Introduction

Regional innovation ecosystems, as a key variant of innovation ecosystems (Chung, 2002), have become a strategic imperative for companies to actively embed themselves within (Li et al., 2022). However, when such ecosystems exhibit low network density and imbalanced population structures, incumbent firms face elevated transaction costs in accessing innovation resources. Furthermore, as complex adaptive systems, their emergence, development, and evolution depend on the coexistence of actors occupying opposing polarities of key tensions, such as competitors and collaborators, or adapters and disruptors. In the absence of actors capable of fulfilling these roles, particularly high-tech entrepreneurial firms, seemingly stable equilibria may lock incumbents into path-dependent trajectories that constrain innovation. By contrast, the entry of high-tech entrepreneurship, which is characterized by high R&D density, rapid creation of new knowledge (Chung et al., 2022), and innovation-driven, can enhance the network density of regional innovation ecosystems and facilitate the integration of innovation resources. Moreover, through knowledge filtering mechanisms, such entrepreneurship generates knowledge spillovers that help incumbent firms overcome path dependence. Thus, theoretically, high-tech entrepreneurship can catalyze the innovative development of regional innovation ecosystems.

Existing literature has examined the economic and non-economic impacts of entrepreneurship across various domains, including economic growth (Acemoglu and Cao, 2015), innovation (Hall et al., 2012; Galindo and Méndez, 2014; Paik et al., 2019; Boone et al., 2019; Ejdemo and Örtqvist, 2020; Chung et al., 2022; Si et al., 2022; Estrin et al., 2022; Wu and Li,2024), innovation efficiency (Chung et al., 2022), employment (Gilbert et al., 2004), firm performance (Zhang et al., 2021), personal well-being (Liu and Zhang, 2024), common prosperity (Liu et al.,2025), and societal challenges such as climate risk and social inequality (Seelos and Mair 2017). However, most studies focus primarily on macro- or micro-level dimensions. Few studies have adopted a meso-level perspective to examine the influence of high-tech entrepreneurial firms(Gumbau Albert, 2017), particularly the specific roles they play within regional innovation ecosystems.

Moreover, most existing studies examine the relationship between high-tech entrepreneurship and innovation from the perspectives of competition and knowledge-filtering mechanisms. However, these frameworks are insufficient in the current era, in which science, technology, and innovation have increasingly shifted toward an ecosystem-based paradigm. Meanwhile, the relationship between high-tech entrepreneurship and innovation is embedded in multiple paradoxes. At the meso level, newly entering high-tech entrepreneurial firms act simultaneously as ecosystem adapters and disruptors. As adapters, they increase network density and strengthen incumbents’ innovation networks. As disruptors, they introduce new technological trajectories, reshape population structures, and ultimately influence incumbent innovation. Correspondingly, at the micro level, high-tech entrepreneurial firms simultaneously serve as co-opetitors that cooperate and compete with incumbents, and as both adaptive innovators that reinforce incumbents’ core technological regimes and disruptive innovators that challenge established trajectories. Therefore, unraveling the relationship between high-tech entrepreneurship and innovation requires an integrated perspective combining innovation ecosystem theory with paradox theory. The study thus addresses the following question: How does high-tech entrepreneurship within a regional innovation ecosystem influence the innovation of incumbent firms?

To address this theoretical gap, we selected firms from Zhongguancun Science Park (ZSP) in China for our study. ZSP, a high-tech park built on technological entrepreneurship, has evolved over 40 years into a thriving innovation and entrepreneurship ecosystem. It has witnessed the emergence and growth of strategic industries, establishing itself as a globally recognized hub for innovation. According to the “Global Startup Ecosystem Report 2024” by the renowned research firm Startup Genome, Beijing ranked fourth globally in terms of startup ecosystems in 2022. However, by 2024, its global ranking dropped to eighth, highlighting the critical real-world importance of promoting the sustained development of the regional innovation ecosystem. Drawing lessons from historical experiences offers a viable approach to addressing this challenge, making ZSP an ideal subject for the study.

Grounded in innovation ecosystem theory and paradox theory, we propose that high-tech entrepreneurship can boost incumbent firms’ innovation. In particular, R&D investment and human capital serve as key pathways through which this effect is realized. Empirically, high-tech entrepreneurship is measured by the proportion and natural logarithm of newly established firms in four-digit industries within ZSP. The use of four-digit industry codes ensures greater precision, while focusing on unlisted small and medium-sized enterprises better reflects the authentic characteristics of entrepreneurial activity. Firm innovation is measured by patent applications, a widely recognized indicator in academic research, enhancing the generalizability and comparability of the results. A two-way fixed effects model is employed, supplemented by robustness checks, including the instrumental variable approach.

Theoretical background and hypotheses

Innovation ecosystem view

The innovation ecosystem, rooted in the metaphor of biological ecology (Moore, 1996; Iansiti and Levien, 2004), provides a lens for analyzing the configuration of interdependent and independent innovation activities. At its core, this perspective seeks to delineate the intricate economic relationships underlying innovation processes. Two dominant schools of thought frame the analysis of innovation ecosystems: the “ecosystem as affiliation” (Iansiti and Levien, 2004; Rong and Shi, 2015) and the “ecosystem as structure” (Adner, 2017; Walrave et al., 2018; Granstrand and Holgersson, 2020). The’ ecosystem as affiliation’ views innovation ecosystems as networks of a large number of loosely coupled and coexisting symbiotic participants around keystone species that form networks for realizing and capturing the value of innovation (Jacobides et al., 2016). This perspective emphasizes factors such as network density, number of partners, and intra- and inter-population relationships, making it particularly suited for meso- and macro-level analyses (Adner, 2017; Walrave et al., 2018; Granstrand and Holgersson, 2020). In contrast, the ‘ecosystem as structure’ perspective focuses on the components and complementary elements essential for focal innovations. It posits that these ecosystems consist of activities, actors, value propositions, and links, all working interdependently to deliver consistent value propositions to end-users (Adner, 2017). This perspective emphasizes the non-hierarchical interdependencies formed around focal innovations and has a comparative advantage in characterizing micro-level innovation ecosystems (Adner and Feiler, 2019; Burford et al., 2022). Importantly, Adner (2017) argues that these two perspectives are not mutually exclusive but are instead complementary, providing a more comprehensive understanding of innovation ecosystems.

