.907	0.913	0.903
Coverage	0.368	0.419	0.202	0.518	0.404	0.223	0.410	0.455
Unique coverage	0.02	0.021	0.041	0.144	0.015	0.031	0.029	0.043
BECONs adjusted distance	0.046	0.064	0.032	0.041	0.067	0.090	0.064	0.058
WICONS adjusted distance	0.243	0.286	0.229	0.279	0.134	0.158	0.179	0.152
Aggregate consistency	0.938				0.945
Aggregate coverage	0.671				0.722

Note: ⬤ indicates the presence of core conditions, • indicates the presence of peripheral conditions, ⊗ indicates the presence of core absent conditions, U indicates the presence of a missing peripheral condition, and a blank cell means that the condition can be either present or absent.

.

4.2.1. Pooled Results

As shown in Table 6, the consistency scores of both individual solutions and the overall solution for high RIER are significantly above the established threshold of 0.75. Specifically, the overall solution exhibits a consistency of 0.938, indicating that the four identified configurations can be considered sufficient combinations for achieving high resilience. Additionally, the coverage of the overall solution is 0.671, suggesting that these four configurations account for 67% of the high-resilience cases, thus demonstrating strong explanatory power. Based on these configurations, this study further explores the differentiated alignment of technological, organizational, and environmental factors in enhancing RIER, revealing the varied mechanisms through which each factor contributes to resilience under different contextual conditions.

(1)

Configurational Analysis of High RIER

Configuration 1: Technology–Organization–Environment Synergistic Type. This configuration constitutes a sufficient condition for enhancing RIER, characterized by the presence of high digital infrastructure, high digital technological innovation, high digital human capital, and high digital attention as core conditions. This path explains 36.8% of the high-resilience cases, and 2% of the cases are uniquely explained by this configuration, indicating strong distinctiveness and representativeness. This configuration reflects the synergistic interplay among technological, organizational, and environmental dimensions, providing systematic support for enhancing resilience. It suggests that, under DT, improving RIER requires integrated and coordinated development: optimizing digital infrastructure deployment, building high-level technology collaboration platforms, improving digital technology diffusion mechanisms, innovating talent development and recruitment models, and reinforcing policy supply mechanisms. Together, these efforts contribute to the formation of a multi-element, co-evolving, and high-resilience innovation ecosystem. Representative cases of this configuration include Guangzhou (2018–2022) and Suzhou (2018–2022), primarily located within the Yangtze River Delta and Pearl River Delta urban agglomerations. Taking Guangzhou as an example, the city has continuously advanced the development of next-generation digital infrastructure, focusing on 5G-A/6G networks, intelligent computing centers, and industrial internet platforms, with full “dual-gigabit” network coverage in districts like Tianhe and Nansha. Redundant infrastructure design has also been strengthened to enhance the system’s ability to withstand shocks. Centered on the Pazhou Artificial Intelligence and Digital Economy Pilot Zone, Guangzhou has built a “government–enterprise–research” collaborative innovation system, with platforms such as the Huawei Ascend AI Innovation Center, focusing on frontier technologies including large AI models and quantum computing, promoting joint R&D on key technologies. To meet the growing demand for high-level digital talent, the city has launched the “Greater Bay Area Digital Talent Green Card” program, offering tax incentives and cross-border data flow facilitation, along with a dynamic digital talent demand map powered by AI algorithms for precise talent–enterprise matching. In addition, Guangzhou has established a municipal Digital Resilience Special Fund, adopting a “grant-investment hybrid” model to support breakthroughs in core technologies. These efforts have gradually shaped a government-led, enterprise-participated, market-driven collaborative governance mechanism, significantly enhancing the resilience of the regional innovation ecosystem and offering a practical paradigm for sustainable urban innovation under DT.

