Technological Innovation and Industrial Upgrading in China’s Automotive Industry: The Dual Mechanisms of Global Value Chain Mediation and Industrial Agglomeration Moderation

Abstract

Amid global economic transformation, technological innovation is widely recognized as a pivotal driver for the sustainable upgrading of the automotive industry. This is particularly critical for China, the world’s largest automotive market, which faces persistent challenges, including technological bottlenecks in core components and confinement to low-value segments within global value chain. This study introduces novelty by systematically integrating and empirically testing the mediating role of Global Value Chain (GVC) and the moderating effect of industrial agglomeration within a unified framework—a focus that remains underexplored in the sector. Using panel data from 28 Chinese provinces (2000–2020), we measure industrial upgrading using the DEA-Malmquist index to capture total factor productivity changes and employ a system GMM approach. The results indicate that technological innovation directly fosters industrial upgrading and indirectly facilitates it by improving the industry’s GVC position. Moreover, industrial agglomeration positively moderates this relationship, thereby amplifying the returns to innovation within geographic clusters. These findings highlight the necessity for integrated policies that simultaneously promote technological innovation, strategic GVC positioning, and synergistic industrial agglomeration to achieve sustainable upgrading. As a paradigmatic emerging economy, China’s experience offers valuable insights for other latecomer economies pursuing industrial upgrading.

1. Introduction

With the profound transformation of the global economy and the rapid advancement of the technological revolution, industrial upgrading has become a cornerstone for achieving sustainable economic development and enhancing national competitiveness globally. Industrial upgrading entails not only increasing the value-added and technological content of products but also securing a position within higher-value segments of the global value chain (GVC) to achieve simultaneous improvements in sustainability and value chain status [1,2].

China’s automotive industry is currently at a pivotal juncture in its transformation and upgrading. Despite having evolved into the world’s leading manufacturer and seller of automobiles over the past two decades, the industry faces formidable challenges under the dual pressures of pursuing “dual carbon” goals and transitioning to new energy sources. It continues to grapple with issues such as a dependence on imported core technologies, which stifles domestic innovation, and confinement to low-end segments of the value chain [3,4]. These challenges severely restrict the technological advancements and competitiveness necessary for sustainable development within China’s automotive industry.

Amid the global new energy revolution and intelligent transformation, technological innovation has emerged as the primary impetus behind industrial upgrading, facilitating the shift from low-value production to high-value innovation that aligns with sustainable practices [5]. However, existing research has yet to provide a systematic explanation of the internal processes by which technological innovation affects industrial upgrading in China’s automotive industry, particularly regarding the roles of GVC and industrial agglomeration.

This study specifically focuses on the roles of Global Value Chain (GVC) and industrial agglomeration for two compelling reasons. First, the GVC lens is critical because the upgrading trajectory of China’s automotive industry is inextricably linked to its position and strategies within global production networks [6]. Technological innovation alone is insufficient; it is the interaction with global knowledge flows, standards, and lead firms that determines how effectively innovation translates into upward mobility within the GVC [7]. Second, industrial agglomeration is selected as a key contextual factor because the geographical concentration of automotive OEMs, suppliers, and R&D centers in specific regions (e.g., the Yangtze River Delta) creates synergistic environments that profoundly shape innovation outcomes through intense knowledge spillovers, specialized labor markets, and shared infrastructure—a category where advanced digital infrastructure has been shown to enhance economic resilience significantly [8,9]. While other factors (e.g., government policy, FDI, and financial development) are undoubtedly important [10,11,12], the combined influence of GVC integration and industrial agglomeration represents a fundamental yet underexplored perspective. This perspective is critical because these two factors form the core structural features of China’s automotive industry, defining the global and local contexts within which innovation and upgrading occur.

GVC serves as a crucial pathway for developing countries, such as China, to enhance their technological capabilities and promote sustainable industrial practices [13]. Nonetheless, a pressing issue remains: how to further improve the value of core segments in the industrial chain through technological innovation, optimize GVC positioning, and achieve sustainable industrial upgrading. Moreover, the effects of industrial agglomeration observed in regional economic development are believed to support technological innovation and industrial upgrading by promoting resource sharing and technology spillovers, which are critical for sustainable growth [14,15]. However, the extent to which industrial agglomeration amplifies the influence of technological innovation on industrial upgrading—specifically, its potential moderating effect—requires further exploration.

Existing research on industrial upgrading primarily emphasizes the direct effects of technological innovation on economic growth and industrial upgrading. However, studies on how technological innovation shapes industrial upgrading via the mediating mechanism of GVCs remain relatively scarce. Furthermore, the moderating effects of industrial agglomeration, a significant characteristic of regional economies, on the relationship between technological innovation and industrial upgrading have been largely underexplored in current scholarship. To fill these gaps, this research seeks to investigate several core questions: Can technological innovation directly promote industrial upgrading in China’s automotive industry? Does technological innovation indirectly facilitate industrial upgrading in China’s automotive industry through the mediating mechanism of the GVC? Additionally, does industrial agglomeration exert a moderating effect on the pathway through which technological innovation influences industrial upgrading in this sector?

This research makes several key contributions to the literature. First, it provides a novel theoretical integration by proposing a unified framework that simultaneously examines the mediating role of Global Value Chain (GVC) positioning and the moderating role of industrial agglomeration, thereby addressing a critical gap in the existing literature, which has predominantly examined these factors in isolation. Second, it elucidates the dual mechanisms through which technological innovation operates, moving beyond merely establishing a direct correlation to empirically demonstrate that GVC positioning acts as a significant mediating pathway while industrial agglomeration serves as a critical contextual amplifier. Thus, it reveals not only that innovation drives upgrading but also the specific ways and contexts under which it does so. Third, it delivers nuanced, sector-specific empirical evidence by applying this framework to China’s automotive industry, offering one of the first quantitative assessments of these complex mechanisms within a strategically vital sector undergoing a profound technological transition, and yielding concrete, evidence-based insights for policymakers and firm managers.

Our empirical analysis provides robust support for this framework. The findings confirm that technological innovation is a direct driver of industrial upgrading. Moreover, we identify a significant mediating effect, revealing that innovation also operates indirectly by elevating the industry’s position in the Global Value Chain. Crucially, we demonstrate that industrial agglomeration acts as a positive moderator, which substantially amplifies the impact of innovation within geographic clusters. These insights underscore the necessity of a synergistic strategy that integrates efforts to strengthen technological innovation, enhance GVC positioning, and promote industrial agglomeration —all to achieve high-quality, sustainable upgrading.

The structure of the subsequent sections is outlined as follows. Section 2 reviews the literature on technological innovation, global value chain, industrial agglomeration, and industrial upgrading, and further develops a conceptual framework and research hypotheses. Section 3 details data description, variable measurement methods, and the model specification. Section 4 presents the empirical findings alongside corresponding analysis, while Section 5 concludes by offering policy implications, acknowledging the research limitations, and proposing possible directions for future investigations.

2. Literature Review

To structure our review and clearly identify the research trajectory, we classify the extensive literature into four interconnected thematic streams: (1) the concept of industrial upgrading; (2) the direct relationship between technological innovation and industrial upgrading; (3) the mediating role of Global Value Chains; and (4) the moderating role of Industrial Agglomeration. This classification enables us to systematically assess the contributions and limitations of each stream systematically. Following this thematic review, we provide a synthesizing discussion to precisely position the integrated contribution of our study by drawing contrasts with prior research.

2.1. Industrial Upgrading

Industrial upgrading refers to the process through which countries, companies, and workers transition from low-value economic activities to those generating higher value within global production systems [2]. This transformation is achieved by augmenting the value-added content of industrial output via improvements in production factors, gains in operational efficiency, enhancements in product quality, and restructuring industrial supply chains [16].

Within the GVC framework, industrial upgrading can be categorized into four types: process upgrading, which involves improving production efficiency and reducing costs through new technologies or organizational practices; product upgrading, characterized by enhancing product quality and technological content to transition from low-margin goods to those that command premium value; functional upgrading, marked by expanding roles in the value chain from simple tasks to more complex activities such as design and marketing; and chain upgrading, which entails shifting from lower-value to higher-value chains through diversification or entry into new industries [17].

However, while this influential typology offers a valuable heuristic for categorization, it has not been immune to critical scrutiny. A principal limitation lies in its implicit assumption of a linear and progressive trajectory from low- to high-value activities. In practice, such progression is neither automatic nor guaranteed. Firms and regions, particularly in developing economies, frequently encounter a “functional upgrading trap” or confront deliberate “barriers to entry” set up by lead firms in GVCs to protect their core, high-value functions [18]. This critique underscores that the pathway to upgrading is inherently contested and shaped by power dynamics within value chains—a critical dimension that the classical typology tends to underemphasize.

Furthermore, recent research has highlighted the critical role of technological advancement and innovation capabilities in driving industrial upgrading [5]. This viewpoint emphasizes that industrial upgrading involves not only structural adjustments but also a comprehensive enhancement of technological content, innovation capabilities, international competitiveness, and sustainable economic development [19].

A second stream of critique focuses on the evolving dimensions of upgrading itself. While traditional and GVC-centric measures prioritize economic value capture, a growing body of literature argues for the integration of social and environmental dimensions into the upgrading concept—often termed “social upgrading” (improving wages and working conditions) and “green upgrading” (adopting environmentally sustainable processes) [20]. The prevailing focus on economic and technological metrics, as seen in much of the literature, risks presenting an incomplete picture of sustainable industrial development. This study acknowledges this broader conceptual debate but maintains an economic-technological focus to allow for a deep, empirical examination of the core drivers within the automotive industry context.

In summary, industrial upgrading is not merely a straightforward process of structural adjustment; it represents a dynamic optimization of the value chain, a strengthening of technological capabilities, and alignment with sustainable development objectives. By improving resource allocation efficiency, expanding the most value-adding stages of the industrial chain, and strengthening technological and innovative capacities, economies can secure a more robust position in regional and global competition.

2.2. Technological Innovation and Industrial Upgrading

Technological innovation is generally regarded as the engine driving industrial upgrading, which can occur through various forms, including technological breakthroughs, market innovation, and full-industrial-chain innovation. Technological innovation acts as a key driver in the transformation and upgrading of industrial structures. Specifically, increased investment in innovation fosters technological progress, significantly promoting structural upgrades across industries and underpinning the shift toward sustainable development [21]. This body of literature, however, often relies on a macro-econometric paradigm that establishes valuable but broad correlations, treating innovation as a homogeneous input and upgrading as a uniform output.

Technological innovation is intrinsically linked to industrial upgrading, manifesting as a process that begins with new ideas and progresses through stages of research and development (R&D), experimentation, trial production, manufacturing, and initial commercialization. This innovation process is not linear; instead, it evolves through various stages characterized by technology-push and market-pull dynamics [22]. While this stage-based model is useful, it represents a simplification that can overlook the recursive, feedback-driven, and ecosystem-dependent characteristics of innovation in complex industries. Using interprovincial panel data from China (2008–2020), Xu et al. [23] found that technological innovation makes a significant contribution to provincial-level industrial upgrading. Such macro-level studies, exemplified by Xu et al. [23], are instrumental in confirming the general relationship at a national or regional scale. Nevertheless, their primary limitation lies in the methodological approach itself, which inherently masks the sector-specific mechanisms, contextual contingencies, and causal pathways that connect innovation to upgrading.

Furthermore, technological innovation drives the development of new production processes and technologies, thereby enhancing productivity. It also enables firms to diversify their products, enter new markets, and reduce dependence on single products or niche markets [24]. Industries gain competitive advantages through the continuous iteration of innovative technologies, which drive internal structural upgrades and foster long-term sustainable industrial development [25].

A critical synthesis of these findings reveals that the literature often presents a sanitized view of the innovation-upgrading nexus. It tends to emphasize the positive outcomes while overlooking the structural barriers and sectoral specificities that can impede or reshape this process. For instance, the persistent challenge of relying on imported core components in industries like automotive underscores a critical disconnect between national R&D investment and holistic upgrading outcomes—a nuance largely missed by the macro-level perspective.

These limitations necessitate a shift in analytical focus from the macro to the sectoral level. The mechanisms of innovation-driven upgrading in the automotive industry are distinct due to its exceptional capital and R&D intensity, complex global supply chains, and the unprecedented ongoing transition towards new energy and intelligent vehicles [26,27]. A significant limitation of the existing macro-level perspective is its inability to address the automotive industry’s persistent challenge of relying on imported core components (e.g., high-end chips, advanced battery management systems) and its unique pathway of upgrading through this technological transition. Therefore, moving from a generalized macro perspective to a targeted investigation within this strategically vital sector is necessary to uncover the specific pathways through which innovation translates into tangible upgrading outcomes.

The proposition that technological innovation drives industrial upgrading is expected to hold with particular force in the automotive industry due to the sector’s distinctive characteristics, including exceptionally high R&D expenditures, long product development cycles, and formidable technological barriers. These characteristics are especially pronounced during the ongoing transition to new energy and intelligent vehicles, making technological innovation a necessity for survival and competitiveness rather than a mere option. The causal pathway operates through several concrete mechanisms. Firstly, innovation drives process upgrading; for instance, advancements in production technology, such as modular platforms and gigacasting substantially boost production efficiency and reduce costs. Secondly, it facilitates product upgrading through breakthroughs in core technologies such as battery energy density and autonomous driving algorithms, which directly increase the value-added content of final products and support a shift into premium market segments. Collectively, these specific technological advancements directly contribute to the growth of total factor productivity (TFP) in the industry. Therefore, we hypothesize:

H1. 

Technological innovation has a positive and significant effect on industrial upgrading in China ’s automotive industry.