The paradoxical relationship between high-tech entrepreneurial firms and incumbent firms within the innovation ecosystem

The integration of paradox theory with innovation ecosystem theory provides a compelling and appropriate lens for explaining the relationship between high-tech entrepreneurship and incumbent innovation. Paradoxes refer to the simultaneous existence and ongoing evolution of contradictory yet interdependent elements (Smith and Lewis, 2011), with contradiction and interdependence as their core characteristics (Schad et al., 2017). Such contradictions generate tensions between opposing elements, which become especially salient in complex and dynamic systems. Embedded within regional innovation ecosystems, high-tech entrepreneurial firms can leverage system-level resources to create value and contribute to ecosystem development. At the same time, due to asymmetries in power and ecological niches, they inevitably experience multiple forms of tension and conflict with incumbent firms. Based on ecosystem theory and paradox theory, these tensions manifest as paradoxical relationships such as legitimacy versus distinctiveness (Taeuscher and Rothe, 2021) and competition versus cooperation.

First, high-tech entrepreneurial firms survive and grow by continually navigating the paradoxical tension between legitimacy and distinctiveness, a core duality that manifests prominently at both the meso and micro levels. At the meso level, as a newly emerging population, high-tech entrepreneurial firms must conform to existing ecosystem rules to gain legitimacy and access essential resources. In doing so, they act as ecosystem adapters, innovative actors whose activities remain largely within prevailing rules, standards, and relational networks. Their strategy focuses on leveraging, optimizing, or incrementally extending the system, enhancing compatibility, occupying ecological niches, and reinforcing incumbent value chains. Their entry increases network density and improves incumbents’ access to innovation resources. Simultaneously, these firms must differentiate themselves through entrepreneurial narratives and disruptive, trajectory-shifting technologies to secure superior ecological positions and resource support. In government-architected regional innovation ecosystems, such distinctiveness often serves as a prerequisite for legitimacy (Taeuscher et al., 2021). When firms leverage this distinctiveness to disrupt incumbents’ markets and contest dominant ecological positions, they trigger an adaptive response among incumbent populations, promoting population-structure renewal and ultimately shaping incumbent innovation. In this phase, high-tech entrepreneurial firms act as ecosystem disruptors, innovative actors who introduce new logics, technologies, or business models to challenge, circumvent, or reshape the core rules and structure of the existing ecosystem. Their strategy focuses on overcoming constraints, creating new value chains, and building alternative networks. At the micro level, to gain legitimacy, high-tech entrepreneurial firms operate as adaptive innovators. They embed themselves in incumbents’ coopetition and value networks and contribute to value co-creation (Haans, 2019). Simultaneously, to demonstrate distinctiveness, they must also assume the role of disruptive innovators that overturn incumbents’ innovation trajectories and business models by offering superior cost–performance solutions or by creating entirely new market demand.

Second, the paradox of competition and cooperation operates at the micro level. Driven by opportunity-seeking and technological breakthroughs (Beckman, 2012), high-tech entrepreneurial firms collaborate with incumbents to access critical resources, opportunities, and markets (Ansari et al., 2016). At the same time, they compete for market share. This persistent co-opetition enriches the co-opetitive network (Corbo et al., 2022) and, in turn, boosts incumbent firm innovation.

The dual impacts of high-tech entrepreneurship on incumbent firms across both the meso and micro levels reveal the limitations of relying solely on either the ecosystem-as-affiliation view or the ecosystem-as-structure view. To more comprehensively capture these dynamics, this study integrates these perspectives with paradox theory to construct a unified framework for understanding how high-tech entrepreneurship shapes incumbent innovation within regional innovation ecosystems.

High-tech entrepreneurship and innovation among incumbent firms within the regional innovation ecosystem

From a meso perspective and based on the perspective of affiliation, the entry of high-tech entrepreneurship into the regional innovation ecosystem can impact the innovation of incumbent firms through influencing network density and population structure.

First, as ecosystem adapters, the entry of high-tech entrepreneurial firms enhances the network density of the regional innovation ecosystem, thereby promoting innovation among incumbent firms. Innovation ecosystems comprise the knowledge economy and the business economy (Jackson, 2011) and are structured around interrelated knowledge, technology, and market networks. Actors embedded in these networks, including firms, organizations, and government entities, constitute the ecosystem’s key nodes. The richness and diversity of these network nodes directly determine the ability of embedded firms to access, integrate, and transform creative, knowledge-based, technological, and market-oriented resources. By recruiting, attracting, and facilitating the entry of high-tech entrepreneurship, the ecosystem enhances its accumulation of knowledge and resources, thereby increasing both resource abundance and network density. This enriched environment enables incumbent firms to leverage synergies across knowledge and technology networks, allowing for rapid, large-scale, and cost-effective integration and recombination of innovation resources. As a result, incumbent firms are better positioned to generate novel ideas, explore new technological trajectories, and ultimately strengthen their innovation capacity (Beltagui et al., 2020). Meanwhile, interactions within market networks accelerate experimentation, commercialization, and scaling of new technologies. Through the iterative integration of knowledge, technology, and market networks, incumbents can mitigate information asymmetries, shorten innovation cycles, and identify more efficient innovation pathways, further promoting technological upgrading.

Second, as ecosystem disruptors, the entry of high-tech entrepreneurial firms optimizes the population structure, thereby stimulating innovation among incumbent firms. Their disruptive effects unfold through two intertwined forms of differentiation: structural differentiation within incumbent populations and niche differentiation between incumbents and high-tech entrepreneurial firms. By shattering the incumbent-dominated equilibrium, high-tech entrepreneurial entrants trigger an ecological bifurcation: strategically flexible incumbents evolve into an ambidextrous firm, while inert firms ossify, facing niche erosion. This differentiation, as an efficient mechanism for selection and resource allocation, both removes innovation barriers for incumbent firms and enhances their adaptability. Meanwhile, high-tech entrepreneurial firms open new technological trajectories and value networks, enabling them to “leap” into pioneer niches centered on high-risk, frontier exploration. This niche expansion simultaneously pressures and releases evolutionary incumbents, allowing them to specialize in “1-to-N” exploitation where they hold comparative advantages. Together, these dual differentiation processes reconfigure and optimize the ecosystem’s population structure, enabling incumbents to concentrate innovation resources on their most competitive domains and ultimately enhancing their innovation performance.