Configuration 2: Technology-Driven, Talent-Supported Type. This configuration is characterized by high digital infrastructure and high digital human capital as core conditions, with high digital technological innovation and high digital finance as peripheral conditions, collectively driving the enhancement of RIER. This path explains 41.9% of the high-resilience cases, with 2.1% of cases being uniquely explained by this configuration, indicating its distinctiveness and representativeness in certain regions. This configuration suggests that, in the digital economy era, regional innovation ecosystems need to prioritize the development of digital infrastructure, while also fostering a highly skilled digital workforce. It emphasizes the synergistic relationship between technology-driven innovation and organizational support to enhance system resilience. Representative cities for this configuration include Hangzhou (2018–2020, 2022), Beijing (2019–2020), and Changzhou (2019–2022), with these cities predominantly located in the Yangtze River Delta and Beijing–Tianjin–Hebei urban agglomerations. Taking Hangzhou as an example, the city has actively promoted the development of 5G IoT infrastructure, becoming one of the first pilot cities for 5G in China, successfully establishing over 20,000 5G base stations. In addition, the Hangzhou Municipal Government has introduced a series of policy measures, including the “Hangzhou Digital Economy Talent Introduction and Training Action Plan (2020–2022)”, to attract and nurture talent in the digital economy sector, enhancing the region’s digital competitiveness. Furthermore, Hangzhou has established a billion-dollar industrial fund, focusing on investing in hard-tech fields such as quantum computing, 6G, and brain-like intelligence, to drive breakthroughs in core technologies. In 2023, Hangzhou ranked sixth globally in the FinTech Index, following cities such as San Francisco and New York. Its cross-border payment volume accounted for 20% of the national total, demonstrating the city’s global competitiveness in the financial technology sector. It is worth noting that although Hangzhou consistently appears as a representative city of the “technology-driven and talent-supported” configuration across multiple years, it is not included in this configuration in 2021. This deviation may be attributed to two main factors. First, it may result from changes in the thresholds used in fuzzy set calibration under the dynamic QCA approach. As calibration anchors are data-dependent, slight shifts in the distribution may cause certain core or peripheral conditions—such as digital human capital or digital finance—to fall short of the specified thresholds in that year. Second, the deviation may reflect policy-related timing effects and transitional fluctuations. As 2021 marks the first year of China’s 14th Five-Year Plan, Hangzhou is undergoing a critical shift from infrastructure development to the deepening of digital application. The full effects of relevant policies may not yet be observable. Additionally, tightened national regulation on fintech in 2021 may have adversely impacted the membership scores of related conditions. From the perspective of configurational theory, this case also illustrates the nonlinear and dynamic nature of RIER. Even cities with strong digital foundations may temporarily fall out of high-resilience configurations due to policy shifts, industrial adjustments, or external shocks.

Configuration 3: Technology–Environment Dual-Driven Type. This configuration is characterized by high digital technological innovation and high digital government attention as core conditions, with high digital infrastructure and high digital finance as peripheral conditions, jointly driving the enhancement of RIER. This path explains 20.2% of the high-resilience cases, with 4.1% of high-resilience cases being uniquely explained by this configuration, indicating its distinctiveness in certain regions. This configuration suggests that, in cities where digital government governance is relatively weak, RIER is primarily driven by enhancing digital technological innovation and increasing government attention to digitalization. By strengthening technological innovation and government policy support, regions can maintain high levels of innovation vitality and resilience, even in the absence of a robust digital governance mechanism. Representative cities for this configuration include Dongguan (2018–2019), Foshan (2018–2019), and Nanjing (2018–2020), primarily located in the Yangtze River Delta and Pearl River Delta urban agglomerations. Taking Dongguan as an example, the city has established a “New Generation Artificial Intelligence Technology Innovation Program” to support enterprises in developing core technologies such as AI chips and algorithms, strengthening basic research and technological breakthroughs, thereby enhancing RIER. Additionally, the Dongguan Municipal Government has implemented the “Dongguan Digital Economy Development Action Plan”, clearly outlining digital economy development goals and key tasks, further providing policy support for regional innovation.