2.3. The Mediating Role of Global Value Chain

GVC framework offers a novel conceptual lens for examining how technological innovation influences industrial upgrading. It characterizes the integrated activities and processes involved in the manufacturing, circulation, and utilization of products and services within a global context [1].

Technological innovation has been shown to significantly enhance the upgrading of China’s manufacturing industry within GVC [13]. Activities associated with technological innovation can elevate an industry’s position within the GVC [28]. This perspective effectively reframes industrial upgrading from a domestic process to a repositioning within global networks. Countries can engage in GVC through backward linkages—which involve importing intermediate goods and services to integrate into upstream production—and forward linkages—which entail exporting intermediate or final products to participate in downstream value creation. A critical aspect of GVC upgrading is the transition from backward to forward integration, where domestic R&D investment plays a pivotal role in facilitating this transition [29]. Domestic R&D investments enable the shift toward forward GVC participation, indicating that domestic innovation efforts are essential for advancing up the value chain [30].

Furthermore, participating in GVC has emerged as a novel approach to achieving industrial upgrading. GVC enhances a country’s or industry’s position in terms of export value added, process upgrading, product upgrading, and skill upgrading, thereby driving industrial upgrading [31]. By facilitating a region’s movement toward the high-value center of GVC, GVC drives industrial upgrading [32]. Moreover, GVC holds significant potential for international knowledge and technology transfer, which has led to their recognition as drivers of productivity and growth in both developed and developing economies [30]. GVC also enables inter-firm knowledge spillovers across the industrial network, a process that fosters upgrading in the manufacturing sector [33]. Embedding in GVC drives manufacturing upgrading, boosts trade gains in capital-intensive industries, and aligns development with global sustainability imperatives [34].

However, a critical analysis of this stream of literature reveals a predominant focus on GVC participation as a direct driver of upgrading, often treating the value chain as a pre-structured arena. This perspective tends to under-theorize the antecedent role of technological innovation in proactively shaping a firm’s or region’s GVC trajectory and position. The literature establishes that being in a GVC can lead to upgrading, and that innovation helps, but it stops short of rigorously modeling and empirically testing GVC positioning as the key mediating mechanism that transmits the effect of innovation to upgrading outcomes.

This theoretical shortcoming is compounded when examined through a sectoral lens. A review of the sector-specific literature reveals a significant gap. This mediating role of GVC is particularly salient in the automotive industry, which is one of the most globally fragmented sectors. However, much of the existing GVC literature on the automotive industry focuses on governance modes or trade patterns [35,36], with limited empirical research quantitatively examining GVC positioning as a mediating pathway between technological innovation and holistic industrial upgrading, especially in the context of China’s dual goals of technological self-reliance and GVC integration.

The mediating role of the GVC is hypothesized to function through specific upgrading pathways that are salient in the globally fragmented automotive industry. We posit that technological innovation elevates a region’s position in the GVC through two tangible shifts. First, it enables product upgrading by developing competitive new energy vehicles (NEVs) or intelligent vehicle features, allowing regions to transition from producing standard components to exporting high-value-added final products. Second, it facilitates functional upgrading; through the mastery of core technologies such as battery packs and driving assistance systems, firms and regions can move beyond simple assembly to higher-value functions like R&D, design, and global supply chain management, thereby reducing reliance on imported core technologies. This enhanced GVC position, characterized by greater value capture, is in itself a direct manifestation and driver of industrial upgrading at the regional level. Consequently, we propose:

H2. 

GVC acts as an intermediary in the relationship between technological innovation and industrial upgrading in China’s automotive industry.

2.4. The Moderating Role of Industry Agglomeration

Industrial agglomeration refers to the phenomenon where firms cluster in a specific location, generating externalities such as economies of scale and localization externalities [37]. Marshall [38] argued that firms achieve the sharing of suppliers, labor pools, and information through agglomeration, which accelerates inter-firm knowledge spillovers and technological diffusion, thereby promoting regional technological innovation, industrial upgrading, and economic sustainability. These classical foundations, while foundational, were conceptualized in the context of 19th and early 20th-century industrial districts, which were largely self-contained and operated within national economies.

Industrial agglomeration generates significant innovation benefits [39]. Porter [40] suggests that firms increase their investments in technological innovation to enhance competitiveness and achieve excess profits. Such competitive agglomeration further stimulates more firms in the agglomeration area to continuously invest in innovation elements, such as funds, technical equipment, and R&D personnel, thereby elevating industry-wide innovation efficiency.

This body of work effectively establishes agglomeration as a catalyst for innovation. However, a critical limitation of this classical and neo-classical literature is its implicit treatment of agglomeration as a primarily market-driven phenomenon emerging organically from firm interactions, with less consideration for the role of strategic state intervention in shaping cluster formation and evolutionary dynamics.

Industrial agglomeration can bring external economies of scale, including knowledge spillovers, resource sharing, and market matching effects, which can enhance the effectiveness of technological innovation [41]. In highly agglomerated regions, firms are more likely to form innovation networks through close cooperation, knowledge transfer, and information exchange, thus significantly enhancing the positive impact of technological innovation on industrial upgrading [42]. Firms within agglomerated regions can generate inter-firm knowledge spillover effects through geographic proximity, accelerating the spread of technological advances and the dissemination of innovative results. As agglomeration intensifies, these technological externalities become more pronounced between neighboring firms, potentially leading to stronger marginal effects on technological innovation performance [14]. Within industrial agglomerations, firms engage in both cooperation and competition. This coopetitive relationship can stimulate more substantial innovation incentives, significantly enhancing firms’ innovation capabilities, production efficiency, and sustainable development capacities [43].

A synthesis of these arguments suggests that the literature provides a robust theoretical expectation that agglomeration should moderate the innovation-upgrading relationship. Yet, this proposition often remains just that—a theoretical expectation. There is a conspicuous gap in empirically testing this moderating effect within integrated analytical models that also account for other critical factors, such as a sector’s embeddedness in Global Value Chains.

This general empirical gap becomes particularly salient and consequential in the context of the automotive sector. The moderating effect of agglomeration is acutely relevant in the automotive industry, given the pronounced geographical clustering of its supply chain. Yet, the question of how this agglomeration explicitly amplifies the effect of innovation on upgrading in the Chinese automotive context, which features both market-driven clusters and state-guided industrial parks—remains underexplored in a systematic, empirical manner.

The moderating role of industrial agglomeration is grounded in the unique ecosystem of automotive clusters. The geographical concentration of original equipment manufacturers (OEMs), suppliers, R&D centers, and skilled labor—as seen in hubs like the Yangtze River Delta—creates a fertile ground for innovation through intense knowledge spillovers, reduced transaction costs, and rapid circulation of tacit knowledge. This agglomeration effect is posited to amplify the relationship between technological innovation and sector-specific upgrading: proximity to leading firms and institutions accelerates the diffusion, adoption, and application of new technologies, such as lightweight materials or battery management systems. Moreover, dense networks lower the risk and cost of collaborative R&D on complex automotive technologies. Thus, the same level of innovation investment is likely to yield a greater return on upgrading in a highly agglomerated environment compared to a dispersed one. Accordingly, we hypothesize:

H3. 

Industrial agglomeration acts as a moderator in the relationship between technological innovation and industrial upgrading in China ’s automotive industry.

2.5. Synthesis and Positioning of the Study

The preceding literature review establishes the foundational relationships between technological innovation, GVC, industrial agglomeration, and industrial upgrading. While these streams of research are well-developed in isolation, a critical synthesis reveals a significant gap in their integration, particularly within sector-specific, contextually nuanced investigations.

A fundamental critique emerging from this review is that the prevailing literature, across its various streams, often adopts a reductionist approach. By examining these drivers in isolation—studying the direct innovation-upgrading link without the mediating context of GVC, or analyzing GVC dynamics divorced from the spatial context of agglomeration, or theorizing agglomeration without situating it within global production networks—the literature inadvertently decomposes a complex, synergistic reality into discrete and incomplete narratives. This reductionism limits the explanatory power of existing models, as they fail to capture the configurational causal relationships where the effect of one factor (e.g., innovation) is contingent on the levels and states of others (e.g., GVC position and agglomeration). Consequently, there is a pressing need for a more relational and integrated theoretical framework that can account for these interdependencies.

The following table (Table 1) systematically compares the focus, methodological approach, and key limitations of selected seminal and contemporary studies against the positioning of the present research. This comparison serves to explicitly clarify how this study builds upon prior work while directly addressing the limitations identified earlier.

As summarized in Table 1, the existing literature offers robust yet fragmented insights. Macro-level studies establish broad correlations but lack sectoral granularity. Qualitative GVC studies offer rich descriptions of governance but do not provide quantitative tests of mediating pathways. Classical agglomeration theories are influential yet require empirical validation in contemporary, globally integrated industrial landscapes. This fragmentation—spanning analytical levels, methodologies, and theoretical contexts—creates a critical gap in understanding how these theoretical constructs interact as a synergistic system to drive industrial development in practice. To move beyond the limitations of this fragmented and reductionist perspective, this study directly addresses these gaps by proposing and testing an integrated framework that simultaneously investigates the mediating role of GVC and the moderating role of industrial agglomeration within the strategically vital and theoretically representative context of China’s automotive industry. Therefore, this research not only applies existing theories but also extends them by examining their interconnectedness in a specific, high-stakes empirical setting, thereby offering a more holistic and mechanistically detailed explanation of industrial upgrading drivers.

This study deliberately selects the automotive industry as its empirical context for investigating the mechanisms of technological innovation-driven industrial upgrading. This choice is predicated on the sector’s distinctive characteristics that render it a strategically critical and theoretically representative case. Firstly, the automotive industry is a technology-intensive pillar sector characterized by a long and complex value chain, high-level R&D investment, and significant spillover effects to upstream (e.g., advanced materials, semiconductors) and downstream (e.g., logistics, digital services) industries. Its upgrading process is inherently dependent on technological innovation, making the causal relationship between the two particularly salient and observable. Secondly, and central to our theoretical framework, the automotive sector is highly representative of the dynamics under study. It is one of the most globally integrated industries, serving as a paradigmatic case for analyzing GVC mediation. Concurrently, its development is geographically concentrated in prominent clusters (e.g., the Yangtze River Delta in China), making it an ideal empirical setting for examining the moderating role of industrial agglomeration. Therefore, while the findings are derived from a specific sector, the core mechanisms of GVC integration and agglomeration economies explored in this study are highly transferable to other complex, globally engaged, and spatially clustered manufacturing industries (e.g., electronics, machinery), thereby providing insights of broad theoretical relevance.

However, a review of the sector-specific literature reveals a gap: while the importance of innovation, GVC, and agglomeration in the automotive industry is acknowledged individually, there is a lack of an integrated empirical framework that simultaneously tests the mediating mechanism of GVC and the moderating role of agglomeration. This gap epitomizes the broader fragmentation identified above, manifesting concretely within the sector that most demands such an integrated analysis. This study aims to fill this gap by providing a systematic analysis of these dual mechanisms within the distinctive and critical context of China’s automotive industry.

Informed by the preceding analysis, this study constructs the following research framework. As depicted in Figure 1, industrial upgrading is designated as the dependent variable, whereas technological innovation is identified as the independent variable. GVC functions as the mediating variable, and industrial agglomeration assumes the role of the moderating variable. To control for extraneous influences on industrial upgrading, variables such as the degree of openness, foreign direct investment (FDI), and government intervention are integrated into the model. This comprehensive variable selection facilitates a more nuanced exploration of the drivers propelling industrial upgrading. By examining the relationships among these variables, this paper investigates how technological innovation drives industrial upgrading toward long-term sustainability in China’s automotive industry.

3. Materials and Methods

This research seeks to explore how technological innovation influences industrial upgrading in China’s automotive industry, specifically examining the mediating role of GVC and the moderating function of industrial agglomeration. A panel dataset comprising 28 provincial administrative units in China (2000–2020) is employed for the analysis. The regions of Qinghai, Ningxia, and Tibet were excluded due to systematically incomplete data on key variables throughout the study period, which preclude their valid inclusion in the econometric analysis. To measure the level of industrial upgrading, the DEA-Malmquist index method is applied. The system GMM model is utilized for regression analysis, and mediation-moderation analytical frameworks are integrated to explore the pathways through which technological innovation affects industrial upgrading in China’s automotive industry.

We opted for the DEA-Malmquist index method to measure industrial upgrading. It is considered a better measure than alternatives, such as labor productivity, for answering our research questions because it separately captures technological progress and efficiency change. This decomposition provides a far richer understanding of how technological innovation promotes upgrading, telling us whether it acts by shifting the technological frontier or by improving technical efficiency—a distinction that simpler metrics cannot capture.

We employed the System GMM estimator for two primary reasons, establishing it as the best choice. First, it accounts for the dynamic nature of upgrading. Second, and most critically, it is the most robust method for addressing endogeneity concerns (e.g., reverse causality). Ordinary regression techniques, such as OLS or standard fixed-effects models, would yield biased results if such endogeneity is present. By using internal instruments, System GMM provides the most reliable estimates, making it the preferred choice for validly testing the direct, mediating, and moderating effects central to our research questions.

The following flowchart (Figure 2) illustrates the sequential research steps involved in data collection, sample construction, variable measurement, and econometric modeling undertaken in this study.

3.1. Variables

3.1.1. Dependent Variable

In this research, industrial upgrading serves as the dependent variable. Following the approaches of Sun and Xi [44] and Xu [45], this study employs total factor productivity (TFP) as the indicator to measure industrial upgrading. TFP is widely used in studies of industrial upgrading because it comprehensively reflects the combined performance of technological progress, resource allocation efficiency, and productivity improvement within an industry. In this study, TFP is measured using the DEA-Malmquist index method. This approach is widely used for analyzing the efficiency of dynamic panel data, as it can effectively decompose the sources of TFP changes—explicitly distinguishing between the impacts of technological change and efficiency change [46,47].