At the micro level, from a structural perspective, high-tech entrepreneurship firms influence incumbent innovation by simultaneously acting as co-opetitors, adaptive innovators, and disruptive innovators within regional innovation ecosystems. First, as co-opetitors, these firms exert both direct and indirect influences on the innovation of incumbent firms. Coopetition naturally occurs in ecosystems (Riquelme-Medina et al., 2022; Cozzolino et al., 2021). Directly, increased inter-firm coopetition intensity enhances the scope and efficiency of sharing, interacting with, and transforming key innovation resources such as data, technology, and talent (Xie et al., 2023). It facilitates the diffusion, dissemination, reproduction, and generation of both explicit and tacit knowledge (Gast et al., 2019). This enables incumbents to access complementary resources and frontier technological knowledge, mitigate innovation delays caused by resource constraints, and improve the efficiency of internal resource allocation, ultimately strengthening innovation capabilities (Bauer and Matzler, 2014; Gernsheimer et al., 2021). Moreover, intense coopetition allows incumbents to engage in deeper innovation-based division of labor with competitors while sharing costs and risks (Bouncken et al., 2020). They can also scan market information from multiple dimensions and respond rapidly to customer needs (Luo et al., 2006), engage in customized innovation and production, and achieve a certain degree of new “demand-driven planned economy”. This improves the accuracy of innovation, overcomes innovation resource mismatches caused by externalities, and avoids innovation waste caused by redundant resources (Xie et al., 2023). The competitive pressure embedded in collaboration also compels incumbents to sustain innovation momentum (Bengtsson and Kock, 2000).

Indirectly, coopetition encourages greater R&D investment. While R&D is essential for innovation, it requires stable and sustained financial support. By embedding themselves in denser coopetitive networks, incumbents can leverage external resource pools to supplement limited internal resources, thereby supporting more stable and long-term R&D efforts. Furthermore, cooperation with competitors that possess similar resources and strategic goals allows incumbents to integrate production lines, product portfolios, and distribution channels, thereby improving efficiency and market performance (Crick and Crick, 2021; Xie et al., 2023). These gains help alleviate executives’ concerns about career risks (Narayanan, 1985) and relax the financial constraints that typically hinder R&D (Hu et al., 2005; García-Quevedo et al., 2018), making incumbents more willing and able to maintain long-term R&D investment.

Second, high-tech entrepreneurship, whether functioning as an adaptive innovator or a disruptive innovator, exerts both direct and indirect positive effects on the innovation of incumbent firms within the regional innovation ecosystem. The direct effect of high-tech entrepreneurship in the regional innovation ecosystem manifests in two key ways. As adaptive innovators, high-tech entrepreneurial firms leverage agility and specialization to address focal innovation needs of incumbents by filling overlooked nodes and modules in incumbent-led value networks, thereby revitalizing and reinforcing incumbents’ core technologies. As disruptive innovators, they typically enter from peripheral or emerging domains and construct value networks distinct from those of incumbents. The resulting collision, substitution, and integrative upgrading between new and existing value networks redefine innovation activities and reshape the ecosystem’s connections. Such redefinition disrupts incumbents’ routines and innovation trajectories, helping them overcome path dependence and rigidity, enhance organizational agility, and explore novel technologies and markets (Bohnsack et al., 2021). Concurrently, the reconfiguration of connections reorganizes knowledge and resource flows, fostering architectural innovation through systemic recombination.

Indirectly, high-tech entrepreneurship enhances incumbent innovation through human capital. Human capital is an important high-level input factor for corporate innovation. When acting as adaptive innovators, high-tech entrepreneurial firms facilitate knowledge spillovers (Huggins and Thompson, 2015; Acs et al., 2009) by reducing knowledge filtering, enabling knowledge exchange and learning among incumbent employees. This process strengthens human capital quality and promotes incumbent innovation through continual knowledge renewal. When acting as disruptive innovators, high-tech entrepreneurs introduce novel perspectives through cumulative feedback processes, increasing the diversity and stock of human capital knowledge within the ecosystem and slowing knowledge depreciation (Boone et al., 2008). Moreover, the disruptions they induce may alter incumbents’ ecological niches, compelling them to reconfigure their human capital structures to adapt, survive, and sustain competitive advantage.

In summary, this paper proposes the following hypothesis:

H1: High-tech entrepreneurship promotes innovation among incumbent firms in the regional innovation ecosystem.

H2a: High-tech entrepreneurship promotes innovation among incumbent firms within the regional innovation ecosystem by enhancing their R&D investment.

H2b: High-tech entrepreneurship promotes innovation among incumbent firms within the regional innovation ecosystem by optimizing their human capital.

Based on the above analysis, this study develops a conceptual framework linking high-tech entrepreneurship, incumbent firms’ innovation, R&D investment, and human capital within a regional innovation ecosystem (see Fig. 1).

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Conceptual framework.

Methodology

Sample and data

This paper employs the panel dataset of the Micro Enterprise Survey of ZSP (2005-2015) to test and estimate the impact of high-tech entrepreneurship on the innovation of incumbent firms within the regional innovation ecosystem. Admittedly, the sample data drawn from Beijing’s ZSP may not fully ensure the global generalizability of this study’s findings. However, ZSP offers several unique advantages that make it an ideal setting for this research. First, ZSP is a globally leading regional innovation ecosystem (Trunina and Ashourizadeh, 2021; Han et al., 2021). As China’s first national high-tech industrial development zone and the country’s inaugural National Independent Innovation Demonstration Zone, ZSP has evolved from being “China’s Silicon Valley” into “the world’s Zhongguancun”. According to the Global Innovation Index (GII) Report 2024, Beijing has maintained its position as the world’s third-ranked global technology city cluster. According to the Global Startup Ecosystem Report 2024, Beijing’s ranking in the global startup ecosystem dropped from third in 2022 to eighth. However, it remains one of the top 10 leading startup ecosystems worldwide. This outstanding performance in innovation ecosystem rankings is largely attributed to the critical role played by ZSP. The data shows that from January to November 2022, the number of patents granted to enterprises in ZSP totaled 89,733, accounting for 64.2% of Beijing, and the number of PCT patent applications reached 8161, accounting for 79.4% of Beijing.