Configuration 4: Technology-Driven, Government-Supported Type. This configuration is characterized by high digital infrastructure and high digital government governance as core conditions, with high digital technological innovation and high digital finance as peripheral conditions, collectively driving the enhancement of RIER. This path explains 51.8% of the high-resilience cases, with 14.4% of high-resilience cases being uniquely explained by this configuration, highlighting its distinctive role in specific urban agglomerations. This configuration indicates that the resilience of regional innovation ecosystems primarily depends on enhancing digital infrastructure development and strengthening digital government governance. This model emphasizes providing robust support and security for innovation ecosystems through the improvement of technological infrastructure and the promotion of digital governance. Representative cities for this configuration include Shanghai (2018–2022), Shenzhen (2018–2022), and Ningbo (2018–2022), with cases distributed across the three major urban agglomerations. Taking Shanghai as an example, the city government has heavily invested in the construction of artificial intelligence (AI) computing centers, providing strong computational power support to drive the development of AI and other technological applications. At the same time, Shanghai has actively advanced digital government initiatives, pioneering the “One Network to Handle All” platform, which integrates municipal government services and enables streamlined access, significantly improving government service efficiency. Additionally, Shanghai has established a unified data resource sharing and exchange platform, facilitating data sharing and collaboration across government departments. Furthermore, Shanghai has become a global hub for financial technology enterprises, attracting leading companies such as Ant Group, Tencent Financial Technology, and JD Digits, further promoting the integration of regional financial innovation and the digital economy.

Overall, Configuration 2 and Configuration 4 exhibit high raw coverage, indicating that these two paths have strong applicability across most cities in the Beijing–Tianjin–Hebei, Yangtze River Delta, and Pearl River Delta urban agglomerations, contributing significantly to the enhancement of RIER. In contrast, Configuration 1 and Configuration 3 show lower raw coverage, suggesting that these paths are applicable only in specific regions, primarily concentrated in the Pearl River Delta and Yangtze River Delta urban agglomerations. Furthermore, the unique coverage for all four configurations is relatively low, indicating that there is a degree of substitutability between the pathways for enhancing RIER in different regions. In other words, the applicability of each pathway is not unique across regions. This reflects the diversified pathways for enhancing RIER, where regions can select the most appropriate combination based on their unique characteristics.

(2)

Configurational Analysis of Non-High RIER

This study further analyzes the digital transformation-related antecedents associated with non-high RIER and identifies four distinct configurations leading to this outcome. The configurational results indicate that the presence of a single digital transformation factor, such as advanced digital innovation capacity, strong digital governance, high digital attention, or well-developed digital finance, is insufficient to enhance RIER when not supported by other complementary conditions. In other words, improvements in individual digital transformation elements alone do not lead to a resilient regional innovation ecosystem. Although the representative cities in these configurations may exhibit a high level of one specific digital transformation factor, such strength is only relative to their own internal digital profile and not indicative of a well-rounded transformation. These cities should therefore adopt context-sensitive strategies by leveraging their comparative advantages and exploring tailored development pathways. More importantly, they should aim to foster synergistic improvements across multiple digital transformation dimensions to achieve comprehensive and coordinated progress toward ecosystem resilience.