In the application of the DEA-Malmquist method, the selection of input and output indicators directly affects the accuracy and scientific nature of efficiency measurement. Considering the unique characteristics of the automotive industry, this study selects the following indicators:

For input indicators, labor input is measured by the year-end number of employees in the automotive industry [48,49], with the unit being 10,000 people. Capital input is measured by the sum of the original value of fixed assets and the total amount of current assets [50], with the unit being billion yuan, and 2000 is used as the base year for deflation adjustment.

For output indicators, the total industrial output value of the automotive industry is used to measure output [51], with the unit being billion yuan, and 2000 as the base year for deflation.

This study follows the approach of Du and Liu [52] by assuming that the TFP in the base year 2000 is 1. Consequently, the TFP for 2001 is calculated by multiplying the TFP of 2000 by the Malmquist productivity index for 2001, and this cumulative calculation process is continued for subsequent years.

3.1.2. Independent Variable

Technological innovation serves as the independent variable in this research. Considering the availability of data, this study adopts the methodology proposed by Yu et al. [53] and Wang et al. [54], which measures technological innovation through internal research and development (R&D) expenditures of above-scale enterprises at the provincial level. Specifically, for the automobile industry, we adapt the methodology of Shen [55] by calculating the ratio of internal R&D expenditures of above-scale enterprises to the provincial GDP, and then multiplying this ratio by the automotive industry’s total industrial output value within each region.

The construction of this indicator is theoretically grounded in the conceptualization of innovation as a process that transforms R&D inputs into economically meaningful outputs. Specifically, we first calculate the ratio of internal R&D expenditures to provincial GDP. This ratio reflects the intensity of regional innovation inputs, capturing the economic priority and resource commitment to R&D activities. We then multiply this R&D intensity by the total industrial output value of the automotive industry in each province. This multiplicative procedure serves to scale and sectoralize the provincial-level innovation effort, effectively allocating it to the automotive sector. The resulting variable, “R&D effort embedded within the automotive industrial output,” provides a conceptually sound and empirically tailored proxy for the sector-specific technological innovation capability of the automotive industry.

This approach is not arbitrary but is justified by prior scholarly work. The methodology of Shen [53], which refines broad industrial-sector indicators into specific industry-level measurements, has been empirically validated and offers a reliable framework for constructing such proxies. By adopting this refined composite measure, we ensure that our operationalization of technological innovation is both theoretically informed and methodologically robust, allowing for a more comprehensive analysis of the automotive industry.

3.1.3. Mediating Variable

GVC serves as the mediating variable in this research. To quantify the position of China’s automotive industry within the GVC, the study employs the GVC position index developed by Koopman et al. [56]. Following the methodologies of Yu and Gu [57], Gonzalez et al. [58], and Liu and Wu [59], we derive the provincial-level GVC position index specifically for China’s automotive industry.

The GVC position index (GVCi) represents the position of the automotive industry in province i within the GVC. The calculation is performed via the following equation:

(1)${\mathrm{G}\mathrm{V}\mathrm{C}}_{\mathrm{i}}=\ln \left(1+{\displaystyle \frac{\mathrm{I}\mathrm{V}{\times \mathrm{V}}_{\mathrm{i}}/\mathrm{V}}{\mathrm{E}\times {\mathrm{T}}_{\mathrm{i}}/\mathrm{T}}}\right)-\ln \left(1+{\displaystyle \frac{\mathrm{F}\mathrm{V}\times {\mathrm{V}}_{\mathrm{i}}/\mathrm{V}}{\mathrm{E}\times {\mathrm{T}}_{\mathrm{i}}/\mathrm{T}}}\right)$

where Vi and V denote the automotive industry’s output value in province i and the national total, respectively, while Ti and T represent the total export value of province i and China as a whole. The data for these indicators are obtained from the “China Statistical Yearbook” and the “China Automotive Industry Yearbook.”

Additionally, IV, FV, and E correspond to the indirect value added from automobile exports, the value added to foreign markets, and the total value added, respectively. These data are extracted from the OECD-TiVA 2023 database (C29 sector), updated through the year 2020. A higher GVC position index indicates a more upstream position within the global value chain, reflecting greater influence on the global economy.

3.1.4. Moderating Variable

In this study, the moderating variable is industrial agglomeration. Following the approach of Zhang et al. [60], we utilize the location quotient (LQ) index to assess the level of industrial agglomeration in China’s automotive industry. LQ provides a clear indication of the relative concentration of a specific industry within a regional economy and is widely used to examine the effects of industrial agglomeration. Its calculation is defined by the formula below:

(2)${\mathrm{I}\mathrm{A}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{A}\mathrm{S}}_{\mathrm{i}}/{\mathrm{S}}_{\mathrm{i}}}{\mathrm{A}\mathrm{S}/\mathrm{S}}}$

where IAi indicates the Location Quotient (LQ) of the automotive industry agglomeration in province i. It is calculated by comparing the proportion of the automotive industry’s output in the province (ASi/Si) to the proportion of the automotive industry’s output nationwide (AS/S). Here, ASi is the output of the automotive industry in province i, and Si is the total industrial output of province i. At the national level, AS is the total output of the automotive industry, and S is the aggregate industrial output of the country.

The LQ index enables the evaluation of the automotive industry’s relative concentration across regions. A higher LQ value corresponds to a stronger degree of regional specialization in this sector. Specifically, an LQ greater than 1 implies that province i exhibits a comparative advantage in the automotive industry relative to the national benchmark. On the other hand, an LQ below 1 signifies that the province’s automotive specialization lags behind the national average, reflecting a relative competitive weakness.

3.1.5. Control Variables

This study incorporates the degree of openness (OPEN), foreign direct investment (FDI), and government intervention (GI) as control variables. The degree of openness is measured as the total value of imports and exports of a province as a percentage of GDP, reflecting the level of international trade engagement [61]. FDI is operationalized following Shen [55], where the total FDI across all industries is multiplied by the ratio of the provincial automotive industry’s output value to the province’s total GDP. Government intervention is proxied by the proportion of general public budget expenditure of a province to its GDP, as governmental actions can influence industrial upgrading through resource allocation and policy guidance [62].

3.2. Model Specification

The empirical analysis in this study utilizes the System Generalized Method of Moments (System GMM) approach. The choice of System GMM is motivated by the dynamic nature of industrial upgrading, which is significantly influenced not only by current factors but also by its historical development levels. Therefore, a dynamic panel model is more appropriate for capturing this process. Moreover, the System GMM method has significant advantages in estimating dynamic panel models. It effectively addresses endogeneity issues arising from lagged dependent variables, mitigates biases due to unobserved individual effects, and handles heteroskedasticity and serial correlation, thereby significantly enhancing the robustness and reliability of the estimation results.

Crucially, the System GMM estimator is specifically designed to address endogeneity concerns, including those stemming from reverse causality and omitted time-invariant confounders. The estimator employs a system of two equations—the difference equation, instrumented by lagged levels, and the level equation, instrumented by lagged differences. This instrumentation strategy, applied to all regressors considered potentially endogenous, facilitates the identification of causal effects by exploiting exogenous variation. The validity of the instruments hinges on the assumption that the lagged values are uncorrelated with the error term. This assumption is empirically supported by the Hansen test for over-identifying restrictions, which fails to reject the null hypothesis of instrument validity across all model specifications, thereby bolstering confidence in the consistency of the estimates.

This research constructs three types of empirical models: the baseline model, the mediation effect model, and the moderation effect model. These models are designed to analyze the direct relationship between technological innovation and industrial upgrading, examine the mediating effect of the global value chain, and investigate the moderating effect of industrial agglomeration, respectively. The empirical investigation utilizes panel data spanning the years 2000–2020, covering 28 provincial-level regions in China. Several provinces were omitted from the sample due to insufficient data availability.

The panel dataset used in this study spans 21 years (T = 21) for 28 provinces (n = 28). While annual data is available, we employ a triennial averaging method, resulting in a balanced panel of 7 time periods. This data transformation is implemented for two primary methodological reasons, and enhances the robustness and validity of the System GMM estimates.

First, it helps to mitigate the problem of instrument proliferation. System GMM estimators can generate a large number of instruments, particularly when T is moderately large. An excessive number of instruments relative to the cross-sectional dimension (N) can overfit endogenous variables and weaken the power of the Hansen test of instrument validity, potentially leading to biased results [63]. Reducing the time dimension to 7 periods effectively controls the instrument count, leading to more reliable specifications.

Second, a three-year averaging interval is well-suited to capture the medium-term trends that are the focus of this study. It smooths out short-term business cycle fluctuations and annual idiosyncrasies that may obscure the fundamental, sustained relationships between technological innovation, GVC, agglomeration, and industrial upgrading. While this approach may reduce the granularity of annual dynamics, the inclusion of a lagged dependent variable in the model itself captures dynamic persistence. The choice of a three-year cycle aligns with typical strategic planning and investment cycles in the automotive industry, making it a conceptually appropriate timeframe for analyzing structural upgrading.

3.2.1. Baseline Model

Initially, a baseline regression model is constructed to examine the direct influence of technological innovation on industrial upgrading in China’s automotive industry. The model is specified as follows:

IU_i,t =α₀ + α₁ IU_i,t−1 + α₂ TI_i,t + α₃ OPEN_i,t + α₄ FDI_i,t + α₅ GI_i,t + ℇ_i,t(3)

In this model, IUi,t denotes the dependent variable, which captures the industrial upgrading level of the automotive industry in province i and year t, as quantified through the DEA-Malmquist index. IUi,t−1 represents the lagged term of the dependent variable, accounting for the dynamic persistence of industrial upgrading. The independent variable, TIi,t captures the technological innovation capacity of the automotive industry in province i in year t. The set of control variables includes OPENi,t, FDIi,t, and GIi,t, denoting the degree of openness, foreign direct investment, and government intervention in province i during year t, respectively.

3.2.2. Mediating Effect Model

To further explore the indirect pathway through which technological innovation affects industrial upgrading via GVC, this study introduces the GVC as a mediating variable. In line with the approach proposed by Yang et al. [64], the mediation model is constructed with the following two equations:

GVC_i,t = β₀ + β₁ GVC_i,t−1 + β₂ TI_i,t + β₃ OPEN_i,t + β₄ FDI_i,t + β₅ GI_i,t + ℇ_i,t(4)

IU_i,t = θ₀ + θ₁ IU_i,t−1 + θ₂ TI_i,t + θ₃ GVC_i,t + θ₄ OPEN_i,t + θ₅ FDI_i,t + θ₆ GI_i,t + ℇ_i,t(5)

Here, GVCi,t acts as the mediating variable, denoting province i’s GVC position index in year t, which measures the relative position of the automotive industry in global production networks. A higher GVC value indicates that the region’s industry is positioned more upstream in GVC and possesses considerable international influence. This model examines the impact of TIi,t on GVCi,t and subsequently assesses how GVCi,t affects IUi,t, thereby verifying the mediating role of GVC position. Yang et al. [64] suggest that if both β2 and θ3 are statistically significant, it indicates that the GVC indeed plays a mediating role in the impact of technological innovation on industrial upgrading.

3.2.3. Moderating Effect Model

To examine how industrial agglomeration moderates the effect of technological innovation on industrial upgrading in China’s automotive industry, a moderation model is developed with industrial agglomeration incorporated as the moderating variable, specified as follows:

IU_i,t = ω₀ + ω₁ IU_i,t−1 + ω₂ TI_i,t + ω₃ TI_i,t × IA_i,t + ω₄ IA_i,t + ω₅ OPEN_i,t + ω₆ FDI_i,t + ω₇ GI_i,t + ℇ_i,t(6)

In this equation, IAi,t represents the moderating variable, which is the level of automotive industry agglomeration in province i and year t. TIi,t × IAi,t denotes the interaction term between technological innovation and industrial agglomeration, capturing their joint effect. If both ω2 and ω3 are statistically significant, it indicates that industrial agglomeration moderates the effect of technological innovation on China’s automotive industry upgrading.

3.3. Data Sources

To investigate how technological innovation influences industrial upgrading in China’s automotive industry and its underlying mechanisms, this study employs panel data spanning the period from 2000 to 2020, covering a 21-year period. The dataset comprises 28 provincial-level regions in China, with three provinces omitted due to systematically incomplete key variable data. The core variables include independent variable (technological innovation), dependent variable (industrial upgrading), mediating variable (global value chain), moderating variable (industrial agglomeration), and control variables (degree of openness, foreign direct investment, and government intervention).

Data for the variables were collected from the following sources: industrial upgrading (the dependent variable) comes from the China Automotive Industry Yearbook and the China Industrial Statistical Yearbook; technological innovation (the independent variable) is drawn from the China Automobile Industry Yearbook and the China Statistical Yearbook; global value chain embeddedness (the mediating variable) is obtained from the OECD-TIVA 2023 Database, the China Automotive Industry Yearbook, and the China Statistical Yearbook; industrial agglomeration (the moderating variable) is sourced from the China Automotive Industry Yearbook, the China Industrial Statistical Yearbook, and the China Statistical Yearbook; control variables rely on the China Automotive Industry Yearbook and the China Statistical Yearbook. A summary of variable abbreviations and corresponding data sources is provided in Table 2.