Second, ZSP serves as a concentrated hub of high-tech industries. According to Xinhua News Agency, ZSP witnessed the establishment of 51,497 new technology-driven firms in 2023, averaging approximately 141 new firms per day. Rapid growth has been observed in industries such as artificial intelligence, robotics, and new energy, with a year-on-year increase of about 30% in newly established firms. Currently, ZSP has developed a trillion-yuan-level industrial cluster in next-generation information technology, along with nine hundred-billion-yuan-level industrial clusters in sectors such as pharmaceuticals and healthcare, integrated circuits, and others. Third, selecting enterprise data from the ZSP enables a clear depiction and examination of the impact of high-tech entrepreneurship on the innovation of incumbent firms within the regional innovation ecosystem. The data, curated by the Zhongguancun Administrative Committee, encompasses extensive enterprise-level information, including legal entity details, operational performance, innovation activities, and access to innovation support policies. This dataset is particularly valuable due to its richness and continuity, with no interruptions in sample coverage during the observation period. These enterprises cluster within critical high-tech domains such as integrated circuits, next-generation information technology, bioengineering, advanced manufacturing, aerospace, new materials, and energy-saving technologies, which offer robust insights for this study. Although the ZSP database has been updated to 2017, this study sets 2015 as the endpoint to mitigate potential biases caused by patent truncation.

Fourth, although technologies and business models in regional innovation ecosystems have evolved rapidly since 2015 with the rise of artificial intelligence and the digital economy, our data extend only to 2015. This temporal limitation does not diminish the study’s theoretical contribution. The core contribution lies in explaining the relationship between high-tech entrepreneurship and incumbent innovation through the integrated lenses of innovation ecosystem theory and paradox theory, which remain highly relevant in contemporary contexts. Moreover, 2015 represents a critical juncture, as China elevated artificial intelligence to a national strategic priority. Accordingly, the study provides an important historical baseline for future research on entrepreneurship and innovation in digital-economy–driven regional innovation ecosystems.

Based on the purpose of the study, this paper treats the raw data as follows: (1) Remove the sample of agricultural enterprises. (2) Delete the data that are inconsistent with accounting standards. That is, the ratio of (total liabilities + owner’s equity)/total assets is greater than 1.2 or less than 0.8. (3) Correct the samples with biased patent data. Specifically, if the total patent applications for a firm in a given year do not equal the sum of invention and non-invention patent applications, sequential matching is performed using the firm name, address, and data from the National Intellectual Property Administration to correct discrepancies. For samples with missing patent application data, the same matching strategy is applied. If no patent application information is found for the firm in that year, the missing value is replaced with zero. (4) Samples with missing or negative values for key indicators such as total assets, R&D investment, tax incentives, subsidies, and total number of employees were excluded. Data with unreasonable entries, such as firms reporting establishment years later than the observed year, were excluded (only 21 observations were deleted). (5) Delete samples with only one period of data survival to meet panel estimation needs. (6) Deflate all nominal variables with the 2004 Beijing PPI data. (7) The main continuous variables are winsorized at the 5% level. The final result is 134,629 firm-year unbalanced panel observations. In the regression analysis, the four-digit industry codes from the National Economic Industry Classification (GB/T4754-2002) range from 1921 to 8790. Among the sample, 123,776 firms, or 91.92%, have fewer than 200 employees, while 2123 firms are listed. This indicates that the sample is predominantly composed of small and medium-sized non-listed enterprises, providing a more accurate reflection of entrepreneurial firms.

Estimation methods

The fixed effects model effectively controls for unobserved individual heterogeneity, reducing endogeneity caused by omitted variables. However, it assumes that individual effects remain constant over time, which may not always hold. To address this concern, we perform a Hausman test to determine whether the random effects model or the fixed effects model is more appropriate. The test result (p-value = 0.0000) rejects the random-effects specification, confirming that the fixed-effects model is the appropriate and consistent choice for our data. Therefore, we adopt the fixed-effects econometric model in our analysis.

$${{\text{ln}}\,patent}_{ift}={\alpha }_{0}+{\alpha }_{1}{wnewstartr}_{it}+\varphi {X}_{ift}+{\lambda }_{f}{+\mu }_{i}+{\varpi }_{t}+{\varepsilon }_{ift}$$
(1)

Drawing on Baron and Kenny (1986), we develop Eqs. (2) and (3), complementing Eq. (1), to rigorously assess the mediating roles of human capital and R&D investment.

$${Mediator}_{ift}={\beta }_{0}+{\beta }_{1}{wnewstartr}_{it}+\varpi {X}_{ift}+{\lambda }_{f}{+\mu }_{i}+{\varpi }_{t}+{\varepsilon }_{ift}$$
(2)
$${{\text{ln}}\,patent}_{ift}={c}_{0}+{{c}_{1}Mediator}_{ift}+{c}_{2}{wnewstartr}_{it}+\kappa {X}_{ift}+{\lambda }_{f}+{\mu }_{i}+{\varpi }_{t}+{\varepsilon }_{ift}$$
(3)

where lnpatentift is the innovation of firm f in industry i in year t, wnewstartrit is the high-tech entrepreneurship of industry i in year t, Mediatorift is the mediator variable, including R&D investment lnrdexpendift and human capital humanift, and Xift is the matrix of control variables. λf is the firm fixed effect, μi is the four-digit code industry fixed effect, \({\varpi }_{t}\) is the year fixed effect, and \({\varepsilon }_{{\rm{i}}ft}\) is the random error term. Equation (1) is used to test the impact of high-tech entrepreneurship on the innovation of incumbent firms within the regional innovation ecosystem. Equation (2) is used to test the effects of high-tech entrepreneurship on the mediator variables of R&D investment and human capital. Equation (3) adds mediator variables based on Eq. (1) and combines them with Eq. (2) to test the mediating roles of R&D investment and human capital in the relationship between high-tech entrepreneurship and the innovation of incumbent firms. Incumbent firms are firms whose age is greater than 1 in the current year.

Variables

High-tech entrepreneurship (wnewstartr). Entrepreneurship is commonly measured in two ways. One relies on survey data, such as whether firms engage in entrepreneurial activities (Galindo and Méndez, 2014). The other uses entry-based measures, including the number or proportion of new entrants (Chung et al., 2022). This study focuses on high-tech entrepreneurship, defined as entrepreneurial activities centered on technological innovation. Given that ZSP is a major hub for high-tech firms, most new ventures in the area are high-tech startups. We therefore adopt the entry-based approach. Specifically, we measure high-tech entrepreneurship using the proportion of high-tech startups in four-digit industries within ZSP (wnewstartr) and the natural logarithm of the number of high-tech startups (lnnew). Measuring the number of high-tech entrepreneurs at the industry level naturally weakens the functional dependence relationship between high-tech entrepreneurship and firm innovation. Additionally, to ensure robustness, we first redefine the age of startups, categorizing them as firms established within 2, 3, 4, or 5 years. We then calculate their industry entry rates or the natural logarithm of total entries as proxies for high-tech entrepreneurship.