4.2.2. Between-Group Result

Inter-group consistency measures whether the configurations of conditions for each year in the sample period are sufficient conditions for the outcome, reflecting the cross-sectional consistency level for each year in the panel data [87]. Although the inter-group consistency-adjusted distances for all four configurational paths are below 0.2, indicating no significant temporal effects, it is still worthwhile to further explore the temporal trends in consistency across the five configurations, as shown in Figure 3. From the figure, it is evident that the consistency levels for the four configurations remained between 0.8 and 1 from 2018 to 2022, with minimal overall fluctuations. Specifically, from 2018 to 2019, the explanatory power of Configuration 3 increased, while the explanatory power of Configurations 1, 2, and 4 declined. From 2019 to 2020, the explanatory power of Configuration 4 increased, whereas the explanatory power of Configurations 1, 2, and 3 decreased. Between 2020 and 2022, all configurations experienced varying degrees of decline in explanatory power. These results challenge the traditional static configurational approach in QCA, addressing the lack of a temporal dimension in conventional QCA. They also illustrate the dynamic evolution of the driving mechanisms of RIER. The collective decline in explanatory power across all configurations in 2020 may be attributed to the COVID-19 pandemic and the external shocks it caused. The sudden outbreak of public health crises—such as the COVID-19 pandemic—disrupted the flow of regional resources, supply chain stability, and innovation collaboration, while simultaneously imposing greater demands on organizational structures, policy responsiveness, and institutional coordination. Under such highly uncertain conditions, the overall effectiveness of DT drivers in enhancing system resilience tends to weaken. Although the consistency-adjusted distances between configurations remain below 0.1 and thus do not substantially undermine the explanatory power of the model, these temporary fluctuations highlight the stage-specific and dynamic nature of RIER. Notably, during crisis periods, all six digital transformation antecedents interact through distinct but complementary mechanisms to jointly activate resilience capabilities. Specifically, digital infrastructure ensures the continuity of innovation activities by supporting remote collaboration and knowledge exchange; digital innovation accelerates deployment in public health and industrial response scenarios, fostering new business models and expanding evolutionary pathways; digital human capital enhances organizational agility and adaptive innovation capacity; digital government governance improves systemic responsiveness through enhanced institutional flexibility and actor coordination; digital attention guides resource allocation and fosters collective consensus, providing a foundation for digital reform; and digital finance alleviates financing constraints amid restricted capital flows, ensuring the efficient distribution of innovation resources. Collectively, these mechanisms reinforce the ecosystem’s ability to sense, absorb, adjust, and reconfigure in the face of disruptions, underscoring the strategic value of DT in strengthening systemic resilience.

4.2.3. Within-Group Result

Intra-group consistency is based on city-level analysis, measuring whether the configurations of conditions for each city during the sample period are sufficient conditions for the outcome. Among the four configurations, the consistency scores for most cities are above 0.75, indicating strong consistency. However, the intra-group consistency-adjusted distances for all configurations are greater than 0.2, suggesting significant differences in the explanatory power of each configuration across cities. From the overall analysis, it is evident that between 2018 and 2022, different cities did not follow a consistent set of conditions, meaning that in certain cities, multiple paths can lead to the enhancement of RIER. For example, cities such as Dongguan, Shanghai, Foshan, Beijing, Nantong, Tianjin, and Ningbo show perfect consistency (scores of 1) in all four configurations, demonstrating extremely high consistency. Zhuhai exhibits high consistency in Configuration 3, while Langfang, Chizhou, Yangzhou, Chuzhou, and Zhenjiang show high consistency in Configuration 4, though Chuzhou and Zhenjiang also exhibit high consistency in Configurations 1 and 2. On the other hand, cities such as Hengshui, Tongling, Ma’anshan, Chengde, and Zhangjiakou show low consistency across all configurations, with consistency scores lower than 0.5. Therefore, further analysis is needed to explore the appropriate paths for different regions across different years in order to more accurately reveal the diverse pathways and dynamic changes in enhancing RIER.

According to the analysis of regional coverage averages (see

Table 7. Geographical coverage.

Region	H1	H2	H3	H4
Beijing–Tianjin–Hebei	0.411	0.376	0.375	0.327
Yangtze River Delta	0.420	0.467	0.332	0.544
Pearl River Delta	0.358	0.327	0.336	0.547