Building upon the theoretical foundations and empirical models specified above, this study integrates the direct, mediating, and moderating effects into a cohesive analytical framework. To provide a clear and intuitive conceptual overview, Figure 3 illustrates the hypothesized pathways through which technological innovation influences industrial upgrading in China’s automotive industry.

The framework delineates three core mechanisms. Firstly, technological innovation directly promotes industrial upgrading by enhancing production efficiency, optimizing resource allocation, and generating knowledge spillovers. Secondly, it exerts an indirect influence by elevating the industry’s position in the global value chain (GVC), where breakthroughs in core technologies enable firms to secure higher value-added segments of the chain. Thirdly, the positive effect of technological innovation is strengthened by industrial agglomeration, as the concentration of firms fosters knowledge spillovers, enables specialized division of labor, and reduces innovation costs, thereby amplifying the impact of innovation on industrial upgrading.

This synthesized framework serves as a visual guide for the empirical tests conducted in the subsequent section (Section 4), which will sequentially examine the direct effect, the mediating effect, and the moderating effect to validate the hypothesized mechanisms outlined above.

4. Results and Discussion

This study empirically examines how technological innovation influences industrial upgrading in China’s automotive industry, with particular emphasis on the mediating function of GVC and the moderating role of industrial agglomeration. The total factor productivity (TFP), measured using the DEA-Malmquist method, serves as the indicator for industrial upgrading. The baseline regression is conducted using the system GMM approach, followed by an examination of the mediating role played by GVC and the moderating influence of industrial agglomeration. This approach elucidates the underlying mechanisms through which technological innovation drives industrial upgrading in China’s automotive industry. This chapter includes the results of descriptive statistics, DEA analysis, baseline regression, mediation and moderation effect tests, as well as robustness tests—all of which collectively support the validation of the study’s hypotheses.

4.1. Descriptive Statistical Results

This study presents descriptive statistics for key variables related to the industrial upgrading in China’s automotive industry. As shown in Table 3, industrial upgrading (IU) has a mean value of 3.0704 and a standard deviation of 2.3097, with values ranging from 0.2768 to 15.1788, indicating significant inter-provincial disparities. For technological innovation (TI), the mean is 2.5193 with a standard deviation of 5.1349, and the values span from 0.0006 to 29.2029, demonstrating marked heterogeneity in innovation capabilities across provinces.

The mediating variable, GVC, shows a mean value of 0.5830, ranging from 0.0132 to 1.3025, suggesting relatively small inter-provincial differences in global value chain positioning. In contrast, the moderating variable, industrial agglomeration (IA), has a mean of 1.1627, with values ranging from 0.0119 to 8.5466, indicating substantial regional variation in industrial agglomeration levels. Regarding the control variables, the degree of openness (OPEN), foreign direct investment (FDI), and government intervention (GI) have means of 0.3170, 2.8040, and 0.1937, respectively. These descriptive statistics reveal the variability across the key variables in the study and provide a foundational basis for the subsequent empirical analysis.

4.2. DEA Results

This research utilizes the DEA-Malmquist index approach with Deap2.1 software to measure China’s automotive industry upgrading level, calculating total factor productivity changes (Tfpch) and decomposing them into technological change (Techch) and efficiency change (Effch). Specifically, Techch measures the outward shift in the production frontier, indicating technological progress or regression. Effch reflects the extent to which decision-making units (DMUs) move closer to the production frontier, given constant technology, thereby indicating improvements in managerial efficiency. Figure 4 summarizes the corresponding results.

Figure 4 indicates that the annual average growth rate of Total Factor Productivity (TFP) in China’s automotive industry is 6.3%, reflecting a positive trend in overall industrial upgrading. The decomposition of TFP reveals that the annual growth rate of technological change (Tech) is 3.4%, which surpasses the annual growth rate of efficiency change (Effch) at 2.8%. Moreover, the figure shows that the trend of TFP changes in China’s automotive industry roughly aligns with the trend of technological change, suggesting that technological advancements are the primary driver of TFP growth in the industry. This finding highlights technological innovation as a key driver of the upgrading in China’s automotive industry.

In temporal terms, the TFP of China’s automotive industry has generally exhibited growth in most years, with the highest growth rate recorded between 2015 and 2016, reaching 59.1%. However, there are certain years, particularly from 2017 to 2020, when TFP experienced negative growth. This volatility underscores the imperative of accelerating industrial upgrading efforts to achieve stable and sustained TFP growth and secure a long-term competitive advantage in the global automotive industry.

Figure 5 illustrates the geographical disparities in TFP growth within China’s automotive industry across 28 provinces. Although a positive upward trend is evident in most regions, a small number of provinces, notably Gansu, exhibited negative growth. Specifically, Beijing recorded the highest TFP growth rate at 15.3%, which stands in stark contrast to Gansu’s growth rate of −7.6%. This significant regional variation in production efficiency underscores the marked unevenness in the progress of industrial upgrading across China’s national automotive industry. Consequently, achieving a more balanced and comprehensive industrial upgrade remains a critical challenge for the sustained development of China’s automobile industry.

4.3. Baseline Effect Results

To address potential endogeneity issues and the dynamic nature of panel data, this study employs System GMM for baseline regression analysis using Stata 17. Here, TFP, calculated through the DEA-Malmquist approach, serves as the dependent variable to represent industrial upgrading. Technological innovation is set as the independent variable, with control variables including the degree of openness (OPEN), foreign direct investment (FDI), and government intervention (GI). The baseline regression results are presented in Table 4.

Table 4 demonstrates that the lagged term of industrial upgrading (L1.IU) in China’s automotive industry exhibits a significantly positive effect on its current level (coefficient = 0.8847, p < 0.01). This finding confirms that the ongoing industrial upgrading in China’s automotive industry is influenced by past upgrading experiences, thereby validating the applicability of the dynamic panel System GMM model used in this analysis. Furthermore, diagnostic tests confirm the validity of the model: the p-values for the AR (1), AR (2), and Hansen tests are 0.051, 0.299, and 0.541, respectively. These results indicate the absence of second-order serial correlation in the model residuals and validate the instrument set, supporting the robustness of the System GMM estimator.

The baseline regression results show that technological innovation has a positive and statistically significant effect on industrial upgrading (coefficient = 0.0343, p < 0.01), providing strong support for Hypothesis 1. This confirms that innovation is a crucial driver of industrial upgrading in China’s automotive industry. This positive impact aligns with established literature, which suggests that innovation enhances productivity through improved efficiency, optimized resource allocation [65], and knowledge spillovers that strengthen industry-wide competitiveness [66].

In the context of China’s automotive industry, technological innovation manifests not only in emerging domains, such as new energy vehicles and intelligent connected vehicles, but also in the optimization and upgrading of traditional manufacturing technologies. Consequently, sustained investment in technological innovation is indispensable for maintaining and enhancing the global market position of China’s automotive industry while supporting its long-term sustainable development.

Regarding the impact of control variables, the coefficients for the degree of openness, foreign direct investment, and government intervention are all positive and statistically significant, indicating that these factors play a vital role in driving industrial upgrading. An increase in the degree of openness enhances market competition, promotes the flow of technology and knowledge, and further stimulates innovation and industry upgrading [67]. Foreign direct investment brings advanced technologies and managerial expertise, which helps enhance firms’ technological capabilities and market competitiveness [68]. Furthermore, moderate government intervention can provide essential support and guidance for industrial development, facilitating the optimal allocation of resources [69].

4.4. Mediating Effect Results

To test the mediating role of the GVC in the relationship between technological innovation and industrial upgrading in China’s automotive industry, this study employs System GMM analysis in Stata 17 for the mediation effect model (Model 2, corresponding to Equations (4) and (5)). Table 5 presents the estimation results.

The diagnostic tests confirm the validity of the model: the p-values for AR (1) and AR (2) indicate no autocorrelation in the residuals (AR (1) < 0.1; AR (2) > 0.1), and the Hansen test (p > 0.1) shows no evidence of over-identification. Therefore, the mediating effect model is correctly specified.

The results in Column (1) indicate that technological innovation has a significant positive effect on the GVC position of China’s automotive industry (coefficient = 0.0064, p < 0.05). This finding highlights the importance of innovation in achieving a more competitive position in the global value chain.

This positive relationship can be explained through three mechanisms. Firstly, breakthroughs in core technologies, such as those related to electric vehicle batteries and autonomous driving, have improved product quality and competitiveness, enabling firms to occupy higher segments of the value chain. These technological advancements also make direct contributions to the automotive industry’s sustainable development objectives [70]. Secondly, the advanced processes and management methods introduced through technological innovation have enhanced production efficiency and reduced costs, allowing companies to maintain price advantages while achieving greater value added [71]. Finally, the knowledge spillovers and learning effects generated by technological innovation accelerate the absorption and application of new technologies, driving not only the upgrading of individual enterprises but also facilitating the entire industry chain’s ascent into higher-value segments [72]. This systemic enhancement supports long-term economic sustainability by fostering a more resilient, innovative, and resource-efficient industrial system.

Column (2) investigates whether China’s automotive GVC position significantly influences industrial upgrading. The results demonstrate that GVC position exerts a significantly positive effect on automotive industry upgrading (coefficient = 1.4568, p < 0.01), indicating that elevating the GVC position of China’s automotive industry can facilitate overall industrial upgrading.

The mechanisms underlying this effect can be summarized as follows: Firstly, an improved position in the global value chain allows Chinese automobile companies to engage more effectively in the international division of labor, optimizing resource allocation. By concentrating limited resources on high-value-added segments, firms can enhance their competitiveness, thereby facilitating overall industrial upgrading [73]. Secondly, as the global value chain position improves, Chinese automobile enterprises gain access to advanced international technologies, enabling them to learn valuable technical and managerial experiences. This technological spillover and learning effect further drive industrial upgrading [74]. Ultimately, a more prominent position within the global value chain facilitates smoother entry into international markets, enabling firms to expand their sales channels. Through interactions with international brands, companies can raise their brand awareness and market influence, further propelling industrial upgrading [34]. This global market integration ultimately contributes to economic sustainability by fostering long-term industrial resilience and facilitating a transition toward higher-value, knowledge-intensive production.

Collectively, the results from Columns (1) and (2) show that technological innovation improves China’s automotive GVC position, which in turn drives industrial upgrading. Baseline model results further demonstrate that technological innovation has a significant impact on automotive upgrading. These findings together validate Hypothesis 2, which posits that the global value chain exerts a mediating effect on the relationship between technological innovation and automotive industrial upgrading.

4.5. Moderating Effect Results

To test the moderating role of industrial agglomeration in the relationship between technological innovation and industrial upgrading in China’s automotive industry, this study estimates Model 3 using the System GMM method in Stata 17. The results are reported in Table 6.

As shown in Table 6, the AR (1), AR (2), and Hansen test results are satisfactory, supporting the validity of the System GMM model. Importantly, the coefficient for the interaction term (TI_IA) is positive and statistically significant at the 1% level. This demonstrates that industrial agglomeration strengthens the effect of technological innovation on industrial upgrading, thus confirming Hypothesis 3 that agglomeration acts as a positive moderator.

To elucidate the economic significance of the moderating effect, we compute the marginal effect of technological innovation on industrial upgrading, derived from the partial derivative of the estimation model: ∂IU/∂TI = β_TI + β_Interaction × IA. Drawing on the coefficients in Table 6 (β_TI = 0.2204; β_Interaction = 0.0530), the analysis demonstrates that the returns to innovation are contingent upon the regional context of industrial agglomeration.

For illustration, in a region with moderate agglomeration (IA = 1.0, near the sample mean), the marginal effect of technological innovation is 0.2734, implying that a one-unit increase in TI is associated with a 0.273-unit rise in industrial upgrading. Conversely, in a highly agglomerated region (IA = 3.0, approximately one standard deviation above the mean), the marginal effect increases to 0.3794—suggesting a 0.379-unit gain in IU for an equivalent innovation effort. This represents a 39% amplification in the efficacy of technological innovation, underscoring the role of industrial agglomeration as a substantive contextual amplifier that significantly enhances the productivity of innovation-driven industrial upgrading activities.

Industrial agglomeration significantly strengthens the positive effect of technological innovation on industrial upgrading, manifested primarily in three aspects: Firstly, the knowledge spillover effect is particularly pronounced within agglomerated regions. The geographical proximity of firms facilitates the rapid flow of technology, talent, and information [75], thereby accelerating the diffusion and absorption of innovation outcomes and amplifying the role of technological innovation. Secondly, the effects of the specialized division of labor and collaboration are significant within agglomerated areas. A well-developed industrial chain, such as upstream parts supply and downstream sales services, reduces innovation costs for firms [76]. This enables technological innovation to be more easily translated into a driving force for industrial upgrading. Lastly, resources within agglomerated regions can complement each other. The innovation network formed by universities, research institutions, and enterprises can integrate complementary resources [77], thereby enhancing the efficiency of technological innovation conversion and strengthening the economic sustainability of regional automotive industries.

4.6. Robustness Check Results

To ensure the validity of the study’s conclusions, this research conducted systematic robustness checks from multiple perspectives. These checks included alternative estimation techniques, model specification tests, corrections for sample selection bias, sensitivity analyses based on data frequency, and a dedicated test for reverse causality using lagged innovation. Additionally, robustness checks were performed on alternative measures of industrial agglomeration. Through these comprehensive assessments, the study strengthened the credibility of its conclusions and ensured the robustness of its findings.

4.6.1. Robustness to Estimation Techniques and Model Specification

We conducted a series of tests to ensure our findings are not driven by specific model choices or variable selection.

First, we addressed potential concerns regarding the endogeneity of control variables. As reported in Table 7a, we sequentially excluded OPEN, FDI, and GI from our baseline System GMM model. The coefficient for Technological Innovation (TI) remains positive, statistically significant at the 1% level, and stable in magnitude across all model specifications. This demonstrates that our core finding is robust to the potential endogeneity of these controls.