Mediating variables

R&D investment (lnrdexpend) and human capital (human). R&D investment is measured by the natural logarithm of the firm’s actual R&D investment in the year plus one. Human capital is measured as the logarithm of the number of employees holding master’s or doctoral degrees.

Controls

First, we control for firm losses, firm size, and firm age, which have been found to influence firm innovation (Bachmann et al., 2021). In particular, firm loss (loss) is a binary variable assigned a value of 0 if firm profit is greater than 0, and vice versa. Firm size (lnasset) is measured as the logarithm of the firm’s real assets, and firm age (lnage) is measured as the natural logarithm of the firm age. Second, we control for firm capital structure (lev). Iqbal et al. (2020) argue that capital structure predicts firm innovation. We use the debt-to-equity ratio to measure a company’s capital structure. Third, resource-based theory suggests that heterogeneous resources are the key to a firm’s core competitive advantage. Under China’s policy framework, firms must meet stringent R&D requirements to be recognized as high-tech enterprises, implying that such firms possess heterogeneous knowledge resources. Moreover, once a company obtains high-tech recognition, it will receive a 15% tax deduction, and only if it continues to meet the R&D criteria can it win the next round of high-tech enterprise recognition. Therefore, high-tech recognition is expected to predict enterprise innovation. We use a 0-1 discrete variable to measure high-tech enterprise recognition (hightech1), which is assigned a value of 1 if the enterprise obtains high-tech recognition, and 0 vice versa. Fourth, the impact of firm subsidies (lnsubsidy) on firm innovation is supported by much of the literature (Marino et al., 2016; Gao et al., 2021). We measure firm subsidies by the logarithmic value of the actual firm subsidy plus one. Fifth, tax relief (lntaxrelief), as an important innovation support policy, has a significant impact on firm innovation (Akcigit et al., 2016). We measure corporate tax relief by the logarithmic value of the actual tax relief received by the firm. Sixth, in the era of the knowledge economy, open innovation is a key mechanism for enterprises to enhance their innovation performance (Chesbrough, 2003; Ardito et al., 2020). Wadhwa et al. (2017) argue that “Extramural R&D involves creative work related to product development performed by another entity for a fee, and excludes expenditure on the acquisition of non-R&D related external knowledge or equipment” (p.880). Therefore, we adopt a 0-1 binary variable to measure open innovation (Coinno) based on whether a company has expenditure towards research institutions or universities. Seventh, we control for the effect of competition on firm innovation. Drawing on Bessonova and Gonchar (2019), we measure competition (Hhi) using the HHI index for the 2-digit code industry.

Table 1 presents the descriptive statistics of the main variables in this study.

Table 1 Descriptive statistics.
Full size table

Results

Result analysis

Table 2 reports the impact of high-tech entrepreneurship on the innovation of incumbent firms within the regional innovation ecosystem. Columns (1)-(2) measure high-tech entrepreneurship using the proportion of new entrants in the four-digit industry codes (wnewstartr), while columns (3)-(4) measure high-tech entrepreneurship using the logarithm value of the number of new entrants in the four-digit industry codes (lnnew). Columns (1) and (3) do not include control variables, while columns (2) and (4) control for these variables. The results show that the regression coefficients of high-tech entrepreneurship are significantly positive at the 1% level of significance, indicating that high-tech entrepreneurship significantly promotes the innovation of incumbent firms within the regional innovation ecosystem, providing support for hypothesis H1. Furthermore, this positive effect remains unchanged when switching high-tech entrepreneurship measures and adding control variables, demonstrating the robustness of hypothesis H1. According to column (2), with the average proportion of high-tech entrepreneurial firms at approximately 2.3%, their presence increases the patent applications of incumbent firms by an average of 0.01%. According to column (4), for every 1% increase in the number of high-tech entrepreneurship firms in the four-digit industry codes, the innovation of incumbent firms improves by 1.1%. This indicates that the entry of high-tech entrepreneurial firms proportionally stimulates innovation among incumbent enterprises. It is important to note that variations in the impact on incumbent innovation arise from differences in the metrics used to measure high-tech entrepreneurship. Overall, sustaining the vitality of innovation ecosystems requires policies and institutional arrangements that attract high-tech entrepreneurial firms. However, because entrepreneurial entry depends on the stock of incumbents, the innovation-enhancing effect may weaken as incumbent density increases. This underscores the importance of well-designed exit mechanisms to maintain balanced entry dynamics and sustain ecosystem vitality.

Table 2 Basic regression results.
Full size table

Moreover, by drawing on data from ZSP and employing a fixed-effects model, we naturally account for time-invariant unobserved heterogeneity, such as initial resource endowments and inherent cultural factors. This strategy enables a cleaner identification of the causal relationship between high-tech entrepreneurship and innovation in incumbent firms. When translating these findings to other regional innovation ecosystems, however, policymakers must carefully assess the commonalities with the Zhongguancun context. A nuanced and precise policy design must account for local resource and cultural factors, which are precisely the influences that our model differs from.

Robustness Test

Alternative Measures of High-Tech Entrepreneurship and Enterprise Innovation

To ensure robustness, we employed three alternative approaches to measurement. First, we redefined high-tech entrepreneurship by using the absolute number of new entrants in the four-digit industry classification (wnewstart) as a proxy variable. Second, we adjusted the measurement of innovation by using the number of granted patents (lnpatent2). Third, we relaxed the age restrictions for defining high-tech entrepreneurial firms, expanding the criteria to include firms established within 2, 3, 4, or 5 years, and recalculated both the proportion of high-tech entrants (newstartr_2, newstartr_3, newstartr_4, newstartr_5) and the natural logarithm of their total numbers (lnnew_2year, lnnew_3year, lnnew_4year, lnnew_5year).

The regression results in Table 3 confirm that H1 remains valid even after adjusting the measures for high-tech entrepreneurship and firm innovation. Notably, the results in Panel B reveal that the positive impact of high-tech entrepreneurship on incumbent firm innovation is sensitive to the age threshold used to define entrepreneurial firms. When firms established within five years are included, the positive effect of high-tech entrepreneurship on incumbent enterprise innovation fails to pass statistical significance. Moreover, the marginal impact of high-tech entrepreneurship on incumbent firm innovation gradually diminishes as the age threshold for defining entrepreneurial firms increases. Furthermore, when high-tech entrepreneurship is measured as the natural logarithm of the total number of entrepreneurial firms, its positive influence on incumbent firm innovation is only evident when entrepreneurial firms are defined as those established within one or 2 years.