), there are significant differences in the distribution of the four configurations across the three major urban agglomerations. Specifically, Configuration 1 (Technology–Organization–Environment Synergy Type) has the widest coverage in the Yangtze River Delta region. This is primarily due to the region’s comprehensive advantages in technology innovation resource aggregation, industrial chain integrity, internationalization level, and regional collaborative development, providing a solid foundation for the synergistic interaction of the technological, organizational, and environmental dimensions, thus driving the enhancement of RIER. Configuration 2 (Technology-Driven, Talent-Supported Type) shows higher coverage in the eastern regions, particularly in the Beijing–Tianjin–Hebei area. This configuration relies on high-level digital infrastructure and high-quality digital human capital, supplemented by the development of digital technological innovation and digital finance, aligning well with the Yangtze River Delta’s strengths in talent mobility, regional collaboration, and balanced digital infrastructure deployment. Configuration 3 (Technology-Driven, Government-Supported Type) has a relatively balanced distribution across the three urban agglomerations, with slightly higher coverage in the Beijing–Tianjin–Hebei region. This suggests that digital technological innovation and the effective implementation of regional digital policies can compensate for the shortcomings in digital government governance. Configuration 4 (Technology–Environment Dual-Driven Type) shows higher coverage in the Pearl River Delta and Yangtze River Delta, while the Beijing–Tianjin–Hebei region faces challenges in digital government governance, such as insufficient regional synergy and a low level of marketization, particularly in Hebei and Tianjin, where digital development lags. This leads to internal development imbalances within the region. This comparison further emphasizes the critical role of digital infrastructure development and government digital governance capabilities in enhancing the resilience of regional innovation ecosystems.

At the theoretical level, this study confirms the embedded influence of organizational and environmental factors on the pathways of DT. Even with similar digital elements, regional differences in institutional receptiveness, resource allocation mechanisms, and coordination efficiency lead to significant variations in outcomes. This finding aligns with the concept of contextual embeddedness [99], emphasizing the importance of accounting for structural disparities and asynchronous development across regions when enhancing resilience. Drawing on dynamic capabilities theory [100], the study highlights the region-level capacities to sense environmental change, seize emerging opportunities, and reconfigure internal and external resources. In advanced regions such as the Yangtze River Delta and Pearl River Delta, strong sensing abilities enable rapid responses to policy, market, and technological shifts, giving rise to complex, multi-dimensional configurations (e.g., Configuration 1). These regions also demonstrate superior integration efficiency through abundant human capital and robust digital infrastructure (e.g., Configuration 2), enhancing flexibility and responsiveness through digital tools. Notably, during major crises such as the COVID-19 pandemic, regions with strong adaptive governance capabilities can rapidly reconfigure collaborative mechanisms, maintain system stability, and even foster new transformation pathways. Institutional theory further underscores the role of institutional arrangements, regulatory norms, and cognitive frameworks in shaping organizational behavior and systemic evolution [101]. First, the maturity of the institutional environment conditions the formation of configurations. Regions with strong digital governance and effective coordination mechanisms tend to exhibit government-led configurations (e.g., Configuration 3), facilitating information sharing, actor collaboration, and risk response. Second, institutional inertia and path dependency constrain the speed of configuration evolution. In some regions, rigid institutional legacies hinder timely policy adjustment, limiting the potential of emergent configurations (e.g., Configuration 4 in parts of the Beijing–Tianjin–Hebei region). Finally, institutional legitimacy plays a critical role in enabling digital configurations to embed and scale. When emerging digital forms—such as innovation platforms or data alliances—receive institutional recognition and support, they are more likely to be incorporated into formal policy frameworks and diffused at scale, promoting the institutionalization of specific configurations.

4.3. Robustness Test

Building on prior research [102], this study conducts a robustness check on the “Technology–Organization–Environment” antecedent condition configurations driving the enhancement of RIER. First, the original consistency threshold and PRI consistency threshold were adjusted to 0.85 and 0.75, respectively. The results show a clear subset relationship between the adjusted configurations and the original configurations. Therefore, after adjusting the consistency levels, the configurational analysis results remain unchanged, confirming the robustness of the findings. Second, the calibration thresholds were adjusted to 0.8, 0.5, and 0.2 [87]. Although the overall solution’s consistency and coverage slightly changed, the configurational results did not undergo any substantial alterations, further validating the robustness of the analysis.