Second, we tested the robustness of our model specification and estimation technique through multiple approaches. As reported in Table 7b, these approaches encompass the addition of control variables, fixed effects modeling, random effects modeling, and the Difference Generalized Method of Moments (DIFF-GMM) technique. The consistency of the test results across all methods validates the study’s primary conclusions, thereby underscoring the robustness and reliability of our findings.

In column (1) of Table 7b, human capital is introduced as an additional control variable, operationalized using average years of education. The results indicate that technological innovation maintains a statistically significant and positive influence on industrial upgrading in China’s automotive industry, consistent with the baseline model. This suggests that controlling for human capital does not affect the primary inference, thereby further corroborating the robustness of the findings.

Columns (2) and (3) present estimates based on fixed effects and random effects models, respectively. The results from both specifications confirm that technological innovation has a significant impact on industrial upgrading in China’s automotive industry, with coefficients that are statistically significant at the 1% level. A Hausman test yields a p-value of 0.3245, supporting the use of the random effects model over its fixed effects counterpart. The alignment of these results across different model specifications strengthens confidence in the robustness of the main conclusion.

Column (4) applies the DIFF-GMM technique for robustness testing. The regression results yield a coefficient of 0.0991 for technological innovation, statistically significant at the 1% level (p < 0.01), affirming its substantial role in promoting industrial upgrading in China’s automotive industry. This finding is consistent with baseline results. Furthermore, the AR (1), AR (2), and Hansen tests all meet the validation criteria, thereby confirming the model’s effectiveness.

Additionally, to address potential concerns regarding the measurement of the dependent variable, we conducted an additional test by replacing Total Factor Productivity (TFP) with Profit Margin (the ratio of total profits to main business income) as an alternative proxy for industrial upgrading [78]. The results, presented in the last column of Table 7b, demonstrate that the coefficient for Technological Innovation (TI) remains positive and statistically significant at the 1% level. This reaffirms that the core conclusion—technological innovation fosters industrial upgrading—holds even under an alternative conceptualization of the upgrading metric.

4.6.2. Robustness to Sample Selection Bias and Outliers

A key concern regarding sample selection bias stems from the exclusion of Qinghai, Ningxia, and Tibet due to data unavailability. To address this issue empirically, we conducted two additional robustness checks, the results of which are presented in Table 8.

First, we constructed a conservative sub-sample that excludes all provinces officially classified under China’s “Western Development” strategy, namely Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, Guangxi, and Inner Mongolia. This approach represents a stringent test, as it removes entire socio-economic and geographic clusters from the analysis. The results, presented in Column (1) of Table 8, demonstrate that the coefficients for technological innovation remain stable and statistically significant, confirming the robustness of our core findings to variations in regional sample composition.

Second, to ensure that our results are not unduly influenced by extreme observations at either tail of the distribution—whether from small-scale economies akin to the excluded provinces or from large-scale industrial hubs—we applied a 1% winsorization to all continuous variables. The regression estimates derived from this winsorized dataset, shown in Column (2) of Table 8, are highly consistent with our baseline results. Specifically, the positive and statistically significant effect of technological innovation on industrial upgrading (at the 1% level) is robustly confirmed. The coherence of results across these two distinct approaches offers strong evidence that the identified relationship is not driven by sample selection bias or outlier effects.

In summary, the core finding of this study—that technological innovation exerts a robust positive effect on industrial upgrading in China’s automotive industry—is consistently supported across a comprehensive set of robustness checks.

4.6.3. Robustness to Data Frequency and Dynamic Specification

To further validate our empirical strategy and address potential concerns regarding data aggregation, we conducted a comparative analysis using the original annual data. The results, presented in Table 9, confirm the consistency of our findings and offer valuable methodological insights.

The comparison yields two key findings. First, the coefficient for technological innovation (TI) remains positive and statistically significant at the 1% level in both specifications, Columns (1) and (2). This persistent significance across specifications provides compelling evidence for the robustness of our core conclusion that innovation drives industrial upgrading.

Second, and more critically, the models differ dramatically in their dynamic specifications. The annual model produces a theoretically implausible negative coefficient for the lagged dependent variable, suggesting a counterintuitive pattern where prior industrial upgrading leads to subsequent decline. This finding contradicts the fundamental economic principle of path dependency. In contrast, our triennial model yields a positive and highly significant lagged effect (0.8847), which is both economically meaningful and statistically robust.

This evidence demonstrates that our triennial averaging approach is not merely a technical convenience but a methodological necessity. It effectively filters out short-term noise that can distort the identification of dynamic relationships, thereby ensuring that our analysis reliably captures the persistent, medium-term structural processes that are the focus of this study.

4.6.4. Robustness to Reverse Causality Using Lagged Innovation

To directly address the potential for reverse causality—particularly the concern that GVC status might reciprocally influence the measurement of technological innovation (TI)—we conduct a series of robustness checks by replacing the contemporary TI with its one-period lagged value (L.TI). The economic rationale is straightforward: employing a lagged measure of TI helps break any potential contemporaneous feedback loop, as past innovation cannot be influenced by current GVC status.

We implement this test in two steps to rigorously validate our core findings. First, we re-examine the mediation effect. As shown in Columns (1) and (2) of Table 10, the coefficient for L.TI remains positive and statistically significant in explaining the mediator, GVC status (Column (1)). Furthermore, when L.TI is included alongside GVC in the full mediation model (Column (2)), GVC itself continues to exert a significant positive influence on industrial upgrading. The sustained significance of both pathways provides robust evidence that the mediated relationship—whereby technological innovation fosters industrial upgrading through GVC participation—remains valid after accounting for potential reverse causality.

Second, to further substantiate the direct effect, we re-estimate our baseline model using L.TI. The result, presented in Column (3) of Table 10, demonstrates that the direct impact of technological innovation on industrial upgrading remains positive and statistically significant at the 1% level, even when employing this pre-determined measure of innovation.

The coherence between results using lagged TI (L.TI) and contemporary TI provides compelling evidence against the alternative explanation of reverse causality. The finding that past innovation (L.TI) robustly predicts current industrial upgrading—with GVC status maintaining its mediating role—strongly supports our core theoretical argument that innovation is a primary driver of this process. Crucially, the persistence of the mediated pathway when using lagged TI further confirms that the identified causal sequence—from innovation to GVC to industrial upgrading—is not an artifact of contemporaneous endogeneity. Taken together, these tests substantially mitigate concerns that our main results are biased by reverse causality from industrial upgrading (or GVC status) to the measurement of technological innovation.

4.6.5. Robustness to Alternative Measures of Industrial Agglomeration

To further strengthen the validity of our findings, we test the robustness of the moderating effect of industrial agglomeration by employing two alternative measurement strategies. This exercise addresses the methodological concern that the results might be sensitive to the specific formulation of the Location Quotient (LQ) index, particularly its static nature.

First, we construct a dynamic variant of the LQ to capture the time-varying intensity of agglomeration. This measure, denoted as LQ_dynamic, is calculated as the product of the original LQ and the logarithmic growth rate of the automotive industry’s output: LQ_dynamic_it = LQ_it × ln(AS_it/AS_it − 1). This composite indicator thus reflects both the existing scale of agglomeration and its recent growth momentum.

Second, we utilize the industrial share of the automotive sector (IA_Share), defined as its output share within a province’s total industrial output (AS_i/S_i). This measure shifts the conceptual focus from regional specialization, as captured by the LQ, to the sector’s absolute economic weight within the provincial industrial structure.

The regression results using these alternative measures are presented in Table 11. The coefficient for the key interaction term between technological innovation and industrial agglomeration remains positive and statistically significant across both specifications. This consistent pattern across different measurement approaches offers strong evidence that the moderating effect of agglomeration is not driven by a specific metric choice, but is robust to alternative conceptualizations of the agglomeration phenomenon.

5. Conclusions and Policy Implications

5.1. Conclusions

This research empirically investigated the mechanisms through which technological innovation shapes industrial upgrading in China’s automobile industry, with a specific focus on the mediating function of GVC and the moderating role of industrial agglomeration. The main findings can be summarized as follows:

Firstly, technological innovation has a significant positive impact on industrial upgrading in China’s automotive industry. It serves as a key driver of total factor productivity (TFP) growth, thereby advancing the industry’s upgrading process. Specifically, technological innovation not only enhances production efficiency by introducing new technologies and management methods but also strengthens the overall competitiveness of the industry through knowledge spillovers and learning effects. These improvements collectively contribute to long-term economic sustainability by cultivating a more productive, efficient, and structurally advanced automotive sector.

Secondly, GVC serves as a statistically significant mediator in the relationship between technological innovation and industrial upgrading within China’s automotive industry. Technological innovation significantly enhances the industry’s GVC position, which in turn exerts a strong positive effect on industrial upgrading. This mediation mechanism operates through the following channels: technological innovation facilitates access to higher-value-added activities through core breakthroughs and knowledge spillovers, improves international resource allocation and division of labor, and strengthens global market integration and learning from advanced international practices. These dynamics not only propel industrial upgrading but also enhance economic sustainability by developing a more globally competitive automotive industry. Overall, these findings highlight that advancing GVC positioning constitutes a critical pathway through which innovation drives industrial upgrading.

Thirdly, industrial agglomeration significantly moderates the positive effect of technological innovation on industrial upgrading. This moderating role operates through three primary mechanisms: intensified knowledge spillovers due to geographic proximity, which accelerate the diffusion of technologies and expertise; cost reduction and specialization efficiency gains derived from well-developed industrial chains; and improved resource complementarity and collaboration within innovation networks involving firms, universities, and research institutes. These synergistic effects collectively enhance the efficiency with which innovation translates into industrial upgrading, thereby supporting regional economic sustainability.

Fourth, our analysis reveals substantial regional heterogeneity in the strength and manifestation of these mechanisms, consistent with the DEA results (Figure 5). For instance, provinces with leading TFP growth (e.g., Beijing, at 15.3%) demonstrate stronger utilization of knowledge spillovers within dense agglomerations and high GVC embeddedness in high-value segments. In contrast, provinces with lagging TFP (e.g., Gansu, with a −7.6% growth rate) face constraints such as weak industrial bases, limited access to global value chains, and insufficient agglomeration effects, which collectively attenuate the impact of technological innovation on upgrading.

When contextualized within the broader international literature, our conclusions reveal significant alignments and a distinctive contribution. The foundational finding that technological innovation is a primary driver of industrial upgrading strongly resonates with studies on advanced automotive nations such as Germany and Japan, where continuous innovation is central to maintaining global competitiveness [79]. Similarly, the mediating role of GVC integration echoes research on industries in emerging economies, such as those in India and Mexico, confirming that leveraging global networks is a critical pathway for latecomer firms to acquire knowledge and upgrade [80]. However, our study underscores a characteristic that is particularly pronounced in the Chinese context: the potent moderating role of large-scale, state-facilitated industrial agglomeration. While agglomeration economies are recognized globally, the scale, speed, and strategic coordination of cluster development in China (e.g., in the Yangtze River Delta) represent a unique model. This contrasts with agglomeration in many Western contexts, which often evolves more organically through market forces. Consequently, this research not only validates universal mechanisms of industrial upgrading but also delineates the significant role of spatially coordinated industrial policy as a distinctive characteristic of China’s development trajectory, thereby enriching the global scholarly conversation with a more nuanced perspective.

5.2. Policy Implications

The following recommendations are explicitly derived from our empirical findings and are designed to be actionable and context-specific, taking into account the mechanisms and regional heterogeneity revealed in our analysis.

5.2.1. Strengthening Investment in Technological Innovation

Building on the confirmed direct driving effect of technological innovation (H1), policy interventions must prioritize enhancing technological innovation through targeted fiscal instruments such as R&D tax credits, subsidies for patent applications, and competitive grants for strategic technologies, including new energy vehicles and intelligent driving systems. These measures can stimulate firm-level investment while mitigating the high risks and positive externalities associated with advanced innovation. Concurrently, establishing structured industry-academia-research partnerships is crucial to facilitate knowledge spillovers and accelerate the commercialization of technological advancements, thereby strengthening the industry’s capacity for sustainable upgrading and global competitiveness.

Policy design should account for regional heterogeneity: In established innovation hubs (e.g., Beijing, Shanghai), priorities should focus on original innovation in core technologies (e.g., chip design, autonomous driving algorithms) to overcome technological bottlenecks and achieve breakthroughs in high-value segments of GVCs. In contrast, in emerging automotive regions (e.g., central China), policies should prioritize the absorption of technology and adaptive innovation (e.g., optimizing battery manufacturing) to better align with local industrial capabilities and development stages.

5.2.2. Enhancing Global Value Chain Positioning

Given the robust evidence that GVC positioning serves as a critical mediating pathway (H2) for innovation-driven upgrading, policy initiatives must actively enable automotive firms to access and master critical technologies and advanced management practices. This necessitates facilitating international technical collaboration, cross-border mergers and acquisitions, and structured technology licensing agreements. Specifically, targeted support is needed to help firms acquire core patents, particularly in emerging fields such as electric vehicle batteries. Concurrently, establishing overseas innovation hubs and providing systematic intelligence on international standards and regulations are crucial for deepening embeddedness in high-value GVC segments. Moreover, a robust intellectual property rights (IPR) protection framework, aligned with international standards, should be strengthened to create a secure environment for technology absorption and to consolidate the industry’s long-term competitive advantage in global markets.