Table 3 Regression results of changing the measures of entrepreneurial ventures and corporate innovation.
Full size table

These findings suggest two key insights. First, younger high-tech entrepreneurial firms exert a stronger influence on incumbent firm innovation. Second, the impact of high-tech entrepreneurship on incumbent innovation within the regional innovation ecosystem is contingent upon the age threshold used to define entrepreneurial firms. And the proportion of high-tech entrepreneurship exerts a more sustained impact on incumbent firm innovation compared to its number.

Replacement of the model, inclusion of lagged terms, and exclusion of the effect of the Zhongguancun National Independent Innovation Demonstration Zone

Firstly, to address the zero-truncated nature of firm patent applications, this study employs Poisson fixed-effects regression for re-estimation. This method is well-suited for count data, ensuring that the discrete, non-negative nature of the dependent variable is properly accounted for while controlling for unobserved heterogeneity across firms. The dependent variable in this analysis is the raw count of firm patent applications. The results, presented in Table 4, column (1), align with the baseline regression results and confirm the robustness of H1.

Table 4 Regression results for replacing the model, adding the lag term, and excluding the effect of the National Independent Innovation Demonstration Zone.
Full size table

Secondly, considering the cumulative nature of firm innovation, we incorporate the lagged variable of firm innovation in the baseline regression to account for the potential cumulative effect that might interfere with the regression results. Column (2) of Table 4 shows that the lag of enterprise innovation significantly promotes enterprise innovation, characterizing the cumulative nature of enterprise innovation. This cumulative effect attenuates the direct impact of high-tech entrepreneurial entry on incumbent firms’ innovation. However, even after controlling for this factor, the promotion effect of high-tech entrepreneurship on incumbent firm innovation within the regional innovation ecosystem remains statistically significant, supporting the validity of H1.

Furthermore, to mitigate estimation bias caused by the reverse causality between high-tech entrepreneurship and innovation, we lag high-tech entrepreneurship by one period, and the results are reported in column (3) of Table 4. The findings indicate that even with a lagged measure of high-tech entrepreneurship, it still induces incumbent firm innovation, further supporting the robustness of H1.

Finally, the Zhongguancun National Independent Innovation Demonstration Zone (ZNIID) was established in 2009. The purpose of this policy is to promote enterprise innovation in ZSP. The policy trial period highly overlaps with the research period of this article, and failing to exclude the interference of this policy may lead to erroneous estimations. To account for the potential influence of this policy and considering the heightened impact of the establishment of the ZNIID on industries with high technological innovation intensity, we follow existing literature by incorporating an additional control variable (post×w1) into the baseline model (1). Here, “post” is a time dummy variable, assigned a value of 0 before 2009 and 1 from 2009 onwards. “w1” represents the innovation intensity of four-digit industries before 2009. It is calculated as the ratio of the annual average total number of invention patent authorizations in the four-digit industries from 2005 to 2008 to the average number of employees in the same industries during the same period. This is done to control for the impact of the establishment of the ZNIID. The results can be seen in columns (4) to (5) of Table 4. The results show that even after accounting for the impact of the establishment of the Zhongguancun National Independent Innovation Demonstration Zone, high-tech entrepreneurship continues to stimulate enterprise innovation, with the magnitude of its effect remaining largely unchanged.

Endogeneity test

Although we have accounted for individual, year, and industry fixed effects and introduced a one-period lag for the independent variable of high-tech entrepreneurship to mitigate potential omitted variable bias and address endogeneity arising from reverse causality, concerns about potential estimation bias related to the measurement of high-tech entrepreneurship persist. To address this concern and further overcome endogeneity issues, this study uses the average net increase of enterprises in the two-digit industry as an instrumental variable for technology entrepreneurship. On one hand, the average net increase of enterprises in the two-digit industry maps the attractiveness of the industry and is related to the quantity of technology entrepreneurship in the four-digit industry. On the other hand, the average net increase of enterprises in the two-digit industry does not directly affect individual firm innovation, thus possessing exogeneity and relevance, making it a suitable instrumental variable.

As shown in Table 5, after rejecting both the null hypothesis of under-identification (Kleibergen-Paap rk LM statistic, p = 0.0000) and the weak instrument hypothesis (Wald F > 16.38), high-tech entrepreneurship is found to exert a positive spillover effect on incumbent firm innovation, consistent with the earlier findings.

Table 5 Regression results of instrumental variables.
Full size table

Expanded analysis

The impact of high-tech entrepreneurship on incumbent firm innovation efficiency in the regional innovation ecosystem

After confirming that high-tech entrepreneurship promotes incumbent firm innovation in the regional innovation ecosystem, we further examine the impact of high-tech entrepreneurship on the innovation efficiency of incumbent firms in the regional innovation ecosystem. To measure innovation efficiency, we employ the ACF method as it can mitigate simultaneity bias (Ackerberg et al., 2015). Specifically, this paper uses the natural logarithm of patents as the desired output, the natural logarithm of R&D personnel as the free input variable, the natural logarithm of actual R&D investment stock as the state variable, and the natural logarithm of actual R&D investment as the proxy variable, and then uses the ACF method to measure the innovation efficiency of enterprises.

The estimation results are shown in column (1) of Table 6. From the results, high-tech entrepreneurship enhances the innovation efficiency of incumbent firms in the regional innovation ecosystem.

Table 6 Regression results of the extended analysis.
Full size table

Can high-tech entrepreneurship promote innovation for all firms in the regional innovation ecosystem?

If high-tech entrepreneurship promotes incumbent firm innovation in the regional innovation ecosystem, can it also promote innovation for all firms? To address this question, we conduct further tests. The dependent variable in the sample is replaced with innovation for all firms, and regression analysis is performed. The results are presented in columns (2) to (5) of Table 6. From the results, it can be seen that high-tech entrepreneurship promotes innovation for firms within the regional innovation ecosystem, and this result holds even when the independent variables are changed. This implies that high-tech entrepreneurship plays an important role in fostering innovation in the regional innovation ecosystem.

Mechanism testing and heterogeneity analysis

Mechanism testing

Theoretical mechanisms suggest that high-tech entrepreneurship promotes innovation in incumbent firms within the regional innovation ecosystem through the R&D investment and human capital channels. This will be tested below.