5. Conclusions and Suggestion

5.1. Conclusions

Building a resilient regional innovation ecosystem is crucial for regions to cope with external shocks and uncertainties, and it is also an inevitable choice for ensuring the long-term, stable, and sustainable development of regional innovation ecosystems [22]. This study addresses the following core question: “Under what digital contexts can combinations of technology, organization, and environment enhance the resilience of regional innovation ecosystems?” Using panel data from 48 prefecture-level cities in the Beijing–Tianjin–Hebei, Yangtze River Delta, and Pearl River Delta urban agglomerations from 2018 to 2022, this research applies the dynamic QCA method and innovatively explores the driving paths for enhancing the resilience of regional innovation ecosystems within the TOE theoretical framework under a digitalization context. The findings suggest that no single technological, organizational, or environmental antecedent condition can independently enhance the resilience of regional innovation ecosystems. Instead, resilience needs to be driven through the synergistic combination of different factors. Nevertheless, digital government attention and digital finance show significant regional differences in their necessity for enhancing resilience. Based on this, a theoretical framework is constructed, with digital infrastructure, digital innovation levels, digital human capital, digital government governance, digital government attention, and digital finance as antecedent factors, further deepening research on innovation ecosystems. The study also finds that there is a causal asymmetry or “same result from different paths” phenomenon between the TOE framework and the enhancement of RIER. This is specifically manifested in four driving paths: a three-dimensional synergy path involving high digital infrastructure, high digital innovation levels, high digital human capital, and high digital government attention; a technology-driven, talent-supported path with a strong emphasis on digital infrastructure under high digital human capital; a technology–environment dual-driven path with the absence of digital government governance but a focus on digital technology innovation and gaining high digital government attention; and a technology-driven, government-supported path with high digital infrastructure and high digital government governance. Among these, three paths center on high digital infrastructure, indicating that digital infrastructure is the prerequisite and foundation for innovation-driven activities in the digital age. Although no significant temporal effects were observed, some configuration solutions in specific years were still influenced by unobserved factors, and the applicability of these configurations varies across regions, further revealing the diversity and complexity of pathways for enhancing RIER.

5.2. Theoretical Contribution

First, in terms of theoretical framework construction, compared to traditional studies that typically explore the factors influencing RIER from a single perspective, this study constructs a “Technology–Organization–Environment” (TOE) framework from a resource allocation perspective. This framework breaks through the limitations of the traditional single-perspective approach and offers a novel analytical lens. It enables a more comprehensive and in-depth understanding of the internal mechanisms driving the enhancement of RIER, laying a solid foundation for further research on innovation ecosystem theory.

Second, in terms of research methodology, this study innovatively applies dynamic QCA with panel data. This method fully accounts for temporal effects, dynamically capturing the complex interactions between various influencing factors and their evolving characteristics over time in the process of enhancing RIER. By doing so, it provides a deeper exploration of the complex driving paths for resilience enhancement. This approach effectively addresses the limitations of previous studies, which relied solely on cross-sectional data or overlooked temporal changes, offering more scientifically rigorous and methodologically robust support for research on RIER.

Finally, in terms of practical guidance, the conclusions drawn from this study have significant practical implications. The findings highlight the differences in development levels, resource endowments, and innovation environments across regions. These results provide strong reference points for policy improvements and innovation strategy adjustments tailored to each region’s development context. They help enhance the resilience of regional innovation ecosystems and promote high-quality innovation-driven development across different regions.

5.3. Policy Implications

First, innovation ecosystem governance is a systemic project, and enhancing the resilience of regional innovation ecosystems should be approached from a synergistic perspective of technology, organization, and environment. Regions should take a holistic configurational perspective, focusing on the interaction and alignment between digital technologies, digital organizations, and digital environmental conditions. Regions can refer to the configurational template for high RIER, selecting development paths that best suit their unique circumstances, leveraging their strengths and addressing weaknesses. For example, Hebei and Tianjin in the Beijing–Tianjin–Hebei region should focus on optimizing digital infrastructure and creating a favorable digital environment, gradually transitioning from local breakthroughs to a more holistic approach, and accelerating the flourishing development of the digital economy. This, in turn, will more effectively drive regional innovation and enhance the resilience of their innovation ecosystems.