Region-specific strategies are critical here: Coastal provinces with strong GVC integration (e.g., Jiangsu, Zhejiang) should target high-value segments (e.g., R&D, after-sales services) through international collaboration; inland provinces, which are more reliant on low-value manufacturing, need policies to embed local firms into regional supply chains (e.g., ASEAN-China networks) as an initial step toward eventual participation in higher-value GVC activities.

5.2.3. Promoting Sustainable Industrial Agglomeration

Capitalizing on the finding that industrial agglomeration acts as a significant moderator (H3), policymakers should strategically promote the formation of specialized automotive clusters, leveraging their demonstrated capacity to enhance innovation efficiency through resource complementarity. Concrete measures include developing integrated upstream-downstream industrial chains to reduce innovation costs, investing in shared infrastructure such as testing facilities and digital platforms to support collaborative R&D, and incorporating green production standards into cluster planning to align agglomeration strategies with both economic and environmental sustainability objectives.

Clustering policies should avoid “one-size-fits-all” approaches: Mature clusters (e.g., Guangzhou’s automotive belt) require regulation to prevent overcrowding and environmental costs, with a focus on upgrading to smart manufacturing clusters. For underdeveloped regions with weak industrial foundations (e.g., Gansu), policies must prioritize investments in foundational infrastructure (e.g., transportation, logistics, and digital connectivity) to catalyze the initial formation of industrial clusters and attract key anchor enterprises.

While our policy implications are derived from the Chinese context, they offer transferable insights for other countries and regions aiming to promote technology-driven industrial upgrading. Firstly, the imperative to strengthen indigenous innovation capacity through R&D incentives is a universal principle, though the specific policy instruments must be adapted to local institutional contexts. Secondly, the strategic focus on GVC positioning is crucial for any latecomer economy, emphasizing the need to move beyond passive participation to active knowledge acquisition and capability building. Finally, the recognition that industrial agglomeration can amplify innovation effectiveness suggests that policymakers elsewhere should consider cluster-based development strategies, even if their implementation differs from the Chinese model. The overarching lesson for other regions lies in constructing integrated policy frameworks that concurrently foster innovation capability, strategic global engagement, and the advantages of economic agglomeration.

In summary, these policy recommendations are designed to establish a synergistic ecosystem where technological innovation, global value chain integration, and spatial agglomeration mutually reinforce one another, thereby promoting sustainable and high-quality development in China’s automotive industry.

5.3. Limitations and Future Research Directions

Despite its comprehensive empirical analysis, this study is subject to several limitations that point to valuable avenues for future research.

First, the reliance on provincial-level panel data constrains the ability to uncover micro-level mechanisms. Enterprise-level data could provide a more nuanced understanding of how technological innovation, industrial agglomeration, and GVC interact to drive upgrading. Future studies should incorporate firm-level data to establish stronger causal claims and elucidate the underlying transmission mechanisms with greater precision.

Second, this study acknowledges the limitations associated with measuring technological innovation. Although our composite proxy is grounded in established literature, more direct indicators—such as sector-specific R&D expenditures, patents, or R&D personnel within the automotive industry—would offer a finer-grained measure. Such granular data are not readily available at the provincial level in China. Therefore, while our measure provides meaningful insights, it may not completely capture the multifaceted nature of technological innovation within the automotive sector. Future research that gains access to disaggregated industry-level data could significantly improve the construct validity of innovation measurement and further verify the robustness of the relationships identified in this study.

Third, this study faces limitations in data coverage and measurement. The analysis does not include data from the provinces of Qinghai, Ningxia, and Tibet, primarily due to the limited availability of data from these regions. Additionally, the measurement of the global value chain (GVC) position relies on the OECD-TIVA database, which presents two constraints: the data are only updated until 2020, thus excluding recent developments; and the robustness of the GVC position index across different statistical standards (e.g., WIOD vs. domestic Chinese input-output tables) could not be tested due to data availability issues (e.g., the public WIOD data ends in 2014, and a continuous set of Chinese provincial input-output tables is unavailable). Future research could aim to address these gaps as more comprehensive and updated databases become accessible.

Fourth, the three-year averaging approach, while methodologically advantageous for mitigating instrument proliferation and recovering theoretically consistent dynamic relationships, inevitably smooths out short-term fluctuations and annual policy shocks. Our comparative analysis reveals that while this approach sacrifices some temporal granularity, it is essential for properly identifying the persistent dynamics of industrial upgrading. Subsequent studies might explore alternative methodologies—such as mixed-frequency models or dynamic factor models—to better capture both short-term variations and medium-term trends in industrial upgrading processes.

Author Contributions

Conceptualization, T.S.; methodology, T.S. and M.A.b.A.; software, T.S.; validation, T.S.; formal analysis, T.S. and M.A.b.A.; investigation, T.S.; resources, T.S.; data curation, T.S.; writing—original draft preparation, T.S.; writing—review and editing, T.S. and M.A.b.A.; visualization, T.S.; supervision, M.A.b.A.; project administration, M.A.b.A. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available upon reasonable request from the corresponding author.

Acknowledgments

The authors express their gratitude to QBU and UNIMAS for the research facilities provided. We also sincerely appreciate the valuable feedback from the anonymous reviewer.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:

DEA	Data Envelopment Analysis
GMM	Generalized Method of Moments
IU	Industrial Upgrading
GVC	Global Value Chain
TI	Technological Innovation
IA	Industrial Agglomeration
FDI	Foreign Direct Investment
TFP	Total Factor Productivity
R&D	Research and Development
GDP	Gross Domestic Product
LQ	Location Quotient
GI	Government Intervention
OPEN	Degree of Openness

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1 Conceptual Framework.

Figure 2 Data Analysis and Empirical Modeling Workflow.

Figure 3 The mechanism map of the impact of technological innovation on industrial upgrading.

Figure 4 Annual means of total factor productivity change and its decomposition in China’s automotive industry (2000–2020). Note: The figure plots the annual indices of total factor productivity change (Tfpch) and its decomposition into efficiency change (Effch) and technological change (Techch). The mean values for the entire period are Tfpch = 1.063, Techch = 1.034, and Effch = 1.028.

Figure 5 Mean values of total factor productivity change (TFPch) in China’s provincial automotive industry. Note: The figure displays the mean TFP change (Tfpch) for each province. Notable high-performers (e.g., Beijing) and low-performers (e.g., Gansu) are identified, illustrating the significant regional disparities discussed in the analysis.

Table 1

Comparison with Existing Literature on Technological Innovation, GVC, Agglomeration, and Industrial Upgrading.

Study	Context/Industry	Key Focus	Methodology	Key Limitations/Gaps	How the Present Study Builds Upon/Diverges
Jiang and Zhao (2020) [21], Xu et al. (2024) [23]	Cross-industry, China (provincial)	Direct Effect of Technological Innovation on Industrial Upgrading	Macro-econometric analysis (Panel data)	Lacks sector-specific mechanisms; Does not account for mediating/moderating variables.	Narrow focus on the automotive industry and test an integrated model with mediation (GVC) and moderation (Agglomeration).
Rísquez and Ruiz-Gálvez (2024) [35]	Automotive Industry (Global)	GVC Governance and Network Structures	Qualitative	Does not quantitatively test the mediating role of GVC positioning in the innovation-upgrading link.	Posits and empirically tests GVC positioning as a mediator between technological innovation and industrial upgrading.
Lampón et al. (2025) [36]	Automotive Industry (Transition)	Impact of new technologies on GVC Governance	Comparative quantitative analysis	Focuses on governance change; does not model agglomeration as a moderator	Integrates the moderating role of industrial agglomeration into a quantitative model, capturing how geographic context amplifies the effects of innovation.
Fu (2023) [34]	Manufacturing, China (Industry-level)	Impact of GVC embedding on manufacturing upgrading and trade gains	Macro-econometric analysis (Industry-level)	Focuses on GVC as a direct driver, not as a mediator; Does not investigate the antecedent role of technological innovation or the contingent role of agglomeration.	Theorizes GVC as a mediating pathway through which technological innovation affects upgrading, and tests a dual-mechanism model that includes agglomeration as a moderator.
Marshall (1890) [38], Porter (1990) [40]	General Theory	Agglomeration externalities and innovation	Theoretical/Conceptual	High-level theory; Lacks empirical testing in the context of modern, globalized value chains and their interaction with innovation for upgrading.	Operationalizes agglomeration theory by quantitatively testing its moderating role within a GVC-mediated innovation model for a specific industry.

Table 2

List of Variables, Abbreviations, and Data Sources.

Variable Properties	Variable Name	Abbreviation	Data Source
Dependent Variable	Industrial Upgrading	IU	China Industrial Statistical Yearbook; China Automotive Industry Yearbook
Independent Variable	Technological Innovation	TI	China Statistical Yearbook;China Automotive Industry Yearbook
Mediating Variable	Global Value Chain	GVC	OECD-TIVA database 2023ed;China Automotive Industry Yearbook;China Statistical Yearbook
Moderating Variable	Industry Agglomeration	IA	China Automotive Industry Yearbook;China Industrial Statistical Yearbook;China Statistical Yearbook
Control Variables	Degree of Openness	OPEN	China Statistical Yearbook
	Foreign Direct Investment	FDI	China Statistical Yearbook;China Automotive Industry Yearbook
	Government Intervention	GI	China Statistical Yearbook

Table 3

Descriptive statistics.

Variable	Type	Mean	Std.dev	Min	Max	Obs
IU	Dependent Variable	3.0704	2.3097	0.2768	15.1788	196
TI	Independent Variable	2.5193	5.1349	0.0006	29.2029	196
GVC	Mediating Variable	0.5830	0.2819	0.0132	1.3025	196
IA	Moderating Variable	1.1627	1.4262	0.0119	8.5466	196
OPEN	Control Variable	0.3170	0.3746	0.0305	1.67560	196
FDI	Control Variable	2.8040	4.4684	0.0002	27.3320	196
GI	Control Variable	0.1937	0.0770	0.0757	0.4575	196

Table 4

Baseline analysis results.

Variable	Model 1
	IU
L1.IU	0.8847 ***
	(0.0239)
TI	0.0343 ***
	(0.0091)
OPEN	0.2859 **
	(0.1100)
FDI	0.0200 *
	(0.0114)
GI	1.4626 ***
	(0.4702)
Constant	0.2897 *
	(0.1457)
Number of groups	28
Number of instruments	25
AR (1)	0.051
AR (2)	0.299
Hansen	0.541

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01), ** (p < 0.05) and * (p < 0.10).

Table 5

Mediating effect results.

Variable	GVC	IU
	Column (1)	Column (2)
L1.IU		0.8460 ***
		(0.0245)
L1.GVC	−0.6585 ***
	(0.1399)
TI	0.0064 **	0.0611 ***
	(0.0026)	(0.0136)
GVC		1.4568 ***
		(0.4090)
OPEN	−0.2105	0.7428 ***
	(0.1275)	(0.1884)
FDI	0.0206 ***	−0.0484 *
	(0.0047)	(0.0241)
GI	−0.1126	2.4186 ***
	(0.3739)	(0.4024)
Constant	0.7941 ***	−0.6519 **
	(0.1449)	(0.2357)
Number of groups	28	28
Number of instruments	12	25
AR (1)	0.082	0.068
AR (2)	0.913	0.301
Hansen	0.303	0.478

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01), ** (p < 0.05) and * (p < 0.10).

Table 6

Moderating effect results.

Variable	Model 3
	IU
L1.IU	0.5999 ***
	(0.0308)
TI	0.2204 ***
	(0.0127)
TI_IA	0.0530 ***
	(0.0083)
IA	0.3720 ***
	(0.0439)
OPEN	2.4421 ***
	(0.1560)
FDI	−0.3760 ***
	(0.0254)
GI	5.8154 ***
	(0.3846)
Constant	−0.4988 ***
	(0.1348)
Number of groups	28
Number of instruments	27
AR (1)	0.040
AR (2)	0.562
Hansen	0.152

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01).

Table 7

(a) Robustness to the Exclusion of Potentially Endogenous Controls. (b) Robustness to Estimation Techniques and Model Specification.

(a)
Variable	Baseline (All Controls)		Excluding OPEN	Excluding FDI	Excluding GI
	(1)		(2)	(3)	(4)
L1.IU	0.8847 ***		0.9186 ***	0.9596 ***	0.9499 ***
	(0.0239)		(0.0400)	(0.0565)	(0.0519)
TI	0.0343 ***		0.0311 **	0.0475 ***	0.0489 ***
	(0.0091)		(0.0137)	(0.0157)	(0.0157)
OPEN	0.2859 **
	(0.1100)
FDI	0.0200 *		0.0345 **
	(0.0114)		(0.0168)
GI	1.4626 ***		0.6152	−0.0016
	(0.4702)		(0.8253)	(1.1092)
Constant	0.2897 *		0.4128 **	0.4311 *	0.4500 ***
	(0.1457)		(0.1623)	(0.2151)	(0.1523)
Number of groups	28		28	28	28
Number of instruments	25		17	12	11
AR (1)	0.051		0.049	0.035	0.039
AR (2)	0.299		0.289	0.309	0.309
Hansen	0.541		0.366	0.189	0.191
(b)
Variable	TFP	TFP	TFP	TFP	Profit Margin
	(1) SYS-GMM	(2) Fixed Effects	(3) Random Effects	(4) DIFF-GMM	(5) SYS-GMM
L1.IU	0.9489 ***			0.4916 ***	0.2562 **
	(0.0268)			(0.0535)	(0.1084)
TI	0.0454 ***	0.1469 ***	0.1673 ***	0.0991 ***	0.0038 ***
	(0.0151)	(0.0557)	(0.0503)	(0.0248)	(0.0007)
OPEN	0.4804 ***	0.4684	0.2839	−0.3138	0.0424 ***
	(0.1555)	(1.0433)	(0.6258)	(0.5861)	(0.0107)
FDI	0.0378 **	0.0396	−0.0003	−0.0840 **	0.0062 ***
	(0.0179)	(0.0689)	(0.0653)	(0.0340)	(0.0010)
GI	1.0623	12.3053 ***	12.4517 ***	5.7419 ***	−0.0885 *
	(0.6320)	(2.0930)	(1.9066)	(0.8076)	(0.0497)
HC	0.3229 **			0.1332 ***
	(0.1292)			(0.0357)
Constant	2.8960 ***	0.0572	0.1479		0.0349 ***
	(0.9606)	(0.5316)	(0.5255)		(0.0096)
Number of groups	28			28	28
Number of instruments	25			19	16
AR (1)	0.057			0.076	0.010
AR (2)	0.281			0.253	0.664
Hansen	0.296			0.115	0.107
Hausman Test			0.3245

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01), ** (p < 0.05) and * (p < 0.10).