Columns (1)-(3) of Table 7 report the regression results of the mediating mechanism of R&D investment. From the results, based on the sequential test, high-tech entrepreneurship significantly increases the R&D investment of incumbent firms within the regional innovation ecosystem (column (2)), and firm R&D investment significantly promotes firm innovation (column (3)), which, combined with columns (1)-(3), suggests that the mediating mechanism of R&D investment is significant. Hypothesis H2a holds. Columns (1), (4), and (5) of Table 7 report the regression results of the mediating effect of human capital. The results show that the mediation mechanism of human capital is significant according to the sequential test. Hypothesis H2b holds.

Table 7 Regression results of the mechanism test.
Full size table

Heterogeneity analysis

Electronic information and non-electronic information industries

The electronic information industry undergoes faster technological iterations compared to non-electronic industries. Moreover, its modular architecture further amplifies the need for specialized and complementary technological contributors to achieve effective value co-creation. Companies in the electronic information industry require a richer coopetition network and new knowledge to adapt to the rapid technological changes and acquire new resources and information. The entry of high-tech entrepreneurship enhances the density of the coopetition network in the regional innovation ecosystem. It also spills over new knowledge and information through knowledge filtering mechanisms, providing greater support for innovation in the electronic information industry. As a result, the effect of high-tech entrepreneurship on incumbent firm innovation within the electronic information industry is more significant.

Based on the classification of enterprise technology fields according to the “Statistical Reporting System for National High-tech Zones Enterprises and High-tech Enterprises” and the “Classification Catalog for Electronic Information Industry”, this study matches the four-digit industry codes to classify industries as either the electronic information industry or the non-electronic information industry. The electronic information industry includes industries such as electronic computers, computer peripheral equipment, information processing equipment, computer network products, computer software products, microelectronics, electronic components, optoelectronic components and their products, broadcasting and television equipment, communication equipment, and other subcategories of electronic information technology. All other industries are classified as non-electronic information industries. Then, through group regression analysis, the impact of technology entrepreneurship on innovation in incumbent firms in different industries is examined. The results are reported in columns (1) and (2) of Table 8.

Table 8 Regression results for industry and research risk stage heterogeneity.
Full size table

Consistent with expectations, high-tech entrepreneurship has a greater inducing effect on innovation in incumbent firms within the electronic information industry. This indicates that the innovation spillover effects obtained through high-tech entrepreneurship are asymmetric between the electronic information industry and the non-electronic information industry. Implementing policies to promote high-tech entrepreneurship is an important measure to accelerate innovation in the electronic information industry.

Research stage and development stage

Enterprise R&D activities can be categorized into two sub-stages, research and development (Czarnitzki et al., 2011; Guan and Pang, 2017), and there are large differences in the R&D risks faced by enterprises in the two stages. Overall, research activities face more technological and market uncertainties and are riskier, whereas firms in the development stage can develop exploitatively on existing technology paths and face relatively lower risks (Czarnitzki et al., 2011).

This raises a question: if high-tech entrepreneurship can increase incumbent firms’ R&D investment and promote innovation, what are the differences in the effects of high-tech entrepreneurship on incumbent firm innovation at different stages of R&D risk? Czarnitzki et al. (2011) have pointed out that the development stage is more often expressed in the form of patents and intellectual property rights, which means that we can distinguish whether a company is in the research or development stage based on the relative change in the number of patents filed by the company. Consequently, we characterize different R&D stages of firms based on their number of patent applications. Specifically, if a firm’s number of patent applications is greater than or equal to the industry average, it signifies that the firm has strong research capabilities and lower risks, positioning it in the low-risk development stage. Conversely, if a firm’s number of patent applications is less than the industry average, it signifies that the firm is in the high-risk exploration stage, with lower R&D output and higher R&D risks, placing it in the high-risk research stage. Based on this characterization, we divide the sample into research and development stages and re-estimate using the grouping regression method, as shown in columns (3)-(4) of Table 8. The results show that high-tech entrepreneurship can simultaneously promote innovation in both development and research stages of incumbent firms. However, in terms of marginal contribution, the marginal impact of high-tech entrepreneurship on incumbent firm innovation is greater in the less risky development stage.

A possible explanation is that, in the exploration stage of research, the resolution of scientific problems and the exploration of technological paths depend on creativity and inspiration. Despite the availability of ample innovation resources, these issues cannot be immediately resolved solely through the strengthening of innovation resources and innovation networks. In contrast, in the utilization stage of development, where the dominant technological path has been essentially determined, abundant innovation resources and networks can provide rich technical experimental scenarios and market opportunities for the dominant technological path, enabling its innovation to undergo substantial development.

Research conclusions and discussion

Conclusions

This study examines the impact of high-tech entrepreneurship within the innovation ecosystem on incumbent innovation, grounded in the theory of the innovation ecosystem and paradox. Using firm-level data from China’s ZSP from 2005 to 2015, we provide robust evidence that high-tech entrepreneurship significantly fosters innovation among incumbent firms, with R&D investment and human capital as critical mediators. The effect is particularly greater in the electronic information industry compared to non-electronic sectors and is stronger for firms in the development stage than in the research stage.

Contributions

The paper offers several key contributions. First, our findings not only enrich research on the relationship between entrepreneurship and innovation but also advance both innovation ecosystem theory and paradox theory. Our findings that entrepreneurship promotes innovation among incumbent firms align with prior research on the entrepreneurship-innovation nexus (Hall et al., 2012; Galindo and Méndez, 2014; Paik et al., 2019; Boone et al., 2019; Ejdemo and Örtqvist, 2020; Chung et al., 2022; Si et al., 2022; Estrin et al., 2022; Dong et al., 2023). However, unlike existing research, we situate our investigation explicitly within the context of regional innovation ecosystems and, by integrating innovation ecosystem theory with paradox theory, reconceptualize the functional roles of high-tech entrepreneurial firms from a paradoxical perspective. Specifically, we characterize these firms as ecosystem adapters, ecosystem disruptors, coopetitors, adaptive innovators, and disruptive innovators. By examining the relationship between high-tech entrepreneurship and incumbent innovation through these multiple, inherently contradictory roles, we move beyond the conventional assumption that high-tech entrepreneurial firms perform a single function within innovation ecosystems. More importantly, we demonstrate that the relationship between high-tech entrepreneurial firms and incumbents is not one of simple substitution or complementarity; rather, it represents a paradoxical symbiosis. This suggests that effective ecosystem governance does not entail eliminating tensions, but instead hinges on deliberately constructing, maintaining, and leveraging such paradoxical tensions through sophisticated institutional design. Doing so can guide the entire ecosystem toward a more advanced state of dynamic equilibrium and prosperity. Thus, our study not only provides new insights for innovation-ecosystem governance but also responds to Schad et al.’s (2017) call to advance research on how contradictions can be harnessed to promote the evolution of complex systems, thereby extending both innovation ecosystem theory and paradox theory.