Second, digital infrastructure plays a universal role in enhancing the resilience of regional innovation ecosystems. Regions should accelerate the improvement of digital infrastructure, build innovation networks centered around digital platforms, and fully leverage the innovative driving force of data integration, analysis, and processing. By improving the efficiency of information flow, optimizing resource allocation, and enhancing system responsiveness, regions can further strengthen their innovation capabilities. Additionally, regions should optimize their innovation services and data governance systems, create high-quality hardware environments, and enhance the resilience, efficiency, and competitiveness of their innovation ecosystems. This will provide a solid foundation for the sustainable development of regional economies and societies.

Third, by strengthening the driving forces of digital innovation, digital talent, and digital government, regions can enhance their digital innovation capabilities, stimulate a digital innovation atmosphere, and facilitate knowledge flow and sharing, significantly boosting the resilience of their regional innovation ecosystems. For example, optimizing talent attraction policies to draw high-level digital talent into the region can inject fresh energy into the innovation ecosystem. Additionally, establishing digital talent exchange platforms and fostering industry–academia–research cooperation mechanisms can promote the flow and sharing of knowledge both within and outside the region, thereby enhancing overall innovation capabilities. These strategies not only contribute to strengthening a region’s resilience to risks but also drive the region’s sustainable development in the digital age, helping it secure a leading position. In this process, local governments play a crucial role and should provide strong support for building the resilience of regional innovation ecosystems through policy guidance, resource investment, and mechanism innovation.

Fourth, from a policy perspective, there is a growing emphasis across countries on advancing digital transformation through regional policy initiatives. Hervas-Oliver et al. (2021) [103] highlight the European Digital Innovation Hub (DIH) initiative, which aims to foster collaborative alliances responsive to local contexts and development needs by building multi-actor platforms that connect both local and non-local stakeholders. This approach accelerates the transition to Industry 4.0 while promoting co-creation and co-governance in regional digital policy design. These platforms primarily function through spatially anchored coordination and policy dialog, facilitating the joint formulation of regionally adapted digital strategies. In this context, regional governments are encouraged to actively promote the construction of innovation platforms that act as intermediaries for knowledge exchange, capacity building, and technological empowerment. By establishing cooperation mechanisms centered on interactive learning, these platforms can enhance multi-actor collaboration among governments, universities, research institutions, and enterprises. At the same time, digital transformation policies should be more closely aligned with regional innovation systems by strengthening spatial embeddedness, local adaptability, and organizational coordination. This alignment supports the development of regionally anchored innovation ecosystems that are demand-driven and network-enabled. Additionally, smart specialization strategies should guide policy implementation by leveraging regional strengths and industrial foundations to promote path renewal and industrial upgrading. Finally, policymakers must continue to improve institutional infrastructures and establish robust support systems for digital technology adoption, thereby ensuring both a stable institutional environment and sustained momentum for regional digital transformation.

5.4. Limitations and Future Research

This study has several limitations that merit future investigation. First, the sample focuses on China’s three major urban agglomerations, which may limit the generalizability of the findings. Future studies should expand to more diverse regions to explore spatial–temporal heterogeneity across different development contexts. Second, variable measurement is constrained by reliance on publicly available data, which may overlook unstructured intra-regional factors such as organizational culture or institutional cognition. The inclusion of big data, survey evidence, or qualitative materials could enhance indicator precision. Third, while this study employs the TOE framework and draws from dynamic capabilities and institutional theory, more complex institutional logics—such as embeddedness, multi-level governance, and collaborative networks—are not fully addressed. Finally, the practical applicability of the proposed configurations and their feedback mechanisms remain underexplored. Future research could incorporate policy experiments or real-world collaborations to assess contextual fit and implementation outcomes.