Table 8

Robustness to Sample Selection Bias and Outliers.

Variable	IU	IU
	(1) Western Excluded	(2) Winsorized Sample
L1.IU	0.9441 ***	0.8438 ***
	(0.0425)	(0.0398)
TI	0.0637 ***	0.0633 ***
	(0.0168)	(0.0171)
OPEN	0.7080 ***	0.7256 ***
	(0.1202)	(0.1836)
FDI	−0.0773 ***	−0.0845 ***
	(0.0213)	(0.0246)
GI	2.0791	2.9173 ***
	(1.6832)	(0.8623)
Constant	−0.0029	0.0549
	(0.3116)	(0.2543)
Number of groups	19	28
Number of instruments	14	17
AR (1)	0.043	0.030
AR (2)	0.591	0.417
Hansen	0.449	0.290

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01).

Table 9

Comparison of System GMM Estimates: Triennial vs. Annual Data.

Variable	IU	IU
	(1) Triennial Model	(2) Annual Model
L1.IU	0.8847 ***	−0.0620 ***
	(0.0239)	(0.0142)
TI	0.0343 ***	0.0082 ***
	(0.0091)	(0.0011)
OPEN	0.2859 **	0.0384
	(0.1100)	(0.0404)
FDI	0.0200 *	−0.0241 ***
	(0.0114)	(0.0026)
GI	1.4626 ***	−0.3703 ***
	(0.4702)	(0.1171)
Constant	0.2897 *	1.3047
	(0.1457)	(0.0380)
Number of groups	28	28
Number of instruments	25	24
AR (1)	0.051	0.028
AR (2)	0.299	0.888
Hansen	0.541	0.142

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01), ** (p < 0.05) and * (p < 0.10).

Table 10

Robustness to Reverse Causality Using Lagged Innovation.

Variable	GVC	IU	IU
	(1)	(2)	(3)
L1.IU		0.6784 ***	0.9009 ***
		(0.0553)	(0.0321)
L1.GVC	−0.4939 ***
	(0.0716)
L1.TI	0.0054 ***	0.1097 ***	0.0487 ***
	(0.0016)	(0.0264)	(0.0114)
GVC		4.8976 ***
		(1.2365)
OPEN	−0.2186	3.0408 ***	0.3693 **
	(0.1334)	(0.7237)	(0.1729)
FDI	0.0238 ***	−0.1696 ***	−0.0204 *
	(0.0041)	(0.0385)	(0.0113)
GI	−0.1555	5.7441 ***	0.9962
	(0.3835)	(1.2742)	(0.7944)
Constant	0.8340 ***	−3.2778 ***	0.4235 **
	(0.1449)	(0.6561)	(0.2034)
Number of groups	28	28	28
Number of instruments	12	17	21
AR (1)	0.053	0.074	0.033
AR (2)	0.982	0.340	0.322
Hansen	0.382	0.101	0.298

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01), ** (p < 0.05) and * (p < 0.10).

Table 11

Robustness to Alternative Measures of Industrial Agglomeration.

Variable	LQ	LQ_dynamic	Industrial Share
	(1)	(2)	(3)
L1.IU	0.5999 ***	0.8547 ***	0.6552 ***
	(0.0308)	(0.0165)	(0.0169)
TI	0.2204 ***	0.1879 ***	0.0667 ***
	(0.0127)	(0.0099)	(0.0217)
TI_IA	0.0530 ***	0.1566 ***	0.7963 ***
	(0.0083)	(0.0142)	(0.1524)
IA	0.3720 ***	0.9324 ***	18.3866 ***
	(0.0439)	(0.0156)	(2.0710)
OPEN	2.4421 ***	1.2360 ***	1.8690 ***
	(0.1560)	(0.0767)	(0.2355)
FDI	−0.3760 ***	−0.2079 ***	−0.2610 ***
	(0.0254)	(0.0160)	(0.0274)
GI	5.8154 ***	2.7129 ***	4.5952 ***
	(0.3846)	(0.2482)	(0.7778)
Constant	−0.4988 ***	−0.2094 *	−0.6839 ***
	(0.1348)	(0.1057)	(0.2027)
Number of groups	28	28	28
Number of instruments	27	27	27
AR (1)	0.040	0.085	0.083
AR (2)	0.562	0.316	0.388
Hansen	0.152	0.183	0.275

Notes: 1. Standard errors are provided in parentheses. 2. Statistical significance is marked by *** (p < 0.01), and * (p < 0.10).

References

1. Kano, L.; Tsang, E.W.; Yeung, H.W.-c. Global value chains: A review of the multi-disciplinary literature. J. Int. Bus. Stud.; 2020; 51, pp. 577-622. [DOI: https://dx.doi.org/10.1057/s41267-020-00304-2]

2. Cao, X.; Furuoka, F.; Rasiah, R. Knowledge mapping of industrial upgrading research: A visual analysis using CiteSpace. Sustainability; 2023; 15, 16547. [DOI: https://dx.doi.org/10.3390/su152416547]

3. Tang, Y. Research on the Transformation and Upgrading of China’s Automobile Industry from the Perspective of Global Value Chain. Proceedings of the International Academic Conference on Frontiers in Social Sciences and Management Innovation (IAFSM 2019); Beijing, China, 28–29 December 2019; [DOI: https://dx.doi.org/10.2991/assehr.k.200207.064]

4. Liu, J.; Liu, G.; Cui, H.; Chen, Y. The Chinese automobile industry’s research and development capability and innovative performance. Front. Psychol.; 2023; 13, 1093305. [DOI: https://dx.doi.org/10.3389/fpsyg.2022.1093305] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36687958]

5. Cheng, G. Research on the pathways and mechanisms of technological innovation driving industrial upgrading from the perspective of new productivity. SHS Web Conf.; 2025; 213, 02051. [DOI: https://dx.doi.org/10.1051/shsconf/202521302051]

6. Lee, K.; Qu, D.; Mao, Z. Global value chains, industrial policy, and industrial upgrading: Automotive sectors in Malaysia, Thailand, and China in comparison with Korea. Eur. J. Dev. Res.; 2021; 33, pp. 275-303. [DOI: https://dx.doi.org/10.1057/s41287-020-00354-0]

7. Eissa, Y.; Zaki, C. On GVC and innovation: The moderating role of policy. J. Ind. Bus. Econ.; 2023; 50, pp. 49-71. [DOI: https://dx.doi.org/10.1007/s40812-022-00255-9]

8. Zhang, J.; Yang, Z.; He, B. Does digital infrastructure improve urban economic resilience? Evidence from the Yangtze River Economic Belt in China. Sustainability; 2023; 15, 14289. [DOI: https://dx.doi.org/10.3390/su151914289]

9. Yang, N.; Liu, Q.; Chen, Y. Does industrial agglomeration promote regional innovation convergence in China? Evidence from high-tech industries. IEEE Trans. Eng. Manag.; 2023; 70, pp. 1416-1429. [DOI: https://dx.doi.org/10.1109/TEM.2021.3084936]

10. Ding, Y.; Zhuang, Y.; Zhang, L.; Yang, G. Industrial policy and potential for export upgrading: Evidence from government-approved projects in China. Econ. Model.; 2024; 133, 106668. [DOI: https://dx.doi.org/10.1016/j.econmod.2024.106668]

11. Tang, L.; Zhang, Y.; Gao, J.; Wang, F. Technological Upgrading in Chinese Cities: The role of FDI and industrial structure. Emerg. Mark. Financ. Trade; 2020; 56, pp. 1547-1563. [DOI: https://dx.doi.org/10.1080/1540496X.2018.1562900]

12. He, B.; Tian, S.; Zhang, X. Does the pilot free trade zone policy increase regional innovation ability? Evidence from China. Appl. Econ. Lett.; 2023; 32, pp. 576-581. [DOI: https://dx.doi.org/10.1080/13504851.2023.2276360]

13. Gao, X.; Li, C.; Elahi, E.; Abro, M.; Cui, Z. Technological innovation, product quality and upgrading of manufacturing value chain: Empirical evidence from China. Sustainability; 2023; 15, 7289. [DOI: https://dx.doi.org/10.3390/su15097289]

14. Lin, X.; Wang, P.; Liu, M. Micro-pathways to technology spillover in industrial agglomerations: Evidence from manufacturing enterprises in Guangdong Province. China Int. J.; 2024; 22, pp. 111-137. [DOI: https://dx.doi.org/10.56159/chn.2024.a920960]

15. Tian, X.; Chen, P. The interplay among energy utilisation efficiency and industrial agglomeration through spatial heterogeneity insights. Sci. Rep.; 2025; 15, 5020. [DOI: https://dx.doi.org/10.1038/s41598-025-89154-2]

16. Wang, F.; Wu, M.; Du, X. Does industrial upgrading improve eco-efficiency? Evidence from China’s industrial sector. Energy Econ.; 2023; 124, 106774. [DOI: https://dx.doi.org/10.1016/j.eneco.2023.106774]

17. Zhang, Q. The impact of digitalization on the upgrading of China’s manufacturing sector’s global value chains. J. Knowl. Econ.; 2024; 15, pp. 15577-15600. [DOI: https://dx.doi.org/10.1007/s13132-023-01698-8]

18. Fuller, D.; Ramirez, P. The limits to upgrading and value capture in R&D global value chains: Indian and Chinese contract R&D firms in the integrated circuit design and pharmaceutical global value chains. Comp. Change; 2023; 28, pp. 144-164. [DOI: https://dx.doi.org/10.1177/10245294231202417]

19. Verhoogen, E. Firm-Level Upgrading in Developing Countries; NBER Working Paper No. 29461 National Bureau of Economic Research: Cambridge, MA, USA, 2021; [DOI: https://dx.doi.org/10.3386/w29461]

20. Cai, B.; Yang, J.; Huang, G. Spatiotemporal dynamics in economic, social, and environmental upgrading in China: Coupling coordination and influencing factors. Sustainability; 2024; 16, 357. [DOI: https://dx.doi.org/10.3390/su16010357]

21. Jiang, J.; Zhao, X. A study of the impact of innovation on industrial upgrading in China: A spatial econometric analysis based on China’s provincial panel data. J. Adv. Comput. Intell. Intell. Inform.; 2020; 24, pp. 272-281. [DOI: https://dx.doi.org/10.20965/jaciii.2020.p0272]

22. Zou, T. Technological innovation promotes industrial upgrading: An analytical framework. Struct. Change Econ. Dyn.; 2024; 70, pp. 150-167. [DOI: https://dx.doi.org/10.1016/j.strueco.2024.01.012]

23. Xu, L.; Shu, H.; Lu, X.; Li, T. Regional technological innovation and industrial upgrading in China: An analysis using interprovincial panel data from 2008 to 2020. Financ. Res. Lett.; 2024; 60, 105621. [DOI: https://dx.doi.org/10.1016/j.frl.2024.105621]

24. Ma, X.; Xu, X.; Wang, Z.; Zhao, X.; Lee, H.; Truskolaski, T. Technological innovation, industrial structure upgrading and mining energy efficiency: An analysis based on the super-efficient EBM model. Resour. Policy; 2024; 98, 105339. [DOI: https://dx.doi.org/10.1016/j.resourpol.2024.105339]

25. Dai, D.; Zheng, Y. The new quality productive force, science and technology innovation, and optimization of industrial structure. Sustainability; 2025; 17, 4439. [DOI: https://dx.doi.org/10.3390/su17104439]

26. American Automotive Policy Council. Leading in Capital Investments and Research & Development: 2021 AAPC Scoreboard on Auto CapEx and R&D. Available online: https://www.americanautomakers.org/sites/default/files/2021 AAPC Scoreboard on Auto CapEx and R%26D.pdf (accessed on 15 August 2025).