Second, our findings contribute to the emerging literature on entrepreneurship, innovation, and ecosystem (Eckhardt et al., 2018; Nambisan et al., 2018; Catala et al., 2023) by providing a holistic theoretical framework on entrepreneurship, innovation, and ecosystem. Existing literature has emphasized the importance of linking entrepreneurship with ecosystems and innovation with ecosystems separately. Recently, some literature has even begun to explore the relationship between entrepreneurship, innovation, and ecosystems. Nambisan et al. (2018) explore how and when open innovation and platforms will promote or inhibit entrepreneurship and their specific mechanisms. Unlike Nambisan et al. (2018), however, which focuses on the impact that opportunities within the innovation ecosystem have on entrepreneurship, our study focuses on the impact of high-tech entrepreneurship on innovation within the innovation ecosystem. Our findings indicate that attracting high-tech enterprises into the regional innovation ecosystem is beneficial for the innovation and development of the entire ecosystem. In other words, the innovative driving force and vitality of the regional innovation ecosystem depend on the institutional design that can attract high-tech entrepreneurship.

Third, our findings also demonstrate the importance of contextual conditions—both industrial and developmental—in shaping the effect of entrepreneurship on incumbent innovation. The electronic information industry and high-tech entrepreneurship share characteristics of high knowledge intensity and rapid technological iteration, making the promotional effect of high-tech entrepreneurship particularly pronounced in this setting. This is consistent with observations from regions such as Silicon Valley and Bangalore, where dense entrepreneurial activity fuels continuous innovation. Moreover, the research stage relies more on inspiration and creativity rather than innovation resources compared to the development stage. Although the new innovation resources brought by high-tech entrepreneurship can stimulate new inspiration through interaction with the innovation resources of incumbent enterprises, the generation of inspiration still has randomness and irregularity. Thus, high-tech entrepreneurship contributes to the innovation of incumbent firms in both the research and development phases, but is more likely to contribute to the innovation of incumbent firms in the development phase. Our research also confirms this view. Our study thus offers new insights into how contextual differences condition the entrepreneurship–innovation–ecosystem relationship.

Policy and management implications

The conclusions of this study are derived from a specific regional innovation ecosystem, yet the policy and managerial implications carry broad generalizability. First, our study confirms that high-tech entrepreneurial entry in regional innovation ecosystems enhances both incumbent innovation and overall ecosystem innovation. Thus, we suggest that the government should strengthen the emphasis on the entrepreneurship cultivation function in the regional innovation ecosystem, or superimpose the design of the regional innovation ecosystem with the regional entrepreneurship ecosystem, so as to form a development logic in which innovation and entrepreneurship stimulate and promote each other. In particular, the government or leading enterprises within the regional innovation ecosystem can act as architects of the entrepreneurship ecosystem. They can attract high-tech entrepreneurs to enter the ecosystem by establishing digital commons, high-tech entrepreneurial hubs, and hard-tech incubators to attract high-tech entrepreneurs into the ecosystem with open entrepreneurial opportunities. Second, our findings indicate that high-tech entrepreneurial promotes incumbent innovation largely by helping incumbents overcome path dependency. Since path dependency is common in mature regional ecosystems and is a major obstacle to their sustainable development. Therefore, for regional innovation ecosystems in advanced economies—where mature markets and powerful incumbents prevail—the governance priority is preventing structural rigidity. Ecosystem governors should design “open system interfaces,” such as interoperable data standards and open-source protocols, while competition authorities should mandate or incentivize large incumbents to open their platforms. Such openness enables entrepreneurial firms to embed seamlessly into incumbent-dominated value networks and helps reshape population relationships within the ecosystem. Third, regional innovation ecosystem governance must reflect industry-specific architectural characteristics. Our findings show that the innovation-enhancing effect of entrepreneurial entry is more pronounced in modular industries such as electronics and information technology. Accordingly, ecosystems anchored in modular industries, such as the digital industry, should emphasize interface standardization and open innovation platforms to lower integration costs, facilitate incumbents’ engagement with specialized startups, and amplify the stage-dependent benefits of entrepreneurial entry.

Limitations and future research

Additionally, there are several limitations in this study that need to be addressed in future research. First, our study did not analyze the impact of high-tech entrepreneurship on the innovation ambidexterity of incumbent firms within the regional innovation ecosystem. Guerrero (2021) calls out ambidexterity as an important missing link in entrepreneurship and innovation research. In the future, the impact of high-tech entrepreneurship on the innovation ambidexterity of incumbent firms within the regional innovation ecosystem can be further explored by categorizing innovation into exploitative and exploratory innovation. Second, our study does not analyze the institutional boundaries through which high-tech entrepreneurship affects the innovation of incumbent firms within the regional innovation ecosystem. Nambisan et al. (2018) state that “Successful enactment of entrepreneurial opportunities may be contingent on a broader set of institutional and infrastructural arrangements” (p.363). Linder et al. (2022) also point out that institutions are crucial for entrepreneurship. Different institutional components may lead to different entrepreneurship, innovation, and ecological relationships. Further research could focus on this dimension in the future.

Third, due to data availability, our analysis is based on data from 2005 to 2015. Although this time window offers an important historical baseline and conceptual reference for subsequent research, the study does not fully elucidate how the “data-sharing versus data-monopolization” paradox within regional innovation ecosystems in the era of artificial intelligence influences the innovation dynamics between high-tech entrepreneurial firms and incumbent firms. Future research could conduct a phase comparison analysis once data for the later period becomes available, allowing for a more comprehensive understanding of these dynamics over time. Fourth, this study relies solely on micro-level data from the ZSP. Consequently, the findings do not capture how factors such as regional resource endowments, cultural contexts, and other place-specific characteristics may influence the innovation relationship between high-tech entrepreneurial firms and incumbents. These factors are essential for designing precise and context-specific regional innovation ecosystem policies. Future research could extend the analysis by examining how industrial resource endowments, cultural attributes, and policy differences across various regional innovation ecosystems shape the dynamics between entrepreneurial and incumbent firms.