27. Qiao, Q.; Zhao, F.; Liu, Z.; Hao, H.; He, X.; Przesmitzki, S.V.; Amer, A.A. Life cycle cost and GHG emission benefits of electric vehicles in China. Transp. Res. Part D Transp. Environ.; 2020; 86, 102418. [DOI: https://dx.doi.org/10.1016/j.trd.2020.102418]

28. Choi, H.; Kim, S.; Jung, T. The role of innovation in upgrading in global value chains. Glob. Econ. Rev.; 2019; 48, pp. 273-283. [DOI: https://dx.doi.org/10.1080/1226508X.2019.1636703]

29. Tian, K.; Dietzenbacher, E.; Jong-A-Pin, R. Global value chain participation and its impact on industrial upgrading. World Econ.; 2022; 45, pp. 1362-1385. [DOI: https://dx.doi.org/10.1111/twec.13209]

30. Stojčić, N.; Matić, M. A journey toward global value chain upgrading: Exploring the transition from backward to forward integration. Technol. Soc.; 2024; 76, 102435. [DOI: https://dx.doi.org/10.1016/j.techsoc.2023.102435]

31. Tian, K.; Dietzenbacher, E.; Jong-A-Pin, R. Measuring industrial upgrading: Applying factor analysis in a global value chain framework. Econ. Syst. Res.; 2019; 31, pp. 642-664. [DOI: https://dx.doi.org/10.1080/09535314.2019.1610728]

32. Murphree, M.; Breznitz, D. Collaborative public spaces and upgrading through global value chains: The case of Dongguan, China. Glob. Strategy J.; 2020; 10, pp. 556-584. [DOI: https://dx.doi.org/10.1002/gsj.1378]

33. Li, W.; Li, Q.; Chen, M.; Su, Y.; Zhu, J. Global value chains, digital economy, and upgrading of China’s manufacturing industry. Sustainability; 2023; 15, 8003. [DOI: https://dx.doi.org/10.3390/su15108003]

34. Fu, Q. The impact of global value chain embedding on the upgrading of China’s manufacturing industry. Front. Energy Res.; 2023; 11, 1256317. [DOI: https://dx.doi.org/10.3389/fenrg.2023.1256317]

35. Rísquez, M.R.; Ruiz-Gálvez, M.E. The transformation of the automotive industry toward electrification and its impact on global value chains: Inter-plant competition, employment, and supply chains. Eur. Res. Manag. Bus. Econ.; 2024; 30, 100242. [DOI: https://dx.doi.org/10.1016/j.iedeen.2024.100242]

36. Lampón, J.F.; Carballo-Cruz, F.; Velando-Rodríguez, M.-E.; Cabanelas, P. A transition in the automotive industry: How are new mobility technologies affecting value chain governance?. Kybernetes; 2025; [DOI: https://dx.doi.org/10.1108/K-12-2024-3383]

37. Yi, K. Explore the influencing factors of manufacturing industry agglomeration. Adv. Econ. Manag. Polit. Sci.; 2023; 28, pp. 91-96. [DOI: https://dx.doi.org/10.54254/2754-1169/28/20231292]

38. Marshall, A. Principles of Economics; Palgrave Macmillan: New York, NY, USA, 1890; pp. 116-262.

39. Wu, M.; Huang, W. Agglomeration and green technology innovation efficiency of industrial enterprises: Based on spatial statistical analysis. Arch. Environ. Prot.; 2022; 48, pp. 3-14. [DOI: https://dx.doi.org/10.24425/aep.2022.140761]

40. Porter, M.E. The Competitive Advantage of Nations; Free Press: New York, NY, USA, 1990; pp. 80-91.

41. Glaeser, E.L. Agglomeration Economics; University of Chicago Press: Chicago, IL, USA, 2010; pp. 277-302.

42. Arcos-Guanga, A.; Flor-Unda, O.; Novillo-Villegas, S.; Acosta-Vargas, P. The impact of knowledge spillovers on economic growth from a national perspective: A comprehensive analysis. Sustainability; 2024; 16, 6537. [DOI: https://dx.doi.org/10.3390/su16156537]

43. Huang, Q.H.; Hu, M.J.; Hu, J.F. Competition or cooperation? The impacts of coopetition strategy on the industrial production efficiency in the Chengdu-Chongqing economic circle. J. Chongqing Univ. Technol.; 2020; 11, pp. 5-18. [DOI: https://dx.doi.org/10.3969/j.issn.1674-8425(s).2020.11.002]

44. Shahid, R.; Shahzad, K.; Shahid, H. How does environmental protection law affect industrial upgrading of Chinese firms: The mediating role of technological innovation?. Asian J. Econ. Bus.; 2025; 6, pp. 45-73. [DOI: https://dx.doi.org/10.47509/AJEB.2025.v06i01.03]

45. Xu, Y. Research on innovative city construction promoting urban industry upgrading. J. Technol. Econ. Manag.; 2020; 11, pp. 110-114.

46. Agostino, M.; Nifo, A.; Ruberto, S.; Scalera, D.; Trivieri, F. Productivity changes in the automotive industry of three European countries. An application of the Malmquist index decomposition analysis. Struct. Change Econ. Dyn.; 2022; 61, pp. 216-226. [DOI: https://dx.doi.org/10.1016/j.strueco.2022.02.017]

47. Liu, X.; Liu, H.; Yang, G.; Pan, J. Productivity assessment of the real estate industry in China: A DEA-Malmquist index. Eng. Constr. Archit. Manag.; 2023; 30, pp. 1243-1270. [DOI: https://dx.doi.org/10.1108/ECAM-07-2020-0561]

48. Jiang, H.; Han, L.; Ding, Y.; He, Y. Operating efficiency evaluation of China listed automotive firms: 2012–2016. Sustainability; 2018; 10, 184. [DOI: https://dx.doi.org/10.3390/su10010184]

49. Qu, L.; Meng, X. Research on the high-quality development of automobile manufacturing industry in the Beijing-Tianjin-Hebei Region from the perspective of total factor productivity. Study Explor.; 2022; 7, pp. 94-101.

50. Fu, S. Analysis and research on total factor productivity of Ningxia manufacturing based on DEA-Malmquist index method. Sci. Technol. Ind.; 2020; 20, pp. 1-7.

51. Lu, Z.; Wen, T. Research on the impact of R&D investment on total factor productivity of automobile manufacturing industry. J. Guangxi Univ. Sci. Technol.; 2018; 29, pp. 106-112.

52. Du, C.; Liu, S. Effect measurement and path analysis of total factor productivity in China’s manufacturing industry empowered by the digital economy. Res. Econ. Manag.; 2023; 44, pp. 43-65. [DOI: https://dx.doi.org/10.13502/j.cnki.issn1000-7636.2023.09.003]

53. Yu, H.; Shen, K.; Sun, L. A Study on impact of VAT reduction on regional manufacturing productivity. Financ. Trade Res.; 2023; 34, pp. 45-56. [DOI: https://dx.doi.org/10.19337/j.cnki.34-1093/f.2023.06.005]

54. Wang, D.; Wang, J.; Zhou, Y. The impact of fiscal and tax incentives on the digital transformation of enterprises. Chin. J. Manag.; 2024; 46, pp. 64-76.

55. Shen, G. Foreign direct investment, opening-up and quality of economic growth: Based on the empirical evidence of China’s auto industry. Inq. Econ. Issues; 2017; 38, pp. 113-122.

56. Koopman, R.; Powers, W.; Wang, Z.; Wei, S.-J. Give Credit Where Credit Is Due: Tracing Value Added in Global Production Chains; Working Paper No. 16426 National Bureau of Economic Research: Cambridge, MA, USA, 2010; [DOI: https://dx.doi.org/10.3386/w16426]

57. Yu, Z.; Gu, H. Global value chain position and regional industry upgrading policies: Taking Northeast China as an example. Sci. Geogr. Sin.; 2016; 36, pp. 1371-1377. [DOI: https://dx.doi.org/10.13249/j.cnki.sgs.2016.09.010]

58. Lopez Gonzalez, J.; Kowalski, P.; Achard, P. Trade, Global Value Chains and Wage-Income Inequality; OECD Trade Policy Papers No. 182 OECD Publishing: Paris, France, 2015; [DOI: https://dx.doi.org/10.1787/5js009mzrqd4-en]

59. Liu, X.; Wu, Y. Export specialization, export diversification and regional economic growth: An empirical study from China’s provincial panel data. J. Manag. World; 2013; 8, pp. 30-40. [DOI: https://dx.doi.org/10.19744/j.cnki.11-1235/f.2013.08.004]

60. Zhang, X.; Pan, Z.; Tang, D.; Deng, Z.; Boamah, V. Impact of environmental regulation and industrial agglomeration on carbon emissions in the Yangtze River Economic Belt. Sustainability; 2023; 15, 8308. [DOI: https://dx.doi.org/10.3390/su15108308]

61. Lee, C.W.; Rabago, A.R.A. Trade openness and economic growth: Evidence from countries in Asia and the South Pacific. Adv. Manag. Appl. Econ.; 2024; 14, pp. 165-178. [DOI: https://dx.doi.org/10.47260/amae/1438]

62. Blancheton, B.; Chhorn, D. Government intervention, institutional quality, and income inequality: Evidence from Asia and the Pacific, 1988–2014. Asian Dev. Rev.; 2021; 38, pp. 176-206. [DOI: https://dx.doi.org/10.1162/adev_a_00162]

63. Roodman, D. How to do xtabond2: An introduction to difference and system GMM in Stata. Stata J.; 2009; 9, pp. 86-136. [DOI: https://dx.doi.org/10.1177/1536867X0900900106]

64. Yang, S.; Wang, J.; Dong, K.; Jiang, Q. A path towards China’s energy justice: How does digital technology innovation bring about a just revolution?. Energy Econ.; 2023; 127 Pt A, 107056. [DOI: https://dx.doi.org/10.1016/j.eneco.2023.107056]

65. Rotabi, S.; Abazi-Alili, H.; Noveski, V.A. The relationship between technological innovation and production efficiency in manufacturing sector across economies. Int. J. Technoentrepreneurship; 2024; 5, pp. 163-177. [DOI: https://dx.doi.org/10.1504/IJTE.2024.142850]

66. Kyrylenko, V.I.; Tkachenko, O.V.; Kotenok, A.G.; Sarkisian, H.O.; Sitkovska, A.O. The role of innovative technologies in ensuring the competitiveness of the enterprise. Nauk. Visnyk Natsionalnoho Hirnychoho Universytetu; 2022; 5, pp. 185-190. [DOI: https://dx.doi.org/10.33271/nvngu/2022-5/185]

67. Marčeta, M.; Bojnec, Š. Trade openness, global competitiveness, and catching up between the European Union countries. Rev. Int. Bus. Strategy; 2023; 33, pp. 691-714. [DOI: https://dx.doi.org/10.1108/RIBS-12-2021-0155]

68. Yue, W. Foreign direct investment and the innovation performance of local enterprises. Humanit. Soc. Sci. Commun.; 2022; 9, 252. [DOI: https://dx.doi.org/10.1057/s41599-022-01274-6]

69. Xu, R.; Shen, Y.; Liu, M.; Li, L.; Xia, X.; Luo, K. Can government subsidies improve innovation performance? Evidence from Chinese listed companies. Econ. Model.; 2023; 120, 106151. [DOI: https://dx.doi.org/10.1016/j.econmod.2022.106151]

70. Feng, B.; Sun, K.; Chen, M.; Gao, T. The impact of core technological capabilities of high-tech industry on sustainable competitive advantage. Sustainability; 2020; 12, 2980. [DOI: https://dx.doi.org/10.3390/su12072980]

71. Rabaia, H. The impact of technological innovation on competition strategies in the technological sector. Pak. J. Life Soc. Sci.; 2024; 22, pp. 11184-11198. [DOI: https://dx.doi.org/10.57239/PJLSS-2024-22.2.00847]

72. Audretsch, D.B.; Belitski, M. The knowledge spillover of innovation. Ind. Corp. Change; 2022; 31, pp. 1329-1357. [DOI: https://dx.doi.org/10.1093/icc/dtac035]

73. McWilliam, S.E.; Nielsen, B.B. Global value chains and development: Redefining the contours of 21st century capitalism. J. Int. Bus. Stud.; 2020; 51, pp. 1347-1350. [DOI: https://dx.doi.org/10.1057/s41267-020-00303-3]

74. Cui, R.; Song, H.; Li, D. Global value chain embeddedness, technology spillover and enterprise innovation. Int. Rev. Econ. Financ.; 2024; 93, pp. 758-771. [DOI: https://dx.doi.org/10.1016/j.iref.2024.03.004]

75. Ye, X.; Li, X.; Tao, C.; Zhou, X. Connected knowledge spillovers, technological cluster innovation and efficient industrial structure. J. Innov. Knowl.; 2022; 7, 100195. [DOI: https://dx.doi.org/10.1016/j.jik.2022.100195]

76. Tian, J.; Tian, Y.; Yi, H. The influence of industrial Synergy Agglomeration on economic resilience: Based on the comparison between Yangtze River Delta Urban Agglomeration and Central Yunnan Urban Agglomeration. J. Innov. Dev.; 2023; 2, pp. 70-77. [DOI: https://dx.doi.org/10.54097/jid.v2i3.7279]

77. Li, J.; Xing, J. Why is collaborative agglomeration of innovation so important for improving regional innovation capabilities? A perspective based on collaborative agglomeration of industry-university-research institution. Complexity; 2020; 2020, 7049606. [DOI: https://dx.doi.org/10.1155/2020/7049606]

78. Sheng, F. The spatial agglomeration of producer services industry and the upgrading of manufacturing industry: Mechanism and experience—A spatial econometric analysis based on the data from 230 cities. Ind. Econ. Res.; 2014; 2, pp. 32-39. [DOI: https://dx.doi.org/10.13269/j.cnki.ier.2014.02.004]

79. Aytekin, A.; Ecer, F.; Korucuk, S.; Karamaşa, Ç. Global innovation efficiency assessment of EU member and candidate countries via DEA-EATWIOS multi-criteria methodology. Technol. Soc.; 2022; 68, 101896. [DOI: https://dx.doi.org/10.1016/j.techsoc.2022.101896]

80. Thakur, R.; Sharma, S. Does global value chain participation and position lead to innovation in an emerging economy? Industry-level evidence from Indian manufacturing industries. J. Bus. Res.; 2025; 186, 114989. [DOI: https://dx.doi.org/10.1016/j.jbusres.2024.114989]